Preliminary analysis of state-space stock assessment model for the chub mackerel in the Northwest Pacific

The technical working group for chub mackerel stock assessment in NPFC has decided to use an operating model for comparing the performance of different assessment models. Stock assessment of Pacific chub mackerel has long been conducted by tuned virtual population analysis (VPA) in Japan, but this method has a few potential problems, such as disregarding catch-at-age error, estimating stock-recruitment relationship exteriorly, highly-variable selectivity due to overfitting, and large uncertainty in abundance estimates in the terminal year. State-space stock assessment model (SAM), a kind of statistical catch-at-age models, can be an alternative approach, which potentially overcomes these problems. Here we report results of preliminary analysis using SAM and its comparison to tuned VPA for the chub mackerel in the Northwest Pacific. SAM exhibited a similar long-term trend of stock biomass estimates to that of tuned VPA, suggesting moderate robustness of relative changes of biomass estimates against model choice. However, SAM estimated lower stock biomass than tuned VPA especially for recent years, highlighting the importance of comparing the accuracy of different stock assessment models. Japan will consider both VPA and SAM as a potential stock assessment model to be evaluated with the operating model.