Quantitative Methods for Classifying and Characterizing Delphinid Whistles

Lead Research Organisation: University of St Andrews
Department Name: Biology


The production and perception of sound is common to all species of animal and is simultaneously a valuable resource for ecological investigation. Passive acoustic monitoring technology (PAM) has enabled researchers to collect greater volumes of ecological data across a wider range of habitats, both terrestrial and marine, than traditional survey methods. However, successful monitoring requires both efficient workflows for analysing acoustic recordings and a thorough knowledge of the vocal repertoires of any subject species. Without these, it becomes difficult to obtain a significant amount of useful information from recordings. Delphinids are highly cognitive mammals that are dependent on sound both for communication and echolocation, making them ideal subjects for passive acoustic monitoring. They are also one of the few known groups of taxa that exhibit vocal learning, that is, dynamic vocal repertoires learnt from peers, which develop with time. This study will use pre-existing acoustic data collected by governmental and non-governmental organisations in Scottish waters, together with newly collected data, to develop and compare methods in delphinid whistle classification. Identification of accurate classification algorithms will contribute to the general aim of improving efficient wildlife monitoring and provide a useful assessment of the important parameters that distinguish species' vocalizations. A second phase of this study will explore whistle variation within and between several North Atlantic and North Sea species, including understudied species such as Lagenorhynchus albirostris, Lagenorhynchus actus, and Grampus griseus. Broadening the state of knowledge on these species' vocal repertoires will facilitate monitoring and may contribute new insights to the overall understanding of marine mammal communication. The marine environment is an inherently difficult and costly site for research and has thus posed limits on the number of studies of wild delphinids. Our study brings the invaluable opportunity to study free ranging North Atlantic and North Sea delphinids, testing cutting-edge detection and classification workflows on vocalizations recorded from wild populations. There is already a large existing dataset of such vocalizations available to this project. This will help accelerate the first phase of the project (development and testing of classification algorithms) before it can later be tested on newly collected data in the second phase.


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
NE/S007342/1 01/10/2019 30/09/2027
2599280 Studentship NE/S007342/1 01/10/2021 31/03/2025 Tristan Kleyn