Audial, visual and statistical discrimination of similar calls made by sei and blue whales

Abstract

In the Gulf of St. Lawrence (GoSL), current near real-time acoustic detection methods for baleen whale calls are misclassifying blue whale (Balaenoptera musculus) acoustic presence as sei whale (Balaenoptera borealis), likely due to their similar calls and the lack of a blue whale call library in the low-frequency detection and classification system (LFDCS) used to inform near real-time analysis protocols. This study aimed to quantify misclassifications using manual analysis of archival data and to determine how LFDCS classifications change with a blue whale call library incorporated. Data were collected from 8 glider deployments in the GoSL from 2020-2023. Manual analysis revealed sei whale acoustic presence was correctly identified by near real-time detection methods 1-3% of the time. Furthermore, analyses with the expanded call library resulted in LFDCS correctly classifying 31% of manually validated blue whale periods. Updating the near real-time analysis protocols will be critical for improving acoustic detections.