Swimming in an ocean of curves: A functional approach to understanding elephant seal habitat use in the Argentine Basin

Abstract

In recent decades, southern elephant seals (SES) have become a species of particular importance in ocean data acquisition. The scientific community has taken advantage of technological advances coupled with suitable SES biological traits to record numerous variables in challenging environments and to study interactions between SES and oceanographic features. In the context of big dataset acquisition, there is a growing need for methodological tools to analyze and extract key data features while integrating their complexity. Although much attention has been paid to study elephant seal foraging strategies, the continuity of their surrounding three-dimensional environments is seldom integrated. Knowledge gaps persist in understanding habitat use by SES, while the representativeness of a predator-based approach to understanding ecosystem structuring is still questioned. In this study, we explore SES habitat use by using a functional data analysis approach (FDA) to describe the foraging environment of five female elephant seals feeding in the Southwestern Atlantic Ocean. Functional principal component analysis followed by model-based clustering were applied to temperature and salinity (TS) profiles from Mercator model outputs to discriminate waters sharing similar thermohaline structures. Secondly, in situ TS profiles recorded by the SES were employed to determine the habitat visited within the range of potential environments identified from the model data. Four Functional Oceanographic Domains (FOD) were identified in the Brazil-Malvinas Confluence, all visited, in varying proportion, by four of the five females studied. We found that the females favored areas where all the FODs converge and mix, generating thermal fronts and eddies. Prey-capture attempts increased in such areas. Our results are in accordance with previous findings, suggesting that (sub-)mesoscale features act as biological hotspots. This study highlights the potential of coupling FDA with model-based clustering for describing complex environments with minimal loss of information. As well as contributing to better understanding of elephant seal habitat use and foraging strategies, this approach opens up a wide range of applications in oceanography and ecology.