| Summary: | Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution September 2018 Highly migratory marine fishes support valuable commercial fisheries worldwide. Yet, many target species have proven difficult to study due to long-distance migrations and regular deep diving. Despite the dominance of oceanographic features, such as fronts and eddies, in the open ocean, the biophysical interactions occurring at the oceanic (sub)mesoscale (< 100 km) remain poorly understood. This leads to a paucity of knowledge on oceanographic associations of pelagic fishes and hinders management efforts. With ever-improving oceanographic datasets and modeling outputs, we can leverage these tools both to derive better estimates of animal movements and to quantify fish-environment interactions. In this thesis, I developed analytical tools to characterize the biophysical interactions influencing animal behavior and species’ ecology in the open ocean. A novel, observation-based likelihood framework was combined with a Bayesian state-space model to improve geolocation estimates for archival-tagged fishes using oceanographic profile data. Using this approach, I constructed track estimates for a large basking shark tag dataset using a high-resolution oceanographic model and discovered a wide range of movement strategies. I also applied this modeling approach to track archival-tagged swordfish, which revealed affinity for thermal front and eddy habitats throughout the North Atlantic that was further corroborated by synthesizing these results with a fisheries-dependent conventional tag dataset. An additive modeling approach applied to longline catch-per-unit effort data further highlighted the biophysical interactions that characterize variability in swordfish catch. In the final chapter, I designed a synergistic analysis of high-resolution, 3D shark movements and satellite observations to quantify the influence of ...
|
|---|