Atlantic mackerel fisheries are of economic importance in EU waters. Triennial surveying of pelagic mackerel eggs is used to assess stocks, which is the main scientific basis for fisheries management and absolute estimates of spawner abundance are obtained from these surveys. However, recent research has suggested that there may be large biases in survey estimates of abundance, partly due to incomplete survey coverage of the spatial and temporal extent of pelagic eggs in the spawning season. It is also unclear if the estimation methods applied adequately account for uncertainties in the estimates of spawner abundance, or if the current management approach is sufficiently robust to address these uncertainties.
To date, the pelagic egg survey is the only fisheries-independent data source available for mackerel and horse mackerel stocks. The calculated egg abundance is entered into a stock assessment model that applies an age-structured population estimation method (VPA). This project aims to combine geostatistical and Bayesian estimation methods to reduce bias in estimates of egg abundance and to better account for uncertainties in these estimates.
The main objectives of the project are:
1) to combine geostatistical and Bayesian statistical methods to improve the scientific basis for the management of Atlantic mackerel stocks
2) to apply Bayesian decision theory to evaluate the potential consequences for fishery management of applying both total allowable catch (TAC) and spatial controls, and to assess the information gathering requirements of these controls.
The research carried out in this project deals with the sequential survey data on pelagic egg densities and aims to improve the estimates of egg production from this data, to better account for uncertainty. Research is not directly targeted at improving the stock biomass estimation methods and does not consider parameters such as fecundity and sex ratios. However, an improved estimate of egg production will have a positive impact on the output of models currently used to estimate stock abundance. Based on this principle, this project also aims to quantify the impact of improved egg production estimates on the stock biomass estimates obtained through conventional modelling techniques, which use the egg abundance estimates as one of the input parameters.
Geostatistical and Bayesian estimation methods are being combined to reduce bias in estimates of egg abundance and to better account for uncertainties in these estimates.
Furthermore, Bayesian decision analysis methods will be applied to identify fishery control measures, information gathering and estimation methods that will ensure that the management methods applied are both adequate in dealing with the uncertainties, and conform to the recently adopted precautionary guidelines for fishery management in the CFP. The project also aims to use the uncertainty calculation methodology developed to design an 'optimal' egg survey design.
Progress to Date
During the first thirty-month period of the project, the planned activities carried out involved:
1) the collection and compilation of the egg survey data and additional information required for the development of the GIS database, the geostatistical Bayesian estimator and the fisheries management scenarios
2) the collection of information on available models for the estimation of mackerel egg density and annual egg production (AEP), and the statistical and geostatistical analysis of the egg densities from the past sampling campaigns
3) work on incorporating the geostatistical Bayesian estimator by studying the parameters of the traditional estimator in a Bayesian framework
4) work on the geostatistical Bayesian estimator by identifying and developing a methodology for addressing the uncertainty associated with the variogram fitting procedure
5) further work on the geostatistical estimator by comparing various methods for obtaining global estimation variance
6) a study of the alternative fishery management options: the construction of a management decision model, and the criteria to be used to evaluate the performance of each fishery management option.
We expect to achieve the following in the project: 1. the development of a mathematical Bayesian framework for geostatistical analysis of sequential spatial data on pelagic egg densities to improve estimates of egg production and better account for uncertainty; incorporate environmental/bathymetric data to improve estimates;
2. the development of Bayesian Monte Carlo integration methods to estimate annual spawner abundance from geostatistical analysis of sequential data for pelagic egg densities;
3. the development of a method for optimising egg survey designs which, for given levels of total annual pelagic egg survey effort, will minimise bias and variance in spawner biomass estimates provided by geostatistical estimator;
4. the assessment of the potential biological consequences of some alternative management options for the western and horse mackerel fisheries that involve different abundance surveying and harvest control methods;
5. the identification of the optimal resource abundance survey and estimation approaches for each combination of methods for controlling fishing mortality;
6. the comparison of the risks of over-depletion and robustness to uncertainties in stock size of the western and horse mackerel fisheries under different methods of controlling fishing mortality, with the quality of information available from different methods for gathering information.
7. the identification of how alternative management strategies can potentially be applied to improve sustainability of the western and horse mackerel fisheries.
FISHERIES AND AQUACULTURE, QUANTITATIVE APPROACHES AND MODELLING
Scientist responsible for the project
Professor SEVKET DURUCAN
SW7 2AZ London
United Kingdom (The) - GB
Phone: +44 2075 947354
Fax: +44 2075 947354
||Imperial College of Science, Technology and Medicine
||01 February 2000
||634 300 €
|Total EC contribution
||465 570 €