Coral reef fish represent a broad range of ecological types and vary considerably in their economic importance. Almost without exception these diverse taxa have a pelagic larval stage, which lasts for weeks to months. After settlement from the plankton, most of these species are relatively site-attached and often associated with particular reef habitats. The pelagic larval stage may function as a mechanism for colonizing patchily distributed habitats or unpredictable environments while spreading the risk of local extinction across a wide geographical range. Thus, at large spatial scales (10's to 1000's of km) reef fish communities can be classified as open non-equilibrial systems where spatially divided sub-populations are inter-connected via larval dispersal. The degree of connectivity among reef systems and its importance to fish population regulation is largely unknown.

A recent paper in the journal Science (Roberts 1997) modeled larval transport envelopes based on general surface current patterns and the average larval duration in an attempt to establish the connectivity of Caribbean coral reefs. Despite criticism that oceanographic processes were oversimplified the model provided a basis for testing specific hypotheses regarding the sources and sinks of larval supply and illuminates the need to link up management initiatives where international boundaries cut across potential envelopes of larval transport. Between 1998 and 2000 CMES tested the degree of larval connectivity of fishery resources between the US and British Virgin Islands with a project funded by the University of Puerto Rico Sea Grant College Program. The two year project to identify the spatial and temporal patterns of larval fish supply and settlement was part of a collaborative effort with the International Center for Living Aquatic Resources Management's (ICLARM) Eastern Caribbean office located in Tortola, BVI.

The importance of this study becomes apparent when one becomes familiar with the biology of reef fishes. The arrival and recruitment of these larval fishes is the primary mechanism for replenishing our fish populations. Thus understanding the patterns of larval fish distribution and settlement is extremely important in maintaining our fishery resources. One of the major difficulties in studying larval biology is that larvae are patchily distributed in the plankton and settlement is often episodic and synchronized with the lunar cycle. The correspondence between the abundance of pelagic larvae, their distribution in the plankton and subsequent settlement patterns requires sampling a collaborative research program conducted at large spatial and temporal scales. Our study determined seasonal and annual variation in larval supply and settlement at three spatial scales: between islands (10-1000 km, Tortola, St. Thomas, Jamaica), between reefs (1-10 km, 3 reefs per island) and within reefs (0.1-1 km, three traps per reef). Larval supply was measured using larval fish light traps constructed of 1/4" plexiglass mounted in an aluminum frame (dimensions: 40 cm x 40 cm x 50 cm). A battery-powered miniature fluorescent light (8 watts) was mounted on the top of each light trap in a waterproof tube and a plastic tub with 0.5 mm mesh was mounted on the bottom to concentrate the catch. Monthly sampling was conducted during seven consecutive nights around the new moon at three sites on each island.

During the new moon period of June to September 1999 three light traps were set near each of three fringing reefs in the British Virgin Islands (BVI) and the United States Virgin Islands (USVI). Separation between traps, reefs and islands was 0.1 km, 2-5 km and 50-60 km, respectively and the BVI sites were upcurrent from the USVI sites. The three sites in the US and British Virgin Islands are shown in Figure 1.

Figure 1. British and United States Virgin Islands, shown with the six study sites marked, (Note the outlying islands of Anegada (British) and St. Croix (United States) are not shown on this map).

Catches varied substantially among traps, sites and months. Hierarchical cluster analyses of total catches showed sites within islands usually grouped together each month. Sample catches from sites within the BVI and USVI were clearly distinct (Figure 2) . In the BVI, the largest numbers of almost all taxa were caught consistently at one site. This site, a proposed marine protected area, may be a local 'hotspot' for settlement. Peak abundance of each family generally coincided at BVI sites. By contrast, in USVI no one site consistently produced more fish and abundances of several families peaked at different sites in different months.

Figure 2. Dendrograms for cluster analysis (using average linkage) of total family abundance divided by number of light traps set for June - September 1999. Note Flay Cay and Sprat Bay were not sampled in June.

Mean light trap catch rates over time for the eight most abundant families showed that for individual taxa, considerable variation existed in both the temporal and spatial supply between regions, but also between sites within regions. Furthermore, the overall processes appear to act differently in BVI and USVI. For example, in BVI, the largest pulse of six out of the eight most abundant families was found at Hans Creek. Peak abundance of each family generally coincided at BVI sites, with September being the month of greatest abundance for most families. By contrast, in USVI abundances of several families peaked at different sites and supply of larvae appears more uniformly distributed over the months July - September. For three families, abundance was particularly different between BVI and USVI. In September, 36.6, 2.4 and 4.3 snappers per trap were caught at the three sites in BVI, while USVI yielded only 0.1, zero and 0.4. Snappers were also relatively abundant at BVI sites in July and August, but were virtually absent from USVI samples for all months. For surgeonfish the pattern reversed, with 0.1, 0.9 and 0.8 fish caught from BVI sites in September compared with 27.6, 11.5 and 10.5 from USVI sites. Surgeonfish were more abundant at USVI sites in July and August than at BVI sites. Similarly, jacks were notably less abundant in all months at BVI sites compared to USVI.

Our results suggest variation in the taxonomic composition of supply of larvae appears to cause the different patterns between and within BVI and USVI, rather than variation in the overall abundance of a set assemblage of species. We noted that two of the three families showing the greatest differences between USVI and BVI were deep-bodied fish. At the largest scale, the strong separation of BVI and USVI sites in the cluster analysis argues against the notion of a single source of fish larvae operating over the whole 50-60 km distance, at least during our surveys. Roberts (1997) considered passive dispersal of fishes and grouped USVI, BVI and Puerto Rico within a single 'one month dispersal envelope'. However, he noted the limitations of this approach and suggested that actual interaction distances might be smaller when active fish behavior was considered.

If the differences we observed at the 50-60 km scale were caused by sequential depletion as larvae from a single upstream source passed over the islands' reefs, fewer fish of all species would be expected in the down stream area. Further, a consistent spatial pattern of relative abundance might even be expected at sites within each region. Alternatively, if larval fish settling on BVI and USVI reefs come from separate sources or active behaviors cause markedly different distributions of larvae, a more haphazard pattern in both numerical abundance and species composition is expected. This is the pattern that emerges.

Thus, while we cannot pinpoint sources for the larvae caught during our surveys, our results support the growing body of evidence that reef fish supply of larvae may be strongly variable over only a few kilometers, and quite different assemblages of fish may be supplied to reefs tens of kilometers apart. Such variation may reflect different sources of larvae supplying reefs at this scale, or a greater degree of local retention than was previously thought plausible. If the latter, supply of larvae to the reefs of the BVI and USVI may be less dependent on each other than previous research suggests.

Watson, M. & R.S. Nemeth (ms in prep). Spatial and temporal distribution patterns of coral reef fish larvae in the Virgin Islands archipelago.

Roberts, C.M. 1997. Connectivity and management of Caribbean coral reefs. Science 278: 1454-1457.