Determining sampling bias and identifying undersampled areas for species of the Phyllomedusidae family in Brazil

Abstract

With 41% of their species threatened, amphibians are the most endangered vertebrate taxon in the world. Brazil, and more specifically the Atlantic Forest, represents a major hotspot of Neotropical biodiversity. The country is home to a high diversity of frogs, particularly those of the Phyllomedusidae family, whose spatial distribution remains poorly understood due to heterogeneous sampling. In this context, I built four objectives: (1) quantifying sampling bias and identifying a potential difference between threatened and non-threatened species, (2) assessing the performance of “Ensembles of Small Models” (ESM), combining bivariate models, in order to improve distribution predictions while reducing overfitting, (3) comparing the rate of undersampling between threatened and non-threatened species, and (4) identifying new suitable areas to survey in order to improve the assessment of the conservation status of Phyllomedusidae.