In a collaborative study, we monitored plant species composition following control of non-native Tamarix trees along southwestern U.S. rivers using defoliation by an introduced biocontrol beetle, and three physical removal methods: mechanical using saws, heavy machinery, and burning. Physical removal favored secondary invasions immediately after Tamarix removal (0–3 yrs.), while in the biocontrol treatment, secondary invasions manifested later (> 5 yrs.). Within this general trend, the response of weeds to control was idiosyncratic; dependent on treatment type and invader. Two annual tumbleweeds that only reproduce by seed (Bassia scoparia and Salsola tragus) peaked immediately after physical Tamarix removal and persisted over time, even after herbicide application. Acroptilon repens, a perennial forb that vigorously reproduces by rhizomes, and Bromus tectorum, a very frequent annual grass before removal that only reproduces by seed, were most successful at biocontrol sites, and progressively spread as the canopy layer opened. These results demonstrate that strategies to control Tamarix affect secondary invasions differently among species and that time since disturbance is an important, generally overlooked, factor affecting response.