Drivers of distance-decay in bryophyte assemblages at multiple spatial scales: Dispersal limitations or environmental control?

Questions Despite the increasing scientific interest in distance decay of compositional similarity in ecology, the scale dependence of geographical versus environmental control on distance decay of biological communities has not been properly addressed so far. The present work highlights the relative importance of niche-based processes versus dispersal limitations on distance decay patterns of epilithic bryophyte assemblages at different spatial scales. Location Serra de Sintra, central Portugal. Methods We adopted a nested sampling design with 32 selected sampling sites in each of which two clusters, each with five rocks, were surveyed. Each cluster was characterized by a set of 15 macroscale variables, which were divided into environmental and anthropogenic. For each rock eight microscale variables were recorded. Partial Mantel tests were used to assess the relative importance of geographical and environmental distance on community dissimilarity for each grain size (site, cluster, rock). Quantile regressions were used to describe the decay patterns of community similarity with respect to geographical and environmental distances. Ordination analyses and variation partitioning techniques were applied to assess the pure and shared effects of measured variables on bryophyte community composition. Results Environmental distance based upon macroscale predictors was significantly correlated to community similarity, while no significant correlation was found for ecological distance calculated for microscale predictors, except at the largest grain size. The decrease of community similarity with geographical and environmental distance was thus consistently strengthened with increasing sample grain. Compositional variation was best explained by anthropogenic variables. Conclusions The relative importance of environmental versus geographical distance on compositional similarity in epilithic bryophyte communities varies with the spatial scale of the predictors and with the sample grain. The decrease of similarity with increasing distance is related to changes in habitat features, especially those driven by human disturbance, while it is weakly affected by variations in substrate features.