Spatial Heterogeneity, Variable Rewards, Tag Loss, and Tagging Mortality Affect the Performance of Mark–Recapture Designs to Estimate Exploitation: an Example using Red Snapper in the Northern Gulf of Mexico

Abstract

Accurate estimates of exploitation rate are essential to the management of exploited fisheries. Tagging studies are often used to estimate exploitation rates, but the performance of these approaches depends strongly on study design characteristics and the magnitude of assumption violations. We simulated a suite of candidate study designs for 1-year high-reward and variable-reward tagging studies, exploring a range of sample sizes (number of tagged fish), exploitation rates, tagging mortality rates, tag loss rates, proportions of double-tagged fish, and spatial variation in fish density, tag releases, and fishing effort. We calculated the uncertainty, biases, and reward costs of these candidate study designs to determine the most cost-effective approach to accurately estimate exploitation rate for Red Snapper Lutjanus campechanus in Alabama waters of the Gulf of Mexico. We also investigated how incorrectly assuming a 100% reporting rate would affect these study results. Our simulations demonstrated that using all high-reward tags provided more accurate and precise exploitation rate estimates than the variable-reward approach but only if 100% reporting could be safely assumed. Further, distributing tags uniformly over the study area when the true spatial distribution of the population and fishing pressure varied over that area drastically biased exploitation rate estimates, suggesting that prior knowledge of the population’s true spatial distribution over the study area is needed to ensure accurate estimates of exploitation rate. The most cost-effective study design involved tagging between 400 and 1,600 fish with high-reward tags, with 40% of the fish double-tagged, and the tags spatially distributed in the same proportion as the population. However, violation of the 100% reporting rate assumption resulted in a proportional downward bias in the estimated exploitation rate. Simulation studies such as this are critical to ensure that cost-effective study designs produce accurate and precise estimates of exploitation rate, particularly for high-value species such as Red Snapper.

Received July 13, 2016; accepted February 24, 2017 Published online April 14, 2017

ACKNOWLEDGMENTS

We thank the ALDCNR Marine Resources Division for providing the funds for this project. We are especially grateful to John Mareska (ALDCNR Marine Resources Division) for consultation and help with Red Snapper data retrieval and to Nicholas Ducharme-Barth (University of Florida) and Kenneth Brennan (National Oceanic and Atmospheric Administration, Southeast Fisheries Science Center) for assistance with population distribution and fishing effort distributions. Additional thanks go to Troy Farmer and Ben Staton (Auburn University) for advice and help with R and to the neighbors on Rainbow Avenue for supplying the computer power necessary to run models. We also thank three reviewers whose comments improved the manuscript.