current biology 2011 garm

8/13/2019 Current Biology 2011 Garm

1/6

Current Biology 21 , 798803, May 10, 2011 2011 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2011.03.054

ReportBox Jellysh Use TerrestrialVisual Cues for Navigation

Anders Garm, 1 ,* Magnus Oskarsson, 2 ,3

and Dan-Eric Nilsson 21 Section of Marine Biology, Department of Biology, Universityof Copenhagen, Universitetsparken 15, 2100 Copenhagen ,Denmark2 Lund Vision Group, Department of Biology, Lund University,So lvagaten 35, 22362 Lund, Sweden3 Centre for Mathematical Sciences, Lund University,So lvagaten 18, 22362 Lund, Sweden

Summary

Box jellysh have an impressive set of 24 eyes of fourdifferent types, including eyes structurally similar to thoseof vertebrates and cephalopods [ 1, 2 ]. However, the knownvisual responses are restricted to simple phototaxis,shadow responses, and object avoidance responses [ 38 ],and it has been a puzzle why they need such a complex setof eyes. Here we report that medusae of the box jellyshTripedalia cystophora are capable of visually guided naviga-tion in mangrove swamps using terrestrial structures seenthrough the water surface. They detect themangrovecanopyby an eye type that is specialized to peer up through thewater surface and that is suspended such that it is con-stantly looking straight up, irrespective of the orientationof the jellysh. The visual information is used to navigateto the preferred habitat at the edge of mangrove lagoons.

Results and Discussion

In an attempt to understand the reason for the many eyes inbox jellysh, we investigated one of the two types of lenseye (the upper lens eye) and found that it is highly specializedfor looking up through the water surface. Two striking special-izations were found: (1) the eye is suspended such that itpassively orients the visual eld straight upward at all times,irrespective of the orientation of the jellysh body, and (2) thesize of the visual eld agrees closely with the angle of 97(Snells window), within which the full 180 terrestrial eld iscompressed by refraction through the water surface. Thesesurprising features strongly suggest that the upper lens eyesare involved in behaviors that exploit either celestial or terres-trial visual cues. We had observed that medusae of our study

species, Tripedalia cystophora , rapidly swim back to thepreferred habitat at the edge of mangrove lagoons after theyare transferred away from the edge, and we went on to showthat this behavior is driven by visual detection of the mangrovecanopy. We further observed that this navigation breaks downat distances predicted by the resolution of the upper lens eyesand when the canopy was obscured from sight.

BackgroundFor well over 100 years it has been known that cubomedusae,or box jellysh, possess a unique visual system [ 9, 10 ]. Theyhave four identical sensory structures, called rhopalia, each

carrying six eyes of four morphological types: the upper and

lower lens eyes, the pit eyes, and the slit eyes [ 1, 1115 ].Two of the eye types, the upper and lower lens eyes, haveimage-forming optics and resemble vertebrate and cepha-lopod eyes [ 2,16,17 ].The role ofvisionin box jellyshis knownto involve phototaxis, obstacle avoidance, and control of swim-pulse rate [ 46, 18 ], but more advanced visually guidedbehaviors have not been discovered prior to this report.

Most known species of box jellysh are found in shallowwater habitats where obstacles are abundant [ 19 ]. Medusaeof thestudy species, T. cystophora , live between theproprootsin Caribbean mangrove swamps[ 8,20 ]. Here they stay close tothe surface [ 8] to catch their prey, a phototactic copepod thatgathers in high densities in the light shafts formed by openingsin the mangrove canopy. The medusae are not found in theopen lagoons, where they risk starvation [ 5]. As a result, their habitat is a restricted zone under the mangrove canopy,typically less than 2 m wide. Here we investigate whether themedusae use vision to nd their preferred habitat at the edgeof the mangrove lagoons and to remain within it.

