Ecoacoustics for assessing ecosystem health and function, from air to soil

Sound is an intrinsic component of an ecosystem, encoding a wealth of information about species presence and behaviours, human activities, and general ecological health. The soundscape of a thriving forest might be a cacophony of bird song, insect chirps, buzzes, howls, squawks, and trills. Below ground, scratches, scrapes, and clicks fill recordings in healthy soil. Lower yourself underwater and you’ll hear a surprising variety of oinks, grunts, boings, crackles, and plops. Anthropogenic activities change the natural soundscape, drowning out wildlife sounds, or causing an unsettling silence over the landscape.

Ecological monitoring using passive acoustic sensors is now a common approach for understanding impacts of environmental change on biodiversity and habitat health, generating bioacoustic data on animal species occurrence, behaviour, and vocal activity, or ecoacoustic data on whole ecosystem sound (the soundscape). Ecoacoustic approaches seek to compare the levels of biotic and anthropogenic sounds and correlate these with biodiversity metrics. We seek to go beyond these correlative methods, towards a processed based approach linking the soundscape to ecosystem function and ecological energy flows. A core aim is to ensure our approach is transferrable across ecosystems and regions, globally, and this hinges on employing cutting-edge machine learning techniques to classify sounds to species or functional groups, anthropogenic, and geophonic sound types.

We and our partners are collecting passive acoustic data in diverse habitats where different nature recovery measures are being tested, including grassland restoration in Oxfordshire, native forest restoration in Scotland, landscape recovery in Ghana, and Savannah recovery in Kenya. Collected alongside traditional and remote sensing ecological monitoring data, we’re able to validate our ecoacoustic ecological energetics estimates and machine learning approach. A challenge in the bioacoustic and ecoacoustic fields is robustly estimating density or abundance of species from their vocalisations or sounds, yet this is key to estimating ecological energy flows from passive acoustic data. We are developing methods to overcome this challenge using arrays of passive acoustic sensors deployed above ground monitoring bats, and below ground, monitoring earthworms.

Until recently, ecoacoustic studies have focussed on above ground or aquatic systems, yet soil health is crucial to nature recovery and can be where much of the ecological energy flows. We are exploring below ground passive acoustic monitoring in Oxfordshire, where most energy cascades through earthworms. However, the application of ecoacoustics to soil health is still in its infancy, largely constrained by limited knowledge on the sources of many below ground sounds. To address this, we are building open sound libraries of soil fauna, starting with earthworms, which will train machine learning models to classify and characterise the underground soundscape.