Search

Behavior represents animals' primary means of responding to environmental variation and adapting to rapid environmental change. Many animals' presence, let alone behavior, is highly cryptic to human observers, presenting a significant barrier in both theoretical and applied behavioral ecology. Bioacoustic signals not only reveal animals' presence, but also encode detailed information about the behaviors in which they are engaging. The study of behavioral bioacoustics has emerged to decipher the context and function of animal sounds and to apply this comprehension to understanding animal behavior across ecological scales and levels of biological organization. Growing capacity for behavioral bioacoustics represents a profound opportunity to understand animal behavior and steward rapidly changing ecosystems in the Anthropocene. Interpreting sound gives powerful insight into the health of ecosystems. Beyond detecting the presence of wildlife, bioacoustic signals can reveal their behavior. However, behavioral bioacoustic information is underused because identifying the function and context of animals' sounds remains challenging. A growing acoustic toolbox is allowing researchers to begin decoding bioacoustic signals by linking individual and population-level sensing. Yet, studies integrating acoustic tools for behavioral insight across levels of biological organization remain scarce. We aim to catalyze the emerging field of behavioral bioacoustics by synthesizing recent successes and rising analytical, logistical, and ethical challenges. Because behavior typically represents animals' first response to environmental change, we posit that behavioral bioacoustics will provide theoretical and applied insights into animals' adaptations to global change. [ABSTRACT FROM AUTHOR]

Individual acoustic signatures, if present within animal species, can provide insights into evolution and behavior, and have great potential for differentiating individuals during monitoring and research. Recent advances in bioacoustic technology combined with acoustic individual identification (AIID) have the potential to revolutionize the study of sound-producing animals; however, methods and application of AIID remain in their infancy. Evidence of individual acoustic signatures across taxa and successes in adjacent acoustic disciplines suggest that opportunities exist for developing AIID. Research and development into AIID should combine deep learning methods with the construction and sharing of labeled training datasets, and should focus on recording and classification methods that maximize the potential applications of AIID. Broadscale implementation of AIID should be achievable in the near future and will allow biologists to answer important ecological and evolutionary questions with less bias and fewer negative population effects and resources than the current approaches. Recent advances in bioacoustics combined with acoustic individual identification (AIID) could open frontiers for ecological and evolutionary research because traditional methods of identifying individuals are invasive, expensive, labor-intensive, and potentially biased. Despite overwhelming evidence that most taxa have individual acoustic signatures, the application of AIID remains challenging and uncommon. Furthermore, the methods most commonly used for AIID are not compatible with many potential AIID applications. Deep learning in adjacent disciplines suggests opportunities to advance AIID, but such progress is limited by training data. We suggest that broadscale implementation of AIID is achievable, but researchers should prioritize methods that maximize the potential applications of AIID, and develop case studies with easy taxa at smaller spatiotemporal scales before progressing to more difficult scenarios. [ABSTRACT FROM AUTHOR]

In recent years, new technologies have emerged that can generate rapid and standardized biodiversity inventories without explicit human guidance (novel community data). The benefits as well as technical challenges of these technologies have been extensively reviewed, and ecologists are currently in the process of incorporating them into their observational studies. So far, however, large novel community datasets are still rare. Consequently, there are still many open questions about how these new data should be optimally used to address fundamental questions in community ecology, macroecology, and conservation. We review the state of the field, highlight the opportunities and analytical tools for advancing ecological research with novel community data, and discuss the implications of these emerging technologies for ecological theory, ecological study design, and environmental management. New technologies for monitoring biodiversity such as environmental (e)DNA, passive acoustic monitoring, and optical sensors promise to generate automated spatiotemporal community observations at unprecedented scales and resolutions. Here, we introduce 'novel community data' as an umbrella term for these data. We review the emerging field around novel community data, focusing on new ecological questions that could be addressed; the analytical tools available or needed to make best use of these data; and the potential implications of these developments for policy and conservation. We conclude that novel community data offer many opportunities to advance our understanding of fundamental ecological processes, including community assembly, biotic interactions, micro- and macroevolution, and overall ecosystem functioning. New technologies for monitoring biodiversity such as environmental (e)DNA, passive acoustic monitoring, and optical sensors promise to generate automated spatiotemporal community observations at unprecedented scales and resolutions. Here, we introduce 'novel community data' as an umbrella term for these data. We review the emerging fields around novel community data, focusing on new ecological questions that could be addressed; the analytical tools available or needed to make best use of these data; and the potential implications of these developments for policy and conservation. We conclude that novel community data offer many opportunities to advance our understanding of fundamental ecological processes, including community assembly, biotic interactions, micro- and macroevolution, and overall ecosystem functioning. [ABSTRACT FROM AUTHOR]

Passive acoustic monitoring (PAM) provides new opportunities to characterise zoonotic and vector-borne disease dynamics in changing landscapes. Acoustic data can inform our understanding of variables that drive transmission risk, including human and wildlife occupancy over space and time and changes in habitat quality. With low-cost hardware, cloud-based computing, and open-source platforms, acoustic data collection and analysis methods are increasingly accessible to nonspecialist users. Key considerations for epidemiologists include the availability of complementary data sources and the technical requirements for acoustic data storage and analysis. Acoustic monitoring is a cost-effective, noninvasive tool which could be effectively combined with existing data to strengthen early warning systems and integrated disease surveillance. Emerging infectious diseases continue to pose a significant burden on global public health, and there is a critical need to better understand transmission dynamics arising at the interface of human activity and wildlife habitats. Passive acoustic monitoring (PAM), more typically applied to questions of biodiversity and conservation, provides an opportunity to collect and analyse audio data in relative real time and at low cost. Acoustic methods are increasingly accessible, with the expansion of cloud-based computing, low-cost hardware, and machine learning approaches. Paired with purposeful experimental design, acoustic data can complement existing surveillance methods and provide a novel toolkit to investigate the key biological parameters and ecological interactions that underpin infectious disease epidemiology. [ABSTRACT FROM AUTHOR]

• We test a simple, scalable acoustic indicator for monitoring animal density. • Call density strongly correlates with distance sampling-based animal density. • Our metric can be quickly derived with classifiers, regardless of classifier quality. • Call density is a sensitive acoustic indicator of animal density to inform management. Passive acoustic monitoring has grown in utility for tracking wildlife populations, although challenges remain when using acoustic detections to monitor population size and density. Distance sampling is considered the 'gold standard' for estimating animal densities but has several important limitations, especially for rare, cryptic, and high-density species. Here, we test the performance of a simple, quickly derived bioacoustic indicator for monitoring population density: call density— the proportion of recording samples containing vocalizations. Over three years, we collected synchronized bioacoustic and point-transect distance sampling data for eight forest bird species native to the Island of Hawai'i, including four endangered species, across diverse ecosystems ranging from subalpine dry woodland to montane rainforest. The species studied exhibit varied population structures, from gregarious flocks to small, territorial family groups. Our results revealed significant, strong correlations between call density and distance sampling-based animal density estimates for all species, demonstrating that call density is a reliable indicator of animal density that can be used independently or in combination with traditional monitoring methods. Our analysis uses a fixed amount of manual validation of machine learning classifier output examples, without requiring prohibitively high classifier performance, and is robust to variation in vocal activity rates across time and space, making it both adaptable and scalable. This approach could enhance passive acoustic monitoring by providing a more sensitive population health indicator than commonly used detection/nondetection methods, facilitating prompt conservation and management decisions, particularly for species that are difficult to monitor with distance sampling. [ABSTRACT FROM AUTHOR]

• The effects of high-pass filters on acoustic index values derived from PAM recordings are not well documented. • High-pass filtering treatments did not consistently strengthen associations between acoustic indices and species richness. • ACI, AR, and CB demonstrated greater resistance to modifications caused by high-pass filters. • CityBioNet (CB) had a notably high correlation with species richness compared to the other examined acoustic indices. High-pass sound filters are a common form of audio manipulation that attenuates or removes low frequency sounds from audio recordings. While high-pass filters are used to reduce anthropogenic noise, there is limited guidance on their optimal application and their effects on acoustic indices—numerical values derived from passive acoustic monitoring (PAM) recordings to summarize acoustic information. Here, we investigated the effects of three high-pass filter treatments (482 Hz, 1 kHz, and 2 kHz) on eight commonly implemented acoustic indices and one less-commonly used convolutional neural network metric. Specifically, we used simulated soundscapes with three levels of traffic noise interference and then applied the filter treatments to field recordings collected throughout Illinois, USA during May 2022–July 2023 and derived acoustic indices to further understand these effects. Our analysis revealed that interactions between acoustic filtering and vehicular noise pollution have diverse effects on the nine acoustic indices, both in simulated soundscapes and empirical PAM recordings. In general, a 1 kHz or 2 kHz filter was necessary in order to produce significant changes in acoustic index values. However, none of the applied filtering treatments consistently strengthened correlations between the examined acoustic indices and avian species richness. The Acoustic Complexity Index (ACI), Acoustic Richness Index (AR), and CityBioNet (CB) demonstrated greater resistance to biologically non-informative changes caused by filter implementation, with CB showing a notably higher correlation with species richness compared to the other examined indices. Our findings suggest that ACI, AR, and CB may be better suited to studies of avian species richness in soundscapes with high levels of anthropogenic noise. Future research is needed to establish best practices for acoustic filtering, understand the behavior of acoustic indices under various environmental contexts, and explore alternative methodologies for avian monitoring in human-modified contexts. [ABSTRACT FROM AUTHOR]

Ants serve as ecosystem engineers that maintain important ecological processes within forests. Given their ecological importance, it is a clear scientific shortcoming that we lack non-invasive methods to survey their behaviour inside common opaque habitats such as mounds, litter, and soil. In this study, we assess if acoustic signals from red wood ant (Formica rufa) mounds are useful to infer temporal changes in ant activity within forested ecosystems. We found that acoustic indices used previously as a proxy for soil fauna in soil ecological studies (Acoustic Complexity Index, Bioacoustic Index) can indeed separate sounds generated by the ant's daily routines (biophony) from other forest sounds. Yet, we also show that these indices are problematic proxies for soil diversity as they increase not only due to an increased number of species but also due to an increased number of the same species. Acoustic measures that incorporated the strength of acoustic signals, Average Power Density (APD) and Peak Power Density (PPD) also increased with increasing ant abundance and constituted the conceptually best proxy for ant activity. For example, the PPD could i) track diurnal changes in Formica rufa activity with a high temporal resolution (minutes) and ii) detect altered behavioural responses to temperature changes. We conclude that microphones detecting biophony can provide high-resolution information about in situ ant behaviours in forested ecosystems. Thus, passive acoustics monitoring offers a promising avenue as a non-invasive monitoring tool for soil macrofauna studies. [ABSTRACT FROM AUTHOR]

