Community Disassembly in a Fragmented Tropical Landscape Driven by Both Deterministic and Stochastic Processes.

Deforestation is a key driver of habitat loss, transforming extensive forested areas into fragmented, isolated patches with reduced biodiversity. While the patterns of species loss from fragmentation are well documented, the underlying processes driving these patterns remain unclear. We sought to identify the community processes driving the disassembly of tropical insectivorous bat communities in response to forest fragmentation in Malaysia. We measured species richness and four functional diversity metrics across assemblages in continuous forests and forest fragments of varying sizes. Eight traits related to prey detection, acquisition, and processing were used to characterize functional diversity based on a global pool of captured species. We found that species-poor assemblages represented nested subsets of species-rich assemblages, indicating that species loss is non-random. This non-random loss led to a collapse of functional trait space between 11 and 8 species before stabilizing at a lower richness. Analyses of functional diversity against null expectations showed that assemblages in continuous forests were structured by environmental filtering and niche packing, whereas persistence in fragments was driven by stochastic processes. This pattern, alongside the random occupation of fragments, suggests that fragmentation-driven disassembly likely arises from a complex interplay between deterministic and stochastic processes. Insights regarding the relative roles of determinism and stochasticity presented herein highlight the collective contribution of habitat fragments to overall landscape-level diversity and underscore the challenges in identifying priority fragments for conservation. They also emphasize the importance of incorporating functional diversity, rather than solely fragment size and species counts, in landscape-level conservation planning.