An innovative study conducted by Oregon State University and the U.S. Forest Service has revealed the potential use of Artificial Intelligence (AI) in analyzing acoustic data to monitor the marbled murrelet and similar species. The marbled murrelet is an endangered bird that inhabits the west coast of British Columbia and Washington, residing in close proximity to puffins and murres. Unlike these species, the murrelets venture deep into mature and old-growth forests, reaching distances of up to 60 miles.
Studying such a reclusive species presents a challenge, as comparable cases are rare, according to Matt Betts, a faculty member at OSU’s College of Forestry and co-author of the study. Betts stated that the next step would involve testing whether the sounds made by murrelets can predict their reproduction and occupancy, although this is still a few steps away.
The research team, led by Adam Duarte from the US Forest Service’s Pacific Northwest Research Station, utilized audio recordings initially installed to monitor populations of northern spotted owls. These recordings were gathered from hundreds of locations managed by federal agencies in the Oregon Coast Range and Olympic Peninsula in Washington. By creating a machine learning system called a convolutional neural network, the researchers were able to detect murrelet calls in the recordings. The findings, published in Ecological Indicators, were tested against known murrelet population data and showed that the recorders and AI achieved an accuracy of at least 90% in identifying murrelets in a given area.
Betts also suggests that the study will pave the way for further research on predicting reproductive success and habitat occupancy based on murrelet sounds.
The marbled murrelet primarily resides in coastal waters, where it feeds on krill, other invertebrates, and forage fish such as herring, anchovies, smelt, and capelin. Successful nests only produce one offspring per year, and the nutrient-rich forage fish are crucial for the proper growth and development of baby murrelets.
These birds are typically found along the West Coast, from Santa Cruz, California to the Aleutian Islands. Currently, the marbled murrelet is designated as a threatened species under the U.S. Endangered Species Act in Washington, Oregon, and California.
However, most of the murrelet detections in this study occurred in areas dominated by mature forests and near coastal habitats, as reported by Duarte. With the constant loss of nesting and resting spots due to predators like Steller’s jays and human-induced habitat degradation, effective monitoring strategies are necessary to protect the marbled murrelet.
Duarte highlights how this AI-driven surveillance system can contribute to species distribution models and serve as a foundation for long-term population monitoring, essential for the conservation of rare and threatened species. By eliminating labor-intensive methods such as telemetry and nest ground searches, this streamlined approach allows for more efficient conservation efforts while maintaining a focus on protecting these species.
