Introduction & Objectives

Previous efforts to document behavioral response probability among pilot whales (Globicephala spp.) in different studies and regions have revealed movement and acoustic responses that are complex and variable on fine spatial and temporal scales. While there has been limited investigation into broader scale spatial movement and avoidance patterns along with changes in diving behavior, several key data sets and new analytical tools developed to examine response in beaked whales now afford the opportunity to do so. These opportunities, including the availability of data on a Navy priority species from multiple operational use areas, warrant further research into the relationship between sonar exposure and behavioral response.

This project leverages two long-term Navy-funded research efforts at the Pacific Missile Range Facility off Kaua‘i in the Pacific and off Cape Hatteras, NC in the Atlantic. A substantial amount of telemetry tag data from both locations allows us to better understand how exposure to various sources of Navy sonar may influence the movements and diving behavior of short-finned pilot whales using movement modeling in conjunction with sound propagation models.

These analyses will achieve several goals: 1) test the application of methods newely developed for beaked whales to a species with different dive and movement behavior; 2) assess the potential for, and magnitude of, behavioral responses by pilot whales to simulated MFAS and various sources of operational MFAS, including surface ships, sonobuoys, and helicopter-dippin sources; and 3) to compare responses across populations from different ocean basins, with different residency patterns and different MFAS exposure histories (e.g., the island-associated population in Hawai’i with known social networks and histories of exposure compared to the open-ocean population in the Atlantic with much more likited past exposure). This will help to determine how sensitive pilot whales might be to MFAS, how their environment or exposure history might shape their response, and if the type of response observed is conserved across the species or is population-specific.

Technical Approach

This project seeks to utilize the newly developed statistical methods from the Atlantic behavioral response study (BRS) and apply them to satellite-tagged pilot whales from both the Atlantic and Pacific. In the Atlantic BRS, the research team has used a class of movement models known as continuous time, discrete space (ctds) models (Hooten and Wikle, 2010; Hooten et al., 2010; Hanks et al., 2015).  These models are used to examine two factors of movements: 1) the speed with which animals move through or reside in a given space, and 2) the choices animals make when moving.  To parameterize the movement response to sonar, the BRS team constructed rasters that represented the distance to the sound-source, and then estimated attractive/repulsive movements in relation to that source. By pairing the model output with estimated received level at the time of exposure, the Atlantic BRS team could then quantify how likely an animal is to move away from a source as a function of received level (Southall et al., In review). We will extend this approach using the ctds model from Johnson et al. (2021) to examine movement response to dynamic covariates. Specifically, we intend to build a dynamic raster cube that represents the ambient wind-driven noise and the sound propagation model for each exposure.

Tasks for this project include:

1) Model received levels of active sonobuoys and helicopter-dipping sonar for PMRF pilot whales, and finish   received level models for hull-mounted MFAS. 

2) Model pilot whale received levels for simulated and operational ship MFAS in Atlantic using methodology  developed for goose-beaked whales.

3) Conduct social network analysis on PMRF pilot whales to examine exposure histories and assess response at the group level.

4) Apply ctds models to all track data from both populations to assess horizontal movement.

5) Apply standard Hidden Markov models to the dive summary data to examine if a received level covariate impacts either state transition probabilities, or state-dependent distributions.

6) Using the results of the ctds models, assess behavioral response at the level of the individual and, when possible, the group (e.g., multiple animals in a group tagged together, or responsiveness in a social network framework; look at responsiveness across MFAS sources (simulated, sonobuoy, helo-dipping, operational hull-mounted).

Progress & Results

Progress will be summarized in annual technical reports and final results will be presented at scientific meetings as well as published in peer-reviewed journals. 

Data preparation tasks – summer 2024-winter 2025

Preliminary analyses tasks – winter-summer 2025

Behavioral analysis tasks – summer 2025-spring 2026