Search

Autonomous recording units (ARUs) paired with signal classification software can be used to detect species‐specific calls, making them useful for evaluating patterns in avian occurrence and activity. However, classification of target signals is not always reliable and may be especially challenging for species that vocalize infrequently. We assessed the use of ARUs to identify and monitor habitat for a cryptic and federally threatened distinct population segment, the western yellow‐billed cuckoo (Coccyzus americanus) in mountainous xeroriparian drainages. Using Kaleidoscope Pro, we developed a call‐classifier and processed acoustic data collected in sites also surveyed using traditional human‐observer methods, applying the same spatial and temporal detection criteria to estimate breeding territories for each method. The classifier detected a total of 4,061 true positive calls at 4 sites, had an overall precision score of 0.07, recall score of 0.09, and F‐score (beta = 1) of 0.08, indicating high false positive and false negative classification rates. Our results were, however, consistent with other ARU studies of rare and cryptic species and ARUs estimated occupancy as effectively as human surveys with as little as 2 hours of daily recording. Total detections varied among sites, likely due to differences in cuckoo population densities and the interaction between topography and ARU detection space. Our results suggest that despite performance shortcomings of call‐classifiers, ARUs can be effective for monitoring cuckoos, with potential for providing higher resolution temporal and spatial information on activity and habitat use and may be particularly effective in remote locations where cuckoos often occur. [ABSTRACT FROM AUTHOR]

Ruffed grouse (Bonasa umbellus) populations are declining throughout their range, which has prompted efforts to understand drivers of the decline. Ruffed grouse monitoring efforts often rely on acoustic drumming surveys, in which a surveyor listens for the distinctive drumming sound that male ruffed grouse produce during the breeding season. Field‐based drumming surveys can fail to detect ruffed grouse when the birds drum infrequently or irregularly, making this species an excellent candidate for remote acoustic sensing with automated recording units (ARUs). An accurate automated recognition method for ruffed grouse drumming could enable effective and efficient use of ARU data for monitoring efforts; however, no such tool is currently available. Here we develop an automated method for detecting ruffed grouse drumming in audio recordings. Our detector uses a signal processing pipeline designed to recognize the accelerating pattern of drumming. We show that the automated recognition method accurately and efficiently detects drumming events in a set of labeled ARU field recordings. In a case study with 56 locations in Central Pennsylvania, we compared detections of ruffed grouse from 4 survey methods: field‐based acoustic drumming surveys, surveys conducted by humans listening to ARU recordings, and automated recognition for both a 1‐day and a 28‐day period. Field‐based surveys detected drumming at 9 of 56 locations (16%), while surveys conducted by humans listening to ARU recordings detected drumming at 8 locations (14%). Using automated recognition, the 1‐day recording period produced detections at 17 locations (30%) and the 28‐day recording period produced detections at 34 locations (61%). Our case study supports the idea that automated recognition can unlock the value of ARU datasets by temporally expanding the survey period. We provide an open‐source Python implementation of the recognition method to support further use in ruffed grouse monitoring efforts. [ABSTRACT FROM AUTHOR]

Hunting can influence the abundance and distribution of animals and act as a source of conflict among recreational user groups. Thus, land managers benefit from tools that can generate information about when and where hunting occurs. We used passive acoustic monitoring to examine spatiotemporal patterns of hunting‐related gunshots at 91 locations in a protected area in Alberta, Canada. We compared 2 methods for detecting gunshots from recordings: a recognizer that used complex pattern recognition and an energy detector that detected loud sounds regardless of their acoustic features. The recognizer primarily detected faint sounds, and multiple observers showed low levels of agreement (37%) with respect to whether sounds were gunshots or not, suggesting it produced ambiguous data. The recognizer also missed many loud, clear gunshots for unknown reasons. The energy detector, in contrast, detected loud sounds upon which observers showed near‐unanimous agreement (99%) on their identity. Gunshots missed by the energy detector could be because they were too quiet (i.e., too far away to be detected). Thus, despite detecting fewer gunshots overall, the energy detector produced higher quality data that were easier to interpret. We analyzed 249 gunshots detected with the energy detector, and found that hunting was concentrated near vehicle access points and peaked on Saturdays, and that hunters largely abided by local regulations prohibiting Sunday hunting. We compared energy detector results with remote cameras, which revealed similar spatiotemporal patterns of hunting effort. Passive acoustic monitoring has the potential to allow hunting activity to be mapped and monitored with unprecedented resolution. [ABSTRACT FROM AUTHOR]

