No description available.

Few studies of tropical fish communities compare fish distributions across the full range of near-shore tropical marine habitats. As a result, our understanding of tropical marine fish communities is often biased towards habitats preferred by researchers. The distribution and habitat preferences of all 136 fish species occurring at 106 stations in three bays off Tortola in the British Virgin Islands were assessed. At a species level, habitat type was often the key factor influencing fish distribution. Of the 44 species occurring at more than 10 stations, 3 were sand specialists, 5 were strongly associated with mangroves, 7 were dependent on seagrass and algal beds and 14 were found only on the forereef. Eleven species were widely distributed both on the forereef and in bays, and 9 of them showed clear evidence of ontogenetic partitioning. The juveniles all preferred bay habitat types and moved onto the forereef as they approached sexual maturity. At a community level, five distinct fish assemblages were found. The assemblage types were classified according to their distribution: 1) forereef, 2) reef flat, 3) non-mangrove associated seagrass, 4) mangrove associated seagrass and 5) eutrophic areas. Forereef stations were the most species-rich with 24 species per station while eutrophic stations had the most depauperate communities with only 4 species per station. Variation in fish species richness at each station was largely explained by a simple habitat complexity index. It accounted for 70% of the variation in fish species richness and 21% of the variation in fish abundance. Rugosity and variety of growth forms were the most important predictors of species richness, but the height of the habitat architecture was the most useful predictor of fish abundance. Artificial reefs were constructed to test the effects of each habitat complexity variable experimentally. Increasing rugosity, variety of growth forms and percentage hard substrate increased the observed number of species but increasing the variety of hole-sizes, and height had no effect. The only complexity variable that had a significant effect on fish abundance was percentage hard substrate. In addition to the static substrate structure, long-spined sea urchins Diadema antillarum affected fish distribution because small fish shelter from predators in their spines. The urchins increase species richness and abundance in low complexity seagrass beds, but on artificial reefs, where shelter was not a limiting factor, the effect was less pronounced. Organic pollution is another factor that negatively affects the fish community by reducing fish species richness and abundance. Poor water quality often alters the natural habitat, confounding observations, but fish species richness was reduced in polluted areas even when artificial reefs were used as habitat controls. The applications of these findings to the management and conservation of fish in the British Virgin Islands are discussed.