Passive acoustic monitoring combined with deep learning-based bird sound classifiers is an effective tool, particularly in remote areas. While self-supervised learning has recently excelled in natural language processing and image recognition, its application to bird sound recognition remains limited. This study proposes an innovative self-supervised learning approach, which leverages vast amounts of passive acoustic recordings for pre-training, followed by fine-tuning of target species. Compared to the three state-of-the-art models based on transfer learning from ImageNet, the proposed method demonstrated improvements in overall recognition performance, with even more significant gains for tail-end species. These results confirm that domain-specific pre-training in self-supervised learning enhances downstream recognition tasks and provides greater robustness, benefiting tail-end species in imbalanced ecological datasets. Our experiments further demonstrate that integrating open-source datasets and data augmentation techniques is the most effective strategy for mitigating data imbalances and cross-domain issues. In addition, introducing a 'catch-all' category into training datasets has been shown to improve model robustness in open set recognition scenarios. We also identified the minimum viable sample size requirements for our proposed model and explored the impact of overlapping bird vocalizations during dawn choruses on model performance. Targeting 31 bird species in the montane regions of subtropical Taiwan, the model achieved a class-wise mean average precision of 0.782 and an overall precision of 85.6 % at the F 0.5 threshold in dawn chorus soundscape recordings. This study confirms the effectiveness and advantages of self-supervised learning in bird sound recognition, supporting long-term monitoring of bird distribution and vocal activity in remote montane areas. [Display omitted] • Self-supervised audio-MAE excels in bird sound recognition, notably for tail species. • Domain-specific tasks enhance recognition, outperforming ImageNet-based models. • Data augmentation mitigate data imbalance and cross-domain challenges. • 'Catch-all' category boosts model robustness in open set recognition. • Dawn chorus overlaps affect FNR rather than FPR in bird sound models. [ABSTRACT FROM AUTHOR]

Underwater passive acoustics is used worldwide for multi-year monitoring of marine mammals. Yet, the large amount of audio recordings raises the need to automate the detection of acoustic events. For instance, the increasing number of Offshore Wind Farms (OWF) raises key environmental and societal issues relating to their impacts on wildlife. In this context, monitoring marine mammals along with information on their acoustic environment throughout the OWF life cycle is crucial. The objective of this study is to evaluate the ability of a single deep learning model to precisely detect and localize, in time and in frequency, the marine mammal sounds over a wide frequency range and classify them by species and sound types. A broadband hydrophone, deployed at the Fécamp OWF (Normandy, France), recorded the underwater soundscape including sounds from marine mammals occurring in the area. To visualize these sounds, 15-s spectrograms were computed. From these images, dolphin (D) and porpoise (P) sounds were manually annotated, including different types of sounds: Click-Trains (D CT , P CT), Buzzes (D B , P B) and Whistles (D W). The spectrograms were then split into five-fold cross-validation datasets, each containing one half of manual annotations and one half of only background noise. A Faster R-CNN model was trained to precisely detect and classify the marine mammal sounds in the spectrograms. Three model output configurations were used: (1) overall detection of marine mammals (presence vs. absence), (2) detection and classification of species (two classes: dolphin, porpoise) and (3) sound types (five classes: D CT , D B , D W , P CT , P B). For the simplest configuration (1) 15.4 % of the spectrogram dataset had detections while missing only 6.6 % of annotated spectrograms. For the more precise configurations, (2) and (3), the mean Average Precision (mAP) achieved were 92.3 % (2) and 84.3 % (3), and the macro average Area under the curve (AUC) 95.7 % (2) and 94.9 % (3). This model will help to speed up the annotation processes, by reducing the spectrogram quantity to be manually analyzed and having time-frequency boxes already drawn. Several model parameters can be adjusted to trade off missed detections and false positives which need to be carefully considered and adapted to the problem. For instance, these adjustments would be particularly relevant depending on the human resources available to manually check the model detections and the criticality of missing marine mammal sounds. These models are promising, ranging from the simple detection of marine mammal presence to precise ecological inferences over the long term. • Deep learning is needed and promising to monitor marine mammal sounds. • An object detection model is proposed to detect and classify underwater sounds. • Presence and behavior of marine mammals around the wind farm can be inferred. [ABSTRACT FROM AUTHOR]

Restoring forests is crucial for safeguarding biodiversity and providing ecosystem functions on degraded lands globally. Despite significant restoration efforts over a half-century in the Loess Plateau of China's prominent dryland region, the impact of restored forests on biodiversity remains less understood. Unlike the abundant studies in the tropics, such understanding is urgently needed for dryland regions given that forest restoration is being scaled up. Here, we assessed the alpha- and beta-diversity of birds from taxonomic and functional dimensions in restored forests, and compared them to that of relatively undisturbed primary forests. We conducted rigorous point-count surveys and acoustic recordings during the avian breeding season in 2021 and 2023. We found that plantations have limited support for bird taxonomic and functional diversity compared to secondary and primary forests. Notably, the biodiversity inference drawn from our acoustic recordings generally supported similar conclusion of bird diversity across studied forest types. Additionally, secondary forests exhibited greater functional richness compared to plantations and contributed complementary functional trait space relative to primary forests. While nestedness dominated functional beta-diversity, spatial turnover dominated the taxonomic beta-diversity between the bird communities of restored forests and primary forests. Importantly, spatial proximity positively influenced the nestedness component of taxonomic beta-diversity, indicating that the distance of plantations to local primary forests (i.e., dispersal limitation) influenced bird community assembly during forest recovery. Overall, our findings provide insights on preserving intact primary forests, delivering biodiversity benefits of forest restoration in dryland regions, and prioritizing restoration locations only when tree planting is necessary. ● Restoring forests is crucial for safeguarding biodiversity and providing ecosystem functions on degraded lands globally. ● Plantations have limited support for multifaceted bird diversity after reforestation in the Loess Plateau. ● Secondary forests support higher taxonomic and acoustic diversity than plantations, along with extra functional advantages. ● The distance of plantations from local primary forests influenced bird community assembly during forest recovery. [ABSTRACT FROM AUTHOR]

Back-reef habitats are important and fragile transition zones acting as nurseries for many coral reef fishes. In this framework, Passive Acoustic Monitoring (PAM) can be an important tool to evaluate the diversity and dynamics of sonic fish community. Here, we investigated the diversity, spatial and diel dynamics of fish sounds in back-reef habitats at Makogai Island in Fiji, South Pacific. Synchronized underwater recorders were deployed in 4 bays collecting data for about 4 days. The abundance of 12 different sub-categories of fish sounds were quantified. Signals were acoustically characterized and the level of discrimination between the sub-categories was evaluated by Discrimination Function Analysis. Generalized Additive Models showed that the abundance of signals was related to the bay and the hour. Moreover, the Shannon Diversity and Equitability Indices were calculated using acoustic and visual census data to describe fish biodiversity of each bay. The two bays with greater biodiversity based on visual census also showed a greater acoustic diversity at dawn and night. Our results highlight the importance of PAM to reveal the diversity of fish community in back-reef habitats, providing a baseline to understand future changes in these crucial environments. • Back-reef habitats showed high levels of fish acoustic diversity. • Twelve categories of fish sounds were statistically discriminated by acoustic features properties. • Fish sounds showed different temporal distributions along daytime. • Bays closer to coral reef crest showed higher fish acoustic diversity. [ABSTRACT FROM AUTHOR]

Crowding is a necessary operation for moving fish out of or between sea cages in Atlantic salmon farming. These operations consist of gradually reducing available volume in the nets to increase the biomass density. This is known to stress the fish, increase risk of injury, and can in extreme cases increase mortality rates. To reduce the negative impacts of crowding, the operation must be adjusted based on real-time monitoring of fish welfare. Today, this is mainly done by manually monitoring surface activity, as the high fish density limits the use of available tools. In this paper a new approach for monitoring was applied to crowding of meso-scale cages containing 40 fish each, using automatic detection of passive acoustic signatures. Four synchronised hydrophones were deployed and the acoustic signature was defined from the recordings. The automatic detection algorithm recorded an extreme increase in registered acoustic events during crowding, as high as 1000 events per 10 min period, compared to only 20 events per 10 min period on the day following crowding. Correlating the number of acoustic events to recorded heart rate and activity measures from implants in the salmon indicate that automatically registering acoustic events is a promising method for monitoring farmed fish. • Detection algorithm applied to monitor acoustic signature of Atlantic salmon during crowding. • Drastic increase of occurrences of acoustic events during crowding. • Increase in number of acoustic events correspond well with increased activity measurements. • Monitoring acoustic events are a promising method for monitoring stress in farmed fish. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) are effective tools for acoustic classification tasks such as species identification. Large datasets of labelled recordings are required to develop CNN classifiers which can be difficult to obtain, particularly if species are rare or vocalise infrequently. Additionally, data often requires manual labelling which can be time consuming requiring expert analysis. Artificially generating data using augmentation can address these challenges, however the impact of data augmentation on CNN performance is poorly understood and often omitted in bioacoustic studies. Here, we empirically test the impact of CNN architecture and 20 data augmentation methods on classifier performance. We use acoustic identification of 18 small mammal species as a case study of a species group that can be effectively surveyed by acoustic monitoring, but recordings for training data are scarce and difficult to collect. Networks that achieved the highest accuracy across all sample sizes was a 10-layer CNN (96.43 %) and a pre-trained ResNet50 model (96.37 %). Overall, all augmentation effects improved ResNet50 model performance and 17 effects improved Conv10 performance, increasing relative change in accuracy (RCA) by 0.021–0.641. Three augmentation effects negatively impacted Conv10 RCA by −0.042 to −0.182. We also show that adding augmented data when the number of original samples is low has the greatest positive impact on accuracy and this effect was larger with ResNet50 models. Our work demonstrates that using data augmentation where few original samples are available can considerably improve model performance and highlights the potential of augmentation in developing acoustic classifiers for species where data are limited or difficult to obtain. • A CNN-based small mammal acoustic classifier was developed to investigate the impact of data augmentation on performance. • Augmenting data can improve classifier accuracy by up to 8.4 % when there are as few as 100 samples per species. • Using augmentation to balance classes with uneven sample sizes reduces classification errors for underrepresented species. • We show that data augmentation can reducce the need for large, labelled datasets for deep learning acoustic classifiers. [ABSTRACT FROM AUTHOR]

