Report of the Study Group on Age-length Structured Assessment Models (SGASAM)

Abstract

The second meeting of the ICES Study Group on Age-length Structured Assessment Models (SGASAM) was held at ICES Headquarters from 14-18 March 2005. There were 12 participants (mainly with expertise in age-length structured modelling and stock assessment) from 10 countries. The main objective of SGASAM is to investigate and evaluate the use of lengthstructured and age-length structured population models in fish stock assessment. The terms of reference for this meeting related to both model development and species-specific applications. Developments in methodologies and applications. An increasing number of age-length structured models are being developed (e.g. GADGET, CALEN, Stock Synthesis) and applied to a wide variety of species with differing life-histories. The acceptance of these for use in stock assessment is increasing, particularly outside of the ICES area. There is growing interest in the development of simpler length-structured models for the assessment of species for which age-structured data are unavailable and in particular, a number of length-structured models have been developed which make use only of lengthstructured survey data to obtain information on stock trends. These are clearly useful for stocks for which commercial catch data may also be unreliable. Incorporating process sub-models. Process models previously developed by the ICES Study Group on Growth, Maturity and Condition in Stock Projections were specifically considered. Many of these are lengthdependent and some, particularly for growth and maturity, have already been included into existing age-length structured modelling frameworks (e.g. GADGET). Further improvements to the implementation of these process models in age-length structured population models (important for the assessment of species where biological and fishery processes are better represented by length) will require greater co-operation between process modellers and agelength structured population modellers. It is therefore recommended that process modellers be encouraged to attend any further meetings of this SG. Investigating complexity. The SG identified two different ways in which age-length structured model frameworks could be used to investigate the performance of models with different levels of complexity. One approach was to consider age-length structured models as operating models to generate data sets and then evaluate other simpler models (e.g. VPA, biomass dynamic) in terms of their performance against the underlying ‘true’ system and perhaps also in terms of relative performance against alternative management regime. The second approach was the comparison of different sub-models within the same framework and comparing their performance in terms of ‘goodness of fit’ to the underlying data. There is a need for development of formal statistical methods to carry out these comparisons. Case studies. The alternative to more complexity is the development of simpler length-based approaches for species for which age-disaggregated data are sparse or unavailable. A number of species for which there are age-reading uncertainties (and hence limited age-based data) were considered by the SG and the development of length-structured models is already in progress for some of these. The SG felt that continuing work on such simpler approaches is important and would be particularly useful for the assessment of species such as Nephrops, redfish, anglerfish and some elasmobranchs.