Capturing vocal communication in a free-living corvid: high-resolution data from low-impact miniaturized tags

Area of Science:

• Ethology
• Bioacoustics
• Animal Behavior

Background:

• Studying wild bird vocal communication is crucial for understanding social behavior and cognition.
• Traditional audio recording methods in field studies present significant challenges.

Purpose of the Study:

• To introduce and evaluate a novel, lightweight multi-sensor biologging device (MiniDTAG) for high-fidelity acoustic and behavioral data collection in medium- to large-sized birds.
• To assess the device's impact on the birds' behavior and reproductive success.
• To explore the application of machine learning for analyzing recorded vocalizations and behaviors.

Main Methods:

• Deployed 52 MiniDTAGs, integrating microphone, accelerometer, magnetometer, and pressure sensors, on free-ranging carrion crows over three breeding seasons.
• Utilized an auto-releasing attachment mechanism for tag recovery.
• Employed a machine learning model (Voxaboxen) to detect and classify over 127,000 vocalizations.
• Analyzed video data to assess the impact of the device on brood feeding rates and reproductive success.

Main Results:

• Successfully recovered 87% of tags, collecting an average of over 83 hours of data per device.
• The machine learning model accurately assigned vocalizations to focal individuals, conspecifics, chicks, and parasitic nestlings.
• Accelerometer data showed potential for identifying specific behaviors like anti-predator mobbing.
• Minimal impact on brood feeding rates and reproductive success was observed.

Conclusions:

• The MiniDTAG is a promising tool for advancing the study of animal communication, capable of capturing a wide range of vocalization amplitudes.
• This technology opens new research avenues for understanding cooperation and information exchange in complex social systems.
• The findings provide a foundation for future comparative studies in corvid communication and other avian species.