Bats are hard to count. They fly at night, move fast and call mostly above the range of human hearing. So between 2021 and 2024, Bat Conservation Trust and Forest Research set acoustic detectors on poles in 310 woodlands across England and let them listen for 9,734 nights. That came to about 90,000 hours of audio, more than ten years of nonstop listening, and 217 terabytes of data.
The clever bit
Researchers at UCL and the University of Edinburgh worked with the survey team to build and run AI that finds bat echolocation calls in the recordings and flags which species made them. To keep it honest, 19 bat-call experts from the trust’s Verifier Network, each of whom passed a standardised identification assessment, checked a subset of the AI’s identifications.
The AI picked out more than 98 million bat calls. Common pipistrelles turned up in all 310 woods. The survey detected every British bat species or species group except the serotine, which is hard to identify by sound, including two UK Red List species, the grey long-eared bat and the barbastelle. Barbastelles also showed up in parts of Cornwall, Devon and Dorset where only a few sparse records existed.
AI has fundamentally changed the scale at which we can monitor wildlife.
— Professor Kate Jones, director of the UCL People and Nature Lab, via forestresearch.gov.uk
Hooray
Three rare bats have a national watch. For the barbastelle, Leisler’s bat and Nathusius’ pipistrelle, this survey is England’s only active national-scale monitoring of any kind, and anyone can explore the data online.
What comes next
Baseline recording continues until 2027. The team plans to model how climate, woodland cover, hedgerows and habitat connectivity shape where bats live, so the results can guide how woodlands are created, restored and managed.