We used microbial lipid analysis to analyze microbial biomass and community structure during 6 years of experimental treatment at the Jasper Ridge Global Change Experiment ( JRGCE), a long-term multi-factor global change experiment in a California annual grassland. The microbial community fingerprint and specific biomarkers varied substantially from year to year, in both control and experimental treatment plots. Possible drivers of the variability included plant growth, soil moisture, and ambient temperature. Surprisingly, background variation in the microbial community was of a larger magnitude than even very significant treatment effects, and this variation appeared to constrain responses to treatment. Microbial communities were mostly not responsive or not consistently responsive to the experimental treatments. Both arbuscular mycorrhizal fungi biomarker abundance (16 : 1 ω5c) and the fungal to bacterial ratio were lower under nitrogen addition in most years. Bacterial lipid biomarker abundances (15 : 0 iso and 16 : 1 ω7c) were higher under nitrogen addition in 2002, the year of largest microbial biomass, suggesting that bacteria could respond more to nitrogen addition in years of better growth conditions. Nitrogen addition and warming led to an interactive effect on the Gram-positive bacterial biomarker and the fungal to bacterial ratio. These patterns indicate that in California grassland ecosystems, microbial communities may not respond substantially to future changes in climate and that nitrogen deposition may be a determinant of the soil response to global change. Further, year-to-year variation in microbial growth or community composition may be important determinants of ecosystem response to global change. [ABSTRACT FROM AUTHOR]