Under the scope of the a4a Initiative a workshop dedicated to studying spatial effects on the stock dynamics of European Atlantic sardine took place in Ispra, Italy, the 14th to the 18th of December 2015, with the objectives of (i) explore a4a methods to assess the Southern sardine stock (Atlanto-Iberian stock) and compare the results with the current ICES assessment carried out with SS3; (ii) explore a4a methods to assess the Northern stock of sardine stock; and (iii) apply a4a to assess putative sub-stock units; discuss local depletion and mixing among sub-units of the stock. Sardine is fished mainly by UK, Netherlands France, Spain and Portugal across ICES areas VII, VIII and IXa. In France, Spain and Portugal sardine has significant social and economic importance to the fishing and canning industries. Sardine shows a complex population structure characterized by spatial heterogeneity in phenotypic characters and life-history traits. Evidence of spatial variability in dynamics does not preclude some fish mixing across the whole region, in agreement with both genetic homogeneity and the similarity in otolith element composition in larger fish. Current knowledge on sardine biology and dynamics is consistent with the hypothesis of a meta-population composed of three populations recruiting in the Bay of Biscay, off northern Portugal, and in the Gulf of Cadiz. The workshop looked into three options of spatial structures (i) current stock structure Bay of Biscay stock (BB; VIIIa,b) and Ibero Atlantic stock (IB; IXa and VIIIc); (ii) three separate sub-units Bay of Biscay, Northwest stock (NW; VIIIc, IXa-North to IXa-Central South) and South stock (S; IXa South); and (iii) a single stock. The a4a stock assessment model was used to estimate the dynamics of each sub-units in each option. To carry out the comparison across sub-units the models used were kept as similar as possible, to mitigate the effect that the choice of model can have on the final results. Uncertainty was estimated using MCMC with the ADMB implementation, which, in the most recent version, can be assessed through the FLa4a package. In the case of the overall stock, a sensitivity analysis about survey's data processing options was carried out, to investigate the robustness of the assessment results. For the Bay of Biscay a bayesian approach has also been implemented. In order to compare both methods (a4a and bayesian) a very simple separable model was selected. A visual evaluation of the consistency between the spatial hypothesis was done based on the SSB estimates, showing that both trends are remarkably similar until 2012, when they start to diverge. In 2012 the Iberian stock (sub-units NW and S) was at a very low level and the migration rates between the Iberian and the Bay of Biscay sub-units may have increased. In such case, the stock assessment model assumption of closed population is less likely to be maintained and the two estimates of SSB diverge. One of the advantages of the approach proposed is to make it possible to look into sub-units of the stock with regards to their productivity and exploitation. These results are shown as time series of fishing mortality and recruitment for each sub-unit.
