Selectivity-induced bias in growth parameter estimates

A short simulation study has been carried out to investigate the size of the bias in estimates of growth parameters resulting from fish of different sizes having different probabilities of being included in catch samples. Fish were included in catch samples from the population with probability equal to an assumed relative selectivity at length. Both the mean growth parameters and the selectivity function used in the simulations closely matched those used for current toothfish assessments in Subarea 48.3. Substantial selectivity-induced bias was demonstrated, with the simulated catch sample data producing considerably lower estimates of L? and higher estimates of K than in the true population. The simulated catch sample data and the growth parameter estimates were similar to those reported by Belchier (2004, WG-FSA- 04/86). This suggests that the growth parameters used in the current toothfish assessments in Subarea 48.3 are not incompatible with the Belchier (2004, WG-FSA-04/86) data.