Incorporating fish food habits data into fish population assessment models

Synopsis

Marine fisheries management is turning more to multispecies assessment models for management advice. Examples of such models include Laevastu’s DYNUMES model, Andersen-Ursin’s multispecies BevertonHolt model, several multispecies VPA models, and singlespecies models. The main links between species in the models are submodels describing diet composition and food intake. These submodels are described in detail and the sampling requirements for estimating the various food habit parameters are discussed. Diet composition parameters are estimated from field collections of stomachs taken over large areas with sampling stratified by fish size, season, area, and time of day. Food intake parameters are usually estimated either from laboratory feeding experiments which determine maintenance and growth rations by temperature or from a combination of field stomach content weight data and laboratory gastric evacuation rate studies. Modelling population level feeding dynamics requires a re-assessment of how food habits data are collected and analyzed. The scale of the sampling program depends on the complexity of questions being asked about the system which in part determines which model will be used to respond to the questions. Sensitivity analysis can provide information about the sources of uncertainty in models due to model structure and model parameter estimates. Sensitivity analysis results can provide guidance as to which model to use and where to allocate research priorities. The ultimate choice of models and sampling programs depends on many variables including the complexity and size of the system, the availability of existing data, and the cost and ease of obtaining the most important data.