uNDERwATER vISuAL FIsHCEnSuS SuRvEyS

Proper use and implementation

Pierre Labrosse, SPC Principal Reef Fisheries Scientist, NoumeaMichel Kulbicki, IRD Research Officer, Noumea

Jocelyne Ferraris, IRD Research Officer, Noumea

Secretariat of the Pacific Community Noumea, New Caledonia

2002

This document has been produced with the financial assistance of the European Community and France.

The views expressed herein are those of SPC and can therefore in no way be taken to reflect the official opinion of the European Community and France.

REef REsources Assessment Tools

ii

© Copyright, Secretariat of the Pacific Community (SPC), 2002

Original text: French

Secretariat of the Pacific Community cataloguing-in-publication data

Labrosse, Pierre

Underwater visual fish census surveys: Proper use and implementation by

Pierre Labrosse, Michel Kulbicki and Jocelyne Ferraris

(REAT: Reef resources assessment tools)

I. Fish stock assessment - Oceania - Handbooks, manuals, etc

2. Diving--Handbooks, manuals, etc

I.Title 2. Secretariat of the Pacific Community 3. Series 4. Kulbicki, Michel 5.

Ferraris, Jocelyne

ISBN 982-203-878-X AACR2

Secretariat of the Pacific CommunityBP D5

98848 Noumea CedexNew Caledonia

Telephone: +687 26 20 00Facsimile: +687 26 38 18

Email: spc@spc.inthttp://www.spc.int

Prepared for publication and printed at the Secretariat of the Pacific CommunityNoumea, New Caledonia

iii

TABLE OF CONTENTSACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .V

FOREWORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .VI

I - INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1I.1 - Why assess resources? Assessment and management . . . . . . . . . . . . . . . . .1I.2 - What kinds of assessment methods are available to fisheries

management personnel? Role of underwater visual census surveys . . . . .2

II - BACKGROUND INFORMATION ON UNDERWATER VISUAL FISH CENSUS SURVEYS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5II.1 - Basic principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5II.2 - Which types of data for exactly what kind of information? . . . . . . . . . . . . .6

II.2.1 - Status of fish stocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7II.2.2 - Environmental factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

II.3 - Sources of error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11II.3.1 - Sources of error due to the diver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11II.3.2 - Sources of error due to fish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13II.3.3 - Sources of error due to sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13

II.4 - How to limit sources of error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16II.4.1 - Observer training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16II.4.2 - Basic rules for designing sampling surveys and selecting transects

or stationary points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20

iv

III - SETTING UP VARIABLE DISTANCE UNDERWATER VISUAL FISHCENSUS SURVEYS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26III.1 - Review of safety rules for diving work . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26III.2 - Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28III.3 - Transects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29III.4 - Stationary points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33

IV - USING THE DATA COLLECTED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37IV.1 - Entering and formatting data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37IV.2 - Calculating densities, biomass and stock (numeric and weighted) . . . . . .37IV.3 - Calculating the accuracy of density and biomass estimates . . . . . . . . . . . .40IV.4 - Mean composition of substrate and living organism cover . . . . . . . . . . . .42IV.5 - Initial analysis of results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42

IV.5.1 - Comparing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42IV.5.2 - Correlations between fish and environmental factors . . . . . . . . . . . . . .45

V - CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47

APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49Appendix 1 - Outlines of the major fish families that are of food

and/or commercial interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49Appendix 2 - List of fish species that are of food and/or commercial interest . . . . .50Appendix 3 - Fish census record sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51Appendix 4 - Environmental factors record sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .52

TO FIND OUT MORE … . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53

v

ACKNOWLEDGEMENTS

This handbook is the result of a team effort and so we would like to thank all those whocontributed to it.

The SPC Publications Section played a major role in this work. Our warm thanks toMuriel Borderie who handled the lay out, Jipé Le Bars who designed its illustrations andKim Des Rochers who edited the manuscript.

We should not overlook the proof-readers, who made significant improvements to themanuscript: Marie-Thérèse Bui, Ben Ponia and Sheryl Mellor and Gilles Kaboha whotranslated the handbook into English.

Finally, we would like to express our special thanks to Mireille Harmelin-Vivien who gaveus permission to take our inspiration from a document about underwater visual censussurveys published in 1985 (see ‘To Find Out More…’ section).

This document is dedicated to Pierre Thollot.

vi

FOREWORD

T his document is the first in a series of publications that respond to requests from fish-ery services in Pacific Island countries and territories for tools and practical advice on

implementing reef fish resource assessment methods. This series is open to all authors andis part of a process of standardising assessment and monitoring methods for reef andlagoon fishery resources and activities at the regional level.

A variety of manuals and publications have dealt with underwater visual census surveymethods and their applications, as well as data processing and analysis. For this reason, thishandbook will simply review their major aspects. Its main purpose is to introduce anunderwater visual census survey method that has been developed and enhanced by theFrench Institute of Research for Development (IRD). This method has been tested in anumber of different locations, including New Caledonia,Tonga and Fiji Islands. It can also beadapted for and has applications in the assessment of invertebrate resources.This bookletis a companion document to the ReACT-FishUVC software, which makes it possible toboth enter the data acquired during surveys using this method and carry out initial calcula-tions on them.

1

I.1 – Why assess resources?Assessment and management

Asteady rise in many Pacific Island popu-lations, particularly those in urban areas

and capitals, is causing an ever-increasingdemand for protein. Given the limited landavailable to small islands and the developingnature of most island economies, it is unlikelythat this nutritional demand will be satisfiedby either agriculture or imports in the nearfuture. Pressure on fisheries, especiallynearshore reef fisheries, will inevitablyincrease.

In addition to local needs, there is a growingdemand in Asia and other areas for specialityreef fishery products, and Pacific Islandexporters are finding new markets on a reg-ular basis. One such example is the marketfor live reef food and aquarium fish. As withall fishing, exploitation has an effect on thereef ecosystem, but the potential risks fromthis particular fishery are quite high.Thismarket demands species that are, for the

I - Introduction

most part, fragile due to their scarcity ortheir biological and ecological characteristics.Due to their high market value, reef fisheriesare also prone to destructive practices (e.g.cyanide or dynamite fishing), which endangerfishermen, coral reefs and their inhabitants.

Despite the risks inherent in such activities,reef resources represent a source of nutri-tion and income and a significant develop-ment opportunity for Pacific Island countries.These communities and their governmentsmust try to balance conflicting demands,development, and maintain ecological integri-ty; however, available information andresources are still insufficient.

Managing means looking ahead. Managingalso involves adapting existing resources tomeet the objectives set for the medium- orlong-term (i.e. beyond three years). Goodmanagement practices are based on havingas much information as possible about thecontext and environment the practices willbe applied to. For that reason, data collection

2

methods that make it possible to describe asclosely as possible the situation at a givenpoint in time, must be determined. Ideally, aspart of the precautionary approach,management plans should be designed frombaseline information and include monitoringmeasures so that action can be takenwhenever human-induced changes becomenoticeable.

This information should include:

estimates of total and exploitable fisherystocks;

analysis of the main structures of thesepopulations (size, trophic and populationstructures);

the health status of the reef and associat-ed ecosystems; and

a good grasp of the dynamics of the fishery.

Acquiring this information involves:1) designing sampling strategies that provide a

reliable picture of the resource for anacceptable level of effort both in terms ofcost and time; and

2) implementing adequately tested methodsto provide high quality information that canbe compared in both space and time,thereby allowing reliable measurement ofany changes that occur.

I.2 – What kinds of assessmentmethods are available tofisheries management personnel?Role of underwater visualcensus surveys

Many types of reef and lagoon resourceassessment methods exist. None of themare perfect and all have advantages and dis-advantages. They all have one point incommon: they are based on the study of asection or subgroup of the population inquestion. For that reason, they require theuse of sampling techniques, which can bedivided into three categories: capturemethods, mixed methods and non-capturemethods.

Such information should make it possibleto adapt management measures to eachnew situation encountered.

3

Capture methods mainly involve record-ing information on fish captured in traps(and, more generally, with bait), in nets bytrawling, and with lines. These methods arebased on an analysis of the catch per unitof effort (CPUE), which is considered to bean index of the study population’s density.Capture methods can be used at any timeand at considerable depths. Moreover, theycan be implemented on a wide scale and atlow cost using scientifically unskilled staff.On the other hand, gear design and selec-tivity, the effect of baits, and the capturabilityof certain species are all factors that havean effect on the comprehensiveness andaccuracy of these methods, which remainlow to moderate. One exception, however,is fishing with explosives or poisons (e.g.rotenone), which gives comprehensiveresults, particularly in terms of species rich-ness, but whose destructive effects are amajor disadvantage, particularly for repeat-ed sampling.

‘Mixed’ methods are somewherebetween capture methods and non-cap-

ture methods. In particular, they includecapture-tag-recapture methods, which aredifficult to assess qualitatively or quantita-tively. On the other hand, mixed methodsare very effective in determining age,growth, movement and behaviour in reeffish populations.