Computational ecoacoustics has seen significant growth in recent decades, facilitated by the reduced costs of digital sound recording devices and data storage. This progress has enabled the continuous monitoring of vocal fauna through Passive Acoustic Monitoring (PAM), a technique used to record and analyse environmental sounds to study animal behaviours and their habitats. While the collection of ecoacoustic data has become more accessible, the effective analysis of this information to understand animal behaviours and monitor populations remains a major challenge. This survey paper presents the state-of-the-art ecoacoustics data analysis approaches, with a focus on their applicability to large-scale PAM. We emphasise the importance of large-scale PAM, as it enables extensive geographical coverage and continuous monitoring, crucial for comprehensive biodiversity assessment and understanding ecological dynamics over wide areas and diverse habitats. This large-scale approach is particularly vital in the face of rapid environmental changes, as it provides crucial insights into the effects of these changes on a broad array of species and ecosystems. As such, we outline the most challenging large-scale ecoacoustics data analysis tasks, including pre-processing, visualisation, data labelling, detection, and classification. Each is evaluated according to its strengths, weaknesses and overall suitability to large-scale PAM, and recommendations are made for future research directions. • Passive Acoustic Monitoring (PAM) enables continuous species monitoring. • We evaluate ecoacoustics datasets for applicability in large-scale PAM. • We compare key ecoacoustics data analysis frameworks/methods. • We discuss the open challenges/opportunities of large-scale PAM data analysis. [ABSTRACT FROM AUTHOR]

• Hunting poses a significant threat to wildlife. • We recorded gunshots in Vietnam and used transfer learning for automated detection in the 'torch for R' ecosystem. • We found that transfer learning architectures led to acceptable performance. Hunting is one of the most serious threats to wildlife populations. Guns can be used to hunt both terrestrial and arboreal species. While many methods and techniques have been developed to monitor wildlife populations, few techniques have been developed that can meet the demand needed to monitor threats to wildlife populations and support intervention activities; this is particularly true in Southeast Asia. Here, we used smartphones to record gunshots at 64 sites that were systematically spaced in Chu Mom Ray National Park, Vietnam from July to November 2019. Our specific goals were to: 1) quantify temporal and spatial patterns in gunshots in the national park; 2) investigate the correlation between gunshots and seven environmental variables including forest quality, elevation, and distance to the nearest village or ranger post; and 3) compare the performance of three convolutional neural network (CNN) architectures for automatically detecting gunshots. We manually annotated spectrograms of gunshots using Raven Pro 1.6. Using the manual detection approach, we identified 115 gunshots at 30 sites. The number of gunshots recorded at each recording site over two days varied from 0 to 11. On average, there were about 0.93 gunshots recorded per day per recording site. Hunting activity showed a strong temporal trend. The number of gunshots detected increased gradually from early morning, reaching the peak at noon, with the most gunshots recorded from 10:00 to 14:00 local time, equivalent to 0.18 gunshots per hour per site. No gunshots were detected from 22:00 to 04:00 in the morning. There were no strong relationships between the number of gunshots and all environmental variables, with all the correlation coefficients smaller than 0.3. Comparing three CNN architectures — AlexNet, VGG16, and ResNet18— implemented in the 'torch for R' ecosystem, we found that both AlexNet and VGG16 architectures led to acceptable performance for automated detection (F1 score > 0.80), but the ResNet18 architecture did not perform well for this task. We found low generalizability, as the models had relatively low performance on an open dataset from Belize (F1 score ∼ 0.52). To implement effective patrols, protected-area managers should make a hunting-risk map. The effort required for this task can be substantially reduced by using a combination of PAM and automated gunshot detection, however these approaches still require creation of a site-specific training and test dataset, along with manual verification of the detections. [ABSTRACT FROM AUTHOR]

In penaeid shrimp, the mandibles are responsible for emitting click sounds when they collide during food intake, which has been used to assess feeding activity by passive acoustic monitoring. However, the acoustic parameters of the clicks could be affected by variations in the thickness of mandible cuticle during the molting cycle. Thus, the aim of this study was to evaluate the effects of the molting cycle on the acoustic characteristics of clicks emitted during the feeding activity of Litopenaeus vannamei fed commercial pelleted diet. Shrimp with an average weight of 8.75 ± 1.05 g were individually maintained in 24 tanks in a recirculating water system. A total of 12 animals had their feeding activity individually recorded soon after ecdysis (postmolt group), while another 12 were recorded in the middle of their molting cycle (intermolt group). Recordings took place in anechoic chambers with a hydrophone connected to a digital recorder (sampling rate of 192 kHz), and the click acoustic parameters were characterized in Raven® 1.5 Pro software. After the recordings, the shrimp had their mandibles removed for length measurement and histological analysis of the mandibular cuticle. Body (weight and length) and mandibular measurements did not differ between shrimp groups, but those in the postmolt stage showed significantly lower cuticle thickness. The acoustic characteristics of the clicks were affected by shrimp molting stages, with significantly lower maximum energy (dB) and maximum frequency (kHz) in postmolt shrimp, probably related to their lower mandibular cuticle thickness. These results may contribute to the assessment of shrimp feeding behavior associated with the molting cycle, as well as to the optimization of algorithms controlling acoustic automatic feeders in farming systems. • Click characterization during the molting cycle is first reported for Litopenaeus vannamei • Acoustic feeding activity was detected in both intermolt and postmolt stage shrimp • Mandible cuticle thickness in postmolt shrimp resulted in clicks with lower maximum energy (dB) and maximum frequency (kHz) • Results may contribute to optimizing algorithms controlling automated acoustic feeders [ABSTRACT FROM AUTHOR]

The global expansion of shrimp aquaculture has positioned this sector as a significant contributor to seafood supply. Efficient production hinges on effective feeding management. This study presents Python-based computational tools for the acoustic monitoring of white shrimp (Penaeus vannamei) feeding activities. We compare the proposed Python implementation for automatic detection of shrimp feeding sounds with a reference detection method using Raven Pro®, examining various shrimp sizes and stocking densities. Our comparative analysis shows substantial agreement between the two methods, particularly for shrimp weighing more than 1 g across different density groups. Although Raven Pro® is a well-established bioacoustic software tested under laboratory conditions, the Python-based methods proposed here offer practical applications in real farming scenarios and can be integrated into dedicated hardware. These techniques not only enhance feeding practices but also demonstrate the potential of computational tools in advancing aquaculture management. • A Python method for detection of shrimp acoustic feeding activity is presented. • The proposed method was compared with a reference detector of Raven Pro®. • An agreement was found between click detectors across shrimp sizes and densities. • Python-based detector enables computation in resource-limited aquaculture settings. [ABSTRACT FROM AUTHOR]

[Display omitted] • Common cuckoo vocalisations reflect bird species richness. • The species list from 10-minute samples weakly correlates with BirdNET's 8-hour data. • BirdNET detected twice as many species as human observers. • Acoustic indices do not reflect species richness in farmland. Surrogate species are often used to represent broader groups in conservation and research. However, whether a single species can truly reflect an entire ecosystem remains debated. The common cuckoo (hereafter cuckoo) Cuculus canorus has been suggested several times as an indicator of biodiversity with various outcomes. In this study, we utilised passive acoustic monitoring to record 48 soundscapes within a farmland area in Poland, aiming to assess the relationship between cuckoo vocalisations and bird species richness. We employed automated software for cuckoo vocalisation detection and manually analysed 10-minute samples from each site to determine species numbers. Additionally, we used BirdNET, an automated recognition algorithm, on 8-hour recordings to compare species composition with results obtained manually. We also utilised acoustic indices to evaluate their reflection of overall species richness. Our results revealed significant differences between manual and automatic species detection methods, with BirdNET detecting twice as many species as human observers. Furthermore, we noted more cuckoo vocalisations at sites with elevated bird species and host richness obtained automatically, indicating its potential as a biodiversity surrogate. However, acoustic indices failed to mirror acoustic diversity in the farmland habitat, showing unified values across all sites. These findings highlight the necessity of enhancing monitoring approaches to ensure a more comprehensive depiction of avian diversity, thus strengthening conservation measures. [ABSTRACT FROM AUTHOR]

• The ecoacoustic index provides a new way to understand biodiversity and ecosystem functioning. • More than 80 ecoacoustic indices have been proposed amidst growing interest. • Maximize the number of index combinations when using ecoacoustic indices to improve assessment. • It is important to identify a set of general treatment processes that are non-subjective and applicable to most habitats. In recent years, ecological acoustic assessment based on Passive Acoustic Monitoring technology has gained widespread attention as a non-invasive method, providing scientific evidence for biodiversity monitoring and wildlife conservation. Although this approach, as a newly established ecological research method, faces numerous challenges to prove its effectiveness, it has gradually established a comprehensive analytical process supported by numerous studies. This study analyzes the application of ecoacoustic indices in assessing terrestrial and underwater environments and biodiversity detection, identifying commonly used indices. After summarizing the principles of these ecoacoustic indices, the study highlights the suitable indices for different application environments and the influencing factors to consider. Additionally, a bibliometric analysis was conducted on related articles and their keywords retrieved from the Web of Science database, visualizing the relationships. Based on this, the study predicts the development trends of ecoacoustic indices and envisions their broader application in other fields. This research provides a clear overview of the use of ecoacoustic indices and offers theoretical references for the development of the field. [ABSTRACT FROM AUTHOR]

