Vernal pools are ephemeral wetlands whose short hydroperiods make traditional amphibian surveys logistically difficult and temporally biased. To assess climate-linked breeding phenology in these habitats, we deployed autonomous recording units at eight vernal pool sites in western Michigan from March to October 2024. Devices recorded 15-second audio snippets every 5 min during dawn and crepuscular periods. Calls of Wood Frogs (Lithobates sylvaticus) and Spring Peepers (Pseudacris crucifer) were detected using a custom BirdNET Analyzer classifier, and daily calling indices were calculated using a hybrid metric that combines peak-confidence extrapolation with probability-weighted binning. Generalized linear mixed-effects models were used to examine relationships between call densities and climate predictors, including temperature, precipitation, and Julian day, with random effects for site and date. Preliminary patterns suggest species-specific phenological responses shaped by seasonal progression and thermal conditions. Our workflow demonstrates the potential of machine-learning-enabled passive acoustic monitoring to track breeding dynamics in cryptic amphibians and offers a scalable approach for monitoring vernal-pool communities under changing climate conditions. Work supported by Grand Valley State University John Ball Zoo, and the Grand Rapids Public Museum.
Jagger Wicker (Wed,) studied this question.