OCEANOGRAPHIC PROCESSES OF CORAL REEFS: Physical and Biological Links in the Great Barrier Reef - Chapter 13 pptx

The Effects of Water Flow around Coral Reefs on the Distribution of Pre-Settlement Fish Great Barrier Reef, Australia John H.. cen-HELIX REEF STUDY Helix Reef 147° 18 E, 18° 38 S is a s

The Effects of Water

Flow around Coral Reefs

on the Distribution of

Pre-Settlement Fish (Great Barrier Reef, Australia)

John H Carleton, Richard Brinkman, and Peter J Doherty

CONTENTS

Introduction 209

Materials and Methods 210

Helix Reef Study 211

Bowden Reef Study 212

Results 213

Helix Reef Study 213

Bowden Reef Study 215

Hydrodynamics 215

Fish Distribution and Abundance 216

Dispersion Model 218

Discussion 219

Acknowledgments 222

References 222

INTRODUCTION

Coral reef fish, with very few exceptions, have planktonic egg, larvae, and pre-set-tlement juvenile stages that vary in duration from weeks to months Most reef fish spawn buoyant eggs that have the potential to be transported many kilometers in wind-driven surface currents before hatching into larvae capable of influencing their dispersal Others lay eggs in protected nests with the subsequent release to the water column of actively swimming larvae or juveniles, thus minimizing the time their off-spring are exposed to the vagaries of ocean or shelf currents and enhancing the

13

209

chances of recruitment back to their natal reef (Jones et al., 1999) In either case, theproblem facing propagules expatriated from coral reefs is one of populationclosure —of finding shallow coral reef habitat suitable for the juvenile/adult phases

in their life cycle

We now know that the physical, chemical, and biological composition of thewater mass in the immediate vicinity of coral reefs is affected by fine-scale currentpatterns generated through the interaction of reef topography with prevailing, far-field currents (see Hamner & Wolanski, 1988 for review) Coral reefs, growing towithin a few meters of the sea surface, act as barriers to the flow of oceanic or shelfcurrents As currents approach they must diverge to flow around the reef edges, cre-ating a zone of relatively stationary water immediately upstream which becomesenriched with nutrients and plankton At the reef face, topographical entrapment oftidal currents by coral buttresses results in the advection of deep water up the reefslope toward the crest On the surface, wave turbulence mixes the chemical and par-ticulate matter from deep layers with shallow wind-driven material just prior to push-ing the mixture across the reef crest and onto the reef flat (Hamner et al., 1988) Asdiverging currents accelerate around the reef, strong longshore currents are generatedclose to and parallel with the reef sides If longshore currents are strong enough, flowseparation occurs adjacent to sharp projections or indentations in the reef margin withthe resultant formation of particulate-rich eddies In the lee of the reef, gyres andeddies, depending on their size, location, strength, and duration, vary in their ability

to retain both neutrally buoyant material such as echinoderm larvae (Black, 1988) andpositively buoyant material such as coral eggs (Willis & Oliver, 1990) A number ofthese reef-associated hydrodynamic processes must also affect the distribution andabundance of pelagic, pre-settlement fish in the near-reef environment, and thereforeimpact on the eventual success/failure of their recruitment back to suitable, coral reefhabitats

To illustrate the role of flow dynamics on the retention of pre-settlement fish, wepresent the findings from two independent studies at two physically distinct platformreefs in the central section of the Great Barrier Reef (GBR) The first of these stud-ies was of short duration and occurred at Helix Reef, a small, topographically simple,oval-shaped reef; and the second, completed over a 3-month period, was at BowdenReef, a considerably larger, topographically more complex, elongated reef.Synchronized light traps moored in close proximity to these reefs produced synopticviews of fish distribution and abundance patterns at various times of the night andstates of the tide By combining information on fish distribution patterns with physi-cal oceanographic data, we gain an insight into which hydrodynamic processes con-tribute most to the retention of juvenile fish near reefs