Passive Acoustic Monitoring (PAM) has emerged as a pivotal technology for wildlife monitoring, generating vast amounts of acoustic data. However, the successful application of machine learning methods for sound event detection in PAM datasets heavily relies on the availability of annotated data, which can be laborious to acquire. In this study, we investigate the effectiveness of transfer learning and active learning techniques to address the data annotation challenge in PAM. Transfer learning allows us to use pre-trained models from related tasks or datasets to bootstrap the learning process for sound event detection. Furthermore, active learning promises strategic selection of the most informative samples for annotation, effectively reducing the annotation cost and improving model performance. We evaluate an approach that combines transfer learning and active learning to efficiently exploit existing annotated data and optimize the annotation process for PAM datasets. Our transfer learning observations show that embeddings produced by BirdNet, a model trained on high signal-to-noise recordings of bird vocalisations, can be effectively used for predicting anurans in PAM data: a linear classifier constructed using these embeddings outperforms the benchmark by 21.7%. Our results indicate that active learning is superior to random sampling, although no clear winner emerges among the strategies employed. The proposed method holds promise for facilitating broader adoption of machine learning techniques in PAM and advancing our understanding of biodiversity dynamics through acoustic data analysis. • Fusion of transfer and active learning reduces annotation time for PAM datasets • We compare uncertainty vs diversity and greedy vs batch active learning methods. • Comparative study of active learning strategies for creating multi-label PAM datasets. • The penultimate layer of BirdNet performs best for transfer learning. • Active learning precision curves rise consistently faster than recall. [ABSTRACT FROM AUTHOR]

Passive acoustic data collected during 2020 and 2021 were used to monitor changes in both terrestrial and underwater soundscapes, as well as human activity from aircraft and vessels. Passive acoustic data were collected at two artificial reefs south of Long Island, as well as along ocean beaches in Southampton, NY. At the artificial reefs, vessel noise was recorded more frequently during 2020 than in 2021. Commercial vessels and multi-user charter fishing vessels were more abundant during 2020. Peaks in power spectral density occurred at 60, 90 and 120 Hz in 2020 and 2021, which are frequencies consistent with noise generated by commercial vessels, suggesting that vessels are a significant contributor to the soundscape of the artificial reefs. In the terrestrial environment, noise generated by aircraft was more common during 2021. Peaks in power spectral density were measured around 160 and 290 Hz at one of the ocean beach sites. These frequencies are consistent with noise generated by aircraft. This study documents the chronic extent of anthropogenic noise in both the underwater and terrestrial environments of Long Island, NY, as well as quantifies the occurrence of various noise sources in these habitats. [Display omitted] • Anthropogenic noise was a common source of noise pollution in coastal soundscapes. • Spectrogram analysis allowed for classification of vessel and aircraft types. • Peaks in power spectral densities align with aircraft or vessel noise. [ABSTRACT FROM AUTHOR]

Marine ecosystems are increasingly subjected to anthropogenic pressures, which demands urgent monitoring plans. Understanding soundscapes can offer unique insights into the ocean status providing important information and revealing different sounds and their sources. Fishes can be prominent soundscape contributors, making passive acoustic monitoring (PAM) a potential tool to detect the presence of vocal fish species and to monitor changes in biodiversity. The major goal of this research was to provide a first reference of the marine soundscapes of the Madeira Archipelago focusing on fish sounds, as a basis for a long-term PAM program. Based on the literature, 102 potentially vocal and 35 vocal fish species were identified. Additionally 43 putative fish sound types were detected in audio recordings from two marine protected areas (MPAs) in the Archipelago: the Garajau MPA and the Desertas MPA. The Garajau MPA exhibited higher fish vocal activity, a greater variety of putative fish sound types and higher fish sound diversity. Lower abundance of sounds was found at night at both MPAs. Acoustic activity revealed a clear distinction between diurnal and nocturnal fish groups and demonstrated daily patterns of fish sound activity, suggesting temporal and spectral partitioning of the acoustic space. Pomacentridae species were proposed as candidates for some of the dominant sound types detected during the day, while scorpionfishes (Scorpaena spp.) were proposed as sources for some of the dominant nocturnal fish sounds. This study provides an important baseline about this community acoustic behaviour and is a valuable steppingstone for future non-invasive and cost-effective monitoring programs in Madeira. • The acoustic fish community was characterized in two important MPAs of Madeira Archipelago. • Fish sounds were catalogued and classified in 43 different sound types. • Fish sound activity presents daily patterns. • Results suggest the existence of temporal and spectral acoustic space partitioning. • Pomacentridae and Scorpaenidae species could be sources of prevalent sound types. [ABSTRACT FROM AUTHOR]

• We analyze 629 environmental soundscapes both with BirdNET and by manual listening. • Confidence thresholds can be reliably used to optimize for either precision or recall. • BirdNET IDs can be highly reliable if a high minimum confidence threshold is used. • F1-scores remain moderate (<0.5) for all datasets and confidence thresholds studied. • Best results are obtained when analyzing acoustic datasets of extended duration. 1. Passive acoustic monitoring has become increasingly popular as a practical and cost-effective way of obtaining highly reliable acoustic data in ecological research projects. Increased ease of collecting these data means that, currently, the main bottleneck in ecoacoustic monitoring projects is often the time required for the manual analysis of passively collected recordings. In this study we evaluate the potential and current limitations of BirdNET-Analyzer v2.4, the most advanced and generic deep learning algorithm for bird recognition to date, as a tool to assess bird community composition through the automated analysis of large-scale ecoacoustic data. 2. To this end, we study 3 acoustic datasets comprising a total of 629 environmental soundscapes collected in 194 different sites spread across a 19° latitude span in Europe. We analyze these soundscapes using both BirdNET and manual listening by local expert birders, and we then compare the results obtained through the two methods to evaluate the performance of the algorithm both at the level of each single vocalization and for entire recording sequences (1, 5 or 10 min). 3. Since BirdNET provides a confidence score for each identification, minimum confidence thresholds can be used to filter out identifications with low scores, thus retaining only the most reliable ones. The volume of ecoacoustic data used in this study did not allow us to estimate species-specific minimum confidence thresholds for most taxa, so we instead used and evaluated global confidence thresholds selected for optimized results when consistently applied across all species. 4. Our analyses reveal that BirdNET identifications can be highly reliable if a sufficiently high minimum confidence threshold is used. However, the inevitable trade-off between precision and recall does not allow to obtain satisfactory results for both metrics at the same time. We found that F1-scores remain moderate (<0.5) for all datasets and confidence thresholds studied, and that acoustic datasets of extended duration seem to be currently necessary for BirdNET to provide a reliable and minimally comprehensive picture of the target bird community. We estimate, however, that the usage of species- and context-specific minimum confidence thresholds would allow to substantially improve the global performance benchmarks obtained in this study. 5. We conclude that a judicious use of AI-based identifications provided by BirdNET can represent a powerful method to assist in the assessment of bird community composition through the automated analysis of ecoacoustic data, especially when applied to acoustic datasets of extended duration. [ABSTRACT FROM AUTHOR]

Continuous underwater Passive Acoustic Monitoring (PAM) has emerged as a strong tool for cetacean research. To handle the vast volume of collected data, it is essential to employ automated detection and classification methods. The recent advancement of deep learning, involving model training and testing, requires a large amount of labeled data. These labels are derived through the manual annotation of audio files often reliant on human experts. Based on an annotation campaign focusing on blue whale calls in the Indian Ocean involving 19 novice annotators and one expert in bioacoustics, this study explores the integration of novice annotators in marine bioacoustics research, through citizen science programs, which could drastically increase the size of labeled datasets and enhance the performance of detection and classification models. The analysis reveals distinctive annotation profiles influenced by the complexity of vocalizations and the annotators' strategies, ranging from conservative to permissive. To address the challenges of annotation discrepancies, Convolutional Neural Networks (CNNs) are trained on annotations from both novices and the expert. The results show variations in model performance. Our work highlights the importance of annotation guidelines encouraging a more conservative approach to improve overall annotation quality. In an effort to optimize the potential of multi-annotation and mitigate the presence of noisy labels, two annotation aggregation methods (majority voting and soft labeling) are proposed and tested. The results demonstrate that both methods, particularly when a sufficient number of annotators are involved, significantly improve model performance and reduce variability: the standard deviation of the area under PR and ROC curves fall under 0.02 for both vocalizations with 13 aggregated annotators, while it was at 0.17 and 0.21 for the Blue Whale Dcalls and 0.05 and 0.04 for the SEIO PBW vocalizations with all annotators separately. Moreover, these aggregation methods enable the training of models using non-expert annotations that achieve performance of models trained with expert annotations. These findings suggest that crowdsourced annotations from novice annotators can be a viable alternative to expert annotations. • Inter-annotators variability influences the performance of automatic detection. • Majority voting and soft labeling are efficient ways to aggregate annotations for model training. • Annotations obtained from several non-experts can be used to train efficient detection models. • Citizen sciences and multi-annotation can be a tool to overcome the lack of annotation. [ABSTRACT FROM AUTHOR]

Marine mammals often aggregate and forage in specific 'hotspot' areas, which are associated with the predictability of prey sources driven by certain oceanographic processes. However, our understanding of the interactions between marine predators and the oceanographic processes remains incomplete, particularly in near-coastal and estuarine regions. This study investigated how predictable oceanographic processes influence the distribution patterns of Indo-Pacific humpback dolphins (Sousa chinensis) foraging in a known hotspot within the Pearl River Estuary, characterized by the complex topography of its manmade harbor and the interplay between strong tidal flows and river plumes. To address the fine-scale spatial distribution pattern of dolphin foraging, we conducted fixed passive acoustic monitoring of foraging 'buzzes' both inside and outside the harbor. Additionally, boat-based transect surveys were conducted to explore the relationship between dolphin foraging and associated water properties and current fields driven by oceanographic processes. Temporal trend analyses of buzzes revealed a pronounced tidal-related pattern in dolphin foraging activity, with increased detections during ebbing tides outside the harbor. Transect surveys indicated areas with elevated chlorophyll and turbidity levels attracted more dolphins during flood tides, whereas regions with high surface salinity gradients were preferred by dolphins during ebb tides. Oceanographic analysis recognized these bio-physical links as indicators of oceanographic features, including chlorophyll-rich river plumes, converging fronts, and shear-induced turbulence. These features enhance prey abundance and availability for humpback dolphins. This study highlights that oceanographic processes are crucial for determining the fine-scale spatiotemporal distribution of foraging in coastal dolphins, and offers insights into the effective prediction of dolphin aggregation. [Display omitted] [ABSTRACT FROM AUTHOR]