Due to hyperdiversity1 and the widerange of reef and lagoon environments,capture methods are often inadequate.The most effective capture method isdestructive and/or disruptive (e.g.dynamite, rotenone), but can occasionallybe useful for calibrating methods basedon observation. This justifies thedevelopment of true non-capture or‘fishery independent’ methods,which include visual censuses andhydroacoustic techniques.These twotechniques are especially useful forassessing pelagic or semi-pelagic fishstocks comprising a limited number ofspecies.They are, however, poorly adaptedto highly diverse benthic populations,although certain applications are possible,

1 The term ‘hyperdiversity’ refers not only to very high species diversity, but also to very high diversity of biotopes, ecological niches,behaviours, genomes and uses.

4

particularly for behavioural studies.Theclarity of tropical water is a factor that hascontributed to the growing use of on-sitevisual assessments.These methods arealso more comprehensive, more accurateand non-destructive. Underwater visualcensuses (UVC) were first used to

measure fish and invertebrateabundances.They were then used tostudy the dynamics of exploited andunexploited populations, and the ecologyand management of natural resources andenvironments.The use of UVC requiresqualified and trained staff.

Comprehen-siveness

Low*

Low

High

Accuracy

Low to moderate

Moderate

High

Coverage

High

Moderate

Low

Bias linked tolife cycle

Yes

Yes

No

Staff training

Low

High

High

Costs

Low

High

Moderate

Quality of data NeedsSampling techniques

Capture

Mixed (combined)

Non-capture

Summary of the various sampling method categories

* except for explosives and poisons

5

II.1 – Basic principles

underwater visual census methods arebased on on-site visual counts of

organisms. Census methods can be donein a variety of ways, the most common ofwhich is by either snorkelling or scuba div-ing. In certain instances, cameras, videocameras or submersible gear can be usedto get around the constraints linked tosampling by divers.

Use of underwater visual censuses is limit-ed by:1. visibility, which must be adequate to

record useful information. If the wateris too turbid (e.g. around rivers estuar-ies, mangroves, mining areas, etc.)and/or there is minimum light (e.g.during night diving, or limited light pen-etration into the water), this method isdifficult, if not impossible, to use;

2. the state of the ocean and, moregenerally, weather conditions; and

3. the diver’s physiology in the aquaticenvironment, in the case ofsnorkeling. In those cases when theyare not able to conduct the censusfrom the surface, observers arelimited by their inability to remainunderwater without breathing. Use ofscuba equipment involves depth andtime limits that cannot be surpassedwithout endangering the diver. Diversare also limited by their inability towithstand cold and fatigue.

For all these reasons, underwater visualcensuses involving snorkelling or scuba arebest carried out in clear, calm and shallowwater (generally between the surface and adepth of 20 m).

Censuses can be conducted in three ways:along random paths (chosen by chance),using quadrats (grids moved along a tran-sect by divers) or transects, or from sta-tionary points. Only the latter two fish

II - Background information onunderwater visual fish census surveys

6

census methods are discussed in this man-ual; the other method is rarely used andcan generate other problems related toresource assessments.

A transect is a rectangular area whoselength and width is clearly defined.The cen-sus, or count, is carried out within theboundaries of the transect, which is general-ly denoted out by a flexible graduated tapemeasure, which is rolled out on the seafloor.Counting is conducted on either side of thistape within a set area. This is one of themost commonly used methods, and is wellsuited to population studies, particularlythose assessing fisheries resources for com-mercial or food purposes.

Censuses can also be done by countingfrom stationary points. In this case, theobserver begins counting from a deter-mined point while slowly turning in a circle.This method is quicker than laying a tran-sect. It is particularly recommended forstudying a species or small group ofspecies, especially in very heterogeneousenvironments, and for isolated complexstructures or large-size formations (coralheads, large boulders).

II.2 – Which types of data for exactlywhat kind of information?

The data collected and the sampling strate-gy used determine the type of informationthat can be obtained. This can differ accord-ing to whether a qualitative or quantitativeinventory of fisheries resources is required.

Information of interest to reef fisheriesmanagement can be divided into twocategories.1. The state and size of food and/or

commercial fish stocks, or those ofinterest due to their ecologicalaspects (biological indicator). Thisincludes analysis of the populations’structure.

2. The state of the reef habitat, whichsupports the resource, includingassociated living organisms.

This information is mainly designed to:1. make it possible to carry out compar-

isons in space (resource location anddetermination of its characteristicsdepending on site, biotope or zonessubject to unequal fishing pressures)and/or time (resource monitoring,

7

changing trends and detectingchanges); and

2. determine correlations between theresource and its environment.

II.2.1 – Status of fish stocks2

a – Identifying and counting speciesIdentifying and counting species providesan estimate of species richness (i.e. thenumber of species), particularly for envi-ronmental inventories. This can be limitedto a sector of the population for foodand/or commercial purposes or else canbe conducted from an ecological point ofview.This is an important parameter toconsider. Any appreciable attack on theenvironment, such as the destruction ofcoral, usually brings about a decrease inspecies richness, which is an indication ofbiodiversity (i.e. number of species, andtheir percentage in the population).

b – Counting individualsIndividuals are counted to estimate abun-dance (number of fish) and density (thenumber of fish per surface area unit) (e.g.individuals per square metre). Abundanceand density are factors that can be affected

by fishing activities and so, in certain cases,are a reflection of fishing intensity.

Counting can be conducted in a fixed rec-tangular area (transect) or a circular area(stationary points), where respectivelengths, widths or diametres have beenpreviously determined (fixed distancecounting). The observer counts the fishwithin the area laid out. For a givenspecies, the estimate of the mean densityD on a transect of width d and length L isexpressed as:

In the case of stationary points, density isexpressed as:

2 The term ‘stock’ is not used in the fisheries sense but rather as a definition of the quantity (in numbers or weight) of fish thatexists in a given environment.

p

Σni

D = i=1

Ldwhere ni: number of fish seen

where ni: number of fish seenr: radius chosen for observation.

p

Σni

D = i=1 Πr2

8

Correctly estimating densities by thismethod assumes:

the observer does not overlook anyfish in the set surface area, an unat-tainable but desirable goal; andindividuals are counted only once(see Section II.3).

This method should especially be used forcounting populations of fairly sedentaryfish (where there is a low risk of errordue to rapid individual movement).Thismethod tends to bring about an under-assessment of density and biomass. Thewidth selected for a transect also has aneffect on results. The wider the transect,the lower the density estimate, which islogical given that the farther away fish are,the less chance they have of beingobserved. This relationship varies accord-ing to species and groups of species, andis fairly significant for large individuals andmore mobile species.

Counting can also be conducted by takinginto account the fish’s distance from thetransect or stationary point at the time ofobservation (variable distance counting).In this case, surface areas are calculated

afterwards. The observer assesses andrecords the perpendicular distance of fishas viewed from the transect or stationarypoint. In the case of schools of fish, twodistances are taken into consideration: thedistance from the transect (or stationarypoint) to the fish closest to it; and the dis-tance from transect (or stationary point)to the fish farthest away from it.

As with the first method, correct estima-tion of density is based on certainassumptions.1. Fish are not counted twice;2. The fish’s distance is determined at

the position where it was firstobserved;

3. The probability of detecting fish with-in the transect is equal to 1, whichmeans it is presumed that all individ-uals along the tape measure havebeen seen and counted.

This method provides good coverage forless mobile species and also limits errorsdue to rapid movements or fish fleeingwhen encountered during fixed-distancecounts (see Section II.3.2). For this rea-son, it is better adapted to resourceassessment.

9

c – Assessing individual sizeSize assessment allows estimation of:

mean sizes;mean weights (from existingsize–weight ratios);biomass, which is fresh weight persurface area unit (e.g. g/m2 or kg/ha);this is calculated using the individualmean weights and abundances;

BASIC CONCEPTS OFTHE SAMPLING-BY-DISTANCE THEORY

Counting individuals while estimating their distance from the transect or stationary point isbased on the fact that not all detectable fish are necessarily seen. It is founded on the theo-ry that detectability — the probability of detecting a fish — decreases with observation dis-tance. In other words, the farther away from the transect or stationary point an individualis, the lower the chances are that it will be observed. Calculating densities depends on thedetectability function g(x). From these observations, it is possible to extrapolate for a givenspecies, family or population, a curve showing the probability of sighting the fish based onits distance from the transect. Estimated density D is expressed as follows:

p

ΣniD = i=1 where a = � g(x)dx

Lani : number of fishL: length of transect dmax: perpendicular distance from the transect to the limit of detectability

stock present in the environment:total weight for a given sampled area.

As with abundance and density, mean fishsizes and biomasses are parameters thatare affected by fishing activities, particular-ly with regards to the most heavily target-ted species. For example, in the specificcase of untouched, or unexploited stock,

dmax

0

10

the introduction of fishing activities willrapidly lead to a decrease in mean sizeand biomass for the largest and mostlong-lived species.

II.2.2 – Environmental factorsDepending on the research question andthe resources available to observers, oneor more environmental factors can berecorded. The morphology of the studyarea has an effect on the population’sorganisation, so it is particularly importantto determine the relationships betweenhabitats (substrate, covering organisms,etc.) and the parameters that describethe resource (species richness, density,biomass).