Vision through the Upper Lens EyesFrom earlier studies [ 2], we were intrigued by the upper lenseyes, of which there is one on each of the four rhopalia. Theseeyes point upwardin vertically oriented medusae, but becauseof the heavy crystal (statolith) and the exible stalk of therhopalium, it seemed possible, as speculated earlier [ 7], thatthey maintain this orientation even when the medusae swimwith a horizontally oriented body axis. To test this, we madeclose-up video recordings of freely swimming T. cystophoramedusae and monitored the rhopalial orientation at differentbody orientations ( Figures 1 A and 1B). This demonstratedthat the upper lens eyes and the pit eyes always point straightupward, with no observable deviation, whereas the lower lenseyes and the slit eyes constantly point obliquely downward.Observations of tethered medusae conrmed that the rhopaliamaintain a strictly vertical orientation, irrespective of the orien-tation of the bell, even when the animal is completely upsidedown (see Figure S1 available online). The muscles presentin the stalk [ 21 ] seemed to have no inuence on the verticalorientation of the rhopalium, and in the close-up video record-ings the rhopalia were never observed to actively move.

The constantly upward-pointing upper lens eye has a retinalgeometry indicating a much smaller visual eld than that of

the downward-pointing large lens eye. Using a previouslydescribed optical model of the upper eye [ 2], we determinedthe visual eld to be close to circular, with a width of 95 100( Figure 1 C; see Experimental Procedures for details). Theprecise orientation of naturally suspended rhopalia ( Figures1 A and 1B), together with the position of the upper lens eyewithin the rhopalium [ 2], allowed us to determine that thecircular visual eldis centeredon thevertical(withan estimatedaccuracy of 6 5 ). This vertically centered visual eld, of justbelow 100 , closely matches Snells window (the 97 circular window through which an underwater observer can see theentire 180 of the terrestrial world compressed by refractionas the light passes through the water surface; Figures 1 C and1D). This, along with the preference of medusae for the top10 cm of the water column, is a strong indication that the upper *Correspondence: [email protected]
http://dx.doi.org/10.1016/j.cub.2011.03.054http://dx.doi.org/10.1016/j.cub.2011.03.054http://dx.doi.org/10.1016/j.cub.2011.03.054http://dx.doi.org/10.1016/j.cub.2011.03.054mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.cub.2011.03.054http://dx.doi.org/10.1016/j.cub.2011.03.054


2/6

lens eye is indeed specialized for looking up through the water surface to exploit terrestrial or celestial visual cues.

With this result, it is tempting to speculate that the upper lens eye is used to detect the mangrove canopy throughSnells window, such that the approximately 1 cm largeanimals can nd their habitat between the mangrove proproots and remain there even in the presence of tidal or storm-water currents. To evaluate the possibility that the upper lenseye detects the position of the mangrove canopy throughSnells window, we made still pictures using a wide-anglelens looking up through Snells window in the natural habitat.The pictures were taken from just under the surface to makeSnells window cover the same area of the surface as seenby the medusae. In the pictures, it was easy to follow themangrove canopy, which shifted from covering most of Snellswindow to covering just the edge of Snells window when thecamera was slowly moved outward to about 20 m away fromthe lagoon edge ( Figure 2 ).

To determine what medusae of T. cystophora would seewith their upper lens eyes, we used the optical model [ 2] of the eye to calculate the point-spread function of the optics at

different retinal locations. Applying these point-spread func-tions to still images of Snells window in the mangrove swamp,we were able to simulate the retinal image formed in the upper lens eyes as a jellysh moves about in the mangrove lagoon.The results ( Figure 2 ) conrm that despite the severely under-focused eyes and blurred image [ 2], the approximately 5 m tallmangrove canopy can be readily detected at a distance of 4 mfrom the lagoon edge and, with some difculty, can be de-tected even at a distance of 8 m (detection depends on theamountof surface ripple and theheight of the mangrove trees).These results thus predict that if T. cystophora medusae usetheir upper lens eyes to guide them to the correct habitat atthe lagoon edge, then they would swim toward this edge if they are closer than about 8 m away from it. Also, if they arefarther out in the lagoon, surface ripple and their poor visual

resolution will prevent detection of the mangrove canopy,and the animals would not be able to determine the directionto the closest lagoon edge.

