Simple mechanistic models can partially explain local but not range‐wide co‐occurrence of African mammals

AIM: One hypothesis to explain strong biodiversity–climate correlations is that climate gradients determine individual species ranges, which then sum up to form species richness patterns. If this is true, then climate gradients should not only predict the local occurrence of species (i.e. patterns in the local community) but also co‐occurrence of species across the entirety of their ranges (i.e. the regional community concept). We test this hypothesis for African mammals. LOCATION: Sub‐Saharan Africa. METHODS: We quantified species richness of African mammals in 10,000‐km² quadrats to assess local community patterns and developed a new metric for range overlap in the assemblage dispersion field to quantify the range‐wide overlap of species ranges. To identify potential underlying processes, we used a mechanistic modelling approach to simulate the effects of environmental determinism, dispersal limitation and range positioning on species co‐occurrence patterns in both local and regional communities. RESULTS: Although mechanistic models that included climate gradients, dispersal limitation and range positioning could emulate local patterns of mammal richness accurately, they were poor predictors of the co‐occurrence of species across their entire ranges. In the regional community, patterns of mammal co‐occurrence were more closely linked to biogeographical regions in Africa. CONCLUSIONS: We found little support for the hypothesis that climate modulates individual species ranges, which then sum up to determine species richness. Instead, climate may determine how many mammal species can persist locally, whereas historical processes and species interactions ultimately determine more general patterns of species co‐occurrence.