Environmental factors can be divided intotwo categories: constant factors and vari-able factors.

a – Constant factors (or those that varyonly slightly)Location of the site (geographiccoordinates + name of site and pos-sibly landmarks)Type of environment or biotope(barrier, intermediate or fringing reef)Reef section (slope, forereef, etc.)Prevailing wind directionDepth (min. and max.)SubstrateType of activity (e.g. regulated orunregulated fishing area, reserve, eco-tourism site)

b – Variable factorsDate and time (beginning and end)VisibilityCurrentSalinityWater temperatureSubstrate-covering organismsAssociated organisms (invertebrates)

11

SUMMARY TABLEDATA

Number of species+

Number of individualsand allowance for distances

(fixed or variable)+

Size estimates

Environmental factors:- constant- variable

INFORMATION

Species richness

Density

Mean sizes,mean weights,

biomass

Fish/habitat correlations

Mean substrate compositionLiving organism cover

II.3 – Sources of error

No assessment method is perfect and underwater visual censuses alsoinclude sources of error. Errors mainlycome from one of three sources: theobserver, fish behaviour, and the sampling method. Understanding thesesources of error is vital for both

minimising them, and taking them intoaccount during analysis and interpretationof the results.

II.3.1 – Sources of error due to the diverThese can be linked to the observer,and/or to interactions between observersand fish. The methods used can also be asource of error.

12

a – Errors caused by the observerDue to diving time limits and the often shynature of the animals surveyed, observersmust be able to record information asquickly as possible, and rapidly identify andestimate sizes and distances with a reason-able level of accuracy. The slightest hesita-tion will result in a loss of data. Observersmay pay more attention to one group offish or a part of the population that inter-ests them more, and this also constitutes asystematic error. Observers may also havea tendency to over or underestimate sizesand/or distances. Moreover, there is alwaysa risk of counting the same fish severaltimes. Finally, in the case of variable dis-tance counts, consideration must be givento the fact that fish detectability tends toincrease with size for almost all species, andwith the size of the school for certain fish(number of individuals in a school).

Hesitation, inattention or paying too muchattention to a certain area, are all increasedwhen the conditions under which the cen-sus is being conducted worsen (e.g. due tostrong currents, fatigue or cold) or whenthe quantity of information to be recordedis too great (too many species, too many

fish). As a result, environmental, psycholog-ical, and physical conditions of observers’work should have as little influence onthem as possible. This means thatobservers should master diving techniquesand not be subject to disturbances thatreduce the acuity of their eyesight and/ortheir motor skills.

b – Observer – fish interactionSuch interactions mainly involve changesin fish behaviour due to the diver’s pres-ence. These changes vary and may resultin either the fish fleeing away from, orbeing attracted to, the diver. For example,some species, such as those from thePlectropomus (coral trout) genus, tend tobe attracted to observers and followthem around. In contrast, spangledemperors (Lethrinus nebulosus) tend tokeep away by remaining at the limit of vis-ibility. Such behaviour also depends onthe individuals’ activity cycle (diurnal vsnocturnal), age and location. The simulta-neous trajectories of the fish and observ-er can bring about either negative or pos-itive biases in size estimation, dependingon whether they are swimming in thesame direction or in opposite directions

13

with an angle of vision that differs morethan 90° from the transect.

The diver’s movement as well as themethod used, also has an influence on fishbehaviour (e.g. the noise from air bubblescoming out of scuba diving equipment).Regular visits to a site, particularly for mon-itoring, tend to decrease attraction or flee-ing reactions, thus minimising these biases.

II.3.2 – Sources of error due to fishThe main sources of error due to fishcome from the distribution of species intime and space. This depends on differentparameters associated with habitat, behav-iour and activity cycles.

Certain sedentary species living in rock orcoral crevices during the day, coming outonly at night, may not be detected byobservers. The same is true for those fishthat only come out of their hiding placesbriefly, or those that are highly mobile orwhich colonise certain biotopes during cer-tain seasons. The probability of encounter-ing species and thus being able to countthem, is influenced by their behaviour andtheir home range or territory. This is why

it is difficult to make a comprehensiveassessment of fish populations. It shouldalso be noted that populations seen andsampled only account for a portion of allthe species that live in the study area.

The various biological and ecological char-acteristics of fish influence measurementsand estimates. Any interpretation of resultsmust take into account that not all speciesare perceived (and therefore estimated) inthe same way.

II.3.3 – Sources of error due to samplingMost sampling errors arise because resultsdepend not only on the elements (all tran-sects or stationary points) that make up thesample (n), but also on the method itself.

A sample is a limited subset of a popula-tion, from which the results obtained fromthe observed data are based. For techni-cal, economic or simply logistical reasons(destruction of specimens, as when fish arecaught by experimental fishing), it is nor-mally not possible to collect data on theentire population. Study of a limited setmakes it possible to increase both thenumber of measurements and their degree

14

of accuracy. Extrapolation of the findingsobtained from sampling generally results inestimates for the entire population, whichhave a reasonable level of acuracy. If twosamples composed of a given number ofelements (set of transects or stationarypoints) are observed, the measurementscalculated for each will be different, but willresult in comparable estimates of popula-tion parameters.The statistical populationis defined as all the N elements fromwhich conclusions are based (i.e. the Nsurface area units corresponding to all pos-sible transects in which censuses can beconducted in the selected zone during the

study period). Samples that are not takenaccording to a strict sampling plan (randomor reasoned) will not be representative ofthe target fish population. The sample isconsidered to be equal to the statisticalpopulation (n=N).

A transect’s position and orientation canbe considered sources of error associatedwith sampling. It is better to have a tran-sect that covers an homogeneous environ-ment, rather than one covering several dif-ferent environments. Transitional areasbetween different biotopes should, there-fore, be strictly avoided.

IN BRIEF

Total error =

observation errors (due to the observer)

+ target population coverage errors (due to observer – fish interactions and to fish)

+ sampling errors (systematic error + random error)

15

REMEMBER…The accuracy of population estimates depends on the size of the sample (number of transects) and variability (dif-ferences between the measurements for each transect).This variability in individual measurements or random error(dispersion) must not be mistaken for systematic error (bias).

In the figures below, the centre of the target represents the parameter to be estimated and the black dots corre-spond to measurements made in the transect. Four model situations are given:

Lack of accuracy can be linked to high bias and/or high dispersion.The ideal situation is shown in thefirst picture (low bias, low dispersion).

Sampling-related bias can be reduced by randomly selecting sample elements.

Errors in observation and representativity do not decrease when the size of the sample increases.

Dispersion depends on the population’s heterogeneity. It is measured by variance.

When dispersion is high (i.e. there are significant differences between transects), better estimates willbe gained by stratifying the population (e.g. by biotope).

Low biasLow dispersion

Low bias High dispersion

High bias Low dispersion

High biasHigh dispersion

16

Acanthuridae (surgeonfish)Siganidae (rabbitfish)Kyphosidae (drummer, chub)Holocentridae (squirrelfish)Haemulidae (sweetlips)Chaetodontidae (butterflyfish), act asecological indicators of coral reefhealth and can also be added to thislist.

The first key to identifying fish is theirshape, which is generally the same foralmost all the species in a given family(Appendix 1).

The visual morphological characteristicsthat identify a species within a family areshape, colour of markings, and any dis-tinctive traits such as spots, lines orstripes, and their location on the bodyand/or fins. Behaviour and preferredbiotopes are also useful information foridentification.These must all be learnedand memorised, which is perhaps themost tedious part of training. Theappearance of some fish changes overthe life cycle; for example, many wrasseand parrotfish have different couloursduring their juvenile and adult phases.

II.4 – How to limit sources of error

II.4.1 – Observer trainingWhile it is relatively difficult, if not impossi-ble, to minimise errors due to fish, it is eas-ier to limit those due to observers throughproper mastery of diving and countingtechniques. This comes about through fol-lowing certain rules during on-site use ofthe method (see Section III) and by keep-ing in mind that divers should have regulartraining to minimise errors caused by poordiving techniques.

a – Identifying fishThis is the first step of training. Most reeffish of interest for marketing and/or con-sumption belong to about 15 families, themain ones are listed below.

Serranidae (grouper, rockcod,seabass)Carangidae (trevally, jack)Lutjanidae (snapper and seaperch)Caesionidae (fusiliers, bananafish)Lethrinidae (sea bream and emperor)Mullidae (goatfish)Labridae (wrasse)Scaridae (parrotfish)

17

In the same way, colours for a singlespecies can differ according to sex.Parrotfish are a good illustration.

Training in fish identification involvesclassroom-learning using available tools(see ‘To Find Out More’ section), and on-site exercises during dives. In order toavoid confusion over the use of commonnames, it is preferable to use eachspecies’ scientific name, which is alwaysmade up of a genus name (e.g. Lethrinus,the genus name for certain breams andemperorfish) followed by a species name(e.g. nebulosus; thus forming the nameLethrinus nebulosus, or spangled emperor).In practice, it is best to begin learningabout 50 species that are easily identifi-able. A list of fish, which are of food andcommercial interest, is given in Appendix2. If fish from a certain family cannot beprecisely identified during a dive, theobserver must be able to rapidly note itsmain features (e.g. shape, colours, mark-ings) so as to identify it through booksafterwards. The use of simple sketches toillustrate and record specific marks onthe body is invaluable.

b – Counting individualsDifficulties counting fish are mainly due to thelimitations of the human eye, which can onlycount four objects at any one time.Moreover, precise counting cannot be carriedout on more than 10 to 20 individuals in thecase of a relatively sedentary school. Takinginto account these limits, and in order tocompensate for them, the most commonlyused technique for counting schools is theso-called group-counting method.This con-sists of counting a ‘group’ of 10 to 20 fish.Thisgroup becomes the basic counting unit andthe observer judges how many groups thereare in the entire area occupied by the schoolof fish. For large schools (> 200 individuals),it may be useful to combine groups intosuper-groups, containing 5 to 10 base groups.

