Summary
A study of wild African savannah elephants found that overlapping rumble choruses carry acoustic cues associated with behavioral context. Rumbles also became more acoustically similar during some multi-caller exchanges.
Wild African savannah elephants’ overlapping rumble calls carry acoustic information associated with what the animals are doing, according to a study published in Scientific Reports. The researchers analysed 902 recordings from adult females in Kenya’s Amboseli National Park and found that choruses could be classified by behavioral context better than chance. In some types of vocal exchange, successive rumbles also became more acoustically similar.
The findings suggest that a chorus is more than a collection of calls happening at once: its individual rumbles retain context-related features, while the changing similarity between calls may add another layer to the group signal.
How the researchers analysed elephant choruses
Elephants use low-frequency rumbles in several ways, including as separate calls, alternating exchanges and choruses in which multiple calls overlap. Because overlapping sounds are difficult to separate and analyse, choruses have received less attention than isolated calls.
The team examined recordings collected in Amboseli between 1986 and 2022. The final dataset contained 319 single rumbles and 285 chorused rumbles from one family group, plus 150 single and 148 chorused rumbles from elephants in other families. All were produced by adult females. Researchers assigned each call to one of six behavioral context types: Contact, Let’s-Go, Cadenced, Coo, Little-Greeting or Greeting.
They extracted acoustic features from each recording and used statistical classifiers to predict its context. The classifiers performed better than chance for both single calls and choruses. Models trained on single rumbles could also classify choruses by context better than chance, indicating that the two call types share some context-related acoustic characteristics.
The predictions were imperfect, and performance fell when models were tested on elephants or family groups not represented in training. That pattern is consistent with calls carrying both context-related features and variation associated with individual callers or groups.
Convergence appeared in some exchanges
To test for acoustic convergence, the researchers examined sequences of rumbles in 92 multi-caller exchanges. They measured acoustic distances between successive calls: a decrease over an exchange indicated that calls were becoming more similar in the features analysed. Permutation tests compared the observed pattern with sequences in which call order or the calls assigned to exchanges had been shuffled.
Contact and Greeting exchanges showed statistically significant convergence in both tests. Cadenced exchanges showed convergence in the order-shuffled test, but the exchange-shuffled test gave p < 0.065, above the conventional 0.05 threshold. The other three context types—Coo, Let’s-Go and Little-Greeting—did not show statistically significant convergence in these analyses.
These results establish a context-dependent pattern of acoustic similarity, not the reason it occurs. The authors suggest that convergence in Contact and Cadenced exchanges could relate to coordination, while convergence in Greeting exchanges could be associated with social bonding. Those explanations remain interpretations of the observed patterns rather than functions tested directly in this study.
What the findings add
The study extends research on elephant communication from individual calls to overlapping group vocalizations. Its results indicate that context-related acoustic cues can remain detectable even when several elephants rumble at once. This could help future studies make use of choruses that are common in natural recordings but often difficult to analyse as separate calls.
The researchers note that the dataset was uneven across individuals, contexts and family groups, and that behavioral context was assigned by experienced observers in the field. Playback experiments—presenting elephants with controlled examples and measuring their responses—could help test how the animals interpret convergent exchanges. The evidence here comes from acoustic analysis of wild elephants’ recorded calls; it does not establish whether callers intentionally copy one another or what social effect convergence has.