Frontiers in Marine Science
Frontiers Media S.A.
Centre for Ecosystems Research
Wildlife Conservation Society, WCS
Holsworth Wildlife Research Endowment
University of St Andrews
University of Queensland, UQ
Australian Geographic Society
Ecological Society of Australia, ESA
This study presents non-song vocalizations of humpback whales (Megaptera novaeangliae) from two migratory areas off the Western Australian coast: Geographe Bay and Port Hedland. A total of 220 sounds were identified as non-song sounds in 193 h of recordings reviewed. Of those, 68 were measured and qualitatively classified into 17 groups using their spectral features. One group (HW-02) had a high level of variation in terms of spectral slope. However, further classification using statistical classification methods was not possible because of the small sample size. Non-song sound frequencies varied from 9 Hz to 6 kHz, with the majority of sounds under 200 Hz. The duration of non-song sounds varied between 0.09 and 3.59 s. Overall, the use of spectral features allowed general classification of humpback whale sounds in a low sample size scenario that was not conducive to using quantitative methods. However, for highly variable groups, quantitative statistical classification methods (e.g., random forests) are needed to improve classification accuracy. The identification and accurate classification of a species’ acoustic repertoire is key to effectively monitor population status using acoustic techniques and to better understand the vocal behavior of the species. The results of this study improve the monitoring of humpback whales by standardizing the classification of sounds and including them in the species’ repertoire. The inclusion of non-song sounds in passive acoustic monitoring of humpback whales will add females and calves to the detection counts of otherwise only singing males. © Copyright © 2020 Recalde-Salas, Erbe, Salgado Kent and Parsons.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.