2001. Variability of abundance indices and its progression through age-structured models: a stochastic simulation with Flemish Cap cod

The assessment of many fish stocks in the North Atlantic are based on age-structured models like XSA or ADAPT. The abundances and fishing mortalities provided by these models are of main interest and their errors are calculated analytically from the catchability relationship. Assumptions on this relationship are explored by Monte Carlo simulation of a XSA analysis of Flemish Cap cod tuned with survey abundance indices for which variance-covariance was previously calculated by bootstrap. Two different Monte Carlo simulations were carried out: with and without covariance among those abundance indices. The results show that the XSA applied to the Flemish Cap cod is quite robust against these assumptions, but the errors are best evaluated by Monte Carlo simulation, which provides more accurate and precise results. Covariance of indices does not promote different results but correlation of the results should be taken into account on short-term projections. The Monte Carlo simulation, or stochastic XSA, also provides directly the probability profile for spawning stock biomass if variance estimates of the maturity ogive and stock weight at age are included. This probability profile could be an important tool to evaluate the risk of reopening fisheries in the context of a precautionary approach