Particle.news
Download on the App Store

Scientists Identify Second Lion Roar, Opening Door to AI-Driven Conservation

Machine learning trained on thousands of recordings separated a newly defined intermediary call, reliably identifying individual lions.

Overview

  • A peer-reviewed study in Ecology and Evolution confirms an acoustically distinct intermediary roar alongside the classic full-throated roar in African lions.
  • Algorithms trained on recordings from Tanzania’s Nyerere National Park and tested on lions in Zimbabwe classified call types with up to 95.4% accuracy.
  • By isolating full-throated roars, the system identified which individual lion was vocalizing and, in at least one population, outperformed human experts.
  • The intermediary roar is shorter, lower-pitched and acoustically flatter than the full-throated call, and it is reported to follow within a roaring bout.
  • Researchers say the method could scale passive acoustic monitoring for conservation, though the function of the intermediary roar remains unknown and further behavior-linked data and broader validation are needed.