Behavioral Assessment of Visual NavigationExperiments were conducted on wild populations of T. cystophora medusae in the mangrove lagoons near LaParguera, Puerto Rico. Preliminary tests demonstrated that if jellysh were displaced about 5 m from their habitat at thelagoon edge, they rapidly swam back to the nearest edge,independent of compass orientation. To make controlledexperiments, we introduced a clear experimental tank consist-ing of a cylindrical wall and a at bottom, open upward, to thenatural habitat under the mangrove canopy. When the tankwas lled with water, it was lightly buoyant such that the wallsextended 12 cm above the external water surface, effectivelysealing off the water around the animals but without affectingthe visual surroundings. A group of medusae was releasedinthe tank, and aslongas the tankremained under the canopy,the medusae showed no directional preference but occasion-

ally bumped into the tank wall. The tank, with the trappedwater and medusae, wasthen slowly towed out into thelagoonfrom theoriginal position under themangrove canopy. In stepsof 24 m, starting at the canopy edge, the positions of themedusae within the tank were recorded by a video camerasuspended under the tank. At all positions, from the canopyedge and outward, the medusae ceased feeding and swamalong the edges of the tank, constantly bumping into it, sug-gesting that they responded to the displacement ( Figure 3 ).Most importantly, their mean swimming direction differedsignicantly from random and coincided with the directiontoward the nearest mangrove trees ( Table S1 ). This behavior was indicated already at the canopy edge but was strongestwhen the tank was placed 2 or 4 m into the lagoon ( Figure 3 ). At 8 m from the canopy edge, the medusae could still detect

Figure 1. Rhopalial Orientation and Visual Fieldof the Upper Lens Eye

(Aand B)In freely swimming medusae,the rhopa-lia maintain a constant vertical orientation. Whenthe medusa changes its body orientation, theheavy crystal (statolith) in the distal end of therhopalium causes the rhopalial stalk to bend

such that the rhopalium remains verticallyoriented. Thus, the upper lens eye (ULE) pointsstraight upward at all times, irrespective of body orientation. The rhopalia in focus are situ-ated on the far side of the medusa and have theeyes directed to the center of the animal.(C) Modeling the receptive elds of the mostperipheral photoreceptorsin theULE (the relativeangular sensitivity of all peripheral rim photore-ceptors are superimposed and normalized ac-cording to the color template). The demarcatedeld of view reveals a near-perfect match to thesize and orientation of Snells window (dashedline).(D) The visual eld of the ULE, of just below 100 ,implies that it monitors the full 180 terrestrialscene, refracted through Snells window. LLEdenotes lower lens eye. Scale bars represent5 mm in (A) and (B) and 500 mm in insets.

Visual Navigation in Box Jellysh799


3/6

Figure 2. Snells Window Seen through the Jellysh Eyes

The rst column shows wide-angle still pictures of Snells window, captured just below the surface at 0, 2, 4, 8,and 12 m from the lagoon edge. The missingpartof Snells windowoppositethe canopywas removed because it contained thephotographer. Thesecond columnshowsthe same imagesprocessed bythe optical model to mimic the view seen through the upper lens eyes of T. cystophora medusae (the gray level represents the calculated relative photoncatch in the receptor cells of the retina). The third column shows the difference between images from the experimental distances and a 20 m image. Greenindicates higherintensities at 20 m.Red lineindicates 12%contrast. Modeling of theimagein theupperlens eyes indicates that, despitethe poor resolution, jellysh vision is good enough to detect t he mangrove canopy through Snells window at 8 m, but not at 12 m.

Current Biology Vol 21 No 9800


4/6

the direction to the nearest canopy, but at 12 m they swamrandomly along the edge of the tank ( Figure 3 ).

Conclusions Visual detection of the mangrove canopy by the upper lens eyeis the only plausible explanation for the behavioral results.Chemical or mechanical cues cannot have guided themedusae because of the enclosed experimental tank. Further,because their navigational ability depends on the distance tothe mangrove trees and is not compromised by the sun beingat zenith, we can rule out celestial or other compass cues. Thepoor visibility in the turbid mangrove lagoon (


5/6

It is surprising to nd such a navigational system in ananimal as basal as a jellysh. The central nervous system of T. cystophora medusae consists of ganglia in the rhopaliaand a ring nerve connecting the rhopalia [ 2124 ]. It is not anentirely trivial task to detect the position and orientation of the border between the dark mangrove canopy and the brightsky in Snells window, but we must assume that the necessaryneural circuitry is present in the rhopalial ganglia. Even thoughthese ganglia are unusually large for a cnidarian, each onlycontains about 1000 neurons [ 22 ], and it subserves ve eyesin addition to the upper lens eye. It remains to be investigatedwhether navigational processing in these ganglia resemblesthat of bilaterian brains.

