Passive acoustic monitoring as a tool for offshore wind farm biodiversity assessments: a pilot study

1. Abstract

Passive Acoustic Monitoring (PAM) has been used to monitor marine mammal species in Scottish waters over the past 10 years. A question remains over whether recordings collected within the scope of marine mammal monitoring can be used to monitor the presence and changes in the communities of species, such as benthic invertebrates and fish. This study aimed to investigate whether bio-and eco-acoustical measurements can be used to assess changes in benthic biodiversity through unlocking PAM data from eight locations along the east coast of Scotland (UK). The focus of this study was to assess fish’s and crustacean’s active (i.e. intentionally produced sounds for communication) and passive (i.e. un-intentionally non-communicative produced sounds) sounds. As part of this study, data around each study location were extracted from the National Biodiversity Network (NBN) atlas, indicating the potential presence of at least 32 species at our sites that can produce either passive or active sounds. However, invertebrates in particular are understudied in the field of bioacoustics, therefore other soniferous species may be present at the study sites. A total of four 10-minute recordings at dusk, midnight, dawn and mid-day were analysed in Raven, R and Matlab. The phonic richness (PR) was measured as the total number of types of sounds, while the acoustic communities are an aggregation of species which interact acoustically in the same acoustic space. Here, active and passive sounds were grouped together to investigate if differences between the phonic richness and acoustic communities can be related to differences in benthic faunal communities and richness. In addition, seven acoustic indices were measured as part of the soundscape analysis approach at the low (50-1000 Hz), higher (1000-2000 Hz) and total (50-2000 Hz) frequency bands. Benthic biodiversity and environmental variables were extracted using QGIS from a variety of open-source platforms, such as EmodNET, the OneBenthic Portal and the essential fish habitat report.

A total of 16 different types of sounds were identified. Redundancy analysis (RDA) showed that the suitability of the area for spawning for a larger number of fish, the current velocity and the modelled benthic richness are the most important drivers of differences in the acoustic community. Kruskall Wallis test showed that the phonic richness of hard substrate habitat types is higher, compared to sandy habitat types. Pearson’s correlation tests were conducted to establish if any of the seven acoustic indices could be used as a proxy for benthic richness, habitat type (i.e. hard or sandy) or phonic richness. Three indices (ACI, AEI and M) were correlated with benthic richness, and five (BI, H, TE, SE, M) with habitat type and phonic richness. Greater benthic richness was associated with an increased acoustic complexity index, reduced unevenness across frequency bands, and quieter recordings. Harder substrates were associated with more uniform signals overall, but not with increased evenness across time bands within recordings. They were also associated with louder recordings but exhibited lower variability across frequency bands. Given that phonic richness correlates with harder substrates, the acoustic indices showed a similar pattern.

Our results indicate that spatial differences in the acoustic communities associated with different habitat types and levels of benthic richness are present, suggesting that PAM data collected for marine mammal monitoring could be ‘unlocked’ to reveal long-term temporal and spatial changes in the biodiversity associated with benthic habitats and serve as a cost-effective tool to understand future impacts of human and climate-driven changes. Although our results indicated that acoustic indices are a promising tool, more data need to be analysed before they could reliably be used as a proxy for phonic or benthic richness or habitat type. If the goal is to analyse variations in species-specific vocalisations or the overall acoustic community, as well as to determine which acoustic properties are relevant to management (such as richness or abundance), then manual annotations or the development of automated sound detectors are essential.