MARINE ECOLOGY PROGRESS SERIES

Vol. 325: 195–203, 2006 Published November 7

Mar Ecol Prog Ser

Large-scale dispersal of the larvae of nearshore

and pelagic fishes in the tropical oceanic waters of
French Polynesia
A. Lo-Yat 1, 2, M. G. Meekan2,*, J. H. Carleton3, R. Galzin1
1
Ecole Pratique des Hautes Etudes, UMR 8046 Centre National de la Recherche Scientifique, Laboratoire d’Ichtyoécologie
Tropicale et Méditerranéenne, Université de Perpignan, 6680 Perpignan CEDEX, France
2
Australian Institute of Marine Science, PO Box 40197, Casuarina MC, Northern Territory 0811, Australia
3
Australian Institute of Marine Science, PMB 3, Townsville MC, Queensland 4810, Australia

ABSTRACT: The larvae of tropical pelagic and reef fishes were sampled from October 1995 to
August 1997 using a very large fry pelagic trawl at sites between 4 and 20° S and 134 and 154° W in
the region of the Society, Tuamotu and Marquesas Archipelagos in French Polynesia. Sites were
selected on the basis of longline catches of adult tuna (a predator of larval reef fishes) and acoustic
surveys of micronekton to a depth of 600 m. A total of 3369 larvae were collected by 93 net tows. Most
were late-stage larvae representing 29 taxa, of which 26 were predominantly reef fishes. Samples
were dominated by acanthurids (43%), pleuronectiform larvae, ostraciids, fistularids, balistids, holo-
centrids, chaetodontids, mullids and carangids. Distance to reefs was the principal factor determining
clustering patterns in classification and ordination analyses, and offshore samples (>150 km from
reefs) were dominated by pelagic species, while collections nearer to reefs were dominated by coral
reef species. There was a linear decline in richness of samples with increasing distance from reefs
and an exponential decline in abundance, so that few reef fish larvae were collected > 300 km from
adult reef habitats. However, abundances of up to hundreds of reef fish larvae per hour of net tow
were collected >100 km from the nearest reef. Our results show that relatively little connectivity is
likely among reef systems separated by > 300 km in French Polynesia, but substantial exchange of
larvae within a restricted range of reef fish taxa might occur at smaller spatial scales.

KEY WORDS: Dispersal · Coral reef · Fish larvae · Plankton net · Connectivity · Reef fish
Resale or republication not permitted without written consent of the publisher

INTRODUCTION the dispersal patterns of larval coral reef fishes (for

reviews see Leis 2002, Mora & Sale 2002, also Taylor &
Traditionally, populations of coral reef fishes were Hellberg 2003, Paris & Cowen 2004).
viewed as ‘open’, where pelagic larvae were ex- Modelling studies have made an important contribu-
changed among patches of habitat occupied by adults tion to this debate, as they provide a means to examine
at broad spatial scales (hundreds to thousands of kilo- dispersal of reef fish larvae at very large spatial scales.
metres). A variety of recent evidence has led to a For example, Roberts (1997) modelled the dispersal
reversal of this idea, so that, today, populations are patterns of reef fish larvae based on surface currents
thought to be substantially ‘closed’ (i.e. self-recruiting) over the entire Caribbean Sea, while Cowen et al.
at scales as small as a single coral reef. This shift in (2000) modelled the dispersal of larval reef fishes over
thought has been driven by the outcome of studies in an area of thousands of square kilometres of the east-
the fields of genetics, physical oceanography and lar- ern Caribbean, centered on Barbados. Incorporating
val behaviour, and of tagging and modelling studies of estimates of larval diffusion and mortality and assum-

*Corresponding author. Email: [email protected] © Inter-Research 2006 · www.int-res.com