MATERIALS AND METHODS

The light traps (Figure 1) were three-chambered devices similar in design to thosedescribed by Doherty (1987) These traps have no moving parts and depend upon thebehaviour of photopositive organisms to effect their capture Fish are attracted into

the upper chamber through a number of tapered slits, then by a vertical array of lating lights moved down through the middle chamber and into the lower chamberwhere most of the fish remain alive until collected Upon trap recovery, the catch iswashed from the lower chamber with filtered seawater, concentrated into a smallervolume and fixed in 100% methylated ethanol The lights were activated for three, 1-h periods each night (21:00 to 22:00, 24:00 to 01:00, and 03:00 to 04:00GMT 10:00) around the new moon between November and January when seasonaland lunar spawning patterns produce the largest catches (Doherty, 1991).

oscil-During the summer, the prevailing longshore currents in the central section of theGBR are driven by the East Australia Current (EAC) and flow from the northwest tothe southeast, parallel to the major isobaths along the continental shelf (Andrews

& Furnas, 1986) Tidal currents which flow across the shelf isobaths, flooding to thesouthwest and ebbing to the northeast (King & Wolanski, 1996), modulate the per-sistent southward flow pushing the resultant current more to the east during fallingtide and more to the west during rising tide (Gay & Andrews, 1994) The interaction

of these far-field currents with the variety of reef shapes and sizes found in the tral GBR results in a complexity of flow pattern through the reef matrix

cen-HELIX REEF STUDY

Helix Reef (147° 18 E, 18° 38 S) is a small (800 m diameter), relatively isolated(10 km to the nearest neighbouring reef), platform reef which rises to the surfacefrom a depth of 55 m (Figure 2) The surrounding seafloor is flat, composed ofmud/sand sediment and devoid of any outcrops These topographical features result

in a relatively simple flow regime The persistent, southerly set current splits aroundthe northern margin of Helix Reef, accelerates along the reef flanks, and sets up acounterclockwise-rotating eddy in the lee (Figure 3, modified from Sammarco

& Andrews, 1988) Although tidal modulation of the shelf currents causes the leeeddy to intensify or relax and to change its actual position, only during moderate tostrong southeasterly winds does the eddy degenerate (Sammarco & Andrews, 1989).Pre-settlement fish were collected from the surface at 16 stations on the south-ern, downstream side of Helix Reef Stations A to C were located around the south-ern reef margin within 50 m of the crest, while the remaining 13 traps were moored

in a regular grid pattern at a spacing of 350 m across a northwest to southeast axis(see Figure 2) From the reef edge to the downstream side of the grid was 800 m andfrom the northeast side to the southwest side was 1.4 km Samples were collectedover three consecutive nights covering the new moon period in January 1992 On thefirst night, an attempt was made to clear all traps after each sampling period Thisproved to be logistically very difficult and on the remaining two nights only the 11traps closest to the reef (A to C, 1 to 8) were cleared after each period

To discern pattern in fish associations, the log transformed abundance data viduals h1of trapping) from stations closest to the reef (A to C, 1 to 8) during thethree individual time periods of each night (21:00 to 22:00, 24:00 to 01:00, and 03:00 to 04:00) were subjected to agglomerative, hierarchical clustering techniques(n 80 samples) Bray-Curtis dissimilarity coefficients (Bray & Curtis, 1957) were

(indi-calculated for every possible pair of samples, the resulting association matrix jected to the Ward’s incremental sum of squares fusion strategy (Belbin, 1987) andthe results summarized by a dendrogram Diagnostic routines developed for use withthe Bray-Curtis metric (Cramer values) were applied to the results from the clusteranalysis to determine the level of fidelity of the various fish species to sample group-ings (Abel et al., 1985) Catch rate, number of species, Shannon-Wiener diversityindex (H), and Pielou’s evenness index (J) (Pielou, 1969 and 1975) were deter-mined for each of the sample groupings.

