296 - Using Birdsong Surveillance and Machine Learning via BirdNET to Help Identify Reservoirs of West Nile Virus

Abstract: The enzootic cycle of West Nile virus requires an avian reservoir host. These birds amplify the virus in their bodies, reinfecting mosquitoes that bite them and allowing the disease to spread to humans. Because some bird species are more competent arboviral reservoirs than others, the community composition of birds is thought to affect the distribution of mosquito-borne disease. To investigate the connection between avian community composition and West Nile virus positivity in mosquitoes, St. Tammany Parish Mosquito Abatement District (STPMAD) placed sixteen audio recording devices near mosquito surveillance sites in the summer of 2023. These devices recorded bird calls, which were identified to the species level using BirdNET machine learning software. A subset of bird calls were manually validated by the research team, and these were used to develop 95% confidence thresholds for the BirdNET results. These bird calls were used to determine avian community composition, including the number of competent West Nile reservoir species detected and species richness. West Nile virus surveillance data from the summer of 2023 was compared to the BirdNET avian surveillance to investigate whether remote audio surveillance can be used to identify potential hotspots for avian arboviral amplification.