Correcting the 2019 survey abundance of Bering-Chukchi-Beaufort Seas bowhead whales for disturbance from powered skiffs

Résumé

The 2019 ice-based survey of Bering-Chukchi-Beaufort Seas bowhead whales was challenged by missed survey effort, unusual ice conditions, and frequent use of motor-powered skiffs by hunters. All three of these likely led to downward bias in the abundance estimate of Givens et al. (2021). Data were collected about boat excursions during the survey period. Indices of short-term whale abundance at the survey perch and short-term boat noise disturbance were computed from the available data, where ‘short term’ refers to a few hours. A generalized additive model (GAM) was fit to the results, predicting short-term whale abundance as a smooth function of the boat noise disturbance index, after controlling for long-term variation in the whale passage rate over the course of the season. The fitted GAM was then used to predict passage with and without the presence of boat noise. The ratio of the integrals of these two predicted passage curves provided a correction factor which can be applied post hoc to the abundance estimate of Givens et al. (2021). Variance of this correction factor was estimated using two approaches, and found to be small. A wide array of sensitivity analyses was conducted to examine the robustness of the result to potential changes in methodology, and the correction factor was found to be quite stable. The estimated correction factor would inflate the original abundance estimate by about 12%, yielding a corrected abundance of 14,025 (CV=0.228). We recommend that this replace the original abundance estimate.