In more complex instances where a schoolof fish is made up of several differentspecies (multispecies school), observers areencouraged to begin counting the mostabundant (or most numerous) species. Thesame applies to schools with a range ofsizes. During training, taking photographs isa good way to evaluate errors made, and tofind out at what level they occurred.

18

c – Estimating individual sizesIn order to obtain acceptable results, sizeestimates must reach a certain level of accu-racy.The relative difference in sizes must notbe more than 20% and should be closer to10% whenever possible. Numerous factorscan affect estimates, including the magnifyingeffect of water (or mask magnification), theangle the fish is viewed from, water clarity,and the fish’s shape.

One of the most commonly used trainingmethods consists of making cut-outs of fishof various shapes and sizes correspondingto the main families to be assessed (seeAppendix I). These cut-outs can be easilymade using marine plywood or PVC. Theyare attached to a 150 m weighted linedropped in the water and moored at bothends. The sizes of the cut-outs must coverthe entire range of likely fish lengths, from10 to 100 cm. The diver swims along theline at a preset distance (generally threemetres at the beginning of training). Hethen notes the number, and estimates thesize of each cut-out. At the beginning oftraining, a few reference cut-outs, with thetrue size indicated, can be used as referencepoints for estimating the sizes of the others.

Once back on land, the diver’s estimates arecompared with the real sizes; this compari-son allows the diver’s performance to beevaluated and shows how much progresshas been made. The simplest way is tomake a dot graph with the various estimat-ed values recorded on the ‘y’ axis, and thereal sizes of the cut-outs on the ‘x’ axis. The

Visual effects linked to the shape of fish

These fish are the same in length and yet thebottom one appears longer!

Acanthuridae

Mullidae

19

dots should adhere as closely as possible tothe straight line where ‘x’ is equal to ‘y’, (i.e.where the estimated values correspond tothe real values). Dots situated above thisstraight line show that the observer overes-timated the sizes of the cut-outs. On theother hand, when the dots are below theline, the diver has understimated the sizes.Generally, it is not useful to analyse eachseparate dot, but rather to note the overalltrend in order to see whether the observerhas a general tendency to overestimate orunderestimate sizes and in which range thisoccurs. Observers may make preferentialerrors, such as systematically underestimat-ing the size of larger fish while correctly esti-mating the sizes of smaller ones.

To go into further detail, performances canbe subjected to statistical analysis (t test onthe correlation between estimated sizesand real sizes). In order to ensure a correctestimate of fish sizes, training must berepeated until reported sizes are closeenough to real sizes (margins of errorunder 20%). For confirmation, a few fishcan be caught during sampling campaigns,measured and their sizes compared withestimates.

Calibration line used to measure an observer’s ability to estimate sizes.

The results of observations recorded on thisgraph must come as close as possible to the

straight line.

100 -

90 -

80 -

70 -

60 -

50 -

40 -

30 -

20 -

10 - - - - - - - - - - -

Estim

ated

siz

es (

in c

m)

10 20 30 40 50 60 70 80 90 100

Real sizes (in cm)

Overestimated sizes

Underestimated sizes

d – Estimating distances (in the case ofvariable distance counts)

As with sizes, observers must pay carefulattention when estimating distancesbecause water has a tendency to magnifyobjects. Water clarity can also lead toassessment errors. For that reason, when

20

the water is clear, there is a tendency tounderestimate distances, whereas distancesare overestimated in turbid waters.

Training in estimating distances perpendicu-lar to the transect can be conducted bylaying out two tape measures. Thesemeasures serve as reference points at pre-determined distances on either side of thetransect, along with objects whose distancefrom the transect is also known… toeveryone but the diver in training. As withfish sizes, results obtained by observers canbe compared with the real distances. Theexercise can be repeated, taking intoaccount errors (under or overestimation)until the margin of error is reduced.

II.4.2 – Basic rules for designing sampling sur-veys and selecting transects or stationary points

A. Methodological approachIt is difficult, if not impossible, to eliminateall the problems associated with sampling.A successful sampling campaign comesfrom following a methodological approachcomprised of specific stages.

a) Thorough planningThe planning phase is indispensable fordetermining justifiable choices. Proper plan-ning allows bias to be minimised, increasesthe study’s accuracy, and avoids problemsassociated with sampling, which are toooften only noted once data have alreadybeen collected.

The planning phase must answer the fol-lowing questions:

Which exact organisms will beobserved in order to respond tothe problem? For example, this can

The ability to correctly estimate sizesand distances decreases with time; forthat reason, training or sampling

must be carried out on a regular basis so asto maintain an acceptable level ofperformance.

The best way to validatemeasurements and verify data is to use and interprete them!

21

involve counting all the fish or onlycounting those fish of commercialinterest;Which approach should be used?The approach can be experimental(the cause of the phenomenon to bestudied is induced, by closing andopening a fishing area) or descriptive(the phenomenon is describedthrough sampling the study popula-tion: this is the case for sampling inareas subject to different fishingintensities);What constitutes the basic elementon which observations will bemade? In our case, this means eithertransects or stationary points.What is the study population? Thiscovers all the elements (transects orstationary points) and is defined byfour factors: its nature (e.g. transect),specific characteristics (e.g. transectlength), location (e.g. southwestlagoon) and the date on which it isobserved (e.g. August 2001);Which variables should be consid-ered? These are the element’s attrib-utes. Variables can be quantitative(e.g. number of fish in the transect,fish size, density, biomass) or qualita-tive (fish species).

Which measurement methodshould be used? In our case, thisinvolves the underwater visual censussurvey method.Which type of sampling strategyshould be selected? This can involvea simple random sampling, one whichis stratified by biotope, systematic,etc.Which parameters are to be esti-mated? This can be the mean ortotal of a quantitative variable (num-ber of fish, density, biomass), or thepercentage of a qualitative variable(% of species), etc.

b) Acquiring high quality dataThis depends on how collection is carriedout and on the sampling strategy.

c) Rational use of the dataThis includes processing and interpretationbut also the conclusions/decisions and theresult formulation phase.

Planning for both sampling and dataprocessing must be carried outsimultaneously so as to better justify

‘why’ and ‘how’ the collection will be made.

22

B. Sample sizeSample size refers to the number of tran-sects needed to arrive at a mean assess-ment representative of the population; thismust include at least two elements inorder to measure dispersion. But two isnot enough; the more heterogeneous thepopulation is, the greater the sample sizemust be.To estimate a population parame-ter to within 10% (p=0.1) — 5 chancesout of 100 of making a mistake in settingthe value of this parameter (alpha=5%) —the theoretical sample size is found by:

the mean: n0=t2S2/P2

a percentage of an infinite population(N>10,000): n0=P(100-P)t2/P2

a finite population (N<10,000):n = n0 / (1+n0 / N)

Where S2: sample variance (see definition of

variance on page 24);

P: percentage of a modality of a qualitative

variable (e.g. percentage of females in the ‘sex’ variable,

which comprises two modalities, males and females)

The sample size can also be calculatedbased on cost. If you take the total studybudget (C) and the unit cost (c) per

observation (each transect), the number oftransects can be found by:

n=C/c

C. Sampling planThe sampling plan is the strategy for defin-ing the population to be assessed. The sam-ple selection method (choice of elements),formulation of the estimator, and the char-acteristics of the study population will all bedetermined by this plan. Bias, cost andaccuracy criteria should be considered dur-ing the selection of a sampling plan.

Randomly selecting sample units ensuresthat selection of one individual does notinfluence selection of others. This meansthat study staff should not be given any lat-itude in selection. Random selection canbe conducted in a number of ways such asby a computerised selection of geographiccoordinates, selecting squares in a griddedarea, or by simply pointing a pencil at amap…blindfolded!

It is sometimes difficult to define a randomsample. Non-random sampling can giveuseful results, providing the selection of

23

DEFINING VARIANCE

Variance is a measurement of dispersion in the study population. In the theoretical examplebelow, the measurements of a sample’s (n) elements are considered to be weights laid out on ahorizontal bar, with each weight associated with an individual value.

X = 8.6

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

X1 X2 X3 X4 X5 X6 X7

The mean (X) corresponds to a point at which it becomes possible to balance negative andpositive differences:

X=X1 + X2 + X3 + X4 + X5 + X6 + X7

n

By replacing each measurement with its numerical value, the formula becomes:

X= 2 + 4 + 5 + 8 + 11 + 14 + 16 = 8.67

Variance (S2) is equal to the mean of the square of the sum of the differences between themeasurement (Xi ) and the mean (X):

S2 =Σ (Xi - X)2

nUsing the values in the example, we get:

S2 =(2 - 8.6)2 + (4 - 8.6)2 + ......+ (16 - 8.6)2

= 247

n

i

24

elements can be controlled (reasonedchoice) and by carefully avoiding any hastygeneralisations. It often represents the mosteconomical method for exploratory studies.

