Abstract
Pre-spill and post-spill passive acoustic data collected by multiple fixed acoustic sensors monitoring about 2400 km 2 area to the west of the Deepwater Horizon oil spill in the northern Gulf of Mexico (GoM) were analyzed to understand long term local density trends and habitat use by different species of beaked whales. The data were collected in the Mississippi Valley/Canyon area between 2007 and 2017. A multistage algorithm based on unsupervised machine learning was developed to detect and classify different species of beaked whales and to derive species- and site-specific densities in different years before and after the oil spill. The results suggest that beaked whales continued to occupy and feed in these areas following the Deepwater Horizon oil spill thus raising concerns about (1) potential long-term effects of the spill on these species and (2) the habitat conditions after the spill. The average estimated local density of Cuvier’s beaked whales at the closest site, about 16 km away from the spill location showed statistically significant increase from July 2007 to September 2010, and then from September 2010 to 2015. This is the first acoustic study showing that Gervais’ beaked whales are predominantly present at the shallow site and that Cuvier’s species dominate at two deeper sites, supporting the habitat division (ecological niche) hypothesis. The findings call for continuing high-spatial-resolution long-term observations to fully characterize baseline beaked whale population and habitat use, to understand the causes of regional migrations, and to monitor the long-term impact of the spill.