Quantifying Vocal Activity and Detection Probability to Inform Survey Methods for Barred Owls (Strix varia)

Abstract

Owls can be difficult to detect due to their secretive behavior, typically low calling rate, and low density on the landscape. Low detection probability during surveys can result in an underestimation of the presence and abundance of a species. Thus, optimizing detection probability of surveys targeting owls is necessary to accurately address ecological questions. We used datasets collected in South Carolina, USA, and Alberta, Canada, to investigate how survey detection can be optimized for Barred Owls (Strix varia). We examined seasonal effects on the detection probability of Barred Owls as determined by playback surveys and autonomous recording unit (ARU) surveys, and whether daily patterns of Barred Owl vocal activity could be used to improve the efficiency of ARU surveys. For each survey method, we estimated the number of survey days needed to obtain a seasonal detection probability ≥ 90% of Barred Owls. We found detection probability with playbacks increased as the breeding season progressed. The effect of seasonality on detection probability with ARUs was dependent on the way encounter history was defined. Barred Owl vocal activity peaked twice per night, with one vocalization peak occurring immediately after sunset and another 7–9 hr after sunset. By targeting these vocalization peaks during surveys, we found that we could reduce ARU survey time by 50% and still retain .82% of the original site detections, thereby reducing survey processing time. Although playback surveys were more efficient than ARU surveys at detecting Barred Owls, ARUs have numerous advantages, such as reducing survey effort and disturbance to the target animal. Ultimately, survey designs are dictated by the budget, personnel capacity, study region, and research objectives, but our findings will help researchers plan studies that optimize detection probability and minimize survey cost and effort.