196 Mar Ecol Prog Ser 325: 195–203, 2006

ing passive dispersal of larvae in surface currents, used to describe the distributions of their micronekton
Cowen et al.’s (2000) model predicted that very few (1 to 10 cm size) prey, which included larval reef fishes.
pelagic larvae would be found >140 km offshore from Unlike traditional approaches to characterising larval
a source population of benthic adults. fish distributions, which sample systematically on a
At present, it is very difficult to validate dispersal spatial or temporal basis, this method effectively used
models, as larvae cannot be tracked in the plankton the open water predators of larval fishes to identify
throughout their entire pelagic lives and distributions places where and times when fish larvae were abun-
of larval reef fishes are rarely sampled in the field at dant in the plankton. This meant that they could be tar-
sufficiently large spatial scales. Some evidence against geted by acoustics and net tows that filtered very large
which model outcomes can be judged is provided by volumes of water. In the present study, we compare dis-
Clarke (1995), who used a large mid-water trawl (3 m tributions of fish larvae sampled using this technique
mouth opening, 6 mm mesh size) to sample larval fish with predictions of models of larval dispersal from reefs
distributions over 40° of latitude in the central equato- and use our results to predict the probable limits of dis-
rial Pacific between Tahiti and Hawai’i. As predicted, persal for some common taxa of reef fishes.
he found that catches of larvae declined with distance
from reefs. However, many taxa were collected up to
300 km from the nearest landmass. Other evidence MATERIALS AND METHODS
that long distance dispersal of reef fish larvae occurs at
least occasionally is shown by the presence of larval Collection. The ECOTAP programme examined the
fish in oceanic waters remote from any reef (Leis 1983, spatial distribution of tunas by longline sampling, and
1984, Victor 1987, Victor & Wellington 2000) and the used acoustics to map the density of the micronekton
colonisation by species (often during climatic events on which tuna feed. Mid-water trawls were then used
such as El Niño) of new reef habitats many hundreds of to characterise micronekton, and the stomach contents
kilometres from potential sources of larval supply of tuna from longlines were examined to confirm the
(Cowen 1985, Robertson & Allen 1996). use of micronekton as prey. Details of the methods and
To date, Clarke’s (1995) study provides the only exam- results of ECOTAP are given in Bertrand et al. (1999,
ple of the distribution of larval reef fishes in tropical 2002). Here, we describe spatial patterns in the compo-
waters at very large scales (hundreds to thousands of nent of larval pelagic and nearshore fishes from net
kilometres); consequently there is little evidence on samples.
which to determine the generality of his results. He sam- The study was conducted from October 1995 to
pled stations at regular intervals (approximately 1° of lat- August 1997 on board the 28 m RV ‘Alis’, between 4
itude) along 3 transects perpendicular to the major cur- and 20° S and 134 and 154° W across the Society,
rent flows in the region. Recent studies show that much Tuamotu and Marquesas Archipelagos (Fig. 1). Larval
small scale (m to km) structure exists in
the distribution of larval fishes in open
water (Kingsford 1993), which such sys-
tematic sampling may miss. Thus, the
extent to which distribution patterns of
larval fishes support recent models of
dispersal and survivorship in the waters
around coral reefs remains unresolved.
Here, we describe the distribution
patterns of the larvae of coral reef fishes
over 16° of latitude and 20° of longitude
in the waters of French Polynesia. Our
data sets were collected as part of a
study on the food and feeding habitats
of tuna (Etude du comportment des
thonidés par acoustique et par pêche:
ECOTAP, Bertrand et al. 1999, 2002).
This program used experimental long-
line fishing to locate tuna. Where this
Fig. 1. Location map. Sampling sites are coded (M: Marquesas; T: Tuamotu; S:
technique was successful, acoustics Society Archipelagos). Oceanic (O) sampling sites were generally > 90 km from
and a very large (17 m mouth opening) the nearest reef. Zones 1 to 3 are defined on the basis of micronekton abun-
midwater fry plankton net were then dance according to Bertrand et al. (1999) (see ‘Materials and methods’)
 Lo-Yat et al.: Dispersal of tropical fish larvae in oceanic waters 197

