Acoustic Surveys to Inform Invasive American Bullfrog Removal Efforts
The Herpetological Research Team is recording audio in areas where American bullfrogs have invaded to learn more about where and when they are breeding and inform managers tasked with controlling their spread.
During breeding season, male bullfrogs produce loud mating calls, and scientists are taking advantage of this acoustic advertisement. USGS researchers and partners from state and federal wildlife management agencies--with help from non-profit organizations and private landowners-- deployed a network of autonomous recording units—or ARUs—in parts of the Oregon spotted frog’s range in Oregon and Washington. The ARUs are programmed to record audio every day for during the evening hours when male bullfrogs are most actively calling. The raw audio recordings are input in a machine learning model that has been trained to recognize bullfrog calls and predict the likelihood that an audio clip contains a bullfrog call. Thousands of hours of audio are being collected across a large region. Machine learning allows us to monitor these thousands of hours for the presence of bullfrogs without the need for human listeners. We are trying to identify patterns between bullfrog mating and geographic features or climate factors. These patterns can help managers better predict when and where invasion is likely to occur to target efforts such as removal of eggs or pre-breeding adults to slow the spread of the invasive bullfrogs.
Invader Removal and Recovery of a Threatened Amphibian
American Bullfrog Invasion and Impacts on Native Species in the Pacific Northwest
Herpetological Research Team (FRESC)
The Herpetological Research Team is recording audio in areas where American bullfrogs have invaded to learn more about where and when they are breeding and inform managers tasked with controlling their spread.
During breeding season, male bullfrogs produce loud mating calls, and scientists are taking advantage of this acoustic advertisement. USGS researchers and partners from state and federal wildlife management agencies--with help from non-profit organizations and private landowners-- deployed a network of autonomous recording units—or ARUs—in parts of the Oregon spotted frog’s range in Oregon and Washington. The ARUs are programmed to record audio every day for during the evening hours when male bullfrogs are most actively calling. The raw audio recordings are input in a machine learning model that has been trained to recognize bullfrog calls and predict the likelihood that an audio clip contains a bullfrog call. Thousands of hours of audio are being collected across a large region. Machine learning allows us to monitor these thousands of hours for the presence of bullfrogs without the need for human listeners. We are trying to identify patterns between bullfrog mating and geographic features or climate factors. These patterns can help managers better predict when and where invasion is likely to occur to target efforts such as removal of eggs or pre-breeding adults to slow the spread of the invasive bullfrogs.