Spiny lobsters are targeted in capture fisheries due the high market demand and this situation has led to overfishing activities. Therefore, a smart monitoring system is imperative to be developed in handling this issue in order to store the distribution of marine spiny lobsters' information such as habitat preferences, population density and biology aspects of the lobsters. This study focused on separating the mixed underwater acoustic sound. An output of the separation system was used for Passive Acoustic Monitoring (PAM) application. The objective of this study is to provide estimated source signals from a recorded mixed acoustic signal. The feasibility of extracting spiny lobster sound as the target sound was observed during the experiment. In this study, Blind Source Separation (BSS) approach was employed to estimate the target sound (i.e., spiny lobsters) from the mixed underwater acoustic signal consisting of the sound of spiny lobsters and several man-made and natural sounds. In this investigation, two different blind source separation methods namely Fast Fixed Point Independent Component Analysis (FastICA) and Non Negative Matrix Factorization (NMF) were implemented. The mixture of one target sound which was spiny lobster and four interferences signals were used as input in this experiment,. The separated source sound by using FastICA with Negentropy, FastICA with Kurtosis and NMF algorithms were compared and evaluated based on bss_eval_sources toolbox metrics. These metrics consisted of signal to distortion ratio (SDR), signal to interferences ratio (SIR) and signal to artifacts ratio (SAR). In conclusion, the FastICA with Negentropy technique provides the best performance in separating mixed signal based on SIR, SAR and SDR measurement results that showed the FastICA with Negentropy generated the highest average values compared to the FastICA with Kurtosis and NMF. [ABSTRACT FROM AUTHOR]

Penaeid shrimp emit click sounds during feeding activity generated by the collision of mandibles. These sounds have been used as a basis for research of feeding behavior and acoustic-based automatic feeders in shrimp farming. However, the use of automatic feeders is generally recommended for shrimp larger than 2 g or more. The aim of the present study was to evaluate the click acoustic parameters during feeding activity of Penaeus vannamei from post-larvae up to 12 g in laboratory. Twelve progressive weight classes (0.01 to 12 g) along nursery and grow-out phases were recorded in tanks internally covered with an acoustic foam. Recordings were carried out using an omnidirectional SoundTrap 300 STD hydrophone (sampling rate: 288 kHz; 16 bits). At the end of the recordings, shrimp were measured (body weight and cephalothorax length) and had their mandibles dissected for length measurements. The clicks emitted by shrimp during acoustic recordings were automatically detect using Raven® software, and 100 clicks per weight class were randomly selected for acoustic sound characterization (low, high and peak frequency, maximum power and duration time). The clicks start to be emitted by P. vannamei at the average weight of 0.01 g, with a band frequency between 15.4 and 70.5 kHz and low maximum power (23.2 dB). The low and peak frequency showed an abrupt decrease until shrimp reached 1.16 g, while maximum power and high frequency presented an opposite pattern. The click sound duration time did not show a clear variation among the different weight classes evaluated. Acoustic parameters showed a significant linear relationship with mandible length, indicating a direct influence of this structure on sound emission. The regression derivative analysis estimated that click maximum power stabilized from 1.48 g onwards, while low, high and peak frequency stabilized from 3.08 g shrimp onwards. The results suggest that it would be possible to anticipate and optimize the use of acoustic-based automatic feeders in P. vannamei farming. • Click emission during feeding activity is first reported for post-larvae of Penaeus vannamei. • The mandible length was correlated with the click acoustic parameters. • The acoustic parameters values stabilized around 1.5 and 3.0 g. • Results indicate the possibility to anticipate and optimize the use of acoustic-based automatic feeders in shrimp farming. [ABSTRACT FROM AUTHOR]

Monitoring threatened species at the individual level is fundamental and crucial in ecology and conservation, especially for critically endangered species. Hainan gibbon (Nomasucs hainanus) is the world's rarest primate species with <40 individuals, urgently needs a labor-efficient and continuous individual-level monitoring system to inform science-based conservation actions. Male dispersal, replacement, and formation of new groups with females are crucial for the population dynamics of gibbons. Their loud and elaborate morning songs provide the potential to automate the monitoring of their dynamics. Based on acoustic recordings from focal recording and 13 autonomous recording units within gibbons' home ranges, we developed an automated system, which successively (a) detected song bouts (duration ≈ 10 min, with ≈ 50 phrases) with temporal parameters using an energy detector with a hidden Markov model (F1 score >75 %), (b) detected and classified phrases (duration ≈ 5 s) within song bouts using a random forest model (F1 score ≈ 80–90 %), and (c) classified known individuals by a linear support vector machine (accuracy >93 %), and recognized unknown individuals by Gaussian mixture models (accuracy >90 %). The system successfully monitored a male immigration and kept tracking a dispersing male for nearly a year since it left the natal group to establish its own group with two females. Our system is effective even in the face of an ever-changing population and can be applied to passive acoustic data. It will contribute to the realization of long-term, large-scale, and continuous monitoring of individual dynamics for vocal animals. • We identified and tracked the dispersal of male Hainan gibbons with an automated system. • The system needs little training data and is robust against data variations. • The system helps to monitor individual dynamics from passive acoustics continuously. [ABSTRACT FROM AUTHOR]

• Passive acoustic monitoring (PAM) detects rare/cryptic species across landscapes. • We trained a convolutional neural network (PNW-Cnet) to identify marbled murrelets. • PAM is used to conduct the first broad-scale marbled murrelet call phenology study. • PAM and PNW-Cnet are powerful tools for marbled murrelet research and monitoring. Population monitoring is an essential component of biodiversity conservation and management, but low detection probabilities for rare and/or cryptic species makes estimating abundance and occupancy challenging. Passive acoustic monitoring combined with machine learning algorithms represents a potential path forward to effectively and efficiently monitor the occurrence of rare vocalizing species across entire forest landscapes. Our objectives were to develop and implement a convolutional neural network (PNW-Cnet) to identify vocalizations of a rare and threatened forest nesting bird species – the marbled murrelet (Brachyramphus marmoratus) – in the Pacific Northwest, U.S.A., 2018–2021. We used PNW-Cnet predictions from broadscale passive acoustic monitoring data to examine spatiotemporal patterns in the distribution of murrelets. PNW-Cnet showed sufficiently high prediction accuracy (overall precision > 0.9) to enable broadscale population monitoring. Spatiotemporal analysis showed that annual peak murrelet call abundance occurs in ordinal weeks 28–32 (late July–Mid August) but this varied by study area. The greatest number of detections typically occurred in the Olympic Peninsula and Oregon Coast Range where late-successional forest dominates and nearer to ocean habitats. We demonstrate that passive acoustic monitoring can be used to understand intensity of use across broad scales for a rare and cryptic species in addition to the typical detection/non-detection data that are often collected. Passive acoustic monitoring combined with PNW-Cnet offers considerable promise for species distribution modeling and long-term population monitoring for rare species. [ABSTRACT FROM AUTHOR]

Fishing is a significant economic sector in Atlantic Canada. Increased fishing vessel density raises ocean ambient noise, adversely affecting the marine ecosystem. This study aimed to evaluate the underwater radiated noise emitted by a typical fishing vessel, identify the noise sources, and determine their respective contributions to the overall noise signature of the vessel. Dipole and monopole source levels are estimated through passive acoustic monitoring during sailing and stationary conditions. This is accomplished by employing a set of hydrophones in conjunction with a numerical propagation loss model and oceanography data. This study also addresses the correlation between structure-borne noise and underwater radiated noise. Over 70% of the noise at frequency bands of 63 and 250 Hz and approximately 40% of the noise above 1 kHz were attributed to the diesel engine, indicating that it significantly contributed to the vessel's noise signature with the propeller. In the unclutched propeller mode, the vessel engine continues to generate noise 40 dB above the background noise and peaks at the 250 Hz band. By addressing this knowledge gap, we can potentially contribute to future endeavours to enhance fishing vessel design and lower the impact of individual noise sources. • Comparative estimation of vessel noise signature with PAM and numerical PL models. • PL model based on range-dependent sound speed profile and seabed characteristics. • Significant contribution of the engine to the vessel noise (> 70% @ 63, 250 Hz). • Relation between underwater noise and onboard velocity levels in different conditions. • Analysis of individual contribution of noise sources to the vessel acoustic signature. [ABSTRACT FROM AUTHOR]

Shipping is one of the largest industries globally, with well-known negative impacts on the marine environment. Despite the known negative short-term (minutes to hours) impact of shipping on individual animal behavioural responses, very little is understood about the long-term (months to years) impact on marine species presence and area use. This study took advantage of a planned rerouting of a major shipping lane leading into the Baltic Sea, to investigate the impact on the presence and foraging behaviour of a marine species known to be sensitive to underwater noise, the harbour porpoise (Phocoena phocoena). Passive acoustic monitoring data were collected from 15 stations over two years. Against predictions, no clear change occurred in monthly presence or foraging behaviour of the porpoises, despite the observed changes in noise and vessel traffic. However, long-term heightened noise levels may still impact communication, echolocation, or stress levels of individuals, and needs further investigation. • Elevated shipping noise and vessel traffic did not influence porpoise occurrence. • No change was observed in either detection rate or foraging behaviour of porpoises. • Porpoises continued to use preferred habitat despite changes in underwater noise. • Long-term impacts on communication, echolocation, or stress remain unknown. [ABSTRACT FROM AUTHOR]

