|Effect of population structure, sampling strategy and sample size on the estimation of length frequency distributions and biological parameters: a case study on brown shrimp (Crangon crangon)|
Redant, F. (1996). Effect of population structure, sampling strategy and sample size on the estimation of length frequency distributions and biological parameters: a case study on brown shrimp (Crangon crangon). CM Documents - ICES, C.M. 1996(D:6). International Council for the Exploration of the Sea: Copenhagen. 18 + tables + figures pp.
Part of: ICES CM Documents - ICES. ICES: Copenhagen. ISSN 1015-4744, more
Codends; Parameters; Population characteristics; Sampling; Marine
In field studies on brown shrimp (Crangon crangon), catches containing large numbers of animals are a common feature, and sub-sampling has become a common practice. So far, however, the cost-benefit-ratio of sampling strategies and sample sizes (in terms of work-load, on the one hand, and statistical soundness of the estimated population parameters, on the other) has never been thoroughly investigated. The present study tries to solve this problem, by means of simulations on theoretical "populations" (= the combined catches of a shrimp cod-end and its cover) with known size distributions, under various conditions with respect to their size composition, and the way they are subdivided into catch fractions (cod-end cover, discards and landings). As far as optimum sample size is concerned, the results of the simulations showed that, in general, samples of 750 animals (all catch fractions combined), will give sound estimates of the mean size of the population, and reasonable estimates of its length-frequency distribution, provided that the original estimates of the numbers-at-length are smoothed with a moving average of order 5. For statistically sound estimates of the size distribution, the total number of measurements has to be increased to at least 1500. With respect to sampling strategy that is to be preferred (samples of a fixed size or proportional samples, weighted according to the relative share of each catch fraction in the population), the outcome of the simulations was much less conclusive. Because of the differences in adequacy between the two methods that were tested, the choice of the optimal sampling strategy will depend on: the kind of information one expects to obtain from the samples; the levels of precision one is aiming for; the size structure of the population in itself; and the way the population is partitioned over the various catch fractions.