Evaluation of alternative stratifications for a stratified random fishery-independent survey

Abstract

Stratified random sampling (StRS) can lead to improved precision and accuracy in estimating key fisheries population parameters, particularly when the spatial distributions of target fish populations have high heterogeneity among different strata and homogeneities within a stratum in the survey area. As the spatial distributions of many fish populations might shift in response to environmental changes and fishing activities, survey designs developed based on previous fish distribution patterns may need to be re-examined. A simulation study was conducted to evaluate the performances and consistency of 12 stratification designs for achieving multiple survey objectives, including the abundance indices of fish species and species diversity indices, for a fishery-independent survey in the coastal waters. Relative estimation error (REE), relative bias (RB) and coefficient of variation (CV) were used to measure the precision and accuracy of mean estimates values for the 12 stratification schemes. The performances of different stratification designs were likely to differ for various survey objectives. As for a multispecies survey, though the current stratification design (design 10) did not always have the lowest REE and absolute RB values, it had the most stable performances for all indices over 3 years. It indicated that the current survey design was still the optimal design according to these measurements. Thus, even though the distribution of fish populations and community composition changed over seasons and years, the current stratification design was robust and could still capture key characteristics of target fish populations and community composition.