Continuous bioacoustic monitoring is an emerging opportunity as well as a challenge, allowing detection of cryptic species' activity while producing high computational demands. In this paper, we present an automated framework that allows the monitoring of a large number of bird species by their vocalizations over extended periods. The framework relies on the BirdNET-Analyzer deep learning model. We applied the framework to >80 species; 20 species with the highest recall scores were selected for further analysis. We used the framework to analyze acoustic signals recorded continuously for over two years using autonomous recorders at various locations in Agmon Hula Lake Park, Israel. During this period there was an acute outbreak of avian influenza in the area. We analyzed differences in acoustic occupancy for various species between two consecutive years (November 2020 to October 2022). We examined between-year population trends for 17 species, both migratory and resident, and found a significant decline in vocal activity between the two years for 10 species. We assume that this decline is related to the avian influenza outbreak, suggesting that the impact of the pandemic may be more widespread and affected a greater number of local species than was previously realized. This further highlights the power and effectiveness of bioacoustic monitoring in detecting cryptic but dramatic dynamics. [Display omitted] • Monitoring bioacoustics over two consecutive years, including an avian-flu outbreak. • An automated framework for tracking bird species occupancy is presented. • 'Vocalogram' - A visualization method of bioacoustic data is proposed. • Yearly comparisons show a significant decline in vocal activity for most species. • The decline in vocal activity may be linked to the avian-flu outbreak. [ABSTRACT FROM AUTHOR]

Passive acoustic monitoring is increasingly used for assessing populations of marine mammals; however, analysis of large datasets is limited by our ability to easily classify sounds detected. Classification of beaked whale acoustic events, in particular, requires evaluation of multiple lines of evidence by expert analysts. Here we present a highly automated approach to acoustic detection and classification using supervised machine learning and open source software methods. Data from four large scale surveys of beaked whales (northwestern North Atlantic, southwestern North Atlantic, Hawaii, and eastern North Pacific) were analyzed using PAMGuard (acoustic detection), PAMpal (acoustic analysis) and BANTER (hierarchical random forest classifier). Overall classification accuracy ranged from 88% for the southwestern North Atlantic data to 97% for the northwestern North Atlantic. Results for many species could likely be improved with increased sample sizes, consideration of alternative automated detectors, and addition of relevant environmental features. These methods provide a highly automated approach to acoustic detection and classification using open source methods that can be readily adopted for other species and geographic regions. • BANTER is an open source, accessible machine learning classifier for data with multiple call types • BANTER can use multiple detectors on single call types for acoustic event classification • BANTER can classify beaked whale species with relatively low sample sizes • BANTER is easy to use and can be easily adopted to address current classification needs [ABSTRACT FROM AUTHOR]

Ecoacoustics is an emerging science that seeks to understand the role of sound in ecological processes. Passive acoustic monitoring is being used to collect vast quantities of soundscape audio recordings to study variations in acoustic community and monitor biodiversity. However, extracting relevant information from soundscape recordings is non-trivial. Recent approaches to machine-learned acoustic features appear promising but are limited by at least three issues: inductive biases, lack of interpretability and crude temporal integration. In this paper we introduce a novel self-supervised representation learning algorithm for ecoacoustics - a convolutional Variational Auto-encoder (VAE) - and directly address these shortcomings. Firstly, we train the network on soundscape recordings from temperate and tropical field sites along a gradient of ecological degradation to provide a more relevant inductive bias than prior approaches. Secondly, we present a new method that allows interpretation of the latent space for the first time, giving insight into the basis of classification. Thirdly, we advance existing methods for temporal aggregation of learned embeddings by encoding latent features as a distribution over time. Under our approach to increase interpretability, we provide insight into how learned features drive habitat classification for the first time: inspection of latent space confirms that varying combinations of biophony, geophony and anthrophony are used to infer sites along a degradation gradient. Our novel temporal encoding method increases sensitivity to periodic signals and improves on previous research that uses time-averaged representations for site classification. This approach also reveals the contribution of hardware-specific frequency response that create a potential bias; we demonstrate how a simple linear transformation can be used to mitigate the effect of hardware variance on the learned representation under our approach. Our novel approach paves the way for development of a new class of deep neural networks that afford more interpretable learned ecoacoustic representations to advance both fundamental and applied science and support global conservation efforts. [Display omitted] • Ecoacoustics needs methods to interpret how soundscapes are embedded in summary descriptors to advance the science. • VAE decoder shows mixed biophony, geophony and anthrophony infer a habitat quality degradation gradient. • Learned differences in recorder hardware normalised using a linear transformation of latent encodings. • Preserving occasional or periodic acoustic signals by binning latent features over time improves site prediction accuracy. [ABSTRACT FROM AUTHOR]

Pacific whiteleg shrimp is the topmost aquacultured species produced in the world. However, shrimp feeding behavior results in unnecessary expenditures, as some of the feed offered is not entirely consumed. Nowadays, aquaculture has become more accepting of using animal by-products such as poultry meal as a protein source to replace fishmeal. Unfortunately, such by-products might affect shrimp acceptance of the offered feed. The present work evaluates the use of a chemosensory feed effector with attractive properties on the feeding behavior, growth performance, and salinity stress tolerance in juvenile Litopenaeus vannamei. Nine diets were formulated to contain either poultry meal (6%), or fishmeal (6% or 12%) as animal protein source, with feed effector supplemented at 0, 0.1, and 0.2%. The various diets were offered to naïve and non-naïve shrimp to evaluate feed consumption. Concurrently, shrimp feeding behavior was monitored using passive acoustic monitoring. Afterwards, a 42-days growth trial was performed followed by a salinity stress trial were performed to evaluate the effect of the feed effector in shrimp diets on animal growth, survival, and stress tolerance. Results of the present study indicate significant positive relationships between the number of clicks and feed consumed after 30 min (p < 0.001, r = 0.46 to 0.69). Additionally, shrimp offered the poultry meal-based diets produced significantly more "clicks" when the diets were supplemented with the feed effector, compared with the rest of the diets. However, the shrimp consumed poultry meal-based diets similarly to the fishmeal-based diets, only when supplemented with the feed effector. Interestingly, no significant differences were observed in terms of final biomass, final average individual weight, weight gain, FCR, or survival. However, salinity stress trial results indicate significantly better survival when offered diets supplemented with the feed effector, regardless of the level of inclusion. Results of the present work suggest that supplementing the feed effector to shrimp diets improves palatability and shrimp acceptance of diets formulated with poultry meal as sole animal protein source, as well as improves shrimp tolerance to salinity fluctuations. In an era of limited resources, supplementation of poultry meal-based shrimp diets with feed effector improves overall profitability by obtaining good shrimp growth with less expensive diets, and despite of stressors. • Feed effectors can improve shrimp acceptance to feed formulated with animal by-products proteins to replace fishmeal. • Passive acoustic monitoring is useful to assess shrimp feeding on animal by-products based feed, with feed effectors added. • Feed effectors have the potential to improve shrimp tolerance to salinity stress. • The feed effector positively influenced the number of clicks emitted by the shrimp when feeding. • Despite not influencing growth or survival, shrimp tolerance to salinity stress improved with feed effector addition. [ABSTRACT FROM AUTHOR]

• Urban ecosystems near cities offer key services, facing urbanisation threats. • Acoustic monitoring assesses habitat status, yet produces vast data. • Neural networks and acoustic indices proposed to refine data analysis. • Short-time acoustic indices boost sound event detection accuracy. • Methodology validates with F1-score of 0.614 for habitat conservation. Urban-natural environments, proximal to rapidly urbanizing cities, provide essential ecosystem functions that benefit both city residents and ecological communities. With escalating urbanization, the resilience of these ecosystems is being progressively challenged, highlighting the need for robust monitoring mechanisms. Acoustic monitoring has emerged as an unobtrusive method to evaluate the status of these environments, capitalizing on indicators that reflect both landscape features and specific acoustic events. Despite potentially offering significant insights, this approach generates a large volume of acoustic data, introducing complexities in subsequent analyses. To mitigate this, we propose integrating artificial neural networks with acoustic indices to enhance data analysis. Our approach emphasizes the usefulness of short-time acoustic indices, computed over finite-duration analysis windows, to enhance polyphonic sound event detection accuracy. Empirical results support the performance of our approach, registering both an F1-Score and an error rate of 0.614. Overall, this study delineates a novel paradigm geared towards enhancing or preserving the biological diversity of urban-natural environments in areas with population growth and urban development. [ABSTRACT FROM AUTHOR]

The use of autonomous recording units (ARUs) for passive acoustic monitoring has recently gained a lot of importance. It is now widely used for scientific purposes and has been increasingly applied in the field of conservation biology. However, while ARUs may greatly increase cost-effectiveness in monitoring, they are not exempt from possible biases. In this study, we compared the performances of three different software in recognizing bi- and tri-syllabic bird songs and tested if their results varied significantly in a real setting. We focused on the interpretation of the relative abundance among habitats and the time of vocal activity, also using different numbers of training records to assess if the results of the software improved. Even with the quality measures of the recognizer being consistent with those of previously published studies, in several cases the results produced significantly different interpretations depending on the software used. While increasing the number of training records slightly improved the overlap in the estimates of activity among the different software, differences in the relative abundance of each song were present even with a fairly solid number of training records. Our results suggest that reliable biological interpretations might only be attained with a large number of training records and very high values of precision, recall, and the F-score. In fact, even with very high precision and F-scores, lower values for recall led not only to statistically significant differences, but also to different interpretations of relative abundance and time of activity between the two best-performing software. Thus, great caution must be taken in interpreting the results of bioacoustics studies that use automated recognition, and standards of minimum quality should be created for recognizers in scientific and technical studies. • Automatic bird song recognition can yield different results depending on software. • Repetability of automatic bird song recognition studies can be compromised. • Increasing number of training records can improve automatic bird song recognition. • High quality measures should be demanded for automatic bird song recognition studies. [ABSTRACT FROM AUTHOR]

