Neutral and niche-based processes driving species turnover at large spatial scales

Beta diversity - the turnover in species among sites - influences biodiversity across spatial scales. Estimating beta-diversity as well as understanding the forces explaining it is therefore a major concern in ecology and conservation biology. The main hypotheses regarding the distribution and abundance of species are: (i) uniformity over large areas (absence of turnover), (ii) spatial autocorrelation driven by dispersal limitation (neutral hypothesis) and (iii) spatial structure controlled by environmental factors (niche-based hypothesis). To test these hypotheses, we gathered spatial data from Mediterranean shrubland communities on four continents. In each community, we established thirty twenty-by-twenty meter quadrats, separated by distances ranging from 20 m to 100 km. Within each quadrat, we collected soils for chemical analyses, recorded slope and aspect, and measured presence/absence and percent cover of plant species. Here, we present results from the study in the Californian chaparral. The uniformity hypothesis gains no support: the plant community is spatially structured and the similarity in community composition decreases significantly with geographic distance. Both redundancy analysis and Mantel tests show no correlation between community composition and environmental factors. This suggests that environmental determinism is weak, while dispersal limitation is the major driver of beta-diversity. However, a large proportion of the variation remains unexplained, and we cannot exclude the possibility that community composition is controlled by unmeasured spatially autocorrelated environmental variables. We complete our analysis with a direct comparison of the observed decay in similarity with distance to its theoretical expectation under the neutral model of biogeography.