Eavesdropping on estuaries: soundscape spatial variation explained by habitat metrics at multiple scales

Abstract Context Understanding how ecological patterns and processes are affected by spatial heterogeneity is a foundational goal of landscape ecology. Recent extensions of this field have explored how habitat-associated soundscapes could inform biodiversity assessments and conservation decisions. Objectives The present study aims to expand understanding of spatial variation in estuarine soundscapes by evaluating its relationship with the surrounding habitat mosaics, nektonic community, and abiotic variables. Methods We sampled 24 habitat mosaics, including seagrass beds, saltmarsh creeks, oyster reefs, and unvegetated mudflats, in Back and Core Sounds, North Carolina, USA, during six sampling periods from May to August 2019. Soundscape summaries, including sound pressure level (SPL) and species-specific calls, were paired with patch- and landscape-scale habitat, community, and abiotic metrics. Differences among the acoustic spectrum at each site were evaluated using the spectral dissimilarity index and visualized using multivariate analyses. Multi-model inference and variance partitioning were used to evaluate the proportion of soundscape variation explained by each ecological metric. Results Landscape-scale habitat metrics were important explanatory variables for all soundscape metrics summarized. Specifically, multivariate analyses revealed four common soundscape types best differentiated by metrics of habitat size and shape at patch and landscape scales and the dominant sources of biological acoustic activity. Variation in SPL across the continuum of habitat heterogeneity sampled was typically explained by landscape-scale habitat metrics such as species richness, seagrass core area, and oyster reef perimeter. Species-specific call rates were explained by habitat and abiotic metrics and were unrelated to nektonic abundance and richness. Conclusions Our results suggest that soundscape metrics could be useful for monitoring estuarine habitat mosaics and acoustic communities over space and time as well as indicating environmental relationships for some soniferous species, though soundscapes may not be an effective proxy for nekton assemblage structure. This study also highlights the value of integrating multiple remote sensing technologies to understand patterns between landscape structure, ecological dynamics, and estuarine soundscapes as well as demonstrates the application of deep learning frameworks for interpretation of soundscape patterns.