fish were sampled with a fry pelagic trawl, with a vari- gates; while Zone 3 was located north of 8° S, had a low
able-opening mouth of up to 17 m and a total length of biomass and few micronekton aggregates (Fig. 1).
50 m. The net was constructed of 80 mm mesh at the Samples were further defined by their proximity to
mouth, which decreased to 5 mm mesh at the cod end. island archipelagos (Society, Tuamotu and Marque-
The net targeted concentrations of micronekton sas). Samples taken from sites between the Tuamotu
detected by an echosounder within the upper 600 m of and Marquesas Archipelagos, but which were remote
the water column. A total of 93 net tows from 10 cruises from these island chains (> 90 km, average distance
were analysed for the present study. On average, these 240 km), were defined as ‘oceanic’ (Fig. 1).
tows were of 48 min duration at a boat speed of 3 knots Following this analysis the same data sets were
and covered a distance of 2.42 nautical miles or 4.5 km. subjected to principal coordinate analysis using the
Sites of each net tow are shown in Fig. 1. The net was Bray-Curtis metric and presence/absence data. One-
sampled during the day and at night as well as at vari- way ANOVAs were used to examine the influence of
able depths, due to the changes in the distribution of physical parameters (average SST, salinity, depth and
micronekton. Catches were frozen once washed from time) on the occurrence (number per hour) of reef fish
the net. In the laboratory, samples were sorted, and taxa during the 10 ECOTAP cruises.
each taxonomic group was counted. Fish were identi-
fied to the family level when possible.
At each site a CDT attached to the net was used to RESULTS
sample the salinity, water temperature and density of
the water column. Estimates of sea surface tempera- Composition
ture (SST) were obtained by averaging values at the
start and end of each tow. A total of 3369 fish larvae were collected during the
Statistical analysis. We did not attempt to convert 93 net tows. The majority of these were late-stage lar-
our samples to concentrations of larvae based on the vae (Table 1). Catches were composed of 29 taxa, of
volume of water filtered by the net. Nearly all the fish which 26 belonged to reef fish families, accounting for
captured by the net were late-stage larvae, which have approximately 30% of the families of adult reef fish
highly developed sensory and locomotory systems that that have been recorded in French Polynesia (Randall
make them capable of avoiding even large towed nets 1985). Samples were dominated by acanthurids (43%
(Choat et al. 1993). Thus, despite targeting of sampling of total numbers), pleuronectiform larvae, ostraciids
and the large net used by the ECOTAP programme, (represented by a single species, Lactoria diaphana),
we have almost certainly underestimated the abun- fistularids, balistids, holocentrids, chaetodontids, mul-
dance of such larvae in oceanic waters. Furthermore, lids and carangids. Together these taxa composed 85%
aggregations of microzooplankton were sampled in of catches (Table 1). As the ostraciid was a species that
the water column, which tended to cause rapid clog- was pelagic throughout its entire life history, it was
ging of the cod end where aggregations were dense excluded from analyses that focused on the abundance
and contained abundances of jellyfish. For these rea- of late-stage larvae as related to distance from shore
sons, we expressed catches in terms of numbers of lar- (Clarke 1995).
vae per hour of sampling. These same problems exist,
to varying degrees, in most other studies that use net
tows to sample late-stage larvae (Heath 1992). Spatial patterns
Association among the taxa of reef fishes collected
during the study was examined by agglomerative hier- Nine groups of samples were defined from the clus-
archical classification using Bray-Curtis dissimilarity ter analysis (Fig. 2). The first split separated samples
coefficients and UPGMA fusion strategy (beta = –0.1) that contained only 1 taxon of reef fish (Group 9), while
based on presence/absence data. To interpret the clus- the second split (Group 8) divided a small group of
ter analysis, samples were identified by season (wet — samples collected from the Marquesas Archipelago
November to April, dry — May to October), day (06:00 during the wet season from all other catches. The next
to 18:00 h), night (18:00 to 06:00 h) and ‘zone’. The lat- split separated samples from micronekton Zones 1 and
ter coding was based on previous results of the ECO- 2. Samples from Zone 3 were spread throughout all
TAP programme (Bertrand et al. 1999), where Zone 1 groups. Catches from Zone 2 formed clusters based on
was located south of 13° S, had a very low biomass and sites sampled in the dry season (Group 1) and a mix of
a small number of micronekton aggregates (as deter- sites sampled during the wet season (Groups 2 and 3).
mined from acoustic data); Zone 2 was located Catches from Zone 1 split into clusters formed primar-
between 8 and 13° S, was characterised by the highest ily from samples from the Society Archipelago and
biomass and a large number of micronekton aggre- samples from a mix of sites collected during the wet
198 Mar Ecol Prog Ser 325: 195–203, 2006

Table 1. Catches of nearshore and pelagic larvae collected by net tows in French Polynesia. Shown are the number of times each
taxon was collected during the study (Tows), the total number (No.), the minimum (Min.) and maximum (Max.) standard length
(SL) of larvae in samples and the minimum and maximum distance from the nearest reef where larvae were collected. Total num-
bers collected <150 and >150 km from the nearest reef are shown for each taxon. A summary of Clarke’s (1995) samples from the
central equatorial Pacific between Hawai’i and French Polynesia are provided for comparison. Only those taxa that were also
collected by the present study are shown

Present study Clarke (1995)

Taxon Tows No. SL (mm) Distance (km) No. No. No. SL (mm) Distance (km)
Min. Max. Min. Max. <150 >150 Min. Max. Min. Max.

