A metacommmunity approach to co-occurrence patterns and the core-satellite hypothesis in a community of tropical arboreal ants

Inferring mechanisms of community assembly from co-occurrence patterns is difficult in systems where assembly processes occur at multiple spatial scales and among species with heterogeneous dispersal abilities. Here, we demonstrate that local scale analysis of co-occurrence patterns is inadequate to fully describe assembly mechanisms and instead utilize a metacommunity and core-satellite approach. We generated a co-occurrence and life-history data set for a community of twig-nesting ants on coffee plants across 36 sites within a tropical agroecosystem to test the following three hypotheses: (1) twig-nesting ant species compete for nest-sites, (2) they are structured as a metacommunity, and (3) core species show segregated patterns, while satellite species show random patterns of co-occurrence. Species were divided into four groups: core species that are well distributed regionally and dominant locally, regional dominants that are well distributed regionally but do not dominate locally, local dominants that are dominant locally but are not widely distributed, and satellites that are neither widely distributed nor dominant locally. Only the most abundant species in the community, Pseudomyrmex simplex, was classified as a core species. Regional dominants, local dominants, and satellite species show random patterns of co-occurrence. However, when P. simplex is included in the co-occurrence matrix, patterns become aggregated for all three species groupings. This suggests that P. simplex “assembles” the community by providing a core metapopulation that other species track. Analyzing co-occurrence patterns among candidate subsets of species, at multiple spatial scales, and linking them to species traits substantially improves the explanatory power of co-occurrence analyses in complex metacommunities.