English abstract
Reliable estimates of fish growth are an essential pre-requisite to understanding the dynamics of a fished stock. The design of any growth sampling scheme rests on two basic considerations: the cost of the age distribution sampling methodology and the precision of the growth estimates required. Although, age-length analyses are more precise than length-based methods, they are costly in terms of money, labor, and effort. Therefore, adopting strategies of cost-effective sampling by reducing sample sizes without compromising reliable parameter estimates is a desired requirement. For the purpose of assessing the effect of reducing sample size on the growth parameter estimates for Scomberomorus commerson harvested from Oman, a reference population of a sample size of 923 was re-sampled seven times with sub-samples equaling 200, 300,..., 800. The process of re-sampling was repeated 1000 times with replacement for each sample size. Associated statistics were estimated and compared to the reference population estimates. The difference of means and their associated standard errors between the reference and the sampled data were marginal with sample size > 600. The accuracy of the growth parameter estimates using both the bootstrap confidence intervals and biased corrected percentile confidence intervals narrowed as sample size increased and stabilized at a sample size of 400. Precision increases marginally after a sample size of 600. Based on the results from our study, sample size 400-600 are considerably sufficient for a future reliable growth estimate of Scomberomorus commerson.
