False killer whale acoustic detection function from towed hydrophone array localizations in the Hawaiian Islands

False killer whale (Pseudorca crassidens) groups organize into dispersed subgroups that can span dozens of miles and be visually cryptic near the surface. These subgroups are used as the unit of detection for collection and analysis of visual line-transect survey data. However, an appropriate acoustic unit of detection and corresponding detection function has not been established. In order to use false killer whale acoustic detections in abundance estimation, we applied inference from visually verified detections to validate acoustic-only detections. Specifically, we validated acoustic-only detections using a classification model with automated species-specific criteria. We analyzed towed hydrophone array acoustic recordings collected during the 2017 Hawaiian Islands Cetacean and Ecosystem Assessment Survey to develop an acoustic-only detection function. Time-difference-of-arrivals (TDOAs) for all false killer whale signals were automatically extracted and then aggregated into what we have defined as “trajectories” based on consistent TDOA trajectories using a multi-target Bayesian framework. We suggest these trajectories represent a minimum of one false killer whale in contrast to the subgroups detected visually. A perpendicular distance to the trackline was computed at the beam for all trajectories (n = 638) to create an acoustic-based detection function. Incorporation of additional acoustic detections of false killer whales may improve abundance estimates.