Lessons learned from testing the Australian weed risk assessment system: the devil is in the details
Daphne A. OnderdonkA, Doria R. GordonB, Alison M. FoxC and Randall K. StockerC
A Department of Botany, PO Box 118526, University of Florida, Gainesville, Florida 32611, USA.
B The Nature Conservancy, Department of Botany, PO Box 118526, University of Florida, Gainesville, Florida 32611, USA.
C Department of Agronomy and Centre for Aquatic and Invasive Plants, PO Box 110500, University of Florida, Gainesville, Florida 32611-0500, USA.
Our test of the Australian Weed Risk Assessment system (WRA) in Florida and comparison of these results with those from tests of the system in other geographies reveal a number of areas where methodological variation may influence results. We demonstrate differences among the tests, such as variability in the base rate of test species and in the evidence required to answer WRA questions. We use the Florida dataset to explore the effects of elements of this variation on accuracy of the WRA and make recommendations to increase consistency in future WRA application and reporting. While we find that the overall accuracy is relatively insensitive to the variation tested, the probability of accurate prediction and comparability of results from different geographies would increase if the system were more consistently and transparently applied.
Keywords: Consistency of WRA tests, invasive species, prediction, screening, variation among WRA tests.
Plant Protection Quarterly (2010) 25 (2) 79-85.