Evaluating sampling strategies for collecting size-based fish fecundity data: an example of Gulf of Maine northern shrimp Pandalus borealis

Fecundity information is critical in determining reproductive potential of a population. Collecting fecundity data, however, can be cost prohibitive or ineffective if a sampling protocol is not well designed. Inappropriate sampling can lead to biased estimates of fecundity, which may result in biased estimate of reproductive potential. Processing egg samples tends to be time-consuming and labour-intensive. For many fish and crustacean species, fecundity is dependent on female sizes. Nevertheless, at extreme size classes, fecundity may decrease or level off due to senescence. In order to account for this maternal effect, female sample of a wide size range need to be collected for developing a complete relationship between fecundity and body sizes. Using the Gulf of Maine northern shrimp, Pandalus borealis, as an example, we evaluated two sampling strategies, simple random sampling and size-based stratified random sampling, with a different number of sampling locations and different number of animals sampled per sampling location or length interval. The study shows that both the sampling strategies, simple random sampling and size-based stratified random sampling, can generate representative samples. However, the simulation analysis suggests that when the population size distribution is skewed with a lack of large and/or small individuals, size-based stratified random sampling is preferred due to lower variation in differences of means and medians between samples and the population. This study provides a simulation framework for identifying a cost-effective sampling protocol that can improve the estimate of fecundity, leading to an improved estimate of fish population reproductive potential.