Invasions by exotic annual grasses (EAGs) are replacing native perennials in semiarid areas globally, including the vast sagebrush‐steppe rangelands of western North America. Efforts to eradicate EAGs and restore perennials have had mixed success, especially in relatively warm and dry areas where EAGs had high dominance prior to intervention. Greater consideration of the ecological sources of variability in EAG treatment outcomes may improve success. We hypothesized that herbicide and restoration outcomes would be influenced by restoration strategy (type of herbicide, seeding or planting, timing of treatment) and underlying spatial variability associated with plant community patch type and litter, all applied in a landscape‐scale experiment in a severely invaded area in Southern Idaho, USA. EAGs, specifically medusahead (Taeniatherum caput‐medusae [L.] Nevski), were strongly reduced for up to 3 years (maximum observation period) by the pre‐emergent herbicide indaziflam, whereas the pre/post‐emergent imazapic reduced EAGs only when applied twice. Indaziflam effects were greater when post‐spray moisture was greater, and also when co‐applied with imazapic, but reapplying indaziflam did not lead to additional reduction of EAGs. Imazapic and indaziflam each stimulated species‐specific, secondary invasion by exotic and/or invasive tall forbs. Application of the broadleaf herbicide aminopyralid provided only a fleeting 1 year of control of a dominant, highly noxious forb skeletonweed (Chondrilla juncea L.). Underlying heterogeneity in plant community patch type (dominant herb species) explained only ∼5% of variation in the herbicide effects, and manipulation of litter prior to spraying had no effect. Several years of seedings and planting resulted in no establishment of native perennials. Herbicides, especially indaziflam, appear to be an effective tool for reducing EAGs for multiple years in the challenging restoration conditions we evaluated, particularly if their application coincides with suitable moisture. However, restoring the perennials required for longer term resistance to reinvasion is a serious challenge that could be avoided with preservation of perennials.

Exotic annual grass invasion into western North America has led to significant loss of native perennials, altering the structure and function of sagebrush-steppe ecosystems. Monitoring and assessment of necessary restoration treatments have provided mixed evidence of success. We hypothesized that treatment outcomes would be influenced by restoration strategy (e.g., the timing of herbicide or drill seeding) and by within-treatment vegetation heterogeneity. We evaluated exotic annual grass and exotic perennial forb response to three replicate treatments of the pre-emergent herbicides indaziflam and imazapic, and a combination treatment of both herbicides, followed with the broadleaf herbicide, aminopyralid, at a highly invaded site in Southern Idaho. A litter removal study was integrated to investigate the effects of thatch cover on herbicide application and two different revegetation methods, drill seeding and hand planting of native perennial seedlings, were nested into herbicide treatments. We accounted for vegetation heterogeneity within treatments by identifying pre-existing plant-community patch types and mapping their locations across the research site using high spatial resolution aerial imagery. We found that imazapic had no detectable effects on exotic annual grass cover, but significantly reduced exotic annual grass seedling density the first two years post-treatment. Indaziflam treatments effectively reduced exotic annual grasses for three years post-treatment, most notably the combination treatment of imazapic and indaziflam. Accounting for vegetation heterogeneity in our predictive models improved our ability to detect exotic annual grass response to treatment by a 5% change in cover. None of the drill seeded plants emerged in either the treatments or controls for the duration of this study and all but a few native seedling plantings failed, precluding any meaningful revegetation comparisons between treatments. We were also unable to detect an influence of residual thatch on herbicide outcomes but did find that precipitation played a significant role in herbicide effectiveness. Overall, our findings suggest that indaziflam can be an effective tool for reducing exotic annual grasses in restoration, particularly when combined with imazapic, and that implementation of multiple sampling methods can provide greater insight into treatment outcomes. Additionally, our results indicate that accounting for plant-community patches in predictive models can improve model accuracy and therefore our ability to detect treatment effects.