From an evolutionary viewpoint, the use of terrestrial cuesdoes not seem to be the most straightforward source of infor-mation for a marine organism, especially not for a jellysh. A possibleexplanationfor this peculiarity is that canopy naviga-tion has evolved by modication of a sun compass. Other species of box jellysh that do not live at the edge of mangrovelagoons have somewhat different upper lens eyes that havebeen implicated as specializations for detecting the solar posi-tion [ 25 ]. Scyphozoan medusae of the genus Mastigas , alongwiththe commonmoonjelly Aureliaaurita , hasalsobeenshownto migrate using the solar position [ 26, 27 ]. Here it is suggestedthat the migration helps the medusae aggregate for reproduc-tion ( Aurelia ) or helps them stay in their saline lakes ( Mastigas ).

Our work demonstrates that despite the lack of a con-ventional brain, box jellysh are able to perform seemingly

sophisticated behaviors such as navigation by terrestrialvisual cues. So far, this navigation is the only known purposeof the upper lens eye, and the lower lens eye seems to havean equally restricted use in repulsion and attraction to nearbyunderwater structures [ 3, 4 ]. Eyes supporting a single visualbehavior presumably represent an early stage in the evolution

of visual systems. Different eyes for different behaviors prob-ably require less neural processing than if information for different behaviors has to pass through the same eye. Thebox jellysh solution may thus be linked to the absence of a central brain, but it defeats the idea that a central brain isa prerequisite for advanced behavior.

Experimental Procedures

Animals All medusae used for the behavioral experiments were adult males andfemales of T. cystophora (bell height of 810 mm). They were experimentallymanipulated in their native mangrove swamp in La Parguera, Puerto Rico.For video recordings, animals were collected and lmed within 2 days.

Measurements of Eye OrientationClose-up video recordings were made to monitor the eye orientation on venonmanipulated medusae swimming freely in a small tank (10 cm height 320cm width 3 2 cmdepth). A Sony PowerHAD video camera, equipped witha Micro Nikkor 105 mm lens and set to a xed shutter time of 0.002 s, wasused for the recordings. A total of 25 min of video was analyzed. Stillpictures were taken with a Nikon D200 camera equipped with the samelens as the video camera. Rhopalial orientation was also observed on teth-ered medusae, where the bell was rotated to different orientations( Figure S1 ).

Optical Modeling A Matlab (2007a, Mathworks) application was made to convert input imagesinto retinal images in the upper eye of T. cystophora medusae. The applica-tion has a hexagonal pixel array corresponding to the approximately 400photoreceptors of the retina. Each pixel samples across a unique point-spread function corresponding to angular sensitivities derived from

a geometrical-optical model of the eye [ 2]. Diffraction can be safely ignoredbecause of the much more severe blurring due to underfocusing [ 2]. Theapplication was fed still images taken in the natural habitat by a high-deni-tion video camera equipped with a wide-angle lens ( w 120 3 80 under-water, GoPro Hero, Woodmans Lab).

Behavioral ExperimentsIn the preliminary trials, three sets of ve medusae were tested. Fivemedusae were collected and tested on the northern shore on the lagoon,ve were collected and tested on the southern shore, and ve werecollected on the northern shore but tested on the southern shore. Themedusae were tested one at a time. In each of the tank experiments, 68medusae were tested simultaneously in a round tank (diameter = 45 cm,height = 10 cm) oating with the walls extending 12 cm above the surface,effectively separating the tank water from the surrounding water. The tankswere lled with water from under the canopy, ensuring that the chemicalcomposition was that of the habitat. A standard video camera, equipped

with a sh eye objective, was mounted under the tank and recorded thebehavior of themedusae.The positionof thetank wassecured by ananchor at successively increasing distances from the canopy edge: 0 m (directlyunder the canopy edge), 2 m, 4 m, 8 m, and 12 m. Both anchor and camerawere attached to the tank with 1 mm thick wire. Every time the tank wasmoved, the medusae were manually dispersed in the tank, and their behavior was then recorded for 2.5 min. Because of the manipulation,only the last 2 min were used in the analyses. During these 2 min, the exper-imenter rested low in the water at least 4 m farther into the lagoon, beyondthe visual range of the medusae. The experimental series, including the vedistances, were repeated three times each with a new group of medusaeand with the nearest canopy in different compass bearings. The videoswere analyzed in a custom-made program for Matlab, which returnedswim trajectories and time spent in predened tank segments, witha temporal resolution of 1 s. Only medusae that could be followed the entire2 min were used in the analysis. All experiments were performed in August2007 on sunny days with the sun position close to zenith ( w 82 88 ). A nal

