Abstract
The intricate relationship between trophic levels in marine ecosystems remains largely unexplored, particularly in the Southern Ocean. To investigate the complex composition of water constituents, notably the phytoplankton community, we used machine learning techniques to classify six bio-optical classes from a comprehensive ocean-color data time-series. In situ data from the Southern Ocean further characterized these classes by predominant phytoplankton types. Bio-logging data from southern elephant seals revealed that 60% of prey-catch attempts were concentrated within the diatom-dominated class, despite its lower regional frequency. Seals exhibited enhanced foraging activity and prolonged residency periods, up to 42 consecutive days, in diatom-rich waters compared to 5-10 days in other classes. These findings suggest a preference among southern elephant seals for diatom-rich waters, indicating a selective foraging activity influenced by the phytoplankton community's structural composition rather than total biomass alone. This preference underscores the intricate interplay between marine top predators and ecosystem dynamics, hinting at mechanisms through which phytoplankton structure shapes seal activity.