[Display omitted] • Recording devices and environmental noise affects acoustic indices. • Further improve the reliability and accuracy of the acoustic index. • Microphone with omnidirectional and flat amplitude frequency response is necessary. • The signal-to-noise ratio of the signal should meet the requirement of threshold. Passive acoustic monitoring serves as a minimally invasive and effective method for biodiversity assessment, particularly in bird monitoring through the application of acoustic indices. However, the use of different recording devices and different types of environmental noise (e.g., rain, wind, stream, traffic noise) lead to signal distortions that affect the application of ecoacoustics indices. Currently, there are no established guidelines specifying the technical requirements for recording devices and the signal-to-noise ratio (SNR) threshold for accurate calculation of acoustic indices. To enhance the accuracy of acoustic indices in bird biodiversity assessments, this study investigated the impact of different recording devices and environmental noise (rain, wind, stream, traffic noise) on acoustic indices. In this study, we selected six acoustic indices: Acoustic Complexity Index, Acoustic Diversity Index, Acoustic Evenness Index, Bioacoustic Index, Acoustic Entropy Index, and Normalized Difference Soundscape Index, and we used four different recording devices to simultaneously record 104 h of bird-sound data at the same location. In addition, 44 h of noisy bird signals with different noise types and intensities were artificially synthesized for comparison. The sound data were then used to analyze the effects of the recording devices and environmental noise on the acoustic indices used for bird biodiversity assessment. Our results showed that (a) all six acoustic indices were affected by the recording device used; (b) each acoustic index had different sensitivities to different noise types; and (c) there was a SNR threshold above which the effect of noise on the acoustic index was negligible. This study provides recommendations for recording device selection and determines SNR thresholds for noisy bird signals, contributing to the refinement of protocols for acquiring and preprocessing signals in the application of acoustic indices for bird biodiversity assessment. These findings aim to establish standardized protocols for signal acquisition and preprocessing in future application of acoustic indices. [ABSTRACT FROM AUTHOR]

The European Marine Strategy Framework Directive (MSFD, 2008/56/CE) aims to achieve and maintain the 'good environmental status' (GES) of marine ecosystems, including the assessment of marine mammal populations. However, assessing their abundance and distribution is challenging due to their mobility, elusiveness, and difficult data acquisition. This article highlights the role of passive acoustic monitoring as a complementary approach to traditional methods (mainly plane and boat surveys). It also presents the conclusions and perspectives of a workshop organized at the 34th conference of the European Cetacean Society in O'Grove, Spain (18th to 20th April 2023). Building on the rich foundation established by the European Cetacean Society (ECS) in 1987, this society plays a pivotal role in advancing marine mammal research and conservation efforts across European waters. The workshop emphasized the need for harmonization of protocols, indicators, and guidelines to ensure comparability of results. The establishment of a dedicated European working group was proposed to coordinate passive acoustic monitoring efforts, foster collaboration, and share best practices. The workshop concluded that the consolidation of scientific knowledge, the development of common guidelines, and the establishment of a transnational working group would improve the understanding and conservation of these vulnerable species and their habitats. [ABSTRACT FROM AUTHOR]

• A novel end-to-end deep learning model, named CicadaNet is proposed for filtering cicada chorus from recordings containing bird sound • Our proposed algorithm achieves the best performance over current state-of-the-art deep denoising models and traditional denoising algorithms • A clean bird sound dataset is built and a large amount of real cicada noise data is collected for model evaluation Passive acoustic monitoring has been an effective tool for bird sound analysis. However, bird sounds often include cicada noise, which is an obstacle for investigating bird sounds. For example, cicada noise can result in large deviations of acoustic index, which will lead to the mismonitoring of species richness trends. Therefore, there is a critical need to filter cicada noise for helping bird sound analysis. We develop a novel end-to-end deep learning model, named CicadaNet for filtering cicada chorus from recordings containing bird sound. CicadaNet utilizes a convolutional encoder-decoder network to encode and decode acoustic features and a conformer module for global and local sequence modeling. We build a clean bird sound dataset and collect a large amount of real cicada noise data for model evaluation. We compare CicadaNet with current state-of-the-art deep denoising models and traditional denoising algorithms. Experimental results show that CicadaNet achieves the best denoising performance (SegSNR is improved by 9.59 dB and SI-SNR is improved by 20.08 dB when the noisy SNR = 0 dB). Meanwhile, CicadaNet achieves good performance for the real-time denoising of cicada noise. Furthermore, CicadaNet achieves bird species-independent noise reduction. We evaluate the effectiveness of CicadaNet for bird diversity survey. CicadaNet achieves the best performance, which can effectively eliminate the deviation caused by cicada noise to the acoustic index. CicadaNet can be easily extended to the cancellation of other environmental noise, and we propose it for the acoustic denoising of other vocalizing animals. [ABSTRACT FROM AUTHOR]

• Acoustic indices have shown inconsistencies in biodiversity assessments. • The Valdivian rainforest is a biodiversity hotspot in South America. • Seven acoustic indices were evaluated as proxies of a biodiversity measure. • None of the indices showed a strong association with bird and anuran richness. • Two acoustic indices showed low to moderate correlation values with bird richness. Passive acoustic monitoring is becoming an extensively used tool to evaluate the status and variation of populations of sound producing animals. The analyses of extensive acoustic recordings for identification and detection of acoustic signals of different species is highly time-consuming, either by traditional audiovisual procedures or by developing effective automated recognizers. These drawbacks in data analysis have promoted research efforts aimed to develop acoustic diversity indices, which are relatively easily obtained by means of different algorithms considering spectral and/or temporal properties of the sounds contained in the recordings. Nevertheless, studies performed in different environments and geographical areas reveal inconsistencies in the association between acoustic diversity indices and biodiversity, suggesting the need of new studies to evaluate commonly used acoustic diversity indices as proxies of the richness of sound producing animal species. The Valdivian rainforest from Chile, South America, is recognized as a biodiversity hotspot because of the high proportion of endemic species and their threatened status associated to anthropogenic activity. As it is imperative to evaluate cost-effective strategies for biodiversity monitoring, in this study we evaluated seven acoustic indices to assess their reliability as proxies of the variation in bird and anuran species richness, two important components of the biodiversity of this threatened environment. Our results indicate that most of the acoustic indices tested fail to describe satisfactorily the variation in species richness. Nevertheless, two indices, namely the Temporal Entropy and the Acoustic Evenness Index, may potentially serve as an indicator of bird richness, but future studies should fine-tune these indices to obtain a robust validation of its use within this environment. We expect that this work will contribute to the understanding of the significance and potential use of acoustic indices within this biodiversity hotspot as well as in other regions of interest for conservation. [ABSTRACT FROM AUTHOR]

Abstract Benthic invertebrates emit broadband transient sounds, that dominate coastal soundscapes and can inform on the environment and its ecological state. Numerous works have studied temporal patterns of Benthic Invertebrate Biophonies (BIB) and linked them to environmental parameters or ecological processes, but there is a lack of knowledge about the spatial variability of BIB. Here, we mapped Benthic Invertebrate Sounds (BIS) in three dimensions using a compact array (2 m × 2 m × 2 m) of four hydrophones within the scope of eco-acoustic s studies. Firstly, we developed tools to localize any individual sound emitted by benthic invertebrates. A theoretical calculation of the Cramer Rao Bounds (CRB) consolidated by in situ active emissions from known positions, demonstrated the accuracy of the localization. Secondly, by accumulating all the positions of the BIS recorded during a night, three methods are proposed to map the BIB at different spatial scales. An in situ measurement campaign carried out on a 300 m × 300 m artificial reef system revealed that the marine fauna was binary spatially distributed: rich and numerous on artificial reefs and poor and weak outside the reefs. At small ranges (10 m), the acoustic maps matched the geometric structures of the reefs with an accuracy of 0.5 m. At large scales (100 m and 300 m), the acoustic hot-spots on the maps corresponded to distant reefs, with an accuracy of 7 m and 10 m respectively. This study provides the first maps of the BIB at ecologically relevant scales and shows that they can be used to identify, quantify, follow and track benthic activity hot-spots. In combination with biological and ecological information, passive acoustic monitoring using such maps can be a powerful complementary tool for ecological studies, such as the quantification of grazing activity or the evaluation of the efficiency of marine ecologic restoration programs. [ABSTRACT FROM AUTHOR]

Highlights • We conducted passive acoustic monitoring (PAM) to detect deer and macaques. • PAM semi-automatically detected the different call types of each species. • PAM detects deer and macaques with higher sensitivity than camera trap. • The PAM detection area was >100 times wider than that of camera traps. • PAM is useful to understand animal community composition and conditions. Abstract In recent years, camera traps have rapidly become popular for the large-scale monitoring of wildlife distribution and population; however, we should not ignore the uncertainty regarding the reliability of camera-based monitoring by inexperienced data gatherers. This study introduces passive acoustic monitoring (PAM) as an easier technique for monitoring terrestrial mammals that uses the sound cues that they produce. To validate the efficacy of PAM, we quantitatively compared the detection areas and rates between sound cues (from PAM) and visual cues (from camera traps) of two mammals—the sika deer Cervus nippon and the Japanese macaque Macaca fuscata —across seven study sites in eastern Japan with different population densities. To collect sound cues, we set up multiple autonomous recording units at the sites and continuously recorded ambient sounds, following a pre-determined schedule. The total recording time reached 9081 h for deer and 8235 h for macaques. We then built sound recognizers to automatically detect eight target call types from the recorded data. To collect visual cues, we also set multiple camera traps at the same sites and for the same observation periods. The key findings were as follows: (1) the fully automated procedures that only used the recognizers to detect sound cues produced numerous false positive detections when the call type possessed vocal plasticity and variations; (2) the semi-automated procedures, which included an additional step to validate the automated detections by manual screening, exhibited a great improvement in the detectability and recall rates of the half of the target calls, reaching >0.70; (3) when using the semi-automated procedures, the frequency of deer and macaque detections per trap-day derived from the sound cues were in most cases approximately dozens of times and several times, respectively, higher than that derived from the visual cues; (4) the main advantage of PAM may be its superior detection areas, which were 100–7000 times wider than those of camera traps; and (5) the current success of the recognition of different call types of each species could broaden the use of PAM, which is not possible for camera traps. PAM could provide socio-behavioral data (i.e., the frequencies and types of inter-individual vocal communications) that could help understand the status of population dynamics and the group compositions, in addition to information related to the presence or absence of species. [ABSTRACT FROM AUTHOR]

