Using transfer learning with a convolutional neural network to detect African manatee (Trichechus senegalensis) vocalizations.

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.

Estimating Florida manatee (Trichechus manatus latirostris) abundance using passive acoustic methods.

Manatees are difficult to detect, particularly cryptic populations that inhabit areas with limited water clarity. The effectiveness of using vocal detections to estimate manatee abundance was evaluated in a clear water spring where manatees congregate seasonally. Vocalizations were extracted by a detection classifier that clustered sounds with similar spectral properties. Vocalization counts from recordings in Blue Spring, FL, USA were strong predictors of manatee abundance. The link between independent visual counts and abundance estimates from passive acoustic monitoring was used to provide an estimate of 1.059 (95% confidence interval 0.963-1.127) vocalizations/manatee/5-min, which might be used elsewhere for cue counting of manatees.

The influence of variations in background noise on Florida manatee (Trichechus manatus latirostris) detection of boat noise and vocalizations.

A manatee's primary modality to detect a vessel on a possible collision course is hearing as underwater visibility is limited in many manatee habitats and their visual acuity is poor. We estimate a Florida manatee's ability to detect the sound of an approaching boat and vocalizations in four different soundscapes in Sarasota Bay, FL. Background noise samples were collected every 5 minutes for a two-week period during winter and summer at each location (2019 or 2020). Sound levels in third octave bands (0.5, 1, 2, 4, and 8 kHz) were measured and compared to manatee auditory hearing thresholds and to sound levels of an approaching boat traveling at a slow, medium, or fast speed. Background sound levels in a wider band (1-20 kHz) were calculated to model vocal communication space at each location. We found that a manatee's estimated ability to detect an approaching boat differs greatly among locations, with time of day, and by season, and that fast boats are predicted to be detected later than slow boats. Latency of boat noise detection is estimated to sharply increase when considering unusually loud background noise levels. We suggest that such uncommonly loud conditions (e.g. 95th percentile sound level), not just typical conditions (median sound level), are important to consider for understanding the problem of manatee-boat collisions. Additionally, background noise impacts estimated vocal communication space and may limit the ability of vocal-mediated mother-calf cohesion. Altogether, a manatee's ability to detect acoustic signals of interest is expected to vary greatly spatially and temporally.

First characterization of vocalizations and passive acoustic monitoring of the vulnerable African manatee (Trichechus senegalensis).

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.

Passive acoustic listening stations (PALS) show rapid onset of ecological effects of harmful algal blooms in real time.

Monitoring ecological changes in marine ecosystems is expensive and time-consuming. Passive acoustic methods provide continuous monitoring of soniferous species, are relatively inexpensive, and can be integrated into a larger network to provide enhanced spatial and temporal coverage of ecological events. We demonstrate how these methods can be used to detect changes in fish populations in response to a Karenia brevis red tide harmful algal bloom by examining sound spectrum levels recorded by two land-based passive acoustic listening stations (PALS) deployed in Sarasota Bay, Florida, before and during a red tide event. Significant and temporally persistent decreases in sound spectrum levels were recorded in real time at both PALS in four frequency bands spanning 0.172-20 kHz after K. brevis cells were opportunistically sampled near the stations. The decrease in sound spectrum levels and increase in K. brevis cell concentrations also coincided with decreased catch per unit effort (CPUE) and species density per unit effort (SDPUE) data for non-clupeid fish and soniferous fish species, as well as increased reports of marine mammal mortalities in the region. These findings demonstrate how PALS can detect and report in real time ecological changes from episodic disturbances, such as harmful algal blooms.