Figure 5. Mangrove Canopy Obscured by a White Sheet

(A) The visual scene in Snells window when a 2 3 5 m white sheet is put upabout 4 m in front of the canopy. The sheet completely blocks sight of thecanopy, and this disrupts the ability of the medusae to navigate.(B) The picture from (A) when processed by the optical model of the upper lens eye. The canopy signature is close to cancelled by the white sheet.(C) Results from the experiments with obscured canopy. The open circle istherelease point of theve meduse; thecolored arrows indicate theapprox-imate initialheading. Thecolored circlesindicate theapproximate endpointof each medusa after 2 min.

Current Biology Vol 21 No 9802


6/6

set of experiments was conducted to test above-water visual cues versusunderwater cues. Here, ve medusae were released one at a time 5 mfrom the canopy, but the canopy was visually obscured by a 2 3 5 m brightwhite sheet placed approximately 4 m from the canopy. The underwater visual scene was left intact. The approximate initial heading of the medusaewas noted, and so was the end point after 2 min of swimming.

StatisticsFor statistical analysis of the behavior, the experimental tank was dividedinto eight segments of 45 , one of which was aligned with the direction tothe nearest mangrove trees. The number of times a medusa was observedin each areawasthen used to createa vectorcorresponding to thepreferredswimming direction of the individual medusae. For each distance to thecanopy, a mean vector was calculated and tested against a random distri-bution with circular statistics in a custom-made program (H 0 : directiondoes not differ from random, n = 12, 21, 22, 15, and 14 at 0, 2, 4, 8, and 12m, respectively). If the mean vector differed signicantly from random, itwas tested against the predicted direction toward the nearest canopy(H0 : direction differs from prediction). For details on the statistics, see[28 ]. Table S1 summarizes the results of the tests.

Supplemental Information

Supplemental Information includes one table and one gure and can befound with this article online at doi:10.1016/j.cub.2011.03.054 .

Acknowledgments

We appreciate the help offered by Henrik Malm, Eric Warrant, Ronald Petie,Linda Parkefelt, and Megan OConnor (Lund University). We would also liketothank Go staNachman (Universityof Copenhagen) forhelp with thestatis-tics. Melissa Coates assisted in taking the photos of Figure S1. The workwas supported by The Carlsberg Foundation (grant 2005-1-74 to A.G.)and The Swedish Research Council (grant 2008-3503 to D.-E.N.).

Received: June 24, 2010Revised: February 3, 2011 Accepted: March 22, 2011Published online: April 28, 2011

References

1. Pearse, J.S.,and Pearse, V.B. (1978). Vision of cubomedusanjellyshes.Science 199 , 458.

2. Nilsson, D.E., Gisle n, L., Coates, M.M., Skogh, C., and Garm, A. (2005). Advanced optics in a jellysh eye. Nature 435 , 201205.

3. Hamner, W.M., Jones, M.S., and Hamner, P.P. (1995). Swimming,feeding, circulation and vision in the Australian box jellysh, Chironex eckeri (Cnidaria, Cubozoa). Mar. Freshw. Res. 46 , 985990.

4. Garm, A., OConnor, M., Parkefelt, L., and Nilsson, D.E. (2007). Visuallyguided obstacle avoidance in the box jellysh Tripedalia cystophoraand Chiropsella bronzie . J. Exp. Biol. 210 , 36163623.

5. Buskey, E.J. (2003). Behavioral adaptations of the cubozoan medusaTripedalia cystophora for feeding on copepod ( Dioithona oculata )swarms. Mar. Biol. 142 , 225232.

6. Garm, A., and Mori, S. (2009). Multiple photoreceptor systems controlthe swim pacemaker activity in box jellysh. J. Exp. Biol. 212 , 3951

3960.7. Berger, E.W. (1900). The cubomedusae. Memoirs from the Biological

Laboratory of the Johns Hopkins University 4, 183.8. Stewart, S.E. (1996). Field behavior of Tripedalia cystophora (class

Cubozoa). Mar. Freshwat. Behav. Physiol. 27 , 175188.9. Berger, E.W. (1898). Thehistologicalstructureof theeyes of cubomedu-

sae. J. Comp. Neurol. 8, 223230.10. Claus, C. (1878). Ueber Charybdea marsupialis. Arb. Zool. Inst. Univ.

