Optimizing Mysis Shrimp Sampling Efforts in Lake Pend Oreille, Idaho

Mysis shrimp Mysis diluvianna were introduced in Lake Pend Oreille (LPO) in the late 1960s to benefit sport fisheries.  Only later, did managers realize mysis’ potential for competition with kokanee Oncorhynchus nerka and propagation of invasive lake trout Salvelinus namaycush.  Because of the central role mysis play in the LPO food web their density has been monitored since the 1970s, using a stratified random design.  Annually, 24-45 samples are equally allocated among the lake’s three management sections.  The goal of this study was to use existing to data to explore methods for improving the accuracy and precision of mysis density estimation.  We evaluated optimal allocation schemes for sampling mysis shrimp in the lake using the Neyman Allocation, and used semi-virtual simulations to compare the mean square error of different measures of central tendency (e.g., arithmetic, geometric, and trimmed means), and to evaluate the benefit of stratification.  Neyman optimization resulted in an average allocation ratio of 1:1:2 among lake sections, geometric means were the best measure of central tendency, and simulations suggest that stratified sampling is superior to simple random sampling only when a sufficient number of samples are allocated per strata to ensure a normal distribution of the sample mean.