A Principal Component Analysis of Vertical Temperature Profiles for Tracking Movements of Large Pelagic Fishes

Abstract

Pop-up satellite archival tag (PSAT) technology that records depth, temperature, and light-level data has expanded the understanding of free-swimming behavior for numerous pelagic animals. Astronomical algorithms using these light-level data have allowed geolocation estimates of daily longitude and latitude. However, many pelagic animals have a crepuscular behavior pattern in which individuals are at depths below the photic layer during the day, thus precluding the use of traditional light-based movement algorithms for geolocation in such species as swordfish. A principal component analysis (PCA) of temperature profiles is described herein that utilizes depth and temperature data rather than light to estimate the horizontal movement between the initial location of tag release and transmission. PSAT data from swordfish (n=4), blue marlin (n=14), white marlin (n=2), and black marlin (n=1) were used to generate daily coordinate estimates. The marlin data provided sufficient light information to derive geolocation estimates using two light-based state space models, while the hydrographic PCA model was used to derive comparison estimates. Comparisons of the two model types show an average root mean square difference of 175.4 km demonstrating that the PCA model can be used to extract the movement of tagged swordfish and other pelagic species demonstrating crepuscular behavior. Integration of this PCA-based geolocation methods with both the best available estimates of the ocean temperature at the time of tag deployment and the existing light-based geolocation models would provide additional information on fine-scale movement of tagged fish.