A difference must be made between aprobability sample survey — where eachindividual of the population has a previouslyknown probability of belonging to the sam-ple (scientifically thorough with a mathemat-ical theory to support it) — and an empiri-cal sampling survey, which does not allowcalculation of the probability of individualinclusion (quotas, standard units). The lattersurvey is less thorough but its use is oftendictated by budgetary considerations.

The main probability sample survey plansare:

Simple random sampling: where allelements of the population have thesame probability of belonging to thesample.Stratified sampling: consists of dividingthe population into more homoge-neous sub-groups. In each stratum, arandom sampling is taken. In order tobenefit from stratification, a strongcorrelation must exist between the

stratification factor or factors and thecharacteristic studied. This means thatthe structure of the population mustalready be known and understood.Cluster sampling (in varyingdegrees): involves carrying out a ran-dom sampling composed of unitswhich are themselves sub-groups ofthe population (or clusters). Thecombined units within each sub-group are always representative ofthe population.Systematic sampling: consists of sys-tematically sampling an ‘n’ number ofelements separated by a constantinterval (interval of time, space, num-ber of lines, etc.). The first elementmust be chosen on a random basis.This type of sampling is frequentlyused in underwater visual censusresource assessments.

The estimates presented below involve sim-ple random sampling plans. They can beextrapolated from systematic sampling (onthe condition that the first element, say atransect, is chosen at random). This can beapplied at different scales and allows com-parisons in space (e.g. between biotopes)and/or time (e.g. between two dates).

25

IN BRIEF …

A sampling strategy comprises several aspects:

1) Selecting the population to which the study's conclusions will be applied (particularly with regards to space and/or time limits).

2) Selecting the sample size, which is determined by cost constraints and the level of accuracy desired.

3) Selecting the (n) elements of the sample (sample design).

4) Determining how information about the sample will be gathered (sampling technique = underwater visual census survey).

26

The methods presented in this sectionare based on the distance sampling

theory. For that reason, they take intoaccount the fish’s distance from the tran-sect or stationary point.

III.1 – Review of safety rules for divingworkDiving is an activity involving a certainnumber of risks that can be minimised byfollowing some basic rules, reviewedbelow.

Before each dive, make certain:1) your diving medical fitness certificate

is still valid (to be renewed annually);2) you get an up-to-date weather

report, if possible;3) your equipment is in good condi-

tion and operates properly, particu-larly your buoyancy vest, tank andregulator ;

III – Setting up variable distanceunderwater visual fish census surveys

4) proper surface security measures andsafety conditions are followed on theboat;

5) vital safety equipment functionsproperly (e.g. VHF radio, emergencyoxygen supply, first aid kit, list ofmedical staff and services to be con-tacted in the case of an emergency).

During the dive:1) NEVER DIVE ALONE (there should

always be a minimum of 2 divers)2) Keep visual contact with your dive

partner(s) at all times and, if possible,do not move away from them.

You must be in good physical andmental condition before each dive. Do not dive if you have a cold, are

tired, have recently eaten a heavy meal,or have consumed alcohol within the last12 hours.

27

3) If you lose your dive partner(s), stayfor a minute where you are and turnaround in a circle (360°); then, slowlyascend looking for the other diver’sair bubbles. Near the surface turncompletely around (360°). As soonas you arrive at the surface, inflateyour vest and make an OK sign tothe dive master. NEVER GO BACKDOWN ALONE. Wait for the oth-ers.

4) Avoid going up and down (‘yoyo’movement).

5) For dives whose maximum depth isgreater than 10 m, systematicallymake a three-minute safety pause atthree metres before coming to thesurface.

6) In general, try to stay under the timelimit beyond which decompression isnecessary.

7) Always ascend slowly (do not gofaster than the speed of the small airbubbles).

After the dive:1) Record your diving parameters (sites,

maximum depths reached, total time,length of pauses, specific facts).

2) Carefully rinse your equipment withfresh water and check its condition.

3) Avoid any strenuous physical effortor free diving for eight hours follow-ing your final dive.

4) Do not take a plane within 24 hoursafter your final dive.

It is worth noting that in almost all recorded diving accidents over the past few years, oneor more of these rules was not followed. Following them would have undoubtedly helped

save many lives. Never forget that all divers are responsible for both their own safety andthat of their fellow divers.

28

III.2 – Equipment

On the boat be sure there are:ID books (see ‘To Find Out More…’section)different coloured pocket folders (1for the blank record sheets, 1 forcompleted sheets)1 portable GPS + spare batterieseraserspencil sharpener

For working underwater, you shouldhave:

3 fifty-metre measuring tapes 1 hard plastic board with clips (atleast two) per diver record sheets (at least 3 stationsheets and 5 fish record sheets)pencils (at least 2 per diver : 1 in thesleeve of the diving suit or knife

HINT

In order to keep books dry and ensure thatthey do not get damaged, keep them coveredwith plastic, and keep water- and moisture-sensitive materials in a waterproof container(such as a cooler or eski).

Fifty-metre measuring tape

Hard plastic clip board

29

sheath + 1 attached by string to theboard)waterproof watchdepth gauge

III.3 – Transects

The method presented below uses stan-dard 50-metre-long transects.

On the boat before the dive:fill in the top part of the stationsheet with general information aboutthe station (e.g. station number, GPSposition, brief description, etc) (seeAppendix 4). Do the same thing onthe fish data record sheet (seeAppendix 3).

HINT

It is better to use wood-free synthetic resinpencils, as these have the advantage of notsplitting and having stonger lead.

During the dive:Before descending, make a finalcheck with your buddy to ensurenothing has been forgotten (doubleverification).Descend.Once you arrive at the bottom,determine the starting point for thetransect.Attach one end of the measuringtape to a rock, or in the sand, using ametal stake and lay the other twomeasuring tapes next to it.Determine the direction for layingout the measuring tape.Wait two or three minutes to givethe fish time to calm down and getuse to your presence.Begin counting the fish.One of the divers unrolls the meas-uring tape as the count progresses.

30

Diver 1

Diver 2

50 m

10 m

20 m

30 m

40 m

3-D view of a visual census survey using transects

Main transect

31

RULES TO FOLLOW DURING THE COUNT

1) It is best to work in sections of about three metres each as this helps avoid counting the samefish more than once.

2) In cases where each diver counts on one side of the transect, care must be taken to ensuredivers proceed at the same rate. If only one diver is conducting the count, he must firstcount on one side and then on the other. The tape is then unrolled by the other diver, whostands by while the observer continues to work.

3) Do not stay in any one spot too long and do not count fish who enter your field of vision toolong a time after you have stopped. If they must be counted, then situate them at the outerlimit of visibility.

4) If possible, begin by counting the most abundant species.

5) Given the theories related to variable distance censuses (see Section II.2.1), systematicallygive greater importance to fish that are near the transects. In practical terms, it is importantat each stop to begin counting in the immediate vicinity of the transect and then pay atten-tion to species farther away (i.e. carry out observations from the transect towards the limit ofvisibility).

6) Always fill out the dive sheet in the same way. Mark the name of the fish (see remarkbelow), then the number, size and finally the perpendicular distances to the transect.

7) Scientific names are sometimes long and tedious to write underwater, so don't hesitate to useabbreviations when these are easily understandable and can be immediately recalled after-wards. For example, you might write Acant xant for Acanthurus xanthopterus, or Sc riv forScarus rivulatus.

32

Once the fish count is done, leavethe main transect in place and unrollthe other two tape measures aboutthree metres on either side of themain transect

3.

Record the visibility, current andother known parameters for this sta-tion in the top part of the stationsheet (see Appendix 4).

2) Don't forget that the transect is 50 metreslong. Care must be taken to not go overone end or the other. Imagine perpendicu-lar lines on the transect at 0 and 50metres, which constitute the count bound-aries. When two divers conduct thecount, both must count within the bound-aries set out by the transect (i.e. the 50 mtape measure). If in doubt about whethera fish has already been counted by yourteam mate, ask.

HOW TO ESTIMATE VISIBILITY?

A simple method consists of placing a clip-board vertically at some point along the maintransect. On your way back to wind up thetape or record the substrate, note the distanceon the tape when the clipboard first becomesvisible to you.

For each of the three transects andin every 10-metre section, identifyonce every metre the nature of thesubstrate and the living organismcover just below the measuring tape.Record your observations on thestation sheet by putting check marksin the corresponding headings (seeAppendix 3)In the one-metre strips on eitherside of each transect, count the num-ber of associated organisms (inverte-brates).On the back of the station sheet,record any special remarks aboutthat station.

3 In the case of fixed distance counting, the three tape measures can be laid out three metres apart at the beginning of the dive.

1) Stay close to the measuring tapeand do not move away from it inorder to avoid distorting estimatesof the perpendicular distances offish from the transect.

33

Make sure the sheets have been filledout fully and accurately.Roll up the three tape measures.Begin your ascent (see recommenda-tions for diving work).

After the dive:

On the boat:Proceed with initial discussionsimmediately, then complete and cor-rect the sheets, if necessary.Take the sheets off the clipboardsand put them into the correspondingplastic pocket folders.