Bowden Reef (147° 56 E, 19° 02 S) is a much larger platform reef than Helix Reef(6.0 km long 3.0 km wide), is crescent shaped with a continuous reef flat along thenorthern, eastern, and southern sides, and has a semi-enclosed, shallow, sandy lagoon(Figure 2) The seafloor surrounding Bowden Reef is flat, smooth, and has a depth of

40 to 50 m

Light-traps were moored at the surface at 13 stations around the circumference

of Bowden Reef and at 4 stations within the lagoon (Figure 2) On the northern, ern, and southern sides, two traps were anchored directly across from each other oneither side of the shallow reef flat with an additional two or three traps moored far-ther out in deeper water, within 100 m of the reef crest The outside near-crest trapswere placed just in front of the breaker zone and the inside traps located just behindthe reef flat Traps were deployed for periods up to seven consecutive nights aroundthe new moon during the months of November to January 1992/1993 The traps wereactivated each night for the same three 1-h periods as in the Helix Reef study, butcleared only once per day and not after each sampling period

east-Concurrently with trap sampling the strength and direction of far-field currentswere measured at half-hourly intervals by two Aanderaa RCM4-S current metersmoored at mid-depth to the west and south of Bowden Reef (Figure 2) Tidal heightdata were collected by Aanderaa tide gauges placed on the seafloor at the base of eachcurrent meter mooring, on the lagoon floor at Bowden Reef, and on two adjacentreefs (Figure 2) In addition, data on the strength of the poleward flowing EAC wereobtained from a current meter moored at a depth of 35 m on the shelf slope seaward

of Myrmidon Reef, approximately 100 km to the northwest Wind data were obtainedfrom a weather station at nearby Davies Reef

The longest continuous sets of far-field current measurements were obtained inNovember and January These data, along with sea level and wind data, were used toforce a two-dimensional, depth-averaged, hydrodynamic numerical model (Wolanski

et al., 1989) This model was considered the most appropriate to simulate the flowfield at the time of biological sampling for a couple of reasons First, the model hassuccessfully reproduced observed current fields at Bowden Reef in previous studies(Wolanski et al., 1989, Wolanski & King, 1990); and second, this relatively simple,two-dimensional model predicts very similar flow patterns at the sea surface, wheremost nocturnal, pre-settlement reef fish occur (Doherty & Carleton, 1996), to com-putationally more complex, three-dimensional models (Wolanski et al., 1997) The

computational grid for the model was a square horizontal mesh of 386 m with the X-axis aligned parallel to the margin of the continental shelf.

To provide an insight into how the reef-associated flow field may affect the tribution of pre-settlement fish around Bowden Reef, the computed current fieldsfrom the two-dimensional hydrodynamic model were applied to a second-orderadvection-diffusion model (Oliver et al., 1992) As in Wolanski et al (1997), the

dis-“simulated fish” were assumed competent and, in addition to passive advection bylocal currents, were given a behavioral repertoire of swimming speeds typical of pre-settlement coral reef fish (0.05, 0.1, and 0.2 ms1: Leis & Carson-Ewart, 1997) Fish,regardless of their size or swimming ability, within 3 km of the northern, eastern, andsouthern sides swam directly toward the reef in response to low frequency soundsgenerated by breaking waves (Leis et al., 1996) and only stopped swimming whenthey were within 800 m of the crest If fish were washed out of this 800-m-wide enve-lope by local currents, they again swam directly toward the crest, stopping when theyreached the seaward margin of the envelope Fish on the western, open-lagoon,

“quiet” side were not allowed a behavioral response to the reef Pre-settlement fishare known to appear in the vicinity of coral reefs immediately following sunset inreadiness for settlement (McIlwain, 1996) To incorporate this behavior into themodel, a plume of pre-settlement fish with a density of 100 fish per cell and extend-ing across the entire width of the upstream model domain was released at 18:30 on

