Article révisé par les pairs
Résumé : Little is known about the combined impact of habitat filtering and dispersal limitation on species turnover patterns. To gain new insights, we constructed a spatially explicit community model wherein we controlled dispersal distances, the strength of habitat filtering, and the grain of habitat heterogeneity to study the distance decay of several (dis) similarity indices. The impact of habitat filtering is dependent on the ratio between the grain of habitats and the mean dispersal distance. The behavior of (dis) similarity indices varies. First, incidence-based measures of species overlap are less affected by habitat filtering than are abundance-based indices. Second, species identity-based indices, derived from population genetics' F-ST, show interesting capacities to infer dispersal processes under neutrality but are also highly sensitive to habitat filtering. All indices except F-ST-related indices under neutrality are very sensitive to overall species richness. Hence, community patterns showing contrasted diversity levels should be compared with caution. Partitioning similarity indices within and between habitats appears to be an efficient approach to assess the strength of habitat filtering, and we show that a torus-translation test is powerful for this purpose. We finally highlight the need for further analytical studies to achieve theoretical expectations of similarity decay under dispersal and niche processes.