SIMULATED ADAPTIVE MANAGEMENT FOR TIMBER AND WILDLIFE UNDER UNCERTAINTY

—A spatially explicit stochastic behavior simulation model for the endangered red-cockaded woodpeckers (Picoides borealis) is coupled with a forest management optimization algorithm to simulate adaptive (feedback) management within an uncertain environment. To update the adaptive forest harvest schedule in a timely manner during each management planning period, a genetic algorithm heuristic is employed. This model is used to evaluate management policies for the production of timber and red-cockaded woodpeckers. Forest management plans are typically based on an “optimal” activity schedule which can be rendered infeasible during implementation due to changes in the resources being managed and management goals. Thus in practice, such an activity schedule is evaluated periodically during implementation, and is often replaced with a newly generated “optimal” schedule. This paper describes an analytical tool for evaluating management strategies that take into account this adaptive implementation of activity schedules. Computer simulation models offer valuable, and often the only, tools for testing and examining intricate theories concerning the response of wildlife to management activities. This is especially true for species that depend on a variety of resources in specific spatial patterns at the landscape level. In this study a spatially explicit wildlife behavior simulation model is coupled with forest management optimization algorithm to simulate adaptive management (feedback) for timber and wildlife benefits. Uncertainty is introduced into the system through the stochastic simulation of changes in the spatial distribution of red-cockaded woodpecker (Picoides borealis) (RCW) groups. The changes in the distribution of the RCW groups influence the decision options for forest management optimization, and forest harvest activities influence the distribution of RCW groups. This results in a feedback loop between simulated wildlife behavior and forest management. To generate near optimal forest harvest schedules in a timely manner, a genetic algorithm (GA) optimization heuristic is applied. The result is a feedback system between RCW groups and forest management actions. This modeling system (Hughell 1996) is used to evaluate adaptive management policies and optimization strategies over a 200-year planning horizon. The management policies define each stand's minimum rotation age as a function of the stand's proximity to a RCW group. The optimization strategies are volume control and the utilization of explicit 200-year near optimal harvest schedules (generated by a genetic algorithm). RCW GROUP BEHAVIOR / FOREST MANAGEMENT MODEL The Red-cockaded woodpecker (Picoides borealis) (RCW), an endangered species associated with the coastal pine forests of the southern U.S., is considered a cooperative breeder because of the manner in which it nests and forages in family groups (which consist of a breeding pair and reproductively mature helpers) (Walters et al. 1988). RCW prefer to nest in live longleaf pine (Pinus palustris) trees over 95 years of age, or in loblolly pine (P. taeda) over 75 years, although they have been found in much younger trees (Hooper 1988). 1 David A. Hughell and Joseph P. Roise, Department of Forestry, North Carolina State University, Box 8008, Raleigh, NC 27695