Comparing the outputs of five weed risk assessment models implemented in Australia: are there consistencies across models?

Lynley M. Stone and Margaret ByrneA

Department of Environment and Conservation, Science Division, Locked Bag 104, Bentley Delivery Centre, Western Australia 6984, Australia, and Future Farm Industries Co-operative Research Centre, The University of Western Australia M081, 35 Stirling Highway, Crawley, Western Australia 6009, Australia.

A Corresponding author.


Weed Risk Assessment models are increasingly used as decision support tools to prioritize weed species for management. Several models are implemented in Australia, but have not previously been compared for consistency of outputs. This study aims to determine if the outputs of four post-border models are comparable with each other and with the predictive outputs of a Border model. Each post-border model determines weed risk by combining three assessment criteria: Invasiveness, Impacts and Distribution. A common set of 24 species were assessed for weed risk to natural ecosystems through four post-border weed risk assessment models and the Australian Border model. Test species were ranked from highest to lowest weed risk to enable a comparison of model outputs. Pearson’s correlation co-efficient and the co-efficient of variation were calculated for all outputs.

Significant positive correlations were observed between the overall outputs of the post-border models. When compared individually, the outputs from the Invasiveness and Impacts criteria also showed significant positive correlations. The distribution criterion was a source of great variability in outputs, producing negative correlations. Outputs from the post-border models also showed significant positive correlations with those from the Border model. The area in greatest need of further development is determining the potential distribution of a species.


Plant Protection Quarterly (2011) 26 (1) 29-35.