Wagner, T., C.S. Vandergoot, and J. Tyson. 2009. Evaluating the Power to Detect Temporal Trends in Fishery-Independent Surveys: A Case Study Based on Gillnets Set in the Ohio Waters of Lake Erie for Walleye. North American Journal of Fisheries Management 29:805-816.
Abstract
Fishery-independent (FI) surveys provide critical information used for the sustainable management and conservation of fish populations. Because fisheries management often requires the effects of management actions to be evaluated and detected within a relatively short time-frame, it is important that research be directed towards FI survey evaluation, especially with respect to temporal trend detection capabilities. Using annual FI gillnet survey data for Lake Erie walleye collected from 1978 – 2006 as a case study, our goals were to (1) highlight the usefulness of hierarchical models for estimating spatial and temporal sources of variation in catch per effort (CPE); (2) demonstrate how resulting variance estimates can be used to examine the statistical power to detect temporal trends in CPE in relation to sample size, duration of sampling, and decisions regarding what data are most appropriate for analysis; and (3) discuss recommendations for the evaluation of FI surveys and the analysis of resulting data to support fisheries management. This case-study illustrated that the statistical power to detect temporal trends was low over relatively short sampling periods (e.g., 5 – 10 years), unless the annual decline in CPE reached 10 – 20%. For example, if 50 sites were sampled each year a 10% annual decline in CPE would not be detected with > 0.80 power until 15 years of sampling, and a 5% annual decline would not be detected with > 0.8 power for approximately 22 years. Because the evaluation of FI surveys is essential to ensure that trends in fish populations can be detected over management-relevant time periods, we suggest that a meta-analysis type approach be used across systems to quantify sources of spatial and temporal variation. This approach can be used to evaluate and identify sampling designs that increase the ability of managers to make inferences about trends in fish stocks.