Passive acoustic monitoring is an emerging approach to wildlife monitoring that leverages recent improvements in automated recording units and other technologies. A central challenge of this approach is the task of locating and identifying target species vocalizations in large volumes of audio data. To address this issue, we developed an efficient data processing pipeline using a deep convolutional neural network (CNN) to automate the detection of owl vocalizations in spectrograms generated from unprocessed field recordings. While the project was initially focused on spotted and barred owls, we also trained the network to recognize northern saw‐whet owl, great horned owl, northern pygmy‐owl, and western screech‐owl. Although classification performance varies across species, initial results are promising. Recall, or the proportion of calls in the dataset that are detected and correctly identified, ranged from 63.1% for barred owl to 91.5% for spotted owl based on raw network output. Precision, the rate of true positives among apparent detections, ranged from 0.4% for spotted owl to 77.1% for northern saw‐whet owl based on raw output. In limited tests, the CNN performed as well as or better than human technicians at detecting owl calls. Our model output is suitable for developing species encounter histories for occupancy models and other analyses. We believe our approach is sufficiently general to support long‐term, large‐scale monitoring of a broad range of species beyond our target species list, including birds, mammals, and others.
Passive acoustic monitoring using autonomous recording units (ARUs) is a fast-growing area of wildlife research especially for rare, cryptic species that vocalize. Northern Spotted Owl (Strix occidentalis caurina) populations have been monitored since the mid-1980s using mark–recapture methods. To evaluate an alternative survey method, we used ARUs to detect calls of Northern Spotted Owls and Barred Owls (S. varia), a congener that has expanded its range into the Pacific Northwest and threatens Northern Spotted Owl persistence. We set ARUs at 30 500-ha hexagons (150 ARU stations) with recent Northern Spotted Owl activity and high Barred Owl density within Northern Spotted Owl demographic study areas in Oregon and Washington, and set ARUs to record continuously each night from March to July, 2017. We reviewed spectrograms (visual representations of sound) and tagged target vocalizations to extract calls from ~160,000 hr of recordings. Even in a study area with low occupancy rates on historical territories (Washington’s Olympic Peninsula), the probability of detecting a Northern Spotted Owl when it was present in a hexagon exceeded 0.95 after 3 weeks of recording. Environmental noise, mainly from rain, wind, and streams, decreased detection probabilities for both species over all study areas. Using demographic information about known Northern Spotted Owls, we found that weekly detection probabilities of Northern Spotted Owls were higher when ARUs were closer to known nests and activity centers and when owls were paired, suggesting passive acoustic data alone could help locate Northern Spotted Owl pairs on the landscape. These results demonstrate that ARUs can effectively detect Northern Spotted Owls when they are present, even in a landscape with high Barred Owl density, thereby facilitating the use of passive, occupancy-based study designs to monitor Northern Spotted Owl populations.
Northern Spotted Owls (Strix occidentalis caurina) are of management and conservation concern in the US Pacific Northwest where populations have been monitored since the 1990s using mark-resight methods. Passive acoustic monitoring has the potential to support monitoring efforts; however, its use is currently primarily restricted to determining species presence rather than breeding status. Distinguishing female from male Northern Spotted Owl vocalizations could facilitate determination of pair status using passive acoustic methods, greatly enhancing inference derived from noninvasive monitoring data. In 2017, we deployed 150 autonomous recording units (ARUs) within 30 5-km2 hexagons overlapping recently occupied owl territories in Oregon and Washington, USA, where mark-resight methods were simultaneously occurring. We collected approximately 150,000 hours of recordings and detected 22,458 Northern Spotted Owl vocalizations at 76 ARUs. We summarized vocalizations by call type and found differences in the proportion of call types made by single, paired, and nesting owls. We used expert opinion to classify 2812 four-note location calls as female or male. We summarized inter-sex variation within 19 acoustic attributes of the four-note location call and its subcomponents, and developed a mixed logistic regression model to classify sex based on call-segment acoustic attributes. Males generally called at lower frequencies than females, with mean fundamental frequencies of 556 Hz and 666 Hz, respectively. Male four-note location calls were also longer than female calls, with signal median times of 1.99 s and 1.71 s, respectively. The middle-two-note and the full-call segment of the four-note location call were both useful for classifying sex of the calling owl. Our top-ranked models were able to predict 82–83% of our testing data consistent with expert classification as either male or female with 98–99% accuracy (17–18% of test set was classified as unknown). Our results suggest that acoustic characteristics of Northern Spotted Owl calls captured with ARUs can be used to identify whether sites have males and/or females present, and we suggest that further investigation into the full repertoire of Northern Spotted Owl call types is warranted.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.