ABSTRACT The Shisiwani National Park (Anjouan, Comoros) is a cetacean biodiversity hotspot. However, the vocal behavior of marine mammals inhabiting the area has never been studied. In 2023, we conducted a 400‐nautical‐mile survey from a 5‐m motorboat to determine their presence. Three dolphin species were encountered: the pantropical spotted dolphin ( Stenella attenuata ), the spinner dolphin ( Stenella longirostris ), and the melon‐headed whale ( Peponocephala electra ). Dolphin vocalizations were recorded using a single hydrophone during each sighting. The visual inspection of spectrograms allowed the identification of 1495 whistles with a good signal‐to‐noise ratio. We extracted the frequency contour of each whistle and measured several spectrotemporal features employing the ROCCA module of PAMGUARD. Using a Random Forest algorithm, we classified whistles according to species, showing that while signals from non‐congeneric species are distinguishable acoustically, those from congeneric species exhibit more subtle differences and are less so. Moreover, within the whistle repertoire of each species, we identified stereotyped frequency contours matching the SIGID criteria, possibly indicating the existence of signature whistles. Our study provides the first characterization of the whistles of three dolphin species in the Comoros archipelago and paves the way for developing the first fine‐tuned tool for their Passive Acoustic Monitoring in these remote areas.
Ahmed et al. (2025) studied this question.