Lion 'accents' reveal a surprisingly complex language

Lions do not just roar – they use two kinds of roars that serve different roles during a bout. A recent study confirms the split and shows that some populations sound measurably different.

Researchers from the University of Exeter and the University of Oxford used field recorders and collar audio to monitor lions in Tanzania and Zimbabwe. 

One number stands out: algorithms separated lion call types with more than 95 percent accuracy.

Two types of lion roars

The work was led by Jonathan Growcott, a doctoral researcher at the University of Exeter. His research focuses on practical bioacoustics and machine learning for wildlife monitoring.

Using field recorders in Nyerere National Park and collar audio from Bubye Valley Conservancy, the study reports two distinct call types inside a roaring bout. 

“Two types of roars exist within a lion’s roaring bout,” wrote Growcott. The first are full-throated roars, the loud, identity rich calls that anchor a bout. The second are intermediary roars, shorter and lower calls that follow later in the sequence.

Earlier research showed that the full-throated roar carries a unique acoustic signature that can identify individuals with 91.5 percent accuracy. 

“The existence of individually unique f0 contours in lion roars indicates a likely mechanism enabling individual lions to identify conspecifics over long distances,” wrote Matthew Wijers, an expert at the University of Oxford.

Making acoustic surveys more accessible

The group trained a Hidden Markov Model, a statistical model for time ordered patterns, to follow the shape of each call as it unfolds over time. That method classified call types with 84.7 percent accuracy from the sound’s lowest pitch contour.

They also tested K-means clustering, a simple algorithm that groups data by similarity. Using only call length and maximum pitch, the approach reached 95.4 percent accuracy when moans were excluded.

Because only full-throated roars carry identity, the team asked whether automatic selection of those calls improves matching. Using the model’s picks raised the F1-score, a balanced measure of precision and recall, from 0.80 to 0.87.

Those numbers matter for field teams that cannot always sit behind cameras or safely approach a pride. They also make acoustic surveys more accessible where computing resources are limited.

Lions develop regional accents

The new work also compared calls from two regions and found clear differences in maximum pitch and duration. 

One Zimbabwe male, known to have come from Botswana’s Tuli Block, stood out in earlier data and skewed automatic classification.

“Lion A4, although resident in the study area, was known to have originated from the Tuli Block in eastern Botswana from where it dispersed more than 60 km before breaking into the fenced Bubye Valley Conservancy,” wrote Wijers. The distance is about 37 miles.

Older field work in Namibia’s Etosha Park reported comparatively shorter roars, a hint that regional style is not new. Differences might be shaped by vegetation, terrain, or typical group spacing.

How lions learn calls

A well-known acoustic adaptation hypothesis proposes that sound structure drifts to fit local habitats. 

Thicker cover can favor lower frequencies that travel farther, while open country can reward other profiles.

Genetics and social learning could both play a part. Lions learn from pride mates, so roaring habits may pass along within family lines.

What this means for counting lions

This work strengthens passive acoustic monitoring, listening with recorders to detect animals over wide areas. 

In places where camera traps struggle, identity rich roars could support capture-recapture style population estimates.

Caution is needed when surveys span multiple regions. If some nomads sound atypical for a site, automatic systems might undercount them unless models are tuned.

When lions roar most

A follow-up analysis found lions roar most just before dawn and often near water. Concentrating effort around those windows can raise detection rates without adding gear.

Dispersal patterns add another layer to the story. Male lions can disperse more than 120 miles, so migrants may carry vocal habits formed elsewhere.

Using region-aware training data and simple features like call duration and maximum pitch helps keep models reliable. That blend of practicality and accuracy is exactly what field teams need.

Future research on lion accents

The next step is to gather recordings across the span from East to southern Africa to see whether regional differences shift gradually or sharply.

Another challenge is to uncover whether intermediary roars communicate something specific.

If they do, conservationists gain more than a monitoring tool – they gain a new behavioral signal from one of the world’s most iconic predators.

The study is published in the journal Ecology and Evolution.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–