Tools for visualizing and integrating pest risk assessment ratings and uncertainties*

Abstract

The application of pest risk analysis (PRA) decision‐support schemes, such as that used by the European and Mediterranean Plant Protection Organization (EPPO), generates many ratings for likelihood or magnitude of risk factors, each with an associated uncertainty. In accordance with the international standard ISPM 11 (FAO, 2004), questions have been devised to assess the key elements of pest risk in the four main sections of pest risk assessment: Entry, Establishment, Spread and Impact. After completing each section, risk assessors are required to give a summary rating and an uncertainty score for that section. The large number of question ratings and uncertainty scores make the task of summarizing each section and its uncertainty quite difficult. Two graphical tools have been developed to aid this task: the PRA Risk score and uncertainty visualizer (Visualizer) and the Rule‐based matrix model (RBMM). The Visualizer presents a case summary graph on a single page in such a way that the risk assessors and peer reviewers can see rating scores and uncertainties in a pictorial manner; the RBMM integrates all the individual questions in the assessment through a hierarchy of rules that attempt to mimic the logic used by the assessors and are arranged in the form of a flow chart to give an overall rating with an accompanying expression of uncertainty.

A quantitative interpretation of the entry section of the EPPO decision‐support scheme for pest risk analysis

“…The EPPO DSS for PRA employs a model based on utility functions which mimic assessor logic to combine the components of risk, not only for the entry section, but also for the whole assessment (Holt et al ., , ; Kenis et al ., ; Schrader et al ., ). Here a quantitative model of the entry section was developed which included the same set of risk components as the EPPO DSS but for the purposes of this study, reduced the number of questions.…”

“…When a rating is defined explicitly as an interval it therefore carries built‐in uncertainty which, depending on the size of the interval, may be quite large. In the EPPO DSS model (Holt et al ., ), low, medium and high uncertainties correspond to the likelihood that the selected rating is correct: 90, 50 and 35% respectively; these values were adapted from IPCC definitions (IPCC, ). A different way to express differences in uncertainty is therefore needed and a possible approach is to allow some greater control of interval width, whilst still retaining a limited set of discrete choices.…”

“…11 (FAO, ), indicates the elements to be included in a PRA but does not state how the analysis should be undertaken or provide a mechanism for combining risk and uncertainty. Recent work to improve European decision‐support schemes for pest risk assessment, ‘Prima phacie’ (MacLeod et al ., , ) and for PRA, ‘PRATIQUE’ (Baker et al ., ; Baker, ), have used models with utility functions in the form of risk matrices to integrate components of risk (Holt et al ., , ). This approach is consistent with the existing risk‐rating systems and accommodates the difficulties of estimating quantitative values or probabilities in the majority of cases.…”

Model of the probability of pest transfer from imported fruit, cut flowers or vegetable produce

“…The model concerns insects and mites and focuses on the stages of transfer and the initiation of a population, defined as egg laying at a site suitable for establishment. Transfer and initiation of a population are arguably under‐represented in pest risk assessment, for example transfer occupies a single question in risk assessment models developed during the PRATIQUE project (Holt et al ., ) and in the associated Entry Section of the EPPO pest risk analysis (PRA) scheme (EPPO, ). This is usually also the most uncertain stage of pest entry, and so it is critical for judgements about entry as a whole.…”

A Regional Decision Support Scheme for Pest Risk Analysis in Southeast Asia

“…In the last decade, several studies have reviewed and enhanced the practice of PRA, but these studies have been limited to high‐income countries . Most suggest incorporating computerized quantitative techniques, where appropriate, to improve the consistency and reduce the uncertainty of risk estimation . These quantitative techniques are highly demanding in terms of data, interpretive skills, time, and effort even in high‐income countries.…”