Kevin Wilkinghoff, Alessia Cornaggia-Urrigshardt, Zheng-Hua Tan

Autonomous recording units (ARUs) are an emerging technology that allows for passive monitoring of soniferous animals and soundscapes. Over the past decade, ARUs have become a popular tool for monitoring birds for their potential to reduce the labor and costs of traditional in-person sampling procedures. However, uncertainty surrounding factors affecting detection of avian taxa using ARUs can inhibit their monitoring efficacy. Eastern Black Rails (Laterallus jamaicensis jamaicensis) are a secretive marsh bird listed as a federally threatened species in the U.S.A. Eastern Black Rail vocalizations are difficult to detect by field personnel, and numerous in-person surveys can be required to confirm their presence at a site. While ARUs are an alternative for detecting Eastern Black Rails, it is unknown at what maximum distance an ARU can detect their vocalizations. We evaluated factors affecting the detection distance of simulated vocalizations for ARUs in four marsh vegetation types under a range of environmental conditions. Detection distances varied across models, vocalization and vegetation types, and call volume. Kickeedo vocalizations were detected at greater distances, and detection distances increased for all vocalization types in open vegetation. High relative humidity increased detection distances, while louder background noise decreased detection distances. High wind speeds in cordgrass (Spartina spp.) decreased detection probability disproportionately relative to other vegetation types. Based on these results, considerations of survey area, vegetation type, and site condition can allow land managers and researchers to optimize Eastern Black Rail monitoring using ARUs. Given the substantial staff time needed to monitor this species, ARUs may increase the likelihood of detection and provide an efficient alternative to in-person monitoring.

Mammals play vital roles in ecological communities, but many are in rapid decline worldwide. Comprehensive monitoring of mammal populations is crucial for effective conservation, but large-scale m...

Above; Recording an early morning dawn chorus in coastal woodland. On this occasion, I simply mounted an ORTF array on a stand and left the ...

Automated bioacoustic analysis is essential for biodiversity monitoring and conservation, requiring advanced deep learning models that can adapt to diverse bioacoustic tasks. This article presents a comprehensive review of large-scale pretrained bioacoustic foundation models and systematically investigates their transferability across multiple bioacoustic classification tasks. We overview bioacoustic representation learning including major pretraining data sources and benchmarks. On this basis, we review bioacoustic foundation models by thoroughly analysing design decisions such as model architecture, pretraining scheme, and training paradigm. Additionally, we evaluate selected foundation models on classification tasks from the BEANS and BirdSet benchmarks, comparing the generalisability of learned representations under both linear and attentive probing strategies. Our comprehensive experimental analysis reveals that BirdMAE, trained on large-scale bird song data with a self-supervised objective, achieves the best performance on the BirdSet benchmark. On BEANS, BEATs$_{NLM}$, the extracted encoder of the NatureLM-audio large audio model, is slightly better. Both transformer-based models require attentive probing to extract the full performance of their representations. ConvNext$_{BS}$ and Perch models trained with supervision on large-scale bird song data remain competitive for passive acoustic monitoring classification tasks of BirdSet in linear probing settings. Training a new linear classifier has clear advantages over evaluating these models without further training. While on BEANS, the baseline model BEATs trained with self-supervision on AudioSet outperforms bird-specific models when evaluated with attentive probing. These findings provide valuable guidance for practitioners selecting appropriate models to adapt them to new bioacoustic classification tasks via probing.

The intersection of artificial intelligence and environmental conservation is rapidly expanding.