Developing a Kalman filter approach to home range estimation: Applied to the Atlantic bluefin tuna (Thunnus thynnus)

Accurate estimation of an animal's home range, or utilization distribution, is of great importance to understanding the animal's role in the ecosystem, and for effective population management. Current methods for home range estimation often do not incorporate uncertainty in the observations of monitored animals. Given days without observations, they also have the potential to omit migration corridors when describing important habitat. Here the Extended Kalman filter is modified to return daily predicted geolocations, creating a most probable estimation of the true path the observed animal followed. Markov Chain Monte Carlo methods were used to map the uncertainty in this path to create a probability of use distribution, representing the animal's utilization distribution. The modified method was applied to Atlantic bluefin tuna (Thunnus thynnus) observed using pop-off satellite archival tags with light-based geolocation. The home range estimation technique developed can be used for any animal with a time-series of locations.