On land:Rinse the sheets one by one in freshwater, then place them separately todry.Once they are dry, take the sheetsand finish or correct them, as need-ed.Begin data entry.

III.4 – Stationary points

The steps to follow before getting into thewater are exactly the same as those men-tioned for transect counting. This alsoapplies to the after-diving rules and the

Diver 1

Diver 2

0 1 2 3 4 5( Sectors )

50 md 1

d 2

d 1 = d 2

3 m

3 m

10 m 20 m 30 m 40 m

2-D view of a visual census survey using transects

Ancilliary transect (substrate + living organism cover)

Main transect (count fish + substrate + living organism cover)

34

important points to follow during the dive.The specific characteristics associated withstationary point counts are as follows:

While treading water, visually deter-mine the observation point.Without standing on the bottom,come upright in the water and countthe number of fish within the centralcircle (i.e. in a one-metre radiusaround the observation point).Once you are done counting withinthe central circle, stand on the seafloor at the observation point andidentify a reference radial so as tocontinue the census.Continue counting from this refer-ence point while slowly turning in acomplete circle (360°). Follow thesame rules as for transects.Once the fish count is done, unroll a10-m tape measure, centring it onthe observation point. Then unrollthe other two 10-m tape measurestwo metres apart on either side of it.In each of the three transects laidout in this way, identify once everymetre the type of substrate and theliving organism cover below the tape.

Record the results of your observa-tions on the station sheet by puttingcheck marks in the correspondingsections (see Appendix 4)

4.

Once identification is done, move theouter two tape measures 2 metresfarther on, repeating the same oper-ation for the two new transects.In each of the transects you havenow laid out, count the number ofassociated organisms (invertebrates)found within a one-metre strip oneither side of the tape measures.

4 Each transect corresponds to a different section on the data entry sheet.

35

d 1 = d 2

0°

180° d 1d 2

Reference radius

C e n t r a l c i r c l e

Stationary observation point

3-D view of a visual census survey using stationary points

36

d 1 = d 2

0

180

d 1

d 2

2 m

10 m

°

C en t r a

l c i r c l e

Substrate + living organism cover transects(1 reading per meter)

Vertical view of a stationary point visual census survey

WORK DOUBLES AS TRAINING…

You should seize every possible opportunity to calibrate your measurements and improve yoursize and distance estimation skills.

1) When a fish is stationary on the sea floor, note the land marks corresponding to the ends ofits head and tail, then measure how far apart they are with the ruled edge of your clip-board.

2) When moving down the transect, make it a habit to estimate the distance to the next stoppingpoint and then compare this estimate to the real distance as shown by the tape measure.

3) When you move away from the transect to identify a species, try to measure the distance sep-arating the fish from the transect by using as a reference, say, the length of your own body.In this case, be careful not to record any fish that were not observed from the transect.

37

IV.1 – Entering and formatting data

Special attention must be paid to dataentry, coding and formatting of data. Theresults of the assessment will depend onhow well this work is done.

The main purpose of data entry is to allowrapid computer processing of data. Therisks inherent in this operation arise mainlyfrom the additional source of error it rep-resents. In those cases where data areentered directly into a spreadsheet pro-gramme, double entry is strongly advised, ifpossible by two different operators. Thetwo sheets can then be compared for veri-fication purposes, thereby making it possi-ble to limit the sources of error associatedwith transferring data into the computer.

IV – using the data collected

HINTIn order to facilitate data entry, it is betterto have each entry line correspond to anobservation and to make a column for eachparameter in the order in which they appearon the site record sheet whenever possible(see Appendix 3).

IV.2 – Calculating densities, biomassand stock (numeric and weighted)

For variable distance transects or stationarypoints, the density and biomass estimatespresented below are based on the calcula-tion of a mean weighted distance (dm) ofthe individuals in the transect or at the sta-tionary point. For most species, this pro-duces results that are comparable to othermethods that require adjusting a mathemati-cal function to the fish detectability curve inorder to estimate the parameter (a) (seeSection II.2.1).This calculation method alsohas the advantage of being simple.

In order to do this, either side of the tran-sect is divided into one-metre-wide corri-dors, the closest at 0 to 1 metres, the sec-ond at 1 to 2 metres, etc. Depending onthe fish’s distance at the time of observa-tion, it will then be associated with one ofthese corridors and given its median value(i.e. the middle of that category). Forexample, if an indiviudal is recorded at aperpendicular distance of 2 metres fromthe transect, it will be considered to have

38

Calculations can be made for eachseparate species, or all the fish from asingle family, or the study

population using data from one or moretransects. The best overall estimates areobtained by considering all the species takenas a whole and not by calculating the sum ofthe per species estimates. Differencesbetween these two types of estimates can befairly large, especially when the number ofspecies is high.

been seen in the 2 to 3 metre corridor.The

distance value used in the calculation will be 2

+ 0.5 = 2.5 metres, which corresponds to the

median of the 2 to 3 metre category.

For that reason, the mean weighted distanceof a species is given by:

where p: total number of observations (occurrences) ofspecies j (one observation can comprise several individu-als)

nij: number of fish in observation (occurrence) i

(generally i = 1, but it can take on a higher value in theevent of schools)

dij: perpendicular distance of fish i from the tran-sect. In the case of schools, it becomes:

dij =(d1+d2)

2

The estimated density can then be obtainedby:

p

Σnij (dij + 0.5)dmj = i=1 (1)

Σnij

p

Σnij

Dj = i=1 (2)dm

jL

p

Σnij . Wij

Bi = i=1 (3)dmjL

The biomass estimate is obtained throughthe use of length-weight ratios:

where Wi: estimated weight of fish i using length–weightratios

The stock estimate (Q) is made by multiplyingthe biomass B by the stock distribution surfacearea:

Qj = Bj.A (4)

where A: surface area of the biotope or zone where thebiomass B estimate was made.

39

Number Length Distance 1 Distance 2(in cm) (in m) (in m)

1 8 4 42 10 1 22 22 2 27 12 2 42 18 4 63 22 4 44 18 3 43 15 5 62 20 3 44 15 3 42 22 3 44 15 5 61 17 4 42 15 5 66 11 2 41 18 4 410 15 0 31 20 3 37 13 6 8

64 = Total number of fish 48 = Total number of fish between 0 - 5 m

Example of calculating the weighted meandistance, density and biomass:The selected example comes from the results ofobservations made during a dive on 28 November1998 at Station #4 in the Namoui Marine Reserveon Niue. A total of 64 individuals of Ctenochaetus

striatus were recorded on both sides of the tran-sect: (see table opposite)

By applying Equation (1), the weighted distance isequal to:

The mean density estimate for this transect can,then, be obtained by using Equation (2) which pro-duces:

D = 64

= 0,151 individuals/m2

2 x 4,23 x 50

Note: If the same operation had been carried outusing a transect with a fixed five-metre width, densi-ty would have been equal to:

D = 48

= 0,096 individuals/m2

2 x 5 x 50

This value is underestimated compared with thatobtained by variable distance counting.

For this species, using the length – weight ratio W =0.0254L3.027 obtained by Letourneur et al. (1999), itis possible to calculate the mean weight for eachindividual or group of individuals. The result of thiscalculation is presented in the following table.

dmj = 1(4+0,5) + 2(0,5+1,5) +....+ 7(7+0,5) = 4,23 m64

40

IV.3 – Calculating the accuracy ofdensity and biomass estimates

It is difficult to simply estimate the accuracyof a measurement by taking into account allthe sources of error while at the same timeintegrating all the sources of variability into asingle clear calculation. It is, however, possibleto approach estimate accuracy through aninitial descriptive analysis of the samples. Incases such as ours, where the sampling unitis a transect, dispersion (or variance)between transects can be calculated for anygiven biotope and/or point in time.Thismakes it possible to calculate the standarddeviation for density or biomass measure-ments and thus establish comparisons forthe sample between prospected biotopesor two points in time.This calculation givesan idea of the accuracy of estimates on thescale of the target population.

Inter-transect variance (S2) for density isfound by (see definition of variance onpage 24):

S2

=Σ

i (Di - D)2

(5)n

Number Length Individual Weight(n) (in cm) weight (W)

(in g) (in g)

1 8 14 142 10 27 542 22 294 5887 12 47 3292 18 160 3203 22 294 8824 18 160 6413 15 92 2772 20 220 4414 15 92 3692 22 294 5884 15 92 3691 17 135 1352 15 92 1846 11 36 2161 18 160 16010 15 92 9221 20 220 2207 13 60 419

Total weight = 7127

By applying Equation (3), the biomass is equal to:

B = 7127 = 16.9 g/m2

2 x 4.23 x 50 n

D

41

where Di: density in transect i D: mean density of transects for a given biotopen: number of transects

For the biomass, it equals:

S2

=Σ

i (Bi - B)2

(6)n

where Bi: biomass in transect i B: mean density of transects for a given biotopen: number of transects

The square root of the variance (standard devi-ation) is used so that work can be carried outin the same measuring unit as that used for themean. Standard deviation for density isobtained by:

SD = √S2 (7)For biomass, standard deviation equals:SB = √S2 (8)

Model calculationDuring a duty travel in 1998, a census on fish ofinterest for consumption and marketing was con-ducted in the Namoui Marine Reserve on Niue.Twobiotopes were selected, the sub-intertidal reef flatand the fringing reef slope. A total of 16 transectswere made (i.e. eight for each biotope). The esti-

n

B

D

B

mated density and biomass values for each transectand all observed species combined are given in thefollowing table:

Sub-intertidal reef flat Fringing reef slopeTransect Density Biomass Transect Density Biomassno. (ind/m2) (g/m2) no (ind/m2) (g/m2)3 0.194 24.8 1 0.197 41.74 0.327 42.7 2 0.257 61.55 0.449 53.0 6 0.245 32.68 0.782 98.0 7 0.305 67.49 0.355 66.1 10 0.389 69.312 0.281 52.1 11 0.560 99.913 0.315 71.1 14 0.217 29.916 0.419 66.5 15 0.455 61.3Average 0.390 59.3 Average 0.328 57.0

Density variance on the sub-intertidal reef flat wascalculated by applying Equation (5). It equalled:

S2=(0.194-0.390)2 + (0.327-0.390)2+ ... +(0.419-0.390)2

= 0.0278

Standard deviation for the density measurementwas calculated through variance by applyingEquation (7), which gave:

SD = √0.027 = 0.165 individus/m2

42

The estimated density for the sample was equal tothe mean of the measurements plus or minus stan-dard deviation (i.e. 0.390 ± 0.165 individuals/m2).

