How do bat, rodent, and marsupial communities respond to spatial and environmental gradients? Insights from a deconstruction of mammal beta diversity from the Atlantic Forest of South America.

Abstract

Space and environment are fundamental in influencing the community structure. However, their relative influences vary according to species' biological characteristics. Here we test whether differences in life-history traits mainly linked to dispersal abilities influence bat, rodent and marsupial beta diversity along spatial and environmental gradients. We expect bat beta diversity to be weakly related with space in comparison to dispersal-limited rodents and marsupials. Using data from communities distributed along the Atlantic Forest of South America, we calculated the total beta diversity and its turnover and nestedness components for each group. We estimated the strength of correlation of beta diversity and its components along spatial and environmental gradients, comparing their importance within and between groups. Space had the higher influence on rodent and marsupial beta diversity. For bats, both gradients influenced similarly their community composition. Between taxa, the influence of these gradients did not differ for rodents and marsupials, while bats presented a stronger relationship with environment compared to non-volant small mammals. Also, all groups presented a similar influence of the spatial gradients on their community structure, despite their differences in dispersal abilities. Our results suggest that differences in biological characteristics partially influence the community structure of these mammals, with their responses along space likely reflecting similar biogeographical dynamics affecting their distribution. Overall, our results improve the understanding of the processes structuring these communities, highlighting the benefits of comparative analyses within a beta diversity perspective to better understand the influence of multiple processes on the community assembly along geographical gradients.