Examples of diagnostic methods implemented for previously completed north Atlantic shortfin mako stock synthesis model runs

A range of model diagnostics were implemented for three previously completed Stock Synthesis assessment models conducted for North Atlantic shortfin mako shark. The objectives were to evaluate stock assessment model fit to data, identify possible model misspecifications, and evaluate model prediction skill by implementing nine diagnostic approaches: 1) Simultaneous visualization of residuals from multiple catch per unit effort (CPUE) indices using Just Another Bayesian Biomass Assessment (JABBA) residual plots; 2) runs tests applied to individual CPUE indices; 3) runs tests applied to size composition data; 4) a runs test applied to estimated recruitment deviations; 5) retrospective analyses; 6) likelihood component profiles; 7) a deterministic age-structured production model (ASPM); 8)Markov Chain Monte Carlo (MCMC); and 9) hind-cast cross-validation. Overall, the diagnostic results were consistent and indicated that changes in abundance after the 1990’s could not be explained by catches alone, that the abundance information, both absolute and relative, contained in the CPUE indices could not be interpreted without modelling fluctuations in recruitment, and that the recruitment deviation estimation had a large influence on the model fit.