Investigating whistle characteristics of three overlapping populations of false killer whales (Pseudorca crassidens) in the Hawaiian Islands

Abstract

Three genetically distinct populations of false killer whales Pseudorca crassidens) reside in the Hawaiian Archipelago: two insular populations (one within the main Hawaiian Islands [MHI] and the other within the Northwestern Hawaiian Islands [NWHI]), and a wide-ranging pelagic population with a distribution overlapping the two insular populations. The mechanisms that created and maintain the separation among these populations are unknown. To investigate the distinctiveness of whistles produced by each population, we adapted the Real-time Odontocete Call Classification Algorithm (ROCCA) whistle classifier to classify false killer whale whistles to population based on 54 whistle measurements. 911 total whistles from the three populations were included in the analysis. Results show that the MHI population is vocally distinct, with up to 80% of individual whistles correctly classified. The NWHI and pelagic populations achieved between 48 and 52% correct classification for individual whistles. We evaluated the sensitivity of the classifier to the input whistle measurements to determine which variables are driving the classification results. Understanding how these three populations differ acoustically may improve the efficacy of the classifier and create new acoustic monitoring approaches for a difficult-to-study species.