polarized light in animal vision: polarization patterns in nature

Polarized Light in Animal Vision

Springer-Verlag Berlin Heidelberg GmbH

Gabor Horveith Dezso Varju

Polarized Light in Animal Vision Polarization Patterns in Nature

With 127 Figures, 16 Plates in Colour

Springer

Dr. habil. GABOR HORV ATH

Department of Biological Physics Eötvös University Pazmany setany 1 H -1117 Budapest Hungary

e-mail: [email protected]

Prof. Dr. DEZSO VARJU

Lehrstuhl Kognitive Neurowissenschaften Universität Tübingen Auf der MorgensteIle 28 72076 Tübingen Germany

e-mail: [email protected]

Cover: Background: Pattern of the angle of linear polarization a of skylight and earthlight displayed on the surface of a sphere and measured by 1800 field-of-view imaging polarimetry in the blue part (450 nm) of the spectrum from a hot air balloon at an altitude of 3500 m. The colour code of a is given in -. colour Fig. 4.5. More details can be found in Chap. 4.2. Foreground: Collection of some representative polarization-sensitive animal species (dragonfly Anax imperator, house cricket Acheta domesticus, red-spotted newt Notophthalmus viridescens, spider Pardosa lugubris and rainbow trout Oncorhynchus mykiss), the polarization sensitivity of which is treated in Part III of this volume.

All figures in this volume were composed by Dr. Gabor Horvath

ISBN 978-3-642-07334-2 ISBN 978-3-662-09387-0 (eBook) DOI 10.1007/978-3-662-09387-0

Library of Congress Cataloging-in-Publication Data

Horv>lth, Gabor, 1963-Polarized light in animal vision: polarization patterns in nature / Gabor Horvath, Dezso Varju.

p.cm. Includes bibliographical references (p.).

1. Vision. 2. Polarization (Light)-Physiological aspects. 3. Animal orientation. 4. Physiology, Comparative. I. Varju, Dezso, 1932- II. Tide

QP481.H652003 152.14--dc22 2003054309

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permit -ted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permissions for use must always be obtained from Springer-Verlag. Violations are liable for prosecution under the German Copyright Law.

springeronline.com

© Springer-Verlag Berlin Heidelberg 2004 Originally published by Springer-Verlag Berlin Heidelberg New York in 2004. Softcover reprint of the hardcover I st edition 2004

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

Production and typesetting: Friedmut Kröner, 69115 Heidelberg, Germany Cover design: design & production GmbH, Heidelberg

3113150 YK - 5 4 3 2 1 0 - Printed on acid free paper

Preface

The subject of this volume is two-fold. First, it gathers typical polarization patterns occurring in nature. Second, it surveys the polarization-sensitive animals, the physiological mechanisms and biological functions of polarization sensitivity as weIl as the polarization-guided behaviour in animals. The monograph is prepared for biologists, physicists and meteorologists, especially for experts of atmospheric optics and animal vision, who wish to understand and reveal the message hidden in polarization patterns of the optical environment not directly accessible to the human visual system, but measurable by polarimetry and perceived by many animals. Our volume is an attempt to build a bridge between these two physical and biological flelds.

In Part I we introduce the reader to the elements of imaging polarimetry. This technique can be efflciently used, e.g. in atmospheric optics, remote sensing and biology.

In Part 11 we deal with typical polarization patterns of the natural optical environment. Sunrise/sunset, clear skies, cloudy skies, moonshine and total solar eclipses all mean quite different illumination conditions, wh ich also affect the spatial distribution and strength of celestial polarization. We present the polarization patterns of the sky and its unpolarized (neutral) points under sunlit, moonlit, clear, cloudy and eclipsed conditions as a function of solar elevation. The polarization pattern of a rainbow is also shown. That part of the spectrum is derived in which perception of skylight polarization is optimal under partly cloudy skies. The reader becomes acquainted with the polarization of the solar corona and can follow how the polarization pattern of the sky changed during a total solar eclipse. We also treat the polarizational characteristics of water surfaces, mirages and the underwater light fleld. We explain why water insects are not attracted by mirages. FinaIly, the occurrence of circularly polarized light in nature is reviewed.

Part III is devoted to the description of the visual and behavioural mechanisms indicating how animals perceive and use natural polarization patterns. Surveying the literature, a detailed compendium of the sensory basis of polarization sensitivity in animals and humans is given. We also present several case studies of known behavioural patterns determined or influenced by

VI Preface

polarization sensitivity. It is shown, for instance, how aerial, terrestrial and aquatic animals use celestial and underwater polarization for orientation. The role of the reflection-polarization pattern of water surfaces in water detection by insects is discussed. We illustrate how reflection-polarization patterns of anthropogeneous origin can deceive water-seeking polarotactic insects. The natural environment is more or less affected by human civilization and is overwhelmed by man-made objects, such as crude or waste oil surfaces, asphalt roads, glass surfaces, or plastic sheets used in agriculture, for instance. We explain why these surfaces are more attractive to water-seeking polarotactic insects than the water surface itself. We explain why mayflies or dragonflies lay their eggs en masse on dry asphalt roads or car-bodies. We show how dangerous open-air oil reservoirs can be for polarotactic insects and why oil surfaces function as insect traps. Some other possible biological functions of polarization sensitivity, such as contrast enhancement, intra- or interspecific visual communication and camouflage breaking are also discussed. Due to the interference of polarization and colour sensitivity, polarization-induced false colours could be perceived by polarization- and colour-sensitive visual systems. We calculate and visualize these false colours by means of a computer model of butterfly retinae, and investigate their chromatic diversity. Finally, a common methodological error is discussed, which is frequently committed in experiments studying animal polarization sensitivity.

Our monograph is in dose connection with the treatise about planets, stars and nebulae studied with photopolarimetry edited by T. Gehreis (1974), the volume on polarized light in nature by Günther P. Können (1985), and the mono graph of Kinsell L. Coulson (1988) on polarization and intensity oflight in the atmosphere. When these volumes were published, the technique of imaging polarimetry was not yet available, thus the polarizational characteristics of natural optical environments were presented in the form of graphs or pairs of photographs taken through linear polarizers with two orthogonal directions of their transmission axes.

Due to imaging polarimetry developed in the last decade, the polarization patterns are visualized in our volume as high resolution colour/grey-coded maps of the degree and angle of linear polarization. All colour figures are placed at the end of the book. They are cited in the text as e.g. ~ colour Fig.1.1.

Considering various kinds of point-source non-imaging polarimeters, induding radar polarimetry, the reader is referred to the monographs of Egan (1985), Kong (1990), Azzam and Bashara (1992), Boerner et al. (1992) and Collett (1994), for instance. All relevant details of the physics of light polarization can be found in the text-books of Shurdiff (1962), Clarke and Grainger (1971), Kliger et al. (1990), Born and Wolf (1999), for example. The early knowledge about the sensory basis of animal polarization sensitivity and its

Preface VII

biological functions was reviewed by Karl von Frisch (1967) and Talbot H. Waterman (1981). Rüdiger Wehner (1976, 1982, 1983, 1984, 1989, 1994,2001) also wrote several important reviews and essays ab out this topic, especially on honeybees and desert ants. In addition to relying on our own contributions to the field, we have liberally quoted from the numerous publications of many other investigators with appropriate references given in each case. While the bibliography at the end of our book is not complete, it is fairly representative of the field.

June 2003, Budapest Tübingen

Gabor Horvath Dezso Varju

Acknowledgements

Financial support for the authors to write the book was given by the German Alexander von Humboldt Fundation, providing Gabor Horvath with a Humboldt research fellowship for 14 months at the Department of Cognitive Neuroscience of the Eberhard Karls University in Tübingen, thus making dose co operation with Dezso Varju possible. Many thanks are due to Professor Hanspeter Mallot, head of the department, who provided us with all the necessary equipment and materials. We appreciate the secretarial and technieal assistance of Mrs. Annemarie Kehrer, Ursula Henique and Dr. Heinz Bendele. The three-year Istvan Szechenyi scholarship from the Hungarian Ministry of Education to G. Horvath is also acknowledged. G. Horvath received further financial support from the Hungarian Science Foundation (OTKA F-014923, T-020931, F-025826).

We are very grateful to the following scientists for reviewing different chapters (numbers in brackets) of the monograph: Kenneth Able (31), Marie Dacke (24), Miehael Freake (29,30), Uwe Homberg (17.6, 17.7), Almut Kelber (17.5), Kuno Kirschfeld (17.2), Thomas Labhart (17.4), Inigo Novales Flamarique and Ferenc Harosi (28), Samuel Rossel (17.1), Rudolf Schwind (18), Nadav Shashar (26), Rüdiger Wehner (17.3) and Hansruedi Wildermuth (18.6). Apart from these scientists, S. Akesson, E.J.H. Bechara, H.1. Browman, M.A.I.M. Coemans, T. Cronin, R.B. Forward, W. Haupt, C.W. Hawryshyn, G.W. Kattawar, G.P. Können, M.E Land, R.L. Lee, D.K. Lynch, E.P. Meyer, v.B. MeyerRochow, ER. Moore, U. Munro, D.E. Nilsson, J.EW. Nuboer, A. Ugolini, K.J. Voss, J.A. Waldvogel, T.H. Waterman, W. Wiltschko and I. Zeil provided us with important literature and information, whieh is acknowledged.

We are also grateful to the following students and colleagues for their continuous help during the polarimetrie investigations in the field and the evaluation and visualization of the polarization patterns: Andras Barta, Balazs Bernath, Ferenc Mizera, Gergely Molnar, Bence Suhai, Gabor Szedenies, Drs. Sandor Andrikovics, J 6zsef Gal, Ott6 Haiman, György Kriska and Istvan Pomozi.

The polarimetrie measurements in Finland, in the Tunisian desert and in Switzerland were possible due to fruitful co operation with Professors Viktor

x Acknowledgements

Benno Meyer-Rochow, Rüdiger Wehner and Hansruedi Wildermuth. Many thanks for their financial support, valuable help and encouragement.

Maria Horvath-Fischer and Janos Horvath rendered indispensable help and support during the field experiments in the Hungarian Great Plain (Kiskunhalas, Kecel and Kunfehert6).

All figures adopted from the cited sources are taken over in this volume with the permission of the publishers.

Many thanks to our wives, Heide Varju and Zsuzsanna Horvath-Tatar who ensured the ideal and quiet familiar atmosphere, which was one of the most important prerequisites of our work. We dedicate this mono graph to them.

Gabor Horvath is very grateful to Professor Rudolf Schwind, who introduced hirn into the wonderful world of polarized light and its role in animal vision during his one-year postdoctoral fellowship at the Institute of Zoology of the University of Regensburg in 1991-1992.

Gabor Horvath acknowledges the inspiring scientific atmosphere at the Department of Biological Physics of the Lorand Eötvös University in Budapest, and the continous support and encouragement of Professor Tamas Vicsek, the head of department.

Last, but not least, we are also very much indebted to Springer-Verlag, especially to Drs. Dieter Czeschlik and Jutta Lindenborn. Dr. Czeschlik agreed without hesitation to publish the book, and from Dr. Lindenborn we received valuable advice while preparing the manuscript.

About the Authors

Gabor Horvath was born in 1963 in Kiskunhalas, Hungary. In 1987 he received his diploma in physics from the Lorand Eötvös University in Budapest. Then he was a research assistant at the Department of Low Temperature Physics of the same university, where he investigated electrical percolation processes in granular superconductors. In 1989 he received a doctoral fellowship in the Biophysics Group of the Central Research Institute for Physics of the Hungarian Academy of Sciences (Budapest), where he developed a mathematical description and computer modelling of retinal cometlike afterimages. He obtained his Ph.D. at the Eötvös University in 1991. His thesis in physiological optics is a computational study of the visual system

XII About the Authors

and optical environment of certain animals. In 1991 he was offered a oneyear postdoctoral position in the Institute for Zoology of the University of Regensburg (Germany), where together with Professor Rudolf Schwind he star ted to study the polarization patterns of skylight reflected from water surfaces. Then he was a postdoctoral fellow at the Department for Biological Cybernetics of the University of Tübingen (Germany) for 1 year. Here, he investigated experimentally the polarization-sensitive optomotor re action in water insects and natural polarization patterns together with Professor Dezso Varju. In 1993 he finished his postdoctoral dissertation in computational visual optics to obtain the degree "Candidate for Biophysical Science" awarded by the Hungarian Academy of Sciences. For this treatise he won the first International Dennis Gabor Award. In 1993 together with Dezso Varju, he won also the biomathematical Richard Bellman Prize from the journal of Mathematical Biosciences. He also received several best paper awards of different Hungarian popular-scientific journals. He won the first prize of the Hungarian Biophysical Society three times. In 1994 he received the Pro Schola award from the Aron Szihidy secondary school, where he studied earlier. Presently he is an associate professor at the Department of Biological Physics of the Eötvös University and leader of the Biooptics Laboratory. He received the Hungarian Istvan Szechenyi (3 years), Lorand Eötvös (9 months), Janos Bolyai (3 years), ZoItan Magyary (1 year) scholarships and the German Alexander von Humboldt fellowship (14 months). His main research interest is studying experimentally as weIl as theoretically the optics of animal eyes, polarization sensitivity of animals and the polarizational characteristics of the optical environment. He developed different kinds of imaging polarimetry, by which he records and visualizes the polarization patterns in nature. He conducted several expeditions and polarimetric measuring campaigns in Hungary, in the Tunisian desert as weIl as in the Finnish Lapland. His wife, Zsuzsanna Tatar-Horvath teaches mathematics and physics in a secondary school in Budapest. His sons, Lorand and Lenard were born in 1991 and 1999, respectively.

About the Authors XIII

Dezso Varju was born in 1932 in Hungary. In 1956 he received his diploma in physics from the Lonind Eötvös University in Budapest. In the same year he left Hungary and joined as graduate student a group of biophysicists headed by the late Werner Reichardt at a Research Institute of the Max Planck Society in Göttingen, Germany. There he was involved in the investigation of movement perception in insects and of phototropic and light growth responses of the slime mold Phycomyces, on both experimental and theoreticallevels. In 1958 he received his Ph.D. from the Georg August University in Göttingen. In the same year the group moved to the Research Institute for Biology of the Max Planck Society in Tübingen. In 1959 he obtained a one-year postdoctoral position at the California Institute of Technology in Pasadena with Max Delbrück, where he continued his investigations into the light and gravity responses of Phycomyces. Returning to Tübingen, he started to study nonlinear signal transformation and binocular interactions in the human pupillomotor pathway at the above-mentioned institution. Afterwards he examined frog retinal ganglion cells. Later, he frequently changed the objects of his investigations, because he was looking for biological problems, the mathematical modelling of which promised to be fruitful, and each new object gave hirn the opportunity to become acquainted with a new chapter in biology., In 1968 the Eberhard Karls University in Tübingen offered hirn aChair for Zoology, which was so on renamed the Department for Biological Cybernetics. The general field of his research during the last 30 years was invertebrate behavioural neurobiology with a special interest for localization and orientation. In

XIV About the Authors

1983 he organized the triannual conference of the German Association for Cybernetics on these topics. His activities included both experimental investigations and mathematical modelling. His experimental animals were the beetle Tenebrio, the stick insect Carassius, the crabs Carcinus, Leptograpsus, Pachygrapsus, the crayfish Cherax, the bugs Triatoma, Gerris, Notonecta and the hawk moth Macroglossum. From 1969 until2001 he was member of the Editorial Board of Biological Cybernetics and since 1993 of the Advisory Board of the Journal of Comparative Physiology A. He spent his sabbaticals in the laboratories of friends in Canberra and Sydney (1980/81, 1986/87, 1991/92). In Tübingen he conducted research with guest scientists from Argentina, Canada, USA, and most frequentlywith Geibor Horveith from Hungary. Since October 1997 he is Professor Emeritus of the University of Tübingen.

Contents

Part I: Imaging Polarimetry

1 1.1

1.2

1.3

1.4 1.5 1.6

1.7 1.8

Polarimetry: From Point-Source to Imaging Polarimeters Qualitative Demonstration of Linear Polarization in the Optical Environment ........... ..... . Elements of the Stokes and Mueller Formalism of Polarization . . . . . . . . . . . . . . . . . . . Polarimetry of Circularly Unpolarized Light by Means ofIntensity Detectors . . . . . . . . . . . . . . . . Point-Source, Scanning and Imaging Polarimetry .. . Sequential and Simultaneous Polarimetry ...... . Colour Coding and Visualization of Polarization Patterns Field ofView of Imaging Polarimetry Polarizational Cameras . . . . . . . . . . . . . . . . . . . .

Part 11: Polarization Patterns in Nature

2

3 3.1

3.2

Space-Borne Measurement of Earthlight Polarization

Skylight Polarization . . . . . . . . . . . . . . . . . . . . . . The Importance of Skylight Polarization in Atmospheric Science . . . . . . . . . . . . . . . . . . . . . Celestial Polarization Measured by Video Polarimetry in the Tunisian Desert in the UV and Green Spectral Ranges

3

3

8

9 10 10 11 11 12

15

18

18

19

XVI

4 4.1

4.2

5

6

6.1. 6.2.

6.3.

7

8

9

10

10.1

10.2

10.2.1

10.2.2 10.2.3 10.2.4 10.2.5 10.2.6

10.2.7

10.2.8

Contents

Principal Neutral Points of Atmospherie Polarization Video Polarimetry of the Arago Neutral Point of Skylight Polarization . . . . . . . . . . . . . . . . . . First Observation of the Fourth Principal Neutral Point

24-Hour Change of the Polarization Pattern of the Summer Sky North of the Arctie Circle

Polarization Patterns of Cloudy Skies and Animal Orientation ...... . Polarization of Cloudy Skies .. . . . Continuation of the Clear-Sky Angle of Polarization Pattern Underneath Clouds Proportion of the Celestial Polarization Pattern Useful for Compass Orientation Exemplified with Crickets

Ground-Based Full-Sky Imaging Polarimetrie Cloud Detection . . . . . . . . . . . . . . . . .

Polarization Pattern of the Moonlit Clear Night Sky at Full Moon: Comparison of Moonlit and Sunlit Skies

Imaging Polarimetry of the Rainbow .

Which Part of the Spectrum is Optimal for Perception of Skylight Polarization? ...... . A Common Misbelief Concerning the Dependence of the Degree of Skylight Polarization on Wavelength

Why do Many Insects Perceive Skylight Polarization in the UV? . . . . . . . . . . . . . . . . . . . . . Is the Celestial Polarization Pattern More Stable in the UV? Was the UV Component of Skylight Stronger in the Past? Relatively Large Proportion of UV Radiation in Skylight? . Mistaking Skylight for Ground-Reflected Light? ..... . Confusion of Motion and Form for Celestial Polarization? Were UV Receptors Originally Skylight Detectors and Only Later Incorporated Into the E-vector Detecting System? ... Maximizing "Signal-to-Noise Ratios" by UV Photopigments Under Low Degrees of Skylight Polarization? In the Spectral and Intensity Domain the Celestial Band of Maximum Polarization is Less Pronounced in the UV than in the BIue .................. .

23

25 27

32

36 36

37

38

41

47

51

53

53

56 56 57 59 60 60

61

61

62

Contents

10.2.9 The Proportion of Celestial Polarization Pattern Useful for Animal Orientation is Higher in the Blue than

XVII

in the Green or Red .. . . . . . . . . . . . . . . . . . . 62 10.2.10 Perception of Skylight in the UV Maximizes the Extent

of the Celestial Polarization Pattern Useful for Compass Orientation Under Cloudy Skies 64

10.3 Resolution ofthe UV-Sky-Pol Paradox .... 68 10.4 E-Vector Detection in the UV also Maximizes

the Proportion of the Celestial Polarization Pattern Useful for Orientation Under Canopies . . . . . . . 69

10.5 Analogy Between Perception of Skylight Polarization and Polarotactic Water Detection Considering the Optimal Spectral Range .......... ....... 71

10.6 Analogy of the UV-Sky-Pol Paradox in the Polarization Sensitivity of Aquatic Animals ....... 71

10.7 Why do Crickets Perceive Skylight Polarization in the BIue? 72 10.8 Concluding Remark . . . . . . . . . . . . . . . . . . . . . .. 73

11 Polarization of the Sky and the Solar Corona During Total Solar Eclipses ................. 74

11.1 Structure of the Celestial Polarization Pattern and its Temporal Change During the Eclipse of 11 August 1999 75

11.2 Origin of the E-vector Pattern During Totality . . . . . . . 78 11.3 Neutral Points of Skylight Polarization Observed

During Totality ................... 80 11.4 Origin of the Zenith Neutral Point During Totality 83 11.5 Origin of Other Neutral Points at Totality 83 11.6 Imaging Polarimetry of the Solar Corona . . . . . . 85

12 Reflection-Polarization Pattern of the Flat Water Surface Measured by 1800 Field-of-View Imaging Polarimetry 88

13 Polarization Pattern of a Fata Morgana: Why Aquatic Insects are not Attracted by Mirages? 92

14 Polarizational Characteristics of the Underwater World 95

15 Circularly Polarized Light in Nature ............. 100 15.1 Circulary/Elliptically Polarized Light Induced

by Total Reflection from the Water-Air Interface . . . . . .. 100 15.2 Circulary Polarized Light Reflected from

the Exoskeleton of Certain Arthropods ..... 101 15.3 Circulary Polarized Light Emitted by Firefly Larvae 102

XVIII Contents

Part III: Polarized Light in Animal Vision

16 16.1 16.2

16.3

16.4 16.5 16.6 16.7

16.7.1 16.7.2 16.7.3 16.7.4 16.7.5

16.7.6

16.7.7

16.7.8 16.8 16.9

17 17.1 17.2 17.2.1 17.2.2 17.2.3

17.2.4 17.3 17.4 17.4.1 17.4.2 17.4.3 17.4.4 17.5

From Polarization Sensitivity to Polarization Vision Forerunners of the Study of Animal Polarization Sensitivity Polarization Sensitivity, Polarization Vision and Analysis of Polarization Patterns ............ . Functional Similarities Between Polarization Vision and Colour Vision . . . . . . . . . . . . . . . . . . How ean Skylight Polarization be Used for Orientation? Possible Functions of Polarization Sensitivity .. How might Polarization Sensitivity Have Evolved? Polarization Sensitivity of Rhabdomeric Invertebrate Photoreeeptors ................ . Hypothetieal Polarizing Ability of the Dioptrie Apparatus Rhabdomerie Polarization Sensitivity . Origin of High Polarization Sensitivity Origin of Low Polarization Sensitivity Rhabdomerie Twist and Misalignment and their Funetional Signifieanee .............. . Ontogenetic Development of Photoreeeptor Twist Outside the Dorsal Rim Area of the Inseet Eye . . . . . . . . Charaeteristies of the Anatomically and Physiologically Specialized Polarization -Sensitive Dorsal Rim Area in Inseet Eyes ......... . Polarization-Sensitive Interneurons in Invertebrates Polarization Sensitivity ofVertebrate Photoreeeptors Polarization Sensitivity in Plants ...... .

Polarization Sensitivity in Terrestrial Inseets Honeybees . Flies .................... . Muscid Flies ............... . Rhabdomeric Twist in the Retina of Flies Musca domestica, Calliphora erythrocephala, Calliphora stygia and Phaenicia sericata Drosophila melanogaster Ants ....... . Crickets Acheta domesticus Gryllotalpa gryllotalpa Gryllus bimaculatus . . Gryllus campestris Lepidoptera: Butterflies and Moths

107 107

108

111 112 115 116

117 118 118 121 122

123

124

125 128 128 130

131 131 143 143 143

144 146 147 156 156 156 157 160 165

Contents

17.5.1 17.5.2

17.5.3

17.6 17.7 17.8 17.9

18 18.1 18.2 18.3 18.4 18.5 18.6 18.7 18.8 18.9 18.10 18.11

19

20

21 21.1 21.2

21.2.1 21.2.2 21.2.3 21.3 21.4

XIX

Papilio xuthus ....... . . . . . . . . . . . . . . . . . .. 166 Polarization-Induced False Colours Perceived by Papilio xuthus and Papilio aegeus ........... 166 Polarized Light Reflected from Butterfly Wings as a Possible Mating Signal in Heliconius cydno chioneus 169 Locusts . . . . 169 Cockroaches . . . . . . . . . . . . . . . . . . . . . . . . . 172 Scarab Beetles ....... . . . . . . . . . . . . . . . . . 173 Response ofNight-Flying Insects to Linearly Polarized Light 176

Polarization Sensitivity in Insects Associated with Water . 178 Velia caprai ............. 180 Corixa punctata .. . . . . . . . . . 180 Non-Biting Midges (Chironomidae) 180 Waterstrider Gerris lacustris .. 181 Backswimmer Notonecta glauca 183 Dragonflies Odonata .. 188 Dolichopodids . . . . . . . . . . 191 Mayflies Ephemeroptera .... 192 Other Polarotactie Water Insects 193 Insects Living on Moist Substrata or Dung 195 Mosquitoes . . . . . . . . . . . . . . . . . . 197

Multiple-Choice Experiments on Dragonfly Polarotaxis 199

How can Dragonflies Discern Bright and Dark Waters from a Distance? The Degree of Linear Polarization of Reflected Light as a Possible eue for Dragonfly Habitat Selection .................... 206

on Reservoirs and Plastic Sheets as Polarizing Insect Traps 215 on Lakes in the Desert of Kuwait as Massive Insect Traps 215 The Waste Oil Reservoir in Budapest as a Disastrous Insect Trap for Half a Century . . . . 219 Surface Characteristies of Waste Oil Reservoirs 220 Insects Trapped by the Waste Oil ........ 221 Behaviour of Dragonflies Above Oil Surfaces . . 222 Dual-Choiee Field Experiments Using Huge Plastie Sheets 223 The Possible Large-Scale Hazard of "Shiny Black Anthropogenie Products" for Aquatie Insects 227

xx

22

22.1 22.2 22.3

22.4 22.5

23

24 24.1 24.2

25 25. 25.2 25.3 25.4 25.5 25.6 25.7 25.8 25.9 25.10 25.11

26 26.1 26.1.1 26.1.2 26.1.3 26.2

27 27.1 27.2 27.3

Why do Maytlies Lay Eggs on Dry Asphalt Roads? Water-Imitating Horizontally Polarized Light Retlected

Contents

from Asphalt Attracts Ephemeroptera ......... 229 Swarming Behaviour of Mayflies above Asphalt Roads 231 Multiple-Choke Experiments with Swarming Mayflies 232 Reflection -Polarizational Characteristks of the Swarming Sites of Mayflies .......... 234 Mayflies Detect Water by Polarotaxis ........ 236 Comparison of the Attractiveness of Asphalt Roads and Water Surfaces to Mayflies . . . . . . . . . . . . 239

Retlection-Polarizational Characteristics of Car-Bodies: Why are Water-Seeking Insects Attracted to the Bodywork of Cars? . . . . . . . . . . . . . . . . . . .. 241

Polarization Sensitivity in Spiders and Scorpions 243 Spiders .. 243 Scorpions 246

Polarization Sensitivity in Crustaceans 247 Mangrove Crab Goniopsis cruentata 249 Fiddler Crabs ............ 249 Copepod Cyclops vernalis . . . . . . 250 Larvae of the Crab Rhithropanopeus harrisi 251 Larvae of the Mud Crab Panopeus herbstii 252 Grapsid Crab Leptograpsus variegatus 253 Crayfish ................ 253 Grass Shrimp Palaemonetes vulgaris 255 Crab Dotilla wichmanni 257 Water Flea Daphnia 259 Mantis Shrimps ..... 263

Polarization Sensitivity in Cephalopods and Marine Snails 267 C~hal~o~ M7 Octopuses ............... 267 Squids ................. 269 European Cuttlefish Sepia officinalis 272 Marine Snails ............. 274

Polarization-Sensitive Optomotor Reaction in Invertebrates 276 Crabs . . . 276 Honeybees 277 Flies. . . . 277

Contents

27.4 27.5

27.6

28 28.1 28.1.1 28.1.2

28.1.3 28.1.4 28.1.5 28.1.6 28.1.7 28.1.8

28.1.9 28.2

28.2.1 28.2.2 28.2.3 28.3 28.3.1

28.3.2

28.3.3

28.3.3.1

28.3.3.2

Rose Chafers . . . . . . . . . . . . . . . . . . . Optomotor Reaction to Over- and Underwater Brightness and Polarization Patterns in the Waterstrider Gerris lacustris Optomotor Response to Over- and Underwater Brightness and Polarization Patterns in the Backswimmer Notonecta glauca ........... .

Polarization Sensitivity in Fish ........... . Fish in wh ich Polarization-Sensitivity was Proposed Sockeye Salmon Oncorhynchus nerka . . . . . . . . . Tropical Halfbeaks Zenarchopterus dispar and Zenarchopterus buffoni Halfbeak Fish Dermogenys pusilus . . . . . Goldfish Carassius auratus . . . . . . . . . African Cichlid Pseudotropheus macrophthalmus Anchovies Engraulis mordax and Anchoa mitchilli Rainbow Trout Oncorhyncus mykiss . . . . . . Juvenile Salmonid Fish Oncorhynchus mykiss, Oncorhynchus, Oncorhynchus nerka and Salvelinus fontinalis ........... . Damselfishes . . . . . . . . . . . . . . . . . . . Fish with Debated Polarization Sensitivity and Fish in which Polarization Insensitivity was Proposed Green Sunfish Lepomis cyanellus ....... . Common White Sucker Catostomus commersoni . Pacific Herring Clupea harengus pallasi . . . . . . Possible Biophysical Basis of Fish Polarization Sensitivity Axially Oriented Membrane Disks in the Photoreceptor Outer Segments as the Basis for Polarization Sensitivity in Anchovies . . . . . . . . . . . . . . . . . . . . . . Embryonic Fissures in Fish Eyes and their Possible Role in the Detection of Polarization ..... . Paired Cones as a Possible Basis for Polarization Sensitivity in Fish . . . . . . . . . . . . . . . . . . . . . . . . Orthogonal Double Cones with Graded Index of Refraction as a Possible Basis for Polarization Sensitivity in the Green Sunfish Lepomis cyanellus Proposed Basis for Polarization Sensitivity in Rainbow Trout due to Internal Reflection from the Membranous

XXI

278

278

287

293 294 294

295 296 297 299 300 301

306 306

307 307 308 308 309

309

311

312

312

Partitions of Double Cones . . . . . . . . . . . . . . . . 314

XXII

29 29.1 29.2 29.3 29.4

30 30.1

30.2 30.3

31 31.1 3l.1.1

3l.l.2

3l.l.3 3l.l.4 3l.l.5 3l.2 31.3

31.3.1

3l.3.1.1

3l.3.2 3l.3.3 3l.4 3l.4.1

3l.4.2

32 32.1 32.2 32.3

Polarization Sensitivity in Amphibians . . . . Tiger Salamander Ambystoma tigrinum Red-Spotted Newt Notophthalmus viridescens Larval Bullfrog Rana catesbeiana ...... . Proposed Mechanisms of Detection

Contents

317 318 320 321

of Polarization in Amphibians .. . . . . . . . . . . . . . . . 322

Polarization Sensitivity in Reptiles ......... . Celestial Orientation in Reptiles and the Polarization-Sensitive Parietal Eye of Lizards Desert Lizard Uma notata Sleepy Lizard Tiliqua rugosa ..

Polarization Sensitivity in Birds . . . . . . . . . Crepuscularly and Nocturnally Migrating Birds White-Throated Sparrow Zonotrichia albicollis and American Tree row Spizella arborea Northern Waterthrush Seiurus noveboracensis and Kentucky Warbier Oporornis formosus . Yellow-Rumped Warbier Dendroica coronata . Blackcap Sylvia atricapilla .......... . Savannah Sparrow Passerculus sandwichensis Day-Migrating Birds ............. . Birds which Might be Polarization Insensitive or not Use Skylight Polarization in their Migratory Orientation Debated Polarization Sensitivity in the Homing Pigeon Columba livia ................. . The Position of the Sun Hidden by Clouds Could also be Determined on the Basis of the Colour Gradients of Skylight Under Partly Cloudy Conditions European robin Erithacus rubecula ........... . Pied Flycatcher Ficedula hypoleuca ........... . Proposed Mechanisms of Avian Polarization Sensitivity Is the Foveal Depression in the Avian Retina Responsible for Polarization Sensitivity? ............... . A Model of Polarization Detection in the Avian Retina with Oil Droplets

Human Polarization Sensitivity Haidinger Brushes Boehm Brushes . Shurcliff Brushes

324

324 325 326

328 330

330

331 332 334 335 340

341

342

348 349 350 351

351

353

355 355 361 361

Contents

33 33.1

33.2 33.2.1

33.2.2

33.2.3 33.2.4

33.3

33.4

33.5

34

Rerences

Polarization-Induced False Colours ...... . Polarization-Dependent Colour Sensitivity and Colour-Dependent Polarization Sensitivity ...... . Polarizational False Colours Perceived by Papilio Butterflies Computation of the Spectral Loci of Colours Perceived bya Polarization- and Colour-Sensitive Retina . . . . . Polarization -Induced False Colours Perceived by a Weakly Polarization-Sensitive Retina ....... . Reflection-Polarizational Characteristics of Plant Surfaces . Do Polarization-Induced False Colours Influence the Weakly Polarization-Sensitive Colour Vision of Papilio Butterflies Under Natural Conditions? ............ . Polarizational False Colours Perceived bya Highly Polarization-Sensitive Retina Rotating in Front of Flowers and Leaves . . . . . . . . . . Camouflage Breaking via Polarization-Induced False Colours and Reflection Polarization Is Colour Perception or Polarization Sensitivity the More Ancient? . . . . . . . . . . . . . . . . .

A Common Methodological Error: Intensity Patterns Induced by Selective Reflection of Linearly Polarized Light from Black Surfaces . .

XXIII

362

362 364

364

369 374

376

377

378

379

381

385

SubjectIndex ................................ 417

Colour Illustrations 425

Part I: Imaging Polarimetry

1 Polarimetry: From Point -Source to Imaging Polarimeters

Biologists dealing with polarization sensitivity of animals, or engineers designing robots using polarization-sensitive imaging detectors, for example, need a technique to measure the spatial distribution of polarization in the optical environment. In the 1980s, 1990s and early 2000s, different kinds of imaging polarimetry have been developed to measure the polarization patterns of objects and natural scenes in a wide field of view. The conventional non-imaging point-source polarimeters average polarization over an area of a few degrees only. The conception of "polarization imagery" or "imaging polarimetry" was introduced by Walraven (1981) to obtain high-resolution information ab out the polarized components of the skylight radiance. Table 1.1 summarizes the most important properties of various imaging polarimeters.

1.1 Qualitative Demonstration of Linear Polarization in the Optical Environment

The presence of linearly polarized light (the most common type of polarization in nature) in the optical environment can be qualitatively demonstrated by the use of a linear polarizer. Looking through such a filter and rotating it in front of our eyes, the change of intensity of light coming from certain directions may be observed. This intensity change is an unambiguous sign of the polarization of light. If we take colour photographs from a scene through linear polarizers with differently oriented transmission axes and compare them, striking intensity and colour differences may occur in those regions, from which highly polarized light originates, furthermore the brightness and colour contrasts may change drastically between different parts of the scene (~ colour Figs. 1.1 and 1.2).

Using triangles cut from a sheet of linearly polarizing filter, Karl von Frisch (1953) constructed a simple device, the so-called Sternfolie (star foil), with which the gross distribution of linear polarization of skylight could be

4 Part I: Imaging Polarimetry

Table 1.1. The most important properties of some imaging polarimeters designed by different authors and used for various purposes. Since aB instruments contain linearly polarizing filter(s) of different types, the polarizers are not mentioned and specified in the column "imaging optics" (10).

Author(s) Type IO DET FOV RES SR Application

Gerharz (1976) FIP CAMO+ PP 12x IS° 535 Polarization Savart distribution of the fIlter + CF circumsolar scatter

field during a total solar eclipse

Dürst (1982) SEQ CAMO+ PE 8xlO° 50x50 600 Polarization pattern PHO 6NF+ of the solar corona

1 CF during a total solar eclipse

Prosch et al. SIM 3lens IT 25x25° 36x36 VIS Ground- and airborne (1983) VID systems remote sensing of

landscape features

Sivaraman SIM four-Iens PE 3x3° 32x32 WL p-pattern of the solar et al. (1984) PHO CAMO corona during a total

solar eclipse

Fitch et al. POR CAMO PE 30x400 512x512 VIS Polarization pattern (1984) SEQ of light reflected from

PHO grain crops during the heading growth stage

POLDER SEQ wide field- CCD 114x114° 242x274 443, Space-borne meas-(1994-1997) VID of-view 670, urement of the polar-Deschamps optics + 865 izational characteris-et al. (1994) filter wheel tics of earthlight

Wolff (1993), SEQ CAMO+ CCD 30x40° 165x192 VIS Polarization patterns Cronin et al. VID 2TNLC (D) of objects and (1994), SUB 240x320 biotopes Shashar et al. (V) (1995a, 1996)

Wolff (1994), SEQ 2 CAMO + CCD 20x20° 165x192 VIS Polarization patterns Wolff& VID PPBS + of objects for robot Andreou (1995) TNLC vision

Wolff& 10 lens PSC - 3x128 VIS Prototype of future Andreou (1995) SIM system 2D polarization

PCC camera chips

Povel (1995) SIM telescope + CCD 0.42'x 288x385 VIS Observation of solar STO PEMs 0.83' magnetic fields

Pezzaniti & MMI lens system CCD 42x42° 512x512 VIS Polarizational proper-Chipman (1995) SEQ + retarders IR ties of static optical

+ laser systems and samples

1 Polarimetry: From Point-Source to Imaging Polarimeters 5

Table 1.1. (Continued)

Author(s) Type 10 DET FOV RES SR Application

North & Duggin SIM four-Iens PE 1800 CIR 300x300 VIS Ground-borne meas-(1997) PHO CAMO+ urement of skylight

spherical polarization mirror

Voss &Liu SEQ FEL CCD 1780 CIR 528xS28 VIS Ground-borne meas-(1997) VID (B) urement of skylight

polarization

Horvath & POR CAMO CCD SOx40° 736x560 VIS Polarization patterns Varju (1997) SEQ of sky, objects and

VID biotopes

Lee (1998) POR CAMO PE 36x24° 550x370 VIS Polarization patterns SEQ of clear skies PHO

Horvath & POR CAMO UV 20x15° 736x560 UV + Polarization patterns Wehner (1999) SEQ IT VIS of sky, objects and

VID biotopes

Bueno & Artal SEQ CAMO+ CCD lxl° 60x60 630 Polarizational proper-(1999), MMI 2TNL+ ties of static optical Bueno (2000) 2 quarter- systems and sampIes

wave plate (e.g. human eye) + laser

Hanlon etal. SIM 3-tube IT 20x30° 512x384 VIS Polarization patterns (1999) VID CAMO+ of moving animals

prismatic beam-splitter

Mizera et al. POR CAMO CCD 50x40° 736xS60 VIS Polarization patterns (2001) SEQ of objects and

STE biotopes VID

Ga! et al. POR FEL+ PE 1800 CIR 670x670 VIS Ground- and airborne (2001 c) SEQ filter measurements of

PHO wheel polarization patterns of the atmosphere, objects and biotopes

Shashar et al. SEQ microscope CCD 5x5° 512x384 VIS Polarization patterns (2001) VID of microscopic targets Horvath et al. POR 3FEL PE 1800 CIR 670x670 VIS Ground-borne meas-(2002a) SIM urements of skylight

PHO polarization


Table 1.1. (Continued)

Author(s) Type IO DET FOV RES SR Application

Pomozi (2002), DPL Laser CCD 256x256 1024x VIS Study of the aniso-Pomozi eta/. SM scanning Ilm 1024 tropic architecture of (2003), Garab microscope microscopic sampIes et a/. (2003) and the interaction of

the sampIe with polar-ized light

Barter et al. SIM CAMO+ CCD 36x36° 640x VIS Patterns of linear (2003) VID 4-way 480 circular polarization

beam- of the optical splitting environment

at 60 Hz frame rate

lD one-dimensional (linear). B binned. CAMO camera optics. CCD charge-coupled device. CP colour filter. CIR circular. D digital. DET detector. DPLSM differential polarization laser scanning microscopy. PEL fisheye lens. PIP forerunner of imaging polarimetry. POV field of view. IR infrared (A > 750 nm). IT imaging tube. MMI Mueller matrix imaging polarimeter. NP neutral density filter. PCC polarization camera chip. PE photo emulsion. PEM piezoelastic modulator. PHO photopolarimeter. POR portable. PP photographic plate. PPBS polarizing plate beam-splitter. PSC polarization-sensitive chip. RES spatial resolution (pixel x pixel). SEQ sequential. SIM simultaneous. SR spectral region (nm). STE stereo. STO imaging Stokes polarimeter. SUB submersible. TNLC twisted-nematic liquid crystal. UVultraviolet. V video. VID video polarimeter. VIS visible (400-750 nm). WL white light.

demonstrated (Fig. 1.3). This pioneering instrument was used by Frisch to investigate qualitatively the degree and angle of polarization of skylight, which was important to interpret the results of his behavioural experiments with honeybees.

What could be demonstrated only qualitatively by Frisch (1953) with his "Sternfolie", nowadays can already be measured quantitatively by different kinds of full-sky imaging polarimeters (North and Duggin 1997; Voss and Liu 1997; Gal et al. 2001a,b,c; Pomozi et al. 2001a,b; Horvath et al. 2002a,b, 2003; Barta et al. 2003). Figure 1.3 and ~ colour Figs. 1.4 and 1.5 (see also ~ colour Figs. 4.3-4.5) demonstrate well the advance of imaging polarimetry in the last 50 years.

Polarimetry: From Point-Source to Imaging Polarimeters 7

South South-West West North-West

Fig. 1.3. A Schematic drawing of a sheet of linearly polarizing filter with cut pattern to construct the "Sternfolie" ("star foil") used to demonstrate the gross distribution of linear polarization of skylight by Karl von Frisch (1953, 1967). The orientation of the transmission axis is shown by double-headed arrows. B The geometry of the "Sternfolie". C Simple instrument - a "Sternfolie" mounted onto a metal holder in such a way that both the elevation and azimuth of the viewing direction through the foil can be changed, - with which Frisch (1953, 1967) investigated qualitatively the polarization of skylight. D View through the "Sternfolie" in eight different directions in the sky with an angle of elevation of 45°. (After Frisch 1953).


1.2 Elements of the Stokes and Mueller Formalism of Polarization

Polarized light can be decomposed into two components vibrating coherently (that is, with a constant phase difference) and perpendicularly to each other. The state of polarization of transversal electromagnetic waves (e.g. light) is usually described by a four-element vector known as Stokes vector $., first introduced by Stokes (1852) with the following components:

.s. = (I, Q, U, V), I = Ir + Ip = 145 + 1135 = Ire + Ilc,

Q = Ir - Ip = I·p·cos(2E)·cos(2a), U = 145 - 1135 = I·p·cos(2E)·sin(2a),

V = Ire - Ilc = I·p·sin(2E) (1.1)

where I is the total intensity of light, Ir and Ip are the intensities of the light components polarized totally linearly in a reference plane and perpendicularly to it, 145 and 1135 are the intensities of the components polarized totally linearly in planes 45 and 1350 to the reference plane, I rc and I lc are the intensities of the components polarized circularly right- and left-handed, p is the degree of linear polarization, E is the ellipticity of polarization, and a is the angle of polarization, which is the angle of the direction of oscillation from a given plane. Q quantifies the fraction of linear polarization parallel to the referen ce plane, U gives the proportion of linear polarization at 450 with respect to the reference plane, and V quantifies the fraction of right-handed circular polarization. The degree of polarization P, the degree of linear polarization p, the angle of polarization a and the ellipticity E can be expressed by the components of the Stokes vector as follows (Shurcliff 1962):

P = (Q2 + U2 + V2)1I2/I,

a = O.5·arc tan(U/Q),

p = (Q2 + U2)1I2/I, 0:::; P,p:::; 1,

(1.2)

A change in the state of polarization of light produced by an optical system, i.e. a transformation of the Stokes vector $.0 = (Jo' Qo' Uo' VoJ of the incident light into a new Stokes vector $. = (I, Q, U, V) by an optical process (e.g. reflection, refraction, scattering, diffraction, birefringence, optical activity) can be expressed as a linear transformation in a four-dimensional space:

(1.3)

where M is a four-by-four matrix called "Mueller matrix" with real elements Mij (i,j=O,I,2,3) containing information on all polarizational properties of light. The 16 elements of the Mueller matrix of a given optical system can be obtained by 16 measurements with independent combinations of states of


polarization (degrees and angles of linear and circular polarization) of the incident light.

1.3 Polarimetry of Circularly Unpolarized Light by Means of Intensity Detectors

Light in the natural optical environment is usually not circularly polarized. The few known exceptions are listed and discussed in Chap. 15. Skylight polarization, for instance, is predominantly linear and the component of circular polarization of skylight can be neglected (Hannemann and Raschke 1974). Thus, the contribution of the Stokes parameter V characterizing circular polarization to the total intensity is negligible in comparison with that of the linearly polarized component. The remaining Stokes vector components I, Q and U can be determined from three intensity measurements, using a rotating linear polarizer in front of a radiometer, for instance. If these three measurements occur at angles of orientation ß = 0, 60 and 120° of the transmission axis of aperfect polarizer (with t = 1 and T = 0, where t and T are the transmittances of the polarizer along the transmission axis and perpendicularly to it), for example, and the state of polarization of light is not changed by other components of the polarimeter, then the transmitted intensities I are (Prosch et al. 1983):

I(ß=OO) == 10 = Idl + p-cos(2a)]/2,

I(ß=600 ) == 160 = Id1 - 0.S-p-cos(2a) + OSp-3112 -sin(2a)]I2,

I(ß=1200 ) == 1120 = Id1 - 0.S-p-cos(2a) - 0.S-p-3 112 -sin(2a))/2, (1.4)

where Ii is the intensity of incident light. The components Qi and Ui of the incident Stokes vector are:

(1.5)

Finally, the intensity I i, degree of linear polarization p and angle of polarization a of incident light can be calculated as follows:

li = 2(10 + 160 + 1120)/3, P = (Q?+U?)1I2/li , a = O.S-arc tan(U/Q). (1.6)


1.4 Point-Source, Scanning and Imaging Polarimetry

The major aim of polarimetry is to measure the four components I, Q, U and V of the Stokes vector $., from wh ich further quantities of the incident light can be derived, according to Eqn (1.2). These measurements can be done either by a point-source polarimeter or by an imaging one. The only principal difference between them is that the former performs measurements in a given direction representing a very narrow field of view within which the optical variables I, Q, U and Vare averaged, while the latter measures the polarization simultaneously in many directions in a wide field of view (~ colour Fig. 1.4). A further development of the latter technique is the stereo video polarimetry (Mizera et al. 2001) which visualizes the polarization patterns in three dimensions (~ colour Fig. 1.5). There is an intermediate technique, the scanning point-source polarimetry between these two extremities. Such apolarimeter scans a given area of the optical environment and measures sequentially the polarization in many directions. However, scanning a greater area of the optical environment with a point-source polarimeter is a troublesome and timeconsuming task. Using imaging polarimetry, the spatial distribution of polarization can be easily and quickly determined.

1.5 Sequential and Simultaneous Polarimetry

If the (at least necessary) three intensity measurements with different orientations of the transmission axis of the polarizer are performed one after the other, we speak ab out "sequential polarimetry". When all these measurements happen at the same time, it is called "simultaneous polarimetry". For the latter at least three separate polarimeters are needed. The advantage of simultaneous polarimetry is that temporally changing radiation fields (e.g.light from doudy skies with rapidly moving douds, or skylight after sunset or prior to sunrise, or measurements from a moving platform) can also be measured with it, if the time needed is not longer than the characteristic period during which considerable changes occur in the radiation field. Its disadvantage is that at least three polarimeters have to be handled simultaneously, which is not a simple task. Furthermore, such a group of polarimeters is heavy, voluminous, its setting up, dismounting and transferring is difficult and time-consuming. These disadvantages frequently make the use of simultaneous polarimetry in the field impossible. The disadvantage of sequential polarimetry is that temporally changing radiation fields cannot be measured with it. Its advantage is that only one polarimeter has to be handled, the setting up, dismounting and transferring of which is much easier and quicker.


1.6 Colour Coding and Visualization of Polarization Patterns

On the basis of the functional similarity between polarization vision and colour vision, Bernard and Wehner (1977) suggested a hue-saturation-brightness visualization method for partially linearly polarized light. This "composite visualization" scheme was used by Wolff and collaborators (e.g. Wolff 1993; Shashar et al. 1995a), for example, who coded the angle of polarization a, degree of linear polarization p and intensity I of partially linearly polarized light by the hue, saturation and brightness, respectively. In their polarization maps, unpolarized light appears achromatic, strongly polarized regions show up chromatically saturated, and the intensity oflight is the brightness regardless of colour. The advantage of this visualization lies in its compactness: it displays the distribution of all three optical parameters (I, p, a) in a single, false-coloured picture. The disadvantage of this coding is that it is difficult to decompose, since in a complex false-coloured picture it is not easy to separate and decode the values of I, p and a from each other. Changes in hue (coding a) appear to the human visual system more strikingly than changes in saturation (codingp). Furthermore, the perception of the hue-saturation-brightness scale is very non-linear (Shashar et al. 1995a).

These problems do not occur if the distributions of I, p and aare displayed in three separate patterns with arbitrary unambiguous colour co ding (~ colour Figs. 1.4 and 1.5). This "separate visualization" of the I-, p- and a-patterns is preferred by Horvath and collaborators (e.g. Horvath and Varju 1997; Horvath and Wehner 1999; Gal et al. 2001c; Pomozi et al. 2001b; Bernath et al. 2002, Barta et al. 2003), for instance.

Other authors (e.g. Dürst 1982; Sivaraman et al. 1984) display the I-,p- or apatterns measured by imaging polarimetry in the form of the conventional contour plots used frequently in the cartography, for example. Although this "contour plot visualization" is the most traditional, it can hardly reproduce the image feature of the spatial distribution of polarization, which is the most important characteristic of the visualization of data gained by imaging polarimetry.

1.7 Field ofView of Imaging Polarimetry

The field of view of an imaging polarimeter is limited by that of the imaging optics used. In the case of common photographic and video cameras, the field of view of the lens system is about 30-50° (horizontal) X 20-40° (vertical) depending on the focallength and the aperture (~ colour Figs. 1.1,1.4 and 1.5). This common field of view can be extended e.g. by decreasing the focal length. A fisheye lens with 8 mm focallength mounted onto anormal photographic camera is an extremum, ensuring a hemispherical field of view with


an aperture angle of 180°, by which the whole hemisphere of the optical environment can be imaged (~ colour Fig. 1.2).

As an alternative, a 180° field-of-view fisheye lens can be replaced by a spherical mirror with a 180° field of view, and the camera can be suspended by a holder above the mirror. A similar construction is used in the full-sky im ag -ing polarimeter designed by North and Duggin (1997). 180° field-of-view imaging polarimetry is ideal to study the polarization patterns of the full sky or the reflection-polarization patterns of water surfaces, for instance.

1.8 Polarizational Cameras

The recently available polarimeters utilize optical imaging systems that are external to the detectors. Compactness of design and speed of generating polarization images can be enhanced greatly by incorporating an array of microscopic polarization filtering optics directly onto a photosensitive chip. Wolff and Andreou (1995) designed a one-dimensional polarization-sensitive chip, in which three adjacent pixels produce one measurement of partiallinear polarization. Two-dimensional polarization-sensitive chips are currently under development. Kalayjian et al. (1996) designed a one-dimensional polarization-contrast retina that can be used as polarimetric scanning sensor for real-time, automated vision tasks.

A common design for colour cameras is to use a non-polarizing beamsplitter that directs equal amounts of incoming light onto three separate CCD sensors for the red, green and blue spectral ranges. If a linearly polarizing filter is placed over every CCD, each filter having a unique direction of its transmission axis, a so-called polarizational camera using a non-polarizing beam-splitter can be built that operates in white light or in a given part of the spectrum, if a colour filter is added in front of the lens system of the camera. This design was suggested by Wolff (1993) and realized by Barter et al. (2003).

Wolff and collaborators are in the process of developing self-contained VLSI (Very Large Scale Integration) versions of polarizational cameras that sense complete states of partial linear polarization on-chip, compute state of linear polarization and visualization or physical information related to sensed polarization. VLSI offers very high computational throughput so that VLSI polarizational cameras enable operations at very high speeds.

A polarizational camera is a generalization of the conventional intensity camera. If necessary, the former can function as the latter. Adding coloursensing capability to a polarizational camera makes it possible to sense the complete set of electromagnetic parameters of light incident on the camera. Polarizational cameras have more general capabilities than standard intensity cameras, and can be applied for different purposes.

2 Space-Borne Measurement of Earthlight Polarization

The POLDER (POLarization and Directionality of the Earth's Reflectance) space-borne sequential imaging polarimeter was designed to measure the directionality and polarization of the earthlight, Le. the sunlight reflected from the earth's surface and scattered by the atmosphere (Deschamps et al. 1994). The POLDER instrument measured the total radiance R = I (Fig. 2.1A), the linearly polarized radiance (the product of the total radi an ce and the degree oflinear polarization p) Rp = pR = (Q2+U2)1I2 (Fig. 2.1B), and the angle of polarization a of earthlight at 443, 670 and 865 nm.

The POLDER system provided new opportunities for estimating atmospheric aerosol content over land surfaces. While radiance reflected from most land surfaces is only slightly polarized, radiance scattered by the molecules and aerosols in the atmosphere is highly polarized. Consequently, the polarization of earthlight measured from space originates primarily from the atmosphere (Fig. 2.1B), and aerosol properties can be derived from polarized reflectance measurements. Computing theoretically the polarized reflectance expected for an aerosol-free atmosphere (Rayleigh scattering only), the difference between the computed and measured polarized radiances corresponds to polarized radiance scattered by the aerosols. Thus, the polarized reflectance measurements by POLDER yielded the aerosol spectral behaviour, which provides an indication of their type (Le. size distribution and refractive index).

POLDER polarization measurements also allowed an estimate of the cloud pressure level. The measured polarized radiance is related to the atmospheric molecular optical thickness above the cloud, assuming that the radi an ce originating from the cloud is negligibly polarized and spectrally neutral. This assumption is not true for particular directions, such as that of the rainbow, which are avoided. In other viewing directions the polarized reflectance is mainly generated by the atmosphere and is nearly proportional to the molecular optical thickness above the cloud. This relationship leads to an estimate of the pressure at the top of clouds. Since the polarization induced by molecular scattering is maximal at 90° from the solar direction, this viewing direction is preferred. Although the aerosols above the cloud layer can also produce

16

A ,,+, B s polarizcd radiancc Rp total radiance R


Fig.2.1. Patterns of the total radiance R (A) and polarized radiance Rp (B) of earthlight measured by the POLDER instrument above Madagascar. In both ( originally coloured) pictures, the radiances for the red, green and blue were measured at 865,670 and 443 nm, respectiveIy. The (originally coloured) Rp -pattern is mainly blue because of the high linear polarization of molecular scattering at 443 nm. The ground surface has a very low contribution to the polarized signal, which depends mainly on the atmospheric light scattering. (After http://ceos.cnes.fr:8100/cdrom-00b2/ceosl/satellit/polder/index.html).

some perturbing polarized radiance, the bulk of atmospherie aerosols is contained in the boundary layer below the doud layer. For this method, polarization measurements at 443 nm were used because the molecular scattering contribution to the polarized reflectance is maximal relative to other contributions.

Cloud type determination and thermodynamie studies of the atmosphere require recognition of the doud phase, a parameter that POLDER polarization measurements could access. Radiative transfer simulations have shown that the polarization of c1oud-reflected radiance in specific directions (e.g. that of the rainbow) is very sensitive to the doud phase, whieh can be either iee or water (Fig. 2.2). Liquid doud droplets are evidenced by the characteristic strong polarization of the rainbow (Fig. 2.2C) exhibited by spherical partides for scattering angles near 1400 from the solar direction. The rainbow characteristic disappears as so on as the scattering partieies depart from spherical geometry. The lack of this characteristic feature in the doud polarization signature, therefore, is indieative of the presence of iee crystals. The method utilizes the polarizational data measured at 865 nm, since this spectral channel is the least polluted by molecular scattering among the other channels. This information is also very useful for polarimetrie doud detection.

Leaf cutide and wax specularly reflect part of the incident solar radiation on the canopy. Because this radiance does not interact with chlorophyll pigments, and hence cannot participate in photosynthesis, it should not be considered when the aim is to remotely sense the vegetation. Since specularly reflected radiance is partially linearly polarized, polarization measurements over land surfaces can be applied to correct for the specular component of the reflectance. POLDER polarization observations also he1ped to characterize the vegetation cover, because they are sensitive to the mieroscale structure of

2 Space-Borne Measurement of Earthlight Polarization 17

Fig.2.2. A The dimensionless polarized radiance Rp of clouds composed of either water droplets (dark grey dots) or ice particles (light grey x) as a function of the scattering angle from the solar direction measured by the POLDER instrument at 865 nm. (After http://ceos.cnesJr:8100/cdromOOb2/ ceos 1/ satellit/polder /index.html; similar graphs can be seen in Goloub et al. 1994). B, C Patterns of the total reflectance rand polarized reflectance rp over stratocumulus clouds measured by the air-borne version of the POLDER instrument at 443 nm. In the rp-pattern the strongly polarized primary and higher order rainbows are clearly discernible. (After Goloub et al. 1994).

+0.07 !"

A I\.

·0.02 u-~8«~~IO«~~1 2~~~~1 4~~~~I&r~~18~~

scanering angle

B

r

o total reflectance r at 443 nm

c

polarized reflectance r, at 443 nm

the canopy (Curran 1982). However, since the polarized reflectance measured from space originates mostly from the atmosphere, accurate atmospheric corrections (subtracting the contribution of atmospheric scattering) are necessary before space-borne polarized reflectance measurements can be used for vegetation monitoring applications.

3 Skylight Polarization

3.1 The Importance of Skylight Polarization in Atmospheric Science

Solar radiation is unpolarized before entering the earth's atmosphere. The skylight is partially linearly polarized through scattering interactions with the atmospheric constituents (gases, aerosol particles, water droplets, ice crystals). Since the discovery of skylight polarization by Arago in 1809, studies of the polarization of skylight and neutral points have been emphasized, as these can be used as indicators of atmospheric turbidity (dust, haze, pollution; Bellver 1987) and surface properties (Coulson 1974).

The clear sky has a characteristic polarization pattern depending on the solar position, the distribution of various components of the atmosphere and the underlying surface properties. The polarization of skylight has been the subject of numerous theoretical and experimental investigations (e.g. Chandrasekhar 1950; Neuberger 1950; Sekera 1957; Holzworth and Rao 1965; Coulson 1988). The principal features of the intensity and polarization of the sunlit sky can be explained in terms of Rayleigh scattering by molecules in the atmosphere (Coulson 1988).

Most ground-based measurements of skylight polarization were performed by means of point -source polarimeters to determine the degree and angle of linear polarization for different wavelengths. The development of full-sky imaging polarimetry (North and Duggin 1997; Voss and Liu 1997; Gal et al. 2001a,b,c; Pomozi et al. 2001a,b; Horv<ith et al. 2002a,b, 2003) offered new methods for observing the distribution of skylight polarization over the whole celestial hemisphere in the visible part of the spectrum quickly and accurately. Although various aspects of the polarization in the sunlit atmosphere have been studied in the past, rapid measurements of the polarization distribution over the entire sky were not possible before the development of full-sky imaging polarimetry. The ability of these imaging polarimeters to provide polarization distribution over the full sky has great potential in studies on atmospheric radiative transfer, because data can be collected in a short time; thus changes in the atmosphere during measurement can be avoided or minimized.

3 Skylight Polarization

3.2 Celestial Polarization Measured by Video Polarimetry in the Tunisian Desert in the UV and Green Spectral Ranges

19

In his paper on skylight navigation in ants, Santschi (1923) wondered why certain species used the sun as a compass, while others relied predominantly on the sun-free parts of the sky. Nowadays, it is known that the decisive aspect of light from the sky perceived by these ants is the E-vector distribution of linearly polarized skylight (see Chap. 17.3). The differences, however that Santschi had observed among different ant genera and species are still a riddIe. Are they really due to differences among different taxonomie groups of ants, or are they caused by characteristies of the habitats occupied by the different species? As the habitats of the Aphenogaster, Messor, Monomorium and Cataglyphis ant species are rather varied and include desert regions in mountains, sand-dune areas, salt pans or coastal inundation plains, several parameters such as water content, haze, frequency of clouds and the turbidity of the atmosphere vary accordingly. All these factors have strong influences on the degree of linear polarization p of scattered skylight.

How do these atmospherie differences in the ants' habitats affect p of skylight that influences the accuracy of navigation (Wehner 1982)? Furthermore, to what extent does p depend on the spectral composition of skylight? The Cataglyphis retina is equipped with UV and green photoreceptors, but only the former are used in E-vector navigation. Do the atmospherie factors mentioned above influence p of skylight more strongly in the UV than in the green, or vice versa?

In order to answer these questions, Horvath and Wehner (1999) measured p and the angle of polarization a of skylight in the UV and green by video polarimetry in several viewing directions (Fig. 3.1) under clear skies in three

Fig.3.1. The positions of the rectangular celestial windows within which videopolarimetrie data were collected by Horvath and Wehner (1999). AS antisolar meridian, S solar meridian. The numbers following these designations indieate the elevation of the camera in degrees. Z zenith. Z-OFF the camera was first rotated by 90" from the solar meridian then elevated by 45° in the plane perpendicu1ar to the solar meridian. The field of view of the camera was 20° (horizontal) x 15° (vertical). (After Horvath and Wehner 1999).

sky

antisolar meridian

z

solar meridian

20 Part II: Polarization Patterns in Nature

Table 3.1. Degree of linear polarization p of skylight measured by video polarimetry in the ultraviolet (UV, 360 nm) and green (G, 450 nm) spectral range (window size: 20x 15°) at three different Tunisian study sites (Tozeur: salt pan, 04.08.1996; Metlaoui: mountains, 05.08.1996 and 06.08.1996; Mahares: coastal area, 08.08.1996 and 10.08.1996). Solar ele-vation es = 35°,local summer time 16:30 (= UTC+l). Mean values ± standard deviations. For direction of view see Fig. 3.1. (After HorvMh and Wehner 1999).

Degree of linear polarization p (%) for solar elevation es = 35°

Direction Spectral Tozeur Metlaoui Mahares ofview range (salt pan) (mountains) (co ast)

AS-20 UV 8.2 ± 3.9 11.6 ± 4.3 10.9 ± 3.2 10.3 ± 4.3

AS-20 G 13.1 ± 5.9 16.1 ± 6.6 14.9 ± 6.3 14.2 ± 6.0

AS-55 UV 17.3 ± 5.1 14.4 ± 4.7 20.3 ± 3.8 18.7 ± 5.2

AS-55 G 30.4 ± 6.2 36.7 ± 5.7 38.9 ± 7.1 32.7 ± 6.0

Z UV 11.4 ± 3.9 10.6 ± 4.2 13.0 ± 3.3 12.3 ± 4.4

Z G 24.2 ± 6.9 26.2 ± 7.9 29.8 ± 9.0 24.8 ± 7.6

Z-OFF UV 16.9 ± 3.9 17.8 ± 3.9 19.9 ± 2.6 17.3 ± 3.5

Z-OFF G 27.9 ± 6.2 29.0 ± 6.9 34.0 ± 7.2 27.9 ± 6.5

S-55 UV 3.8 ± 2.1 5.1 ± 3.1 4.8 ± 2.4

S-55 G 6.0 ± 4.1 6.3 ± 3.5 7.5 ± 4.2

Tunisian habitats occupied by different species of desert ants (genera Cataglyphis, Messor, Aphenogaster, Monomorium):

1. Within the vast expanses of the salt-pan area of the Chott el Djerid (site "Tozeur")

2. In the extremely arid and vegetation-free highland area of the south-eastern parts of the North African Dorsale (site "Metlaoui")

3. In the coastal inundation plains of the Tunisian Sahel zone (site "Mahares")

In Tables 3.1 and 3.2 pis given in the UV and green for two solar elevations at these three types of habitat. None of the three study sites exhibited significantly high er or lower p than any of the other sites. Whenever two measure-

3 Skylight Polarization 21

Table 3.2. As Table 3.1 for solar elevation es = 70° at local summer time 14:00 (= UTC+ 1). (After Horv<ith and Wehner 1999).

Degree of linear polarization p (0/0) for solar elevation es = 70°

Direction Spectral Tozeur Metlaoui Mahares ofview range (salt pan) (mountains) (coast)

AS-20 UV 20.4 ± 3.6 22.0 ± 4.1 19.1 ± 3.2 22.4 ± 4.1 23.8 ± 3.5

AS-20 G 26.2 ± 4.5 28.6 ± 3.4 20.1 ± 5.1 25.1 ± 3.7 29.9 ± 7.6

AS-55 UV 16.5 ± 3.6 16.5 ± 3.7 17.5 ± 2.9 16.5 ± 4.2 19.1 ± 3.2

AS-55 G 24.1 ± 4.6 25.4 ± 5.7 23.5 ± 3.9 23.7 ± 5.3 31.5 ± 6.8

Z UV 4.9 ± 2.5 5.1 ± 2.9 4.8 ± 2.1 5.1 ± 2.9 4.8 ± 2.1

Z G 8.1 ± 3.9 7.5 ± 3.9 7.6 ± 3.7 7.7 ± 4.0 9.5 ± 4.8

Z-OFF UV 13.2 ± 3.7 14.2 ± 4.7 14.9 ± 6.3 13.3 ± 4.4 17.1 ± 3.2

Z-OFF G 18.2 ± 6.5 16.0 ± 6.9 24.3 ± 3.2 16.3 ± 6.6 22.2 ± 8.8

ments were performed at the same site on two subsequent days, p differed, sometimes remarkably, from one day to another, even though the human ob server could not detect any obvious differences in the appearance of the sky. The day-to-dayvariations at one particular site always exceeded the variations that were due to the geographicallocation of that site. In all celestial windows, p was always higher in the green than in the Uv, which is a typical characteristic of the normal atmosphere (Coulson 1988). Even in the cloudless sky vaulting a subtropical desert lands cape, p never exceeded mean values of 60, and 75 % in individual celestial directions.

Thus, the question posed above cannot be answered in the affirmative. p, wh ich is affected much more by atmospheric disturbances and surface reflections than a, does not vary systematically among the different types of desert habitat. The day-to-day fluctuations of p are much larger than the habitatbased variations. Hence, Santschi's (l923) early observation that for navigation some ant species inhabiting particular geographical regions relied more on scattered skylight than direct sunlight, cannot be explained on the basis of the distinctness of skylight polarization available to the ants in these different


habitats. Instead, the interspecific and intergeneric differences might be caused by peculiarities of the ants' species-specific navigational systems.

4 Principal Neutral Points of Atmospheric Polarization

The most important optical characteristics of the clear sunlit sky are well described by the Rayleigh theory (Coulson 1988). The fine details of skylight polarization, however, differ from the ideal Rayleigh model. This failure, called the polarization defect, is caused by multiple scattering, molecular anisotropy, scattering by aerosol particles, size distribution and particle shapes of aerosol, and the light reflected from the ground. One of the most remarkable features of this defect is the phenomenon of the neutral points where the degree of linear polarization is zero.

In the clear sunlit normal sky, only three loci exist, the Arago, Babinet and Brewster neutral points, where the skylight is unpolarized. The antecedents date back to 1809 when the French astronomer Dominique Francois Jean Arago discovered the partial linear polarization of skylight, and soon thereafter, above the antisun he observed a neutral point which nowadays bears his name (see BarraI1858). In 1840 the French meteorologist Jacques Babinet discovered a second neutral point situated above the sun (Babinet 1840). Since a neutral point existed above the sun, from considerations of symmetry, the Scottish physicist David Brewster predicted a third point of zero degree of polarization positioned at a similar angular distance below the sun along the solar meridian. This celestial point, called nowadays the Brewster neutral point, was found later at the theoretically predicted position (Brewster 1842). Only in 1846 could Babinet confirm the existence of the Brewster point (Brewster 1847). Figures 4.1A and 4.1B show the relative positions of the Arago, Babinet and Brewster points in the sky. Although he provided a succinct theoretical explanation for the maximum polarization of skylight at 90° from the sun in his pivotal paper on sky colour and polarization, Strutt (187l), alias Lord Rayleigh surprisingly did not mention the neutral points observed in 1810,1840 and 1842 along the solar and antisolar meridian by Arago, Babinet and Brewster. Occasionally, some secondary neutral points have also been observed under special conditions associated with reflections from water surfaces (e.g. Brewster 1847; Soret 1888), turbid atmospheres after volcanic eruptions (e.g. Cornu 1884), or total solar eclipses (Pomozi et al. 2001a; Horvath et al. 2003).

24

D. F. J. Arago

carth 's shadow

AS

74TH

c ZE

A

A

F space

a AS

earth

Part II: Polarization Patterns in Nature

B J. W. Strutt I (Lord Rayleigh) 'foI

almosphere down ward skyJighl

/ +08

4TH

u

- - - -0 invisible 10 observer _. _ .... c geomclrical points

and direclions

ZI':

Fig. 4.1. A, B Schematic diagram of the normal positions of the Arago (AR), Babinet (BA) and Brewster (BR) neutral points of skylight polarization in the vertical plane induding the ground-based observer (OB), sun (5U), zenith (ZE), antisolar point (A5), and nadir (NA). From the ground, only two neutral points are visible simultaneously: either the Arago and Babinet points (A, for lower solar elevations), or the Babinet and Brewster points (B,for higher solar elevations). From the ground, the fourth neutral point (4TH) is not visible. The in sets represent the portraits of Dominique Francois Jean Arago (1786-1853), Jacques Babinet (1794-1872) and David Brewster (1781-1868), the discoverer of the neutral points. The portrait of John William Strutt, alias Lord Rayleigh (1842-1919) who developed the first theory of skylight polarization, is also shown as an inset. C The fourth neutral point cannot even be observed after sunset, because the atmosphere below the antisun is then in the shadow of the earth. D Increasing the altitude of observation, above a certain height three neutral points can be observed simultaneously: the Arago, Babinet and Brewster points. E At an appropriately high altitude, all four neutral points can be observed simultaneously. Then, the Arago and Babinet points are the neutral points of downwelling polarized skylight, while the Brewster and fourth neutral points are the neutral points of upwelling polarized earthlight. F Geometry of the spaceborne observation of the Arago and fourth neutral points. (After Horvath et al. 2002b).

4 Principal Neutral Points of Atmospheric Polarization 25

With Brewster's discovery, the three principal neutral points, and the only ones now bearing the names of their discoverers, became known. Since they were first observed, they have been the subject of many ground-based investigations because their positions have been proven to be sensitive indicators of the amount and type of atmospheric turbidity (Coulson 1988). In the second half of the twentieth century, however, the neutral points have lost their importance in applied meteorology and became a neglected tool in meteorological research (Neuberger 1950).

4.1 Video Polarimetry of the Arago Neutral Point of Skylight Polarization

Horvath et al. (1998b) measured the spatial distribution of the degree p and angle a of linear polarization of skylight in the area of the Arago point by video polarimetry under clear sky conditions in Tunisia (Fig. 4.2). The position of a neutral point is not correlated with the radi an ce of skylight. In contrast, the Arago point is clearly discernible in the p- and a-patterns: skylight is unpolarized (p = 0 %) at the Arago point, and p gradually increases with increasing angular distance from the neutral point. The direction of polarization is more or less vertical, i.e. the polarization is negative between the Arago point and the antisun, but above the Arago point the E-vectors are more or less horizontal indicating positive polarization. Hence, polarization switches from negative to positive as one passes the neutral point parallel to the antisolar meridian.

In Fig. 4.2 the measured angular distance ß of the Arago point from the antisun is ßred = 24.4°, ßgreen = 22.4°, ßblue = 29.3° in the red, green and blue, respectively. Hence at the time of recordings by Horvath et al. (1998b), the Arago point was farthest away from the antisun in the blue and slightly closer to the antisun in the green than in the red. This exceptional situation was caused by the ground reflection of light. Reflection from rough ground surfaces can introduce more or less vertically polarized light into the atmosphere. This effect enhances the region of negative polarization of the sky in those spectral ranges in wh ich the reflectivity of the ground is high. At the site of the measurements made by Horvath et al. (1998b) the ground had a typical reddish brown colour. Thus, the ground reflection was high in the red. The consequence was that in the red a considerable amount of vertically polarized light was reflected from the ground, wh ich enhanced the contribution of negative polarization in the atmosphere. Thus, the Arago point shifted slightly further away from the antisun in the red.

A similar shift of the position of the Arago point due to reflection from snow or bright sand has also been reported by other authors (e.g. Können 1985; Coulson 1988). In such cases, however, the shift of the Arago point was

26

c.. c

.3 0: N .;:

'" ö c.. ...... o ., ~ Oll .,

"0

ö

.§

.~ '" Ö 0-'o ., Ob c "'

video picture

A

B ~

backscattered sun glow

mountain ridge

650 11m (red)

~

-.....

antisun

c


550 nm (green)

degree of linear pol.riz.tion p

-9 +90·

+ 135" 180"

angle of polari zation 0-mcasured rrom the vcrt ic.al

D 450 11m (blue) ,-i!"

tt..- ....

1

2

3

Fig. 4.2. Video-polarimetrie imaging of the Arago neutral point of skylight polarization. A Video picture of the sky around the Arago point. The position of the antisun is indieated by a dot. The horizon is demareated by a mountain ridge. B-D The patterns of radianee I, degree of linear polarization p and angle of polarization a of skylight measured by video polarimetry at 650, 550 and 450 nm. In the a-patterns any particular black or white bar represents the loeal E-vector orientation as averaged over a small reetangular region around the bar. The positions of the Arago point are indicated by white dots. (After Horv<ith et al. 1998b).


observed in all ranges of the spectrum, because the ground reflection was high in all spectral ranges due to the whiteness of snow or sand. Reflection of light from huge water surfaces (lakes or sea) affects the position of the Arago point in the opposite direction. Light reflected from water surfaces intro duces horizontally polarized light into the atmosphere in all spectral ranges, which enhances the region of positive polarization. This effect results in a shift of the Arago point towards the antisun in every part of the spectrum (Können 1985; Coulson 1988).

4.2 First Observation of the Fourth Principal Neutral Point

Theoretically, for reasons of symmetry, a fourth neutral point should exist below the antisun (Fig. 4.1). However, it cannot be observed from the ground, because the region below the anti solar point is either under the horizon after sunrise (Fig. 4.1A,B), or after sunset it is in the shadow of the earth (Fig. 4.1C), thus the sub-antisolar region is not illuminated by direct sunlight, which is the prerequisite for the occurrence of the fourth neutral point. The fourth neutral point can be observed only in the sunlit atmosphere and at appropriately high altitudes in the air (Fig. 4.1E) or space (Fig. 4.1F) somewhere below the antisun. The light field in the atmosphere can be divided into two components (Fig.4.1):

1. The radiation scattered downward to the earth's surface (downwelling light field) from the sunlit sky is called "skylight".

2. The radiation directed to space (upwelling light field) and originating from scattering of sunlight in the atmosphere and reflection of light from the earth's surface is termed "earthlight" (Coulson 1988).

For a ground-based observer, the Arago, Babinet and Brewster points are the neutral points of polarized skylight (Fig. 4.1A,B), while for an air- or space-borne observer the fourth and Brewster points are the neutral points of polarized earthlight (Fig. 4.1E,F). Up to 2001, no observation of the fourth neutral point had been reported and so it has remained nameless. Apparently, the fourth neutral point has been overlooked in observational atmospheric optics, even though theoretical considerations (e.g. Rozenberg 1966) or model computations (Adams and Kattawar 1997; Breon et al. 1997) have predicted its existence. It has been mentioned only occasionally in the literature. Rozenberg (1966), for example, called it the "point observable from above", but it is not even mentioned in the most famous monographs on polarized light in nature (Gehreis 1974; Können 1985; Coulson 1988).

With this in mind, Horvath et al. (2002b) (~colour Fig. 4.5D) performed two hot air balloon flights over Hungary under clear sky conditions immedi-


ately after sunrise1• Using 1800 field-of-view imaging polarimetry, they measured the patterns of the degree p and angle a of linear polarization of skylight and earthlight at 650, 550 and 450 nm as functions of the altitude and solar elevation (---7 colour Figs. 4.3, 4.4, 4.5 and Figs. 4.6, 4.7). The results of these baHoon-borne imaging polarimetrie measurements of Horvath et al. (2002b) provided the first experimentallobservational evidence for the existence of the fourth principal neutral point within the clear sunlit atmosphere. The fourth neutral point was observed from different altitudes between 900 and 3500 m during and immediately after sunrise at the theoretieaHy predieted position, at about 22-400 below the antisun along the antisolar meridian, depending on the wavelength. The fourth neutral point has similar characteristies as the Arago, Babinet and Brewster points (---7 colour Figs. 4.3, 4.4, 4.5 and Figs. 4.6,4.7, Table 4.1):

• It is located along the antisolar meridian at the edge of the areas of positive and negative polarization of earthlight.

• At sunrise, it is at about the same angular distance below the antisun as the Brewster point is below the sun.

• Its nadir angle decreases with decreasing wavelength. • Its position and the polarizational characteristies of earthlight around it

are influenced by ground reflection, the effect of whieh decreases as the altitude increases and/or the wavelength decreases.

• Its nadir angle is decreased by multiple scattering on atmospherie aerosols increasing the areas of negative polarization of earthlight.

In the clear atmosphere, a neutral point occurs if the radiance of the normaHy positively polarized sky- or earthlight is matched exacdy by an equal quantity of negatively polarized light. Multiple scattering of light by dust, haze and other aerosol partieies in the atmosphere introduces positive or negative polarization, depending on characteristies of the partieies and the incident radiation. Under clear atmospherie conditions, multiple scattering causes more negative than positive polarization, thus the net p of skylight is reduced. The stronger the multiple scattering (i.e. the shorter the wavelength), the more negative polarization is introduced in the atmosphere (resulting in the

1 At sunrise and sunset the contribution of light reflected from the ground is small to the earthlight which is dominated by atmospheric light scattering, especially for shorter (UV, blue) wavelengths. Furthermore, at sunrise and sunset the antisun has a minimal (zero) elevation resulting in a maximal distance between the aerial observer and the earth's surface in the predicted direction of the fourth neutral point (about 20-35° below the antisun). The first prerequisite of observation of the fourth neutral point is an appropriately thick air layer below the antisun in which the sunlight can be backscattered towards the aerial observer (Fig. 4.1E). The second prerequisite is that this backscattered light must not be suppressed by the ground-reflected light. Thus, sunrise and sunset are ideal periods to observe the fourth neutral point.


70"10

70%

"" 60% ,....... c

~ .g 50% 13

:5'; . ~ 40% V"l Ö

';' ,:: 30% <I.l 0

~ ~ 200/0 e.o~

."

earthl ight

A

-30" 0" 30" 60" 90" nadir angle 9

O%+-__ ~~ __ ~ __ ~ __ ~~ .. ~~~ -9()" -60" -30" 3()" 6()" 90'

nad ir angle e 70%

Co 60%

c

E .~ 50% c , ~ o J§ 40% V"l 0 'D Co

:;- ~ 30%

~ ~ 20"10 ... "0

10%

" '0 Co

E ~ = 8. " ~ " >-o

0"10~ __ ~ __ -. ____ ,-__ -r __ ~~ __ ~

-90' -60' -30" 30" 60" 90" nadir angle e

horizon nadir horizon

t t t antisolar meridian solar meridian

70% skylight 60%

B

1\ 1 50%

40%

30%

20%

10% / \ 1

O%+-~~----r---~---.----~~~

70%

60%

50%

40%

30%

20%

10%

-90" -60" -30" O' 30" 60" zen ith angle 9

0% +-"'E,---~--~---.---'-:"'c.....i.I.,

70%

60%

50%

40%

30%

20%

10%

-90" -60' -30" ()o 30' 60' 90'

zenith angle 9

F

0% +--...Jto~ __ ~_~_--. ____ ~.L..~

-90" -60" -30" O· 30" 60" 900 zenith angle e

horizon zenith horizon

t t t ant i solar meridian solar rncridian

Fig.4.6. Degree oflinear polarization p of earthlight (A, C, E) and skylight (B, D, F) along the solar and antisolar meridian measured at 450, 550 and 650 nm as a function of the nadirlzenith angle e. Data for earthlight and skylight originate from the polarization patterns in ~ colour Figs. 4.3 and 4.4, respectively. (After HorveHh et al. 2002b).

30

neg"ive earthlight negative polariUlion polariU lion

( )( .. .. )( ) 180' pOSlIlVe polam.alion

"C " ~ .. ---------------~

~ ~ 160· 'S ':.

I 11:: Jt~~ .. ..i1fL-c-'-" .N ~ 80b C __ • ~ e G) ~ ~ • ~ ! :E8. ~60' ~ - Ji ~

.... u ~ ~ 0-5 40. : - t 1i, E 20' - • > "e : ~ 0

~~ ~ ~~-~~-~~---~~--~---T~"~~~ -90' -60' -30' O' 30' 60' 90'

nadir angle a

40'

20' "'.

0'_90' -60' -30' O' 30' 60' 90' nadir angle e

20' 0' ~Ao-,--_.----_.----,.--..;:IIIJ_ .......

-90' -60" -30' O' 30' 60' 90' nadir angle e

horizon nadir horizon

" -A antisolar meridian solar meridian .;.


ncgm ivc polarization

( ) (

skylight

po itive polariulion

neg'live polarization )( )

O' 160'

140'

120'

100'

80'

60'

40'

~~--------------- --------~

B

20'

O' ~--,--___,-""",-"""T"-_-~

180'

160'

140'

120'

100'

80'

60' 40'

20'

-90' -60' -30' 0' 30' 60' 90' zenith angle a

-------------- ---- - ---~-

D

-1 ~.M''''J------.--~ ~r

: 0'_90' -60' -30' O' 30' 60' 90'

180'

160'

140'

zcnith angle e ----------- - ---------~--~ . - ,

F

:~: Ir;" -i~ 80' o · .:v.-. d NI{öi .g f~-ea :- ::öa 60' « ß I ~ 40' \ ~

! ~ 20' ~ ! ~ O' ...... ~._____.-__T-"""T"-____fll.-.....

-90' -60' -30' 0' 30' 60' 90' zcnilh angle a

horizon zenilh horizon

" . .. -A I ·d· ';' antlsolar mendmn so ar mCn Jan

Fig.4.7. As Fig. 4.6 for the angle of polarization a of earthlight and skylight measured from the solar/antisolar meridian. (After Horvath et al. 2002b).


Table 4.1. The angular distance of the Arago, Babinet, Brewster and fourth neutral points from the nadir or the zenith as determined on the basis of the patterns of the degree and angle oflinear polarization measured at 650, 550 and 450 nm at different altitudes A. (After Horvath et al. 2002b).

Neutral point

Skylight

Arago (from zenith)

Babinet (from zenith)

Earthlight

Brewster (from nadir)

Fourth (from nadir)

Spectral range

650 550 450 650 550 450 650 550 450 650 550 450 (nm) (nm) (nm) (nm) (nm) (nm) (nm) (nm) (nm) (nm) (nm) (nm)

A =3s00m -A = 1340 m -A = 900m A=Om

56.3° 51.1° 46.6° 65.3° 49.8° 49.2° 53.4° 50.2° 53.4° 55.6° 56.9° 54.0° 62.4° 62.4° 62.4° ss±so 55.3° 57.9°

decrease of p), and the more the neutral points are displaced from the sun or antisun. The amount of multiple scattering is strongly affected by atmospherie turbidity.

The fourth neutral point was not observed during earlier air- or space-borne polarimetrie experiments and/or it escaped the attention of researchers, because

- some of these measurements were performed at longer (red or infrared) wavelengths in order to minimize the contribution of molecular scattering at shorter (UV and blue) wavelengths;

- the previous techniques were not adequate to register neutral points; - the routinely used non-imaging point-source scanning polarimeters were

not pointed towards the fourth neutral point; - unpolarized points did not show up explicitly in the polarization maps due

to an inadequate, disadvantageous colour coding and displaying of the measured polarizational data.

According to Coulson (1988, p. 242), more attention has been paid to the measurement of the positions of the Arago, Babinet and Brewster points than to any other feature of skylight polarization. This statement is now rounded off by the first observation, visualization and characterization of the fourth neutral point reported by Horvath et al. (2002b) 193 and 162 years after the discovery of the Arago point and the Babinet point, and 160 years following the first observation of the Brewster point.

5 24-Hour Change of the Polarization Pattern of the Summer Sky North of the Arctic Circle

Using full-sky imaging polarimetry, Gal et al. (2001c) measured the polarization pattern of the summer sky in Sodankylä (Finnish Lapland), north of the Arctic Circle. Since at the pI ace and time of their registrations the sun did not set, they could measure the 24-h change of the celestial polarization. This gave the opportunity to demonstrate how variable the degree oflinear polarization p of skylight can be (Fig. 5.1A) and the position of the neutral points (Fig. 5.2) within 24 h on a cloudless, visually clear day.

During the 24-h period investigated, p was always lowest in the red (Fig. 5.1A). For certain solar elevations e,p was high er in the blue than in the green, while for other es the relation was the contrary. Although the temporal change of p was non-monotonous in all three spectral ranges, there was a general trend that p decreased with decreasing e, especially in the red. There was a characteristic hysteresis in the temporal change of p: from 2-13 h the change of p in all three spectral ranges was characterized by different graphs in comparison with the case between 14 and 24+ 1(=25) h. The most anomalous feature in Fig. 5.1 A is the relatively low p at 650 nm. A decrease of p of skylight at the longer wavelengths is typical of hazy conditions, and the relatively high albedo of vegetated surfaces (pine forest in Sodankylä) in the longer wavelength range is an additional contributing factor. Generally, the turbidity (e.g. haze or dust) of the atmosphere strongly reduces the maximum of p, particulady at the longer wavelengths (Coulson 1988, p. 289).

Independently of the wavelength, the angle of polarization a of skylight was always within the range of 80° < a < 100° with an average of 90° (Fig. 5.1B), i.e. the direction of polarization was always approximately perpendicular to the antisolar meridian as expected from the Rayleigh theory. There was no systematic temporal change of a.

The most important characteristics of the Arago and Babinet neutral points were the following (Fig. 5.2):

5 24-Rour Change of the Ce1estial Polarization Pattern 33

100% ,-----------------, 180·,---------------,

A B ... . - -

2\l " 19 ...... ..... ...

7 6 l

• blue • green 0 red 0% ~--~--~--~-~

• blue • green 0 red

O·~_-~_-~_-~ _ ____l

45· solar zenith angle e 85' 45· solar zenith angle e 85'

Fig. 5.1. Spectral dependence of the degree p (A) and angle a (B) of linear polarization of skylight measured at 450 nm, 550 nm and 650 nm at 90° from the sun along the antisolar meridian versus the solar zenith angle e from 2 h (local summer time = UTC+ 3) to 13 h on 25 June 1999 in Sodankylä (67°25'N, 26°30'E, Finnish Lapland). The numbers around the graphs indicate the hours of recording. In the first and second half of the course, the neighbouring points of the graphs are connected with solid and dashed straight lines, respectively. (After GM et al. 2001c).

• The sm aller the solar zenith angle, the smaller or the larger was the zenith angle of the Babinet or Arago point, respectively.

• The longer the wavelength of skylight, the larger was the zenith angle of the Arago and Babinet points.

During the 24 h studied, there was also a hysteresis in the temporal change of the zenith angle of the Arago and Babinet points: when the sun moved along the first half of its arc in the sky, the change of the zenith angle of the neutral points in all three spectral ranges was characterized by different graphs in comparison with the case when the sun moved along the second half of its arc.

Figures 5.1A and 5.2 demonstrate that p of skylight and the zenith angle of the neutral points can considerably change within some hours even if the sky is visually clear. The rather unsettled temporal variation of p of light from clear skies and the hysteresis of this variation seen in Figs. 5.1A and 5.2B,C show that the p-pattern of the clear sky is temporally unstable (i.e. for the same solar position at different times significantly different p-values can occur at a given point of the clear sky) in comparison with the relatively stable a-pattern (i.e. the a-values at a given point of the clear sky are approximately the same at different times if the solar position is the same). Thus, it is not surprising that polarization-sensitive animals which orient with the aid of celestial polarization use the a-pattern rather than the p-pattern (e.g. Wehner 1976,1994).

34

D. F. J. Arago (17 6-1853)

D_ Brewster (1781-1868)

A

90·,.-~---'i----------,

Arago neutral point 55· .j--.-~-~-~-.-~----1

50' solar zenith angle e 8S'

Part I1: Polarization Patterns in Nature

borizon

J. Babine' (1794-1 72)

60"~---------~~

c

• bluc • green

o red

Babinet neutral point

solar zCl1ith ang le e 8S'

5 24-Hour Change of the Celestial Polarization Pattern 35

Fig. 5.2. A Hourly positions of the sun and Arago, Babinet and Brewster neutral points of skylight polarization on the firmament evaluated from a 24-h series of the celestial polarization patterns measured by full-sky imaging polarimetry on 25 June 1999 in Sodankylä. The positions of the sun are indicated by dots, and next to them the tirnes of recording are shown. The positions of the Arago and Babinet neutral points measured in the red (650 nm), green (550 nm) and blue (450 nm) spectral regions are indicated by white, grey and black dots, respectively. At a given solar position, the Babinet point is placed on the solar meridian while the Arago point on the antisolar meridian. Black squares represent the predicted positions of the Brewster neutral point. For a few hours the positions of the Arago and Babinet points could not be evaluated from the recordings. The ellipses represent the trajectories of the sun and the Arago and Babinet points fitted to their hourly measured positions by the method of least squares. The insets in the corners show the portraits of Arago, Babinet and Brewster. East is on the left of the cornpass rose, because we are looking up through the celestial dome rather than down onto a map. B, C The change of the zenith angle of the Arago and Babinet points for the red, green and blue spectral ranges as a function of the solar zenith angle (J. The nurnbers around the graphs indicate the hours (local summer time, UTC+3) of recording. In the first and second half of the course the neighbouring points of the graphs are connected with solid and dashed straight fines, respectively. (After Geil et al. 2001 c).

6 Polarization Patterns of Cloudy Skies and Animal Orientation

6.1. Polarization of Cloudy Skies

The polarization of light originating from an area of the sky covered by clouds (termed "cloudlight") consists of two components:

1. The first originates from the cloud itself. White light illuminating the cloud remains white, but becomes partially linearly polarized after scattering on the cloud partieies (iee crystals or water droplets).

2. The second component is caused by the scattering of light within the air column between the cloud and the observer. This column scatters bluish and partially linearly polarized light.

Apart from very high or distant clouds, the intensity of the first component is much higher than that of the second. When the clouds and the atmosphere underneath them are directly lit by the sun (in a partly clouded sky, under thin clouds or in fog), the angle of polarization a of cloudlight follows the same geometrieal rule as in the case of blue sky. Because of the randomizing effect of multiple scattering within clouds, the degree of linear polarization p of the first component is usually much lower than that of the clear sky. In general, the first component dominates, so the net p of cloudlight is rather low and usually reaches maximal values of approximately 40 % at 90° from the sun (Können 1985, pp 40-41). As there are many different types of clouds, and as p of cloudlight depends on a multitude of factors, p may differ from cloud to cloud: it is usually lower for denser clouds because of the randomizing effect of diffuse scattering by the cloud particles.

In contrast to iee-clouds, water-clouds are strongly polarized not only at 90°, but also at approximately 145° from the sun (due to rainbow scattering), where p can reach 60 %, i.e. potentially higher values than in the background skylight (Können 1985, pp 42-43). If the clouds are not thin and/or parts of them are not directly illuminated by the sun, their polarizational characteristies differ from those discussed above. Under a heavily overcast sky, when the cloud layer is several km thick, the illumination comes more or less from all

6 Polarization Patterns of Cloudy Skies and Animal Orientation 37

directions and, hence, p of cloudlight is strongly reduced (Können 1985, pp 42-43). More light comes from the zenith, where the clouds look thinnest, than from the horizon, and the cloudlight is horizontally polarized. p of this cloudlight reaches maximal values of 10-20 % just above the horizon and decreases rapidly towards the zenith, where it is 0 %. A similar polarization pattern occurs in fog not illuminated by direct sunlight. When the clouds are very thick and the visibility is poor (e.g. during rain), the illumination is extremely diffuse, so that p of cloudlight is reduced to zero.

6.2. Continuation of the Clear-Sky Angle of Polarization Pattern Underneath Clouds

Stockhammer (1959) hypothesized that the scattering of direct sunlight between clouds and the earth's surface may generate an E-vector pattern that continues the pattern present in the cloudless celestial regions. Brines and Gould (1982) as well as Pomozi et al. (2001b) and Horv<ith et al. (2002a) showed by measurements that the angle of polarization a is the most stable and predictable parameter of skylight even under a wide range of atmospheric conditions, whereas p, radi an ce and spectral composition are highly variable and hence, less reliable as orientation cues for animals. Furthermore, a of skylight undergoes only minor changes with wavelength, while p is strongly dependent on wavelength, especially in the blue and UV (Coulson 1988).

Using full-sky imaging polarimetry, Pomozi et al. (2001b) showed that the a-pattern underneath certain clouds approximates that of the clear sky (~ colour Fig. 6.1). Thus, the celestial E-vector pattern continues underneath clouds under certain atmospheric conditions, such as when the air columns beneath clouds or parts of clouds are lit by direct sunlight (1) obliquely from above for smaller solar zenith angles, (2) from the side as with bright white cumuli, or (3) from below as sometimes occurs at dawn and dusk. Apart from heavy overcast skies with multiple cloud layers, such conditions occur frequently if the sky is partly cloudy. The reasons for this are the following: (1) If part of the air column underneath clouds is lit directly by the sun, the distribution of a of scattered light originating from the sunlit part of the column is the same as that of the clear sky. (2) The scattering of direct sunlight on the cloud particles results in the same E-vector pattern as that of the blue sky (Können 1985).


6.3. Proportion of the Celestial Polarization Pattern Useful for Compass Orientation Exemplified with Crickets

It is known from electrophysiological recordings from the polarization-sensitive interneurons in the cricket's (Gryllus campestris) medulla that these neurons respond reliably to E-vectors if p > 5 % and that the standard deviation for the reliability of the E-vector detection of these neurons is approximately ±6S for 5 ~ P ~ 10 % and ±4° for p > 10 % (Labhart 1988,1996). If P < Pthresh

old = 5 %, crickets cannot perceive the skylight polarization. The polarizationsensitive visual system of crickets determines the direction of the sun from the distribution of a of the clear sky (ac/ear sky). If in a cloudy region of the sky, the angle of polarization ac/oud differs considerably from ac/ear sky' the use of ac/oud will reduce the accuracy of the determination of the sun's direction if P

> Pthreshold. Crickets are not confronted with such a reduction in accuracy if the difference between ac/ear sky and ac/oud is below a certain threshold L1athreshold'

which is not lower than the reliability (±4-6S) of the E-vector detection of their polarization-sensitive neurons. Hence, regions of the sky that provide reliable compass information are characterized by P > Pthreshold = 5 % and/or I ac/earsky - ac/oudl ~ L1athreshold = 4-6.5°.

On the basis of these two conditions, Pomozi et al. (2001b) calculated the proportion q of the celestial polarization pattern that can be used by crickets for reliable E-vector orientation under different meteorological conditions. ~ Colour Fig. 6.2 presents two examples derived in this way using the data in row 1 of ~ colour Fig. 6.1A-C and 6.1F-H. ~ Colour Fig. 6.2 demonstrates that surprisingly large parts of a cloudy sky can be used by the insect for compass orientation. Pomozi et al. (2001b) also investigated the wavelengthdependency of q (Tables 6.1,6.2). The following results were obtained for both clear and cloudy ski es:

1. Because of the spatial distribution of p, q is smaller in the solar than in the antisolar half of the celestial hemisphere.

2. The greater the amount ofhaze and/or aerosol concentration, the sm aller is p, and hence the smaller q iso

3. In general, in clear skies, q is always very high (> 80 %). It is influenced by the spectral content, the solar zenith angle and, of course, the meteorological conditions.

4. The lower the solar elevation, the larger q iso

For cloudy skies a further conclusion was drawn (Tahle 6.2, columns F-H in ~ colour Fig. 6.1):

5. In general, q increases with decreasing wavelength for cloudy skies.

6 Polarization Patterns of Cloudy Skies and Animal Orientation

Table 6.1. Proportion q (in %) of the polarization pattern of the clear sky useful for cricket navigation at 650 (R), 550 (G) and 450 nm (B) evaluated from the clear-sky polarization patterns in ~ colour Fig. 6.1A-C. The degree of linear polarization p > 5 %. Number of pixels for the entire sky = 543000. The union of overexposed regions of skies studied in the different spectral ranges was not included. Row: row number in ~ colour Fig. 6.1 (After Pomozi et al. 2001b).

Row R G B

1 98.2 98.9 98.5 2 99.9 99.9 99.8 3 97.8 98.9 97.7 4 98.5 98.9 98.7 5 90.7 94.3 94.6 6 87.5 92.9 93.1 7 83.6 90.7 92.9

39

Only if parts of the clouds and the air columns beneath them are not directly lit by sunlight, does q decrease. This can result from a low p (rows 2,6 and 7 of ~ colour Fig. 6.1) and/or from situations in which the clear-sky apattern does not extend into the air columns underneath clouds (row 6 of ~ colour Fig. 6.1). The closer the sun to the horizon, the larger the cloudy-sky values of q, because the low solar elevation increases the chance that the air volumes underneath clouds are directly illuminated by the sun.

It is a rather wide-spread belief that animals using celestial polarization compass can orient themselves solely by means of the polarization pattern of the clear, blue regions of the sky when the sun is not visible. The reason for this is the assumption that the clouds reduce the extent of sky polarization pattern useful for animal orientation by decreasing p and causing large disturbances in a. However, we have seen above that the celestial a-pattern continues below the clouds under certain atmospheric conditions. This phenomenon can apparently help animal orientation, because not only the a-pattern of clear, blue sky regions, but also the a-pattern underneath certain clouds enables polarization-sensitive animals to determine the position of the invisible sun, if p of the cloudlight is not lower than the perceptual threshold of the visual system. Hence, clouds decrease the extent of skylight polarization useful for animal orientation much less than assumed earlier.

40 Part 11: Polarization Patterns in Nature

Table 6.2. Proportion q (in %) ofthe polarization pattern of the clear-sky regions and the clouds useful for cricket navigation at 650 (R), 550 (G) and 450 nm (B) evaluated from the polarization patterns of cloudy skies in ~ colour Fig. 6.1F-H. For clear-sky regions, the degree of linear polarization p > 5 %. For cloudy regions p > 5 % and I ac/ear sky - ac/oudsl

:-::; 6S, where ais the angle of polarization. Number of pixels for the entire sky = 543000. The union of overexposed regions of skies studied in the different spectral ranges was not included. (Row: row number in ~ colour Fig. 6.1) (After Pomozi et al. 2001b).

Clear sky regions Clouds

Row R G B R G B

1 95.5 93.8 77.8 47.6 59.3 62.3 2 52.8 53.9 68.4 12.9 11.1 18.4 3 90.5 93.9 96.0 27.6 32.6 42.6 4 95.2 97.4 99.3 21.8 21.4 29.4 5 94.7 97.5 97.8 19.9 21.6 24.2 6 92.9 94.6 94.1 6.8 8.3 12.9 7 71.7 76.8 86.6 3.1 2.9 9.4

7 Ground-Based Full-Sky Imaging Polarimetrie Cloud Detection

In many meteorologie al stations the accurate determination of sky conditions, especially the detection of clouds, is a desirable yet rarely attainable goal. Traditionally, sky conditions are reported by human observers with considerable discrepancies between individual and subjective reports. In practiee, employing human observers is not always feasible due to budgetary constraints. Human observers can be replaced by automatie full-sky imager systems, like the Scripps-produced Whole Sky Imager, or the TSI-880 Total Sky Imager produced by the Yankee Environmental Systems, Inc. (YES 2001). These systems provide real-time processing and display of daytime sky conditions using common image processing algorithms, whieh detect the clouds radiometrieally by filtering the colour picture of the sky so that the approximate value of the cloud cover fr action can be calculated.

Using the additional information obtained by evaluating both the degree and angle of polarization patterns of cloudy skies measured by full-sky im aging polarimetry in the red (650 nm), green (550 nm) and blue (450 nm) spectral ranges, the algorithms of radiometric cloud detection can be significantly improved. Horvath et al. (2002a) developed such an efficient combined radiometrie and polarimetrie algorithm whieh performs the detection of clouds more efficiently and reliably as compared with an exclusively radiometric cloud detection algorithm (Figs. 7.1, 7.2; Table 7.1). In the future, similar polarimetrie algorithms can accomplish cloud detection with ground-based automatie instruments, which could be a new generation of the presently existing ground-based automated total sky imagers using exclusively radiometrie algorithms for cloud detection.

Using full-sky imaging polarimetry, one obtains the values of nine optieal variables for every pixel of the sky image: Ir,Ig, Ib, Pr' Pg, Pb' ar, ag, ab' i.e. radian ce I, degree of linear polarization P and angle of polarization a measured in the red (r), green (g) and blue (b) spectral ranges. The essence of the cloud detection algorithm of Horvath et al. (2002a) is that for every pixel of the sky pieture seven decisions are made: (1) Analysing the values of Ir' Ig and Ib, the colour of the pixel is determined, and it can be decided whether the pixel belongs to a colourless cloud or to a blue sky region. (2)-(7) Using the values

42

photograph

red (650 nrn)

visually detected cloud

PCC-56. 1%

PU0=2.4%

green (550 nrn)


radiornetrically detected clouds

PC0=50.7%, PSDC-4.5%, PCDS=6.0% PUO=22 .6%

PUO~7 .2%

blue (450 nrn)

cloud clear sky

aunder- or V overexposure

Fig.7.1. A Photo graph of a partially cloudy sky. B Cloudy (white) and clear (black) sky regions detected visually by the naked eye in picture A. C Clouds detected radiornetrically, where the under- or overexposed celestial areas are chequered. D-I Clouds detected polarirnetricallyat 650, 550 and 450 nrn using the patterns of the degree or angle of linear polarization rneasured by full-sky irnaging polarirnetry. PCC Proportion of cloud cover deterrnined by the different detectors IRGB, PR, PG, PB, aR, aG and aB. PSDC Proportion of (clear) sky detected (erroneously) as cloud; PCDS proportion of clouds detected (erroneously) as (clear) sky; PUD proportion of under- or overexposure. (After Horvath et al. 2002a).

7 Ground-Based Full-Sky Imaging Polarimetrie Cloud Detection 43

of Pr' Pg, Pb' ar, ag or ab' it can again be decided whether the pixel is part of a cloud or a clear sky region.

Every decision is the outcome of its specific subalgorithm, called "detector". Detector (1) is symbolized by IRGB, since it uses the I-values measured in the red (R), green (G) and blue (B) spectral ranges. Detectors (2)-(7) are symbolized by PR, PG, PB, aR, aG and aB, because they use the measured values of Pr' P g' Pb' ar, ag or ab' respectively. If detector IRGB identifies a pixel as "cloud", the pixel qualification is weighted by 3. The total weight of a pixel qualification is i, if it is identified as "cloud" by i detectors among detectors PR, PG, PB, aR, aG, aB. The partial weight is 0 in every case when the pixel is identified as "clear sky" by a given detector. If the investigated pixel is under- or overexposed in at least one of the R, G, B spectral ranges, detector IRGB is inactive resulting in a 0 partial weight value. Similarly, any other detector is inactive, if the pixel is under- or overexposed in the corresponding spectral range.

The partial weights are summed up, thus finally the investigated pixel has a total weight n ranging from 0 to 9. n tells how many times the pixel was identified as "cloud"; n is called the "number of cloud identification". At a given number m of active detectors, n is proportional to the likelihood of cloud: the high er n is, the greater the prob ability that the pixel belongs to a cloud in the pieture. The authenticity (or reliability) of n is proportional to m. The distributions of the n- and m-values in the sky can be represented by colour- or grey coded maps (Fig. 7.2).

In the case of "radiometric cloud detection" only detector IRGB is used. "Polarimetrie cloud detection" uses only detectors PR, PG, PB, aR, aG and aB. We speak about "combined (radiometrie and polarimetrie) cloud detection" if all seven detectors are used. The combined cloud detection algorithm has seven control parameters. Setting their values appropriately, certain types of clouds can be reliably detected.

The optimal values of these control parameters are empirically determined. In the digitized colour picture of a given cloudy sky the clouds are visually identified by inspection with the naked eye and each pixel is marked accordingly. The resulting cloud pattern serves as a "control pattern" (Fig. 7.1B). Changing the value of the control parameter of a given detector, the visually detected control clouds are compared with the clouds recognized by the detector. The following quantities are calculated: (1) the proportion PCDS of clouds detected (erroneously) as (clear) sky, (2) the proportion PSDC of (clear) sky detected (erroneously) as cloud, (3) the proportion PCC of cloud cover, (4) the proportion PUO of under- or overexposed pixels, (5) the proportion PED = PCDS + PSDC of erroneous detection. That value of the control parameter of a given detector is considered as optimal at whieh PED is minimal, i.e. where the correlation between the pixels of the algorithmieally and visually detected clouds and clear sky regions is maximal.

In Table 7.1 the lower (PCCmin) and upper (PCCmaJ limits of the proportion of cloud cover PCC determined by the radiometrie, polarimetric and com-

44

map ofcloud likelihood

combined map of authenticity and cloud likelihood

numbcr n of cloud identification o 1 2 3 4 5 6 7 8 9

111(9)*=5

11 likclihood ofcloLld ~ ",~,( m)

O --------------~~~) 1

J ( 0 likclihood of clcar sky = 1 __ 11_

,,_,(m)


B

map of authenticity

number 111 of active dctectors (authcnticity of detcct ion)

o 2 4 9

radiometrically and polari metrically dctected clouds

PCC 53 . S~ •. I'SDl' 6.1 %.I'l'DS- 7.M% I'UO-O.~%

i CIOlld

clcar sky undcr- 01' ovcrcxposurc

7 Ground-Based Full-Sky Imaging Polarimetrie Cloud Deteetion 45

Fig.7.2. A Grey-eoded map of the number n of cloud identifieation ealculated for the partially cloudy sky in Fig. 7.1A. B Grey-eoded map of the number m of aetive (neither underexposed nor overexposed) deteetors ealculated for the partially cloudy sky in Fig. 7.1A; m is proportional to the authenticity of the (cloud or clear sky) deteetion. C Map eombining maps A and B.At agiven m-value, n/nmaim) is the likelihood of cloud, while I-n/nmaim) is the likelihood of clear sky. D Cloudy (white) and clear (black) sky regions are deteeted by the eombined (radiometrie and polarimetrie) algorithm sueh that pixels with larger or smaller n(m) than n(m)* were eonsidered to belong to clouds or clear sky regions, respeetively. For n(2) * = 1, n(4)* = 3 and n(9) * = 5 (the positions of whieh are indieated by white vertical bars in the grey palette) the proportion of erroneous deteetion PED = PCDS + PSDC is minimal. The under- or overexposed sky regions (m = 0) in the maps are chequered. (After Horv<ith et al. 2002a) .

....

bined cloud detection algorithms are eompared. The real value (PCCr) of PCC is somewhere between these two extrema. As an approximate value of PCCr' 56.1 % was obtained by visual cloud deteetion in Fig. 7.1B. The reliability of a cloud deteetion algorithm is eharaeterized by the differenee .1PCC = PCCmax -PCCmin: the smaller .1PCC is, the high er is the reliability. We ean see in Table 7.1 that .1PCC is largest (33.1 %) for the radiometrie, smaller (20.8 %) for the polarimetrie and smallest (14.7 %) for the eombined cloud deteetion. In the ease of the eombined cloud deteetion, the interval in whieh PCCr ean be, is about half of that obtained for the radiometrie cloud detection. This demonstrates well that the combined algorithm ean deteet clouds more reliably than the exclusively radiometrie or the purely polarimetrie algorithm alone. The power of the eombined cloud deteetion algorithm in eomparison with the radiometrie one is ensured by the faet that the information contributed by the radiometrie subalgorithm (deteetor IRGB) to the final decision (cloud or clear sky) is only one third by weight. The polarimetrie algorithm is more reliable than the radiometrie one, beeause the former is based on the use of twiee as many deteetors as the latter.


Table 7.1. The results of cloud deteetion for the cloudy sky in Fig. 7.1A. The number of pixels of the entire sky is N = 346207, to whieh all pereentage values are related. (After Horvath et al. 2002a).

PCCdet - PSDC = PCCmin :<:; PCCr:<:; PCCmax = PCCdet+PCDS + PUO L1PCC (%) (%) (%) (%) (%) (%) (%) (%) (%)

PR 45.9 9.7 = 36.2 :<:; 56.1 :<:; 80.8 45.9 + 18.7 + 16.2 44.6

PG 38.2 6.4 3l.8 :<:; 56.1 :<:; 62.9 38.2 +22.3 + 2.4 31.1

PB 49.9 8.5 4l.4 :<:; 56.1 :<:; 68.3 49.9 + 11.2 + 7.2 26.9 aR 59.7 13.8 45.9 :<:; 56.1 :<:; 84.8 59.7 + 8.9 + 16.2 38.9

aG 55.1 16.1 39.0 :<:; 56.1 :<:; 72.6 55.1 + 15.1 + 2.4 33.6

aB 51.6 16.0 = 35.6 :<:; 56.1 :<:; 75.8 = 51.6 + 17.0 + 7.2 40.2

IRGB 50.7 4.5 46.2 :<:; 56.1 :<:; 79.3 50.7 + 6.0 + 22.6 33.1

POL 59.4 12.1 = 47.3 :<:; 56.1 :<:; 68.1 59.4 + 7.9 + 0.8 20.8

COM 53.5 6.1 47.4 :<:; 56.1 :<:; 62.1 53.5 + 7.8 + 0.8 14.7

IRGB radiometrie deteetor; POL polarimetrie deteetor involving deteetors PR, PG, PB, aR, aG and aB; COM eombined deteetor eombining the IRGB and POL deteetors; PUO proportion of under- or overexposure; PED proportion of erroneous deteetion; PCDS proportion of clouds deteeted (erroneously) as (clear) sky; PSDC proportion of (clear) sky deteeted (erroneously) as cloud; PED = PCDS + PSDC, PCC proportion of cloud cover; PCCr "real"value of PCC deteeted visually; PCCdet value of PCC deteeted by a given algorithm; PCCmin lower limit of PCC; PCCmax upper limit of PCe. In the first row of the seeond (widest) eolumn of the table the inequality PCCmin :<:; PCCr:<:; PCCmax is seen, where PCCmin = PCCdet - PSDC and PCCmax = PCCdet + PCDS + PUo. The pereentage values of these terms are given in the table for the different types of deteetor. Differenee L1PCC = PCCmax - PCCmin gives the uneertainty or error of PCCr: the value of PCCr is somewhere between PCCmin and PCCmax•

8 Polarization Pattern of the Moonlit Clear Night Sky at Full Moon: Comparison of Moonlit and Sunlit Skies

The light of the night sky originates from the following main natural sources (in order of brightness): (1) moonlight, (2) stars and planets, (3) the Milky Way, (4) zodiacallight, (5) airglow, and (6) the light from these sources scattered by the earth's atmosphere. Light pollution from artificial city lighting also affects many sites' night skies. Most of these sources are weakly polarized, the airglow is unpolarized (Wolstencroft and Brandt 1974), and the scattered skylight may be strongly polarized. At full moon, in extended regions of the sky around the moon and antimoon, the moonlight scattered and polarized in the atmosphere overwhelms all other sources of the night sky.

Moonlight has nearly the same spectral composition as sunlight, but with a shift somewhat toward the red (Kopal 1969). While the sunlight is always unpolarized, the moonlight is slightly partially linearly polarized, and at full moon it is unpolarized (Pellicori 1971). With the increase of the phase angle ß, its polarization is negative, i.e. its direction of polarization is parallel to the plane of sight. The degree p of negative polarization culminates at ß = 11 0 ,

then it decreases. At ß = 23.51 0 the moonlight is unpolarized again. With the further increase of ß, the polarization of the moonlight becomes positive, i.e. its direction of polarization becomes perpendicular to the plane of the direction of sight. In the positive range of polarization, the maximum p = 8.7 % is reached on the first or second day following the fourth quarter. Then p decreases again until it becomes zero 3 days prior to and after the new moon. Following the new moon, p increases until its maximum of 6.6 % 1 day prior to the first quarter. Following this stage, in the same way as after the full moon, p of the moonlight decreases to zero, then its angle of polarization turns over again.

Using full-sky imaging polarimetry, Gal et al. (2001a) compared the polarization patterns of moonlit and sunlit skies (Fig. 8.1) as weIl as the positions of the lunar/solar Arago and Babinet neutral points of clear moonlit/sunlit night/day skies (Fig. 8.2). They confirmed experimentally that the polarizational characteristics of a moonlit night sky are practically identical with

48

A

radianc,c I

C angle of

po tari2.at ion Cl

moonlit night sky

4


orth

~.+-South 2

0%

100"10 degr •• of

po lari~tion p

181r

angle of polariwtion Cl

measured from the loeal meridian

® overcxpo urC

4

5

6

8

9

o

radiance I

E degrce of

polarization p

F angle of

polariwtion Cl

sunlit day sky

Fig. 8.1. A-C Spatial distribution of radiance, degree and angle of linear polarization over the entire clear moonlit night celestial hemisphere approximately at full mo on with a phase angle of9°4' measured in white light for different zenith angles of the moon. D-F As columns A-C for the clear sunlit day sky for approximately the same zenith angles of the sun as those of the moon. The positions of the moon and sun are indicated by black or white dots. East is on the left of the compass rose because we are looking up through the celestial dome rather than down onto a map. (After Ga! et al. 2001a).

8 Polarization Pattern of the Moonlit Clear Night Sky at Full Moon 49

EaSI ~-----------------------~

SOlllh

Fig.8.2. Positions of the moon (crosses), sun (circles), lunar (black dots) and solar (grey triangles) Arago and Babinet neutral points of the clear moonlit/sunlit night/day sky evaluated from the measured celestial polarization patterns in Fig. 8.1B,C/E,F. The numbers next to the dots/triangles correspond with the row numbers in Fig. 8.1. (After Ga! et al. 2001a).

those of a sunlit sky if the position of the moon is the same as that of the sun. This follows from the theory of light scattering: considering the structure of the celestial polarization pattern, it is all the same if the mo on or the sun is the light source.

These results may have functional significance for arthropod navigation. For example, Kerfoot (1967) reported that the foraging activities of nocturnal bees last as long as the moon stays above the horizon. Furthermore, several insect and crustacean species have been shown to use the mo on as a navigational aid. Desert ants of the genus Cataglyphis are exclusively diurnal foragers, but when experimentally tested at night, they take the full moon for the sun, and navigate according to what a time-compensated sun-compass would predict (Wehner 1982).


However, what about nocturnal insects, i.e. species that usually forage at night? Even if they do not possess a moon-compass, they could use the moon and the nighttime a-pattern as a short-term compass which is calibrated anew each time the animals start a foraging journey (see e.g. Lambrinos et al. 1997). The relatively low radi an ce of the moonlit night sky may not be a serious limiting factor of the navigation by nocturnal insects, because field crickets (Gryllus campestris), for instance, can respond to polarization at intensiti es that are lower than that of the clear, moonless night sky (Herzmann and Labhart 1989). Crickets are also active at night and may orient on the basis of celestial polarization patterns during dawn and dusk, or even at night when the sky is lit by the moon. They perceive the skylight polarization in the blue with the dorsal rim area oftheir compound eyes (see Chap.17.4).

Talitrid sandhoppers use a sun- and moon-compass in their offshoreonshore orientation (during their movements perpendicular to the shoreline), and it was suggested that they can use the celestial polarization pattern for this task if the sun is not visible (Ugolini et al. 1999). The same could also be true at night during full moon, when the moon is occluded by clouds.

The dung beetle Scarabaeus zambesianus orients by means of the polarization pattern of the moonlit night sky if the moon is not visible (Dacke et al. 2003b).

9 Imaging Polarimetry of the Rainbow

The rainbow, a coloured eircular band visible at about 42° from the antisolar point if sunlight falls onto water droplets underneath clouds, is one of the most spectacular phenomena in nature (Lee and Fraser 2001). One of its peculiar characteristics is that rainbow light is strongly polarized with the direction of polarization tangential to the bow, as discovered by the French physieist, Jean Baptiste Biot in 1811.

Lee (1991) isolated the rainhow's intrinsic colours exploiting the fact that rainbow light is highly linearly polarized compared with light from the background. Barta et al. (2003) performed the first imaging polarimetrie investigation of the rainbow (Fig. 9.1). In the p-patterns ofFig. 9.1 the rainbow shows up most strikingly at 650 nm, while at 450 nm it is hardly visible l . Crossing the primary rainbow upward, there is an abrupt decrease of p. Moving off downward from the primary rainbow, p gradually decreases with some oseillations due to the supernumerary rainbows. At all three wavelengths, in the a-patterns of Fig. 9.1 the rainbow does not show up, which demonstrates that there is no contrast in a between the rainbow and its celestial background.

The strong polarization of the rainbow is the consequence of the path that the beams of light gene rating the rainhow follow in the drops (Können and de Boer 1979; Können 1985): In the primary or secondary rainbow, the beams suffer one or two reflections in the drop, respectively. Since these reflections happen at angles very near the Brewster angle, the reflected light is highly polarized, and the direction of polarization is always perpendicular to the scattering plane, i.e. tangential to the are of the bow. Since the light below the primary rainhow and above the secondary one rises in the same way as the light of the rainbow itself, its E-vector direction is also tangential to the bow.

1 The explanation ofthe observation, that in the p-patterns ofFig. 9.1 the rainbow shows up best in the red, but is hardly visible in the blue, is that the background light was unpolarized or only very weakly polarized due to multiple scattering, and it was most intense in the blue due to Rayleigh scattering. Therefore it could most strongly desaturate and depolarize the coloured and polarized rainbow light in the blue. Furthermore, the partially polarized light scattered in the air column between the observer and the rainbow, and being most intense in the blue, partly overwhelmed the rainbow light, especially in the blue.

52

total radiancc

blue (450 nm)

p ,----I -

a


0%

degree of polarizalion p

100%

green (550 nm)

-90' +90'

angle of polariz'lion a mcasurcd from lhe radius

red (650 nm)

Fig. 9.1. Patterns of the total radiance, the degree of linear polarization p and the angle of polarization a of a rainbow above the sea surface measured at 450, 550 and 650 nm by 1800 field-of-view imaging polarimetry for solar elevation of 290 41'. At a given point of the a-patterns, a is measured clockwise from the radius of the bow. (After Barta et al. 2003).

Though the scant light from the Alexander's band between the primary and secondary rainbow arises from reflections from the outside surface of the water droplets, background skylight plays a large role also. This light is also tangentially polarized with respect to the bows.

10 Which Part of the Spectrum Is Optimal for Perception of Skylight Polarization?

In many insect species the perception of skylight polarization is mediated by a group of anatomically and physiologically specialized ommatidia in an upward-pointing narrow dorsal rim area (DRA) of the compound eye. The ommatidia in the DRA contain two sets of monochromatic and highly polarization-sensitive photoreceptors with orthogonal microvilli (see Chap. 16.7). The spectral type of the receptors in the DRA is ultraviolet (UV) in flies, honeybees, bumblebees, desert ants, scarab beetles and spiders, blue in crickets, desert locusts and cockroaches, and green in the beetles Melolontha melolontha and Parastizopus armaticeps (Table 10.1).

However, the perception of skylight polarization in the UV is rather surprising, because both the radi an ce land the degree of linear polarization p of skylight in the UV are considerably lower than in the blue and green (Figs. 10.1 and 10.2). Furthermore, the atmospheric absorption of light is much high er in the UV than for longerwavelengths (e.g. Henderson 1970). We call this the "ultraviolet paradox of perception of skylight polarization", simply"UV-sky-pol paradox" further on. In the opinion ofWehner (1994, p. 110), there is no particular region of the spectrum predestined to be used preferentially for detection of the E-vector of skylight under all possible sky conditions.

