UC San Diego

The southern California Current Ecosystem has experienced extreme marine heatwaves and El Niño events over the last two decades, causing mass die-offs and species range shifts. The region is also one of the most well-monitored marine ecosystems, through the California Cooperative Oceanic Fisheries Investigations (CalCOFI) program, which has conducted marine mammal visual and passive acoustic monitoring (PAM) since 2004. This dissertation applies deep learning and statistical methods to CalCOFI visual and acoustic data and to long-term High Frequency Acoustic Recording Package (HARP) data to characterize cetacean distribution and density across this period of change. Chapter 1 trains a deep neural network to classify two echolocation click types produced by Pacific white-sided dolphins and applies it to 57 years of HARP recordings from 14 sites between the Gulf of California and the Gulf of Alaska. Spatial, seasonal, and diel patterns are consistent with two acoustically distinct populations that experienced possible range shifts after the 2014-2016 marine heatwave. Chapter 2 trains a deep-learning object detection algorithm to classify five blue and fin whale call types from two decades of CalCOFI PAM recordings and combines these detections with acoustic propagation modeling to produce 20-year call density estimates, showing inshore-offshore differences and interannual variability linked to ocean conditions. Chapter 3 estimates long-term density trends for ten cetacean species from CalCOFI visual surveys conducted between 2004 and 2025, finding declines in Dall's porpoise and Pacific white-sided dolphin and increases in long-beaked common dolphin and humpback whale, consistent with range contractions, expansions, and recovery. Community composition shifted between pre- and post-heatwave periods, driven by species turnover rather than a change in diversity. Chapter 4 develops a joint Bayesian model integrating CalCOFI visual and acoustic data through a shared latent density field to estimate spatial-seasonal density surfaces and acoustic cue rates for blue, fin, and humpback whales. We find species-specific habitat associations, spatial-seasonal partitioning, and seasonally offset peak singing periods between the three migratory baleen whale species. These chapters document biogeographic change linked to ocean warming and provide methods for estimating cetacean density from long-term survey data to inform conservation and management as the region continues to change.