21 November and at 18:45 on 21 January To calculate the time-integrated abundancearound the reef, propagules were counted every 15 minutes during simulation runsalong 13, evenly spaced, 1-km-long transects which projected seaward at right anglesfrom the reef margin These counts were added to running totals for each transect.Replicate abundance data were tested by three-way, fixed-factor analysis of vari-ance for differences between November and January (those months in which flowpatterns were modeled), among stations around the reef circumference (stations 1 to

11, 16, and 17, Figure 2) and among three size classes (small10 mm, medium 10and15 mm, large 15 mm) Prior to analysis, all data were log transformed to sta-bilise variances (Sokal & Rohlf, 1981) and tested for heteroscedasticity by Cochran’sprocedures (Winer, 1971)

RESULTS

HELIX REEF STUDY

Over the three nights of sampling more than 15,000 individual fish were caught, resenting more than 160 species belonging to 31 families However, a large number

rep-of the species occurred only once or twice The dataset was dominated by Clupeidsthat accounted for 68% of all individuals, followed by Pomacentrids (10%), Nomeids(7%), Apogonids (4%), Carangids (3%), and Gobies (2%)

Classification of all 80 samples identified four sample groupings highlightingboth spatial and temporal distribution patterns (Figure 4) The fish community struc-ture at stations close to Helix Reef was consistent through time All samples from sta-tion A, regardless of time of night or night of sampling, clustered together in a distinct

group, while all the samples from station B and approximately 50% of those from tions C, 1, and 2 formed a second near-reef group Fish community structure at sta-tions farther from the reef (stations 3 to 8 and the remaining 50% of samples fromstations C, 1, and 2) varied with time of night Approximately 70% of all late eveningsamples (21:00 to 22:00) formed a cluster distinct from the remaining far-grid sam-

sta-ples Diagnostic routines indicated Spratelloides larvae, Apogonids, and Gobies tributed most in the characterization of near-reef communities and that Psene arafuensis was instrumental in distinguishing the late evening, far-grid community.

con-The relevance of these key taxa in defining fish associations is evident from the

composition of the four sample groupings (Table 1) Spratelloides larvae dominated

the two near-reef groups, Apogonids and Gobies occurred in relatively large numbers

only at station A, and P arafuensis dominated the early evening, far-grid group The

catch rate at station A was almost an order of magnitude higher than the other reef group, which in turn was five to eight times higher than the far-grid groups.Although species richness was highest at station A, the species diversity index was

near-TABLE 1

Composition of the Four Species Groupings from the Cluster Analysis

(H ⴕ ⴝ Shannon-Wiener diversity index and Jⴕ ⴝ Pielou’s evenness index)

the lowest (H 0.75) due to the dominance of Spratelloides larvae (J 0.19).

Species diversity indexes were highest at the far-grid groups (H 2.30 and 2.25)due to an even proportioning of abundance among species (J 0.66 and 0.69).Detailed scrutiny of catch rate data for each of the key taxa illustrates the con-

sistency of spatial and temporal patterns The distribution of Spratelloides larvae in

the lee of Helix Reef becomes evident when the catch data from all three samplingperiods is integrated over the night of 5 to 6 January (Figure 5) In this figure thecolumns and discs represent different information The height and colour of thecolumns are proportional to the number of individuals collected at each station at aparticular time (the numbers over each column are the actual catch rates) The diam-eter of the discs represents the proportion of the total catch, from all stations over the

3 days of sampling, taken at each station, whereas the colour of the disc representsthe percentage of the station catch which was captured at that particular time For

example, the relatively small disc at station 10 indicates that few Spratelloides larvae

were taken at this station—in fact, only one fish—and the bright red colour indicatesthat the single individual captured on the night of 5 to 6 January represents 100%

of the station catch A quick scan of disc diameters indicates that Spratelloides

larvae were most abundant at stations A, B, and 1 Although there was some ral variation in catch rates, the spatial distribution pattern remained fairly con-sistent (Animation 1) Apogonids and Gobies had similar distribution patterns to

tempo-Spratelloides larvae occurring primarily at stations A and B (Animations 2 and 3)