From the acoustic data acquired by the RHUM-RUM (Réunion Hotspot and Upper Mantle Réunions Unterer Mantel) Ocean Bottom Seismometer (OBS) network between October 2012 and November 2013, this study revealed baleen whale occurrence in the western Indian Ocean (IO). Low-frequency songs from three species (Antarctic Blue Whales, Pygmy Blue Whales and Fin Whales) as well as P-calls (or Spot-calls) from an unknown species were recorded on the dataset. The wide arrangement of the OBS network (2000 km × 2000 km) provided valuable information to draw seasonal patterns of occurrence and distribution all over the area. These species occurred sympatrically in the western IO, at least during austral autumn months emphasizing the importance of this region for these populations. This data set helped to refine the knowledge on their spatio-temporal distribution and complete the picture built by previous studies. A tighter sub-network of 8 OBSs deployed on the South West Indian Ridge provided ideal inter-sensor spacing for whale tracking. We demonstrated the capability of such array of detecting and tracking the three different whale species up to 50 km and for several hours. As a result and to understand the effect of acoustic wave propagation, songs from the tracking were described at a close and remote distance of the sensor. This work could also help to understand the local behavior of these species during austral autumn months in this area of the western Indian Ocean. [ABSTRACT FROM AUTHOR]

The urban acoustic environment (AE) provides comprehensive acoustic information related to the diverse systems of urban areas, such as traffic, the built environment, or biodiversity. The decreasing cost of acoustic sensors and rapid growth of storage space and computational power have fostered the collection of large amounts of acoustical data to be processed. However, despite the extensive information that is recorded by modern acoustic sensors, few approaches are established to capture the rich complex dynamics embedded in the time-frequency domain of the urban AE. Quantitative methods need to account for this complexity, while effectively reducing the high dimensionality of acoustic features within the data. Therefore, we introduce complex networks as a tool for analyzing the complex structure of large-scale urban AE data. We present a framework to construct networks based on frequency correlation matrices (FCMs). FCMs have shown to be a promising tool to depict environment specific interrelationships between consecutive power spectra. Accordingly, we show the capabilities of complex networks for the quantification of these interrelationships and thus, to characterize different urban AEs. We demonstrate the scope of the proposed method, using one of the world's most extensive longitudinal audio datasets, considering 3-min audio recordings (n = 319,385 ≙ 665 days) from 23 sites. We construct networks from hour-of-day specific audio recordings for each site. We show that the average shortest path length (ASPL) as an indicator for dominance of sound sources in the urban AE exhibits spatial- and temporal-specific patterns between the sites, which allows us to identify four to seven clusters of distinct urban AEs. To validate our findings, we use the land use mix around each site as a proxy for the AE and compare those between and within the clusters. The identified clusters show high intra- and low inter-cluster correlations of ASPL diel cycles as well as strong intra-similarities in land use mix. Our results indicate that complex networks are a promising approach to analyze large-scale audio data, expanding our understanding of the time-frequency domain of the urban AE. • The introduction of complex networks for the analysis of the acoustic environment. • The application to a dataset of approx. One million minutes of audio recordings. • Average shortest path length as indicator for the acoustic dominance. • Complex networks capture temporal characteristics of the acoustic environment. • Complex networks capture spatial features of the acoustic environment. [ABSTRACT FROM AUTHOR]

A novel framework for acoustic detection and species identification is proposed to aid passive acoustic monitoring studies on the endangered Indian Ocean humpback dolphin (Sousa plumbea) in South African waters. Convolutional Neural Networks (CNNs) were used for both detection and identification of dolphin vocalisations tasks, and performance was evaluated using custom and pre-trained architectures (transfer learning). In total, 723 min of acoustic data were annotated for the presence of whistles, burst pulses and echolocation clicks produced by Delphinus delphis (~45.6%), Tursiops aduncus (~39%), Sousa plumbea (~14.4%), Orcinus orca (~1%). The best performing models for detecting dolphin presence and species identification used segments (spectral windows) of two second lengths and were trained using images with 70 and 90 dpi, respectively. The best detection model was built using a customised architecture and achieved an accuracy of 84.4% for all dolphin vocalisations on the test set, and 89.5% for vocalisations with a high signal to noise ratio. The best identification model was also built using the customised architecture and correctly identified S. plumbea (96.9%), T. aduncus (100%), and D. delphis (78%) encounters in the testing dataset. The developed framework was designed based on the knowledge of complex dolphin sounds and it may assists in finding suitable CNN hyper-parameters for other species or populations. Our study contributes towards the development of an open-source tool to assist long-term studies of endangered species, living in highly diverse habitats, using passive acoustic monitoring. [Display omitted] • Novel framework to detect and identify dolphin sounds in long-term recordings. • Detection model achieved 89.5% accuracy for sounds used in ecological studies. • Audio segments length and images dpi were important for both models' accuracy. • Best multi-class species classifier model achieved 96.9% accuracy for S. plumbea. [ABSTRACT FROM AUTHOR]

• The global threshold scheme to generate the binary spectrogram makes ADI sensitive to noise. • The proposed FADI is robust to noise within a much wider SNR range as compared with ADI. • FADI has a potential to expand the spatial–temporal coverage in PAM applications. In recent years, passive acoustic monitoring (PAM) has been widely applied in the field of soundscape ecology. Meanwhile, acoustic indices, as an important tool for rapid biodiversity assessment (RBA), have also attracted increasing attention. However, considerable evidence demonstrates that existing acoustic indices are susceptible to noise, which poses a challenge for wider applications of acoustic indices. In this work, we firstly provide insights into how noise affects the widely used acoustic diversity index (ADI), revealing that the global threshold scheme (i.e., −50 dBFS for all time–frequency bins) to generate the binary spectrogram (spectrogram containing time–frequency bins denoted as '1′ with power above the threshold and '0′ below) is the key factor that makes ADI sensitive to noise. To address this problem, we propose a revised ADI version named frequency-dependent acoustic diversity index (FADI), which employs a new threshold scheme to reduce noise impacts on its values. Specifically, considering that the noise floor along frequency axis usually varies in practice, i.e., a colored noise rather than a white noise exists in field recordings, floating thresholds adapted to the noise level at each frequency bin were employed and further incorporated with the global threshold of ADI. Controlled computational experiments using field recordings were conducted where manipulated noise levels were introduced as different signal-to-noise ratios (SNR). It was shown that the proposed FADI was significantly robust to noise within a much wider SNR range as compared with ADI. Furthermore, a singular example using a real-world recording also verified its effectiveness and highlighted the promising potential of FADI in applications of PAM. [ABSTRACT FROM AUTHOR]

• Identification of four types of songs of western black-crested gibbons. • A deep neural network based speech enhancement network is proposed. • Gate Recurrent Unit has been added for continuous song recognition. • Distributed batch audio segmentation based on Bayesian information criterion. • The location of the gibbons was identified using sound intensity. The extraction of species-specific calls from passive acoustic recordings is a common preliminary step in ecological analysis. But for many species, especially those occupying noisy, acoustically variable habitats, the call extraction process remains largely manual, a time-consuming, and increasingly unsustainable process. Deep neural networks have been shown to provide excellent performance in a range of acoustic classification applications. We take as an example the recognition of four songs of one of the rarest mammals in the world, the western black-crested gibbon (Nomascus concolors).The process of recognition in this paper which includes distributed BIC ambient sound segmentation based on the OpenPAI platform; DNN-based western black-crested gibbon song enhancement processing; data pre-processing, labeling samples; proposed DNN + ResNet34 + CBAM + GRU + Attention recognition model; comparing other classical neural network models.Our best model converts acoustic recordings into spectrogram images on the mel frequency scale and uses these images to train convolutional neural networks.Our proposed model proved to be very accurate in predicting the segmented western black-crested gibbon songs with an accuracy of 99.8%, and almost a few western black-crested gibbon songs were incorrectly identified when all segmented data were recognized. In the four consecutive years of the acoustic monitoring system deployed in the Ailao Mountain National Nature Reserve, the western black-crested gibbon was most active near the monitoring site from March to August each year, and least active in January and February. Based on call sound intensity analysis, we monitored a total of two different western black-crested gibbon groups (G1 and G2) during the monitoring cycle. We demonstrate that passive acoustic monitoring combined with CNN classifiers is an effective tool for the remote detection of one of the rarest and most threatened species in the world. [ABSTRACT FROM AUTHOR]

[Display omitted] • Acoustic monitoring: key tool for biodiversity assessment and ecological studies. • Harnessing metadata enhances deep bioacoustics classifier. • Geographical prior shows promise in leveraging citizen databases for bioacoustics. • Multi-branch CNN: significant improvements in bird song classification. • Metadata enables a 63% decrease in false positives for gibbon call detection. Bioacoustics, the exploration of animal vocalizations and natural soundscapes, has emerged as a valuable tool for studying species within their habitats, particularly those that are challenging to observe. This approach has broadened the horizons of biodiversity assessment and ecological research. However, monitoring wildlife with acoustic recorders produces large volumes of data that can be labor-intensive to analyze. Deep learning has recently transformed many computational disciplines by enabling the automated processing of large and complex datasets and has gained attention within the bioacoustics community. Despite the revolutionary impact of deep learning on acoustic detection and classification, attaining both high detection accuracy and low false positive rates in bioacoustics remains a significant challenge. An intriguing yet unexplored avenue for enhancing deep learning in bioacoustics involves the utilization of contextual information, such as time and location, to discern animal vocalizations within acoustic recordings. As a first case study, a multi-branch Convolutional Neural Network (CNN) was developed to classify 22 different bird songs using spectrograms as a first input, and spatial metadata as a secondary input. A comparison was made to a baseline model with only spectrogram input. A geographical prior neural network was trained, separately, to estimate the probability of a species occurring at a given location. The output of this network was combined with the baseline CNN. As a second case study, temporal data and spectrograms were used as input to a multi-branch CNN for the detection of Hainan gibbon (Nomascus hainanus) calls, the world's rarest primate. Our findings demonstrate that adding metadata to the bird song classifier significantly improves classification performance, with the highest improvement achieved using the geographical prior model (F1-score of 87.78% compared to 61.02% for the baseline model). The multi-branch CNNs also proved efficient (F1-scores of 76.87% and 78.77%) and simpler to use than the geographical prior. In the second case study, our findings revealed a decrease in false positives by 63% (94% of the calls were detected) when the metadata was used by the multi-branch CNN, and an increase of 19% in gibbon detection. This study has uncovered an exciting new avenue for improving classifier performance in bioacoustics. The methodology described in this study can assist ecologists, wildlife management teams, and researchers in reducing the amount of time spent analyzing large acoustic datasets obtained from passive acoustic monitoring studies. Our approach can be adapted and applied to other calling species, and thus tailored to other use cases. [ABSTRACT FROM AUTHOR]