RESUMEN
African manatees (Trichechus senegalensis) are vulnerable, understudied, and difficult to detect. Areas where African manatees are found were acoustically sampled and deep learning techniques were used to develop the first African manatee vocalization detector. A transfer learning approach was used to develop a convolutional neural network (CNN) using a pretrained CNN (GoogLeNet). The network was highly successful, even when applied to recordings collected from a different location. Vocal detections were more common at night and tended to occur within less than 2 min of one another.
Asunto(s)
Trichechus manatus , Trichechus , Animales , Redes Neurales de la Computación , Aprendizaje AutomáticoRESUMEN
Even among the understudied sirenians, African manatees (Trichechus senegalensis) are a poorly understood, elusive, and vulnerable species that is difficult to detect. We used passive acoustic monitoring in the first effort to acoustically detect African manatees and provide the first characterization of their vocalizations. Within two 3-day periods at Lake Ossa, Cameroon, at least 3367 individual African manatee vocalizations were detected such that most vocalizations were detected in the middle of the night and at dusk. Call characteristics such as fundamental frequency, duration, harmonics, subharmonics, and emphasized band were characterized for 289 high-quality tonal vocalizations with a minimum signal-to-noise ratio of 4.5 dB. African manatee vocalizations have a fundamental frequency of 4.65 ± 0.700 kHz (mean ± SD), duration of 0.181 ± 0.069 s, 97% contained harmonics, 21% contained subharmonics, and 27% had an emphasized band other than the fundamental frequency. Altogether, the structure of African manatee vocalizations is similar to other manatee species. We suggest utilizing passive acoustic monitoring to fill in the gaps in understanding the distribution and biology of African manatees.
