Abstract

Amphibians are one of the most imperiled groups of vertebrates; many species worldwide are intrinsically susceptible to extinction due to habitat loss, climate change, disease, and various other anthropogenic factors. Deterministic models often fail to capture the complex and diverse nature of uncertainty exhibited in ecological data, particularly for species with limited data. Our study presents a Bayesian modeling framework that estimates extinction risk in critically endangered amphibians, utilizing input from both prior ecological knowledge and limited observational data to produce probabilistic estimates of extinction risk. We developed hierarchical models to generate an unpredictable extinction risk based on species-specific life-history traits, fragmentation indices, and exposure to threats. The Bayesian framework is advantageous as it accounts for the uncertainty of the data and provides an updated extinction risk estimate with new information as it becomes available, which is crucial for the adaptive management of conservation. The model we applied to explore extinction risk across 50 critically endangered amphibian species in various parts of the globe illustrates considerably different extinction risks. Disease prevalence and microhabitat specialization were the two primary predictors of extinction risk for a highly threatened group of vertebrates. We demonstrate the application and utility of Bayesian modeling in the context of developing extinction risk in conservation biology. It affords a statistically robust, transparent, and flexible means to advance the protection of extinct species by prioritizing species and acting with targeted mitigation measures under significant uncertainties