Particle.news
Download on the App Store

AI Uncovers a Second Lion Roar, Opening New Paths for Conservation

The peer-reviewed analysis reports machine learning accuracy approaching 95% in classifying roars across two lion populations.

Overview

  • The study, published in Ecology and Evolution, identifies an acoustically distinct 'intermediary roar' alongside the classic full-throated roar.
  • Researchers analyzed more than 3,000 calls drawn from tens of thousands of hours of audio in Tanzania and collar recordings from Zimbabwe.
  • The intermediary roar is shorter, flatter and lower in maximum frequency than the full-throated roar within a roaring bout.
  • Models classified call types with reported accuracy up to 95.4% and identified individual lions, outperforming human experts in tests.
  • Scientists say passive acoustic monitoring could help estimate populations and track individuals, though the new roar’s function remains unclear and further automated, context-rich validation is needed.