Likelihood and consequences: reframing the Australian weed risk assessment to reflect a standard model of risk

Curtis C. DaehlerA and John G. VirtueB

A Department of Botany, University of Hawai'i, Honolulu, HI 96822, USA.

B Department of Water Land and Biodiversity Conservation, Adelaide, South Australia 5001, Australia.


Whereas standard risk assessment models express risk as a function of likelihood and consequences, the Australian weed risk assessment (WRA) system expresses risk as a single score based on the answers to 49 questions, with no explicit differentiation of likelihood and consequences. We identified WRA questions that most closely reflect likelihood of spread and those that reflect consequences of spread (impacts) to determine whether explicit consideration of likelihood and consequences could reduce WRA mis-classifications or provide insights into why species may be misclassified. Data from a previously published test of the Australian WRA system in Hawai'i were reanalysed. As expected, most major weeds had high scores for both likelihood and consequences components of risk, while both scores tended to be low for non-weeds. Major and minor weeds did not differ in terms of consequence scores, but major weeds averaged higher likelihood scores. A composite score obtained by multiplying likelihood and consequences was better at identifying weeds than the original WRA score while correctly identifying non-weeds at the same rate as the original WRA score. Explicit consideration of likelihood and consequences is a promising approach to improve WRA, but questions remain about non-independence of these components.

Keywords: Impacts, likelihood, risk assessment, spread.


Plant Protection Quarterly (2010) 25 (2) 52-55.