Novel approaches to plant pest risk assessment

Zhu, Lihong (2009) Novel approaches to plant pest risk assessment. PhD thesis, University of Greenwich.

Preview

PDF - Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (16MB) | Preview

Abstract

Pest risk assessment is an essential yet problematic stage in pest risk analysis (PRA) that concerns the likelihood and consequences of pest introduction. The aim of this study was to develop methodologies for risk assessment and to explore different approaches that could lead to the development of new methods for practical PRA in line with the requirement of "scientific justification" by World Trade Organisation and Food and Agriculture Organisation of the United Nations.

Current international practices were discussed and research reviewed on qualitative and quantitative approaches to risk assessment. It was proposed that risk assessment be divided into two steps: Pest risk identification (PRI) and pest risk evaluation (PRE). Mind Mapping was a valuable tool for PRI that reduced ambiguity and increased transparency. Approaches to PRE were proposed that facilitated the scoring and weighting of risk factors, and the subsequent combining of risk scores. Several methods were developed to incorporate weighting into PRA, which included subjectively assigned weighting and Delphi technique-derived weighting. Metrics for combining risk scores into an overall risk value were also explored, compared and evaluated.

Correlation and interaction between risk factors were analysed, which revealed that some risk factors were highly correlated and some were relatively independent, which meant there was some information redundancy, and therefore simplification of risk assessment was possible. Cluster analysis was applied to risk factor scores and different clusters of risk factors were identified: some more appropriate for preliminary assessment; some for determining the level of risk; and some could be eliminated.

A method to apply Principal Components Analysis (PCA) to derive weighting for individual risk factors was developed. PCA could be applied to historical data of pest introductions, previous PRA cases, or expert opinion. Genetic algorithms implemented in the software BEAGLE, were applied to PRA data. The rules obtained could distinguish high-risk situations with high accuracy, which was useful in predicting the risk of an organism by using a simplified set of conditions.

The results showed that weightings and rules differed for different taxonomic groups. Therefore it was implausible to develop a generic scheme in this way. However, it may be possible to develop patterns based on taxonomy. The results of applying several different techniques all suggested that by grouping risk factors for different purposes, risk assessment could be simplified without compromising rigor, because a) some factors were redundant; b) some factors are more important than others; and c) high risk situation could be predicted with a few key factors.

Actions (login required)