Soils harbor large, diverse microbial communities critical for local and global ecosystem functioning that are controlled by multiple and poorly understood processes. In particular, while there is observational evidence of relationships between both biotic and abiotic conditions and microbial composition and diversity, there have been few experimental tests to determine the relative importance of these two sets of factors at local scales. Here, we report the results of a fully factorial experiment manipulating soil conditions and plant cover on old‐field mesocosms across a latitudinal gradient. The largest contributor to beta diversity was site‐to‐site variation, but, having corrected for that, we observed significant effects of both plant and soil treatments on microbial composition. Separate phyla were associated with each treatment type, and no interactions between soil and plant treatment were observed. Individual soil characteristics and biotic parameters were also associated with overall beta‐diversity patterns and phyla abundance. In contrast, soil microbial diversity was only associated with site and not experimental treatment. Overall, plant community treatment explained more variation than soil treatment, a result not previously appreciated because it is difficult to dissociate plant community composition and soil conditions in observational studies across gradients. This work highlights the need for more nuanced, multifactorial experiments in microbial ecology and in particular indicates a greater focus on relationships between plant composition and microbial composition during community assembly.