One of community ecology’s main goals is to better understand community assembly, which involves teasing apart the ecological, evolutionary, and biogeographic processes that structure communities. Essentially, local communities are a subset of potential colonizers from a regional species pool that successfully pass through abiotic and biotic filters, and ultimately coexist within a given environment. If these local communities are potentially linked together by dispersal, they form a metacommunity. Within the metacommunity framework, the species-sorting model of community assembly proposes that environmental and dispersal filters interactively structure communities. However, the relative roles of these filters remain unclear, and are rarely investigated beyond the scope of a single metacommunity. Using variation partitioning, we tested whether there is universal evidence for the species-sorting model, i.e. whether environmental and dispersal filters jointly structure communities in a consistent way across 10 terrestrial mammal island metacommunities around the globe (the Alexander Archipelago, Indonesia and the Philippines, Melanesia, the Maluku islands, the Greater and Lesser Antilles, the Japanese archipelago, and archipelagos in the Gulf of California, the Adriatic Sea, and the western Mediterranean Sea). We also assess the relative importance of environmental filtering and dispersal filtering both separately and interactively, as well as the influence of stochasticity in the community assembly of these 10 archipelago metacommunities.
We did not find universal evidence supporting the species-sorting model across the 10 archipelago metacommunities. However, we found that environmental and dispersal filters explain a significant portion of compositional turnover between communities in 5 of the 10 metacommunities, including the Alexander Archipelago, the Gulf of California archipelago, the western Mediterranean archipelago, Indonesia and the Philippines, and Melanesia. Environmental and spatial variables explain between 14-59% of beta-diversity within these metacommunities, indicating some variation in the relatively high strength of species-sorting processes. The relative importance of environmental and dispersal filters, both separately and interactively, also differed among these archipelagos. We therefore did not find universal evidence for the species-sorting model, but detected species-sorting processes in half of the considered metacommunities. The mechanisms underlying community assembly therefore appear to vary across the 10 archipelagos. We suggest that these inconsistencies could be related to the composition of the different regional pools, which could modulate the availability of species with suitable traits for local niches and therefore determine the strength of species-sorting processes.