Which characteristics are important for species identification of birds in flight? Results from a survey of Norwegian birdwatchers

Cover photo: Great Skua Catharacta skua. Photo: Terje Lislevand. A better understanding of how birdwatchers identify species of birds in flight may support the development of machine learning algorithms for automated identification from camera-tracking systems for bird monitoring and mitigation. Norwegian birdwatchers scored the importance of 18 criteria for identifying species of birds in flight in an online anonymous survey. Responses were analysed using an Analytical Hierarchy Process and Bayesian Belief Networks. Species identification was first affected by a seasonal expectation as to which species may be observed during a birding trip. Criteria linked to bird’s appearance were most important for species identification, including plumage colouration or patterns; body and wing shape; beak, neck and tail shape. However, flight pattern and speed may provide additional information. A hierarchical approach to categorisation and species identification may improve processing time of automated algorithms.