A comparison of acoustic and visual metrics of sperm whale longline depredation

Annual federal stock assessment surveys for Alaskan sablefish also attempt to measure sperm whale depredation by quantifying visual evidence of depredation, including lip remains and damaged fish. A complementary passive acoustic method for quantifying depredation was investigated during the 2011 and 2012 survey hauls. A combination of machine-aided and human analysis counted the number of distinct “creak” sounds detected on autonomous recorders deployed during the survey, emphasizing sounds that are followed by silence (“creak-pauses”), a possible indication of prey capture. These raw counts were then adjusted for variations in background noise levels between deployments. Both a randomized Pearson correlation analysis and a generalized linear model found that noise-adjusted counts of “creak-pauses” were highly correlated with survey counts of lip remains during both years (2012: r(10) = 0.89, p = 1e-3; 2011: r(39) = 0.72, p = 4e-3) and somewhat correlated with observed sablefish damage in 2011 [r(39) = 0.37, p = 0.03], but uncorrelated with other species depredation. The acoustic depredation count was anywhere from 10% to 80% higher than the visual counts, depending on the survey year and assumptions employed. The results suggest that passive acoustics can provide upper bounds on depredation rates; however, the observed correlation breaks down whenever three or more whales are present.