Autonomous recording units (ARUs) allow for collection of extensive acoustic data sets, while reducing costs and time associated with traditional surveys used to determine gobbling chronology of male wild turkeys (Meleagris gallopavo). A challenge with ARUs is efficiently locating and identifying calls of interest, so autonomous call recognition (ACR) software such as Raven Pro have traditionally been used to identify wild turkey gobbles. However, ACR software often produces high false positive detections, requiring substantive time to verify selections as gobbles. We used ARUs across 3 study sites in the southeastern United States to collect 107,580 hours of ambient sound. We developed a convolutional neural network (CNN) to autonomously identify wild turkey gobbles and compared results of our CNN to results gathered using the commercially available program Raven Pro. After processing of ambient sound, the CNN detected 15,793 more gobbles than Raven Pro, and did so with 5,716,718 fewer selections. Collectively, our CNN improved precision from 0.01 to 0.32 relative to Raven Pro, while decreasing the time required for validation from 4,452 hours to 219. We found precision of our CNN varied across ARUs primarily due to differences in occurrence of ambient sounds similar to gobbles. Thus, we recommend that additional site‐specific training data should be considered when developing CNNs. Our results suggest that researchers interested in describing gobbling activity by male wild turkeys should consider developing and applying CNNs for automated call recognition. [ABSTRACT FROM AUTHOR]

Turbine‐related bat mortality at commercial wind energy facilities may threaten populations of migratory tree‐roosting bat species in North America. Industry stakeholders and regulatory agencies alike are investigating strategies to reduce risk of population‐level consequences as the wind energy industry grows. Bats collide with turbines only when turbine rotors are spinning and curtailing turbine operation at low wind speeds can effectively reduce bat fatality rates. Nonetheless, few quantitative data exist to determine appropriate threshold wind speeds below which turbine operations should be curtailed. Carcass monitoring is labor‐intensive and does not provide information on factors linked to bat fatality rates on any scale finer than nightly. We tested whether acoustic bat data recorded at turbine nacelles could provide a more precise and sensitive measure of fatality risk to bats by analyzing acoustics, weather, turbine operation, and carcass data collected at 2 commercial wind energy facilities in West Virginia over 7 years. Each wind facility implemented several distinct curtailment treatments during our study, allowing us to compare fatality rates and acoustic bat activity across multiple operational strategies. We found that bat passes exposed to turbine operation explained close to 80% of the variation in carcass‐based estimates of bat fatality rates and accounted for significant variation in raw carcass counts per turbine and probability of finding bat carcasses during individual turbine searches. Conversely, bat activity occurring when turbines were not operating had little or no relationship to fatality rates. We also found that patterns in bat activity exposure could be predicted accurately among turbines and years. Our results demonstrate that measuring exposure of acoustic bat activity provides a quantitative basis for designing, evaluating, and adaptively managing curtailment strategies. This is an important advance towards using curtailment to reduce bat fatality rates strategically while allowing for increased generation of renewable energy. © 2021 The Authors. Wildlife Society Bulletin published by Wiley Periodicals LLC on behalf of The Wildlife Society. We demonstrated that the subset of acoustic bat activity exposed to turbine operation (e.g., recorded at wind turbine nacelles when turbine rotors are spinning) is directly correlated with turbine‐related bat fatalities on multiple temporal scales. Accordingly, exposed acoustic bat activity provides a novel, quantitative metric that wind facility operators and wildlife regulators could use to characterize and manage risk to vulnerable bat species. [ABSTRACT FROM AUTHOR]

