Abstract:
|
In fisheries management, stock size estimation is the basis for key policy decisions. We look at design-based sample survey methods for hydroacoustic fisheries surveys, seeking to improve estimation when the target stock has a patchy spatial distribution. We examine the efficiency and feasibility of adaptive cluster sampling (ACS). A simulation experiment compared the relative efficiency of ACS and traditional sampling designs in a hydroacoustic survey setting. Fish densities with known spatial covariance were generated and subjected to repeated sampling; distributions of the different estimators are compared. Distribution of the estimator for traditional designs based on one-stage cluster sampling was markedly skewed. ACS designs were optimal for all stocks with small-scale spatial correlation in fish density, yielding estimates that were not skewed, with lower variance and fewer large errors than traditional designs. The uncertainty in the final ACS sample size can be partially controlled by applying ACS within a stratified design. For stocks with an aggregated or patchy spatial distribution, ACS can provide a more precise estimate of stock size than traditional survey methods.
|