Estimating fish stock population parameters from bottom trawl surveys: sources and magnitude of uncertainty

Abstract

Fisheries surveys are widely used in the stock assessment process by providing time series of relative abundance and recruitment strength. Surveys also provide biological data that have the potential to improve stock projections by providing alternatives to the use of spawning stock biomass as a measure of stock reproductive potential. The objective of the present work is to illustrate various sources of error in survey data, using examples from bottom trawl surveys and to provide innovative methods for identifying and minimising these errors. In Chapter 1, the current work is framed within the existing literature and an overview is given of the different sources of error. The next six chapters consist of papers that have been published, are in press, or under review with peer-reviewed journals. These papers consist of case-studies, each exploring a different source of error. In Chapter 2, a method for summarising the precision of a length-frequency distribution is presented and recommendations are given on the sample sizes required to achieve a particlar precision level. In Chapter 3, the variability in the assignment of maturity stages of plaice (Pleuronectes platessa, L.) and whiting (Merlangius merlangus, L.) is examined, using statistical techniques developed in the social sciences. In Chapter 4, the spatial variability in the age-length structure of haddock (Melanogrammus aeglefinus, L.) is explored and a new method for comparing age-length keys is developed. In Chapter 5, the spatial structure in the length-weight relationships of whiting and haddock are analysed using geostatistical tools. In Chapter 6, the spatial trend in the sex ratio of megrim (Lepidorhombus whiffiagonis, Walbaum) is explored using generalized additive modelling techniques. In Chapter 7, spatial trends in the proportion of mature cod (Gadus morhua, L.) are investigated. In the concluding Chapter 8, the scale of various errors is reviewed in the light of the current work and recommendations for sampling design are proposed.