Reef fish larvae

Acanthuridae 51 1411 5 51 2 341 931 480 25 2 30 35 994
Pleuronectiforms 45 289 10 45 2 341 169 120 25 5 138 44 555
Holocentridae 18 224 12 49 4 185 208 16 8 3 8 44 993
Chaetodontidae 14 133 5 29 4 185 125 8 3 5 12 35 525
Mullidae 15 123 16 94 4 133 123 0 5 11 20 740 1131
Fistularidae 21 98 20 172 2 270 66 32 2 84 86 44 140
Pomacanthidae 8 44 10 22 37 272 29 16 8 7 16 44 496
Lutjanidae 8 40 17 121 2 167 22 18 2 9 10 35 158
Priacanthidae 7 27 9 43 4 211 24 3 1 9 9 272 272
Scorpaenidae 6 25 10 16 2 326 19 6 28 4 29 44 987
Apogonidae 2 22 20 25 98 106 22 0 5 4 25 44 522
Synodonthidae 3 16 38 46 2 144 16 0 8 20 34 24 140
Tetraodontidae 10 12 9 32 4 243 8 4 1 21 21 140 140
Hemiramphidae 8 10 42 57 41 363 1 9 – – – – –
Eleotrididae 2 9 22 33 4 65 9 0 4 5 7 140 140
Sphyraenidae 1 8 32 41 37 37 1 0 – – – – –
Kyphosidae 3 6 18 68 111 211 2 4 – – – – –
Monacanthidae 2 6 28 42 4 96 6 0 – – – – –
Zanclidae 1 6 45 45 15 15 6 0 – – – – –
Lethrinidae 2 6 30 48 22 41 6 0 1 23 23 993 993
Pomacentridae 3 5 15 22 61 241 1 4 13 7 16 35 987
Malacanthidae 2 4 32 60 4 22 4 0 – – – – –
Dactylopteridae 2 3 14 18 22 241 2 1 – – – – –
Serranidae 2 3 23 29 41 211 2 1 89 3 29 24 994
Aulostomidae 2 2 99 107 74 150 1 1 – – – – –
Siganidae 1 1 32 32 18 18 1 0 – – – – –
Balistidae 19 232 7 48 2 211 220 12 1 5 5 1140 1140
Carangidae 15 83 10 80 22 211 22 61 9 3 20 35 1131
Pelagic larvae
Ostraciidae 35 279 6 41 4 444 77 202 6 7 18 140 994
Unidentified 30 240
Total 3369 244

season (Group 7). The cluster of samples from sites in Society Archipelago. The number of taxa (richness)
the Society Archipelago further disassociated into was significantly higher in Group 4, which was mostly
groups of samples collected during the night in the dry composed of samples from sites near the Society
(Group 4) and wet (Group 5) seasons and samples col- Islands (Fig. 4A).
lected during the day (Group 6). Groups 4 and 5 had the greatest occurrence of reef
The results of ordination analysis confirmed the pat- fish larvae (26 and 10% of composition, respectively).
terns identified by cluster analysis (Fig. 3). Dimension These were mostly sampled at night in Zone 1, princi-
1 accounted for 27.5% of the variability in catches. pally at sites in the Society Islands. The lowest propor-
Acanthurids, ostraciids (Lactoria diaphana) and pleu- tion of larval reef fishes occurred in Groups 3 and 9 (0.5
ronectiforms were the 3 larval fish taxa that con- and 0.7%, respectively), both of which were composed
tributed most to the separation of groups in the ordina- of sites at large distances from the nearest reef.
tion. Acanthurids were present in more samples from Distance to the nearest reef was highly correlated
Zone 1 (Groups 4, 5, 6 and 7), while ostraciids were with oceanic sites and the family Ostraciidae, while
present in oceanic samples (Group 1), and occurrences SST, salinity and depth of sampling had lower correla-
of pleuronectiform larvae were higher in Group 2. The tions with the ordination axes. Groups of samples
majority of vectors of taxa pointed towards Groups 4, 5 taken at sites that were the greatest distance from
and 6, which were dominated by samples from the reefs were to the left of the centre of the plot (Groups 1,
 Lo-Yat et al.: Dispersal of tropical fish larvae in oceanic waters 199

significant differences in either rich-

ness or abundance between samples
collected during the day and night
(t-tests, p = 0.32 and p = 0.50, respec-
tively).
Analysis of catches of nearshore fish
larvae (excluding the pelagic ostraciid)
showed a significant linear decline in
richness of samples with increasing
distance offshore (Fig. 6A). The de-
cline in abundance of nearshore fish
larvae on this gradient was even more
striking, with a significant negative
exponential trend in the data set, so
that while 100s of nearshore fish lar-
vae (per hour of tow) were collected
close to reefs, very few were captured
beyond 300 km from the nearest adult
habitat (Fig. 6B).

