Residual Diagnostics for Indian Ocean yellowfin tuna Stock Synthesis models

This paper presents additional residual diagnostics for the 2018 Stock Synthesis 4-Area reference case model and the newly proposed 2019 2-Area reference model for Indian Ocean yellowfin tuna. The set of residual diagnostics comprise runs tests to evaluate the randomness of residuals, three-sigma limits to identify outliers, and “JABBA residual plots” together with associated Residual-Mean-Squared (RMSE) values. In addition, a summary of runs test results is provided for the 24 Stock Synthesis models of the 2018 reference grid. The residual diagnostic results indicate a slightly higher runs test passing rate for the 2018 4-Area reference case model, but showed overall no clear evidence in favor of either reference case model. On closer inspection of the individual grid model runs, the scenarios that estimate a second catchability coefficient (q2) were generally associated lower runs test passing rates for CPUE residual than alternative formulations. The generally poor runs test passing rates of mean lengths residuals may be interpreted in support of the relative low weight that currently assigned to the length composition for both reference case models.