Nomeids and Pomacentrids were distributed quite differently Psene arafuensis

avoided Helix Reef and was always most abundant at far-grid stations (Figure 6).Although this spatial pattern was evident at all times, specific catch rates differedconsistently among sampling periods The late evening samples (21:00 to 22:00)always contained considerably more of this species (Animation 4) Pomacentridswere distributed across the entire sampling grid but were most abundant at stations

A, B, and C Catch rates were highest during the late evening and declined steadilyover the subsequent sampling periods (Animation 5)

Hydrodynamics

Tides during the sampling periods in November and January were similar in boththeir amplitude (~2.5 m) and semi-diurnal nature During November, winds werefrom the north to northwest at 5.5 ms1and a persistent, southeast flow at 0.47 to 0.59 ms1 was recorded on the shelf slope seaward of Myrmidon Reef DuringJanuary, easterly winds prevailed (2.8 to 12.0 m1) and the persistent, southeast cur-rent was slightly weaker (0.26 to 0.49 ms1)

For both November and January, computed current fields around Bowden Reefwere dominated by tidal forcing (Animation 6) and displayed similar characteristics

to previous observations and numerical studies (Wolanski et al., 1989) Current nitudes of 0.3 to 0.4 ms1occurred in the far field during both maximum ebb andflood tides producing zones of strong lateral velocity shear (Figure 7) During flood

mag-tide, the net southerly current bifurcated at the northern end of the reef, acceleratedalong the eastern and western flanks, and recombined to the south, resulting in

a relatively narrow region of reduced velocity immediately adjacent to the reef (Figures 7a and d) During ebb tide, the net northeasterly current bifurcated to thewest of the reef, accelerated around the northern and southern ends, and recombinedsome distance to the east, producing a wide region of still water along the reef face(Figures 7b and e)

The size and strength of tidally generated hydrodynamic features were lated by prevailing winds and the low-frequency, southeasterly shelf flow, both ofwhich varied between November and January The combination of northerly windsand stronger southward shelf flow in November resulted in an enhanced net south-ward flow which, during ebb tide, produced a number of re-circulation featuresincluding the formation of a closed eddy to the southeast of the reef (Figure 7b).Features of this strength were not evident in the computed circulation for January.Removal of tidal effects by temporal averaging of the time-varying currents over atidal cycle revealed a stronger southward flow during November, with a more clearlydefined convergence zone and associated region of relatively reduced velocity to thesoutheast of the reef (Figures 7c and f)

modu-Fish Distribution and Abundance

Almost 50,000 individual fish, representing 45 families and over 300 species, werecaptured at Bowden Reef during the new moon sampling periods in November andJanuary As in the Helix Reef study, a large number of the species occurred only once

or twice However, unlike the Helix Reef study, Pomacentrids were the most dant family comprising approximately 40% of the catch, followed by Clupeids (9%),Apogonids (8%), Blennies (6%), and Gobies (3%)

abun-Clupeids and Nomeids, families instrumental in defining near-reef and far-gridcommunities at Helix Reef, were of less significance at Bowden Reef Nomeids werepoorly represented, occurring only in extremely low numbers at the more exposed

stations, while Clupeids, comprised primarily of Spratelloides gracilis and S catulus, were ubiquitous (Animations 7 and 8) As the effect of local currents on pre-settlement fish was of primary interest, the Clupeidae were removed from allsubsequent analyses

deli-A shift in size frequency distribution toward larger fish in January (p0.001,

2

1059; Figure 8) resulted in a highly significant interaction between months

and sizes (p 0.001, Table 2) Pre-settlement Pomacentrids and Blennies wereslightly larger in January but juveniles of the more open water families, Scombrids,Carangids and Monocanthids, were substantially larger The larger size inPomacentrids appears to be due primarily to a shift in species composition, from a

greater number of smaller species in November (e.g., Chrysiptera rollandi and Dischistodus spp.) toward more medium-sized species in January (e.g., Pomacentrus bankanensis), although some Pomacentrids, such as P coelestis, were larger in