DISCUSSION

Fig. 2. Dendrogram from classification analysis of larval pelagic and nearshore The ECOTAP programme collected
fishes obtained from net samples in French Polynesia. Zones 1 to 3 defined on > 3000 predominantly late-stage fish
the basis of micronekton abundance according to Bertrand et al. (1999) (see larvae in the waters of French Polyne-
‘Materials and methods’). For descriptions of Groups 1–9 see ‘Results’. Dry: dry
season samples; Wet: wet season; Ocean: oceanic, sample sites > 90 km from the sia. As predicted by dispersal models,
nearest reef; Ma: Marquesas Archipelago; Society: Society Archipelago; ML: there was an exponential decline in
mixed locations; Dep: depauperate sample containing only 1 taxon of reef fish abundance of catches with increasing

2, 3, 8 and 9), while groups closer to reefs were to

the right (Groups 4, 5, 6 and 7). Fig. 4B shows the
average distance for each of these groups.
Average SST at our sampling sites ranged from
26.6 to 29.5°C. Richness of catches was significantly
higher at the lowest temperature (26.6°C) than at
mid- or high temperatures (1-way ANOVA, p =
0.025; Fig. 5A). SST had no significant influence on
the number of fish larvae captured per hour of
sampling (1-way ANOVA, p = 0.71).
Richness of catches increased in samples col-
lected at higher average salinities (1-way
ANOVA, p = 0.00004; Fig. 5B). There was a simi-
lar trend in the number of larvae captured per
hour, although this was non-significant (1-way
ANOVA, p = 0.31).
The depth of net tows ranged from 59 to 274 m.
Richness was significantly lower in the deepest
samples, while samples collected at mid- (140 m)
and shallow (59 m) depths did not differ in rich-
ness (1-way ANOVA, p = 0.02; Fig. 5C). Again, Fig. 3. Results of ordination analysis of larval pelagic and nearshore
fishes obtained from net samples in French Polynesia. Groups (Gr.
there was a similar (but non-significant) pattern in 1 to 8) correspond to clusters of samples formed by classification
the numbers of larvae collected per hour with analysis (see Fig. 2). Physical factors and larval fish taxa contribut-
depth (1-way ANOVA , p = 0.16). There were no ing most to variation in data sets are shown
200 Mar Ecol Prog Ser 325: 195–203, 2006

(A) sampled in the central equatorial Pacific between

Hawai’i and French Polynesia. At smaller spatial scales
(10 to 50 km), a number of tropical studies have
reported declining numbers of larvae along cross-shelf
transects (e.g. Young et al. 1986) or in oceanic waters
offshore of isolated islands (Clarke 1991) or barrier reef
systems (Leis 1991, 1993). To date, one of the major
problems of interpretation of such patterns has been
that these surveys of abundance have been sampled
using small nets that filtered low volumes of water.
Because of the relatively large volumes in which larvae
(B) are diluted offshore, gradients in abundance may sim-
ply reflect relatively low sampling efficiency at off-
shore sites (Clarke 1991, 1995). Furthermore, larvae
dispersed into oceanic water will tend to be older, well
developed forms, which are capable of avoiding nets
(Choat et al. 1993, Clarke 1995), in contrast to the
newly hatched and preflexion forms that dominate
catches closer to shore (Leis 1991, 1993).
The ECOTAP programme confirms the results of these
previous studies and suggests that strong offshore gradi-
ents in abundance of larvae are a real phenomenon, at
Fig. 4. (A) Average richness (number of taxa ± 95% CI), least on a scale of hundreds of kilometres. Unlike earlier
(B) average distance to the nearest reef (km, ± CI) of each work, ECOTAP used a massive plankton net that filtered
group of samples (Gr. 1 to 8) identified by classification analy-
sis (see Fig. 2). Number of samples in each group is shown huge volumes of water to capture larval fish. Sampling
in (A) was also structured to include larval fish predators (tuna)
and sonar, so that the net targeted aggregations of larvae
in the water column. Consequently, our results are ro-
distance from the shore, so that very few reef fish lar- bust as they are based on abundances of larvae that are
vae were collected > 300 km from the nearest adult an order of magnitude greater than those of previous
habitats. This result shows that under the oceano- studies (e.g. Clarke 1995).
graphic conditions encountered by the ECOTAP study, Similar to abundance, richness of our catches also
relatively little connectivity is likely in this region declined with increasing distance from reefs, albeit in
among reef systems separated by such distances. a linear rather than an exponential fashion. Overall,
A steep decline in abundance of larval fishes from richness of our collections was relatively low when
reef habitats with increasing distance offshore appears compared with the study of Clarke (1995). This may
to be typical of many tropical localities. Similar to our have been due to the large mesh size of the net
study, Clarke (1995) recorded a sharp decline in abun- (80 mm, 5 mm mesh in cod end), which was principally
dance of reef fish larvae > 300 km from reefs at sites designed to capture juvenile tunas. Smaller fish larvae

