Analysis of groundfish survey abundance data: combining the GLM and delta approaches

Abstract

This paper describes a method for the analysis of groundfish survey data by incorporating zero and non-zero values into a single model. This is done by using a model which modifies the delta-distribution approach to fit into the GLM framework and uses maximum likelihood to estimate parameters. No prior assumptions of homogeneity are used for the structure of the zero or non-zero values. The method is primarily applicable to fixed-station designs, although extensions to other designs are possible. The maximum likelihood estimation reduces to fitting a GLM to 0/1 values and another GLM to the positive abundance values. The new model is tested on Icelandic groundfish survey data. It is seen that the model can be used for evaluating the effect of different factors on catch rates as well as estimating abundance indices. Results from different models are compared on the basis of tuned VPA runs.