10.1 A Common Misbelief Concerning the Dependence of the Degree of Skylight Polarization on Wavelength

Some researchers are not aware of the UV-sky-pol paradox due to an erroneous belief considering the wavelength dependency of polarization of light from the clear sky. In the literature of animal polarization sensitivity, a frequently occurring misbelief is that p of scattered skylight is the highest in the UV. Several biologists (e.g. Waldvogel 1990, p. 352; Beason and Semm 1991, p. 107; Helbig 1991b, p. 288; Tovee 1995, p. 456; Shashar et al. 1995b, p. 215) tried to explain by this misinformation why certain animal species detect the


Table 10.1. Wavelengths Ama/OL at which the sensitivity of photoreceptors detecting skylight polarization is maximal in insects and spiders.

Species

Calliphora erythrocephala, Musca domestica (flies)

Apis mellifera (honeybee)

Bombus hortorum (bumblebee)

Cataglyphis bicolor (desert ant)

Cataglyphis setipes (desert ant)

Lethrus apterus, Lethrus inermis (scarab beetles)

Pachysoma striatum (desert dung beetle)

Drassodes cupreus (spider)

Gryllus campestris (field cricket)

Ama/OL (nm) Reference

330-350 Smola and Meffert (1978), Hardie et al. (1979), Hardie (1984), Philipsborn and Labhart (1990)

345-350 Helversen and Edrich (1974), Labhart (1980)

353,430 Meyer-Rochow (1981)

380-410 Duelli and Wehner (1973)

380-400 Frantsevich et al. (1977)

350 Frantsevich et al. (1977)

350 Dacke et al. (2002)

350 Dacke et al. (1999)

433-435 Labhart et al. (1984), Brunner and Labhart (1987), Herzmann and Labhart (1989)

Schistocerca gregaria (desert locust) 450 Eggers and Gewecke (1993)

Loesel and Homberg (2001) Leucophaea maderae (Madeira cockroach)

<471

Melolontha melolontha (cockchafer) -520

Parastizopus armaticeps (beetle) -540

Labhart et al. (1992)

Bisch (1999)

skylight polarization in the UV. From the context it was always obvious that these researchers wrote about polarization of light from the clear sky rather than ab out light from clouds. However, polarization measurements (e.g. Coulson 1988, p. 285) have shown that under clear atmospheric conditions, P of scattered skylight decreases with decreasing wavelength Ä (Fig. 10.1A). The reason for this has already been explained in Chap. 4. Figure 10.1B shows the dispersion of PmaiÄ} of skylight for a turbid, dusty atmosphere. Dust considerably reduces Pmax in the long-wavelength (orange, red, infrared) range, while in the short-wavelength (blue, UV) range it has only a minor effect, P in the UV being essentially the same as that for the clear atmosphere. Due to this reduction,Pmax in the UV is about as low as Pmax for A > 650 nm. However, for Ä< 650 nm,Pmax is the lowest in the UV.

10 Which Part of the Spectrum is Optimal for Perception of Skylight Polarization? 55

Fig.lO.1. A Degree oflinear polarization Psc versus wavelength A of scattered light from the clear sky measured under at a solar elevation of 10°. B As A measured in an atmosphere when it was turbid, dusty (continuous) and clear (dashed) for dif-ferent solar elevations es' (After Coulson 1988).

Fig.l0.2. Relative radiances as a function of the wavelength A. Is/A} radiance of scattered skylight measured by Hess (1939) at 90° from the sun under clear sky conditions. IclA} radiance of white cloudlight measured by Coemans et al. (l994b) at an elevation of 40° under a thick cloud deck. Ica(l) radiance of green canopylight transmitted through the leaves of cottonwood (Populus deltoides) (after Gates 1980).

:i >' 100 ~ ~ i ;

Q...!!

" '" o " 90 ': .c

.~ E '" 0 -.:: 8.", '-cn o .-u~ ~..=! ,,-" """" E 2 " l< E 0; .- " " ~ ~~ 60

A

P~(A.)

300 400 500 600 700 00 900 wavelength I.. (nm)

9 ~ 45" ----- ' .; - - - cle;r -

wavclength I.. (nm)

wavelength, I.. (nm)


10.2 Why Do Many Insects Perceive Skylight Polarization in the UV?

Several hypotheses have been published which tried to solve the UV-sky-pol paradox. In this section, we describe these hypotheses and their criticism, finally we expound a convincing solution of the paradox. However, we should emphasize that several insect species perceive the skylight polarization in the blue or green (Table 10.1). Why do these insects not use UV-sensitive photoreceptors for detection of skylight polarization? Some arguments are presented below for the advantage of perception of celestial polarization in the blue and especially in the uv. However, other important physical, biologicalor environmental factors mayaiso exist, which determine the optimal wavelength range of detection of skylight polarization in a particular animal species. Sometimes it has simply been dedared without any explanation that the UV spectral range is the least or the best reliable for perception of skylight polarization (e.g. Able 1989, pp. 252-253). Such dedarations cannot help to solve the UV-sky-pol paradox.

10.2.1 Is the Celestial Polarization Pattern more Stable in the UV?

Zdenek Sekera daimed in his communication to Karl von Frisch (1967, p. 382) that UV wavelengths are the least sensitive to "atmospheric disturbances". Relying on this suggestion, Frisch (1967) postulated that UV polarization patterns of skylight might be more advantageous as cues for orientation not only because the UV E-vector orientation ought to approximate simple theory most precisely, but also because it might be most stable during "marginal sky conditions", unlike patterns in longer wavelengths which may be easily disrupted. In other words, atmospheric disturbances may affect the E-vector direction of skylight, and such disturbances may have an increased effect on the longer wavelengths, and the least influence on UV skylight. Although in the 1960s neither the experimental nor the theoretical basis was available for this assumption and the evidence for this conjecture was very slim, this idea has become widely accepted in the literature because any strategy that could extend the conditions under which successful orientation is possible would certainly constitute a major selective advantage. The statement of Sekera (cf. Frisch 1967, p. 382) that the celestial polarization pattern might be the least sensitive to atmospheric disturbances in the UV spectral region, has been frequently cited in the literature (e.g. Duelli and Wehner 1973, p. 50; Wehner 1976, p. 110; Dacke et al. 2002, p. 215).

However, the major problem with such too general statements is that these "atmospheric disturbances" have never been precisely defined. It remained undear what "susceptibility to atmospheric disturbances" or "stability of the


celestial polarization pattern under different weather conditions" exactly mean.

10.2.2 Was the UV Component of Skylight Stronger in the Past?

It is a logical assumption that the sensitivity maximum of the monochromatic polarization-sensitive photoreceptors perceiving polarized skylight may be adapted to the wavelength where the radi an ce of skylight is maximal. Such spectral adaptations of certain receptors to the dominant radiation field of the optical environment are common in visual systems (e.g. Lythgoe 1979). Since in the present atmosphere of the earth the radiance of skylight is maximal in the blue (Fig. 10.2), the receptors detecting polarized skylight should be blue-sensitive.

Brines and Gould (1982) suggested that a possible reason why UV wavelengths are used by skylight detectors in certain animals may be that in the era when polarization sensitivity has evolved in these animals, the UV component of skylight might have been stronger than it is today. The reasons could be that the atmosphere might have attenuated the UV flux of sunlight to a lesser degree than it does today, and/or the UV radiation of the sun might have been more intensive during earlier epochs of evolution. Although the total energy emitted by the sun fluctuates by a tiny 0.1 % over an 11-year solar cyde, and solar UV radiation changes three times as strongly during a cyde as total radiation (Pearce 1998) and furthermore, the composition of the earth's atmosphere has dramatically changed during the his tory of the biosphere, the major problem with this hypothesis is that from the past there are no reliable data about the temporal change of the UV radiation reaching the earth's surface. Thus, this idea is hard to evaluate. If we accepted this hypothesis, the period during which the UV component of skylight might have been stronger than nowadays should have been in the near past on the time scale of evolution, otherwise the sensitivity maximum of skylight detectors would have adapted to the present situation, namely to the blue maximum of skylight radiation (Fig. 10.2).

Let us estimate in which spectral range a monochromatic crossed-analyzer in the DRA of insects would function optimally. If the E-vector of partially linearly polarized incident light is parallel (par) or perpendicular (perp) to the microvilli, the amount Q of skylight absorbed by a polarization-sensitive photoreceptor can be calculated as follows (Horvath et al. 2002 c):

Qpar = C oI<x A(A) I(A) [PS+ 1 +(PS-l) p(A) 1 dA,

Qperp = c or A(A) I(A) [PS+ 1-(PS-1) p(A) 1 dA, (10.1)

where cis a constant,.A is the wavelength of light, A(.A) is the absorption spectrum of the receptor (Fig. 1O.3A),I(.A) and p(.A) are the radi an ce (Fig.10.2) and


the degree of linear polarization (Fig. 10.lA) of scattered skylight, PS is the polarization sensitivity describing the fact that if the E-vector of totally linearly polarized light is parallel to the microvilli, then a receptor absorbs PStimes more light than in the case when the E-vector is perpendicular to the microvillL The difference between logQpar and logQperp is:

L1(logQ) = 10gQpar -logQperp = 10g(Qpa/Qperp)' (10.2)

The basis of E-vector contrast sensitivity of crossed-analyzers in the DRA is to compare logQpar and logQperp' Le. to evaluate the difference L1(logQ). The greater this difference, the better the detection of skylight polarization functions. Thus, maximizing L1(logQ) is optimal for receptors in the DRA. Using Eqns. (10.1) and (10.2), L1{logQ(Ama)] was calculated as a function of Amax for the graphs p(A) and frA) in Figs.10.1A and 10.3A, where Amax is the wavelength at which the receptor's absorption A(A) (Fig. 10.3A) is maximal. The result is shown by graph 1.0 in Fig. 1O.3B for PS = 7, which is a typical value for crickets, for example. One can see that this graph has a maximum at 458 nm. Hence, under the recent atmospheric radiation circumstances, the most effective polarization-sensitive photoreceptor has an absorption maximum in the

A (; Fig. 10.3. AReal and imaginary relative .<: ~ Ci. d)::::

:.: c 3 radiances In(A), n = 1.0,1.5,2.0,2.5,3.0 of <> " ~ Ec 0 scattered light from the clear sky as a -'" E ]-::-"'C ,-:;: ~-'" c. '::: function of the wavelength A. The Gauss-~e <> :;

B~ ~ ~ ian function A(A,Amax) with 50 nm half o ~ .: '" g~ bandwidth is the absorption function of a o ~ .. ;<~ . ":$ photoreceptor, the sensitivity maximum , ~o , " - " . .2 of which is at Amax' li.oCA) Radiance of blue s-= . § E ' ' e. . .

skylight today (after Seliger et al. 1994). . - 0 uv 1/ . 'bl Sl ]tt: -'" ~' VISI e ~)..._. " In(A) (n = 1.5-3.0) Imaginary radiances of _____ - r----

300 350 400 450 500 550 blue skylight obtained in such a way that wavelenglh )... (nm) the UV part (A < 400 nm) of li.arA) is mul-

tiplied by a factor n ranging from 1.5 to B 3.0. B The difference ,1(logQ) = logQpar -

i: logQperp of the logarithms of the amounts CI of skylight absorbed by a polarization-00 ~~ sensitive (PS = 7) photoreceptor with I ·c

I " microvilli direction parallel (par) and ~c .2~ perpendicular (perp) to the E-vector of 11 ~

8~ polarized skylight as a function of Amax "" , calculated for the se ries of InCA) shown in 0 I

~ I A. The maxima (marked by vertical bars) UV : visible ------- of ,1 (logQ) i.O' ,1 (logQ) i.5' ,1 (logQ) 2.0' 300 350 400 450 500 550 ,1(logQ)2.5 and ,1(lOgQ)3.0 are at Amax = 458,

A.... (nm) 442,404,390 and 380 nm, respectively.


blue. In spite of this fact Hymenoptera and Diptera, for example, use UV (A < 400 nm) receptors for this purpose. Thus, the functioning of these receptors is not as effective as could be in this regard.

Following the hypothesis of Brines and Gould (1982), let us estimate the necessary magnitude of the "ancient level" of UV radiation of scattered skylight, which would ensure that the maximum of L1[logQ(AmaxYl of photoreceptors in the DRA is shifted to the UV part of the spectrum. L1[logQ(AmaxYl was calculated for the series of In' n = 1.0, 1.5,2.0,2.5, 3.0 shown in Fig. 10.3A, where Il .o is the re cent radiance of skylight, while 11.5' 12.0' 12.5 and 13.0 are im aginary radi an ce curves derived in such a way that the UV part (A < 400 nm) of Il.iA) was multiplied by factor n. The results are shown by graphs 1.5,2.0,2.5 and 3.0 in Fig.l0.3B (for PS = 7), from which one can read that if the total radian ce of UV skylight were ab out twice as high as today, the maximum of L1[logQ(AmaxYl would be shifted to the UV.

Hence, if the ancient UV level of skylight had been at least two times higher than the recent one, it would have been advantageous for the skylight detectors to function in the UV. However, the minuseule periodic variation of the solar flux could not account for a considerable (e.g. twice) increase of the UV level of skylight in the past. Much greater variations in the UV radiation from the sun are improbable during evolution in the case of a star like the sun. Thus, it is very unlikely that an earlier enhancement of the solar UV radiation could be the clue to the solution of the UV-sky-pol paradox.

What about the absorption of UV light in the earth's atmosphere? The attenuation of the UV flux of solar radiation in the atmosphere is governed predominantly by the concentration of ozone (03) in the stratospheric ozone layer, which is greater, the higher the oxygen (02) level. The link between UV level and atmospheric oxygen concentration is that UV radiation converts oxygen into ozone. However, this cannot be the clue for the solution of the UVsky-pol paradox, because the detection of skylight polarization in animals functions between 345-400 nm (Table 10.1), and in this range of the spectrum the light absorption by ozone is practically zero. In the visible spectrum, the ozone has one absorption maximum at 600 nm, while in the UV range there are three maxima at 255, 314 and 344 nm (Rozenberg 1966). Due to absorption by the ozone layer, 300 nm is practically the effective wavelength cut-off for UV light incident on the earth's surface. Thus, the change in the oxygen and ozone concentration in the atmosphere does not influence the UV level of sunlight and skylight in that part of the spectrum where the detection of UVskylight polarization by animals happens.

10.2.3 Relatively Large Proportion of UV Radiation in Skylight?

According to Hawryshyn (1992, p. 166), even though it is potentially harmful, the relatively large proportion of ultraviolet radiation in scattered light at


least partiaHy accounts for the use of these wavelengths for the detection of polarization. However, Fig. 10.2 shows that the radiance of skylight is much lower in the UV than in the blue, where it is maximal. The maximal proportion of blue radiation in scattered skylight could account for the use of blue (rather than UV) wavelengths for the detection of skylight polarization.

10.2.4 Mistaking Skylight for Ground-Reflected Light?

Mazokhin-Porshnyakov (1969) suggested that by using UV wavelengths, animals would be fairly sure that they use polarized skylight for orientation rather than polarized light reflected from the ground, whieh is rieher in longwavelengths than skylight. In other words, using UV light might help to distinguish phototactieaHy "sky" from "ground". This argumentation was also taken over by other researchers (e.g. Wehner 1982, p. 88, 89, 123; 1983b, pp. 360-361; 1984, pp. 285-286; 1994, p. 125). In reality, this idea originates from Frisch and Lindauer (1954), who first discussed the concurrence of sky and earth in the orientation of honeybees.

Since skylight and ground-reflected light can always re ach the eye from above and below, respectively, an appropriate regionalization of photoreceptors in the eye can simply eliminate the confusion of skylight with groundreflected light (earthlight). The skylight analysers and earthlight detectors should be separately arranged in the eye in such a way that only the former can see the sky, while the latter can view only towards the ground. Then, both the skylight and earthlight detectors can function in the same visible range of the spectrum, and using the UV wavelengths for skylight detection is unnecessary. Hence, the anatomieal separation of the adequate photoreceptors can simply solve the problem of distinguishing "sky" from "ground", independently of receptor wavelength sensitivity. Indeed, this is the usual case in insects: it is only the DRA of insect compound eyes whieh is sensitive to skylight polarization and this area is oriented towards the sky, thus the ambiguities envisaged by Mazokhin-Porshnyakov (1969) and Wehner (1982, 1983, 1984, 1994) do not arise. Thereby, confusion of sky with ground would not occur, because they are viewed by different eye regions.

10.2.5 Confusion of Motion and Form for Celestial Polarization?

In the opinion ofWehner (1976), UV wavelengths may be used for orientation by means of skylight polarization so that polarization can be analysed separately from motion and form, the detection of whieh is mediated by receptors sensitive for longer wavelengths. However, an appropriate division of labour between receptors in the eye as weH as an appropriate eye regionalization can eliminate the confusion of information from motion and form with polariza-


tion information from the sky. If there are separate skylight polarization detectors as well as motion/form detectors in separate eye regions, both detector types can function in the same spectral range, and using other (e.g. UV) wavelengths for skylight detection is not necessary. Indeed, the detection of motion and form is mediated by receptors being distinct from receptors in the polarization-sensitive DRA, so that the confusion envisaged by Wehner (1976) does not arise.

10.2.6 Were UV Receptors Originally Skylight Detectors and only later Incorporated into the E-vector Detecting System?

According to Wehner (1989, p. 80), it is very likely that UV receptors evolved originally as a me ans of detecting skylight rather than for extending the spectral range of the insect's colour vision system. Wehner (1982, 1994) hypothesized that UV receptors might have been incorporated into the E-vector detecting system only later. Bees, for example, take a UV, but unpolarized beam of light for the sky, particularly for a point lying within the anti solar half of the celestial hemisphere. In contrast, an unpolarized green beam of light is taken for the sun (Brines and Gould 1979; Edrich et al. 1979; Rossel and Wehner 1984b). Furthermore, phototactie escape responses exhibited by many insect species have their sensitivity maxima in the UV (Wehner 1981). The major problem with this hypothesis is that it does not explain why the photoreceptors used originally as simple "photometrie" skylight detectors should have been sensitive to UV instead of blue or green, for instance. The radi an ce of skylight in the UV is much weaker than in the blue and green (Fig. 10.2), whieh feature should be rather disadvantageous to UV photometrie skylight detectors.

10.2.7 Maximizing"Signal-to-Noise Ratios"by UV Photopigments under Low Degrees of Skylight Polarization?

In a theoretieal approach, Seliger et al. (1994) surmised that photopigments that are most efficient under conditions of high p (under clear ski es) would have their maximum sensitivity at 450 nm, whereas UV (350 nm) photopigments would maximize the signal-to-noise ratio under low p (under cloudy skies), where the biologieally significant "signal" is the net plane-polarized single-scattered Rayleigh skylight, while the "noise" is the unpolarized, multiply-scattered skylight.


10.2.8 In the Spectral and Intensity Domain the Celestial Band of Maximum Polarization Is less Pronounced in the UV than in the Blue

It was proposed that using UV receptors in skylight navigation might also be advantageous in exploiting spectral gradients across the sky (e.g. Wehner 1984,p. 286). According to Wehner and Rossel (1985,p. 20) and Wehner (1989, p. 80), those parts of the (anti-solar) sky that exhibit the most saturated UV tinge are also the ones that exhibit maximum polarization, and this physical property of skylight patterns has been incorporated into the bee's and ant's visual system. However, the same is true for the blue wavelengths of skylight (e.g. Hess 1939; Coulson 1988): Those parts of the sky that exhibit the most saturated blue tinge are also the on es that exhibit maximum p. What is more important, the intensity gradients of skylight are much higher in the blue than in the Uv, therefore in the UV the sky is more homogeneous than in the blue (Coemans et al. 1994b). Thus, the cited argument is not sound and unable to explain just why the UV-sensitive photoreceptors are used for detection of skylight polarization. The UV sensitivity of the polarization-sensitive photoreceptors in the DRA would be rather disadvantageous in detecting the celestial UV intensity gradients. Using the blue part of the spectrum would be more advantageous due to the fact that in the blue the sky is more heterogeneous than in the UV. Consequently, the celestial band of maximum p (at 90° from the sun) is more pronounced in the blue than in the UV. In the UV this band merges into the homogeneous UV surrounding.

10.2.9 The Proportion of Celestial Polarization Pattern Useful for Animal Orientation Is Higher in the Blue than in the Green or Red

Pomozi et al. (2001b) proved experimentallythat the proportion q of the celestial polarization pattern available for use in animal navigation measured in the red, green and blue spectral ranges are greater than about 80 % for dear skies. Thus, under dear sky conditions there is no selective advantage for the shorter wavelengths, because the extent of the polarized dear sky usable for orientation is great enough in all parts of the visible spectrum (see Chap. 6). More serious is the consequence of the wavelength -dependency of q if the sky is doudy, because under such frequently occurring meteorological conditions q can often be considerably reduced. Pomozi et al. (2001b) have also proven that in the visible spectrum and under partly doudy skies, the shorter the wavelength, the greater the proportion q. This phenomenon may have a selective advantage for shorter wavelengths. Hence, the extension of the E-vector pattern of dear sky into celestial areas covered by douds is more useful for an E-vector compass when the skylight is perceived in the blue rather than in the green or red.

The above features of doudy and dear skies are demonstrated in Figs. 10.4 and 10.5, where the patterns of the angle of polarization of a partly doudy sky

10 Wh ich Part of the Spectrum is Optimal for Perception of Skylight Polarization? 63

~ und er- or overexposed

blue (450 nm) green (550 nm)

-90"

0"

+90'

+135' 180'

angle of polariz3lion 0;

mcasurcd from lhe local meridian

red (650 nm)

Fig. 10.4. Patterns of the angle of polarization of a partly cloudy sky measured by fuHsky imaging polarimetry in Tunisia in August 1999. The sun is occluded by clouds, but its position is within the overexposed (chequered) region of the sky.

0) O' u

'" .~ ~ lmder- or overexposed "0 ·90' +90' e S

135' 180'

2 angle of polarizalion 0;

I11casurcd frOll1 lhe local meridian

Ö

'" .S1

"" N 'C co

Ö 0. ..... 0 0)

ÖJJ

'" co

blue (450 nm) green (550 nm) red (650 11m)

Fig.10.5. Patterns of the angle of polarization of a clear sky measured by fuH-sky imaging polarimetry in Tunisia in August 1999 approximately at the same solar zenith angle as in the case of the cloudy sky in Fig. 10.4. In the circular pictures the radial bar with a small disk is the sun occulter. The sun is positioned behind the disko


and a corresponding clear sky measured by full-sky imaging polarimetry in the red, green and blue are shown. We see in these figures that the E-vector pattern of the cloudy sky is most similar to that of the corresponding clear sky in the blue. This conclusion is based on many similar full-sky imaging polarimetrie measurements. Figures 10.4 and 10.5 demonstrate quantitatively what Karl von Frisch and Zdenek Sekera could have only suspected (see Chap. 10.2.1): the celestial E-vector pattern at shorter wavelengths is most stable and less disrupted under cloudy conditions. In other words, the shorter the wavelength, the weaker the disturbing effect of clouds on the E-vector distribution of skylight, at least in the visible part of the spectrum.

10.2.10 Perception of Skylight in the UV Maximizes the Extent of the Celestial Polarization Pattern Useful for Compass Orientation under Cloudy Skies

According to Brines and Gould (1982), under partly cloudy meteorological conditions, or under extensive vegetation! UV wavelengths may have advantages over Ion ger ones in animal polarization orientation, because both spuriously polarized and unpolarized light resulting from reflections from the clouds or the vegetation may cause more troublesome interference at longer wavelengths. They proposed that the UV sensitivity of the E-vector detection in many animals may be at least partly an adaptation for perceiving celestial polarization patterns under conditions when useful scattering can occur only relatively close to an anima!. They argued that under clear sky conditions there may be no selective advantage for a visual system that detects skylight polarization at wavelengths where p is high. They suggested that the necessary selection pressure to use UV-sensitive skylight polarization detectors has been provided by light scattering beneath the clouds, because these scattering events produce E-vector patterns with nearly the same E-vector orientation seen in a clear sky, and result in higher p in the UV.

Barta and Horvath (2004) formed this idea into a quantitative model which gave a satisfactory solution to the UV-sky-pol paradox. They have shown that the perception of skylight in the UV maximizes the proportion q(A) of the celestial polarization pattern useful for the polarization compass, as suggested by Brines and Gould (1982), who, however, were not able to determine q(A) quantitatively in the full sky. With full-sky imaging polarimetry, Pomozi

1 Bees must often fly with most of their view of the sky obscured by vegetation. This is a constant problem for the tropical honeybees (the ancestors of all bees) living and dancing on exposed limbs in the dense tropical forests (Wilson 1971, p. 266). Brines and Gould (1982) hypothesized that under many circumstances, typical and biologically significant E-vector patterns may exist against overhead vegetation at UV wavelengths.


et al. (2001b) could measure celestial polarization patterns and calculate q(A) only in the visible part of the spectrum, because UV light was not transmitted by their fisheye lens. As long as full-sky polarization measurements are not available in the UV, model calculations can provide the relation between q(UV) and q(blue) for cloudy skies.

Let us consider the model of Barta and Horvlith (2004). Since under partly cloudy conditions the E-vector pattern of cloudy celestial regions is approximately the same as that of the corresponding clear sky regions as shown experimentally by Brines and Gould (1982) as well as Pomozi et al. (2001b), q(A) is essentially determined only by the degree oflinear polarization PSkiA) of skylight. If P ski A) at a particular direction in the sky is higher than the threshold of polarization sensitivity, the skylight from this direction can be used for a polarization compass. The high er the PSkiA) in the whole sky, the larger q(A). The skylight originating from a cloudy celestial region and reaching a groundbased ob server is composed of (1) the cloudlight with a wavelength-dependent radiance 1iA) and degree of linear polarization piA), and (2) the scattered light with 15/ A) and P seC A) from the air layer between the clouds and the ground (Fig. 10.6). Since the E-vector direction of both components is approximately the same, the net degree of linear polarization PSkiA) of skylight is the net polarized radi an ce aiA-,h)piA)IiA) + aJA-,h)pJA)1JA) divided bythe total radi an ce aiA-,h)1iA) + as/A-,h)1JA):

Psky(A)= [ad(A,h)p d(A )Id(A )+asc(A,h)psc(A) Isc(A) ]/[ aiA,h) Id(A )+asc(A,h) Isc(A)],

(10.3)

where aiA-,h) and aJA-,h) are factors describing the wavelength-dependent effect of the thickness h of the air layer underneath the clouds. Due to the absorption of cloudlight in the atmosphere, the larger the h, the lower the relative contribution ac/A-,h) of the radiance 1c/A) of cloudlight reaching the observer. On the other hand, increasing the thickness h of the air layer

Fig. 10.6. Schematic representation of the two components of cloudlight reaching a ground-based observer. Unpolarized sunlight is scattered in the air and/or in a cloud. Direct cloudlight is nearly unpolarized, (apart from the direction of rainbow scattering in water clouds), while light scattered in air is partially linearly polarized.

ground surface

66 Part I1: Polarization Patterns in Nature

between a cloud and the observer, the number of scattering events increases resulting in the increase of the relative contribution asJA,h) of the radiance Is/A) of light scattered in the air beneath clouds. Since at all wavelengths the degree of polarization P JA) of cloudlight is practically zero due to the diffuse scattering of light by the cloud particles apart from the direction of rainbow scattering in water clouds, the following approximation can be made:

Psky(A, a) ::::: asc(A,h)psc(A)Isc(A)/[acl(A,h)Icl(A) + asc(A,h)Isc(A)] = = aPsc(A)Isc(A)/[Icl(A) + aIsc(A)], where a = asc(A,h)/acl(A,h), 0 $; a

(10.4)

Since measurements of a = as/A,h)laJA,h) are not available yet, as a first approximation we assurne that the quotient aCh) is independent of A.Although the dependence of aCh) on h is also unknown, it is clear that a increases with increasing h:

- If a cloud is in the immediate vicinity of the observer, then the contribution oflight scattered in air beneath the cloud would be zero, thus a(h=O) = o.

- When a cloud is at a huge distance from the observer, then the contribution of cloudlight is weak in comparison with that of light scattered in the air between the observer and the cloud.

Figures 10.IA and 10.2 show the measured functions Ps/A) and Is/A) of scattered light from the clear sky at 90° from the sun. Figure 10.2 shows also the spectrum IJ A) of cloudlight measured by Coemans et al. (1994b) under a thick cloud deck, when the total intensity aIs/l)+Id(l) of skylight is practically the same as the intensity Id(l) of cloudlight (because a ::::: 0). Using these particular functions without any loss of generality, Fig.1O.7 A shows PSklA) calculated on the basis of Eqn. (10.4) for different values of the control parameter a. We can see in Fig. 10.7 that

• If a < 2.5 (when the cloudlight dominates, i.e. the air layer between the clouds and the observer is thinner than a threshold),psklA,a) is maximal in the UV;

• If a > 2.5, the maximum of psklA,a) is in the visible part of the spectrum; • If a > 1O,psklA,a) approximates PsJA) ofthe clear sky (Fig.l0.1A).

The reason for this is the following: although the polarized radi an ce apsisc of skylight is more intense in the blue (B) than in the ultraviolet (UV) because PsJB) > ps/UV) and Is/B) > Is/UV), in the uv the radi an ce IclUV) of cloudlight is much weaker than the radi an ce aIs/UV) of light scattered in the air beneath clouds. In other words, chan ging the wavelength A from blue to Uv, the denominator of the expression of psklA,a) given in Eqn. (10.4) decreases


100 ,-------------.-------------------------~

A 80

60

40

20

I I

elear sky P",(A)

a ~ 1000

UV i visible

O~------~·~====t'====~·~----~~------._----~ 300 350 400 450 500 550 600

wavelength Ä (nm)

450 B

1~ 400 1~-------= -----------------------------------------

350~

o 2 3 4 5 6 7 8 9 10 control parameter 0

Fig.l0.7. A The degree of polarization p'kyCA,a) of light from cloudy sky regions calculated on the basis of Eqn. (10.4) for different values of the control parameter a, using the functions P,JA) in Fig. 10.lA, and I,/-\) and IJA) in Fig. 10.2. Increasing a-values mean increasing the proportion of the polarized radiance of light scattered underneath clouds. The positions of the maxima of the curves are marked by dots. B Wavelength Amax where p'kyCA,a) is maximal as a function of the control parameter a. Amax is optimal for orientation by means of skylight polarization.

68

Table 10.2. Average degree oflinear polarization Psky of skylight measured by Brines and Gould (1982) at three wavelengths A against 20 different small cumulus clouds under hazy and clear atmospheric conditions.

Psky(%) A(nm) Sky condition

10 350 Hazy 7 500 Hazy 6 600 Hazy 37 350 Clear 23 500 Clear 17 600 Clear


more drastically than the nominator, resulting in psk/UV,a) becoming high er than psk/B,a). Figure 10.7B shows the wavelength Amax where psk/A,a) is maximal as a function of the control parameter a. Amax is optimal for orientation by means of skylight polarization.

The measurements ofBrines and Gould (1982) confirm that the above theoretical prediction is correct. They measured Psky against several isolated cumulus clouds at 350, 500, 600 nm and obtained that Psky was the highest in the UV (Table 10.2).

10.3 Resolution of the UV-Sky-Pol Paradox

The essence of the resolution of the UV-sky-pol paradox proposed by Brines and Gould (1982), Seliger et al. (1984), Pomozi et al. (2001b) as well as Barta and Horvath (2004) is the following:

1. There is no favoured wavelength for perception of skylight polarization under clear ski es, because the proportion of the celestial polarization pattern useful for orientation is large enough at all wavelengths in the UV and visible parts of the spectrum.

2. Under partly cloudy skies, the E-vector patterns characteristic for clear skies approximately continue beneath the clouds, especially for blue and UV wavelengths.

3. If the clouds are near enough to the ground-based observer and the air columns under clouds are partly sunlit, the degree of linear polarization of skylight originating from the cloudy regions is the highest in the UV, because the nearly unpolarized UV-deficient cloudlight dilutes the polar-


ized light scattered in the air beneath the clouds least. Thus, detection of skylight polarization in the UV maximizes the extent of the celestial polarization pattern useful for a polarization compass under cloudy skies.

10.4 E-Vector Detection in the UV also Maximizes the Proportion of the Celestial Polarization Pattern Useful for Orientation under Canopies

Let us consider the influence of the weighting of unpolarized green light transmitted through vegetation and linearly polarized light scattered in the air beneath the foliage on the degree of linear polarization pJA,} of downwelling light under a canopy, if the air beneath the foliage is illuminated partly by direct sunlight, as is usual in forests, for example. This is important for those insects with polarization-sensitive DRA that live under canopies and orient by means of the E-vector pattern of downwelling light. Under canopies, the same calculation can be performed as under clouds, but in the former case the intensity Ic/(A} of white cloudlight should be replaced by the intensity Ica(A} of green light transmitted by the canopy, called "canopylight" further on:

(10.5)

Figure 10.2 shows the intensity Ica(A} of canopylight transmitted through the leaves of cottonwood (Populus deltoides). Similarly to the cloudlight, the canopylight is most deficient in the UV and is practically unpolarized due to the diffuse scattering in the leaf tissue (e.g. Gates 1980). Consequently, the same phenomenon occurs as under clouds, as shown in Fig. 10.8: the degree of linear polarization Pca(A,a} of light from the canopy (composed of the partially linearly polarized bluish light scattered in the air layer beneath the canopy and the UV-deficient unpolarized greenish canopylight) is maximal in the UV if a < 0.1. Hence, detection of polarization of downwelling light in the UV also maximizes the extent of the celestial polarization pattern useful for a polarization compass under canopies.

In this chapter, we showed how the weighting (described by the control parameter a) of unpolarized white cloudlight or unpolarized green canopylight and linearly polarized light scattered in the air beneath clouds or canopies affects the degree of linear polarization p(A,a) of downwelling light under clouds or canopies, respectively. The only important difference between the effects of clouds and canopies is that clouds can also be at huge (practically infinite) distances from the ground-based observer (meaning great a-values), while the distance of canopies from the ground can range between 0 m and only ab out some 10 m (meaning small a-values). Thus,


clear sky

80

350 400 450 500 550 600

wavelength A (nm)

480~-------------------------------------,

B 460

440

420 11 400 l~- ------------------------------------------------------------380

360

340~--_.--_,--_.----.---r_--,_--_r--_.--_.--~

o 2 3 4 5 6 7 8 9 10 control parameter a

Fig.1O.8. As Fig. 10.7 for the downwelling light under a canopy calculated on the basis of the expression of Pea (l,a) given in Eqn. (10.5) using the functions Pea (A) in Fig.10.1A, as weil as 1,c(A) and 1ea(A) in Fig. 10.2. Increasing a means increasing the proportion of the polarized light scattered underneath the green foliage.


under canopies,Pca(A,a) is maximal always in the UV. However, the question is whether the maximum of pca(A,a) is high er than the threshold p* for polarization sensitivity (about 5 % for crickets and 10% for honeybees). In other words, the question is whether the polarized light scattered in the thin air layer beneath the canopy can be intense enough (relative to the unpolarized canopylight) to ensure that Pca(A,a) > p*. The experimental spectro-polarimetric study of this question could be an interesting task in future work.

10.5 Analogy Between Perception of Skylight Polarization and Polarotactic Water Detection Considering the Optimal Spectral Range

The spectral aspects of the detection of polarization of light reflected from water surfaces are discussed in Chaps. 18,20. Here, we mention only that the majority of the known polarotactic water-seeking insect species exploit UV wavelengths to seek for water (see Table 19.2), because the amount of light originating from the underwater region is minimal in the UV, thus P of light reflected from the water surface is maximal in the UV. However, some polarotactic water insect species also detect water in the visible part of the spectrum (Table 19.2). Some reasons for this are discussed in Chap. 18.

Note that considering the optimal wavelength range, there is an analogy between perception of skylight polarization for orientation and detection of the polarization of light reflected from water surfaces to find water bodies. Both tasks are most efficient in the UV, the reason for which is the same: P of both skylight and water-reflected light is highest in the UV if there is a background - a cloud or canopy in the sky and the bottom or particles suspended in water -, wh ich reflects nearly unpolarized light. The amount of light originating from this background is minimal in the UV, thus the net P of the biologically relevant light (downwelling skylight and water-reflected light) is highest in the UV.

10.6 Analogy of the UV-Sky-Pol Paradox in the Polarization Sensitivity of Aquatic Animals

Interestingly, UV sensitivity is frequendy coupled with sensitivity to linear polarization also in aquatic animals. Several fish species (e.g. Hawryshyn 1992) as well as mantis shrimps (Marshall et al. 1991a,b) use their UV photoreceptors to perceive underwater polarization. However, the role of UV polarization sensitivity in the underwater world by these animals is as yet unknown.


What is the unique property of the UV part of the spectrum that has such importance to underwater polarization sensitivity? The common answer to this question is: "The underwater UV light field undergoes fewer changes during the day, and was more stable on an evolutionary scale than other regions of the visible range. This stability is important when polarization sensitivity is used for navigation" (Shashar 1995, p. 203). This recalls the similarly frequently cited opinion that the skylight polarization should be more stable against atmospheric dis turban ces mentioned in Chap. 10.2.1. Unless these "changes", "stabilities" and "disturbances" are not exactly defined and their existence and importance experimentally are not proven, one can do nothing with such hypotheses.

Note, however, that crustaceans generally perceive polarization between 440 and 580 nm (e.g. Goldsmith 1972; Schwind 1999), and the polarizationsensitive photoreceptors of cephalopods are maximally sensitive ne ar 500 nm (see Chaps. 26 and 27). According to Cronin and Shashar (2001), this may be explained by the fact that P of underwater light increases with increasing wavelength, at least above 450 nm (Ivanoff and Waterman 1958b). The question is whether this trend also continues below 450 nm.

10.7 Why Do Crickets Perceive Skylight Polarization in the BIne?

We can see in Fig.l0.7 A that psk/A,a) is always relatively high in the violet and blue (400 nm < Ä < 470 nm) for a given a-value. Thus, under partly doudy conditions the violet-blue wavelength region is the second optimal spectral range to detect skylight polarization for orientation. Crickets perceive the celestial polarization in the blue; the reason for this is still unknown. Using the blue part of the spectrum may have the following advantage against the UV range under dear skies, when the degree of skylight polarization is high enough for all wavelengths: The intensity of the UV component of sunlight and light from the dear sky is low relative to that of the blue and green components (Fig. 10.2). At twilight under dear ski es, the absolute light intensity is more likely to fall below the sensitivity threshold of a polarization-sensitive visual system operating in the UV than in the blue.

In the context of the detection of skylight polarization, the finding that the photoreceptors in the DRA of the twilight-active field cricket Gryllus campestris operate in the blue rather than in the UV, has been interpreted in this way by Labhart et al. (1984) as well as Herzmann and Labhart (1989). Crickets (Acheta domestica, Gryllus bimaculatus and Gryllus campestris) are active not only during the day, but also during crepuscular periods (dusk and dawn) as well as at night and all have highly polarization-sensitive blue receptors in their DRA specialized to perceive skylight polarization for orientation.

10 Wh ich Part of the Spectrum is Optimal for Perception of Skylight Polarization? 73

According to Zufall et al. (1989), the combination of blue sensitivity and polarization sensitivity in the DRA may be a common adaptation of insects that are active at very low light intensities, as opposed to day-active insects (e.g. honeybees, desert ants and flies) which predominantly use UV receptors as detectors for skylight polarization (Table 10.1).

However, the question is whether this "intensity argument" holds also for cloudy conditions: On the one hand, since under cloudy skies the UV component of skylight is much weaker than under clear skies (Fig. 10.2), detection of skylight may be more disadvantageous in the UV than in the blue. On the other hand, under cloudy skies the degree of linear polarization Psky of skylight is the highest in the UV (Fig. 10.7), thus perception of skylight polarization could be more advantageous in the UV than in the blue. The question is, which effect is the stronger.

10.8 Concluding Remark

The question why insects differ in their spectral channel used for polarization detection cannot be answered at the present time, because too little data are available. One would have to correlate the spectral channels of a large number of insect species with their biology and ecology (e.g. under what sky conditions they are normally active) to obtain an answer. Theory alone will not clarify the situation. Clearly, honeybees, for instance, have an advantage in that they can exploit the weak UV (but stronger than blue) polarization under clouds, whereas under clear skies the polarization is normally strong enough at all wavelengths. But why do other insects not take advantage of this? The explanation of this remains an interesting future task.

11 Polarization of the Sky and the Solar Corona During Total Solar Edipses

During a total solar eclipse, the sun is completely covered by the moon for some minutes, and this immediately transforms the aspect of the sky completely. The sky is not lit up by the radiance of the solar corona alone; the main source of skylight is light coming from outside the area where the totality is taking place and where the sun is still shining (Können 1985). During a total eclipse, a particular type of twilight occurs: most light is seen near the horizon where parts of the atmosphere are stilllit by the partially eclipsed sun outside the zone of totality, and the sky is darkest in the zenith.

Since the beginning of the 1960s, several atmospheric optical phenomena associated with total solar eclipses have been the subject of extensive studies. As the radi an ce and colour distribution of light of the sky is immediately transformed at totality, so also is the polarization of skylight. Apart from the very scant light of the solar corona, the skylight is then produced almost entirely by secondary and high er order scattering (Fig. 11.3A), thus the degree of linear polarization p of skylight is very low.

In spite of the scientific popularity of total solar eclipses, appearing almost every year somewhere on the earth, the empirical knowledge accumulated about the polarization pattern and neutral points of eclipse skies is rather limited, since the earlier polarization measurements were restricted to single points in the sky (Piltschikoff 1906; de Bary et al. 1961; Moore and Rao 1966; Dandekar and Turtle 1971; Rao et al. 1972; Miller and Fastie 1972; Coulson 1988) or at most to the solar and antisolar meridian (Shaw 1975a).1t has been known since the observation by Piltschikoff (1906) that at the beginning of the totality of a solar eclipse, the polarization of the sky decreases drastically at 90° from the sun. De Bary et al. (1961) measured the temporal change of p of skylight at 90° from the obscured sun during the total solar eclipse of 15 February 1961 in Viareggio (Italy). Dandekar and Turtle (1971) performed skylight polarization measurements in the blue and red spectral ranges at a point 90° from the sun during the total eclipse of 7 March 1970 in Kinston (USA). There were great technical bravura when Shaw (l975a) was able to scan the sky with a rotating -analyzer point -source polarimeter along the solar and antisolar meridian during the total eclipse on 30 June 1973 in Northern Kenya. He observed the approximate cylindrical symmetry of the distribution

11 Polarization of the Sky and the Solar Corona During Total Solar Eclipses 75

of P of the eclipse sky and near the zenith a local minimum of p. Using a numerical model, Können (1987) explained quantitatively several polarizational characteristics of the eclipse sky.

The forerunner of imaging polarimetrie studies of the eclipse sky was Gerharz (1976), who took photographs about the celestial circumsolar area of 12° x 15° through a modified Savart filter and a green interference filter during the total solar eclipse of 7 March 1970 near Williamston (USA). From the photographed interference bands, he deduced the degree and angle of polarization of light scattered from the circumsolar region of the eclipse sky and demonstrated a slight polarization asymmetry around the eclipsed sun.

Although the main characteristies of the normal polarization of the firmament are well-known, the same cannot be said ab out the fine structure of the celestial polarization pattern and its temporal change during total solar eclipses. This gap was partially filled by the pioneering full-sky imaging polarimetrie measurements of Pomozi et al. (2001a) during the total solar eclipse on 11 August 1999.

11.1 Structure of the Celestial Polarization Pattern and its Temporal Change During the Eclipse of 11 August 1999

During the solar eclipse on 11 August 1999, detectable differences in the celestial p- and a-patterns occurred only between 12:50:00 (preeclipse, 98 % obscuration of the solar disk) to 13:01:00 (posteclipse, 89 % obscuration) in comparison with the normal sky. From Fig. 11.1 it is evident that the celestial polarization pattern suffered a sudden and dramatie change at the moment of the beginning and the end of totality. Immediately prior to and after totality, the qualitative characteristics of the polarization pattern of the sky were very similar to those of the normal sky. During totality, however, the distribution of p of skylight became roughly cylindrieally symmetrie with respect to the zenith (Fig. 11.1B3-5). P gradually increased from the horizon, then reaching a maximum, it gradually decreased towards the zenith where it was approximately zero. During totality, the distribution of a of skylight remained asymmetric with respect to the zenith (Fig. 11.1C3-5). However, comparing the preeclipse (Fig. 11.1 C1,2) and posteclipse (Fig. 11.1 C6,7) a-patterns, during totality the region of negative polarization (where -45° ~ a ~ +45° relative to the local meridian) considerably extended at the cost of the area of positive polarization (where 45° < a ~ 135°).

The change in p was no greater than ±24 % within a few minutes immediately prior to and after totality (Fig. 11.1D1, 11.1D6). The same was true for the period of totality (Fig. 11.1D3,4). After the second (Fig. 11.1D2) and third (Fig. 11.1 D 5) contacts, however, in a considerable area of the sky (for angular distances from the sun greater than about 55°) lL1pl > 24 % differences

76

A

... ., 0 .

C~ g>. '"


neutral points 12,SI,oo. IBO:OO 12,SI,)4 - I"S I,OO 12,S2,oo· 12,SI:34 I"S"30 · 1"S2:00 12:S9:00· ''''2:30

differences between the subsequent celestial polarization patterns during the eclipse

*-umbra S

Fig. 11.1. Temporal change of the celestial pattern of radiance I (A), degree of linear polarization p (B) and angle of polarization a with respect to the local meridian (C) measured at 450 nm in Kecel (Hungary) during the total solar eclipse on 11 August 1999. Values of time and pereent geometrie obseuration of the solar disk are given above every column. D, E Differences t1p and t1a between the subsequent polarization patterns calculated for the entire sky apart from the overexposed areas and the landmarks/vegetation near the horizon. The photographs of the sky in row A do not represent correctly the real radiance of skylight, because they were taken with different times of exposure and apertures. (After Pomozi et al. 2001a).

occurred. The a-pattern suddenly changed at the moment of the second (Fig. IUE2) and third (Fig. lUES) contacts, otherwise its change was rather modest (Fig. 11.1EI,3,4,6). For zenith angles greater than about 20°, the values of lL1al were smaller than 38°. Changes in a greater than ±38° occurred only around the zenith at the second (Fig. 11.1E2) and third (Fig. lUES) contacts.

11 Polarization of the Sky and the Solar Corona During Total Solar Eclipses

"~ ~" ] ;~ Ci.

... " .s. ~~ ö ,;j -

" -

A degree of polarization p B conl<ol (Tuni.ia) cc lip'c (Hungary)

O%~~~~~~-- ~~~~~~ __ 100%

Qo-~II! .. ", 4 .,.,,:!e.i: ~ ~ •• COCA uo. ~~ ,tIIC)(~

100%

·90"

"911'

horizon

O' D' meridian

zcn.ith

N

E~ ?r~ umbrn S

+90'

+911'

) honzon

« Iip'" (Hungary)

180"

c angle ofpolarization 0: D cclips< (Hungary) control (Tuni, ia)

• ,0 ~~~::::~;"

! .lI ol.4A4IA44

•• ~O';C~ OIll .... .. .. .. .. " · :~~lI:X)(x ... J(J' oDoo::~~9P~

D' ~~~~-----~~--~--~--180' ..

" i~ ~~~~. !~ A

6 4 A.Ilo AD. 4

~:~~~~~Dg6 ." gO )(x"',o''''''

fjjI~.~~J';~ .. t · .:;.4I Q

Jo: O(] .. 4 64

K aa~

K~~ lOl(:::~)t; 4 oc OlJO;!l9 1ii1

0" L.A. ................ .!.... __ ~ L __ --''-·-· .......:'c..' -' --180'

DgDtlO~Ii!'Iö! .. 0" L...A .................. ~ __ -" ~ __ --''-._' -''c..' _' __

180'

• • ° cQQ. OQgClClOOtl(

i"''l()(x!ll:l'~ x;

... ol .a.lIo,o,A64.Aa

44 4 4 . ~" ~~~6~::: 6 0 • .. ::::;~~; ; 0" '-___ . ....... 0"---' _' _·'_ '--'-_.....:' c.c ....... c:.' _' __

180'

d ~ " e •

H "8.

. 04 lIo 0 K 7 !l ~~e~:~~)(::

.' 440 4 Z1.4oa 6

F

conlrol (Tuni,ia)

77

Fig.l1.2. Spatial change of p and a (measured from the local meridian) of skylight as a function of time (1-7) along four differently oriented meridians (coded with x, square, + and triangle in E and F) of the Hungarian eclipse skies (B, C) and the Tunisian control skies (A, D) measured at 450 nm. The shape of the data points in the diagrams coincides with the shape of the symbols coding the different meridians indicated in E and F. Every data point represents a value averaged on p- or a-values measured in 33 neighbouring celestial points along a given meridian. The position of the neutral point near the zenith occurring during totality is marked by a vertical dashed line in diagrams B3-B5. (After Pomozi et al. 2001a).


During the preeclipse and posteclipse periods the celestial distribution of p was not rotationally symmetric (Fig. 11.2Bl,2,6,7): in the antisolar half of the celestial hemisphere higher p-values occurred than in the solar half. At the second contact the p-pattern became approximately rotationally symmetrie, a feature whieh remained throughout the totality (Fig. 11.2B3-5). The celestial distribution of p was, however, not exactly cylindrically symmetrie to the zenith during totality: in the anti solar half of the eclipse sky lower p-values occurred than in the solar half. The change in a along the different meridians of the sky was rather complex (Fig. 11.2C), but the change in a along all meridians during totality (Fig. 11.2C3-5) was substantially different from that during the preeclipse (Fig. 11.2Cl,2) and posteclipse (Fig. 11.2C6,7) periods. As the umbra moved across the observation point, the celestial polarization varied somewhat during totality due to the chan ging geometry of atmospherie light scattering (Figs. 11.1, 11.2).

Previous studies (e.g. Sharp et al. 1971; Coulson 1988) have indieated that up to approximately 98 % geometrie obscuration of the solar disk eclipse phenomenology can be interpreted in terms of attenuated, but otherwise essentially unchanged, sunlight. For very high obscuration ratios, greater than about 98 %, multiple scattering predominates, and the distribution of colour, intensity and polarization over the sky hemisphere changes rapidly and dramatieally.

11.2 Origin of the E-vector Pattern During Totality

Figure 11.3A represents schematieally how the originally unpolarized sunlight illuminating the atmosphere reached an observer after primary and higher order scattering events during totality. Light from a first order scattering event (A) could reach the observer only at very small view angles with respect to the horizon. At greater viewing angles, light could reach the observer only due to second (B) or higher order scattering events. The observed celestial polarization was the result of these high er order scattering events.

As a first approximation, a of skylight during totality can be qualitatively explained solelyon the basis of first (A, Bj ) and second (B2) order scattering events. First order scattering would result in the well-known Rayleigh pattern of skylight polarization. In Fig. 11.3B 1 and 11.3B3 the single-scattering Rayleigh pattern calculated for a solar zenith angle of 32° (corresponding with the solar zenith angle during totality on 11 August 1999 in Hungary) can be seen. According to the model, during totality the atmosphere was illuminated only by the single-scattering Rayleigh skylight from outside the umbral region. The atmospherie scattered centres (B2) in the umbra scattered the rays of this Rayleigh skylight towards the observer (Fig. 11.3A). If the observer


A

+ d light with

ground-r~flec~~ nositive negatiVe y .

polarizatlOn

B

g ground

reflection

regions receiving negatively polarized ground-retlected light

antisolar ~ solar direction direction

2 antisolar meridian ,

_ P!'1i~vc:po~af~7..a !~Qn_ ------->-' ---- --, -- . . ----- r -- - - -~~~-- ( --

.... obSc!'Ver" • zCßllh '" ... .... .... / ;' " I r I ~ " " ....

I I I I "~h~J \ \ \ I f I , ; 'Sub \ \ \ '

'I'cl,a.!v{k.larila,jlm \ \ f I ri< - 1- ~ - \. - -

' I I , I I

solar I meridian

measured ecl ipse sky

3

79

Fig. 11.3. A Sehematie representation of the geometry of primary (A, B 1) and seeondary (B2) seattering as weH as ground refleetion of sunlight in the atmosphere during a total solar eclipse. For the qualitative explanation of the origin of the loeal minimum of p and the neutral points observed approximately along the antisolar meridian ne ar the horizon during totality if primary seattering events of negatively (n _' m J or positively (n+, m+) polarized ground-refleeted light are taken into aeeount. B For the qualitative explanation of the origin of the regions of positive (BI, B2) and negative (B2, B3) polarization in the sky observed during totality. The single-seattering Rayleigh pattern was ealculated for the position of the sun during totality (solar zenith angle = 32°); the alignment of the bars represents the loeal direetion of polarization and their length is proportional to p. (After Pomozi et al. 2001a).

looked towards the antisolar half of the umbra (Fig. 11.3B 1), this region of the atmosphere was illuminated mainly by highly polarized scattered Rayleigh skylight (B j ), the E-vectors of which were approximately perpendicular to the scattering plane (the local meridian). This more or less perpendicularly polarized skylight was scattered (B2 ) towards the observer. This is the reason for the fact that during totality mainly positive (E-vectors more or less normal

80 Part I1: Polarization Patterns in Nature

to the scattering plane) skylight polarization was observed in the antisolar half ofthe sky (top half ofFig.l1.3B2).

If the observer looked towards the solar half of the umbra (Fig. I1.3B3), this region of the atmosphere was illuminated mainly by highly polarized scattered Rayleigh skylight (B j ), the E-vectors of which were approximately parallel to the local meridian. This more or less parallel polarized skylight was scattered (B2) towards the observer. This is the reason for the fact that during totality mainly negative (E-vectors more or less parallel to the scattering plane) skylight polarization was observed in the solar half of the sky (bottom half of Fig. 11.3B2). In Fig. 11.3B2 we can see that a suffers a sudden change when the zenith is crossed, and ILial is about 90° if we cross the zenith parallel to the solar/antisolar meridian.

11.3 Neutral Points of Skylight Polarization Observed During Totality

Horvath et al. (2003) reported on the neutral points of the sky observed during the totality of the solar eclipse on 11 August 1999 in Hungary (Figs. 11.4-11.7). Earlier, only de Bary et al. (1961) observed zero degree of linear polarization p of skylight during totality at 90° from the obscured sun on the antisolar meridian.

On the p-patterns measured at 12:51:34 (Figs. 11.4A, 11.6B) and 12:52:00 (Figs. II.4B, 11.6D) at 450 nm on 11 August 1999 a point is discernible near the zenith where p = 0 %. This point is called "zenith neutral point type-2". It is classified as "type-2", because it can be considered a point where p passes through aminimum, rather than a real neutral point such as the Arago, Babinet, Brewster and fourth (Horveith et al. 2002b) neutral points of the normal sky classified as "type-I". In the type-2 zenith neutral point of the eclipse sky the absence of polarization is analogous to the absence of polarization of the sunlit sky straight in the direction of the sun.At 12:52:00 approximatelyat the position of the zenith neutral point, a local minimum of p occurred at 550 nm (Figs.ll.4C, 11.6F). The local minimum of p in the immediate vicinity of the zenith can also be seen in the graphs of Fig. 11.6A, 11.6C and 11.6E. At 12:51:34 at 450 nm (Figs. 11.4A and 11.6A), two neutral points of type-l occurred approximately along the antisolar meridian near the horizon. They arise because the Stokes parameter Q (if U = 0) for single and multiple (double) scattering cancels out; this also occurs in the usual Arago, Babinet and Brewster neutral points. At 12:52:00, a neutral point of type-3 was observed at 450 nm (Figs. II.4B, 11.6C) and a local minimum of p occurred at 550 nm (Figs. 11.4C, 11.6E) approximately at the position of the two type-l neutral points.

It is interesting that there was no switch of a crossing the type-3 neutral point along a meridian. This unique celestial point is characterized by the


p

100

12:51 :34, 450 nm (blue)

two neutral points NI of type I near the horizon

~ overcxposurc

12:52:00. 450 nm (blue)

neutral point N3 of type 3 near the horizon

E~ ~ .. umbra S

12:52:00, 550 nm (green)

loeal minimum MH of p near the horizon

sun

81

Fig. 11.4. Celestial patterns of the degree of linear polarization p of skylight measured with full-sky imaging polarimetry during totality of the solar eclipse on 11 August 1999 in Hungary at different times and wavelengths. A 12:51:34 (local summer time = UTC+2), 450 nm; B 12:52:00,450 nm; C 12:52:00,550 nm. The values of p are rounded to integers (0,1,2,3, ... ,100 %). The neutral points are marked by dots. (After Horveith et al. 2003).

o·

+90·

· 135· 180. + 135·

angle of polarization (l measurcd from the

loeal meridian

~ overcxposurc

12:5 1 :34, 450 nm (blue) 12:52:00, 450 nm (blue)

E~ ~ .. umbra S

12:52:00, 550 nm (green)

loeal mi nimum MH of p near the horizon

Fig. 11.5. As Fig. 11.4 for the angle of polarization a of skylight measured from the local meridian. (After Horveith et al. 2003).

82

12:51:34, 450 nm (blue)

50%

"" o

il .~

! o

~ oll

lWO ne\l1f1i1 point' Nlorlypc:-I

zc:nith l1~rthchorizoll


12:52:00, 450 nm (blue)

z.cnith

nt.v, ... 1 pomt J orlypc-}

l1~rtheOOrizon

12:52:00, 550 nm (green)

kJeaJminimum MHo(pn~r

~nith tM horizon

'~Imjmmwn F MZo(pnQr

Ibezcmlh

-

Fig.l1.6. Graphs of the degree of linear polarization p measured along different meridians crossing the zenith Z of the eclipse sky at different times and wavelengths. A, B 12:51:34 (UTC+2), 450 nm; C, D 12:52:00,450 nm; E, F 12:52:00,550 nm. The continuous lines represent the curves fitted by the method of least squares to the measured values of p, while the dashed lines show the upper and lower limits, between which 90 % of the pvalues falls. The circular insets show how the scans are loeated relative to the cireular patterns in Fig. 11.4. A Scan through the two neutral points NI of type-l near the horizon. B Sean through the neutral point N2 of type-2 near the zenith. C Sean through the neutral point N3 of type-3 near the horizon. D Sean through the neutral point N2 of type-2 ne ar the zenith. E Sean through the loeal minimum MH of p near the horizon. F Scan through the loeal minimum MZ of p near the zenith. (After Horvath et al. 2003).

abolition of P (Figs. 11.4B, 11.6C) in a celestial area where the a-pattern is homogeneous, i.e. positive polarization occurs on both sides of the neutral point (Figs. 11.5B, 11.7C). The latter feature distinguishes this unique type-3 neutral point from the other neutral points of type-l and type-2 of the eclipse sky as well as from the normal Arago, Babinet, Brewster and fourth (Horvath et al. 2002b) neutral points, which are characterized by a sudden change of 90° of a (Figs.11.5A, 11.5B, 11.7A, 11.7B, 11.7D).


11.4 Origin of the Zenith Neutral Point During Totality

The essenee of qualitative explanation of the origin of the neutral point or loeal minimum of p observed near the zenith during totality on 11 August 1999 is the following (Pomozi et al. 2001a; Horvath et al. 2003): sinee the atmospheric seattering eentres at or ne ar the zenith (above the ob server in the umbra) were illuminated by seattered skylight with all possible E-vector direetions eoming from outside the umbra, the atmosphere at or near the zenith seattered E-veetors with all possible alignments towards the observer. This resulted in zero or almost zero net p, Le. unpolarized skylight or skylight with very low p near the zenith. The exaet position of the zenith neutral point or loeal minimum of p near the zenith depended on the wavelength (due to the dispersion of polarization of seattered skylight) as well as on the timedependent geometry of the lunar shadow with respect to the earth's surfaee and the observer's position. Shaw (1975a) observed a similar minimum of p near the zenith of the eclipse sky, which phenomenon was quantitatively explained by Können (1987).

11.5 Origin of Other Neutral Points at Totality

The prerequisite of formation of a neutral point or a loeal reduction of p in the positively polarized antisolar half of the sky during totality is a meehanism that introduees negatively polarized light into the umbral region of the atmosphere. One of sueh meehanisms is the refleetion of light from the ground.1

Figure 11.3A shows sehematieally the situation when negatively or positively polarized ground-refleeted light was introdueed into the atmosphere during the total eclipse. The degree of positive polarization of multiple seattered skylight from the antisolar half of the sky in the umbra was more or less redueed by the negatively and enhaneed by the positively polarized light refleeted by

1 Natural soil surfaces reflect more or less partially linearly polarized light, the p of which depends on the type (roughness, albedo and spectral characteristics) of the surface (Coulson 1974). It is a general rule that the higher the albedo of a rough reflecting surface in a given spectral range, the lower the p of reflected light. This phenomenon is called the Umow effect (Umow 1905). The polarization of the ground-reflected light is negative or positive if the angle of reflection measured from the direction of incidence is smaller or greater than a threshold angle y, respectively (Fig. I1.3A). Angle y is dependent on the characteristics of the reflecting surface, but its typical value is about 20° for bright sandy and grassy terrains, as was the terrain at the place of the polarimetrie measurements of Horvath et al. (2003) in the surroundings of Kecel (Hungary). The degree of negative polarization of light reflected from such a surface changes from zero to several percent if the angle of reflection with respect to the direction of incidence decreases from yto zero.


the ground from outside the umbral region if it was scattered towards the observer by umbral atmospheric scattering centres (n _ and n+ in Fig. 11.3A). At a given direction of view depending on the relative intensity of the positively and negatively polarized skylight (the Stokes parameter Q if U = 0), the following situations can be imagined:

1. If at wavelength A the negatively polarized skylight intensity IJA} is lower than the positively polarized skylight intensity I+(A} in all directions of view, a local minimum of the celestial degree of positive polarization can be observed at the zenith angle of the maximum reduction of p (Figs. 11.4C, 11.6E).

2. If IJA} = I+(A} at a certain zenith distance, a neutral point occurs in this direction (Figs.l1.4B, 11.6C).

3. If IJA} > I+(A}, a negatively polarized "island" is seen in the region of positive polarization (Figs. 11.4A, 11.5A, 11.6A, 11.7 A). Then two neutral points appear at the border of this celestial island of negative polarization. In these neutral points, the positive polarization switches to negative polarization as in the case of the neutral point observed near the zenith during totality, or of the normal Arago, Babinet, Brewster and fourth (Horvath et al. 2002b) neutral points.

Due to the moving lunar shadow, both the p- and a-patterns changed during totality for any location of the observer while the eclipse proceeded from second to third contact. These changes depended on the wavelength and were determined by the observer's view through a varying slant range of air in the umbra, before the directly scattered sunlight was encountered. The site and points of time of the measurements of Horvath et al. (2003) in relation to the moving umbra as weH as the wavelengths (450,550 nm) of their observations were so fortunate during the totality of the eclipse on 11 August 1999 that they could observe aH the above-mentioned three different situations (Figs. 11.4-11.7). In these cases, the maximum reduction of p happened approximately along the antisolar meridian, because the thickness of the umbral region of the atmosphere receiving negatively polarized ground-reflected light was the greatest in this direction at the time of recording.

During the total eclipse on 11 August 1999 in Kecel, the degree of negative polarization of multiple scattered skylight from the solar half of the sky in the umbra was more or less enhanced by the negatively and reduced by the positively polarized light reflected by the ground from outside the umbral region and scattered towards the observer by umbral atmospheric scattering centres (m _ and m+ in Fig. 11.3A). During totality the negatively polarized light dominated in the solar half of the firmament, thus here negatively polarized skylight with slightly greater p than in the antisolar half was observed.

The main cause of the slight drift of the neutral points from the solar-antisolar meridian may be the changing geometry of the umbra with respect to


12:51 :34, 450 nm (blue) 12:52:00, 450 nm (blue)

'" 180 .§ ~ 'e ~ 90" -o u

~ 0'

'" 180' ~ o

.~ ~90' o u

'" ~ 0"

!>on ... mmdJIII1 ....

two naJlnl points l oftypc'-I

zenith neu thI= OOrizon

- - ::--~_A_

- - \

nc'I.I1ra1POlnt 2: o"ype-2

neulhc lC11itb

_B_ j ...

I I 11 --_ ...

-

-

;\4l:J.rtul' or ky

umlli nnllwl;!r tuJlr n;rit(lil h(M'ib)t'l

0" ofsky +90" -90'"

N3

c

-- -

- - - , D

IK\Jlr<11l poinl N} ofl)'pc-J

IKlIIrtllc:horizoo

- -~- -I

nWIJ1II point Nlor.ype-2

ntar\he z(nitß

I

\ - - -- -

12:52:00, 550 nm (green)

~~Iminimum MliofpnRr

""ith Ihchorizon

E

- - - -

F

Ir-I I

local minimum \\ IZo(p nc:u

lbe1.enith

Fig. 11.7. As Fig. 11.6 for the angle of polarization a measured from the loeal meridian. (After Horveith et al. 2003).

the ob server as the eclipse proceeded. A second cause may be distant clouds, which may disturb the distribution of singly scattered light around the ob server. A third factor is the polarization of singly scattered light.

11.6 Imaging Polarimetry of the Solar Corona

The light of the corona is scattered sunlight: Thomson scattering takes place on free electrons surrounding the sun like a cloud. The corona visible in the neighbourhood of the sun from the earth's surface during total solar eclipses, or through coronographs ofhigh altitude astronomical observatories or satellites has been the subject of many investigations (e.g. Newall1906; Sivaraman et al. 1984). The scattered corona light is partially linearly polarized with Evectors approximately perpendicular to the radii from the solar centre. The

86

B

\ c

-90" +90' • overe.posure

-135" 180. +135"


Q, 40%

= 0 30% "..,

" ~ .... os Ö 20% '-0 ., 1! 10% bIl ., ."

0% 1.0

dcgrcc of polarizarion p

0"/0 100% e overexposure

0; 120' Ö "€ § .,

"l:l » 1:l ., 90' "5 oS - s o 0 o.~ '-." 60' ~ ~ 01)",

red" • • " grec:n __ bl",, __

D

relative radius p

E

a :l s 30' -L---',-:-~~-'-:---'-:---L-:---'-:----I._-' 0' 45' 90' 135" 180" 225' 270" 315" 360'

angle ß measured from the vcrtical

Fig. 11.8. A Photo graph of the solar corona taken during the total solar edipse on 11 August 1999 in Kecel (46°32'N, 19°16'E, Hungary). B, C Patterns of the degree of linear polarization p and the angle of polarization a rneasured at 550 nrn with rotating-analyzer photo polarirnetry. ais rneasured frorn the vertical. The black bars in the a-pattern show the local directions of the E-vectors. The obscured solar disc is replaced by a white disc, in which the North and South poles of the sun are rnarked. D Radial change of p of the corona light along the radius (thick line) shown in the inset rneasured at 650, 550 and 450 nrn. p = rlrSU" is the relative radius, where r is the radius frorn the cent re of the sun and rsun is the radius of the solar disCo E Tangential change of a of the corona light along the circle shown in the inset (thick line) rneasured at 650, 550 and 450 nrn. (After Horv<ith et al. 2001).


maximum degree of polarization p is about 40 % at a distance of a quarter of the solar diameter from the edge of the sun (Können 1985). Farther away p gradually decreases. Taking the corona as a whole, the directions of polarization more or less neutralize each other with the result that the total radiance is very weakly polarized, if at all. On the basis of the polarization pattern of the solar corona, the astronomers and solar physicists can calculate the electron density around the sun, which is an important parameter in solar physics. The net p of the corona, if not zero, is also an important quantity characterizing the solar atmosphere.

Using rotating-analyzer video and photo polarimetry, Horvath et al. (2001)

studied the polarizational characteristics of the corona during the total eclipse on 11 August 1999. Figure 11.8 shows that the polarization pattern of the corona was approximately rotationally symmetrical. The maximum p of corona light was ab out 30-35 % at a relative radial distance of ab out p = 1.70-1.75. The angle of polarization a of corona light was practically independent of the wavelength in the visible part of the spectrum, and the E-vectors of the corona were approximately perpendicular to the radial direction.

12 Reflection-Polarization Pattern of the Flat Water Surface Measured by 1800 Field-of-View Imaging Polarimetry

Using 180° field-of-view imaging polarimetry, Gal et al. (2001b) recorded the reflection-polarization pattern of the flat water surface under a dear sky at sunset (Figs. 12.1, 12.2). Due to this technique, they could experimentally test and prove the validity of the theoretical predictions of Schwind and Horvath (1993) and Horvath (1995). Comparing the theoretical (Fig. 12.2B-D), semiempirical (Fig. 12.2E-G) and measured (Fig.12.2H-J) reflection-polarization patterns of the flat water surface, a remarkable resemblance can be established between them. The reason for this dose similarity is that the strong repolarization ability of the water surface overwhelms the slight differences between the polarization of the single-scattered Rayleigh skylight (Figs. 12.1A,B) and the real skylight (Figs. 12.1C,D). There are small, irrelevant differences between the measured (Figs.12.2H-J) and predicted (Figs.12.2E-G) reflection-polarization patterns, because the water surface was slightly undulating, and some light was scattered inside the water, then returned (this reemitted radiation from the water was not taken into consideration in the prediction). By analysing the fine details of the reflection-polarization patterns in Fig. 12.2, we can establish the following:

• The maximum (approximately 100%) degree of linear polarization p of reflected skylight is located in a characteristic annular band, called the Brewster zone, from which the light is reflected with an angle of 53° relative to the vertical, called the Brewster angle (Fig. 12.2B,E,H). When the sun is on the horizon, the Brewster zone with a strong horizontal polarization is maximally extended towards and away from the sun and becomes narrowest perpendicular to this direction.

• At sunset or sunrise, the light reflected from the flat water surface is mainly horizontally polarized. The angle of polarization is 45° $; a $; 135° with respect to the vertical, both in the direction of the sun and opposite to it. Apart from the horizontally polarized Brewster zone it is mainly vertically polarized, i.e. 0° $; a< 45° and l35° $; a $; 180° at right angles to the mirror solar meridian (Fig. 12.2C,F,I) just like the blue sky itself (Fig. 12.1B,D). At twilight the mainly vertically polarized region of the water surface is 8-

12 Refleetion -Polarization Pattern of the Flat Water Surfaee

degree of polarization p

0% =::::1 • •• 100% p

angle of polarization 0.

_90·

180·

+90·

+135·

89

overexposure

Fig. 12.1. A, B Spatial distribution of the degree p and angle a of linear polarization of skylight ealculated on the basis of the single-seattering Rayleigh theory for sunset. C, D Patterns of p and a of skylight measured by 1800 field-of-view imaging polarimetry at 550 nm at sunset. E Radianee I of the sunset sky during the measurement of patterns C and D. a is measured from the loeal meridian. The position of the setting sun is represented by a dot and the horizon is the perimeter of the circular patterns. (After GM et al. 2001b).

shaped within the Brewster zone and takes an extended bow-shape outside the Brewster zone.

• At twilight under a clear sky, there are several neutral points on the water surface (Fig. 12.2B,E,H). Inside the Brewster zone (Bz) there exist two neutral points (a, b) positioned at about 45° from the nadir at right angles to the mirror solar meridian. There are two additional neutral points (c, d) outside the Brewster zone perpendicularly to the mirror solar meridian, and two further neutral points, the positions of which co in eide with the mirror sun (e) and the mirror antisun (j). These neutral points are the regions of the water surfaee where the horizontal polarization of refleeted skylight switehes to vertical.

• The pattern of refleetivity R of the flat water surfaee has a quasi-eylindrical symmetry for R > 7 %, Le. for direetions of observation larger than 65° from the vertieal. The eontour lines of equal R-values are elongated perpendicularly to the mirror solar meridian. The two eentral patehes in

90

p 0% =:::1 •••• 100%

degrcc of polari7.3tion p


_90·

-135°

overcxposurc

angle of polarization a measurcd from the vcnical

radiancc I

rcncctivilY R

12 Retlection-Polarization Pattern of the Flat Water Surface 91

Fig. 12.2. A Mirror image of the sky retlected from the tlat water surface at 550 nm at sunset. The triangular region on the right hand side is the railing at the end of the jetty from which the recording was taken. B, C, D Theoretical patterns of the degree of linear polarization p, the angle of polarization a and the retlectivity R of the tlat water surface calculated for single-scattered Rayleigh skylight (Fig. 12.1A,B) with the use of the Fresnel formulae. a,b,c,d,e,J neutral points on the water surface. Bz Brewster zone. E, F, G Retlection-polarization patterns of the tlat water surface calculated for the measured skylight pattern (at 550 nm; Fig. 12.1C,D,E) with the use of the Fresnel formulae. H, I, J Retlection-polarization patterns of the flat water surface measured by 1800 field-of-view imaging polarimetry at 550 nm. The co ding of the retlectivity values R is the following: the two centralB-shaped black patches in patterns D, G and J represent R .,; 2 %. The concentric oval and annular, alternately black and white narrow zones around these patches represent R = 3,4, ... ,9,10 % towards the periphery. The outermost annular wide white or chequered zone represents R > 10 %. a of light reflected from the water surface is measured from the vertical. The position of the mirror image of the sun is represented by a dot and the horizon is the perimeter oJ the circular patterns. (After Gal et al. 2001b) . ...-:

Fig. 12.2D,G,J show those regions of the water surface where R < 2 %. These two dark patches can be seen on the water surface at 90° from the sun when it is near the horizon. The surface is clearly more transparent at these patches. The occurrence of these patches is the result of the fact that R of the water surface is lower for vertically polarized incident light than for horizontally polarized light .

• The reflection-polarization patterns visible over the flat water surface under a clear sky at sunset or sunrise have characteristic gradients of R, p and a. These different gradients are associated with the same regions of the water surface: where the gradient of R is high, so too are the gradients of p and a. This can be seen, for example, in the case of the characteristic 8-shaped pattern inside the Brewster zone in Fig. 12.2C,F,I. The tips of this 8-shaped pattern coincide with the two neutral points (a, b) of the p-pattern (Fig. 12.2B,E,H) and with the centre of the two dark patches of the R-pattern (Fig. 12.2D,G,J).

13 Polarization Pattern of a Fata Morgana: Why Aquatic Insects Are not Attracted by Mirages?

It is a weIl-known phenomenon that on hot days mirages, also called Fata Morgana, may appear on roads. Such mirages are also seen on hot plains. There seems to be a pool of shiny water in the distance, which dissolves on approach. The sky, different landmarks and objects are mirrored in this "pool". In addition, the chaotic vibration of amirage due to the turbulent flows of hot air imitates the wind-generated undulation of a water surface. Water insects, however, do not detect water on the basis of its brightness and colour, but by means of the horizontal polarization of reflected light (see Chap. 18). Hence, the quest ion arises, whether mirages can deceive waterseeking polarotactic insects.

To answer this question, Horvath et al. (1997a) investigated and compared the polarizational characteristics of amirage and areal water surface (Fig.13.1). They showed that generallythere is no contrast in the degree p and angle a of linear polarization between the sky and its mirage (Fig. 13.1A-C), while usually there are sharp radi an ce as weIl as p- and a-contrasts between the water surface and the sky (Fig.13.1D-F). Flat water surfaces usually reflect more or less horizontally polarized light, while undulating water surfaces reflect light, the E-vector of which is perpendicular to the plane through the observer, the point of reflection and the sun. If the water surface is far away from the observer, p is relatively low due to the grazing direction of view. 1 On the other hand, p and a of skylight depends on the solar position and the direction of view. Furthermore, skylight reflected from water surfaces becomes repolarized (Fig. 13.2A).

Mirages are not reflections, but are formed by gradual refraction and a total reflection of light (Fig. 13.2B). Fata Morganas are generated above hot plains. Close to the ground, the air is warmer and its index of refraction is smaller.

1 If unpolarized incident light is reflected from a flat water surface under angles of incidence over 89.4°, p of horizontally polarized reflected light is not high er than about 2 % (Guenther 1990). However, if one approaches the water surface,p of reflected light increases abruptly as the angle of observation approaches the Brewster angle. In contrast, amirage can never be reached by an observer, so that the direction of observation remains always the same, i.e. nearly horizontal.

13 Polarization Pattern of a Fata Morgana 93

A

c

D

E

F

I'," , ' " \. sky

nlOuJ1lalll • I ' . . ' ... ly·~ Illlmgc

~all pan

\ " • "'{ " skv

~ - . .... . ; ~ .. ~ '.. : '" -. ~)ou;;t"ir'l

. ~ky\, mlrJ.b~ , '. - . " . .. .... -' -... . ~ ,

, ,

, snlt pan

c:>. '- c 00 u ';:: .., ~ 6"0 . ~

f~ ~-ä

0%

degree of polarizalion p

100%

+90. angle of polarizalion Cl measured from Ihe vertical

c:>.

Fig.13.l. A Picture of amirage occurring above a desert landscape, the salt pan Chott el Djerid in southern Tunisia. The dark grey cone-shaped band in the middle right is a mountain, tapering to the left. Below the mountain, the shiny stripe represents the mirage of the sky which merges in the real sky on the left. The lower half of the picture is occupied by the sandy floor of the salt pan. The vertical angular extension of the landscape shown is about IS. B,C Spatial distribution of p and a of the landscape portrayed in A and measured byvideo polarimetryat 550 nm, D-F Same as A-C,for a seaside landscape near Mahares, Tunisia. The uppermost part of the picture is filled with clear sky, the middle part is occupied by the sea and the lowermost part by the shore. (After Horvath et al, 1997a),

94

A reflection

partially horizontally unpolarized ~ polanzed incident light _ ~ reflected light ~

"_cr~y~ water ______ --

B mirage unpolarized ><j total-reflected light

observer

air

unpolarized incident light

hot plain /yyyyyyyyyyyyy ....


Fig. 13.2. A Unpolarized incident light becomes partially horizontally linearly polarized when reflected from a water surface. B Formation of amirage above a hot plain, where the air temperature decreases exponentially as height above ground increases. As a consequence, the refractive index of air increases abruptly with height above ground, and grazing rays of light refract and after total reflection bend gradually into the eye of the observer. This gradual refraction and the total reflection do not alter the polarizati on of light. The inset figures represent polarization ellipses. The E-vector orientations are shown by double-headed arrows. (After Horvath et al. 1997a).

Thus, the direction of grazing rays of light is gradually changed to such an extent that the rays do not reach the ground, but after total reflection they are deflected upward (Fig. 13.2B). This gradual deflection provides an ob server with the same impression as mirroring does. However, such gradual refractions and total reflection of light do not change the state of polarization (Können 1985).

Horvath et al. (l997a) proposed that mirages can imitate water surfaces only for polarization-blind visual systems. A polarization-sensitive waterseeking insect should be able to detect the polarizational characteristics of a mirage. Since these characteristics differ considerably from those of water surfaces, these animals cannot be deceived by and attracted to mirages.

14 Polarizational Characteristics of the Underwater World

In qualitative submarine visual observations down to 15 m, Waterman (1954) found that aquatic animals of the photic zone are surrounded by complex polarization patterns. He used a diving heImet and a hand-held polarization analyser, in which the light first passed through a plate of a uniaxial crystal cut perpendicularly to its optical axis, then through a quarter-wave plate and a linearly polarizing filter. The intensity of unpolarized light passing through this polariscope simply decreases. If the incident light is partially linearly polarized, a brightly coloured interference pattern of concentric, broken rings appears in the polariscope, from which the presence of polarization, the Evector direction as weIl as a rough estimate of the degree of linear polarizati on p can be deduced. The interruptions of the broken rings occur in orthogonal quadrants, and the axis of one of the pairs of these opposed quadrants is parallel to the E-vector of incident light. Both the intensity of the pattern and the number of concentric rings vary with p. With this polariscope, the E-vector direction could be obtained with an accuracy of about ±3°.

Waterman (1954) found the following characteristics of the submarine polarized light field (Fig. 14.1): Underwater there are two polarization patterns, one inside and one outside the Sn eIl window, which is visible within the critical angle of the refractive cone. l Due to refraction at the water surface, the

1 The boundary of the Snell window extends up to ßsw = arctan[nj(nw2-n/)1I2] = 48.5° measured from the zenith, where na = 1 and nw = 1.333 are the refractive indices of air and water, respectively. Due to refraction, the abovewater world visible through the Snell window is distorted (Horv<ith and Varju 1991). A point of the firmament with a zenith angle ß is apparently seen in direction ß* = arctan[nasinß/(nw2-n/sin2ß)1I2] from the vertical. The apparent horizon corresponds to the boundary of the Snell window. Light from the Snell window in shallow waters contains most of the components of the spectrum available to terrestrial animals. Outside the Snell window, the light from deeper water layers is reflected, it is dirn and its spectral range is restricted especially in open waters. At the boundary of the Snell window, light from ne ar the abovewater horizon is split into a rainbow due to dispersion (Jerlov 1976). An abovewater object direcdy overhead suffers litde refractive distortion when seen from the water, but the image of objects near the horizon is substantially compressed (Horvath and Varju 1991). When the water is flat, the boundary of the Snell window is sharp, and


Fig.14.1. The underwater polarization pattern forms a virtual sphere that surrounds the observer. The distribution of polarization underwater depends on the position of the sun and the relative stillness of the water surface. The highest degree of linear polarization occurs in a band that runs along a great circle of this sphere perpendicularly to the refracted sunlight, and the E-vector is always perpendicular to the scattering plane. (After Hawryshyn 1992).

entire 1800 field of view above the water is eompressed into a eone with 48.50

half-angle. The polarization pattern of the sky is visible within the Snell window. Horvath and Varju (1995) ealculated this pattern in detail. Outside the Snell window is another polarization pattern ereated by the seattering of sunlight entering the water. Both of these underwater polarization patterns are eomplex and eontain information about the loeation of the sun. An aquatie animal able to maintain a stable spatial orientation relative to gravitation and eapable of analyzing E-veetor direetion would have a sun eompass available even if the glitter pattern at the depth of the animal prevented it from seeing the sun direetly.

The underwater polarization pattern forms a virtual sphere that surrounds the observer. The highest p oeeurs in a band along a great circle of this sphere perpendicularly to the refraeted sunlight, and the E-veetor is always perpendicular to the seattering plane (Fig. 15.4). On cloudy days most of the underwater polarization ean be attributed to the seattering of light by water molecules, with little eontribution from the polarized skylight.

Small particles suspended in the water seatter the UV and blue light most strongly, while water molecules have absorption bands in the UV and red

I (Continued) there is a strong contrast between the bright scene above and the darker reflections from deep water. Some plankton-feeding fish living near the surface have an area of enhanced acuity on that part of their retinae where the boundary of the Snell window comes to lie (Munk 1970). One of the two foveae of the compound eye in the water bug Notonecta glauca also looks in the direction of the edge of the Snell window when the animal rests upside down below the water surface (Schwind 1983b, 1985b).


(Jerlov 1976). Dissolved organic materials absorb UV light extensively, thus it is most attenuated. Attenuation of the blue and red wavelengths can be accredited mostly to absorption by various chlorophylls (Wetzel 1975). p is highest near the water surface (Ivanoff and Waterman 1958b) and decreases rapidly within the first 40 m depth. The underwater polarization is influenced above a critical depth by the sky as well as the relation between the observer's line of sight and the direction of the underwater rays. Waterman (1954) hypothesized that under totally overcast skies, the E-vector of the underwater polarized light is overall horizontal and the polarization pattern in deep waters is similar to that near the surface on a heavily overcast day.

After the pioneering observations oflight polarization in the ocean (Waterman 1954), a huge amount of experimental (e.g. Waterman and Wes tell 1956; Ivanoff and Waterman 1958a,b; Timofeeva 1961,1962,1969,1974; Jerlov 1976; Ivanoff 1974; Novales Flamarique and Hawryshyn 1997a) and theoretical (e.g. Kattawar 1994) information has been accumulated about the underwater polarized light field. Submersible point -source scanning polarimeters with different colour filters made it possible to collect data in shallow as well as deep marine waters (e.g. Ivanoff and Waterman 1958b; Tyler 1963; Ivanoff 1974). These results have been reviewed and discussed thoroughly by Jerlov (1976). The most complete description of underwater polarization in the visible part of the spectrum combining laboratory and field experiments was given by Timofeeva (1961, 1962, 1969, 1974). In milky solutions in the laboratory, she studied p and a of underwater scattered light as functions of the direction of observation and the azimuth angle of the light source. Ivanoff and Waterman (1958b) as well as Timofeeva (1961) found that p was highest for both milky solutions and ocean waters with the highest absorption and lowest dispersion, regardless of the azimuth angle of the source. Timofeeva (1969, 1974) also observed submarine neutral points in the vertical plane through the observer and the sun. Waterman (1955), Waterman and Westell (1956), Ivanoff and Waterman (1958b) as well as Timofeeva (1969, 1974) found that p decreases with increasing depth, and at a critical depth it reaches a constant maximum value horizontally, when the radiance distribution no longer varies with direction of observation and the downwelling light becomes more vertical. This critical depth depends on the optical properties of the medium and varies from 40 m (Ivanoff and Waterman 1958b) to 200 m (Waterman 1955) in very clear waters. p near the surface of clear water was found to reach 60 % depending on the solar azimuth, but at greater depths p dropped to 30 % (Ivanoff and Waterman 1958b). Tyler (1963) calculated that even under cloudy skies, p of light in deep water can reach 30-40 %. The lowest p occurs at approximately 470 nm, at which light is least attenuated in clear seawater (Hawryshyn 1992).

Lythgoe and Hemmings (1967) reported that when the transmission axis of a linearly polarizing filter was oriented in front of their eyes to exclude the maximally polarized underwater spacelight, the apparent brightness of small


fish (Sparidae and Atherina) was reduced less than the background spacelight, and thus fish had a higher contrast against their background. Furthermore, distant fish, invisible to the naked eye became visible using the polarizer. Photographing underwater white, black and grey panels through a linear polarizer with two orthogonal transmission axes from different distances and at various depths under cloudless skies, Lythgoe and Hemmings observed that the unobstructed water background undergoes a greater brightness change than the targets. One target, whieh was brighter than its background with a given orientation of the polarizer, became darker when the polarizer was rotated by 90°.

In another experiment, apolarizer was fixed to the outer surface of a diving heImet and oriented to exclude the maximally polarized underwater light. The horizontal distances at whieh different grey underwater targets just became invisible seen through the heImet with and without the polarizer were measured. Lythgoe and Hemmings found that the polarizer increased or decreased the range at whieh underwater objects brighter or darker than the water background could be seen. They suggested that polarization sensitivity could enable aquatie animals to see distant objects in clear waters. However, Luria and Kinney (l974) have argued that there are so many drawbacks in using polarizers as a means of enhancing contrast that underwater vision with polarizers is not reliably superior to vision without them.

Novales Flamarique and Hawryshyn (l997a) measured the spectral distribution of the underwater polarized light field at a depth of 4 m under clear skies in the upper photie zone of meso-eutrophie waters, i.e. blue-green waters containing medium to high chlorophyll-A concentrations. They found that the maximum degree of polarization Pmax during the day was 35-40 %, but at dawn and dusk it increased to 67 %. The reason for this is that during the crepuscular period underwater polarization is mainly determined by the incident light from the sky, whieh is maximally polarized when the sun is on the horizon. At dawn and dusk the relative contribution of blue and UV wavelengths to the total spectrum also increased.

The above-mentioned pioneering results of Waterman (1954) were confirmed also by the polarimetrie investigations of Cronin and Shashar (2001). They measured the spatial and temporal variation of the radiance I, P and a of light in clear, tropieal marine waters under partly cloudy skies from 350 to 600 nm throughout the day on a coral reef at a depth of 15 m. They used a submersible rotating-analyzer point-source sequential polarimeter based on a spectrometer with sensitivity in the UV and visible spectral ranges. Light ente ring a collector aperture and passing through a rotatable linearly polarizing filter was conducted to the spectrometer by an optical fibre. Polarizational characteristies of the downwelling light were investigated in different directions in the upper hemisphere. They found that both p and a varied only slightly with wavelength. Light was sometimes less polarized in the UV and maximally polarized in the visible range, or vice versa, depending on the view-


ing direction, solar elevation and sky cloudyness. Thus, they could find no particular optimum wavelength range for polarization-sensitive photoreceptors of underwater animals. p was always less than 50 %.

The complex underwater intensity and polarization patterns are modified by spatio-temporal variations oflight distribution due to surface waves. They focus sunlight at different depths depending on their wavelength. Surface rippIes focus light at depths of a few centimetres, longer waves at greater depths (Schenck 1957). Wave-focusing causes flicker, the frequency of which decreases with increasing depth.

15 Circulary Polarized Light in Nature

Circularly polarized light is rare in nature. It occurs only at total reflection of underwater linearly polarized light from the water-air interface outside the Snell window, at the reflection of light from the exoskeleton of certain insects and crustaceans, as well as in the bioluminescent light emitted by firefly larvae. Dinoflagellates are also known to induce circular polarization of light passing through their bodies (cf. Shashar et al. 1995a). However, optical structures which could selectively absorb and/or reflect left- or right-handed circularly polarized light have not been discovered until now in animal dioptrie apparatuses. Furthermore, the photoreceptors do not discriminate between left- and right-handed circularly polarized light. Animals perceive circularly polarized light as if it was unpolarized, while elliptically and partially linearly polarized light with the same degree and angle of polarization cannot be discriminated by them. Thus, it is questionable whether circular/elliptical polarization of light in nature could have any biological importance.

15.1 Circulary/Elliptically Polarized Light Induced by Total Reflection from the Water-Air Interface

At total reflection of underwater linearly polarized light from the water-air interface, both horizontally and vertically polarized components are completely reflected. However, after total reflection, vertically polarized light suffers a phase-shift, which depends on the angle of incidence and the index of refraction of water (Können 1985, p. 149). The phase-shift is largest at a particular angle of incidence, and the high er the refractive index of water, the larger is the phase-shift. Due to this phase-shift, part of the incident partially linearly polarized light is converted after total reflection into elliptically polarized light, while the intensity of reflected light remains the same. This elliptically polarized light was observed by Waterman (1954) as weH as Ivanoff and Waterman (1958a) underwater, elose to the surface, looking upward just beyond the critical angle. There, within a narrow ring around the Snell window, linearly polarized light produced by the scattering of directional rays of sunlight became elliptically polarized by total reflection from the underside

IS Circulary Polarized Light in Nature 101

of the water surface. The elliptically polarized totally reflected light is righthanded or left-handed whenever the angle of polarization of the linearly polarized incident light is 0° < a < +90° or -90° < a < 0° measured counterclockwise from the water surface, respectively. Since for total reflection the angle of incidence is always larger than the Brewster angle, the E-vector direction of the linearly polarized component of reflected light is the same as that of the incident light and is, therefore, not mirrored. Totally linearly polarized light is only converted entirely into circularly polarized light, if the phaseshift is 90° and the a of the incident light is 45° to the water surface. Since water has a relatively low index of refraction, a phase-shift of 90° cannot be produced with one total reflection. For water, multiple reflections of light are required for the complete conversion of linearly polarized light into circularly polarized light. The possible biological relevance, if any, of this underwater elliptically/circularly polarized light has not been studied yet.

15.2 Circulary Polarized Light Reflected from the Exoskeleton of Certain Arthropods

Michelson (1911) discovered that the light reflected from the beetle Plusiotis resplendens was circularly polarized like the light reflected from the exoskeleton of certain other beetles, butterflies and the feathers of some birds. He concluded that "the effect must therefore be due to a screw structure of ultramicroscopic molecular dimensions". Robinson (1966), investigating the chemical structure and optical properties ofliquid crystals and observing circularly polarized light reflected by liquid crystals, became fascinated by the studies of Michelson and obtained a variety ofbeetles, with which he repeated Michelson's investigations. Using a quarter-wavelength retarder,he found that the reflected light from these beetles was circularly polarized. He emphasized that "it would be of interest to consider what survival value can account for the occurrence of this most unusual property in so many species".

The "metallic" cuticles of certain scarabaeid beetles are able to reflect selectively left-handed circularly polarized light as can be shown by observing them with a circular polarizer (e.g. Caveney 1971). According to Können (1985, p. 84), some tropical butterflies have a coloured gloss caused by many reflections on the wingscales. In this case, an accidental combination of reflections can convert unpolarized light into circularly polarized light in tiny spots on the wing depending on the angle of incidence. Rose chafers (Cetonia aurata), cock chafers (Melolontha sp.), summer chafers (Rhizotrogus solstitialis), garden chafers (Phyllopertha horticola) and some other beetles belonging to the family of Scarabaeidae, are partly or entirely dark with a greenish or yellowish left-handed completely circularly polarized metallic gloss irrespective of whether or not the incident light is linearly polarized


(Können 1985, pp. 84-85, plates 55 and 56 on p. 83). Certain mutant specimens reflect not left -handed, but right -handed circularly polarized light. The circularly polarized gloss spreads all over the body of certain beetles, e.g. rose chafers, and is retained after the death of the animals. This circular polarizati on is caused by the helical structure of the molecules of the chitinous cuticle (Können 1985, p. 139-140). The direction of rotation of the E-vector of circularly polarized reflected light depends on the sense of rotation of the helix of the molecules. In living organisms, the ability to produce a given helical molecule is restricted to one sense of rotation, which has been fixed at a very early stage in evolution. Thus, apart from some mutants, this sense is the same for allliving organisms. Since the exoskeletons of all beetles reflecting circularly polarized light consists of the same substance, the sense of rotation of the Evector of reflected light is the same, left-handed, for all of them.

Using a simple circularly polarizing filter composed of a retarder (.1 = 540 nm) and a linear polarizer, Wolken (1995, p. 189) found that the light reflected from the June beetle (Scarab phyllophaga) and Japanese beetle (Popillia japonica) is circularly polarized. These beetles are of bright metallic colours, but viewing them through a circularly polarizing filter they appear black as the circularly polarized reflected light is extinguished.

The observations that circularly polarized light is reflected from the cuticle of certain insects raises the question whether the eyes of these animals are able to detect this circularly polarized light. The answer is unknown. If they do perceive circularly polarized light, they need a retarder that changes circularly polarized light to plane-polarized light. The cornealienses of many insects are birefringent (e.g. Meyer-Rochow 1973), but it is unknown whether they function as a retarder. If such an optical system existed, it would enable the insects to detect circularly polarized light. The biological function, if any, of this phenomenon is completely obscure. One function, for example, could be to filter the left-handed circularly polarized reflection gloss of the cuticle bya right -handed circularly polarizing filter within the visual system in order to perceive the species-specific colour and brightness patterns on the exoskeleton.

According to Neville and Luke (1971), the birefringent cuticle of certain crustaceans also reflects circularly polarized light.

15.3 Circulary Polarized Light Emitted by Firefly Larvae

While investigating circular polarization in luminescence, chemiluminescence and bioluminescence, Wynberg et al. (1980) found that the left and right lanterns of live larvae of the fireflies Photuris lucicrescens and Photuris versicolor emit left-handed and right-handed circularly polarized light, respectively, at a peak wavelength of 540 nm (Fig. 15.1). To explain the origin of the

15 Circulary Polarized Light in Nature

Fig.15.1. Schematic representation of a firefly larva in ventral view. The two grey spots at the end of the abdomen mark the lanterns emitting right- and left-handed circularly polarized light. (After Wynberg et al. 1980).

le ft lanlem cmilling le ft-handcd circularly

polarizcd lighl

103

righl lanlem cmilling righl-handcd c irc ularly

polarized lighl

circular polarization of bioluminescence, it was assumed that either a circularly dichroic medium partially absorbs the emitted light, or a chiral excited state is involved in the mechanism of bioluminescent light mediated by luciferin. To explain the opposite handedness of circularly polarized light emitted by the left and right lanterns, Wynberg et al. (1980) proposed that the light from a bioluminescent emitter may be partially linearly polarized due to anisotropy of an absorbing medium or by inhomogeneous formation of excited states due to local molecular organization within a photocyte. This linearly polarized light will become circularly polarized when it passes through a linearly birefringent medium. Orientated biopolymers can act as such a medium. On a macroscopic scale the larvae, like many other organisms, have a longitudinal symmetry plane dividing the lanterns. This mirror image of the membrane structure and orientation of the emitters may be responsible for the handedness of the emitted circularly polarized light. The possible biological function, if any, of this phenomenon is unknown. It is also unclear whether firefly larvae can perceive circularly polarized light.

16 From Polarization Sensitivity to Polarization Vision

16.1 Forerunners of the Study of Animal Polarization Sensitivity

The forerunners of the study of polarization sensitivity in animals are Santschi (1923), Crozier and Mangelsdorf (1924), Verkhovskaya (1940) and Frisch (1949). In one ofhis behavioural experiments, Santschi depolarized the skylight with a ground glass disk held above an ant running on the ground to its nest. He observed that the ant suddenly stopped and searched around randomly. However, from this disorientation he did not draw the conclusion that the ant could perceive the polarization of skylight (Wehner and Rossel 1985, p. 13). This hindered hirn to discover that certain ants can perceive the skylight polarization and use it for navigation.

Crozier and Mangelsdorf (1924) were the first to use Nicol prisms to study the responses of blowflies and beetle larvae to linearly polarized light. In their pioneering laboratory experiments, the animals had to choose between a polarized and an unpolarized beam of light of the same intensity. They did not observe the ability of the animals to distinguish polarized light from unpolarized light. This ability was observed first in Drosophila and daphnid crustaceans byVerkhovskaya (1940). Frisch (1949) first demonstrated that the polarization of skylight is the most important compass cue for orientation in honeybees when the sun is occluded, but patches ofblue sky are visible. Stockhammer (1956) was the first to suggest that insects might be sensitive to the E-vector direction of linearly polarized light because they possess dichroically absorbing photoreceptors. The first training experiments using linearly polarized light were performed also by Stockhammer (1956) with honeybees. In this third part of the volume, we survey the results of researchers who continued the investigation of animal polarization sensitivity after these pioneering forerunners.

108 Part III: Polarized Light in Animal Vision

16.2 Polarization Sensitivity, Polarization Vision and Analysis of Polarization Patterns

In the literature, "vision or perception or sensing of light polarization" can me an at least two quite different things (Kirschfeld 1973b):

1. Certain animals with photoreceptors mediating a polarization-sensitive reaction may not perceive polarization as a special sensory quality, rather only intensity differences due to the receptors' inherent dichroism.

This kind of polarization-sensitive visual system is analogous to a camera, wh ich records the optical environment through a linearly polarizing filter. Changing the direction of the transmission axis of the polarizer results in intensity and colour variations in the recorded pictures (see ~ colour Figs. 1.1 and 1.2).

2. The polarization-sensitive visual system can perceive polarization as a unique sensory quality distinct from other characteristics of light, such as wavelength and intensity.

This type of polarization-sensitive visual system is analogous to an imaging polarimeter, which records and visualizes the spatial distribution of the intensity, degree oflinear polarization and E-vector direction oflight in given ranges of the spectrum (see e.g. -7 colour Figs. 1.4, 1.5 and 4.4). However, the majority of the experiments performed with animals do not reveal whether the polarization-sensitive response or behaviour observed belongs to the first or the second category. According to Kirschfeld (1973b), it is pertinent to use the following nomenclature and basic terms in connection with polarization sensitivity, polarization vision and polarization-guided behaviour:

• Spontaneous Orientation to Polarization (Basopolarotaxis). When certain animals are illuminated from above with linearly polarized light, they may orient spontaneously at certain angles (most frequently 0/180° and/or ±45° and/or ±900) relative to the E-vector direction.

In laboratory experiments, many aquatic arthropods, e.g. Daphnia pulex (Hazen and Baylor 1962), Cyclops vernalis (Umminger 1968), Hyalella azteca, Arrenurus sp., and Bidessus flavicollis (Jander and Waterman 1960) and others (e.g. Baylor and Smith 1953; Bainbridge and Waterman 1957) displayed behavioural responses to linearly polarized light and oriented relative to the E-vector independently of their phototactic condition. Similar polarotaxis was observed, for instance, in the ants Tapinoma sessile and Solenopsis saevissima, the housefly Musca domestica, the firefly Photurus pennsylvanicus and

16 From Polarization Sensitivity to Polarization Vision 109

the Japanese beetle Popillia japonica (Wolken 1995, p. 186, 187). On the basis of these experiments, the phenomenon of basopolarotaxis was established (Waterman 1981) as a behaviour distinct from the well-known phototaxis.

Many studies reinforced the occurrence of polarotaxis in aquatic arthropods, but only a few researchers investigated the role of underwater polarization (Waterman 1988). Young and Taylor (1990) proposed a reasonable explanation for the polarotactic orientation they observed. They suggested that by perception of E-vector direction filter-fee ding cladocerans can avoid water they had previously filtered. In contrast, predators of these cladocerans have the opposite orientation relative to the E-vector of polarized light, thus maximizing their encounter rate with the prey. Several other functions of polarotaxis have been proposed, including (1) contrast enhancement between an object and its background (e.g. Lythgoe and Hemmings 1967), (2) guiding vertical migration of copepods (e.g. Umminger 1968), (3) maintenance of body position (Waterman 1981), and (4) orientation (e.g. Bardolph and Stavn 1978), but only partly proven.

• Optomotor Response to Moving Polarization Patterns. Certain animals respond by turning their head and/or body to moving patterns containing exclusively appropriate E-vector contrasts, i.e. segments with different Evector directions.

This reaction is thoroughly dealt with in Chap. 27.

• Menotactic Orientation to Polarization (Menopolarotaxis). Some animals are able to determine the bearing of the sun occluded by clouds on the basis of the polarization of light from cloudfree patches of the sky.

The menopolarotactic behaviour has been most intensively studied in the honeybee Apis mellifera (Chap. 17.1) and the desert ant Cataglyphis bicolor (Chap. 17.3). The grass shrimp Palaemonetes vulgaris was the first aquatic arthropod in which orientation by means of the celestial polarization pattern visible from water through the Snell window was convincingly proven (Chap.25.8).

• Successive Determination of the E-vector Direction. The animal rotates its polarization-sensitive receptor(s) relative to the E-vector direction. This may be performed by rotation of the eye, the head or the body. With this successive method, a single polarization-sensitive receptor is sufficient to determine the state of polarization, since from the absorption extremes Imax and Imin both the angle of polarization a and the degree of linear polarization p can be obtained unequivocally: p = (Imax-Imin)l(Imax +Imin), and a coincides with the angle of the transmission axis of the analyser relative to a reference direction when I = Imax.


Note that this method is in strict analogy to the technique of sequential polarimetry (Chap. l.5). In a special kind of the successive method, in the "scanning model" an animal determines the symmetry plane of the celestial polarization pattern by scanning the skywith its polarization-sensitive dorsal rim areas, experiencing maximal excitation whenever the retinal array of receptors matches the celestial E-vector pattern (Rossel and Wehner 1986). Although one polarization-opponent interneuron integrating E-vector information from the whole dorsal rim area would be sufficient to achieve this task, additional units directed forward and backward could serve to discriminate between solar and antisolar direction based on the different degrees of polarization of skylight and/or they could playa role as auxiliary units in finetuning the animal's alignment.

• Simultaneous Determination of the E-vector Direction. To determine the state of polarization the animal would not have to rotate its ommatidia relative to the E-vector direction, if every ommatidium or neighbouring ommatidia with overlapping fields of view contained at least three polarization-sensitive receptors of the same spectral sensitivity with three different microvilli directions (Kirschfeld 1972, 1973b). Another way of unambiguous determination of the E-vector direction would be possible with the binocular co operation of two ommatidia, one in the left and another in the right eye, within which there are two receptors of the same spectral type with two different microvilli orientations (Menzel and Snyder 1974). In an alternative model, one polarization-sensitive receptor and two intensitydetecting and polarization-insensitive receptors of the same spectral type in every ommatidium are postulated (Wehner et al. 1975). Another possibility could be the existence of two polarization-sensitive receptors with different microvilli directions and either the perception of the intensity (Menzel and Blakers 1975) or the degree oflinear polarization (Glas 1977).

All these models are analogous to the technique of simultaneous polarimetry (Chapt. l.5) and share the common feature that at least three independent measurements are required to determine the state of polarization instantaneously by comparing the responses of three or more polarization-sensitive receptors with different microvilli directions (Edrich and Helversen 1987). This prerequisite follows from the fact that the optical characteristics of linearly polarized monochromatic light can unambiguously be described by three independent parameters, e.g. by the degree and angle of polarization and intensity (see Chap.1). One could regard the three polarization-opponent interneurons found in field crickets (Labhart 1988) as "macroreceptors" giving input into such a three-channel simultaneous E-vector detecting system.

At present, it is unknown whether animals use the simultaneous or the successive method for E-vector detection. Certain researchers (e.g. Edrich and Helversen 1987; Kirschfeld 1988b) prefer the former hypothesis, while others


(e.g. Rossel et al. 1978; Rossel and Wehner 1982, 1984a,b, 1986; Rosse11993; Wehner 1982,1983,1989,1994,2001) favour the latter one.

According to Bernard and Wehner (1977, p.1 020), "polarization vision" is the ability to discriminate between two monochromatic lights of the same intensity I, but of a different angle of polarization a and/or degree of linear polarization p. On the other hand, a visual system "has true polarization vision" if it can perceive a independently of variations in I, p and spectral composition. The existence of true polarization vision, in its strictest sense defined above, has never been proven in any animal. In the majority of behavioural experiments studying animal polarization sensitivity, usually only a has been varied between two linearly polarized light stimuli and both land p were fixed. Thus, true polarization vision cannot be deduced from these experiments. The most elaborate and best-studied polarization-sensitive visual systems are those in honeybees (Chap. 17.1) and desert ants (Chap. 17.3). The information these systems provide is not the individual E-vector directions, but the compass courses, which are most probably derived from the global processing ofE-vector gradients in the sky (Wehner 2001, p. 2595).

16.3 Functional Similarities Between Polarization Vision and Colour Vision

Moody and Parriss (1961) first showed that a polarization-sensitive visual system with two orthogonal microvilli directions inevitably confuses every angle of polarization a with -a. Kirschfeld (1973b) was the first to point out that a fixed simultaneous system with only two independent measurements is insufficient for determination of a if I or p are independent parameters. He showed that three independent receptors with different microvilli directions can simultaneously determine a independently of p and 1.

According to Bernard and Wehner (1977, p. 1020), polarization vision (PV) is analogous to trichromatic colour vision (CV), which is the ability to discriminate lights of the same intensity I, but of different dominant wavelength Ä and/or colour purity c. The foundation of this analogy between PV and CV is their common three-dimensionality: PV(I,a,p), CV(I,Ä,c). Thus, colours are analogous to the state of linear polarization. Bernard and Wehner considered a and p to be analogous to Ä and c, respectively. They also concluded that three independent receptors must cooperate to determine a independently of p and 1. Thus, a PV system based on only two receptors is not polarizationblind, but is characterized by "neutral points"l and "confusion states" if the

1 Note, however, that these neutral points have nothing to do with the Arago, Babinet, Brewster and fourth (Horvath et al. 2002b) neutral points of skylight polarization, at which p = 0 (see Chap. 4).


eye is fixed. A PV system with vertical and horizontal microvilli of the same PS-value, for example, cannot discriminate between E-vectors aligned ±45° from the vertical, independently of p. The E-vectors with a = +45° and a = -45° are called "neutral points" by Bernard and Wehner. If I is constant and a and p vary in such a way that the amount of light absorbed by the two receptors is also constant, then such polarizational states cannot be discriminated from each other by a two-channel PV system. These states of polarization are named "confusion states" by Bernard and Wehner. Similarly, a dichromatic CV system is characterized by "spectrally neutral points" and "confusion colours".

16.4 How Can Skylight Polarization Be Used for Orientation?

Since the celestial cues are practically at infinity and thus not subject to motion parallax, their retinal position does not change when the animal moves without rotation. The entire pattern of skylight polarization rotates around the motionless pole point, where the earth's axis of rotation intersects the celestial hemisphere (coinciding with the daytime invisible pole star Polaris in the northern hemisphere of the globe). Brines (1980) proposed that useful information is available to animals, which can utilize the dynamic properties of skylight polarization. The following information could not be obtained by simple observation of the static celestial polarization pattern alone:

• The pole point that could provide a fixed locus for an animal which could see the uniform rotation of the entire pattern of skylight polarization around this point with an angular speed of I5°/h.

• The azimuth of the pole point, which provides true compass direction. Its elevation is the observer's latitude, and the E-vector direction at the pole point provides true local solar time.

• The dynamic celestial polarization pattern, wh ich might be used to calibrate another orientation systems by providing time, solar declination or other navigational variables. This information would be very useful for animals migrating long distances and thus exposed to strong changes in these variables.

Animals might locate the pole point approximately by detecting the celestia! band of maximum degree of linear polarization which rotates around this point. At the equinoxes, this band passes through the pole point. The maximal angular distance of this band is 23.so from the pole point.

The question, whether the above celestial cues are indeed used by animals, remains open until direct measurements of both the threshold at which an


animal can detect E-vector rotation and the capability to discriminate between slightly different E-vector directions are performed. An important biological question is whether animals can detect changes in celestial polarization patterns at rates as low as 15° /h, and if so, how much time is needed to attain various degrees of accuracy. We shall see in this part of the volume that honeybees and many other arthropods use the celestial polarization pattern only to determine directions relative to the plane of the slowly rotating solar meridian, rather than to a fixed reference point such as the pole point.

We have seen in Chap. 10 that whereas the celestial gradients of intensity, spectral composition and degree of linear polarization are highly susceptible even to weak atmospheric dis turban ces like haze, pollution or thin clouds, the gradient of E-vector directions is not, and thus it provides the most reliable cue for navigation. This is the reason why animals use the celestial E-vector pattern as a compass cue.

Kirschfeld et al. (1975) emphasized that the perception of the E-vector direction of skylight at any celestial point alone is insufficient to determine the solar position. To reduce the number of possible solar positions from infinity to two or one, the E-vector directions in at least two points of the sky, not too near to each other, are needed, or in addition to the E-vector direction in one celestial point, at least one other parameter, for example the solar elevation, is necessary. In this case, usually there are two solutions to the problem, since the great circle through the sun and the celestial point observed crosses the horizontal height-circle through the sun (called almucantar) generally at two points. To select the true solar position from these two points, additional information, e.g. the degree of linear polarization or intensity of skylight is needed, except the two special cases when the two mentioned celestial circles are tangential to each other, or the celestial point observed has the same elevation as the sun. The hidden assumption behind these considerations is that the E-vector is always perpendicular to the scattering plane, i.e. the skylight polarization pattern coincides with the single-scattering Rayleigh pattern.

In principle, an animal orienting by means of celestial polarization should first perceive the directions of particular E-vectors of skylight and then rely on precise knowledge of all possible E-vector patterns at different solar elevations (Wehner 1994). To determine E-vector direction, a polarization-sensitive photoreceptor must be able either to rotate about its optical axis, or to compare its output with that of adjacent receptors with different preferred Evector directions. If the animal were able to derive correct compass information even from small parts of the celestial E-vector pattern, as gene rally believed, it should be provided either with a general rule of how to infer any particular compass direction from individual celestial points, or the information about all possible celestial E-vector patterns gathered by evolutionary or individual experience. On the other hand, Gould et al. (1985) assumed that honeybees could remember the celestial E-vector pattern last seen and could


later match it to the E-vector pattern in patches observed in the sky. It will be demonstrated in Chapt. 17.1 and 17.3 that at least honeybees and desert ants are not able to orient by these methods.

Inspired by the polarized skylight compass of honeybees, desert ants and field crickets, Lambrinos et al. (1997) constructed a mobile robot, called Sahabot, which navigated by means of the celestial polarization pattern. The robot had three pairs of polarization-sensitive units consisting of two photodiodes below two linear polarizers with orthogonally oriented transmission axes. These sensors were based on neurobiological findings in Apis mellifera, Cataglyphis bicolor and Gryllus campestris and they mimicked the polarization-opponent neurons found in field crickets (see Chap. 17.4). With the signals from these sensors, the robot was able to determine its body orientation. Three models for extracting compass information from skylight polarization were tested in the Tunisian desert, where the ants of the genus Cataglyphis live. In the "scanning model", the robot had only one polarization-opponent sensor unit, scanned the sky by rotating about its vertical axis, and detected the solar-antisolar meridian by finding the maximum response of the sensor unit. A similar mechanism has been suggested by Wehner (1989) in honeybees and desert ants as well as by Labhart (1999) in field crickets. In the "extended scanning model", the robot detected the region of the response maximum by means of the signal from one of its three sensor units, and the signals from the two other "helper units" were used only for finding the response maximum more precisely. An analogous mechanism has been proposed for hypothetical polarization-sensitive neurons in honeybee navigation by Rossel (1993). In the "simultaneous model", the robot derived its orientation directly by continuously comparing the outputs of the three zenith-centred sensor units without scanning. Such a mechanism was proposed by Kirschfeld (1972, 1988a).

During the field experiments, the robot had to steer in a given direction, travel a certain distance, and then return to the starting position along the same route, relying solelyon skylight polarization and proprioception under clear ski es. All three models were suitable for navigation, and their performance did not change when the sky was loosely scattered with clouds. The largest directional error was about 5°. The average errors were approximately 1.7,1.2 and 0.7° for the scanning, the extended scanning and the simultaneous model, respectively. In comparison, note that the homing errors of desert ants are about 1° (Wehner and Wehner 1986). Hence, the robot apparently performed better with the simultaneous than with the scanning mechanisms. In the latter cases, the robot depended entirely on proprioception during its journey, i.e. between the sky scans, and orientation errors could be corrected only during the scans. In the former case, the robot was guided by the celestial polarization pattern during its entire journey and could correct errors at any moment. Performance with the scanning models may be improved by more reliable proprioception or more frequent sky scanning.


16.5 Possible Functions of Polarization Sensitivity

Polarization sensitivity could have several possible functions, and responses to polarization are always task-dependent performances in animals:

- Skylight polarization for honeybees and desert ants, for instance, is a cue for steering compass courses (e.g. Wehner 1994).

- Some aquatic insects detect their habitat by the horizontal polarization of light reflected from the water surface (e.g. Schwind 1991).

- Certain insects, e.g. Gerris lacustris (Schneider and Langer 1969) and Sympycnus lineatus (Trujillo-Cen6z and Bernard 1972) may employ dichroic photoreceptors to reduce glare resulting from polarized reflections on the water surface and wet surfaces, thereby improving the detection of prey against shiny backgrounds.

- Polarization might enhance course control by assisting the maintenance rather than the selection of a course, for example, by providing the animal with some kind of overhead stabilizer (e.g. Bardolph and Stavn 1978; Wolf et al. 1980; Brunner and Labhart 1987) or artificial horizon (Laughlin 1976).

- In underwater vision, polarization sensitivity may increase the visibility of objects by enhancing the contrast between them and their background (e.g. Lythgoe and Hemmings 1967; Umminger 1968).

- Communication by polarized light signals has been suggested in certain crustaceans (e.g. Neville and Luke 1971), the insects Sympycnus lineatus (Trujillo-Cen6z and Bernard 1972) and Poecilobothrus nobilitatus (Land 1993), the squid Loligo pealei (Shashar and Hanlon 1997; Hanlon et al. 1999; Shashar et al. 2001) and the European cuttlefish Sepia officinalis (Shashar et al. 1996).

- An aquatic prey near the water surface is particularly vulnerable to underwater predators, because it is silhouetted against the background spacelight (Bone and MarshallI982). Many mesopelagic fish species have upwardly directed eyes to detect their prey when it is silhouetted against the bright downwelling light (Marshall 1979). Even transparent aquatic animals are visible, since the refractive indices of their tissues are generally different from the refractive index of water and, therefore, they appear somewhat darker or brighter than their background. In mid-water at least two kinds of radiance matching have been observed: mirror camouflage and downward-directed bioluminescence (Lythgoe 1979). Aquatic predators have different possibilities to detect camouflaged preys, one of them is polarizati on sensitivity. The polarization of light reflected downwards from the body of a prey (Denton and Nicoll965) and the polarization of the emitted bioluminescent light (Herring 1977) differs from that of the skylight within the Snell window. A polarization-sensitive predator could, therefore, per-

116 Part I1I: Polarized Light in Animal Vision

ceive the camouflaging prey against the refraction-polarization pattern of skylight. In addition, the sensitivity to the direction of polarization could enhance the contrast of an underwater target. For this the water surface need not be flat and for this kind of contrast enhancement, not only the refraction-polarization pattern of skylight within the Snell window, but also the underwater polarization pattern outside the Snell window might playa role.

In this part of the volume, we present several further examples how the polarization sensitivity in animals can guide different types of behaviour. We show in Chap. 33 that in certain circumstances polarization sensitivity could also be more of a handicap than an advantage. Then the detection of plant surfaces is improved when the visual processing is not disturbed by polarization of light reflected from shiny plant surfaces. For identifying the hue of colours, it is better to use polarization-insensitive photoreceptors, so that the polarization-induced false colours do not disturb the perception of real colours.

16.6 How Might Polarization Sensitivity Have Evolved?

Shaw (1969) and Snyder (1973) have shown that the overall absorption efficiency for unpolarized light is significantly high er in fused and layered (e.g. crustacean) rhabdoms with interdigitating orthogonal microvilli than in open and non-Iayered (e.g. dipteran) rhabdoms with a single microvilli direction, if the length and diameter of the fused and open rhabdoms are the same. Shaw (1969) proposed first that the prevalence of the fused rhabdom with opposing microvilli in the arthropods may be attributed to enhance the absorption of unpolarized light and not necessarily to any advantage conferred by possession of a polarization-sensitive mechanism as suggested by Rutherford and Horridge (1965), for example.

Snyder (1973) calculated that in open rhabdoms high absolute sensitivity is associated with low polarization sensitivity (PS), while in fused rhabdoms high absolute sensitivity is associated with high PS. Snyder and Laughlin (1975) concluded that in fused rhabdoms the dipole axes of the photopigment moleeules must be aligned in parallel to the microvillus long axis to provide the rhabdom with a maximum absorption of unpolarized light. The exact parallel alignment would result in a maximal dichroic ratio of ab out 20. They suggested that photoreceptor membrane dichroism may only be a consequence of an adaptation to provide the photoreceptor with a maximum sensitivity to unpolarized sunlight rather than with polarization sensitivity. Inspired by these facts, Wehner (1994) proposed the following hypothetical evolutionary scenario:


1. Since small rhabdomeric photoreceptors catch more quanta even from unpolarized light, arthropods might have first evolved densely packed stacks of microvillar membrane to increase their light sensitivity. Photoreceptor cells with microvilli might have been an ancient evolutionary innovation, predating mid-Cretaceous periods when angiosperm plants evolved and certain hymenopterans became flower-visiting insects. As a by-product, such receptors have become dichroic and thus sensitive to polarization.

2. Since the detection of the direction of polarization has become advantageous and could have been used for various purposes, polarization sensitivity of the photoreceptors might have been increased by strictly aligning the microvilli within short rhabdomeres to counterbalance the inevitable effect of self-screening, or by combining several rhabdomeres into a single waveguide structure, the rhabdom, to allow for serial or lateral dichroic 61-tering.

3. When polarization sensitivity became an unwanted side-effect of enhanced quantum catch in colour vision after the appearance of angiosperm plants, it was reduced in flower-visiting insects by introducing structural irregularities into the microvilli stacks, e.g. causing rhabdomeric twist or wobble, by enhancing self-screening or by electrical coupling of photoreceptors with different microvillar directions.

In the opinion of Wehner (1994), evolution has worked either way, to increase or to decrease polarization sensitivity, depending on the particular needs of a given species.

We describe in this third part of the volume how a variety of insects, crustaceans, cephalopod mollusks, squids and octopuses as well as vertebrates are sensitive to the linear polarization of light. Since arthropods have compound eyes, cephalopod mollusks and vertebrates possess single lens eyes with quite different retinal and central nervous system architectures, it is pertinent to suppose that polarization sensitivity has evolved in animals in at least three independent ways. Based on fine structural disparities in the design of ommatidia in the anatomically and physiologically specialized dorsal rim area serving for perception of skylight polarization in the compound eyes of different insect groups, Labhart and Meyer (1999) hypothesized that polarization sensitivity may have arisen polyphyletically in insects.

