Nature
PUBLISHED
How do you tell large dangerous animals apart while staying safe? Well, one way for lions has been to identify them through their roar. For a long time it was believed that lions only had one kind of roar, a full-throated noise that was a pretty iconic part of the African soundscape. Now, thanks to AI technology, researchers have identified a new kind of roar that could even be used to aid conservation efforts.
Lions have a whole heap of sounds for different occasions including the famous full-bodied roar, chuffs, mews, and snarls. The roar is mostly used for long-distance communication to maintain contact with other members of the pride or to let other prides know about their territory.
"The full-throated roar is an explosion of sound. Loud and complex and arcs in pitch," lead author Jonathan Growcott from the University of Exeter told Live Science. "The intermediary roar differs by being a flatter sound with less variation."
These roars can even be used as a way of estimating lion population sizes and identifying individual lions. The team suggest in their paper “that within a lion's roaring bout there are four distinct call types: moan, full-throated roar, intermediary roar (which has not previously been classified), and grunt”.
“Lion roars are not just iconic – they are unique signatures that can be used to estimate population sizes and monitor individual animals. Until now, identifying these roars relied heavily on expert judgment, introducing potential human bias,” said Growcott in a statement. “Our new approach using AI promises more accurate and less subjective monitoring, which is crucial for conservationists working to protect dwindling lion populations.”
To look at this in more detail, research took place in Nyerere National Park in southern Tanzania. By setting up 50 recording units for 62 days, the team were able to record all sorts of lion vocalizations. Biologgers were also fitted to five male and three female lions in Zimbabwe to collect vocalization recordings from individual lions.
Analysis of the data showed that lions produce four different call types within a roaring bout. This includes two types of roars, the full-throated roar and the intermediary roar that was not previously known about. This suggests that lion roars are more complex than previously known, which is similar to recent discoveries about calls in hyenas.
By using machine learning to classify the lion vocalization data, the team suggest that this approach could be used by non-experts to examine different datasets, or even that the framework could be applied to other species' vocalizations.
“We believe there needs to be a paradigm shift in wildlife monitoring and a large-scale change to using passive acoustic techniques. As bioacoustics improve, they’ll be vital for the effective conservation of lions and other threatened species,” concluded Growcott.
The study is published in Ecology and Evolution.
1
