Field evaluation for playback surveys: species-specific detection probabilities and distance estimation errors in a nocturnal bird community

Abstract

ABSTRACT Capsule During playback experiments, the distance from the surveyor to the call influences the chances of detection across nocturnal species in farmland and woodland habitats. Aim To evaluate how distance affects surveyor detection capability, expressed as the probability of hearing broadcast calls and of estimating their distances correctly, in a nocturnal bird community. Methods We conducted a playback field experiment in farmland and woodland areas within an agricultural landscape in winter and summer 2020. Vocalizations of five species (Little Owl Athene noctua, Tawny Owl Strix aluco, Long-eared Owl Asio otus, Common Nightingale Luscinia megarhynchos, Water Rail Rallus aquaticus) were broadcast at various distances to a surveyor, who attempted to detect them and, if successful, to classify them into predefined distance zones. Binomial GLMMs were used to estimate detection probability as a function of distance, and the effects of habitat and season on this relationship. Results The distance of the broadcast call from the surveyor had a significant effect on detection probability in both habitats. In woodland, the probability of hearing calls was significantly higher in winter, while estimating distances correctly was generally higher in summer. An increase in field experience improved our detection capability, mainly in terms of distance estimation, for which errors were mostly overestimates. Conclusions The probability of hearing calls and distance estimation accuracy varied between species. Most Little and Tawny Owl calls were detected within a 200 m radius of the surveyor, while this was 100 m for the other species. For a multi-species community-level study, playback surveys are thus likely to be most representative of a 200 m radius surrounding the surveyor where the probability of detection is highest, while estimates of distance from the observer are likely to be inaccurate in most cases. Field evaluations such as this should be implemented prior to actual playback surveys.