Tractable practices for soil microbial restoration in tallgrass prairies reclaimed from agriculture are a critical gap in traditional ecological restoration. Long‐term fertilization and tilling permanently alter soil bacterial and fungal communities, requiring microbe‐targeted restoration methods to improve belowground ecosystem services and carbon storage in newly restored prairies. These techniques are particularly important when restoring for climate‐ready ecosystems, adapted to altered temperature regimes. To approach these issues, we conducted a multi‐factorial greenhouse experiment to test the effects of plant species richness, soil amendment and elevated temperature on soil microbial diversity, growth, and function. Treatments consisted of three seedlings of one plant species (Andropogon gerardii) or one seedling each of three plant species (A. gerardii, Echinacea pallida, Coreopsis lanceolata). Soil amendments included cellulose addition, inoculation with a microbial community collected from an undisturbed remnant prairie, and a control. We assessed microbial communities using extracellular enzyme assays, Illumina sequencing of the bacterial 16S rRNA gene, predicted bacterial metabolic pathways from sequence data and phospholipid fatty acid analysis (PLFA), which includes both bacterial and fungal lipid abundances. Our results indicate that addition of cellulose selects for slow‐growing bacterial taxa (Verrucomicrobia) and fungi at ambient temperature. However, at elevated temperature, selection for slow‐growing bacterial taxa is enhanced, while selection for fungi is lost, indicating temperature sensitivity among fungi. Cellulose addition was a more effective means of altering soil community composition than addition of microbial communities harvested from a remnant prairie. Soil water content was typically higher in the A. gerardii treatment alone, regardless of temperature, but at ambient temperature only, predicted metagenomics pathways for bacterial carbon metabolism were more abundant with A. gerardii. In summary, these results from a mesocosm test case indicate that adding cellulose to newly restored soil and increasing the abundance of C4 grasses, such as A. gerardii, can select for microbial communities adapted for slow growth and carbon storage. Further testing is required to determine if these approaches yield the same results in a field‐level experiment.