Ranking metapopulation extinction risk: From patterns in data to conservation management decisions

Viability analyses of large metapopulations are often hampered by difficulties in the parameter estimation. This leads to high uncertainty in parameter values and model outputs and complicates the formulation of clear recommendations for conservation management. We present a comprehensive. procedure that is able to process spatiotemporal patterns of metapopulation occupancy to rank management scenarios. The first step of the procedure involves the. formulation of the stochastic metapopulation model and the estimation of parameter Values with a Bayesian approach, using a Markov chain Monte Carlo algorithm. In the second step, the model is used to predict the effects of different management actions, taking into account the uncertainty in the parameter estimates. Finally, in the third step, decision analysis is used to evaluate and aggregate the results of the previous step into a simple rank order of management scenarios. The procedure was applied to a metapopulation of the Glanville fritillary, Melitaea cinxia. Although the amount of available occupancy data was considerable, the uncertainty in the estimated model parameter values was so large that a precise estimate of the extinction risk of the metapopulation could not be made. However, the procedure was able to produce a rank order of management scenarios that was extraordinarily robust to, the uncertainty. Application of the procedure to two other case studies revealed that, even though robust rankings cannot always be obtained, the results of the procedure are helpful in assessing the degree of uncertainty in the ranking and, pointing to those factors most responsible for the lack of robustness. The results of this paper demonstrate very clearly, by way of example, both the-limitations and the possibilities of model-based metapopulation viability analysis.