16.7 Polarization Sensitivity of Rhabdomeric Invertebrate Photoreceptors

Baylor and Smith (1958), Kalmus (1958, 1959) and Waterman (1981) proposed that in certain cases the reactions of some animals to polarization could be


elucidated by the intensity pattern induced due to selective reflection or scattering of linearly polarized light from the optical environment. This problem is thoroughly discussed in Chap. 34.

16.7.1 Hypothetieal Polarizing Ability of the Dioptrie Apparatus

Stephens et al. (1953), Baylor and Smith (1953), Hazen and Baylor (1962) as well as Skrzipek and Skrzipek (1974) suggested that the analysers for polarization may be located somewhere within the dioptrie apparatus of the compound eyes in certain invertebrates. Baylor and Smith (1953) supposed that the polarization sensitivity of invertebrate photoreceptors may be due to the Fresnel reflection characteristies of the overlying optieal media inc1uding the eye surface. In the so-called Brewster-Fresnel reflection-refraction model, one or more refractions and internal reflections at the lens surface in the eye alter preferentially the intensity of light polarized parallel to the plane of inciden ce. Stephens et al. (1953) proposed such a model, relying upon a single refraction, for the Drosophila eye. A modified version of this model, relying on internal reflection from the lentieular surface, was hypothesized by Baylor and Smith (1953) for Daphnia magna. Baylor and Kennedy (1958) as well as Kennedy and Baylor (1961) considered it unlikely that the rhabdomerie retinula cells are the polarization analysers in honeybees.

Since the dioptrie apparatus within an insect's ommatidium proved to be isotropie for natural on-axis illumination (Stockhammer 1956), these hypotheses have been refuted. Hence, the optics of the compound eye do not serve as apolarizer. Nevertheless, in the anatomieally and physiologieally specialized highly polarization-sensitive dorsal rim area of various insect species, the cornealienses are markedly specialized for the enhancement of the visual field of the polarization-sensitive receptors.

Interestingly, the eye of the extinct trilobites had hundreds of lenses consisting ofhighly birefringent calcite crystals (e.g. Clarkson 1979). These crystals were arranged in such a way that they could not have ac ted as polarization analysers (Horvath et al. 1997b). Of course, this does not exc1ude the possibility that the trilobites were able to use the polarization of the underwater light field for orientation in their marine habitat.1t only means that trilobite calcite lenses were not analysers located in front of the photoreceptors.

16.7.2 Rhabdomerie Polarization Sensitivity

Autrum and Stumpf (1950) first suggested that polarization sensitivity may be an intrinsie property of invertebrate photoreceptors. The presence of diehroic filters in the eye of the honeybee was suggested by Autrum (cf. Frisch 1950) and has been supported by Stockhammer (1956, 1959). The hypothesis


of Autrum and Stumpf (1950) that mierovillar photoreceptors can respond to different E-vector directions of linearly polarized light has been proven by Kuwabara and Naka (1959) in Lucilia caeser and Burkhardt and Wendler (1960) in Calliphora erythrocephala with intracellular recordings from photoreceptors. These were the first recordings whieh could prove that the intrinsie mierovillar diehroism rather than the polarizing ability of the dioptrie apparatus is responsible for insect polarization sensitivity.

The amount of light absorbed by a visual pigment molecule is the greater, the narrower the angle between the E-vector of polarized light and the dipole axis of the pigment molecule. Thus, a pigment molecule absorbs maximally or minimally if the E-vector is parallel or perpendieular to its dipole axis. This dependence of the absorption of polarized light on the E-vector direction is called linear diehroism. De Vries et al. (1953) proposed first that if the microvilli of a photoreceptor were parallel to each other (Fig. 16.1A), and if the visual pigments embedded in the photoreceptor membrane exhibited diehroie absorption, preferential alignment of the absorption axes of the pigment molecules within the mierovillar membranes (Fig.16.1B) would be sufficient to explain the polarization sensitivity.

The first electron-mierograph of an insect retina was made by FernandezMoran (1956), who demonstrated that the photoreceptor membranes are rolled into narrow tubes called mierovilli, whieh are oriented at right angles to the optieal axis of the cell (Fig. 16.1A,B). Shaw (1967) demonstrated that photoreceptors located within the same ommatidium, and hence receiving linearly polarized light through the same dioptrie apparatus, differ in their preferred E-vector direction.

In invertebrates polarization sensitivity arises from the diehroie properties of mierovilli. In terrestrial insects the photoreceptor cells have non-overlapping microvilli (e.g. Labhart and Meyer 1999). Among aquatie invertebrates, mierovilli may be arranged in a mostly non-overlapping fashion, e.g. in some aquatic beetles, cladocerans and other zooplankton (Waterman 1981; Schwind 1995), or in stacks with alternating sections of orthogonal microvilli in cephalopods and decapod crustaceans (e.g. Moody and Parriss 1961; Waterman and Horch 1966; Waterman 1981).

Microvillar photoreceptors are inherently polarization sensitive. The tubular arrangement of the photoreceptor membrane already results in some diehroism, even if the pigment molecules are oriented randomly (Fig. 16.1C) within the tangent planes of the tube-shaped mierovilli membrane (Stockhammer 1956; Moody and Parriss 1961; Laughlin et al. 1975). Such a mierovillar structure absorbs maximally twiee as much totally linearly polarized light if the E-vector is parallel to the mierovilli (Fig. 16.1D,E). This so-called "form diehroism" is relatively low resulting in diehroie ratios 6 ~ 2 if the orientation of the dipole axes of the photopigment molecules is random (Fig. 16.1C) rather than parallel (Fig. 16.1B) to the tangent planes of the mierovilli membrane. However, electrophysiologieally measured polarization sensitivity in a

120

photoreceptor with microvilli

, \

c


B

D

Fig.16.1. A Schematic structure of a rhabdomeric photoreceptor in the compound eye of invertebrates with finger-like protrusions of the receptor membrane called microvilli, the long axes of which are perpendicular to the incident light. B An enlarged part of the rhabdomere with the dipole axes of pigment moleeules indicated by double-headed arrows. If the dipole axes of the pigment molecules are parallel to the long axis of the microvilli, then the selectivity of absorption of linearly polarized light is maximized. CA single microvillus is intrinsically selective to linearly polarized light, even if the dipole axes (bars) of the visual pigment moleeules are randomly oriented in the membrane, because more pigment molecules are aligned along the tubulus than across it. D, E This is more obvious in two orthogonal section of the microvillus, where twice as many photopigment moleeules (bars) must be aligned parallel to the long axis compared with either orthogonal direction. (After Wehner 1976 and Land 1991).


variety of arthropod species usually reach values as high as PS = 5-60. The prerequisite for the explanation of such high PS-values is the assumption that the chromophores are preferentially aligned relative to the long axis of the microvillus. With microspectrophotometry on isolated single stacks of crayfish microvilli (Orconectes and Procambarus), Goldsmith and Wehner (1977) proved that this is the case. They observed the absence of both translational diffusion and free Brownian rotation of the photopigment molecules in the membrane. They estimated that the dipole axes of the pigment molecules are aligned within about ±50° relative to the microvillar axis and are tilted into the surface of the membrane at an average angle of about 200 •

Since in vertebrates the visual pigment molecules undergo rotational and translation al (or lateral) diffusion within the photoreceptor membrane at such a high rate that any photo-induced dichroism is randomised within less than 0.1 ms (e.g. Cone 1972), the arthropod visual pigments should be kept somehow aligned. Viscosity characteristics of the membrane, the cytoskeleton within the microvilli (e.g. BIest and Stowe 1990) and membrane linkages between adjacent microvilli (e.g. BIest et al. 1982) have been proposed as possible stabilizing mechanisms for molecules and microvilli. The viscosity of the photoreceptor membrane is likely to be high er in invertebrate rhabdomeric photoreceptors than in vertebrate rod outer segments, for example as a consequence of the higher content of cholesterol in invertebrate photoreceptor membranes (Mason et al. 1973). Asymmetries in the shape of the visual pigment molecules and the incorporation of the molecules into a membrane that is bent into a narrow tube favour molecular alignment parallel to the microvillar axis by restricting the movement of the pigment molecules to the long axis of the tube (Laughlin et al. 1975). The diameter of a microvillus is about 50 nm and thus only approximately ten times that of the photopigment molecule. The cytoskeleton within the microvilli of rhabdomeric photoreceptors consists of an F-actin axial filament linked to the plasma membrane by side arms, which most probably consist of myosin heads (Arikawa et al. 1990). Due to this structural organization, the microvillar cytoskeleton is more likely to be involved in dynamic processes like membrane turnover and shedding (BIest and Stowe 1990) than in creating a paracrystalline array with specific optical properties. Thus, most probably, the extracellular bridges between the microvilli stabilize the microvillar structures.

16.7.3 Origin of High Polarization Sensitivity

The polarization sensitivity (PS) of a retinula cell is not simply related to the dichroic properties of its individual microvilli. Instead, it is determined by the properties of the entire rhabdom structure, particularly the arrangement of microvilli along its cross section and length (Snyder 1973). There are several mechanisms by which PS can be increased beyond the level defined by rhab-


domeric dichroism. Maximum polarization sensitivities of the photoreceptors as high as PS = 9 (Shaw 1969) and 15 (Stowe 1980) were found in crabs, 18 in honeybees (Labhart 1980),60 in certain butterflies (Horridge et al. 1983),

but only 1.5-3.5 in flies (McCann and Arnett 1972; Järvilehto and Moring 1976). The following mechanisms have been proposed for achieving such high polarization sensitivities:

- High dichroism of the visual pigment molecules in the microvillar membrane. A PS higher than 2 implies a preferential alignment of the visual pigment molecules in the membrane, and theoretically the highest possible PS is 20 when the dipole axis of all pigment molecules is exactly parallel to the long axis of the microvillus (Moody and Parriss 1961; Snyder and Laughlin 1975).

- Screening by an overlying rhabdomere with orthogonally oriented microvilli acting as a linearly polarizing filter, as is the case for the visual cell R9 in the honeybee (Menzel and Snyder 1974), for example.

- Reciprocal inhibitory interactions between visual cells with phase-shifted PS functions. A possible mechanism for this was proposed by Shaw (1975b) assuming electrical field interactions, and a probable example was found in the butterfly retina by Horridge et al. (1983).

- Rhabdoms (as weIl as vertebrate rod and cone outer segments with elongated cross sections) may preferentially carry E-vectors of one direction due to their waveguide properties. Such sensors may act as polarizers (Snyder and Love 1983), thus increasing PS.

- Due to waveguide effects, the smaller the diameter of open (e.g. dipteran) rhabdoms, the high er the PS (Snyder 1973).

- Off-axis illumination of open (e.g. dipteran) rhabdoms leads to a high er PS than on-axis illumination (Snyder 1973).

16.7.4 Origin of Low Polarization Sensitivity

There are also various mechanisms by wh ich PS is decreased or even abolished:

- Random alignment of photopigment molecules within the microvilli membrane relative to the long axis of the microvilli (Moody and Parriss 1961).

- Self-screening in long open (e.g. dipteran) rhabdoms. The longer the rhabdomeres and/or the higher the photopigment concentration within them, the lower the PS, but the higher the absolute sensitivity (Snyder 1973).

- Twist of the microvillar direction along the length of the rhabdom (Snyder and Mclntyre 1975)

- Intercellular electrical coupling between visual cells with phase-shifted PS functions (Shaw 1969).


- Arthropod rhabdoms (as well as vertebrate rod and cone outer segments) act as narrow waveguides conducting light in discrete waveguide mo des. Related modes out of phase with each other could partially depolarize light propagating along the rhabdom, thus leading to a decrease in ps. This effect is called the "mode polarization beating" (Snitzer and Osterberg 1961).

- Depolarization, and thus decrease of PS, can also be caused by light scattering or strong birefringence (Mclntyre and Snyder 1978).

- Screening by a pigment incorporated into the photoreceptor membrane, which absorbs linearly polarized light preferentially at right angles to the dipole axis of the visual pigment. This is the case in one dass of photoreceptors, cells R7y in the fly Calliphora erythrocephala. In this visual cell type a blue-absorbing photostable pigment occurs wh ich has its dipole axes oriented perpendicularly to those of the visual pigment. As usual, the latter are aligned parallel to the microvillar long axes. The dichroism of the accessory pigment explains the surprising finding that in this type of photoreceptor the absorption is maximal when light is polarized perpendicularly to the microvilli (e.g. Kirschfeld and Franceschini 1977). However, the sensitivity, as obtained electrophysiologically, is maximal when light is polarized parallel to the microvilli (Hardie et al. 1979).

16.7.5 Rhabdomeric Twist and Misalignment and Their Functional Significance

Earlier, there was a controversy about the twist of rhabdomeres along their longitudinal axes in insect eyes. It has been found outside the dorsal rim area of the compound eyes in Odonata (e.g. Ninomiya et al. 1969), Hymenoptera (e.g. Menzel and Blakers 1975; Wehner and Meyer 1981) and Diptera Ce.g. Smola and Tscharntke 1979; Williams 1980). Ribi (1979) has daimed that rhabdomeric twist could result from artefacts. In his opinion, the twist of rhabdoms and rhabdomeres described in earlier studies may be a result of destroying the mechanical stability of the retina either by inappropriate fixing solutions or by preparation of the retina without taking special care to avoid damage to the cornea and the basement membrane. However, in an electronmicroscopic study Smola and Wunderer (1981) found that the rhabdomere section of visual cells R2 of Calliphora erythrocephala presented by Ribi (1979, Fig. 2) was from a twisted rhabdomere. Later, it has been unequivocally proven that rhabdomere twist in several insects is not an artefact, but rather an in vivo structure (e.g. Wehner et al. 1975; Smola and Wunderer 1981; Wehner and Bernard 1993).

According to Smola and Tscharntke (1979), the functional significance of rhabdomere twist may be the improvement of the absorption of unpolarized light with the reduction of self-screening. Wehner and Bernard (1993) proposed another hypo thesis. In their opinion, the rhabdomeric twist reduces or


even abolishes polarization sensitivity in order to avoid the perception of polarization-induced false colours (see Chap. 33).

As the theoretical models of Moody and Parriss (1961) and Laughlin et al. (1975) indicated, every microvillus in a rhabdomere possesses an intrinsic dichroism that depends both on the shape of the microvillus and on the arrangement of the visual pigment molecules. Absorption in a single microvillus with aligned dipoles is higher than with randomly oriented dipoles. Another feature that greatly affects polarization sensitivity is the orientation of the microvilli over the entire length of the rhabdomere. In long, non-twisted rhabdomeres with aligned dipoles, the total absorption of unpolarized light is less than in corresponding rhabdomeres with randomly oriented dipoles, because of self-screening. When rhabdomeres having microvilli with aligned dipoles are twisted, self-screening is prevented. The result is that the high absorbance of the microvilli is effective all along the rhabdomere. Sensitivity to unpolarized light is increased, but the high polarization sensitivity that would appear in rhabdomeres with microvilli of such high dichroic absorption arranged in parallel is eliminated. Finally, we mention that Eguchi (1999) and Warrant et al. (1999) have found strongly curved microvilli in certain sphingid moths. Such microvilli structure could reduce or even abolish polarization sensitivity.

16.7.6 Ontogenetic Development of Photoreceptor Twist Outside the Dorsal Rim Area of the Insect Eye

In Apis mellifera, the twist of the photoreceptors outside the dorsal rim area of the eye originates early in the development of the retina (e.g. Wagner-Boller 1987). It occurs already in the preommatidia before the rhabdomeric microvilli have reached their differentiated state and before the rhabdom has been formed. In these early stages the full amount of twist about 1800 has already been reached, but the photoreceptors are much shorter and, hence, the twist rates much high er. During the subsequent elongation of the photoreceptors, the twist rate decreases to its final value of about 1 o/flm. The axon bundles leaving the ommatidia are twisted as well, but this axonal twist runs counter to the twist of the cell bodies within the retina (e.g. Sommer and Wehner 1975). The basement membrane penetrated by the axon bundles separates the upper ommatidial twist from the lower axonal twist. According to Wehner (1994), this membrane might well serve as the pivotal point about which the axons rotate early in development when the axon bundles are already surrounded by and tightly coupled to dense sheets of glial cells, and when the basement membrane is formed. This hypo thesis could explain that ommatidial and axonal twists are equal in amount, but opposite in sign.

16 Frorn Polarization Sensitivity to Polarization Vision 125

16.7.7 Characteristics of the Anatomically and Physiologically Specialized Polarization-Sensitive Dorsal Rim Area in Insect Eyes

The anatomically and physiologically specialized polarization-sensitive dorsal rim area (DRA) of the compound eyes in various insect species (Fig. 16.2) is generally thought to mediate orientation by means of the celestial polarization pattern. The DRA of insects share the following common features not found in other eye regions (Wehner 1994; Labhart and Meyer 1999):

- All rhabdomeric microvilli of a particular visual cell are oriented parallel to each other. Neither do the photoreceptors twist nor the microvilli wobble along the long axis of the receptor, as outside the DRA. Thus, PS is high in the DRA and weak or absent in the remainder of the eye.

- In the DRA each ommatidium has two sets of polarization-sensitive photoreceptors with orthogonal preferred E-vector directions. The hypothesis is that the outputs of such receptor pairs with orthogonal microvilli are compared antagonistically in order to enhance polarization sensitivity. Similar orthogonal microvilli arrangement occurs in the ventral eye region ofthe backswimmer Notonectaglauca (Schwind 1983b).

- All orthogonally polarization-sensitive photoreceptors in the DRA are monochromatic, Le. contain the same type of visual pigment.

- Within the DRA the microvillar directions of the polarization-sensitive photoreceptors rotate from front to back in a fan -shaped pattern (Figs. 17.5 and 17.9D).

- The DRAs always look contralaterally towards the zenith and/or around the zenith.

Sympelrum vulgalum (Odonala)

Grylllls campestrls

(OrlhOplera)

Melolontha melolon/ha

(Coleoptera)

u~ uv

@ffi@ Apis

mellifera (Hymenoplera)

Calaglyphi$ bicolor

(Hymenoplera)

Drosophila melanogasler

(Diplera)

Fig. 16.2. Schernatic drawings of the cross sections of rhabdoms within the dorsal rirn area (upperrow) and the dorsal region (lower row) in Sympetrum vulgatum (dragonfly), Gryllus campestris (field cricket), Melolontha melolontha (cockchafer), Apis mellifera (honeybee), Cataglyphis bicolor (desert ant) and Drosophila melanogaster (fruitfly). (After Labhart and Meyer 1999).


- The cornealienses in the DRA are penetrated by light-scattering pore canals (Apis mellifera: Meyer and Labhart 1981), or extensive cavities (Andrena, Ammophila: Aepli et al. 1985), or bubble-like inclusions (Melolontha melolontha: Labhart et al. 1992) surrounding the clear centres of the corneal facets. Their surfaces might lack the facet -array structure (Gryllus campestris: Labhart et al. 1984), or they might be underlain by unpigmented primary or secondary pigment cells ( Gryllus campestris: Labhart et al. 1984; Melolontha melolontha: Labhart et al. 1992). All these corneal specializations lead to highly enlarged visual fields of the underlying polarization-sensitive photoreceptors.

The presence of only two orthogonal microvilli directions in the DRA has the dis advantage of "neutral points" in the determination of E-vector direction, if the analysis happens locally and instantaneously (see Sect.16.3). However, this problem could be overcome by sequential scanning or by integrating information over the entire DRA where the microvillar axes vary systematically in a fan-shaped array.

According to Wehner (1994, pp. 131-132), it seems hard to believe that the geometrical similarity between the retinal fan-shaped microvilli array in the DRA (Fig. 17.1-17.5 and 17.9) and the celestial E-vector pattern originates from a directional fine-tuning occurring separately in each individual retinula cello In fact, the microvilli directions of receptors Rl and R5 in Apis mellifera and Cataglyphis bicolor follow approximately the course of the meridians of the compound eye. As these meridians converge at the dorsal pole of the eye, they automatically create a fanlike arrangement that resembles the Evector pattern of the sky, at least if the sun is near the horizon. The fan is optically converted, and the partial match between the pattern of preferred E-vectors in the eye and the E-vector pattern in the sky is achieved.

In order to find the reason for the enhanced PS in the DRA of crickets and desert ants, Nilsson et al. (1987) compared the optical properties and photoreceptor design of the DRA and other eye regions of Gryllus campestris and Cataglyphis bicolor. They found qualitatively that rhabdom waveguide effects do not affect the PS. Measurements of depolarization of light in the retina demonstrated that all parts of the eye retain the state of polarization very well through the full retinal depth. The factor limiting inherent PS of receptor cells must thus be the dichroic absorption in the rhabdomeres, wh ich is determined by the dichroic ratio of microvilli and the degree of microvillar alignment. A simple theoretical model of light propagation in a dichroic rhabdom revealed a strong influence of random microvillar misalignment on PS. The values of misalignment in crickets are ±9° in the DRA and ±18° in the dorsal eye region, while in desert ants ±3.2° in the DRA and ±17S in the dorsal eye region. The significantly weaker microvillar misalignment partly explains the higher PS of receptors in the DRA.


In several hymenopteran species (e.g. Pamphilius inanitus, Chrysis ignita, Pseudogenia carbonaria, Polistes gallicus, Vespa crabro, Pemphredon unicolor) belonging to various genera (e.g. Amblyteles, Ammophila, Andrena, Bombus, Crossocerus, Halictus, Paravespula, Prosopis, Psityrus, Rhophalum) Aepli et al. (1985) found similar anatomical specializations in the DRA as in Apis mellif era:

• The cornea is penetrated by pore canals or air cavities resulting in extreme porosity, which may affect the optics of the ommatidia by scattering the light falling into the eye. This could lead to wide visual fields of the DRA ommatidia. The cornea also contains such pore canals outside the DRA. These canals are much finer and structurally different from those in the DRA. They may not cause considerable light scattering due to their small size. Their function, if any, is unknown yet.

• Each retinula contains nine long receptor cells apposed to eight long ones in the adjacent dorsal area, and the rhabdom area is approximately doubled compared to that in other eye regions.

In the ants Camponotus ligniperda, Cataglyphis bicolor, Formica cunicularia, Manica rubida, Myrmecia gulosa, Nothomyrmecia macrops and Polyergus rufescens there are no corneal, but only retinal specializations, e.g. only two orthogonal microvillar directions in the DRA (Aepli et al. 1985). DRA was also found in the orthopteran species Bullacris membracoides and Pholidoptera griseaptera as well as in the wood cockroach Ectobius silvestris (Labhart and Meyer 1999).

The failure to find anatomically specialized DRA in some insect species -e.g. in the grasshopper genera Chorthippus and Tetrix as well as in the bushcricket genus Conocephalus (Burghause 1981 b), in the beetles Parastizopus armaticeps (Bisch 1999), Amphimallon solstitiale (June bug) and Melanophila accuminata, in the German cockroach Blattella germanica or in the backswimmer Notonecta glauca (Labhart and Meyer 1999) - suggests that orientation by means of skylight polarization is not omnipresent in insects.

Finally, we should emphasize that although the presence of a DRA may indicate that an insect species could be sensitive to the polarization of skylight, only electrophysiological recordings can provide evidence for the polarization sensitivity of the DRA photoreceptors, and exclusively behavioural experiments can prove that the insect indeed uses the celestial polarization pattern for orientation mediated by the DRA. Astrange example is the tenebrionid beetle Parastizopus armaticeps, in which DRA has not been found, but there is behavioural evidence for E-vector navigation (Bisch 1999). Other remarkable examples are certain tabanid flies, which have untwisted central rhabdomeres with well-aligned microvilli in the whole eye (Smith and Butler 1991).


16.7.8 Polarization-Sensitive Interneurons in Invertebrates

Although sensitivity to the direction of polarization has been reported for visual interneurons of certain beetles (Zolotov and Frantsevich 1977), this result turned out later to be due to experimental artefacts (Frantsevich and Zolotov 1979). However, there is evidence in the mantis shrimp Scylla serrata (Yamaguchi et al. 1976; Leggett 1976), the field cricket Gryllus campestris (Labhart 1988), the desert ant Cataglyphis bicolor (Labhart 2000), the cockroaches Periplaneta americana (Kelly and Mote 1990a) and Leucophaea maderae (Loesel and Homberg 2001) and the locust Schistocerca gregaria (Vitzthum et al. 2002) that the optic lobe contains interneurons sensitive to changes of E-vector direction. In Gryllus, Cataglyphis, Schistocerca, Periplaneta and Leucophaea these interneurons are polarization opponent, because they receive antagonistic inputs from two polarization-sensitive photoreceptors with orthogonal microvilli in each ommatidium of the DRA.

16.8 Polarization Sensitivity ofVertebrate Photoreceptors

The long axis of the photoreceptors of vertebrates is usually oriented toward the pupil of the eye (e.g. Laties et al. 1968). This orientation serves to maximize the amount of light guided to the outer segments, since off-axis light is ineffectively trapped by the photoreceptor inner segments. However, such orientation of the photoreceptors toward the pupil also minimizes the receptors' potential use of the dichroism inherent in the photopigments and outer segment disks for polarization sensitivity. Thus, the outer segments of vertebrate photoreceptors are usually axially illuminated (Laties and Enoch 1971) and have membrane disks, the surface of which is perpendicular to the longitudinal axis of the receptor (Fig. 16.3).

In the disk membrane the visual pigment moleeules can freely rotate with a relaxation time of about 20 }ls (Cone 1972) and translate (Poo and Cone 1974) due to the high fluidity of the liquid-crystal-like membrane with a viscosity as low as that oflight oils. This phenomenon results in the random orientation of the visual pigment moleeules from the axial direction of view. This is the reason why the whole receptor will be insensitive to the direction of linear polarization ofaxially incident light. Although the absorption of an individual pigment moleeule depends on the E-vector direction ofaxially incident linearly polarized light, the randomly oriented population of these molecules absorbs, on average, the same amount of light independently of the E-vector direction. However, because these pigment moleeules are aligned almost parallel to the disk membrane, and hence perpendicular to the longitudinal axis of the receptor, the receptor is sensitive to polarization if the incident light is transversally scattered (Fig. 31.3) or reflected (Fig. 28.3) onto the

16 From Polarization Sensitivity to Polarization Vision

Fig. 16.3. Schematic structure of a retinal rod photoreceptor in the eye of vertebrates with membrane disks, the surface of which is perpendicular to the long axis of the visual cell, and thus also to the axially incident light. The dipole axes of the visual pigment molecules, indicated by double-headed arrows, are randomly oriented in the disk membrane, but approximately parallel to its surface. Thus, an axially incident light hits pigment molecules with all possible directions of their dipole axes which are perpendicular to the direction oflight propagation. Consequently, the receptor is insensitive to the E-vector direction ofaxially incident linearly polarized light. If the light is incident perpendicularly to the long axis of the photoreceptor, it hits pigment molecules, the dipole axes of which are always parallel to the membrane surface. The consequence is that the receptor is sensitive to the E-vector direction of off-axis incident linearly polarized light. (After Wehner 1976).

rc tinal rod ccll

ax ially incident light

129

rod mcmbranc disks

ax ially

ii' tran vcrsally i ne idem light

dipo lc 3XC' of pigmcnt moleeules hit by the incidcnt light

outer segment (e.g. Schmidt 1935; Denton 1959; Waterman 1975) or if the outer segment has axially oriented disks (Fig. 28.2). Visual cell dichroism for transverse illumination was first demonstrated in frog rod outer segments by Schmidt (1935) and in other vertebrate rods by Denton (1959).

Several vertebrate species are polarization-sensitive. In the majority of these animals the mechanism of polarization sensitivity is yet unknown, although some hypotheses have been proposed to explain this ability of vertebrates. There are at least three ways how the polarization insensitivity of vertebrate photoreceptors to axially incident light might be overcome:

1. To lie some of the receptors on their sides. The photopigment molecules are then aligned in the plane of each disk membrane, which is parallel to the receptor's long axis. Similar arrangement was found in anchovy fish (see Chaps. 28.l.6 and 28.3.1), in which the cones have their disk membranes lying almost parallel to the long axes of the cells, not across them. Adjacent cones have orthogonal membrane stacks, forming a basis for polarization sensitivity (Fineran and NicoI1978).

2. To provide each receptor with some kind of polarizing filter in front of the photopigments.

3. To guide light transversally onto the outer segments of the photoreceptors by scattering or reflection.


In this third part of the volume, we will discuss what kinds of mechanism have been proposed to explain the polarization sensitivity of photoreceptors of different vertebrate species.

16.9 Polarization Sensitivity in Plants

Finally, we mention that animal visual systems are not the only place where photoreceptor molecules are oriented parallel and where the system is designed for a differential response to linear polarization. Haupt (1973), for example, described the photocontrol of chloroplast orientation in the filamentous green alga Mougeotia. Although the pigment that mediates this response is phytochrome, the system shares several features with the arthropod rhabdom:

1. The pigment molecules are located in the cortical cytoplasm, either in the plasma membrane or attached to its inner surface.

2. The transition moments of pigment molecules are not randomly disposed in the membrane, but have a fixed, in this case helical pattern around the surface of the cylindrical cells.

3. Photoconversion of the form F660 of pigment molecules absorbing in the red (660 nm) to the form F730 absorbing in the far-red (730 nm) is accompanied by a significant reorientation of the absorption vector. In the F660

form, the absorption vector lies in the tangent plane of the cylindrical cell, whereas in the F730 form it is tipped more nearly normal to the plasma membrane, becoming radially oriented with respect to the cylindrical cello There is a relatively rigid anchoring of the pigment molecules in a matrix either associated with, or identical to the plasmalemma.

Further details about polarization sensitivity in plants can be read in Haupt (1991) and Wada et al (1993), for instance.

17 Polarization Sensitivity in Terrestrial Insects

17.1 Honeybees

Karl von Frisch (1948, 1953, 1967) observed that honeybees Apis mellifera in form their hive mates ab out the existence and abundance of a food source or during swarming about a new nesting place farther than 100 m away from the hive by means of the so-called tail-wagging dance. They communicate the direction in which the food lies through a straight wagging walk. The bees utilize the solar azimuth as a reference direction and transpose the azimuth angle, which has to be kept while flying to the feeding-place with respect to the solar azimuth, onto the direction of gravity. During the dance, this is usually on the vertical comb in the dark hive. Then the vertical direction pointing upwards corresponds to the solar direction. On a horizontal surface the bees point the direction of their wagging walk directly toward the feeding-place if they see at least a clear patch of the sky.

In one of his seminal papers, Frisch (1949) showed that the dances of bees on a horizontal surface become disoriented in total darkness, but they are correct1y oriented as so on as a lamp emitting white light taken as the sun or a piece of blue sky are visible to the dancing bees. The dan ces on a vertical surface of a comb are correctly oriented in total darkness to gravity. When a patch of blue sky with an aperture of ab out 10-15° becomes visible, a deviation occurs in the direction of the dance in the sense of an immediate orientation toward the direction of the sun. This deviation is maximal (about 40-50°) at sunrise and sunset and minimal (about 0-5°) at solar culmination. If the blue skylight is deflected by a mirror, the orientation of the dancing bees is deflected accordingly. When the sky is totally overcast by very thick clouds, the bees become disoriented. From these findings, Frisch concluded that the position of the sun could be recognizable for the bees from the blue sky, even from a spot of clear sky.

After discussing these observations with the physicists Otto Kiepenheuer (Freiburg) and Hans Benndorf (Graz), who suggested the possibility that skylight polarization could be a cue for orientation in bees, Frisch drew the conclusion that bees can determine the solar position by means of the polarization pattern of the clear sky when the sun is not visible. This hypothesis has


been confirmed in such a way that he could influence the direction of the bees' dance on a horizontal surface by means of a linearly polarizing filter. When the polarizer was turned above the dancing bees, which could see the sky through the polarizer, the dan ce direction changed correspondingly. If the turning angle of the transmission axis of the polarizer was larger than 50-60°, the dancing bees became disoriented. They also oriented randomly if the degree of linear polarization of skylight was lower than about 11 %. From experiments in which bees danced on a horizontal comb under different colour filters, Frisch (1953, 1967) concluded that they may perceive the polarization in the UV and/or blue. The polarization sensitivity in honeybees was also demonstrated by Seletskaya (1956), who recorded evoked potentials in the optic lobe of bees as a response to rotation of the E-vector of totally linearly polarized light.

Frisch et al. (1960) have observed that the tail-wagging dan ce was sometimes correcdy oriented even if the sky was totally covered by clouds and the sun was not visible to humans. The bees oriented correcdy if the region of the overcast sky at and around the solar position was visible, but they disoriented if other cloudy celestial areas were visible only. They demonstrated that in photographs taken from cloudy skies in the UV when the bees oriented correctly, the sun was still discernible, since it was brighter by ab out 5 % than its surroundings. However, when the bees were disoriented under overcast skies with a very thick cloud layer, the sun was no longer discernible in the UV photographs. From this Frisch et al. concluded that bees are able to see the sun in cloudy ski es in the UV if the thickness of the cloud layer is below a threshold. However, the problem with this interpretation is that the direct sunlight is least intense in the UV. Furthermore, the UV light is strongly scattered due to the Rayleigh and Mie scattering and absorbed by the cloud particles, the consequence of which is that the cloudlight is the least intense in the UV (Fig. 10.2). Just this phenomenon resolves the so-called UV paradox of polarization sensitivity as we have shown in Chap. 10. Unfortunately, Frisch et al. (1960) did not test whether the sun was also slightly brighter in the visible part of the spectrum when their bees oriented correcdy, as one expects on the basis of the above argumentation. Perhaps the latter might have been the case, which could explain their observations.

Since these pioneering discoveries of Frisch (1948, 1949), many details of the polarization sensitivity and orientation in bees have been revealed. The results obtained before 1967 are summarized in the famous book of Frisch (1967) about the dance language and orientation of bees. In the remaining part of this chapter we give a survey of the most relevant results achieved after 1967.

Apis mellifera has three different types of photoreceptors, which are maximally sensitive at about 530-550, 440-450 and 340-350 nm (Autrum and Zwehl 1962). Zolotov and Frantsevich (1973) demonstrated that bees dancing on a horizontal comb can orient faultlessly even if the view of the clear

17 Polarization Sensitivity in Terrestrial Insects 133

sky is limited to a solid angle of 0.157 steradian (corresponding to about 2.5 % of the celestial hemisphere) in the band of maximum degree of linear polarization at 90° from the sun, which was about Pmax = 30-40 % at 460 nm. The accuracy of orientation decreased with decreasing diameter of this sky window. The increase of p to 100 % by covering the sky window with a linear polarizer with the same E-vector direction as seen through the window improved only slightly the precision of orientation. Zolotov and Frantsevich (1973) concluded that P of skylight above a threshold of 10-15% is not a dominating cue for orientation. Bees could orient correctly if the sky window was larger than a threshold of about 0.027-0.039 steradian. When the clear sky was exposed to the bees through four small windows with total area of 0.157 steradian, but each one being of threshold size 0.039 steradian, the accuracy of orientation diminished immediately with the increase of spacing between these tiny windows. Edrich and Helversen (1976) showed that a visual angle of about 1 ° of totally linearly polarized UV light from the zenith enables bees to orient their dance correctly on a horizontal combo Dances under a very bright totally polarized white light were still well oriented even if the field of view was only os.

With electrophysiological recordings from single retinula cells Menzel and Snyder (1974) found that the green (530-550 nm) receptors of Apis mellifera are only weakly sensitive to polarization with PS < 2.4. The PS of blue (420-460 nm) receptors varied between l.27 and l.43. They found two different types of UV receptors. The more frequent UV cells have a distinct spectral sensitivity at longer wavelengths over 450 nm and low polarization sensitivity PS < 1.4, while the rare UV receptors with no sensitivity to longer wavelengths have PS = 5 on average with a maximum of 9. They proposed that the short retinula cell R9 in the proximal third of the fused rhabdom may be the highly polarization-sensitive UV receptor type. They suggested that the E-vector detection in honeybees may happen in the dorsal eye region in such a way that the signals of cells R9 in neighbouring ommatidia with strongly overlapping visual fields and/or ommatidia in the two eyes sharing the same field of view are compared.

Helversen and Edrich (1974) observed bees dancing on a horizontal comb hidden from the sun and sky, but exposed to monochromatic totally linearly polarized light of variable wavelength and intensity within a 17° circular window at the zenith. The bees oriented correctly as if they could see a corresponding area ofblue sky at the zenith, when the wavelength of polarized light was shorter than 430 nm. The accuracy of their orientation was improved with increasing intensity of polarized light at a given wavelength. Their orientation was bimodal, since they could not discriminate between the solar and antisolar meridian due to the artificial homogeneous polarized light field at the zenith. When the E-vector direction of polarized light was turned by a given angle, the direction of orientation changed accordingly. From these experiments Helversen and Edrich could establish that bees are maximally sensitive

l34 Part II1: Polarized Light in Animal Vision

to polarization at ab out 345 nm, and thus the UV receptors mediate polarization sensitivity.

The electron-microscopic investigations by Schinz (1975) revealed that the ommatidia in the dorsal rim area (DRA) of the compound eye are anatomically specialized and distinct from ommatidia in other eye regions. In the DRA nine visual cells contribute microvilli to the rhabdom over its fulliength, which is much shorter than outside the DRA. Within the rhabdoms in the DRA Schinz found three different microvilli directions, and he mentioned that this seems to fit the simultaneous method of E-vector detection, which requires comparisons between the signals of polarization-sensitive photoreceptors with three different microvilli directions (Kirschfeld 1972, 1973b). Although later anatomical studies proved that in the DRA of bees there are only two orthogonal microvilli directions, the merit of Schinz (1975) is that he suggested first that the anatomically specialized DRA may serve for detection of skylight polarization.

In an anatomical study, Wehner et al. (1975) showed that, apart from the DRA of the eye of honeybees where the retinula cells are non-twisted and each ommatidium contains nine long visual cells Rl-R9, in the dorsal area the retinulae are twisted and each ommatidium contains eight long cells RI-R8 while the short cell R9 occurs only proximally. In the distal part of the fused rhabdom, the microvilli of RI-R4 are vertical while those of R5-R8 are horizontal. R9 and the two overlying long twisted cells Rl and R2 are UV-sensitive. There are two randomly distributed types of twisted rhabdoms with mirror-imaged geometry. One type twists clockwise, while the other twists counter-clockwise with a rate of ab out 10/~m. The total twist angle is about 180° in the long RI, R2 UV receptors and 40° in the short R9 UV cells. The twist of the ommatidia is already present in the prepupal stage (WagnerBoIler 1987). On the basis of theoretical considerations, Wehner et al. (1975) hypothesized that the polarization sensitivity of the long twisted UV cells RI and R2 may be significantly reduced, while PS of the short UV cells R9 may remain relatively high and the preferred E-vector directions of the two types of R9 cells, twisted clockwise and counter-clockwise, may differ by about 36°. They suggested that unambiguous detection of single E-vector directions in the UV by the dorsal eye region could be mediated by two twist-types of the R9 receptors as two polarization-sensitive channels in two adjacent ommatidia, and one of the cells Rl and R2 as a polarization-insensitive channel. This model lost its validity when it turned out that E-vector detection is mediated exclusively by the DRA.

On the basis of anatomical studies, Sommer and Wehner (1975) proposed that in each ommatidium of the DRA the two opposite receptors RI and R5 with parallel microvilli are UV-sensitive like the cell R9 with microvilli perpendicular to those of Rl and R5. This was supported by Wehner and Bernard (1980), who suggested that in the dorsal and ventral eye regions R2 and R6 may be blue while R3, R4, R7 and R8 are green receptors. They investigated the


pupillary responses of the eye to linearly polarized light from only a single spectral type of photoreceptor. They found that the UV pupil exhibited high polarization sensitivity PSuv = 8, the blue pupil had medium PSB = 4 and the green pupil very weak PSG < 1.3 both in the dorsal and ventral eye regions.

The microvillar direction of retinula cells Rl and R5 changes in a fanlike way in the DRA of honeybees (Figs. 17.1 and 17.2). With intracellular recordings Labhart (1980) studied the spectral, polarization and angular sensitivities of photoreceptors in the DRA of Apis mellifera. He found that each ommatidium in the DRA contains nine non-twisted long retinula cells RI-R9. These ommatidia look upwards in directions at and around the zenith during free flight. The UV cells with a secondary sensitivity peak at 540 nm have average PS ranging from 3.8 to 5.6, while pure UV receptors possess an average PS =

dorsal rirn area - - -_ - / f ~ I I '\ \ \, " , , ..... / .... -/ /' / -: ~ /,' :,', \ I'"~ \ ~ ,'~ :::: :::~..:- -

",;"'/"'/,//1' ,',' 1'1\\\ \ ,,' .... , ..... _-..... .:-_ / / '" / / / I I I, I, I I \ \ " I \ •• ' \ , ..... " : • - ..........

/ / /1·. I I , • •• '.' • •••• • ••• \ •••• : ~ - .--// / . . .. .... .... .. \ /////..... ...... . ... '

/ /. .. ....... .. . ./ .. . .. .. . . \ ... , ..

• • • • • • • • • • • • • • • • • • Fig.17.1. The fan-shaped array of rhabdomeres in the dorsal rim area (DRA) of the compound eye in the honeybee Apis mellifera. The bars indicate the directions of the parallel microvilli of retinula cells Rl and R5. The DRA is shaded by grey. Black dots represent the positions of twisted retinula cells outside the DRA. (After Wehner 1982).

Fig.17.2. Visual field of the left eye of Apis mellifera displayed by grey on a sphere, the upper and lower half of which represents the celestial and terrestrial hemisphere, respectively. The ommatidia in the dorsal rim area (DRA) view the dark grey shaded region of the sky. The DRA of the left eye shares the visual field of the remainder of the dorsal region of the right eye, and vice versa. (After Rossel and Wehner 1984b).

136 Part II1: Polarized Light in Animal Vision

13±4.5. The overall average of PS of UV cells is 6.6. Outside the DRA the UV cells have PS < 2. In the DRA the green cells possess an average PS = 1.8±0.3, while in other eye regions PS = l.3±OA. In the DRA there are two populations of UV receptors with orthogonal preferred E-vector directions. The photoreceptors in the DRA have wide and strongly overlapping receptive fields with aperture angles of about 40-60°, with high sensitivity in a narrow central spot and low sensitivity at the periphery. Outside the DRA the receptors have fields of view below 10°. Labhart concluded that E-vector detection is performed by the UV receptors of the DRA.

In an anatomical study Meyer and Labhart (1981) found that the corneal lenses in the DRA are clear only centrally, but at the periphery they are grey and cloudy due to numerous fine rugged-walled pore canals penetrating the cornea from the inside and ending a few micrometres below its outer surface. The wide visual fields of receptors in the DRA may be due to light scattering by these canals. This way, light hitting the facets offaxis can also reach the rhabdom, without considerable depolarization as manifested by the PS measurements of Labhart (1980) using off-axial stimuli. The central sensitivity peak of the photoreceptors in the DRA is due to the clear centre of each corneal facet, whereas the flat surround of this peak of the angular sensitivity function is a result of light scattering on the pore canals. Similar light scattering pore canals have been found by Aepli et al. (1985) in other hymenopteran species including yellow jackets (Paravespula, Vespoidea), sand wasps (Ammophila, Sphecoidea) and sweat bees (Halictus, Apoidea).

In a thorough histological study, Wehner and Meyer (1981) proved that the rhabdoms in the dorsal retina outside the DRA of the eyes in Apis mellifera and the Indian dwarf honeybee Apis florea really twist along their lengths in vivo. This twist has been earlier questioned and held as a preparative or fixational artefact by Ribi (1979).

Brines and Gould (1979) observed the dance ofbees on a horizontal comb and provided an artificiallight source as the only visual cue for orientation. They systematically changed the elevation, aperture angle, intensity, wavelength distribution, degree of linear polarization p and E-vector direction of the light source. They found that bees use the extent of a visual stimulus, its relative UV content and p to distinguish whether the observed cue is the sun or apart of the sky. Depending on p, a light source with a visual angle below about 18-25° and UV content lower than about 20-40 % is considered as the sun. A light source with p > 10% and relative UV content over ab out 20-40 % is taken as apart of the sky depending on its visual angle. The elevation is important only when the light source is at the zenith. This rule results in certain sky patches being identified as sun. This "sun-sky rule" corresponds roughly to the characteristics of natural sunlight and skylight. Sunlight contains only ab out 8 % UV, while the UV content of skylight ranges from 20 to 35 % (Hess 1939). Sunlight is unpolarized and the skylight from the immediate surroundings of the sun is only very weakly polarized (p < 5 %), while p of


light from the rest of a dear sky is higher than about 15 % in the UV. If the light source is judged as apart of the blue sky, bees identify by means of the Evector direction which part of the sky they see, and by this they orient their dances to the food source. Brines and Gould (1979) surmised that bees might have three further simple rules during this process, and conduded that they may not use the Rayleigh scattering relationships at all. However, later more precise behavioural studies ofWehner and co-workers (see below) refuted the validity of these rules.

In behavioural experiments, Wehner and Strass er (1985) found that the photoreceptors in the DRA of at least one eye are necessary and sufficient for the detection of skylight polarization and for derivation of compass information from the celestial E-vector pattern. In these experiments, the orientation of bees dancing on a horizontal comb was recorded. They were exposed to a patch of dear sky or a beam of artificially polarized light, and different eye regions were selectively painted over. They oriented bimodally if a small spot of blue sky at the zenith was visible. Only when this zenith patch was increased, did they unambiguously select the proper compass course. When the DRA was painted over, they were unable to derive compass information from a small patch of polarized light at the zenith, even if other dorsal ommatidia outside the DRA could see the full zenith patch of sky. When the zenith patch of blue sky was increased, bees were well oriented even if their DRA was ocduded. In this case they did not rely on skylight polarization, but instead on intensity and spectral gradients. When several dorsal ommatidia dosest to, but outside the DRA were painted over, they could orient correctly under a small patch of blue sky at the zenith. In this case only the ommatidia in the DRA were stimulated by polarized skylight. From this it was conduded that the DRA is not only necessary, but also sufficient for E-vector detection in the zenith. The accuracy of the orientation decreased with decreasing number of non-painted ommatidia in the DRA. When at least ab out 16-20 ommatidia in the DRA were left open in one eye, the bees could correctly orient by means of the blue sky patch at the zenith. In these experiments, in some cases also the ocelli were ocduded. The orientation did not depend on whether or not the ocelli were painted over. Thus, the ocelli are not necessary for the detection of skylight polarization. Wehner and Strasser (1985) called the specialized DRA introduced first by Labhart (1980) as the "POL-area".

Edrich and Helversen (1987) demonstrated that Apis mellifica on a horizontal comb can correct1y orient their dances in a homogeneous field (4.7°) of linearly polarized UV light from the zenith even if p is modulated from 0 to 100 % at frequencies between 0.05 and 25 Hz at constant E-vector direction and at an intensity fluctuating by about ±19 % sinusoidally from an average value. From this they suggested that bees analyse the E-vector direction simultaneously with three or more receptors with overlapping fields of view and different preferred E-vector directions in the DRA, rather than with one polarization-sensitive receptor analysing sequentially by scanning the sky.


These findings seem to contradict the results of later experiments of Rossel and Wehner (1986). A possible resolution of this discrepancy is discussed below.

In behavioural experiments, Rossel et al. (1978) found that if a honeybee was trained to fly in a certain direction to a food source, the direction of its wagging dance on a horizontal comb pointed directly towards the goal provided that it was able to see the sky. However, if it was only allowed to view a single E-vector direction in a patch (angular diameter 5 or 10°) of the sky or in a spot (5 or 10°) of artificial totally linearly polarized UV light, it oriented bimodally. One of the preferred directions pointed closely, but not exactly towards the goal. By systematically chan ging the elevation 0 and angle of polarization a clockwise from the local meridian of the stimulating patch at a given solar elevation Os at noon, when the change rate of Os is the lowest, Rossel et al. could measure the directional errors as functions of a and 0, and reconstruct at which azimuth angle ep from the solar meridian a given a presented at elevation ° is expected. If a horizontal E-vector was presented, the bees oriented unimodally in the correct direction, and they always interpreted horizontal E-vectors as lying along the antisolar meridian. If a vertical E-vector was presented, they oriented bimodally in such a way that they always expected the vertical E-vector to be at ±90° from the antisolar meridian, even when vertical E-vectors did not occur at the given elevation in the sky. Whatever E-vector direction was presented at a given 0, the bees oriented bimodally and consistently, even if the E-vector direction did not naturally occur at that elevation. One of two preferred directions, pointing always toward the antisolar half of the sky where p is high er, was chosen much more frequently.

From these experiments, Rossel et al. (1978) deduced that bees rely on a rather generalized and simple internal representation ("template") of the celestial E-vector pattern, which can be described as follows: abee(ep) = ep - 90° + 9°·sin2ep (Fig. 17.3). Hence, the internal map abee( ep) of angle of polarization does not depend on the elevation ° and the solar elevation Os' The bee expects a constant angle of polarization abee( ep) along a given meridian with azimuth ep irrespective of ° and Os. The internal celestial E-vector map abee(ep) is roughly similar to the E-vector distribution ask/ep,O,Os) of the clear sky near the zenith (0::::: 90°), especially when the sun is ne ar the horizon (Os::::: 0°). This internal representation is called byWehner (1989) a "matched filter": matched to the spatial properties of the navigational problem to be solved.

Rossel and Wehner (1982) suggested that bees use their internal celestial Evector map invariably throughout the day, ignoring the spatial details of the celestial E-vector pattern and instead reacting to the band of maximal p at 90° from the sun. They demonstrated that the bee's internal celestial E-vector map approximately reflects the me an E-vector distribution of the maximally polarized band of the sky during the day at elevations larger than ab out 30°. This internal E-vector map of sky, confining the antisolar half of the sky, co-


Fig.17.3. The bee's internal representation of the sky consisting of the sun emitting unpolarized light dominated by long wavelengths, and the antisolar half of the sky (shaded by grey) emitting linearly polarized light dominated by short wavelengths. The bars indicate the bee's internal representation of the celestial E-vector pattern, which depends only on the azimuth angle, but is independent of the elevation and the solar elevation. (After Rossel and Wehner 1984b).

139

rotates with the solar azimuth. All E-vectors in the skyare interpreted as lying within the antisolar half of the celestial hemisphere (Fig. 17.3).

Rossel and Wehner (I984a) exposed bees dancing on a horizontal comb to large horizontal stripes of clear skies at high solar elevations ranging from 58 to 63°. From the directional mistakes made in these situations, they suggested that all E-vectors of skylight are equally important regardless of p if P > 10 %. They hypothesized that under natural conditions, when orientation errors do not occur, the bee could orient in the following way: it ratates around its vertical body axis, during which it may perceive the celestial polarization in the form of temporal brightness and/or colour modulations. Colour modulations could occur if the polarization-sensitive UV receptors would take part in the colour coding system, where their response would be compared with those of polarization-insensitive or only weakly polarization-sensitive blue and green receptors. Fram the maxima and/or minima of these brightness and/or colour modulations, the bee may be able to derive the orientation of the solar and/or antisolar meridian and subsequently the direction of the intended compass course. They also assumed that the internal celestial E-vector map may be determined by the fan-shaped arrangement of the orthogonal microvilli directions of polarization-sensitive receptors in the DRA (Fig. 17.1). In their opinion, celestial polarization is not perceived as an extra quality of skylight, but simply as the modulation of brightness and/or colour while the bee rotates. Hence, bees might not have true polarization vision or polarization vision at all.

The hypothesis that E-vector perception could be based on a colour discrimination mechanism was first suggested by Glas (1975). According to hirn, the celestial polarization pattern may be converted into colour pattern by means of polarization-sensitive UV receptors and polarization-blind blue and green receptors. In his experiments the bees dancing on a horizontal comb were exposed to different overhead artificial unpolarized colour patterns in which long- and short-wavelength colours blended in a gradual tran-


sition from one to the other while the geomagnetic field vector was changed to vertical. Under these conditions they oriented bimodally as if the colour patterns provided a clue to the sun's bearing. Glas interpreted the observed orientation in such a way that the colour patterns simulated the polarizationinduced false colour patterns. Although these results can be explained also such that the colour patterns mimicked the colour gradients of the solar and antisolar halves of the sky (see Rossel and Wehner 1984b), the merit of the experiments of Glas is two-fold. On the one hand, his experiments demonstrated that bees can derive compass information from spectral gradients of unpolarized light. On the other hand, Glas (1975) was the first to propose that polarization-induced false colours can be perceived by a colour-sensitive visual system if it is also sensitive to linear polarization (see Chap. 33).

Rossel and Wehner (1984b) stimulated bees dancing on a horizontal comb with unpolarized or linearly polarized light. E-vector information was prevented by painting over the DRA and/or the ocelli. They showed that the dorsal eye region outside the DRA is sufficient for mediating spectral and intensity information for orientation independently of whether the ocelli are or are not occluded. Spectral cues are used to discriminate between sun and sky. A long-wavelength stimulus is taken as the sun. A short-wavelength stimulus is expected to lie anywhere within the antisolar half of the sky. More detailed directional information is obtained only when the patch of sky provides additional E-vector information. This corresponds to the fact that skylight is increasingly dominated by short wavelengths with increasing angular distance from the sun. However, spectral and intensity cu es alone do not provide precise compass information except when a full celestial colour and intensity gradient is available including the solar and antisolar meridian. Since the celestial spectral and intensity gradients are symmetrical to the solar-antisolar meridian if the sky is clear, the latter symmetry axis can be determined also by means of the colour and intensity pattern of the full blue sky. Wehner and Rossel (1985) hypothesized that bees are not able to determine E-vector orientations locally and instantaneously in any particular part of the sky. Instead, they seem to use the celestial polarization merely to determine the direction of the solar or antisolar meridian, and seem to do so by means of some kind of global scanning strategy.

Rossel and Wehner (1986) hypothesized that for the perception of E-vector directions of skylight the bee sweeps its retinal analysers, the polarizationsensitive UV photoreceptors in the DRA, across the celestial E-vector pattern by performing a full turn around its vertical axis and perceives wh ich part of the retina is maximally active during the sweep, or compares the outputs of receptors with perpendicularly oriented microvilli. However, the latter comparison is not necessary, because only one analyser is sufficient per ommatidium, if polarization is analysed by the scanning method. Of course, for this task at least one continuous ring of analysers in the DRA is necessary. This scanning model predicts that a patch of unpolarized light arranged to evoke a


peak output within a particular part of the retina will cause the bee to orient as if the light were polarized in the appropriate direction.

To test this prediction of the scanning hypo thesis, Rossel and Wehner (1986) studied the orientation of trained bees dancing on a horizontal comb during which an artificiallight source (aperture: 10°, elevation: 60°) provided the only directional cue by which they could orient their wagging dance. When given a single patch of polarized light in such a situation, the orientation of dance varies with the E-vector direction of skylight. Rossel and Wehner have mimicked this behaviour by means of a beam of unpolarized UV light, the intensity of which was temporally modulated according to the orientation of the body long axis, such that the intensity of the beam peaked when it stimulated a predetermined part of the retina. Maximal and minimal intensities were presented when the body long axis was aligned either parallel or perpendicularly to a given reference direction on the horizontal combo The eyes were covered with a small piece of linearly polarizing filter with transmission axis parallel to the transverse axis of the eyes. It was assumed that due to this polarizer one population of the UV receptors with a given microvilli direction might have been stimulated more strongly by the unpolarized intensity-modulated UV light than the other receptor population with orthogonal microvilli.

As a control, some bees were equipped with a polarizer with transmission axis at 45° from the transverse axis of the eyes. In this case, it was assumed that the two receptor populations with orthogonal microvilli might have been stimulated approximately equally. The light was modulated such that the intensity was maximal when it fell on one of two different retinal positions, which, according to the scanning hypothesis, contain detectors for one of two oblique E-vectors. Under these conditions, with the transmission axis of the polarizer parallel to the transverse axis of the eyes, the dan ces oriented as if the bees were presented with one of the two oblique E-vector directions. In contrast, with the polarizer oriented at 45° from the transverse axis of the eyes, this behaviour disappeared and the bees behaved as if they were stimulated by a patch of unpolarized, non-modulated UV light. The observation that the intensity-modulated unpolarized UV light could not simulate oblique E-vectors when the eyes were not covered by the polarizer suggests that the output signals of two photoreceptors with orthogonal microvilli are compared. If only one of the two receptor populations with orthogonal microvilli played a role in the E-vector compass, the intensity-modulated UV light should have elicited the same behaviour as an oblique E-vector direction. These results were believed by Rossel and Wehner (1986) to be experimental evidence for the scanning hypothesis. They also suggested that this scanning method may also be gene rally used by other invertebrates which detect skylight polarization for orientation.

The results of the experiments of Rossel and Wehner (1986) seem to contradict the results and interpretation of the similarly elegant and convincing


experiments of Edrich and Helversen (1987), who proved that the dan ce orientation is not affected by modulation of p of the stimulus up to a modulation frequency of 25 Hz. The latter was believed to be experimental evidence for the simultaneous E-vector analysis. This contradiction between the results and interpretations of these two crucial experiments can be resolved only by the assumption that the scanning E-vector compass system does not need pas information for orientation. This seems to be so, because according to the behavioural experiments of Wehner and co-workers, each E-vector direction of skylight is equally important for the celestial E-vector compass, irrespective of p. p of skylight is not used as a compass cue by bees, and it is important merely in the sense that it must be high er than the polarization sensitivity threshold of about 10 %. Thus, the experiment ofEdrich and Helversen (1987) could only prove that the E-vector compass is insensitive to the modulation of p, rather than prove that bees detect the E-vector direction by a simultaneous method or by a sequential method, the probing frequency of which is higher than 25 Hz.

According to the scanning theory of orientation by means of the celestial polarization pattern proposed by Rossel et al. (1978) and Rossel and Wehner (1982, 1984a, 1986), orientation can be erroneous and lead to the wrong course of a recruited bee in search of the foraging site whenever only a small patch of the blue sky is visible. The orientation mistakes occurred in the experiments of Rossel and Wehner, because the sky seen during foraging and during the test was different. During foraging the bee experienced that the feeding station was in a given direction relative to the solar azimuth. In the test, this direction was communicated with the wagging dan ce relative to the compass reference defined by the E-vector direction in the patch of sky seen during the test. The consequence was an orientation error.

Kirschfeld (1988a) pointed out that even with such a wagging direction not pointing to the feeding station, recruited bees would still be able to follow the correct course, since once on the wing, their compass reference on the blue sky is the sun again. If they immediately adopt the solar direction as a new reference direction, they can find the food source correctly. Hence, orientation under natural conditions is not erroneous if the compass reference is variable in time, but equally defined for both scout bees and recruits. The concept of a variable compass reference was also considered by Frisch (1967). This concept can only be applicable if the sky parameters do not change considerably during the 20-120 s period when the directional information is transferred from the scout to a recruited bee.

Kirschfeld (1988b) has shown that the scanning strategy to obtain compass information from the celestial polarization pattern while the bee scans the sky by rotating its field of view as suggested by Rossel et al. (1978) and Rossel and Wehner (1982, 1984a, 1986) can work only in apart of the sky dose to the zenith with zenith angles smaller than about 200 • The ommatidia in the DRA cannot detect the polarization of light from regions of the sky with zenith


angles over 20°, because they do not point toward these celestial areas in spite of their wide (40-60°) visual fields. However, several experiments dearly demonstrated that bees are capable of perceiving the E-vector direction of light coming from a small celestial patch, not only if the patch is dose to the zenith, but also if it is as low as 30° above the horizon (e.g. Frisch 1967; Rossel et al. 1978). Thus, bees need a different strategy for regions of the sky far from the zenith.According to Kirschfeld (1988b), it could also be that they are capable of analysing the E-vector direction with the ommatidia in the DRA using a simultaneous method as proposed earlier (Kirschfeld, 1972, 1973b).

Honeybees perceive also the geomagnetic field and use it for compass orientation (Frisch, 1953, 1967). In behavioural experiments, Leucht and Martin (1990) investigated the interactions between E-vector orientation and weak steady magnetic fields in Apis mellifica. The orientation of bees dancing on a horizontal comb was studied. They could see a circular patch (20-40°) of the dear sky at the zenith under three magnetic conditions: natural magnetic field, magnetic field with reversed polarity but the same intensity as the natural field, magnetic field with intensity less than 4 % of the natural field. Leucht and Martin found that bees are ahle to detect the polarity of the magnetic field, and dancing bees with a restricted view of the E-vector pattern of the zenith sky are able to use the magnetic field as a reference for their orientation based on skylight polarization. However, they exhibit pronounced response latency to the change of the magnetic field.

17.2 Flies

17.2.1 Muscid Flies

In pioneering, but qualitative behavioural experiments, Wellington (1953) observed that the muscid fly Sarcophaga aldrichi changed the orientation of its body long axis when a linear polarizer was turned above it. He found that long and straight flights of this fly seem to require skylight polarization from a doudless zenith area.

17.2.2 Rhabdomeric Twist in Flies

Smola and Tscharntke (1979) showed by means of electron-microscopy that the rhabdomeres of the peripheral visual cells Rl-R6 in the middle eye region helow the eye's equator of male Calliphora erythrocephala and Drosophila melanogaster are twisted along their longitudinal axes. In the proximal and distal regions of the rhabdomeres the twist is slight (-0.52°/flm) or absent, whereas in the middle the twist rate is high (-2.4°/flm). The twist of the rhab-


domeres of Rl-R3 is congruent and codirectional, and the rhabdomere structure of R4-R6 is the mirror image of that of RI-R3. It was shown that the dichroic absorption ratio of a single microvillus of RI-R6 must be high er than 2, and thus the absorbing dipoles of the visual pigment molecules must be more or less parallel to the microvilli axes (see Chap.16.7). The twist of the rhabdomeres Rl-R6 prevents self-screening despite high microvillar absorption. According to Smola and Tscharntke, rhabdomeres RI-R6 are especially effective in absorbing unpolarized light due to their twist. In an electronmicroscopic study, Smola and Wunderer (1981) found that the rhabdomeres of the visual cells of Musca domestica also twist in the mann er typical of Calliphora erythrocephala and Drosophila melanogaster. Williams (1980) has noted that the rhabdomeres in the eye of a tipulid fly (Ptilogyna spectabilis) also twist.

Using electron-microscopy, Altner and Burkhardt (1981) investigated the fine structure of the ommatidia and the occurrence of twist in the open rhabdome in the dorsal eye of male March flies Bibio marci. They found that in the left eye the peripheral rhabdomeres R3, R5 and R6 twist clockwise along their longitudinal axes, while rhabdomeres Rl, R2, R4 and R8 twist counter-clockwise. Opposite twisting occurs in the right eye. The twist rate varies along the length ofthe rhabdomeres. The maximal twist rate is 2-5°/jlm in RI-R6 and even higher in the central rhabdomere R8. In a proximal region, the direction of twist is reversed, but the initial microvillar orientation is not re-established. The short central rhabdomere R7 does not twist. Rhabdomeric twist was not found in the ventral eye region.

17.2.3 Musca domestica, Calliphora erythrocephala, Calliphora stygia and Phaenicia sericata

Seitz (1969) performed polarization optical studies on the eye of the blowfly Calliphora erythrocephala, and Strausfeld and Wunderer (1985) studied the optic lobe projections of photoreceptors in the dorsal rim area (DRA) of Calliphora. With electrophysiological recordings and optomotor experiments, McCann and Arnett (1972) studied the spectral and polarization sensitivity of wild-type adult Musca domestica, Calliphora erythrocephala and Phaenicia sericata. They found that the peripheral retinula cells RI-R6 as well as the sustaining and on-off units in the first optic ganglion, belonging completely to the RI-R6 subsystem, have two approximately equal spectral sensitivity maxima at 350 and 480 nm. Both central retinula cells R7 and R8 are maximally sensitive at 465 nm, and it was suspected that they are responsible for detection of polarization, because the microvilli of their rhabdomeres are oriented orthogonally.Although all retinula cells RI-R6 have different preferred E-vector directions, the sustaining and on-off units in the first optic ganglion are insensitive to polarization. This verifies the indication based on the neural


anatomy of the first optic ganglion that the activity of retinula cens Rl-R6 is indiscriminately integrated in the first optic ganglion. Motion detection and optomotor responses are insensitive to the E-vector for long spatial wavelength stimuli in an three fly species, while they are polarization-sensitive for short spatial wavelengths in Musca domestica, for which the central subsystem R7/R8 may be responsible.

In an electron -microscopic study Wada (197 4b) found that Calliphora erythrocephala also has a specialized DRA. Järvilehto and Moring (1974) measured the polarization sensitivity of individual retinula cens and neurons in the compound eye of wild-type adult Calliphora erythrocephala by means of intracellular recordings. They found both polarization-sensitive and -insensitive visual cells among the peripheral photoreceptors RI-R6. In the lamina ganglionaris, all fibres belonging to Rl-R6 were insensitive to polarization. On the soma of the central retinula cell R7 no polarization sensitivity was found. The second order monopolar neurons LI and L2 in the lamina, which are thought to be in contact with each of Rl-R6, were also polarization -insensitive.

By means of intracellular recording, Hardie et al. (1979) studied the spectral and polarization sensitivity of the central retinula cells R7 and R8 in the middle eye region of wild-type Calliphora stygia, white-eyed mutants Calliphora erythrocephala and Musca domestica. Cells R7 had spectral sensitivity with a single peak at either 340 nm (with PS = 1.5-2.3) or 360 nm (with PS = 2.3-5.2). Cells R8 had a major peak at 520 nm and a small secondary peak at 360 nm (with PS = 1.7-1.8).

In similar experiments, Hardie (1984) investigated the properties of the central photoreceptors R7 and R8 in the specialized ommatidia in the DRA of female adult wild-type Musca domestica and Calliphora erythrocephala. He found that both R7 and R8 are UV receptors with high polarization sensitivity (6< PS < 19) and sensitivity maximum at 330-350 nm. The preferred Evector direction of R7 is approximately perpendicular to that of R8. The rhabdomeres of R7 and R8 are greatly enlarged in cross section and reduced in length in the DRA ommatidia compared to R7 and R8 in other eye regions. In contrast to other rhabdomeres in the eye of Calliphora erythrocephala, the R7 and R8 rhabdomeres in the DRA do not twist, the microvilli of R7 are perpendicular to those of R8 and they are specialized for detection of skylight polarization (Wunderer and Smola 1982). Hardie (1984) suggested that there may be reciprocal inhibitory interactions between R7 and R8.

In a similar open-Ioop behavioural experiment as Brunner and Labhart (1987) performed with crickets, Philipsborn and Labhart (1990) found that tethered adult male and female Musca domestica walking on an air-suspended Styrofoam ball oriented spontaneously to the E-vector of wide (87°) overhead field of totally linearly polarized white and UV light. A slowly rotating E-vector induced periodic changes in the strength and direction of turning tendency. Periods of left and right turns alternated and were separated by moments of no turning. This behaviour is similar to the E-vector response


observed in crickets by Brunner and Labhart (1987). The flies preferred Evectors approximately perpendicular to their body long axis and avoided Evectors directed nearly parallel to it.

From the existence of the turning response to the rotating E-vector in both white and UV light and from its complete absence in yellow light, Philipsborn and Labhart (1990) concluded that the specialized DRA with its central R7 and R8 UV receptors with orthogonal microvilli might mediate polarization sensitivity in Musca domestica. Note that this is in co nt rast to the findings of Wolf et al. (1980) that in Drosophila melanogaster receptors RI-R6 may mediate polarization sensitivity. In control experiments by Philipsborn and Labhart (1990), the rotating polarizer was replaced either by a diffuser or by an intensity pattern. The latter was produced by the diffuser overlain with a pattern of neutral density films simulating the intensity distribution, which can be observed when the polarizer is viewed from the position of the walking fly. A turning response was elicited neither by the rotating diffuser nor by the rotating intensity pattern. The periodic turning response of the flies under the rotating polarizer was therefore not due to possible intensity patterns induced, for example, by selective reflection of polarized light from the surrounding walls and bottom of the experimental chamber.

Considering the fanlike arrangement of the microvillar orientations in photoreceptors R7 and R8 in the DRA of the eye of Calliphora erythrocephala (Wunderer and Smola 1982), Philipsborn and Labhart (1990) suggested that the polarization sensitivity in Musca domestica might playa role in the stabilization of flight course. Temporal variations in the excitation pattern of cells R7 and R8 in the DRA could be used to correct deviations from a straight flight path.

The DRA ommatidia in the flies Musca domestica and Calliphora erythrocephala have their counterparts in honeybees and desert ants, for example. Despite these striking similarities at the retinal level between honeybees, desert ants and flies, evidence for polarotactic behaviour in flies is limited (Wolf et al. 1980; Philipsborn and Labhart 1990), and whether flies can make similar use of their DRA photoreceptors must be investigated in further behavioural experiments.

17.2.4 Drosophila melanogaster

Stephens et al. (1953) studied the spontaneous orientation of the fruitfly Drosophila melanogaster to plane-polarized light. In an electron-microscopic study Wada (1974a) found that fruitflies also have a specialized DRA. The polarization sensitivity of the optomotor response in Drosophila melanogaster is thoroughly treated in Sect. 27.3.

In the opinion of Wolf et al. (1980), at least under clear sky conditions Drosophila could use the celestial polarization pattern in flight orientation if


p > 15 %. Polarization sensitivity is not restricted to the dorsal part of the eye. It seems to possess polarization-sensitive optomotor response for most if not all parts of its visual fleld. Therefore, it might utilize polarization sensitivity to distinguish different surface properties on the ground, searching for water, or avoiding water, distinguishing wet and dry surfaces which are vital functions for such a small insect that can dry out and thus perish easily. Alternatively, also with its ventral eye region it could use a polarization pattern for course control.

Considering the polarization sensitivity of the peripheral RI-R6 retinula cells in flies, there is a controversial problem. On the one hand, the polarization sensitivity of receptors mediating the optomotor response has been demonstrated for both Musca (Kirschfeld and Reichardt 1970) and Drosophila (Heisenberg 1972; Wolf et al. 1980; Coombe et al. 1989). On the other hand, it was argued that due to neural superposition in the dipteran lamina ganglionaris, behavioural responses mediated by retinula cells RI-R6 should be almost invariant to the rotation of the E-vector of polarized light in Musca (Kirschfeld 1971). All these six cells, the activity of which is superimposed onto the same monopol ar cells in the lamina, differ in their planes of maximal absorption of linearly polarized light by ab out 60° from the neighbouring cells. Thus, polarization sensitivity should be lost at the level of the lamina if the six retinula cells have the same polarization sensitivity and if they contribute equally to the activity of the monopolar cells. In light of this argument, Kirschfeld (1973c) suggested that receptors RI-R6 may not be involved in the polarization-sensitive optomotor response. However, this is not the case in Drosophila. Coombe et al. (1989) showed that although the optomotor response mediated unambiguously by retinula cells RI-R6 is strongly polarization-sensitive, the large monopolar neurons in the lamina are not sensitive to polarization. The optic lobe projections of polarization-sensitive photoreceptors were investigated by Fortini and Rubin (1991).

17.3 Ants

According to Gronenberg (1996), ants are gene rally referred to as olfactory and tactile insects, because they have evolved sophisticated pheromone and tactile communication. Many ant species live in a world of scent and touch and have tiny eyes or may even be blind. Yet there are also predatory ant species with relatively large eyes and accordingly developed visual centres in the brain. The desert ant genus Cataglyphis has become a paradigm for a facet of polarization sensitivity research. Since usually only few landmarks are available in their habitats, desert ants rely on compass navigation and use the complex E-vector pattern of skylight.


In 1911 the Swiss physician Felix Santschi (1911) found that harvester ants (Messor sp.) can orient by sun compass. Later, Santschi (1923) observed that the harvester ant Messor barbarus could maintain its homeward course after occluding the sun and the landmarks on the horizon by a black cylinder moved with the walking ant and providing it with onlya small celestial patch at the zenith. Santschi hypothesized that the ants might have been able to infer the position of the sun from the intensity gradient potentially visible in the top opening of the screening cylinder. This experiment of Santschi (1923) indicated for the first time that an animal can derive compass information not only from the direct sunlight, but also from the skylight (Wehner 1982, p. 25, 148). About 70 years later, it has been shown that the desert ant Cataglyphis bicolor is indeed able to navigate by means of the spectral and polarization patterns of the sky as well (Wehner 1982). The older literature on the sensitivity of ants (e.g. Formica rufa, Solenopsis saevissima, Cataglyphis bicolor) to linear polarization is summarized by Wehner (1982).

Cataglyphis bicolor is a common ant species in the desert areas of northern Africa and southwest Asia (Wehner et al. 1994). It inhabits underground nests, which are indicated on the surface only by small holes surrounded asymmetrically by a low pile of soil that has been carried out of the nest, and also carapax shells of the insects that the ants have collected and eaten (Harkness and Wehner 1977). The ant forages randomly until it finds food and then runs straight back to its nest to minimize the time spent on the hot desert ground (the temperature of which can be as high as 70 Oe) in order to reduce the risk of desiccation (Lighton and Wehner 1990). In evolutionary terms there has been a high selection press ure for such astrategy, because during the hottest time of the day in their habitats the ants will perish if they have to walk for more than 2 h outside the nest (Harkness and Wehner 1977).

The straight homeward run demonstrates that during its excursion Cataglyphis should continually store all angles turned and all distances travelled, and integrate these angular and translational components of movement into aglobai vector by path-integration, that is, by true vector integration. This vector links the ant's actual position to the nest entrance (e.g. Wehner et al. 2002).

As a long-range navigator and predatory and solitary hunter Cataglyphis never moves along scent trails as so many other ant species do. This would be very ineffective on the hot desert ground due to the rapid evaporation of any scent mark. Thus, Cataglyphis relies upon visual signals. Its desert habitat has only in exceptional cases conspicuous landmarks, so that it must rely almost exclusively on skylight. Single ants can be easily trained to go to different feeding places more than 200 m away from the nest entrance (Wehner and Wehner 1990).

In behavioural field experiments, Wehner (1968) and Wehner and Menzel (1969) demonstrated that after the ants are displaced from their actual foraging course, neither chemical nor vibrational stimuli are decisive for homing.


They also showed that Cataglyphis bicolor orient by means of the solar azimuth as well as by landmarks whenever they are available. Foraging ants orient to terrestrial cues as long as possible, and only after these have become ineffective do they switch over to a time-compensated sun compass (Wehner and Lanfranconi 1981). Whereas the sun moves along its arc with uniform angular velo city of 15°/h, the solar azimuth does not. Its angular velo city depends on time of day, time of year and geographicallatitude. The ants are informed more or less accurately about the rotation of the solar azimuth during their foraging (Wehner and Lanfranconi 1981). However, they slightly underestimate the highest rotation rates of the solar azimuth at noon and overestimate lower rotation rates at sunset or sunrise. They use an internal circadian dock to correlate time-linked positions of the sun with an earthbound system of reference (Wehner and Müller 1993).

In field experiments, Wehner and Duelli (1971) studied the orientation of Cataglyphis bicolor between sunset and sunrise. As foraging is usually restricted to daytime in this ant species, the ants have been trained under daylight conditions and were afterwards tested during the night and twilight. On moonless nights they orient anemomenotactically, i.e. by means of the wind direction, similarly to daytime, if the compound eyes and the ocelli are covered with black paint. After elimination of the antennae, the ants search around randomly. On moonlit nights, they orient by means of a moon compass in the same way as if the mo on were the sun. During dusk at dirn light they orient anemomenotactically irrespectively of whether the moon is visible or not. About half an hour before sunrise and after sunset, they orient by means of the celestial polarization pattern. Due to the two-fold mirror symmetrical polarization pattern of the sky at this time (see ~ colour Fig.4.4), the ants orient bimodally. After sunrise and before sunset they use a sun compass if the sun is visible. They always decide alternatively between celestial and anemomenotactic courses and never perform compromising directions. The hierarchy between these different orientation mechanisms is the following: orientation by means of the celestial polarization pattern> anemomenotactic orientation > moon compass, where the symbol ">" means predominancy. The reason why the ants can orient only about 35 min prior to sunrise and after sunset is the disappearance and gradual fading of the skylight polarization pattern characteristic to daylight and twilight rather than the low light intensity, since the ants light sensitivity would also allow longer navigational periods (see Fig. 16, p. 39 in Wehner 1982). Under total overcast skies the ants perform their foraging courses only in the immediate vicinity of the entrance hole of their nest, since in this case neither the sun nor the polarization compass can be used (Wehner 1982).

Duelli and Wehner (1973) found in behavioural field experiments that Cataglyphis bicolor is able to orient by means of the celestial polarization pattern even without perceiving information regarding the solar position. The accuracy of artificially induced homing courses of the ants was not influenced


by performing the experiments just before sunrise and after sunset, when the ants could not see the sky near the horizon up to an elevation of 15°, or by occ1uding the sun during daytime. After destroying the celestial polarization pattern by an UV-transmitting pseudodepolarizer, the ants could orient exc1usively on the basis of the solar azimuth angle. The accuracy, however, was significantly lower than in the case when the celestial polarization pattern was the only available cue. If the sun was not visible and the skylight was unpolarized, the ants were disoriented. By shifting the solar azimuth with a mirror either during the foraging excursion or the return, the sun's azimuth competed with the celestial polarization pattern. In this situation the latter had preference and the solar azimuth had no influence on direction finding. It was demonstrated by mirror experiments that the solar altitude does not influence the orientation, but the accuracy of the compass reading decreases with increasing solar elevation (for theory and empirical data, see Wehner 1994, pp. 106-107). Using different colour filters carried horizontaUy by means of a troUey above the running ants during their foraging excursion, DueUi and Wehner have shown that Cataglyphis bicolor can orient by means of the polarization pattern of the sky only if the wavelength of skylight is between 380 and 410 nm, irrespectively of whether the sun is visible or not. Thus, only the UV receptors are involved in polarization sensitivity.

In two Ph.D. theses of the Zurich group (Duelli 1974; Fent 1985, see also Duelli 1975; Fent 1986) the importance of different eye regions for the celestial polarization compass was tested by restricting the E-vector information in the UV to specific parts of the visual field of the ant. The final result was that the uppermost dorsal part of the eye is both necessary and sufficient for deriving compass information from the E-vector pattern in the sky. Furthermore, the degree of polarization turned out not to influence the compass courses. Frantsevich et al. (1977) observed that the desert ant Cataglyphis setipes can also orient by means of the celestial polarization pattern in the Uv.

In addition to multifaceted lateral compound eyes, Cataglyphis bicolor also has three frontal ocelli, each possessing a single lens. The ocelli of certain locusts (Wilson 1978) and dragonflies (Stange 1981) function as horizon detectors involved in the visual stabilization of course. Wellington (1974b) suggested that the western bumblebee Bombus terricola occidentalis might be able to detect the skylight polarization by means of its ocelli. In a behavioural field experiment, Fent and Wehner (1985) showed that the ocelli of Cataglyphis bicolor can provide compass information from an isolated patch of the polarized c1ear sky. When the compound eyes were occ1uded and both sun and landmarks were obscured, the ants could orient with the help of their ocelli. However, in this case the accuracy of orientation was significantly lowered. When they could perceive only with their ocelli a single spot of totally linearly polarized light (diameter 40°, elevation 45°), they could select the proper compass course with respect to the E-vector direction. The ocelli are not necessary for this orientation, since painting over them has no influence


on the homing accuracy if the compound eyes can see the blue sky. Although it is unknown in what respect the celestial compasses provided by ocelli and compound eyes differ from each other, it is clear that the exclusively UV-sensitive and highly polarization-sensitive ocelli (Mote and Wehner 1980) can help in guiding the ants back horne. Electrophysiological (electroretinogram) measurements have shown that the ocelli look at sky regions that are closer to the horizon than to the zenith and that within the horizontal plane the visual axes of the left and right lateral ocelli deviate by 90° from the visual axis of the median ocellus (Fig. 21 in Wehner 1982). Fent and Wehner (1985) suggested that the three ocelli may function as a three-detector system that scans the sky for compass information. A scanning strategy involving widely separated detectors corresponds well to the tortuous walking of the ants, the compound eyes of wh ich have been occluded.

The retina of Cataglyphis bicolor can be subdivided into three regions (Herrling 1976): (1) the small dorsal rim area (DRA), (2) the dorsal area (DA) and (3) the ventral area (VA). By painting different eye regions black and testing the effect of this manipulation on the orientation, it was shown that also Cataglyphis bicolor perceives the skylight polarization with the anatomically and physiologically specialized DRA (Wehner 1976). Electrophysiological recordings from the photoreceptors by Labhart (1986) confirmed the behavioural finding that polarization sensitivity is mediated by the strongly polarization-sensitive UV receptors in the DRA. Orientation by landmarks is mediated by the VA and the DA, the sun compass by the DA and DRA. In the central eye region, Wehner and Toggweiler (1972) as well as Mote and Wehner (1980) found only two spectral types of photoreceptors with a sensitivity maximum at 380 or 550 nm. According to Herrling (1976), the retinula cells R2, R4, R6, R8 are green receptors, while Rl, R3, R5, R7, R9 are UV receptors.

Cataglyphis stabilizes its head position relative to pitch and roll movements about its transverse and longitudinal body axis, respectively, whenever it takes a compass reading by scanning the sky (Wehner 1982). Irrespective of the load carried, it maintains a constant angular (pitch) direction of its head during walking. When trained to walk on tilted surfaces, it compensates this tilt by counter-tilting its head. Only when the tilt exceeds the limit up to which the ant is able to properly adjust its head orientation, do navigational errors occur that are in accord with the misalignment between the DRA of the compound eye and the celestial hemisphere (Wehner 1992).

By means of intracellular recording, Labhart (1986) studied the characteristics of the photoreceptors in the DRA, the DA and the VA of the compound eye of Cataglyphis bicolor. The DRA is looking in a contralateral-dorsal direction (Wehner 1982), Le. the DRA of the left eye looks toward the right visual field and vice versa (Fig. 17.4). Labhart found two spectral types of receptors with sensitivity maxima at 350 and 510 nm in all three eye regions. In the DRA the UV cells are larger than the green cells. This is in contrast to the DRA of flies and crickets, in which only a single spectral type occurs, UV in honey-

152

I ft ey right eye

• ••• •• ••• " •••••• "0 , .. . -;;-:

""" • 'IU' ..- .... u:::: -_.. "" - ••• :::l

- • •• <I') .... - ••• '> _.... 1:

••• eJJ ... .;:

• • • •• • •• •


Fig. 17.4. Analyser array and contralateral field of view of the dorsal rim area (DRA) in the eye of the desert ant Cataglyphis bicolor. Bars show the analyser directions of receptors Rl and R5 in the DRA, circles represent the rhabdoms of retinal area adjacent to the DRA. The filled symbols within the grey area and the open symbols within the white area mark ommatidia that look at the right and left half of the celestial hemisphere, respectively. (After Wehner 1994).

bees and flies, while blue in crickets. The polarization sensitivity of UV cells in the DRA of Cataglyphis is PS = 6.3±2.4, wh ich is significantly higher than that in the DA and the VA, where PS = 2.9±1.6. The polarization sensitivity of the green cells in the DA is PS = 2.7±0.4 and in the VA PS = 1.9±O.7. In the DRA there are two populations of polarization-sensitive UV cells with orthogonal microvilli and preferred E-vector directions in each retinula cello The photoreceptors have narrow visual fields in all three eye regions. The acceptance angle of photoreceptors varies from 3,3.4,3.5 to 4.30 in the frontal eye part, VA and DA, respectively, to 5.40 in the DRA. Although the width of the visual field does not differ substantially across the eye, it gradually increases from the frontal eye region towards the DRA. This regional distribution of the acceptance angles corresponds closely to that of the interommatidial angles (see Fig. 2A in Wehner and Srinivasan 1984). The relatively narrow field of view of receptors in the DRA of Cataglyphis is in contrast to that in honeybees and crickets, in which the visual fields are strongly extended. Thus, enlarged visual fields of the polarization analyser cells in the DRA seems not to be a necessary condition for polarized skylight orientation.

In behavioural field experiments, Fent (1986) also demonstrated that Cataglyphis bicolor is able to use the celestial polarization pattern for orientation. Trained ants were confronted with single spots of artificially and naturally polarized light during their homing course beneath a trolley which allowed control of the celestial optical cues available to them. The vehicle prevented the ants from seeing landmarks and the sun as weIl as detecting wind direction. When exposed to a single spot with artificial horizontal E-vector, Cataglyphis was always oriented correctly toward the trained horne direction throughout the day. Irrespectively of the solar elevation, the ants always interpreted the horizontal E-vector as lying along the antisolar meridian. Pronounced and


consistent orientation errors occurred when other E-vector directions were presented. This indicates that the E-vector positions expected bythe ant do not coincide with the E-vector positions actually realized in the sky.

In a second experiment of Fent (1986), the ants were exposed simultaneously to two spots separated by 90° from each other and with different artificial E-vector directions. They oriented correct1y in ahomeward direction when one of the E-vectors was horizontal and the other vertical, though this E-vector constellation does not occur in reality. However, when the actual Evector constellation was displayed to the ants, they exhibited an orientation error. This indicates that each E-vector was equally important to the ants. Fent conduded that Cataglyphis bicolor largely ignores the dynamic feature of the celestial polarization patterns arising from the varying solar elevation, and it has no detailed knowledge of the actual azimuth angles of the celestial E-vectors. Instead, it assigns E-vectors stereotypically to fixed azimuth angles with respect to the solar meridian and is relying on some simpler means of E-vector navigation (see the "simplified celestial map" of honeybees in Fig. 17.3). Thus, it seems very unlikely that polarized skylight orientation in Cataglyphis is based on spherical geometry as suggested by Kirschfeld et al. (1975) or on some kind of memorised sky pattern when the animal would have to match the stored pattern last seen with the current one as proposed by Gould et al. (1985).

In a third experiment of Fent (1986), the ants were alternatively confronted with the same natural E-vector direction in spots of the solar and antisolar half of the sky. Usually, these two corresponding celestial spots with identical E-vector directions did not confuse the ants, though the degree of polarization, colour and intensity of skylight were different in the two spots. In a fourth experiment, in which the ants could see the full natural sky, they oriented correctly homeward even at sunset when the celestial polarization pattern is ambiguous due to the two orthogonal axes of mirror symmetry, solar/antisolar meridian and perpendicularly to it (see ~ colour Fig. 4.4). This was also the case when a small patch of natural sky was presented 50° from the solar meridian. This suggests that the ants use additional celestial cues, such as the patterns of the degree of polarization, colour or intensity of skylight, to find their way horne when the sun is obscured.

The findings of Fent (1986) are in dose analogy to those in honeybees (see Chap. 17.1). However, there is one difference between the two insects: While dancing bees assign all E-vectors to the antisolar half of the sky (Brines and Gould 1979), the walking Cataglyphis does not, since it interprets spots of natural skylight from the solar half of the sky readily as lying there. This suggests that the simplified celestial map of the ant consists of both the solar and the antisolar half of the sky, and the ant may distinguish between the two celestial halves by means of additional celestial cues mentioned.

From large series of experiments (Fent 1985; Müller 1989) one general condusion can be drawn: in desert ants navigational errors occur only when the


stimuli provided by skylight polarization vary between the training and testing situation. The usual experimental paradigm applied by Wehner and his students in studying the ant's E-vector compass was to train the animals under the full celestial E-vector pattern and to test them under restricted skylight conditions: with a small patch of sky (e.g. Fent 1985) or a strip-like sky window (e.g. Müller 1989). Then systematic navigational errors occurred, which, however, vanished whenever the E-vector patterns presented during training and testing coincided.

Based on these behavioural results as well as on the physiological finding that the outputs of the retinal polarization analysers (UV receptors) are sampled by only a few polarization-sensitive interneurons ("POL-neurons"), Wehner (1997) discussed two compass mechanisms. He star ted his discussion by stating that "it is tempting to speculate about how the ant's (and bee's) celestial compass system might work; but in following this temptation we must be cautious" (Wehner 1997, p. 177). According to his first hypothesis, the ant could use the E-vector pattern in the sky merely to read a reference direction (e.g. the solar or antisolar meridian) from the sky. For example, when the animal rotates about its vertical body axis and thus scans the sky with its array of polarization analysers, a balance between the outputs of the two dorsal rim areas would be reached when the animal is oriented parallel to the symmetry line of the skylight pattern, the solar vertical. This hypothesis corresponds to the "scanning hypothesis" as proposed for honeybees (Rossel and Wehner 1986). Any compass course other than the solar (or antisolar ) meridian must then be determined by other means.

The second hypothesis implies that the E-vector patterns in the sky are used to determine any particular point of the compass, and that the information about the points of the compass is encoded in the response pattern of the few POL-neurons in the ant's second visual neuropil, the medulla. As simulated by the software of an "ant robot" (Sahabot, Lambrinos et al. 1997,2000), the animal could acquire some kind of "neural look-up table" correlating POL-neuron response patterns with azimuthai positions. If the celestial Evector pattern varies from one foraging trip to another, andif the POL-neuron response patterns vary accordingly, the ant could always recalibrate its sky compass when starting a new foraging excursion. In many Cataglyphis species Wehner observed that when the ants left the nest on a foraging trip, they often perform rotatory movements on the spot (e.g. Wehner 1992). Such movements could be used in the calibration process. In this respect, Wehner (1997) warned that it might not be useful to invoke too strong a dichotomy between a "simultaneous" and "successive" mode of compass orientation.

In an electron-microscopical study Meyer and Domanico (1999) found that in the dorsal and ventral eye regions of Cataglyphis bicolor the retinula cells form a cylindrical rhabdom, where 2-3 neighbouring photoreceptors bundle their microvilli to form a "microvillar tuft" in which all microvilli are parallel, but they are sometimes slightly curved. The microvillar tufts of putative UV


receptors Rl, R3, R5, R7, R9 change their orientation repetitively every 1-4 pm for up to 90°, jumping between two positions at ir regular intervals, while the putative green receptors R2, R4, R6, R8 keep their microvillar orientation rigidly. Calculations based on waveguide optics revealed (Nilsson et al. 1987) that even relatively slight misalignments of the rhabdomeric microvilli cause a strong decrease in polarization sensitivity. A mean microvillar misalignment of ±30° reduces the polarization sensitivity to about PS = 2. The cross section of the rhabdomeres changes substantially along the length of the rhabdom due to periodical variations of the microvillar lengths. We mention that in bulldog ants, Myrmecia gulosa, irregular photoreceptor twist destroys or largely reduces the polarization sensitivity of the receptors (Menzel and Blakers 1975). Hence, in contrast to honeybees, flies and bulldog ants, where the rhabdomeres of all photoreceptors within the ommatidia outside the DRA are misaligned by the twist (bees: Wehner et al. 1975; flies: Smola and Tscharnke 1979; ants: Menzel and Blakers 1975), the misalignment in Cataglyphis is restricted to particular cells (Rl, R3, R5, R7,R9),namely the UV cells according to Herrling (1976). Cells Rl and R5 send their axons to the second visual neuropil, the medulla. All other photoreceptor axons terminate in the first visual neuropil, the lamina (Meyer 1984). The green receptors contribute many more microvilli to the rhabdom than do the UV receptors and may be involved in movement perception like in the honeybee (Kaiser 1974). Although the green receptors do not exhibit misalignments, their PS = 2.2 (Labhart 1986) is rather low. Mote and Wehner (1980) and Labhart (1986) proposed that this low PS may be caused by coupling between cells of different microvillar orientation.

On the other hand, the rhabdoms of the DRA have a rectangular or dumbbell-shaped cross section, with a constant area along the entire length of the cell, within which only two microvillar directions occur. The microvilli of receptors Rl and R5 are perpendicular to those of receptors R2-R4, R6-R9, and they are strictly parallel in a given receptor. The rhabdom cross section is approximately constant. The above findings regarding the microvilli orientations in Cataglyphis also hold qualitatively for the wood ant, Formica polyctena (Meyer and Domanico 1999).

Cataglyphis bicolor inhabits steppe-like regions, in which the desert floor consists of chalk- and gypsum-rich material, whereas Cataglyphis fortis lives on the often white and shiny salt-pan areas (Wehner et al. 1994). Meyer and Labhart (1998) demonstrated that large sand grains and salt crystals give rise to polarized reflections. Ronacher and Wehner (1995) showed that Cataglyphis ants can deduce some information about distances travelled from selfinduced image flow within their ventral field of view. Since polarized reflections might impair such perception, Meyer and Domanico (1999) suggested that one of the functions of the microvillar misalignment of the UV receptors in the ventral eye region of Cataglyphis may be to reduce the sensitivity of the eye to polarized UV light reflected from the desert floor. The other possible


function could be to eliminate polarization-induced false colours wh ich could impair colour vision (see Chap. 33). However, colour vision in Cataglyphis has not yet been proven.

With electrophysiological recordings from polarization-sensitive interneurons (POL-neurons) in the optic lobe (medulla) of Cataglyphys bicolor, Labhart (2000) found that the POL-neurons exhibit a characteristic polarization opponency and receive monochromatic input exclusively from the UV receptors of the DRA. The POL-neurons have spontaneous spiking activity in the dark, but with polarized light, spike frequency is approximately a sinusoidal function of E-vector orientation with alternating portions of excitation and inhibition. The E-vector orientation eliciting maximal spike frequency is approximately perpendicular to that eliciting minimal spike frequency. Thus, the POL-neurons of Cataglyphis are polarization-opponent units receiving antagonistic input from two E-vector-sensitive analyser channels with orthogonal orientation of maximal sensitivity. These polarizationally opponent neurons are represented in the retina by the two sets of polarization-sensitive UV receptors with orthogonally arranged microvilli in each ommatidium of the DRA (Herrling 1976). This antagonism has two advantages. It enhances sensitivity for E-vector contrasts, and makes the response independent of absolute light intensity (Labhart 1988, 1999). Removing the UV component from the stimulus abolishes the E-vector response. Thus, the POLneurons are driven by UV light in Cataglyphis. This is the reason why Cataglyphis depends on the polarized UV radiation of the sky for E-vector navigation (Duelli and Wehner 1973). Monochromacy of the polarization sensitivity, which implies colour blindness, prevents interference between spectral and E-vector information from the sky (Wehner and Bernard 1993).

17.4 Crickets

17.4.1 Acheta domesticus

A specialized dorsal rim area (DRA) possessing exclusively blue receptors was found in adult domestic crickets Acheta domesticus (cf. Zufall et al. 1989),contrary to the results of Gribakin et al. (1980). Behavioural experiments should show in the future, whether domestic crickets use their DRA for orientation by means of skylight polarization.

17.4.2 Gryllotalpa gryllotalpa

The burrowing and prevalently nocturnal mole cricket Gryllotalpa gryllotalpa lives in damp habitats, and it is a good swimmer. Ugolini and Felicioni (1991)


investigated the onshore orientation of mole crickets dropped into water. In this situation the main motivation for reaching the bank in the shortest time possible may be the danger of predation by fish and/or the difficulty in swimming or breathing. Using only celestial cues (sun and/or skylight polarization), Gryllotalpa gryllotalpa can correctly orient towards the bank it originally came from when released in water. The orienting mechanism is time-compensated.

17.4.3 Gryllus bimaculatus

According to the structural studies of Burghause (1979), there is an anatomically specialized DRA in the compound eye of the cricket Gryllus bimaculatus. Each ommatidium in this eye region contains two sets of receptor cells having orthogonal microvilli (lower inset in Fig. 17.5) and the cross section of the rhabdom is strongly increased compared with the rhabdoms in other eye regions. Distally, the receptor cells Rl, R2, R5, R6 and R7 form the rhabdom. Proximally, the short cell R8 also contributes to the rhabdom between Rl and R7, while R3 and R4 do not have any microvilli. Screening pigment and corneal facets, possessing lenses with a flat outer surface (Burghause 1979; Ukhanov et al. 1996), are missing in the DRA, which widens the visual fields of individual receptor cells. Due to the weak curvature of their entrance surfaces, the cornealienses in the DRA are slightly underfocused (Ukhanov et al. 1996). In contrast, in the remainder of the eye the rhabdom is formed by all the eight visual cells RI-R8, and the retina is strongly pigmented.

In a behavioural experiment of Burghause (1979), head flicks in tethered Gryllus bimaculatus were evoked by sudden rotation of a linear polarizer. This response disappeared when the DRA was painted over. Burghause concluded that Gryllus bimaculatus perceives the E-vector of linearly polarized light, although he did not exclude the possibility that the animals responded to inherent brightness patterns. The rhabdoms in the DRA are approximately twice as long as those in the central eye region with circa 50 % larger cross section (Burghause 1979; Hoff 1985), and almost double their volume at night in a circadian rhythm of membrane turnover (Hoff 1985). Similar specializations of the DRA have been found also in the Canarian cricket Cydoptiloides canariensis, in the ommatidia of which the crystalline cones are missing in the DRA (Egelhaaf and Dambach 1983).

Zufall et al. (1989) studied by means of intracellular recordings the spectral and polarization sensitivity of the retinula cells in adult Gryllus bimaculatus. They found three spectral types of photoreceptors with a sensitivity maximum at 332,445 or 515 nm. The green receptors have a secondary sensitivity maximum at 350 nm. Blue receptors are only in the DRA and respond comparatively slowly to visual stimuli. UV and green receptors occur in the dorsal eye region, while in the ventral region only green receptors were found. The


dorsal ------------- : nrn

;rorr:fiiJ:o= area

(DRA)

posterior

Fig.17.5. Neural integration in the POL-neurons of the field cricket Gryllus campestris. All three types of POL-neurons (POL-Ni, -N2, -N3) with different preferred E-vector directions (double-headed arrows) are connected with ommatidia in the anterior, middle and posterior sections of the dorsal rim area (DRA). Each POL-neuron type receives input from ommatidia, in which the orthogonal microvilli orientation (symbol Tin the inset below) is directed approximately parallel to the preferred E-vector direction, as exemplified bythe large bold Tsymbols in each box. The left-hand side inset below shows the distal cross section of a retinula in the DRA with two orthogonal microvilli directions. Here the short retinula cell RB, existing only proximally, is not visible. (After Labhart et al. 2001).


polarization sensitivity of the blue cells (PSB = 6.5±1.5) is much higher than that of UV (PSuv = 1.76±0.05) and green (PSG = 2.4±1) receptors. The blue receptors of the DRA send their axons either to the lamina and medulla (long visual fibres) or only to the lamina (short visual fibres). The axons of green receptors project with short visual fibres to the lamina, and the UV receptors with long visual fibres through the lamina to the medulla. Since the blue receptors in the DRA are about 60 times more sensitive than the green receptors, Zufall et al. (1989) suggested that the former are designed to extract Evector information under low illumination, e.g. at dawn, dusk or night. This is of particular importance, because crickets make foraging and mating excursions and also migrate in dirn light and even fly at night (Rost and Honegger 1987). Thus, crickets could orient to the celestial polarization pattern not only during walking, but also when flying during crepuscular and nocturnal periods. The calling and locomotor activity of the field cricket Gryllus campestris, for example, is spread throughout the whole day and night and has a minimum in the early morning (Rost and Honegger 1987).

Being hemimetabolous insects, crickets already have well-developed compound eyes after first hatching as in star larvae. During development new ommatidia differentiate within a growth zone situated at the anterior margin of the eye. By means of light- and electron-microscopy, Labhart and Keller (1992) investigated the fine structure and growth of the DRA in the compound eye of larval Gryllus bimaculatus. They found that in larval crickets a specialized DRA is already present. During growth the total number of both corneal facets and DRA rhabdoms increases exponentially with an average growth rate of 1.4 and 1.48 per instar. Thus, the DRA grows rather isometrically with respect to the whole eye. The proportion of DRA rhabdoms varies between 7.7 and 10.4% for instars 1-9 and increases slightly to 15.4% in the adults. During growth both the length and the cross section of the rhabdoms gradually increase. Hence, cricket larvae are equipped with the same sensory structures for polarization sensitivity as adults. It remains to be tested by behavioural experiments from which developmental stage the sensory potential for E-vector detection is actually exploited. Labhart and Keller (1992) hypothesized that the absolute sensitivity and precision of E-vector discrimination in larval crickets may be reduced compared with adult crickets due to sm aller rhabdoms and the low number of ommatidia in the DRA.

Gribakin and Ukhanov (1993) hypothesized that light-scattering could be one of the reasons for high polarization sensitivity of the photoreceptors in the DRA. This assumption dates back to Waterman (1981), who presumed that a scattering polarizer may underlie polarization sensitivity in some animals. Yet Ukhanov et al. (1996) have shown that in Gryllus bimaculatus the cornealienses of the DRA are neither linear polarizers nor light scattering elements.


17.4.4 Gryllus campestris

Labhart et al. (1984) and Blum and Labhart (2000) investigated by means of intracellular recordings the spectral, angular and polarization sensitivities of photoreceptors, the ommatidial array and the receptor axon projections in the DRA of adult field crickets Gryllus campestris. These characteristics of receptors in the DRA were compared with those in the adjacent dorsal area (DA) of the eye. The optical axes of dorsal rim ommatidia are directed to the contralateral side of the upper visual field of the eye, i.e. 20-50° from the zenith. The DRA is flat and within it the interommatidial angle is very narrow (_1°). The narrow visual field of ommatidia in the DA has an average acceptance angle of 6°, while the field of view of the DRA receptors ranges from 6 to 67° along the long axis of the DRA, and 6 to 37° perpendicularly to it with a medium value of 20°. These very wide fields of view are the consequence of the flat cornealienses and the absence of screening pigments. Due to the wide field of view and narrow interommatidial angle, the visual fields of dorsal rim ommatidia overlap extensively, such that a given portion of the sky is viewed simultaneously by a large number of ommatidia.

There is no rhabdomeric twist either in the DA, nor in the DRA. The DRA contains exclusively blue receptors with a sensitivity peak at 435 nm and high polarization sensitivity PS = 9.8, whereas the DA is composed of green (510 nm,PS = 2.6) and UV (340 nm,PS = 2.9-4.2) receptors. The microvilli in the DRA exhibit a relatively high degree of alignment along the rhabdom. Within each rhabdomere the microvilli orientation varies by only ±9° as compared to ±18° in the adjacent DA (Nilsson et al. 1987). The receptors Rl, R2, R5 and R6 project to the lamina, whereas R7 projects to the medulla. The microvilli directions of these two projection types differ by 90° (lower inset in Fig. 17.5), indicating the two analyser channels that give antagonistic input to the polarization-sensitive interneurons. The microvilli direction in the short proximal receptor R8 is perpendicular to that of R7. The volume of the rhabdomere of R8 is only about 2.5 % of that of R7 and only 1 % of the total rhabdome re volume of Rl, R2, R5 and R6. Therefore, R8 is about two orders of magnitude less sensitive than any of the two analyser channels, and its contribution to polarization sensitivity can probably be neglected (Blum and Labhart 2000).

Due to the wide field of view and slow light response of the DRA receptors (Labhart et al. 1984; Blum and Labhart 2000), disturbances of the celestial polarization pattern in space and time are already filtered away at the receptor level. According to Popov (1975), field crickets change burrows several times during their adult life. While foraging they may stray up to 2 m from their burrows. When disturbed, they return horne on a direct course. These observations suggest that they are capable of visual navigation. Labhart et al. (1984) suggested that the possible role of the DRA in field crickets may be the analysis of skylight polarization.


In an open-Ioop behavioural experiment, Brunner and Labhart (1987) found that tethered adult male and female Gryllus campestris walking on an air-suspended Styrofoam ball spontaneously responded to the E-vector of polarized light from above. The stimulus was a wide (83°) overhead field of totally linearly polarized white light. When the E-vector direction rotated continuously, the turning tendency of the crickets changed periodically with the rotation of the E-vector. The direction and amount of turning tendency was a function of the actual E-vector direction. Control experiments showed that the crickets reacted to the E-vector and not to any weak intensity patterns of the polarizer. Experiments in which different eye regions of the crickets were covered with black paint suggested that the response to the E-vector is mediated by the DRA. Brunner and Labhart (1987) thus concluded that the DRA plays a key role in cricket polarization vision. The crickets seem to prefer certain E-vector directions, suggesting polarotactic behaviour. In similar laboratory behavioural experiments, Nilges (1990) studied the orientation of Gryllus campestris to linearly polarized light. Although polarization sensitivity was clearly demonstrated, it remains to be shown whether this ability of the crickets is used indeed for celestial orientation. Although celestial orientation was also demonstrated in the crickets Nemobius sylvestris (Beugnon 1983) and Pteronemobius lineolatus (Beugnon 1986), it was not established whether the cues were the solar position, or the pattern of polarization and/or intensity and/or colour of skylight. The polarization-sensitive DRA could also play an important role in the offshore-onshore escape reaction (movements perpendicular to the shoreline) of the cricket Pteronemobius lineolatus. Beugnon (1986) showed that this cricket uses celestial cues in its orientation during landward swimming. This behaviour could also be governed by the celestial polarization pattern when the sun is not visible, like the offshore-onshore movements of certain other animals.

With the same method as Brunner and Labhart (1987) also Herzmann and Labhart (1989) found that the sensitivity of adult Gryllus campestris to totally linearly polarized light is maximal in the blue. This supports the thesis that polarization sensitivity in field crickets is mediated by the highly polarization-sensitive blue receptors in the DRA. The absolute sensitivity of the E-vector detection system is very high. The threshold irradiance at 433 nm (where the blue receptors are maximally sensitive) for totally linearly polarized light was 2.5.107 quanta cm-2s-1, which is lower than the effective quantum flux under clear, moonless night skies between 380 and 500 nm (Höhn and Maffeo 1967). A similar value was obtained for the absolute sensitivity threshold of the polarization-sensitive interneurons. For partially polarized light with lower degrees of polarization p, threshold intensities in crickets may be somewhat high er.

Polarization-sensitive neurons in the optic lobe of crickets were studied electrophysiologically by Labhart (1988), Labhart and Petzold (1993), Petzold (2001) and Labhart et al. (2001). Stimulating the DRA with flashes of totally


linearly polarized light with different E-vector directions, Labhart (1988) found polarization-opponent interneurons (POL-neurons) in the optic lobe of adult Gryllus campestris. These neurons receive dendritic input in the dorsalmost part of the medulla, have their cell body in the proximal part of the medulla and send a long axon through the brain to the contralateral optic lobe. The spike activity of these neurons depends on the E-vector direction, but is independent oflight intensity variations above a criticallevel. The sensitivity of POL-neurons is maximal in the blue. The spike frequency of POLneurons is a sinusoidal function of E-vector direction with an excitatory and an inhibitory part (Fig. 17.6, diagram lower right). The E-vectors eliciting maximal and minimal spike frequencies are perpendicular to each other. POL-neurons are thus polarization-opponent: they receive antagonistic input from two sets of polarization-sensitive blue receptors with orthogonal microvilli that are present in each ommatidium of the DRA (Fig. 17.6). Although polarization-sensitive interneurons have also been reported earlier in the visual system of crabs (Leggett 1976) and goldfish (Waterman and Aoki 1974; Waterman and Hashimoto 1974), this was the first demonstration of polarization-opponent neurons!. There are three types of POL-neuron characterized by preferred E-vector directions eliciting maximal spike frequency at 10,60 and 130° relative to the body long axis (Fig. 17.5). The orientation of the rhabdoms within the DRA varies in a fanlike fashion such that each section of the DRA contains differently oriented microvilli in a semicircular arrangement (Fig.17.5). Each type of POL-neuron receives input from a large number of ommatidia from the whole DRA, collecting input from ommatidia in which the microvilli orientations approximate the preferred directions of maximal sensitivity of the POL-neurons (Fig. 17.5). All three POL-neuron types have large and almost fully overlapping visual fields with angular diameters of ab out 60° (Fig. 17.7). Thus, the celestial polarization pattern is not repre sen ted in fine spatial detail, but by just three types of interneurons, or even three individual neurons.

The wide visual fields of the polarization-sensitive photoreceptors in the DRA are an adaptation to the coarse spatial gradient of E-vector direction in the sky, which makes high spatial resolution unnecessary. This wide-field optical integration makes the system less susceptible to disturbances such as clouds or plants partly obstructing the celestial polarization pattern. The polarization-opponent mechanism makes the system insensitive to changes of skylight intensity. Furthermore, it enhances effective E-vector contrast. With monochromatic antagonistic receptors the system is colour-blind, which eliminates interference with spectral modulations.

1 Although Zolotov and Frantsevich (1977) reported also on polarization-sensitive interneurons in the beetle Geotrupes, later they (Frantsevich and Zolotov 1979) withdrew this observation admitting that their method was unreliable.


Fig.17.6. Left Schematic representation of a polarization-opponent interneuron (POL-neuron) in the field cricket Gryllus campestris. The two analyser channels from the retinula ceU R7 and R 1 + R2 + R5+ R6 of the dorsal rim area of the compound eye act antagonisticaUy on a POL-neuron. Right E-vector response functions (spike frequencies) of receptors (upper graph) and POL-neurons (lower graph). The E-vector angle is measured from an arbitrary reference direction. The response of the POL-neuron is the difference between the logarithm of responses ofR7 and Rl+R2+R5+R6. (After Labhart 1988).

163

RI + R2+ 1;\ + 1 R7 R5+ R6 "' l / ...... 'x' ....... \ .. .." / .::. ...... "',.., - E-vector orientation ~ 0 90·

orienlalion

By means of intracellular recordings from POL-neurons, Labhart (1996) measured the threshold of the degree of linear polarization p of blue light that can elicit E-vector response in the POL-neurons of adult Gryllus campestris. Elliptically polarized light, which is physiologically the same for an insect photoreceptor as partially linearly polarized light with the same degree of polarization p, was produced by combining a linear polarizer with an optical retarder, i.e. a quarter-wave plate. Ellipticity and thus p is a function of the angle between the E-vector produced by the linear polarizer and the principal axis of the retarder. As this angle increases from 0 to 45°, P decreases from 100 to 0 % and the major axis of the polarization ellipse also varies. p varied between 2.5 and 98 %. Labhart (1996) found that the POL-neurons can signal E-vector information if p > 5 % for blue light. The precision with which the Evector direction is defined by the PO L-neurons was ±3 to ±4 ° if P > 10% and ±6° for 5 < P < 10 %. This allows the polarization-sensitive system of field crickets to exploit skylight polarization for orientation even under unfavourable meteorological conditions, when the sky is densely cluttered with clouds. After sunset, p of light from the clear twilight sky first increases with time and then decreases with decreasing solar elevation.

Labhart (1999) developed a field-proof opto-electronic model of the POLneurons of Gryllus campestris. The model neuron implemented the basic physiological properties of the polarization-opponent interneurons, such as the size and direction of view of the receptive field, the spectral sensitivity and the polarization-opponent mechanism. With this model neuron, both the strength of the celestial polarization signal and the directional information available to POL-neurons were assessed under different meteorological conditions. The model neuron scanned the sky along a circumzenithal path producing a response function with two maxima indicating horizontal E-vectors. The azimuth angles lPm of the two local maxima of its output signal were


determined, wh ich define the azimuth angles of the solar and antisolar meridians. The azimuth error L1<p = l<Pr - <Pm I , defined as the difference between the real solar/antisolar azimuth angle <Pr and the measured azimuth angle <Pm' was used as a measure of precision by which the model neuron can determine the direction of the solar/antisolar meridian, the most important information for the celestial compass system. Under cloudy ski es the standard deviation of the azimuth error L1<pwas only ±2.3° for the solar half of the sky and ±3.4° for the antisolar half, and it was never over 12°. The strength of the polarization signal, i.e. p integrated over the visual field of the model, did not exceed 50 % even under clear skies with low solar elevation. The median integrated p for aH measurements (including cloudy skies) was only 13 and 28 % for the solar and the antisolar half of the sky, respectively. For integrated p-values over 5 % the error was independent of the strength of the polarization signal in the sky. This demonstrates weH that the polarization signal as experienced by the POL-neurons is very robust, indicating that the polarization compass of crickets is reliable even under cloudy conditions. This is due to the special filtering properties of these neurons, such as polarization antagonism, spatial lowpass, monochromacy, and to the relatively stable E-vector pattern of the sky (Brines and Gould 1982; Pomozi et al. 2001 b). Although the large visual field (Fig. 17.7) reduces the strength of the polarization signal (since it integrates E-vectors over a large area of the sky), it acts as a spatiallow-pass filter serving to even out local disturbances in skylight polarization.

visual tield

po terior

Fig.17.7. Position and size of the visual fields of POL-neurons in the field cricket Gryllus campestris mapped in the polar coodinate system of the celestial hemisphere. The visual fields are indicated by grey circular areas for the POL-neurons of both the right and the left optic lobe. (After Labhart et al. 2001).


17.5 Lepidoptera: Butterflies and Moths

About the polarization sensitivity of Lepidoptera (butterflies and moths), there are only scarcely available data in the literature. Warrant et al. (2003) briefly reviewed some of these data. The majority of relevant publications deals exc1usive1y with the anatomy of the retina of certain lepidopteran species (e.g. Meinecke 1981; Eguchi 1999). From these studies only the rhabdom structure, the microvillar orientations and the existence or absence of rhabdomeric twist or misalignments can be deduced, but it remains uncertain whether the photoreceptors are indeed polarization sensitive.

According to Meinecke (1981), in moths the rhabdoms of most ommatidia are usually curved. Eguchi (1999) and Warrant et al. (1999) also found strongly curved microvilli in certain sphingid moths. Such curved microvilli could reduce or even abolish polarization sensitivity. In a few lepidopteran species anatomical specializations have been found in the dorsal rim area (DRA) of the superposition compound eyes. These ommatidia are characterized by untwisted rhabdoms with strict orthogonal microvilli directions. Meinecke (1981) has reported on DRA in the African armyworm moth Spodoptera exempta and the sphingid moth Deilephila elpenor. DRAs were found also in the saturniid moth Antheraea polyphemus (Anton-Erx1eben and Langer 1988), in the tortricid moth Adoxophyes reticulata (Hämmerle and Kolb 1987), in the moth Manduca sexta (Warrant et al. 2003), in the nymphalid butterflies Aglais urticae (Kolb 1986), Pararge aegeria (Hämmerle and Kolb 1996), Danaus plexippus, (Warrant et al. 2003) and in the papilionid butterfly Papilio aegeus (Warrant et al. 2003). Rhabdoms in the DRA of butterflies are gene rally much shorter than in the remainder of the eye (Kolb 1986).

Behavioural or electrophysiological evidence for polarization sensitivity of the DRA in butterfly and moth eyes as well as for the use of this eye region in celestial orientation are yet lacking. Nevertheless, certain butterfly species can navigate by the sun compass. Baker (1968), for example, showed that some British butterflies orient by the solar position during their migration. The monarch butterfly Danaus plexippus has also been proven to use a sun compass (Perez et al. 1997). Orientation by means of skylight polarization has only been studied in Danaus plexippus by Hyatt (1993). This species may use the celestial polarization pattern for orientation during migration. In this chapter we survey the sporadic information about polarization sensitivity and polarization-guided behaviour in butter flies and moths.


17.5.1 Papilio xuthus

With electron-microscopy and intracellular recording from single photoreceptors, Bandai and Arikawa (1990) as well as Bandai et al. (1992) studied the polarization sensitivity of photoreceptors of different spectral types in the ommatidia of both male and female Japanese yellow swallowtail butterflies Papilio xuthus. This butterfly has five types of photoreceptor with sensitivity maxima at 364,402,460,522 and 601 nm. There are two types of green receptors with and without a secondary absorption peak at 364 nm. An ommatidium contains nine receptors (RI-R9), in each of wh ich parallel microvilli form a non-twisted rhabdom, and thereby the receptors are polarization-sensitive. The microvilli of receptors Rl, R2 and R9 are oriented dorso-ventrally, i.e. vertically in the animal's normal posture, while those of R3 and R4 are parallel to the horizontal antero-posterior axis of the body. The microvilli of receptors R5 and R7 are directed at 135°, while those of R6 and R8 at 45° from the dorso-ventral axis. The Uv, violet and blue receptors are maximally sensitive to the vertical E-vector, while both types of green receptors to the horizontal E-vector. The double-peaked green receptors also have a preference for the horizontal E-vector in the UV. The red receptors are most sensitive to Evector direction at ab out 35° from the vertical. The polarization sensitivities of the ultraviolet (UV), violet (V), blue (B), single-peaked green (SG), doublepeaked green with Amax=522 nm (DG522), double-peaked green with Amax=364 nm (DG364) and red (R) receptors are: PSuv = 1.62±0.13, PSv =

1.79±0.12, PSB = 2.59±0.21, PSSG = 1.93±0.11, PSDG522 = 1.89±0.07, PSDG364 = 4.14±O.65, PSR = 2.35±O.57. Bandai et al. (1992) concluded that Rl and R2 are either Uv, violet or blue receptors, while R3 and R4 are green, and some R6 and R8 are red receptors (Fig. 17.8). All green receptors have horizontal microvilli and horizontal preferred E-vector direction. The same horizontal alignment of green receptor microvilli was found in the cabbage butterfly Pieris rapae (Shimohigashi and Tominaga 1991). Waterman (1981) hypothesized that perhaps reflection of polarized green light from plant leaves may contain some important information for certain butterflies.

17.5.2 Polarization-Induced False Colours Perceived by Papilio xuthus and Papilio aegeus

The females of the Australian orchard butterfly Papilio aegeus lay their eggs on the underside of the shiny leaves of Rutaceae or citrus plants. Multiple- and dual-choice experiments showed that females of Papilio aegeus prefer to oviposit on artificial green substrata (Kelber 1999a). In a laboratory dualchoice experiment, Kelber (l999b) studied the influence of totally linearly polarized light on the colour sensitivity of Papilio aegeus. Females had to choose between two stimuli, each passing a vertical coloured filter that was


Fig. 17.8. Microvilli directions of the photoreceptors of different spectral types in the ommatidium of the Japanese yellow A swallowtail butterfly Papilio xuthus (after Bandai et al. 1992).

B

c

167

illuminated from behind and covered by a totally linearly polarizing filter of different directions of the transmission axis. When the green colour of the two stimuli was the same, the butterflies preferred horizontally polarized light over diagonally or vertically polarized light. When a bright green and a dark green stimulus were simultaneously presented with horizontal E-vector, the butterflies preferred the dark green colour. However, this preference was reversed by changing the E-vector direction of the dark green stimulus from horizontal to vertical.

In similar dual-choice laboratory experiments, Kelber et al. (2001) showed that Papilio aegeus and Papilio xuthus discriminate between vertically, obliquely and horizontally polarized light of the same colour regarding oviposition and fee ding. The discrimination depends on the spectral compo-


sition of the stimuli. In the blue and probably in the green, discrimination does not depend on light intensity. Females of Papilio aegeus spontaneously preferred horizontally polarized over vertically or obliquely polarized light of the same colour, and obliquely polarized light over vertically polarized light of the same colour. The butterflies did not discriminate light polarized at 45° or 135° relative to the vertical. The preference for horizontally versus vertically polarized light depended on colour. Papilio xuthus trained to associate vertically polarized blue light with food strongly preferred this stimulus in unrewarded tests. Animals trained to vertically polarized red light preferred this stimulus in tests, in which both stimuli were equally intense. Papilio xuthus trained to horizontally polarized red light did not learn the stimulus. The choice behaviour of butterflies trained to red stimuli depended strongly on light intensity. Red-trained animals preferred the blue vertically polarized stimulus to the blue horizontally polarized one. Animals trained to blue vertically polarized light preferred red vertically polarized light against red horizontally polarized light.

