Passive acoustic monitoring outperforms observer-based methods for Australian frog communities

Sebastian Hoefer , Slade Allen-Ankins, Donald T McKnight, Eric Nordberg, Lin Schwarzkopf

Effective biodiversity monitoring is fundamental for evaluating conservation status and detecting population declines, yet traditional observer-based monitoring (OBM) is often constrained by high costs and logistical challenges resulting in limited spatial and temporal coverage. Passive acoustic monitoring (PAM) offers a scalable alternative, but its efficacy for frog biodiversity assessments remains largely unexplored. In this study, we compared the effectiveness of PAM (combined with BirdNET embeddings) to OBM for assessing frog biodiversity across six open eucalypt woodland sites in eastern Australia. Using embeddings from the BirdNET deep-learning model, we efficiently analysed over 300,000 hours of continuous audio data, detecting 34 frog species. While OBM proved more effective over short-term (28-day) periods due to visual detections, long-term PAM significantly outperformed OBM in total species richness, detecting 48% more species overall. We found that frog activity was highly seasonal, with species accumulating fastest during spring and summer. Financially, PAM was far more cost-effective for long-term monitoring, costing approximately 5 times less than OBM by the end of the study. However, we found that monitoring methods were complementary rather than interchangeable. Consequently, we propose a hybrid monitoring design with short-term targeted OBM surveys to capture the species and individuals that are difficult to detect acoustically, and long-term PAM deployment to capture the full breadth of acoustic diversity. This integrated approach maximises the strength of both monitoring methods, ensuring comprehensive and cost-effective frog biodiversity assessments.