Identification of forest songbirds at the Chalk River Laboratory

Experts in bird identification, supported by leading-edge software, analyzed 290 hours of birdsong audio recordings to identify the presence of 131 species at the Canadian Nuclear Laboratories site in Chalk River, Ontario. This work can contribute significantly to a broader understanding of bird populations and habitats in Canada.

The Chalk River Laboratories site in Chalk River, Ontario is a key location where the federal government is advancing efforts to better understand bird populations in Canada. Situated on the border between Ontario and Quebec approximately 200 km northwest of Ottawa, the 3,700-hectare area provides heavily wooded habitat for a wide range of bird species. The data gathered there will support conservation efforts in both provinces.

CIMA+ was retained by Canadian Nuclear Laboratories to process and analyze avian acoustic data collected by 34 autonomous recording units (ARUs) located across the site. Deployed strategically each spring in a diverse range of habitats and retrieved in mid-July, the ARUs capture bird vocalizations during the primary breeding period. CIMA+ interpreted the audio files obtained from the ARUs, identified unique bird vocalizations, and produced data on species detections and breeding evidence. The data collected from 2023 to 2025 will contribute to the Ontario Breeding Bird Atlas (OBBA), a large-scale, volunteer-based scientific survey that maps bird distribution and breeding status across the province.

Blending human expertise with leading machine-learning technologies

Acoustic monitoring of birds is a powerful tool for assessing ecosystem health, understanding the impacts of climate change and identifying the habitats of Species at Risk (SAR). ARUs allow for the collection of large volumes of data at large spatial and temporal scales for a better understanding of species distribution and habitat requirements, especially in remote areas.

To efficiently and accurately process the ARU data, CIMA+ combined the expertise of trained and experienced ornithologists with advanced machine learning tools. The team used an in-house song analysis tool, paired with software like BirdNET and Raven Pro, to accurately and cost-effectively process the large datasets. The software detected 47,943 vocalizations of 131 bird species, including 7 SAR. Human experts played a critical role in validating automated detections, resolving ambiguous recordings, and confirming the presence of sensitive species, ensuring the reliability and accuracy of the results.

Contributing to the protection of SAR

A key target of the project was to confirm the presence of any federally listed SAR birds at the site. These species are classified as Endangered, Threatened or of Special Concern under the federal Species at Risk Act. Data was analyzed following OBBA protocols so that it can be easily incorporated into OBBA models, publications and outputs. This will support the federal government in evaluating critical habitat for SAR and making species presence/absence determinations on project sites across Ontario.