Retrospective analysis identifies largely effective seabird mitigation strategies with some opportunities for improvement Fisheries Research

Oceanic seabirds, including albatrosses and petrels, are, in part, threatened by bycatch in pelagic longline fisheries. Gear and operational requirements are often implemented to reduce seabird interactions and their severity. Decades of at-sea monitoring facilitated a retrospective evaluation of the effectiveness of mitigation measures in Hawaiʻi's pelagic shallow- and deep-set longline fisheries. We investigated how mitigation measures influence the probability of interaction and disposition (alive vs. dead) of Laysan (Phoebastria immutabilis) and black-footed (P. nigripes) albatrosses using categorical ensemble random forest models. Our analysis found that most mitigation measures efficiently reduce seabird bycatch, but few influence whether interactions are fatal. North of 23° N, seabird interactions increased in the deep-set fishery, and we concluded that continuing to use this boundary to activate additional seabird mitigation measures was appropriate. In the shallow-set fishery, side-setting and stern-setting proved equally effective, increasing operational flexibility. Night-setting, a shallow-set mitigation measure since 2001, is highly effective at reducing interactions, though interaction and mortality rates were 10-15 times higher during full moon phases, with most of the shallow-set effort occurring during these periods. Several measures yielded unanticipated results. The use of blue-dyed bait and strategic offal/bait discharge were associated with increased interactions, possibly reflecting the fishers' response to elevated seabird presence. These findings supported the 2024 regulatory removal of the blue-dyed bait requirement in the deep-set fishery. This study provides a framework for adaptive management by quantifying the most effective mitigation measures and informing domestic and international policy aimed at minimizing seabird bycatch in longline fisheries.