Biogeography of western Mediterranean butterflies: combining turnover and nestedness components of faunal dissimilarity

Aim Unpartitioned dissimilarity indices such as the Sorensen index (beta(sor)) tend to categorize areas according to species number. The use of turnover indices, such as the Simpson index (beta(simp)), may lead to the loss of important information represented by the nestedness component (beta(nest)). Recent studies have suggested the importance of integrating nestedness and turnover information. We evaluated this proposition by comparing biogeographical patterns obtained by unpartitioned (beta(sor)) and partitioned indices (beta(simp) and beta(nest)) on presence data of western Mediterranean butterflies. Location Western Mediterranean. Methods We assessed the regionalization of 81 mainland and island faunas according to partitioned and unpartitioned dissimilarity by using cluster analyses with the unweighted pair-group method using arithmetic averages (UPGMA) combined with non-metric multidimensional scaling (NMDS). We also carried out dissimilarity interpolation for beta(sor), beta(simp), beta(nest) and the beta(nest)/beta(sor) ratio, to identify geographical patterns of variation in faunal dissimilarity. Results When the unpartitioned bsor index was used, the clustering of sites allowed a clear distinction between insular and mainland species assemblages. Most islands were grouped together, irrespective of their mainland source, because of the dominant effect of their shared low richness. bsimp was the most effective index for clustering islands with their respective mainland source. bsimp clustered mainland sites into broader regions than clusters obtained using bsor. A comparison of regionalization and interpolation provided complementary information and revealed that, in different regions, the patterns highlighted by bsor could largely be determined either by nestedness or turnover. Main conclusions Partitioned and unpartitioned indices convey complementary information, and are able to reveal the influence of historical and ecological processes in structuring species assemblages. When the effect of nestedness is strong, the exclusive use of turnover indices can generate geographically coherent groupings, but can also result in the loss of important information. Indeed, various factors, such as colonization-extinction events, climatic parameters and the peninsular effect, may determine dissimilarity patterns expressed by the nestedness component.