(A) (B) (C)

Fig. 5. Average richness (number of taxa ± 95% CI) of larval pelagic and nearshore fishes at differing average (A) sea surface
temperature, (B) surface salinities and (C) depths collected by net tows in French Polynesia. Numbers of tows are shown over
each bar. Results of least square difference tests are also shown: *significantly different at p < 0.05
 Lo-Yat et al.: Dispersal of tropical fish larvae in oceanic waters 201

(A) larvae, rather than younger forms. However, as the

survivors of demographic processes occurring in the
plankton, patterns in the abundance and distribution
of older larvae may impart more information on disper-
sal than those of younger larvae that are still under-
going, or have yet to experience, significant mortality.
Although there was a strong declining gradient in
abundance of nearshore fish larvae with increasing
distance from the shore, there were still large numbers
(occasionally hundreds per hour) of larvae collected by
the ECOTAP programme >100 km from the nearest
reef. This result suggests that a widespread dispersal
(B) and exchange of larvae is likely among patches of
habitat separated by spatial scales of < 300 km. At this
spatial scale, demographic connectivity of reef fish
populations via larval dispersal may be far more com-
mon than conceded by modelling studies and prevail-
ing views of reef fish population dynamics.
This conclusion does not necessarily apply to all taxa
of reef fishes, as the larvae we collected were a
restricted subset of the species that occur in reef habi-
tats, and, furthermore, included only those that had
larval forms large enough to be retained by the net.
Fig. 6. (A) Richness (number of taxa) and (B) abundance (note Acanthurids, pleuronectiform larvae, ostraciids, fistu-
log scale) of larval nearshore fishes (pelagic larvae excluded)
larids, balistids, holocentrids, chaetodontids, mullids
versus increasing distance from the nearest reef (n = 70).
Least squares regression equation for richness: 4.815 – 0.011 × and carangids together composed 85% of catches,
distance to nearest reef (r = –0.562, p < 0.00001). Least while acanthurids, holocentrids, fistularids, pomacan-
squares regression equation for log(1 + abundance): 1.564 – thids, lutjanids and pleuronectiform larvae accounted
0.004 × distance to nearest reef (r = –0.589, p < 0.00001) for 95% of catches of reef fishes >150 km offshore
(Table 1). Some evidence suggests that the life history
traits of these taxa predispose them to long distance
may have escaped, restricting catches to larger taxa dispersal. For example, otolith studies show that larvae
and pre-settlement larvae. Conversely, larger larvae of acanthurids often have long pelagic lives relative to
were probably captured with greater efficiency those taxa that were not present offshore, such as
through targeted sampling with a large net. Some evi- pomacentrids. Wilson & McCormick (1999) found that
dence for this is provided by a comparison of minimum pelagic larval durations of newly settled acanthurids
and maximum size ranges of larvae. Of the 21 taxa collected on the Great Barrier Reef and in the
common to both, the minimum size of larvae from 18 Caribbean were often > 90 d. In contrast, larval dura-
taxa was smaller (by an average of 7.5 mm) in Clarke’s tions of pomacentrids at these localities were rarely
study than in the ECOTAP collections, while the maxi- > 30 d. Moreover, there seems to be considerable
mum size of larvae from 19 taxa was greater (by an intraspecific variation in larval durations of acan-
average of 31 mm) in our study than in Clarke’s (1995). thurids. Bergenius et al. (2002) found that the pelagic
The escape of small larval forms may also explain the larval duration of Acanthurus chirurgus collected the
lack of wrasse larvae in the ECOTAP catches. Larval morning after settlement varied from 69 to as long as
wrasse have been found in oceanic waters remote 90 d, while Doherty et al. (2004) found a similar range
from reefs (Leis 1983, 1984, Victor 1987), are known to in pelagic duration (64 to 97 d) within a single settle-
be transported large distances by current systems ment cohort of the acanthurid Naso sp. collected on
(Hare & Cowen 1991) and have previously been cap- only a few nights of sampling in French Polynesia.
tured in oceanic waters near French Polynesia (Clarke Lengthy and variable pelagic durations may predis-
1995). However, most species attain relatively small pose larvae to long distance dispersal by extending the
larval sizes (<12 mm standard length, Lo Yat unpubl. period they reside in the plankton where they may be
data) and typically have shallow body depths; thus, entrained by currents away from reef habitats into
they may not have been retained by our nets. Due to open water.
the possibility of significant escapement, our study Variation in the duration of larval life may reflect the
focused on the distribution patterns of late-stage capacity to delay metamorphosis into juvenile forms
202 Mar Ecol Prog Ser 325: 195–203, 2006

