Patch and landscape effects on forest-dependent dung beetles are masked by matrix-tolerant dung beetles in a mountaintop rainforest archipelago.

Naturally fragmented landscapes provide suitable scenarios through which to investigate patch and landscape effects on biodiversity patterns in areas that are isolated from the disturbances usually associated with human-made fragments. We aimed to investigate the patch and landscape effects on the diversity of forest-dependent and matrix-tolerant dung beetles in a naturally fragmented landscape. We also assessed the influence that seasonal and vegetation variations had on these dung beetles. We sampled dung beetles during two summers and two winters in 14 forest islands of various sizes and shapes within a natural mountainous forest archipelago in southeast Brazil. We measured the patch and landscape variables based on high-resolution multispectral images of circular sectors with radii of 100, 250, and 500 m. We used generalized linear mixed models to relate dung beetle metrics to patch and landscape attributes. The interaction between canopy cover and season influenced both species' richness and abundance of the dung beetle metacommunity. The forest-dependent species' richness increased with greater canopy cover, regardless of the season. Patch attributes (e.g., size, canopy cover, distance to the closest patch, and distance to continuous forest) and landscape attributes (e.g., percentage of forest in the landscape, total edge, number of patches, distance to the nearest neighbor, and shape complexity) had small general effects on dung beetle species as a whole and on matrix-tolerant species in particular. However, these values strongly influenced forest-dependent species' richness, abundance, and temporal beta diversity. The matrix-tolerant species, therefore, mask the effects of patch and landscape effects on forest-dependent species within the mountainous forest archipelago. In other words, the changes in these patch and landscape attributes influenced forest-dependent and matrix-tolerant species differently. Therefore, the evaluation of entire metacommunities may not be helpful when evaluating species-specific responses in mixed landscapes-a fact that impairs the conservation of forest-dependent species.