Kelber (I999b) and Kelber et al. (2001) proposed that Papilio aegeus and Papilio xuthus do not perceive polarization and colour separately, and thus they may perceive polarization-induced false colours (see Chap. 33). This was the first observation which indicated the possible perception of such polarizational false colours. Kelber and coworkers suggested that such detection of polarizational false colours may help butterflies to find optimal oviposition sites. They proposed that with the help of polarization-induced false colours butter flies could, for example, discriminate between shiny and matt leaves before landing on them. Leaves reflect different amounts of polarized light, so the colours of glossy and matt leaves may look different to a butterfly, and the colour could change when the butterfly passes by. This property might be an indicator of the quality of leaves as a food source for the caterpillar. Kelber and collaborators also hypothesized that horizontally oriented leaves would probably be more attractive than vertically oriented ones to an egg-Iaying butterfly, because they would provide a better landing and offer more protection for the eggs and young larvae. To an approaching Papilio female, the shiny leaves of a citrus bush may have different colours depending on their orientation. A horizontally oriented leaf could look more green, whereas a vertically oriented one more blue-green or reddish. Perhaps the leaf orientation could be coded by the hue and saturation of the perceived polarizational false colours, wh ich could help to select the approximately horizontally oriented leaves already from a distance. Due to the polarization-induced false colours, as a whole, a plant with shiny leaves would look more colourful than one that does not reflect polarized light. This could help the females to choose the larval food plant from a distance. Caterpillars of Papilio aegeus grow faster on young leaves, which may be preferred by females for oviposition, and females can chose between old and young leaves within a plant by colour vision (Kelber 1999a). The selection of


young leaves can perhaps be improved by perception of polarizational false colours. However, some serious arguments against the significance of the role of polarization-induced false colours in the colour vision of weakly polarization-sensitive butterflies were proposed by Horv<ith et al. (2002 c; see Chap.33).

17.5.3 Polarized Light Reflected from Butterfly Wings as a Possible Mating Signal in Heliconius cydno chioneus

Some iridescent scales of certain butterflies reflect highly linearly polarized light at certain angles (Vukusic et al. 2000). In an outdoor insectary Sweeney et al. (2003) investigated the response of closely related male Heliconius cydno chioneus and Heliconius melpomene malleti butterflies to polarized and depolarized views of wings of female butterflies, which were conspecific to the males. Males approached the polarizing wings of female H. cydno chioneus significantly less often when the wings were displayed behind a depolarizer rather than the non-depolarizer. On the other hand, there was no significant difference in the response of male H. melpomene malleti to the non-polarizing female wings for either condition. Sweeney et al. hypothesized that the polarization sensitivity of Heliconius butterflies may have adaptive value in dense forest, where both the intensity and colour of the illumination can vary strongly.

Unfortunately, it was not studied by optical measurements how the depolarizing and non-depolarizing filters used by Sweeney et al. affected the intensity, colour and polarization of light reflected from the butterfly wings. It remained unknown whether the two female wings of H. cydno chioneus displayed in a given double-choice experiment differed indeed only in the polarization of reflected light. It is unclear if the depolarizer completely removed the linear polarization signal and left colour and brightness intact, and whether the non-depolarizer indeed left both polarization and colour unaffected. Thus at present there is no convincing evidence that polarized light reflected from wings is involved in butterfly mate recognition.

17.6 Locusts

Electrophysiological recordings have shown that the locust photoreceptors are polarization sensitive (Shaw 1967). Eggers and Weber (1993) observed in laboratory experiments menotactic orientation of locust larvae walking on a Kramer treadmill with respect to the E-vector of totally linearly polarized light from above. Selective occlusion of various eye regions and/or the ocelli indicated that this E-vector-dependent orientation in locusts is mediated by


photoreceptors in the dorsal rim area (DRA) of the compound eye. Intracellular recordings from photoreceptors in the DRA of the desert locust Schistocerca gregaria revealed average polarization sensitivity PSB ::::: 6 in the blue and PSuv ::::: 2 in the UV. Most retinula cells were blue-sensitive (Eggers and Gewecke 1993).

Homberg and Paech (2002) studied the ultrastructure and orientation of ommatidia in the DRA of Schistocerca gregaria. The DRA contains about 400 ommatidia with a visual field of 16-30° contralaterally. The DRA facets are flat and half the size of other facets. Pore canals are often present in the cornea resulting in a grey cloudyness, probably due to light scattering. Screening pigment is missing in the region of the dioptrie apparatus in the DRA indieating large receptive fields. According to Eggers and Gewecke (1993), the visual field of DRA photoreceptors is about 30°. The DRA rhabdoms are shorter, but about four times larger in cross section than those of other ommatidia. Rhabdom shortening favours polarization sensitivity by reducing self-screening, while widening the rhabdomerie cross section increases absolute sensitivity (Nilsson et al. 1987).

In the DRA of Schistocerca gregaria eight retinula cells contribute to the rhabdom. The mierovilli of retinula cell R7 are perpendieular to those of Rl, R2, R5, R6 and R8 throughout the entire rhabdom without twist. The mierovilli in the minute rhabdomeres of R3 and R4, in contrast, have no particular alignment. Photoreceptors Rl, R2, R5, R6, R7 and R8 may be bluewhile R3 and R4 could be UV-sensitive. Microvillar orientations are arranged in a fanlike pattern across the DRA (Fig. 17.9). Photoreceptor axons project to distinct areas in the dorsal lamina and medulla. These morphologieal specializations in the DRA most likely maximize the polarization sensitivity and indieate that in Schistocerca gregaria this eye region makes possible the analysis of the celestial polarization pattern, perhaps for maintaining migratory directions during changing wind conditions. This hypothesis is supported by observed polarotaxis during stationary flights in laboratory experiments (cf. Homberg and Paech 2002). This response was abolished after painting over the DRA. Kennedy (1951) provided evidence for celestial compass orientation in mass migrating Schistocerca gregaria. He was able to divert the walking path of locust larvae as well as the flight course of individuals in a swarm by deflecting the solar azimuth with a mirror. However, it is unclear whether the sun alone or the sky polarization pattern was responsible for these reactions.

With intracellular recordings from the optic lobe combined with Lucifer dye injections, Homberg and Würden (1997) found polarization-sensitive neurons in the medulla of adult Schistocerca gregaria when the dorsal eye region was stimulated by totally linearly polarized white light with different E-vector directions. This was the first report on polarization-sensitive neurons in the locust visual system. One population of the medulla tangential neurons with projections to the lamina was maximally sensitive to E-vector


Fig. 17.9. A Schematic drawing of the dorsal part of the head of the desert locust Schistocerca gregaria with the compound eyes in which the dorsal rim areas are designated with DRA. B Photo graph of the dorsal part of the head of the locust. C Scanning electron micrograph of the transition zone of the eye between the DRA and the dorsal area (DA). D Rhabdom cross sections and microvilli orientations of retinula ceHs R7 in the DRA of a male desert locust. The microvilli are arranged in a fan-shaped pattern. (After Homberg and Paech 2002).

direction about 100° from the body long axis, while the other population responded maximally to ab out 20° E-vector direction. The polarized light response of these neurons was independent of intensity above a certain threshold. Two medulla tangential neurons with contralateral optic-lobe projections were maximally sensitive to E-vector directions ab out 80° respectively 60° from the body long axis. With dirn background illumination, the responses to polarized light were usually excitatory, but one neuron was polarization-opponent. Such polarization-opponent neurons were found in the field cricket Gryllus campestris (Labhart 1988), the desert ant Cataglyphis bicolor (Labhart 2000) and the cockroaches Periplaneta americana (Kelly and Mote 1990a) and Leucophaea maderae (Loesel and Homberg 2001). While the cricket polarization-opponent neurons receive inputs from the DRA, the polarization-sensitive neurons of Schistocerca gregaria do not receive direct input from polarization-sensitive photoreceptors, but require intercalated interneurons between the DRA and the medulla.


Vitzthum et al. (2002) found polarization-opponent interneurons in the central complex of the brain in Schistocerca gregaria. Receptive fields of these neurons are in the dorsal field of view with some neurons receiving input from both eyes and others only from the ipsilateral eye. These neurons have no preferences for particular E-vector directions. This contrasts with the three classes of polarization-opponent neurons with three different preferred E-vector directions found in field crickets (Labhart 1988). Vitzthum et al. (2002) proposed that the central complex in the brain of Schistocerca gregaria may be a centre for direction perception and spatial navigation and, therefore, is likely to exploit all information available for this task including, in particular, the celestial polarization pattern.

17.7 Cockroaches

Polarization sensitivity has been demonstrated in the nocturnal cockroach Periplaneta americana, the retina of which has two photoreceptor classes, one maximaHy sensitive at 365 nm and the other at 510 nm (Butler and Horridge 1973). With intracellular recording KeHy and Mote (1990a) found two polarization-sensitive neurons in the optic lobe of adult male Periplaneta americana. One of these neurons received either excitatory or inhibitory input from both UV and green photoreceptors. The other neuron was colour-opponent, i.e it was inhibited by UV and excited by green receptors. These neurons responded to the direction of rotation of the E-vector of totaUy linearly polarized violet (410 nm) light. The significance of polarization sensitivity in a nocturnal insect is not clear. Kelly and Mote (1990a) admitted that in their electrophysiological experiments spurious reflections from the light source as well as unrecognised intensity fluctuations could have occurred through polarizer rotation. Thus, it was not excluded that the observed polarization sensitivity was simply an artefact. Behavioural experiments have shown that this cockroach avoids light transmitted through a rotating linear polarizer (KeHy and Mote 1990b). However, according to KeHy and Mote (1990b), the reason for this avoidance might be the sequential activation of receptors due to motion sensitivity rather than sensitivity to the Evector direction.

Using intracellular recordings combined with dye injections, Loesel and Homberg (2001) found polarization-sensitive interneurons in adult males of the Madeira cockroach Leucophaea maderae. Two types of commissural neurons with projections to the contralateral optic lobe were sensitive to the E-vector direction of totally linearly polarized blue light. These neurons coded light intensity and were not polarization-opponent. They were most sensitive to stimulation in the ipsilateral field of view. This suggests that, in addition to ommatidia in a small dorsal rim area of the eye, more lateral eye


regions may contribute to the polarization-selective responses. Loesel and Homberg (2001) hypothesized that polarization sensitivity in the Madeira cockroach might playa role in contrast enhancement at low light intensities in addition to a possible function in orientation by means of skylight polarization.

17.8 Scarab Beetles

In behavioural field experiments, in which the animals were shaded from the sun and/or covered with a linear polarizer and/or a colour filter, Frantsevich et al. (1977) studied the celestial orientation of the wingless scarab beetles Lethrus apterus and Lethrus inermis. These beetles wander from their hole to search randomly for vegetable food during repeated excursions of ab out 1 m from the hole. They return along approximately straight lines with a maximal error of less than about 10 cm, which indicates that they perform path integration. In a new excursion the beetles do not keep to the earlier searching direction. They are active only in sunny weather and select their homeward direction relative to the sun or to the polarization pattern of the sky when the sun is not visible. The beetles perceive the skylight polarization in the UV. They can maintain their proper homeward direction under a linear polarizer if its transmission axis coincides with the E-vector direction of skylight from the celestial band of maximum degree of polarization at the antisolar meridian. Turning the polarizer's transmission axis by 90°, the beetles also turn accordingly or are disoriented. They have green (530 nm) and UV (350 nm) receptors in their superposition compound eyes adapted to diurnal vision. In both the dorsal and ventral eye regions crustacean-like rhabdoms occur, in which there are two stacks of microvilli, the directions of which are perpendicular to each other.

Labhart et al. (1992) investigated the anatomical specializations of ommatidia in the dorsal rim area (DRA) of the refracting superposition compound eyes of the crepuscular/nocturnal cockchafer Melolontha melolontha. In the DRA the peripheral regions of cornealienses are non-transparent, but contain light-scattering, bubble-like inclusions. The fused rhabdom is composed by the rhabdomeres of retinula cells RI-R7, and the untwisted rhabdom is approximately twice as long as that outside the DRA. On the other hand, the cross section of DRA rhabdoms is about half of that in other eye regions. Within each rhabdomere the microvilli are parallel to each other. The microvilli of Rl are perpendicular to those of R2-R7. The tracheal sheets around the retinulae are missing. They separate optically the rhabdoms outside the DRA. These anatomical specializations indicate that the DRA photoreceptors in Melolontha melolontha are highly polarization-sensitive and may have strongly overlapping wide visual fields. In the DRA the electrophys-


iologically measured polarization sensitivity ranges from PS = 3 to 10, and Evector detection may be mediated by the green receptors with maximal sensitivityat about 520 nm. Behavioural evidence suggests that in the dusk- and night -active beetle Parastizopus armaticeps the green receptors are also involved in polarization sensitivity (Bisch 1999). The rhabdoms within the DRA have a fan-shaped arrangement. Labhart et al. (1992) suggested that Melolontha melolontha may also detect skylight polarization for celestial orientation during its migratory flights between breeding and feeding areas by means of the DRA. After emergence at the breeding site cockchafers fly to feeding places which can be several kilometres away. This flight seems to be dominated by hypsotaxis, Le. the insects direct themselves towards forested hills silhouetted on the horizon (Schneider 1967).After feeding and oogenesis the beetles return to their breeding area, during wh ich they rely on celestial compass orientation (e.g. Couturier and Robert 1957). The flight activity of cockchafers is crepuscular (Schneider 1967), yet the celestial polarization persists until about 1 h after sunset.

There are indications for similar anatomical specializations in the DRA of other scarab beetles, such as the dung beetles Geotrupes auratus (Gokan 1989), Onthophagus lenzii (Gokan, 1990) as well as the Christmas beetle Anoplognathus (Labhart et al. 1992). Birukow (1954) observed that the dung beetle Geotrupes silvaticus can orient by means of the sun position and skylight polarization.

The diurnal flightless desert dung beetle Pachysoma striatum lives on firm coastal sands in the deserts of the Southern African West Coast, where they wander over long distances searching for dry dung. After finding dung, the beetle digs a burrow to store the resource.1t walks back and forth between its burrow and the food source along a straight path. Rearranging landmarks around the nest or occlusion of the sun by clouds do not affect this behaviour, as long as the sky is clear around the zenith (Scholtz 1989). This suggests that these scarab beetles can navigate by means of skylight polarization. Dacke et al. (2002) studied the DRA of the superposition compound eye in adult Pachysoma striatum. Morphological studies and intracellular recordings indicated that the DRA cover a considerable part (40%) ofthe dorsal eye. The parallel microvilli of photoreceptors Rl, R4 and R5 are perpendicular to those of R2, R3, R6 and R7, and they are oriented in a fan-shaped pattern across the eye. The fused rhabdoms in the DRA have a slightly larger cross-sectional area than those outside the DRA. No other features, e.g. differences in the optics, screening pigment density or rhabdom length between the DRA and other eye regions, which increase the visual fields of polarization-sensitive photoreceptors in the DRA of other insects, were found in Pachysoma striatum. The average polarization sensitivity of single-peaked UV (350 nm) photoreceptors in the DRA was PSuvs = 12.8±1.2, while PSUVD = 6.5±1.l for doublepeaked UV cells with a secondary sensitivity maximum at 550 nm. PSc-uv = 9.6 was found in one green receptor with a secondary sensitivity maximum in


the UV. Dacke et al. (2002) suggested that these specializations make the DRA of Pachysoma striatum suited for detection of the polarization of skylight in the UV. However, they argued that in Pachysoma striatum an optieally unspecialized DRA for detection of skylight polarization may not be used exclusively for navigation. The beetle could also detect avian predators by means of the DRA, in whieh the dioptrie apparatus possesses good optieal properties.

The dung beetle Scarabaeus zambesianus starts foraging on the wing for fresh dung at around sunset. After locating a source of fresh droppings, it forms a ball of dung and rolls it away at high speed in a straight line radial from the source. This is the most efficient path for escaping aggressive competition for food in the dung pile. The ball is finally buried in a suitable place to be consumed in secure solitude, either by the beetle itself or by a beetle larva. In behavioural field and laboratory experiments Dacke et al. (2003a) showed that the crepuscular dung beetle is able to roll along a straight path by using the polarization pattern of the clear evening sky. In the field the rolling direction turned by approximately 90° when the beetle entered an area covered by a linearly polarizing filter, the transmission axis of whieh was perpendieular to the E-vector of skylight at the zenith. When the transmission axis of the overhead polarizer was parallel to the zenith E-vector direction, no change of the rolling direction occurred. On completely overcast evenings the beetles were inactive.

In another field experiment Dacke et al. (2003b) found that on moonlit nights Scarabaeus zambesianus rolls radially in straight lines away from the dung, but on nights without a mo on, it no longer follows a straight line path. If the mo on is not visible, the beetle orientates by using the polarization pattern of the moonlit night sky, whieh is practieally the same as that of the sunlit day sky for the same position of sun and mo on (Gal et al. 2001a, see Chap. 8). By using the polarization of the moonlit sky for orientation, s. zambesianus is able to extend its foraging time.

In S. zambesianus skylight polarization is detected by the photoreceptors in the dorsal rim area (DRA) of the superposition compound eye (Dacke et al. 2003a). The transverse axis of the DRA rhabdoms is oriented in a dorsally converging fan-shaped pattern across the eye. The structure of the DRA ommatidia differs from that outside the DRA. They have non-twisted rhabdoms with two orthogonal mierovilli directions. A reflecting tracheal sheath and a lack of screening pigments makes these large ommatidia weIl adapted for detection of skylight polarization at low light levels.


17.9 Response of Night -Flying Insects to Linearly Polarized Light

Light traps are important tools in investigations of insect phenology and flight activity as well as in comparative studies, particularly in the case of insect spedes of economic importance. First, Kovorov and Monchadskiy (1963) attracted insects with light traps emitting linearly polarized light. Their aim was to increase the effidency of light traps and they found that polarizing light traps have caught at least twice as many Lepidoptera than unpolarizing traps.

The possibility that in night-flying insects polarized light is a cue for orientation during their dispersal and migration has rarely been considered. Danthanarayana and Dashper (1986) hypothesized that the correlation of the flight activity peaks of certain mosquitoes and moths during the lunar cyde with the degree of linear polarization of moonlight may be explained by the polarization sensitivity of these insects. They examined the response of some night -flying insects to polarized light in the field by means of three light traps of the same type arranged in an equilateral triangle 10m apart, operating simultaneously and emitting equal amounts of white light. In two traps the light sources were covered with linearly polarizing filters. One of them emitted totally vertically and the other horizontally polarized light. The third trap emitted unpolarized white light of approximately the same intensity as that of the polarizing traps. The relative positions of the traps were changed daily. Significantly more Trichoptera and Diptera, particularly Chironomidae were attracted to the horizontally than to the non-polarizing or vertically polarizing traps. However, more Dermaptera were attracted by vertically than by horizontally or non-polarized light. Nearly equal, but low numbers of Coleoptera, Hemiptera, Hymenoptera and Neuroptera were captured in each trap type. More Lepidoptera were caught by the non-polarizing trap, which is in contrast to the above-mentioned results of Kovorov and Monchadskiy (1963).

In another experiment, Danthanarayana and Dashper (1986) studied the response of nocturnal tortridd moths Epiphyas postvittana and Laspeyresia pomonella to linearly polarized light. The night activity of these moths was compared in a glass cylinder illuminated by a vertical white light beam, which was unpolarized or totally linearly polarized with the same intensity. The activity of both moth spedes was significantly high er in polarized light than in unpolarized light or in total darkness.

Polarized light increases the sensitivity of compound eyes by about 15-30 % in comparison to unpolarized light of the same intensity (MazokhinPorshnyakov 1969). Danthanarayana and Dashper (1986) suggested that this increase in sensitivity may be useful for insects depending on and/or activated by natural partially linearly polarized dirn light, such as twilight and


moonlight. Perhaps this phenomenon explains the high er attractiveness of polarizing light traps to night -flying insects. In insects having open rhabdoms high absolute sensitivity is associated with low polarization sensitivity (PS), while in insects possessing fused rhabdoms high absolute sensitivity goes together with high PS (Snyder 1973). Consequently, polarizing light traps may enhance the number of attracted insects with highly polarization-sensitive fused rhabdoms, since for these insects the polarized light emitted by the trap could be brighter. The species-specific preference of horizontally or vertically polarizing light traps could be explained by the possibly different eye structure. For insects possessing predominantly vertically aligned microvilli, the vertically polarizing light trap could appear brighter than the horizontally polarizing one and thus the former would be phototactically more attractive than the latter. On the other hand, polarotactic aquatic insects, preferring horizontally polarized light for water detection, would be obviously attracted to the horizontally polarizing light trap during their nocturnal flight. Of course, on the basis of experiments with polarizing light traps it is difficult to determine whether a given insect species is sensitive to polarization or its response to polarized light is governed by simple phototaxis.

18 Polarization Sensitivity in Insects Associated with Water

Insects associated with water may possess a ventral polarization-sensitive eye region, which could be involved in the detection of water and horizon, or optomotoric compensation for rotation and displacement of the body. Evidence for the existence of such a specialized "ventral POL-area" (VPA) was found in the waterstrider Gerris lacustris (Schneider and Langer 1969) and the backswimmer Notonecta glauca (Schwind 1983b, 1985b). Walton (1935) demonstrated that flying backswimmers (Notonecta sp.) identify the water by visual cues. Schwind (I983a) discovered that Notonecta glauca detects water by means of the horizontal polarization of light reflected from the water surface. Schwind (1985b, 1991) has shown that many other aquatic insects also find their habitat by means of polarotaxis (Table 18.1).

Polarotactic detection of water surfaces has a lot of sense, because the light returned from water bodies can possess complex patterns of intensity and colour because of the interference of the light returned by particles suspended in water and the light reflected from the surface (Fig. 18.1). Due to these complicated colour and intensity patterns, the phototactic detection of water surfaces would be unreliable for water insects.

Some insects, however, may make use of other cues to detect water bodies. Insects inhabiting running waters, for example, may not locate their habitat by means of reflection polarization, because polarization is distorted by waves (e.g. Zwick 1990). Aquatic insects and insects living on moist substrata are influenced in their habitat choice not only by polarization, but also by the intensity and colour of reflected light as weH as by non-optical factors, e.g. temperature, humidity, odours, flavours (e.g. Westphal 1984). Shortly after dropping onto a substratum, they may find it unsuitable and leave again. Their visual system, however, is the first that becomes operative as a long-distance water detector before other factors become operative (Schwind 1991).

18 Polarization Sensitivity in Insects Associated with Water 179

Table 18.1. Polarotactic insects detecting water or moist substrata by means of the horizontal polarization of reflected light studied by rnultiple-choice field experiments of Schwind (1991, 1995). The known spectral ranges in which the polarization of reflected light is perceived are given in brackets. (After Schwind 1995).

HETEROPTERA

Corixidae:

Pleidae:

Saldidae:

EPHEMERIDAE:

COLEOPTERA Dytiscidae:

Haliplidae:

Hydrophilidae:

Hydraenidae:

Sphaeridiinae:

Sigara nigrolineata (360 nrn), Sigara lateralis (360 nrn)

Plea leachi

Saldula saltatoria

Cloeon sp. (450-480 nrn)

Agabus bipustulatus (480-520 nrn), Bidessus nasutus, Guignotus pusillus (360 nrn), Hydroporus sp. (390-420 nrn), Laccophilus minutus (430-450 nrn), Potamonectes sp., Rhantus pulverosus (500 nrn), Scarodytes halensis

Neohaliplus lineatocollis (530-550 nrn), Haliplinus lineolatus (530-550 nrn), Peltodytes caesus

Anacaena limbata (390-420 nrn), Enochrus quadripuctatus, Helophorus aquaticus ( < 360 nrn), Helophorus brevipalpis « 360 nrn), Helophorus jlavipes « 360 nrn), Helophorus griseus « 360 nrn), Helophorus minutus « 360 nrn), Helochares lividus (380-390 nrn), Hydraena sp., Hydrobius fuscipes (370-400 nrn), Laccobius sinatus (370-390 nrn), Limnoxenus niger

Limnebius crinifer (360-380 nrn)

Megasternum boletophagum, Cryptopleurum minutum, Cercyon sp.

Fig.18.1. Photo graph of a smalllake under a partly cloudy sky dernonstrating that the light returned frorn water bodies can possess cornplex patterns of intensity (and colour, not visible in this picture) due to the interference of the light returned by particles suspended in water and the light reflected frorn the surface. On the surface of this lake we can distinguish the following typical regions: sunlit areas, shady regions, rnirror image of clouds, patches of the clear sky and vegetation on the shore. Due to this cornplex intensity pattern, the phototactic detection of water surfaces would not be a reliable rnethod for water insects. This is the reasan why these insects detect the water polarotactically.


18.1 Velia caprai

In a behaviourallaboratory experiment, Rensing (1962) found that the nightactive water bug Velia caprai preferred to orient spontaneously its body long axis at 01180, ±30, ±60 and ±90° relative to the E-vector of totally linearly polarized white light from an overhead source in two different situations: (1) the body was fixed and the animal could run in situ on a small plastic sphere; (2) the animal could freely move on the water surface in a circular arena. The biological relevance, if any, of this reaction to downwelling polarized light in Velia is unclear.

18.2 Corixa punctata

Rensing and Bogenschütz (1966) studied the spontaneous orientation of the body long axis of the water bug Corixa punctata lying on a rough white underwater substratum in the laboratory. The insect could move freely in a water-filled dish illuminated from above with totally linearly polarized white light produced by an incandescent lamp. The light passed through a diffuser and a polarizer. The arena was surrounded by a vertical black curtain. The long axis of the insect was oriented predominantly 0, ±45 and ±90° with respect to the E-vector. Since the orientation ofboth positively and negatively phototactic individuals was the same, their orientation might not have been governed by unwanted brightness patterns induced by selective reflection of polarized light from the black curtain. Covering the ventral eye region with black paint did not influence the orientation, while painting the dorsal eye region resulted in random orientation. Increasing the intensity of polarized light led to the decrease of the standard deviation of directions of orientation o and ±90° relative to the E-vector and the increase of the standard deviation of the directions of orientation ±45°. The biological relevance, if any, of this re action of motionless Corixa is completely unknown.

18.3 Non-Biting Midges (Chironomidae)

Danthanarayana and Daspher (1986) equipped their light traps for night-flying insects with linear polarizers and found that especially non-biting midges (Chironomidae) were much more strongly attracted by horizontally than by vertically polarized light when both light sources were equally intense. Here, the polarization-sensitive visual system serving water detection may have been involved. Schwind (1991) observed that non-biting midges oviposited


exclusively on a glass pane underlain by a matt black cloth, clearly discriminating it from a much brighter glass pane underlain by a rough aluminium sheet reflecting weakly vertically polarized light. From this he concluded that Chironomidae may be polarotactic insects possessing intensity invariant discrimination of polarized light with horizontal E-vector from unpolarized light.

18.4 Waterstrider Gerris lacustris

Waterstriders (Gerridae, Heteroptera), which live during spring and summer on the water surface of small ponds and creeks, have compound eyes with open rhabdoms, a typical design for heteropteran insects. According to Schneider and Langer (1969), within the apposition compound eye of the waterstrider Gerris lacustris there are two types of ommatidia (Fig. 18.2): dorso-Iateral and ventral type. In both types the open rhabdom consists of six peripheral rhabdomeres RI-R6 around the central rhabdomeres R7 or R8. The axes of the microvilli in R2, R3, RS and R6 are horizontal (cranio-caudal), parallel to each other and lie perpendicular to the vertically (dorso-ventrally) orientated microvilli of Rl and R4. In the dorso-Iateral ommatidium type, with optical axes pointing from 30° ventral to 90° dorsal, the central part of the rhabdom is tiered, consisting of a distal and a proximal visual cello Rhabdomere R7 with horizontal microvilli is formed only distally, while the proximal cell extends only up to the middle of the rhabdom and forms two rhabdomeres R8 with vertical microvilli, lying parallel to one another. Hence, the microvilli axes of R7 and R8 are perpendicular to each other. In the ventraltype ommatidia, where the optical axes point 30° or more ventral, the central

Fig. 18.2. Schematic cross section of the ommatidium in different regions of the compound eye of the waterstrider Gerris lacustris. The hatched areas represent the microvilli ofretinula cens R1-R8 and their orientation in the open rhabdom. (After Bohn and Täuber 1971).


rhabdom is formed totally by the two parallel rhabdomeres R8 which have nearly the same length as Rl and R4, while R2, R3, R5 and R6 are shorter, existing only distally. The field of view of the ventral eye region extends laterally and medially up to 60° and -15° from the vertical, respectively. Thus, the animal has a binocular field of view of 30° beneath its head (Schneider and Langer 1969; Bohn and Täuber 1971). The geometrical centrum of the eyes is at a distance of ab out 2.5 mm from the water surface. The whole field of view of the ventral regions of the (wo eyes scans an oval area with a radius of about 5-6 mm on the water surface.

The ratio between the cross sections of distal rhabdomeres with horizontal and vertical microvilli is 1:1 in the dorso-Iateral eye region and 1:4 in the ventral one. Thus, the rhabdomeres with vertical microvilli predominate in the ventral ommatidia. Similar anatomical specializations occur in the ventral eye region of the long-Iegged fly Sympycnus lineatus (Trujillo-Cenoz and Bernard 1972). In the ventral eye region of Gerris, the proportion of the peripheral rhabdomeres R2, R3, R5 and R6 with horizontal microvilli is 45, 22 and 0 % distally, medially and proximally, respectively. The corresponding proportions of Rl and R4 with vertical microvilli are 55,78 and 100 % (Schneider and Langer 1969).

Bohn and Täuber (197l) found two maxima of responses to rotating E-vector of totally linearly polarized white and coloured (360, 4l3, 500, 570 nm) light at 0/180 and ±90° relative to the dorso-ventral plane of the eyes in the electroretinogram (ERG) of the lateral eye region in Gerris lacustris. ERG responses from the ventral eye region had one maximum at 0/180° from the dorso-ventral plane. No amplitude modulation of the ERG to rotating E-vector was registered in the dorsal eye region.

Hamann and Langer (1980) found the prerequisites of a trichromatic colour vision system with sensitivity maxima at 350, 460 and 525 nm in the dark-adapted equatorial eye region of Gerris lacustris. These results agree with the findings of Bartsch (1995), who studied the polarization and spectral sensitivity of photoreceptors by means of intracellular recordings in lateral-equatorial and lateral-dorsal ommatidia with azimuth angles from 50° to 120° relative to the frontal direction and elevation angles between -10° and +50° with respect to the eye equator. He found that the average polarization sensitivity at 455 nm is PS = 6.7 (PSmax = 14.3), while at 515 nm PS = 6.9 (PSmax = 16.6). The blue receptors are maximally sensitive to horizontally polarized light, whereas the green receptors either to horizontally or vertically polarized light. Thus, the green receptors could form an orthogonal polarization-sensitive system. In the lateral and dorsal eye regions R7 may be blue-sensitive, R8 UV-sensitive and RI-R6 green-sensitive. The function of the high PS of the green receptors is probably related to contrast enhancement of objects which lie on or near the water surface. The function of the high PS of the blue receptors is unclear, and nothing is known about the PS of the UV receptors.


Schneider and Langer (1969) suggested that the ventral rhabdom specialization in the Gerris eye has a screening function against horizontally polarized surface-reflected light to get a better view into the water. Their argumentation was that light reflected from the water surface is always horizontally polarized, so it is filtered by the rhabdomeres with vertical microvilli. This vertical microvilli direction, on the other hand, is optimal for perceiving light from objects below water, which is vertically polarized because of refraction at the water surface.

A further function of the polarization sensitivity and structure of the ventral eye region is the water detection. Schwind (1991, 1995) found that in Gerris lacustris water detection happens in the UV (360-370 nm). Thus, water detection is probably mediated by the UV-sensitive receptor R8. This is rather surprising, because the vertical microvilli orientation of R8 argues against a sufficient water detection due to the predominantly horizontal polarization of the water-reflected light. In the ventral eye region, an ideal water detector could be the subsystem RI-R6 with vertical and horizontal microvilli. However, this subsystem is in alllikelihood green-sensitive.

The optomotor response of Gerris lacustris to different oscillating polarization patterns and its possible roles were studied by Horvath et al. (unpublished) and are described in detail in Chap. 27.5.

18.5 Backswimmer Notonecta glauca

Recording sum potentials from a dorsal eye part, Lüdtke (1957) demonstrated polarization sensitivity in the eye of the backswimmer Notonecta glauca. The recorded potentials were highest if the E-vector of the stimulating linearly polarized light was parallel to the microvilli of the rhabdomeres R3 and R4. These early experiments were performed soon after the discovery of Frisch (1949) that bees use the celestial polarization pattern for navigation.

Later on, Rudolf Schwind discovered that polarization sensitivity of the visual system of Notonecta plays a role in a completely different behavioural context: in aseries of experiments Schwind (1983a,b; 1984a,b; 1985a,b) showed that flying backswimmers detect bodies of water by the horizontal polarization of UV light reflected from the water surface. In a first behavioural experiment with backswimmers flying in the laboratory towards a vertical lightening surface, Schwind (1983a) observed that weak UV light emitted by a horizontal surface below the flying Notonecta glauca elicited a vertical turn of the flight path toward the horizontal surface even though a vertical surface above the horizontal one emitted much more intense UV light. This response is polarization sensitive. Linearly polarized UV light from below with E-vector perpendicular to the median plane of the flying


animal is more attractive than UV light with the same intensity and E-vector parallel to the median plane. These early experiments led to the idea that water surfaces are detected by the high intensity contrast of the reflected polarized UV light.

Although in nature there is always more UV light radiated from the sky than reflected from water surfaces, the surface-reflected UV light is predominantly horizontally polarized, while the E-vector direction of skylight depends on the solar elevation as well as the direction of view. Furthermore, plants reflect less UV than green light (Gates 1980) with E-vectors perpendicular to the random plane of reflection due to the random orientation of the leafblades. Thus in the Uv, water surfaces are more distinct from their surroundings than in the visible spectral range, and their reflection-polarizational characteristics contrast prominently with the adjacent landscape for a UV- and polarization-sensitive visual system (e.g. see Fig. 5 ofSchwind 1983a,p. 91).

Later experiments (Schwind 1984a, 1985a), however, have shown that it is not the intensity contrast alone that leads the backswimmers to water, but the polarization of reflected UV light can be discriminated from unpolarized UV light, and polarization is perceived invariantly from intensity as a unique sensory quality.

Backswimmers were set into flight in a room with a homogeneously illuminated ceiling and a light-emitting platform on the floor (Schwind 1984a). With these experimental conditions, which were more natural than the former ones, polarized UV light from the platform was more attractive than unpolarized UV light even if the latter was several times more intense than the former. Experiments with an array of baffles restricting the directions from which the polarizing platform could be seen showed that the polarized UV light can cause the animals to descend only if the horizontal E-vector stimulating the forward- and downward-Iooking parts of the eyes is perpendicular to the median sagittal plane of the flying animal. Schwind concluded that polarized UV light with horizontal E-vector is distinguished by backswimmers as a special sensory quality from unpolarized UV light regardless of intensity. Threshold intensity of the re action to polarized UV light is independent of the overall intensity of UV light in the surroundings (Schwind 1985a). Unpolarized UV light attracts the animals only when the intensity is high er than that of the surroundings, which shows that unpolarized UV light elicits only the positive phototactic behaviour. In nature the intensity of the UV component of skylight is always several times higher than that of UV light reflected from water surfaces. Thus, it is logical that the descent to water is elicited exclusively by the polarized component of the reflected UV light. Furthermore, the animals displayed a complex re action in the flight room above the polarized UV light-emitting surface as well as above a natural water surface.

Recording the characteristic diving movement of Notonecta glauca into water with a high-speed camera, Schwind (1984b) found that typical flight


paths rise slightly before the animals turn down toward the horizontal surface emitting polarized light with E-vector perpendicular to the median plane of the flying animals. At the beginning of the ascent the body long axis raises head-up, then tilts downward while spreading out the rowing legs and closing the wings. During the des cent phase the body axis points nearly vertically downward (Fig. 18.3A). This "plunge reaction" was also observed by Walton (1935) in Notonecta maculata in the field above water surfaces. The angle of the body axis to the horizontal is about 15° during normal forward flight, while it is 36° when the body is maximally raised during the plunge reaction. Schwind (1984b) suggested that one of the possible functions of raising the body before the plunge may be to see the highly and horizontally polarized Brewster zone of the water surface (Fig. 18.3B).

The ventral eye region is specialized for the detection of the horizontal polarization of UV light reflected from the water surface (Schwind 1983b). Without rising the head the ommatidia of this ventral eye region do not look toward the Brewster zone. However, when the body is raised to an angle over about 33° from the horizontal, as in the plunge re action, these ommatidia look towards the Brewster zone. The sudden incidence of strongly polarized light in the ventral part of the eye could contribute to the immediate elicitation of the next phases of the plunge. In the laboratory the plunge re action is performed in a similar way above a polarizing filter, but the duration of the phase of flying over an almost totally linearly polarized light source with the body raised, before the actual plunge, is longer than over real water surfaces. The

Fig.18.3. A Two examples (a, b) of the plunge reaction of the backswimmer Notonecta glauca landing on a horizontal linearly polarizing filter with transmission axis perpendicular to the median plane of the flying animal in the laboratory. c Orientations of the body long axis during the plunge reaction over an outdoor pool. B Silhouette of a backswimmer flying above water, and the field of view of its specialized ventral eye region during forward flight (shaded with light grey) and while the body long axis is raised in the plunge reaction (shaded with dark grey). The arrow indicates the direction of the Brewster angle. (After Schwind 1984b).

water surfacc

-~ -.. .. , \ ,

I I

•

A

\ \

\

\


reason for this may be that the light transmitted through the polarizer lacks the gradients of the degree of linear polarization that would correspond to a water surface.

The intensity invariant discrimination of polarized UV light with horizontal E-vector from unpolarized UV light in backswimmers is mediated by the special orthogonal arrangement of the vertical and horizontal microvilli in the rhabdomeres of the UV-sensitive central visual cells R7 and R8 in the ventral part of the eye (Fig. 18.4). Although with only two analyzers, Le. photorecetors of the same spectral (UV) type with vertical and horizontal microvilli, the state of polarization of linearly polarized light cannot be determined unambiguously. The two-channel system of backswimmers can reliably distinguish horizontally polarized UV light reflected from the water surface from unpolarized or obliquely (including vertically) polarized UV light, if the signals Sver and Shor of the two receptors with vertical and horizontal microvilli are compared with each other, respectively, by an appropriate neural circuit:

• If the stimulating light is unpolarized, or polarized with E-vector tilted ±45° from the vertical, then Sver = Shor'

• If the angle of polarization a of the incident light is -45° < a < +45° from the vertical, then Sver > Shor'

• If 45° < a< 135°, then Sver < Shor'

In the apposition compound eye of Notonecta glauca, the six peripheral rhabdomeres RI-R6 are open and grouped around the two central fused rhabdomeres R7 and R8 (Fig. 18.4). In the median eye region, with optical axes pointing from 60° ventral to 80° dorsal, the microvilli of the central rhab-

Fig. 18.4. Schematic representation of the cross section of the rhabdom in the ventral region of the compound eye of Notonecta glauca with the average orientations of the microvilli (thick bars) and their standard deviations (thin bars). These ommatidia look immediately above the water surface through the Snell window at its margin. The contours ofretinula cells RI-R8 are also shown. Vertical bar at Zeft orientation of the dorso-ventral plane of the animal. S, Average direction of the symmetryaxis of the rhabdoms; G, green receptor; UV/B, ultraviolet or blue receptor. (After Schwind 1985b).


dome res are horizontal and parallel to each other (Fig.18.5). In the ventral eye region, where the optical axes of the ommatidia point 700 or more ventral, the microvilli of rhabdomeres R7 and R8 are horizontally and vertically aligned, respectively (Schwind 1983b). Both in the dorsal and ventral eye regions, the receptors RI-R6 are maximally sensitive at 560 nm and serve as a scotopic system under water. In the dorsal eye region, the R7 and R8 are either UV-sensitive (346 nm) or blue-sensitive (445 nm). In the ventral eye region R7 and R8 are only UV-sensitive (Schwind et al. 1984).

Hence, receptors R7 and R8 are of the same spectral type (UV), their microvilli are perpendicular to each other, they are of about the same length and cross section without twist over their entire length. The predominance of UV-sensitive central receptors in the ventral part of the eye is thought to be significant in the location of a water body during flight (Schwind et al. 1984). It was suggested that blue receptors are present in the dorsal and perhaps absent in the ventral eye region, because when the animal is swimming with the ventral side upward, the part of the eye looking into the water, owing to the spectral composition of the underwater light field, receives less UV light than the ventral part looking out of the water.

Until now, any function of the polarization sensitivity of the ventral polarization-sensitive area (VPA), looking upwards when Notonecta is hanging beneath the water surface, could not be found. In the opinion of Schwind (1983b), the special pattern of the microvilli in the ventral part of the eye might not be related to an ability of the animal to orient itself by skylight polarization under water. Because of the small size of the habitat of an animal swimming in a pool, an orientation mechanism based on polarized skylight seems quite unnecessary. If the only function of the backswimmer's VPA were

Fig.18.5. A Directions of view of the different eye regions and the microvilli directions (hatched) of the central UV- or bluesensitive photoreceptors R7 and R8 in the backswimmer Notonecta glauca hanging upside down under the water surface in its normal resting posture; B the animal is flying above a water surface (after Schwind 1983b).

A

B

water surface

R74!\1l)R8

iiiR7 RB

water surface

R7€®R8

body lang ax;s

body long axis


the detection of water, it would be more efficient if this VPA extended over the whole ventral eye region and not only over 70° or more ventral from the longitudinal body axis. In this case, water could already be detected from a remote distance by the VPA, and the raising of the body axis prior to the plunge reaction would be unnecessary.

The longitudinal body axis of Notonecta encloses an angle of 30° in the normal resting position relative to the water surface (Fig. 18.5A). In this position the visual field of the backswimmer is subdivided into three different regions: (1) the VPA looks into the air through the Snell window; (2) the eye region directly below the VPA looks at the region of total reflection of the water-air interface; (3) the remaining part of the eye looks at the underwater world. The eye is adapted to this zonation of the optical environment (Schwind 1983b). Notonecta spends most of its time in water and the field of view of its VPA coincides with the Snell window. Hence, in resting position below the water surface, its VPA perceives the refraction-polarization pattern of skylight and light from the shore vegetation visible through the Snell window.

An orientation based on the celestial polarization pattern is indeed very unlikely in Notonecta. However, the polarized light reflected from objects on the shore may be significant in another behavioural context. Notonecta inhabits not only still ponds, but also creeks and other kinds of slowly flowing waters, where the animals often try to maintain a nearly stationary position. Resting under the surface of calm waters, they can also be drifted or rotated by the wind. Similarly to waterstriders (Junger and Varju 1990), backswimmers also compensate for passive rotation and translation (Blanke 1996), and these reactions are also mediated predominantly by visual cues. Thus, another function of the VPA under water might be the enhancement of contrast of terrestriallandmarks serving optomotoric compensation of passive rotation of the body. The optomotor response of Notonecta glauca to different oscillating polarization patterns and its possible role were studied by Horvath et al. (unpublished) and are described in detail in Chap. 27.6.

18.6 Dragonflies Odonata1

Meyer and Labhart (1993) examined the fine structure of dorsal rim ommatidia in the compound eyes of the dragonflies Sympetrum striolatum, Aeshna cyanea and Ischnura elegans. These specialized ommatidia have very short rhabdoms, thereby minimizing self-screening. They are not twisted, indicat-

1 According to the International Commission on Zoological Nomenclature (International code of Zoological Nomenclature. 1985, Charlesworth & Co, Huddersfield, p. 338), Odonata (ordo) = dragonflies, including both Zygoptera (= damselflies = small dragonflies) and Anisoptera (= great dragonflies).


ing high polarization sensitivity, and contain only two orthogonal microvillar orientations that can enhance sensitivity for E-vector contrasts. In Sympetrum striolatum, the optics of dorsal rim ommatidia is modified in a way that enlarges the visual field. This adaptation to the coarse spatial gradients of E-vector direction in the sky suppresses noise and enhances the information capacity of the polarization detection system. Since the dorsal rim ommatidia of several other insect species are known to be anatomically specialized in a similar way and to be responsible for the analysis of skylight polarization, the dorsal rim area of the dragonfly compound eyes may playa similar role. Although it is very probable that dragonflies may also exploit celestial polarization for orientation in their habitat and for control of flight direction during migration, there is no behavioural evidence for their polarization-based celestial orientation. The Odonata are a phylogenetically primitive insect order, thus, according to Meyer and Labhart (1993), the use of skylight polarization for navigation may have evolved early in insect phylogeny.

Spectral and polarization sensitivities of dark adapted retinula cells in the ventral retina of the dragonfly Hemicordulia tau were investigated by Laughlin (1976), and the structure of the ventral retina by Laughlin and McGinnes (1978). In the apposition compound eye of Hemicordulia tau there are eight retinula cells. However, only the so-called full-size cells contribute to the fused rhabdom. The ventral retina is tiered possessing a single distal typical cell Rl. The ventral distal ommatidium is delineated by the contribution of cell Rl to the rhabdom. Its rhabdomere occupies up to half the rhabdom cross section ne ar the cone tip, at deeper levels its contribution dwindles and deeper still, its microvilli are withdrawn. Rl is thought to be the polarization-sensitive UV cell in the ventral distal retina. It has either horizontal or vertical microvilli, and these two orientations occur with similar frequency. Hence, the subsystem Rl may be an orthogonal polarization analyser. The medial region of the ventral ommatidium is defined by the contributions of cells R2-R5 to the rhabdom. Rhabdomeres of cells R6 and R7, having almost the same alignment with nearly vertical microvilli, dominate in the ventral proximal rhabdom.

In the ventral retina of Hemicordulia tau the retinula cells form two distinct spectral sensitivity classes, single-pigment cells and linked-pigment cells. The former have narrow spectral sensitivity functions with a maximum either at 360,440 or 510 nm. They occur only in the distal retina. The UV and green cells are far more common than the blue cello The linked-pigment cells have broadened and often double peaked spectral sensitivity functions and at least three rhodopsins contribute to their response. The single-pigment blue and UV distal cells of the ventral ommatidia are sensitive to polarization and have high PS-values (PSuv = 7.1, PSb1ue > 3), whereas the single-pigment green cell and the linked-pigment cells are insensitive to polarization (Laughlin 1976). Since the UV cells are maximally sensitive to either horizontal or vertical E-vector, Laughlin (1976) hypothesized that the UV cells may compose a


two-channel orthogonal system to analyse polarization. Although the rarer single-pigment blue ceH in the ventral distal retina is polarization-sensitive, its role is not yet clear. In the dragonfly Hemianax papuensis, for instance, the distal blue ceH (with PSb1ue = 3-6) is commoner than the UV. This could mean that the blue ceH substitutes for the UV ceH in a proportion of ommatidia and its relative frequency is species-specific (Laughlin 1976).

The function of the polarization-sensitive cells in the ventral retina of Hemicordulia tau is not clear either. In the opinion of Laughlin (1976), these ceHs may form a polarization-sensitive horizon detector system for dragonflies flying over water. The UV-sensitive orthogonal subsystem Rl could be used for maintaining a horizontal orientation of the head maximizing the difference between their outputs. Alternatively, these cells could act as water detectors or help in contrast enhancement of territoriallandmarks. Until now, this hypothesis has not yet been behaviourally tested.

Note that, to our knowledge, Laughlin (1976) was the first who suggested that an animal could use the polarization sensitivity of its ventral eye region for water detection. This capability was later hypothesized also by Wolf et al. (1980) in fruit flies and demonstrated behaviourally by Schwind (1983a) in the backswimmer Notonecta glauca.

According to Laughlin (1976), the ceHs within the same fused rhabdom of the ventral retina in Hemicordulia tau have three different functions: (1) a trichromatic colour vision is based on the single-pigment UV, blue and green receptors of the distal retina. (2) Detection and analysis of the polarization of light reflected from water surfaces may be performed by the orthogonal UVsensitive subsystem RI also in the distal retina. (3) The linked-pigment ceHs in the medial and proximal retinae are high acuity contrast perceiving units.

In behavioural multiple-choice field experiments, Horvath et al. (1998a) and Wildermuth (1998) showed that dragonflies indeed recognize the water of mating and oviposition sites primarily by horizontaHy polarized reflected light. Since the results of Horvath et al. (1998a) are presented in Chap. 19, here we deal only with the findings of Wildermuth (1998), who performed multiple-choice experiments also with Somatochlora arctica (Wildermuth and Spinner 1991) and Aeshna juncea (Wildermuth 1993). He tested the attractiveness of horizontal reflecting surfaces - dark brown Plexiglas, shiny colourless and black plastic sheets, matt white and black cloths, aluminium foil and dry vegetation with different reflecting and polarizing properties - to the dragonfly species Coenagrion puella, Pyrrhosoma nymphulla, Aeshna juncea, Somatochlora alpestris, Cordulia aenea, Libellula depressa and Libellula quadrimaculata, differing in their ecological requirements. At one of the corners of the test surfaces vertical sticks were placed providing perches for the males of territorial Anisoptera species, which frequently settle on such perches at natural oviposition sites. Typical responses, such as lowering the flight altitude, hovering, circling and looping above the surfaces, scouring the edges, aggression, dipping movement, perching, patrol flight, copulation, egg-


laying, inspection flight, landing on the surface, surface touching with legs and wings, air fight of the different species to natural and dummy sites were recorded and compared.

In contrast to the automatie collection of attracted aquatie insects in the field experiments of Schwind (1991, 1995), during the choiee experiments of Wildermuth (1998) it was possible to observe the responses of dragonflies in their natural habitat in detail and to show that both sexes exhibit all elements of reproductive behaviour at the dummies. Analyzing and comparing the physieal properties of the dummies, temperature, odour, colour and intensity of reflected light could be excluded as cues explaining the observed high attractiveness of the brown Plexiglas and the shiny black plastie sheet. The differences in the reactions of dragonflies to the dummies could be explained only by polarotaxis. Horizontally polarized reflected light was the primary cue that attracted these dragonflies.

It is unknown in what spectral ranges dragonflies detect the horizontal polarization of light reflected from the water surface. We have already mentioned above that in the ventral eye region of Hemicordulia tau, the singlepigment blue and UV distal cells are highly polarization-sensitive (Laughlin 1976). On the other hand, only the red-sensitive photoreceptors Rl and R4 with horizontal mierovilli in the ventral eye region of the reddish Sympetrum rubicundulum were reported to be polarization-sensitive (Meinertzhagen et al. 1983). In dragonfly species differing in their habitat requirements, polarization sensitivity might operate in various spectral regions like in aquatie insects (Table 18.1). The species-specific visual habitat recognition by dragonflies is surely based not only on the polarization of reflected light, but also on colour and structural properties, e.g. vegetation of the water surface. In addition, the verdiet based on sight is confirmed by mechanieal tests with the abdominal tip (as shown in Somatochlora arctica byWildermuth and Spinner 1991) or the legs (as demonstrated in Perithemis mooma by Wildermuth 1992b). Thus, dragonflies approaching the mating or oviposition sites recognize the place by different visual and tactile cues, polarotaxis being an essential part of the process.

18.7 Dolichopodids

The specializations in the ventral retina of the long-legged fly Sympycnus lineatus (Doliehopodidae) are believed by Trujillo-Cenoz and Bernard (1972) to be vertieal polarizing filters for eliminating water surface glare, a possible hin dran ce in prey capture. As we have seen above, the same function of similar specializations in the ventral retina of waterstriders Gerris was suggested by Schneider and Langer (1969).


18.8 Mayflies Ephemeroptera

Schwind (1995) found that in a Cloeon species water detection happens in the blue (450-480 nm) part of the spectrum. Kriska et al. (1998) showed that also mayflies detect the water polarotactically. Their results are described in detail in Chap.22.

18.9 Other Polarotactic Water Insects

Schwind (1991) tested light polarized by reflection in the field for its attractiveness to some flying insects. He compared the attractiveness of different artificial reflecting surfaces, "water-dummies", i.e. water-covered horizontal glass panes underlain by sub strata, e.g. aluminium foil, coloured or colourless plastie sheets, matt black cloth. Insects dropping onto the pane were swept into an automatie collecting trap by continuously flowing water on the glass surface with defined spectral and reflection-polarizational characteristies. He showed that the horizontal polarization of reflected light attracts a variety of insects: (1) bugs living in water (Corixidae, Notonectidae, Pleidae), or on the water surface (Gerridae),or on the shore (Saldidae); (2) aquatie beetles (Hydrophilinae, Dytiscidae, Haliplidae, Hydraenidae); (3) beetles inhabiting moist substrata (Sphaeridiinae); (4) Chironomidae among other nematocerans.

Irrespective of its intensity, unpolarized reflected light did not attract any of these insects. The multiple-choice experiments of Schwind (1991) showed that glitte ring from water ripples is not necessary to attract these insects, and intensity gradient is not decisive in recognizing water. The latter was proven by test surfaces with matt black side walls, whieh restricted the visibility of the surfaces to a narrow field of view around the nadir. Polarization is the only factor to explain the attractiveness of light reflected from water surfaces to these insects, whieh do not displayasimple positive phototactie behaviour, and olfaction is also excluded as the main cue in their water detection. Certain insects tried to lay eggs onto the test surfaces. The reactions of these insects to the polarization of reflected light are variable. Schwind (1991) could distinguish three main response groups:

1. The first group is attracted whenever the degree of linear polarization p is high in the Uv, irrespectively of p in other parts of the spectrum and of colour or brightness of the background beneath the polarizing, reflecting surface. The polarization sensitivity of these insects operates in the UV. This type is able to detect water with a bright bottom as weH as dark waters, because the light returned by these water bodies is highly polarized in the UV.


2. The second insect group is attracted only by the reflecting surface over a dark background, where the reflected light of all wavelengths visible to insects is highly polarized. Here, there are two subsystems. If the insect polarization sensitivity operates at wavelengths Ion ger than 500 nm, then the POL-system is advantageous to insects living in deeper turbid ponds, where the intensity of UV light is low. With this visual system they only find the appropriately dark waters attractive. If the POL-system operates in the short wavelength range and different receptors react to light with wavelengths longer than 500 nm, then too high intensity of reflected light causes avoidance of water. In some species seasonal changes in their preference were observed. Helophorus griseus, for example, behaves in spring and early summer like a member of the first group and prefers brighter waters, perhaps for feeding and breeding. In autumn it behaves as a member of the second group and prefers dark ponds, whieh are usually deep enough not to freeze down to the bottom. The Sphaeridiinae inhabit moist sub strata and belong to the first group, while the Hydrophilinae live in water and belong to the second group. The former have to recognize shiny polarizing surfaces on a bright background, the latter do not. To detect their habitat, the former must have a POL-system operating in the UV, but this need not necessarily be the case with the latter.

3. The third insect group is between the above two extreme types. Here there are again two subsystems. The POL-system operates either in the blue (400 nm < A < 500 nm), or in the Uv, and water is avoided if the light intensity is too high in the blue.

In further multiple-choiee experiments, Schwind (1995) determined the spectral regions in whieh certain aquatie insects perceive the polarization of reflected light (Table 18.1). He compared the attractiveness of different water-imitating glass reflectors to flying aquatie insects in the field. p of light reflected from these water-dummies was reduced selectively in different narrow wavelength ranges from UV to red in such a way that the intensity of light reflected from the underlying sub strata was enhanced in different spectral ranges. In the red, yellow or the whole visible part of the spectrum the glass pane was underlain by red, yellow or white substratum. In the blue or UV a UV-transmitting Plexiglas pane was underlain by a coloured glass transmitting only blue or UV light, and these two glass layers were underlain by aluminium foil reflecting diffusely in both the UV and visible spectral ranges.

The degree of attractiveness was determined by the number of individuals landing on the water-dummies, swept by the continuously flowing water film on their surface and trapped by the automatie collector. The attractiveness of the test surfaces to a given insect species was then compared with the calculated wavelength-dependent effective degree of linear polarization Pefj and intensity Iefjof reflected light as well as the effective reflectance Reffof the sur-


faces as perceived by receptor systems with broad-band absorption curves. Effective curves PelA}, IelA}, RelA} were derived from the measured functions p(A}, I(A}, R(A} convoluted by the receptors' absorption curves. Since there was no region of the spectrum in whieh the attractiveness of the test surfaces varied according to IelA} or RelA}, Schwind (1995) conc1uded that the trapped insects (Table 18.1) recognize water surfaces predominantly not by the intensity of reflected light. There was always one partieular wavelength A * at whieh the attractiveness of all test surfaces corresponded approximately to PelA*} above a threshold value T= 35 %. From this Schwind conc1uded that the attracted insects (Table 18.1) detect water mainly polarotactieally at A *, at whieh the attractiveness of the test surfaces matched best the value of

PelA*}-T. Schwind (1995) found that various aquatie insect species perceive the

polarization of light reflected from water in diverse parts of the spectrum ranging from the UV to the yellow-green (Table 18.1). The visual ecologieal significance of this diversity are the following:

• Figure 18.6 demonstrates that in limnetie habitats P of reflected light increases with decreasing wavelength, thus it is highest in the Uv, while in dark waters even the reflected light at long wavelengths is strongly polarized. Thus, the polarization of light reflected from water bodies with a bright subsurface can decrease so severely toward the longer wavelengths that from a certain wavelength on it cannot be detected polarotactieally. Insects with polarization sensitivity in the UV can detect many different optieal types of water reflecting maximally polarized light always in the UV. Shallow waters with a very light bottom, for example, can only be detected polarotactieally with such a UV-POL visual system. However, species with a UV-POL system differ in their preferences for optieally different habitats. The Sigara species and Helophorus griseus, for instance, accept both dark and bright waters, while Helophorus aquaticus, Helophorus flavipes and Hydrobius fuscipes avoid bright waters. Selection among the optieally different habitats is independent of the UV-POL system and is mediated by receptors that operate at Ion ger wavelengths.

• POL-systems operating at wavelengths longer than 400 nm are not adapted to detect the most highly polarized component of light reflected by water, but to the spectral composition of the underwater and/or abovewater light fields: Agabus bipustulatus (480 nm < A * < 520 nm), Neohaliplus lineatocollis and Haliplinus lineolatus (530 nm < A * < 550 nm), for example, would avoid highly eutrophie waters, because they cannot perceive the very weak polarization oflight reflected from them (see Figs. 18.6 and 20.3). To detect such eutrophie waters a PO L-system with A * < 470 nm would be necessary. Light with Ion ger wavelengths is attenuated in water less than light with shorter wavelengths (Lythgoe 1979; Gates 1980; Schwind 1995). Thus, for insects not restrieted to the uppermost water levels a shift of A * of their


Fig. 18.6. Degree of linear polarization p of light reflected at the Brewster angle from a pond with clear water and dark bottom (dark water) and an eutrophie green-brown pond (green water) versus the wavelength A (after Schwind 1995).

Cl,

§ '.g N .~

"0

+50%

horizontal E-vector

~ 0%

" :5 vertical '8 E-vector

" ~ -50% wavelength A (nm) ~+=~--~--~~--~--~~--4

300 400 500 600 700

POL-system toward longer wavelengths could be an adaptation to the higher underwater light intensity at longer wavelengths, if the receptors for polarotactic water detection can also function underwater, at least as intensity detectors. Waterstriders, e.g. Gerris lacustris (360 nm < A * < 370 nm) or mayflies, e.g. Cloeon sp. (450 nm < A* < 480 nm) do not submerge themselves into water, thus their POL-system should be adapted to the aeriallterrestriallight fields.

Ecologists use efficient traps built around glass panes slightly tilted from the horizontal and underlain by different substrata to monitor the flight activity of aquatic insects (e.g. Landin 1968). The efficiency of these traps is due to the horizontal polarization of reflected light. There are numerous observations of water insects being deceived by artificial shiny surfaces such as glass panes, car roofs or wet asphalt streets (e.g. Popharn 1964; see also Chaps. 19, 21-23). This phenomenon can be explained also by the attractiveness of the horizontal polarization of reflected light to polarotactic insects. Ecologists frequently use coloured dishes filled with colourless fluid to test colour preferences of insects (e.g. Kirk 1984). According to Schwind (1991), perhaps some of the insects caught in these traps are not attracted by colour, but by the polarization of reflected light, the degree of polarization of which is high in wavelength ranges within which the coloured dish does not reflect.

18.10 Insects Living on Moist Substrata or Dung

Schwind (1991) observed that certain insects living on moist sub strata or dung detect polarization of reflected light in a similar way to the water bug Notanecta glauca (Schwind 1985b) and many other water insects (Schwind 1995). The beetles Megasternum boletophagum, Cryptopleurum minutum and aCereyon species of the subfamily Sphaeridiinae of the family Hydrophilidae were also attracted by the polarization of reflected light. Sphaeridiinae inhabit


moist substrata such as plant debris and cow-dung. The polarization sensitivity of these insects operates in the UV; they belong to the response group 1 defined by Schwind (1991).

Gal (1997) demonstrated how strong the reflection polarization of fresh cow-dung can be under different illumination conditions in the field. The dung in Fig. 18.7 had a similar brightness and colour as the ground of the grassy surrounding. However, the degree p and angle a of linear polarization of light reflected by the dung differed from those reflected by the ground, especially in the blue because of the blueness of the incident skylight. p of light reflected specularly from fresh dung is high er than that from the background with a rough surface. Furthermore, the distribution of a of fresh dung is more homogeneous than that of the rough background. However, p of dung decreases versus time as the moisture is gradually lost (Horvath and Gal 1997, unpublished data). Dry dung possesses a rough surface reflecting light diffusely, thus p is low and the distribution of a is heterogeneous. Thus, dry dung has similar reflection-polarizational characteristics as the surroundings.

total radiance I

red (650 nm)

0%

degree of polarizat ion p

100%

green (550 nm)

O·

-90· +90·

180· angle of polarization (X

measured from the vertical

blue (450 nm)

Fig.IS.7. Reflection-polarization patterns of a sunlit greenish-brown fresh cow-dung in a meadow measured by video polarimetry in the red, green and blue under a clear sky (after Gal 1997).


Dung insects prefer fresh dung, because they (1) can feed on it, suck its liquid components or liek its wet surface, (2) can lay their eggs easily in the wet and soft substrate, where (3) their larvae can develop before the dung becomes dry and hard. Thus, they must find fresh dung as soon as possible. From a remote distance the smell of dung, even if it is quite fresh, may not be intensive enough to attract a large number of insects. In this case, optieal cues are more effective for the explanation of the observed high attractiveness of fresh dung to certain insects. Fresh dung can be detected on the basis of the strong polarization oflight reflected by its wet surface (Fig. 18.7).

18.11 Mosquitoes

The polarization sensitivity of mosquitoes is controversial. In a laboratory experiment Kalmus (1958) investigated the rotatory optomotor re action of three male adult Aedes aegypti under an oscillating horizontal sheet of linear polarizer. The mosquitoes were enclosed between two watchglasses, and their optieal surroundings were composed of (1) a vertieal black cylinder covered by a linear polarizer illuminated from above by a light bulb and (2) a horizontal black annular diaphragm shielding the insects from stray light. The rotation of the transmission axis of the polarizer induced clear optomotor response only if the hollow of the lower watchglass was coated with black gum strip, whereas with a light background either no optomotor reaction or only a weak one occurred. Kalmus concluded that the observed optomotor response in mosquitoes was not induced by the rotation of the E-vector of the downwelling linearly polarized light directly perceived, but by the brightness pattern of polarized light reflected from the dark background and/or substratum, which pattern co-rotated with the overhead polarizer (see also Chap. 34).

In a qualitative and very subjective field studyWellington (l974a) observed the diurnal activity of adult Aedes and Culex mosquitoes. He found that resting mosquitoes attacked as soon as the ob server approached them closely, whether or not the zenith sky was clear or clouded independently of the solar elevation. In contrast, mosquitoes made long, roving flights only when the zenith sky was clear and the solar elevation was low. Roving flight stopped whenever clouds passed through the zenith sky or the solar zenith angle was narrow. According to Wellington, these changes in mosquito flight might be induced by the change of zenith polarization, since clouds at the zenith or narrow solar zenith angles are associated with the disruption of zenith polarization, while clear zenith sky at wide solar zenith angles me an strong zenith polarization. He suggested that the polarization of the zenith sky could be utilized by mosquitoes traveling to and from fee ding or oviposition sites near sunrise and sunset. However, from these observations ofWellington the polar-


ization sensitivity of mosquitoes cannot be deduced, because it is unclear whether the behaviour of the insects was governed by the change of the celestial pattern of intensity, or colour or polarization at the zenith.

Hence, it remained to be investigated whether the dorsal eye region of adult mosquitoes can perceive the skylight polarization and whether celestial polarization is used for orientation. Since mosquito larvae develop in water, it would be interesting to study, whether adult mosquitoes detect the water bodies polarotactically, and if underwater polarization plays any role in the swimming patterns of mosquito larvae.

19 Multiple-Choke Experiments on Dragonfly Polarotaxis

Kennedy (1917) gave an account of many individuals of the dragonfly Anax junius having been killed as a result of mistaking an open surface of crude oil for water. Puschnig (1926), Fraser (1936) and Whitehouse (1941) reported that dragonflies Ophiogomphus forcipatus, Ictinogomphus ferox, Macromia magnifica and several species of Chlorogomphus patrolled along asphalt roads instead of rivers and showed a typical water-touching behaviour above the asphalt surface. Kennedy (1938) cited cases in which dragonflies were attracted to pools of petroleum. Horvath and Zeil (1996) reported that dragonflies were deceived, attracted and trapped in large numbers by crude oil lakes in the desert of Kuwait. Horvath et al. (1998a) observed the same behaviour of dragonflies at a waste oillake in Budapest (Fig. 19.1).

Muller (1937) observed the females of Orthetrum dragonflies laying eggs on a shiny cement floor and Copera marginipes made repeated egg-laying movements in a dirty seam on a shiny black bench. Wyniger (1955) reported on the egg-laying of Libellula depressa onto a glass pane of a greenhouse. Neville (1960) experienced that mature individuals of Pantala flavescens performed sexual behaviour and oviposition movements over shiny roofs of tents. Kennedy (1938) reported on instances in which dragonflies were attracted to shiny roofs of automobiles.

Such examples demonstrate that dragonflies1 respond to shiny surfaces, and also that their response is elicited by particular misleading cues. Horvath and Zeil (1996) suggested that the reason why crude oil deceives, lures and traps insects on a large scale might be that an oil surface looks like an "exaggerated", strongly horizontally polarizing water surface, making oil visually more attractive than water to water-loving insects, the visual system of wh ich is sensitive to the polarization of reflected light. Horvath et al. (1998a) have tested and supported this hypothesis in multiple-choice field experiments with dragonflies. They compared the numbers of dragonflies being caught in water, crude oil (Fig. 19.2) and salad-oil traps with different reflection-polar-

1 Dragonflies: all members of üdonata, including both Anisoptera and Zygopteraj the latter commonly known as damselflies.


Fig.19.1. Carcasses of a copulating pair of Sympetrum sanguineum (left) and a pigeon (right) trapped by the waste oi! lake in Budapest. (After Horv<ith et al. 1998a)

izational characteristics. They showed that polarotaxis is the most important mechanism which guides dragonflies during their habitat choice and oviposition site selection, and this is the reason why dragonflies can be deceived by, and attracted to crude and waste oil, tar or asphalt.

In the first choice experiment of Horvath et al. (1998a), two trays were filled with water, respectively black crude oil. They were placed on a large field about 500 m away from a smalliake. The bottom of the water-filled tray was covered by a thin layer of grey, sandy soil to imitate the typical bottom of alkaline puddies in the biotope. In order to trap all insects that touched the water, the common ecological method of catching and monitoring insects was used (Southwood 1966): the surface tension of water was reduced with a detergent. The position and orientation of the trays was changed randomly. The dragonflies trapped by the trays were collected and identified. Male dragonflies were trapped about twice as frequently as females, and black crude oil was significantly more attractive than water on a light grey background. This observation constituted the experimental evidence for the hypothesis put forward by Horvath and Zei! (1996). The light reflected from the oil-trap had a degree of linear polarization p = 33 % with horizontal E-vector, while the light returned by the water-trap had p = 4 % with vertical E-vector (Fig. 19.2). Although the reflection-polarizational characteristics of the trays depend on the angle of view, solar zenith angle and meteorological conditions, Fig. 19.2 demonstrates well the fact that crude oil is a more effective polarizer than water with a bright bottom, even relatively far away from the Brewster angle (570 from the vertical for crude oil and 530 for water). Thus, the light reflected from crude oil is a supernormally polarized stimulus for water-seeking polarotactic dragonflies.

In the second choice experiment of Horvath et al. (1998a), five plastic trays were filled with transparent, yellowish salad-oil. The bottom of four traps was covered by shiny plastic sheets with different shades ranging from black, through dark and light grey to white. The bottom of the fifth trap was a shiny aluminium foil. Although the black trap reflected about twice so highly polarized light (p = 69.6 %) as the dark grey trap (p = 33.7 %), there was no significant difference between the number of dragonflies trapped by them (Table

19 Multiple-Choice Experiments on Dragonfly Polarotaxis 201

A radiance I B degrec 01' polarizalion p 0 -1.

o w

\yatcrsurr~cc -........................................................ .

_ erude oil $urface : -- / - -- -

....... ,. -

c angle of polarizalion IX D Fig.19.2. The refleetion-polarizational eharaeteristies of the two trays used by Horvath et al. (l998a) in their first ehoiee experiment, filled with blaek erude oil (0) and detergent -treated water (W) measured by video polarimetry at 450 nm under a clear sky. A Distribution of the radianee 1. B Pattern of the degree of linear polarization p. C Pattern of the angle of polarization a. (BZack vertieal E-veetor, white horizontal E-veetor). D Histogram of p ealculated for the surface area of the two trays. The Zelt (p = 4 %) and right (p = 33 %) peak of the distribution eorresponds to the water and erude oil surfaee, respeetively. Viewing direetion was 70° from the vertieal and at a right angle to the solar meridian. (After Horvath et al. 1998a).

"19.1). The light grey and white traps with low p trapped significantly fewer dragonflies and the aluminium trap with very variable direction of polarization was the least attractive. Males were trapped again much more frequently than females.

Similarly to Horvath et al. (1998a), Wildermuth (1993) also observed that dragonflies (Aeshna juncea) do not land when the test surface composed of black plastic sheet or cloth heated up too much in full sunshine. To exclude the role of temperature and radiance of the surfaces in the choice of dragonflies, Horvath et al. (unpublished) performed a third choice experiment, in which six sheets (1 m 2) covered by different colourless materials were laid on the ground (Table 19.2). The surfaces were positioned along a rectangle, and their order was changed randomly. The different response types (air fight,


Table 19.1. Row 2 The total number (J juvenile) and sex (F fern ale, M male) of dragonflies (Ischnura elegans, Erythromma viridulum, Lestes macrostigma, Enallagma cyathigerum, Orthetrum cancellatum, Libellula quadrimaculata, Sympetrum sanguineum) trapped by the salad-oil-filled trays during the second choice experiment of Horvath et al. (I998a). Rows 3-5 The relative radiance I, degree of linear polarization p and direction of polarization of light reflected from the trays and measured by video polarimetry at 450 nm from a direction of view of 70° from the vertical. (After Horvath et al. 1998a).

Tray Black Darkgrey Light grey White Aluminium

Number of trapped 8 F (5 J) + 12 F (3 J) + 11 F (7 J) + 7 F (5 J) + 5F(4J)+ dragonflies 62 M (7 J) 63 M (9 J) 32 M (3 J) 21 M (2 J) 10 M (2 J)

Relative radiance I 22.1 % 35.6 % 52.4 % 100% 42.3 % (variable)

Degree of linear 69.6 % 33.7 % 10.3 % 4.6 % 20.2 % (variable) polarization p

E-vector direction Horizontal Horizontal Horizontal Vertical Horizontal (variable)

Table 19.2. The total number and the frequencies relative to the total number of a given re action counted above all six test surfaces of the different behaviour types of dragonflies (Ischnura elegans, Erythromma vividulum, Lestes macrostigma, Enallagma cyathiqerum, Orthetrum cancellatum, Sympetrum sanquimum) in the third choice experiment of Horvath et al. (unpublished) repeated 11 times. The relative radiance, degree of linear polarization and angle of polarization with respect to the vertical of the light reflected from the test surfaces and measured by video polarimetry at 470 nm from a direction of view of 70° relative to the vertical.

Behaviour type

Air fight, hovering, protection

Surface touching

Settling down

Egg-Iaying

Relative radiance

Degree of polarization

Angle of polarization

Shiny black plastic

922

(40.1 %)

343 (46.3 %)

5 (6.3 %)

Shiny grey plastic

486 (21.2 %)

190 (25.6%)

9

(11.4%)

Matt white cloth

189 (8.2%)

22 (3%)

30 (38%)

Shiny white plastic

397 (17.3 %)

120

Matt black cloth

91 (4.0%)

13

Shiny aluminium

212 (9.2%)

53 (16.2%) (1.8%) (7.1 %)

20 14 (25.3%) (1.3%) (17.7%)

4 8 0 4 0 4 (20 %) (40 %) (20 %) (20 %)

52.9 % 63.5 % 99.7 % 100 % 22.2 % 78.8 % variable

73.5% 9.1 % 3.4% 2.8% 21.5% 7.0% variable

90° 90° 51.3° 0° 51.6° 80.3° horizontal horizontal vertical variable


hovering and protection of the territory against intruders; surface touching; settling down; egg-Iaying) of flying dragonflies to these test surfaces were visually observed and their relative frequencies were counted (Table 19.2). The number of egg-Iaying was very low, thus one cannot draw any firm conclusion considering the oviposition. Egg-Iaying occurred only on the shiny surfaces. Settling down2 occurred also only sporadically, but significantly more often than egg-Iaying. For settling down the brighter test surfaces, matt white cloth, shiny white plastic, shiny aluminium with low p were preferred independently of their shiny or matt appearance and of the angle of polarization of reflected light.

Surface touching was the second more frequent behaviour type, wh ich is typical above water surfaces when dragonflies inspect the surface to select the optimal oviposition site. The shiny black plastic sheet with the highest degree (p = 73.5%) ofhorizontal polarization was the most attractive (46.3%). The shiny grey plastic sheet with a much lower degree (p = 9.1 %) of horizontal polarization was less attractive (25.6 %). Since the direction of polarization was not horizontal, the matt white and black cloths, independently of their low or high p, and the shiny aluminium were practically unattractive.

The most frequently observed types of behaviour were the air fight, hovering and protection, which are typical again only above water surfaces3• Considering these aerial territorial behaviours, the shiny black plastic sheet was again the most attractive (40.1 %). The shiny grey plastic sheet was less attractive (21.2 %), the matt white and black cloths as weIl as the shiny aluminium were practically unattractive, while the shiny white plastic sheet was only slightly attractive (I 7.3 %). Hence, for air fight, hovering and proteetion the relative attractiveness of the different test surfaces was similar to that obtained for the surface touching behaviour. These reactions demonstrate that the high er the degree of horizontal polarization, the greater is its attractiveness to dragonflies.

Analysing Table 19.2, we can establish that the radi an ce I does not play an important role in the choice of surfaces. The dragonflies were attracted predominantly to the shiny black plastic sheet, and the very dark matt black and

2 This reaction recaIls the settling down behaviour onto the matt light grey ground observed frequently in the study site. This behaviour is apparently different from the water-seeking behaviour, because dragonflies never settle down directly onto the water surface. According to Corbet (1999), settling down to a brighter surface may serve thermoregulation or simply resting. Many dragonfly species must warm their thorax to a temperature weIl above the ambient temperature before they can fly, and they do this by behavioural and physiological means. Behavioural warming occurs as basking in sunlight, sometimes gaining additional heat by sitting on hot rocks or the ground (Corbet 1999).

3 Males frequently hover in their territories, or when they search for females elsewhere. Hovering serves to advertise the presence of a male in his territory. Females also hover when they inspect oviposition sites (Corbet 1999). Both females and males inspect the surface, or males protect their territory during air fights against intruders.


too bright white and aluminium surfaces were unattractive. The preferred shiny black plastic sheet reflected highly and horizontally polarized light. The matt test surfaces scattered light diffusely and the reflected light was practically unpolarized. Thus, the strong reflection polarization of light remains as the only explanation for the fact that dragonflies preferred the black plastic surface exclusively. This conclusion is in agreement with the results of Wildermuth and Spinner (1991) and Wildermuth (1993, 1998).

Since the smell of all salad-oil traps was the same and the test surfaces used in the third choice experiment were odourless, one can conclude that olfaction is not relevant for detection of water by dragonflies. Because the temperature of the black shiny plastic and black matt cloth, like the temperature of the white plastic and white cloth, were approximately the same, the temperature was irrelevant in the choice of dragonflies. It can also be established that the radiance of reflected light did not play an important role in their choice, as otherwise either the darkest matt black cloth, the white cloth, or the brightest aluminium would have been the most attractive.

To confirm that strongly polarized reflected light is very attractive to waterseeking dragonflies, Horvath et al. (1998a) performed a fourth field experiment. Half of a shiny aluminium test surface(0.2 m2) was covered by a common linearly polarizing filter, while the other half was uncovered. The two halves were separated by a narrow matt black cloth, wh ich was unattractive. The polarizing filter was neutral grey with a transmissivity of 80 % in the visible range of the spectrum and absorbed entirely UV light. The number of the different behaviour types above the two different halves of the aluminium test

Table 19.3. Rows 1-4 The total number of the different behaviour types of dragonflies (Ischnura elegans, Erythromma viridulum, Lestes macrostigma, Enallagma cyathigerum, Orthetrum cancellatum, Sympetrum sanguineum) above the two test surfaces in the fourth choice experiment ofHorvlith et al. (1998a) repeated five times. Rows 5-7 The relative radiance, degree of linear polarization and angle of polarization with respect to the vertical of the light reflected from the test surfaces and measured by video polarimetry at 450 nm from a direction of view of 70° relative to the vertical. (After Horvath et al. 1998a).

Behaviour type

1 Air fight and hovering 2 Surface touching 3 Settling down 4 Egg-Iaying

Optical characteristics 5 Relative radiance 6 Degree of linear polarization 7 Angle of polarization

Polarizing filter on shinyaluminium

412 (59.6 %) 115 (85.8 %) 6 2

40% 100% 90° (horizontal)

Shiny aluminium

279 (40.4%) 19 (14.2 %) 3

100% 30 % (variable) 65° (variable)


surface have been counted and compared (Table 19.3). The amount of egg-laying and settling down was very low on both halves. About 60 % of the most frequent reactions, the air fight and hovering, happened above the polarizing filter, which thus was not significantly more attractive than the aluminium surface. The reason for this is that the territories of the observed small dragonflies are usually about 1 m2 (Corbet 1999). Thus, both halves of the aluminium belonged to their territory. About 86 % of the surface touching, the second most frequent reaction, happened on the polarizing filter. This difference is highly significant and demonstrates that both males and females select strongly horizontally polarizing surfaces as habitats and oviposition sites.

On the basis of the above results, one can conclude that dragonflies detect water by means of the horizontal polarization of light reflected from water (polarotaxis), like many other water insects. In the choice experiments of Horvath et al. (1998a), chiefly males were trapped. Wildermuth and Spinner (1991) and Wildermuth (1993) have also observed that female dragonflies visited black, shiny plastic sheets and natural oviposition sites less frequently. This can be explained by the operational sex ratio at breeding sites which is strongly biased towards males, because females spend much time elsewhere (Corbet 1999).

20 How Can Dragonflies Discern Bright and Dark Waters from a Distance? The Degree of Linear Polarization of Reflected Light as a Possible Cue for Dragonfly Habitat Selection

While monitoring the dragonfly fauna of dark ponds in peatland and bright ponds in gravel pits, Bernath et al. (2002) observed that eertain species prefer exdusively bright water bodies, while other species prefer only dark waters, and some species are ubiquitous, ehoosing dark and bright waters with equal frequencies (Table 20.1). It is a well-known optical phenomenon that two water bodies, being bright and dark to the human eye viewing downwards perpendicularly to their surfaee, eannot be distinguished from eaeh other from a distanee. Then the angle of view with respeet to the water surfaee is very narrow ("grazing" angle of view) , and the amounts of refleeted light are practically equal for both dark and bright waters. How can dragonflies distinguish a bright from a dark pond before they get sufficiently elose to pereeive brightness differenees?

Sinee many dragonfly species find their aquatic habitat by polarotaxis (Horvath et al. 1998a; Wildermuth 1998; Bernath et al. 2001b), one ean hypothesize that eertain dragonflies ean seleet from far away their preferred dark or bright water bodies, at least partlyon the basis of refleetion polarization. In order to test this hypothesis, Berncith et al. (2002) investigated the refleetion-polarizational eharacteristies of a number of dark and bright ponds inhabited by different dragonfly species (Figs. 20.1-20.3,20.5-20.8). Their field studies were earried out at two loealities: (1) a former gravel pit, and (2) a moorland area with former peat diggings; the two sites being situated 22 km apart from each other. The Odonata fauna was monitored in a sampie of six ponds at eaeh loeality. The water bodies at study site 1 appeared bright to the human eye viewing downwards perpendicularly to their surfaee, those at site 2 dark. "Bright" means shallow and dear water with a bright substratum, "dark" refers to shallow and elear water with a dark substratum, from which light refleetion is limited. The bright ponds had a diameter of 5-10 m and a maximum depth of 0.3-0.4 m. Their surfaee was sparsely eovered by aquatic vegetation. The eolour was bright beige and the bottom eonsisted of gravel and day. All dark ponds were situated in peatland. Their diameter

20 How can Dragonflies Discern Bright and Dark Waters from a Distance? 207

Table 20.1. Dragonflies inhabiting bright and/or dark ponds as adults and/or larvae. (After Bernath et al. 2002).

Enallagma cyathigerum Anax imperator Libellula depressa Orthetrum cancellatum Orthetrum brunneum

Pyrrhosoma nymphula Coenagrion puella Aeshna cyanea Libellula quadrimaculata Sympetrum striolatum

Lestes virens Lestes sponsa Lestes viridis Coenagrion pulchelIum Aeshna juncea Cordulia aenea Somatochlora flavomaculata Leucorrhinia pectoralis Sympetrum sanguineum

Six bright ponds in gravel pits Adults Exuviae

++ ++ ++ ++ ++ ++ ++ ++ ++ ++

+ + ++ ++ ++ ++ ++ ++ ++ ++ - -- -- -- -- -- -- -- -

- -

Six dark ponds in peatland Adults Exuviae

(+ ) -+ (+) (+) -- -- -

+ + ++ ++ ++ ++ ++ ++ + +

++ ++ ++ ++ + + + + + + + + + + ++ ++ + +

Abundance classes: ++ = common, + = regular, (+) = sparse, - = absent.

ranged from 4 to 8 m and they had a maximum depth of 0.4-0.8 m. The ponds were sparsely or moderately overgrown with emergent vegetation. Their colour was dark brown and the substratum consisted of peaty mud.

In the visible (red, green, blue) part of the spectrum there were no significant differences in radiance I between dark and bright water bodies towards the sun (Fig. 20.IA). The same was true in the UV away from the sun and perpendicularly to it (Fig. 20.2A,B). In the green and red, however, I of light reflected from bright waters was significantly higher than that reflected from dark ponds away from the sun (Fig. 20.IB) and perpendicularly to it (Fig.20.IC).

The degree of linear polarization p of light coming from bright or dark waters was the highest in the blue for any direction of view (Figs. 20.ID-F, 20.2C,D). Independently of the wavelength and the viewing direction, p of light reflected from dark water bodies was significantly high er than that from bright waters (Figs. 20.ID-F, 20.2C,D). The differences were smallest in the blue.

Independently of the wavelength as weH as the viewing direction, the average direction of polarization of light reflected by waters was generally hori-

blu

e to

war

ds t

he s

lIn

~ :6·~

nm)

I :~

~

20

10

gree

n (5

50 n

m)

red

(65

0n

m)

ee ~t?

A

NS

NS

NS

O~'~~------~------~

du

k

brig

hl

dark

br

iJ!:hl

dn

rk

brig

hl

rloQ'~

~ oD $ C

l ~

20

~

gree

n re

d

o I

S·

S"

· S

·"

bluc

gr

een

re<l

bluc

g NS

blue

~ s·

· bluc

away

fro

m t

hc s

un

perp

endi

clIl

ar

to t

he s

un

gree

n re

d bl

uo

gree

n re<

l

9~ +

Q ~O

9~

~~

B

C

s· S

· N

S S

· S··

gree

n re<

l bl

uc

gree

n re

d

E

F

~ ~

BQ )~

s •

• $ s

•• ~

:.g s·

·

gn:c

n re

d 1

blu"

gr

een

red

.. ]

10

0·

g li

"':;;:

:.. 90

0-,~

.s ..!

::

SO

g. !~

~ 'l'!

70·

'iD~

a ~

fH

E

:~~G8

~6 ~~

~-~6 :

~~I~~

",O~

sO"

J., _

__

__

__

__

__

_

Fig

.20.

1. M

eans

(ho

rizo

ntal

bar

s in

box

es),

qua

rtil

es (

boxe

s), 5

th a

nd

95t

h pe

rcen

tile

val

ues

(ver

ti

cal

bars

pro

trud

ing

fram

box

es)

of

the

rela

tive

rad

ianc

e I,

deg

ree

of

line

ar p

olar

izat

ion

p an

d an

gle

of p

olar

izat

ion

a w

ith

resp

ect

to t

he v

erti

cal

mea

sure

d b

y v

ideo

pol

arim

etry

at

sunl

it s

ur

face

s o

f six

bri

ght a

nd s

ix d

ark

Swis

s po

nds

in th

ree

diff

eren

t vie

win

g di

rect

ions

: tow

ards

the

sun,

aw

ay f

rom

the

sun

an

d p

erpe

ndic

ular

to t

he s

un a

t 65

0, 5

50 a

nd

450

nm

. Dat

a fo

r d

ark

or

brig

ht

pond

s ar

e sh

own

by d

ark

grey

or

whi

te b

oxes

, re

spec

tive

ly. t

-tes

t (S

achs

197

4) w

as u

sed,

S*

p <

0.

05, S

** P

< 0

.01,

S**

* P

< 0

.001

, NS

not s

igni

fica

nt. (

Aft

er B

erna

th e

t al.

2002

).

N

0 00

'"t:I

P' ... .... ..... ~ ~

Pi"' ... N·

'" P- t"'

ciQ. g- 5·

:;. i:l S· e:..

S CI> o· i:l


Fig. 20.2. I, P and a in two different viewing directions away from the sun and perpendicular to the sun at 360 nm. Other conventions as in Fig. 20.1. (After Bern.lth et al. 2002).

75 away from the sun - 70 ,,~

g 'l§ 65 A '" :>

t '5 c 60 ~ ~ 55 > .-.= -e 50 .. .. e~ 45

S 40

dark bright

55 ~ 50 ? ~ ~ 45

C g.2 40

~ß 35 } .", 'C 30 ..

&. 25 ... 20

dark bright

.. 94· I:! .;1 I .~ ~ 90°·. ···············1··· ;; C;;I

.~ -5 86° .. '" &.,g 82' E

~ ~ 78° Co~ a 3 74°L..-__ N_S __

E dark bright

perpendicular to the stln

B

t t S

dark bright

? D

S··· ? dark bright

F f S

dark bright

zontal for both bright and dark waters (Figs. 20.1G-I, 20.2E,F). However, its standard deviation was small towards (Fig. 20.1G) and away from the sun (Figs. 20.1 H, 20.2E), while it was large perpendicularly to it (Figs. 20.11, 20.2F). The direction of polarization of light reflected from bright water changed from horizontal to vertical fra m the shorter wavelengths towards the langer ones (Fig. 20.3), since the amount of vertically polarized light emanating from the subsurface overwhelmed the amount of horizontally polarized surfacereflected light (Fig. 20.4) for longer wavelengths. A similar change in the direction of polarization did not occur in the case of dark water bodies (Fig.20.5).

Shadows also have a considerable effect on the reflection-polarizational characteristics of water bodies. In the case of dark waters, the horizontally polarized surface-reflected light always dominates, and thus the direction of polarization is always horizontal for both the shaded and sunlit regions (right column in Fig. 20.5). The middle column in Fig. 20.5 shows that p of light reflected from the shaded regions of dark waters is lower than that reflected from the sunlit regions, because in the shaded areas the amount of horizontally polarized surface-reflected light is reduced. Figure 20.6 shows an exampIe for a bright pond, in the sunlit or shaded regions of which the E-vector is horizontal or vertical, and p is higher or lower, respectively. Figure 20.7 presents another bright pond, where the contrasts of a and p are contrary to


>_---:" • ..t;;:-...... . " , .. ~.' ~.~ . :' '. '.

~- . ~ . -"; ... -\...... .......

100

,~ . ,

- .~.,. ,

~,

- ~ .. ~. ,

100 .... - .' --- . ycllow

OL-__ --'-____ ~~~~!!!.-------

'1=& .~

100

uhraviolel

400 550 700 wavclcnglh J.. (nm) radiance I

lransmiuivily curvcs oflhe colour fihers

degree of polarization p angle of polarization a. measured from the vertical

0% <r +45'

p +90' (J.

100"10 180"

Fig.20.3. Transmittivity T( A) versus wavelength A of the eolour filters used in the videopolarimetrie measurements (left column) and refleetion-polarization patterns (relative radianee I, degree of linear polarization p and angle of polarization a from the vertieal) of light refleeted from a bright pond in a gravel pit measured by video polarimetry in seven different ranges of the speetrum perpendicular to the sun with a viewing angle of 20° from the horizontal. (After Bernath et al. 2002).


Fig.20.4. Schematic representation of the reflection polarization of unpolarized incident light at a body of water. The returned light has two components: the partially horizontally polarized light reflected from the water surface, and the partially vertically polarized refracted light coming from the water.

those in Fig. 20.6. In Fig. 20.7 in the sunlit or shady regions of the bright pond the E-vector is vertical or horizontal, and pis lower or higher, respectively. p of bright water bodies is always much less than that of dark waters.

The reflection-polarizational characteristics of dark and bright water bodies are also influenced by the roughness of the water surface. Under windy conditions the water surface undulates, which more or less distorts the reflection-polarization patterns as seen in Fig. 20.8.

Bernath et al. (2002) conduded that from long distances, Le. at grazing angles of view, dark waters cannot be distinguished from bright ones on the basis of the radi an ce or the direction of polarization of reflected light. However, even at grazing angles of view, dark waters reflect light with a significantly higher degree of linear polarization than bright waters in both the visible and UV spectrum and in any direction of view with respect to the sun. It does not contradict the fact that the human visual system discriminates between dark and bright waters by intensity differences. However, this distinction can be done only if the water bodies are relatively dose to the observer, so that the viewing angle with respect to the horizontal is large.

Water bodies possess many physical, chemical and biotic features. Although mechanical (Wildermuth 1992b), thermal (Sternberg 1990) and even olfactory (Steiner 1948) characteristics can be used in the localization of oviposition sites, dragonflies recognize their habitat mainly by visual cues (Wildermuth and Spinner 1991; Wildermuth 1993), one of them being the partially and horizontally linearly polarized light (Horvath et al. 1998a; Wildermuth 1998). In Coenagrion mercuriale, Platycnemis pennipes and Leucorrhinia pectoralis, for example, it was shown that structural features of the habitat, such as emergent vegetation, are also important for the choice (e.g. Wildermuth 1992a) The degree of linear polarization of reflected light is a physical property that can be perceived from great distances and provides some information about the quality of the habitat. Thus, it may be the visual cue for polarization-sensitive dragonflies enabling them to discern dark and

212

radiance I



0%

p

100%

angle of polarization a measured from the vertical

O'

+90' (l

180'

Fig. 20.5. As Fig. 20.3 for a dark pond in a peat digging measured away from the sun. (After Bernath et al. 2002).

20 How can Dragonflies Discern Bright and Dark Waters from a Distance?

0%

100%

a O' -4-' <45·

-90" - 90"

-135' 180" <135'

.2

'" R 'o

2 ot,

" -0

llnlil

hndy

213

0%

100%

a 0" -4. • +45"

-90" 190"

-135" I~O" 1135"

Fig, 20.6. Reflection-polarization patterns Fig. 20.7. As Fig. 20.6 for another shady of a sunlit bright pond with shady regions bright pond with sunlit regions measured measured by video polarimetry at 450 nm away from the sun. (After Bermith et al. perpendicular to the sun with a viewing 2002). angle of 20° from the horizontal. (After Bernath et al. 2002).

Fig. 20.8. As Fig. 20.6 for a dark lake with undulating surface measured away from the sun.

0%

100%


bright water bodies from a distance. Future studies applying structural manipulations of natural substrata and choice experiments using dummies should prove whether dragonflies indeed use p of reflected light in their habitat selection.

21 on Reservoirs and Plastic Sheets as Polarizing Insect Traps

21.1 on Lakes in the Desert ofKuwait as Massive Insect Traps

During the Gulf War in early 1991, Iraqi occupation forces blas ted oil weHs and pipelines in the desert of Kuwait, forming more than 900 oil ponds. Several years later, the majority of these oillakes still existed (Pearce 1995) and continued to trap a variety of animals, mainly insects (Pilcher and Sexton 1993; Horvath and Zeil 1996). Reductions in the oil level created distinct bands of insect carcasses at their edges (Fig. 2l.l). Bands of dead dragontlies and ground-beetles retlected arrivals of migrating insects in autumn and spring. Mass arrivals of aeschnid dragontlies were witnessed by Jochen Zeil in October 1994 and February 1995, many females being trapped while attempting to lay eggs in the oil. Different species of water beetles (Dytiscidae, Coleoptera), giant water scorpions (Belostoma sp., Nepidae, Heteroptera), mole crickets (GryHotalpidae, Orthoptera) as weH as sphingid moths, Vanessa buttertlies, solifugid spiders, scorpions, reptiles, birds and mammals were found at the edge of the oil ponds.

These oil lakes trap different animals in different ways: in summer and autumn, when water has already evaporated, terrestrial an im als become entrapped by chance at the edges of the ponds during foraging or migration. During winter and spring, they may have been attracted to the water that overlies the taro Water-seeking birds (e.g. herons, egrets) and many tlying insects (e.g. buttertlies, dragontlies) land directly on the surface or at the edge of the oillakes also during times of the year when the surface is completely covered by oil. While some insects might have crashed into the oil, others have been clearly attracted to the oillakes.

Horvath and Zeil (1996) suggested that certain insects were attracted by the strong horizontal polarization of light retlected from the surface of the oil ponds. There are a number of reasons why oil surfaces may be even more attractive than water surfaces to animals sensitive to the polarization of retlected light:

216

.. Ddy bore

/ band. ofln. 'cl cDrC1lSSCS


Fig.21.1. The edge of an oillake in the vicinity of Kuwait City photographed in September 1995 by Jochen Zeil. Two distinct bands of mainly aeschnid dragonfly carcasses are visible on the old tar deposits high up at the bank of the lake.

1. erude oil is a better polarizing refleetor, beeause it has a high er refraetive index than water. The index of refraetion of clear water is 1.33 for the middIe range of the visible speetrum. Oil has a refraetive index of 1.39-1.49 depending on its eomposition, but the refraetive index of tar ean be as large as 1.57 (Levorsen 1967).

2. The higher viseosity of oil has the effeet that the reflection polarization of light is less distorted by wind-indueed ripples.

3. Sinee erude oil is not transparent, the light refleeted from flat oil surfaees is always horizontally and more strongly polarized than that refleeted from transparent water bodies (see Fig. 20.4).

To demonstrate the stronger polarizing ability of an oil surfaee relative to that of a water surfaee, Horvath and Zeil (1996) eompared the refleetionpolarizational eharaeteristics of erude oil and transparent/translueent water surfaees. The water-filled dishes in Fig. 21.2 demonstrate that light emanating from water is vertically polarized whenever the refraeted light dominates (top half, shady) and that the light is horizontally polarized when the main eontribution sterns from surfaee-refleeted light (bottom half, illuminated). A similar effeet eannot oeeur in an oil pond, beeause the penetrating light is entirely absorbed by the dark pigments of oil. The E-veetor refleeted from a flat oil surfaee is therefore always horizontal. The larger the degree of linear polarization p and the smaller the deviation of the E-veetor of refleeted light from the horizontal, the greater is the attraetiveness of the surfaee to waterseeking polarotaetic inseets. The eonsequenee is that a erude oil pond is more attraetive to water inseets than a water lake. Thus, for animals sensitive to polarization, oillakes eould appear as a supernormal optical stimulus.

Figure 21.3 shows the reflection-polarizational eharaeteristics of a Kuwait oillake in summer when the surfaee eonsisted of flat, low-viseosity oil. Figure 21.4 displays the same patterns for another oillake in winter when rainwater and oil formed eomplex surfaee features. In the foreground of the pieture in Fig. 21.4 there is a clear water surfaee; the dark oil surfaee beyond is broken by wind into bands and ehannels. The tar bands have a rough texture

21 Oil Reservoirs and Plastie Sheets as Polarizing Inseet Traps

Fig.21.2. The refleetion-polarizational eharaeteristies of three Petri dishes filled with blaek erude oil (Zeft), clear water (middle) and milky water (right) measured by video polarimetry at 450 nm. The top half of the dishes is in shadow, the bottom half is illuminated by unpolarized diffuse light from an overeast sky. Viewing direetion is 55° relative to the vertieal. (After Homith and ZeilI996).

Fig.21.3. The refleetion-polarization patterns of an oillake in the desert of Kuwait measured on 9 May 1995 at 450 nm. The polarimeter viewed towards north from a direetion of 75° relative to the vertieal. The surfaee of the lake was flat due to the low viseosity of oil beeause of the high air temperature. (After HorvMh and ZeilI996).

217

0%

p

100%

Cl O' -4 ' +45"

-90· +90"

-135' I RO' +135'

0%

p

100%

Cl O' -4 -, +45'

-90" +90'

-135' IRO' + 135'

218

Ce:

.2

.~ .. 8. .... o

i

total radiance

red (650 nm)


100010

degree of polarization

p

green (550 nm) ...---,.-...-=

O·

-90· +90·

+ 135· 180·

angle of polarization a measured from the vert ical

blue (450 nm)

Fig.21.4. As Fig. 21.3 measured on 15 January 1995 towards east and from 70° relative to the vertical at 650, 550 and 450 nm. Rain water accumulated on the high-viscosity oil during the winter months. Slabs of oil swam on the water surface and sand has settled on it. (After Horvath and Zeil1996).

and reflect light diffusely because sand has settled on the surface. Water and clean oil surfaces are highly polarizing, whereas p of light from the sand-covered tar is very low. The direction of polarization of light from both flat water and oil surfaces is horizontal, while areas with a rough surface reflect tangentially polarized light with respect to the sun. The high p of light from the region where oil and water meet (Fig. 21.4) again demonstrates the effect of transparency: the sandy bottom lies in the shadow of a floating oil slab and horizontally polarized surface reflection thus dominates in the blue due to the blueness of skylight. As with the water-filled dishes, however, p is reduced in the bright areas of water puddies, because the polarization of surface-

21 Oil Reservoirs and Plastic Sheets as Polarizing Insect Traps 219

reflected light is degraded by the refracted, vertically polarized light returning from the sandy bottom.

21.2 The Waste on Reservoir in Budapest as a Disastrous Insect Trap for Half a Century

Unfortunately, in many countries plenty of temporary inland oil spills exist as a by-product of the oil industry (exploitation, transport, storage and refinery of the oil). Since 1951 there has existed an open-air waste oil reservoir (Fig. 2l.5) in a suburb of Budapest, for example. Bernath et al. (2001a,b) observed that this oil reservoir deceived, attracted and trapped insects (Figs. 2l.6, 2l.7 A) in large numbers. They also measured the reflectionpolarizational characteristics of the waste oil surface versus time (Fig. 2l.8). This oil reservoir acted as a disastrous insect trap for 50 years. Its uniqueness was that it has existed for half a century in a densely populated suburb of a city (Fig. 2l.5A), where there was no natural water surface within 3 km. Bermith et al. (2001a,b) observed that even quite a small and shallow oil spill with an area of a few dm2 and a depth of a few mm can attract, trap and kill water insects.

In addition, some ancient natural asphalt seeps in the Earth's his tory have acted as massive animal traps, and their fossil remains play an important role in palaeontology. Cases in point are the famous Rancho La Brea tar pits at Hancock Park in Los Angeles (Akersten et al. 1983). In Rancho La Brea 9S % of the entrapped animal species belong to insects. Similar fossil deposits asso-

Fig. 21.5. The waste oil reservoir in Budapest in 1997. A On this satellite photograph the oil reservoir is the black polygonal patch at the right-hand side marked with a white arrow. B Schematic map of the reservoir compo~ed of seven smaller oillakes. C Aerial photo graph of the oil reservoir. D In summer (July) the oil surface was flat and shiny. E In autumn (September) the oil surface became gradually dull as the air temperature decreased. F In winter (December) the oil surface became matt and wrinkled, and rainwater accumulated in small pools on it. (After Bern<ith et al. 2001a).

. ~ -' '....... ::: .... ~ ' ...... "~~


ciated with natural oil reservoirs are the tar pools at Starunia in Western Ukraine, the Talara tar seeps in Peru, and the tar pits in Binagadin ne ar Baku in Azerbaidjan. Most of the insect fossil remains found in Starunia were water beetles belonging to the genus Helophorus (Angus 1973).

It is a general view in palaeontology that Rancho La Brea and Starunia animals might have stumbled accidentally across these tar pools, wh ich may have been camouflaged by dust or leaves (e.g. Akersten et al. 1983). Alternatively, these asphalt seeps may have been covered by rainwater from time to time, and thus attracted animals which then sank into the underlying tar, became entrapped and fossilized. Horvath and Zeil (1996) suggested that some waterseeking polarotactic animals were deceived by and attracted to the tar pits, even in the absence of water, by the horizontal polarization of light reflected from the oil surface mimicking a supernormally attractive body of water.

21.2.1 Surface Characteristics ofWaste Oil Reservoirs

The waste oil reservoir in Budapest was located in the 18th district of the Hungarian capital (47°27'N, 19°17'E) and consisted of seven oil pools situated within an approximately 220 X 110 m area (Fig. 21.5). Before the Second World War this area was a pebble mine. During the war the mining operations were stopped, then from 1951 the remained pits were used to store waste, spent and refused oil. The pebble pits were separated by hindrances resulting in the lakes. After appropriate recycling the oil was consumed as fuel, hut according to the decrease of demand the plant was closed. Its final removal happened hetween 1988 and 2001.

The reflectivity of the surface of the waste oil had a characteristic seasonal cycle: In summer the oil surface was usually mirror flat and shiny (Fig. 21.5D). This was disturbed only rarely by rainfall or cool weather. The viscosity of the oil remained low due to the great thermal inertia of the oil mass. The oil also kept its fluidity in cooler periods, consequently, the denser rainwater sank down in the lighter oil. Thus, the shiny and flat appearance of the black oil surface remained a characteristic feature throughout the warm months (from April to September). In autumn the oil surface became gradually dull (Fig. 21.5E) as the air temperature dropped and the oil became more viscous. In winter the surface of the oil became matt and wrinkled, and rainwater accumulated in small pools on it (Fig. 21.5F). Then the oil surface looked like gluey asphalt if it was not covered by snow.

In summer (from June to August) the oil surface reflected highly and horizontally polarized light (rows 1 and 6 in Fig. 21.8). In autumn (from September to November) the average p of reflected light gradually decreased as the air temperature decreased (row 2 in Fig. 21.8). The average E-vector direction of reflected light differed considerably from the horizontal in the dull regions of the gradually stiffening oil surface. In winter (from December to February)


the average p of reflected light was very low, and the average E-vector directi on was not horizontal and changed from point to point (row 3 in Fig. 2l.8). In spring (from March to May) as the temperature gradually increased, the average p increased and the average direction of polarization approximated the horizontal (rows 4 and 5 in Fig. 2l.8).

21.2.2 Insects Trapped by the Was te on

Dragonflies, mayflies, water bugs, water beetles and butter flies were trapped en masse by the waste oil in spring, summer and autumn at the time of their swarming and migration (Fig. 2l.6, Table 21.1). Usually, the insects landed or plunged directly onto the sticky oil surface and became immediately entrapped. Pairs of insects, e.g. dragonflies (Fig. 19.1) and mayflies, were trapped frequently by the oil during copulation and/or egg-Iaying. Depending on the viscosity of the oil, the trapped insects sank within more or less time. The higher the oil temperature, the lower the oil viscosity, and the shorter the time interval between landing and submergence of an insect.

Filtrates from the waste oil contained insects associated with water in large numbers (Table 21.1). Ephemeropterans, Trichopterans and Corixidae were abundant. Nematocerans were found most frequently; 44 % of them could be classified as Chironomids. Hymenopterans were found in large numbers, many of them were swarming ants. It is evident that the waste oil reservoir trapped a huge number of insects during its existence of half a century. Certain insects, e.g. Mantis religiosa and Oryctes nasicornis holdhausi, probably became entrapped by the oil during their walk when they reached the shore of the oil reservoir, where the soil and the pebbles were covered by the sticky oil. Although some of the larger insects, like great silver diving beetles (Hydrophilus piceus), were able to crawl out from the oil to the shore or to a pebble

Fig.21.6. Some typical representatives of the insects collected from the waste oil reservoir in Budapest and photographed on the oily shore (A, B, E, F) and on the oil surface (C, D). A A dragonfly (d, Anax imperator), a long-bodied water scorpion (b, Ranatra linearis) and two waterstriders (g, Gerris lacustris); B a moth (Lepidoptera sp.); C a mayfly (Cloeon dipterum); Da water beetle (Dytiscus sp.); E a great silver diving beetle (Hydrous piceus); F a dragonfly (d Sympetrum vulgatum), and scavenger beetles (w, Hydrophiliidae sp.). (After Bernath et al. 2001b).


Table 21.1. Insect species, the carcasses of which were collected from the waste oillake in Budapest and could be identified. (After Bernath et al. 2001b).

ODONATA: Aeshna mixta, Anax imperator, Sympetrum vulgatum

HETEROPTERA: Callicorixa concinna, Corixa punctata, Gerris lacustris, Gerris pallidus, Hesperocorixa linnei, Notonecta glauca, Ranatra linearis, Sigara falleni, Sigara lateralis, Sigara striata

COLEOPTERA: Acilius sulcatus, Coelambus impressopunctatus, Colimbetes fuscus, Copelatus ruficollis, Cybister laterimarginalis, Dytiscus circumflexus, Dytiscus marginalis, Elaphrus riparius, Hydaticus transversalis, Hydrophilus atterinus, Hydrophilus piceus, Hyphydrus ovatus, Ilybius subaenus, Oryctes nasicorniS, Longitasrus tabidus, Rhanatus punctatus

MANTODEA: Mantis religiosa

(Fig. 21.6E), they so on perished because their trachea openings became filled by oil. This is one of the reasons why so many carcasses of large insects could be found on the shore of the waste oil. The other reason is that the level of the oil gradually decreased from year to year, so that many insect carcasses became exposed on the shore (Fig. 21.6A,B,F).

21.2.3 Behaviour of Dragonflies Above on Surfaces

Horvath and Zeil (1996) as weIl as Bernath et al. (2001b) observed the behaviour oflarger dragonflies above the crude oillakes in Kuwait and the waste oil reservoir in Budapest. Male dragonflies frequently patrolled above the flat oil surface and protected their territory against all intruders (Fig. 21.7 A). They often sat guard on the tip of perches at the shore (Fig. 21.7B). Copulating pairs of dragonflies were frequently observed flying above the oil surface or trying to lay eggs into the oil. They became trapped during water-touching manoeuvres or egg-Iaying. In the latter case sometimes only the female became entrapped when the tip of her abdomen was dipped into the oil (Fig. 21.7C). In many cases, however, the male was also carried along with the female into the oil (Fig. 19.1). Touching the surface by dragonflies observed often at oil lakes is areaction which is typical only above water surfaces when dragonflies inspect the surface to select the optimal habitat or oviposition site (Wildermuth 1993; Corbet 1999). The most frequently observed behaviour types of dragonflies above the oil surface were the air fight, hovering and protection, which again are typical only above water surfaces (Horvath et al. 1998a; Wildermuth 1998; Corbet 1999).

21 Oil Reservoirs and Plastic Sheets as Polarizing Insect Traps

Fig.21.7. Dragonflies observed at different oil surfaces and trapped by oil. A Male dragonfly (Anax imperator) patrolling above the shiny flat surface of the waste oil reservoir in Budapest. B Male dragonfly sitting guard on the tip of a perch at the shore of a crude oillake in the desert of Kuwait photographed in September 1995 by Jochen Zeil. C Dragonfly (Anax parthenope) that was trapped in the oil moments before the picture was taken in Kuwait in September 1995 by J. Zeil. D-F When a large dragonfly touched the surface of a water pool with a thin oillayer in the desert of Kuwait (D), it was entrapped (E) and drowned (F by courtesy of J. Zeil).

223

After rain, pools of water formed on the shore of the waste oil reservoir. The surface of these pools was covered by a thin iridescent oillayer. Dragonflies or other insects that touched the surface of a water pool with a thin oil layer frequently became entrapped and drowned.! Even if the insect could crawl out from the water, its trachea openings were blocked by the oil. The same was observed by Jochen Zeil at the crude oil lakes in the desert of Kuwait (Fig. 21.7D-F). In the case of unpolluted water bodies, dragonflies and larger water beetles can easily fly out if they are dropped into water (Corbet 1999).

21.3 Dual-Choice Field Experiments Using Huge Plastic Sheets

Bernath and Horvath (1999) as well as Bernath et al. (2001a,b) performed dual-choice field experiments with insects and birds. Two huge plastic (polyethylene) sheets (20 x 30 m) were laid on the ground in a large alkaline field at about 500 m from a smaller alkaline lake. Such plastic sheets are commonly used in agriculture against weeds, and/or to keep the soil warm in order to speed up the sprouting. One of the sheets was black and the other milky translucent. The lower surface of the latter dimmed in some minutes following unfolding. Since billions of tiny condensed water drop lets scattered the incident light diffusely, the translucent plastic sheet became brilliant white.

1 A thin layer of oil has been used for many decades for mosquito control.

224

A

2

3

. - ., ~ - - ....

" ,.,- -'.

4

5

6

total radiance I

B

,... " - .

.. , ,' . . :4~'1~ ~.

--. -\. . . - -


0% P 100%

Part I1I: Polarized Light in Animal Vision

C 0 T OIir tempcralurc rC)

-5 q 5 15 25 3S 45 01 .0'.1'"

.. ' T August ' , p .......

S , ' I-e-I 4J 1 _QIiI_ I~7 \

Septembe r . , t---I

'. I-t ~. I-t

01.10. 19'97 " , " • Oetober « I-t ,

01.11 . 1991 ". 111 ........ ovcmbcr .. "

01.12 .1997 ) ' ..

Decc l11 ber , .. , .

01.01. ' ''8 . -: , ' January

I .~

" , I OI.02 . 1 99~ t. Fcbruary " .

',. OI .O,l.I9911 ' ,

. .1 March .:' 111 .. I-t

01 .04. 199:11 ; ..... April • • \' litJ

' . OI .05 . 1 ~;g . .. '

' . IIII-t

May , ; litJ I-e-I

OI ,Ob IIN$. .' J I-e-I

Junc ',0 PM .....

01.07. 199& :1 " July \ . T .' I

angle of polarization 01.08.1 998

o 5 15 25 35 45 55 65 75 Cf. from the vertical dcgree of polarization p (%)

Fig.21.8. Reflection-polarizational characteristics of the surface of the waste oil reservoir in Budapest versus time measured by video polarimetry at 450 nm under clear skies for a solar zenith angle of 60°. A Radiance I of the oil surface; B degree of linear polarization p of reflected light; C angle of polarization a of reflected light measured from the vertical (black vertical, a = 0°, white horizontal, a = 90°); D the daily maximal air temperature Tin Budapest and the average (calculated for the entire picture) P average of light reflected from the oil surface as a function of time. The horizontal bars represent the standard deviation of p; sampie size (= number of pixels in the p-pattern) = 560 x 736 = 412160. Viewing direction of the camera was 60° relative to the vertical and perpendicular to the solar meridian. (After Bernath et al. 2001b).


The black plastic sheet was used to imitate the shiny dark surface of oil or tar surfaces, and the white plastic sheet mimicked the surface of brighter bodies of water. The distance between the two plastic sheets was 30 m. In every experiment in the first half of the period the white plastic sheet was doser to the lake, and in the second half of the period the two sheets were transposed with each other. The vegetation beneath the sheets was mown. The sheets were stretched out horizontally as tightly as possible and were pinned down by bricks at the edges. Because of wind-generated wrinkles and thermal dilatation in sunshine the surface of the sheets sometimes became uneven, which was compensated by repeated spanning of the sheets, usually at sunrise, noon and sunset. After rain the water was removed from the plastic sheets. The insects attracted to the plastic sheets were documented (Figs. 21.9, 21.10; Table 21.2).

The light reflected from the black plastic sheet has much high er p than the light reflected from the white plastic sheet (Fig. 21.11). On the other hand, the white plastic sheet reflects a much greater amount of light. The direction of polarization of the horizontal black plastic sheet is always horizontal, while the white horizontal plastic sheet reflects vertically or obliquely polarized light (Horveith and Pomozi 1997). Figure 21.11 demonstrates well that the shiny black plastic sheet is a more effective polarizer than the white plastic sheet, even relatively far away from the Brewster angle. Thus, the light reflected from the shiny black plastic sheet acts as a supernormally polarized stimulus for polarotactic water-seeking insects. The optical characteristics of the shiny black plastic sheet are practically the same as those of wet, marshy soil or dark, deep water bodies and black crude/waste oil or asphalt surfaces. The reflection-polarizational characteristics of the shiny white plastic sheet are very similar to those ofbright -bottomed shallow dear water bodies or turbid white (e.g. alkaline) water. Because the oil surface and the plastic sheets are colourless, their reflection-polarizational characteristics are practically independent of the wavelength.

Certain water-Ioving insects can be easily deceived by and attracted to shiny black plastic sheets or asphalt roads (Fig. 21.9). Figure 21.10 shows an example for the behaviour of a great diving beetle (Dytiscus marginalis) on a

Fig.21.9. Some examples how insects associated with water can be deceived by and attracted to a shiny dry asphalt road (A) and a shiny black plastic sheet used in agriculture (B-D). A, Ba female large stonefly (Perla burmeisteriana); C a male stone fly (Nemoura cinerea); D a fe male caddish fly (Trichoptera sp.). (After BernMh et al. 2001b).

(

--~.

B . L.~" ,~.~ ~ ,. 'i •

;'" .tP y-"


Fig.21.1O. Behaviour of a great diving beetle (Dytiscus marginalis) on a shiny dry black plastic sheet at sunset. A landing on the plastic sheet; B touching, trying and probing the sheet; C flying up from the plastic sheet and looking for another place; D after landing again, the beetle tried to swim, crawled, or crept on the smooth plastic sheet; E after half an hour the beetle was entirely exhausted, it could not fly away, although it tried to fly up several times; F after an hour, the beetle perished. (After BernMh et al. 2001b).

Table 21.2. Insect species attracted to the shiny black plastic sheet used in the dualchoice field experiments. (After BernMh et al. 2001b).

EPHEMEROPTERA: Baetis rhodani, Cloeon dipterum, Ecdyonurus venosus, Epeorus silvicola, Ephemera danica, Haproleptoides confusa, Rhitrogena semicolorata

PLECOPTERA: Perla burmeisteriana

COLEOPTERA: Acilius sulcatus, Anacaena limbata, Besorus luridus, Copelatus ruficollis, Cybister laterimargnalis, Cymbiodita marginella, Dytiscus dimidatus, Hydaticus transversalis, Hydrobius fusipes, Hydrochara caraboides, Hydrochara Jlavipes, Hydrophilus piceus, Hyphydrus ovatus, Laccophilus obscurus, Phylidrus bicolor, Rhanatus punctatus, Spercheus emarginatus

HETEROPTERA: Aigara assimilis, Corixa affinis, Cymatia rogenhofferi, Hesperocorixa linnei, Notonecta glauca, Sigara falleni, Sigara lateralis, Sigara striata

shiny dry black plastic sheet at sunset. Bernath et al. (2001b) found that only the black plastic sheet laid onto the ground attracted insects associated with water (Table 21.2), and the white plastic sheet was totally unattractive to them. All these insects showed similar behavioural elements on and above the black plastic sheet: landing, flying up, touching, crawling, egg-laying, copulating, reproductive activity. Finally, all of them dried out and perished within some hours. Butterflies, flies, bees, wasps and dragonflies were also attracted to both plastic sheets, but they did not perish on them.

Almost every sunset, Bernath et al. (2001b) heard the black plastic sheet rattle with asound like the pattering of raindrops. The reason for this was thousands of Corixidae landing on and crashing into the black plastic sheet, then jumping repeatedly up and down. They did not leave the optical trap, and did not fly away from the visually attractive black plastic sheet; they remained on it throughout the night and perished. At the white plastic sheet a similar

21 Oil Reservoirs and Plastic Sheets as Polarizing Insect Traps

Fig.21.11. The reflection-polarization patterns of an aluminium foil (left) , a white plastic sheet (middle) and a black plastic sheet (right) laid on the ground and measured by video polarimetry at 450 nm under a clear skyat sunset. The viewing direction of the camera was 70° relative to the vertical and perpendicular to the solar meridian.

--:; ö .~ c ;:: o >'

227

0%

100%

.~ ~ 0: 0" ";: -::: -4 " +45" .g g -90" +90" ~;--_ .... " ~ -135' 180' + 135'

.. altt~~ -::::: ~

" " E

effect was not observed. At sunset the black plastie sheet was as cool as the white one. Apparently, the strongly polarizing black plastie sheet was more attractive to these polarotactie water-seeking insects than the weakly polarizing white one.

21.4 The Possible Large-Scale Hazard of"Shiny Black Anthropogenie Products" for Aquatie Insects

Horvath and Zeil (1996) as wen as Bernath et al. (2001a,b) identified a seldom addressed conservation and animal welfare issue, the possible large-scale hazard for aquatie insects of an strongly polarizing "shiny black anthropogenie products" induding oil reservoirs, asphalt roads and plastie sheets used in agrieulture. They recognized a general problem, the attraction of polarotactic aquatie insects to oillakes, black plastie sheets and other black and shiny surfaces. These insects show the same behaviour at these surfaces as at real water surfaces; these typieal water-specific behavioural elements involve touching the water surface (e.g. at egg-laying or probing the oviposition site), or landing on the water surface, or plunging into the water. All of these reactions are fatal for insects in the case of oil surfaces, because the stieky oil traps the insects. Even a thin oil layer on the water surface can hin der insects from


escaping, the consequence of whieh can be drowning. It is important to emphasis that water insects laid their eggs on the surface of plastie sheets, so these surfaces may endanger the renewal of their populations too. Although the dry shiny black plastie sheets used frequently in agrieulture cannot mechanieally trap the attracted insects, the polarization of light reflected from such smooth, shiny surfaces are so strong that polarotactie insects are visually compelled to remain on the dry plastie sheets in spite of the fact that other senses signal that these surfaces are not water. The consequence of this reaction is drying out and perishing.

On the basis of the above, it is clear that if the degree of linear polarization p of reflected light could be somehow reduced and it could be ensured in some way that the direction of polarization of reflected light differs from the horizontal, then the dangerous visual attractiveness of "shiny black anthropogenie products" to polarotactie insects could be reduced or even eliminated. A general rule is that the brighter and rougher a surface, the lower is the p of reflected light (Umow 1905) and the more the direction of polarization can deviate from the horizontal. Thus, Bernath et al. (2001b) suggested the following environmental protective arrangements that should be taken in the vicinity of habitats of insects associated with water:

• In agrieulture, the huge shiny black plastie sheets should be replaced by matt grey or white plastie sheets. It has to be investigated still whether these sheets would perform their agrieultural function as properly as black sheets, and therefore be acceptable to farmers .

• Until their removal, the surfaces of open-air oil reservoirs, spills and seeps should be covered by a thin layer composed of finely granulated white polystirol spheres, for example. This layer may require frequent renewal, because the spheres themselves can become coated all over.

22 Why Do Mayflies Lay Eggs on Dry Asphalt Roads? Water-Imitating Horizontally Polarized Light Reflected from Asphalt Attracts Ephemeroptera

At sunset, Kriska et al. (1998) observed that individuals of several mayfly (Ephemeroptera) species swarmed and mated above and landed on dry asphalt roads (Fig. 22.1A-C) and shiny black plastie sheets used in agrieulture (Fig. 22.1D-I) in the immediate vicinity of ephemeropteran emergence sites (mountain streamlets), and that after copulation the females laid their eggs en masse on dry asphalt roads instead of laying them on the water surface. The behaviour of male and female mayflies swarming and mating above asphalt roads is similar to that above water surfaces. Previous descriptions of ephemeropteran swarming, mating and egg-Iaying (reproductive) behaviour have misinterpreted this phenomenon: (1) It has gene rally been assumed, that asphalt roads serve as "markers"! for swarming mayflies. (2) Oviposition by mayflies on asphalt roads has been simply explained by the shiny appearance of wet roads whieh may lure the insects like the surface of real water bodies. The first explanation, however, cannot apply to the observed egg-Iaying on asphalt roads, because mayflies never oviposit onto objects serving as markers. The second interpretation cannot explain why egg-laying by Ephemeropte ra also frequently occurs on totally dry asphalt surfaces.