January The other two highly significant interactions (months X stations, and

sta-tions X sizes, p0.001, Table 2) reveal more as to the possible role of hydrodynamicprocesses in the distribution of pre-settlement fish

The interaction between months and stations indicates a significantly differentdistribution pattern around Bowden Reef between November and January InNovember, fish were more abundant at stations 1, 5, 6, 7, and 11 and less abundant atstation 3 (Figure 9) The November flow pattern averaged over a tidal cycle (Figure7c) clearly shows an area of divergence and reduced flow near station 1; a large region

of reduced flow along the reef face surrounding stations 5, 6 and 7; and a narrow band

of still water immediately to the south near station 11 Station 3, to the north of thereef, lies in an area of relatively strong currents The tidally averaged flow regime inThe interaction between stations and sizes denotes a size-dependent distributionpattern around the reef circumference Small fish (10 mm) had the greatest variabil-ity in abundance among stations, large fish (15 mm) had the most uniform distribu-tion, and medium-sized fish (10 to 15 mm) had a distribution similar in pattern tosmall fish but less variable (Figure 10) Small fish were most abundant at stations 7 to

11 and at station 1 During flood tide, stations 8 to 11, which lie along the southern gin, are in a clearly defined convergence zone with an associated reduction in flowvelocity (Figures 7a and d) During ebb tide, these southern stations are in zone of flowseparation and subsequent eddy formation, and station 7, moored on the eastern reefface, is surrounded by still water (Figures 7a and d) As noted previously, the tidallyaveraged flow regime places station 1 in an area of divergence The distribution ofmedium-sized fish closely parallels that of small fish (Figure 10), but differences inabundance among stations are much smaller Large fish, with the exception of station

mar-3, are uniformly distributed Catch data for Chrysiptera rollandi and Pomacentrus coelestis in November illustrate the distribution of a small/medium (mean 9.98 mm)and large (mean 15.09 mm) fish (Animations 9 and 10) A plot of mean size at eachstation for November and January data combined (Figure 11) shows that most fish werecaptured at station 11 and that their mean size was significantly smaller than anywhere

else (p0.05, Games and Howell multiple range procedure; Sokal & Rohlf, 1981)

Note: ns not significant.

** p 0.01.

*** p 0.001.

January, although similar in pattern, was substantially weaker (Figure 7f)

Dispersion Model

The theoretical distribution of pre-settlement fish around Bowden Reef, as predicted

by the second-order, advection-diffusion model, varied between November andJanuary (Figure 12) With the exception of passive, non-swimming particles inNovember, fish were retained along the reef face from the northeast sector (location0.3, Figure 12) around to the southern sector (location 0.8) However, the exact loca-tion of maximum retention, the most abundant size class of fish, and the total num-ber of propagules retained differed between months In November, the primary peak

in retention for all swimming speeds occurred in the northeast sector; a secondary,weaker peak was located on the eastern face near stations 5, 6, and 7; and abundancedeclined steadily around the southeast reef margin toward the southern end (Figure12) The difference in height between the primary and secondary peaks was most pro-nounced for large, faster-swimming fish, indicating a fairly variable distribution, andwas least pronounced for small, slower fish, indicating a more even distribution Themodel also predicted a greater abundance of large rather than medium-sized or smallfish, and a total lack of retention in either the northwest or southwest sectors

In January, a single peak was located to the southeast and a greater number ofmedium-sized and small fish were retained (Figure 12) Within the sector of maxi-mum retention, medium-sized fish (swimming speed of 0.1 ms1) were most abun-dant, while at other locations, such as in the northeast sector, predicted abundancewas proportional to swimming ability Generally, the overall level of retention washigher—the total number of fish was higher, retention occurred in the northwest andsouthwest sectors, and a greater number of passive particles were trapped, especiallyalong the southern margin

