We selected 204 non-invaders, minor-invaders,
and major-invaders to develop and validate the
new model, and compare its performance to the
Australian model using the same set of species.
Performing better than the Australian model, our new
model accurately identified 94.1% of major-invaders
and 97.1% of non-invaders, without committing any
false positives or false negatives. The new secondary
screening tool we developed reduced the number of
species requiring secondary evaluation from 22 to
12%. We expect that the new weed risk assessment model should significantly enhance the United State’s timeliness and accuracy in regulating potential weeds.