Analytical approaches for evaluating passive acoustic monitoring data: A case study of avian vocalizations

Abstract

1. The interface between field biology and technology is energizing the collection of vast quantities of environmental data. Passive acoustic monitoring, the use of unattended recording devices to capture environmental sound, is an example where technological advances have facilitated an influx of data that routinely exceeds the capacity for analysis. Computational advances, particularly the integration of machine learning approaches, will support data extraction efforts. However, the analysis and interpretation of these data will require parallel growth in conceptual and technical approaches for data analysis. Here, we use a large hand-annotated dataset to showcase analysis approaches that will become increasingly useful as datasets grow and data extraction can be partially automated.2. We propose and demonstrate seven technical approaches for analyzing bioacoustic data. These include the following: (1) generating species lists and descriptions of vocal variation, (2) assessing how abiotic factors (e.g., rain and wind) impact vocalization rates, (3) testing for differences in community vocalization activity across sites and habitat types, (4) quantifying the phenology of vocal activity, (5) testing for spatiotemporal correlations in vocalizations within species, (6) among species, and (7) using rarefaction analysis to quantify diversity and optimize bioacoustic sampling.3. To demonstrate these approaches, we sampled in 2016 and 2018 and used hand annotations of 129,866 bird vocalizations from two forests in New Hampshire, USA, including sites in the Hubbard Brook Experiment Forest where bioacoustic data could be integrated with more than 50 years of observer-based avian studies. Acoustic monitoring revealed differences in community patterns in vocalization activity between forests of different ages, as well as between nearby similar watersheds. Of numerous environmental variables that were evaluated, background noise was most clearly related to vocalization rates. The songbird community included one cluster of species where vocalization rates declined as

Refining manual annotation effort of acoustic data to estimate bird species richness and composition: The role of duration, intensity, and time

“…The aim of our study, therefore, is to analyze richness and compositional differences between acoustic subsetting scenarios across three choice gradients, revealing the potentially missed detection opportunities and across biologically relevant time periods. We not only replicate previous research for a central European forest context for how effort allocation scenarios (hereafter scenarios) affect species richness (Cook & Hartley, 2018 ; de Araújo et al, 2021 ; La & Nudds, 2016 ; Symes et al, 2022 ; Wimmer et al, 2013 ; Wood et al, 2021 ) but we also additionally analyze how these scenarios affect revealed assemblage composition. To do this we considered three main parameters of acoustic file selection: (1) recording intensity, i.e., number of minutes within 1 h; (2) day phase, i.e., period of the morning in relation to sunrise (dawn, morning, or both); and (3) duration, i.e., number of days included in the subset.…”

“…To our knowledge, no studies exist that also directly address how perceived bird species composition is affected by acoustic file selection choices, but such investigations are recommended (Symes et al, 2022). Wood et al (2021) addressed species composition insofar as they used simulated bird assemblages to investigate how assemblage structure affects richness estimates depending on how many theoretically rare species comprise the assemblage.…”

“…In other cases, there is a dearth of annotated data from certain regions (de Araújo, 2013 ; xeno‐canto, 2022 ), which is required to build accurate classifiers; in such cases, manually annotated data will be a necessary precursor to scalable machine learning campaigns in those areas. For these reasons, manual annotation is and will continue to be an important link as our knowledge is slowly transitioning to digital and automated monitoring methods (Symes et al, 2022 ). Manual annotations are, however, time‐consuming (Rempel et al, 2005 ; Swiston & Mennill, 2009 ; Wimmer et al, 2013 ), and researchers must maximize this time by subsetting acoustic data efficiently to identify the maximum number of unique species without repeatedly identifying species already observed in that area.…”

“…Manual annotations are, however, time‐consuming (Rempel et al, 2005 ; Swiston & Mennill, 2009 ; Wimmer et al, 2013 ), and researchers must maximize this time by subsetting acoustic data efficiently to identify the maximum number of unique species without repeatedly identifying species already observed in that area. Whether manual annotations from PAM are used to directly estimate species richness or to validate the predictions of machine learning algorithms (reducing false positives), it is recommended that this time be strategically allocated to maximize accuracy for either objective (Symes et al, 2022 ).…”

“…In the case of bird assemblages, PAM has the potential to be a wellspring of information for assessing the impact of human activities on assemblage homogenization or heterogenization (also concerning functional diversity) and thereby guide bird conservation decision‐making (Gasc et al, 2017 ). To our knowledge, no studies exist that also directly address how perceived bird species composition is affected by acoustic file selection choices, but such investigations are recommended (Symes et al, 2022 ). Wood et al ( 2021 ) addressed species composition insofar as they used simulated bird assemblages to investigate how assemblage structure affects richness estimates depending on how many theoretically rare species comprise the assemblage.…”