In an attempt to clarify the causes of reproductive behaviour of mayflies over asphalt roads, Kriska et al. (1998) conducted a study on six species of mayflies, (Ephemera danica, Ecdyonurus venosus, Epeorus silvicola, Baetis rhodani, Rhithrogena semicolorata, Haproleptoides confusa) using visual observations, video recordings, multiple-choiee experiments and videopolarimetrie measurements in the field. On the basis of these investigations,

1 Flight is virtually the only form oflocomotion in winged mayflies, therefore their mating takes place in the air. Mating is preceded by swarming, during which a group of males maintains a stationary position with respect to a striking visible object called "marker" (Harker 1992). Markers can be trees, bushes, the shore of lakes, dirt roads, rows of plants on the littoral, for example (Savolainen 1978). Due to the very short adult stage and because the newly moulted insects can dry out quickly, during swarming the mayflies remain relatively elose to the water. Thus, it is essential for the markers to be near water.


Fig.22.1. Examples of mayflies deceived by and attracted to a dry asphalt road (A-C) and a shiny black plastic sheet (D-I) in the immediate vicinity of a mountain creek near Budapest. A male Rhithrogena semicolorata; B female Epeorus silvicola; C one female and two males Epeorus silvicola attempting to mate; D male Rhithrogena semicolorata; E copulating pair of Rhithrogena semicolorata; F female and two males Rhithrogena semicolorata attempting to mate; G ovipositing Rhithrogena semicolorata; H ovipositing Ephemera danica; I egg-packets of Ephemera danica. (After Kriska et al. 1998).

they proposed a new interpretation for the peculiar behaviour of Ephemeroptera over asphalt roads. According to their explanation, asphalt surfaces at sunset mimic a highly and horizontally polarizing water surface to water-seeking mayflies which, as Kriska et al. (1998) showed, detect water by means of the horizontal polarization of reflected light, like many other water insects.

The study site was the bank of a section of a Hungarian mountain creek, from which mayflies emerged in large numbers and where they swarmed at sunset. In the immediate vicinity (1-5 m) of the creek an asphalt road ran between trees and bushes. The creek itself ran in a valley under trees and bushes and was usually completely shadowed by riparian vegetation, except where the road crossed it. The road was several metres high er than the creek, and above the road the sky was open. The surface of the asphalt road was relatively smooth and dark grey, but there were severallighter grey patches with a rough er surface.

In the multiple-choice experiments, rectangular (1 m x 2 m) test surfaces (shiny black plastic sheet, shiny white plastic sheet, shiny aluminium foil, slightly shiny black cloth, matt black cloth, matt white cloth) were laid on the

22 Why do Mayflies Lay Eggs on Dry Asphalt Roads? 231

asphalt road near different reaches of the creek where mayflies swarmed at sunset. The numbers of mayflies landing on and swarming immediately above the test surfaces at a height of no more than 1 dm within a I-dm2 rectangular region were counted. The position of the test surfaces was changed randomly. The experiments were always carried out under clear skies at sunset. At the beginning of an experiment, the landscape was illuminated by the setting sun and, after sunset, by skylight from above.

22.1 Swarming Behaviour of Mayflies Above Asphalt Roads

At the study site, depending on the species, the swarming of mayflies began prior to and after sunset every evening from the beginning of May until the end of June. After the emergence of the newly hatched insects from the mountain creek, the males gathered in several diffuse swarms in the air at a distance of approximately 4-5 m from the ground. At the beginning of swarming, these relatively diffuse swarms were observed everywhere above the streamlet, asphalt road, dirt roads and clearings in the vicinity of the emergence sites. Generally, these swarms developed at places where the sky was visible. As time elapsed, the swarms gradually approached the ground and more females flew through them to copulate with the males. After mating, the females returned to the streamlet or landed on the asphalt road and laid their eggs on the water or asphalt surface.

Later, as the air temperature and intensity of ambient light decreased, the swarms gradually left the dirt roads and clearings. Thereafter, swarming mayflies were observed exclusively above the asphalt road and the reaches of the creek open to the sky. In these swarms, both the males and females flew periodically up and down displaying the species-specific nuptial dances, or flew parallel to the water or asphalt surface against the prevailing breeze. They frequently touched the water or asphalt surface, or dropped onto it for a few seconds. When the air temperature decreased below 14-15 oe and the light intensity became too low, mayfly swarming suddenly ceased, and the insects disappeared from both the water and asphalt surfaces. They then landed on the leaves of neighbouring trees, bushes and grass to roost.

All six mayfly species observed behaved similarly above and on the asphalt road as at the water surface. The density of swarming, mating and ovipositing mayflies was highest above those patches of the asphalt where the surface was smooth and dark. No courting and egg-laying occurred above the relatively light grey or rough spots of the asphalt. One of the most typical reactions of female mayflies to the black and smooth asphalt patches was the following: after aerial copulation the females arrived above one of these patches. First, they flew over the patch, then suddenly turned back at its border, and in the presence of a gentle breeze all of them flew into the


breeze. Females touched the patch several times and landed on it to lay their eggs. Thus, Kriska et al. (1998) assumed that the darker and smoother the asphalt, the greater is its attractiveness to water-seeking mayflies. Above the asphalt road they observed the following two flight types for the six Ephemeroptera species, which are typical flight manoeuvres usually found only above water surfaces:

1. Egg-Iaying flight of females: the females, generally facing into the slight breeze, flew to and fro, parallel to and immediately above the asphalt surface, dancing up and down in a zig-zag pattern and sometimes touching the asphalt. This type of flight was performed only by females above the middIe part of the asphalt road. During egg-Iaying flight, the females performed a typical, species-specific stereotypical flight pattern, which resembled the nuptial dan ce of the swarming males and occurred simultaneously with it. As egg-Iaying flight progressed, an increasing number of eggs was pressed out from the genitalia of the females in the air. At the end of this flight, the females landed on the asphalt and laid their egg-packet (Fig. 22.1H). In the case of Ephemera danica, the females landed on the asphalt and remained on it until their elongated egg-packet was pressed out and laid (Fig. 22.11). The functions of egg-Iaying flight are finding an optimal site for oviposition, and/or allowing a larger number of eggs to be pressed out, and/or acting as a defence against attacks by swarming males (Fischer 1992).

2. Surface-touching manoeuvres of males: the males also periodically touched the asphalt surface during their flight, usually facing into the wind. Some of the individuals touched the asphalt periodically only with their cerci while flying up and down immediately above the road. Others landed on the asphalt, stayed on it for a few seconds and then took off, to land again some seconds later. Similar water touching by male mayflies (e.g. Baetis vernus, Ecdyonurus venosus, Rhithrogena semicolorata and Ephemera danica) was observed by Fischer (1992) above water surfaces at ephemeropteran emergence sites. According to hirn, such touching of the water surface by male Ephemeroptera allows them either to drink or to test the height above the water surface using their cerci.

22.2 Multiple-Choke Experiments with Swarming Mayflies

In the first multiple-choice experiment, Kriska et al. (1998) found that the shiny black plastic sheet reflecting light specularly was the only attractive surface for all six mayfly species; the shiny white plastic sheet and the aluminium foil were unattractive. In a control experiment, matt white cloth and slightly shining black cloth were used, which reflected light diffusely. Again, the shiny

22 Why do Mayt1ies Lay Eggs on Dry Asphalt Roads? 233

black plastic sheet was significantly more attractive than the cloths. The white cloth was unattractive. The slightly shiny black cloth attracted a small number of mayflies.

In the second multiple-choice experiment, Kriska et al. (1998) used totally matt black cloth as one of the control surfaces. The matt black and white cloths were unattractive and the shiny black plastic sheet was the only attractive to mayflies. The mayflies observed on the black plastic sheet were mainly single males or egg-laying females, but copula were also abundant. At the beginning of swarming only a few mayflies landed on the shiny black plastic sheet, but later practically every member of the swarm landed on it periodically. At the end of swarming more individuals had settled onto the plastic than remained flying above it.

The landing of mayflies on the black plastic sheet was so intensive that one could hear the loud strikes of the insect bodies similar to rain drops rattling on the plastic. If any part of the black plastic sheet was covered by a piece of any other test surface, the reproductive activity of mayflies ceased above this region. The swarming mayflies followed the slowly moving black plastic sheet if it remained horizontal. If the black plastic sheet was held vertically, the mayflies did not swarm over or next to it, nor did they follow its movement.

Using a hand net, mayflies (single males and females, egg-laying females, copulae) swarming above the black plastic sheet were captured and released onto one of the other test surfaces. These mayflies did not continue their reproductive activity on the new test surface, but left it and returned to the black plastic. However, if these mayflies were transfer red to another black plastic sheet, they began their reproductive behaviour again, showing that the captured mayflies did not fly away from the new test surface because of the netting procedure, but because of the unattractive or repellent nature of the test surface.

The water temperature in the creek was 12-14 oe and did not change during swarming on a given day. The air temperature above the creek decreased from 20-22 to 14-15 oe between the start and end of swarming each day. The air temperature above the asphalt road decreased from 25-26 to 16-17 oe during swarming. The warrnest location was always the asphalt road and the test surfaces on it. No temperature differences were found between the test surfaces, the temperature of which was always high er than that of the air above the asphalt road.

The swarming of mayflies began immediately prior to or after sunset when the air temperature was still relatively high above both the asphalt surface and the creek. The higher air temperature above the asphalt road prolonged the reproductive behaviour of mayflies by 15 min in comparison to the reaches of the creek from wh ich the sky was visible, presumably making the asphalt more attractive to mayflies than the creek. However, since there was no temperature difference among the test surfaces, the different reac-


tions of mayflies to the different test surfaces cannot be explained by their thermal perception.

When Kriska et al. (1998) laid the test surfaces onto the ground beneath trees and bushes on the bank of the creek, after swarming the mayflies landed en masse not only on the shiny black plastic sheet, but also on other test surfaces independently of their type. The behaviour of roosting mayflies was, however, quite different from that observed during their swarming. Roosting mayflies did not dance, fly up and down, or oviposit on the test surfaces, but simply settled on them and remained motionless, apparently using the test surfaces as roosting pI aces and not as reproduction sites. Because of the lower temperature, the roosting of mayflies on the shore of the creek began earlier than at the border of the warmer asphalt road.

22.3 Reflection-Polarizational Characteristics of the Swarming Sites of Mayflies

Figure 22.2 shows the reflection-polarizational characteristics of three different re ach es of a creek from which mayflies emerged and where they swarmed, mated and oviposited in large numbers at sunset. In the first reach of the creek (Fig. 22.2A), the water was relatively slow and calm and a small pond was present in the shadow of trees. Through the foliage, skylight illuminated the water surface from above and to the right. The degree of linear polarization p was high only in those regions of the water surface that were illuminated by skylight. The other regions of the water and the shore reflected practically unpolarized light. Because of the undulation of the water surface, the degree p and angle a of polarization changed strongly from site to site on the water surface, giving a relatively wide distribution of these variables. The Evectors of light reflected from the water surface were approximately horizontal, but because of the ripples on the water surface, they could diverge strongly from this direction.

The second reach of the creek was exposed to skylight from above (Fig. 22.2B). The water flowed slowly among stones and pebbles. Here, p of light reflected from the undulating surface of the turbulent water was also relatively low, and the dry stones and pebbles were unpolarizing. Thus, the spatial distributions of p and a were patchy and the histograms of these variables were again relatively wide.

In the case of the third reach (Fig. 22.2C), the creek flowed under trees, but its surface was illuminated by skylight from the side. Consequently, p of light reflected from the water surface was relatively high. However, similarly to the first and second reaches, both p and a of the water-surface-reflected light changed strongly because of ripples and their histograms were again wide.

22 Why do Mayflies Lay Eggs on Dry Asphalt Roads?

A

degrec of po tar; znt ion p

B


angle of polarization (X

235

c

1

2

3

4

5

Fig.22.2. The reflection-polarizational characteristics of three different reaches of a mountain creek (a typical emergence and swarming site of the maytlies studied) measured by video polarimetry at 450 nm. All three scenes were recorded from a direction of view of the camera of 60° relative to the vertical. A In this relatively slow and calm reach of the creek, a small pond was shadowed by trees. Through the foliage skylight illuminated the water surface from above and from the right. BAreach of the creek illuminated from above by the clear sky where the water flowed slowly among stones and pebbles. C Areach where the creek flowed under trees, but its surface was illuminated by skylight from the side. Rows 4 and 5 show the frequencies of the degree p and angle a of linear polarization of reflected light calculated for the rectangular windows in row 1. (After Kriska ct al. 1998).


Table 22.1. The relative radiance I, degree oflinear polarization p and angle of polarization a of light reflected from the test surfaces measured by video polarimetry. (After Kriska et al. 1998).

SI S2 S3 S4 S5 S6 S7 S8

1(%) 38.8±3.4 99.7±5.4 100±5.7 97.6±4.3 24.4±2.8 17.6±3.2 22.6±2.4 26.0±3.1 p (%) 50.9±3.4 3.3±O.9 3.2±1.l 7.7±1.5 9.1±2.1 15.1±2.8 55.0±5.4 30.6±3.4 a (0) 89.1±1.4 58.8±4.3 57.7±2.1 91.3±1.l 81.9±5.4 73.1±4.9 90.5±1.2 90.9±1.3

I is calculated relative to the shiny aluminium foi1; ais measured from the vertical. Values are me an ± standard deviation (N = 560 x 736 = number of pixels in a video picture). SI Wet asphalt; S2 matt white cloth; S3 shiny aluminium foil; S4 shiny white plastic sheet; S5 matt black cloth; S6 slightly shiny black cloth; S7 shiny black plastic sheet; S8 dry asphalt.

Figure 22.3 shows the reflection-polarization patterns of three different sections of the asphalt road above and on which mayflies swarmed, mated and oviposited. The distribution of p and a oflight reflected from the asphalt road was narrow; the E-vector of reflected light was predominantly horizontal and, apart from the lighter and rougher patches of the asphalt surface, p was relatively high, in spite of the fact that the surface was dry. The reflectionpolarizational characteristics of wet asphalt were similar to those of dry asphalt, but p was significantly higher. Table 22.1 contains the measured reflection-polarizational characteristics of the asphalt and test surfaces.

22.4 Mayflies Detect Water by Polarotaxis

Analyzing the reflection-polarizational characteristics of the test surfaces and their attractiveness to mayflies, Kriska et al. (1998) concluded that: (1) olfaction plays an unimportant role in the attractiveness of asphalt roads to mayflies. (2) The roles of wind, air humidity and colour in the choice by mayflies can be excluded. (3) Mayflies were not guided by phototaxis to the asphalt surface. (4) The higher air temperature above asphalt roads relative to that above waters prolongs their reproductive activity. However, it is unlikely to be the high er temperature that attracts mayflies to asphalt roads. (5) Polarization of reflected light is the most important variable explaining the attractiveness of asphalt roads. Only horizontally polarized light attracts mayflies, and the higher the degree of polarization, the greater is the attractiveness. (6) Mayflies detect the water predominantly by means of the horizontal polarization of reflected light.

These are in accordance with the results of Schwind (1985b, 1991, 1995), whose test surfaces also attracted Cloeon species (Ephemeroptera). He found

22 Why do Mayflies Lay Eggs on Dry Asphalt Roads?

O'

A

0%

dcgrcc of polarization p

100%

180' O'

B

angle of polarizalion a


237

c

2

3

4

5

180'

Fig.22.3. The reflection-polarizational characteristics of three different sections of the asphalt road above and on which mayflies swarmed, mated and oviposited. In each case, the asphalt surface was dry, and the scenes were recorded from a direction of view of the camera of 60° from the vertical. A A long section of the asphalt road illuminated by direct light from the setting sun under a clear sky. The camera viewed towards the solar meridian. B A short, smooth and dark section of the asphalt road illuminated by direct sunlight prior to sunset. The camera viewed towards the solar meridian. CAshort secti on of the asphalt road with smooth and rough, bright and dark patches illuminated by skylight from above after sunset. Other conventions as in Fig. 22.2. (After Kriska et al. 1998).


that the probable spectral range where Cloeon is sensitive to polarization is between 450 and 480 nm. In the experiments of Kriska et al. (1998), the slightly shiny black cloth with p = 15 % was slightly attractive while the matt black cloth (p = 9.1 %) was relatively unattractive to all mayfly species investigated. This indicates that the threshold of polarization sensitivity of their visual system is between 9 and 15 %.

Kriska et al. (1998) observed that the six mayfly species studied were deceived by and attracted to asphalt roads not only under clear skies, but also under partially cloudy skies or even totally overcast skies, when the UV component of skylight is strongly reduced or even absent. From this, they suggested that the polarotactic water detection of the Ephemeroptera studied functions in the blue, as in Cloeon.

According to Schwind (1991), insects inhabiting running waters, e.g. plecopterans living ne ar brooks, may not locate their habitats by polarotaxis, because polarization is reduced or even completely distorted by waves. Nevertheless, the observations and multiple-choice experiments of Kriska et al. (1998) show that this is not true for the six mayfly species studied.

Most works published on the ephemeropteran visual system deal predominantly or entirely with the dorsal (or turban) eyes (e.g. Horridge and McLean 1978). The function of the dorsal eyes2 is to distinguish the female in flight, because females usually enter the top of a swarm, and the turban eyes allow the males to search the area above them for females. With the lateral eyes both the males and females orient in relation to the landscape. Although in some mayfly species the microvilli structure of the retina in the dorsal as weIl as lateral eye was investigated (e.g. Burghause 1981a), nothing is known about the polarization sensitivity of the visual system in mayflies.

The experiments of Kriska et al. (1998) demonstrate that a shiny black plastic sheet can be efficiently used for the investigation of reproductive behaviour in Ephemeroptera. In the field, under natural conditions it is often difficult to observe mayfly swarms, because they are formed in unapproachable sites, above the water surface, or at high altitudes in the air. The placement of a shiny black horizontal plastic sheet of a few m2 can attract the whole swarm, allowing the study of mayflies, or their capture. This simple method could facilitate field studies on Ephemeroptera.

2 One of the few exceptions is Ephemera danica, the males of which do not possess dorsal eyes; the morphology of their eyes is very similar to that of the females.

22 Why do Mayflies Lay Eggs on Dry Asphalt Roads? 239

22.5 Comparison of the Attractiveness of Asphalt Roads and Water Surfaces to MayBies

Since the asphalt is black or dark grey and non-transparent, an asphalt road is an efficient specular reflector and polarizer if its surface is smooth. At sunset it always reflects horizontally polarized light, the p of which is almost 100 % near the Brewster angle (57S). Light penetrating into the asphalt has no effect on the polarization, because it is totally absorbed. On the other hand, an ephemeropteran emergence site returns horizontally polarized light when the surface-reflected light dominates while vertically polarized light if the light from the water dominates (see Fig. 20.4), and the greater the contribution of the second component, the lower is the net p. The highly and at sunset always horizontally polarized light reflected from asphalt roads with a relatively homogeneous distribution of p and a (Fig. 22.3) can, therefore, be much more attractive to mayflies than the surface of ephemeropteran emergence sites (Fig. 22.2).An asphalt road can polarize light in such a way that the reflected light becomes a supernormal stimulus for water-seeking polarotactic mayflies in comparison to the light reflected from water. This was also observed by Kriska et al. (1998)

when mayflies swarming above the asphalt road were attracted to the highly polarizing shiny black plastic sheet after it was laid onto the road.

In the attractiveness of asphalt roads to mayflies at sunset it is of particular importance that at dusk the E-vector of polarized light reflected from the rough asphalt surface is always horizontal due to the diffuse illumination from the sky. Mayflies (like many polarotactic water-seeking insects) are attracted only by horizontal E-vector. If the asphalt surface is illuminated by direct sunlight during the day, the E-vector of light reflected from the asphalt is usually not horizontal, hut perpendicular to a plane through the sun, the ob server and the site of reflection. In this case asphalt roads are no longer attractive to mayflies, even if p of reflected light is high.

One of the prerequisites of mayfly mating is to swarm above places where the sky is visible3, because the females usually are detected visually and captured by the males from below (Brodskiy 1973). The sky is generally open above highways and asphalt or dirt roads; thus, in this respect roads near the emergence site of mayflies provide a good swarming place. Hence, asphalt roads can be attractive to mayflies because (1) the sky above them is visible, (2) their frequently strong and horizontal polarization at sunset mimics a water surface, and (3) at sunset the air temperature above them is high er than in the surroundings.

It would be important to prevent the oviposition by mayflies on asphalt roads and plastic sheets, since the eggs laid onto these surfaces do not survive.

3 An exception is Ephemera danica without dorsal eyes and with a ventrally oriented chasing flight.


The following measures could be taken to prevent the egg-Iaying by mayflies on asphalt surfaces: one should treat the sections of the asphalt roads running near the emergence sites of Ephemeroptera in such a way that their surface becomes relatively light and rough to reduce their reflection polarization. This could be performed by rolling down small-sized lighter gravel on the asphalt surface. This treatment of the asphalt surface significantly reduces p of reflected light, which abolishes its attractiveness to mayflies. Furthermore, it would be advisable to forbid the farmers to use huge shiny black plastic sheets near the swarming and egg-Iaying sites of Ephemeroptera.

23 Reflection-Polarizational Characteristics of Car-Bodies: Why Are Water-Seeking Insects Attracted to the Bodywork of Cars?

--t Colour Fig. 1.5 demonstrates that the shiny bodywork of cars possesses regions from which strongly and horizontally polarized light is reflected. These regions can be attractive to polarotactic water-seeking insects. The smooth upper surface of the bodywork reflects the whole spectrum of light, while the paint beneath it returns light only in a certain spectral range. The light returning from the underlying paint becomes vertically polarized after refraction at the surface. This vertically polarized light reduces the degree of linear polarization p of the horizontally polarized light reflected by the surface itself (see Fig. 20.4). Thus, the net pis low or even abolished in that spectral range in which the paint reflects light efficiently. The light reflected from cars with red paintwork, for example, is less polarized in the red, but highly polarized in the blue and UV, if the paint absorbs shorter wavelengths. Many insects do not perceive red, but blue and UV light. Thus, red cars are highly attractive to water-seeking polarotactic red-blind insects. The bodywork of cars with metallized paint reflects light with low p in a wide region of the spectrum, where the metal particles reflect light efficiently.

There are many observations (e.g. Popham 1964; Watson 1992; Mizera et al. 2001) that insects associated with water are deceived by and attracted to the shiny windscreen or bodywork of cars (Fig. 23.1). Walton (1935), for example, observed that flying backswimmers (Notoneeta sp.) landed on the shiny roof of a car. Stevani et al. (2000a,b, 2001) observed that certain common Brazilian dragonflies are attracted by the reflecting surface of cars and lay eggs on them.

At the high temperatures of the car surface exposed to the sun the insect eggs can cause damage (Fig. 23.2), the chemical mechanism of which is similar to that caused by acid rains. In experiments on resin-coated plates, Stevani et al. (2000a,b, 2001) showed that cysteine and cystine residues present in dragonfly eggs are oxidized during the egg-hardening process, sclerotization, which releases hydrogen peroxide, a cysteic acid derivative, a strong acid capable of catalysing the hydrolysis of acrylo/melamine clearcoat polymer.


Fig. 23.1. Some examples how insects associated with water can be deceived by and attracted to the windscreen (A) or the red bodywork (B, C) of cars. A, B male Baetis rhodani, C a scavenger beetle (Hydrochara sp.).

A

Fig.23.2. Scanning electron-microscopic photographs of the acrylic/melamine resin after 3 h at 90 oe in contact with dragonflyeggs (A), cysteine of 10 mmol/L (B), and cystine of 10 mmol!L (C). (after Stevani et al. 2000b).

Cysteic acid was indeed identified in dragonfly egg extracts submitted to oxidation by H2ü 2 followed by acid digestion. These acids, destroying the clearcoat above 70°C, originate from proteins of the egg-shell, chorion, as products of chemical reactions. The temperature of car-bodies can often rise above 70°C in sunshine. Then, eggs laid onto the car surface can damage the resin just like acid rain (Fig. 23.2).

24 Polarization Sensitivity in Spiders and Scorpions

24.1 Spiders

Finding the way back to their web or burrow or shelter, or orienting within the web are complex tasks for which spiders also rely on their visual system. Spiders have two sets of simple eyes, a pair of anterior-median principal eyes directed forward, and three pairs of secondary eyes usually with a reflecting tapetum lining the back of the eye (Fig. 24.1A). These eyes are specialized for different tasks. All spider eyes possess microvillar photoreceptors, in certain species with orthogonally arranged microvilli (e.g. Schröer 1974; Dacke et al. 1999). Kovoor et al. (1993) studied the anatomy of the anterior median eyes and its possible relation to polarization sensitivity (PS) in Lycosa tarentula. They suggested that PS in Lycosa tarentula is mediated by the ventral part of the retina, where the photoreceptors bear rhabdomeres aligned in parallel series and successive lines of rhabdoms are orthogonal to each other. They also hypothesized that the analysis of polarization may be a successive process using a twisting of the retina due to the action of two muscles: the alternating contraction of these muscles can generate rotation and, to some extent, up and down movements of the retinal cup. The E-vector analysis by such a successive mechanism in spiders was first proposed by Schröer (1974) for Agelena gracilens.

Although the organization of the retinae in most spiders would permit the analysis of polarization, this ability was demonstrated only in a few species. PS and orientation by means of celestial polarization in their web or during homing were demonstrated in certain agelenid and lycosid spiders (e.g. Tretze11961; Papi and Tongiorgi 1963). In behavioural experiments PS was proven in the ground-dwelling lycosid spider Arctosa variana (e.g. Magni 1966), the agelenid funnel web spider Agelena labyrinthica (e.g. Görner and Claas 1985) and the gnaphosid spider Drassodes cupreus (Dacke et al. 1999). Only the latter appears to detect the direction of polarization by its secondary eyes. The former two may do this by their principal eyes. Magni (1966) demonstrated with electroretinography that the eyes of Arctosa variana are polarizationsensitive.

Papi and Tongiorgi (1963) and Magni (1966) observed that in an arena illuminated from above by totallY linearly polarized light, certain wolf spiders


Fig.24.1. A The cephalothorax of the spider Drassodes cupreus indicating the position of the pair of principal eyes (P) and the three pairs of secondary (S) eyes. B Tangential section through the retina of both poste rio-median (PM) secondary eyes showing the arrangement of rhabdomeral plates of the main photoreceptors and microvilli directions. C The V-shaped mirror and the retina in the PM secondary eye (after Dacke et al. 1999).

tended to move in a direction that would lead them back to their permanent retreat. When the E-vector direction of polarized light changed, the spiders altered their preferred direction accordingly. By selective ablation of the eyes, Magni (1966) demonstrated that the principal eyes were primarily responsible for polarization detection. Wolf spiders are "sit-and-wait" predators, and often build a permanent burrow or retreat and wait nearby for prey. Henton and Crawford (1966) found that the mygalomorph spider Apbonopelma californica can discriminate different E-vector directions of totally linearly polarized light. Experiments on single receptor cells of the posterior lateral eyes of a Plexippus spider species showed a weak re action to the E-vector rotation (Hardie and Duelli 1978).

In behavioural experiments Ortega-Escobar and Munoz-Cuevas (1999) studied the celestial cues used by the spider Lycosa tarentula for orientation and homing. Experiments (in all cases with occlusion of the sun) were performed under clear skies, under totally overcast skies and under clear ski es seen through a depolarizer (a plastic sheet which changed the polarization from linear to circular). They found that the celestial polarization pattern is sufficient for accurate homing. Painting over selectively the anterior median eyes (AMEs) or all other eyes revealed that the former are responsible for the perception of polarization. Electroretinography of all eyes confirmed that only the AMEs are sensitive to the E-vector direction.

24 Polarization Sensitivity in Spiders and Scorpions 245

In laboratory experiments Dacke et al. (2001) let lycosid wolf spiders Pardosa tristis run on an air levitated ball. The spiders were fixed at their cephalothorax and illuminated by a divergent totally linearly polarized light beam from above subtending an angle of 30°. The spiders turned rapidly after the commencement of continuous rotation of the overhead polarizer. Unfortunately, it was not excluded that they responded to the refraction-polarization-induced weak intensity pattern visible through the extended circular overhead rotating polarizer.

In the lycosid wolf spiders Geolycosa godeffroyi, Alopecosa pulverulenta and Pardosa prativaga, Dacke et al. (2001) found a narrow strip in the ventral retina of the principal eyes that views the zenith and has photoreceptors tiered into two layers. In a given layer the receptors have parallel microvilli, the direction of which is perpendicular to that in the other layer. The photoreceptors in this strip have wider inter-receptor angles (maximum of about 5°) than in other retinal areas with typical values of ab out IS. Furthermore, they lack any isolation from their neighbours by screening pigment, suggesting wide receptive fields. Dacke et al. (2001) suggested that this strip may mediate polarization detection in an analogous way as the dorsal rim area of the insect compound eye.

At sunset the gnaphosid spider Drassodes cupreus leaves its well-hidden nest to search for prey, then returns to its shelter. In laboratory experiments, the homing of these spiders was only successful if there was an overhead artificiallinear polarization pattern similar to that of the evening sky (Dacke et al. 1999). As soon as this pattern was removed, or the secondary eyes were covered with black paint, the spiders could not find their way horne. This suggests that they detect skylight polarization with their secondary eyes. The majority of the photoreceptors in a secondary eye have microvilli parallel to the long axis of the eye such that the whole eye is maximally sensitive to Evector parallel to the long axis (Fig. 24.1B). These UV (350 nm) receptors are highly polarization-sensitive with an average PS = 9.1. There are two other distal receptors with the microvilli direction perpendicular to the long axis of the eye. In the eye only green (500 nm) and UV (350 nm) receptors occur. Together, the three pairs of secondary eyes have a field of view that covers the entire sky. Since they are oriented at different angles, each eye responds maximally to a given E-vector direction. The fields of view (about 125°) of the postero-median (PM) secondaryeyes (Fig. 24.1B) are entirely overlapping and cent red on the zenith, where the highest degree of skylight polarization occurs at dusk. These eyes have very flat transparent lenses ensuring a wide receptive field and forming underfocused images. The left PM secondary eye has a preferred E-vector direction, which is nearly orthogonal to that of the right PM eye (Fig. 24.1B). Dacke et al. (1999) suggested that these two PM secondary eyes may integrate signals from a larger part of the zenith sky and cooperate to analyse the polarization of skylight. The V-shaped reflecting tapetum with two flanks at an angle of about 90-100° in these eyes (Fig. 24.1 C)


could act as a linear polarizer to enhance the PS of the entire eye. Light reflected from the tapetum becomes partially linearly polarized with E-vector parallel to the long axis of the eye as well as to the microvilli direction. Similar secondary eyes have been found also in several other spider species (Dacke et al. 2001). The extremely wide field of view of 1250 of the PM secondary eyes suggests that these eyes could enable navigation by means of skylight polarization only if the sun is ne ar the horizon, when the celestial E-vector pattern is rather simple and homogeneous and could elicit great enough differences in the output signals of the two PM eyes.

24.2 Scorpions

The sand scorpions Hadrurus arizonensis and Paruroctonus mesaensis are nocturnal insectivores inhabiting desert burrows by day and appearing on the dune surface only at night. As strictly nocturnal animals they have poor visual acuity, relying instead on mechano- and chemosensory systems for orientation to prey and potential mates. In laboratory behavioural experiments Brownell and Weber (1995) observed that these animals are unexpectedly polarization sensitive and direct their locomotory movements as a function of the E-vector direction of linearly polarized light.

Using a locomotion compensator (Kramer sphere), Brownell and Weber (1995) recorded changes in direction and velocity of walking Hadrurus arizonensis and Paruroctonus mesaensis as the E-vector direction of totally linearly polarized UV or white light from an overhead source changed. To ensure that orientation of the animals was not caused by variations in light intensity across the overhead linearly polarizing filter, a detectable intensity gradient was held at constant orientation while the E-vector direction was varied. Direction of locomotion varied with the E-vector direction, which shows that the an im als are indeed sensitive to polarization.

Scorpions have two sets of eyes that could mediate PS: (1) the large dorsal medial eyes and (2) 3-4 pairs of smaller, lateral eyes positioned ne ar the anterior margin of the carapace. When the medial eyes were covered by black paint, orientation under polarized light was lost. Since painting over the lateral eyes had no effect on this orientation, the medial eyes contain the polarization-sensitive photoreceptors wh ich mediate this behaviour.

The behavioural advantage, if any, of PS in nocturnal scorpions is unknown. Brownell and Weber (1995) hypothesized that the polarized light from night skies may not be sufficiently intense to be perceived by scorpions.

25 Polarization Sensitivity in Crustaceans

To our knowledge, the first an im als in wh ich polarization sensitivity has been demonstrated were certain entomostracan crustaceans (Verkhovskaya 1940) and the horseshoe crab Limulus (Waterman 1950). However, so far the behavioural or ecological significance of this ability in these arthropods is unknown.

In aseries of qualitative behavioural experiments, Baylor and Smith (1953) found that the Cladocera species Bosmina obtusirostris, Ceriodaphnia reticulata, Chydorus globosus, Daphnia magna, Kurzia latissima, Leptodora kindtii, Moina affinis, Sida crystallina, Simocephalus serrulatus and Simocephalus vetulus swam back and forth perpendicularly to the E-vector of totally linearly polarized white light in an aquarium. The outside surface of the glass walls of the aquarium filled with turbid pond water was covered by black wax, and the aquarium was illuminated from above or from one transparent side. When the horizontal E-vector of the vertical beam was slowly rotated, the population of Cladocera swam in a circle following the rotating E-vector. In a horizontal light beam with horizontal E-vector the animals swam upright, while with vertical E-vector they maintained their orientation by swimming on their sides perpendicularly to the E-vector.t Unfortunately, it was not excluded that the animals responded to the intensity patterns induced by selective reflection of polarized light from the black side walls of the aquarium.

Baylor and Smith (1953) suggested that the typical oscillation of the cladoceran eye may be a scanning mechanism for the detection of intensity changes and/or polarization. They also hypothesized that the possible functi on of polarization sensitivity may be either to find polarotactically the brighter areas with more food on the basis of the partially polarized light scattered laterally from the light beams penetrating into water between the surface vegetation, or to guide the an im als toward the water surface at morning and evening by means of the highly polarized skylight from the zenith at low

1 In these experiments water mites and caddisfly larvae responded to a vertical beam of polarized light in the same way as the Cladocerans. Mosquito larvae swam in every direction under polarized light, but they turned when the E-vector was rotated.


solar elevations. It is known that zoo plankton have such a vertical diurnal migration, i.e. during the crepuscular periods and at night they swim towards the water surface and during the day to deeper water layers.

The use of skylight polarization in the onshore-offshore orientation of Talitrus saltator, Tylos latreillii and Lygia italica was reported by Pardi and Papi (1953), Pardi (1954) and Menzel (1975), respectively.

In a behavioural experiment, Bainbridge and Waterman (1957) observed that in turbid sea water the marine crustacean Mysidium gracile swam in a horizontal direction and mainly perpendicularly to the E-vector of a totally linearly polarized verticallight beam. This response disappeared when the water was clear, but when yeast powder was added to the clear sea water, it was re-established (Bainbridge and Waterman 1958). Waterman (1960) found that Mysidium gracile swam approximately perpendicularly to the E-vector of a vertical beam of totally linearly polarized white light in clear sea water. The same response was observed when the experimental vessel was filled with clear sea water and surrounded by a larger concentric outer vessel filled with turbid water. The orientation became more precise with increasing turbidity.

Waterman and Horch (1966) proposed a simple model to explain the polarization sensitivity in crustaceans. Stowe (1983) gave a theoretical explanation of intensity-independent variation of polarization sensitivity in crustacean retinula cells. Ommatidia in crustacean eyes contain a rhabdom with two sets of strictly parallel microvilli, wh ich are aligned orthogonally to each other. According to the model ofWaterman and Horch (1966), polarization discrimination is a function of the relative photon catch between cell types with orthogonal microvilli.

The validity of this model has been confirmed for several crustaceans by electrophysiology and electron-microscopy. Shaw (1966), for example, recorded intracellularly the responses of retinula cells in the decapod crab Carcinus maenas to flashes of totally linearly polarized white light as a function of the E-vector direction. He found two groups of cells, the preferred Evector directions of which were perpendicular to each other and corresponded to the orthogonal microvilli directions in the rhabdom. Leggett (1976) found polarization-sensitive interneurons in the optic lobe of the crab Scylla serrata having photoreceptors with PS ::::: 10. There were two types of interneurons. The first type responded to the rotation of the E-vector, but was insensitive to the direction of rotation, while the second type was also sensitive to the direction of rotation. Leggett (1976) suggested that these polarization-sensitive interneurons may playa role (1) in contrast enhancement of underwater objects and/or (2) in stabilization of swimming direction with respect to the polarized background and/or (3) in detection of a polarized horizon in a featureless underwater environment and/or (4) in navigation by means of the underwater polarization pattern.

25 Polarization Sensitivity in Crustaceans 249

25.1 Mangrove Crab Goniopsis cruentata

Schöne (1963) tested the orientation of the mangrove crab Goniopsis cruentata in a circular arena with matt white walls under a vertical beam of totally linearly polarized white light with different E-vector directions. The crabs oriented with respect to the changing E-vector direction in such a way that they changed their orientation to polarized light by amounts roughly corresponding to and compensating for the change of the E-vector direction of skylight at the zenith. This suggests that the animals have a time-compensated celestial compass. By this orienting process, they walked mainly perpendicularly to the shoreline, where they were captured. However, when an additional artificial sun with azimuth always perpendicular to the chan ging overhead E-vector direction was presented, the animals preferred the shady half of the arena, which was doser to the sun. This reaction was interpreted as an avoidance of the brighter, and thus hotter and drier, parts of the substratum.

25.2 Fiddler Crabs

Fiddler crabs inhabit intertidal sand- and mudflats in dense colonies and are active during low tide. Male fiddler crabs have one massively enlarged and conspicuously coloured daw, which they use in waving displays and in fights against other males. Their eyes are long vertical stalks high above the body, and thus they see conspecifics in the ventral visual field below the local visual horizon and against the mudflat surface as background.

Altevogt and Hagen (1964) demonstrated that the European fiddler crab Uca tangeri is polarization sensitive. Korte (1965) observed polarizationinduced optomotor responses in Uca tangeri. Korte (1966) found that Uca tangeri only made an oriented response to artificial polarized light when it entered the apical ommatidia. The sand fiddler crab Uca pugilator has the ability to learn an orientation direction under an artificial field of linearly polarized light (Herrnkind 1968), which indicates that it may be able to use the pattern of skylight polarization. These observations prove that fiddler crabs are sensitive to polarization. Zeil and Hofmann (2001) found that areas with high polarization contrast in mudflat scenes include highly polarized specular reflections on the smooth and wet cutide of the fiddler crab Uca vomeris. Although both the spectral and polarization sensitivities of Uca vomeris are at present unknown, they hypothesized that reflection-polarization patterns of the carapace, especially the enlarged waving claw, may playa role in intraspecific signalling.


25.3 Copepod Cyclops vernalis

In a laboratory experiment, Umminger (1968) observed an endogenous rhythm in the orientation of the freshwater copepod Cyclops vernalis to a vertical beam of totally linearly polarized white light in clear water. The outer walls of the aquarium were covered either with black or white paper to discover if any light contrast reactions were present. In addition, the walls were shaded from direct light with a circular black or white diaphragm. When white side walls and diaphragm were used, the copepods swam perpendicularly to the E-vector at the beginning (0-1 h, "experimental dawn") and end (11-12 h, "experimental dusk") of a 12-h diurnal light period followed by a 12-h dark period, but they swam parallel to the E-vector midway through the light period (6-7 h, "experimental noon"). At intermediate times the animals oriented randomly to the E-vector. With black side walls and diaphragm, they swam perpendicularly to the E-vector during the 12-h light period.

In a second experiment, Umminger (1968) covered the vertical side walls of the aquarium by linear polarizers with horizontal transmission axes illuminated by a horizontal beam of unpolarized white light. The aquarium was additionally illuminated from above by unpolarized white light, the intensity of which was about ten times higher than that of the lateral light. This pattern of the surrounding light field crudely simulated the simplest natural underwater pattern near the water surface when the sun is ne ar the zenith. Under this condition the copepods always swam perpendicularly to the horizontal Evector. In natural waters, the ratio of the vertical to horizontal light intensity decreases with depth (Terlov 1976). In the laboratory, the deep-water situation was simulated by making the light intensities from above and from the sides equal. Under this condition, the copepods swam more vertically upwards and downwards perpendicularly to the E-vector at the beginning and the end of the 12-h light period, and more horizontally parallel to the E-vector during the middle of the light period. This indicates that a uniform light intensity pattern induces a diurnal rhythm in the polarotactic response. Under unpolarized light the animals swam mainly horizontally at all times of the day under both surface-water and deep-water conditions, but their orientation was not precise. The presence of horizontal E-vector increased the precision of horizontal swimming and induced a periodic vertical swimming component. If the light from the sides was vertically polarized, the diurnal rhythm in polarotaxis still persisted, but then the copepods swam horizontally, but henceforward perpendicularly to the E-vector.

In a third experiment, Umminger (1968) either decreased or increased light intensities from above and/or from the sides to approximate the temporal change of illumination at sunset and sunrise, respectively. In unpolarized light, intensity changes did not cause net displacements of the copepod population. Swimming remained primarily horizontal. However, when the lateral


light was horizontally polarized, increasing and decreasing intensity caused a net downward and upward movement, respectively, in vertically swimming individuals, while horizontally swimming animals were not affected, regardless of the time of day and whether the intensity of overhead and laterallights changed at the same or at different rates. This indicates that the cue of changing intensity is stronger than the rhythm in polarotaxis that occurs when intensity is constant.

From the above results Umminger (1968) proposed the following, ecologically important scenario: in the morning with increasing light intensity, the polarization-governed swimming of copepods has a net downward component both at the surface and in deep water, thus the population tends to sink. At midday, those copepods that sank into deep water in the morning move primarily horizontally, maintaining their position. Those copepods that remained at the surface during the morning still tend to move vertically up and down at midday. In the evening with decreasing light intensity, polarization-guided swimming has a net upward component both at the surface and in deep water, which causes the population to rise. Umminger suggested that horizontally polarized light provides the copepods with a verticallhorizontal reference direction for spatial orientation in the water where other orienting cues, such as landmarks, are sparse. The diurnal rhythm in polarotaxis operates only in deep water and helps to orient in a horizontal plane at midday, thus inhibiting vertical swimming which occurs at dawn and dusk and might carry the copepods to the surface at that time.

These trends in polarization-based swimming of Cyclops vernalis are consistent with the pattern of vertical migration observed in many planktonic Crustacea. The ecological functions of this movement pattern may be the avoidance of visual predators ne ar the water surface and/or protection against the UV radiation which is dangerous for animals without protecting, UVabsorbing pigments (e.g. Rhode et al. 2001). However, the major problem with the interpretation ofUmminger (1968) is that the polarization pattern used in his experiments occurs only at low latitudes around noon, when the sun is ne ar the zenith. Thus, under both lower solar elevations and latitudes, the proposed mechanism could not explain the vertical migration of polarization-sensitive zoo plankton.

25.4 Larvae of the Crah Rhithropanopeus harrisi

Via and Forward (1975) studied the ontogeny and spectral sensitivity of polarotaxis in the planktonic larvae of the crab Rhithropanopeus harrisi. The larvae were illuminated from above by a vertical beam of totally linearly polarized white light with different E-vector directions. Only light-adapted larvae in the developmental stages 11 and III responded with oriented swim-


ming to polarization. Stage 11 larvae swam perpendicularly to the E-vector, while stage III larvae swam parallel to it. Larvae in developmental stages land IV oriented randomlywith respect to the E-vector. Swimming under unpolarized light was random in all developmental stages. Polarotaxis disappeared in dirn light in stage 11 larvae, and at intensities high er than 2· 10-1 W 1m2 all animals swam randomly in polarized light. The polarotaxis was the strongest at 499 nm and a secondary peak occurred at 420 nm. The intensity threshold at 499 nm for polarotaxis was lO-c 10-3 W 1m2 in stages 11 and III.

In a control experiment the vessel was surrounded by alternating black and white quadrants. Under unpolarized light, stage 11 larvae oriented toward the white sectors with positive phototaxis. Under polarized light with E-vector parallel to the opposite white sectors, this phototactic behaviour was reduced by polarotactic swimming toward the black sectors placed perpendicularly to the E-vector. This demonstrates that the larvae responded to the E-vector and not to intensity patterns produced by selective reflection of polarized light from the walls of the experimental chamber. Via and Forward (1975) hypothesized that the onset of polarotaxis in stage 11 may result from functional alterations associated with morphological and neural changes between the sessile eyes of stage land the stalked eyes of stage 11. Such structural changes in the eyes are not known between stages III and IV. They concluded that Rhithropanopeus harrisi larvae in developmental stages 11 and III can perceive and orient to the E-vector of polarized light.

25.5 Larvae of the Mud Crab Panopeus herbstii

In a laboratory experiment, Bardolph and Stavn (1978) studied the polarization sensitivity of larvae of the mud crab Panopeus herbstii in their developmental stage I. The larvae swam freely in a cubic cuvette filled with clear sea water and illuminated from above and from its four sides by white light. The illumination conditions were: (1) unpolarized light (depolarizers on all four sides and the top of the cuvette), (2) horizontally polarized light (the side walls of the cuvette covered by linear polarizers with horizontal transmission axes and the top with a depolarizer), (3) horizontal and tilted E-vectors (two opposite side walls of the cuvette covered by linear polarizers with horizontal transmission axes and the other two side walls by polarizers, the transmission axes of which were tilted by 20° from the horizontal, and the top was covered with a depolarizer).

The body long axis of swimming larvae was mainly horizontal under all three illumination conditions, but there was a significant difference in angular orientation distributions of their body axis under the conditions of unpolarized versus horizontally polarized light. The average body axis turned by 1-5° in the direction of the E-vector tilt when the larvae were exposed to the


horizontal and tilted E-vectors. From these observations, Bardolph and Stavn conc1uded that Panopeus herbstii stage I larvae may be sensitive to polarization. They also speculated that the possible ecological significance of this capability may be a more precise orientation for maintenance of body axis and for vertical migration.

25.6 Grapsid Crah Leptograpsus variegatus

With intracellular recordings from retinula cells Doujak (1984) investigated the photoreceptor membrane dichroism and polarization sensitivity in the dorsal-lateral region of dark-adapted eyes of the shore crab Leptograpsus variegatus. He found that the rhabdomeric dichroic ratio was between 1.9 and 5.63 with an average of 5.0, and the polarization sensitivity varied between 1.8 and 7.5 with an average of 5 both at 499 and 621 nm. Hence, the average dichroic ratio equals the average polarization sensitivity in the same retinula cells.

25.7 Crayfish

With intracellular recording in the anterior and dorsal quadrant of the compound eye of the crayfish Procambarus clarkii, Waterman and Fernandez (1970) found two types of photoreceptors, which are maximally sensitive either to blue (440 nm) or yellow-orange (594 nm) light. Blue-sensitive receptors were found only in the anterior quadrant of the eye. Both spectral types of receptors have sensitivity maxima to vertical and horizontal E-vectors. These E-vector directions correspond roughly to the two microvilli directions occurring in the retina. The polarization sensitivity of both spectral types of cells ranges from 1.2 to 11.9 with an average of 3.1 at the wavelength of their sensitivity maximum.

With cell dye injection, Sabra and Glantz (1985) found that the polarization-sensitive photoreceptors terminate in the two plexiform layers of the first optic ganglion (lamina ganglionaris) of Procambarus clarkii. Photoreceptors Rl, R4 and R5 are maximally sensitive to the horizontal E-vector and receptors R2, R3, R6 and R7 to the vertical E-vector. They project their terminals to the distal and proximal plexiform layer, respectively. There is no spatial overlap in the locations of these two functional c1asses of receptor terminals. Hence, there are two separate polarization-sensitive channels. Differential photoreceptor sensitivity to the E-vector direction is correlated with a differentiallocalization of receptor terminals in the lamina. The PSvalue of vertically polarization -sensitive receptors changed from 1.5 to 9.17


with an average PSv = 2.65, while for the horizontally polarization-sensitive receptors the range was 1.49 to 2.81 with a mean PSh = 2.13.

By means of intracellular recordings and broad-field illumination, Glantz (1996) examined the polarization sensitivity and polarization responsiveness of photoreceptors and lamina monopolar neurons in two species of crayfish, Procambarus clarkii and Pacifasticus leniusculus (Table 25.1). He found that the change of sensitivity ofboth photoreceptors and lamina monopolar neurons to totally linearly polarized light is well described by cos2a functions, where ais the angle of polarization. Of the lamina neurons 78 % are polarization sensitive and both the PS and the direction of preferred E-vector orientation are similar to those of photoreceptors. The dynamic properties of lamina monopol ar neurons are consistent with either an opponency mechanism or strong, but polarization-insensitive lateral inhibition. On the other hand, medullary neurons in stomatopods (Yamaguchi et al. 1976), crabs (Leggett 1976) and crayfish (Waterman 1984) appear to be insensitive to the E-vector directions oflight pulses or to stationary E-vector orientation.

The four classes of polarization-sensitive interneurons, lamina monopolar cells, tangential cells, sustaining fibres and dimming fibres in the visual system of the crayfishes Procambarus clarkii and Pacifasticus leniusculus are not only sensitive to polarization, but also respond to unpolarized intensity patterns. Furthermore, many of these cells, especially the sustaining fibres respond more strongly to changing E-vectors than to stationary E-vectors. While all four cell types respond modestly to light flash es with an E-vector perpendicular to the preferred orientation, the dynamic response to a changing E-vector is weak or absent. Rence, polarization sensitivity interacts with normal intensity contrast vision throughout the visual system. The polarization-sensitive sustaining fibres provide afferent input to optomotor neurons which are involved in compensatory reflexes for body pitch. These optomotor neurons are maximally sensitive to a flash of vertically polarized light. The motor neurons also respond maximally to a continuously rotating linear polarizer if the E-vector is vertical. Therefore, Glantz (2001) proposed that polarization sensitivity at the most peripheral stages of the crayfish visual

Table 25.1. Polarization sensitivity and responsiveness of photoreceptors and lamina monopolar neurons in two species of crayfish, Procambarus clarkii and Pacifasticus leniusculus (after Glantz 1996). Polarization sensitivity PS = S( amax)/S( amax -90°), where S( a) is the sensitivity at angle of polarization a. Polarization response ratio Rps = R( amaxY/R(amax-900), where R( a) is the response at angle of polarization a.

Crayfish species

Procambarus clarkii Pacifasticus leniusculus

Photoreceptors PS Rps

3.1 ± 0.5 2.8 ± 0.7 4.3 ± 1.8 3.9 ± 0.9

Lamina neurons PS Rps

3.2 ± 0.7 3.5 ± 0.4 4.5 ± 2.4 6.2 ± 3.0


system, lamina ganglionaris and medulla externa, enhances the contrast of objects where intensity differences are absent, and thus may contribute to motion detection in low-contrast environments.

Interestingly, beside reflection polarization, the cuticle itselfhas polarizing ability at certain parts of the body of crayfish (Neville and Luke 1971) similarly to the carapace of mantis shrimps (Marshall et al. 1999).

25.8 Grass Shrimp Palaemonetes vulgaris

Ritz (1991) reported on polarization-dependent responses in the grass shrimp Palaemonetes vulgaris. Goddard and Forward (1991) investigated whether Palaemonetes vulgaris can use the celestial polarization pattern visible from water within the Snell window for sun compass orientation, espedallY to orient the offshore escape response in the vegetation-free area. This celestially guided escape response is a rapid movement perpendicular to the local shoreline to deeper water. It is elidted by the approach of a shoreline predator (Goddard and Forward 1989). The shrimps are capable of learning the relationship between the local offshore direction and celestial cues, either the sun or the blue sky alone, after only a 2.5-4 hexposure to an unfamiliar shoreline (Goddard and Forward 1990). The behaviour, which is a time-compensated celestially guided menotactic orientation, persisted up to 7 h after the shrimps had been maintained under constant conditions, but decayed after 7-24 h. Thus, the shrimps must relearn their offshore direction after prolonged periods of complete cloud cover and each morning after sunrise. Both the decay and learning of the actual offshore direction is rapid relative to similar behaviour of other arthropods. Since Palaemonetes vulgaris live along convoluted shorelines, it is likely that they will encounter new offshore directions to learn on a daily or even hourly basis. Their rapid learning ability may be necessary for effident avoidance of shoreline predators.

In a laboratory experiment, Goddard and Forward (1991) found that Palaemonetes vulgaris can perceive polarization. It oriented parallel to the E-vector of downwelling totally linearly polarized light and followed the E-vector rotation regardless of the phototactic state.2 In this experiment, the arena was illuminated from above by downwelling light of a tungsten bulb passing through a filter composed of two layers of waxed paper and one layer of a linear polarizer sandwiched between sheets of clear Plexiglas. Thus, the downwelling light within the arena was totally linearly polarized if the polarizer was face down,

2 Phototactic state means the following: when approached by a predator, the shrimp inhabiting dock sites escape to dark regions beneath a dock, while shrimp living in the open shore escape in the offshore direction towards a bright area relative to the dark shoreline grasses (Goddard and Forward 1989).


or unpolarized if the waxed paper was face down. It was proven that the reaction of the shrimps was not elicited by intensity patterns created by the refraction and scattering of polarized light in the water or by differential reflection from the waHs of the arena, which must always be tested.

In the outdoor experiments performed by Goddard and Forward (1991) under natural illumination conditions, the arena was covered by a dome made of clear, acrylic Plexiglas having approximately 100 % transmittance and not disrupting the polarization of transmitted light for wavelengths above 325 nm. Through the clear dome the sun and the brightness and colour distribution as weH as the polarization pattern of the sky were almost unalterably available. To study the role of the different celestial cues in the offshore escape response, the availability of these cues was manipulated on the dome:

• The sun was blocked by a thick circular "sky-blue" plastic sheet. • The polarized skylight was fuHy depolarized with a wax layer painted over

a layer of Parafilm attached to the dome. • If the dome was completely waxed, fuHy depolarizing skylight, the sun

appeared as a bright spot through the wax layer, and the shrimps could use it for orientation.

• To determine whether the shrimps can orient by means of blue sky cues at different elevations above the horizon or at different azimuth directions relative to the sun, the dome was completely waxed except for test sections of the sky. Horizontal annular sections with different elevations, or one, two or three circular windows were centred at 45° elevation and at various azimuths relative to the solar position.

• To determine the role of polarization in the orientation, the polarization pattern of the sky available in circular windows on the completely waxed dome was manipulated using linearly polarizing filters taped over the test windows. Presented in each window was either the correct E-vector directi on which appeared naturaHy in that section of the sky at the time of the test, or different artificial E-vector directions. It was assumed that these polarizers simulate more or less natural skylight polarization.

Goddard and Forward (1991) found that in the absence of other environmental cues, such as slope or landmarks, Palaemonetes vulgaris orients by means of the sun or blue sky in the offshore direction, but becomes disoriented under completely cloudy skies. The shrimps can orient by means of blue sky alone from any elevation above 17° and from azimuths in the antisolar, but not the solar half of the celestial hemisphere. Some differences in the cues available in the solar versus antisolar half of the sky account for this difference. The degree of linear polarization p is minimal in the directions of the sun and the antisun and increases to a maximum at 90° from these points. Thus, apart from the case when the sun is at the zenith or on the horizon, the average p of skylight is always lower in the solar than in the


antisolar half. This difference may affect the orientation, if it depended upon p. Small circles of blue sky, subtending a 27° angle, are sufficient for orientation. This ability is useful for the shrimp under partly cloudy skies, when at least patches (angular diameter ~ 27°) of the blue sky are seen. When the Evector direction of the natural celestial polarization pattern visible in such circular windows was changed with linear polarizers, the shrimps' orientation changed accordingly.

The orientation of Palaemonetes vulgaris might possibly be supported by the skylight intensity and colour pattern. Neither cue is sufficient for orientation alone. There is a hierarchy of them: the sun can either be used without other celestial cues, or the polarization is dominant over the intensity and colour pattern. The latter can be the case on partly cloudy days, when the sun is blocked, but patches of blue sky are visible, or on windy days, when ripples and waves on the water surface displace the apparent position of the sun while the celestial polarization pattern remains relatively unaffected (Waterman 1988). The relative role of non-celestial cues, such as landmarks, water surface waves and slope of the sea bottom, is unknown. However, these may be essential on completely cloudy days or during the period when the shrimps are learning a new shoreline and a new offshore escape direction (Goddard and Forward 1990).

25.9 Crab Dotilla wichmanni

Many crustacean species are highly mobile covering considerable distances during their far-ranging excursions, and some species are able to return to well-defined horne locations, like the fiddler crabs of the genus Uca (e.g. Wehner 1992). Several other crab species usually do not move more than a few centimetres away from their horne burrows. One of these highly sedentary crab species is Dotilla wichmanni (Luschi et al. 1997). These small crabs with carapace length of ab out 1 cm are common on tropical sandy shores, where they feed at diurnallow tides by sorting the organic material deposited on the sand around their self-dug burrows. The feeding activity is centred on the burrow. It is restricted to the immediate vicinity of the hole, and performed by systematic sampling and sorting the sand. The large food abundance near the burrow allows the crabs to remain near their shelters. While fee ding, the crabs move along shallow trenches radiating from the burrow and produce pseudofaecal pellets which are amassed over the already excavated area (Fig. 25.1). Thus, the feeding trench is walled on one side by a row of pellets and on the other side by the untouched sand. Any particular sand surface area is sorted only once during a fee ding period. Since the crabs scrap the sand in front of the trench and deposit the discarded pellets at the rear, the position of the fee ding trench changes from time to time, rotating clockwise or counter-

258

ü o

aClua l ---'- feeding

trencll


Fig.25.1. Schematic representation of the feeding area of the crab Dotilla wichmanni ploughing through the sand substratum around its burrow counter-clockwise. The actual feeding trench is lined along its edge with pseudo-faecal pellets covering the already sorted and treated sector of the substratum. One of the former positions of the feeding trench is indicated by a dashed white line. (After Luschi et al. 1997).

dockwise around the burrow like the pointer of a dock (Fig. 25.1). The average maximum length of the feeding trenches is about 8.3 cm (Luschi et al. 1997). When disturbed, the crabs rapidly return into their burrows, where they remain hidden for a while. Upon re-emerging, they recommence feeding moving along the same trench as before, in order to continue the systematic exploitation of the substratum. The orientation of the first fee ding trench is random.

However, even under these spatially restricted conditions Dotilla wichmanni can encounter situations requiring relevant orientation capabilities, e.g. after heavy rain showers or when the fee ding area is washed out by larger sea waves. Then, the crabs interrupt their feeding activities and find refuge in their holes. When they re-emerge, the area around the burrows is uniform and they have to choose where to feed without being guided by cues derived from their previous activities. Luschi et al. (1997) found that Dotilla wichmanni correcdy chooses the same feeding orientation held before retreating even in such situations. To study the cues used enabling this orientation, Luschi et al. (1997) manipulated the area near the burrows:

• If the area around the burrows was deaned by sweeping away the pellets from the feeding area, taking away the surface layer of the sand around the hole and then covered with new layers of untouched wet sand taken from the neighbouring area so that any sign of the animals' previous activities was removed, the crabs were able to select and maintain correcdy the feeding direction they had chosen before treatment.

• When the deaned feeding area was covered by two concentric translucent plastic domes, l3/15 cm wide, 6/11 cm high, which depolarized the light reaching the animals and prevented them from viewing both sky and landscape, and the sun was screened by a plywood board, the crabs continued to feed in random directions after re-emergence. Hence, the orientation of the crabs is based neither on any feature of the area ne ar the burrow (e.g. the slope of the sand surface) nor on the geomagnetic field, nor on idiothetic information.


• If the crabs were prevented from viewing the visuallandmarks of the surrounding landscape by placing a white conical plastic screen with its wider opening upward around the burrow, they were able to correctly re-orientate their fee ding direction only under blue skies. They oriented randomly under overcast skies. Thus, visuallandmarks around the feeding area are not necessary for correct orientation as long as the blue sky is visible.

• When the sun was screened by a plywood board and deflected by 180° with a mirror, the crabs correctly maintained their previous feeding directions if they could see the blue sky in the opening of the conical screen around their burrows. Thus, the crabs do not orient by means of the solar azimuth.

• If the sun was screened and the upper opening of the conical screen around the burrow was covered by a linear polarizer, above which three sheets of translucent graph paper were placed as depolarizers, and the crabs were left to feed a few minutes under this artificial E-vector pattern, they continued to feed in the same direction as previously under the artificial polarization pattern. If the polarizer was rotated by 90°, the crabs' feeding direction followed this rotation, but became bimodal. Hence, the crabs rely mainly on the skylight polarization pattern, but other celestial cues, e.g. celestial spectral and radi an ce distributions allowing unimodal orientation are also necessary to distinguish between the solar and antisolar directions. However, visual landmarks near the feeding area playa role when celestial cues provide no useful information under overcast skies. Both cues are used as a reference to choose the feeding direction before retreat.

Among the polarotactic orientation mechanisms of animals, the E-vector orientation of Dotilla wichmanni represents a special case, since the celestial polarization is used he re to orient short-range (maximum about 8 cm) feeding movements in the immediate vicinity of the burrows. Taking the previous feeding direction after a retreat has the significance that the crabs avoid feeding over an already excavated sector, and optimise their fee ding activity, which must take pI ace in the limited time window of the daytime intertidal period, even in those instances when the signs of the crabs' previous activity have disappeared, e.g. due to heavy rain, or large sea waves.

25.10 Water Flea Daphnia

Verkovskaya (1940) found that Daphnia presented with two narrow-field lights at two opposite sides of their visual field behaved positive phototactically as if the polarized light source had two to three times the intensity of the unpolarized light source. The E-vector direction was not reported in the paper of Verkovskaya. This experimental approach, presenting polarized


stimuli at the side of the visual field of Daphnia, was taken over only much later again by Schwind (1999).

Baylor and Smith (1953) have shown that when the water fleas Daphnia magna are in a vertical beam of totally linearly polarized white light, they preferentially swim perpendicularly to the E-vector. That daphnids utilize an intra-ocular polarization analyser was established by Smith and Baylor (1960) with half-wave plates to distinguish intra-ocular and extra-ocular polarization analysers. Hazen and Baylor (1962) found that both positively and negatively phototactic (the latter induced by drug or UV light) D. pulex illuminated simultaneously by three beams of white light (one totally linearly polarized and vertical, the other two unpolarized and horizontally opposed parallel to the E-vector of the overhead light) in clear water swam horizontally and perpendicularly to the E-vector of the overhead polarized light when its intensity was 20 times higher than that of the horizontal beams. When the intensity of the overhead polarized beam was less than this critical value, the daphnids swam toward the brighter source of non-polarized horizontal light.

Waterman (1960) observed that D. schodleri swam approximately perpendicularly to the E-vector of a vertical beam of totally linearly polarized white light both in clear and turbid waters and the precision of orientation was higher in turbid water. Later experiments with a usually vertical beam of totally linearly polarized light confirmed the polarization sensitivity of Daphnia, but failed to verify an adaptive significance of this ability or any evidence for its role in course steering (e.g. Jander and Waterman 1960; Waterman 1981, p. 322).

However, according to Schwind (1999), vertical beams oflinearly polarized light are not typical stimuli that Daphnia normally encounters. The sunlight that directly penetrates the water is not polarized. The polarized light wh ich Daphnia can see is diffuse, covering a wide field of view. It is either the part of the polarized skylight visible through the Snell window of the water surface, or the more or less horizontally polarized light produced by Rayleigh and Mie scattering in the water for high er solar elevations. For lower solar elevations there are considerable departures of the E-vector direction from horizontal (Waterman 1981, p. 300). The maximum degree of linear polarization p of this sidewelling light is ab out 40 %.

In a behavioural experiment, Schwind (l999) found that adult water fleas D. pulex, illuminated by a wide-field horizontally polarized white light with different p on the two sides, swim horizontally towards the side with high er p. When the E-vector of the lateral stimuli was vertical, Daphnia moved away from it. The response during horizontal swimming towards horizontally polarized sidewelling light is intensity invariant, because Daphnia swim in the direction of maximal p regardless of on which side the light intensity is high er or lower. This lateral polarotaxis is not influenced by a vertical light beam with different E-vector directions and light intensities.


As a result of Rayleigh and Mie scattering of light in apond, the light field surrounding the Daphnia is partially linearly polarized and has approximately a horizontal E-vector for higher solar elevations. According to the measurements of Schwind (1999) under overcast skies ne ar the shore, the distribution of polarization is inhomogeneous. The light coming from the direction of open water has higher p than that from the direction of the shore. Near the shore, p of light coming horizontally from the shore increases with the distance from the shore, reaching a maximum at several met res from it. This is because the diffuse light reflected horizontally from the oblique bottom of the shore is unpolarized, which reduces the net p of sidewelling light. The relative contribution of this unpolarized light to the polarized light scattered laterally in water decreases with increasing distance from the shore. Thus, in apond, swimming horizontally towards the lateral direction of maximal p leads Daphnia away from the shore, and polarized skylight from above does not interfere with this response. This offshore response explains a mechanism that underlies the well-known phenomenon of"shore flight" ("Uferflucht", Siebeck 1968). The ecological sense of this offshore re action is probably that it removes the animals from the vicinity of shoreline-inhabiting predators and/or reduces the disturbing effect of waves and water currents at the shore.

The polarization analysers in the eyes of Daphnia are also the rhabdomeres of the receptor cells, wh ich are so arranged that the microvilli in different cells are orthogonal to each other (Waterman 1981). Because a Daphnia floating in the water has no fixed orientation relative to the E-vector direction of the ambient polarized light field, the analyser system will usually have to be rotated to perceive the degree of polarization. Schwind (1999) suggested that this may be one of the functions of the marked rotatory movements of the eyes of Daphnia (Frost 1974; Consi et al. 1990).

The results of Schwind (1999) could partly explain the response of Daphnia to linearly polarized light with high intensity, when the animals swim horizontally and perpendicularly to the E-vector of the verticallight beam (Baylor and Smith 1953; Waterman 1960; Jander and Waterman 1960; Waterman 1981, p. 322). The vertically incident polarized light produces straylight which, if it is strong enough, can be perceived by Daphnia in the sideward directions of view. This straylight is differently polarized and even its intensity is variable when viewed from different lateral directions. In a lateral direction of view perpendicular to the E-vector of the downwelling polarized light, the E-vector of straylight is horizontal and its intensity is high, while in a lateral direction parallel to the E-vector of the downwelling light, the E-vector of straylight with a lower intensity is vertical. The polarized wide-field straylight with horizontal E-vector which is seen by the animals in two opposite directions in their lateral field of view could cause the animals to swim perpendicularly to the E-vector of the vertical beam of polarized light. Since the intensity distribution of straylight is inhomogeneous too in the lateral directions of view, it


would be also necessary to exclude that the animals merely show positive phototaxis.

The response of Daphnia in the polarized verticallight beam was more pronounced in turbid than in clear water (Waterman 1960), which indicates the implication of scattered light, since it is more intense in turbid water. In the experiments ofJander and Waterman (1960), Daphnia swam horizontally and perpendicularly to the E-vector of verticallight beam only if the aquarium walls were black. When these side walls were bright, the response became indistinct and other swimming directions were also observed. This also shows the significance of laterally scattered polarized light. The response became indistinct because the degree of polarization of the scattered light was reduced by the unpolarized light reflected diffusely from the bright side walls.

In laboratory experiments Novales Flamarique and Browman (2000) found that D. magna and D. pulex respond in different ways to various polarized light stimuli. In downwelling totally linearly polarized white light both closely related species oriented bimodally and perpendicularly to the E-vector independently of intensity gradients. In D. magna this behaviour was not influenced by totally polarized and less intense sidewelling light with arbitrary Evector direction. This suggests that it is the downwelling light field which is most important for this species in nature. Polarotactic orientation in D. magna may be an adaptation for displacements during crepuscular periods, when the sun is no longer visible. As described by Schwind (1999), under wide-field stimuli, D. pulex swam toward a sidewelling totally horizontally polarized light field irrespective of the downwelling polarized light field. From a vertically polarized sidewelling light source in contrast, it swam away.

Novales Flamarique and Browman (2000) found that the polarotaxis in D. magna is wavelength-independent. Under blue (455 nm), green (515 nm) and red (570 nm) totally linearly polarized illumination D. magna oriented perpendicularly to the E-vector. This contrasts with the response of D. pulex, in which polarotaxis is wavelength-dependent. In totally polarized blue (455 nm) light, D. pulex swam perpendicularly to the E-vector. Under green (515 nm) polarized light it swam randomly, while it oriented parallel to the Evector in red (570 nm). From these experiments Novales Flamarique and Browman concluded that D. magna uses a green-sensitive pigment for polarization detection, and D. pulex uses more than one visual pigment for this purpose.

However, until now it could not be concluded from these experiments that there were two different sets of analysers with different visual pigments. We mention he re only one example for another possible explanation: it is known that Daphnia can discriminate different colours (e.g. Smith and Macagno 1990). Daphnia pulex could recognize the background colour with its colour vision system and then decide to swim either parallel or perpendicularly to the E-vector of polarized light. Both courses could be maintained by one and the same system of orthogonal analysers, containing only one visual pigment.


If, however, the assumption of Novales Flamarique and Browman (2000) could be confirmed, then D. pulex would be the first animal shown to use more than one pigment in a polarization analyser system.

The functional and ecological significance of polarotactic orientation in Daphnia is not yet completely understood. Hazen and Baylor (1962) proposed aggregations into lines perpendicularly to the local E-vector as a mechanism to avoid mutual interference while feeding in food-rich surface waters. Under these conditions, polarization-induced swarms may minimize predation risk, because predators sometimes hesitate to attack aggregated prey (Neill and Cullen 1974), and on average, the individual is better protected (Hamilton 1971). Furthermore, optical mechanisms such as thin-film interference by cuticle layers when oriented perpendicularly to the wave front may operate to reduce spectral contrast to predators at short wavelengths (Giguere and Dunbrack 1990). In addition, Rayleigh scattering by microscopic algae could produce horizontally polarized light. Swimming perpendicularly to the direction of this polarization would lead Daphnia to the algal patches.

Schwind (1999) has elegantly shown that swimming towards the place of the highest degree of polarization leads to "shore flight" in D. pulex. The same may be the case with other small crustaceans that display this re action (e.g. Siebeck 1968). However, it is unknown whether other cues are also used in this response. In addition, not all Daphnia species respond in the same way to polarized light (see Novales Flamarique and Browman 2000). Further experiments, especially field experiments are necessary to explain fully the ecological and functional significance of polarotaxis in different Daphnia species.

25.11 Mantis Shrimps

Stomatopod crustaceans, commonly known as mantis shrimps, inhabit shallow tropical waters where they live within cracks and crevices of coral reefs or in burrows in muddy or sandy substrata. They are usually brightly coloured, hunt and capture prey with their two raptorial appendages, and possess a suite of sense organs, including compound eyes, which enable the detection and recognition of animals at a distance. The stalked eyes can perform largely independent free movements, sweeping their visual field through the optical environment (Land et al. 1990). Their apposition compound eyes have several thousand ommatidia and are divided into three parts: the dorsal region, the ventral region and the midband running around the equator of each eye. In members of the superfamilies Gonodactyloidea and Lysiosquilloidea the midbands are composed of six rows of ommatidia. Many ommatidia of both hemispheres and all six rows of ommatidia in the midband sampie the same narrow band in space, providing the possibility of stereoscopic vision in a single eye (Marshall and Land 1993).


Marshall (1988) and Marshall et al. (1991a) have shown that both the anatomicaUy and physiologicaUy specialized midband of Odontodactylus scyllarus, Gonodactylus chiragra, Gonodactylus oerstedii, Gonodactylus ternatensis, Gonodactylus smithii, Pseudosquilla ciliata, Coronis scolopendra, Lysiosquilla sulcata, Lysiosquilla scabricauda and Lysiosquilla tredecimdentata are adapted for polarization sensitivity. The rhabdoms of the lower and upper hemisphere of the eyes are typical of many crustaceans and consist of eight retinula ceUs RI-R8 arranged in two tiers, with ceU R8 producing the upper very short layer and RI-R7 with orthogonal microvilli the remainder. Midband rhabdoms are slightly longer and twice as wide as those of the hemispheres, midband rows 5 and 6 are two-tiered but the upper R8layer is much longer than that in the hemispheres. There are three tiers in midband rows 1 and 4, and five in rows 2 and 3. In the latter two midband rows, aseries of carotenoid colour filters screen the photopigment and potentially provide input for contrast-enhanced colour vision (Marshall et al. 1996). The colour filters are blocks of red, orange, yellow, purpie, pink or blue droplets within the rhabdom.

In the midband rows 5 and 6 the microvilli of the distal R8 receptors have only one direction and these highly parallel microvilli in row 5 are perpendicular to those in row 6. The R8 cells in the midband rows 5 and 6 are longer than in the rest of the eye. In row 5 the microvilli of proximal cells Rl, R2, R5, R6 are perpendicular to those of R3, R4, R7. In row 6 the proximal microvilli Rl, R2, R5 are perpendicular to the R3, R4, R6, R7 microvilli. Axons from these two receptor sets converge to high er order interneurons (Cronin and Marshall 2001), which may explain how polarization opponency arises in mantis shrimp vision (Yamaguchi et al. 1976). All these proximal microvilli are oriented at ±45° with respect to the distal R8 microvilli.

Marshall (1988) suggested that the orthogonal R8 microvilli may be part of a two-channel polarization-analyser system with output comparison between rows 5 and 6. Alternatively, both rows 5 and 6 may contain aseparate threechannel polarization-analyser system with output comparisons, in row 5 between R8, RI-R2-R5-R6 and R3-R4-R7, while in row 6 between R8, RI-R2-R5 and R3-R4-R6-R7. The spectral sensitivities of R8 and RI-R7 cells in rows 5 and 6 of the midband are maximal near 350 and 500 nm, respectively (Cronin and Marshall 2001). Thus, in both types of the threechannel polarization-sensitive system suggested by Marshall (1988) polarization-induced false colours would be inevitably generated (see Chap. 33). This suggests that such a three-channel system may not be in operation here (Marshall et al. 1991a). However, to prevent such confusion between polarization and colour in the ommatidia specialized for spectral analysis the rhabdoms are polarization insensitive, or only weakly polarization sensitive, having randomly or bi-directionally oriented microvilli (Marshall et al. 1991a).

Marshall et al. (1999) trained Odontodactylus scyllarus and Gonodactylus chiragra by operant conditioning to feed from white Plexiglas cubic or cylin-


drical containers, on one side of which a linearly polarizing filter was cemented. When presented with a choice of three randomly arranged containers, none of which contained food and only one of which offered the Evector orientation or pattern to which animals had been trained, stomatopods chose the container to which they had been trained, at levels significantly above chance. In the first experiment, Odontodactylus scyllarus was trained to a polarization contrast pattern composed of two adjacent triangles of linear polarizer with orthogonal transmission axes. This pattern was backed by the white Perspex square face of a cube. In the test, the shrimps had to choose between the polarizing cube or two different neutral density cubes. The animals could learn to choose a polarization contrast in preference to a brightness contrast. Since the polarizing cube may have appeared as a brightness pattern to the polarization-sensitive visual system of stomatopods, and/or subtle differences between the polarizer and neutral density filters may have been visible to the animals, this experiment could not conc1usively demonstrate E-vector discrimination.

In the second experiment of Marshall et al. (1999), with containers covered by different neutral density filters, it was tested whether Odontodactylus scyllarus and Gonodactylus chiragra could discriminate between containers on the basis of intensity alone. They could not. In the third experiment, Odontodactylus scyllarus and Gonodactylus chiragra were trained to feed from a cube or cylinder, on one end of which a single round piece of polarizer was cemented. Individual animals were trained to horizontal or vertical E-vector. In test, they had to choose from three cubes or cylinders, one with the trained E-vector direction, the other two with E-vectors orthogonal to this. Both species chose the trained E-vector direction significantly above chance levels. From these Marshall et al. (1999) conc1uded that Odontodactylus scyllarus and Gonodactylus chiragra can be trained to a particular E-vector direction.

With video polarimetry Marshall et al. (1999) found strong polarizing ability and optical activity on the telson and the antennal scales of Odontodactylus scyllarus and Gonodactylus chiragra. They suggested that one function of polarization sensitivity in these stomatopod species may be visual signalling and communication.

Marshall et al. (1991a) as well as Cronin and Marshall (2001) hypothesized that the distal receptor R8, with its parallel microvilli, in midband rows 5 and 6 could serve as a quarter-wavelength retarder. If so, it could convert circularly polarized light, e.g. reflected from the birefringent cutic1e of certain crustaceans (Neville and Luke 1971), to elliptically polarized light, which could then be analysed by the polarization-sensitive cells RI-R7 of the proximal part of the rhabdom.

Each eye of mantis shrimps can be rotated by up to 70° about the eyestalk axis (Land et al. 1990). This will change the orientation of the microvilli relative to a given fixed E-vector direction of linearly polarized light. Theoretically, this could enable the animals in combination with a single-direction


microvillar array to analyse polarized light patterns in aserial manner (Kirschfeld 1973b).

The retina of mantis shrimps has a great diversity of visual pigments and spectral receptor classes. Cronin et al. (2000) found that in 12 species of gonodactyloid stomatopods from a variety of depths of photic environments, the wavelength Amax of maximum sensitivity of the peripheral photoreceptors, being outside the midband and responsible for spatial vision and motion detection, decreases with increasing habitat depth. This trend, wh ich is common in aquatic animals (Lythgoe 1979), is an adaptation to the feature of aquatic optical environments, in which the wavelength of maximal light intensity decreases with increasing depth. The receptors of midband rows 1-4 are narrowly tuned for maximum coverage of the spectrum of irradiance available in the habitat of each studied mantis shrimp species. On the other hand, the photoreceptors of midband rows 5 and 6, which are specialized for perception of polarization, are maximally sensitive at Amax ::::: 500 nm in all investigated species, independently of their habitat depth. This approximately constant Amax of the polarization-sensitive receptors is rather surprising, because also here a tendentious decrease of Amax with increasing habitat depth could be expected. The reason for this is still unknown. Note that the polarization sensitivity of stomatopod crustaceans does not function in the UV, unlike in several fish species (see Chap. 28), although UV-receptors also exist in stomatopod retinas (Marshall et al. 1991a).

26 Polarization Sensitivity in Cephalopods and Marine Snails

26.1 Cephalopods

AIthough cephalopods can perceive linear polarization, it remains to be discovered whether colour-blind cephalopods can discriminate between intensity and polarization.1t has been suggested that certain polarization-sensitive cephalopods could communicate with each other by the polarization patterns of their body surface. However, the actual use of polarization signals for communication has not been confirmed, nor has the information mediated by these signals (if any) been identified.

26.1.1 Octopuses

Moody and Par riss (1960, 1961) trained octopuses (Octopus vulgaris) by food reward and electrical shock to attack a small underwater torch emitting totally linearly polarized light with different E-vector directions. The animals could discriminate between vertical and horizontal E-vectors as well as between 45° and 135° oblique E-vectors with respect to the horizontal, but they discriminated between the latter two E-vector directions significantly less well than the former ones. Moody and Parriss could not rigorously exelude the possibility that the animals responded to unwanted brightness patterns induced either by selective reflection of polarized light from the floor and walls of the test tank, or by selective scattering of polarized light in the turbid water within the tank. They suggested that polarization sensitivity of octopuses may be useful to find the solar direction by means of the celestial polarization pattern.

The retina of Octopus vulgaris consists of pigment-containing retinula cells separated from each other by rhabdomes (e.g. SchuItze 1869). Moody and Parriss found that the square rhabdome is composed of four rhabdomeres, each consisting of a elose-packed group of parallel microvilli (Fig. 26.1). The rhabdomeres of two opposite retinula cells have parallel microvilli, to wh ich the

268

A

microvi ll i

direct ion of incident light

retinula cell

microvill i

pigment granules


Fig.26.1. A Schematic structure of an octopus photoreceptor. B Tangential cross section of the receptor layer in the octopus retina showing a square rhabdome composed of four rhabdomeres, each consisting of a dose-packed group of parallel microvilli. Two rhabdome res have microvilli axes pointing in one direction, and the microvilli of the other two rhabdomeres are perpendicular to this direction. (After Moody and Parriss 1961).

microvilli of the other two opposite rhabdomeres are perpendicular. This structure is responsible for the polarization sensitivity of Octopus vulgaris. The orientation of the microvilli is approximately vertical or horizontal, and this may be the reason why the an im als could discriminate significantly less between E-vectors tilted at 45° and 135° from the horizontal than between vertical and horizontal E-vectors.

Rowell and Wells (1961) as well as Tasaki and Karita (1966) have shown that the octopus retina is able to discriminate between different E-vector directions of totally linearly polarized light. By means of electrophysiological recordings from single retinular cells, Sugawara et al. (1971) found that octopus photoreceptor responses encode the direction of polarization.

Shashar and Cronin (1996) trained octopuses (Octopus vulgaris and Octopus briareus) to distinguish between underwater targets on the basis of the presence or absence ofE-vector contrasts. Targets were constructed from vertical rectangular colourless linearly polarizing filters, to the back of which a wax-paper depolarizer was attached. A small circ1e was cut from the centre of each filter and was replaced in such a way that the angle ße of its transmission axis from the vertical could be changed relative to the angle ßb of the transmission axis of the rectangular filter serving as background. The test tank was illuminated from above bywhite light. The animals could detect E-vector contrasts (ße t:- ßb)' They did not discriminate between E-vector contrasts when the depolarizer side of the targets faced them. Octopuses recognized a polarization contrast Iße-ßbl = 90° within a single target, when ß/ßb was 0/90° and 90/0° or 45/135° and 135/45°, respectively.

26 Polarization Sensitivity in Cephalopods and Marine Snails 269

These results are in accord with those obtained earlier by Moody and Parriss (1960, 1961) and Moody (1962). The major difference occurred in the method: Shashar and Cronin (1996) presented a given pair of different E-vectors simultaneously, while Moody and Parriss presented it sequentially. Shashar and Cronin expanded earlier work from stimuli with homogeneous E-vector patterns to stimuli with heterogeneous E-vector patterns. They concluded that octopuses can recognize objects on the basis of their polarization pattern, wh ich has also been shown by Moody and Parriss. The novel result achieved by Shashar and Cronin was that octopuses are able to transfer their learning to new polarization patterns and the minimal detectable angle of polarization difference Ißc-ßbl is between 10° arid 20°. They suggested that polarization sensitivity in octopuses may enhance the contrast between the partially linearly polarized background light field and underwater objects, and this may help to identify prey and predators, or to communicate with other octopuses. However, until now these hypotheses have not yet been tested behaviourally.

26.1.2 Squids

Jander et al. (1963) demonstrated polarization-based body orientation in two Hawaiian squid spedes, Euprymna morsei and Sepioteuthis lessoniana. In the retina of the long-finned squid Loligo pealei, each outer segment of the photoreceptors comprises two microvillar regions separated by a cytoplasmic core (Fig. 26.2). In any local area of the retina, the microvilli of all outer segments are orientedin one oftwo orthogonal directions (e.g.Zonana 1961).On the basis of electrophysiological recordings and the orthogonal orientation of microvilli in outer segments of adjacent photoreceptors, Saidel et al. (1983) suggested that in Loligo pealei any particular E-vector is perceived as a different intensity. They proposed that squid vision is based on two complementary views of the world, each determined through the transformation by polarization-sensitive receptors into complementary intensity scales. They suspected that changes in postural orientation that follow changes in the E-vector direction of polarized light underwater may aid visual perception by allowing the animal to orient for the most favourable view to observe objects against the partially linearlypolarized light field. A visual system based on this transformation would lead to enhanced underwater contrast and visualization of object details obscured by confounding specular reflections (Rowe et al. 1995).

By means of video polarimetry, Hanlon et al. (1999) studied the linear polarization patterns of the body surface of Loligo pealei, which is pelagic, active day and night and exhibits a complex range of sodal interactions, accompanied by various body pattern displays. Squids have a highly developed system of visual communication. This takes pi ace mainly through the chromatophore system of the skin, which can quickly change the colour pat-


Fig. 26.2. Tangential cross section of the retina in the long-finned squid Loligo pealei showing the orthogonal arrangement of the microvilli, the average directions of which are indicated by doubleheaded arrows (after Saidel et al. 1983).

tern of the body for camouflage or communication. Loligo pealei, for example, has yellow, red and brown chromatophores. Expansion of these chromatophores darkens the skin, while their retraction and the resultant expression of underlying iridophores brightens the skin. Iridophores also play a very significant role in the intensity patterns on squids. Squids probably perceive these intraspecific visual signals monochromatically, because cephalopods are thought to be colour-blind. Hanlon et al. (1999) described the following four components of surfacial polarization in Loligo pealei:

1. The arms can reflect highly polarized light with maximum degree of linear polarization Pmax::::: 75 %. The E-vector direction of reflected light can be the same or different in the arms, and it can change quickly within 1 s without any movement of the animal or change in colouration.

2. The dorsal part of the mantle occasionally reflects partially linearly polarized light.

3. The iridophore cells surrounding the eyes can reflect partially linearly polarized light.

4. The skin surface can also reflect partially linearly polarized light with Pmax

< 50%.

The physical mechanisms of the first two components are unclear. In squid iridophores there are layers of intracellular platelets that are parallel to each


other (Mirow 1972). The colour oflight reflected by these platelets can change from reddish to bluish and depends on the distance between the platelets, their orientation as well as the direction of view (Mäthger and Denton 2001). These platelets can strongly polarize the reflected light if the angle of inciden ce is equal or near to the Brewster angle. In squid dermis, such iridophores have been found beneath the layer of chromatophores. Iridophores are found in many parts of the squid skin. In most species they are especially abundant on the mantle. Shashar et al. (2001) observed strongly polarized light reflected from specific lines along the arms of Loligo pealei associated with blue or pink colours, and they observed iridophores in the skin tissue at these locations above the chromatophores. They found also a correlation of neurological stimulation of iridophores with changes in reflection polarization. The rapid change of the polarization pattern within 1-2 s suggests neural rather than hormonal control. Hanlon et al. (1999) hypothesized that squids may use their surfacial polarization patterns for intraspecific communication. Furthermore, these polarization patterns may playa role in the vision of polarizationsensitive predators which may be looking for polarization contrasts to locate squid prey.

Shashar and Hanlon (1997) also described a few specific polarization patterns of the body surface in another squid species, Euprymna scolopes and correlated them with the distribution of iridophore cells in the skin. Reflected light with Pmax = 80 % was measured from all body parts of Euprymna scolopes, but no specific pattern could be identified. Euprymna scolopes is a predominantly nocturnal and solitary predator, which mates at night, and apart from this brief event, the animal does not seem to engage in complex social interactions. The function of the polarization pattern of light reflected from the body surface of squids, if any, is unclear, since they appear unrelated to the animal's behaviour. Polarization patterns were also recorded on the body surface of octopuses (Cronin et al. 1995).

Contrast enhancement and detection range increase have long been suggested as possible roles for polarization sensitivity (e.g. Briggs and Hatcett 1965. Lythgoe and Hemmings (1967) reported an increase of about 20% in target detection range when the target is viewed in such a way that it is sandwiched between linearly polarizing filters with orthogonal transmission axes. Recording the polarizational characteristics of a vertical Plexiglas covered by strip es with different transmission and reflection properties as a function of the target distance in a shallow mangrove channel, Shashar et al. (1995a) determined that the underwater target detection range can increase by up to 82 % for transparent objects, which depolarize light, while only by 12-14 % for objects with an intensity contrast. They hypothesized that a similar improvement in detection range is likely to exist for transparent organisms such as zoo plankton, if their tissues depolarize light.

Many planktonic animals are largely transparent, and therefore hard to detect visually by predators (e.g. Johnsen and Widder 1999). The muscle


fibres, rhabdomeric photoreceptors, the antennae and cutide of certain zooplanktonic species slightly polarize the transmitted light. Polarization sensitivity could enable predators to improve detection of their transparent prey, if there is polarization contrast between the prey and the background. In a behavioural experiment, Shashar et al. (l998a) found that adult Loligo pealei preferred transparent heat -stressed beads (of 1 cm diameter polarizing the underwater light passing through them) over non-stressed beads (of the same size affecting only slightly the polarization of transmitted light). In a waterfilled tank illuminated through a side window with either linearly polarized or depolarized light, hatchling Loligo pealei attacked living planktonic prey under illumination with polarized light at a 70 % greater distance than under illumination with depolarized light.

Long-finned squids (Loligo pealei) migrate perpendicularly to the shoreline each year in the NW Atlantic. To test whether these onshore/offshore movements could be governed by the polarization of skylight, Shashar et al. (2002a) placed hatchlings, juvenile and adult Loligo pealei in a circular tank illuminated from above by totally linearly polarized white light with different E-vector directions (0/180,45/225,90/270, 135/315°) relative to a reference direction. To limit possible brightness artefacts, the tank was painted black and white in alternating sectors of 15°. In this laboratory experiment no preferred swimming and body orientation of the squids was found with respect to the overhead E-vector direction.

In electrophysiological experiments, Shashar et al. (2002b) found that the change Lla of the E-vector direction of totally linearly polarized light stimulus which can elicit a single spike in a photoreceptor of Loligo pealei ranges from 2 to 9°, or from 5 to 22° when considering receptor noise. Lla was lowest (2 or 5°) and highest (9 or 22°) at 45 and 0° from the microvilli direction, respectively. These limits could be explained partly by the variability of the alignment of microvilli in the photoreceptors around a dominant axis.

26.1.3 European Cuttlefish Sepia officinalis

Cuttlefish are diurnal animals and colour-blind (Hanlon and Messenger 1996) having only one visual pigment maximally sensitive to wavelength A = 490 nm (Shashar et al. 1998b). Electron-microscopical examination of the retina of the European cuttlefish Sepia officinalis revealed the typical orthogonal orientation of the microvilli in the rhabdomeres of neighbouring photoreceptors, suggesting polarization sensitivity (Shashar et al. 1996). Electron-microscopical studies of the skin of the arms of cuttlefish revealed that the iridophores present throughout the skin contain numerous parallel platelets, which are expected to induce partial polarization of reflected light. Using underwater video polarimetry, Shashar et al. (1996) observed prominent horizontally polarized patterns on the arms, around the eyes and on the forehead


of Sepia officinalis while the animal was cruising, hovering or lying alert on the bottom. The polarizing regions corresponded to the pink iridophore arm stripes, the eye ring and the posterior head bar. These polarization patterns disappeared when the animals were camouflaged on the bottom and also during extreme aggression, attacks on prey, copulation and egg-laying in females.

In a behavioural experiment, the responses of Sepia officinalis to their images reflected from a mirror changed significantly when the polarization of reflected light was distorted (Shashar et al. 1996). Depolarization was performed by a filter made of Pyrex glass previously heated and cooled several times to create stress within it. This filter did not affect the intensity and colour of transmitted light. While cuttlefish tended to retreat from their reflected image with polarized light, they more often stayed in place without a noticeable response when the polarization was distorted. Shashar et al. (1996) suggested that the polarization sensitivity of the eyes and the polarization patterns of the body surface of cuttlefish may serve as intra-specific recognition and communication.

The silvery scales of fish, reflecting a broad band of the spectrum, are considered to function in radi an ce matching, resulting in reduced detection by predators or prey (e.g. Denton and NicolI966). Light reflected from fish scales can possess polarizational properties which are different from those of the background underwater light field (e.g. Rowe and Denton 1997). Shashar et al. (1995 c) demonstrated such a polarization contrast between a fish and its surroundings. Denton and Nicol (1965) suggested that polarization sensitivity could be used to detect fish with a distinct polarization pattern. Tyo et al. (1996) showed that a polarization-sensitive sensor can double the detection range of reflecting and polarizing targets, as compared to an intensity sensor. They demonstrated that polarization sensitivity is especially useful if a great portion of the illumination reaching the sensor originates from scattering of light in water. Various fish species are an important food source of adult cuttlefish.

In a behavioural experiment, Shashar et al. (2000) presented freshly killed and frozen butterfish (Peprilus triacanthus) to adult Sepia officinalis. The side and dorsal part of these fish reflected partially linearly polarized light with p = 30-50 % and 70-95 %, respectively. The fish were moved horizontally back and forth and could be seen by the cuttlefish through two layers of Roscolux transparent filters. When two such filters are attached in such a way that their optical axes are parallel, they depolarize totally linearly polarized incident light by 6 %. When the optical axes of the layers are at 45° to each other, they depolarize through partial circular polarization the totally linearly polarized light by 70 %. Sepia officinalis preyed preferably on fish with only slightly depolarized reflection-polarization patterns over fish, the reflection-polarization patterns of wh ich were largely distorted by depolarization. Shashar et al. (2000) suggested that polarization sensitivity may help cuttlefish to detect reflecting fish scales and improve their predation success by breaking the


countershading camouflage of silvery fish. This hypothesis should still be tested.

26.2 Marine Snails

The polarization sensitivity in certain marine snails is controversial. In behaviourallaboratory experiments Burdon -J ones and Charles (1958), Baylor (1959) and Charles (1961a,b,c) investigated the spontaneous polarotaxis of the marine snails Littorina littorea, L. littoralis, L. neritoides, L. saxatilis and Nassa obsoleta. The snails responded when crawling on horizontal surfaces in air or sea water, upright or upside down to totally linearly polarized white light incident from above. When the light was unpolarized, they oriented randomly. The direction of their track changed approximately by 90° when the Evector direction was rotated by 90°. Photonegative or photopositive snails crawled parallel or perpendicularly to the overhead E-vector, respectively. When the overhead source of polarized light was invisible but the substratum visible, they moved perpendicularly to the E-vector.

Unfortunately, the optical surroundings of the animals abounded in black reflecting surfaces and the illuminating light reflected from a tilted plane mirror was not depolarized before passing the polarizer. An orientation by means of the brightness pattern of light reflected from the black surrounding surfaces can explain the surprising and biologically absurd difference in the orientation of phototactically positive and negative animals relative to the E-vector direction: Photopositive snails oriented perpendicularly to the E-vector, because from this direction much amount of light was reflected from the inner black surface of the vertical cylinder, while photonegative snails crawled parallel to the E-vector, because from this direction the black cylinder reflected less amount oflight (see Chap. 34). From these experiments it is clear that under the polarizer the snails oriented by means of the intensity pattern of polarized light reflected from the surrounding surfaces and not direcdy by means of the E-vector direction.

Charles (1961a,b,c) hypothesized that Littorina species perceive the polarization by means of an extraocular mechanism based on Fresnel refraction: the amount of linearly polarized light reaching the retina (supposed to be polarization-blind) is the larger, the narrower the angle between the E-vector and the plane of refraction at the refracting surfaces of the eye. The role of such a mechanism in animal polarization sensitivity was later refuted (see Sect. 16.7.1). Finally, if the alleged spontaneous polarotaxis of the above-mentioned marine snails indeed existed, its biological function (if any) would be completely unknown.

On the basis of the above, the following conclusion can be drawn: Both the polarization sensitivity and the spontaneous polarotaxis of the mentioned


marine snail species are not proven. All orientation experiments with these snails should be repeated under optically well-controlled conditions, at which the intensity patterns induced by selective reflection of the incident linearly polarized light should be eliminated (see Chap. 34).

27 Polarization-Sensitive Optomotor Reaction in Invertebrates

If a pattern of vertical black and white stripes is rotated around an animal, it usuaUy displays a turning reaction. The tendency for the animal to turn in the direction of motion of a pattern is caUed "optomotor response", which demonstrates that the animal is able to detect the movement of the optical environment on the basis of brightness cues. This behaviour serves to stabilize the animal's orientation with respect to the environment, and helps it to maintain a straight course during locomotion. The optomotor reaction of insects to black-and-white (B&W) patterns has been intensively studied (e.g. Hassenstein and Reichardt 1956; Varju 1959). Studies of the dependence of motion perception on the wavelength of light demonstrated that the visual subsystems performing directionaUy-selective movement detection are usuaUy colour-blind (e.g. Kaiser 1974; Lehrer et al. 1990).

If these visual subsystems are sensitive to polarization, it is to be expected that an optomotor response can also be elicited by the movement of stripes of linearly polarizing filters with alternating orientations of their transmission axes. The strength and phase difference of the optomotor response to rotating B&W patterns should depend on the orientation of the transmission axis of a polarizing filter positioned between the animal and the pattern. Such experiments with crabs, honeybees, flies, backswimmers and waterstriders have shown that depending on the orientation of the polarization-sensitive microvilli system in the eyes of these animals, optomotor responses can be elicited by different E-vector patterns. In this chapter, first the published results are briefly surveyed. Then the results of the experiments performed by G. Horvath, H. Blanke, H.-J. Dahmen and D. Varju (unpublished), who studied the polarization sensitivity of the optomotor response of the backswimmer Notonecta glauca and the waterstrider Gerris lacustris, are presented.

27.1 Crabs

Korte (l965) observed polarization-induced optomotor response in the European fiddler crab Uca tangeri. According to Kirschfeld (l973b, p. 291), opto-

27 Polarization-Sensitive Optomotor Reaction in Invertebrates 277

motor response can be elicited in the crab Carcinus with white, blue or orange light, if the alternating E-vectors of the moving polarization pattern are parallel and perpendicular to the dorso-ventral plane of the eyes. However, if the E-vector directions were ±45° with respect to the dorso-ventral plane, optomotor response was not observed.

27.2 Honeybees

De Vries and Kuiper (1958) investigated the optomotor reaction ofhoneybees Apis mellifera to a moving pattern of stripes of linear polarizers with alternating vertical and horizontal E-vectors. In this experiment the bees did not have optomotor response to the E-vector contrast, since the polarizationinsensitive lateral and frontal eye regions were stimulated. In Apis mellifera the optomotor response appeared onlywhen the E-vectors were oriented ±45° from the dorso-ventral plane of the eyes (Kirschfeld 1973a). In this reaction the blue- and/or UV-sensitive receptors participate. Alternating horizontal and vertical E-vectors in the frontal eye region did not elicit optomotor response.

27.3 Flies

Kirschfeld and Reichardt (1970) recorded optomotor reactions under open loop conditions of walking houseflies Musca domestica as a function of the Evector orientation of the stimulus light and found a sinusoidal modulation of the strength of the response. The E-vectors of the stimulating white, blue or orange light had to be ±45° from the dorso-ventral plane of the eyes to produce this reaction. Alternating E-vectors parallel and perpendicular to the dorso-ventral plane in the lateral eye region elicited no optomotor response. With electrophysiological recordings in optomotor experiments, McCann and Arnett (1972) studied the spectral and polarization sensitivity of wild-type adult Musca domestica, Calliphora erythrocephala and Phaenicia sericata (see Chap.17.2).

Polarization sensitivity of the optomotor response in flying fruitflies Drosophila melanogaster was investigated by Heisenberg (1972) in a pilot experiment, in which a rotating cylinder composed of two polarizing filters with E-vectors ±45° to the vertical was used as a stimulus. Repeating the optomotor experiment of Kirschfeld and Reichardt (1970) with Drosophila, Wolf et al. (1980) thoroughly studied the polarization sensitivity of course control and optomotor re action of fruitflies. They found that polarization sensitivity is mediated by the peripheral retinula cells R1-R6 in the ommatidia. Although


the amplitude of the optomotor response of walking Drosophila was a sinusoidal function of the E-vector orientation, the phase and amplitude did not reflect directly the polarization sensitivity of the photoreceptors mediating the re action. This suggests that Drosophila has an inner representation of the E-vector orientation, which is abstracted from the alignment of the dichroic microvilli. According to the original interpretation of Kirschfeld and Reichardt (1970), the modulation of the optomotor response due to the change of E-vector orientation should correspond to the change in the perceived brightness and thus to the polarization sensitivity of the receptors. Obviously, this is not the case in Drosophila, for which the polarization sensitivity estimated on the basis of the optomotor response is PS = 20-60, values considerably higher than typical PS-values in insect photoreceptors.

In a closed-Ioop situation, in which the fruitflies were illuminated from above by linearly polarized light and were allowed to turn the orientation of the E-vector relative to their body axes by their yaw torque, the animals could maintain their optomotor balance, Le. they could use the polarization to fly straight. Above a rotating polarizer covering a fraction of the ventral visual field with an angular diameter of 45° just underneath the animal, Drosophila displayed a significant optomotor response at 550 nm.

27.4 Rose Chafers

Mischke (1984) investigated the polarization sensitivity of optomotor reacti on in the African rose chafer Pachnoda marginata. The lateral, middle regions of the superposition eyes were stimulated by different oscillating (6 Hz) patterns containing intensity and/or colour and/or E-vector contrasts. The scarab beetle displayed optomotor response exclusively to intensity contrasts. Mischke concluded that Pachnoda is insensitive to polarization contrasts and was surprised, because the related scarab beetle Lethrus can orient menotactically by means of polarization (Frantsevich et al. 1977). Since in the experiment of Mischke the dorsal rim area of the eye of Pachnoda marginata was not stimulated, only the polarization insensitivity of the lateral, middle eye regions can be concluded.

27.5 Optomotor Reaction to Over- and Underwater Brightness and Polarization Patterns in the Waterstrider Gerris lacustris

Waterstriders (Gerrids) have trichromatic colour vision (Hamann and Langer 1980; Bartsch 1991) and polarization sensitivity (Bohn and Täuber 1971;


Bartsch 1995). The high polarization sensitivity of their photoreceptors to vertieally and horizontally polarized light is not restrieted to a special eye region or to a distinguished spectral region. One of the functions of this polarization sensitivity is to find the aquatie habitat by means of the partially and horizontally polarized light reflected from the water surface (Schwind 1991). This task requires a ventral polarization-sensitive eye region. Schwind (i985b) proposed that this ventral polarization -sensitive visual pathway is UV-sensitive in Gerrids. The high polarization sensitivity in the whole eye of waterstriders (Bartsch 1991,1995) raises the question of its functional signifieance in the dorsal and lateral eye regions.

Waterstriders compensate for displacement and rotation of their body due to water flow or wind by two distinct types of behaviour (Junger and Varju 1990; Junger 1991; Junger and Dahmen 1991):

1. To compensate for linear displacement, they periodieally jump against the direction of drift such that, on average, they maintain their position relative to the surroundings over a long time.

2. For rotation they compensate by a precisely combined rotation of head and body, such that the gaze is stabilized in space.

Both behaviour types are visually controlled. With this in mind, the function of the polarization sensitivity of the lateral and dorsal eye regions in waterstriders could be to enhance the contrast of objects in the optieal environment used for motion detection. The optieal environment of waterstriders, composed of the water surface below the animal, the vegetation on the shore and the sky above the animal, is rieh in polarization patterns with different degrees and angles of polarization, whieh could serve for contrast enhancement. To test whether these polarization patterns can be exploited for motion detection, Horvath et al. (unpublished) investigated the optomotor response of Gerris lacustris to different polarization patterns and compared it with the optomotor reaction to B&W patterns.

Male and fern ale Gerris lacustris were collected from ponds and kept in a watertank. They were fed on wingless Drosophila from a culture. Before an experiment the animals were narcotised by COz for some seconds. Then they were fixed by a wax drop on their pronotum to a 2-mm-Iong vertieal piece of a needle, whieh was glued to one end of a horizontal thin plastie strip of about 2.5 cm length and 5 mm width. The other end of the strip was adjusted by a holder to an appropriate height above the water surface. In this way, the animals on the water surface were allowed to roll, pitch and rise to some degree in order to accommodate themselves, but were prevented from yaw and displacement of their body. On the head of the animal, a pin, whieh was a hair of about 1 cm length of a shaving brush, was fixed by a wax drop let. The function of this pin was to display the head orientation during video recording (Fig.27.1).

280

A

L

o o o o


lateral

1

T

c ~D dorsal ~

stimulation ~ CL

Fig. 27.1. Schematic diagrams of the arrangement of the optomotor experiments for stimulation of the lateral (A), ventral (B) and dorsal (C) eye regions in waterstriders Gerris lacustris. V, video camera; W, water; PM, plane mirror; P, pin (on the head of the animal); CM, conical mirror; T, tube; Cl, first plexi cylinder (fixed); C2, second plexi cylinder (vertically moveable, to alter the water level in Cl); H, oscillating cylindrical holder with two different stimulus patterns above (in a given experiment only one of them was visible to the animal) and a black and white striped pattern below (invisible to the animal); 0, occulter (vertically moveable by strings, to occlude one of the two stimulus patterns on the holder); F, colour filter (green or blue); K, moveable keeping (to change the vertical position of the animal following the change of the water level); L, incandescent lamps emitting white light; D, diffusor (white milky plexi glass); CL, condenser lens.


The animals were placed in the centre of a plexi cylinder (diameter 9.4 cm, height 12 cm), which was connected to a second one by a silicone tube (Fig. 27.1A). After the system was half-filled with water, the water level in the first cylinder could be changed by lowering or raising the second cylinder. To stimulate the ventral and dorsal eye regions, a horizontal disc consisting of 12, e.g. black and white, sectors was oscillated sinusoidally with an amplitude of 5° and a frequency of 1 Hz around its vertical symmetry axis immediately below (Fig. 27.1B) and above the animal (Fig. 27.1C). Ventrally and dorsally, this disc occupied a cone with an aperture of 120° and 90°, respectively (Fig. 27.2). The black and white sector pattern (B&Wsec, Fig. 27.2) consisted of matt black and thin semitransparent white paper providing a brightness

dor al stimuli

lateral stimuli

B&WIJ

VVPOL ITIT]

VhPOL[IB

±45POL kl'l <-x -> 30' 30'

~~I-B& Wsec ~ ~ POLsec

~-I_I

30'~~

ventral

ventral stimul i

Fig. 27.2. Frontal view of the visual field of the eyes of Gerris lacustris (black eyes and head) showing the different eye regions stimulated by various patterns. One of the two patterns on a given cylindrical pattern holder (Fig. 27.1A) stimulate the lateral eye region ±30° from the horizon due to the mirror image of the pattern at the water surface. The symbols outside the rectangle represent the different stimuli. B& Wsec, horizontal disc with alternating black and white sectors; POLsec, horizontal disc composed of sectors of linearly polarizing filters with alternating orthogonal transmission axes; B& W, vertical alternating black and white strip es; vvPOL, vertical stripes of polarizers with vertical transmission axes; vhPOL, vertical stripes of polarizers with alternating vertical and horizontal transmission axes; ±45POL, vertical stripes of polarizers with transmission axes alternating ±45° to the vertical. The orientation of the transmission axes of the polarizers is shown by bars.


contrast of 80 %. The polarizing sector pattern (POLsec) consisted of sec tors cut from a neutral density linearly polarizing filter (HN32, Polaroid) with the transmission axes of adjacent sectors perpendicular to each other (Fig. 27.2).

In order to stimulate the lateral eye region ±30° from the horizontal direction (Fig. 27.2), a vertical cylindrical pattern holder with two different patterns, one on the top and another below, composed of 12 segments in a panoramic arrangement, was sinusoidally oscillated around the animal (Fig. 27.1A). Two pattern holders with four different patterns were used (Fig. 27.2). In an experiment one of the two patterns on a given pattern holder was occluded by a cylindrical matt black occulter, which could be vertically lowered and raised (Fig. 27.1A). The water level with the animaion it was adjusted in such a way that the non-occluded stimulus pattern on the holder was seen by the animal at 0-30° above the horizon. Due to reflection at the water surface, the animal could also see the mirror image of the pattern at 0-30° below the horizon (Fig. 27.2). A change between the two stimulus patterns could be achieved by cautiously lowering or raising the water level together with the animal for 10-15 s without larger disturbance to the animal. This was important, because these reactions of the animal to two different patterns could be recorded without changing the physiological state of the an im al. The experiments were performed in white, green and blue light. In the latter two cases cylindrical colour filters (Käsemann, Germany) were used around the animal (Fig. 27.1A). The intensity of light emitted by the white lamps was adjusted in such a way that the intensities transmitted by these two colour filters were the same (Fig. 27.3).

The oscillation of the stimulus patterns and the reactions of the animals were video-recorded from above or below through tilted plane and conical mirrors (Fig. 27.1). After a frame-by-frame analysis, the orientations qJ of the stimulus pattern and the pin fixed onto the head of the animal were ascertained as a function of time. The resulting stimulus and response data were fitted by sinusoid functions with amplitude A, from which the closed loop gain gc = Arespons/Astimulus and the closed loop phase difference LlqJc = qJresponse -

1.0

~ 0.8 fr

~ '-

0.6 0

·go: OA '"

.g <l.l

0.2 .~ 1ii ""§

300 350 400 450 500 550 wavelength Ä (nm)

1.0

0.8

0.6

OA

0.2

0.0

600 650

Fig.27.3. Relative light intensities (dashed lines) transmitted through the blue and green filters versus wavelength I used in the optomotor experiments, as weH as the relative absorption curves A(l, continuous lines) of the blue and green receptors R1-R6 in the eye of Gerris lacustris. (After Bartsch 1995).


l(Jstimu!us were obtained. Then the open loop gain go = gj[1 +g/-2gccosL1l(J}1I2 and the open loop phase difference L1l(Jo = are tan(sinL1l(JjeosL1l(Jc - g) were calculated.

The stimulation of the ventral eye region of Gerris from nadir angle 0° to ±60° (Fig. 27.2) in white light did not induce an optomotor reaction for both the B&W sec and POLsec patterns, although about 30 % of the ommatidia look into these directions and see the underwater world (Varju and Horvath 1989; Dahmen 1991). This corresponds to the finding that Velia caprai does not respond to underwater brightness stimuli at all (Meyer 1970,1971). When the dorsal eye region of Gerris from zenith angle 0 to ±4S0 (Fig. 27.2) was stimulated in white light, a weak optomotor response (with an open loop gain go = 2S±3 % and phase difference L1l(Jo = 62±SO averaged for eight animals) was elicited by the B&Wsec pattern, but not by the POLsec pattern.

Stimulating the lateral eye region between 0 and 30° above the horizon and because of reflection at the water surface also from 0 to 30° below the horizon in white light, Gerris exhibited an optomotor response to the B&W pattern (Fig. 27AA) and the vhPOL pattern (Fig. 27AB), but did not respond to the ±4SPOL pattern (Fig. 27AC) and the vvPOL pattern (Fig. 27AD). The vvPOL pattern was used as a control to test whether the inevitable low brightness contrast at the borders of adjacent polarizing filters elicited a response. As we can see in Fig. 27 AD, there was no such areaction, even if the amplitude of the stimulus was sometimes quite large. The spatial resolution of the compound eyes of Gerris (Dahmen 1991) was apparently not high enough to perceive the low brightness contrast at the borders of the polarizing filters. The optomotor response to the vhPOL pattern (Fig. 27AB) is, therefore, exclusively due to the E-vector contrast.

As has been mentioned, in the experiments waterstriders could also see the mirror image of the vhPOL pattern (Fig. 27.2). The intensity of vertically polarized light reflected from the water surface is slightly weaker than that of horizontally polarized light, because the reflectivity of water is slightly lower for vertical polarization (Guenther 1990). Thus, after reflection the vhPOL pattern should have been partially transformed to a brightness pattern with a weak contrast. We repeated the experiment without water beneath the animals and with a double-height vhPOL pattern stimulating the lateral eye region at ±30° from the horizon. The average gain and phase difference of the optomotor response to this modified vhPOL pattern were the same as in the original case. Thus, the weak brightness contrast in the mirror image of the vhPOL pattern did not affect the optomotor response of Gerris.

The open loop gain and phase difference of the optomotor response of Gerris to vhPOL and B&W patterns in white, green and blue light averaged over eight animals are shown in Fig. 27.5 together with their standard deviations. In Fig. 27.6A the open loop gain of responses to the B&W pattern in white, green and blue light is plotted versus that of the vhPOL pattern. The strongest response is found in white light, the weakest in blue. The ratio


0.2

0.1

0

stimulus -0.1

0.2

0.1

0

-0.1 --1--~-~_-""::':=::;'::::::"-~-~--....----r--~---1

0.35 .....-------------------------.

0.25

0.15

0.05 o

-0.05

--t---_____________ response

stimulus -0.15 ---L-=:..-.,.-=,r-----=:r--~~__._=___,~-..___-__r_-__._-__1

response 0.6

0.2

0

-0.2

-0.6 0 4 8 12 16 20

time (sec)

A

IJ B&W

B

vhPOL

c 1/1,,1 ±45POL

D

[ill] vvPOL

Fig.27.4. Typical examples of the optomotor response of Gerris lacustris to different lateral stimuli in white light. The stimulus type is indicated by its symbol defined in Fig. 27.2. The abscissa is the time (in seconds) and the ordinate is the oscillating orientation (in radian) of the stimulus and the head. The response is practically zero to the ±45POL and vvPOL stimuli.


gainB&wfgainYhPOL is relatively constant independently of colour and physiological state of the animal (the latter varied only slightly between subsequent stimulations). Figure 27.6B shows the average open loop gains for the three different light fields. The open loop gain ratio, as indicated by the regression line, is gainB&wfgainYhPOL = 2.564. It is clear from Fig. 27.5B that waterstriders foHowed the lateral stimuli with the shortest delay in white light and with the longest delay in blue light. Furthermore, the phase difference for the vhPOL pattern was always larger than that for the corresponding B&W pattern.

In order to interpret these findings, we refer to the ultrastructure as weH as the colour- and polarization-sensitivity of the ommatidia in Gerris lacutris described in Chap.18.4 and shown in Fig.18.2. Since the light sources in these optomotor experiments did not emit UV light and the polarizing filters did not trans mit UV light, the UV-receptors cannot be involved in the optomotor response of Gerris. This and the result that waterstriders do not respond to ventral rotating brightness or polarization patterns suggest that self-rotation is detected by a visual pathway, wh ich is separate from that mediating polarotactic water detection. The contrast of the vhPOL pattern is

CvhPOL = (PS-l)/(PS+l):::: 75% (27.1)

both for the green and blue receptors due to the similar PS-values of about 7. The intensity of the stimulating green and blue light was adjusted to be the same (Fig. 27.3). In his electrophysiological recordings, Bartsch (1991, 1995)

Fig. 27.5. The open Ioop gain go (A) and phase difference ,1CPo (B) of the optomotor response of Gerris lacustris to the vhPOL (white columns) and B&W (grey columns) stimulus in white, green and blue light averaged over eight animais. The bars show the standard deviations.

200%

100%

0%

90'

45'

0"

.,j

" 'i=; co c. ] c: ~ 0

s. <l

" " " 2 @

" fi " -ä c. 0 .2 c ~ 0

."'''} A

green

ITE .::J ITE .::J ITE .::J

,t.POL u&w

286

350% ..-------,-,

• 300%

'" " 1250%

A

~ 200% • ~ ... o

":0 150%

"§ ~ 100% .§

.. • .. . •

•

• white il 50% g. IJ green .. blue

0% +--~-~----l 0% 50% 100% 150% 0% 50% 100% 150%

open loop gain g. for vhPOL stimulus IIE


Fig. 27.6. A The open loop gain go for B& W stimulus versus go for vhPOL stimulus of the optomotor response of Gerris lacustris in white (dots), green (rectangles) and blue (triangles) light for eight animals. Every symbol represents two subsequent optomotor responses to vhPOL and B& W stimuli. B The average of the above-mentioned data. The horizontal and vertical bars show the standard deviation of the open loop gains in the case of vhPOL and B& W stimuli, respectively. The slope of the regression line passing through the origin and fitted to the data points is gainB&wlgainvh_ POL = 2.564.

found a great variation of the absolute sensitivity. Nevertheless, the blue receptors tend to be slightly more sensitive than the green ones. If the blue receptors contributed remarkably to the optomotor response, a similar or even a somewhat higher gain could be expected for the blue compared to the green stimuli in these experiments. Instead, we found a gain ratio

gainB&wgreen/gainB&Wblue = 2.82. (27.2)

The green receptors were also stimulated by the blue light (Fig. 27.3). The ratio of the amounts of green and blue light absorbed by the green receptor (calculated from the spectral sensitivity of the green receptor and the equal transmitted light intensity of the green and blue filters, Fig. 27.3) is

(27.3)

Although nothing is known about the relation between stimulus intensity and gain of the optomotor response of Gerris, comparison of the ratios given in Eqns. (27.2) and (27.3) suggests that the blue receptor may not contribute to the optomotor re action, otherwise the gain ratio gainB&vlreen/gainB&Wblue should be less than 1.58. Thus, the optomotor response of Gerris is mediated by the RI-R6 green receptors, and the motion-sensitive visual pathway is colour-blind. This agrees wen with the findings that in many insect species the primary input to the optomotor pathway is via receptors RI-R6 (Coombe et al. 1989), and in honeybees and butterflies, for example, the optomotor response is green sensitive (Kaiser 1974; Horridge et al. 1983). For waterstriders, living in an optical environment dominated by green foliage on the shore, green receptors seem to be optimal to the task of detecting self-motion, or motion of dark objects (e.g. enemies) against a green background.


Since in Gerris the ±45POL pattern did not elicit optomotor response, whereas the vhPOL pattern did (Fig. 27.4), all rhabdomeres involved in this reaction should have either horizontal or vertical microvilli in the normal posture of the animal. This conclusion is in accordance with the anatomical findings of Schneider and Langer (1969) and with the electrophysiological findings ofBartsch (1991, 1995).

27.6 Optomotor Response to Over- and Underwater Brightness and Polarization Patterns in the Backswimmer Notonecta glauca

The backswimmer Notonecta glauca is polarization sensitive (see Sect. 18.5) and detects water by means of the horizontally polarized light reflected from the water surface (Schwind 1983a,b, 1984a,b, 1985a,b). Both of its above- and underwater optical environments are rich in polarization patterns:

• viewing from water through the Snell window, the celestial polarization pattern is seen, which is modified by refraction polarization (Horvath and Varju 1995);

• the foliage on the shore reflects strongly polarized light, which also can be seen through the Snell window;

• the light scattered in turbid water is strongly polarized.

The last phenomenon was demonstrated in the following way: a small aquarium was filled with turbid water from a pond with a dense growth of green-yellowish phytoplankton inhabited by backswimmers. The middle part of the aquarium was illuminated by a vertical, slightly divergent white light beam of an incandescent lamp. The polarization of light scattered by the suspended particles in water was measured by video polarimetry from the side. Figure 27.7 shows the obtained polarization patterns. The light scattered by the suspended phytoplankton is strongly polarized, and its E-vector is horizontal, Le. perpendicular to the incident light beam. Thus, in small turbid ponds the spatial distribution of scattering polarization is quite similar to that in lakes and seas.

In turbid waters the characteristic strongly polarized ring at 90° from the refracted sun light (see Fig. 14.1) develops within some centimetres around a backswimmer, light intensity progressively decreases due to absorption, and brightness contrasts are significantly reduced due to scattering (Lythgoe 1979). In such a contrast-poor, dimly lit, turbid optical environment the mentioned highly polarized ring around backswimmers could be well exploited for orientation. In their natural habitat backswimmers often hang upsidedown at the water surface in such a way that the angle of their longitudinal

288

<> u

" " 'Ö e -;:; § 1°·,


Fig.27.7. Scattering-polarizational characteristics of turbid water in an aquarium, the middle part of which is illuminated by a slightly divergent vertical white light beam and the polarized light scattered by suspended particles in water was measured from the side by video polarimetry at 550 nm in such a way that the optical axis of the video camera was horizontal (perpendicular to the incident light beam). Demonstration of the strong scattering polarization of light within some centimetres in turbid water with a dense growth of green-yeUowish phytoplankton from a pond inhabited by backswimmers Notonecta glauca.

bodyaxis is 31 0 from the surface (Fig. 18.5A). In this position they precisely compensate for passive translations and rotations of their body induced by water flow or wind. In this behaviour they rely mainly on visual cues (Blanke and Varju 1995; Blanke 1996). To elucidate whether the abovewater and underwater polarization patterns can be exploited for motion detection in backswimmers, Horvath et al. (unpublished) investigated the optomotor response of Notonecta glauca to different above- and underwater polarization patterns and compared it with the optomotor reaction to B&W patterns.

