Influence of forest cover on amphibian assemblages: a priori grouping of species in multi-species occupancy models

Abstract

The conversion of native forest to human activities reduces and isolates natural habitats, and is linked to biodiversity loss at multiple-scales. Assemblages' responses to forest loss are complex and not unidirectional as some species are negatively affected by forest loss, while others species are favored by forest reduction. A key issue to investigate the effect of forest loss on assemblages is to identify common traits and/or ecological requirements shared by a grouping of species that may lead to a negative or positive response along a gradient of forest cover. The classification of species by a priori grouping is a common approach in the fields of ecology, conservation, and management and could be a useful tool in understanding how communities are distributed and respond to habitat loss. Over the last two decades, amphibians have shown non-random global decline rates that exceed historical extinction levels, and some species are believed to be especially sensitive to human modification of the landscape. However, most studies on this issue use incidence data to fit models, with the assumption that all species are observed without error in the field survey. In reality, species are detected imperfectly and thus the detection probabilities of nearly all species is less than one. If this is not modeled properly, models may have low accuracy and the results can be misleading. Hence, our main goal here is to investigate how a priori groups of amphibians respond to a gradient of natural forest cover, from broad (watershed) and fine scales (catchment), within the Atlantic Rainforest using an approach that accounts for detection error. (AU)