Vocal identification and monitoring of individual Australian Boobook owls ( Ninox boobook )

The identification of individuals from their vocalisations offers a non-invasive approach to population monitoring, but only if individual vocal signatures are distinct and stay stable over time. We tested whether territorial calls of Boobook owls could meet these requirements and support individual acoustic monitoring in eucalypt open-forest. Six acoustic features were measured in 208 calls from eight reference owls (21–30 calls per bird) and analysed using linear discriminant analysis (LDA). Leave-one-out cross-validation achieved 82.7% classification accuracy, with mean frequency and spectral entropy (Hf) showing the greatest potential for vocal individual discrimination. The trained model was then tested in a field application using recordings from 11 consecutive nights of passive acoustic monitoring at a single study site. We evaluated 22 bouts (sequences of repeated territorial calls) from owls of unknown identity using predefined assignment criteria. A local resident owl was repeatedly re-identified, and two other owls were detected, one previously recorded locally and the other at a neighbouring site. These findings show that these owl calls contain individually distinctive acoustic signatures with potential for longer-term population monitoring, supporting the integration of passive acoustic monitoring, acoustic analysis, and machine learning into conservation and management practices, with potential applications to other species.