Cues to individuality in Greylag Goose faces: algorithmic discrimination and behavioral field tests

Abstract

Cues to individuality, and the corresponding capacity for individual-level discrimination, can allow individually specific investment by conspecifics into offspring, partners, neighbors or competitors. Here we investigated possible cues to individuality via faces in an ancient avian lineage, the Greylag Goose (Anser anser). Konrad Lorenz could famously name each Greylag Goose in ‘his’ flock from a photograph. Confirming this anecdotal observation, we developed facial recognition software that can reliably (~ 97% accuracy) assign a goose face to a goose ID within a database, using bill morphology normalized during photo preparation. To explore conspecific detection of individuality cues, we erected life-size photos of geese and measured subjects’ responses to photos of themselves (unfamiliar goose), their partner, and another flock mate. Geese displayed significantly greater affiliative response to photos of their partners, providing evidence that geese can use two-dimensional images as cues to determine social category (partner/non-partner) and/or individual-level recognition. Our methods provide novel approaches to automatically detect and monitor geese and to test avian cognition. Our approach may also create new opportunities for species monitoring approaches more generally using photographic images and citizen-science engagement.