Global climate and atmospheric changes may interact in their effects on the diversity and composition of natural communities. We followed responses of an annual grassland to three years of all possible combinations of experimentally elevated CO 2 (+300 µL/L), warming (+80 W/m 2 , +∼1°C), nitrogen deposition (+7 g N·m –2 ·yr –1 ), and precipitation (+50%). Responses of the 10 most common plant species to global changes and to interannual variability were weak but sufficiently consistent within functional groups to drive clearer responses at the functional group level. The dominant functional groups (annual grasses and forbs) showed distinct production and abundance responses to individual global changes. After three years, N deposition suppressed plant diversity, forb production, and forb abundance in association with enhanced grass production. Elevated precipitation enhanced plant diversity, forb production, and forb abundance but affected grasses little. Warming increased forb production and abundance but did not strongly affect diversity or grass response. Elevated CO 2 reduced diversity with little effect on relative abundance or production of forbs and grasses. Realistic combinations of global changes had small diversity effects but more marked effects on the relative dominance of forbs and grasses. The largest change in relative functional group abundance (+50% forbs) occurred under the combination of elevated CO 2 + warming + precipitation, which will likely affect much of California in the future. Strong interannual variability in diversity, individual species abundances, and functional group abundances indicated that in our system, (1) responses after three years were not constrained by lags in community response, (2) individual species were more sensitive to interannual variability and extremes than to mean changes in environmental and resource conditions, and (3) simulated global changes interacted with interannual variability to produce responses of varying magnitude and even direction among years. Relative abundance of forbs, the most speciose group in the community, ranged after three years from >30% under elevated CO 2 + warming + precipitation to <12% under N deposition. While opposing production responses at the ecosystem level by different functional groups may buffer responses such as net primary production (NPP) change, these shifts in relative dominance could influence ecosystem processes such as nutrient cycling and NPP via differences between grasses and forbs in tissue chemistry, allocation, phenology, and productivity. Corresponding Editor: S. Smith