D-PorCCA, a new tool to study the behaviour of harbour porpoises

Cosentino, Mel and Schwarzbach, Patrick and Tougaard, Jakob and Nairn, David and Guarato, Francesco and Jackson, Joseph and Windmill, James F.C. (2019) D-PorCCA, a new tool to study the behaviour of harbour porpoises. In: World Marine Mammal Conference, 2019-12-07 - 2019-12-12.

[img]
Preview
Text (Cosentino-etal-WMMC2019-D-PorCCA-a-new-tool-to-study-the-behaviour-of-harbour-porpoises)
Cosentino_etal_WMMC2019_D_PorCCA_a_new_tool_to_study_the_behaviour_of_harbour_porpoises.pdf
Final Published Version

Download (2MB)| Preview

    Abstract

    The behaviour of harbour porpoises can be deduced from the variation pattern of their vocalisations. They produce only narrow-band high-frequency (NBHF) clicks, emitted in so-called click trains. Therefore, available acoustic recordings can be used to increase our understanding of their behaviour. Behavioural studies, however, are time consuming and require an accurate, automated classifier and a click train identifier. Here, we present D-PorCCA, a graphical user interface to study recordings from harbour porpoises in the wild. D-PorCCA is an independent tool developed in Matlab and includes an impulsive-sound detector and a new high-accuracy porpoise click classifier (PorCC). PorCC classifies each signal as either: noise (N), low-quality (LQ), or high-quality (HQ) porpoise click. The user can decide on the echolocation events they want to visualise, specifying the length of the click train (including both HQ and LQ clicks) and the minimum separation time between echolocation events. These events are plotted as time vs amplitude, vs inter-click interval, and vs centroid frequency (or direction of arrival, if available). Additionally, the waveform, power spectrum, and spectrogram of each click within the echolocation event is available. Click trains of interest can be easily selected by the user and extracted for further analysis. These events are then automatically cleaned of echoes and other noise sources, after which the pattern of the click train is automatically investigated to determine whether there is one or more animals vocalising simultaneously (i.e., overlapping click trains), as well as which behaviour the animal was potentially engaged in (e.g., feeding). D-PorCCA is a user-friendly tool with potential for behavioural studies of wild harbour porpoises as well as other NBHF species, as it focuses on clicks trains, and can be used to fill knowledge gaps of these elusive species. Moreover, it has potential for application in large monitoring project, such as SAMBAH.