Historical and ongoing species losses and invasions have altered the diversity and dominance of plant functional traits found in most California grasslands.  As a result, these grasslands are now dominated by early-season exotic annual grasses and have lost members with late-season, deep-rooted, perennial, or branching/ prostrate growth traits.  We are exploring the hypothesis that non-random species losses and compositional changes affect ecosystem functioning more than expected by chance because species response traits (vulnerability) are positively correlated with effect traits (such as spatial and temporal patterns of resource use).  We propose that restoration of both ecosystem functions and native species communities can be enhanced by understanding the dynamics that control functional group composition and diversity as communities dis- and re-assemble.  At both sandstone and serpentine grassland sites in California, nested subset analyses indicate that species disappear in non-random sequences along gradients of declining species richness.  As richness declines, species with certain functional traits tend to decline and disappear early.  We explored the implications of these non-random species and functional losses in sandstone grassland for resistance to invasion by yellow starthistle (Centaurea soltitialis), a noxious exotic that continues to spread throughout the western US.  We found vulnerable native species with similar resource use traits to the invader to be disproportionately effective at suppressing invader performance.  With microcosm and field experiments, we are now exploring opportunities to restore grassland and enhance resistance to yellow starthistle through low-density seeding of one such species, the native tarweed Hemizonia congesta.  Targeted restoration of species with key functional contributions to desired ecosystem services is one of many ways that attention to species traits and community dynamics can enhance restoration practice.