Backswimmers Notonecta glauca were collected from ponds, and were placed in the centre of a vertical plexi cylinder (diameter: 9.5 cm, height: 12 cm) half-filled with water. Translation of the animal's body was prevented by means of a small water-filled plexi cylinder (diameter: 3 cm, height: 6 cm) within the large one. In the small cylinder the animal could rotate freely around the vertical axis. The stimulation of the ventral, dorsal and lateral eye regions happened in white light with the same apparatuses and patterns as in the optomotor experiments with waterstriders (Fig. 27.1). Figure 27.8 shows the angular extensions of the stimuli within the visual field of Notonecta. The patterns were sinusoidally oscillated with 50 amplitude and 0.2 Hz frequency. The orientation of the longitudinal body axis and the angular position of the stimulus pattern were evaluated frame by frame, from which the closed and open loop gains and phase differences were calculated.

Stimulation of the dorsal and ventral eye regions induced optomotor reaction only to the B&W sec pattern. To the ventral B&W sec pattern there were

27 Polarization-Sensitive Optomotor Reaction in Invertebrates

overwaler (OW) lateral stimuli

B&WIJ

vvPOL[ili]

vhPOL[jE

±45POLkl, 1

B&WIJ

vvPOL[ili]

vhPOL[jE

±45POLI±] underwaler (UW)

lateral stimuli

\ \

\

\

\ \ ,

OW ,

\

\

B&WSeC~ ~ POLsec

overwater (OW) ventral stimuli

\ OW dors~a~1 ---;-_O_W dorsal

\ 32' 45'

\

2----------UW laleral

289

Fig. 27.8. Frontal view of the visual field of the eyes of Notonecta glauca (black eyes and head) showing the different eye regions stimulated by various patterns. The patterns on the cylindrical pattern holder (Fig. 27.1A) stimulate the lateral eye region ±30° from the horizon due to total reflection at the water-air interface. The symbols outside the rectangle represent the different stimuli, which are the same as in Fig. 27.2. At the right hand side of the figure the incident, refracted and reflected rays of light from the borders of the different stimulus patterns are traced by continuous lines, while at the left hand side the dashed lines show the corresponding directions of view of the Notonecta eye.

only weak responses (go = 7±2.7 %, t1qJo = 43±21.7°). The responses were relatively strong (go = 54±28.7 %, t1qJo = 33±18.3°) to the dorsal B&Wsec pattern. Backswimmers did not respond to the ventral and dorsal POLsec patterns.

The open loop gain and phase difference of the optomotor response to lateral B&W, vhPOL and vvPOL patterns are shown in Fig. 27.9. Stimulating the lateral eye region, backswimmers exhibited strong optomotor responses to the B&W pattern and a considerably weaker reaction to the vhPOL pattern. The optomotor responses to vvPOL and ±45POL patterns were the same and veryweak. The vvPOL pattern was used to control the weak response induced by the weak inevitable brightness contrast at the border lines between adjacent polarizing filters. Although the response to the vvPOL pattern was very weak, it was not zero, as in the case of waterstriders. Due to the high er spatial


resolution of their compound eyes (Schwind 1978,1980, 1983b), backswimmers could perceive the low brightness contrast at the borders of the polarizing filters. In order to obtain the real gains of the optomotor response to the vhPOL and ±45POL patterns, the gain of this "border response" to the vvPOL pattern must be subtracted from them. After this subtraction we obtain that Notonecta did not respond to the ±45POL pattern, while its response to the vhPOL pattern was weak, but definitely significantly above zero. Backswimmers followed the lateral B&W stimulus pattern with shorter delay than the vhPOL pattern (Fig. 27.9B).

The responses to overwater lateral stimulations were always slightly weaker than those for the corresponding underwater stimulations. Due to totalreflection of light at the water surface, backswimmers could also see the mirror image of the underwater stimulus pattern at 0-30° above their horizon (Fig. 27.8). Thus, the underwater stimulus patterns were doubled in their vertical angular extension. On the other hand, because of refraction of light at the water surface the apparent vertical angular extension of the overwater stimulus patterns was compressed to about 8°. These two effects explain why backswimmers had a weaker optomotor response to overwater patterns than to underwater ones.

According to Schwind (l985b), in the ommatidia of Notonecta glauca the six peripheral rhabdomeres RI-R6 are open and grouped around the two central fused rhabdomeres R7 and R8 (Fig. 18.4, 18.5). In the medial eye region, with optical axes from 60° ventrally to 80° dorsally from the horizontal direction, the microvilli of the central rhabdomeres are horizontal. In the

0·-'---- U

B&W ovcrwalcr(OW) ;-~-:~

undcrW3ICr UW) IIJ c_~_:

Fig.27.9. The open loop gain go (A) and open loop phase difference !J<po (B) of the optomotor response of Notonecta glauca to the B&W, vhPOL and vvPOL patterns stimulating the lateral eye region below (UW) and above (OW) the water surface. Columns average data obtained with eight animals. Bars, standard deviations.


ventral eye region, 70° or more ventrally from the horizontal, the microvilli of receptors R7 and R8 are horizontally respectively vertically aligned. We call this part of the eye the "ventral PO L-area". Both in the dorsal and ventral eye regions, receptors Rl-R6 are green sensitive. In the dorsal eye region, cells R7 and R8 are either UV or blue sensitive, while in the ventral POL-area they are exclusively UV sensitive. In the normal resting position the visual field of the animal is subdivided into three different regions (Fig. 18.5A):

- the ventral POL-area looks into the air through the Snell window; - the eye region directly below the ventral POL-area looks at the region of

total reflection on the water-air interface; - the remaining part of the eye looks at the underwater world.

In the optomotor experiments the light sources did not emit UV light, and the polarizing filters did not trans mit UV light. Therefore, in the ventral POLarea only the peripheral green receptors RI-R6 were stimulated, while in the lateral and dorsal eye regions also the central blue receptors. Thus, in the optomotor response of Notonecta to overwater stimuli the green receptors of the ventral part of the eye must have been involved. Since it is very unlikely that in the other eye regions the motion perception would be mediated by another receptor type than in the ventral POL-area, we conclude that in backswimmers only the green-sensitive visual pathway is responsible for motion detection, which is distinct from the UV-sensitive pathway involved in polarotactic water detection. In honeybees, blowflies, droneflies and goldfish (Kaiser 1974; Tinbergen and Abeln 1983; Srinivasan and Guy 1990; Schaerer and Neumeyer 1994), for example, the visual subsystem performing directionallyselective movement detection responds also only in the green.

Since the microvilli of the central receptors R7 and R8 in the ventral POLarea of Notonecta are always vertical or horizontal (Fig. 18.4, 18.5), the contrast of the vhPOL pattern is CvhPOLblue = (PS-1)/(PS+ 1) ::::: 75 %. In Notonecta PS-values are unknown, therefore PS = 7 measured in waterstriders (Bartsch 1991, 1995) was assumed for the blue receptors of Notonecta. Because the B&W pattern provided a contrast of about CB&W = 90 %, the ratio of the contrast of the vhPOL and B&W patterns is CvhPOLblue/CB&W = 0.83.

On the other hand, the effective contrast of the vhPOL pattern perceived by the green-sensitive visual pathway is reduced, because the microvilli orientations of the peripheral green receptors are slightly disordered (Fig. 18.4). The perceived contrast of the vhPOL pattern is Cvhpolreen(ß) = !cos2ßI(PS-1)1(PS+ 1) for a green receptor, where ß is the angle of the microvilli axis measured from the vertical. In the ventral POL-area of the Notonecta eye the me an values of ß for the peripheral green receptors RI-R6 are (Fig.18.4) ßl = 56°, ß2 = 15°, ß3 = 71°, ß4 = 68°, ßs = 20°, ß6 = 30°, and the relative cross sections of the corresponding rhabdomeres are Al = 0.17,A2 = 0.83,A3 = 0.83,A4 = I,As = I,A6 = 0.17 (Schwind 1985b). Supposing that the contribution of the green receptors to the


effective contrast is proportional to their cross section and to the perceived contrast, the average effective contrast of the vhPOL pattern is C\hPoLgreen = J:6n=l An CvhPolreen(ßn)lJ:6n=l An = 56 %. From he re the ratio of the contrasts of the vhPOL and B&W patterns perceived by the green receptors is C\hPoLgreenICB&W = 0.62. The contrast of the vhPOL pattern perceived by both the blue and green receptors is, thus, lower than that of the B&W pattern. This is in accordance with our findings that backswimmers responded stronger to the B&W stimulation. In the case of the ±45POL stimulation, the green-sensitive subsystem perceived a considerably reduced contrast, and this is the reason why Notonecta displayed practically no optomotor reaction in this situation.

The role of the UV-sensitive ventral POL-area in water detection is weIl understood (Schwind 1985b). In the opinion of Schwind (I983b), the special pattern of the central microvilli in the ventral part of the eye might not be related to the ability of the animal to orient itself by polarization below the water surface. On the other hand, the above results suggest the functional significance of polarization sensitivity of the lateral eye region in the optomotor re action to polarization patterns in the green range of the spectrum. We propose that the function of this polarization sensitivity may be a slight contrast enhancement for motion perception in the course of compensation for passive drift and rotation of the body. We admit that the relatively disoriented microvilli structure of the peripheral green receptors (Fig. 18.4) is not ideal for this task, because the effective polarization sensitivity of this subsystem is not as good as it could be. The consequence is that the effective contrast of polarization patterns perceived by the green-sensitive visual pathway is generally reduced in comparison with the contrast perceived by the UV- or bluesensitive central receptors. In spite of this, the motion perception is mediated by the green receptors, perhaps because in the aquatic habitat of backswimmers brightness and polarization contrasts occur mainly in the visible, especially in the green part of the spectrum, e.g.light reflected from foliage on the shore, water plants and phytoplankton in water.

28 Polarization Sensitivity in Fish

Polarization sensitivity (PS) has been studied in several groups of teleost fish with different methods, ranging from orientation experiments to single-unit recordings from the optic tectum. Kawamura et al. (1981), for instance, found that by changing the direction of polarization, the heart -beat rate decreases in several fish species. This effect is called "bradycardia". The significance of this physiological change is not clear, but the response has been proven useful to investigate PS in fish.

Although PS has been proposed in certain species of severallargely unrelated fish groups, and polarization insensitivity has been suggested only in a few species, PS may not be a general feature of the visual system of teleost fish, since the majority ofbehavioural studies on fish PS lack the adequate controls to eliminate light intensity as a possible confounding cue (see Chapt. 33). On the other hand, at present there is no convincing proof of a polarization-sensitive detector system in any vertebrate except anchovy fish. In addition, most electrophysiological studies suffer from artefacts or the use of techniques in wh ich the response to polarization is of the order of the error range of the technique (see e.g. the criticism given by Novales Flamarique 2001). Without a proven polarization detection system (receptor), vertebrate PS (except anchovies) remains a problematic facet of animal vision. For these reasons, several works on fish PS cannot be trusted and should be repeated.

Some fish species have been shown to orient by means of the solar direction, light polarization and/or electric as well as magnetic fields in experimental arenas. However, it is not clear yet how important these information sources are. Some species are capable of a time-compensated sun compass orientation (e.g. Goodyear and Bennett 1979). This ability implies that from their underwater vantage point fish are able to determine the solar azimuth and possibly also the sun's altitude. Close to the surface and in a flat calm these fish could directly observe the solar disk and/or the sky polarization. Otherwise, the solar azimuth could be derived either from the radiance distribution in water or from the underwater polarization pattern, both of which directly depend on the solar position (see Chap. 14).

Polarization sensitivity in fish could guide the periodic movements perpendicular to the shore-line, usually termed y-axis orientation, or zonal ori-


entation, or on/offshore reaction. Adult largemouth bass Micropterus salmoides (Loyacano et al. 1977) and immature bluegill sunfish (Goodyear and Bennett 1979), for example, use the sun compass to orient their movements perpendicularly to the shoreline, while Stepanov et al. (1979) hypothesized that certain salmon species may rely on the sun compass during the homeward migration.

Denton and Nicol (1965) suggested that polarized light reflected from the scales of a fish may enable other fish to detect it against diffusing backgrounds reflecting less polarized light. In fish scales there are plenty of tiny platelets composed of a stack of thin, flat guanine crystals separated by cytoplasm. Since the optical properties of such platelets are dominated by constructive and destructive interference of light rays reflected at the different boundaries between the layers within the stack, the platelets function as biological mirrors. Near the Brewster angle of 53° from the normal vector of the platelets for a guanine-cytoplasm stack the light reflected by the platelets is highly polarized with E-vector perpendicular to the plane of reflection. At angles of incidence over about 70°, the platelets become highly reflecting and non-polarizing at all wavelengths.According to Rowe and Denton (1997), such reflecting and polarizing structures in the skin of certain fish may be useful, especially to schooling fish, to mediate information on relative positions, orientations and movements of neighbours.

28.1 Fish in Which Polarization-Sensitivity Was Proposed

We would like to emphasize that the only case in which fish PS can be trusted is that of the anchovies. Thus, it would be misleading to say that PS is common in fish.

28.1.1 Sockeye Salmon Oncorhynchus nerka

In a pioneering laboratory experiment, Groot (1965) found that young sockeye salm on Oncorhynchus nerka changed their swimming orientation with respect to the E-vector direction of totally linearly polarized light passing through a polarizer covering their test tank, from which they could always see the sky, but they oriented only at dusk when they had an activity peak near the water surface. However, it was not excluded that the salmon oriented to the polarization-induced brightness patterns. Sockeye salmon smolt completely before reaching the ocean, where they reside for 1-4 years before migrating back to the natal river to spawn and die. The migration of newly emerged parr (i.e. young animals with vertical "parr" marks along their bodies) and smolts (a stage characterized by the loss of "parr" marks and silvering of the body)

28 Polarization Sensitivity in Fish 295

are rapid and both celestial and magnetic cues are apparently utilized (e.g. Groot 1965; Dill 1971). Novales Flamarique and Hawryshyn (1996) found by means of electrophysiological recordings that sockeye salmon smolts are maximally sensitive to horizontally polarized UV (380 nm) light in front of a white crepuscular background.

28.1.2 Tropical Haltbeaks Zenarchopterus dispar and Zenarchopterus buffoni

Waterman and Forward (1970) monitored the spontaneous orientation ofthe body long axis of the surface-living hemirhamphid tropical haltbeaks Zenarchopterus dispar and Zenarchopterus buffoni. Single fish were placed in a cylindrical transparent plastic vessel screened laterally and downward by white screens. Natural submarine illumination, including the sun and sky within the Snell window, was visible to the animals in the upper hemisphere of their visual field. They oriented unimodally at azimuth angle fJ = 120° from the solar meridian. When a linearly polarizing filter with different angles ß of its transmission axis with respect to the solar meridian was placed over the vessel, the preferred direction of orientation changed. For ß = 0,45 and 90° the fish oriented unimodally at fJ = 167,87 and 139°,respectively.Although Waterman and Forward (1970) concluded that Zenarchopterus dispar and Zenarchopterus buffoni are sensitive to polarization, the relative importance of the sun, sky polarization, underwater polarized light field or other possible factors in this orientation as well as its biological sense, if any, could not have been assessed.

In a second study with a similar experimental setup as described above, Waterman and Forward (1972) found that juvenile Zenarchopterus dispar spontaneously oriented their body long axis mainly perpendicularly to an overhead E-vector. This response was most intense when the imposed E-vector differed maximally (by 60-90°) from that predominant in the natural illumination. When the sun was obscured by clouds, the orientation was significantly weakened even though the polarization pattern generated by the polarizer remained unchanged. Individual fish reacted differently. Certain individuals oriented unimodally and strongly parallel and perpendicularly to the imposed E-vector, some fish oriented bimodally and obliquely, while another responded weakly to the E-vector of polarized light. The biological significance, if any, of this polarization-sensitive behaviour of Zenarchopterus dispar is unknown like the relative role of the different celestial and environmental optical cues.

In a third field experiment, Forward et al. (1972) studied the spontaneous heading preference of juvenile Zenarchopterus dispar swimming near the water surface under natural illumination as well as under six different E-vector directions of imposed linearly polarized light. Single fish were placed in a


cylindrical transparent plastic container screened laterally and downward by white screens. The fish had a free view of the sky through the water surface, but were prevented from seeing surrounding landmarks. A linearly polarizing filter with different directions (0,30,60,90,120,150°) of the transmission axis relative to the solar meridian was placed over the vessel in the afternoon and morning. In this experiment the animals tended to orient unimodally and exclusively parallel to the overhead E-vector, a response which was quite different from that observed in the above-mentioned two submarine experiments. Since Zenarchopterus did not or only weakly oriented when the sun was obscured by clouds, it is doubtful whether PS is essential for orientation by means of the celestial polarization pattern. The biological function, if any, of this re action is unknown like the reasons for the significant differences between the responses of an im als to the artificial E-vector in the above-mentioned three experiments.

These pioneering, but methodologically erroneous experiments could not prove the PS of Zenarchopterus convincingly, since the overhead polarizer drastically changed the distribution of both intensity and colour of polarized light from the sky and water. On the other hand, the cloudiness and the water clarity changed considerably from day to day during the second experiment in relation to tidal phase and meteorological conditions. Thus, it is not clear whether the change of orientation of the individual fish after the placement of the polarizer over the experimental vessel was induced by the perceived change of the overhead intensity and/or colour pattern, or by the change of illumination condition, or by the perception of the E-vector of polarized light. Furthermore, the possibility that the animals responded to unwanted intensity patterns induced by selective reflection of polarized light from the walls of the experimental vessel rather than to the E-vector of polarized light, has not been excluded (see Chap. 34).

28.1.3 Halfbeak Fish Dermogenys pusilus

In a laboratory experiment, Forward and Waterman (1973) monitored the spontaneous body orientation of adult haltbeak fish Dermogenys pusilus in a vertical beam of unpolarized and totally linearly polarized white light. A depolarizer above the linear polarizer guaranteed that unpolarized light with constant intensity entered the polarizer. To test whether the animals responded to unwanted intensity patterns produced by selective reflection of polarized light from the walls of the test arena, the responses of the fish were recorded with either a vertical uniform white screen surrounding the experimental vessel or with a screen divided into black and white alternating quadrants. The fish oriented randomly under unpolarized light with the white surroundings. Under polarized light with the white surroundings, the animals oriented parallel to the E-vector. With the black and white quadrants com-


bined with unpolarized overhead light, preferential orientation was observed toward the white sectors (positive phototaxis).

Since maximum reflection of linearly polarized light occurs at the vertieal walls, the plane of whieh is parallel to the E-vector direction, the positive phototactic response of Dermogenys excluded that under polarized light the orientation of the animal toward walls with perpendieular plane to the E-vector direction and minimum reflection was governed by unwanted intensity patterns (see Fig. 34.1). This was the first behavioural experiment in whieh it was convincingly shown that the primary orientation of a fish to the E-vector is orthogonal to that expected if the animal were reacting to intensity patterns of differentially reflected light. Tests with polarized light combined with the black and white surroundings indieated that phototaxis in Dermogenys predominated over polarotaxis. Forward and Waterman (1973) concluded that Dermogenys pusilus can respond to the E-vector orientation. The biologieal sense, if any, of spontaneously orienting the body long axis parallel to the Evector in Dermogenys is unknown.

28.1.4 Goldfish Carassius auratus

Kleerekoper et al. (1973) monitored the locomotor orientation of the goldfish Carassius auratus in the laboratory under a depolarized and a totally linearly polarized overhead white light field. Polarized light was produced by a linear polarizer below a waxpaper illuminated from above by white light. It was depolarized by placing the waxpaper below the polarizer to maintain the same spectral and intensity characteristies of polarized and depolarized light. Individual fish were placed in a cylindrieal tank with 16 peripheral compartments to which the animals had free access. The tank together with the light source was rotated slowly to eliminate compartmental and geographie bias in the orientation. The orientation of Carassius auratus was random under depolarized light. Under polarized light the fish oriented bimodally and parallel to the E-vector. When the E-vector was rotated abruptly by 90°, the animals continued to orient parallel to it. Kleerekoper et al. (1973) concluded that Carassius auratus can respond to the E-vector orientation. The biologieal relevance, if any, of maintaining the swimming orientation parallel to the E-vector is unclear, since according to Kleerekoper et al. (1973, p. 35), "this species does not appear to need orientation cues based on polarized light patterns". The possibility that the animals responded to unwanted intensity patterns induced by selective reflection of polarized light from the acrylic walls of the test arena was not excluded by control experiments.

Harosi and MacNiehol (1974) found that the outer segments of side-illuminated red- (625 nm), green- (530 nm) and blue- (455 nm) sensitive cones in Carassius auratus are dichroic with a diehroie ratio of 2-3. With intracellular


recordings of bipolar, ganglion, amacrine and horizontal cells in the retina of Carassius auratus, Waterman and Hashimoto (1974) found no evidence for PS in isolated retinal fragments stimulated by linearly polarized flash es with different E-vector directions. With a polarization-microscope dichroism could not be demonstrated in the cornea, lens or other dioptric elements of the goldfish eye. On the other hand, almost all single units examined with extracellular recording in the optic tectum were sensitive to the E-vector direction when the retina was stimulated by totally linearly polarized white, red (620 nm), green (540 nm) or blue (460 nm) flashes. Since the preferred E-vector directions (0,30,60,90,120,1500 relative to the body long axis) eliciting a maximal response varied from unit to unit, an analyzer without discrete channels for particular E-vector directions may be present quite differently from that in many rhabdom-bearing eyes. The E-vector direction also affected colour coding in the goldfish, since all red, green and blue cones appeared to be involved in the polarization sensitivity, the PS-value of which ranged from 1.4 to 31.7 with an average of 8.2.

With the same electrophysiological technique as described by Waterman and Hashimoto (1974), Waterman and Aoki (1974) found that both the direction of strongest response and the degree of polarization sensitivity of the tectal units in Carassius auratus form a regular and distinctive pattern across the tectal area, which corresponds to the visual field map overlying the tectum. These patterns are strongly affected by changes in direction of the stimulus beam relative to the optical axis of the eye.

Hawryshyn and McFarland (1987) showed that the UV (380 nm) cones in adult goldfish are most sensitive to vertically polarized light, the green (540 nm) and red (660 nm) cones are most sensitive to horizontally polarized light, while the blue (460 nm) cones are insensitive to polarization. The UV, green and red cones have an average PS = 4.68, which is roughly two times higher than the dichroic ratios of 2-3 for transversely illuminated vertebrate photoreceptors (Harosi and MacNichoI1974). This suggests that the longitudinal axes of polarization-sensitive cones in the goldfish are either obliquely oriented with respect to the centre of the pupil, or these receptors are receiving reflected or scattered light along the transverse axis of the receptor (Waterman 1975). This,however,does not explain whyPS is so high. There are also other reports on PS exceeding the dichroic ratio of receptors (e.g. Shaw 1969; Labhart 1980). Fein and Szuts (1982) suggested that synaptic inhibition between classes of photoreceptors could increase PS in a fashion comparable to that of neural mechanisms responsible for contrast enhancement (see Sect. 16.7.3). Furthermore, reflection from the tapetum lucidum, a multilayer interference mirror underlying the retina (Fineran and Nicol1978) could result in a double pass of the polarized light through the photoreceptor outer segments, thereby augmenting absorption.

Hawryshyn and McFarland (1987) used a heart-beat rate conditioning technique to measure increment thresholds of immobilized goldfish for lin-


early polarized, narrow-band (10 nm half bandwidth) spectral stimuli as a function of the direction of polarization. The activity of a given spectral type of cone photoreceptors was isolated 1 by chromatie adaptation suppressing the sensitivity of other spectral types of cones.

Hence, for detection of polarization goldfish have more than one spectral type of cones, the absorption spectra of whieh overlap in the UV. This suggests that goldfish may be able to discriminate between different E-vector directions ofUV light. The UV PS may be important also for the enhancement of brightness and colour contrasts (Lythgoe and Hemmings 1967). On the other hand, according to Mare and Sperling (1976) the polarization-blind blue cones could playa role in monitoring the veiling radi an ce or ambient illumination, since it has the lowest degree of linear polarization at 450 nm (Ivanoff and Waterman 1958b). In dear freshwater lakes, the underwater downwelling light field is dominated by blue light gradually increasing with depth. The overall sensitivity to the background would thus be adjusted through the blue cones permitting the Uv, green and red cones to operate at a high er signal-tonoise ratio. However, the problem with this interpretation is that cyprinids (e.g. goldfish) live in mesotrophie to eutrophie lakes, in whieh the ambient intensity peaks in the green (520-540 nm) with depth. According to Hawryshyn and McFarland (1987), PS in goldfish cones may have two functions: (1) enhancement of brightness differences of targets, (2) discrimination of UV stimuli on the basis of E-vector direction.

28.1.5 African Cichlid Pseudotropheus macrophthalmus

In a behaviourallaboratory experiment, Davitz and McKaye (1978) tested whether the African cichlid fish Pseudotropheus macrophthalmus can perceive the direction of polarization. A group of 19 fish was kept in a tank, the sides of whieh were covered by white cardboard, and its top with a commercial lighting hood. Two opposite walls of the tank were illuminated by linearly polarized white light with the same degree of polarization of 89 % and the same intensity, but with different, horizontal and vertical E-vectors. The fish were trained with operant conditioning to discriminate between

I A possible way of isolation of a given spectral type of receptors to eliminate the influence of other spectral types of receptors is chromatic adaptation. This technique is based on the principle that the sensitivity of a photoreceptor is approximately inversely proportional to the background radiance. A background illumination, with a spectrum in which a narrow spectral band at wavelength Ab is lacking or suppressed, enhances selectively the sensitivity of photoreceptors which are maximally sensitive to wavelength Ab' while reducing the sensitivity of all other spectral types of receptors. For example, a blue-green-yellow background containing blue, green, yellow, orange and red wavelengths reduces selectively the sensitivity of the bIue, green and red receptors, thereby isolating spectrally the UV receptors.


horizontal and vertical E-vectors. Any fish swimming towards the horizontally polarizing tank wall was rewarded by the release of a small quantity of food suspension. Swimming towards the vertically polarizing tank wall was not rewarded.

In the first 6 days after training the fish did not discriminate between horizontally and vertically polarized light. However, during the period between the 13th and 20th days following training they oriented to the horizontal Evector.Although Davitz and McKaye (1978) admitted that in their experiment unwanted intensity patterns might have occurred due to selective reflection of polarized light from the tank walls, they concluded that Pseudotropheus macrophthalmus could discriminate between horizontal and vertical E-vectors.

Cichlids possess some type of sun compass and migrate vertically as well as horizontally during their reproductive and feeding excursions (Fryer and Iles 1972). Davitz and McKaye (1978) hypothesized that Pseudotropheus macrophthalmus could use polarization information during migration. However, this was not tested in behavioural experiments.

28.1.6 Anchovies Engraulis mordax and Anchoa mitchilli

By me ans of compound action potential recordings from the optic nerve, Novales Flamarique and Hawryshyn (1998b) found that under photopic conditions the retina of adult northern anchovies Engraulis mordax possesses a spectral sensitivity curve with a single peak at 500 nm and maximum sensitivity to horizontally polarized light. Sensitivity to UV light was not found. The central region of the retina has long and bilobed cones. Novales Flamarique and Hawryshyn suggested that during clear-sky crepuscular periods anchovies may use their PS to detect planktonic prey.

Using microspectrophotometry, Novales Flamarique and Harosi (2002) found that the rods and cones in adult bay anchovies Anchoa mitchilli have dichroic absorption of light in accordance with the lamellar orientation of the photoreceptors, and that both the bilobed and long cones contain the same photopigment with peak absorbance at 540 nm, while the rods contain a photopigment with absorption maximum at 501 nm. The monochromacy of the cones in Anchoa mitchilli ensures the elimination of perception of polarization-induced false colours (see Chap. 33), as perhaps in the colour-blind cephalopods.

The retinal cone system in anchovies is unique in the animal king dom and the only truly demonstrated polarization detection system in vertebrates. The mechanism of anchovy PS is described in detail in Chap. 28.3.1.


28.1.7 Rainbow Trout Oncorhyncus mykiss

The rainbow trout Oncorhyncus mykiss (formerly Salmo gairdneri) has been known to have PS since the work of Kawamura et al. (1981). Hawryshyn et al. (1990) demonstrated that immature rainbow trout with a body weight of 30-40 gare capable of orienting to the E-vector of an overhead broad-band (white+UV) polarized light field. Rainbow trout were trained to swim to a re fuge located at one end of an elongated training tank under an overhead polarized light field. The E-vector of totally linearly polarized light was oriented parallel or perpendicularly to the long axis of the training tank. Trained fish were then released in a circular test tank and their angular responses recorded. They responded correspondingly to the training. When the E-vector in the test tank was rotated by 90°, the preferred orientation of trout followed the new E-vector direction. Under an unpolarized light field trout oriented randomly.

When a patch of diffusing screen was placed on the skin over a large area including and surrounding the pineal gland, the trout oriented also to the direction predicted from the trained E-vector alignments. Thus, the orientation of rainbow trout to polarization is mediated by retinal and not pineal photoreceptors. This is consistent with the result of Kawamura et al. (1981), who showed that heart-rate responses of rainbow trout to linear polarization were not diminished when the pineal organ was covered. Under a white polarized light field lacking UV, rainbow trout could not orient by means of the Evector. This indicates that UV polarized light is necessary to discriminate the E-vector. Hawryshyn et al. (1990) demonstrated that rainbow trout weighing 50-60 gare incapable of orientation by means of the E-vector direction of an overhead polarized light field. This ontogenetic loss in PS is not due to an ontogenetic loss of a general cognitive ability for spatial orientation. Rainbow trout remain in fresh water and undergo a transformation somewhat similar to smoltification. In the life stage leading up to and during sexual maturity, rainbow trout may conceivably possess the capability of UV polarization sensitivity2.

In similar experiments as performed by Hawryshyn et al. (1990), Hawryshyn and Bolger (1990) showed that the lower the degree oflinear polarization p of a downwelling UV +white light field, the more inaccurate the polarotactic orientation of immature ( < 50 g) rainbow trout. Below p = 65-75 % rainbow trout cannot detect the E-vector direction. Hawryshyn and Bolger

2 UV wavelengths may be used by fish to detect food which either absorbs, scatters or reflects strongly in the UV relative to the background. In fish, UV receptors tend to be found in juveniles, which live dose to the surface and feed on small planktonic organisms. Such planktonic food particles will preferentially scatter short wavelengths, and UV perception has been shown to playa role in plankton predation (e.g. Browman et al. 1994).

302 Part IIl: Polarized Light in Animal Vision

(1990) decreased p by increasing the concentration of a solution of polystyrene monodispersed latex beads suspended in water in a UV -transmissive Plexiglas cuvette positioned below the UV -transmissive linear polarizer.

Certain Pacific and AtIantic salmonid fish migrate during their life. Anadromous salmonids hatch and forage in the freshwater environment until they smoltify. Then they migrate out to sea. During or prior to smoltification, there is a change in the photoreceptor mosaic causing a local decrease of UV cones in some species (e.g. Novales Flamarique 2001). Then, they spend a fewyears in the open ocean where they feed and grow, and finaHy return to their natal streams. The distance traveHed varies from species to species and stock to stock, but it may be thousands of kilometres. There have been a variety of indications to explain homing in fish, including sun compass orientation (Schwassmann and Hasler 1964), olfactory imprinting (Hasler et al. 1978) as weH as navigation by means of celestial and magnetic cues (Dill 1971). However, it remains to be demonstrated how the actual sensory systems of a fish receive, encode and process the relevant cues for spatial guidance. As long as rainbow trout do not lose their abilityto orient polarotactically, their sensitivityto polarization ofUV light may playa role in the sun compass navigation as one component of a probably multimodal navigational system during their migration.

The finding of Hawryshyn and Bolger (1990) that rainbow trout are capable of orienting to the overhead E-vector of linearly polarized light with p > 65-75 % suggests that such polarotactic orientations need not be restricted to dawn or twilight hours, when p of skylight is maximal. The zenith blue skyat dawn or dusk may reach p = 70-80 % (Coulson 1988). Hawryshyn and Bolger (1990) found that the proportion of trout that can orient to polarization is proportional to p, and even at as low p as 65 %, there were individuals that were capable to orient polarotacticaHy.

Measuring multi-unit ganglion ceH responses from axons in the optic nerve of the rainbow trout, Parkyn and Hawryshyn (1993) found that under scotopic conditions trout do not respond to the E-vector of polarized light, because their rods do not mediate PS. The optic nerve response is the summed response of a large number of ganglion cells as observed in the optic nerve by means of a relatively large electrode. Under photopic conditions, trout weighing 8-10 gare polarization sensitive in two orthogonal channels. A UV (380 nm) stimulus on a white background evokes maximal responses to vertical and horizontal E-vectors, and minimal responses at E-vector directions 30 and 1500 from the vertical. PS = 3.31 under photopic conditions. The green and red cones have maximal sensitivity to the horizontal E-vector independentIy of whether their a-absorption band in the visible spectral range is stimulated or their ß-absorption bands in the near UV. The UV cones are maximaHy sensitive to the vertical E-vector as a result of a-band absorbance of UV light. The blue cones are insensitive to polarization. Hence, the rainbow trout has two channels for the detection of polarization of UV light. Parkyn and Hawryshyn (1993) proposed that a major role of the UV cones may be the


perception of E-vector, because UV light may poorly contribute to image formation. The reasons for the latter are manifold:

1. The wavelength dependency of the refractive indices of the cornea and lens increases logarithmically with decreasing wavelength (Lythgoe 1979) resulting in strong chromatic aberration of the lens in the short-wavelength range (Sivak and Mandelman 1982).

2. UV light is strongly scattered in the water (Jerlov 1976). 3. The UV cone mechanism has a relatively poor brightness contrast sensitiv

ity (Hawryshyn 1991a). 4. The large receptive-field diameter of UV cones results in decreased spatial

resolution and lower acuity in the UV (Hawryshyn 1991b).

Coughlin and Hawryshyn (1995) examined the PS of chromaticallyisolated single units in the torus semicircularis, a subtectal visual area in the midbrain with high UV sensitivity (Coughlin and Hawryshyn 1994), with inputs from each of the four (UV, blue, green and red) cone types in juvenile rainbow trout. Coughlin and Hawryshyn (1994) found two main types of colour-coding units in the torus semicircularis of juvenile rainbow trout showing phasic response properties to visual stimuli: biphasic units with OFF inputs from the red and green cones, and with ON inputs from the UV and blue cones and triphasic units with ON inputs from the UV, red and blue cones, and with OFF input from the green cone. UV ON-response units are maximally sensitive to vertical E-vector (PSUlN = 3.13±0.77), while ON-response units of the blue, green and red cones are polarization insensitive (Coughlin and Hawryshyn 1995). OFF-response units of the green (PSGOFF = 3.59±0.95) and red (PSROFF = 3.34±0.45) cones are maximally sensitive to the horizontal E-vector. E-vector sensitivity was observed in colour-coding units, which receive inputs from more than one cone type and have ON or OFF responses at different wavelengths. Biphasic units with ON input from UV and blue cones and OFF inputs from green and red cones are polarization opponent.

Coughlin and Hawryshyn (1995) believe that the interaction of orthogonally polarization-sensitive receptors within biphasic colour-coding units in the torus semicircularis may provide a possible basis for PS. The hypothesis that the torus semicircularis of juvenile rainbow trout may be the primary location for PS is consistent with its other known roles: it integrates inputs from several sensory systems, including visual, auditory, lateralline and, in some fish, electroreceptive organs (Schellart 1990). The torus semicircularis is thought to contribute to object detection. The PS may contribute to prey, predator and/or conspecific location or to navigation, or could simply increase the contrast of visual cues.

Although PS in trout is believed to be driven by a comparative interaction of two orthogonally opposed polarization-sensitive receptors, the dominance of an ON-response to the vertical E-vector and OFF-response to the horizon-


tal E-vector in the ventral retina indicates that vertical polarization may play an important role. In the natural underwater optical environment, such vertically polarized light could originate due to reflection of light from the sides of fish, for example.

The orthogonally polarization-sensitive, two-channel system has neutral points or confusion states in which a fish would be unable to determine the Evector direction (Bernard and Wehner 1977). E-vector discrimination with chromatically different detectors is not optimal due to polarization-induced false colours (see Chap. 33). While the chromatic and polarization channels may be separate, within the polarization units there would be inherent conflict between responses to E-vector and intensity at different wavelengths. The visual system in certain insects, for example, avoids this conflict by a specific monochromatic group of polarization-sensitive receptors.

By means of operant conditioning in a behaviourallaboratory experiment, Degner and Hawryshyn (2001) trained juvenile rainbow trout to swim in an elongated tank under two identical neighbouring light sources emitting totally linearly polarized light with the same E-vector directions, which were either parallel or perpendicular to the long axis of the training tank. Then the fish were tested in a circular tank in seven different situations, in which the two E-vector directions of the overhead light patches differed by 0,15,30,45, 60,75 and 90°. The animals oriented bimodally parallel or perpendicularly to the E-vector as trained, when the directions of the two E-vectors differed by 0, 15,30,45 and 90°, suggesting that the animals perceived the two stimuli as being the same. They did not orient along the trained axis when the two Evector directions differed by 60 and 75°. From this Degner and Hawryshyn (2001) conc1uded that Oncorhynchus mykiss may be able to discriminate two E-vector directions which differ at least by 60-75°.

This low level of E-vector discrimination capability is inappropriate for tasks requiring contrast enhancement of spatially distinct targets. Rather, it seems to be more appropriate for orientation. When the animals were exposed to an overhead depolarized light field, they oriented randomly. To exclude orientation by the geomagnetic field, a 2.5-cm-thick metal plate was placed beneath the testing tank. However, all experiments were conducted in a room with walls and ceiling painted flat black. Furthermore, a black feIt curtain was placed around the optical system, the Plexiglas training tank, and the testing tank was painted flat grey inside. The use of such highly polarizing black surfaces was the worst possible choice. Degner and Hawryshyn (2001) did not exclude by control experiments that the rainbow trout responded to intensity patterns induced by selective reflection of polarized light from the black surrounding surfaces. Unfortunately, all the behavioural orientation experiments of Hawryshyn and collaborators did not have appropriate controls to account for variations in intensity due to scattering and/or reflection. Therefore, rainbow trout could have been orienting to intensity patterns in these experiments (see Chap. 34).


Small rainbow traut feed on zoo plankton. The majority of these prey organisms contain lipids and carotenoid pigments which absorb UV and blue light. Thus, zoo plankton become more visible to a UV- and blue-sensitive predator, like the rainbow trout, when the background illumination contains UV and/or blue components, as is the case in nature (Novales Flamarique and Hawryshyn 1997a). Perhaps as a result of this enhanced contrast, several zoo plankton species have evolved cuticle layers, which efficiently reflect UV and blue light (Giguere and Dunbrack 1990), thereby reducing their contrast. Zoo plankton may be further conspicuous to polarization-sensitive predators because of the birefringence of their exoskeletons composed of calcium carbonate. Linearly polarized light passing through their bodies is scattered and retarded depending on the wavelength and path of the light, and the E-vector direction relative to the optical axes of the various birefringent structures. Consequently, in polarized light, zoo plankton may exhibit high er contrast to a polarization-sensitive predator.

Novales Flamarique and Browman (2001) tested the hypothesis that polarized light can improve the prey location ability of free-swimming juvenile rainbow traut in laboratory aquaria. They observed rainbow traut foraging on Daphnia magna under unpolarized and totally linearly polarized UV +visible light oflow intensity as well as under low-intensity short-wavelength (390-4l3 nm) light with degrees of polarization p = 52, 71, 85 and 97 %. They found that prey location distances were significantly Ion ger and significantly more prey attacks were directed downwards than upwards under totally linearly polarized than under unpolarized white light. The latter observation is consistent with the higher density of polarization-sensitive UV cones in the dorsal retina of young rainbow trout. The average frequency distribution of the azimuth angle of prey attack was more bimodal with maxima in the interval 20-80° on either visual field under totally polarized light than under unpolarized illumination. This suggests that the centratempora-dorsal retina with UV- and polarization-sensitive double and single cones may be the primary site for polarization detection in young rainbow trout. Under short-wavelength illumination, prey location distances were significantly longer for p = 97 and 85 % than for p = 52 %. This suggests a threshold of 63 % < Pthreshold < 72 % for PS. These results indicate that PS improves location of prey due to the enhanced contrast of them under polarized illumination.

Several studies (e.g. Hawryshyn et al. 1990; Parkyn and Hawryshyn 1993; Coughlin and Hawryshyn 1994) have indicated that the rainbow trout loses UV sensitivity due to the loss ofUV cones and the associated PS when the animal transforms from a small juvenile to a larger smolt stage. In contrast to these previous investigations, Novales Flamarique (2001) showed that juvenile rainbow trout lack UV cones throughout the lower half of the ventral retina. Because in both electrophysiological and behavioural previous experiments, the stimulating illumination was directed primarily at the ventral


retina the reported age-dependent changes in UV or polarization sensitivities can be explained by differences in the area of UV cones that was illuminated, and not necessarily by a loss of UV cones as weIl as UV and polarization sensitivities. According to Novales Flamarique (2001), the central retina may be a potential site for PS in rainbow trout.

28.1.8 Juvenile Salmonid Fish Oncorhynchus mykiss, Oncorhynchus darki darki, Oncorhynchus nerka and Salvelinus fontinalis

By means of multi-unit recording from the optic nerve, Parkyn and Hawryshyn (2000) compared the UV PS of juvenile rainbow trout and steelhead (both Oncorhynchus mykiss), cutthroat trout (Oncorhynchus clarki clarki), kokanee (Oncorhynchus nerka) and brook char (Salvelinus fontinalis). Rainbow trout and steelhead are the freshwater and ocean-going races of the same fish species. Kokanee is the freshwater variant of the sockeye salmon (Oncorhynchus nerka). The coastal cutthroat trout is a moderately migrating species that makes use of estuaries during part of the year only. It has been hypothesized that the visual sensitivity of these fish might differ because the habitat of anadromous (ocean-going) juvenile and adult salmonids is vastly different from the small coastallakes and streams of their non-anadromous (freshwater-inhabiting) counterparts. All these species were differentially sensitive to both vertical and horizontal E-vector of linearly polarized UV light. The ON-responses were maximal to both vertically and horizontally polarized UV light, whereas the OFF-responses were elicited only by horizontally polarized UV light. Hence, the UV PS appears to be similar among these species of salmonid fish, regardless of whether they are migratory or nonmigratory. This suggests that for these species PS has probably a meaningful biological role other than the possibility of supporting orientation.

This hypothesis is supported by the observation of Parkyn et al. (2003) that juvenile rainbow trout trained to orient parallel to an overhead E-vector in the laboratory oriented parallel to the bearing of the band of maximum degree of linear polarization of the sky.

28.1.9 Damselfishes

Using electroretinogram recording, Hawryshyn et al. (2003) found UV PS with a 60° periodicity of the E-vector direction in the three-spot damselfish (Dascyllus trimaculatus) and the blacktail damselfish (Dascyllus melanurus), furthermore UV PS with a 45° periodicity of the direction of polarization in the blue-green chromis (Chromis viridis). These coral reef fishes may have 3-or 4-channel UV PS, the most complex PS recorded for any vertebrate.

28 Polarization Sensitivity in Fish

28.2 Fish with Debated Polarization Sensitivity and Fish in Which Polarization Insensitivity Was Proposed

28.2.1 Green Sunfish Lepomis cyanellus

307

Cameron and Pugh (1991) investigated the PS of the green sunfish Lepomis cyanellus with a classieal conditioning technique. Three fish learned to suppress their heart rate if totally linearly polarized light with a given E-vector direction and dominant wavelength of 604 nm was presented. Conditioning happened with a mild electrieal shock to negative responses. The heart rate was recorded by electrocardiography in a movement-restrieting opaque black plastie box with a frontal port for the light beam. The polarized stimuli were imaged onto the temporal retina of both eyes from a target directly in front of the fish. Green sunfish were maximally sensitive to medio-laterally and dorsoventrally polarized long-wavelength (.A > 540 nm) light and less sensitive to light with oblique E-vector direction with respect to the body long axis. Cameron and Pugh (1991) suggested that the function of PS in the green sunfish could be contrast enhancement, whieh would help the visual discrimination of underwater objects. However, by control experiments they did not exclude the possibility that their animals responded to intensity patterns induced by selective reflection of polarized light from the walls of the surrounding black plastic box, and/or inherent weak partial polarization of the light. Since the results were published in Nature, there was not enough room to thoroughly and appropriately describe the methods in this experiment, and therefore many important technieal details remained unclear. Consequently, Cameron and Pugh (1991) could not convincingly prove the PS in the green sunfish.

This conclusion is supported by the fact that in contrast to the findings of Cameron and Pugh (1991), the multi-unit optic nerve recordings by Novales Flamarique and Hawryshyn (1997b) did not indieate PS in the post-larval stage ofboth the green sunfish Lepomis cyanellus and the pumpkinseed sunfish Lepomis gibbosus. In both species red (620 nm) twin cones and green (530 nm) single cones were found whieh were insensitive to polarization. The geometrie arrangement of the red twin cones forms a square retinal mosaie in which their long axes are locally aligned along either of a pair of nearly orthogonal axes (Fig. 28.1). The polarization insensitivity of the twin cones in sunfish is supported also by Novales Flamarique et al. (1995), who found by mierospectrophotometry that the transmission of polarized light through these twin cones revealed only minute optieal anisotropies.

Hence, as in the case of the homing pigeon Columba livia (see Chap. 31.3.1), the PS in the green sunfish is also a debated and controversial sensorial capability. Additional studies are needed to decide whether Lepomis cyanellus is indeed sensitive to polarization.

308

B dorsal

f 5;i nglc cone

incidel1l lighl \'cillml


Fig.28.1. A Axial transmission oflinearly polarized light in the red (R) twin cones of the green sunfish Lepomis cyanellus proposed by Novales Flamarique et al. (1998). B Arrangement of the red (R) twin cones and green (G) single cones in the retinal mosaic of Lepornis cyanellus. Double-headed arrows indicate the dominant E-vector direc-ti on. (After Novales Flamarique et al. 1998).

28.2.2 Common White Sucker Catostomus commersoni

As we have seen above, certain species in two families of fish, the Cyprinidae and Salmonidae, appear to have both UV and polarization sensitivity. Thus, one could expect that UV-sensitive fish are simultaneously polarization sensitive. However, by means of electrophysiological recordings, Novales Flamarique and Hawryshyn (1998a) found that the retina of the common white sucker Catostomus commersoni is insensitive to polarization, though it posses ses four different cone types similar to those in certain species of cyprinid and salmonid fish. The four cone types have maximum sensitivity to the ON component ofthe stimulus in the UV (380 nm), blue (460 nm),green (540 nm) and red (640 nm). Under daylight conditions, the ON response is mediated primarily by the blue and red receptors, but during the late evening, the UV and red receptors are the most sensitive. The OFF response under all adapting backgrounds is dominated by the green receptors. However, unlike the cyprinids and salmonids, Catostomus commersoni has a random cone mosaic in the centro-temporal region of the retina: random mixtures of large and small single cones and thinner rods occupying extensive areas between double cones. Novales Flamarique and Hawryshyn suggested therefore that PS in UV-sensitive fish may require an ordered double-cone mosaic in the centrotemporal area of the retina.

28.2.3 Pacific Herring Clupea harengus pallasi

Novales Flamarique and Hawryshyn (1998b) recorded compound action potential from the optic nerve and found that under photopic conditions the


spectral sensitivity curve of the retina of juvenile and adult Pacific herrings Clupea harengus pallasi has a single peak at 520 nm and the animals are insensitive to polarization. UV sensitivity was not found. The central region of the retina has twin cones arranged in rows with the same orientation, and tangentially arranged disks in the outer segment of the photoreceptor. The lack of PS in the Pacific herring suggests that twin cones may not be polarization-sensitive, or that one orientation of polarization-sensitive receptors is insufficient to detect the E-vector direction. The model proposed by Novales Flamarique et al. (1998) based on reflection at the partition separating paired cones may explain the lack ofPS in fish with twin cones. In the central part of the retina of adult Pacific herrings there is only one twin cone orientation and there are no single cones.

28.3 Possible Biophysical Basis of Fish Polarization Sensitivity

Among fish, only in certain anchovy species is the biophysical mechanism of PS of photoreceptors dear (Fig. 28.2). In spite of several studies since the 1960s, the mechanism underlying the detection of polarization in other fish species is not weIl understood, and the role of PS in the vision and behaviour of fish is equally undear. The experimental support of current models of the biophysical basis of PS in fish is limited. The common feature of these models is that all suppose orthogonally opposing E-vector-sensitive cones recalling the two-channel orthogonally polarization-sensitive system proposed by Waterman (1966) for invertebrates. In this section we briefly describe these models. In the putative polarization-sensitive fish, some cones are sensitive to the E-vector and others are not. Thus, it is improbable that preretinal anisotropies could somehow be responsible for detection of polarization, since, in this case, all cones would be expected to have similar responses to polarized light stimuli regardless of the adapting and stimulating conditions.

28.3.1 Axially Oriented Membrane Disks in the Photoreceptor Outer Segments as the Basis for Polarization Sensitivity in Anchovies

The retinae of the bay anchovy Anchoa mitchilli (e.g. Fineran and Nico11978), Black Sea anchovy Engraulis encrasicholus (Zueva 1980) and northern anchovy Engraulis mordax (Novales Flamarique and Hawryshyn 1998b) have two unique photoreceptor types, the so-called bilobed cones and long cones. The long cones are flanked on either side by the shorter cones with bilobed outer segments. The outer segments of the long and bilobed cones strongly overlap (Fig. 28.2) and contain multiple layers of membrane disks usually oriented parallel to the long axis of the photoreceptors, and the planes of the

310

tapetum IIiCidulII

distal platelet stacks

outer segment of long cone

inner egment of long cone

1

outer segment of hort bilobed cone

proximal

inner egment

of short

bilobed co ne


platelet stacks

Epcrp» Epo,

multi layer reflection

\.!!""==.,~I dichroic ab orption

Fig.28.2. Dichroic absorption by the visual pigment embedded in the lipid disks of the outer segments of cones, and multilayer reflection from the platelet stacks and the tapetum lucidum underlying the retina and functioning as ideal, dielectric, multilayer, interference mirror, which likely contributes to the polarization sensitivity in the retina of the anchovies Engraulis mordax and Anchoa mitchilli proposed by Novales Flamarique and Hawryshyn (l998b) and Novales Flamarique and Harosi (2002). In the figure, two cone units are seen composed of long cones and short bilobed cones, the outer segments of which strongly overlap. In the outer segments of both cone types, the membrane disks represented here as lamellae are oriented parallel to the longitudinal axes of the cones, and the disks of the two cone types are oriented perpendicularly to each other. The overlapping regions of these cone units may act as orthogonal dichroic filters to axially incident light. The multilayer interference mirrors in the retina are slightly more reflective for polarized light with E-vectors Eperp perpendicular to the plane of reflection than for parallel E-vectors Epar• The dichroic absorption in the outer segment of short cones favours transmission of the perpendicular component Eperp of incident light. (After Novales Flamarique and Hawryshyn 1998b).


disks of the two co ne types are oriented perpendicularly to each other. This axial disk orientation in the outer segments is distinct from that in all other known vertebrate rods and cones, where the lamellae are stacked transversely with their planes perpendieular to the incident light. Fineran and Nicol (1978) suggested that the two mutually perpendicular sets ofaxially oriented lamellae segregated into bilobed and long cones in anchovies could function as orthogonal diehroie filters to axially incident linearly polarized light resulting in PS. Judging from their orientation in the retina, the long cones and bilobed short cones are expected to be most sensitive to horizontally and vertieally polarized light, respectively.

Another peculiarity in retinal design of anchovies is the close association between the highly reflecting tapetallayer and the long cone outer segments, where two reinforcing mechanisms come into play (Novales Flamarique and Hawryshyn 1998b; Novales Flamarique and Harosi 2002): anisotropie absorption by lamellar diehroism and anisotropie reflection at dielectrie interfaces (Fig. 28.2). The dorso-ventral faces of the long cone outer segments are wedge-shaped with nearly horizontal lamellae facing V-shaped pigment epithelial processes containing flat guanine crystals (Fineran and NieoI1978). These platelets preferentially reflect horizontally polarized light. They and the tapetum lucidum reflect light in the blue-green range of the spectrum where the absorption maximum of the photopigments also lies. The successive plates in the platelet stacks and tapetum lucidum are arranged so that they function as ideal multilayer interference mirrors (e.g. Land 1972). The platelets are tilted 10° relative to the vertieal, approximately parallel to the surface of the outer segments of long cones. Since these multilayer mirrors, as all dielectrie interference mirrors, possess a slightly larger reflectivity for polarized light with perpendieular E-vectors relative to the plane of reflection than for parallel E-vectors, they could function as an analyzer of polarization near the wavelength of the absorption maximum of the photopigment.

Although the lamellae of the cone outer segments in the above-mentioned anchovies are parallel to the long axis of the photoreceptors, Zueva (1980) found some cones with lamellae oriented perpendieularly to the long axis of the receptors in the Black Sea anchovy. Occurrences of such cones have also been reported in the peripheral growth zone of the retina in the larval bay anchovy (Fineran and Nieo11978) and in the northern anchovy (Novales Flamarique and Hawryshyn 1998b).The PS in these species was, however, not investigated.

28.3.2 Embryonic Fissures in Fish Eyes and Their Possible Role in the Detection of Polarization

It has been proposed that polarization detection might occur in the embryonie fis sure, a prominent structure in many fish species (Kunz and Callaghan


1989). In this region of the retina, light may strike photoreceptors from the sides, which might allow the use of the dichroism conferred by the stacking of the disks and the transition moments of the photopigments. We mention also that Snyder (1973) has proposed that teleosts could be able to detect polarization by exploiting the weak component of incident polarized light that is perpendicular to the rod outer segments, but no PS has been found under scotopic conditions for many fish species.

28.3.3 Paired Cones as a Possible Basis for Polarization Sensitivity in Fish

The idea that the mosaic of paired cones may mediate PS in fish was first suggested by Gary D. Bernard (cf. Waterman and Forward 1970). Paired cones are composed of two photoreceptor cells which are closely apposed along the length of their inner segments, sharing a wide and often flat double-membrane partition. If the two photoreceptor cells are identical in morphology and, usually, visual pigment content, the paired cone is termed a "twin cone", while if the two members are morphologically different, it is called a "double cone". The twin cones of the green sunfish have been proposed to be an orthogonal set of polarization-sensitive receptors (Cameron and Pugh 1991). A biophysical mechanism for polarization detection based on hypothetical birefringent waveguide properties of the twin cones has also been suggested by Rowe et al. (1994). Later, optical and electrophysiological studies have not supported PS in green sunfish, or the proposed birefringent waveguide hypothesis for the detection of linear polarization (Novales Flamarique and Hawryshyn 1997b; Novales Flamarique et al. 1995,1998).

28.3.3.1 Orthogonal Double Cones with Graded Index of Refraction as a Possible Basis for Polarization Sensitivity in the Green Sunfish Lepomis cyanellus

Cameron and Pugh (1991) hypothesized that the inner segment of the red -sensitive twin cones with roughly elliptical cross section in the green sunfish Lepornis cyanellus functions as a birefringent, polarization-sensitive dielectric waveguide (Fig. 28.1A), and that these twin cones may be opponent pairs of orthogonally polarization-sensitive photoreceptors (Fig. 28.1B). These twin cones are morphologically identical and contain the same red-sensitive visual pigment in both outer segments. Cameron and Pugh (1991) hypothesized that the twin cone inner segment might be a unitarywaveguide. The almost elliptical tangential cross section of the twin cone inner segment may endow the waveguide with some degree of geometrical birefringence, i.e. to enable it to trap and propagate light with one E-vector direction more efficiently than light with other E-vector directions to the outer segments. The effective dimensions


of the waveguide cross section are probably decreased by the graded index of refraction of the inner segments of the twin cones. Cameron and Pugh (1991) found that the response of the twin cones in the green sunfish to the E-vector direction oflinearly polarized light with wavelengths longer than 540 nm has a 90° periodicity. The twin cones are aligned in an orthogonal array and the planes of maximum behavioural PS correspond to the axes of this array. Furthermore, PS in green sunfish was found by Cameron and Pugh (1991) to be restrieted to wavelengths A > 540 nm, where the twin cones with Amax = 621 nm (and no other photoreceptors) have their maximum sensitivity.

The results of Cameron and Pugh (1991) have been considered to formulate the following model to explain vertebrate PS based on waveguide properties of paired cones in general (Rowe et al. 1994): the index of refraction inside vertebrate photoreceptors is slightly higher than that of the intercellular medium. Due to this difference, photoreceptors behave as miniature fibreoptie cables, guiding light from their inner segments into their outer segments, where phototransduction occurs. In the retina of many fish, the dose apposition of the inner segments of the two cones that combine to form a double cone causes the pair of cone inner segments to guide light as a unitary structure, the transverse seetions of which are roughly elliptieal. The refractive index in the ellipsoid region of the inner segments of the double cones in the green sunfish is high er in the centre than at the periphery. On the basis of a two-dimensional waveguide model of the double-cone inner segment, Rowe et al. (1994) hypothesized that the elliptieal cross section and parabolie refractive-index gradient could result in differential PS on double cones in such a way that light waves linearly polarized parallel to the major axis of the inner segment cross section are more effectively guided than waves polarized parallel to the minor axis. With such geometrie birefringence the double cones could act as miniature, weakly polarizing filters. For a given incident light intensity, the fraction of power guided into the photoreceptor outer segments could be the function of the angle of polarization of incident light. In the retina of the green sunfish the major axes of the transversal cross sections of two adjacent double cones are approximately perpendieular to each other. Rowe et al. (1994) have also hypothesized that two neighbouring double cones with orthogonal major axes could function as a two-channel polarization analyzer serving as the input to a neural network that computes a local polarization difference at each point in the retina.

Rowe et al. (1994) calculated the polarization contrast, as a measure of the differential PS of adjacent orthogonal photoreceptors in the square mosaie of double cones in the retina as a function of wavelength A. Per definition, the polarization contrast is proportional to the relative difference in energy absorbed by two double cones oriented with their shortest axes orthogonal to each other and stimulated by the same linearly polarized light. It was found that the polarization contrast is 1-5 % for 550 nm < A < 750 nm for axially incident polarized light with E-vector direction parallel to the major or minor


axis of the elliptical cross section of the double cones. In this part of the spectrum, the polarization contrast of the graded-index double cone model was approximately five times high er than that of a homogeneous cone model with the same size and average refractive index.

The major problem with this model is that the computed polarization contrast as low as 1-5 % seems to be too slight to explain the PS of fish investigated behaviourally and physiologically. On the other hand, this model was based on data obtained for the green sunfish Lepomis cyanellus, the PS of which is debated and controversial: Lepomis cyanellus was first found to be polarization-sensitive (Cameron and Pugh 1991), but later electrophysiological measurements could not verify PS (Novales Flamarique and Hawryshyn 1997b).

28.3.3.2 Proposed Basis for Polarization Sensitivity in Rainbow Trout due to Internal Reflection from the Membranous Partitions of Double Cones

Certain species of Cyprinidae and Salmonidae possess a variety of row or square arrangements of cones and/or their dividing partitions in the centrotemporal region of the retina, an area where PS has been proven (e.g. Hawryshyn and McFarland 1987; Parkyn and Hawryshyn 1993; Coughlin and Hawryshyn 1995). The unit of the square mosaic is composed of polarizationsensitive double green/red cones (with horizontal preferred E-vector) lining the sides of the square, polarization-sensitive UV cones at the corners (with vertical preferred E-vector), and a polarization-insensitive blue cone in the cent re of the square (Fig. 28.3B). In the mosaic, the middle partitions of the double cones in the central retina still define a square.

Novales Flamarique et al. (1998) and Novales Flamarique and Hawryshyn (l998a) proposed a model to explain the biophysical basis of the PS in rainbow trout (Fig. 28.3). This model was supported by comparative linear birefringence and linear dichroism measurements as weIl as histological studies on the cones of the pumpkinseed sunfish Lepomis gibbosus and rainbow trout Oncorhynchus mykiss. Further support of this hypothesis came from largescale optical models and simple theoretical calculations (Novales Flamarique et al. 1998). The basis of the model is the spatial pattern of cones in the retina and the ultrastructural properties of the double cones, which have a partitioning membrane that separates their two elements. The partitioning membrane of the double cones has a bulge at the distal end of the inner segments where the membrane tilts by 10-25° towards neighbouring UV cones (Fig. 28.3A). The cone of rays incident on this tilted surface is ab out 20° and thus the partitioning membrane could reflect anisotropically polarized light onto the outer segment of the neighbouring UV cones if its index of refraction considerably differs from that of the intracellular medium. The tilted segment of the double cone partitioning membrane is oriented such that axial reflection of polarized light is directed onto adjacent UV cones. When the E-vector


dorsal G B i

accessory R UV

'E 23 Q) Vl

E -0 CJ)

t Q) Vl ..... B

G ® G :::; 0

~ I Q)

I :: Ö I f := .D 50 I 5 <> I '" ..... I E 0) I CJ) :: :: I c: 'E

I

:~ <U :=

UV R I 5iJ I t:: 0)

~ I tIl

l G '" ..... I 0.. <U

:: ~

I G R ::

E I ventra l I x

2 0.. inc i dent l i g h t

Fig.28.3. A Axial transmission of linearly polarized light in a retinal unit of the rainbow trout Oncorhyncus mykiss composed of a red (R), green (G) and ultraviolet (UV) cone proposed by Novales Flamarique et al. (1998). B Arrangement of the UV, B (blue), G and R cones in the centro-temporal retinal mosaic of Oncorhyncus mykiss. Eh horizontal Evector, Ev vertical E-vector. Double-headed arrows indicate the dominant E-vector direction. (After Novales Flamarique et al. 1998).

of incident polarized light is rotated, light would be anisotropically reflected from the double cone partitioning membrane and the UV cones would be transversely illuminated with an E-vector directed parallel to the axis of the membrane disks. The tight spacing of cones in the retina of rainbow trout would further facilitate this process.

Figure 28.3B shows the highly regular arrangement of receptors in the centro-temporal retina of rainbow trout, from which one can read that axial reflection from tilted partitioning membrane surfaces happens in two directions since double cones are arranged quadrilaterally in the square mosaic. The two directions of reflection are approximately perpendicular to each other. Within a square cone mosaic unit, vertically polarized light would transversely strike the outer segments of two UV cones obliquely positioned in the corners of the unit, and horizontally polarized light would strike the outer segments of the two other UV cones obliquely positioned, but in opposite corners of the unit. While the UV cones receive the transversely reflected light, the outer segments of the double cones receive the residual polarized


light distal to the partitioning membrane reflection. Blue cones positioned centrally in the square mosaic unit do not receive transversely reflected light from the double cone partitioning membranes and, therefore, are insensitive to polarization. Possible signal integration by interneurons within and across mosaic units could amplify PS and conceivably code the E-vector direction.

In the goldfish retina the cone mosaic is much less regular than in rainbow trout. Moreover, the members of goldfish double cones are usually of unequal size, and the partitions between them are not as flat as in rainbow trout. Because in goldfish the UV cones have not been shown to be positioned appropriately relative to the double cone partitions, one cannot propose a polarization detection similar to that in rainbow trout.

In the polarization-blind retina of the common white sucker the different cone types are randomly arranged, which disrupts the cone-specific reflection patterns hypothesized in rainbow trout, and would resuIt in polarization insensitivity (Novales Flamarique and Hawryshyn 1998a). This lack ofPS may resuIt from an approximately equal number of receptors capturing the same amount of light for any particular direction of polarization due to their random distribution in the central retina.

We have already mentioned that green sunfish Lepomis cyanellus have been reported by Cameron and Pugh (1991) to possess PS. However, these data could not be replicated by Novales Flamarique and Hawryshyn (I997b). Figure 28.1 illustrates that the partitioning membrane between members of twin cones in green sunfish do not have any tiIt or protrusion and, therefore, cannot transversely reflect polarized light onto neighbouring cones. Furthermore, Novales Flamarique et al. (1995,1998) could not find birefringent inclusions in the inner segments of these twin cones.

29 Polarization Sensitivity in Amphibians

Adult amphibians perform periodic migrations and homing between the sites of reproduction, nutrition, shelter and hibernation. The presence and the extent of site fidelity vary considerably among different amphibian species, and the homeward orientation seems to be restricted almost entirely to the area of previous migratory experience (Sinsch 1992). Several species of amphibians are able to orient by means of celestial cues, e.g. the toad Bufo fowleri (Ferguson and Landreth 1966), newts (Landreth and Ferguson 1967), the bullfrog Rana catesbeiana (Ferguson et al. 1968), the southern cricket frog Acris gryllus (Taylor and Ferguson 1970) and the tiger salamander Ambystoma tigrinum (Taylor and Adler 1978). The predictable effects of dock-shift on the directional choice of amphibians using the celestial compass prove that they possess a circadian dock for the necessary time compensation of celestial rotation (e.g. Taylor and Ferguson 1970). In nocturnal amphibians, star or lunar compasses have been demonstrated (e.g. Plasa 1979). However, the optical cues guiding the celestial orientation and the underlying perceptual systems have not been identified yet in most cases.

In amphibians the perception of optical cues takes place in both the lateral eyes and the pineal complex (Adler 1976). The lateral eyes of most amphibians possess muscles of accommodation and a retina containing two kinds of rods and two types of cones (Duellman and Trueb 1986). The sensitivity of these eyes is high and can reach that of owls (Hailman 1982). The pineal complex of anurans consists of the intracranial pineal body (epiphysis cerebri) and the frontal organ, whereas the latter is lacking in urodeles and caecilians. Both the pineal body and the frontal organ contain photoreceptors, wh ich are sensitive to visible and UV light (Adler 1976). The pineal body and the frontal organ are implicated in several responses, such as pigmentary adaptation (e.g. Bagnara and Hadley 1970), sun compass orientation (e.g. Taylor and Fergusson 1970; Taylor and Adler 1978), and synchronization of circadian locomotor rhythms (e.g.Adler 1969). The pineal body has a low number of photoreceptors resembling the cones and rods in the lateral eyes. The membrane disk of the outer segments of these intracranial photoreceptors are parallel to the incident light and thus polarization sensitive, unlike the cones and rods in the retina, where the disks are perpendicular to the incident light.


Among amphibians, polarization sensitivity and spatial orientation on the basis of polarization have been demonstrated until now only in adult tiger salamanders, larval and adult red-spotted newts as well as larval bullfrogs.

29.1 Tiger Salamander Ambystoma tigrinum

Taylor and Adler (1973) studied the spatial orientation of adult tiger salamanders Ambystoma tigrinum under an artificial source of polarized light in indoor experiments. The animals were trained in an elongated waterfilled galvanized metal tank, in wh ich an artificial shore was provided at one end. The shoreline was illuminated by totally linearly polarized light from above in such a way that the E-vector was parallel to the shore and perpendicular to the long axis of the tank. During training, the animals were removed from the shore, placed in the water at the opposite end of the tank, and allowed to return to the shore. The animals were tested in a circular arena (in aquatic tests with water and in terrestrial tests without water), over the cent re of which the source of polarized light was suspended. The E-vector was rotated by 90° from its direction in training to reduce the possible influence of geographical clues. In the test arena untrained salamanders oriented randomly under both unpolarized and totally polarized light. This indicated that there was no spontaneous orientation with respect to the Evector. The trained salamanders moved perpendicularly to the E-vector. Their orientation was as expected bi-directional. From this Taylor and Adler concluded that tiger salamanders can learn to orient with respect to the Evector direction, and that the animals may orient by means of the celestial and/or the underwater polarization pattern when the sun is not visible. Tiger salamanders migrate after sunset to shallow breeding pools, when both the temperature and the risk of predation are lower than during daytime. In wooded areas the sun is obscured by landmarks for longer periods during the day. Thus, the ability to perceive the celestial and/or underwater polarization pattern would provide a relatively continuous knowledge of the solar direction throughout the day and even for a time after sunset.

When the eyes of salamanders were removed, also they oriented bimodally and perpendicularly to the E-vector (Adler and Taylor 1973). However, if the skull of the animals was covered by opaque plastic between the eye sockets and from immediately behind the nostrils to the posterior end of the skull, whether seeing or blinded, their orientation was random under polarized light. If the plastic cover was transparent, bi-directional orientation perpendicularly to the E-vector was observed. From this, Adler and Taylor concluded that the perception of polarization is mediated by extraocular photoreceptors rather than by the lateral eyes.

29 Polarization Sensitivity in Amphibians 319

Hence these polarization-sensitive extraocular receptors are assumed to be intracranial. Polarized light passing through the various layers of the skin and skull may change considerably before it reaches the receptors. In Ambystoma the skin of the head is isotropie and strongly reduces the intensity of transmitted light (Adler and Taylor 1973). The cartilaginous skull also reduces light intensity and has optieal axes parallel to the body axes and functions as a wave-retarding plate. Thus, when the incident light is linearly polarized, the characteristies of transmitted light can vary from almost completely linear to elliptieal polarization, depending on the E-vector direction of incident light relative to the orientation of the skull's wave-plate axes.

According to Adler and Taylor, the polarization-sensitive intracranial photoreceptors of salamanders may be located in the pineal body on the dorsal surface of the diencephalons, direcdy beneath that part of the skull, whieh was covered with plastie in certain experiments. They hypothesized that the outer segments of the pineal photoreceptors of salamanders can perceive polarization, since the incident light may pass through them non-paraxially with respect to their longitudinal axis, to whieh the membrane disks are perpendieularly oriented. For example, Oksche (1965) found outer segments, whieh protrude horizontally into the lumen of the epiphysis. This orientation of the outer segments allows light to pass through them transversely.

The major problem with the experimental apparatus of Taylor and Adler (1973) is that they used black plastie covers around the training and testing tanks (see Chap. 34). The entire training tank, including the area direcdy above it, was enclosed with opaque black plastie curtains such that no light penetrated from outside. The test arena, a circular pale blue plastie pool, was also completely surrounded by an opaque black plastie curtain. No differences in light intensity were measured directly beneath the polarizing filter when its transmission axis was rotated by 90° from its original direction. However, the brightness distribution around the training tank and test arena was not measured. Thus, it cannot be excluded that the salamanders utilized brightness patterns induced by selective reflection of polarized light from these black coverings. According to Taylor and Adler, the utilization of such brightness patterns is unlikely, because ''Ambystomatids typieally are negatively phototactie and, as such, their typical response should have been parallel, not perpendieular, to the e-vector since the region of maximum reflection and thus brightness is perpendieular to the e-vector." However, this argumentation loses its validity if both in the training and testing situation, reaching the shore by positive phototaxis towards the brightest region of the optieal environment (perpendieular to the E-vector) was more important for the salamanders than finding a shelter by negative phototaxis towards the darkest region (parallel to the E-vector). Hence, the observed reaction of salamanders, bi-directional orientation perpendieularly to the E-vector, can also be explained exclusively by a response to the brightness pattern, i.e. bimodal orientation towards the brightest directions associated with the shore during


training, the directions of which are always perpendicular to the E-vector. Control experiments should have proven that the salamanders responded indeed to the polarization and not to the concomitant brightness patterns.

29.2 Red-Spotted Newt Notophthalmus viridescens

Taylor and Auburn (1978) studied the orientation of larval (eft stage) and adult red-spotted newts Notophthalmus viridescens. Outdoor experiments were performed at sunset, when the sun was too low to be visible. The animals were trained in outdoor elongated tanks under clear skies by placing them in the shallow end of the tank, from which they could swim back to deep water (adult newts), or in the case of the efts, they were placed in deep water and allowed to move back to shore. Testing was done in an outdoor circular arena filled with shallow water and enclosed by an opaque wall, which obscured all landmarks, but allowed a view of the sky. Efts were tested with completely black surroundings, and also with white panels placed over opposite quarters of the test arena, either perpendicularly or parallel to the celestial E-vector at the zenith. Adult newts trained to a given geographical direction oriented in a direction that would have returned them to deep water had they been in their training tank, when they were tested both at midday and after sunset under clear skies. They oriented randomly when tested after sunset under overcast skies. Efts oriented in the predicted direction under clear skies after sunset, but they oriented randomly after sunset under overcast skies. These tests were made with the white panels in place, either perpendicularly or parallel to the E-vector. This indicated that the efts were responding to the E-vector and not to the brightness patterns induced by selective reflection of polarized light.

In another experiment, adult newts were trained to two different compass directions in the outdoor tanks and then tested indoors under artificial totally linearly polarized light. One group was tested in the circular arena surrounded with black curtains. A second group was tested first with white quadrants placed parallel to the inside overhead E-vector and then with white quadrants placed perpendicularly to the E-vector. This second group was also tested with white quadrants in pi ace, but under unpolarized light. In all three tests under polarized light the newts oriented as predicted relative to the imposed E-vector, Le. they responded to the E-vector as if it were the primary celestial E-vector outdoors at that time of day.

From these experiments, Taylor and Auburn (1978) concluded that larval and adult red-spotted newts can perceive the E-vector of linearly polarized light and use the celestial polarization pattern for orientation.


29.3 Larval Bullfrog Rana catesbeiana

Taylor and Auburn (1978) as weIl as Auburn and Taylor (1979) demonstrated that bullfrog tadpoles Rana catesbeiana can perceive linear polarization for orientation. Tadpoles were trained indoors to move perpendicularly to the shoreline illuminated by totally linearly polarized light (produced by a waxpaper depolarizer and a polarizer) with E-vector either parallel or perpendicular to the shoreline. During training they were placed on the shore at one end of an elongated tank, from which they were allowed to swim back to deep water parallel to the long axis of the tank. Animals trained under polarized light with the E-vector perpendicular or parallel to the shoreline oriented bimodally parallel or perpendicularly to the E-vector when tested indoors in a circular arena illuminated by totally linearly polarized light. Rotation of the E-vector resulted in the predicted change in the bi-directional orientation. These animals oriented randomly under unpolarized light, when the order of the waxpaper depolarizer and polarizer was inverted. This was to be expected if tadpoles can orient by means of the E-vector. Animals trained under unpolarized light oriented randomly in the test. Thus, there was no innate preference for movement direction with respect to the E-vector. Hence, such a preference must be learnt.

Control experiments were performed in order to exdude the possibility that the animals were merely responding to the brightness pattern induced by selective reflection of polarized light from the black walls of the training and testing apparatuses. In these experiments alternating curved black and white quadrants were placed around the test arena in such a way that the surface of the white quadrants at their centre was either perpendicular or parallel to the E-vector. The latter case simulates qualitatively the brightness distribution induced by reflection of polarized light from the walls around the arena (Fig. 34.1). In the test with this black-and-white pattern under unpolarized light, the responses were not significantly oriented toward either the black or the white quadrants, nor was there any evidence of an oriented response to any particular direction. This shows that under polarized light the orientation of tadpoles was not governed by the reflection-induced brightness pattern. When in the test under polarized light the black-andwhite pattern conflicted with the brightness pattern created by the selective reflection of polarized light, the animals oriented bimodally as expected if they respond to the polarization.

Tadpoles trained to move toward a compass direction outdoors oriented predictably at sunrise and sunset under dear skies with the sun out of view, but randomly under overcast skies. Untrained tadpoles tested direct1y after collection from a restricted area of the shore of a pond oriented similarly. The involvement of light polarization in this response was demonstrated by experiments in which tadpoles, collected within a small area of apond, trained


outdoors under clear skies and tested indoors under artificial polarized light, oriented bimodaHy in a particular direction relative to the E-vector. This direction was the same as the offIonshore direction in the horne pond relative to the E-vector of polarized skylight at the zenith outdoors at the time of the test. The off- and onshore responses are characteristic movements perpendicular to the shoreline in the horne pond towards deep and shallow water, respectively.

Auburn and Taylor (1979) suggested that the possible significance of the ability of buHfrog tadpoles to perceive polarization and learn the E-vector direction as weH as to use it for orientation may be in their offIonshore response. This movement may have several functions, including temperature regulation, foraging and shelter from predators. This type of spatial orientation is often termed "y-axis orientation". The solar/antisolar meridian deduced from the celestial polarization pattern could serve as a useful reference for determination of the preferred horne offIonshore direction, which must be learnt, since it varies with location in the pond.

Bullfrog tadpoles also possess a pineal organ as weIl as a frontal organ (Adler 1976), and eyeless tadpoles can orient at sunset under clear sky (Auburn and Taylor 1979). On the basis of orientation experiments with tadpoles, the eyes as weH as pineal and frontal organs of which were manipulated (cut, ectomized or covered), Taylor and Auburn (1978) as weH as Auburn and Taylor (1979) suggested that the polarization-sensitive photoreceptors in tadpoles may be positioned in the pineal and/or frontal organs. Although adult bullfrogs exhibit sun-compass orientation (Ferguson et al. 1968), it is unknown whether they can also perceive polarization and use it for orientation.

29.4 Proposed Mechanisms of Detection of Polarization in Amphibians

• According to Phillips et al. (2001), the detection of both light polarization and magnetic fields may be mediated by the same biophysical mechanisms, at least in amphibians. They suggested that photoreceptor-based magnetoreceptors involving a radical-pair re action proposed, e.g. by Ritz et al. (2000) mayaiso be sensitive to polarization. In such a mechanism, the constraints on the alignment of the moleeules in which the magnetic interaction occurs are likely to be more stringent than in a typical vertebrate photoreceptor. In the latter, the photopigment moleeules are non-randomly aligned with respect to the plane of the membrane in the outer segments, but free to rotate within this plane. The importance of the ordered array of the light-absorbing moleeules in a radical-pair mechanism sterns from the complex three-dimensional pattern of response generated by this type of


mechanism (Ritz et al. 2000). For example, at some field intensities, this pattern would consist of a large circular field with or without surrounding concentric rings at either end of the magnetic field lines. Deriving directional information from such a pattern would require either sequential sampling with a single receptor or simultaneous sampling with a threedimensional array of receptors (Ritz et al. 2000). In either case, variation in the alignment of molecules involved in the radical-pair reaction within an individual photoreceptor would degrade the three-dimensional pattern and, consequently, decrease or eliminate the dependence of the receptor's response on magnetic field alignment. Therefore, if a radical-pair-based magnetoreceptor is present in an organism, selection should favour a reduction of variation in the alignment of the light -absorbing molecules, ultimately resulting in a highly ordered linear array. One consequence of such a highly ordered array would be strong dichroic absorption of light, resulting in sensitivity to the E-vector of polarized light in addition to the alignment of the magnetic field .

• Phillips et al. (2001) also suggested that photoreceptor-based magnetoreceptors involving freely rotating magnetic particles proposed by Edmonds (1996) mayaiso be sensitive to light polarization. The mechanism involves needle-shaped single-domain particles of magnetite suspended in a liquid crystal made up of elongated carotenoid molecules, like those found in the oil droplets in some vertebrate photoreceptors. Edmonds (1996) has shown that the alignment of the carotenoid molecules can be affected by the mechanical strain energy resulting from the rotation of the single-domain particles by the geomagnetic field. The minimum strain energy occurs when the molecules are aligned parallel to the long axes of the particles and, thus, parallel to the axis of the magnetic field. Because the absorption of light by carotenoid molecules is strongly anisotropie, transmission of light through such a magnetite-containing oil droplet would vary with alignment of the magnetic field. Magnetic fields aligned parallel to the long axis of the photoreceptor and causing the magnetite particles as weIl as associated carotenoid molecules to be aligned along this axis would result in a higher light intensity reaching the outer segment. On the other hand, magnetic fields aligned perpendicularly to the long axis of the receptor and causing the particles as weIl as carotenoid molecules to be aligned perpendicularly to the photoreceptor axis would result in a lower light intensity reaching the outer segment (Edmonds 1996). In addition, changes in the alignment of the magnetic field (and corresponding changes in the alignment of the carotenoid molecules) in the plane normal to the long axis of the photoreceptor would cause differences in the transmission of polarized light with different E-vectors. As in the case of a radical-pair-based mechanism (e.g. Ritz et al. 2000), therefore, the mechanism proposed by Edmonds would be likely to exhibit both magnetic and polarization sensitivity.

30 Polarization Sensitivity in Reptiles

30.1 Celestial Orientation in Reptiles and the PolarizationSensitive Parietal Eye of Lizards

Although there is much information on the homing capacity of reptiles, the role of celestial cues in time-compensated orientation has been studied only in certain freshwater turtles, land tortoises, alligators, snakes and lizards (Chelazzi 1992). The freshwater turtles Terrapene carolina, Chrysemys picta and Trionyx spinifer use celestial cues to orient homeward and may have a sun compass (DeRosa and Taylor 1980), like the land tortoise Gopherus polyphemus (Gourley 1974). Murphy (1981) also found that the sun compass governs the on/offshore movements of juvenile American alligators (Alligator mississipiensis). Some evidence for celestial orientation in displaced snakes has been presented by Gregory et al. (1987). The western diamondback rattlesnake Crotalus atrox has the ability to orient by solar cues (Landreth 1973), and the water snakes Regina septemvittata and Natrix sipedon can use the sun compass for on/offshore orientation (Newcomer et al. 1974). In these snakes a sun compass mechanism based on an internal dock was demonstrated, which capacity seems to be related to orientation to water bodies, not necessarily involved in homing. Sun compass orientation has been demonstrated in the lizard Lacerta viridis (Fischer 1961). However, the homing mechanisms of these reptil es are far from being darified. The possibility that skylight polarization is used by sea turtles, freshwater turtles, land tortoises, crocodiles, alligators, gavials and snakes has not yet been investigated. Lawson (1985) suspected a role of skylight polarization in the orientation of the snake Thamnophis radix, but this hypothesis was never tested experimentally.

Until now, among the a pproximately 6000 living species of reptiles only two lizard species have been proven to be sensitive to polarization. Evidence for a sky polarization compass in these species originates from laboratory experiments, in which fringe-toed lizards (Uma notata) were able to use skylight polarization patterns as celestial compass (Adler and Phillips 1985), and sleepy lizards (Tiliqua rugosa) were able to orient by means of polarization (Freake 1999). Certain lizard species - e.g. the North American iguanid lizards Sceloporus orcutti (Weintraub 1970), Dipsosaurus dorsalis (Krekorian 1977) and Sceloporus jarrovi (Ellis-Quinn and Simon 1989) - are able to horne suc-

30 Polarization Sensitivity in Reptiles 325

cessfully after displacements of about 200-300 m. Sceloporus jarrovi is the first lizard species in which field displacement experiments have dearly revealed some of the sensory mechanisms available to them (Ellis-Quinn and Simon 1991). This lizard can orient homewards by means of a celestial compass. Covering the parietal eye disrupts homing.

In addition to the photoreceptive intracranial pineal body or Epiphysis cerebri, there exists in most lizards a parietal eye, often called the «third eye", located in a foramen between the parietal bones in the skull roof (e.g. Eakin 1973). This eye has a layered cornea fused to a slightly curved lens and also has a well-developed retina with cone-like photoreceptors (Hamasaki and Eder 1977). The principal function of the parietal eye is to act as a photo-neuroendocrine transducer, converting photic stimuli into a neuroendocrine signal, wh ich is melatonin (Firth and Kennaway 1980). It plays an important role in thermoregulation by acting as a radiation dosimeter (Eakin 1973), and in maintaining circadian rhythms possibly through an antagonistic chromatic mechanism in parietal eye photoreceptors, which act as a dawn and dusk detector to determine the beginning and end of daylight (Solessio and Engbretson 1993).

Hamasaki and Eder (1977) proposed that the arrangement of photoreceptors in the parietal eye of lizards may be suitable for analysing the skylight polarization. The photoreceptor outer segments protrude into the parietal eye lumen, and so there is a ring of photoreceptors that lie in a plane perpendicular to the light ente ring the eye. Vertebrate photoreceptors are sensitive to the E-vector direction of off-axis polarized light, with the maximum response occurring when the E-vector is parallel to the disks' membrane, Le. perpendicular to the longitudinal axes ofthe receptor cells (e.g. Schmidt 1935; Denton 1959; Adler 1976). Since each photoreceptor in the ring is therefore maximaHy sensitive to one particular E-vector direction, this circular array of receptors could theoretically be used to determine the angle of polarization of skylight.

30.2 Desert Lizard Uma notata

The first reptile species in which an ability to orient by means of light polarization was demonstrated is the desert-dwelling fringe-toed lizard, Uma notata (Adler and Phillips 1985). Lizards were trained to orient in a predictable direction under natural skylight in outdoor elongated tanks. During outdoor tests in a circular arena shifting the phase of the internal dock of the animals by -6 hand +6 h induced 90° shifts dockwise and counter-dockwise, respectively, in the compass direction that would have been necessary to reach a shaded chamber at one end of their training tank. This was expected if a time-compensated celestial compass is involved. In another experiment, the


lizards were allowed to freely move along an elongated training tank with a shaded chamber at one of its ends under natural skies. When they were tested indoors in a circular arena under an overhead artificial totally linearly polarized light source, abimodal E-vector-dependent orientation was observed. This was evident when the response directions were transposed to match the E-vector rotation of skylight at the zenith, Le. the expected direction of movement with respect to E-vector was corrected for the time of day. The possible role of room-specific cues, brightness patterns associated with differential reflection of polarized light from the arena walls, or geomagnetic field in the response of the lizards was thus precluded. On the basis of these experiments, it was not possible to determine whether the sun and/or the pattern of skylight polarization were utilized, since both require time-compensation. Nevertheless, the results of these experiments show that polarization can be used alone by these lizards and in a manner wh ich compensates for the regular daily rotation of the celestial E-vector pattern. Although the underlying mechanism is unknown,Adler and Phillips (1985) suggested that the parietal eye may be involved.

Fringe-toed lizards live in aeolian deserts with loose, wind-blown sand terrains having few landmarks. Although some fixed landmarks, e.g. vegetation in their natural habitat are potentially available as orientation cues, these are only observable when the animals are at the edges or on the crests of dunes. These lizards are prey for numerous terrestrial and aerial predators. They escape from open, vegetationless sand dunes to refuge beneath bushes or in burrows at the base of bushes. During escape they move along relatively straight paths even if the burrow entrance cannot be seen. Since the substratum configuration is modified continuously by wind, this habitat has virtually no stable surface features. For the diurnal directed escape movements of the lizards in such a habitat the constantly available and reliable solar direction and celestial polarization pattern are of particular importance. These lizards could be considered as the vertebrate analog of the desert ant Cataglyphis.

30.3 Sleepy Lizard Tiliqua rugosa

Freake (1999) trained adult Australian sleepy lizards (Tiliqua rugosa) to move in a direction under natural skylight in outdoor pens. When tested under clear skies in the late afternoon without a view of the sun, the lizards exhibited symmetrical bimodal orientation along the trained direction. From this Freake (1999) concluded that the bimodality of the response is consistent with the possibility that the lizards were orienting with respect to the E-vector of skylight. Nevertheless, alternative explanations were not excluded, for exampIe, bimodal orientation relative to the geomagnetic field, or to the patterns of intensity and/or colour of skylight. To confirm that the lizards can orient by

30 Polarization Sensitivity in Reptiles 327

me ans of an overhead polarization pattern, animals were trained in indoor pens to move in one direction under a totally linearly polarized light source. When tested in a circular indoor arena illuminated from above by totally linearly polarized light, the lizards used the E-vector to orient bimodally along the trained direction. They were able to determine E-vector direction and learn its spatial relationship relative to a goal. There was no evidence that the lizards were using any room-specific cues, or brightness patterns associated with differential reflection of polarized light from the arena walls (see Chap. 34), or geomagnetic field to orient in the training direction. These results support the hypothesis that the lizards can use the E-vector direction of skylight as a compass.

In field experiments, male and female sleepy lizards were displaced 250-800 m from their horne either with or without access to terrestriallandmarks and celestial optical cues. Their homeward orientation was significandy worse if visual cues were not available (Freake 2001). When lizards were displaced with their field of view restricted to the sky, their homeward orientation was as good as that of lizards displaced with no such visual restriction. These experiments of Freake (2001) suggest that sleepy lizards use celestial cues to determine the compass bearing of their outward journey, and reverse this bearing to orient in the homeward direction. Lizards oriented randomly with respect to the homeward direction when their parietal eye was entirely covered with a patch during the displacement and return, while lizards which were fitted with a sham patch (a hole cut in the patch exposed the parietal eye to unobstructed skylight) were weIl oriented homewards. In both groups, the lateral eyes were unobstructed and had complete access to celestial cues and landmarks. Hence, the parietal eye plays a significant role in the homing behaviour of sleepy lizards, perhaps mediating a time-compensated sky compass sense.

Sleepy lizards are slow-moving animals with no climbing ability and are thus unable to elevate their field of view by more than approximately 10 cm. As a consequence, in their normal habitat they would often have difficulty observing landmarks, because these are obscured by relatively tall vegetation. Thus, celestial compass orientation is a particularly useful orientation mechanis m during trips within and beyond the large horne ranges.

31 Polarization Sensitivity in Birds

Birds can use several environmental cues - sun, stars, geomagnetic field, wind direction, odours, landmarks, sound sources, patterns of the intensity, colour and polarization of skylight - for orientation and navigation (Schmidt-Koenig 1979). Until now, only the homing pigeon (Columba livia) has been shown convincingly to possess a sun compass. The magnetie compass seems increasingly to be central to most avian orientation behaviour. Young migratory birds enter the world with two innate representations of the migration direction, one coded with respect to the geomagnetie field, the other to the celestial rotation (Wiltschko and Wiltschko 1991). A functional magnetic orientation capability develops in birds that have never seen the sky (e.g. Wiltschko et al. 1998). The preferred migration direction based on magnetie information develops early in life and it may be modified and calibrated by celestial rotation, observed either in clear daytime or at night at least during the first few months of the life of naive migratory birds. This calibration phenomenon was discovered by Bingman (1983) in Savannah sparrows (Passerculus sandwichensis) '. Celestial rotation provides information about geographie directions during both day and night. At night, a rotating pattern of artificial stars in laboratory experiments provided a sufficient stimulus to calibrate magnetie orientation in young Savannah sparrows (Able and Able 1990a). According to Able (1993), the orientation of migratory birds and homing pigeons is aided by multiple stimuli. In short-term orientation of many migratory birds, the geomagnetie field takes precedence over stellar patterns. Celestial optieal information at sunset overrides both of these cues, and the skylight polarization pattern might to be the primary cue in dusk orientation.

The possibility that migratory and homing birds could perceive skylight polarization and utilize it in their homing flight was raised first by Donald R. Griffin in 1950 in a talk (cf. Montgomery and Heinemann 1952). Since that time this hypothesis has been tested and confirmed in several bird species. By means of the dynamie pattern of polarization, a bird might localize the axis of rotation of the celestial hemisphere, as suggested by Brines (1980). Unfortu-

I The Savannah sparrow is named after the city of Savannah (Georgia, USA), thus "Savannah" should always be capitalized.

31 Polarization Sensitivity in Birds 329

nately, it is technieally not possible to simulate this dynamic pattern in laboratory experiments and thus the Brine's model is not testable.

On the other hand, it is possible that birds are not using skylight polarizati on as an indirect means of performing sun compass orientation, i.e. using the polarized skylight to indieate where the invisible sun actually iso They may use the celestial polarization as an independent compass learned by the same sort of mechanism (i.e. via celestial rotation) that is involved in the development of the star compass.

Birds might also perceive the location of geographie north by averaging the sunrise and sunset positions of the band of maximum degree of linear polarization of the sky, as suggested by Phillips and Waldvogel (1982, 1988). When the sun is on the horizon at sunrise and sunset, the celestial band of maximum polarization comprises an are from horizon to horizon through the zenith. The band itself and the E-vector of polarized skylight within it are aligned orthogonally to the solar-antisolar meridian (Fig. 31.1B). At the equinoxes (21st March and 23rd September), when the sun rises exactly east and sets exactly west, the band and its E-vector will be aligned geographieally northsouth and will intersect the horizon vertieally only in these two compass directions. Because sunrise and sunset azimuths are always symmetrie al with respect to geographie north, geographie north could be localized during the entire year by determining the azimuth midway between the sunrise and sunset vertieal intersection points on the horizon. This method would enable migrating birds to utilize this polarized light calibration system without reference to a time-compensation mechanism.

Possessing several different compasses provides protection against spatial and temporal variability in the quality and availability of the physieal factors upon which the compasses are based. Celestial cues are often obscured by clouds, and the geomagnetic field exhibits spatial and temporal changes. This high degree of redundancy in the compass systems for avian migration demonstrates the strong selection for accurate migratory orientation (Alerstarn 1990). The present richness of orientation mechanisms in birds may reflect both traits that have evolved within the context of avian migration and conserved phylogenetieally.

However, firm generalizations are weakened due to possible species-specific differences in the weighting of orientation cues, and to differences in experimental designs. This may be the reason for the controversial results on the ability of certain bird species to perceive the magnetic field and skylight polarization. For example, it is technieally not possible to accurately simulate the celestial polarization patterns. Covering orientation cages with linearly polarizing filters confronts birds with a visual stimulus that differs profoundly from the natural one. In these cases it is unclear whether the tests under such polarizers simulate the natural role of skylight polarization correcdy. Eliminating the polarization of skylight with depolarizing materials is a better experimental option. According to Martin (1991), the principal weak-


ness of studies of polarization sensitivity in birds has been the failure of investigators to find a satisfactory way in which light with different planes of polarization could be discriminated.

Although several bird species can perceive UV light (e.g. Tovee 1995), in the majority of avian behavioural experiments under manipulated linearly polarized light fields stimuli were applied within the visible (400 nm < A < 750 nm) range of the spectrum, since the polarizers absorbed strongly or totally UV (A < 400 nm) light. Thus, whereas one cannot say that the investigated birds do not respond to polarization in the UV, the behavioural data show that these birds are responsive to polarization within the visible part of the spectrum.

In many cases it is only known that certain avian species use celestial cues for orientation during daytime or in the period between sunset and the first appearance of the brightest stars. Yet in most cases it is unknown whether the dominant cue is the pattern of intensity or colour or polarization of skylight, or the sunset glow. We mention only one case without convincing evidence for polarization sensitivity. Wiltschko et al. (1998) supposed that Australian silvereyes Zosterops lateralis used the pattern of skylight polarization for orientation at sunset. However, from their experiments performed with manipulated magnetic fields under natural evening skies beginning half an ho ur before sunset, one could have only conduded that these birds can orient by means of celestial cues.

31.1 Crepuscularly and Nocturnally Migrating Birds

31.1.1 White-Throated Sparrow Zonotrichia albicollis and American Tree Sparrow Spizella arborea

Able (1982, 1989) investigated the orientation of night-migrating whitethroated sparrows Zonotrichia albicollis and American tree sparrows Spizella arborea between sunset and the first appearance of the brightest stars, a period of about 50 min, under differently manipulated skylight polarization patterns. Under dear skies in orientation cages covered by linearly polarizing filters with transmission axes either 45° dockwise or 45° counter-dockwise from the sunset azimuth, white-throated sparrows oriented bimodally parallel to the artificial overhead E-vector with a slight bias towards the sunset. American tree sparrows oriented unimodally parallel to the E-vector with a strong bias towards the sunset. Under dear skies white-throated sparrows oriented unimodally towards the sunset under polarizers with transmission axes either parallel or perpendicular to the solar meridian, but in the latter case with a much greater angular spread and larger deviation from the solar meridian. Under totally overcast skies at dusk in orientation cages covered by polarizers with transmission axes aligned at 20° intervals in different compass


directions, white-throated sparrows oriented bimodally and parallel to the actual overhead E-vector.

Able (1982, 1989) interpreted the orientation of these birds parallel to the E-vector by the fact that during spring and autumn migration in temperate northern latitudes the natural twilight zenith E-vector provides an approximately N -S axis, corresponding to the main flow of bird migration. In a control experiment, the orientation cages with white walls were divided into eight equal sectors and alternate sectors were covered with grey blotter paper. White-throated sparrows tested under dear skies and under polarizers with transmission axes ±45° from the solar meridian oriented bimodally parallel to the overhead E-vector irrespective of whether the grey sectors were parallel or perpendicular to the E-vector. From this, Able (1989) conduded that white-throated sparrows were not responding to brightness patterns induced by selective reflection of polarized light from the cage walls. When the birds were tested under dear skies at dusk in orientation cages covered with pseudodepolarizers, their orientation was unimodal and seasonally appropriate and the same as that of control birds viewing an unaltered dear sky. For the functioning of pseudodepolarizers see Shurdiff (1962).

Able (1989) conduded that although skylight polarization patterns are not necessary for nocturnal migratory orientation in white-throated sparrows and American tree sparrows, these birds respond to changes of the E-vector direction, and thus are sensitive to polarization in the visible range of the spectrum. These birds may use skylight polarization at dusk as one of their multiple compass capabilities. The relevant stimulus is the E-vector orientation rather than other parameters of skylight, e.g. patterns of intensity, colour or degree of linear polarization.

31.1.2 Northern Waterthrush Seiurus noveboracensis and Kentucky Warbier Oporornis formosus

Night-migrating birds sometimes fly during early morning. The function of this morning flight can be either to correct for wind drift sustained during nocturnal migration, or it can represent migratory restlessness. In the latter case, nocturnal migrants crossing ecological barriers initiate flight at night, but necessarily complete their flight in daytime. Morning flights of typically night-migrating passerines were observed in the northern waterthrush Seiurus noveboracensis and the Kentucky warbier Oporornis formosus by Moore (1986). When these birds were tested in orientation cages on dear mornings with the horizon glow from the rising sun visible, they oriented northward, which was a seasonally appropriate direction. The direction of their orientation was influenced in a predictable way by manipulation of the pattern of skylight polarization. The polarization experiments were performed under dear skies before the solar disc was visible to the caged migrants. The top of


the orientation cages was covered with linear polarizers in such a way that in the first and second group of test birds the transmission axis of the polarizers was oriented 45° counter-clockwise and 45° clockwise from the azimuth of sunrise, respectively. The two groups were thus exposed to skies of identical patterns of colour, intensity and degree of linear polarization of skylight, but the E-vectors differed from each other by 90° and from the natural zenith sky E-vector by ±45°. Under these conditions, both bird groups oriented bimodally and approximately parallel to the E-vector. This was expected when the birds perceive the band of maximum polarization of skylight and use it for orientation, since at sunrise on time average the E-vector from this band is approximately parallel to the N -S direction at sunrise and sunset. Thus, Moore (1986) concluded that these two species can perceive skylight polarization and use it for orientation at sunrise.

31.1.3 Yellow-Rumped Warbier Dendroica coronata

In cue-conflict experiments, Moore and Phillips (1988) examined the relative role of the sunset position and the pattern of skylight polarization during dusk orientation of the yellow-rumped warbier Dendroica coronata in the local geomagnetic field. In a first series of experiments, the top of the orientation cages was covered with a linearly polarizing filter. In the first and second bird groups the transmission axis of this polarizer oriented 60° counterclockwise and 60° clockwise from north, respectively. At sunset, under clear skies these birds oriented bimodally and approximately parallel to the overhead E-vector. Birds of the third (control) group could see the clear sky at dusk without the polarizer, and they oriented unimodally in a seasonally appropriate migratory direction, approximately parallel to the E-vector of the celestial band of maximum degree oflinear polarization at sunset. Hence, yellow-rumped warbiers are sensitive to the polarization of skylight.

In a second series of experiments of Moore and Phillips (1988), the view of the zenith sky was blocked by a sheet of plywood and yellow-rumped warbIers were looking out to the sky near the horizon through side windows of hexagonal enclosures with an orientation cage at their centre. Light entering each window of the hexagon passed through a pseudodepolarizer and then through a polarizer transmitting light of wavelengths over 400 nm and producing totally linearly polarized light. The depolarizer consisted of a sheet of high wave-retardation Melinex mylar sandwiched between four sheets of commercial mylar. This arrangement functions as a pseudodepolarizer, Le. the E-vector of transmitted light is rotated by different angles in different narrow wavelength bands. Thus, broadband sensors, such as the avian visual photoreceptors, will see the transmitted light as depolarized with a maximal degree of polarization as low as ab out 8 % in the visible and near-UV regions of the spectrum. Two of the polarizers 180° apart were oriented to produce a


vertical E-vector to simulate the intersection of the celestial band of maximum degree of polarization with the horizon at sunset. The remaining four polarizers produced horizontal E-vectors. The azimuth direction of the opposite windows with vertical E-vector were rotated either clockwise or counterclockwise from north with the same angle. The apparent position of sunset was either not changed, or it was rotated by 90° south from its normal westerly azimuth with mirrors attached at 45° from the vertical cage windows. Although the birds could also see the non-mirrored or mirrored sunset glow, in both cases the majority of them oriented bimodally and approximately parallel to the azimuth direction of the opposite side windows with vertical Evector. However, a few individuals oriented phototactically toward the reflected sunset position. Hence the majority of birds used only the E-vector pattern for orientation rather than integrating information from both the polarization pattern and some other source of directional information, e.g. from the sunset glow or the geomagnetic field.

In another mirror experiment of Moore and Phillips (1988), the light through the side windows was depolarized by reversing the filter arrangement in such a way that the mylar depolarizer was placed on the interior of the polarizer, and the transmission axis of all six polarizers were oriented horizontally. (Note that this arrangement resulted in that the side windows facing nearly perpendicularly to the mirrored solar-antisolar meridian transmitted less amount of depolarized light, since the mirrored incident skylight was vertically polarized). Then yellow-rumped warbiers oriented unimodally and always westerly rather than parallel to the actual azimuth of vertically polarized opposite side windows, as expected if they used the reflected sunset cues when polarized light cues are absent. From these experiments, Moore and Phillips (1988) concluded that yellow-rumped warbiers orient preferentially on the basis of polarization information when both the sunset and polarization cues are visible, but surprisingly the birds do not use other cues, e.g. sunset glow or magnetic field, to resolve the ambiguity of the celestial polarization pattern.

In another series of experiments with the same design, Phillips and Moore (1992) exposed two groups of yellow-rumped warbiers to a polarization pattern in which the azimuth of the two opposite side windows with vertical E-vector was 60° clockwise and 60° counter-clockwise from the N-S direction, respectively. When tested for 2 h during the sunset period in an unaltered local geomagnetic field, birds oriented bimodally and parallel to the azimuth of the two opposite windows with vertical E-vector. Then, birds were tested with vertical transmission axes of the polarizers in all six side windows, with which directional polarized light information was eliminated. In this case, the birds continued to orient bimodally in a direction coinciding with that in the previous experiment. In a following experiment, mirrors were attached at 45° from the cage windows to rotate the apparent position of the setting sun by 90° counter-clockwise with directional polarized light


cues absent (all six windows with vertical E-vector). In this case, the orientation of the birds rotated nearly 90° counter-clockwise to maintain the same angle relative to the setting sun. Phillips and Moore (1992) concluded that celestial polarization patterns at sunset calibrate the sun compass in yellowrumped warbiers.

31.1.4 Blackcap Sylvia atricapilla

Helbig and Wiltschko (1989) studied the nocturnal migratory orientation of the blackcap Sylvia atricapilla under manipulated artificial E-vector patterns. Orientation tests were carried out after sunset until the first appearance of the brightest stars. The test birds could see the entire sky and the horizon with distant landmarks. The orientation cages were covered by linearly polarizing filters with transmission axes oriented E-W, or NE-SW, or NW -SE. Blackcaps oriented bimodally, except when the E-vector direction was E-W, and approximately parallel to the actual E-vector of the overhead polarizer with a slight bias towards the sunset. In the simultaneous control experiments there were no polarizers on the top of the orientation cages, and the birds could see the natural clear sky. In this case, they oriented unimodally towards SW, which is the seasonally appropriate migratory direction. The E-vector direction of the celestial band of maximum degree of polarization changed by 31 ° in the course of the test period.

Helbig and Wiltschko (1989) concluded that blackcaps can perceive skylight polarization, probably in the visible range of the spectrum. Note, however, that according to Coemans et al. (1994a, p. 108), Helbig and Wiltschko (1989) did not exclude the possibility that a type of phototaxis may have been responsible for their results.

Helbig (1990) found that migratory blackcaps, which could see the clear evening sky after sunset, orient appropriately in the absence of meaningful information from the geomagnetic field fluctuating randomly between 5400 and 42,000 nT with average inclinations between 54 and 78° caused by an air conditioning system which made the magnetic field unsuitable for orientation. When, however, the degree of linear polarization of natural skylight was reduced with two sheets of an optically clear pseudodepolarizer, the test birds with unobstructed view of the sky oriented randomly, and their migratory restlessness in autumn drastically decreased. Birds were also disoriented under overcast skies and in a fluctuating magnetic field, while at magnetically undisturbed sites they oriented well without a view of the clear sky. The effect of the depolarizer on the orientation resembled that of a total cloud cover, although the birds could see the clear evening sky and the horizon glow of the setting sun. These results have already provided convincing evidence for the use of natural skylight polarization patterns for orientation in blackcaps, independently of the horizon glow of the setting sun.


However, Bletz et al. (1996) proved that blackcaps from a southern German population are also capable of developing migratory orientation based solely on information from the geomagnetic field. On the other hand, blackcaps from another, Baltic population required celestial cues for correct magnetic orientation (Shumakov and Zelenov 1988). The reason for this populationspecific difference is still unclear, and it demonstrates well the complexity of avian orientation mechanisms.

31.1.5 Savannah Sparrow Passerculus sandwichensis

Able and Able (1990b,c) studied the ontogeny of the nocturnal migratory orientation in the Savannah sparrow Passerculus sandwichensis, which is a nocturnal migrant nesting across North America and migrates southward to northern Central America in winter. In the laboratory, hand-raised Savannah sparrows typically exhibit bimodal migratory orientation. Individual birds sometimes orient for a while in a unimodal direction, but often switch to more or less opposite directions on subsequent nights. When raised indoors without exposure to the sky, they exhibit NW -SE magnetic orientation (Bingman 1983).

Groups of hand-raised young birds were given controlled experience with the clear daytime sky during the first 3 months of their life (Able and Able 1990b,c). The directional relationships between sun azimuth, skylight polarization patterns and magnetic directions were manipulated. In the first experiment conducted in an unaltered geomagnetic field, the rearing cages were covered by linear polarizers, the transmission axis of which was rotated every half an hour in such a way that during periods of exposure to the sky it was always either +90 or +45 or -450 from the solar azimuth. The birds could see the whole sky down to the horizon. In tests at dusk under solid overcast skies during the first autumn migration period in orientation cages covered by polarizers with random directions of the transmission axis, birds reared under the three different E-vector directions oriented bimodally and in directions as predicted if they had learned a compass direction approximately N-S with respect to the imposed E-vectors. In tests between sunset and the first appearance of the brightest stars under clear skies in orientation cages covered by pseudodepolarizers, birds of all three groups oriented unimodally towards the sunset azimuth as if their orientation had been governed by positive phototaxis. In tests under clear dusk skies in cages from wh ich birds could see for the first time both the natural skylight polarization pattern and the sunset position, members of the three groups oriented unimodallywith strong phototactic bias towards sunset and in directions as predicted if their orientations were based on the calibrated natural celestial polarization pattern.

In a second experiment of Able and Able (1990b,c), the E-vector produced by overhead polarizers was always 450 clockwise from the solar azimuth, and


the magnetie field was either unaltered or shifted by a given angle from the direction of magnetie north. In tests under natural clear dusk skies and with a vertieal magnetie field offering no directional cue, birds oriented as if they used the polarization pattern of zenith sky calibrated by the magnetie field with a phototactie bias towards sunset. Able and Able concluded that young Savannah sparrows can learn to perform migratory orientation using skylight polarization, whieh seems to be their orientation cue of first choiee, and this polarized light compass appears to be calibrated during ontogeny by the magnetie field.

In another series of experiments, Able and Able (1993) kept hand-raised Savannah sparrows in anormal geomagnetie field without view of the outdoor optieal environment until the birds became 50-70 days old. Then, in outdoor orientation cages the birds were exposed to the clear sky several times from sunrise to sunset under natural or depolarized skylight and in shifted or unshifted magnetie fields. During their first autumn migratory period the magnetic orientation of the young birds was recorded indoors in orientation cages covered with white translucent plastie sheets in the normal geomagnetie field and in a field in whieh magnetic north was shifted to geographie west. The following results were obtained:

1. Birds reared with exposure to the natural clear daytime sky in the normal geomagnetie field showed abimodal N/NW -S/SE magnetie orientation.

2. Birds reared under natural skies within a shifted E-W magnetie field (magnetic north rotated to geographie west) exhibited abimodal E-W magnetic orientation.

3. The magnetie orientation ofbirds reared in such a way that they could view the clear daytime sky through pseudodepolarizers within the normal geomagnetic field was bimodal and N/NW -S/SE, like that of the first group.

4. Birds reared with exposure to depolarized skylight within an E-W shifted magnetic field showed abimodal N -S magnetic orientation, like the third group.

Hence, altered magnetie migratory orientation was observed only in the young birds, which had visual access to the natural sky, including the sun and skylight polarization, during their rearing period. Birds of all four groups had similar opportunities to see the sun and its movement in the sky. From these, Able and Able (1993) concluded that the pattern of skylight polarization is necessary to calibrate the magnetic orientation in young Savannah sparrows, and the sun is not involved in this process.

In similar experiments, Able and Able (1995b) showed that the calibration of magnetie orientation found in very young Savannah sparrows also occurs in older individuals. However, the experiments did not explicitly test whether light polarizatin or any other specific stimulus was responsible for the calibration of magnetic orientation in older birds, since they were exposed to


both daytime and night skies with no manipulations of visual cues. Nevertheless, polarization could be necessary for calibration of magnetic orientation during daytime. Celestial rotation, accessed via other stimuli (e.g., stars) is an effective calibrating stimulus at night.

Hence, the plasticity to recalibrate magnetic orientation by celestial polarization may persist throughout life, enabling adult Savannah sparrows to compensate for variability in magnetic declination that may be encountered as the birds migrate. For example, in the vicinity of the magnetic equator the magnetic orientation is difficult because the magnetic field lines are horizontal, and birds crossing the magnetic equator must reverse their directional response to the field in order to continue southward. Able and Able (1996) showed that the calibration of magnetic orientation by celestial cues in Savannah sparrows performed during the autumn migration is unexpectedly not evident during migration in the following spring. The reasons for this could be that the calibration might be transient, i.e. it can be forgotten, or the spring and autumn migration directions may be coded independently in young birds.

Able and Able (1995a) allowed hand-raised Savannah sparrows to see the clear sky for 1 h prior to sunrise and for 1 h following sunset in modified Emlen funnel cages. Each funnel cage was covered with a clear acrylic dome to the outside surface of which was affixed a band of linearly polarizing material spanning the dome from horizon to horizon and passing the zenith (Fig. 31.1A). Light transmission through these two layers of material was approximately constant across the visible spectrum and down to 370 nm in the near-UV, below which it dropped off sharply. The remainder of the dome surface was covered with a paraffin layer to depolarize incoming light. The birds never saw the sun or stars. The transmission axis of the polarizer was aligned parallel to the long axis of the band through which the birds could see the sky.

The control group of birds could see the sky through polarizers with the transmission axis aligned 90° from the azimuth of sunrise and sunset (Fig. 31.1B). This alignment simulates roughlythe natural dawn and dusk pattern of polarized light in the part of the sky visible to the birds. Other groups of birds could view the sky through polarizers with E-vector aligned at different fixed angles clockwise or counter-clockwise from the azimuth of sunrise or sunset, i.e. the transmission axis of the polarizers was rotated by a given angle ß from the azimuth clockwise or counter-clockwise relative to the Evector direction presented at sunrise or sunset to the birds of the control group (Fig. 31.1C). When the birds came into migratory mood, their directions of magnetic orientation were recorded in covered Emlen funnel orientat ion cages during indoor tests in a wooden frame building in normal and shifted magnetic fields. Birds were prevented from seeing the outside environment by translucent white plastic sheets covering the cages. The birds of the control group oriented bimodally magnetic N -S, direction 0 1 in


Fig. 31.1B, in both shifted and unshifted magnetic fields. The birds which observed the sky through polarizers with E-vector rotated by angle ß from the E-vector axis for the control group oriented in a direction 02 (Fig. 31.1C), which was rotated approximately by angle ß from the direction of orientation ° 1 of the control group. Hence, the orientation direction followed the rotation of the transmission axis of the polarizers. These results indicate that a fairly simplified, static polarization pattern viewed a limited number of times only at dawn and dusk is sufficient to calibrate the preferred magnetic migratory orientation direction. This suggests that Savannah sparrows analyse the complex sky pattern by some rather simple rules.

In these experiments, Savannah sparrows behave as if the determination of their migration orientation were based on the following algorithm:

• During migration the two directions of the E-vector of the celestial band of maximum degree of linear polarization must be observed and memorized at sunrise and sunset.

• The preferred migration direction (in these experiments north) is the bisector of these two directions.

• If these two directions are equal at equinoxes, then the preferred migration direction coincides with them.

• The magnetic sensation, i.e. the angle of the preferred direction from the magnetic north must be memorized when facing the visually determined preferred migration direction.

• If later the sky polarization pattern will not be visible at sunrise or sunset due to clouds, the preferred direction of migration can be determined by recalling the memorized angle from magnetic north.

• If, indeed, Savannah sparrows used this algorithm, their magnetic orientation could not be calibrated when they saw the celestial polarization pattern only at sunrise or only at sunset. This could be simply tested by a behavioural experiment similar to that conducted by Able and Able (1995a). These manipulations show that the static sunrise and sunset patterns of skylight polarization strongly affect the development of a preferred magnetic direction for migratory orientation. However, there is no information indicating which part of the sky may be most important in this process. Likewise, nothing is known ab out the relative importance of visual experience at sunrise versus sunset. However, later studies (Able and Able 1997) indicated that the sunset orientation develops in Savannah sparrows entirely independently of any magnetic influence.


A

mlcn Funncl cage

GN gcographical nonh prcfcrrcd dircClioll

of oricl1l3tion

0 1

band of transparent l inear polarizer

G

" ,""",' 'X.v // E S5

I

339

Fig.31.1. A Emlen funnel cage with a dome cover, one part of which is covered by a translucent, depolarizing wax layer and the other part by a band of transparent linear polarizer. Such orientation cages were used by Able and Able (1995a) in their behavioural experiments with Savannah sparrows Passerculus sandwichensis, in which the patterns of skylight polarization at dawn and dusk were manipulated. B, C Viewed from above, the directions of the E-vector (EjSR, E/S) oflight passing through the polarizer relative to the directions of sunrise (5R), sunset (55) and geographie north (GN). If EjSR and EjsS were perpendicular to the directions 5R and 55, respectively, the direction of orientat ion Oj preferred by the birds was the bisectrix between the directions Ep and E/s. C If the original E-vector direction (E/R, E/S) was rotated by angle ß clockwise or counterclockwise (EP, ElS), the preferred direction of orientation 02 rotated also approximately by ß clockwise or counter-clockwise from the original direction Oj' (After Able and Able 1995a).


31.2 Day-Migrating Birds

Whether day-migrating birds can also derive information from the complex diurnal patterns of celestial polarization was studied by Munro and Wiltschko (1995) in the Australian yeHow-faced honeyeaters Lichenostomus chrysops in laboratory experiments. These day-migrating birds use the magnetie field and celestial cues for orientation. In the natural geomagnetie field under dear skies, their migratory course is normal. Manipulation of the E-vector by depolarizing the skylight or rotating the E-vector direction failed to affect the orientation of these birds as long as the natural geomagnetie field and the sun were perceptible. When deprived of magnetie information, the birds kept their normal migratory direction as long as either the sun or skylight polarization was visible. These findings indieate that skylight polarization can be used for orientation when no other cues are available. However, in the hierarchy of cues the celestial polarization pattern dearly ranks lower than information from the geomagnetie field. This is in contrast to the predominant role of sky polarization as an orienting factor in some nocturnal migrants (e.g. Able 1982; Moore 1986; Moore and Phillips 1988; Helbig 1991a).

The pattern of skylight polarization varies considerably with season and solar elevation. Hence, its use as a reference requires rather complex mechanisms to find the diurnal migratory direction with sufficient accuracy. The celestial polarization pattern may, together with the sun, represent apart of a general skylight compass mechanism, as has been proposed for insects, in particular für hüneybees and ants (e.g. Frisch 1949, 1967; Wehner 1982). In that case, a limited view of the blue sky might provide birds with similar information as the sun. Clock-shift experiments with nocturnal migrants at dusk (Helbig 1991a) indieated that the internal dock is involved in the orientation during sunset and might be taken to support such a view.

There is a difference between the day-migrating yeHow-faced honeyeaters and nocturnal migrants. In the former, the role of the sun and the celestial polarization pattern is largely equivalent, whereas in European robins as weH as in blackcaps, the polarization pattern of the sky could guide orientation in the absence of magnetie information, while the view of the sun itself could not (Helbig 1991a). The same was true for the long-term effect of the daytime sky on Savannah sparrows during ontogeny (Able and Able 1993). Here, sky polarization at sunrise and sunset mayprove crucial (Able 1991), i.e. patterns whieh are much simpler than those formed during the day when the sun is high in the sky. The minor role of skylight polarization in the orientation of the daymigrant yellow-faced honeyeaters may be caused by the complexity of the pattern and its dependency on the time of the day, season and geographie latitude. Day-migrants would have to adjust this mechanism repeatedly to their current position, whieh drastieally limits the use of skylight polarization as a cue for migratory orientation over long distances.


31.3 Birds Which Might Be Polarization Insensitive or not Use Skylight Polarization in Their Migratory Orientation

341

First, rock doves, alias homing pigeons Columba livia were found to be insensitive to polarization by Montgomery and Heinemann (1952). Later, Kreithen and Keeton (1974) as wen as Delius et al. (1976) found polarization sensitivity in pigeons, which was questioned by Coemans and Vos (1989), Coemans et al. (1990, 1994a) and Vos et al. (1995), who attributed these results to artefacts resulting from minute differences in brightness and/or flicke ring caused by the polarizers themselves or by selective reflections of light in the experimental chamber. The findings of Helbig (1991a) on polarization perception in European robins Erithacus rubecula contradicted the results of Sandberg (1988). The main problem in handling light polarization is that it is easy to translate a linearly polarized stimulus into a brightness difference (Martin 1991). The mentioned studies on pigeon polarization sensitivity also may make questionable the results of several other field experiments, from wh ich it was concluded that the celestial polarization pattern influences the orientation behaviour of the investigated species. Although there is no doubt that the orientation behaviour of some, but not an, investigated birds was influenced by the direction of the transmission axis of linearly polarizing filters, in the light of the findings by Coemans and Vos (1989), Coemans et al. (1990, 1994a) and Vos et al. (1995) it is prudent not to assurne that the response was influenced simply by the plane of polarization (Martin 1991).

After the publication of the conclusion that homing pigeons do not perceive polarization (Coemans and Vos 1989; Coemans et al., 1990, 1994a; Vos et al., 1995), the investigation of polarization sensitivity in birds became a totaHy neglected field of research. As far as we know, between 1996 and 2003 no papers have been published about this subject.1t is edifying to follow the temporal change of trust in avian polarization sensitivity in general:

- In 1990, Helbig (1990, p. 755) wrote: "Despite some early failures to demonstrate the ability of birds to perceive the plane of polarization of linearly polarized light (Montgomery and Heinemann 1952), it has now been firmly established that birds are able to differentiate e-vector directions (Kreithen and Keeton 1974; Delius et al. 1976; Able 1982):'

Note that Montgomery and Heinemann (1952), Kreithen and Keeton (1974) as weH as Delius et al. (1976) studied the polarization sensitivity of one bird species, the homing pigeon (Columba livia), while Able (1982) that of the white-throated sparrow (Zonotrichia albicollis). Hence, when Helbig (1990) wrote the cited words, polarization sensitivity has been "firmly established" only in two avian species, namely in Columba livia and Zonotrichia albicollis among the several hundreds of other known bird species. In spite


of this fact, Helbig concluded that birds generally are able to perceive polarization. On the other hand, as an irony of fate, just in 1990, Coemans et al. (1990) convincingly proved that homing pigeons do not perceive polarization (see below), which was later confirmed by Coemans et al. (1994a) and Vos et al. (1995).

