Unstructured spatial gradients in anthropogenic landscapes drive contrasting rodent and bat metacommunities, underscoring the need for multiscale conservation planning

Anthropogenic disturbances disrupt natural gradients, increasing environmental heterogeneity and creating spatially unstructured gradients. These changes alter habitat availability and connectivity, affecting species dispersal, isolating populations, and reshaping community structure through species redistribution, population increases, declines, or losses. Understanding these impacts at multiple scales is crucial for biodiversity conservation in Neotropical forests. Rodents and bats, highly abundant and speciose mammal groups with key ecological roles and contrasting ecological strategies, should be a research priority in anthropogenic gradients. This study examines spatial patterns and mechanisms underlying bat and rodent metacommunity structures along a Neotropical anthropogenic gradient. Over two years, we sampled bat and rodent communities at 13 sites in the tropical forests of Calakmul, Mexico. Contrasting metacommunity structures revealed key factors shaping species distribution in anthropogenic landscapes. Beta diversity was high and driven by species turnover. Rodents followed a Gleasonian structure, while bats exhibited a quasi-Clementsian pattern, with herbivorous and animalivorous bats responding differently. Rodents were more influenced by local habitat attributes, whereas bats were mainly structured by landscape-scale variables. Then, functionally diverse metacommunities exhibit high beta diversity due to species turnover, driven by species sorting, mass effects, and ecological drift. These processes may intensify under anthropogenic disturbance, leading to nested patterns, increased species isolation, stronger mass effects, heightened vulnerability in low-diversity metacommunities, and reduced resilience. These findings highlight the need for (1) integrated landscape-level conservation, (2) preserving diverse patches and secondary forests, and (3) avoiding oversimplified metacommunity models (v.g. analyzing a single spatial scale) that weaken predictive power and conservation planning.