COS 258-4 - Passive acoustic monitoring of mature and immature life stages helps reveal optimal habitat for supporting the persistence of Great Horned Owls (Bubo virginianus) in the eastern United States

Abstract: Preserving habitat which can optimally support wildlife is crucial in maintaining stable populations of animals, but determining which habitat is best is a fundamental challenge of conservation planning. Habitat quality assessments may rely on estimates of presence or abundance, or they may incorporate additional demographic information. Surveys for both types of data have their advantages: presence or abundance is usually easier to estimate than more specific demographic data whereas demographic data is more likely to predict population outcomes. The increasing availability of automated recording units (ARUs) may be able to increase the scalability of gathering demographic data by tapping into signatures of the age or reproductive status of recorded wildlife. Therefore, we have developed a strategy for assessing the reproductive success in birds using adult songs and juvenile food begging calls in ARU recordings. We applied this strategy by investigating adult and reproductive habitat requirements in the Great Horned Owl (Bubo virginianus) found in the eastern United States. We hypothesized that in this species, we would find juveniles in a non-random subset of points compared with where adults are found with respect to characteristics of habitat, suggesting certain requirements of habitat for successful reproduction.

We used breeding season ARU recordings from 490 points in our study area which consisted of private managed forest spanning nearly 900 miles from North Carolina to Maine. We developed Convolutional Neural Networks (CNNs) to identify the audio clips most likely to contain adult and juvenile Great Horned Owl sounds. We manually verified the ten highest scoring clips from each ARU for evidence of adults and juveniles. We used separate static, single-season occupancy models to estimate the effect of habitat covariates on occupancy of adults and juveniles. We found that adult occupancy of Great Horned Owls was significantly correlated with high forest cover, low elevation, and low tree basal area whereas juvenile occupancy was significantly correlated with low forest cover. 95% confidence intervals for the estimates of the effect of forest cover on adult and juvenile occupancy did not overlap which shows a significant difference in the effect of forest cover on these two life stages. These results suggest that habitat requirements of the species are more complex than adult occupancy alone shows. ARUs and CNNs allowed for the rapid assessment of ~20,700 hours of audio in just under 13, proving the viability of our strategy as a scalable method for gathering wildlife demographic data.