Identifying and predicting occurrence and abundance of a vocal animal species based on individually specific calls

Abstract

AbstractPassive acoustic monitoring (PAM) offers opportunities to collect data on the occurrence of vocal species for long periods of time, at multiple locations, and under a range of environmental conditions. Some species emit individually distinctive calls, including bottlenose dolphins (Tursiops truncatus) that produce signature whistles. Our study used PAM to determine the seasonal occurrence of bottlenose dolphins and utilized individually specific signature whistles to (1) track individuals spatially and temporally, (2) assess site fidelity off Maryland (MD), USA, (3) estimate the minimum abundance of dolphins in the study area, and (4) develop a dynamic habitat‐based relative abundance model applicable as a real‐time dolphin relative abundance prediction tool. Acoustic recorders were deployed at two sites offshore of Ocean City, MD, and at one site in the upper Chesapeake Bay, MD. Acoustic recordings from 2016 to 2018 were analyzed for signature whistles, and re‐occurrences of individual whistles were identified using a combination of machine learning and manual verification. A habitat‐based density model was created using the number of signature whistles combined with environmental conditions. A total of 1518 unique signature whistles were identified offshore of Maryland and in the upper Chesapeake Bay. There were 184 re‐occurrences of 142 whistles, with a mean of 135 d between re‐occurrences (range = 1–681 d). These repeated detections of the same individuals occurred most frequently at the site near Ocean City, MD, indicating the highest site fidelity. Re‐occurrences were recorded among all three sites, indicating movement of dolphins between the Chesapeake Bay and off the Atlantic coast of Maryland. The weekly number of individual dolphins detected off the Atlantic coast was significantly related to two environmental variables: sea surface temperature and chlorophyll a concentration. This habitat model could be used to predict relative dolphin abundance offshore of Maryland and inform management within the region, including in relation to offshore wind energy development and other stakeholders.