Passive acoustic monitoring is a common method of studying bats that involves recording echolocation calls of bats in their natural environment. Call sequences are then identified to species using automated acoustic analysis software. One limitation of acoustic software programs, particularly for long‐term monitoring efforts, is that newer versions may provide results that are not directly comparable with older versions. However, there is little available information regarding how much or in what ways the versions differ, or which versions are most accurate. We evaluated 2 software programs used for automated bat call identification by testing a common set of echolocation call files across multiple versions of each program. We quantified the level of agreement on identification results and compared accuracy rates among the versions of each program. Level of agreement varied by species, recording location, and the software versions being compared. Overall percent agreement ranged from 28–82%. Newer versions were more conservative, in that they assigned fewer species‐level identifications. However, newer versions were not substantially more accurate than older versions. Our conclusions suggest that bat researchers should be attentive to what software versions and settings are used as they plan and perform data analyses. Software developers could assist software users by providing more detailed information about their testing procedures and results, and what changes are associated with new versions. © 2021 The Wildlife Society. We compared different versions of automated bat call classifiers and found that the level of agreement between classifiers varied widely depending on version and species. Newer versions of classifier software assigned fewer species‐level identifications and generally did not show improved accuracy rates over older versions. [ABSTRACT FROM AUTHOR]

Over the past 13 years, White‐nose Syndrome (WNS) has caused North American bat population declines and shifted community structure towards species less or unaffected by the disease. Mist‐netting, acoustic surveys, and cave count data have been used to document changes in bat presence and activity through site‐specific, pre‐ and post‐WNS studies. Management and survey guidance often must be applied at a combined landscape and site‐specific scale. Our objective was to explore the relationships among WNS impact, influence of available hibernacula, and environmental factors for the nightly presence of 3 WNS‐affected bats: the Indiana bat (Myotis sodalis), northern long‐eared bat (M. septentrionalis), and big brown bat (Eptesicus fuscus). We used recordings from 10 acoustic monitoring study areas, each with 3 survey locations across the states of Virginia, West Virginia, Ohio and Kentucky to assess changes in nightly bat presence during the summer of 2017. There were significant positive and negative correlates of broad land‐cover categories for presence of all 3 bat species. Our findings also corroborated trends in abundance and distribution patterns found in prior, smaller‐scale studies, supporting the relevance of land cover categories in a large‐scale acoustic monitoring framework. We observed a negative association between WNS impact‐years and nightly northern long‐eared bat presence, but low occurrence and patchy distribution reduced our ability to infer strong relationships. Big brown bat presence showed a significant positive relationship with WNS occurrence on the landscape, providing evidence that big brown bats are maintaining populations after years of exposure. Indiana bats were the least‐documented species, limiting the strength of our conclusions, but we did observe significant temporal patterns in nightly presence, with higher probabilities of presence earlier in the summer. Our results show the potential efficacy of using a WNS impact metric to predict summer bat presence, inform current U.S. Fish and Wildlife Service acoustic monitoring guidelines, and highlight which environmental variables are relevant for large‐scale acoustic monitoring. © 2021 The Wildlife Society. This article has been contributed to by US Government employees and their work is in the public domain in the USA. : Broad habitat categories are relevant in designing and interpreting acoustic sampling for bats. Within White‐nose Syndrome impact areas, proximity to and time since the advent of the disease also are important parameters. [ABSTRACT FROM AUTHOR]