- Five years later, the tables were turned due to the negative results of Coemans and co-workers with the polarization sensitivity of homing pigeons. Then Tovee (1995, p. 457) already wrote: "It seems that birds cannot detect differences in the angle of polarization (Coemans et al. 1994a), and therefore do not use this mechanism for navigation. However, the position of the sun can also be determined from intensity and wavelength gradients across the sky (Coemans et al. 1994a):' In the opinion of Coemans et al. (1994a, p. 107): "we were, however, not able to demonstrate any directional response [of homing pigeons 1 that was caused by the E-vector orientation of the illumination. These results throw doubt on the suggested polarization-sensitivity of birds in general".

However, these opinions are nothing else but poorly established generalizations. From the finding that Columba livia, a non-migrating bird species, cannot perceive polarization (Coemans and Vos 1989; Coemans et al. 1990, 1994a; Vos et al. 1995), one must not conclude that birds in general, including all migrating bird species, are unable to detect the E-vector direction of linearly polarized light. As we have seen above, the polarization sensitivity of certain migrating bird species and its role in their orientation were more or less convincingly demonstrated in several elegant behavioural experiments. These experiments, especially those using depolarizing material, cause many researchers to remain convinced that migratory birds are really using skylight polarization as a compass. However, it would be of great importance to have some solid physiological work to back up this belief.

31.3.1 Debated Polarization Sensitivity in the Homing Pigeon Columba livia

In a conditioning experiment, Montgomery and Heinemann (1952) found three homing pigeons (Columba livia) to be unable to discriminate between orthogonal E-vectors of totally linearly polarized blue light. These pigeons were rewarded with food or unrewarded if they pecked a key illuminated from behind by polarized blue light with a given E-vector direction or an Evector orthogonal to this direction. In the same experimental situation the birds were able to discriminate between blue brightness cues. From this, Montgomery and Heinemann concluded that "if homing pigeons can dis-


criminate at all among patterns of polarized light, they can do so only with extreme difficulty", and that "it is highly unlikely that homing pigeons make use of patterns of polarized sky light as cues in their homing flights". This was a logical conclusion in 1952, since then it was already known that the skylight is most intense in the blue and strongly polarized at 90° from the sun, while the fact that pigeons can also perceive UV light with sensitivity maximum at 370 nm (ehen et al. 1984) was still unknown.

Kreithen and Keeton (1974) reinvestigated the polarization sensitivity of pigeons, since "Montgomery and Heinemann (1952) attempted unsuccessfully to train pigeons to discriminate patterns of polarized light". They found that of 12 homing pigeons tested, only 4 could be trained to discriminate between a linearly polarized light source with a rotating E-vector and the same light source with a stationary E-vector. Initially, all 12 pigeons were trained to discriminate between rotating and non-rotating crosshairs parallel and perpendicular to the transmission axis of the polarizer, wh ich were gradually faded until only polarized light remained. The final polarization tests were performed with the crosshairs removed from the projector. The response was a classically conditioned increase in heart rate preceding an electric shock if the polarizer rotated. In a control experiment, the polarizing filter was replaced by a neutral density glass filter. With this control it could be excluded that the birds responded to the motor noise or relay clicks. As light source, a modified slide projector was used. In front of the projection lens a rotatable linear polarizer was mounted. An aperture, 68 mm in diameter in front of the eye, through very thin material allowed a view of the source, the polarized light of which was projected from behind onto a non-depolarizing re ar projection screen.

Kreithen and Keeton (1974) were not able to answer satisfyingly the question, why only one third of the pigeons tested could be trained to respond to the rotating polarizer. A further problem is that the pigeons could also respond to the rotation of brightness inhomogeneties of the polarizing filter, the presence of which was not convincingly excluded. Furthermore, the rotating E-vector of the polarized light stimulus should have induced a rotating brightness pattern at the edges of the circular aperture due to selective reflection of polarized light. Neither was it convincingly excluded that the birds did not respond to this rotating brightness pattern.

Delius et al. (1976) also conducted aseries of behavioural experiments to test whether homing pigeons are able to orient to the E-vector of polarized light. They used an octagonal box, equipped with four pecking keys, and an overhead polarized light source. The pigeons could learn to peck a key at a specific azimuth angle with respect to the E-vector of the stimulus above them. Delius et al. supplemented their findings with an electrophysiological examination. The stimulus consisted of flashes of totally linearly polarized white light. They obtained a consistent relationship between the shape of the ß-wave of the electroretinogram (ERG) and the orientation of the transmis-


sion axis of the polarizer when the eye was stimulated axially. For a given angle of the E-vector with respect to a stereotactic baseline, the summit of the ß-wave was reported to be sharp and single-peaked, whereas the wave became more flattened, or even double-peaked, when the E-vector was turned by 90°. This effect was described as being unmistakably recognizable in four of the five pigeons tested. In control experiments the possibility was excluded that this change in the shape of the ERG was brought about by weak, unintentional intensity variations. This type of response was even more pronounced when the retina was stimulated ventrally, 40° below the optical axis of the eye. In addition, experiments with broadband colour filters showed that flashes of green or red light yielded ERGs, the shape of which depended on the E-vector direction, although in a less apparent, colour-specific manner. This was not found with broadband blue stimulation. From both behavioural and electrophysiological studies Delius et al. (1976) as well as Delius and Emmerton (1979) concluded that sensitivity in pigeons to the E-vector direction may reside in the ventral part of the retina, the so-called yellow field, 20° below the fovea.

Coemans et al. (1990) retested the polarization sensitivity of homing pigeons both behaviourally and electrophysiologically. In their behavioural experiments they avoided the following pitfalls frequently encountered when working with polarized light (see also Chap. 34):

• Common light sources, e.g. such as slide projectors, produce partially linearly polarized light due to internal reflections from the built-in mirrors, prisms and lenses. Illuminating a rotating polarizer with this polarized light yields a sinusoidal intensity modulation besides a rotating E-vector. Since pigeons are very sensitive to intensity differences (e.g. Hodos et al. 1985), the experiment of Coemans et al. (1990) was set up as a discrimination task between fixed, non-rotating E-vector directions. In the experimental apparatus of Kreithen and Keeton (1974), the light passing through the polarizer was not depolarized in advance, so that rotating the polarizer could have led to the mentioned intensity modulations.

• The correct handling of polarized light reaching the observer after one or more reflections is even more important. When a linearly polarized light is reflected, its amount alters by chan ging the E-vector direction. The intensity of reflected light is minimal or maximal when the E-vector of incident light is parallel or perpendicular to the plane of reflection. To avoid polarization-induced intensity differences of reflected light, the reflective surfaces should be as rough as possible, because such surfaces depolarize the incident light due to diffuse reflection.

• It is often not taken into account that the surface reflectivity plays an important role in determining the remaining polarization-induced intensity differences. Within highly reflective white media, light is reflected many times causing a strong depolarization, while dark media strongly


reduce the intensity of light reflected many times within it. What remains in the latter case is light reflected only a few times which therefore retains a large fraction of its original polarization (Umow 1905). Such dark surfaces can in fact be used as an analyzer, since they transform the E-vector directions of the illumination into brightness variations. Hence, when studying reactions to linearly polarized light, the use of white, depolarizing rough surfaces are necessary to avoid the disturbing reflection-polarizationinduced intensity variations.

In behavioural experiments with pigeons, Coemans et al. (1990) used an experimental chamber covered on the inside with white blotting paper minimising unwanted selective reflections. The stimulus consisted of light passing through a depolarizer, a white diffusing plate and a linear polarizer. Since pigeons produce tallow, the blotting paper could gradually lose its white matt appearance, reintroducing the unwanted E-vector-dependent reflection differences between adjacent walls. Coemans et al. experienced that this process was apparent in the responses of the pigeons: the birds' ability to discriminate between two directions of the transmission axis of the polarizer gradually improved. Renewal of the blotting paper or provision of an extra indirect illumination of the walls abolished this ability. Coemans et al. concluded that the pigeons utilized brightness instead of E-vector differences in these cases. Therefore, the blotting paper was regularly changed. This observation suggests an explanation for the earlier results of Delius et al. (1976). The experimental chamber in that study had matt black walls, wh ich implies that intensity cues must have been present.

Coemans et al. (1990) also examined the variation of the pigeon's electroretinogram in response to flashes of totally linearly polarized light in quite a similar manner as Delius et al. (1976) did. When interference filters were used, a depolarizer was placed in front of the polarizer, which was essential because light is always polarized to some extent after passing such filters. If this precaution is not taken, changing the E-vector direction will vary the light intensity, which in turn influences the ERG, even if pigeons are not sensitive to polarization. The test stimuli (393,477,575 nm) consisted of series of flashes of linearly polarized light, and each test stimulus was followed by an unpolarized reference stimulus. There was no relationship between the ERG characteristics and the E-vector direction.

Coemans et al. (1990) also replicated the experiments of Delius et al. (1976). They modified their experimental setup and stimuli making it as similar to that of Delius et al. as possible. They obtained the same result. The ERG features did not vary systematically with the E-vector direction. The electrophysiological results of Coemans et al. completely contradicted those of Delius et al. Since Delius et al. neither quantified their electrophysiological data nor attempted to test them statistically, the results of Coemans et al. (1990) are more reliable.


In two further behavioural experiments, Coemans et al. (1994a) tested the sensitivity of four mature homing pigeons to the E-vector direction of totally linearly polarized light. The first test was done in a symmetrical Y-maze. In the centre of the maze, the E-vector of an overhead linearly polarized light source matched the E-vector of a linear polarizer on the ceiling of one of the corridors illuminated from above. Pigeons had to choose this matching corridor. The second test was performed in a single-key Skinner box with an overhead source of totally linearly polarized light. Pigeons were rewarded with food when they pecked a key if the polarized light stimulus had a chosen Evector direction. Pecking the key when the E-vector direction was perpendicular to this led to a time-out. In both experiments, in which secondary artifidal cues were rigorously avoided, the pigeons were not able to utilize the information provided by the E-vector direction.

Using the same techniques of stimulation and recording as Delius et al. (1976), Vos et al. (1995) found no evidence for polarization sensitivity in the electroretinogram of 19 adult female and male homing pigeons. They compared the electroretinographical response to flash es of totally linearly polarized light with angles of polarization 0,45,90 and 135° from a reference direction followed by the same set of stimuli in reverse order, directed at the yellow field of the pigeon's retina with that to flashes of unpolarized light. This was carried out for white (360 nm < A < 660 nm) light and monochromatic light of various wavelengths (477,575,632 nm), induding UV (393 nm). A = 393 and 575 nm are dose to the wavelengths of the two local maxima of sensitivity of the ventral retina in pigeons. In addition, responses to slow rotation of the Evector of polarized light with angular velo city of 0.5-4 Hz projected onto the ventral part of the retina, 20° below the area centralis, were recorded for white and monochromatic (494 or 575 nm) light.

In experiments with light flashes, polarized light was produced by a quartz, air-spaced Glan-Thompson polarizer. Because the light leaving the monochromator was partially polarized, rotating the polarizer would give changes in intensity. To prevent this, a calcite depolarizer, suitable for depolarizing monochromatic as well as broadband light, was placed between the monochromator and the polarizer. In experiments with rotating E-vectors, a linearly polarizing filter, located after a depolarizer, was used, the plane of which was exact1y perpendicular to the inddent beam of light. In control experiments, a second static polarizer was inserted into the path of stimulating light, in front of the first one, yielding a stimulus with sinusoidally modulated intensity. Neither the presence or absence of polarization, nor the E-vector direction influenced any of the electrophysiological variables in these experiments. From this one can conclude that the investigated regions of the retina of homing pigeons are not sensitive to polarization. Vos et al. (1995) concluded that "the electrophysiological observations reported by Delius et al. (1976) must have been erroneous. Because the laboratory behavioural experiments that were alleged to have proven the pigeon's ability to perceive the


angle of the E-vector (Kreithen and Keeton 1974; Delius et al. 1976) have not been void of parasitie cues (caused by selective reflection, ... ), and field experiments do not yield direct proof (Phillips and Waldvogel 1988), there is no evidence that pigeons are sensitive to polarized light."

According to Vos et al. (1995), "to incorporate sensitivity to polarized light into avian navigation al models in a meaningful mann er it is imperative to have detailed data concerning the sensitivity of the birds to this stimulus. Physiologieal variables, such as the just discernible angle between two linearly polarized light stimuli or the minimum perceptible degree of polarization, must be determined. Only on the basis of such knowledge is it possible to make quantitative predietions."

The well-designed behavioural and electrophysiologieal experiments of Coemans and Vos (1989), Coemans et al. (1990, 1994a) and Vos et al. (1995) convincingly proved that homing pigeons are not able to discriminate between different E-vector directions and are therefore insensitive to polarization. The earlier reports on polarization perception in pigeons are based on errors in the experimental design. Thus, all interpretations of the reactions of homing pigeons in terms of the alleged sensitivity to skylight polarization are apparently erroneous. Here we mention only four examples:

- Phillips and Waldvogel (1982, p. 197-198) proposed that "pigeons could be averaging the position of the vertical band of maximum polarization at sunrise and sunset to obtain an accurate estimate of the geographie North-South axis", although they admitted that (p. 201) "we do not as yet have direct evidence that polarized light is involved".

- Phillips and Waldvogel (1988) presented "data showing that the seasonal pattern of variation in the initial orientation of short-term deflector loft birds closely parallels changes in the influence that the deflector panels have on polarization patterns visible within the lofts at sunrise and sunset". According to them, these results indieate that the pigeons paid particular attention to the vertieal intersection of the E-vector with the horizon and used this reference to identify geographie north and to calibrate their sun and perhaps magnetic compasses. Furthermore, this implicates a UV-sensitive photoreceptor for polarization by homing pigeons.

- Waldvogel et al. (1988) described measurable changes in the orientation behaviour of short-term deflector loft pigeons, and suggested a correlation between these changes and the presence of a massive upper-atmospherie dust cloud originating from the 1982 eruption of the volcano EI ChiehOn in the Yucatan of Mexieo, whieh significantly altered natural skylight polarization in 1982 and 1983 (Coulson 1983).

- Waldvogel (1990, p. 353) wrote that "In pigeons, only that section of the retina that normally views the sky (called the inferior yellow field) responds to polarization stimuli ... Other areas of the retina, such as the frontal (binocular) field, are insensitive to polarization. Thus, only the rele-


vant portion of the retina is adapted for this highly specialized and perhaps expensive neural task. Interestingly, the inferior yellow field is known to include an abundance of double cones, whieh is another hint that this class of cells may be the polarized-light transducer in birds:'

31.3.1.1 The Position ofthe Sun Hidden by Clouds Could also be Determined on the Basis of the Colour Gradients of Skylight Under Partly Cloudy Conditions

Many animal species possess a time-compensated sun compass in their spatial orientation. When the sun is occluded by clouds, but some patches of the blue sky are visible, some of them may be able to infer the solar position by the distribution of the intensity and colour or the degree and direction of linear polarization of skylight. The major subject of Part III of this volume is how certain animals use the celestial polarization pattern for orientation. However, we would like to emphasize that the UV-sensitive an im als could determine equally reliably the solar position exclusively on the basis of the colour gradients of the sky.

The scattering of sunlight in the atmosphere accounts for the distribution of the degree and direction of polarization, the intensity and the colour of skylight (Coulson 1988). Due to the Rayleigh and multiple scattering, the intensity distribution along the sky for short-wavelength light is far more isotropie and homogeneous than for long-wavelength light. Moreover, the intensity of long-wavelength light increases strongly towards the horizon (this phenomenon is called "horizon brightening") due to the long optieal path-Iengths of sunlight in the atmosphere.

Measuring the distribution of the spectral characteristies of skylight as functions of the solar elevation and the meteorologieal conditions, Coemans et al. (1994b) showed that the colour of light from the clear sky depends only on the angular distance between the sun and the point of observation. Consequently, from the spectral composition of a patch of blue sky the angular distance of the sun from the patch can be determined. They hypothesized that homing pigeons could compare the output of their short-wavelength-sensitive system with the output of their long-wavelength system to process the celestial colour gradients to determine the solar position under partly cloudy conditions. They proposed that the relatively slow UV-sensitive subsystem in the eye of pigeons may not be apart of the system whieh optimally analyses spatial detail in the environment (e.g. for finding seeds), but is instead ideally suited to process wide-angle chromatie information, such as the celestial colour gradients. This may be the reason why pigeons can guess the position of the sun occluded by clouds under partly cloudy conditions even if they are polarization-blind.


31.3.2 European Robin Erithacus ruhecula

The role of skylight polarization in the orientation of nocturnally migrating European robins Erithacus rubecula is also controversial, since on the one hand Sandberg (1988) found that skylight polarization does not affect the migratory orientation of these birds, but on the other hand Helbig (1991a) showed just the contrary. In addition, this example demonstrates wen that comparison of experimental results from various authors is often hindered by crucial differences in experimental design and paradigm. Therefore, great care must be taken when attempting to synthesize existing "evidences".

Behavioural experiments performed by Sandberg (1988) with European robins involving manipulations of the visible sky polarization patterns in the local unaltered geomagnetic field failed to reveal any influence of E-vector direction on the migratory orientation under clear ski es after sunset. These experiments were performed with screening shields restricting the view of the sunset sky overhead to ab out 90° centred at the zenith. The orientation cages were covered by linear polarizers, the transmission axes of which were either parallel or perpendicular to the sunset azimuth. For both E-vector directions, total overcast was simulated by covering the test cages with a sheet of opaque diffusing Plexiglas on top of the polarizers. The diffuser-polarizer combination thereby produced a homogeneous E-vector pattern in an otherwise visually cueless situation. These artificial polarization patterns did not change the orientation of the birds in any way. Sandberg concluded that European robins do not use skylight polarization for migratory orientation.

Helbig (1991a) found that in the absence of meaningful magnetic information due to random fluctuation of the magnetic field at the test site caused by a permanently running air conditioning system, European robins are able to use sun-related cues for dusk orientation. The birds responded to both forward and backward clock-shifts with the expected shifts of the direction of orientation, indicating the use of a time-compensated compass, and the disregard of the sunset position as a fixed directional reference. Their orientation was influenced by linear polarizers with different transmission axes and by depolarization of the natural skylight. In the polarization experiments the orientation cages were covered by linear polarizers with attenuation of 60-65 % in the visible spectrum and alm ost 100 % at 380 nm, and transmission axes were directed either E-W, or NW -SE, or NE-SW. In the simultaneous control experiments the orientation cages were covered with clear Plexiglas. The test birds could see the entire sky and the horizon with a few landmarks. Tests ran from sunset to the first appearance of the brightest stars. In the depolarization experiments the orientation cages were covered by a double-Iayered colourless pseudodepolarizer, which decreased the degree of linear polarization of transmitted light to about 10 %, but did not alter significantly the intensity distribution of skylight. During the spring migratory period, under clear skies with non-manipulated optical cues, the birds ori-


ented towards NE, a seasonally appropriate migratory direction, while under totally overcast skies their orientation was random, as in the case when the light from the dear sky was depolarized. During the autumn migratory period, under dear skies with non-manipulated optical cues, the birds oriented unimodally towards SW, also a seasonally appropriate migratory direction. Under linear polarizers, the birds oriented always approximately parallel to the actual artificial E-vector, and the orientation was unimodal for E-W and NW -SE E-vectors, while bimodal for the NE-SW E-vector.

For the disagreement between the results of Sandberg (1988) and Helbig (1991a), the following major differences in the experimental design have been held responsible by Helbig:

- Sandberg shielded the sky up to 450 elevation from the bird's view, while Helbig gave the birds a full view of the sky.

- The tests of Sandberg were conducted in an undisturbed magnetic field, while those of Helbig in a strongly fluctuating magnetic field. The latter may have forced the birds to rely entirely on visual cues.

- Finally, the results of the overcast and depolarization experiments of Helbig suggest that European robins may derive directional information from the natural celestial polarization pattern at dusk. According to Helbig (1991a, p. 320), "polarizer tests cannot prove this, because they impose a highly artificial stimulus .... The depolarizers, however, take away only the natural polarization pattern without significantly altering the light intensity."

31.3.3 Pied Flycatcher Ficedula hypoleuca

In outdoor orientation cage experiments, Akesson and Bäckman (1999) investigated the autumn migration of juvenile pied flycatchers Ficedula hypoleuca at dusk, from 10 to 70 min after sunset. During some of the experiments, the birds could see the natural sunset glow and the first brightest stars at the end of the test period. The birds were tested under several different experimental conditions in which the geomagnetic field was either natural or vertical and the visual cues available to the birds for orientation were manipulated: (1) natural dear sky with a field of view of 1600 centred at the zenith, (2) simulated total overcast, (3) natural pattern of skylight polarization and stars screened off by opaque diffusing Plexiglas sheets, (4) normal solar azimuth, (5) solar azimuth shifted by 1200 counter-dockwise with mirrors, (6) skylight polarization reduced by an optically dear and colourless pseudodepolarizer, (7) placing linearly polarizing filters onto the top of the orientation cages with transmission axes aligned +45 or -450 relative to geographic north.

The birds oriented towards slightly south of west when they could perceive a combination of natural skylight and geomagnetic cues. Their orientation


direetion was shifted along with the defleetion of the sunset azimuth by mirrors. Redueed skylight polarization had no signifieant effeet on their orientation either in the loeal or in a vertieal magnetie field. The birds tended to orient approximately parallel to the artificial E-veetor, but only when the E-veetor direetion was NW -SE. The orientation under simulated overeast in both the vertieal and the loeal magnetie field was random. Akesson and Bäekman (1999) eoncluded that both visual and magnetie eues seem to be important for young pied flyeatehers during their first autumn orientation.

Unfortunately, from these experiments one eannot establish whether juvenile pied flyeatehers are indeed sensitive to the polarization of skylight and ean use it for orientation, sinee the overhead E-veetor pattern was manipulated awkwardly: the two E-veetor directions (±45° relative to geographie north) were not symmetrie al to the solar meridian at sunset; they were 26° and 57° from the sunset azimuth. The skylight was not depolarized (e.g. by the pseudodepolarizer used in another experiment) before passing through the linear polarizer. Thus, the birds saw differently altered eelestial intensity and eolour patterns through the polarizers in the ease of the two E-veetor direetions. Henee, it was not excluded whether they oriented by means of these eues rather than by the skylight polarization.

Akesson and Bäekman (1999) should have oriented the two E-veetors symmetrieally to the sunset azimuth (rather than symmetrieally to geographie north) in order to ensure that the birds eould see the same eelestial intensity and colour patterns. If the skylight polarization had been important for the birds, then the redueed skylight polarization should have had a signifieant effeet on their orientation either in the loeal or in the vertieal magnetie field. However, the eontrary was observed. The shift of the birds' orientation direeti on by rotating the overhead E-veetor direetion was very weak: rotating the E-veetor by 90° resulted in only a 22° rotation of the orientation direetion. Finally, the used pseudodepolarizer did not eaneel out the skylight polarization pattern eompletely, as Akesson and Bäekman (1999) noted.

In our opinion, from these experiments neither the polarization sensitivity of Ficedula hypoleuca nor the importanee of eelestial polarization in its orientation eould be eoncluded.

31.4 Proposed Mechanisms of Avian Polarization Sensitivity

31.4.1 Is the Foveal Depression in the Avian Retina Responsible for Polarization Sensitivity?

If isolated vertebrate photoreeeptors are illuminated at angles different from their longitudinal axis, they can respond differentially to varying E-veetor direetion oflinearly polarized light (e.g. Sehmidt 1935; Denton 1959). Aeeord-

352

incident lighl opl ie nerve


Fig.31.2. Cross section of the retinal tissue from a bird's eye showing the layered organization of the cells. The deep eleft in the retina is the fovea, a region of high visual acuity due to the high photoreceptor density. At the surface of the fovea incident light is refracted in a pattern that tends to magnify the image projected onto the photoreceptors. (After Waldvogel 1990).

ing to Kreithen and Keeton (1974, p. 89), in the eyes ofbirds (1) it may be possible that light strikes a photoreceptor cell at oblique angles on the sloping walls of the foveal pit (Fig. 31.2), or (2) perhaps light reflected from the opposite slopes of the foveal depression strikes the other side at oblique angles different from the photoreceptors' long axes. In either case, Kreithen and Keeton (1974) hypothesized that the fovea can act as a radial analyser and respond differentially to E-vector directions.

According to Vos et al. (1995), the first hypothesis is ill-founded, because the foveal pit is confined to the celllayers vitreal to the visual cells; in birds, the outer segments of the foveal cones are aligned towards the incoming light. The second hypothesis is highly unlikely, because the foveal depression is relatively shallow, depending on the species, and the difference in refractive index between the vitreous humour and the neurallayer of the retina is slight, so only small amounts of light can be reflected. Using Fresnel formulae and assuming totally linearly polarized incident light, a slope of 30° of the wall of the foveal pit and a refractive index difference of 0.5 between the neurallayer and the vitreous humour, the values of which are much greater than the real ones, Vos et al. (1995) calculated that the reflectivity of the wall of the foveal depression is 1.8 or 0.8 % for E-vectors perpendicular, respectively parallel to the plane of incidence. Because the photoreceptors receive this reflected light in addition to direct light, the intensity difference will cause a contrast modulation as low as 0.05 %. Taking into account that the above-mentioned values are strong overestimates, this mechanism is unlikely to be responsible for avian polarization sensitivity. A further argument against this hypothesis is that the cones in the light-adapted retina of many day-active birds, e.g. pigeons, are surrounded by epithelial pigments, wh ich effectively screen them from light entering at oblique angles. This is particularly true for foveal cones, the outer segments of which are very well protected from stray light in the photopic state (Vos et al., 1995). On the other hand, light reflected from the foveal walls would impair visual acuity.


31.4.2 A Model of Polarization Detection in the Avian Retina with on Droplets

353

At the junction of the inner and outer segments of cone photoreceptors in the retina of all investigated birds and certain fish, amphibians, reptiles and mammals there are one or more oil droplets, in pigeons with typical diameters of 2-4 pm (Muntz 1972). The coloured oil droplets are considered to screen visual pigments (Bowmaker 1977), while the colourless oil droplets to have a light gathering function (Baylor and Fettiplace 1975).

Using the Mie theory of light scattering on the colourless avian retinal oil droplet, Young and Martin (1984) computed the electromagnetic field in the

Fig.31.3. Mechanism of polarization sensitivity of the avian double co ne due to sideways scattering of linearly polarized incident light from the oi! drop let between the inner and outer segments of the principal co ne towards the outer segment of the accessory co ne as proposed by Young and Martin (1984). The photopigments are embedded in an intricately folded membrane that fills the outer segment of the cells. The oi! drop let in the principal co ne lies just below this membrane. The accessory cone usually has no oi! droplet. Here, the double cone of the domestic chicken (Gallus gallus) is represented schematically. (After Waldvogel 1990).

double co ne

out r segment

oil droplet

inner segment

principal cone

incident l ight

outer segment

sideway scattered

light

inner segment

co ne


region dose to the droplets. They showed that these non-absorbing droplets perform significant light collection in cone photoreceptors and thus enhance the photon capture rate of the outer segment. The morphology of avian double cones, particularly the lack of melanin screening pigments between the two outer segments (Mariani and Leure-Dupree 1978) implies that light might pass from one cell to the other. On the basis of this feature, Young and Martin (1984) proposed that scattering by the oil droplet of the principal cone of a double cone pair combined with dichroic absorption for transverse illumination in the accessory cone, may be a potential mechanism mediating avian polarization sensitivity (Fig. 31.3). Light scattered from the principal cone's oil droplet crosses the accessory cone outer segment sideways, and the state of polarization of this light retains that of the incident light. Since the outer segment of vertebrate photoreceptors is dichroic when illuminated sideways (e.g. Harosi and MacNichoI1974), the accessory cone signals could vary accordingly to whether the E-vector of scattered light is perpendicular or parallel to the disc membran es.

32 Human Polarization Sensitivity

In 1844 Wilhelm Karl von Haidinger (1795-1871) an Austrian physicist, geologist and mineralogist discovered that the human eye is able to perceive the linear polarization of light due to an entoptic phenomenon that was later named after hirn. Since the human photoreceptors themselves are insensitive to the E-vector direction of paraxially incident light, human polarization sensitivity seems to be simply a by-product of the dichroic and/or birefringent properties of the ocular media and/or the foveal region of the retina (Figs. 32.1-32.4) without any biological function.

32.1 Haidinger Brushes

According to Haidinger (1844), if one gazes for a few seconds at an unstructured area emitting strongly linearly polarized white light with horizontal Evector, then glances at a white field in wh ich the E-vector is vertical, a faint pattern is seen consisting of two opposing small yellowish brushes with bluish intervening areas (Fig. 32.2). The faint entoptic image called "Hai dinger brushes" are a bow-tie or propeller-shaped pattern subtending about 5° that co-rotates about the fixation point with the rotation of the E-vector. Usually, a little practice is needed to see the Haidinger brushes. The optimal condition to perceive them is when the E-vector rotates slowly, since otherwise, adaptation may cause the brushes to fade. Recording the evoked potentials in response to a rotating E-vector oflinearly polarized blue light in the human eye, Dodt and Kuba (1990) found that the response disappeared when E-vector rotation ceased. Haidinger brushes have the best co nt rast when the degree of linear polarization is 100 %. They can sometimes be observed when one stares at the cloudless blue sky at an angle of 90° from the sun, the light from which is maximally linearly polarized, and then glances at a white surface reflecting unpolarized light.

The exact mechanism of the origin of Haidinger brushes is unclear. Liebman et al. (1974) found that isolated rod outer segments of the human retina have a greater transmission for the perpendicular than for the parallel E-vec-


tor to the membrane discs in the case of off-axis stimulation. Since this transmission difference increases from near zero at 400 nm to a peak at 500 nm and decreases to near zero at 600 nm, this type of spectral change in diehroism cannot account for the yellow-blue Haidinger brush pattern. Gribakin and Govardovskii (1975), who pointed out that only tilted cones would be necessary for retinal polarization effects, explained Haidinger brushes with diehroism in the photoreceptor outer segments. However, their model could not explain the blue and yellow colours of the brushes.

A model for the explanation of the Haidinger brushes suggested by Summers et al. (1970) is based on the assumption that the human retina has the properties of an anisotropie absorbing crystal. Weale (1975) has attributed Haidinger brushes to light scattering in the retina, with photoreceptors detecting light passing laterally through them. He measured the amount of light needed to see Haidinger brushes and found that it follows the cone sensitivity curve. With enough light the Haidinger brushes can be seen at any wavelength within the sensitivity range of cones. He thought that the scattering takes place in the layer of the Henle fibres (Fig. 32.1B). In the fovea there are two groups of fibres emanating radially (Fig. 32.1): (1) Henle fibres are the numerous, closely packed axons extending radiaHy from the co ne photo receptors toward synapses in the outer plexiform layer. (2) Ganglion ceH axons within the nerve fibre layer emerge from the peripheral fovea also with radial

ret inal nerve fi ber layer A

nerve head

Fig. 32.1. A Schematic representation of the human retinal nerve fibre layer. Since all retinal ganglion cell axons converge to the blind spot (optic disk), the fibre pattern is approximately radial and cylindrically symmetrical. (After Dreher et al. 1992). B Schematic drawing of the retinal nerve fibre axon arrangement at the fovea as an examiner would see in the right eye. The cent re of the foveola, marked by an asterix, which is nearly devoid of nerve fibres, includes the central 0.35 mm of the fovea (1.20 of the visual field), located 4 mm temporally and 0.8 mm inferior to the cent re of the optic disko (After Hunter et al. 1999).


Fig. 32.2. A Photo graph of a human retina with the fovea, optic disk (blind spot) and blood vessels. B Computergraphical visualization of the Haidinger brushes as could be seen by a human observer looking at strongly linearly polarized light, the E-vector of which is represented by a double-headed arrow. When the E-vector rotates, the yellow and blue orthogonal brush pattern also co-rotates.

357

symmetry, but then extend toward the optic disk (the "blind spot") in a more complieated arrangement.

Hochheimer (1978), applying linearly polarized light, photographed a Maltese cross-shaped intensity pattern in the rhesus monkey eye. He attributed this cross to the same structure that pro duces Haidinger brushes in the human eye. In contrast to the Haidinger brushes, however, this cross appeared not only with short-wavelength light, but also on photographs taken with red light where the macular pigment has zero absorption. He concluded that the cross, and therefore also Haidinger brushes, could not be related to diehroism of the macular pigment. Shute (1978) suggested that the predominant orientation of collagen in corneal stroma is responsible for this cross pattern.

Several authors (e.g. Naylor and Stanworth 1954) have attributed Haidinger brushes to oriented diehroic elements in the macula. Dimmer (1894) dem onstrated that the Henle layer may be dichroie and it is radially symmetrie over the macular area. The Henle nerve fibre layer stretches over the entire macula lutea and contains the axons of the foveal cones (Fig. 32.1). Radial orientation of these fibres causes form birefringence, and the lutein pigment molecules


attached to the fibre framework have intrinsie birefringence (Bone and Landrum 1984). De Vries et al. (1953) proposed that Haidinger brushes are due to a fraction of the symmetrie macular pigment molecules. Lutein, whieh is probably the pigment involved, has a long chain molecule whieh will absorb strongly if the incident light is polarized parallel to the molecule axis. To account for the Haidinger brushes, a fraction of the otherwise randomly aligned molecules must be arranged tangentially with respect to the foveal cent re (Fig. 32.3). An alternative model proposed by Hemenger (1982) does not require the directional organization of the macular pigment molecules, but rather that of the medium in whieh they are located. This is the Henle fibre layer, and the molecules are assumed to be randomly oriented between the fibres.

In order to reveal the role of the macular pigment in the detection of polarization, Bone (1980) measured the diehroism of this pigment and the birefringence of the human cornea. He also suggested that the long molecular axes of the lutein pigment may be aligned tangentially with respect to concentrie circles about the foveal centre (Fig. 32.3). He showed that when the corneal birefringence is taken into consideration, the results are consistent with a symmetrie al arrangement of a fraction of the macular pigment molecules. The diehroism responsible for Haidinger brushes has spectral characteristies virtually indistinguishable from those of the macular pigment. Apparent variations in the diehroism with orientation of the E-vector of incident light follow a trend to be expected as a result of birefringence of the cornea. A comparison of the strength of the diehroism and the macular pigment density revealed that the number of preferentially oriented pigment molecules per unit area of the retina did not vary much among them. Tests indicated that all ob servers possessed similar areal densities of such symmetrically arranged molecules. The macular pigment is found between the outer and inner limiting membranes of the retina, a region containing the radially

m~~ I I I I I I I I I I foveola I I I I I I I I I I

~~I~~ Fig.32.3. Schematic representation of the hypothesized preferential arrangement of the macular pigment molecules bounded partially to the membrane of the radially oriented Henle fibres in the human fovea. According to different models, Haidinger brushes are attributed either to this ordered tangential arrangement of the pigment molecules, or to the radially arranged Henle fibres. (After Bone 1980).

32 Human Polarization Sensitivity 359

arranged nerve fibres.1t is conceivable that these fibres may be responsible for producing the alignment of the pigment molecules, in much the same way as a stretched sheet of polyvinyl alcohol acts as the aligning matrix for polymerie iodine in the Polaroid polarizer (Land 1951). If the molecules are membrane-bound to the fibres, only a fraction of them is aligned. The remaining molecules could freely rotate with random orientation between the inner and outer limiting membranes.

The axons of the foveal cones, constituting the Henle fibre layer, are arranged in a fan-shaped pattern. The regular makeup of pro tein fibres in the surrounding cytoplasm of Müller cells with mutually different refractive indiees will cause uniaxial form birefringence. The optieal axis is directed along the fibres and represents the slow axis, and so the fibre layer shows positive birefringence. Bone and Landrum (1984) suggested a regular alignment of the lutein pigment molecules, presumed to be the macular pigment, attached to the Henle fibre framework in order to explain the Haidinger brushes, and demonstrated the linear diehroism of these molecules. They also assumed that the fraction of the diehroie lutein molecules attached to the Henle fibre membrane are perpendieularly aligned to the radially oriented fibres, and thus the molecules are tangentially oriented along concentrie cirdes centred at the fovea (Fig. 32.3). They demonstrated also that the model suggested by Hemenger (1982) is inconsistent with the spectroscopie properties of the macular pigment.

Haidinger brushes became the basis for a technique of several ophthalmologists to subjectively study and diagnose a variety of diseases and disorders affecting various segments of the human fundus oculi, especially the macula and its surrounding structure (e.g. Sloan and Naquin 1955). In these studies, the tests were based on the patient's ability or inability to visualize Haidinger brushes. Later, an objective photographic test applying "retinal polarization cross or brush patterns", replaced the subjective one (e.g. Chisholm 1975).

Taking photographs of the retina with crossed polarizers in the stimulating and recording light paths, a cross or brush pattern of 4-50 diameter centred at the fovea can be observed overlying the macula (where cone density is maximal) depending on the orientations of the transmission axes of the polarizers. This retinal polarization-induced pattern can be seen between 400 and 700 nm, but cannot be observed at wavelengths longer than 700 nm (Hochheimer and Kues 1982). Delori et al. (1979) conduded that macular birefringent structures alone produce the cross pattern, and when the birefringence of the cornea is considered, a brush pattern is obtained. The cross occurs when the polarizing axis is aligned along either the slow axis in the inferior nasal direction or fast axis perpendieular to the slow axis of the cornea. Experiments with dark- and light-adapted monkeys and normal and colour-blind human subjects revealed that the retinal cross or brush pattern is due to the birefringence in cone photoreceptor outer segments. This pattern occupies the same retinal area where Haidinger brushes are seen, and


both effects may be due to the cone outer segments, as suggested by Hochheimer and Kues (1982).

According to Hochheimer and Kues (1982), if both the dichroism and the dispersion of the birefringence of these outer segments are taken into account, a simple explanation can be given for the visibility of Haidinger brushes. The combined intrinsic (pigment) and form (layered structure) birefringence in the outer segments has dispersion (Liebman et al. 1974). From 400 to 500 nm the birefringence is stronger than average, while from 500 to 600 nm it is less. If light penetrates laterally through the photoreceptor, it will be differentially retarded. This differential retardation, together with the dichroism of the outer segments, will account for both the yellow and blue segments of the Haidinger brushes. The birefringence of the cornea plays only a minor role. Helmholtz (1866) stated that, when viewing Haidinger brushes, the dark horizontal brushes are much narrower at their centre than the vertical dark brushes. This may be due to the birefringence of the cornea. The absence of a photographically visible polarization pattern is an indication of macular dysfunction due to senile macula degeneration, angioid streaks or diabetic retinopathy, and thus the phenomenon can be useful for diagnosing diseases affecting the macula.

Hochheimer and Kues (1982) tested several healthy human subjects and several patients with diseased retinae whether they could see Haidinger brushes. A small darkroom safelight was fitted with a blue filter and a slowly rotating linear polarizer. The subjects were asked whether they could see a pattern, and if yes, they had to describe its appearance and indicate the darkblue brush orientation as the pattern slowly rotated. All those who could see Haidinger brushes had an easily photographically discernible retinal polarization pattern, and those who could not see them did not display any retinal polarization pattern. Hence, the perception of Haidinger brushes is normal and common in humans rather than an exceptional phenomenon. Perception of Haidinger brushes may indicate a healthy eye and the inability of percepti on of these brushes indicates certain visual dysfunctions.

In sum, we can conclude that the origin of Haidinger brushes is a mechanism, which involves a radial and/or tangential arrangement of dichroic elements with the fovea as the centre of cylindrical symmetry. Then the simplest physical model of the Haidinger brushes would be a special linear polarizer, in which the transmission axes are either tangential along concentric circles (Fig. 32.4A) or radial from a centre (Fig. 32.4B). Looking at a linearly polarized light field with homogeneous E-vector direction and degree of linear polarization (Fig. 32.4C) through such filters, we could see a cross pattern in which two bright arms are perpendicular to two dark arms (Figs. 32.4D,E). The resulting brightness pattern is very similar to the Haidinger brushes. If the spectral characteristics of the polarizer and the light field depend appropriately on colour, the bright arms could appear in yellow while the dark arms in blue, which would be a better model of the Haidinger brushes.


Fig. 32.4. Two simple physical models, which simulate the Haidinger brushes. Looking through a linear polarizer in which the transmission axes are either tangential to concentric circles (A, corresponding to the tangential arrangement of macular pigment molecules) or radial from a centre (B, corresponding to the radial arrangement of Henle fibres) at a linearly polarized light field with homogeneous E-vector direction and degree of linear polarization (C), we could see a cross pattern, in which two bright arms are perpendicular to two dark arms (D, E).

32.2 Boehm Brushes

361

tangential pol-filter radial pol-filter (tangential macular (radial Henle fibers) pigment molccules)

There is another entoptic phenomenon, the so-called Boehm brushes, wh ich are analogous to the Haidinger brushes. The Boehm brushes are parafoveal entoptic images visible to humans when a rapidly rotating E-vector of linearly polarized light is observed. A small bright linearly polarized source fixated about 15-20° parafoveally with its E-vector rotating at about 1 Hz is optimal to elicit this phenomenon, wh ich was discovered and described by Boehm (1940). Boehm brushes are believed to depend on differential scattering mainly in the retina (Boehm 1940; Vos and Bouman 1964). Such differential scattering should be strongest near the axis of the beam scattered and decrease sharply with increasing angular distance from the axis.

32.3 Shurcliff Brushes

Shurcliff (1955) reinvestigated the phenomenon of Haidinger brushes and found that circularly polarized light can also produce similar brushes such that an observer can tell the handedness of circular polarization. Righthanded circularly polarized light produces brushes for most observers. He presumed that linear dichroism occurs in the yellow pigmented macula of the eye, bringing about birefringence effects in the refracting cornea and lens.

33 Polarization-Induced False Colours

33.1 Polarization-Dependent Colour Sensitivity and Colour-Dependent Polarization Sensitivity

Glas (1975) hypothesized first that false colours could be induced in an insect visual system by polarized light. He proposed that the honeybee Apis mellifera may perceive the polarization of skylight not as a distinct entity, but rather as "polarizational false colours". In his model, all Uv, blue and green receptors are involved in the perception of polarization. Since, according to the model, the UV and blue receptors are polarization-sensitive, their output signals depend on the degree p and angle a of linear polarization. Consequently, the model retina perceives "polarization-induced false colours".

However, in the honeybee Apis melIifera the polarization-sensitive UV receptors detecting skylight polarization are gathered in an upward-pointing narrow dorsal rim area (DRA) of the eye. Except for this specialized eye region, the retina of honeybees is composed of photoreceptors that are twisted about their longitudinal axes (Wehner et al. 1975), so that their polarization sensitivity (PS) is almost abolished (Labhart 1980). Similar twisted photoreceptors, or receptors in which the microvilli of the rhabdomeres are not aligned consistently in a single particular direction, were found also in many other insects (see Chap. 16.7.5). In these receptors the PS is weak, whereas the untwisted receptors located at the dorsal rim of the eye and used exclusively for detecting skylight polarization exhibit high PS.

Wehner and Bernard (1993) proposed that the functional significance of the photoreceptor twist is to avoid the polarization-induced false colours of natural surfaces, such as leaves and petals of flowers, which reflect partially linearly polarized light. p and a of reflected light depend on how smooth the plant surfaces are and how they are oriented relative to the incoming light at the direction of view. For a flower-visitor this could cause difficulties, because the absorbing photopigments responsible for colour vision are contained in receptors with different microvillar orientations. Thus, each receptor gives a signal that depends not only on intensity and wavelength but also on p and a. If the sensors of a colour vision system are also polarization sensitive, the system generates false colours that may obscure the real colours defined by the spectral properties of the object.

33 Polarization-Induced False Colours 363

As Wehner and Bernard (1993) pointed out: " ... when zig-zagging over a meadow with all its differently inclined surfaces ofleaves, the bee would experience pointillistic fireworks of false colors that would make it difficult to impossible to detect the real colors of the flowers". The twist of photoreceptors of the colour vision system eliminates the ability to respond selectively to the plane of polarization. This allows each type of receptor to take part unambiguously in the bee's trichromatic colour vision system. To demonstrate the false colour problem, Wehner and Bernard (1993) computed the shift of perceived colour caused in the bee's colour triangle when it views reflections from a dan deli on leaf at different angles.

Marshall (1988) suggested that each of rows 5 and 6 of the midband in the compound eyes of mantis shrimps may contain aseparate three-channel polarization-analyzer system with output comparisons: in row 5 between retinula cells R8, RI-R2-R5-R6 and R3-R4-R7; in row 6 between R8, RI-R2-R5 and R3-R4-R6-R7, which receptor trip lets have different microvilli directions. Since the spectral sensitivities of R8 and Rl-R7 cells are maximal at 350 and 500 nm, respectively (Cronin and Marshall 2001), in both types of the three-channel polarization-sensitive system polarization-induced false colours would be inevitably generated. However, to prevent such confusion between polarization and spectral information, in the ommatidia specialized for colour vision the rhabdoms are polarization insensitive due to randomly oriented microvilli (Marshall et al. 1991b).

Mammals represent an extreme case in the solution of the problem of polarization-induced false colours: they eliminate polarizational false colours in such a way that they have colour vision, but are polarization-blind. Certain cephalopod (squid, cuttlefish and octopus) species are the other extremum: they are colour-blind, but polarization-sensitive (Hanlon and Messenger 1996). Hypotheses about salmon PS also predict an interaction between colour and polarization (Novales Flamarique et al. 1998). UV, green and red cones appear to contribute to both spectral and polarization sensitivity in several salmonid fish (Parkyn and Hawryshyn 2000). Thus, these fish mayaiso perceive polarization-induced false colours. The counterpart of polarizationdependent colour vision, namely colour-dependent PS has been found in Daphnia pulex (Novales Flamarique and Browman 2000).

Horvath et al. (2002 c) developed a quantitative model to calculate correctly such polarizational false colours with the use of polarization patterns measured by imaging polarimetry. In this chapter some results obtained with this model are presented.


33.2 Polarizational False Colours Perceived by Papilio Butterflies

Kelber (l999b) and Kelber et al. (2001) suggested that the butterflies Papilio aegeus and Papilio xuthus might not process polarization and colour separately, and thus they may perceive polarization-induced false colours due to their weakly polarization-sensitive photoreceptors (see Chap. 17.5.2). Since Kelber and collaborators worked with artificial stimuli having an unnaturally high p = 100 % which is not characteristic for light reflected from plant surfaces, no published behavioural data so far support that there is a significant influence of polarization on butterfly colour vision under natural conditions, when the receptors are stimulated by partially linearly polarized light with frequently low p. Since the PS of photoreceptors in Papilio species, ranging between 1.3 and 2 (Bandai et al. 1992; Kelber et al. 2001), is very low1, the questions arise:

• Can the often low p of light reflected from plant surfaces induce sufficiently strong polarizational false colours in Papilio butterflies to influence their colour vision significantly?

• How do these polarization-induced false colours depend on the different parameters of the butterfly retina (microvillar directions, PS, orientation of the eye), on the characteristics of the optical stimuli (p and a of reflected light) and on the illumination conditions (alignment of the plant surface relative to the direction of view and to the solar direction; plant surface in direct sunshine or in shadow)?

33.2.1 Computation of the Spectral Lod of Colours Perceived by a Polarization- and Colour-Sensitive Retina

The numerical values of the retina model of Horvath et al. (2002 c; Figs. 33.1A,B) are characteristic to the butterfly Papilio xuthus (Kelber et al. 2001). The model retina contains polarization-sensitive photoreceptors of spectral types red (R), green (G) and blue (B), with sensitivity maxima at wavelengths AR' = 600 nm, Ac' = 520 nm and AB' = 460 nm. In the retina model, angle ß is the direction of the microvilli clockwise from the dorso-ventral meridian of the compound eye (Fig. 33.1 C). For the microvilli of the blue photoreceptors ßB = 0°, in the green receptors ßc = 0,35,90, 1450 and in the red

I Note that PS = 1 for polarization-insensitive receptors, and PS = 1.3-2 are such low values that many researchers consider a photoreceptor with such PS as polarization insensitive. Laughlin (1976, p. 227), for example, wrote about the polarization-insensitive receptors of the dragonfly Hemicordulia tau: ''Alllinked pigment ceIls and the single pigment green ceIls are notable for their lack of PS « 2.5) at peak wavelength:'


A

400 500 : 61)0

1..;=460 ",":=520 1...'=600 "ll\'clcnglh '" (nm)

c

polarization ellipse

B rnicrovilti oricnlulions of Papilio bUlIcrny pholoreccplOrs

rcd rcccplor gn .. "Cn rl.~CepIOr bluc rcccplor

I ß. = O' I ßa = O'

/ 13, = 35' / ß~ = 35'

-- ßG= 90'

\ 13.= 145' \ 13,,= 145'

I 13, = O'

D inpul for ~IC model rClina blue: red

Co

!l,.;H fl A..' ) - - - - - - -/r---'~--i ~ B ~ , flA)

" &. - flA,) - - - - -~'o~ flA.')I-~~-I ~t~ o~ ~

I; 0

365

). 00 ),,- 500 , ,-600 '.- 00

E

1..,'=450 A..'=550 1..,'=650 ~\·.a.~elength h (tim)

G

Iriangle

cqui latcral colour triangle

B~-------G--b-a-sc~------~R

Fig, 33.1. A Relative absorption functions of the blue, green and red photoreceptors of Papilio xuthus. B Microvilli orientations ß clockwise from the eye's dorso-ventral meridian in the photoreceptors of different spectral types (red, green, blue) in the butterfly Papilio xuthus. C Definition of the different parameters of partially linearly polarized light and a polarization-sensitive photoreceptor. The direction of hatching indicates the microvilli orientation ß. The angle of the eye's dorso-ventral meridian is X clockwise from the vertical. a is the angle of polarization of light clockwise from the vertical. The arrows represent the maximum (Ernax ) and minimum (Ernin) of the electric field vector (the major and minor axes of the polarization ellipse) and their components that are parallel (EPar rnin' EPar rnax) or perpendicular (EPerp rnin' EPerp rnax) to the microvilli. D Replacement ofthe blue (400-500 nm), green (500-600 nm) and red (600-700 nm) parts offunctionf(,l) (f = I (intensity), or f = p (degree oflinear polarization), or f = a (angle of polarization)] by discrete constant values f(,l,c) (r=B,G,R) measured by video polarimetry at wavelengths ,l,c. E Position of a visual stimulus C with spectral components MR, MG and MB within the equilateral colour triangle (the center of which is marked by +) of a colour-sensitive visual system with photoreceptor types R, G and B. (After Horvath et al. 2002c).


receptors ßR = 0,35,1450 (Fig. 33.1B). The colour vision system of Papilio butterflies is pentachromatic (Arikawa et al. 1987). Treating the short-wavelength receptors (UV, violet, blue) as one receptor type, allows us to demonstrate false colour effects in a plausible way by indicating the shifts of colour Iod in the equilateral colour triangle (Fig. 33.1E). No prindpally different false colour effects are expected by including all five receptor types in the retina model.

If the electric field vector E of totally linearly polarized inddent light is parallel (par) to the longitudinal axes of the microvilli, a polarization-sensitive photoreceptor of type r (=R,G,B) absorbs Pr-times the number of photons as when the E-vector is perpendicular (perp) to the microvilli. Thus, the relationship between the numbers of absorbed quanta is: q,rar = Prq;erp, where Pr is the polarization sensitivity ("PS-value") of the receptor, and qr is the quantum absorption. Pr of the photoreceptors in Papilio xuthus ranges from l.3 to 2 at peak wavelengths (Kelber et al. 2001). In the retina model PB = P G = PR = 2 are chosen, by which the average PS is slightly overestimated.

Let the angle of the eye's dorso-ventral meridian be X clockwise from the vertical (Fig. 33.1C). If receptor r receives partially linearly polarized light with intensity [(A), degree of linear polarization p(A), angle of polarization arA), minimum and maximum E-vectors Emin(A) and EmaiA), the quantum absorption qr can be calculated as follows:

q = k frx A (A) [P Epar (A)2 + Eperp (A)2 + P Epar . (A)2 + Eperp . (A)2]dA r 0 r r max max r mm mm ,

(33.1)

where k is a constant, A is the wavelength of light, A,(A) is the relative absorption of the receptor (Fig. 33.1A), EPar maiA), EPerp maiA) and EPar min(A), EPerp miiA) are the parallel and perpendicular components of the electric field vectors EmaiA) and Emin(A) with respect to the microvillar direction. From Fig. 33.1C one can read:

Eparmax(A) = Emax(A)COS[a(A)-X-ßr]' Eperp max(A) = Emax(A)sin[a(A)-X-ßrJ,

Eparmin(A) = -Emin(A)sin[a(A)-X-ßr]' Eperp min(A) = Emin(A)COS[a(A)-X-ßr]'

(33.2)

The relationship between ß2 miiA) , ß2 maiA) and p(A) is:

(33.3)

[(A) can be expressed with EmiiA) and EmaiA) as follows:

I(A) = k' [E2max(A) + E2min(A)]/2 = k' E2 max(A)/[l+p(A)], k' = constant. (33.4)


Using Eqns. (33.1)-(33.4), one can obtain:

r = R, G, B; k" = constant. (33.5)

Constants k, k' and k" involve different electrodynamical constants. We omit to give their expressions, because they all are eliminated in the final expressions describing the spectralloci of colours perceived by a polarization- and colour-sensitive retina. Since with video polarimetry (Horv<ith et al. 2002 c) could measure the spatial distribution of I, p and a of light reflected from plant surfaces only at wavelengths ABc = 450 nm, AGc = 550 nm and ARc = 650 nm, in the calculations the following approximations are taken (Fig.33.1D):

f(400 nm ~ A ~ 500 nm) = f(ABe) = fb1ue'

f(500 nm < A < 600 nm) = f(AGe) = fgreen'

f(600nm ~ A ~ 700 nm) = f(ARe) = fred, f = I, p, a, (33.6)

i.e. in the spectral range s = red, green, blue I, p and aare considered to be constant. This approximation can be applied, because the maxima and half bandwidths of the red, green and blue sensitivity functions of the camera of the imaging polarimeter fall dose to those of the corresponding red, green and blue absorption functions A(A) (Fig. 33.1A) of the butterfly retina modelled. Then:

r = R, G, B; s = red, green, blue;

Ab! = 400 nm, Ab2 = Ag! = 500 nm, Ag2 = Ar! = 600 nm, Ar2 = 700 nm.

(33.7)

In the literature on colour vision, there are two different conventions to give the relative absorption functions A(A) of photoreceptors: they possess either equal amplitude AmaiA) = 1, or equal integrals l A(A)dA = 1. The first convention is called "amplitude normalization" (Fig. 33.1A). The second convention, called "integral normalization", corresponds to the assumption that the quantum absorptions of receptors of different spectral types are the same if the incident light is unpolarized [p(A) = 0] and physically white [I(A) = constant]. This has the consequence that "physical (or optical) white" coincides with "physiological (or perceptional) white", in other words, the locus of both physical and physiological white is positioned at the colourless centre of the equilateral colour tri angle (Fig. 33.1E). In this case the receptor absorption


curves are normalized by setting their integral to 1, i.e. the quantum absorption qr of receptor type r is divided by the quantum absorption

(33.8)

of the receptor for unpolarized (Ps = 0) and physically white light (Is = Iwhite = arbitrary constant). Then, the normalized quantum absorption is:

mr = q/qrwhite =

{k"Ls Is [P/I +p.)+ I-ps -2Ps(Pr -1)sin2(as-X-ßr)] ASJAS2 Al\')dA}/

/[k" Iwhite(Pr + 1) Ls ASJAS2 A/A)dA]. (33.9)

The three coordinates of the spectrallocus of the perceived colour within an equilateral colour triangle (Fig. 33.IE) are

for amplitude normalization, and

(33.11)

for integral normalization. Note that the constants k" and I white are eliminated from the expressions of M R' MG and MB' as mentioned above. The calculations were performed for both amplitude and integral normalizations, but both conventions provided very similar results. The only significant difference between them is that for integral normalization the colour loci remain elose to the white point (centre of the colour tri angle ), i.e. the colours are extremely pale, while for amplitude normalization all colour loci slightly shift towards the red-green border of the colour triangle. The reason for the latter shift is that the integral of AlA) is the greatest among the integrals of the absorption curves of the red, green and blue receptors (see Fig. 33.IA). Hence, when amplitude normalization is used, the quantum absorption qG of the green receptors is the largest resulting in that the component MG will be the greatest. If integral normalization is used, the relative differences in the quantum absorptions qR' qG and qB of the R, G and B receptors are reduced, which decreases the colour saturation. In this chapter only the results obtained with the more common integral normalization are presented, which puts white in the intuitively correct location in the middle of the colour triangle. The values of Is'Ps and as originate from the reflection-polarization patterns measured by imaging polarimetry in the 5 = red, green, blue ranges of the spectrum.


33.2.2 Polarization-Induced False Colours Perceived by a Weakly Polarization-Sensitive Retina

369

~ Colour Fig. 33.2 shows the reflection-polarizational characteristics of the red flower petals and green leaves of Campsis radicans (trumpet vine, Bigniniaceae). In ~ colour Fig. 33.3A the real colours of Campsis radicans are shown as perceived by a polarization-blind retina. ~ Colour Fig. 33.3B-E shows the false colours of the plant perceived by the weakly polarization-sensitive retina of Papilio xuthus as a function of the alignment X of the dorso-ventral symmetry plane of the eye with respect to the vertical, when a given set of photoreceptors rotates in front of the plant. Rotating the polarization-sensitive receptor set by 180°, the perceived false colours shift continuously in the colour triangle: in cases B, C, D and E of ~ colour Fig. 33.3 the false colour of the leaves is slightly blue-greenish, bluish, reddish and greenish, respectively. These colours are, however, more or less masked by the whitish reflected light (see ~ colour Fig. 33.2A). Similar shifts of the perceived colour occur if the relative position of the plant surface with respect to the receptor set changes because of rotation and/or translation. In the case of Papilio xuthus, the chromatic distances of the polarization-induced false colours from the real colour are short due to the relatively low PS = 2 of the retina. These chromatic distances are shorter for the matt petals reflecting light with lower p than for the shiny leaves reflecting light with much high er p. ~ Colour Fig. 33.3 also demonstrates how the real and the polarization

induced false colours of leaves depend on the orientation of leaf blades. Although the average alignment of leaf bl ades is approximately horizontal, there are considerable deviations from this direction (see ~ colour Fig. 33.2D; the E-vector alignment of specularly reflected light is always perpendicular to the plane of reflection). The more or less randomly curved leaf bl ades are more or less randomly oriented around the horizontal direction, thus both p and a change from site to site. The consequence is that the homogeneously green real colour of the leaves being independent of p and a (see the narrow colour distribution around the most frequent real green colour ofleaves in the right colour triangle of ~ colour Fig. 33.3A of the leaves) becomes more heterogeneous for a polarization-sensitive retina resulting in different colour hues ranging from (although partlywhite-masked) violet through blue,green, yellow, orange to red (see the relatively wide false colour distribution around the most frequent green false colour of leaves in the colour triangles of ~ colour Fig. 33.3B-E). This shows one of the consequences of the PS of colour vision: due to the high variation of p and a of light reflected from plant surfaces, the perceived polarizational false colours are more diverse than the real colours. This phenomenon makes it more difficult to recognize a given real colour and demonstrates a disadvantage of the perception of polarizationinduced false colours.


Figure 33.4D shows how the polarization-induced false colours of an Epipremnum aureum plant (golden pothos, Aracea, having a large petal-imitating shiny red leaf called spathe) perceived by Papilio xuthus depend on the microvilli directions ßR and ßG' The false colours are scattered within areas, the dimensions of wh ich are similar for both the spathe and the leaf, because both are shiny and reflect strongly polarized light (Fig. 33.4B).

Figure 33.5 demonstrates the dependence of the polarization-induced false colours on the polarization sensitivity PB = P G = PR = P as a function of ßG and ßR' When P increases from 1 to 20, all false colours shift to some degree from the real unsaturated, bluish-green colour (locus a) of the leaf towards relatively saturated red, orange, yellow or green colours (lod b-m). The chromatic distance of the false colours from the real colour can be considerable if the PS is strong enough.

Figure 33.6 shows the dependence of the polarization-induced false colours on p of reflected light. This dependence is qualitatively the same as that on the PS of the photoreceptors (Fig. 33.5). The only essential quantitative difference between Figs. 33.5 and 33.6 is that in the latter case the chromatic shifts (the lengths of the arrows) are much sm aller than in the former case in spite of the very high p-values of 78, 75 and 99 % in the red, green and blue spectral ranges. ~ Colour Fig. 33.7 displays how the spectral and reflection-polarizational

characteristics of a sunlit leaf of a Ficus benjamina tree (weeping fig, Ficaceae) depend on the direction of the sunlight at a given solar elevation, and how they change if the leaf is shaded from direct sunlight. The colour of the sunlit leaf is always greenish (~ colour Fig. 33.7 A,C,E,G) due to the diffuse scattering and selective absorption of white sunlight in the green subcuticular leaf tissue. This greenish hue is, however, more or less masked by strong specular reflection of white sunlight, if the leaf is viewed in the direction of the sun (~colour Fig. 33.7G). The colour of the shaded leaf (~colour Fig. 33.7B,D,F,H) is always bluish, because it is illuminated by blue skylight. Due to the non-planar, curved shape of the leaf blade, p and a of reflected light change from point to point.

In Fig. 33.8 we can see that under the clear blue sky the hues of shaded leaves are always nearer to the blue-green parts of the colour tri angle than those of sunlit leaves. In the left window of the leaf in ~ colour Fig. 33.7 the false colour shifts (represented by arrows in Fig. 33.8) towards red, orange, yellow or green hues for both shaded and sunlit leaves. Since in the right window of the leaf in ~ colour Fig. 33.7 the orientation of the leaf blade is different (vertical) from that (horizontal) in the left window, the colour shifts in the right window differ from those in the left window. Apart from case E, in the right window the false colours shift toward the green hues for both shaded and sunlit leaves. In case E the colour shift is very small. Note that although in ~ colour Fig. 33.7G the entire leaf is lit by direct sunshine, both the left and right windows are placed in a local shaded region due to the


total rad iance 1 degree ofpolarization p

leur spalhc w indow window

b-rn: polarization-scnsi tive retina p. = p,. = p" = 2; X = 0°, ß. = 0° ß(i

a

e e

b

O· 35· 90· 145·

ßR { 3~: b c d e f g h

145· k 1 m

a: polarization-blind ret ina p. = p,. = p" = I

ß., ß,., ß. = arbitrary

- - -

Blue

angle of polarization <X

from the vertical

a

D

371

Red

Fig. 33.4. A-C Patterns of the total radiance I, degree p and angle a oflinear polarization of Epipremnum aureum (golden pothos, Aracea) measured by video polarimetry at 450 nm. D Colours (MR> MG> MB) of E. aureum perceived by a polarization-blind retina with PB = P G = PR = 1, and ßR' ßG> ßB = arbitrary (a), and by a polarization-sensitive retina with PB = P G = PR = 2, X = 0°, ßB = 0° as a function of the microvillar directions ßG and ßR of the green and red receptors (b-m). Every microvilli situation is designated by a lower case letter ranging from a to m. The corresponding spectralloci (designated by letters a-m) of two details of the picture, one on a leaf blade (white) and one on the spathe (black) marked by rectangular windows in patterns A-C, are plotted within the equilateral R-G-B colour triangle, the colourless cent re of which is represented by +. (After Horvath et al. 2002 c).

372

BIue

a: polarization-blindness PR=PG=PB=P=!


Green

polarization sensitivity X=O', ßB=O'

0'

ßR { 35' 145'

0' 35' 90' 145'

b c d e f g h

k 1 m

PR=PG=PB=P=!

P=2 ! P=5

P=lO

P=20

Red

Fig.33.5. Dependence of the polarization-induced false colour (MR, MG' MB) perceived bya retina with X = 0°, ßB = 0° on the polarization sensitivity PB = P G = PR = Pas a function of the microvillar directions ßG and ßR of the green and red receptors (designated by lower case letters b-m) plotted within the equiIateral R-G-B colour triangle, the colourless centre of which is represented by +. The colours are calculated for a typical point on a leaf of Campsis radicans. The arrows start from the spectrallocus a of the real colour when PB = PG = PR = P = 1, meaning polarization-blindness, while the arrowheads point to the spectrallocus of perceived false colours if PB = P G = PR = P = 20. The spectralloci of false colours for P ranging between 1 and 20 are placed along the straight arrows, on which the loci for P = 2,5 and 10 are marked by bars. (After HOfV<ith et al. 2002 c).


a: polarization-blindness PR=PG=PB=1

polarization sensitivity X=O°, ßB=O°

P R=PG=PB=2

almost totally polarized light n = 1.28

1 unpolarized n = 0

light

p(R,G,B) = n· Po(R,G,B)

ßR { 3~: 1450

Green

----~

373

ßG -------00 350 90° 1450

b c d e f g h

k I m

Fig. 33.6. Dependence of the polarization-induced false colour (MR, MG> MB) perceived bya polarization-sensitive retina with PB = P G = PR = 2, X = 0°, PB = 0° on the degree of linear polarization p(R,G,B} of reflected light as a function of the microvillar directions PG and PR of the green and red receptors (designated by lower case letters b-m) plotted within the equilateral R-G-B colour triangle, the colourless centre of which is represented by +. The colours are calculated for a typical point of a leaf of Campsis radicans. The degrees of linear polarization of reflected light are calculated as p(R,G,B} = n. Po(R,G,B}, where Po is the original degree oflinear polarization and n is an arbitrary factor. The arrows start from the spectrallocus a of the real colour when n = 0 (unpolarized light) and PB = PG = PR = P = 1 (polarization-blindness), while the arrowheads point to the spectrallocus of perceived false colours for n = 1.28 (almost totally polarized light in all three spectral ranges). The spectralloci of false colours for n ranging between 0 and 1.28 are placed approximately equidistant along the straight arrows. (After Horvath et al. 2002 c).


polarizmion-blindncss ------;10 polarizillion-scnsilivilY right window

P,-P.,=P.= 1 P.=P.,-P,- 2. x= 0' ß., ß.,. ß.= arbitrnry ß.- 14S", ßr,= 35", ß,= O'

len window

shady

~, +

~

"-______________ .... Red

Fig.33.8. Spectralloci (designated by capitals A - H, representing the situations A -H in ~ colour Fig. 33.7) of the leaf areas marked with a lett and a right small rectangular window in ~ colour Fig. 33.7 plotted within the equilateral R-G-B colour triangle, the colourless centre of which is represented by +. The arrows start from the spectrallocus of real colours perceived by a polarization-blind retina with PB = PG = PR = 1 and ßR' ßG' ßB = arbitrary, while the arrowheads point to the spectrallocus of false colours perceived bya polarization-sensitive retina with PB = PG = PR = 2, X = 0°, ßR = 145°, ßG = 35°, ßB = 0°. (After Horveith et al. 2002 c).

curved leaf blade. Thus, both the left and right windows in case G represent a shaded situation.

33.2.3 Reflection-Polarizational Characteristics of Plant Surfaces

The darker a plant surface in a given spectral range, the higher the p of reflected light. The reason for this is the following: p of light reflected by the cuticle or epidermis of plants depends on the incident angle, but is almost independent of the wavelength. The direction of polarization of reflected light is parallel to the surface. The colour of plant surfaces arises from the selective absorption and diffuse scattering of light in the tissue below the transparent cuticle. The diffuse light emanating from this tissue is originally unpolarized, but it becomes partially polarized after refraction at the epidermis. The Evector of the tissue-scattered light is perpendicular to the cuticle because of refraction polarization. Hence, the net degree and direction of polarization of a plant surface are determined by the superposition of the epidermisreflected and the subcuticle-scattered light. If the former dominates (e.g. in sunlit shiny leaves observed from the direction of specular reflection), the direction of polarization is parallel to the cuticle; otherwise, the E-vector is perpendicular to it (e.g. sunlit leaves observed from behind, when the leaftransmitted light is perceived). In those spectral regions where the subcuticle-


scattered light has a considerable contribution, the net p of returned light is reduced or even abolished. All these are demonstrated in ~ colour Fig. 33.2.

Surfaces of petals have a matt finish, making them much better diffuse reflectors than leaves, which have a shiny, smooth cuticle (Kay et al. 1981). Thus, petals usually reflect diffuse and only weakly polarized light, while leaves reflect more specularly and the reflected light is generally highly polarized if the direction of view is near the Brewster angle. The E-vector of light reflected from a plant surface is always perpendicular to the plane of reflection.

Horvath et al. (2002 c) proposed that the major function of the surface roughness of petals is not the reduction of p of reflected light (and thus the reduction of polarization-induced false colours), but to reduce the white glare of the surface, which would overwhelm the petal-tissue-backscattered coloured light and would make it more difficult to perceive the real, attractive and striking colour of the petal. An appropriately rough petal surface functions as a Lambertian reflector, which reflects light uniformly in all directions independently of the angle of incidence. As a by-product, the light reflected by a Lambertian surface is unpolarized. The intensity and colour of such a (matt) Lambertian surface is the same from all directions of view. If the surface of a petal were smooth, like the red spathe in Figs. 33.4A and ~ colour Fig. 33.11, it would function as a Fresnel reflector. Then the intensity and colour of the petal-tissue-backscattered coloured light would be overwhelmed by the white glare from the smooth cuticle if the direction of view coincides with the angle of reflection. This problem would not arise for other directions of view. Hence, the reduction of p of reflected light seems to be the consequence, and not the aim of the surface roughness of petals. The roughness of petal surfaces is of great importance for all colour vision systems, independently of polarization-blindness or polarization-sensitivity, which must efficiently detect and distinguish the colours of flowers.

In columns 2 and 3 of ~ colour Fig. 33.7, we can see that at a given illumination direction and in a given (e.g. blue) part of the spectrum the gross features of the patterns of p and a of the Ficus leaf are similar for both the sunlit and the shaded cases, although the colours of the sunlit and shaded leaf differ considerably. The reason for this is that the smooth Ficus leaf is similar to a Fresnel reflector, and the leaf bl ade was tilted so that sunlight could not be reflected specularly from it towards the camera (apart from certain small curved areas). Thus, the sunlight reflected specularly from the leaf bl ade is not visible and does not add to the leaf-tissue-backscattered light. Large differences between the reflection-polarizational characteristics of sunlit and shaded leaves occur only if the direction of view coincides with or is near the direction of specular reflection. This is seen at those regions of the Ficus leaf in rows G and H of ~ colour Fig. 33.7 where the sunlight is specularly reflected.


33.2.4 Do Polarization-Induced False Colours Influence the Weakly Polarization-Sensitive Colour Vision of Papilio Butterflies Under Natural Conditions?

~ Colour Fig. 33.3, Figs. 33.4-33.6 and 33.8 clearly show that for a weakly polarization-sensitive retina the polarization-induced false colours of plants fall near the real colours perceived by a polarization-blind retina even if they reflect strongly polarized light. Horvath et al. (2002 c) showed that in plant parts with dominating diffuse reflection, the colour saturation is relatively high, but p is low. Although in this case hue discrimination will be good, the false colour effect is minute. On the other hand, plant surfaces with high p possess low colour saturation due to the white specularly reflected light. Thus, under natural conditions the weak PS of the photoreceptors might not interfere with the colour vision at all. This may be the reason why the average PS of the photoreceptors in proven colour-sensitive insects is not reduced to 1.0, but was found to be about 2.0-2.5 (Cataglyphis bicolor: Labhart 1986; Papilio: Kelber et al. 2001; Drosophila melanogaster: Speck and Labhart 2001; other fly species: Hardie 1985). Only in honeybees is the PS significantly lower than 2 (Labhart 1980).

The complete destruction of the polarization sensitivity in a microvillar photoreceptor is not a trivial task, but calls for a random or systematic misalignment of the microvilli along the rhabdom. In honeybees, the rhabdom twists by about 1800 which re duces the PS to lower values than in other insects (Wehner et al. 1975; Labhart 1980). This might be taken as an indication that the exquisite colour vision system of honeybees might be more sensitive to small colour differences than that of other insect species and, thus, more compelled to avoid polarizational false colours.

At first glance, the findings of Kelber and collaborators that polarization influences the colour choices of Papilio butterflies, seems to contradict the conclusion of Horvath et al. (2002c) that colour vision is quite insensitive to reflection polarization. However, in their behavioural tests, Kelber and collaborators used stimuli that had both a very high p (:::::: 100 %) and a high degree of colour saturation, a situation that does not occur under natural conditions. Using this hyperstrong polarization/colour saturation combination, Kelber (1 999b ) and Kelber et al. (2001) confirmed behaviourally the polarization sensitivity of Papilio photoreceptors that was previously investigated electrophysiologically by Bandai et al. (1992). Thus, one can assume that this receptor property plays only a minor role in reallife.

To demonstrate that the polarization sensitivity of the colour vision system can indeed ease certain vital tasks in a butterfly's life, further behavioural experiments with Papilio exposed to stimuli with natural combinations of p and colour saturation are needed. For an eye with PS = 2, even for almost totally polarized light reflected from a leaf of Campsis radicans, the false colour shifts in Fig. 33.6 should be smaller than those induced by the totally


polarized and highly colour saturated stimuli of Kelber and co-workers, because the light reflected from leaves has rather low colour saturation. At the moment it is unknown how large a false colour shift needs to be in order to be just detectable, and thus useful in a behavioural context. Although Horvath et al. (2002 c) did not claim that their calculations prove that Papilio is incapable of detecting false colours under natural conditions, they did predict that the calculated colour shifts in the simulated Papilio retina may not be large enough to be seen. The question of whether Papilio could perceive spectral shifts comparable to the calculated polarizational false colour shifts, can be answered only by further studies.

33.3 Polarizational False Colours Perceived by a Highly Polarization-Sensitive Retina Rotating in Front of Flowers and Leaves

Insects hover frequently in front of flowers and leaves, or approach the landing site on plant surfaces along oscillating flight paths, and their body axis more or less rotates to and fro simultaneously. ---7 Colour Figs. 33.9, 33.10, 33.11 and 33.12 demonstrate how the polarization-induced false colours perceived by a highly polarization-sensitive visual system (PS = 20) change in this situation. ---7 Colour Fig. 33.9 demonstrates well that the polarizational false colours of shiny, strongly polarizing leaves usually differ much more from their real colours than for matt, weakly polarizing flower petals. Rotating the head, the false colours of the leaves change more drastically (from violet through bluish and greenish to orange) than those of the petals (remaining in the reddish-orange spectral range). There is no colour change in those regions of the flower where the reflected light is unpolarized. Here, the colours perceived by the polarization-sensitive retina are the same as the real colours.

The leaf blades of grass are randomly oriented and curved, thus both p and a of reflected light change gradually from site to site (---7 colour Fig. 33.10A). The consequence is that the real green colour of the grass becomes very heterogeneous resulting in practically all possible colour shades and hues ranging from violet through blue, yellow, orange to red (---7

colour Fig. 33.10B). Thus, the original relatively homogeneous green grass surface looks kaleidoscopic with randomly alte ring tiny false-coloured patches for a polarization-sensitive visual system, and what is more, the colours of the spots in this kaleidoscope change chaotically as the head rotates or the viewing direction alters.

---7 Colour Fig. 33.11 demonstrates that both the shiny green leaf and the flower-imitating shiny red spathe of Epipremnum aureum reflect highly polarized light, and induce striking false colours, the hues of which change drastically as the eye rotates in front of the plant. If the eye regions viewing the


flowers in visitors were polarization-sensitive, then colour-related behaviour, innate colour preferences, learned associations between nectar and flower colours, flower fidelity, colour constancy, chromaticity contrast between flowers and their background, nectar guides and other spectral flower-pattern components, floral colour changes, true colour vision, or colour mimicry of flowers would lose their sense because of the disturbing effect of polarizationinduced false colours. The consequence of these false colours would be that such flower visitors could not distinguish visually the flowers from other nonfloral objects or the vegetable background.

33.4 Camouflage Breaking via Polarization-Induced False Colours and Reflection Polarization

Generally, polarization-induced false colours would be disadvantageous for perception of colour signals in the plant-pollinator interaction. However, ~ colour Fig. 33.12 presents an example when polarizational false colours could be advantageous in camouflage breaking for a predator. The black carapace of the beetle in ~ colour Fig. 33.12A reflects green light from the surrounding vegetation and blue light from the sky. Thus, the carapace has a greenishbluish appearance, which reduces the colour contrast between the beetle and the underlying green leaf. This effect results in a moderate colour camouflage of the carapace for a polarization-blind visual system. However, this camouflage is broken for a highly polarization-sensitive visual system perceiving the striking polarizational false colours of the carapace, which differ considerably from those of the leaf bl ade. The polarization-induced false colours change dramatically as the eye rotates, which further enhances the break of the colour camouflage. We can see in ~ colour Fig. 33.12B that there is also a remarkable p- and a-contrast between the carapace and the leaf blade, because the horizontal and slightly rough surface of the leaf reflects horizontally polarized light with relatively low p, while the strongly curved and smooth carapace reflects highly polarized light with spatially changing a. Hence, the break of the camouflage of the carapace occurs not only for a highly polarization-sensitive colour vision system, but also for a visual system with true polarization vision which is able to perceive the large p- and a-contrasts between the beetle and the substratum.

Hence, camouflage breaking via polarization-induced false colours or reflection-polarization is a possible visual phenomenon, which would be worth testing in the future. The possible breaking of the brightness and colour camouflage of hidden animals (e.g. caterpillars or frogs on leaves) by means of reflection polarization was suggested by Shashar et al. (1995 c).


33.5 Is Colour Perception or Polarization Sensitivity the more Ancient?

379

Honeybees need accurate colour vision to properly find the flowers as nectar and pollen sourees. They also need PS to orient by means of the celestial polarization pattern. The polarization-sensitive receptors with non-twisted microvilli in the DRA and the polarization-insensitive receptors with regularly twisted microvilli in other eye regions of honeybees and some other insects support the hypothesis that perception of polarization is a more ancient visual capability than colour perception:

• We have seen above that unambiguous colour vision can be ensured only by means of polarization-insensitive photoreceptors. Morphologically, a random microvilli orientation would be the simplest way to abolish the PS. Having exclusively polarization-blind photoreceptors, bees could precisely discriminate the colours and find the adequate flowers in a meadow, but they could not find the way backwards to their nest (with the celestial polarization) if the sun is occluded by clouds.

• Contrary to this, however, if the non-twisted photoreceptors were polarization-sensitive due to the parallel orientation of the microvilli in the entire eye, bees could orient by means of the celestial polarization even if the sun is not visible, and could navigate back to their nest. The cost of this would be ambiguous colour perception due to the polarization-induced false colours. Although in this case bees could not easily find the proper flowers on the basis of their characteristic colour patterns, this imperfection could be partly compensated by using other sensory cues, such as the characteristic shape or odour of flowers. Although polarization-sensitive compound eyes could be frequently deceived by the polarizational false colours, the resulting wrong landing on plant leaves or inappropriate flowers instead of landing on the sought flower petals with given colours would not be deadly. This would only decrease more or less the efficiency of nectar and pollen gathering in bees.

On the basis of the above, the following evolutionary scenario can be proposed:

1. Since the absorption spectrum of photopigments is relatively narrow, receptors with different spectral sensitivities were needed to perceive the near-UV and visible ranges of the spectrum. Thus, UV, blue and green receptors evolved in the ancestors of bees.

2. In the ancestors ofbees polarization-sensitive, non-twisted photoreceptors with parallel microvilli evolved in the eye in order to orient by means of the celestial polarization pattern.


3. In a later stage of evolution, apart from the DRA, the polarization-induced false colours of plant surfaces were eliminated by the proper regular twist of photoreceptors. Polarization sensitivity remained only in the DRA, where polarizational false colours do not cause any problem, because here only UV receptors remained.

Polarization-sensitive receptors need only to be twisted in order to abolish their inherent polarization sensitivity. Would the colour perception be the more ancient, the microvilli of ancient photoreceptors would have originally been randomly oriented (as the simplest solution) to ensure polarizationblindness and unambiguous colour discrimination. Later, polarization-sensitive receptors with parallel microvilli should have gradually evolved from such receptors, if the ancestors of bees had survived the lack of the capability of orientation by means of celestial polarization.

The above proposal is consistent with the findings of Chittka (1996) that the essential components of bee's colour vision predated the evolution of flower colour, because the spectral receptor sets of bees are indistinguishable from those of many members of arthropod taxa, whose evolutionary lineages diverge from those ofbees before there were flowers. The Cambrian ancestors of extant insects and crustaceans already possessed UV, blue and green receptors. Insects were well preadapted for flower colour co ding more than 500 million years ago, about 400 million years before the extensive radiation of the angiosperm plants which started in the middle Cretaceous (100 million years ago), although the origin of the angiosperms might have to be placed in the Triassic. According to Chittka (1996), flower colours had no impact on wavelength positioning of bee photoreceptors. In contrary, because bee colour vision is optimally suited to code flower colour, flower colours should adapted to insect vision.

34 A Common Methodological Error: Intensity Patterns Induced by Selective Reflection of Linearly Polarized Light from Black Surfaces

In behavioural laboratory experiments studying animal orientation, black surfaces are traditionally used to minimize the influence of light reflected from the surfaces surrounding the an im al. The same tradition has been adopted by the majority of researchers investigating polarization sensitivity of animals. However, in these cases the use of black surfaces is the worst choice, as shown in this chapter and as already mentioned in Chap. 31. Coemans and Vos (1989), Coemans et al. (1990, 1994a) and Vos et al. (1995) have called the attention of the scientific community to the need for elimination of such strongly polarizing black surfaces. This has lead to the change of paradigm in behavioural experiments on animal polarization sensitivity.

The forerunners of this subject are Baylor and Smith (1958), Kalmus (1958, 1959) and Waterman (1981). Baylor and Smith (1958) suggested an extra-ocular mechanism of polarization sensitivity in honeybees. According to their hypothesis, an appropriate (e.g. black) substratum, serving as a dancing place for bees, can function as polarization analyzer, since vertical beams oflinearly polarized light illuminating the substratum give rise to reflections with minimum and maximum intensities parallel and perpendicular to the E-vector direction, respectively. On the other hand, vertical rays of unpolarized light illuminating the substratum give rise to reflections of light with uniform intensities in all directions. Although later anatomical, electrophysiological, behavioural and theoretical studies (e.g. Autrum and Stumpf 1950; Stockhammer 1956; Shaw 1967; Seitz 1969) have proven that the rhabdomeres are responsible for polarization sensitivity in bees and many other arthropods, the merit of the hypothesis of Baylor and Smith (1958) was that they as first called the attention of researchers dealing with animal polarization sensitivity to the possibility that spurious unwanted intensity patterns induced by selective reflection of linearly polarized light from the dark surfaces surrounding the animal can also serve as a cue for orientation. The same was emphasized by Kalmus (1958, 1959), who investigated the responses of insects to linearly polarized light in the presence of dark reflecting surfaces. He


observed that certain optomotor reactions to plane-polarized light disappeared when the experimental situation was redesigned so that the unwanted scattering- and reflection-induced intensity patterns could not arise. Waterman (1981) is also one of the few researchers who has taken this problem seriously and pointed out the necessity of testing whether areaction of an animal to linearly polarized light is not elicited by such unwanted brightness patterns. In spite of these warnings, many experimenters studying the responses of animals to polarized light have left them out of consideration. In these cases the polarization sensitivity of the investigated animals cannot be considered as proven. Throughout this volume we have always mentioned when this methodological mistake has occurred in different behavioural experiments.

The light reflected from a non-metallic partially transparent material with a flat surface has two components. The first is reflected directly from the surface, the second one originates from the subsurface, i.e from the inner layers of the material. When a partially linearly polarized light with a given degree of polarization p and angle of incidence is reflected from a flat dielectric material, its amount depends on the angle of polarization a measured from the plane of reflection. The smaller ais, the less the amount of surfacereflected light. Thus, more light is reflected from the surface if the E-vector is perpendicular to the plane of reflection than when it is parallel to this plane (Fig. 34.1A). This selective surface-reflection depends only slightly on the wavelength A in the near-UV and visible ranges of the spectrum, because the index of refraction of dielectric materials is usually only slightly dependent on A in these spectral ranges. The light penetrating into the material is randomly, diffusely scattered and absorbed, the consequence of which is depolarization and decrease of intensity. Both the absorption and the diffuse scattering-induced depolarization may strongly depend on A, which results in that the subsurface-returned component will be more or less depolarized and will possess more or less changed spectral composition.

In the case of dark partially transparent materials with smooth surfaces, the surface-reflected component dominates in those regions of the spectrum in which the subsurface-returned component is reduced by strong absorption. If a dark material strongly absorbs UV light, for example, the surfacereflected component will dominate in the UV, where the amount of reflected light will vary with the change of the E-vector direction relative to the plane of reflection. The high er the degree of polarization, the stronger this variation. In the UV the dark surface is then seen apparently darker or brighter if the Evector of incident light is perpendicular or parallel to the surface, respectively (Fig. 34.1A). Such reflection-polarization-induced UV-brightness differences can occur in all dark chambers, rooms or tanks, for instance, the walls of which are covered by black or grey UV-absorbing paint, or composed of black UV-absorbing Plexiglas or black UV-absorbing cloth curtain. The walls receiving more or less perpendicularly polarized light reflect less light in the

34 A Common Methodological Error

A

npparenlly brighlcr

383

t 0 p v i e w

darker t-.

! er ~ .., ..c (lÖ ' B eJ) ::l" 'C ;; .0 ..,

darker

I äi brighter p-I -"'.... .. .. ~ C «I :>;-"C ~

brighter

Fig. 34.1. A Selective reflection of linearly polarized light from the vertical walls SUf

rounding an imaginary observer. The E-vectors of light are represented by doubleheaded arrows, the length of which is proportional to the intensity. The differences in the grey shades demonstrate qualitatively the brightness differences perceived by the observer. B, C The relative positions of the brighter and darker regions of the walls of a rectangular room, chamber or tank as perceived by an observer in the centre for two orthogonal directions of the E-vector of linearly polarized light illuminating the scene from above.

UV than the walls receiving more or less parallel polarized light (Fig. 34.1A). If the E-vector of ambient light in these chambers, rooms or tanks is rotated, the reflection-induced UV-brightness pattern will follow this rotation (Figs.34.1B,C).

All these problems can be avoided if white materials with as rough surfaces as possible are used in the experiments instead ofblack or grey materials with smooth shiny surfaces. In the case of bright, UV +white-reflective materials with rough surfaces the polarized surface-reflected component, causing all the troubles, can be strongly depolarized and suppressed by the intense and unpolarized subsurface-returned component, which drastically reduces or even eliminates any brightness differences induced by reflection polarization.

References

Able KP (1982) Skylight polarization patterns at dusk influence migratory orientation in birds. Nature 299:550-551

Able KP (1989) Skylight polarization patterns and the orientation of migratory birds. J Exp Biol141:241-256

Able KP (1991) The development of migratory orientation mechanisms. In: Orientation in birds. P Berthold (ed) Birkhäuser, Basel Berlin Boston, pp 166-179

Able KP (1993) Orientation cues used by migratory birds: a review of cue-conflict experiments. Trends Ecol Evol8:367-371

Able KP, Able MA (1990a) Calibration of the magnetic compass of a migratory bird by celestial rotation. Nature 347:378-380

Able KP, Able MA (1990b) Ontogeny of migratory orientation in the Savannah sparrow, Passerculus sandwichensis: calibration of the magnetic compass. Anim Behav 39:905-913

Able KP, Able MA (1990 c) Ontogeny of migratory orientation in the Savannah sparrow, Passerculus sandwichensis: mechanisms at sunset.Anim Behav 39:1189-1198

Able KP, Able MA (1993) Daytime calibration of magnetic orientation in a migratory bird requires a view of skylight polarization. Nature 364:523-525

Able KP, Able MA (1995a) Manipulations of polarized skylight calibrate magnetic orientation in a migratory bird. J Comp Physiol A 177:351-356

Able KP, Able MA (1995b) Interactions in the flexible orientation system of a migratory bird. Nature 375:230-232

Able KP, Able MA (1996) Migratory orientation: autumn calibration of magnetic orientation is not evident in spring. Naturwissenschaften 83:517-518

Able KP,Able MA (1997) Development of sunset orientation in a migratory bird: no calibration by the magnetic field. Anim Behav 53:363-368

Adams JT, Kattawar GW (1997) Neutral points in an atmosphere-ocean system. 1. Upwelling light field. Appl Opt 36:1976-1986

Adler K (1969) Extraoptic phase shifting of circadian locomotor rhythms in salamanders. Science 164:1290-1292

Adler K (1976) Extraocular photoreception in amphibians. Photochem Photobiol 23:275-298

Adler K, Phillips JB (1985) Orientation in a desert lizard (Uma notata): time-compensated compass movement and polarotaxis. J Comp Physiol A 156:547-552

Adler K, Taylor DH (1973) Extraocular perception of polarized light by orienting salamanders. J Comp Physiol87:203-212

Aepli F, Labhart T, Meyer EP (1985) Structural specializations of the cornea and retina at the dorsal rim of the compound eye in hymenopteran insects. Cell Tiss Res 239: 19-24

Akersten WA, Shaw CA, Jefferson GT (1983) Rancho La Brea: status and future. Paleobiol 9:211-217

Akesson S, Bäckman J (1999) Orientation in pied flycatchers: the relative importance of magnetic and visual information at dusk. Anim Behav 57:819-828

386 References

Alerstam T (1990) Eeological eauses and eonsequenees of bird orientation. Experientia 46:405-415

Altevogt R, Hagen HO von (1964) Über die Orientierung von Uca tangeri Eydoux im Freiland. Z Morphol Ökol Tiere 53:636-656

Altner I, Burkhardt D (1981) Fine strueture of the ommatidia and the oeeurrence of rhabdomeric twist in the dorsal eye of male Bibio marei (Diptera, Nematoeera, Bibionidae). Cell Tiss Res 215:607-623

Angus PB (1973) Pleistoeene Helophorus (Coleoptera, Hydrophilidae) from Borislav and Starunia in the Western Ukraine, with areinterpretation of M. Somnieki's speeies, deseription of a new Siberian species and comparison with British Weiehselian faunas. Phil Trans Roy Soe Lond 265:299-326

Anton-Erxleben F, Langer H (1988) Functional morphology of the ommatidium in the eompound eye of the moth, Antheraea polyphemus (Insecta, Saturniidae). Cell Tiss Res 252:385-396

Arikawa K, Hieks JL, Williams DS (1990) Identifieation of actin filaments in the rhabdomeral microvilli of Drosophila photoreeeptors. J Cell Biol11 0: 1993-1998

Arikawa K, Inokuma K, Eguchi E (1987) Pentachromatie visual system in a butterfly. Naturwissenschaften 74:297-298

Auburn JS, Taylor DH (1979) Polarized light pereeption and orientation in larval bullfrogs Rana eatesbeiana. Anim Behav 27:658-668

Autrum H, Stumpf H (1950) Das Bienenauge als Analysator für polarisiertes Licht. Z Naturforsch 5b:116-122

Autrum HJ, Zwehl V von (1962) Die Sehzellen der Insekten als Analysatoren für polarisiertes Licht. Z vergl Physio146:1-7

Azzam RMA, Bashara NM (1992) Ellipsometry and polarized light. North-Holland, Amsterdam, New York

Babinet J (1840) Sur un nouveau point neutre dans l'atmosphere. Compt Rend 11:618-620

Bagnara JT, Hadley ME (1970) Endoerinology of the amphibian pineal. Am Zoologist 10:201-216

Bainbridge R, Waterman TH (1957) Polarized light and the orientation of two marine Crustaeea. J Exp Bio134:342-364

Bainbridge R, Waterman TH (1958) Turbidity and the polarized light orientation of the erustaeean, Mysidium. J Exp Biol 35:487-493

Baker RR (1968) Sun orientation during migration in some British butterflies. Proc Roy Entomol Soe Lond A 143:89-95

Bandai K, Arikawa K (1990) Arrangement of speetral receptors in the ommatidia of Papilio xuthus (Lepidoptera) determined by measurement of the polarization sensitivity. Zool Sei 7:1029

Bandai K,Arikawa K, Eguehi E (1992) Localization of spectral receptors in the ommatidium of butterfly eompound eye determined by polarization sensitivity. J Comp PhysiolA 171:289-297

Bardolph M, Stavn RH (1978) Polarized light sensitivity in the stage I zoea of the mud erab Panopeus herbstii. Mar Bio146:327-333

Barral MJA (1858) Oeuvres de Franeois Arago I-V., Gide - Paris, Weigel- Leipzig Barta A, Horvath G (2004) Why is it advantageous for animals to deteet eelestial polar

ization in the ultraviolet? Skylight polarization under c10uds and canopies is strongest in the uv. Theor Biol (in press)

Barta A, Horvath G, Bernath B, Meyer-Roehow VB (2003) Imaging polarimetry of the rainbow. Appl Opt 42:399-405

Barter JD, Thompson HR, Riehardson CL (2003) Visible-regime polarimetrie imager: a fully polarimetrie, real-time imaging system. Appl Opt 42: 1620-1628

References 387

Bartsch K (1991) Die funktionelle und strukturelle Regionalisierung des Auges der Wasserläufer. PhD thesis, Univ Tübingen, Germany

Bartsch K (1995) Polarization-sensitive photoreceptors of different spectral types in the compound eye of waterstriders. Naturwissenschaften 82:292-293

Bary E de, Bullrich K, Lorenz D (1961) Messungen der Himmelsstrahlung und deren Polarisationsgrad während der Sonnenfinsternis am 15.2.1961 in Viareggio (Italien). Geofisica Pura et Applicata 48: 193-198

Baylor ER (1959) The responses of snails to polarized light. J Exp BioI36:369-376 Baylor DA, Fettiplace R (1975) Light path and photon capture in turtle photoreceptors. J

Physiol 248:433-464 Baylor ER, Kennedy D (1958) Evidence against a polarizing analyzer in the bee eye.Anat

Rec 132:411 Baylor ER, Smith FE (1953) The orientation of Cladocera to polarized light. Am Natural

ist 87:97-101 Baylor ER, Smith FE (1958) Extra-ocular polarization analysis in the honeybee.Anat Rec

132:411-412 Beason RC, Semm P (1991) Neuroethological aspects of avian orientation. In: Berthold P

(ed) Orientation in birds. Birkhäuser, Basel Boston Berlin, pp 106-127 Bellver C (1987) Influence of particulate pollution on the positions of neutral points in

the sky at Seville (Spain). Atmos Environ 21:699-702 Bernard GD, Wehner R (1977) Functional similarities between polarization vision and

color vision. Vis Res 17:1019-1028 Bernath B, Horvath G (1999) Visual deception of a Great White Egret by shiny plastic

sheets. Ornis Hungarica 8/9:57-61 Bernath B, Szedenics G, Molmir G, Kriska G, Horvath G (2001a) Visual ecological impact

of a peculiar waste oillake on the avifauna: dual-choice field experiments with waterseeking birds using huge shiny black and white plastic sheets. Arch Nat Conserv Landsc Res 40(1):1-28

Bernath B, Szedenics G, Molmir G, Kriska G, Horvath G (2001 b) Visual ecological impact of"shiny black anthropogenie products" on aquatic insects: oil reservoirs and plastic sheets as polarized traps for insects associated with water. Arch Nat Conserv Landsc Res 40(2):89-109

Bernath B, Szedenics G, Wildermuth H, Horvath G (2002) How can dragonflies discern bright and dark waters from a distance? The degree of polarization of reflected light as a possible cue for dragonfly habitat selection. Freshwater Bio147:1707 -1719

Beugnon G (1983) Terrestrial and celestial cues in visual orientation of the wood cricket Nemobius sylvestris (Bosc). Biol Behav 8: 159-169

Beugnon G (1986) Learned orientation in landward swimming in the cricket Pteronemobius lineolatus. Behav Proc 12:215-226

Bingman VP (1983) Magnetic field orientation of migratory naive Savannah sparrows with different first summer experience. Behav 87:43-53

Birukow G (1954) Photogeomenotaxis bei Geotrupes silvaticus Panz. und ihre zentralnervöse Koordination. Z vergl PhysioI36:176-211

Bisch SM (1999) Orientierungsleistungen des nachtaktiven Wüstenkäfers Parastizopus armaticeps Peringuey (Coleoptera: Tenebrionidae). PhD thesis, Univ Bonn, Germany

Blanke H (1996) Visuelle Detektion von Eigenbewegung beim Rückenschwimmer Notonecta glauca. PhD thesis, Univ Tübingen, Germany, p. 79

Blanke H, Varju D (1995) Visual determination of self motion components: Regionalization of the sensitivity to rotation in the backswimmer Notonecta. In: Elsner N, Menzel R (eds) Göttingen Neurobiol Rept 1995. Proc 23rd Göttingen Neurobiol Conf, vol II, p 427, G Thieme, Stuttgart

388 References

BIest AD, Stowe S (1990) Dynamic microvillar cytoskeletons in arthropod and squid photoreceptors. Cell Motil CytoskeI17:1-5

BIest AD, Stowe S, Eddey W, Williams DS (1982) The local deletion of a microvillar cytoskeleton from photoreceptors of tipulid flies during membrane turnover. Proc Roy Soc Lond B 215:469-479

Bletz H, Weindler R, Wiltschko R, Wiltschko W, Berthold P (1996) The magnetic field as reference for the innate migratory direction in blackcaps, Sylvia atricapilla. Naturwissenschaften 83:430-432

Blum M, Labhart T (2000) Photoreceptor visual fields, ommatidial array, and receptor axon projections in the polarization-sensitive dorsal rim area of the cricket compound eye. J Comp Physiol A 186:119-128

Boehm G (1940) Über ein neues entoptisches Phänomen im polarisierten Licht. "Periphere" Polarisations büschel. Acta Ophthalmol18: 143-169

Boerner WM, Brand H, Cram LA, Holm WA, Stein DE, Wiesbeck W, Keydel W, Giuli D, Gjessing DT, Molinet FA (eds) (1992) Direct and inverse methods in radar polarimetry. Dordrecht, Kluwer

Bohn H, Täuber U (1971) Beziehungen zwischen der Wirkung polarisierten Lichtes auf das Elektroretinogramm und der Ultrastruktur des Auges von Gerris lacustris 1. Z vergl Physiol 72:32-53

Bone RA (1980) The role of the macular pigment in the detection of polarized light. Vis Res 20:213-220

Bone RA, Landrum JT (1984) Macular pigment in Henle fiber membran es: a model for Haidinger's brushes. Vis Res 24: 1 03-1 08

Bone Q, Marshall NB (1982) Biology of fishes. Blackie, London Born M, WolfE (1999) Prieiples of optics. 7th ed, Cambridge Univ Press, Cambridge, UK Bowmaker JK (1977) The visual pigments, oil droplets and spectral sensitivity in the

pigeon. Vis Res 17:1129-1138 Brewster D (1842) On the existence of a new neutral point and two secondary neutral

points. Rept 12th Meet British Assoc Adv Sei (Manchester, June 1842), J. Murray, London, 1843, p. 13

Brewster D (1847) On the polarisation of the atmosphere. Phil Magaz J Sei 31 :444-454 Breon FM, Deuze JL, Tanre D, Herman M (1997) Validation of spaceborne estimates of

aerosolloading from Sun photometer measurements with emphasis on polarization. J Geophys Res D 102:17187-17195

Briggs RO, Hatcett GL (1965) Techniques for improving underwater visibility with video equipment. Ocean Sei Ocean Engin 1-2:1284-1308

Brines ML (1980) Dynamic patterns of skylight polarization as dock and compass. J Theor BioI86:507-512

Brines ML, Gould JL (1979) Bees have rules. Seience 206:571-573 Brines ML, Gould JL (1982) Skylight polarization patterns and animal orientation. J Exp

BioI96:69-91 Brodskiy AK (1973) The swarming behavior of mayflies (Ephemeroptera). Entomol Rev

52:33-39 Browman HI, Novales Flamarique I, Hawryshyn CW (1994) Ultraviolet photoreception

contributes to prey search behaviour in two speeies of zooplanktivorous fishes. J Exp BioI186:187-198

Brownell P, Weber T (1995) Polarization vision in nocturnal sand scorpions. In: Burrows M, Matheson T, Newland PL, Schuppe H (eds) Nervous systems and behaviour. Proc 4th Int Cong Neuroethol, Sept 3-8,1995, P 281, G Thieme, Stuttgart

Brunner D, Labhart T (1987) Behavioural evidence for polarization vision in crickets. Physiol Entomol12: 1-10

Bueno JM (2000) Polarimetry using liquid-crystal variable retarders: theory and calibration. J Opt A (Pure Appl Opt) 2:216-222

References 389

Bueno JM, Artal P (1999) Double-pass imaging polarimetry in the human eye. Opt Lett 24:64-66

Burdon-Jones C, Charles GH (1958) Light reactions of Littoral gastropods. Nature 181:129-131

Burghause FMHR (1979) Die strukturelle Spezialisierung des dorsalen Augenteils der Grillen (Orthoptera, Grylloidea). Zool Jhrb, PhysioI83:502-525

Burghause FMHR (1981a) The structure of the double-eyes of Baetis and the uniform eyes of Ecdyonurus (Ephemeroptera). Zoomorphol 98:17-34

Burghause FMHR (1981b) Die Struktur der Komplexaugen von der Gewächshausschrecke, Tachycines asynamorus. Zool Beitr NF 27:185-197

Burkhardt D, Wendler L (1960) Ein direkter Beweis für die Fähigkeit einzelner Sehzellen des Insektenauges, die Schwingungsrichtung polarisierten Lichtes zu analysieren. Z vergl Physio143:687 -692

Butler R, Horridge GA (1973) The electrophysiology of the retina of Periplaneta americana 1. II. Receptor sensitivity and polarized light sensitivity. J Comp Physiol 83:279-288

Cameron DA, Pugh EN Jr (1991) Double cones as a basis for a new type of polarization vision in vertebrates. Nature 353:161-164

Caveney S (1971) Cuticle reflectivity and optical activity in scarab beetles: the role of uric acid. Proc Roy Soc Lond B 178:205-225

Chandrasekhar S (1950) Radiative transfer. Clarendon Press, Oxford Charles GH (1961a) The orientation of Littorina species to polarized light. J Exp Biol

38:189-202 Charles GH (1961b) The mechanism of orientation of freely moving Littorina littoralis

(1.) to polarized light. J Exp BioI38:203-212 Charles GH (1961 c) Orientational movements of the foot of Littorina species in relation

to the plane of vibration of polarized light. J Exp Biol 38:213-224 Chelazzi G (1992) Reptiles. In: Papi F (ed) Animal homing. Chapman and Hall, London,

pp 235-255 Chen D, Collins JS, Goldsmith TH (1984) The ultraviolet receptor of bird retinas. Science

225:337 -340 Chisholm IH (1975) Haidinger's brushes: a photographic record. In: Proc Macular

Worksh, Bath, England, 27 Apr.-3 May 1975, Section 1-23 Chittka L (1996) Does bee color vision predate the evolution of flower color? Naturwis

senschaften 83: 136-138 Clarke D, Grainger JF (1971) Polarized light and optical measurement. Pergamon Press,

Oxford Clarkson ENK (1979) The visual system of trilobites. Palaeontology 22: 1-22 Coemans MAJM, VOS HJJ (1989) On the perception of polarized light by the homing

pigeon, Columba livia 1. In: Kulikowski JJ, Dickonson CM, Murray IJ (eds) Seeing contour and colour. Pergamon Press, Oxford, pp 632-638

Coemans MAJM, Vos HJJ, Nuboer JFW (1990) No evidence for polarization sensitivity in the pigeon. Naturwissenschaften 77:138-142

Coemans MAJM, Vos HJJ, Nuboer JFW (1994a) The orientation of the E-vector of linearly polarized light does not affect the behaviour of the pigeon Columba livia. J Exp BiolI91:107-123

Coemans MAJM, Vos HJJ, Nuboer JFW (l994b) The relation between celestial colour gradients and the position of the sun, with regard to the sun compass. Vis Res 34:1461-1470

Collett E (1994) Polarized light: fundamentals and applications. Marcel Dekker Inc, New York

390 References

Cone RA (1972) Rotational diffusion of rhodopsin in the visual receptor membrane. Nature 236:39-43

Consi TR, Passani MB, Macagno ER (1990) Eye movements in Daphnia magna. Regions of the eye are speeialized for different behaviors. J Comp Physiol A 166:411-420

Coombe PE, Srinivasan MV, Guy RG (1989) Are the large monopolar cells of the insect lamina on the optomotor pathway? J Comp Physiol A 166:23-35

Corbet PS (1999) Dragonflies: behaviour and ecology of Odonata. Harley Books, Martins, Great Horkesley, Colchester, Essex

Cornu MA (1884) Observations relatives a la couronne visible actuellement autour du Soleil. Compt Rend 99:488-493

Coughlin DJ, Hawryshyn CW (1994) The contribution of ultraviolet and short-wavelength cone mechanisms to color vision in rainbow trout. Brain Behav Evol 43:219-232

Coughlin DJ, Hawryshyn CW (1995) A cellular basis for polarized-light vision in rainbow trout. J Comp Physiol A 176:261-272

Coulson KL (1974) The polarization oflight in the environment. In: GehreIs T (ed) Planets, stars and nebulae studied with photopolarimetry. Univ Arizona Press, Tucson, Arizona, pp 444-471

Coulson KL (1983) Effects of the EI Chich6n volcanic cloud in the stratosphere on the polarization of light from the sky. Appl Opt 22: 1 036-1050

Coulson KL (1988) Polarization and intensity oflight in the atmosphere.A Deepak Publishing, Hampton, Virginia, USA

Couturier A, Robert P (1957) Recherches sur la faculte d'orientation du hanneton commun (Melolontha melolontha L). Compt Rend Acad Sei 245:2399-2401

Cronin TW, Marshall J (200l) Parallel processing and image analysis in the eyes of mantis shrimps. Biol Bu1l200:177-183

Cronin TW, Marshall N], Caldwell RL (2000) Spectral tuning and the visual ecology of mantis shrimps. Phil Trans Roy Soc Lond B 355:1263-1267

Cronin TW, Shashar N (200l) The linearly polarized light field in clear, tropical marine waters: spatial and temporal variation of light intensity, degree of polarization and evector angle. J Exp BioI204:2461-2467

Cronin TW, Shashar N, WolffL (1994) Portable imaging polarimeters. Proc 12th IAPR Int Conf Pattern Recogn. Jerusalem, Israel, October 9-13, 1994, pp 606-609

Cronin TW, Shashar N, Wolff L (1995) Imaging technology reveals the polarized light fields that exist in nature. Biophotonics Int 2(2):38-41

Crozier WJ, Mangelsdorf AF (1924) A note on the relative photosensory effect of polarized light. J Gen PhysioI6:703-709

Curran PJ (1982) Polarized visible light as an aid to vegetation classification. Rem Sens Environ 12:491-499

Dacke M, Doan TA, O'Carroll DC (200l) Polarized light detection in spiders. J Exp Biol 204:2481-2490

Dacke M, Nilsson DE, Scholtz CH, Byrne M, Warrant EJ (2003b) Insect orientation to polarized moonlight. Nature 424: 33

Dacke M, Nilsson DE, Warrant EJ, BIest AD, Land MF, O'Carroll DC (1999) Built-in polarizers form part of a compass organ in spiders. Nature 401:470-473

Dacke M, Nordström P, Scholtz CH (2003a) Twilight orientation to polarised light in the crepuscular dung beetle Scarabaeus zambesianus. J Exp BioI206:1535-1543

Dacke M, Nordström P, Scholtz CH, Warrant EJ (2002) A specialized dorsal rim area for polarized light detection in the compound eye of the scarab beetle Pachysoma striaturn. J Comp Physiol A 188:211-216

Dahmen HJ (199l) Eye specialisation in waterstriders: an adaptation to life in a flat world. J Comp Physiol A 169:623-632

References 391

Dandekar BS, Turtle JP (1971) Day sky brightness and polarization during the total solar eclipse of7 March 1970. Appl Opt 10:1220-1224

Danthanarayana W, Dashper S (1986) Response of some night -flying insects to polarized light. In: Danthanarayana W (ed) Insect flight: dispersal and migration. Springer, Berlin Heidelberg New York, pp 120-127

Davitz MA, McKaye KR (1978) Discrimination between horizontally and vertically polarized light by the cichlid fish Pseudotropheus macrophthalmus. Copeia 1978(2):333-334

Degner SL, Hawryshyn CW (2001) Orientation of rainbow trout (Oncorhynchus mykiss) to multiple patches of linearly polarized light. Canadian J Zool 79:407-415

Delius JD, Emmerton J (1979) Visual performance of pigeons. In: Granda AM, Maxwell JH (eds) Neural mechanisms of behavior in the pigeon. Plenum Press, New York, pp 51-70

Delius JD, Perchard RJ, Emmerton J (1976) Polarized light discrimination by pigeons and an electroretinographic correlate. J Comp Physiol Psycho190:560-571

Delori FC, Webb RH, Parker JS (1979) Macular birefringence. Invest Ophthalmol Vision Sci (Suppl,ARVO Abstr) 19:53

Denton EJ (1959) The contributions of the oriented photosensitive and other molecules to the absorption of the whole retina. Proc Roy Soc Lond B 150:78-94

Denton EJ, Nicol JAC (1965) Polarization oflight reflected from the silvery exterior of the bleak, Alburnus alburnus. J Mar Biol Assoc UK 45:705-709

Denton EJ, Nicol JAC (1966) A survey of reflectivity in silvery teleosts. J Mar Biol Assoc UK 46:685-722

DeRosa CT, Taylor DH (1980) Homeward orientation mechanisms in three species of turtles (Trionyx spinifer, Chrysemys picta, and Terrapene carolina). Behav Ecol Sociobio17:15-23

Deschamps PY, Bn!on FM, Leroy M, Podaire A, Bricaud A, Buriez JC, Seze G (1994) The POLDER mission: instrument characteristics and scientific objectives. IEEE Trans Geosci Rem Sens 32:598-615

Dill PA (1971) Perception of polarized light by yearling sockeye salmon (Oncorhynchus nerka). J Fish Res Board Can 28:1319-1322

Dimmer F (1894) über die entoptischen Erscheinungen in der Gegend der Macula lutea. Centralbl Physiol 7:266-268

Dodt E, Kuba M (1990) Visually evoked potentials in response to rotating plane-polarized blue light. Ophthal Res 22:391-394

Doujak FE (1984) Electrophysiological measurement of photoreceptor membrane dichroism and polarization sensitivity in a Grapsid crab. J Comp Physiol A 154:597-605

Dreher AW, Reiter K, Weinreb RN (1992) Spatially resolved birefringence of the retinal nerve fiber layer assessed with a retinallaser ellipsometer. Appl Opt 31 :3730-3735

Duelli P (1974) Polarisationsmusterorientierung bei der Wüsten ameise Cataglyphis bicolor (Formicidae, Hymenoptera). PhD thesis, Univ Zurich, Switzerland

Duelli P (1975) A fovea for e-vector orientation in the eye of Cataglyphis bicolor (Formicidae, Hymenoptera). J Comp Physiol102:43-56

Duelli P, Wehner R (1973) The spectral sensitivity of polarized light orientation in Cataglyphis bicolor (Formicidae, Hymenoptera). J Comp Physio186:37-53

Duellman WE, Trueb L (1986) Biology of amphibians. McGraw-Hill, New York Dürst J (1982) Two colour photometry and polarimetry of the solar corona of 16 Febru

ary 1980. Astronom Astrophys 112:241-250 Eakin RM (1973) The third eye. Univ California Press, Berkeley, Los Angeles Edmonds DT (1996) A sensitive optically-detected magnetic compass for animals. Proc

Roy Soc Lond B 263:295-298

392 References

Edrich W, Helversen 0 von (1976) Polarized light orientation of the honey bee: the minimal visual angle. J Comp Physiol A 109:309-314

Edrich W, Helversen 0 von (1987) Polarized light orientation in honey bees: is time a component in sampling? Biol Cybern 56:89-96

Edrich W, Neumeyer C, Helversen 0 von (1979) "Anti-sun orientation" of bees with regard to a field of ultraviolet light. J Comp Physiol l34: 151-157

Egan WG (1985) Photometry and polarization in remote sensing. Elsevier, Amsterdam NewYork

Egelhaaf A, Dambach M (1983) Giant rhabdomes in a specialized region of the compound eye of a cricket: Cycloptiloides canariensis (Insecta, Gryllidae). Zoomorphol 102:65-77

Eggers A, Gewecke M (1993) The dorsal rim area of the compound eye and polarization vision in the desert locust (Schistocerca gregaria). In: Wiese K, Gribakin FG, Popov AV, Renninger G (eds) Sensory systems of arthropods. Birkhäuser, Basel,pp 101-109

Eggers A, Weber T (1993) Behavioural evidence for polarization vision in locusts. In: Eisner N, Heisenberg M (eds) Gene, brain, behaviour. Thieme Verlag, Stuttgart, p 336

Eguchi E (1999) Polarized light vision and rhabdom. In: Eguchi E, Tominaga Y (eds) Atlas of arthropod sensory receptors. Springer, Berlin Heidelberg New York, pp 33-46

Ellis-Quinn BA, Simon CA (1989) Homing behaviour of the lizard Sceloporus jarrovi. J HerpetoI23:146-152

Ellis-Quinn BA, Simon CA (1991) Lizard homing behaviour: the role of the parietal eye during displacement and radio-tracking, and time-compensated celestial orientation in the lizard Sceloporus jarrovi. Behav Ecol SociobioI28:397-407

Fein A, Szuts EZ (1982) Photoreceptors: their role in vision. Cambridge Univ Press, Cambridge

Fent K (1985) Himmelsorientierung bei der Wüstenameise Cataglyphis bicolor. Bedeutung von Komplexaugen und Zellen. PhD thesis, Univ Zurich, Switzerland

Fent K (1986) Polarized skylight orientation in the desert ant Cataglyphis. J Comp PhysiolA 158:145-150

Fent K, Wehner R (1985) Ocelli: a celestial compass in the desert ant Cataglyphis. Science 228:192-194

Ferguson DE, Landreth HF (1966) Celestial orientation of Fowler's toad, Bufo fowleri. Behav 26:105-123

Ferguson DE, McKeown JP, Bosarge OS, Landreth HF (1968) Sun-compass orientation of bullfrogs. Copeia 1968:230-235

Fernandez-Moran H (1956) Fine structure of the insect retinula as revealed by electron microscopy. Nature 177:742-743

Fineran BA, Nicol JAC (1978) Studies on the photoreceptors of Anchoa mitchilli and A. hepsetus (Engraulidae) with particular reference to the cones. Phil Trans Roy Soc Lond B 283:25-60

Firth BT, Kennaway DJ (1980) Plasma melatonin levels in the scincid lizard, Trachydosaurus rugosus: the effects of parietal eye and lateral impairment. J Exp Biol 85:311-321

Fischer C (1992) Evolution des Schwarmfluges und Flugverhalten der Ephemeropteren. PhD thesis, 1992/3291, Friedrich Alexander Univ Erlangen-Nürnberg, p 171

Fischer K (1961) Untersuchungen zur Sonnenkompassorientierung und Laufaktivität von Smaragdeidechsen (Lacerta viridis Laur.). Z Tierpsych 18:450-470

Fitch BW, Walraven RL, Bradley DE (1984) Polarization of light reflected from grain crops during the heading growth stage. Rem Sens Environ 15:263-268

Fortini ME, Rubin GM (1991) The optic lobe projection pattern of polarization -sensitive photoreceptor cells in Drosophila melanogaster. Cell Tiss Res 265:185-191

References 393

Forward RB Jr, Horch KW, Waterman TH (1972) Visual orientation at the water surface by the teleost Zenarchopterus. Biol Bull143:112-126

Forward RB Jr, Waterman TH (1973) Evidence for E-vector and light intensity pattern discrimination by the teleost Dermogenys. J Comp PhysioI87:189-202

Frantsevich L, Govardovski V, Gribakin F, Nikolajev G, Pichka V, Polanovsky A, Shevchenko V, Zolotov V (1977) Astroorientation in Lethrus (Coleoptera, Scarabaeidae). J Comp Physiol A 121:253-271

Frantsevich LI, Zolotov VV (1979) Criticism of the method for identification of polarization sensitive neurons in insects. Neurophysiol (Kiev) 11:326-337 (in Russian)

Fraser FC (1936) The fauna of British India. Odonata vol3, London, Taylor and Francis Freake MJ (1999) Evidence for orientation using the e-vector direction of polarised light

in the sleepy lizard Tiliqua rugosa. J Exp BioI202:1159-1166 Freake MJ (2001) Homing behaviour in the sleepy lizard (Tiliqua rugosa): the role of

visual cues and the parietal eye. Behav Ecol SociobioI50:563-569 Frisch K von (1948) Gelöste und ungelöste Rätsel der Bienensprache. Naturwis

senschaften 1:12-23,2:38-43 Frisch K von (1949) Die Polarisation des Himmelslichtes als orientierender Faktor bei

den Tänzen der Bienen. Experientia 5: 142-148 Frisch K von (1950) Die Sonne als Kompass im Leben der Bienen. Experientia 6:210-221 Frisch K von (1953) Aus dem Leben der Bienen (5th ed) Springer, Berlin Göttingen Hei

delberg Frisch K von (1967) The dance language and orientation of bees. Harvard Univ Press,

Cambridge, Massachusetts (USA) Frisch K von, Lindauer M (1954) Himmel und Erde in Konkurrenz bei der Orientierung

der Bienen. Naturwissenschaften 41:245-253 Frisch K von, Lindauer M, Schmeidler F (1960) Wie erkennt die Biene den Sonnenstand

bei geschlossener Wolkendecke? Naturwiss Rundsch 5:169-172 Frost J (1974) Eye movements in Daphnia pulex (de Geer). J Exp BioI62:175-187 Fryer G, !les TP (1972) The cichlid fishes of the Great Lakes of Africa: their biology and

evolution. Oliver and Boyd, Edinburgh Garab G, Pomozi I, Javorfi T, Menczel L, Weiss G, Jürgens R (2003) Differential polariza

tion laser scanning microscope. Biophys J (submitted) Gates DM (1980) Biophysical ecology. Springer, Berlin Heidelberg NewYork Ga! J (1997) Some biophysical applications of videopolarimetry. Diploma thesis, Eötvös

Univ, Dept Atomic Phys, Biophys Group, Budapest, p 96 (supervisor: G Horv<ith) Ga! J, Horv<ith G, Barta A, Wehner R (2001a) Polarization of the moonlit clear night sky

measured by full-sky imaging polarimetry at full moon: comparison of the polarization of moonlit and sunlit skies. J Geophys Res D 106:22647-22653

Ga! J, Horv<ith G, Meyer-Rochow WB (2001b) Measurement of the reflection-polarization pattern of the flat water surface under a clear sky at sunset. Rem Sens Environ 76:103-111

Ga! J,Horvath G,Meyer-RochowVB, Wehner R (2001 c) Polarization patterns of the summer sky and its neutral points measured by full-sky imaging polarimetry in Finnish Lapland north ofthe Arctic Circle. Proc Roy Soc Lond A 457:1385-1399

Gehreis T (ed) (1974) Planets, stars and nebulae studied with photopolarimetry. Univ Arizona Press, Tucson, Arizona

Gerharz R (1976) Appearance of the atmospheric scatter field during a solar eclipse. J Geophys 42:163-167

Giguere LA, Dunbrack RL (1990) Thin layer interference may reduce the visibility of transparent phantom midge larvae (Chaoborus trivittatus) to predators. Can J Fish Aq Sei 47:1043-1046

394 References

Glantz RM (1996) Polarization sensitivity in crayfish lamina monopolar neurons. J Comp Physiol A 178:413-425

Glantz RM (2001) Polarization analysis in the crayfish visual system. J Exp Biol 204:2383-2390

Glas HW van der (1975) Polarization induced colour patterns: a model of the perception of the polarized skylight by insects. 1. Tests in choice experiments with running honey bees, Apis mellifera. Netherlands J Zoo125:476-505

Glas HW van der (1977) Models for unambiguous E-vector navigation in the bee. J Comp Physiol A 113:129-159

Goddard SM, Forward RB Jr (1989) The use of celestial cues in the offshore escape response of the shrimp, Palaemonetes vulgaris (Say). Mar Behav Physiol16:11-18

Goddard SM, Forward RB Jr (1990) The decay and learning of an y-axis orientation behavior: the offshore escape response of the shrimp, Palaemonetes vulgaris (Say). J Exp Mar Biol Ecol142:137-150

Goddard SM, Forward RB Jr (1991) The role of the underwater polarized light pattern in sun compass navigation of the grass shrimp, Palaemonetes vulgaris. J Comp Physiol A 169:479-491

Gokan N (1989) The compound eye of the dung beetle Geotrupes auratus (Coleoptera, Scarabaeidae). Appl Entomol Zool 24: 133-146

Gokan N (1990) Fine structure of the compound eye of the dung beetle, Onthophagus lenzii (Coleoptera, Scarabaeidae). Japan J Entomo158:185-195

Goldsmith TH (1972) The natural history of invertebrate visual pigments. In: Dartnall HJA (ed) Handbook of sensory physiology. VIIIl, Springer, Berlin Heidelberg New York, pp 685-719

Goldsmith TH, Wehner R (1977) Restrietions of rotation al and translational diffusion of pigment in the membranes of a rhabdomeric photoreceptor. J Gen Physiol 70:453-490

Goloub P, Deuze JL, Herman M, Fouquart Y (1994) Analysis of the POLDER polarization measurements performed over cloud covers. IEEE Trans Geosci Rem Sens 32:78-88

Goodyear CP, Bennett DH (1979) Sun compass orientation of immature bluegill. Trans Am Fish Soc 108:555-559

Gould JL, Dyer FC, Towne WF (1985) Recent progress in the study of the dance language. In: Hölldobler B, Lindauer M (eds) Experimental behavioural ecology. Fortsehr Zool 31:141-161

Gourley EV (1974) Orientation of the gopher tortoise Gopherus polyphemus. Anim Behav 22:158-169

Görner P, Claas B (1985) Homing behaviour and orientation in the funnel-web spider, Agelena labyrinthica Clerck. In: Barth FG (ed) Neurobiology of arachnids. Springer, Berlin, pp 275-297

Gregory PT, Macartney JM, Larsen KW (1987) Spatial patterns and movements. In: Seigel RA, Collins JT, Novak SS (eds) Snakes: ecology and evolutionary biology. Macmillan, New York, pp 366-395

Gribakin FG, Govardovskii VI (1975) The role of the photoreceptor membrane in photoreceptor optics. In: Snyder AW, Menzel R (eds) Photoreceptor optics. Springer, Berlin Heidelberg New York, pp 215-236

Gribakin FG, Ukhanov KY (1993) Effect of light scattering on visual input in arthropods. In: Wiese K, Gribakin FG, Popov AV, Renninger G (eds) Sensory systems of arthropods. Birkhäuser, Basel, pp 110-118

Gribakin FG, Vishnevskaya TM, Polyanovskii AD (1980) Polarization and spectral sensitivity of single photoreceptors of the domestic cricket. Neurophysio15:358-365

Gronenberg W (1996) Neuroethology of ants. Naturwissenschaften 83:15-27

References 395

Groot C (1965) Orientation in relation to plane polarized light. In: On the orientation of young sockeye salmon (Oncorhynchus nerka) during their seaward migration out of lakes. Behav, Suppll4, E J Brill, Leiden, Chap lOC, pp. 146-158

Guenther RD (1990) Modern optics. Duke Univ, John Willey and Sons, Inc Haidinger W (1844) Über das direkte Erkennen des polarisierten Lichts und der Lage

der Polarisationsebene. Annal Phys (Leipzig) 63:29-39 Haiiman JP (1982) Extremely low ambient light levels of Ascaphus truei. J Herpetol

16:83-84 Hamann B, Langer H (1980) Sehfarbstoffe im Auge des Wasserläufers Gerris lacustris 1.