and extend the period of ‘competence’, when larvae at- we thank Dr. François-Xavier Bard, Dr. Arnaud Bertrand,
tain the ability to settle into the benthic habitats Christophe Misselis, Stephen Yen Kai Sun and Arsène Stein.
We thank Drs. Mark McCormick, Jeff Leis and the anony-
of adults. Field experiments by McCormick (1999)
mous referees for discussions and comments that substan-
demonstrated delayed metamorphosis in the absence of tially improved early drafts of this manuscript.
close physical association to reefs of late-stage larvae of
the convict surgeonfish Acanthurus triostegus in
French Polynesia. This behaviour is thought to be LITERATURE CITED
shared by other taxa of reef fishes that were also abun- Bergenius MAJ, Meekan MG, Robertson DR, McCormick MI
dant in our offshore samples or have been frequently (2002) Larval growth predicts the recruitment success of a
collected in oceanic waters remote from reefs, notably coral reef fish. Oecologia 131:521–525
pleuronectiform and wrasse larvae (Victor 1986, 1987, Bertrand A, Le Borgne R, Josse E (1999) Acoustic characteri-
sation of micronekton distribution in French Polynesia.
Cowen 1991, May & Jenkins 1992, Cowen & Sponaugle
Mar Ecol Prog Ser 191:127–140
1997). For these taxa, evidence of delayed metamor- Bertrand A, Bard F-X, Josse E (2002) Tuna food habits related
phosis comes from otolith analysis, which reveals re- to the micronekton distribution in French Polynesia. Mar
duced growth during the later stages of larval life (see Biol 140:1023–1037
Leis & McCormick 2002 for review). Species from other Choat JH, Doherty PJ, Kerrigan BA, Leis JM (1993) A compar-
ison of towed nets, purse-seine, and light-aggregation
taxa that do not appear to have the ability to delay devices for sampling larvae and pelagic juveniles of coral
metamorphosis may achieve the same goal of extend- reef fishes. Fish Bull 91:195–209
ing competence in open water by having a ‘staged’ pro- Clarke TA (1991) Larvae of nearshore fishes in oceanic waters
cess of settlement in which the late-stage larvae and ju- near Oahu, Hawaii. NOAA Tech Rep NMFS 101
Clarke TA (1995) Larvae of nearshore fishes in oceanic
veniles associate with floating objects or jellyplankton
waters of the central equatorial Pacific. Pac Sci 49:
prior to settling on reefs (Kingsford 1993). 134–142
The offshore dispersal of a restricted suite of reef Cowen RK (1985) Large scale pattern of recruitment by the
species has important implications for the management labrid Semicossyphus pulcher: causes and implications.
of reef fish populations. Assuming that these larvae are J Mar Res 43:719–742
Cowen RK (1991) Variation in the planktonic larval duration
capable of returning to reefs and surviving the juvenile of the temperate wrasse Semicossyphus pulcher. Mar Ecol
stage, our study suggests that there may be great dif- Prog Ser 69:9–15
ferences in connectivity and, thus, the effective stock Cowen RK, Sponaugle S (1997) Relationship between early
size of fishes inhabiting the same reefs. This implies life history traits and recruitment among coral reef fishes.
In: Chambers RC, Trippel EA (eds) Early life history and
that it is unlikely that a single size of marine protected
recruitment in fish populations. Chapman & Hall, New
areas will be suitable for all reef species. Additionally, York, p 423–449
widely dispersed species are more likely to be resilient Cowen RK, Lwiza KMM, Sponaugle S, Paris CB, Olson DB
to local extinction and disturbance events than those (2000) Connectivity of marine populations: open or closed?
species whose larvae may only be found close to natal Science 287:857
Doherty PJ, Dufour V, Galzin R, Hixon MA, Meekan MG,
reefs, such as most pomacentrids. Conversely, widely Planes S (2004) High mortality during settlement is a pop-
dispersed species, such as acanthurids, chaetodontids ulation bottleneck for a tropical surgeonfish. Ecology 85:
and holocentrids, may require management strategies 2422–2428
that encompass very large spatial scales if they are to Hare JA, Cowen RK (1991) Expatriation of Xyrichtys novacula
(Pisces: Labridae) larvae: evidence of rapid cross-slope
be effective. For this reason, it is important that the
exchange. J Mar Res 49:801–823
frequency of broad-scale connectivity is assessed in Heath MR (1992) Field investigations of the early life stages of
reef fish populations and that the viability and post- marine fish. Adv Mar Biol 28:1–174
settlement survivorship of larvae that have undergone Kingsford MJ (1993) Biotic and abiotic structure in the pelagic
offshore dispersal is determined. Recent developments environment: importance to small fish. Bull Mar Sci 53:
393–415
in genetics and the analysis of otolith chemistry (e.g. Leis JM (1983) Coral reef fish larvae (Labridae) in the East
Swearer et al. 1999) provide a means to investigate Pacific Barrier. Copeia 1983:826–828
the fate of such larvae in reef habitats. Leis JM (1984) Larval fish dispersal and the East Pacific Bar-
rier. Oceanogr Trop 19:181–192
Leis JM (1991) The pelagic stage of reef fishes: the larval
Acknowledgements. The authors thank the scientists, staff biology of coral reef fishes. In: Sale PF (ed) The ecology
and crew of the vessels participating in the fieldwork for of fishes on coral reefs. Academic Press, San Diego, CA,
ECOTAP. We acknowledge the Service de la Pêche de la p 183–227
Polynésie Française, the Institut de la Recherche et du Leis JM (1993) Larval fish assemblages near Indo-Pacific
Développement and the Institut Français de Recherche pour coral reefs. Bull Mar Sci 53:362–392
l’Exploitation de la Mer, which supported the ECOTAP pro- Leis JM (2002) Pacific coral reef fishes: the implications of
gramme. We acknowledge the generosity of the scientists behaviour and ecology of larvae from biodiversity and
who allowed us access to data and to the larval fish samples conservation, and a reassessment of the open population
that were collected as part of this programme. In particular, paradigm. Environ Biol Fishes 65:199–208
 Lo-Yat et al.: Dispersal of tropical fish larvae in oceanic waters 203

