Three experimental designs for analysis and prediction of plant invasions from the traits of alien species and the traits of potentially invasible ecosystems were compared. Two designs that involve manipulation, i.e., mimicking conditions experienced by invaders in a growth chamber and simulating invasion in synthetic communities, are most useful for testing specific mechanisms of invasions, as well as for identifying traits of potential importance. Field designs without manipulation are more suitable to test the value of these traits for predictive frameworks.

Additional index words: Invasibility, invasion success, invasiveness, synthetic ecosystems.