Researchers discovered an "intermediate" lion roar — shorter, lower in pitch and consistently following the full-throated roar — after analysing 3,149 calls recorded with 50 microphones and collar sensors. An AI model classified vocal types with over 95% accuracy and revealed this previously unrecognized category. Recognizing the intermediate roar should help researchers isolate true full-throated roars and improve acoustic population estimates for the roughly 23,000 African lions remaining in the wild.
Scientists Identify a New "Intermediate" Lion Roar — Shorter, Lower and Always Follows the Classic Roar
Researchers discovered an "intermediate" lion roar — shorter, lower in pitch and consistently following the full-throated roar — after analysing 3,149 calls recorded with 50 microphones and collar sensors. An AI model classified vocal types with over 95% accuracy and revealed this previously unrecognized category. Recognizing the intermediate roar should help researchers isolate true full-throated roars and improve acoustic population estimates for the roughly 23,000 African lions remaining in the wild.
04:00 PM, 11/21/2025Science
Researchers have identified a previously unrecognized lion vocalization they call the "intermediate roar": a shorter, lower-pitched call that consistently follows the classic full-throated roar. The finding reveals greater complexity in lion vocal behaviour and could improve acoustic methods for estimating population sizes.
How the discovery was made
The study, published Nov. 21 in the journal Ecology and Evolution, combined passive acoustic monitoring and animal-borne sensors. Teams deployed 50 custom microphones across Tanzania's Nyerere National Park and fitted acoustic collars to five lions in Zimbabwe's Bubye Valley Conservancy. In total the researchers collected 3,149 African lion (Panthera leo) vocalizations.
AI reveals a new roar type
Using an artificial intelligence model to analyse the recordings, the research team identified a distinct vocal category they call the intermediate roar. The AI classified lion vocalization types with over 95% accuracy, substantially reducing subjective bias introduced by manual labelling and making it easier to isolate and compare roar types.
"The full-throated roar is an explosion of sound — loud, complex and arcing in pitch. The intermediary roar differs by being a flatter sound with less variation," said Jonathan Growcott, the study's lead author and a doctoral student in mathematics and statistics at the University of Exeter.
Why it matters
Previous research showed that full-throated roars contain individual signatures reflecting a lion's sex, age and other traits, which makes acoustic monitoring a promising tool for population surveys. Recognizing the intermediate roar as a separate category should help researchers more consistently identify true full-throated roars and extract more reliable identity cues from recordings.
There are an estimated 23,000 African lions remaining in the wild, and traditional population-estimation methods such as camera traps and spoor surveys can be resource intensive and sometimes less accurate. Improved acoustic classification and data-driven identification of roar types could produce more accurate population density estimates that better inform conservation strategies.
Next steps
The authors recommend applying the AI-based classification method to additional populations and environments to verify how widespread the intermediate roar is and to refine acoustic monitoring techniques for conservation use.