Interbenchmark process on Sardine (Sardina pilchardus) in the Bay of Biscay (IBPSARDINE)

This report documents a Bay of Biscay sardine inter-benchmark tasked with evaluating the stock assessment focusing on retrospective bias, data revisions and updating reference points. The working group (WG) used standard model diagnostics to evaluate a series of interventions de-signed to evaluate the models and to determine causes of and corrections for the retrospective bias.The retrospective bias could be corrected by several straightforward interventions. First, fixing selectivity at asymptotic improved model fit and reduced bias. Second, invoking a very weak stock–recruitment relationship (steepness=0.99) and commensurate bias correction ramping on recruitment deviations coupled with not estimating terminal year recruitment, further reduced the bias. Such a treatment of terminal year recruitment and penalizing poorly informed recruit-ment deviations is common assessment practice. Additional concerns were raised by the esti-mated catchability coefficients above one for the PELGAS and BIOMAN surveys. There are a number of reasons why these surveys could estimate higher abundance than the assessment model. These include mismatch of timing given the rapid population dynamics, overestimation of acoustic biomass, mismatch of assumed selectivity of the survey as well as many other com-mon issues that support the standard practice of treating most surveys as relative rather than absolute. Once the decision to use these indices as relative inputs, the absolute value of catcha-bility is meaningless as the index could simply be scaled to a mean of one with the same impact in the model.Given the substantial reduction in retrospective bias achieved through straightforward model interventions and the solid diagnostic performance of the WG-preferred model, the WG recom-mends the assessment be upgraded from category 2 to category 1. Nonetheless the model cannot estimate MSY-based reference points and this requires proxies. Based on considerations of life history, the WG recommends a proxy of SPR35%. Recommendations for future work include explicitly modelling variability in growth reflecting the declines in mean weight-at-age, incorpo-rating length composition and considering a management procedure approach as the majority of catch comes from ages 1 and 2 which are very poorly informed in catch projection due to the time lag between the assessment and the provision of management advice.