BULLETIN OF MARINE SCIENCE, 62(3): 919–935, 1998 919 CORAL REEF PAPER PREDICTABILITY OF FISH ASSEMBLAGES ON ARTIFICIAL AND NATURAL REEFS IN BARBADOS Mark Tupper and Wayne Hunte ABSTRACT Fish assemblages on four large (100 m 2 ) artificial reefs were censused in May, June and July, 1986 and again in May 1988. The mean coefficient of variation of species’ relative density decreased from 1986 to 1988, while the mean percent similarity of fish assemblages increased from 1986 to 1988. Fish assemblages on two of the large artificial reefs and on equal areas of the nearest natural coral reef were censused biweekly for 12 mo. Neither reef type (artificial or natural) nor reef location appeared to have an effect on assemblage structure. Fish assemblages on six small (1 m 2 ) concrete block reefs were censused biweekly for 12 mo; three of the six reefs were denuded at 3-mo intervals. Assemblages on these small reefs were less predictable than those on the large artificial and natural reefs. Predictability of fish assemblages on large and small reefs generally increased with number of species present, but was not consistently related to total num- ber of individuals present. Values of assemblage similarity varied from agreement with Australian studies, which generally espouse recruitment-limited, non-equilibrial assem- blage structure, to agreement with other Caribbean studies, which generally espouse stable, resource-limited assemblage structure. It is concluded that the dichotomy of results be- tween Australian and Caribbean studies of reef fish assemblage structure is primarily a result of different sampling methodology and analysis coupled with different species pools, and requires no difference in biological mechanism. A central question in the study of community ecology is determining the relative role of competition, predation and abiotic factors in structuring communities. Some ecolo- gists argue that competition between ecologically similar species for limited resources, usually living space or food, is the major process structuring communities and results in an orderly, predictable species composition (Werner and Hall, 1977; Werner and Gilliam, 1984; Drake, 1990). Others advocate that predation, natural disturbance, or stochastic recruitment variation prevent populations from reaching densities at which intense inter- specific competition occurs (Conner and Simberloff, 1979, 1986; Sousa, 1979, 1981; Grossberg et al., 1982). Chance events thus play a large role in structuring communities, resulting in unpredictable species composition (Roughgarden, 1989; Levin, 1992; Sale, 1991). Coral reef fish assemblages are widely recognized for their high species diversity, which can reach 150 species in a single collection from <50 m 2 (Goldman and Talbot, 1976). It remains unclear how so many species are able to coexist in a limited area. The arguments presented above have led to the formation of several competing hypotheses concerning the processes regulating assemblage structure of coral reef fish. The first, known as the “single-species equilibrium” model, argues that interspecific competition has led to the evolution of precise, species-specific niche requirements that decrease the probability of competitive exclusion, allowing stable coexistence of many species on one reef (Ogden and Ebersole, 1981; Bohnsack, 1983). Therefore, the population density of each species will be limited by the availability of the particular resources that it requires and competi- tion will occur mostly between conspecifics (Mapstone and Fowler, 1988). The conse-