Evaluation of the status and risk of overexploitation of the Pacific billfish stocks considering non-stationary population processes

Fish population processes could exhibit non-stationary behaviour as a stochastic biological process with temporal autocorrelation that may be influenced by environmental changes. Here we developed a Bayesian autoregressive state-space surplus production modelling framework to explore potential non-stationarity in population processes. We then evaluated the consequence of non-stationary population processes on the future risk of overexploitation for three Pacific billfish stocks (striped marlin, Kajikia audax; blue marlin, Makaira nigricans; and swordfish Xiphias gladius) that are formally assessed on a regular basis by a Regional Fisheries Management Organization in the Pacific Ocean. The results showed evidence of non-stationary population processes for Western and Central North Pacific Ocean (WCNPO) striped marlin, and to a lesser extent, Pacific blue marlin and WCNPO swordfish. Trends in the theoretical maximum sustainable yield and intrinsic growth rate were observed as oscillating regimes for swordfish, and as long-term directional changes for striped marlin. The non-stationary population processes did not strongly influence the forecasted biomass trend at the current catch level for any of the three stocks. However, the future risk of overexploitation (Prob[B < BMSY]) was sensitive to changes in the population processes for striped marlin (increased the risk by 20%). This work illustrates that the inclusion of non-stationary population processes could impose challenges for developing a stock rebuilding plan and provides a framework to account for non-stationary population processes for the billfish stocks in the Pacific Ocean. © 2019 Elsevier Ltd