Detailed examination of the interaction between local currents and variousswimming abilities may help explain the resultant distribution pattern generated bythe diffusion model If the “simulated fish” were treated as passive particles, theywere either swept past the reef or retained temporarily in regions of reduced velocity,but without an increase in concentration (Animations 11a and 12a) If the fish wereallowed to swim to and remain near the reef, then “hot spots” of increased concen-tration appeared at size-specific locations around the reef Larger, strong-swimmingfish reached the reef first, quickly aggregating at a point on the upstream end in con-centrations exceeding an order of magnitude above that of the incident plume.Smaller, slow-swimming fish spent more time in the prevailing currents and wereadvected farther downstream before making their first contact with the reef Theseaggregations developed more slowly, but were still at concentrations above that of theincident plume

Once near the reef, all fish, regardless of size, were subjected to the same suite

of tidally dominated local currents Multiple, size-specific aggregations which hadformed as the incident plume was wrapped around the reef, were washed back andforth along the reef face by the ebb and flood of the tide Variability in currentstrength pulled at and deformed these aggregations, but still they maintained theircohesion After a period of time, similar distribution patterns emerged for the “hotspots” in each month, although concentrations within size-specific aggregations

differed Converging spatial patterns were most evident in January (Animation 12)and, to a lesser extent, between medium-sized and large fish in November (Animation11) Differences in abundance between size classes were due to a disproportionatelygreater number of small- and medium-sized fish lost from the reef by fast currentssweeping past the southern end (for example: 8:15 to 9:45 on 24/11/1992, Animation

11 and 10:00 to 12:00 on 24/1/1993, Animation 12) The zones of strong, velocity shear were strongest around the ends of the reef in November, resulting in agreater loss of smaller fish at that time Although the location of fish aggregations wasgenerally restricted to still water along the reef face, smaller concentrations of large,stronger-swimming fish were able to maintain their position on the more exposedwestern side

lateral-Inter-month hydrodynamic variability resulted in distinctive dispersal patterns ineach month During the November simulations, passive particles were not retained inthe vicinity of the reef for more than 2 days, zones of aggregation remained on theeastern face, and there was very little trapping on the western side or in the lagoon.However, during January, passive particles were retained along the eastern perimeterduring the entire simulation, aggregations of fish were advected to the western side

by an anticyclonic re-circulation, and there was persistent trapping in the lagoon,albeit at a low concentration

From an Eularian, fixed-reference perspective, the most persistent zones ofretention were in the south and southeast sectors where aggregations of fish generallyremained at or above the concentration of the incident plume

DISCUSSION

The findings of these two independent studies suggest that fine-scale hydrodynamicfeatures generated through the interaction of reef topography with prevailing, far-field currents have a considerable impact on the distribution and abundance of pre-settlement fish around coral reefs

At Helix Reef, station A, located near the indentation in the southern margin,consistently had the highest catch rates and greatest number of species Overlayingour station grid with the flow pattern generated by the time-averaged hydrodynamicmodel of Sammarco and Andrews (1988) places station A at the centre of an eddy(Figure 13) Although this flow regime was developed with the environmental condi-tions that persisted during the mass coral spawning event in November 1983, thecombination of constant northerly winds, prevailing southerly flow, and falling tide

at the time of our sampling would most likely generate a very similar flow pattern.Wind stress on the sea surface has the greatest influence on surface flow and the con-stant northerly winds during this study (2.2 to 8.3 m s1) would have accelerated theflow around Helix Reef and enhanced the counterclockwise-rotating, lee eddy(Sammarco & Andrews, 1988 and 1989) It is most likely, therefore, that the highercatch rates and larger number of fish species at station A are due to the aggregatingeffect of the eddy This is not surprising as eddies are known to concentrate manyforms of meroplanktonic and holoplanktonic organisms behind reefs (Black, 1988;