A new study has found, with the help of artificial intelligence, that lions have a less powerful type of roar called an “intermediate roar.” The discovery could be useful in conservation efforts for the species.
New research in the journal Ecology and Evolution revealed a kind of lion’s roar hitherto unknown, which calls into question old assumptions about the way the king of the jungle communicates.
The discovery, which was made possible by artificial intelligence, shows that the Lions’ vocal behavior is more complex than previously believed and could transform the methods used to monitor population decline.
For decades, scientists thought that lions produced only one type of roar: the iconic powerful roar that resonates across the African savannas, used mainly for long-distance communication and territorial defense. Lions too produce a variety of shorter soundsincluding grunts, growls and meows, but the powerful roar has always been considered its characteristic vocalization.
A team led by Jonathan Growcott, from the University of Exeter, has now identified a second type of roar, dubbed “intermediate roar“. Unlike the powerful, dramatic, pitch-variant roar, this newly discovered vocalization has a flatter tone and contains fewer variations. “The powerful roar is an explosion of sound,” Growcott explained. “The intermediate roar differs in that it is a sound flatter and with less variation.”
According to the study, lions actually produce four distinct types of vocalizations within a single sequence of roars: the groan, the full roar, the recently described intermediate roar, and the grunt. These subtle distinctions, once overlooked, have significant potential to improve the monitoring of lion populations, explains the .
The researchers installed 50 passive acoustic recording units in Nyerere National Park, Tanzania, for 62 days, capturing a wide variety of lion vocalizations in the wild. To complement this, eight lions in Zimbabwe (five males and three females) were equipped with bio-loggers that recorded individual vocalizations.
Traditionally, the recognition and classification of lion roars depended largely on the interpretation of experts, being subject to human bias. Growcott’s team used machine learning algorithms to analyze thousands of vocalization recordings, enabling a more consistent and objective classification. “Our new approach, which uses AI, promises a more accurate and less subjective monitoring“, he noted. “This is crucial for conservationists working to protect lion populations, which are dwindling.”
Growcott argues that the future of wildlife tracking lies in passive acoustic monitoring combined with advanced bioacoustic analysis. “We believe a paradigm shift is needed in wildlife monitoring,” he said. “As bioacoustics improves, will become vital for the conservation effective against lions and other threatened species.”
