Linguistic variables as fuzzy sets to model uncertainty in the combined efficacy of multiple phytosanitary measures in pest risk analysis

摘要

To support pest management innovation in international trade, a model is provided to calculate the efficacy of potential combinations of phytosanitary measures each of which has an uncertain effect. The efficacy of each measure was considered in terms of residual infestation, or infestation remaining (proportion), following implementation of the measure. The cumulative result is the product of that for the individual measures, described by an interval of the base variable, the proportion of infested units in the export lot of a commodity. The residual infestation associated with the measures was described by linguistic terms, Negligible to Very high, also corresponding to intervals in the base variable, calculated using three principles: 1. a consistent efficacy ratio across linguistic terms, so Very low has the same relationship to Negligible as Low to Very low, and so on; 2. a definition of Negligible based on pest detection probability; and 3. the anchoring of Medium to 50% efficacy. The intervals may overlap depending on their perceived uncertainty. The cumulative result of all measures implemented has some level of overlap with one or more linguistic terms and this indicates the extent to which the result fits concepts like Negligible, Very Low, Low, etc. With reference to a published trade case analysis of Thrips palmi control on Thai orchid exports, the degree of overlap of suggested management alternatives with the concept of Negligible infestation was calculated, illustrating how the support can be provided for judgements made by risk managers when giving opinions about the potential effectiveness of phytosanitary measures.