Validating automated click detector dolphin detection rates and investigating factors affecting performance.

Abstract

Passive acoustic monitoring (PAM) is a widely used technique for studying the distribution and habitat use of cetaceans. The C-POD, an acoustic sensor with an onboard automated click detector, has been deployed in diverse acoustic environments, but studies verifying its offshore detection rates and factors affecting detection probability are scarce. To empirically evaluate the performance of C-PODs in detecting bottlenose dolphins (Tursiops truncatus), C-PODs were deployed alongside archival acoustic recorders 12-30 km offshore in the Northwest Atlantic Ocean. The C-POD and acoustic recordings, post-processed using PAMGUARD software, were compared for a period of 6852 h. C-POD false positive rates were very low (mean 0.003%), and positive hourly detection accuracy was very high (mean 99.6%). Analysis of the acoustic environment and dolphin click characteristics revealed that true positive detections by C-PODs were significantly more likely to occur when PAMGUARD detected more clicks and there was increased high frequency noise (>20 kHz), likely from distant or unclassified clicks. C-PODs were found to be reliable indicators of dolphin presence at hourly or greater time scales. These results support the application of C-PODs in PAM studies that aim to investigate patterns of dolphin occurrence, such as those related to offshore windfarms.