Assessment of the South African sardine resource using data from 1984-2011: further results for a two stock hypothesis

Abstract

As part of the process of updating the assessment of the South African sardine resource, a model of the sardine two mixing-stock hypothesis is being developed. This hypothesis postulates a “west” stock distributed west of Cape Agulhas and a “south” stock distributed south-east of Cape Agulhas with movement from the “west” to the “south” stock in November as recruits age to 1 year olds. de Moor and Butterworth (2012a) presented some initial results for a two mixing-stock model, but cautioned that those results may not yet have fully converged and that the lack of a Hessian prevented MCMC simulation to estimate posterior distributions for key model parameters. Work has continued on this model to try to ascertain in which areas the model may be overparameterised and which “unimportant” parameters can be fixed at their estimated values without influencing results. In this document further results for the model of a two sardine mixing-stock hypothesis are presented. The current fit is improved from that of de Moor and Butterworth (2012a) and a Hessian is estimated by ADMB. The main changes to the model are detailed below. Some ideas for further work are also listed.