Wien 1, 156.11. Laska, G., and Hu ndgen, M. (1982). Morphologie und ultrastruktur der

lichtsinnesorgane von Tripedalia cystophora conant (Cnidaria,Cubozoa). Zool. Jb. Anat. 108 , 107123.

12. Garm, A., Andersson, F., and Nilsson, D.E. (2008). Unique structure andoptics of the lesser eyes of the box jellysh Tripedalia cystophora . Vision Res. 48 , 10611073.

13. Piatigorsky, J., and Kozmik, Z. (2004). Cubozoan jellysh: An Evo/Devomodel foreyes andothersensory systems. Int. J. Dev.Biol. 48 , 719729.

14. Kozmik, Z., Swamynathan, S.K., Ruzickova, J., Jonasova, K., Paces, V., Vlcek, C., and Piatigorsky, J. (2008). Cubozoan crystal lins: Evidence for convergent evolution of pax regulatory sequences. Evol. Dev. 10 , 5261.

15. Conant, F.S. (1898). The Cubomedus. Memoirs from the BiologicalLaboratory of the Johns Hopkins University 4, 161.

16. Martin, V.J. (2004). Photoreceptors of cubozoan jellysh. Hydrobiologia530/531 , 135144.

17. Koyanagi, M., Takano, K., Tsukamoto, H., Ohtsu, K., Tokunaga, F., andTerakita, A. (2008). Jellysh vision starts with cAMP signaling mediatedby opsin-G(s) cascade. Proc. Natl. Acad. Sci. USA 105 , 1557615580.

18. Garm, A., and Bielecki, J. (2008). Swim pacemakers in box jellysh aremodulated by the visual input. J. Comp. Physiol. A Neuroethol. Sens.Neural Behav. Physiol. 194 , 641651.

19. Coates, M.M. (2003). Visual ecology and functional morphology of Cubozoa (Cnidaria). Integr. Comp. Biol. 43 , 542548.

20. Werner, B., Cutress, C.E., and Studebaker, J.P. (1971). Life cycle of Tripedalia cystophora Conant (Cubomedusae). Nature 232 , 582583.

21. Garm, A., Ekstro m, P., Boudes, M., and Nilsson, D.E. (2006). Rhopaliaare integrated parts of the central nervous system in box jellysh. CellTissue Res. 325 , 333343.

22. Skogh, C., Garm, A., Nilsson, D.E., and Ekstro m, P. (2006). Bilaterallysymmetrical rhopalialnervous systemof thebox jellysh Tripedaliacys-tophora. J. Morphol. 267 , 13911405.

23. Satterlie, R.A. (2002). Neural control of swimming in jellysh: A compar-

ative story. Can. J. Zool. 80 , 16541669.24. Yamasu, T., and Yoshida, M. (1976). Fine structure of complex ocelli

of a cubomedusan, Tamoya bursaria Haeckel. Cell Tissue Res. 170 ,325339.

25. OConnor,M., Garm,A., andNilsson,D.E. (2009). Structure andoptics of the eyes of the box jellysh Chiropsella bronzie . J. Comp. Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 195 , 557569.

26. Hamner, W.M., Hamner, P.P., and Strand, S.W. (1994). Sun-compassmigration by Aurelia aurita (Scyphozoa): Population retention andreproduction in Saanich Inlet,British Columbia.Mar. Biol. 119 , 347356.

27. Hamner, W.M., and Hauri, I.R. (1981). Long-distance horizontal migra-tions of zooplankton (Scyphomedusae: Mastigias ). Limnol. Oceanogr. 26 , 414423.

28. Zemek, R., and Nachman, G. (1999). Interactions in a tritrophic acarinepredator-prey matapopulation system: Prey location and distancemoved by Phytoseiulus persimilis (Acaari: Phytoseiidae). Exp. Appl. Acarol. 23 , 2140.

Visual Navigation in Box Jellysh803
http://dx.doi.org/doi:10.1016/j.cub.2011.03.054http://dx.doi.org/doi:10.1016/j.cub.2011.03.054

current biology 2011 garm

Documents