SPATIALLY AND TEMPORALLY PREDICTABLE FISH COMMUNITIES ON CORAL-REEFS

Spatial and temporal variation in the distribution and abundance of si te-attached fish species inhabiting small, isolated patches of coral r eef has been attributed to variability in larval recruitment. However, the relative importance of settlement and postsettlement processes sp ecies inhabiting large sections of contiguous reef, remains to be dete rmined. Here, we examine the degree of spatial and temporal variation in community structure and population density of fish inhabiting secti ons of coral reef varying in size and connectivity. To investigate spa tial variability in fish community structure and population density, w e surveyed 36 sites on contiguous reef and 39 patch-reef sites varying in size and isolation from neighboring patches. Ordination and regres sion analyses indicated that the structure of fish assemblages inhabit ing contiguous reef varied predictably along habitat gradients. In con trast, intrinsic habitat characteristics, such as the shelter availabi lity and the composition of the substratum, were apparently unrelated to the structure of fish assemblages on patch reefs. For sites on cont iguous reef, multiple regression models explained a significant propor tion of spatial variation in the population density of 10 site-attache d and vagile species (including 90% of variation in the density of Pom acentrus moluccensis, a site-attached damselfish). For patch-reef site s, models of spatial variation in population density were significant for six species, five of which were not site attached. The overall pat tern across most species was that patch-reef models were characterized by a reduced r(2) relative to corresponding models of contiguous-reef populations, but the reduction was substantially less for vagile spec ies than for site-attached species. Ordination analysis of temporal va riability in community structure over two years suggested that none of the fish communities at the sites examined was in a consistent state of community succession. For most sites, community structure varied ra ndomly over time, although at some sites, the structure of resident fi sh communities tended towards a stable position in multivariate commun ity space. Across all sites, temporal change in community structure wa s significantly higher during periods of recruitment than at other tim es of the year, although there was little evidence to suggest that rec ruitment was the sole source of temporal variation. At most sites, the structure of fish assemblages fluctuated considerably during nonrecru itment periods. Patterns of temporal variation in the population densi ty of four site-attached species indicated that population increases c orresponding with pulses of recruitment were modified by postsettlemen t processes. For site-attached and moderately vagile species, overall temporal variability in assemblage structure increased significantly a s sites became smaller and more isolated. Temporal variability in asse mblages of highly vagile species was unrelated to survey area and conn ectivity. Overall, the results of the analyses of spatial and temporal variability in fish community structure suggest that species' vagilit y and reef connectivity strongly influence the relative importance of recruitment and postrecruitment processes in determining local populat ion density. Individuals of highly vagile species are able to move amo ng isolated patches in response to habitat preferences or resource ava ilability. Conversely, the continuous shelter provided by contiguous r eef may allow sedentary species to migrate to more favorable areas. We suggest that for many fish species, including vagile species on patch y reef and site-attached species on contiguous reef, patterns in distr ibution and abundance established at recruitment are modified by posts ettlement migration, which enhances the relationship between populatio n density and habitat structure. Thus, while recruitment patterns may explain much of the spatial and temporal variation in populations of s ite-attached fish on small, isolated patch reefs, this result cannot n ecessarily be extrapolated to fish communities inhabiting large sectio ns of contiguous reef.