Verhandl Dtsch Zool Ges, p 337 Hamasaki DI, Eder DJ (1977) Adaptive radiation of the pineal system. In: Creseittelli F

(ed) Handbook of sensory physiology: the visual system in vertebrates. Springer, Berlin Heidelberg New York, VII/5, pp 498-548

Hamilton WD (1971) Geometry for the selfish herd. J Theor Bio131:295-311 Hanlon RT, Maxwell MR, Shashar N, Loew ER, Boyle KL (1999) An ethogram of body pat

terning behavior in the biomedically and commercially valuable squid Loligo pealei off Cape Cod, Massachusetts. Biol BuIl197:49-62

Hanlon RT, Messenger JB (1996) Cephalopod behaviour. Cambridge Univ Press, Cambridge

Hannemann D, Raschke E (1974) Measurements of the elliptical polarization of sky radiation: preliminary results. In: Gehreis T (ed) Planets, stars and nebulae studied with photopolarimetry. Univ Arizona Press, Tucson,Arizona, pp 510-513

Hardie RC (1984) Properties of photoreceptors R7 and R8 in dorsal marginal ommatidia in the compound eyes of Musca and Calliphora. J Comp Physiol A 154:157-165

Hardie RC (1985) Functional organization of the fly retina. In: Autrum H et al (eds) Progress in sensory physiology. Springer, Berlin Heidelberg New York, vo15, pp 2-79

Hardie RC, Duelli P (1978) Properties of single cells of posterior lateral eyes of jumping spiders. Z Naturforsch 33 c:156-158

Hardie RC, Franceschini N, McIntyre PD (1979) Electrophysiological analysis of fly retina. H. Spectral and polarisation sensitivity in R7 and R8. J Comp Physiol A 133:23-39

Harker JE (1992) Swarm behaviour and mate competition in mayflies (Ephemeroptera). J Zool Lond 228:571-587

Harkness R, Wehner R (1977) Cataglyphis. Endeavour (New Series) 1(3/4):115-121 Hasler AD, Scholz AT, Horrall RM (1978) Olfactory imprinting and homing in salmon.

Am Sci 66:347-355 Hassenstein B, Reichardt W (1956) Systemtheoretische Analyse der Zeit-, Reihenfolgen

und Vorzeichenauswertung bei der Bewegungsperzeption des Rüsselkäfers Chlorophanus. Z Naturforsch llb:513-524

Haupt W (1973) Role of light in chloroplast movement. Bioscience 23:289-296 Haupt W (1991) Phytochrome and cryptochrome: co action or interaction in the control

of chloroplast orientation. In: Riklis E (ed) The science and its applications. Plenum Press, London New York, pp 479-489

Hawryshyn CW (1991a) Light-adaptation properties of the ultraviolet-sensitive co ne mechanism in comparison to the receptor mechanisms in goldfish. Visual Neurosci 6:293-301

Hawryshyn CW (1991b) Spatial summation of cone mechanisms in goldfish. Vis Res 31:563-566

Hawryshyn CW (1992) Polarization vision in fish.Am Sei 80:164-175 Hawryshyn CW, Arnold M. G., Bowering E. and Cole R. 1. (1990) Spatial orientation of

rainbow trout to plane-polarized light: the ontogeny of E-vector discrimination and spectral sensitivity characteristics. J Comp Physiol A 166:565-574

396 References

Hawryshyn CW, Bolger AE (1990) Spatial orientation of trout to partially polarized light. J Comp PhysiolA 167:691-697

Hawryshyn CW, McFarland WN (1987) Cone photoreceptor mechanisms and the detection of polarized light in fish. J Comp Physiol A 160:459-465

Hawryshyn CW, Moyer HD, Allison WT, Haimberger TJ, McFarland WN (2003) Multidimensional polarization sensitivity in damselfishes. J Comp Physiol A 189:213-220

Hazen WE, Baylor ER (1962) Behavior of Daphnia in polarized light. Biol Bull 123:243-252

Harosi FI, MacNichol EF Jr (1974) Visual pigments of goldfish cones: spectral properties and dichroism. J Gen PhysioI63:279-304

Hämmerle B, Kolb G (1987) Structure of the dorsal eye region of the moth, Adoxophyes reticulata Hb. (Lepidoptera: Tortricidae). Int J Insect Morphol EmbryoI16:121-129

Hämmerle B, Kolb G (1996): Retinal ultrasrtucture of the dorsal eye region of Pararge aegeria (Linne) (Lepidoptera: Satyridae). Int J Morphol EmbryoI25:305-315

Heisenberg M (1972) Comparative behavioral studies on two visual mutants of Drosophila. J Comp Physiol 80: 119-136

Helbig AJ (1990) Depolarization of natural skylight disrupts orientation of an avian nocturnal migrant. Experientia 46:755-758

Helbig AJ (1991a) Dusk orientation of migratory European robins, Erithacus rubecula: the role of sun-related directional information. Anim Behav 41:313-322

Helbig AJ (1991b) Experimental and analytical techniques used in bird orientation research. In: Berthold P (ed) Orientation in birds. Birkhäuser, Basel Boston Berlin, pp 270-306

Helbig AJ, Wiltschko W (1989) The skylight polarization patterns at dusk affect the orientation behaviour of blackcaps, Sylvia atricapilla. Naturwissenschaften 76:227-229

HeImholtz H von (1866) Handbuch der Physiologischen Optik. Leopold Voss, Hamburg Helversen 0 von, Edrich W (1974) Der Polarisations empfänger im Bienenauge: ein

Ultraviolettrezepor. J Comp Physiol 94:33-47 Hemenger RP (1982) Dichroism of the macular pigment and Haidinger's brushes. J Opt

Soc Am 72:734-737 Henderson ST (1970) Daylight and its spectrum. American Elsevier, New York Henton WW, Crawford FT (1966) The discrimination of polarized light by the tarantula.

Z vergl Physiol 52: 26-32 Herring PJ (1977) Bioluminescence of marine organisms. Nature 267:788-793 Herrling PL (1976) Regional distribution of three ultrastructural retinula types in the

retina of Cataglyphis bicolor Fabr. (Formicidae, Hymenoptera). Cell Tiss Res 169:247 -266

Herrnkind WF (1968) Adaptive visually-directed orientation in Uca pugilator. Am Zool 8:585-598

Herzmann D, Labhart T (1989) Spectral sensitivity and absolute threshold of polarization vision in crickets: a behavioral study. J Comp Physiol A 165:315-319

Hess P (1939) Die spektrale Energieverteilung der Himmelsstrahlung. Gerlands Beitr Geophys 55:204-220

Hochheimer BF (1978) Polarized light retinal photography of a monkey eye. Vis Res 18:19-23

Hochheimer BF, Kues HA (1982) Retinal polarization effects. Appl Opt 21 :3811-3818 Hodos W, Bessette BB, Macko KA, Weiss SRB (1985) Normative data for pigeon vision.

Vis Res 25:1525-1527 Hoff R (1985) Diurnal ultrastructural changes in the compound eye of Gryllus bimacu

latus (Orthoptera, Grylloidea) in particular the dorsal rim area. Zool Beitr NF 29:87-102

Holzworth GC, Rao CR (1965) Studies of skylight polarization. J Opt Soc Am 55:403-408

References 397

Homberg U, Paech A (2002) Ultrastructure and orientation of ommatidia in the dorsal rim area of the locust compound eye. Arthr Struct Develop 30:271-280

Homberg U, Würden S (1997) Movement-sensitive, polarization-sensitive, and light-sensitive neurons of the medulla and accessory medulla of the locust, Schistocerca gregaria. J Comp NeuroI386:329-346

Horridge GA, Marcelja L, Jahnke R, Matic T (1983) Single electrode studies on the retina of the butterfly Papilio. J Comp Physiol A 150:270-294

Horridge GA, McLean M (1978) The dorsal eye of the mayfly Atalophlebia (Ephemeroptera). Proc Roy Soc Lond B 200: 137 -150

Horvath G (1995) Reflection-polarization patterns at flat water surfaces and their relevance for insect polarization vision. J Theor BioI175:27-37

Horvath G, Barta A, Geil J, Suhai B, Haiman 0 (2002a) Ground-based full-sky imaging polarimetry of rapidly changing skies and its use for polarimetrie cloud detection. Appl Opt 41:543-559

Horvath G, Bermith B, Molnar G (1998a) Dragonflies find crude oil visually more attractive than water: multiple-choice experiments on dragonfly polarotaxis. Naturwissenschaften 85:292-297

Horvath G, Bernath B, Suhai B, Barta A, Wehner R (2002b) First observation of the fourth neutral polarization point in the atmosphere. J Opt Soc Am A 19:2085-2099

Horvath G, Clarkson ENK, Pix W (1997b) Survey of modern counterparts of schizochroal trilobite eyes: structural and functional similarities and differences. Hist BioI12:229-263

Horvath G, Gal J, Labhart T, Wehner R (2002 c) Does reflection polarization by plants influence colour perception in insects? Polarimetrie measurements applied to a polarization-sensitive model retina of Papilio butterflies. J Exp BioI205:3281-3298

Horvath G, Geil J, Pomozi I, Kriska G, Wehner R (2001) Polarized glance at the sky during a total solar eclipse: the peculiar polarization patterns of the sky-dome and the solar corona on 11 August 1999. Fizikai Szemle (Physical Review) 51:229-238 (in Hungarian)

Horvath G, Gal J, Pomozi I, Wehner R (1998b) Polarization portrait of the Arago point: video-polarimetrie imaging of the neutral points of skylight polarization. Naturwissenschaften 85:333-339

Horvath G, Gal J, Wehner R (1997a) Why are water-seeking insects not attracted by mirages? The polarization pattern of mirages. Naturwissenschaften 84:300-303 [Erratum 85:90 (1998) 1

Horvath G, Pomozi I (1997) How celestial polarization changes due to reflection from the deflector panels used in deflector loft and mirror experiments studying avian navigation. J Theor BioI184:291-300

Horvath G, Pomozi I, Geil J (2003) Neutral points of skylight polarization observed during the total eclipse on 11 August 1999. Appl Opt 42:465-475

Horvath G, Varju D (1991) On the structure of the aerial visual field of aquatic animals distorted by refraction. Bull Math BioI53:425-441

Horvath G, Varju D (1995) Underwater refraction-polarization patterns of skylight perceived by aquatic animals through Snell's window of the flat water surface. Vis Res 35:1651-1666

Horvath G, Varju D (1997) Polarization pattern of freshwater habitats recorded by video polarimetry in red, green and blue spectral ranges and its relevance for water detection by aquatic insects. J Exp BioI200:1155-1163

Horvath G, Wehner R (1999) Skylight polarization as perceived by desert ants and measured by video polarimetry. J Comp Physiol A 184: 1-7 [Erratum 184:347-349 (1999) 1

Horvath G, Zeil J (1996) Kuwait oillakes as insect traps. Nature 379:303-304 Höhn DH, Maffeo GF (1967) Spektrale Strahldichte des klaren und des bedeckten

Nachthimmels im Wellenlängenbereich von 0.5 bis 1.1 pm. Infrared Phys 7:221-245

398 References

Hunter DG, Patel SN, Guyton DL (1999) Automated detection of foveal fixation by use of retinal birefringence scanning.Appl Opt 38:1273-1279

Hyatt MB (1993): The use of sky polarization for migratory orientation by Monarch butterflies. PhD thesis, Univ Pittsburgh, USA

Ivanoff A (1974) Polarization measurements in the sea. In: Jerlov NG, Steemann-Nielsen E (eds) Optical aspects of oceanography. Academic Press, London, pp 151-175

Ivanoff A, Waterman TH (1958a) Elliptical polarization of submarine illumination. J Mar Res 16:255-282

Ivanoff A, Waterman TH (1958b) Factors, mainly depth and wavelength, affecting underwater polarized light. J Mar Res 16:283-307

Jander R, Waterman TH (1960) Sensory discrimination between polarized light and light intensity patterns by arthropods. J Cell Comp PhysioI56:137 -160

Jander R, Daumer K, Waterman TH (1963) Polarized light orientation by two Hawaiian decapod cephalopods. Z vergl PhysioI46:382-395

Järvilehto M, Moring J (1974) Polarization sensitivity of individual retinula cells and neurons of the fly Calliphora. J Comp PhysioI91:387-397

Järvilehto M, Moring J (1976) Spectral and polarization sensitivity of identified retinula cells of the fly. In: Zettler F, Weiler R (eds) Neural principles in vision. Springer, Berlin Heidelberg New York, pp 214-226

Jerlov NG (1976) Marine optics. Elsevier, Amsterdam Johnsen S, Widder EA (1999) The physical structure of transparency in biological tissue:

ultrastructure and the minimization oflight scattering. J Theor BioI199:181-198 Junger W (1991) Waterstriders (Gerris paludum E) compensate for drift with a discon

tinuously working visual position servo. J Comp Physiol A 169:633-639 Junger W, Dahmen HJ (1991) Response to self-motion in waterstriders: visual discrimi

nation between rotation and translation. J Comp Physiol A 169:641-646 Junger W, Varju D (1990) Drift compensation and its sensory basis in waterstriders (Ger

ris paludum E). J Comp Physiol A 167:441-446 Kaiser W (1974) The spectral sensitivity of the honeybees's optomotor walking response.

J Comp Physiol 90:405-408 Kalayjian ZK, Andreou AG, Wolff LB, Sheppard N (1996) A polarization contrast retina

that uses patterned iodine-doped PVA film. In: Proc ESSCIRC '96, Neuchatel, September 1996,pp 1181-1184

Kalmus H (1958) Responses of insects to polarized light in the presence of dark reflecting surfaces. Nature 182:1526-1527

Kalmus H (1959) Orientation of animals to polarized light. Nature 184:228-230 Kattawar GW (1994) Polarization oflight in the ocean. In: Spinrad RW, Carder KL, Perry

MJ (eds) Ocean optics. Oxford Univ Press, Oxford, pp 203-225 Kawamura G, Shigata A, Yonemori T (1981) Response of teleost to the plane of polarized

light as determined by the heart-beat rate. Bull Japan Soc Sei Study Fish 47:727-729 Kay QON, Daoud HS, Stirton CH (1981) Pigment distribution, light reflection and cell

structure in petals. Bot J Linnean Soc 83:57-84 Kelber A (1999a) Ovipositing butterflies use a red receptor to see green. J Exp Biol

202:2619-2630 Kelber A (1999b) Why'false' colours are seen by butterflies. Nature 402:251 Kelber A, Thunell C, Arikawa K (2001) Polarisation-dependent colour vision in Papilio

butterflies. J Exp BioI204:2469-2480 Kelly KM, Mote MI (1990a) Electrophysiology and anatomy of medulla interneurons in

the optic lobe of the cockroach, Periplaneta americana. J Comp Physiol A 167:745-756 Kelly KM, Mote MI (1990b) Avoidance of monochromatic light by the cockroach Peri

planeta americana. J Insect Physio136:287-291

References 399

Kennedy CH (1917) Notes on the life history and ecology of the dragonflies (Odonata) of central California and Nevada. Proc US Nat Mus 52:483-635

Kennedy CH (1938) The present status of work on the ecology of aquatic insects as shown by the work on the Odonata. Ohio J Sei 38:267-276

Kennedy JS (1951) The migration of the desert locust (Schistocerca gregaria Forsk.). I. The behaviour of swarms. II. A theory of long range migrations. Phil Trans Roy Soc Lond B 235:163-290

Kennedy D, Baylor ER (1961) Analysis of polarized light by the bee's eye. Nature 191:34-37

Kerfoot WB (1967) The lunar periodicity of Sphecodogastra texana, a nocturnal bee (Hymenoptera: Halictidae).Anim Behav 15:479-486

Kirk WDJ (1984) Ecologically selective coloured traps. Ecol Entomol 9:35-41 Kirschfeld K (1971) Aufnahme und Verarbeitung optischer Daten im Komplexauge von

Insekten. Naturwissenschaften 58:201-209 Kirschfeld K (1972) Die notwendige Anzahl von Rezeptoren zur Bestimmung der Rich

tung des elektrischen Vektors linear polarisierten Lichtes. Z Naturforsch 27b:578-579 Kirschfeld K (1973a) Optomotorische Reaktionen der Biene auf bewegte "Polarisations

Muster". Z Naturforsch 28 c:329-338 Kirschfeld K (1973b) Vision of polarised light. In: Symp papers 4th int biophys cong, int

union pure appl biophys,Acad Sei USSR, Pushchino, 19 zak 181, pp 289-296 Kirschfeld K (1973 c) Das neurale Superpositionsauge. Fortsehr Zoo121:229-257 Kirschfeld K (1988a) Navigation and compass orientation by insects according to the

polarization pattern of the sky. Z Naturforsch 43 c:467 -469 Kirschfeld K (1988b) The role of dorsal rim ommatidia in the bee's eye. Z Naturforsch

43 c:621-623 Kirschfeld K, Franceschini N (1977) Photostable pigments within the membrane of pho

toreceptors and their possible role. Biophys Struct Mechan 3:191-194 Kirschfeld K, Lindauer M, Martin H (1975) Problems of menotactic orientation accord

ing to the polarized light of the sky. Z Naturforsch 30 c:88-90 Kirschfeld K, Reichardt W (1970) Optomotorische Versuche an Musca mit linear polar

isiertem Licht. Z Naturforsch 25b:228 Kleerekoper H, Matis JH, Timms AM, Gensler P (1973) Locomotor response of the gold

fish to polarized light and its e-vector. J Comp Physio186:27-36 Kliger DS, Lewis JW, Randall CE (1990) Polarized light in optics and spectroscopy.Acad

emic Press, San Diego, USA Kolb G (1986) Retinal structure in the dorsal rim and large dorsal area of the eye of

Aglais urticae (Lepidoptera). Zoomorpholl06:244-246 Kong JA ( ed) (1990) Polarimetrie remote sensing. Progress in Electromagnetics

Research, Elsevier,Amsterdam London NewYork Kopal Z (1969) The moon. Dordrecht, Reidel, Holland Korte R (1965) Durch polarisiertes Licht hervorgerufene Optomotorik bei Uca tangeri.

Experientia 21 :98 Korte R (1966) Untersuchungen zum Sehvermögen einiger Dekapoden insbesondere

von Uca tangeri. Z Morphol Ökol Tiere 58:1-37 Kovoor J, Munoz-Cuevas A, Ortega-Escobar J (1993) Microanatomy of the anterior

median eyes and its possible relation to polarized-light reception in Lycosa tarentula (Araneae, Lycosidae). Boll Zoo160:367-375

Kovorov BG, Monchadskiy AS (1963) The possibility of using polarized light to attract insects. Entomol Rev 42:25-28

Können GP (1985) Polarized light in nature. Cambridge Univ Press, Cambridge Können GP (1987) Skylight polarization during a total solar eclipse: a quantitative

model. J Opt Soc Am A 4:601-608

400 References

Können GP, de Boer JH (1979) Polarized rainbow.Appl Opt 18:1961-1965 Kreithen ML, Keeton WT (1974) Detection of polarized light by the homing pigeon,

Columba livia. J Comp Physiol 89:83-92 Krekorian C (1977) Homing in the desert iguana, Dipsosaurus dorsalis. Herpetologica

33:123-127 Kriska G, Horvath G, Andrikovics S (1998) Why do mayflies lay their eggs en masse on

dry asphalt roads? Water-imitating polarized light reflected from asphalt attracts Ephemeroptera. J Exp BioI201:2273-2286

Kunz YW, Callaghan E (1989) Embryonic fissures in teleost eyes and their possible role in the detection of polarized light. Trans Am Fish Soc 118:195-202

Labhart T (1980) Specialized photoreceptors at the dorsal rim of the honeybee's compound eye: polarizational and angular sensitivity. J Comp Physiol A 141:19-30

Labhart T (1986) The electrophysiology of photoreceptors in different eye regions of the desert ant, Cataglyphis bicolor. J Comp Physiol A 158:1-7

Labhart T (1988) Polarization-opponent interneurons in the insect visual system. Nature 331:435-437

Labhart T (1996) How polarization-sensitive interneurones of crickets perform at low degrees of polarization. J Exp BioI199:1467-1475

Labhart T (1999) How polarization-sensitive interneurones of crickets see the polarization pattern of the sky: a field study with an optoelectronic model neurone. J Exp Biol 202:757-770

Labhart T (2000) Polarization-sensitive interneurons in the optic lobe of the desert ant Cataglyphis bicolor. Naturwissenschaften 87:l33-l36

Labhart T, Hodel B, Valenzuela I (1984) The physiology of the cricket's compound eye with particular reference to the anatomically specialized dorsal rim area. J Comp Physiol A 155:289-296

Labhart T, Keller K (1992) Fine structure and growth of the polarization-sensitive dorsal rim area in the compound eye oflarval crickets. Naturwissenschaften 79:527-529

Labhart T, Meyer EP (1999) Detectors for polarized skylight in insects: a survey of ommatidial specializations in the dorsal rim area of the compound eye. Microsc Res Techn 47:368-379

Labhart T, Meyer EP, Schenker L (1992) Specialized ommatidia for polarization vision in the compound eye of cockchafers, Melolontha melolontha (Coleoptera, Scarabaeidae). Cell Tiss Res 268:419-429

Labhart T, Petzold J (1993) Processing of polarized light information in the visual system of crickets. In: Wiese K, Gribakin FG, Popov AV, Renninger G (eds) Sensory systems of arthropods. Birkhäuser, Basel, pp 158-169

Labhart T, Petzold J, Helbling H (2001) Spatial integration in polarization-sensitive interneurones of crickets: a survey of evidence, mechanisms and benefits. J Exp Biol 204:2423-2430

Lambrinos D, Maris M, Kobayashi H, Labhart T, Pfeifer R, Wehner R (1997) An autonomous agent navigating with a polarized light compass. Adapt Behav 6: l31-161

Lambrinos D, Möller R, Labhart T, Pfeifer R, Wehner R (2000) A mobile robot employing insect strategies for navigation. Robotics Autonom Syst 30:39-64

Land EH (1951) Some aspects of the development of sheet polarizers. J Opt Soc Am 41:957-963

Land MF (1972) The physics and biology of animal reflectors. In: Butler JAV, Noble D (eds) Progress in biophysics and molecular biology. Pergamon Press, Oxford New York, pp 75-106

Land MF (1991) Polarizing the world of fish. Nature 353:118-119 Land MF (1993) The visual control of courtship behaviour in the fly Poecilobothrus

nobilitatus. J Comp Physiol A 173:595-603

References 401

Land MF, Marshall NJ, Brownless D, Cronin TW (1990) The eye-movements of the mantis shrimp Odontodactylus scyllarus (Crustacea: Stomatopoda). J Comp Physiol A 167:155-166

Landin J (1968) Weather and diurnal periodicity of flight by Helophorus brevipalpis Bedel (Col. Hydrophilidae). Opusc Entomo133:28-36

Landreth HF (1973) Orientation behavior of the rattlesnake, Crotalus atrox. Copeia 1973:26-31

Landreth HF, Ferguson DE (1967) Newts: sun-compass orientation. Science 158: 1459-1461

Laties AM, Enoch JM (1971) An analysis of retinal receptor orientation. 1. Angular relationship of neighboring photoreceptors. Invest Ophthalmol Visual Sei 10:69-77

Laties AM, Liebman PA, Campbell CEM (1968) Photoreceptor orientation in the primate eye. Nature 218: 172-173

Laughlin SB (1976) The sensitivities of dragonfly photoreceptors and the voltage gain of transduction.J Comp PhysiolA 111:221-247

Laughlin SB, McGinness S (1978) The structures of dorsal and ventral regions of a dragonfly retina. Cell Tiss Res 188:427-447

Laughlin SB, Menzel R, Snyder AW (1975) Membranes, dichroism and receptor sensitivity. In: Snyder AW, Menzel R (eds) Photoreceptor optics. Springer, Berlin Heidelberg NewYork,pp 237-262

Lawson MA (1985) Preliminary investigations into the roles of visual and pheromonal stimuli on aspects of the behaviour of the western plains garter snake, Thamnophis radix haydeni. MSc thesis, Univ Regina, Saskatchewan, Canada

Lee RL Jr (1991) What are 'all the colors of the rainbow'? Appl Opt 30:3401-3407 Lee RL Jr (1998) Digital imaging of clear-sky polarization. Appl Opt 37:1465-1476 Lee RL Jr, Fraser AB (2001) The rainbow bridge: rainbows in art, myth, and seien ce.

Pennsylvania State Univ Press Leggett LMW (1976) Polarised light-sensitive interneurones in a swimming crab. Nature

262:709-711 Lehrer M, Srinivasan MV, Zhang SW (1990) Visual edge detection in the honeybee and its

chromatic properties. Proc Roy Soc Lond B 238:321-330 Leucht T, Martin H (1990) Interactions between e-vector orientation and weak, steady

magnetic fields in the honeybee,Apis mellifica. Naturwissenschaften 77:130-133 Levorsen AI (1967) Geology of petroleum. (2nd ed, FAF Berry) Freeman, San Francisco,

London Liebman PA, Jagger WS, Kaplan MW, Bargoot FG (1974) Membrane structure changes in

rod outer segments associated with rhodopsin bleaching. Nature 251:31-37 Lighton JRB, Wehner R (1990) Ventilation patterns in thermophilie desert ants,

Cataglyphis bicolor, during rest and activity. Verh. Dtsch Zool Ges 83:620-621 Loesel R, Homberg U (2001) Anatomy and physiology of neurons with processes in the

accessory medulla of the cockroach Leucophaea maderae. J Comp Neurol 439:193-207

Loyacano HA Jr, Chappell JA, Gauthreaux SA (1977) Suncompass orientation in juvenile largemouth bass Micropterus salmoides. Trans Am Fishery Soc 106:77-79

Luria SM, Kinney JAS (1974) Linear polarising filters and underwater vision. Undersea Biomed Res 1:371-378

Luschi P, DeI Seppia C, Crosio E (1997) Orientation during short-range feeding in the crab Dotilla wichmanni. J Comp Physiol A 181:461-468

Lüdtke H (1957) Beziehungen des Feinbaues im Rückenschwimmerauge zu seiner Fähigkeit, polarisiertes Licht zu analysieren. Z vergl Physio140:329-344

Lythgoe JN (1979) The ecology of vision. Clarendon Press, Oxford Lythgoe JN, Hemmings CC (1967) Polarized light and underwater vision. Nature

213:893-894

402 Referenees

Magni F (1966) Analysis of polarized light in wolf spiders. In: Bernhard CG (ed) The funetional organization of the eompound eye. Pergamon, Oxford, pp 171-186

Mare RE, Sperling HG (1976) The ehromatie organization of the goldfish eone mosaic. Vis Res 16:1211-1224

Mariani AP, Leure-Dupree AE (1978) Photoreeeptors and oil droplet eolours in the red area ofthe pigeon retina. J Comp NeuroI182:821-838

Marshali NB (1979) Explorations in deep-sea biology. Blandford, Po oie, UK Marshall NI (1988) A unique eolour and polarization vision system in mantis shrimps.

Nature 333:557-560 Marshall I, Cronin TW, Shashar N, Land M (1999) Behavioural evidenee for polarisation

vision in stomatopods reveals a potential ehannel for eommunieation. Curr Biol 9:755-758

Marshall NI, Iones IP, Cronin TW (1996) Behavioural evidence for colour vision in stomatopod crustaceans. I Comp Physiol A 179:473-481

Marshall NI, Land MF (1993) Some optical features of the eyes of stomatopods. I. Eye shape, optical axes and resolution. I Comp Physiol A 173:565-582

Marshall NI, Land MF, King CA, Cronin TW (1991a) The compound eyes of mantis shrimps (Crustacea, Hoplocarida, Stomatopoda). I. Compound eye structure: the detection of polarized light. Phil Trans Roy Soc Lond B 334:33-56

Marshall NI, Land MF, King CA, Cronin TW (1991b) The compound eyes of mantis shrimps (Crustacea, Hoplocarida, Stomatopoda). H. Coloured pigments in the eyes of stomatopod crustaceans: polychromatic vision by serial and lateral filtering. Phil Trans Roy Soc Lond B 334:57-84

Martin GR (1991) The question of polarization. Nature 350:194 Mason WT, Fager RS, Abrahamson EW (1973) Characterization of the lipid composition

of squid rhabdom outer segments. Bioehim Biophys Acta 306:67-73 Mazokhin-Porshnyakov GA (1969) Inseet Vision. Plenum Press, NewYork Mäthger LM, Denton EI (2001) Reflective properties of iridophores and fluorescent 'eye

spots' in the loliginid squid Alloteuthis subulata and Loligo vulgaris. I Exp Biol 204:2103-2118

MeCann GD, Arnett DW (1972) Spectral and polarization sensitivity of the dipteran visual system. I Gen PhysioI59:534-558

McDonald CG, Hawryshyn CW (1995) Intraspecifie variation of spectral sensitivity in three spine stiekleback (Gasterosteus aculeatus) from different photic regimes. I Comp Physiol A 176:255-260

McIntyre P, Snyder AW (1978) Light propagation in twisted anisotropie media: applicati on to photoreceptors. I Opt Soc Am 68: 149-157

Meinecke CC (1981) The fine structure of the compound eye of the African armyworm moth, Spodoptera exempta Walk. (Lepidoptera, Noctuidae). Cell Tiss Res 216:333-347

Meinertzhagen IA, Menzel R, Kahle G (1983) The identification of spectral receptor types in the retina and lamina of the dragonfly Sympetrum rubicundulum. I Comp PhysiolA 151:295-310

Menzel R (1975) Polarized light sensitivity in arthropods. In: Evans GC et al. (eds) Light as an ecological factor H., pp. 289-303, Blackwell, Oxford

Menzel R, Blakers M (1975) Functional organization of an insect ommatidium with fused rhabdom. Cytobiolll:279-298

Menzel R, Snyder AW (1974) Polarized light deteetion in the bee, Apis mellifera. J Comp PhysioI88:247-270

Meyer EP (1984) Retrograde labelling of photoreceptors in different regions of the compund eyes ofbees and ants. I Neurocytol13:825-836

Meyer EP, Domanico V (1999) Microvillar orientation in the photoreceptors of the ant Cataglyphis bicolor. Cell Tiss Res 295:355-361

References 403

Meyer EP, Labhart T (1981) Pore canals in the cornea of a functionally specialized area of the honeybee's compound eye. Cell Tiss Res 216:491-501

Meyer EP, Labhart T (1993) Morphological specializations of dorsal rim ommatidia in the compound eye of dragonflies and damselflies (Odonata). Cell Tiss Res 272:17-22

Meyer EP, Labhart T (1998) Polarized reflections from sand grains: implications fro photoreceptor fine structure in desert ants. In: Eisner N, Wehner R (eds) Proc 26th Göttingen Neurobiol Conf. Thieme, Stuttgart New York, p 400

Meyer HW (1970) Reizwechselfrequenz und Auslösung des Beutefangs beim Bachwasserläufer (Velia caprai). Naturwissenschaften 57:313

Meyer HW (1971) Visuelle Schlüsselreize für die Auslösung der Beutefanghandlung beim Bachwasserläufer Velia caprai (Hemiptera, Heteroptera). I. Untersuchung der räumlichen und zeitlichen Reizparameter mit formverschiedenen Attrappen. Z vergl Physiol72:260-297

Meyer-Rochow VB (1973) The dioptrie system in the beetle's compound eye. In: Horridge GA (ed) The compound eye and vision of insects. Oxford University Press, Oxford, London, pp 299-310

Meyer-Rochow VB (1981) Electrophysiology and histology of the eye of the bumblebee Bombus hortorum. J Roy Soc NZld 11:123-153

Michelson AA (1911) On metallic colouring in birds and insects. Phil Magaz 21:554-567 Miller RE, Fastie WG (1972) Skylight intensity, polarization and airglow measurements

during the total solar eclipse of 30 May 1965. J Atmos Terrest Phys 34:1541-1546 Mirow S (1972) Skin color in the squids Loligo pealii and Loligo opalescens. Z Zellforsch

125:176-190 Mischke U (1984) The stimulus efficiency of intensity contrast, spectral contrast and

polarization contrast in the optomotorics of Pachnoda marginata (Coleoptera: cetoniinae). In: Varju D, Schnitzler HU (eds) Localization and orientation in biology and engineering. Springer, Berlin Heidelberg, pp 156-159

Mizera F, Bernath B, Kriska G, Horvath G (2001) Stereo videopolarimetry: measuring and visualizing polarization patterns in three dimensions. J Imag Sci Technol 45:393-399

Montgomery KC, Heinemann EG (1952) Concerning the ability of homing pigeons to discriminate patterns of polarized light. Science 116:454-456

Moody MF (1962) Evidence for the intraocular discrimination of vertically and horizontally polarized light by Octopus. J Exp BioI39:21-30

Moody MF, Par riss JR (1960) Discrimination of polarized light by Octopus. Nature 186:839-840

Moody MF, Parriss JR (1961) The discrimination of polarized light by Octopus: a behavioural and morphological study. Z vergl PhysioI44:268-291

Moore FR (1986) Sunrise, skylight polarization, and the early morning orientation of night-migrating warbiers. Condor 88:493-498

Moore FR, Phillips JB (1988) Sunset, skylight polarization and the migratory orientation of yellow-rumped warbiers, Dendroica coronata. Anim Behav 36:1770-1778

Moore JG, Rao CRN (1966) Polarization of the daytime sky during the total solar eclipse of 30 May 1965. Annal Geophys 22:147-150

Mote I, Wehner R (1980) Functional characteristics of photoreceptors in the compound eye and ocellus of the desert ant, Cataglyphis bicolor. J Comp Physiol13 7:63-71

Muller HRA (1937) Een zonderlinge vergissing van Copera marginipes. De Tropische Natuur 26:95-98

Munk 0 (1970) On the occurrence and significance of horizontal band-shaped retinal areas in teleosts. Vidensk Medd Dan Naturhist Foren 133:85-120

Munro U, Wiltschko R (1995) The role of skylight polarization in the orientation of a day-migrating bird species. J Comp Physiol A 177:357-362

404 References

Muntz WRA (1972) Inert absorbing and reflecting pigments. In: Dartnall HIA (ed) Handbook of sensory physiology, Springer, Berlin, Heidelberg, VIII 1 , pp 529-565

Murphy PA (1981) Celestial compass orientation in juvenile American alligators (Alligator mississippiensis). Copeia 1981(3):638-645

Müller M (1989) Mechanismus der Wegintegration bei Cataglyphis fortis (Hymenoptera, Insecta). PhD thesis, Univ Zurich, Switzerland

Kuwabara M, Naka K (1959) Response of a single retinula cell to polarized light. Nature 184:455-456

Naylor EI, Stanworth A (1954) Retinal pigment and the Haidinger effect. I Physiol 124:543-552

Neill SRSI, Cullen IM (1974) Experiments on wh ether schooling by their prey affects the hunting behaviour of cephalopods and fish predators. I Zooll72:549-569

Neuberger H (1950) Arago's neutral point: a neglected tool in meteorological research. BuIlAm Meteorol Soc 31:119-125

Neville AC (1960) A list of Odonata from Ghana, with notes on their mating, flight and resting sites. Proc Roy Entomol Soc Lond A 35:124-128

Neville AC, Luke BM (1971) Form optical activity in crustacean cutide. I Insect Physiol 17:519-526

Newall HF (1906) Polarization phenomena in the solar corona 1905,Aug. 30. Month Not Roy Astronom Soc 66:475-481

Newcomer RT, Taylor DH, Guttman SI (1974) Celestial orientation in two species of water snakes (Natrix sipedon and Regina septemvittata). Herpetologica 30:194-200

Nilges C (1990) Orientierung der Feldgrille, Gryllus campestris L., auf linear polarisiertes Licht. PhD thesis, Univ Munich, Germany

Nilsson DE, Labhart T, Meyer E (1987) Photoreceptor design and optical properties affecting polarization sensitivity in ants and crickets. I Comp Physiol A 161:645-658

Ninomiya N, Tominaga Y, Kuwabara M (1969) The fine structure of the compound eye of a damsel-fly. Z Zellforsch 98:17-32

North IA, Duggin MI (1997) Stokes vector imaging of the polarized sky-dome. Appl Opt 36:723-730

Novales Flamarique I (2001) Gradual and partialloss of corner cone-occupied area in the retina of rainbow trout. Vis Res 41:3073-3082

Novales Flamarique I, Browman HI (2000) Wavelength-dependent polarization orientation in Daphnia. I Comp Physiol A 186:1073-1087

Novales Flamarique I, Browman HI (2001) Foraging and prey-search behaviour of small juvenile rainbow trout (Oncorhynchus mykiss) under polarized light. J Exp Biol 204:2415-2422

Novales Flamarique I, Hawryshyn CW (1996) Retinal development and visual sensitivity of young Pacific sockeye salmon (Oncorhynchus nerka). I Exp BiolI99:869-882

Novales Flamarique I, Hawryshyn CW (1997a) Is the use of underwater polarized light by fish restricted to crepuscular time periods? Vis Res 37:975-989

Novales Flamarique I, Hawryshyn CW (1997b) No evidence of polarization sensitivity in freshwater sunfish from multi-unit optic nerve recordings. Vis Res 37:967-973

Novales Flamarique I, Hawryshyn CW (1998a) The common white sucker (Catostomus commersoni): a fish with ultraviolet sensitivity that lacks polarization sensitivity. J Comp PhysiolA 182:331-341

Novales Flamarique I, Hawryshyn CW (1998b) Photoreceptor types and their relation to the spectral and polarization sensitivities of dupeid fishes. I Comp Physiol A 182:793-803

Novales Flamarique I, Hawryshyn CW, Harosi FI (1998) Double-cone internal reflection as a basis for polarization detection in fish. I Opt Soc Am A 15:349-358

References 405

Novales Flamarique I, Harosi FI (2002) Visual pigments and dichroism of anchovy cones: a model system for polarization detection. Visual Neurosci 19:467-473

Novales Flamarique I, Hendry A, Hawryshyn CW (1992) The photic environment of a salmonid nursery lake. J Exp BiolI69:121-141

Novales Flamarique I, Oldenbourg R, Harosi FI (1995) Transmission of polarized light through sunfish double cones reveals minute optical anisotropies. Biol BuH 189:220-222

Oksche A (1965) Survey of the development and comparative morphology of the pineal organ. Progr Brain Res 10:3-29

Ortega-Escobar J, Munoz-Cuevas A (1999) Anterior median eyes of Lycosa tarentula (Araneae, Lycosidae) detect polarized light: behavioral experiments and electroretinographic analysis. J Arachnology 27:663-671

Pardi L (1954) l' orientamento diurno di Tylos latreillii (Aud. & Sav.) (Crustacea, Isopoda terrestria). Bollettino dell'Instituto e Museo di Zoologia, Universita di Torino 4:167-196

Pardi L, Papi F (1953) Ricerche sull'orientamento di Talitrus saltator (Montagu) (Crustacea,Amphipoda) 1. 1'orientamento durante il giorno in una popolazione deI litorale tirrenico. Zeitsehr vergl Physiol 13:459-489

Papi F, Tongiorgi P (1963) Innate and learned components in the astronomical orientation of wolf spiders. Ergeb Bio126:259-280

Parkyn DC, Austin JD, Hawryshyn CW (2003) Acquisition of polarized-light orientation in salmonids under laboratory conditions. Anim Behav 65:893-904

Parkyn DC, Hawryshyn CW (1993) Polarized-light sensitivity in rainbow trout (Oncorhynchus mykiss): chracterization from multi-unit responses in the optic nerve. J Comp Physiol A 172:493-500

Parkyn DC, Hawryshyn CW (2000) Spectral and ultraviolet-polarisation sensitivity in juvenile salmonids: a comparative analysis using electrophysiology. J Exp Biol 203:1173-1191

Pearce F (1995) Devastation in the desert. New Scientist 146 (No 1971):40-43 Pearce F (1998) Sunny side up. New Scientist 159 (No 2142):45-48 Pellicori SF (1971) Polarizing properties of pulverized materials with special reference to

the lunar surface. Appl Opt 10:270-285 Perez SM, Taylor OR, Jander R (1997) A sun compass in monarch butterflies. Nature 387:

29 Petzold J (2001) Polarisationsempfindliche Neuronen im Sehsystem der Feldgrille, Gryl

lus campestris: Elektrophysiologie, Anatomie und Modellrechnungen. PhD thesis, Univ Zurich, Switzerland

Pezzaniti JL, Chipman RA (1995) Mueller matrix imaging polarimetry. Opt Engin 34:1558-1568

Philipsborn A von, Labhart T (1990) A behavioural study of polarization vision in the fly, Musca domestica. J Comp Physiol A 167:737-743

Phillips JB, Deutschlander ME, Freake MJ, Borland C (2001) The role of extraocular photoreceptors in newt magnetic compass orientation: paralleis between light-dependent magnetoreception and polarized light detection in vertebrates. J Exp Biol 204:2543-2552

Phillips JB, Moore FR (1992) Calibration of the sun compass by sunset polarized light patterns in a migratory bird. Behav Ecol Sociobio131:189-193

Phillips JB, Waldvogel JA (1982) Reflected light cues generate the short-term deflectorloft effect. In: Papi F, Wallraff HG (eds) Avian navigation. Springer, Berlin Heidelberg, pp 190-202

Phillips JB, Waldvogel JA (1988) Celestial polarized light patterns as a calibration reference for sun compass of homing pigeons. J Theor Biol 131 :55-67

406 References

Pilcher CWT, Sexton DB (1993) Effects of the Gulf War oil spills and well-head fires on the avifauna and environment of Kuwait. Sandgrouse 15:6-17

Piltschikoff N (1906) Sur la polarisation du eiel pendant les eclipses du solei!. Compt Rend Sei Paris, Seance 18 juin 1906, p 1449

Plasa L (1979) Heimfindeverhalten bei Salamandra salamandra (1.). Z Tierpsych 51:113-125

Pomozi I (2002) Polarization patterns measured by wide and narrow field -of-view imaging polarimetry with atmospheric optical and biological applications. PhD thesis, Eötvös Univ, Dept Biol Phys, Biooptics Lab, Budapest, p 103 (supervisor: Horvath G) (in Hungarian)

Pomozi I, GaI J, Horvath G, Wehner R (2001a) Fine structure of the celestial polarization pattern and its temporal change during the total solar eclipse of 11 August 1999. Rem Sens Environ 76:181-201

Pomozi I, Horvath G, Wehner R (2001b) How the clear-sky angle of polarization pattern continues underneath clouds: full-sky measurements and implications for animal orientation. J Exp BioI204:2933-2942

Pomozi I, Javorfi T, Menczel L, Garab G (2003) Birefringence mapping of magnetically oriented chloroplasts, determined by a differential polarization laser scanning microscope. Biophys J (submitted)

Poo MM, Cone RA (1974) Lateral diffusion of rhodopsin in the photoreceptor membrane. Nature 247:438-441

Popham EJ (1964) The migration of aquatic bugs with special reference to the Corixidae (Hemiptera Heteroptera). Arch HydrobioI60:450-496

Popov AV (1975) Acoustic behavior and migrations of field crickets Gryllus campestris. L Zool Zh 54:1803-1809

Povel H (1995) Imaging Stokes polarimetrywith modulators and charge coupled-device image sensors. Opt Engin 34:1870-1878

Prosch T, Hennings D, Raschke E (1983) Video polarimetry: a new imaging technique in atmospheric science. Appl Opt 22: l360-1363

Puschnig R (1926) Albanische Libellen. Konowia 5:33, 1l3, 208, 313 Rao CRN, Takashima T, Moore JG (1972) Polarimetry of the daytime sky during solar

eclipses. J Atmos Terrest Phys 34:573-576 Rensing L (1962) Beiträge zur vergleichenden Morphologie, Physiologie und Ethologie

der Wasserläufer (Gerroidea). Zool Beitr 7:447-485 Rensing L, Bogenschütz H (1966) Vorzugsrichtungen von Corixa punctata Illig. in polar

isiertem Licht. Zool Jb Physiol72:123-l35 Rhode SC, Pawlowski M, Tollrian R (2001) The impact of ultraviolet radiation of the ver

tical distribution of zooplankton of the genus Daphnia. Nature 412:69-72 Ribi WA (1979) Do the rhabdomeric structures in bees and flies really twist? J Comp

Physiol134: 109-112 Ritz DA (1991) Polarized light responses in the shrimp Palaemonetes vulgaris (Say). J

Exp Mar Biol EcoI154:245-250 Ritz T,Adem S, Schulten K (2000) A model for photoreceptor-based magnetoreception in

birds. Biophys J 78:707-718 Robinson C (1966) The cholesteric phase in polypeptide solutions and biological struc

tures. Mol Cryst 1 :467 -494 Ronacher B, Wehner R (1995) Desert ants Cataglyphis fortis use self-induced optic flow

to measure distances travelled. J Comp Physiol A 177:21-27 Rossel S (1993) Navigation by bees using polarized skylight. Comp Biochem Physiol A

104:695-708 Rossel S, Wehner R (1982) The bee's map of the e-vector pattern in the sky. Proc Natl

Acad Sci USA 79:4451-4455

References 407

Rossel S, Wehner R (1984a) How bees analyse the polarization patterns in the sky. J Comp Physiol A 154:607-615

Rossel S, Wehner R (1984b) Celestial orientation in bees: the use of spectral cues. J Comp Physiol A 155:605-613

Rossel S, Wehner R (1986) Polarization vision in bees. Nature 323: 128-131 Rossel S, Wehner R, Lindauer M (1978) E-vector orientation in bees. J Comp Physiol A

125:1-12 Rost R, Honegger HW (1987) The timing of premating and mating behaviour in a field

population of the cricket Gryllus campestris L. Behav Ecol Soeiobio121:279-289 Rowe DM, Denton EJ (1997) The physical basis for reflective communication between

fish, with special reference to the horse mackerei, Trachurus trachurus. Phil Trans Roy Soc Lond B 352:531-549

Rowe MP, Engheta N, Easter SS Jr, Pugh EN Jr. (1994) Graded-index model of a fish double cone exhibits differential polarization sensitivity. J Opt Soc Am A 11 :55-70

Rowe MP, Pugh EN Jr, Tyo JS, Engheta N (1995) Polarization difference imaging: a biologically inspired technique for observation through scattering media. Opt Lett 20:608-610

Rowell CHF, Wells MJ (1961) Retinal orientation and the discrimination of polarized light by octopuses. J Exp Bio138:827-831

Rozenberg GV (1966) Twilight: a study in atmospheric optics. Plenum Press, New York Rutherford DJ, Horridge GA (1965) The rhabdom of the lobster eye. Quart J Microsc Sei

106:119-130 Sabra R, Glantz RM (1985) Polarization sensitivity of crayfish photoreceptors is corre

lated with their termination sites in the lamina ganglionaris. J Comp Physiol A 156:315-318

Saidel WM, Lettvin JY, MacNichol EF Jr (1983) Processing of polarized light by squid photoreceptors. Nature 304:534-536

Sandberg R (1988) Skylight polarization does not affect the migratory orientation of European robins. Condor 90:267-270

Santschi F (1911) Observations et remarques critiques surle mecanisme de l'orientation chez les fourmis. Rev Suisse Zool19:303-338

Santschi F (1923) L' orientation siderale des fourmis, et quelques considerations sur leurs differentes possibilites d'orientation. 1. Classification des diverses possibilities d'orientation chez les fourmis. Mem Soc Vaudoise Sci Nat 4:137-175

Savolainen E (1978) Swarming in Ephemeroptera: the mechanism of swarming and the effects of illumination and weather. Ann Zool Fennici 15:17-52

Schaerer S, Neumeyer C (1994) Motion perception in goldfish is "color blind". In: Elsner N, Breer H (eds) Proc 22nd Göttingen Neurobiol Conf. G Thieme-Verlag, Stuttgart, vol II,p 484

Schellart NAM (1990) The visual pathways and central non-tectal processing. In: Douglas RH, Djamgoz MB (eds) The visual system of fish. Chapman and Hall, NewYork, pp 345-372

Schenck H (1957) On the focusing of sunlight by ocean waves. J Opt Soc Am 47:653-657 Schinz RH (1975) Structural speeialization in the dorsal retina of the bee, Apis mellifera.

Cell Tiss Res 162:23-34 Schmidt WJ (1935) Doppelbrechung, Dichroismus und Feinbau des Außengliedes der

Sehzellen vom Frosch. Z Zellforsch 22:485-522 Schmidt-Koenig K (1979) Avian orientation and navigation. Academic Press, London Schneider F (1967) Schwärmbahnen der Maikäfer. In: Hediger H (ed) Die Strassen der

Tiere. Vieweg and Son, Braunschweig, pp 256-278 Schneider L, Langer H (1969) Die Struktur des Rhabdoms im "Doppelauge" des Wasser

läufers Gerris lacustris. Z Zellforsch 99:538-559

408 References

Scholtz CH (1989) Unique foraging behaviour in Pachysoma (= Scarabaeus) striatum Castelnau (Coleoptera: scarabaeidae): an adaptation to arid conditions? J Arid Environ 16:305-313

Schöne H (1963) Menotaktische Orientierung nach polarisiertem und unpolarisiertem Licht bei der Mangrovekrabbe Goniopsis. Z vergl Physio146:496-514

Schröer WD (1974) Zum Mechanismus der Analyse polarisierten Lichtes bei Agelena gracilens C. 1. Koch (Araneae, Agelenidae). I. Die Morphologie der Retina der vorderen Mitteaugen (Hauptaugen). Z Morphol Tiere 79:215-231

Schultze M (1869) Die Stäbchen in der Retina der Cephalopoden und Heteropoden. Arch mikroskAnat 5:1-24

Schwassmann HO, Hasler AD (1964) The role of the sun's altitude in the sun orientation of fish. Physiol Zoo137:163-178

Schwind R (1978) Visual system of Notonecta glauca: a neuron sensitive to movement in the binocular visual field. J Comp Physiol A 123:315-328

Schwind R (1980) Geometrical optics of the Notonecta eye: adaptations to optical environment and way of life. J Comp Physiol A 140:59-68

Schwind R (1983a) A polarization-sensitive response of the flying water bug Notonecta glauca to UV light. J Comp PhysiolI50:87-91

Schwind R (1983b) Zonation of the optical environment and zonation in the rhabdom structure within the eye of the backswimmer, Notonecta glauca. Cell Tiss Res 232:53-63

Schwind R (1984a) Evidence for true polarization vision based on a two-channel analyser system in the eye of the water bug, Notonecta glauca. J Comp Physiol A 154:53-57

Schwind R (1984b) The plunge reaction of the backswimmer Notonecta glauca. J Comp PhysiolA 155:319-321

Schwind R (1985a) A further proof of polarization vision of Notonecta glauca and a note on threshold intensity for eliciting the plunge reaction. Experientia 41:466-467

Schwind R (1985b) Sehen unter und über Wasser, Sehen von Wasser. Naturwissenschaften 72:343-352

Schwind R (1991) Polarization vision in water insects and insects living on a moist substrate. J Comp Physiol A 169:531-540

Schwind R (1995) Spectral regions in which aquatic insects see reflected polarized light. J Comp Physiol A 177:439-448

Schwind R (1999) Daphnia pulex swims towards the most strongly polarized light - a response that leads to 'shore flight'. J Exp Bio1202:3631-3635

Schwind R, Horv<ith G (1993) Reflection-polarization pattern at water surfaces and correction of a common representation of the polarization pattern of the sky. Naturwissenschaften 80:82-83

Schwind R, Schlecht P, Langer H (1984) Microspectrophotometric characterization and localization of three visual pigments in the compound eye of Notonecta glauca 1. (Heteroptera). J Comp Physiol A 154:341-346

Seitz G (1969) Polarisationsoptische Untersuchungen am Auge von Calliphora erythrocephala Meig. Z Zellforsch 93:525-529

Sekera Z (1957) Light scattering in the atmosphere and the polarization of skylight. J Opt Soc Am 47:484-490

Seletskaya LI (1956) Contribution to the preception of polarized light by the compound eye of the bee. Biophysica 1:155-157 (in Russian)

Seliger HH, Lall AB, Biggley WH (1994) Blue through UV polarization sensitivities in insects: optimizations for the range of atmospheric polarization conditions. J Comp Physiol A 175:475-486

Sharp WE, Silverman SM, Lloyd JWF (1971) Summary of sky brightness measurements during eclipses of the sun. Appl Opt 10: 1207 -1210

References 409

Shashar N (1995) UV vision by marine animals: mainly questions. In: Ultraviolet Radiation and Coral Reefs (D Gulko, PL Jokiel, eds.) HIMB Techn Rep 41:201-206

Shashar N, Addessi L, Cronin TW (1995 c) Polarization vision as a mechanism for detection of transparent objects. In: Gulko D, Jokiel V (eds) Ultraviolet radiation and coral reefs. HIMB Techn Rep 41:207-211

Shashar N, Borst DT,Ament SA, Saidel WM, Smolowitz RM, Hanlon RT (2001) Polarization reflecting iridophores in the arms of the squid Loligo pealeii. Biol Bull 201:267-268

Shashar N, Cronin TW (1996) Polarization contrast vision in Octopus. J Exp Biol 199:999-1004

Shashar N, Cronin TW, Johnson G, Wolff LB (1995a) Portable imaging polarized light analyzer. Proc Soc Photo-Optical Instrumentation Engineers (SPIE) (9th Meet Opt Engin Israel) 2426:28-35

Shashar N, Cronin TW, Johnson G, Wolff LB (1995b) Designs for submersible imaging polarimeters. In: Gulko D, Jokiel PL (eds) Ultraviolet radiation and coral reefs. HIMB Techn Rep 41:213-218

Shashar N, Hagan R, Boal JG, Hanlon RT (2000) Cuttlefish use polarization sensitivity in predation on silvery fish. Vis Res 40:71-75

Shashar N, Hanlon RT (1997) Squids (Loligo pealei and Euprymna scolopes) can exhibit polarized light patterns produced by their skin. Biol Bull193:207-208

Shashar N, Hanlon RT, Petz A deM (1998a) Polarization vision helps detect transparent prey. Nature 393:222-223

Shashar N, Harosi FI, Banaszak AT, Hanlon RT (1998b) UV radiation blocking compounds in the eye of the cuttlefish Sepia officinalis. Biol Bull195:187-188

Shashar N, Milbury CA, Hanlon RT (2002a) Polarization vision in cephalopods: neuroanatomical and behavioral features that illustrate aspects of form and function. Mar Freshwat Behav Physiol35:57-68

Shashar N, Rutledge PS, Cronin TW (1996) Polarization vision in cuttlefish. A concealed communication channe!? J Exp Biol199:2077-2084

Shashar N, Saidel WM, Schmolesky MT, Hanlon RT (2002b) Sensitivity to orientation of linear polarization in isolated photoreceptor cells of the squid, Loligo pealeii. Vis Res (submitted)

Shaw GE (I975a) Sky brightness and po1arization during the 1973 African eclipse. App1 Opt 14:388-394

Shaw SR (1966) Polarized light responses from crab retinula cells. Nature 211 :92-93 Shaw SR (1967) Simultaneous recording from two cells in the locust retina. Z vergl Phys

iol55:183-194 Shaw SR (1969) Sense-cell structure and interspecies comparisons of polarized-light

absorption in arthropod compound eyes. Vis Res 9:1031-1040 Shaw SR (1975b) Retinal resistance barriers and electrical lateral inhibition. Nature

255:480-483 Shimohigashi M, Tominaga Y (1991) Identification of UV, green and red receptors, and

their projection to lamina in the cabbage butterfly, Pieris rapae. Cell Tiss Res 263:49-59

Shumakov ME, Zelenov NP (1988) In: Abstr 12th Eastern Baltic Ornithol Conf, Vilnius, p 255

Shurcliff WA (1955) Haidinger's brushes and circularly polarized light. J Opt Soc Am 45:399

Shurcliff WA (1962) Polarized light: production and use. Harvard Univ Press, Cambridge, Massachusetts, USA

Shute CCD (1978) Haidinger's brushes. Vis Res 18:14-67 Sieb eck 0 (1968) "Uferflucht" und optische Orientierung pelagischer Crustaceen. Arch

Hydrobiol Suppl35:1-118

410 References

Sinsch U (1992) Amphibians. In: Papi F (ed) Anima! homing. Chapman and Hall, London, pp 213-233

Sivak JG, Mandelman T (1982) Chromatic dispersion of the ocular media. Vis Res 22:997 -1003

Sivaraman KR, Jayachandran M, Scaria KK, Babu GSD, Bagare SP, Jayarajan AP (1984) Brightness, polarization and electron density of the solar corona of 1980 February 16. J Astrophys Astron 5:149-158

Skrzipek KH, Skrzipek H (1974) Die spektrale Transmission und die optische Aktivität des dioptrischen Apparates der Honigbiene (Apis mellifera). Experientia 30:314-315

Sloan LL, Naquin HA (1955) A quantitative test for determining the visibility of the Haidinger brushes: dinical applications. Am J Ophthalmol 40:393-406

Smith FE, Baylor ER (1960) Bees, Daphnia, and polarized light. Eco141:360-363 Smith WC, Butler JF (1991) Ultrastructure of the Tabanidae compound eye: unusual fea

tures for Diptera. J Insect Physio137:287-296 Smith KC, Macagno ER (1990) UV photoreceptors in the compound eye of Daphnia

magna (Crustacea, Brachiopoda). A fourth spectral dass in single ommatidia. J Comp Physiol A 166:597-606

Smola U, Meffert P (1978) A single UV-receptor in the eye of Calliphora erythrocephala. J Comp Physioll03:353-357

Smola U, Tscharntke H (1979) Twisted rhabdomeres in the dipteran eye. J Comp Physiol A 133:291-297

Smola U, Wunderer H (1981) Fly rhabdomeres twist in vivo. J Comp Physiol A 142:43-49 Snitzer E, Osterberg H (1961) Observed dielectric waveguide modes in the visible spec

trum. J Opt Soc Am 51:499-505 Snyder AW (1973) Polarization sensitivity of individual retinula cells. J Comp Physiol

83:331-360 Snyder AW, Laughlin SB (1975) Dichroism and absorption by photoreceptors. J Comp

Physiol100:101-116 Snyder AW, Love JD (1983) Optical waveguide theory. Chapman and Hall, London Snyder AW, McIntyre P (1975) Polarization sensitivity of twisted fused rhabdoms. In:

Snyder AW, Menzel R (eds) Photoreceptor optics. Springer, Berlin Heidelberg New York, pp 388-391

Solessio E, Engbretson EA (1993) Antagonistic chromatic mechanisms in photoreceptors of the parietal eye of lizards. Nature 364:442-445

Sommer EW, Wehner R (1975) The retina-lamina projection in the visual system of the bee, Apis mellifera. Cell Tiss Res 163:45-61

Soret MJL (1888) Influence des surfaces d'eau sur la polarisation atmospherique et observation de deux points neutres a droite et a gauche de Soleil. Compt Rend 107:867-870

Southwood TRE (1966) Ecological methods. Chapman and Hall, London Speck M, Labhart T (2001) Carotinoid deprivation and rhodopsin alignment in RI-6

photoreceptors of Drosophila melanogaster. In: Elsner N, Kreutzberg GW (eds) Proc 28th Göttingen Neurobiol Conf. Thieme, Stuttgart, p 514

Srinivasan MV, Guy RG (1990) Spectral properties of movement perception in the dronefly Eristalis. J Comp Physiol A 166:287-295

Stange G (1981) The ocellar component of flight equilibrium control in dragonflies. J Comp Physiol A 141:335-347

Steiner H (1948) Die Bindung der Hochmoorlibelle Leucorrhinia dubia Vand. an ihren Biotop. Zool Jhrb, Systematik 78:65-96

Stepanov AS, Churmasov AV, Cherkashin SA (1979) Migration direction finding by pink salmon according to the sun. Soviet J Mar Bio15:92-99

Stephens GC, Fingerman M, Brown FA Jr (1953) The orientation of Drosophila to plane polarized light. Ann Entomol Soc Am 46:75-83

References 411

Sternberg K (1990) Autökologie von sechs Libellenarten der Moore und Hochmoore des Schwarzwaldes und Ursachen ihrer Moorbindung. PhD thesis, Univ Freiburg, Freiburg, Germany

Stevani CV, de Faria DLA, Porto JS, Trindade DJ, Bechara EJH (2000a) Mechanism of automotive clearcoat damage by dragonfly eggs investigated by surface enhanced Raman scattering. Polym Degrad Stabil 68:61-66

Stevani CV, Liria CW, Miranda MTM, Bechara EJH (2001) Cysteic acid is the chemical mediator of automotive clearcoat damage promoted by dragonfly eggs. J Appl Polym Sci 81:1549-1554

Stevani CV, Porto JS, Trindade DJ, Bechara EJH (2000b) Automotive clearcoat damage due to oviposition of dragonflies. J Appl Polym Sci 75:1632-1639

Stockhammer K (1956) Zur Wahrnehmung der Schwingungsrichtung linear polarisierten Lichtes bei Insekten. Z vergl Physiol 38:30-83

Stockhammer K (1959) Die Orientierung nach der Schwingungsrichtung linear polarisierten Lichtes und ihre sinnesphysiologischen Grundlagen. Ergebn BioI21:23-56

Stokes GG (1852) On the composition and resolution of streams of polarized light from different sources. Trans Cambr Phil Soc 9:233-258

Stowe S (1980) Spectral sensitivity and retinal pigment movement in the crab Leptograpsus variegatus (Fabricius). J Exp BioI87:73-98

Stowe S (1983) A theoretical explanation of intensity-independent variation of polarization sensitivity in crustacean retinula cells. J Comp PhysioI153:435-441

Strausfeld NJ, Wunderer H (1985) Optic lobe projections of marginal ommatidia in Calliphora erythrocephala specialized for detecting polarized light. Cell Tiss Res 242:163-178

Strutt JW (Lord Rayleigh) (1871) On the light from the sky, its polarisation and colour. Phil Magaz 41:107-120, 274-279

Sugawara K, Katagiri Y, Tomita T (1971) Polarized light responses from octopus single retinular cells. J Facult Sci Hokkaido Univ (Ser VI Zoology) 17:581-586

Summers DM, Friedmann GB, Clements RM (1970) Physical model for Haidinger's brush. J Opt Soc Am 60:271-272

Sweeney A, Jiggins C, Johnsen S (2003) Polarized light as a butterfly mating signal. Nature 423:31-32

Tasaki K, Karita K (1966) Intraretinal discrimination ofhorizontal and vertical planes of polarized light by Octopus. Nature 209:934-935

Taylor DH, Adler K (1973) Spatial orientation by salamanders using plane-polarized light. Science 181:285-287

Taylor DH, Adler K (1978) The pineal body: site of extraocular perception of celestial cues for orientation in the tiger salamander (Ambystoma tigrinum). J Comp Physiol A 124:357-361

Taylor DH, Auburn JS (1978) Orientation of amphibians by linearly polarized light. In: Schmidt-Koenig K, Keeton WT (eds) Animal migration, navigation and homing. Springer, Berlin Heidelberg New York, pp 334-346

Taylor DH, Ferguson DE (1970) Extraoptic celestial orientation in the southern cricket frog Acris gryllus. Science 168:390-392

Timofeeva VA (1961) On the problem of polarization oflight in turbid media. Izvestiya Akad Nauk SSSR, Geophysics 5:766-774

Timofeeva VA (1962) Spatial distribution of the degree of polarization of natural light in the sea. Izvestiya Akad Nauk SSSR, Geophysics 6:1843-1851

Timofeeva VA (1969) Plane of vibrations of polarized light in turbid media. Izvestiya Akad Nauk SSSR, Atmospheric and Oceanic Physics 5:603-607

Timofeeva VA (1974) Optics of turbid waters - results oflaboratory studies. In: Tedov N, Steeman-Nielsen E (eds) Optical aspects of oceanography. Academic Press, New York, pp 177-218

412 References

Tinbergen J, Abeln RG (1983) Spectral sensitivity of the landing blowfly. J Comp Physiol A 150:319-328

Tovee MJ (1995) Ultra-violet photoreceptors in the animal kingdom: their distribution and function. Trends Ecol Evol10:455-460

Tretzel E (1961) Biologie, Ökologie und Brutpflege von Coelotes terrestris (Wider) (Araneae,Agelenoidae). L Biologie und Ökologie. Z Morphol Ökol Tiere 49:658-745

Trujillo-Cenoz 0, Bernard GD (1972) Some aspects of the retinal organization of Sympycnus lineatus Loew (Diptera, Dolichopodidae). J Ultrastruct Res 38:149-160

Tyler JE (1963) Estimation of per cent polarization in deep oceanic water. J Mar Res 21:102-109

Tyo JS, Rowe MP, Pugh EN Jr, Engheta N (1996) Target detection in optically scattering media by polarization-difference imaging. Appl Opt 35: 1855-1870

Ugolini A, Felieioni S (1991) Celestial orientation in Gryllotalpa gryllotalpa. Physiol EntomoI16:355-360

Ugolini A, Melis C, Innocenti R, Tiribilli B, Castellini C (1999) Moon and sun compasses in sandhoppers rely on two separate chronometrie mechanisms. Proc Roy Soc Lond B 266:749-752

Ukhanov KY, Leertouwer HL, Gribakin FG, Stavenga DG (1996) Dioptrics of the facet lenses in the dorsal rim area of the cricket Gryllus bimaculatus. J Comp Physiol A 179:545-552

Umminger BL (1968) Polarotaxis in copepods LAn endogenous rhythm in polarotaxis in Cyclops vernalis and its relation to vertical migration. Biol Bull135:239-251

Umow N (1905) Chromatische Depolarisation durch Lichtzerstreuung. Phys Z 6:674-676 Varju D (1959) Optomotorische Reaktionen auf die Bewegung periodischer Hel

ligkeitsmuster: anwendung der Systemtheorie auf Experimente am Rüsselkäfer Chlorophanus viridis. Z Naturforsch 14b:724-735

Varju D, Horvath G (1989) Looking into the water with a facet eye. Biol Cybern 62:157-165

Verkhovskaya IN (1940) The influence of polarized light upon the phototaxis of certain organisms. Bull Moscow Nat Hist Soc Biol Sect 49:101-113 (in Russian)

Via SE, Forward RB Jr (1975) The ontogeny and spectral sensitivity of polarotaxis in larvae of the crab Rhithropanopeus harrisi (Gould). Biol Bull149:251-266

Vitzthum H, Müller M, Homberg U (2002) Neurons of the central complex of the locust Schistocerca gregaria are sensitive to polarized light. J Neurosci 22: 1114-1125

VOS JJ, Bouman MA (1964) Contribution of the retina to entoptic scatter. J Opt Soc Am 54:95-100

VOS HJJ, Coemans MAJM, Nuboer JFW (1995) No evidence for polarization sensitivity in the pigeon electroretinogram. J Exp BioI198:325-335

Voss KJ, Liu Y (1997) Polarized radiance distribution measurements of skylight. L System description and characterization. Appl Opt 36:6083-6094

Vukusic P, Sambles JR, Lawrence CR (2000) Colour mixing in wing scales of a butterfly. Nature 404:457

Vries HL de, Kuiper JW (1958) Optics of the insect eye.Ann NY Acad Sei 74:196-303 Vries HL de, Spoor A, Jielof R (1953) Properties of the eye with respect to polarized light.

Physica (Utrecht) 19:419-432 Wada S (1974a) Spezielle randzonale Ommatidien der Fliegen (Diptera: Brachycera):

Architektur und Verteilung in den Komplexaugen. Z Morphol Tiere 77:87-125 Wada S (1974b) Spezielle randzonale Ommatidien von Calliphora erythrocephala Meig.

(Diptera: Calliphoridae): Architektur der zentralen Rhabdomeren-Kolumne und Topographie im Komplexauge. Int J Insect Morphol EmbryoI3:397-424

Wada M, Grolig F, Haupt W (1993) Light-oriented chloroplast positioning: contribution to progress in photobiology. J Photo ehern PhotobioI17:3-25

References 413

Wagner-BoIler EC (1987) Ontogenese des peripheren visuellen Systems der Honigbiene (Apis mellifera). PhD thesis, Univ Zurich, Switzerland

Waldvogel JA (1990) The bird's eye view. Am Seient 78:342-353 Waldvogel JA, Phillips JB, Brown AI (1988) Changes in the short-term deflector loft effect

are linked to the sun compass ofhoming pigeons.Anim Behav 36:150-158 Walraven RL (1981) Polarization imagery. Opt Engin 20:14-18 Walton GA (1935) Field experiments on the flight of Notonecta maculata Fabr.

(Hemipt.). Trans Soc Brit Entomologists 2:l37-144 Warrant E, Bartsch K, Günther C (1999) Physiological optics in the hummingbird hawk

moth: a compound eye without ommatidia. J Exp Bio1202:497-511 Warrant E, Kelber A, Kristensen NP (2003) Eyes and vision. In: Kristensen NP (ed) Hand

book of zoology. Lepidoptera II, De Gruyter, Chap 12 (in press) Waterman TH (1950) A light polarization analyzer in the compound eye of Limulus. Sei

ence 111:252-254 Waterman TH (1954) Polarization patterns in submarine illumination. Science

l20:927-932 Waterman TH (1955) Polarization of scattered sunlight in deep water. Deep Sea Res 3:

426-434 Waterman TH (l960) Interaction of polarized light and turbidity in the orientation of

Daphnia and Mysidium. Z vergl Physio143:149-172 Waterman TH (1966) Mechanisms of polarized light perception. Seience 154:467-475 Waterman TH (1975) Natural polarized light and e-vector discrimination by vertebrates.

In: Evans GC, Bainbridge R, Rackham 0 (eds) Light as an ecological factor. Blackwell, Oxford, vol II, pp 305-335

Waterman TH (l981) Polarization sensitivity. In: Autrum H (ed) Handbook of sensory physiology. VII/6B Vision in invertebrates. Comparative physiology and evolution of vision in invertebrates B: invertebrate visual centers and behavior I, Springer, Berlin Heidelberg New York, pp 281-469

Waterman TH (l984) Natural polarized light and vision. In: Ali MA (ed) Photoreception and vision in invertebrates. Plenum Press, New York, pp 63-114

Waterman TH (1988) Polarization of marine light fields and animal orientation. Proc Soc Photo-Opt Instrum Engin (SPIE Ocean Optics IX) 925:432-437

Waterman TH,Aoki K (1974) E-vector sensitivity patterns in the goldfish optic tectum. J Comp Physio195:l3-27

Waterman TH, Fernandez HR (l970) E-vector and wavelength discrimination by retinular cells of the crayfish Procambarus. Z vergl Physiol 68: 154-174

Waterman TH, Forward RB Jr (l970) Field evidence for polarized light sensitivity in the fish Zenarchopterus. Nature 228:85-87

Waterman TH, Forward RB Jr (l972) Field demonstration of polarotaxis in the fish Zenarchopterus. J Exp Zoo1180:33-54

Waterman TH, Hashimoto H (l974) E-vector discrimination by the goldfish optic tecturn. J Comp Physio195:1-12

Waterman TH, Horch KW (l966) Mechanism of polarized light perception. Science 154:467-475

Waterman TH, WesteIl WE (1956) Quantitative effect ofthe sun's position on submarine light polarization. J Mar Res 15:149-169

Watson JAL (1992) Oviposition by exophytic dragonflies on vehicles. Notul Odonatol 3:l37

Weale RA (1975) A new way of testing macular (as distinct from foveal) function. In: Proc Macular Worksh, Bath, England, 27 Apr - 3 May 1975, Section II-3

Wehner R (1968) Optische Orientierungsmechanismen im Heimkehrverhalten von Cataglyphis bicolor (Formicidae, Hymenoptera). Rev Suisse Zoo175:1076-1085

414 References

Wehner R (1976) Polarized-light navigation by insects. Sci Am 235(7):106-115 Wehner R (1981) Spatial Vision in Arthropods. In: Autrum H (ed) Handbook of sensory

physiology VII/6C Comparative physiology and evolution of vision in invertebrates C: Invertebrate visual centers and behavior H, Springer, Berlin Heidelberg New York, pp 287-616

Wehner R (1982) Himmelsnavigation bei Insekten. Neurophysiologie und Verhalten. Neujahrsbl Naturforsch Ges Zürich 184:1-132

Wehner R (1983) The perception of polarised light. In: Co sens DJ, Vince-Price D (eds) The biology of photoreception. Soc Exp Biol Symp 36, Cambridge Univ Press, Cambridge, London, pp 331-369

Wehner R (1984) Astronavigation in insects.Ann Rev EntomoI29:277-298 Wehner R (1989) The hymenopteran skylight compass: matched filtering and parallel

coding. J Exp BioI146:63-85 Wehner R (1992) Arthropods. In: Papi F (ed) Animal homing. Chapman and Hall, Lon

don, pp 45-144 Wehner R (1994) The polarization-vision project: championing organismic biology. In:

Schildberger K, Elsner N (eds) Neural basis ofbehavioural adaptations. Gustav Fischer Verlag, Stuttgart Jena New York, Fortschr ZooI39:103-143

Wehner R (1997) The ant's celestial compass system: spectral and polarization channels. In: Lehrer M (ed) Orientation and communication in arthropods. Birkhäuser, Basel, pp 145-185

Wehner R (2001) Polarization vision - a uniform sensory capaeity? J Exp BioI204:2589-2596

Wehner R, Bernard GD (1980) Intracellular optical physiology of the bee's eye. H. Polarizational sensitivity. J Comp Physiol A 137:205-214

Wehner R, Bernard GD (1993) Photoreceptor twist: a solution to the false-color problem. Proc NatlAcad Sei USA 90:4132-4135

Wehner R, Bernard GD, Geiger E (1975) Twisted and non-twisted rhabdoms and their significance for polarization detection in the bee. J Comp Physiol104:225-245

Wehner R, Duelli P (1971) The spatial orientation of desert ants, Cataglyphis bicolor, before sunrise and after sunset. Experientia 27:1364-1366

Wehner R, Gallizzi K, Frei C, Vesely M (2002) Calibration processes in desert ant navigation: vector courses and systematic search. J Comp Physiol A 188:638-693

Wehner R, Lanfranconi B (1981) What do ants know about the rotation of the sky? Nature 293:731-733

Wehner R, Menzel R (1969) Homing in the ant Cataglyphis bicolor. Science 164:192-194 Wehner R, Meyer E (1981) Rhabdomeric twist in bees - artefact or in vivo structure? J

Comp PhysiolA 142:1-17 Wehner R, Müller M (1993) How do ants acquire their celestial ephemeris function?

Naturwissenschaften 80:331-333 Wehner R, Rossel S (1985) The bee's celestial compass - a case study in behavioural neu

robiology. In: Hölldobler B, Lindauer M (eds) Experimental behavioral ecology and sociobiology. Gustav Fischer Verlag, Stuttgart, New York, Fortschr ZooI31:11-53

Wehner R, Srinivasan MV (1984) The world as the insect sees it. In: Insect communicati on (T Lewis ed) Royal Entomol Soc Lond, pp 29-47, Academic Press, London

Wehner R, Strasser S (1985) The POL area of the honey bee's eye: behavioural evidence. Physiol EntomoI1O:337-349

Wehner R, Toggweiler F (1972) Verhaltensphysiologischer Nachweis des Farbensehens bei Cataglyphis bicolor (Formieidae, Hymenoptera). J Comp Physiol 77:239-255

Wehner R, Wehner S (1986) Path integration in desert ant. Approaching a long-standing puzzle in insect navigation. Monitore Zoologico Italiano (New Series) 20:309-331

Wehner R, Wehner S (1990) Insect navigation: use of maps in Ariadne's thread? Ethol Ecol EvoI2:27-48

References 415

Wehner R, Wehner S, Agosti D (1994) Patterns of biogeographic distribution of the North African desert ant, Cataglyphis Foerster 1850 (Insecta: Hymenoptera: Formieidae). Senckenberg BioI74:163-191

Weintraub J (1970) Homing in the lizard Sceloporus orcutti.Anim Behav 18:132-137 Wellington WG (1953) Motor responses evoked by the dorsal ocelli of Sarcophaga

aldrichi Parker, and the orientation of the fly to plane polarized light. Nature 172:1177-1179

Wellington WG (1974a) Changes in mosquito flight assoeiated with natural changes in polarized light. Canad EntomoI106:941-948

Wellington WG (1974b) Bumblebee ocelli and navigation at dusk. Seience 183:550-551 Westphal U (1984) Die Besiedlung künstlicher Kleingewässer in Abhängigkeit von

Fläche und Substrat. PhD thesis, Univ Marburg/Lahn, Görich & Weiershäuser, Marburg, Germany

Wetzel RG (1975) Limnology. WB Saunders and Co, Philadelphia, USA Whitehouse FC (1941) A guide to the study of dragonflies of Jamaica. Bull Inst Jamaica

Sei Ser No 3, pp 1-69 Wildermuth H (1992a) Habitate und Habitatwahl der Grossen Moosjungfer (Leucor

rhinia pectorialis) Chap.1825 (Odonata, Libellulidae). Z Ökol Natursch 1:3-22 Wildermuth H (1992b) Visual and tactile stimuli in choice of the oviposition substrates

by the dragonfly Perithemis mooma Kirby, 1899 (Anisoptera: Libellulidae). Odonatologica 21:309-321

Wildermuth H (1993) Habitat selection and oviposition site recognition by the dragonfly Aeshna juncea (L.): an experimental approach in natural habitats (Anisoptera: Aeshnidae). Odonatologica 22:27-44

Wildermuth H (1998) Dragonflies recognize the water of rendezvous and oviposition sites by horizontally polarized light: a behavioural field test. Naturwissenschaften 85:297-302

Wildermuth H, Spinner W (1991) Visual cues in oviposition site selection by Somatochlora arctica Zetterstedt (Anisoptera: Corduliidae). Odonatologica 20:357-367

Williams DS (1980) Organisation of the compound eye of a tipulid fly during the day and night. ZoomorphoI95:85-104

Wilson EO (1971) Insect Societies. Harvard Univ Press, Cambridge, Massachusetts, USA Wilson M (1978) The functional organisation of locust ocelli. J Comp Physiol A

124:297-316 Wiltschko W, Wiltschko R (1991) Magnetic orientation and celestial cues in migratory

orientation. In: Berthold P (ed) Orientation in birds. Birkhäuser, Basel Boston Berlin, pp 16-37

Wiltschko W, Wiltschko R, Munro U, Ford H (1998) Magnetic versus celestial cues: cueconflict experiments with migrating silvereyes at dusk. J Comp Physiol A 182:521-529

Wolf R, Gebhardt B, Gademann R, Heisenberg M (1980) Polarization sensitivity of course control in Drosophila melanogaster. J Comp Physiol A 139:177-191

Wolff LB (1993) Polarization camera technology. Proc DARPA Image Understanding Works, April 1993, Washington DC, USA, pp 1031-1036

Wolff LB (1994) Polarization camera for computer vision with a beam splitter. J Opt Soc Am A 11:2935-2945

Wolff LB, Andreou AG (1995) Polarization camera sensors. Image Vision Comp 13:497-510

Wolken JJ (1995) Light detectors, photoreceptors, and imaging systems in nature. Chap 13, Polarized light in nature: detection by animals. Oxford Univ Press, Oxford, New York

Wolstencroft RD, Brandt JC (1974) Multicolor polarimetry of the night sky. In: Gehreis T (ed) Planets, stars and nebulae studied with photopolarimetry. Univ Arizona Press, Tucson, Arizona, pp 768-780

416 References

Wunderer H, Smola U (1982) Fine structure of ommatidia at the dorsal eye margin of Calliphora erythrocephala Meigen (Diptera: calliphoridae): an eye region specialized for the detection of polarized light. Int J Insect Morphol Embryol11:25-38

Wynberg H, Meijer EW, Hummelen JC, Dekkers HPJM, Schippers PH, Carlson AD (1980) Circular polarization observed in bioluminescence. Nature 286:641-642

Wyniger R (1955) Beobachtungen über die Eiablage von Libellula depressa L. (Odonata, Libellulidae). Mitt Entomol Ges Basel NF5:62-63

Yamaguchi T, Katagiri Y, Ochi K (1976) Polarized light responses from retinula cells and sustaining fibers in the mantis shrimp. Biol J Okayama Univ 17:61-66

YES (2001) Description of the TSI-990 Total Sky Imager, Yankee Environmental Systems, Inc., Airport Industrial Park, 101 Industrial Blvd., Turners Falls, MA 0l376, USA, http://www.yesinc.com. e-mail: [email protected]

Young SR, Martin GR (1984) Optics of retinal oil droplets: a model oflight collection and polarization detection in the avian retina. Vis Res 24:129-l37

Young SR, Taylor VA (1990) Swimming tracks in swarms of cladoceran species. Anim Behav 39:10-16

Zeil J, Hofmann M (2001) Signals from 'crabworld': cuticular reflections in a fiddler crab colony. J Exp BioI204:2561-2569

Zolotov V, Frantsevich L (1973) Orientation of bees by the polarized light of a limited area of the sky. J Comp PhysioI85:25-36

Zolotov VV, Frantsevich LI (1977) Polarized light sensitive interneurons in insects. Neurophysiol (Kiev) 9:397-401 (in Russian)

Zonana HV (1961) Fine structure of the squid retina. Johns Hopkins Hosp Bull 109:185-205

Zueva LVZ (1980) Retinal cones of the black sea anchovy Engraulis encrasicholus - an analyzer of polarized light in vertebrates. Zhurnal Evoliutsionnoi Biokhimii i Fiziologii 17:602-605

Zufall F, Schmitt M, Menzel R (1989) Spectral and polarized light sensitivity of photo receptors in the compound eye of the cricket (Gryllus bimaculatus). J Comp Physiol A 164:597-608

Zwick P (1990) Emergence, maturation and upstream oviposition flights of Plecoptera from Breitenbach, with notes on the adult phase as a possible control of stream insect populations. HydrobioI194:207-223

Subject Index

A absorption vector 130 Acheta domestica 72, 156 Acris gryllus 317 Adocopyes reticulata 165 Aedes aegypti 197 Aeshna cyanea 188 Aeshna juncea 180,201 African cichlid fish 299 Agabus bipostulatus 179,194 Agelena gracileus 243 Agelena labyrinthica 243 agelenid spiders 243 Aglais urticae 165 Alexander's band 52 Alligator missisipiensis 324 almucantar 113 Alopecose pulverulenta 245 Ambystoma tigrinum 317pp American alligators 324 American tree sparrows 330 Amphimallon solstitiale 127 Anax imperator 221,223 Anax parthenope 223 Anchoa mitchilli 300,309pp anchovies 300 ancovy fish 129,293 ancient UV level of skylight 59 Andrena 126 anemomenotactic orientation 149 Anoplognathus 174 anterior median eyes 244 Antheraea polyphenus 165 ant robot 154 ants 19,62,107,108,127,147pp Anucaena limbata 179 Apbonopelma californica 244

Aphenogaster 19,20 Apis florea 136 Apis mellifica 143 Apis mellifera 54,109,114, 124pp,

131pp, 277, 362 apposition compound eye 186,263 aquatic animals 71,95,98 Arago neutral point 23,27,33,35,80 Arctosa variana 243 Arrenurus 108 arthropod navigation 49 asphalt roads 199, 229pp Atherina 98 atmospheric disturbances 56 Australian silvereyes 330 avian double cone 353

B Babinet neutral point 23,27,33,35,80 backswimmer 287 Baethis rhodani 229 Baethis vernus 232 basopolarotaxis 108, 109 bayanchovy 309 bees 61,62,64 Bibio marci 144 Bidessus flavicollis 108 Bidessus nasutus 179 bilobed short cones 311 bioluminescence 102, 103 birefringent cuticle 102 blackcap 334,335 Black Sea anchovy 309 Blattella germanica 127 blowflies 107,144 Boehm brushes 361 Bombus terricola occidentalis 150

418

Bosmina obtusirostris 247 bradicardia 293 Brewster angle 51,88,92,101,185 Brewster neutral point 23,27,35,80 Brewster zone 88, 89 bright and dark waters 206pp brook char 306 Bufo fowleri 317 Bullacris membracoides 127 bullfrog 317 bullfrog tadpoles 321,322 bulldog ants 155 bumblebee 150 butterfish 273 butterflies 221

C cabbage butterfly 166 Calliphora erythrocephala 54,119,123,

143pp,277 Calliphora stygia 144,145 Campis radicans 369pp, 376, 437, 439,

442 Camponotus ligniperda 127 Canarian cricket 157 canopylight 55,69 Carassius auratus 297 Carcinus maenas 248,277 carotenoid eolour filters 264 Cataglyphis 19,20,49,114,18,149,150,

171 Cataglyphis bicolor 54,109,114,126,

127,128,376 Cataglyphis setipes 54 Catostomus commersoni 308 CCD (charge-coupled device) 8,12 celestial polarization 56 celestial polarization pattern 57,244 celestially guided menotactic orientation

255 Clupea harengus pallasi 309 cephalopods 267pp cephalothorax 244 Cercyon 179 Ceriodaphnia reticulata 247 Cetonia aurata 101 Chironomideae 176,192 circular polarization 8,9,102,103 Chlogomphus 199 Chrysemys picta 324 Chrysis ignita 126 Chydorus globosus 247

Subject Index

circularly polarized light 100,101,102, 103

circular polarizer 101 Cladocera 247 clock-shift experiments 340 Cloeon 179,192,221 cloud detection 41,42,43,44,45 cloudlight 55,65,66,69, cock chafers 54,101,125,173 cockroaches 53 Coenagrion mercuriale 211 Coenagrion puella 190 communication by polarized light 115 confusion colours 112 confusion states 112 Columba livia 328,341 pp copepod 250 Copera marginipes 199 Cordulia aena 190 Corixa punctata 180 Corixidae 179,192,221,226 Coronis scolopendra 264 cottonwood 55,69 crabs 276 crayfish 121,253 crickets 38,53,72,126,146,151,152,

156pp Crotalus atrox 324 erude oillakes 199 crustaceans 115, 247pp Cryptopleurum minutum 179,195 Cyclops vernalis 108,250 Cycloptiloides canariensis 157 Culex 197 cutthrout trout 306 cuttlefish 272pp cyprinids 299,314

D damselflies 215 Danaus plexippus 165 Daphnia magna 118,247,260,262,305, Daphnia pulex 108, 259pp, 363 Daphnia schodleri 260 daphnid crustaceans 107 day-migrating birds 340 Deilephila elpenor 165 Dendroica coronata 332 Dermogenys pusilus 296 desert ants 43,53,54,109,111,114,115,

125,126,147 desert locusts 53, 54

Subject Index

desert dung beetle 54 diamondback rattelsnacke 324 dichroic filters 118 dichroic ratio 253 Dinoflagellates 100 Diptera 59,123,176 Disposaurus dorsalis 324 dorsal rim area (DRA) 53,58,60,61,69,

72, 110, 117, 123, 125pp Dotilla wichmanni 257 dragonfly 125, ISO, 188pp, 199pp, 215,

221 Drassodes cupreus 54, 243pp Drosophila 107,118, 143pp, 277, 278,

376 dung 195,196,197 dung beetles 174, 175 dynamic celestial polarization pattern

112 Dytiscidae 179,192,215,221

E earthlight 27,29,30 Ecdyonorus venosus 229,232 eclipse 74 egg-laying 203 egg-laying flight 232 electrocardiography 307 elliptically polarized light 101 ellipticity of polarization 8 embryonic fissure 311 Emlen funnel cage 339 Enallagma cyathigerum 202 Engraulis encrasicholus 309pp Engraulis mordax 300,309pp Enochrus quadripunktatus 179 Epeorus silvicola 229 Ephemera danica 229,232,239 Ephemeridae 179 Ephemeropterans 221, 230pp epiphysis cerebri 325 Epiphyas postvittana 176 Epipremnum aureum 371,377,445 Erithacus rubecula 341,349 Erythromma viridulum 202 Euprimna morsei 269 Euprimna scolopes 27l European robins 341,349 E-vector compass 62 E-vector navigation 19 E-vector orientation 38 E-vector pattern(s) 62, 276pp

F Fata Morgana 92 Ficedula hypolenca 350,351 Ficus benjamina 370,441 fiddler crabs 249,257

419

field cricket 50,54,110,114,125, 160pp filter-feeding cladocerans 109 fireflies 108 fireflylarvae 100,102,103 flies 53,54,146,151,152 Formica cunicularia 127 Formica polyctena ISS Formica rufa 148 fourth neutral point 27,28,31,80 freshwater turtles 324 Fresnel formulae 91,352 Fresnel reflection 118 frog 129 fruitfly 125 full-sky imaging polarimeter/polarime

try 6,18,37,41,47,63,64 fundus oculi 359 fused rhabdoms 116

G Gallus gallus 353 garden chafers 10 1 Geolycosa godeffroyi 245 Geotrupes 162,174 Gerridae 192 Gerris lacustris 115,178, 181pp, 195,

221,278pp goldfish 172,297 Goniopsis cruentata 249 Gonodactyloidea 263 Gonodactylus chiragra 264,265 Gonodactylus oerstedii 264 Gonodactylus smithic 264 Gonodactylus ternatensis 264 Gopherus polyphemus 324 grass shrimp 109,255 great silver diving beetle 221 green sunfish 307, 312, 316 ground-beetles 215 ground-reflected light 83 Gryllotalpa gryllotalpa 156,157 Gryllus bimaculatus 72, 157pp Gryllus campestris 38, SO, 72,114,126,

128, 160pp Guignotus pusillus 179

420

H Hadrurus arizonensis 246 Haidinger brushes 355pp,361 Haliplinus lineolatus 179,194 halfbeak fish 296 Haliplidae 179,192 Haproleptoides confusa 229 harvester ants 148 heart rate 343 Heliconius cydno chioneus 169 Heliconius melpomene malleti 169 Helophorus aquaticus 194 Helophorus brevipalpis 179 Helophorus flavipes 179, 194 Helophorus griseus 179,193,194 Helophorus minutus 179 Hemicordulia tau 189 Henle nerve-fibre layer 357 heteroptera 179,215 homing pigeon 328,341 pp honeybee 53,54,73, 107,109, 111pp,

118,122, 125, 131pp, 151 housfly 108 hovering 203 Hyadella azteca 108 Hydraena 179 Hydraenidae 179,192 Hydrobius fuseipes 194 Hydrophilidae 179,192,195 Hydrophilinae 192,193 Hydrous piceus 221 Hydroporus 179 Hymenoptera 59, 123 hypsotaxis 174

ice-clouds 36 Ictionogomphus ferox 199 imaging polarimeter/polarimetry 3,4,

10,108 Indian dwarf honeybee 136 intensity gradients of skylight 62 intracranial photoreceptors 319 internal representation of the sky 139 iridophores 270,271 Ischnura elegans 188,202

J Japanese beetle 102,109 June beetle 102

K Kentucky warbier 331 kokanee 306 Kramer treadmill 169 Kurzia latissima 247

L Laccobious sinatus 179 Laecophilus minutus 179

Subject Index

lamina ganglionaris 145,147,253,255 lamina monopolar neurons 254 land tortoise 324 larval crickets 159 Laspeyresia pomonella 170 lateral eyes 246 Lazerta viridis 324 Lepidoptera 165,221 Lepomis cyanellus 306pp, 312, 314, 316 Leptodora kindtii 247 Lepomis gibbosus 307,314 Leptograpsus variegatus 253 Lestes macrostigma 202 Lethrus apterus 54, 173 Lethrus inermis 54, 173 Leucophaea maderae 54,128, 171pp Leucorrhinia pectoralis 211 Libellula depressa 190,199 Libellula quadrimaculata 190,202 Limnebius crinifer 179 limnetic habitats 194 Limnoxenus niger 179 Limulus 247 Littorina littoralis 274 Littorina littorea 274 Littorina neritoides 274 Littorina saxatilis 274 lizard 324 locomotion compensator 246 locust 150 Loligo pealei 115,269,270,272 long cones 311 Lucifer dye injections 170 Lucilia caeser 119 lycosid spiders 243 Lycosta tarentula 243,244 Lygia italica 248 Lysiosquilla scabricauda 264 Lysiosquilla sulcata 264 Lysiosquilla tredecimdentata 264 Lysiosqulloidea 263

Subject Index

M Macromia magnifica 199 Madeira cockroach 54 Manduca sexta 165 Manica rubida 127 mantis shrimp 71 Mantis religiosa 221 March fly 144 mayfly 221 Megasternum boletophagum 179,195 Melanophila accuminata 127 Melolontha melolontha 53,54,101,126,

173,174 Messor 19,20,148 menopolarotactic behaviour 109 mesopelagic fish species 115 Mie scattering l32 milky solutions 97 mirages 92, 93, 94 mirror camouflage 115 mobil robot 114 moist substrata 195 Mongeotia l30 Monomorium 19,20 moon compass 50 moonlight 47 mosquitoes 197,198 Mueller matrix 8 Musca domestica 54,108, 144pp muscid fly 143 Myrmecia gulosa 127,155

N negative polarization 47,80 Nemobius sylvestris 161 Neohaliplus lineatocollis 194 Nepidae 215 neutral points 23 Nicol prism 107 night-flying insects 176 night skies 47 nocturnal bees 49 nocturnal insects 50 non-biting midges 180 Nothomyrmecia macrops 127 Notonecta glauca 96, 125, 127, 178,

183pp, 190, 195 Notonecta maculata 185 Notonectidae 192

o ocelli l37, 150, 151 Odonata 188pp oillake 215pp oil-trap 200

421

Oncorhynchus clarki clarki 306 Oncorhynchus mykiss 301pp, 314, 315 Oncorhynchus nerka 394,306 Onthophagus lenzii 174 open rhabdom 116,181 Ophiogompus forcipatus 199 optomotor response 197 Orconectes 121 Orthetrum cancellatum 202 Orthoptera 215 Oryctes nasicornis holdhausi 221

p

Pachnoda marginata 278 Pachysoma striatum 174,175 Pacifasticus leniusculus 254 paired cones 313 Palaemonetes vulgaris 109, 255pp Pamphilius inanitus 126 Panopeus herbstii 252,253 Pantala flavescens 199 Papilio 122 Papilio aegeus 165pp,364 Papilio xuthus 166pp,364pp Papillia japonica 102, 109 Pararage aegeria 165 Parastizopus armaticeps 53,54,127,174 Pardosa prativaga 245 Paradosa tristis 245 Paroroctonus mesaensis 246 path-integration 148 Passerculus sandwichensis 328,335 Periplaneta americana 128, 171pp Pemphredon unicolor 126 Peprilus triachanthus 273 Perithemis mooma 191 Phaemecia sericata 144 Phaenicia sericata 277? Pholidoptera griseaptera 127 phototactic escape responses 61 phototactic state 255 Photuris lucicrescens 102 Photuris versicolor 102 Photurus pennsylvanicus 108 Phyllopertha horticola 101 pied flycatchers 350,351 Pieris rapae 166

422

pineal body 317,325 plastie sheets 223pp,229 Platycnemis pennipes 211 Pleidae 179,192 Plexippus 244 Plusiotis resplendens 101 Poecilobothrus nobilitatus 115 point -souree polarimeter 10 point-souree seanning polarimeter 97 point-souree sequential polarimeter 98 polarimetrie cloud deteetion 43 polarizational eamera 12 polarization eompass 39,65,164 polarization-indueed false eolour 124,

156, 168, 169, 264, 300, 362pp, 370, 378,

polarization-indueed optomotor response 249

polarization opponeney 156,264 polarization-opponent neuron 1l0, 114,

162,163 polarization-sensitive horizon deteetor

190 polarization -sensitive interneurons

128,154,156,248 polarization sensitivity in plants 130 polarized refleetanee 15,17 polarizing light traps 177 polarotactic water detection 178 polarotaxis 178, 236pp plunge reaetion 185 POLDER 15 Poliergus rufesceus 127 Polistes gallicus 126 Populus deltoides 69 pools of petroleum 199 positive polarization 28, 30 posteclipse 75,78 Potamonectes 179 preeclipse 75,78 Procambarus 121,253 Pseudochilla ciliata 264 Pseudogenia carbonaris 126 Pseudotrocheus macrphthalmus 299 Pteronemobius lineolatus 161 Ptylogyna spectabilis 144 pumpkinsed sunfish 314 Pyrrhosoma nymphulla 190

Q quarter-wavelength retarder 101

Subject Index

R radiometrie cloud deteetion 43 rainbow 51 rainbow light 51 rainbow trout 301 pp, 314 Rana catesbiana 317,321,322 Ranatra linearis 221 Rayleigh pattern 113 Rayleigh seattering 18,132,137 Rayleigh skylight 61,78,79,80,88 red-spotted newts 320 refraetion-polarization pattern 8 refractive eone 95 Regina septemvittata 324 reptiles 329pp retarder 163 rhabdomerie twist 123 Rhantus pulverosus 179 Rhitrogena semicolorata 229,232 Rhitropanopeus harrisi 251,252 Rhizotrogus solstitialis 101 rose ehafers 101,102

S Sacrophaga aldrichi 143 Saldidae 179,192 Saldula saltatoria 179 Salmonidae 314 Salvelinus fontinalis 306 sand fiddler erab 249 Savanna sparrows 328, 335pp Savart filter 4,75 seanning point-souree polarimetry 10 scarabaeid beetles 53, 10 1, 173pp, 278 Scarabaeus zambesianus 175 Scarab phyllophaga 102 Scarodytes halensis 179 seavenger beetle 221 Sceloporus jarrovi 324,325 Sceloporus oncutti 324 Schistocerca gregaria 54, 128, 170pp seorpions 246pp Scylla serrata 248 seeondary eyes 245,246 Seiurus noveboracenis 331 Sepia officinalis 115,272pp Sepiotentis lessoniana 269 sequential polarimetry 10, 110 shore erab 253 shore flight 251 Shurcliff brushes 361 Sida cristallina 247

Subject Index

Sigara lateralis 179 Sigara migrolineata 179 Simocephalus serrulatus 247 Simocephalus vetulus 247 skylight naviagation 62 skylight polarization 9, 18,56 simultaneous polarimetry 10,110 sleepy lizards 326 Snell window 95,96, 100 sockeye salmon 294,306 solar corona 85,86,87 solar ec1ipse 74 solar meridian 19 Solenopsis saevissima 108,148 Somathochlora arctica 190,191 Somathochlora alpestris 190 soutern cricket frog 317 Sparidae 98 Sphaerridiinae 179,192,193,195 Sphingid moth 165 Spodoptera exempta 165 spiders 53, squids 269 Spizella arborea 330 steelhead 306 stereo video polarimetry 10 Sternfolie (star foil) 3,6,7 Stokes parameter 9,80 Stokes vector 8,9, 10 stomatopod crustaceans 263,266 submarine neutral points 97 submarine polarized light field 95 summer chafers 10 1 sunlit sky 47 swallowtail butterfly 166 swarming of mayflies 231 pp Sylvia atricapilla 334 Sympetrum rubucundulum 191 Sympetrum striolatum 188, 189 Sympetrum sanguineum 200,202 Sympetrum vulgatum 221 Sympycnus lineatus 115, 182, 191

T tabanid flies 127 tail-wagging dan ce 131 Talitrus salta tor 248 tapetum lucidum 298,311 Tapinoma sessile 108 teleost fish 293pp Terrapene carolina 324 Thamnofix radix 324

Thomson scattering 85 tiger salamander 317pp Tiliqua rugosa 324,326,327 time-compensated celestial compass

249 time-compensated sun compass 348 tipulid fly 144 toad 317 torus semicircularis 303 totality 75,77,78 total reflection 100, 101 total solar ec1ipse 74,76,79 totricid moths 176 transparent aquatic animals 115 Tricoptera 170 Trichopterans 221 trichromatic colour vision 111 trilobites 118 Trionix spinifer 324 tropical butter flies 101 tropical halfbeaks 295 tropical honeybees 64 twin cones 309,312 Tylos latreillii 248

U Uca pugilator 249 Uca tangeri 249,276 Uca vomeris 249 ultraviolet paradox (UV-sky-pol para

dox) 53,56,59,64,68,132 Uma notata 324pp

423

umbral atmospheric scattering 78, 79, 80

Umow effect 83,228,345 underwater polarization patterns 96, 99 UV cones 302,303

V vector integration 148 Velia caprai 180,283 ventral POL-area (VPA) 178,187,291 Vespa crabo 126 VLSI (very large scale integration) 12

W wagging dance 141,142 waste oil resevoir 219pp water beetles 221 water bugs 96,221 water-c1ouds 36 water-dummies 192

424

water scorpion 221 water-seeking insects 71 water snakes 324 waterstrider 221,278pp wawe-retardation Melinex mylar 332 white sucker 30 white-throated sparrow 341 wolf spiders 243 wood ant 155

y

yellow-faced honeyeaters 340 yellow-rumped warbler 332,333

Subject Index

Z Zenarchopterus buffon 295 Zenarchopterus dispar 295 zenith neural point 80,83 zenit polarization 197 Zonotrichia albicollis 330,341 zoo plankton 248 Zosterops lateralis 15

Colour Illustrations

426 Colour Illustrations

A

B

c

D

E

F


colour Fig.l.l. Triplets of colour pictures of various scenes from which highly polarized light originates. The pictures are taken by a video camera through a linearly polarizing filter with three different orientations of the transmission axis shown by white or black bars. A A dark brown bottomed pond, the surface of which reflects blue light from the clear sky. B A bright yellow bottomed pond with some plants on its surface under a clear sky. C The flower of Epipremnum aureum (Aracea) possessing a shiny petal-imitating red leaf called spathe. In the background there are shiny green leaves illuminated by light of a full clear sky from above through the glass panes of a greenhouse. D Surface of a grey asphalt road under a clear sunset sky. The upper half of the road is rough and light grey, the lower half is smooth and dark grey, the left half is dry, the right half is wet. E Stripes of shiny black plastic sheets used in agriculture laid onto a plough-Iand under a clear sky. F A car under a clear sky.


colour Fig. 1.2. Triplets of colour photographs of various scenes with highly polarized light taken by a 1800 field-of-view fisheye lens through a linear polarizer with three different orientations of the transmission axis shown by white bars. The optical axis of the lens is vertical, the periphery of the circular pictures is the horizon, while the centre is the zenith (A, B) or nadir (C, D). A A dear sky at sunrise (rising sun at 3 o'dock). B A partly doudy sky with sun ocduded by a metal sheet. C Sunlight reflected from the ground and scattered in the atmosphere below a hot air balloon photographed at sunrise (rising sun at about 9 o'dock) at an altitude of 4000 m. D A dark lake with smooth surface photographed at sunset (setting sun at 6 o'dock) from a jetty.


colour video picture

B radiance I

A

c degree of li near po lariza tion p D

429

" (J.

.,: '\, . ,

, ,

. , ' , ,

: t •

• p < 15%

angle of polarizal ion CI measu red from the vert ical

colour Fig. 1.4. A Video picture of a collection of fruits and vegetables of different colours. B-D The patterns of the intensity I, degree of linear polarization p and angle of polarization a of light reflected from the collection measured by video polarimetry in the red (650 nm), green (550 nm) and blue (450 nm) ranges of the spectrum. The different numerical values of I, p and aare coded by different colour and grey tones as shown in the insets. In the a-patterns black represents areas with p < 15 %. WF white fennel root, 00 ochre orange, RT red tomato, RP red paprika and red pepper, YP yellow paprika, GP green pepper, GA green avocado, VE violet egg-fruit. (After Horvath and Varju 1997).


left eye right eye

colour video pieture

100%

p

0%

degree of linear polarization p


radiance I dark brighl

radiance I and degree oflinear polarization p

a. 00

_4rfJ0 +450

-90 +900

-1450 ! +1450

1800

radiance I and angle of polarization a


colour Fig. 1.5. The reflection-polarizational characteristics of a car with a shiny bodywork represented in parallel view stereo format and the corresponding colour and grey palettes encoding the numerical values of the degree of linear polarization p and angle of polarization a. First row Stereo pair of the colour video picture of the car. Second row Stereo pair of the p-pattern measured in the green (550 nm). Third row Stereo pair of the a-pattern at 550 nm. Fourth row Stereo pair of the combined patterns of p and radiance 1. The higher the p-value, the deeper the red hue. If I< 20 % = Ithreshold' then pis not represented by red. We used Ithreshold' otherwise, due to the inevitable small noise of p at low I-values, erratic deep red patches or pixels would occur in the picture. Fifth row Stereo pair of the combined patterns of I and a. The higher the I-value, the brighter the colours coding a. The same Ithreshold is used as in row 4 to remove the noise of a at low I-values.


upwelling earthlight measured from balloon at an altitude of 3500 m

~- . ~ :::l

Ü ."'-... :::l o "0 u

. •

,

radiance I

•

- l . ~/


o\"Cru pos.cd

180" angle of pohni?;:nion CL

mcasurcd from Ihe loe. I merid i" n

colour Fig.4.3. 1800 field-of-view photograph of the landscape below the gondola of the hot air balloon (A) and the patterns of radianee I (B, E, H), degree of linear polarization p (e, F, I) and angle of polarization a (D, G, J) of upwelling earthlight. Measurements were taken by using 1800 field-of-view imaging polarimetry at an altitude of 3500 m and a solar elevation of 20 at 450, 550 and 650 nm immediately after loeal sunrise (05:12; loeal summer time = UTC+2; 28 June 2001, Hungary). The position of the sun and the neutral points are indicated by dots. (After Horvath et al. 2002b) .


downwelling skylight measured on the ground at sunrise

radiance I

C

E "

dcgrcc of polarization p

• angle of polarizalion Cl

mcasurcd from lhc localmcrid ian

433

eolour Fig. 4.4. As -7 eolour Fig. 4.3 for the full clear sky measured from the ground at loeal sunrise (26 August 1999,06:00 = loeal summer time = UTC+ 1, solar elevation = 0°; salt pan Chott el Djerid, Tunisia). Note that on the compass rose East and West are transposed, beeause we are looking up towards the eelestial dome rather than down towards the ground. (After Horv<ith et al. 2002b).

434

·90· +90· lIIIjk'orpol4nl"'I(Ift(1 melJW~ fffllllthe

Iac ... J nwndun


colour Fig. 4.5. Perspectivic representation of the three-dimensional spatial distribution of radiance and colour (A), degree (B) and angle (C) of linear polarization as weIl as the Arago and fourth neutral points on the surface of spheres. Similar patterns can be observed/measured around a hot air balloon in the blue (450 nm) spectral range at an altitude of 3500 m. Every sphere in the picture is the combination of the patterns in ~ colour Figs. 4.3A,C,D and 4.4A,C,D. D Portraits of Gdbor Horvdth, Baldzs Berndth, Bence Suhai and Andrds Barta, who first observed the fourth neutral point during two hot air balloon flights, the pilot of which was Attila Bakos. The 1800 field-of-view imaging polarimeter was lent by Rüdiger Wehner. (After Horv<ith et al. 2002b).


clear sky

11

12

colour photograph

degree of polarization

p

angle of polarization

a

100%

0"10


435

Rayleigh sky cloudy sky degree of

polarization p

o D


a

E

eolour degree of photograph polarization

p


a

_ overe.xposure

angle of polarizalion a measured from lhe

local meridian

colour Fig. 6.1. A-C Spatial distribution of radiance and colour, degree of linear polarization p and angle of polarization a (from the local meridian) over the entire dear sky measured by full-sky imaging polarimetry at 450 nm for different hourly positions of the sun - from sunrise (06:00 h = UTC+ 1) to no on (12:00 h) - on 26 August 1999 in the Tunisian Chott el Djerid. D, E Patterns of p and a of skylight calculated using the singlescattering Rayleigh model for the same solar positions as those in A-C. F-H Patterns for doudy skies at different places in Tunisia between 27 August 1999 and 4 September 1999 measured at 450 nm for approximately the same solar zenith angles as those in A-C. The positions of the sun are indicated by dots. The radial bar in the pictures is the wire of the sun occulter. Bast is on the left of the compass rose, because we are looking up through the sky-dome rather than down onto a map. (After Pomozi et al. 2001b).

436

clear sky


cloudy sky ~------~ ~----~~

c10uds

c1ear ky

useful for navigation

inappropriate to navigation

useful for navigation inappropriate to navigation

overexpo ed region

colour Fig. 6.2. The dear (A) and doudy (B) sky shown in row 1 of ~ colour Fig. 6.1A and in row 1 of ~ colour Fig. 6.1F, respectively. C, D Regions of the dear (C) and doudy (D) sky with polarization patterns useful for or inappropriate to reliable cricket navigation calculated on the basis of the celestial polarization patterns measured by full-sky imaging polarimetry at 450 nm. Blue (useful for navigation) regions of the dear sky where the degree of linear polarization p > 5 %. Yellow (inappropriate for navigation) regions of the dear skywhere p ~ 5 %. Green (useful for navigation) regions of the douds where p > 5 % and lac/earsky - ac/oudsl ~ 6.5°, where ais the angle of polarization. Red (inappropria te for navigation) regions of the douds where p ~ 5 % and/or lac/earsky - ac/oudsl > 6.5°. Black region of the sky where the photo emulsion was overexposed. The numerical values of p, ac/ear sky and ac/ouds originate from quantitative full-sky measurements. (After Pomozi et al. 2001b).


. -. --- .. . - ,

• . , . ...... '" - ..

.... \. '... '" . , . . . . , , , ~ " .

• ..... I . . .

~ "'- ~

. . "" .

.... ", , J . . . ,.

D §ca

',p u

437

100%

0%

.~ .~ -90' 90' '" > 8..5 '0 Ei .2,.g .p < 10% ~ö

~ colour Fig. 33.2. A Colour picture of red flowers and green leaves of Campsis radicans (trumpet vine, Bigniniaceae). B-D Patterns of intensity I, degree of linear polarization p and angle of polarization a (from the vertical) of the plant surfaces in A measured by video polarimetry at 650, 550 and 450 nm. Number of pixels = 560x736 = 412160. In row D regions are black where p < 10 %. (After Horveith et al. 2002 c).


real colours perceived by a polarization-blind retina with PR=PG=PB=l; ßR' ßG> ßB= arbitrary

B colour triangle R eolom di 'tribu tion of the flli l pielure

B

c

D

E

polarization-induced [alse colours perceived by a polarization-sensitive retina with

PR=PG=PB=2; ßR=145°, ßG= 35°, ßB= 0°

directioll or the cyc's

dorso-vcnt ra I mcridi,1Il


~ colour Fig. 33.3. A Left Equilateral R-G-B colour triangle filled with the isoluminant colour shades used. Middle Real colours of Campsis radicans in -t colour Fig. 33.2A as perceived by a polarization-blind retina with polarization sensitivity PR = P G = PB = 1 and microvillar directions ßR' ßG' ßB = arbitrary. Right Relative frequency distribution of perceived colours (MR, MG' MB) within the colour triangle calculated for the full rectangular picture. B-E Polarization-induced false colours of C. radicans perceived by a polarization-sensitive retina with PR = P G = PB = 2, ßR = 145°, ßG = 35°, ßB = 0° and their relative frequency distribution in the colour triangle as a function of the alignment X of the eye's dorso-ventral symmetry plane (indicated by red arrows in the circular insets) measured from the vertical. Note that the isoluminant rectangular images and the isoluminant colour triangle on the left in row A give information on colour alone; intensity information is missing. (After Horvath et al. 2002 c).

440

colour video piclurc ( intensity and real colour)

A

ASM ,unlit

B

ASM shady

c WPSM ,unlit

D

wrSM shady

E

EP M sunlit

F

EPSM shady

G

SM sunlit

H

SM hady

2 degl'\.'<! of

polurizmion p

103

3 "ngle of

polarizalion a


horilC>n

IllignulC'ntCtund ili ~ar'ldllnJ. ~ ... in1 ion

WPSM l!PS\I

97'±22'

... ic,"ing din."C,.1ion orlhc (Ufltl!'rn lW.'Cn rml11 l100,'C

100%

p

0%

• ovcrcxposurc

a

13

-90· +90·

+135· 180·

• ovcrexposurc


~ colour Fig. 33.7. Spectral and reflection-polarizational characteristics of a leaf of a Ficus benjamina tree (weeping fig, Ficaceae) as functions of the illumination conditions in the open. The leaf was mounted in front of the camera on a horizontal rod, which rotated in a horizontal plane along a vertical axis together with the camera (insets 11 and 12). The solar elevation was es = 55° and the leaf was illuminated by direct sunlight (A,C,E,G) or shaded with a small screen (B,D,F,H) which just occluded the sun and exposed the leaf to the full clear sky. In the sm all rectangular left and right windows, the leaf blade is approximately horizontal and vertical, respectively. Inset I3 shows the four different horizontal directions of view of the camera with respect to the solar azimuth. ASM antisolar meridian, SM solar meridian, EPSM eastwardly perpendicular to the solar meridian, WPSM westwardly perpendicular to the solar meridian. Column 1 Colour pictures of the leaf. Column 2 Patterns of the degree oflinear polarization p of the leaf measured by video polarimetry at 450 nm. Column 3 Patterns of the angle of polarization a (from the vertical) of the leaf at 450 nm, where the average E-vector alignment of the leaf blade is represented by a double-headed solid arrow, while the standard deviations are shown by double-headed dashed arrows. (After Horvath et al. 2002 c).


real colours 650 nm (red) 550 nm (green) 450 nm (blue)

A

0"

.90" +90"

180"

polarization-induced false colours B preceived by a polarization-sensitive retina

PR=PG=PB=20; ßR=145°, ßG= 35", ßB= 0"

~ colour Fig. 33.9. A Reflection-polarizational characteristics of a reddish flower and a green leaf of Campsis radicans measured by video polarimetry in the red, green and blue. B Brightness and polarization-induced false colours of the same plant perceived by a highly polarization-sensitive retina with PR == PB == PG == 20, ßR == 145°, ßG == 35°, ßB == 0° as a function of the alignment X of the eye's dorso-ventral meridian with respect to the vertical. In the circular insets the red arrow shows the actual value of X.


real colours 650 nm (red) 550 nm (green) 450 nm (blue)

polarization-induced false colours B preceived by a polarization-sensitive retina

PR=PO=PB=20; ßR=145', ßo= 35', ßB= 0'

~ colour Fig. 33.10. As --7 colour Fig. 33.9 for shiny green grass leaves in a meadow.


polarization-sensitive retina polarization-blind retina

,Q

" 0

~ Ci "

Q;) '" '" " ~ OJ)

2 "d §

~ oS 0

"

polarization-sensitive retina polarization-sensitive retina

,Q

" 0

~ 0

Ci "

~ '"

g '" " Ei ..c: OJ)

'e .0 "0

" " ~ 0 ö "


;... Ö 0

~ 0

Ci "

@ '"

G) '" " ß OJ)

] "0

" " ~ 0

Ci "


~ colour Fig. 33.11. Colours as weIl as colours and brightness of Epipremnum aureum, the reflection-polarizational characteristics of which are given in -7 colour Fig. 33.2, perceived by a polarization-blind (PR = PB = P G = 1, ßR' ßG' ßB = arbitrary) and a highly polarization-sensitive (PR = PB = P G = 20,ßR = 145°,ßG = 35°,ßB = 0°) retina as a function of the alignment X of the eye's dorso-ventral meridian with respect to the vertical. In the circular insets the red arrow shows the actual value of X.

446

A

B


polarization-sensitive retina polarization-blind retina

~ 0

~ 0 ö u

g 00

'" " S ..c:: bJl

·C .D "0

" '" ~ ..s 0 u


~ 0

~ 0 ö u

G 00 00

" S ..c:: bJl

·C .D "0 §

~ 0 ö u

polarization-sensitive retina


angle of polarization a from the vertical

... ~ ... . ,.

. . ,.

- .

@

polarization-sensitive retina


--+ colour Fig. 33.12. AAs -t colour Fig. 33.11 for a beetle with shiny black carapace on a green leaf blade of Helianthus annuus. The recording was taken under a dear sky. The scene is illuminated by direct sunlight and the originally colourless (shiny black) carapace of the beetle reflects blue skylight and green light from the surrounding vegetation. Thus, the carapace has a greenish-bluish appearance, which reduces the colour contrast between the beetle and the leaf blade. This results in a moderate colour camouflage of the carapace for a polarization-blind visual system. However, this camouflage is broken for a highly polarization-sensitive visual system perceiving the striking polarizational false colours of the carapace, which differ considerably from those of the leaf blade. The polarization-induced false colours change dramatically as the eye rotates, wh ich further enhances the break of colour camouflage. B Patterns of the degree p and angle a of linear polarization of the scene measured by video polarimetry at 550 nm.

polarized light in animal vision: polarization patterns in nature

Documents