As part of a broader trial of noninvasive methods to research wild wolves (Canis lupus) in Minnesota, USA, we explored whether wolves could be remotely monitored using a new, inexpensive, remotely deployable, noninvasive, passive acoustic recording device, the AudioMoth. We tested the efficacy of AudioMoths in detecting wolf howls and factors influencing detection by placing them at set distances from a captive wolf pack and compared those recordings with real‐time, on‐site howling data between 22 May and 17 June 2019. We identified 1,531 vocalizations grouped into 428 vocal events (236 solo howl series and 192 chorus howls). The on‐site AudioMoth correctly recorded 100% of chorus and solo howls that were also documented in real‐time. The remote array detected 49.5% of chorus and 11.9% of solo howls (≥1 unit detected the event). The closest remote AudioMoth (0.54 km, 0.33 mi) detected 37% of choruses and 8.9% of solo howls. Chorus howls (9.4%) were detected at the farthest unit (3.2 km, 2.0 mi). Favorable wind (carrying source howls to the remote units) and calm (no wind) conditions increased detectability and detection distance of chorus howls. Temperature was inversely related to detection. Given the detection distances we observed, AudioMoths are probably useful in studying specific sites during periods when wolves move less frequently (e.g., during late spring and summer at homesites or potentially during winter at kill sites of very large prey). AudioMoths would also be useful in a passive sampling array (e.g., occupancy studies), especially when used in concert with other methods such as camera‐trapping. Additional research should be conducted in areas with different environmental variables (e.g., wind, temperature, habitat, topography) to determine performance under varying conditions and also when fitted with a parabolic dish. © 2020 The Wildlife Society. AudioMoths are a useful tool for noninvasive, remote, acoustic monitoring of wild wolf populations up to 3.2 km away from source howls. We assessed solo and chorus howl detectability and detection distance, factors that should be considered when studying wild wolves. [ABSTRACT FROM AUTHOR]

ABSTRACT Many anuran monitoring programs now include autonomous recording units (ARUs). These devices collect audio data for extended periods of time with little maintenance and at sites where traditional call surveys might be difficult. Additionally, computer software programs have grown increasingly accurate at automatically identifying the calls of species. However, increased automation may cause increased error. We collected 435 min of audio data with 2 types of ARUs at 10 wetland sites in Vermont and New York, USA, from 1 May to 1 July 2010. For each minute, we determined presence or absence of 4 anuran species ( Hyla versicolor, Pseudacris crucifer, Anaxyrus americanus, and Lithobates clamitans) using 1) traditional human identification versus 2) computer-mediated identification with software package, Song Scope® (Wildlife Acoustics, Concord, MA). Detections were compared with a data set consisting of verified calls in order to quantify false positive, false negative, true positive, and true negative rates. Multinomial logistic regression analysis revealed a strong ( P < 0.001) 3-way interaction between the ARU recorder type, identification method, and focal species, as well as a trend in the main effect of rain ( P = 0.059). Overall, human surveyors had the lowest total error rate (<2%) compared with 18-31% total errors with automated methods. Total error rates varied by species, ranging from 4% for A. americanus to 26% for L. clamitans. The presence of rain may reduce false negative rates. For survey minutes where anurans were known to be calling, the odds of a false negative were increased when fewer individuals of the same species were calling. Published 2016. This article is a U.S. Government work and is in the public domain in the USA. [ABSTRACT FROM AUTHOR]

ABSTRACT Manufacturers of acoustic bat detectors use proprietary microphones with different frequency responses, sensitivities, and directionality. Researchers implement various waterproofing strategies to protect microphones from inclement weather. These factors cause different detector models to have unique sampling areas and likely results in dissimilar data recording. We tested whether SM2BAT+ (SM2) and Anabat SD2 (Anabat) bat detectors record dissimilar data when weatherproofed in a manner suitable for long-term passive sampling. We deployed detectors at 71 random points near Indianapolis, Indiana (USA), from May to August, 2012-2013. We used 45° polyvinyl chloride tubes to weatherproof directional Anabat microphones and the stock foam shielding to cover omnidirectional SM2 microphones. Anabat and SM2 microphones were paired at 2-m and 5-m heights. We adjusted file parameters to make Anabat and SM2 data comparable. We identified files to phonic group (low, midrange, and Myotis) using Bat Call ID software. The effects of detector type, phonic group, height, and their interactions on mean files recorded per site were assessed using generalized estimating equations and least-significant-differences pairwise comparisons. Anabats recorded more low and midrange files, but fewer Myotis files per site than did SM2s. When comparing the same model of detectors, deployment height did not affect data acquisition. Weatherproofing may limit the ability of Anabats to record Myotis, but Anabat microphones may have greater detection ranges for low and midrange bats. We demonstrated that the ability to record bat calls in different frequency ranges varies with microphone type and weatherproofing strategy, which implies that best practices for presence-absence surveys may also vary for bats in different phonic groups. © 2015 The Wildlife Society. [ABSTRACT FROM AUTHOR]