IV.4 – Mean composition of substrateand living organism cover

A total of 150 features are recorded fortransects and 50 features for stationarypoints. For each category of substrate andliving organism cover, the mean composi-tion is expressed as a percentage of thetotal number of observations. It isobtained by:

CM = nr x100NR

Where: nr : number of readings for each categoryof substrate or covering organismNR: total number of readings

Biotope Sub-intertidal reef flat Fringing reef slopeParameter Density Biomass Density BiomassVariance 0.027 411.2 0.014 462.0Standard deviation 0.165 20.3 0.120 21.5

For the mean substrate composition,the sum of the percentages from allcategories must equal 100%.

IV.5 – Initial analysis of results

IV.5.1 – Comparing results

a) Density and biomassAn initial interpretation of the results caneasily be conducted by plotting on a graphthe means and standard deviation valuesfor the biomasses and densities obtainedfrom different environments or at differenttimes. In the figure below, the mean bio-mass values have been calculated for thevarious reef biotopes that make up theeastern and western lagoons of theNorthern Province of New Caledonia.These means are symbolised by dots ortriangles and the confidence intervals bybars whose ends indicate the minimumand maximum values on either side of themean. On small island reefs, the fact thatthe bars overlap suggests that biomassesare not significantly different. On the otherhand, for the other two environments(fringing and barrier reefs), the valuesobtained on the west coast are significantlyhigher than those on the east coast.Thissame graphic exercise can be conductedusing the results obtained in the model cal-culation in Section IV.3.The sample does

43

not show any differences in density or bio-mass between the values estimated foreither the sub-intertidal reef flat or thefringing reef slope.This type of analysis canallow simple initial comparison of biotopesor sampled areas as well as comparisonsbetween regular time intervals within a sin-gle environment (resource monitoring).Differences can be validated statistically ata later time.

Biom

ass

(in g

/m2 )

Graphic representation of biomass by reef types studied in theeastern and western lagoons of the Northern Province of

New Caledonia.The bars correspond to standard deviationsaround the mean (dots or triangles)

b) Mean composition of the substrate andcovering organisms

Initial comparisons can be made by simplyconverting the values into bar graphs. Inthe figures below, the values for eachcomponent of the substrate and coveringorganisms were estimated on the slopesand flats of two sections of the fringingreef around the island of Niue (Avateleand the Namoui Marine Reserve). Theseresults were expressed in terms of per-centages and then displayed as bar graphs,and show there were no significant differ-ences in the mean composition of thesubstrate and the living organism coverbetween the sites and biotopes. The sub-strate was composed mainly of hard bot-tom with a high predominance of rock(more than 80%). Living organism coverwas about 50% of the total surface area,mainly in the form of small branchingcorals. Coral heads, encrusting corals andlarge branching corals accounted forlower percentages.

400

350

300

250

200

150

100

Fringing reefs Small island reefs Barrier reefs

Eastern zone

Western zone

44

0 0

Graph of the mean substrate’s composition on two sites of the island of Niue

Namoui Marine Reserve Avatele (outside reserve)

0

0

0

0

0

0

Graph of the living organism cover (in %) at two sites of the island of Niue

Namoui Marine Reserve Avatele (outside reserve)

Slope Flat Slope Flat

Slope Flat

60

50

40

30

20

10

0

60

50

40

30

20

10

0

100

80

60

40

20

0Slope Flat

100

80

60

40

20

0

Sponges

Millepora coral

Coral head or encrusting coral

Small branch coral

Large branch coral

Alcyonaria

Brown seaweed

Green seaweed

Seagrass

Coral

Slab

Rock

Large boulders

Small boulders

Debris

Gravel

Coarse sand

Fine sand

Mud

45

IV.5.2 – Correlations between fish and environ-mental factors

Analysing correlations between fish andenvironmental factors makes it possible todetermine possible links between theparameters of the population or section ofthe population (mean species richness, den-sity or biomass) and certain basic elementsof the habitat and environment (substrateor living organism cover components).

An initial empirical approach can be devel-oped very simply. This consists of analysingon a graph the dispersion of dots denotingestimated values of selected populationparameters (e.g. mean species richness,density or biomass) for a group of stations(each dot represents the results from onestation) on the ‘y’ ordinate in relation tothe percentage of a component of thesubstrate or a category of covering organ-isms on the ‘x’ axis. Two cases are possible.In the first instance, the scatterplot is dis-persed and does not reveal any kind ofcorrelation. In the second instance, thedots seem to be non-randomly distributedand it is possible to make a curve with

most of them, thereby revealing a trend.This initial analysis can be strengthened byanalysing the correlation between parame-ters taken in pairs and then trying to arriveat a curve which passes through a maxi-mum number of dots (e.g. by carrying outa regression through the least-squaresmethod). A correlation coefficient canthen be calculated and tested statistically.

46

ySp

ecie

s ri

chne

ssbi

omas

s de

nsity

xSubstrate component or living organism cover

category (e.g. coral cover)

1st case(R = 0.20)

y

x

2nd case(R = 0.93)

Theoretical representations of estimated fish population parameters by the nature of the sub-strate and covering organisms. First case: dispersed scatterplot, no trend; second case: the scat-

terplot shows a trend demonstrated by the straight line.(R = correlation coefficient)

47

V - CONCLUSIONS

The use of variable distance underwatervisual census survey methods can pro-

vide appropriate responses to questionsraised during the assessment of reef andlagoon fish resources.

However, implementing these methodsinvolves paying attention to certain rulesand precautions in order to obtain usableand, therefore, acceptable results. Theserules can seem too numerous or too com-plicated, but this is not the case at all. Theycan be learned easily and gradually.Thereafter, training and regular practice arethe only way of guaranteeing that the skillsare acquired and maintained. Any skill willbecome rusty if it is not practiced on aregular basis.

As for data processing, a large range ofpossible statistical processing and analysistechniques exists. Depending on the infor-mation sought and the questions raised bythe assessment, it is always possible torefer to specialised works in order tobegin using them. But if this should not

prove possible, do not hesitate to requestthe advice of specialists, who can alwaysprocess the data if they are of adequatequality.

IN SUMMARY…Always remember that information collectionand processing are two sides of the same coinand are, therefore, interdependent.

48

49

Acanthuridae

Carangidae

Kyphosidae

Holocentridae

Haemulidae

Caesionidae

Labridae

Lethrinidae

Lutjanidae

Mullidae

Scaridae

Serranidae

Siganidae

Appendix 1. Outlines of the major fish families that are of food and/or commercial interest

Source: Fishes of the Great Barrier Reef and Coral Sea. Randall et al. 1997

50

Appendix 2. List of fish species that are of food and/or commercial interest

ACANTHURIDAEAcanthurus achillesAcanthurus albipectoralisAcanthurus blochiiAcanthurus dussumieriAcanthurus lineatusAcanthurus mataAcanthurus nigricansAcanthurus nigricaudaAcanthurus nigrofuscusAcanthurus olivaceusAcanthurus pyroferusAcanthurus triostegusAcanthurus xanthopterusCtenochaetus binotatusCtenochaetus striatusNaso annulatusNaso brachycentronNaso brevirostrisNaso hexacanthusNaso lituratusNaso tuberosusNaso unicornisNaso vomerParacanthurus hepatusZebrasoma flavescensZebrasoma scopasZebrasoma veliferum

SERRANIDAEAnyperodon leucogrammicusCephalopholis argus Cephalopholis boenack Cephalopholis miniataCephalopholis sexmaculataCephalopholis sonneratiCephalopholis urodeta Cromileptes altivelis Epinephelus areolatus Epinephelus caeruleopunctatusEpinephelus coioidesEpinephelus cyanopodus Epinephelus fasciatus

Epinephelus fuscoguttatus Epinephelus hexagonatusEpinephelus howlandi Epinephelus macrospilosEpinephelus maculatus Epinephelus malabaricus Epinephelus merra Epinephelus ongusEpinephelus polyphekadion Epinephelus rivulatusEpinephelus suilusEpinephelus tauvinaPlectropomus laevis Plectropomus leopardus Plectropomus maculatusVariola louti

