ALBERT

Description: Numerous flying and swimming animals constantly need to control their heading (that is, their direction of orientation) in a flow to reach their distant destination. However, animal orientation in a flow has yet to be satisfactorily explained because it is difficult to directly measure animal heading and flow. We constructed a new animal movement model based on the asymmetric distribution of the GPS (Global Positioning System) track vector along its mean vector, which might be caused by wind flow. This statistical model enabled us to simultaneously estimate animal heading (navigational decision-making) and ocean wind information over the range traversed by free-ranging birds. We applied this method to the tracking data of homing seabirds. The wind flow estimated by the model was consistent with the spatiotemporally coarse wind information provided by an atmospheric simulation model. The estimated heading information revealed that homing seabirds could head in a direction different from that leading to the colony to offset wind effects and to enable them to eventually move in the direction they intended to take, even though they are over the open sea where visual cues are unavailable. Our results highlight the utility of combining large data sets of animal movements with the "inverse problem approach," enabling unobservable causal factors to be estimated from the observed output data. This approach potentially initiates a new era of analyzing animal decision-making in the field.