A simple implementation of the statistical modelling framework Gadget for cod in Icelandic waters

Abstract

Gadget is a statistical modelling framework that can be used to assess individual fish stocks and to create multispecies, multi-fleet and multi-area models. The development of a complex multispecies model requires understanding of the modelled single-species population. Simple single-species models are essential in evaluating whether more complex models increase understanding of the population dynamics. In this paper, Gadget is illustrated using a single-species case study with two stock components on one area. Features of the implementation include immature and mature stock components, a maturation process, along with commercial and survey fleets. Parameter estimation is done using maximum likelihood based on a variety of datasets. The estimated parameters relate to growth, maturation, fleet selection, recruitment and the initial population. The data types used to calculate the likelihood include survey indices and biological samples from the commercial catch and surveys. Optimisation of the model, along with a protocol to estimate appropriate weighting of the likelihood components, is described. A range of different assumptions and estimation methods are evaluated including the effect of reduced data availability on parameter estimation by excluding age data from the likelihood components.