SCARIDAEBolbometopon muricatumCetoscarus bicolorChlorurus bleekeriChlorurus sordidusHipposcarus longicepsScarus altipinnisScarus chameleonScarus flavipectoralisScarus forsteniScarus frenatusScarus ghobbanScarus globicepsScarus longipinnisScarus microrhinosScarus nigerScarus ovicepsScarus psittacusScarus quoyiScarus rivulatusScarus rubroviolaceusScarus schlegeliScarus spinus

LETHRINIDAEGnathodentex aureolineatusGymnocranius euanusGymnocranius grandoculisLethrinus atkinsoniLethrinus erythracanthusLethrinus genivitattusLethrinus harakLethrinus kallopterusLethrinus lentjanLethrinus nebulosusLethrinus obsoletusLethrinus olivaceusLethrinus rubrioperculatusLethrinus variegatusLethrinus xanthochilusMonotaxis grandoculis

LUTJANIDAEAprion virescensLutjanus adetiiLutjanus argentimaculatusLutjanus boharLutjanus fulviflammaLutjanus fulvusLutjanus gibbusLutjanus kasmiraLutjanus lutjanusLutjanus monostigmaLutjanus quinquelineatusLutjanus rivulatusLutjanus russelliLutjanus sebaeLutjanus vittaMacolor macularisMacolor nigerPristipomoides multidensSymphorus nematophorus

MULLIDAEMulloides flavolineatusMulloides vanicolensisParupeneus barberinoides

Parupeneus barberinusParupeneus bifasciatusParupeneus ciliatusParupeneus cyclostomusParupeneus dispilurusParupeneus heptacanthusParupeneus indicusParupeneus multifasciatusParupeneus pleurospilosParupeneus pleurostigmaParupeneus spilurusUpeneus tragulaUpeneus vittatus

SIGANIDAESiganus argenteusSiganus corallinusSiganus doliatusSiganus fuscescensSiganus lineatusSiganus oraminSiganus puellusSiganus punctatusSiganus spinusSiganus vulpinus

HAEMULIDAEDiagramma pictumPlectorhinchus chaetodonoidesPlectorhinchus diagrammusPlectorhinchus gibbosusPlectorhinchus goldmanniPlectorhinchus obscurumPlectorhinchus orientalisPlectorhinchus picusPomadasys argenteus

CAESIONIDAECaesio caerulaureaCaesio cuningPterocaesio trilineataPterocaesio tile

51

Appendix 3. Fish census record sheet

LOCATION: PAGE:

DATE

REC. SCIENTIFIC NAME CODE NUM LGT ST D1 D2

LAT. LONG. TRA. DI.

Key: LAT: latitude; LONG: longitude;TRA: transect or stationary point number; DI: diver identification number;REC: record number (optional); CODE: fish code; NUM: number of fish seen; LGT: estimated length (in cm); ST:observation sector (from 0 to 4); D1: distance 1 (in m); D2: distance 2 (in m).

52

Appendix 4. Environmental factors record sheet

Site no.: __________ Site name: _______________________ Visibility (m): __________Latitude (deg/min/sec): __________ Longitude (deg/min/sec): __________Date (day/month/year): ______ Current (none: 0, weak: 1, strong: 2): ____Fringing reef:_________ Middle reef:_______ Barrier reef: ______ Seagrass bed: ______Others (explain): _____________ Windward: ___________________ Leeward:___________Site description (15-20 words) :___________________________________________________

Depth.

Min:

...............

Max:

...............

Min:

...............

Max:

...............

Substrates

Mud . . . . . . . . . . . . . .Sand . . . . . . . . . . . . . . .Gravel and debris 2 mm-5 cm . . . . . . . .Small boulders 5-30 cmLarge boulders 30-100 cmRock . . . . . . . . . . . . . .Slab . . . . . . . . . . . . . . .Dead coral . . . . . . . . .

Mud . . . . . . . . . . . . . .Sand . . . . . . . . . . . . . . .Gravel and debris 2 mm-5 cm . . . . . . . .Small boulders 5-30 cmLarge boulders 30-100 cmRock . . . . . . . . . . . . . .Slab . . . . . . . . . . . . . . .Dead coral . . . . . . . . .

Mud . . . . . . . . . . . . . .Sand . . . . . . . . . . . . . . .

%

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Living Organisms

Seagrass . . . . . . . . . . . . . . . . .Brown algae . . . . . . . . . . . . .Other algae . . . . . . . . . . . . .Alcyonaria (soft coral) . . . . .Encrusting coral . . . . . . . . . .Massive or sub-massive coralBranch coral . . . . . . . . . . . . .Table coral . . . . . . . . . . . . . .Leaf coral . . . . . . . . . . . . . . .Millepora . . . . . . . . . . . . . . . .Mushroom coral . . . . . . . . . .

Seagrass . . . . . . . . . . . . . . . . .Brown algae . . . . . . . . . . . . .Other algae . . . . . . . . . . . . .Alcyonaria (soft coral) . . . . .Encrusting coral . . . . . . . . . .Massive or sub-massive coralBranch coral . . . . . . . . . . . . .Table coral . . . . . . . . . . . . . .Leaf coral . . . . . . . . . . . . . . .Millepora . . . . . . . . . . . . . . . .Mushroom coral . . . . . . . . . .

Seagrass . . . . . . . . . . . . . . . . .Brown algae . . . . . . . . . . . . .

%

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Other

Diademasea urchins

Other seaurchins

Sea cucumbers Other

Diademasea urchins

Other seaurchins

Sea cucumbers Other

Diademasea urchins

No

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

S

0

1

2

53

To find out more…… about tropical fish and invertebrates

In EnglishFishes of the Great Barrier Reef and Coral Sea. John E. Randall, Gerald R. Allen and Roger C.Steene. University of Hawai’i Press, 2840 Kolowalu Street, Honolulu, Hawaii 96822.ISBN 0-8248-1895-4 (1997).

Coral Reef Fishes, Indo-Pacific and Caribbean. Ewald Lieske and Robert Myers. Harper Collins pub-lishers, 77-85 Fulham Palace Road, London W6 8JB, UK. ISBN 0-00-219974-2 (1994).

Micronesian Reef Fishes. Robert F. Myers. Coral Graphics, P.O. Box 21153, Guam Main Facility,Barrigada,Territory of Guam 96921, USA. ISBN 0-3621564-5-0 (1999).

Tropical Pacific Invertebrates. Patrick L. Colin and Charles Arneson. Coral Reef Press, 270 NorthCanon Drive, Suite 1524, Beverly Hills, California 90210, USA. ISBN 0-9645625-0-2 (1995).

FishBase 2000. CD ROM. c/o ICLARM, MCPO Box 2631, 0718 Makati City, Philippines. (This data-base can also be consulted on the Internet at the following URL: http://www.fishbase.org).

Length-Weight Relationship of Fishes from Coral Reefs and Lagoons of New Caledonia – AnUpdate. Letourneur,Y., M. Kulbicki, and P. Labrosse. NAGA, the ICLARM quarterly (21)4: 39-46(1998).

FISHEYE, database on fish in the South Pacific. IRD (formerly ORSTOM). (This database can also beconsulted on the Internet at the following URL: http://noumea.ird.nc/BASE/FISHEYE).

54

In FrenchPoissons de Nouvelle-Calédonie. Pierre Laboute et René Grandperrin. Editions Catherine Ledru, 13 rueVictor Hugo, Baie de l’Orphelinat, 98 800 Nouméa, Nouvelle-Calédonie. ISBN 2-9505784-3-8 (2000).

FISHEYE, base de connaissances sur les poissons du Pacifique Sud. IRD (anciennement ORSTOM),base de données consultable sur Internet à l’adresse suivante :http://noumea.ird.nc/en/BasCo.html

Guide des poissons des récifs coralliens, région Caraïbes, océan Indien, océan Pacifique, merRouge. Ewald Lieske et Robert Myers. Delachaux et Niestlé (Eds), 79 route d’Oron, 1000 Lausanne21, Suisse. ISBN 2-603-00982-6 (1995).

… about Underwater Visual CensusIn EnglishManual for Assessing Fish Stocks on Pacific Coral Reefs. Melita Samoilys (ed). Department ofPrimary Industries, GPO Box 46, Brisbane Qld 4001, Australia. ISBN 0812-000, ISBN 0-7242-6774-3(1997).

Survey Manual for Tropical Marine Resources (2nd Edition). S. English, C.Wilkinson and V. Baker(eds). Australian Institute of Marine Science,Townsville, Queensland, Australia. ISBN 0-642-25953-4(1997).

Evaluation of Sampling Methods for Reef Fish Populations of Commercial and Recreational Interest.Cappo M., Brown I.W. CRC Reef Research Technical Report n°6. ISBN 1 876054 06 9 (1996).

In FrenchÉvaluation visuelle des peuplements et populations de poissons : méthodes et problèmes.M.L. Harmelin-Vivien, J.G. Harmelin, C. Chauvet, C. Duval, R. Galzin, P. Lejeune, G. Barnabé, F. Blanc, R.Chevalier, J. Duclerc, G. Laserre. Revue d’Écologie (Terre Vie),Vol. 40, 1985.