Understanding and predicting infection dynamics for an endangered amphibian using long-term surveys of wild and translocated frogs

Amphibians are a prominent component of Earth's sixth mass extinction and the fungal pathogen Batrachochytrium dendrobatidis (Bd) is a primary driver of declines. Although Bd dynamics are well studied, the environmental drivers, exacerbating risk factors, and value of conservation interventions like translocations remain challenging to predict. Here, we present results from two decades of Bd monitoring for mountain yellow-legged frogs (Rana muscosa) in the southern California Transverse and Peninsular mountain ranges. We describe Bd prevalence and infection intensity across sites; model how variables associated with climate, habitat, and populations relate to prevalence; and integrate Bd data from wild and translocated frogs to test whether a machine learning system can predict infection prevalence at new sites. Our findings indicate substantial spatiotemporal variation in Bd dynamics. Bd was present at all sites but prevalence and infection intensities were often low. Environmental features including temperature, precipitation, vegetation, and shortwave radiation explained significant variation in Bd prevalence, but their predictive value varied across mountain ranges. Although clear environmental predictors across populations remain elusive, we provide evidence for the importance of warmer and wetter springs and winters, with implications of increased risk under climate change predictions. We also found evidence for higher Bd prevalence among translocated than wild frogs. Although our machine learning model predicted a Bd prevalence threshold with relatively high accuracy, understanding the factors driving within- and between-population Bd dynamics is complex. Taken together, our findings provide new insights into the complicated role of Bd in amphibian declines and suggest revised management approaches.