Leis JM, McCormick MI (2002) The biology, behavior, and fishes fauna from Clipperton Atoll. Coral Reefs 15:
ecology of the pelagic, larval stage of coral reef fishes. 121–131
In: Sale PF (ed) Coral reef fishes — dynamics and Swearer S, Caselle J, Lea D, Warner R (1999) Larval retention
diversity in a complex ecosystem. Academic Press, and recruitment in an island population of a coral-reef
London, p 171–199 fish. Nature 402:799–802
May HMA, Jenkins GP (1992) Patterns of settlement and Taylor MS, Hellberg ME (2003) Genetic evidence for local
growth of juvenile flounder Rhombosolea tapirina deter- retention of pelagic larvae in a Caribbean reef fish.
mined from otolith microstructure. Mar Ecol Prog Ser 79: Science 299:107–109
203–214 Victor BC (1986) Delayed metamorphosis with reduced larval
McCormick MI (1999) Delayed metamorphosis of a tropical growth in a coral reef fish (Thalassoma bifasciatum). Can
reef fish (Acanthurus triostegus): a field experiment. Mar J Fish Aquat Sci 43:1208–1213
Ecol Prog Ser 176:25–38 Victor BC (1987) Growth dispersal and identification of plank-
Mora C, Sale PF (2002) Are populations of coral reef fish open tonic labrid and pomacentrid reef-fish larvae in the east-
or closed? Trends Ecol Evol 17:422–428 ern Pacific Ocean. Mar Biol 95:145–152
Paris CB, Cowen RK (2004) Direct evidence of a biophysical Victor BC, Wellington GM (2000) Endemism and the pelagic
retention mechanism for coral reef fish larvae. Limnol larval duration of reef fishes in the eastern Pacific Ocean.
Oceanogr 49:1964–1979 Mar Ecol Prog Ser 205:241–248
Randall JE (1985) Fishes. In: Delesalle B, Galzin R, Salvat B Wilson DT, McCormick MI (1999) Microstructure of settle-
(eds) Fauna and flora, a first compendium of French Poly- ment-marks in the otoliths of tropical reef fishes. Mar Biol
nesia sea dwellers. Proc 5th Int Coral Reef Congr, Vol 1. 134:29–40
Antenne Museum — EPHE, Moorea, p 462–481 Young PC, Leis JM, Hausfeld HF (1986) Seasonal and spatial
Roberts CM (1997) Connectivity and management of distribution of fish larvae in waters over the North West
Caribbean coral reefs. Science 278:1454–1457 Continental Shelf of Western Australia. Mar Ecol Prog Ser
Robertson DR, Allen GR (1996) Zoogeography of the shore 31:209–222

Editorial responsibility: Howard Browman (Associate Editor- Submitted: March 1, 2005; Accepted: March 7, 2006
in-Chief), Storebø, Norway Proofs received from author(s): October 6, 2006