Abstract
Tuna fisheries catch over three million tonnes of skipjack tuna (Katsuwonus pelamis) each year, the majority of which come from purse-seine vessels targeting fish associated with man-made fish aggregating devices (FADs). A significant challenge for fisheries management is to maximize the efficiency of skipjack tuna catches whilst minimizing the bycatch of small and immature bigeye (Thunnus obesus) and yellowfin (T. albacares) tuna, for which long-term sustainability is uncertain in 75% of the world's stocks. To better manage the issues common with this fishing method, an improved understanding of tuna behaviour around FADs is necessary. We probabilistically classified the vertical behavioural patterns of 50 bigeye and 35 yellowfin tuna (mean fork length 72cm and 70cm, respectively) electronically tagged throughout the western and central Pacific Ocean into shallow and deep states, using a state-space modelling approach. The occurrence of surface-association behaviours, defined as an individual remaining in a shallow state for 24-hours, was examined in relation to known capture events and FAD density. In general, surface-association events for both species were short and lasted on average less than three days, although events as long as 28 days were observed, and were more common in yellowfin when in archipelagic waters. Events were longest immediately following tagging in 62% and 17% of bigeye and yellowfin, respectively. Surface-association behaviour was not generally estimated just prior to recapture, being either non-existent or shorter than two days for 85% of bigeye and 74% of yellowfin. Current management measures in purse-seine tuna fisheries involve periodic or spatial closures for FAD use. If the chief benefit to purse-seine fishers of surface-association around floating objects is in locating schools in horizontal space at short-term time-scales, rather than holding fish near the surface for extended periods, controlling the number of sets made on FADs should be explored further as an additional management tool.