Classification of Florida manatee (Trichechus manatus latirostris) vocalizations

Abstract

The vocal repertoire for the Florida manatee is quantitatively categorized from a sample of 1114 calls recorded from 3 different manatee habitats in Florida. First, manatee vocalizations were categorized into five call categories based on visual inspection of spectrograms and following descriptions provided in previous studies. Second, based on measurements of 17 acoustic parameters, the subjective classification scheme was validated using classification and regression trees (CARTs) and model-based cluster analysis paired with silhouette coefficients. CART analysis revealed that these five broad call categories can be successfully distinguished based on correct classification scores of 41.6%–62.5%. Silhouette coefficients determined that the manatee vocal repertoire is highly intergraded. This study supports and expands upon existing subjective categorization schemes by providing a quantifiable methodology for describing the Florida manatees' vocal repertoire. These findings contribute to the increasing number of studies suggesting many animal species vocal repertoires contain graded call types.

Bioacoustic Classification of Antillean Manatee Vocalization Spectrograms Using Deep Convolutional Neural Networks

“…This work complements and improves previous works in the classification of manatee vocalizations in the context of manatee detection and and individual identification schemes, as presented in [5]. Moreover, together with a recently published article by Brady et al [50], in which Floridian manatee vocalization are studied using classification and regression tree (CARTs) and cluster analysis, are few of the examples of using machine learning and deep learning applied to the classification of manatee vocalizations.…”

Humpback whale call repertoire on a northeastern Newfoundland foraging ground

“…This phenomenon has been described in a variety of other species including koalas ( Phascolarctos cinereus ; Charlton, 2015), Risso's dolphins ( Grampus griseus ; Corkeron & Van Parijs, 2001), killer whales ( Orcinus orca ; Ford, 1989), wild boars ( Sus scrofa ; Garcia et al, 2016), Old World monkeys (superfamily Cercopithecoidea; Gautier & Gautier, 1977), and great apes (family Hominidae; Marler & Tenaza, 1977). This mixed repertoire suggests that some, or all, calls in each of the broad classes may be better classified in a continuum‐based manner, rather than into discrete groups, as done, or suggested, for other species (e.g., manatees, Trichechus manatus latirostris ; Brady et al, 2020; southern right whales, Eubalaena australis ; Clark, 1982).…”

“…Variation in the forms within a call type provides support for gradation in the repertoire (Brady et al, 2020) by…”

“…Variation in the forms within a call type provides support for gradation in the repertoire (Brady et al, 2020) by showing that the boundaries between stereotyped forms of call types are blurred by transitional forms. The presence of transitional forms resulted in difficulties in clearly differentiating the boundaries among the call types (Figures 3 and 5) during AV classification and likely contributed to misclassifications in the CART and RF.…”

Contribution to unravel variability in bowhead whale songs and better understand its ecological significance

“…Units were positioned along a continuum on the MDS axes, following a gradation in the frequency, modulation and duration parameters of this unit type, which complicated the classification into subtypes. Graded vocal repertoire has been described for many mammalian taxa such as primates 53 , elephants 54 , sirenians 55 , odontocetes [56][57][58] , and mysticetes including humpback 59 and right whales 60,61 . Graded repertoires are challenging for classification tasks since there is no clear boundary where a grouping begins or ends.…”