Prioritizing Invasive Species Threats Under Uncertainty

Agricultural and Resource Economics Review, Apr 2006 by Moffitt, L Joe, Osteen, Craig D

In addition to the concerns raised by Horan et al. (2002), a serious practical limitation of risk-based models is that it may be difficult or impossible to estimate or interpret probabilities and the novel circumstances that surround the introduction and establishment of potential invasive species, particularly those that have not yet entered or those that are poorly understood. Ouchi (2004) argued that there is no formally established methodology for treating expert judgment and that Bayesian and other approaches suffer from limitations in practical application. Even in cases where a historical record may suggest likelihoods, the prospect of intentional introductions or related challenges to biosecurity may render historical records largely unusable. Accounting for the effects of human activity on new pest entry and the susceptibility of agro-ecosystems to pests, as well as the endogenous effect of mitigation and adaptation activities on probabilities characterized in Shogren (2000), will increase the complexity of probability estimation.

Developing Criteria for Decision Making Under Uncertainty

When probabilities of events are not reliably measured or are not appropriate for representing uncertainty, as discussed above, traditional criteria for decision making under uncertainty include the maximin and maximax criteria, which represent polar extremes of optimism and pessimism, and the Laplace and Hurwitz criteria, which require information similar to probabilities (Render, Stair, and Hanna 2006). Info-gap methods are also applicable (Ben-Haim 2006). We selected the minimax criterion to represent a cautious approach to decision making regarding the adverse effects of invasive species and responses to those species when probabilities of events are unknown or are inappropriate for representing uncertainty, but other criteria could be used with different results. The following discussion shows how decision rules using that criterion follow from the traditional risk model through simplifying assumptions and economic logic when probability estimates and other important information, such as the cost-effectiveness of management options, are not available. As information becomes scarcer, decision making changes from selecting optimal actions to prioritizing pests.

Uncertainty criteria are often depicted in the context of decision analysis. The risk model, as represented in Shogren (2000), can be interpreted in a decision analysis framework by characterizing the choice set as a finite number of alternatives and basic scientific uncertainty as a discrete random variable. [A decision table based on Shogren's model is shown in Moffitt and Osteen (2004).] The choice set contains J possible pairs of investments in adaptation and mitigation, ((x^sub 1^,Q^sub 1^), (x^sub 2^,Q^sub 2^), ..., (x^sub J^,Q^sub J^)). The basic scientific uncertainty concerning establishment of an introduced species, θ, is represented by a discrete random variable that takes on K different values denoted by θ^sub 1^, θ^sub 2^, ..., QK, with the probability of the kth value denoted by p(θ^sub k^). The optimal solution is the adaptation and mitigation pair corresponding to the largest EU. To develop a ranking of potential invaders for budgetary purposes, equation (1) would be solved to find the optimal action or program, (x,Q), for each potential invader, and the ranking would be based on the relative welfare improvements due to the optimal actions as compared to doing little or nothing, (0, 0). Due to large information requirements, expression (1) may be very difficult to apply to many potential invaders, but it provides a foundation for modeling preparedness under uncertainty for invasive species.