Back

Whistle variability and social acoustic interactions in bottlenose dolphins

Mustun, F.; Semenzin, C.; Rance, D.; Marachlian, E.; Guillerm, Z.-L.; Mancini, A.; Bouaziz, i.; Fleck, E.; Shashar, N.; de Polavieja, G. G.; Sumbre, G.

2024-10-16 animal behavior and cognition
10.1101/2024.10.15.618471 bioRxiv
Show abstract

Bottlenose dolphins exhibit a sophisticated social structure, known as a fission-fusion society. To sustain this complex system, dolphins rely on a rich vocal repertoire: clicks exclusively used for echolocation, burst-pulse sounds associated with emotions during social interactions, and whistles, including signature whistles that serve as individual-specific identifiers ( names). How dolphins maintain their complex social structure based only on a limited repertoire of sounds remains elusive. Previous studies suggest that contextual information can be transferred by the modulation of the whistles. Here, we investigated the whistle variability using a comprehensive computational approach, and studied the structure of the interactions between the whistle variants. Using a unique large dataset, acquired in a natural environment, we observed that signature whistles exhibit variations in their frequency contours. Unsupervised clustering revealed that signature whistles could be classified into sub-categories (signature whistle variants). The existence of these categories, and their independence on the emitter dolphin, indicate that these variations are not random. Analysis of pairwise interactions between sub-categories revealed a clustered structure similar to that of their social hierarchy. Network analysis of this structure showed that whistle sub-categories had different functional roles: some acted as hubs, others as bridges, and certain were used for turn-taking between the main whistle categories. We also found that the dolphins emit signature whistles of their deceased mothers, a phenomenon only observed in human language. Overall, these findings suggest that dolphins modulate signature whistles to transmit additional information beyond individual identity, and that they engage in "dialogue-like" acoustic interactions.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.