Incorporating time-series structure in medium-term recruitment projections

Abstract

One of the key tasks in current fisheries research is to improve the performance, in terms of accuracy and utility, of projections of recruitment-driven population dynamics in the medium-term. Reliable indications of the median level and variability of recruitment over a five- to ten-year time-scale would be invaluable in the determination of appropriate levels of fishing mortality, in order to attempt to maintain sustainable fish stocks. Building upon the stochastic simulation approach currently adopted within ICES stock assessment working groups, this paper investigates the use of time-series models to characterise the historical development of residuals to fitted stock-recruitment models. We use the probability of SSB (spawning stock biomass) falling below Bpa (precautionary value of spawning stock biomass) over a range of multipliers on imposed fishing mortality as a diagnostic statistic to compare projections. Case studies of commercially-important fish stocks are presented (North Sea cod, haddock and whiting), and the potential implications of the new approaches for fisheries management are discussed.