Accounting for positional uncertainty when modeling received levels for tagged cetaceans exposed to sonar

Exposure to anthropogenic sound can have a range of negative behavioral and physical effects on marine species and is of increasing ecological and regulatory concern. In particular, the response of marine mammals, and notably the family of cryptic deep-diving beaked whales, to military sonar is a timely and complex issue. To make inference on aspects of response by individual whales to noise of any type, it is critical to either measure or systematically estimate what received levels (RLs) the animal actually experienced. Various tools and techniques exist to monitor RLs and associated responses, each with advantages and disadvantages. Most behav-ioral response studies to date have used relatively short-term (hours to a few days), high-resolution acoustic tags that provide direct RL measurements. Because of their short duration, these tags do not allow for assessments of longer-duration base-line behavior before and following a disturbance that may tell us more about the nature of response within a broader context for tagged individuals. In contrast, longer-duration (weeks to months), satellite-transmitting tags lack high-resolution kine-matic data and the ability to directly measure RL. Herein, we address these issues and efforts to derive robust statistical RL characterizations using animal movement and fine-scale, site-specific sound propagation modeling for longer-duration tags in the context of a behavioral response study off Cape Hatteras, North Carolina. In the autumn of 2017, we tagged nine Cuvier's beaked whales and three short-finned pilot whales and conducted controlled exposure experiments using simulated and operational military mid-frequency active sonar. We used sound propagation modeling methods and modeled positions of individual animals to estimate RLs in four dimensions and to statistically describe uncer-tainty within volumes of water space where animals were predicted to occur during exposure periods. By properly accounting for positional error in this study, it is clear that previous studies using single median RL estimates drastically underestimate the full range of plausible values; ranges in estimated RLs here often exceeded 40 dB. We also demonstrate how ancillary data from visual focal follows of tagged individuals can significantly narrow estimated RL ranges. Further, we compared measured RLs on a calibrated acoustic tag to modeled RLs at the same position to evaluate our volumetric modeling results. While satellite tags record data over longer time frames, their substantial geospatial error coupled with the unique deep-diving behavior of beaked whales means that estimates of RL can vary broadly and, consequently, that single point estimates from less robust approaches may be substantially in error. Accounting for this uncertainty using robust statistical modeling is critical to fairly characterize variance and effectively assess exposure-response relationships.

Duke Scholars

Author Robert Schick Marine Science and Conservation

Author Douglas Nowacek Marine Science and Conservation