Abstract
Despite technical advances in archival tag attachment and functional longevity, long-term tracking of individuals of some fish species remains prohibitively difficult. Here we combined high resolution stable isotope sampling of otoliths, genetic assignment, growth increment analyses and a simple movement model to inform a hidden Markov model of the location of individual wild Atlantic salmon ( Salmo salar L.) at sea. The model provided a reconstruction of North Atlantic migration for a species which is extremely difficult to track throughout its marine life-stage. We show that plausible emigration and return migration patterns can be achieved from wild fish. Simulations of simplified northward and westward movement patterns in the North Atlantic were used to quantify precision and accuracy of the model which differed between these two directional scenarios. Because otolith-derived migratory information does not rely on capture, release and subsequent recapture of the individual, this can be retrieved from any fish. Thus these techniques offer a powerful tool for improving our understanding of the processes that govern movement and survival of individual fish during a protracted and largely unobservable life-stage in the oceanic environment.