Multi-Criteria Decision Analysis for Recreational Trout Fisheries in British Columbia, Canada: A Bayesian Network Implementation

Abstract

One of the key challenges in recreational fisheries management is to provide a rational basis for decisions in the face of conflicting objectives, such as improving angling opportunities, maintaining satisfaction across a diverse fishing client base, controlling costs, and conserving wild populations. We developed a multi-criteria decision analysis approach for managing recreational trout fisheries (e.g., Rainbow Trout Oncorhynchus mykiss). The approach was implemented in a Bayesian decision network. The decision support tool, called “Stock-Optim,” provides a user-friendly interface for predicting fishery performance from alternate stocking prescriptions. The tool integrates survey information on angler typology and satisfaction with previously developed models for fish biology and fishery dynamics to more fully consider the biological and social outcomes of management decisions. Specifically, the tool evaluated alternative stocking options given three performance criteria: angler effort, angler satisfaction, and the cost of the stocking program. Predicted effort was highest for fish that were released in the size range of 8–20 g and at stocking densities of 200–500 fish/ha. Effort maximization at these rates and sizes is a result of compromise between the conflicting preferences of Rainbow Trout enthusiasts and occasional anglers toward fish size and harvest. Furthermore, lowering the stocking program’s costs will lead to lower stocking rates and thereby favor the enthusiasts. Currently, stocking levels in British Columbia are lower than levels that would maximize effort and are most consistent with a policy of maximizing satisfaction for Rainbow Trout enthusiasts and minimizing costs. Stock-Optim will allow managers to compare predicted outcomes from the current and alternative regimes with stated lake-specific or region-specific management objectives and regional averages and thereby more closely meet these objectives in the future. Lastly, the model was validated by comparing predicted effort with observed effort in stocked lakes.

Received February 1, 2016; accepted July 14, 2016 Published online November 28, 2016

ACKNOWLEDGMENTS

D.A.V. was supported by an industrial postdoctoral fellowship from the Natural Sciences and Engineering Research Council of Canada, a Mitacs Elevate fellowship, and the FFSBC. We are grateful to Sara Northrup and Marcus Boucher (FFSBC) for their help in using the small-lakes database. We also thank biologists with the British Columbia MFLNRO for their comments and suggestions on the model. We extend our gratitude to Kornelia Dabrowska and Wolfgang Haider (Simon Fraser University) for familiarizing us with their choice modeling study so that their results could be used in this paper.