Complementary analyses of behavioral responses to sonar in blue whales (Balaenoptera musculus)
Southall, B, S DeRuiter, A Friedlaender, E Hazen, J Goldbogen, A Stimpert, R Langrock, C Harris, L Thomas, G Schorr, A Allen, G Gailey, E Falcone, D Moretti, and J Calambokidis. 2015. Complementary analyses of behavioral responses to sonar in blue whales (Balaenoptera musculus). Abstract (Proceedings) 21st Biennial Conference on the Biology of Marine Mammals, San Francisco, California, December 14-18, 2015.
We used field experiments to measure potential changes in behavior from noise exposure to blue whales off southern California from 2010-2014. High-resolution movement and acoustic data were obtained from DTAGs (n=43) while surface positions and behavioral observations were made through dedicated visual focal follows. Controlled exposure experiments were used to obtain direct measurements of behavior before, during, and after simulated and actual military mid-frequency active sonar (MFAS), pseudorandom noise (PRN), and no noise controls. For a subset of foraging animals, active acoustic measurements of prey around tagged whales were obtained as contextual covariates in response analyses. Potential changes in behavior resulting from exposure were analyzed using a combination of approaches. To assess potential responses across individuals exposed within various conditions, we used two approaches: principal component analyses with generalized additive mixed models (with and without prey covariates) to identify differences among related behavioral parameters, and hidden Markov models to evaluate behavioral state switching as a consequence of exposure. To investigate potential changes within individuals, we used parallel analyses of time-series data for selected behavioral parameters (e.g., diving, horizontal movement, feeding). This included both expert scoring of responses according to a specified behavioral severity rating paradigm and quantitative change-point analyses using Mahalanobis distance statistical metrics. The suite of results indicated that some blue whales (who predominantly use low frequency signals) responded to mid-frequency anthropogenic noise, but responses were generally brief, highly dependent on exposure context (particularly behavioral state and prey distribution), and did not follow a simple dose-response model based on received exposure level. Using multiple analytical methods to investigate different aspects of potential responses both across and within individuals provides a complementary and powerful means of evaluating both broader questions about species-typical response and the individual exposure-response data needed to inform probabilistic functions for assessments of potential human impact.