Combining acoustic localisation and high-resolution land cover classification to study predator vocalisation behaviour

ABSTRACT Context. The ecology of cryptic animals is difficult to study without invasive tagging approaches or labour-intensive field surveys. Acoustic localisation provides an effective way to locate vocalising animals using acoustic recorders. Combining this with land cover classification gives new insight into wild animal behaviour using non-invasive tools. Aims. This study aims to demonstrate how acoustic localisation – combined with high-resolution land cover classification – permits the study of the ecology of vocalising animals in the wild. We illustrate this technique by investigating the effect of land cover and distances to anthropogenic features on coyote and wolf vocal behaviour. Methods. We collected recordings over 13 days in Wisconsin, USA, and triangulated vocalising animals’ locations using acoustic localisation. We then mapped these locations onto land cover using a high-resolution land cover map we produced for the area. Key results. Neither coyotes nor wolves vocalised more in one habitat type over another. Coyotes vocalised significantly closer to all human features than expected by chance, whereas wolves vocalised significantly further away. When vocalising closer to human features, coyotes selected forests but wolves showed no habitat preference. Conclusions. This novel combination of two sophisticated, autonomous sensing-driven tools permits us to examine animal land use and behavioural ecology using passive sensors, with the aim of drawing ecologically important conclusions. Implications. We envisage that this method can be used at larger scales to aid monitoring of vocally active animals across landscapes. Firstly, it permits us to characterise habitat use while vocalising, which is an essential behaviour for many species. Furthermore, if combined with additional knowledge of how a species’ habitat selection while vocalising relates to its general habitat use, this method could permit the derivation of future conclusions on prevailing landscape use. In summary, this study demonstrates the potential of integrating acoustic localisation with land cover classification in ecological research.