Cooperative Fish and Wildlife Research Units Program: Georgia
Education, Research and Technical Assistance for Managing Our Natural Resources

Moore, C. T., C. J. Fonnesbeck, K. Shea, K. J. Lah, P. M. McKenzie, L. C. Ball, M. C. Runge, and H. M. Alexander. 2011. An adaptive decision framework for the conservation of a threatened plant. Journal of Fish and Wildlife Management: 2:247-261.


Mead's milkweed Asclepias meadii, a long-lived perennial herb of tallgrass prairie and glade communities of the central United States, is a species designated as threatened under the U.S. Endangered Species Act. Challenges to its successful management include the facts that much about its life history is unknown, its age at reproductive maturity is very advanced, certain life stages are practically unobservable, its productivity is responsive to unpredictable environmental events, and most of the known populations occur on private lands unprotected by any legal conservation instrument. One critical source of biological uncertainty is the degree to which fire promotes growth and reproductive response in the plant. To aid in its management, we developed a prototype population-level state-dependent decision-making framework that explicitly accounts for this uncertainty and for uncertainties related to stochastic environmental effects and vital rates. To parameterize the decision model, we used estimates found in the literature, and we analyzed data from a long-term monitoring program where fates of individual plants were observed through time. We demonstrate that different optimal courses of action are followed according to how one believes that fire influences reproductive response, and we show that the action taken for certain population states is informative for resolving uncertainty about competing beliefs regarding the effect of fire. We advocate the use of a model-predictive approach for the management of rare populations, particularly when management uncertainty is profound. Over time, an adaptive management approach should reduce uncertainty and improve management performance as predictions of management outcome generated under competing models are continually informed and updated by monitoring data.