Spatial pattern of the fish assemblage inhabiting a littoral rocky reef was examined in relation with its habitat, by means of spatial (correlograms, Mantel tests) and regression (GLM, RDA) analysis techniques. Several biological variables showed an aggregated spatial pattern, although with different patch sizes -180 m for total abundance and similar to 400 to 500 m for species richness. Habitat complexity, as measured by rugosity, number of medium-sized boulders and verticality, and Posidonia cover showed a small-scale spatial pattern (with a patch size of similar to 120 to 200 m), while the alternation of rock (covered or not by Posidonia) and sand, and also wind exposure, showed a larger scale aggregation (400-500 m). Regarding the relationship among biological and environmental variables on a small spatial scale (10(1) m), the 17 more frequent species showed significant, and usually linear relationships, with descriptors either of complexity, heterogeneity, or both, while habitat structure accounts for a important portion (12-40%, depending on the analysis) of the observed variability in fish assemblage structure. Therefore, spatial structure of biological data would be determined, to some extent, by that of habitat. Habitat heterogeneity of rocky substrate seems to promote species richness in the studied area, so that each bottom type (patches of sand, clumps of Posidonia) embedded in the rocky matrix would provide specific resources for more or less specialized species. For its part, habitat complexity favors at the same time a higher number of species and higher abundances. Habitat structure is thus likely to drive a large part of spatial variability in the distribution and abundance of Mediterranean rocky reef fishes, especially when abundance is assessed on small spatial scales. We argue that knowledge about this "natural" variability can help to separate the influence of habitat from other sources of variation, such as the effect of protection from fisheries.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below