Modeling species composition and mortality rates of sea turtles incidentally captured in the Portuguese pelagic longline fishery of the equatorial and tropical northeast Atlantic

Sea turtles frequently interact with pelagic longline fisheries, where they are incidentally captured and subsequently discarded. In this study, multinomial and binomial models were developed to estimate the species composition and at-vessel mortality rates of sea turtles captured in a pelagic longline fleet, using the Portuguese fleet operating in the equatorial and tropical Northeast Atlantic as a case study. A total of 2095 sea turtle interactions were recorded between 2003 and 2023, comprising leatherbacks, olive ridleys, loggerheads and green turtles. Species composition conditional on capture varied spatially, with olive ridleys dominating the equatorial waters, loggerheads in tropical waters closer to the African continent, and leatherbacks across both regions. At-vessel mortality rates conditional on capture differed significantly among species, being lowest for leatherbacks (4.3%) and highest for olive ridleys (32.9%) and loggerheads (26.8%). Both the multinomial and binomial models showed good goodness-of-fit and predictive performance. The developed models can be used to infer species composition and at-vessel mortality given a sea turtle interaction with pelagic longline fisheries in the equatorial and tropical Northeast Atlantic, particularly when species level identification and fate data are unavailable, which is common in logbook data. This work contributes to the improvement of the species-specific information and fate of sea-turtle species captured in pelagic longline fisheries, which is important for the evaluation of the status of those species and monitoring of the fisheries impacts on those populations over time. • Sea turtles are incidentally captured in pelagic longline fisheries. • Multinomial models were used to predict species composition. • Binomial models were used to estimate mortality rates. • Both models perform well and allow extrapolation to logbook data. • This improves monitoring of fisheries impacts on sea turtle bycatch.