Horizontal movements of bigeye tuna (Thunnus obesus) near Hawaii determined by Kalman filter analysis of archival tagging data

Abstract

Geolocation data were recovered from archival tags applied to bigeye tuna near Hawaii. A state-space Kalman filter statistical model was used to estimate geolocation errors, movement parameters, and most probable tracks from the recovered data. Standard deviation estimates ranged from 0.5° to 4.4° latitude and from 0.2° to 1.6° longitude. Bias estimates ranged from -1.9° to 4.1° latitude and from -0.5° to 3.0° longitude. Estimates of directed movement were close to zero for most fish reaching a maximum magnitude of 5.3 nm day-1 for the one fish that moved away from its release site. Diffusivity estimates were also low, ranging from near zero to 1000 nm2 day-1. Low values of the estimated movement parameters are consistent with the restricted scale of the observed movement and the apparent fidelity of bigeye to geographical points of attraction. Inclusion of a time-dependent model of the variance in geolocation estimates reduced the variability of latitude estimates. The state-space Kalman filter model appears to provide realistic estimates of in situ geolocation errors and movement parameters, provides a means to avoid indeterminate latitude estimates during equinoxes, and is a potential bridge between analyses of individual and population movements.