Constructing stock abundance indices from catch and effort data: Some nuts and bolts

Abstract

Construction of annual indices of stock abundance based on the standardisation of catch and effort data remains central to many fisheries assessments. However, while the use of advanced statistical methods has helped catch rates to be standardised accounting for many explanatory variables, some of the more routine aspects of constructing abundance indices, such as how the index is constructed from the model parameters, receives little explanation in many analyses. This has lead to a lack of understanding as to how the indices are constructed and in some instances incorrect techniques have been applied. This lack of understanding can be a particular issue when interactions are used in the standardising model, especially those which contain interactions with the temporal effects over which the time-series of the abundance index is required. Other issues include the use of weighted model fits, the influence of anomalous data values, how best to impute missing values when required, the consequences of model mis-specification, and the use of random-effects. In this study, the basic approach for constructing abundance indices is outlined and a worked example using simulated data is presented to explore the nature of these issues and several techniques are suggested for dealing with them and to overcome potential biases. Finally, the methods presently used by ICCAT scientists for constructing abundance indices are reviewed.