Ecosystem models commonly use stable‐state assumptions to predict responses of soil microbial functions to environmental change. However, past climatic conditions can shape microbial functional responses resulting in a ‘legacy effect’. For instance, exposure to drier conditions in the field may shape how soil microbial communities respond to subsequent drought and drying and rewetting (DRW) events. We investigated microbial tolerance to low moisture levels (‘resistance’) and ability to recover after a DRW perturbation (‘resilience’) across a steep precipitation gradient in Texas, USA. Although differences in precipitation regime did not result in differences in resistance and resilience of soil microbes, microbial communities appeared to be generally resilient and resistant across the gradient, suggesting that frequent exposure to drought had characterised the trait distributions of microbial communities. Moreover, microbial communities from historically drier sites used carbon more efficiently during a DRW perturbation suggesting that long‐term drought history leaves a legacy effect on microbial functions. This may have been due to an indirect effect of drought caused via precipitation‐induced differences in primary productivity, influencing the availability of soil organic matter to microbes. Alternatively, different exposures to drought might have shaped the microbial ‘readiness’ to cope with the DRW disturbance. Microbial community composition was also linked to drought history, but was unrelated to variation in function. Synthesis. Exposure to drought can have both direct and indirect effects on soil microbial communities, which can result in lasting legacy effects on the functions they control.

Soil microbiome is one of the most heterogeneous biological systems. State‐of‐the‐art molecular approaches such as those based on single‐amplified genomes (SAGs) and metagenome assembled‐genomes (MAGs) are now improving our capacity for disentailing soil microbial‐mediated processes. Here, we analysed publicly available datasets of soil microbial genomes and MAG's reconstructed from the Amazon's tropical soil (primary forest and pasture) and active layer of permafrost, aiming to evaluate their genome size. Our results suggest that the Candidate Phyla Radiation (CPR)/Patescibacteria phyla have genomes with an average size fourfold smaller than the mean identified in the RefSoil database, which lacks any representative of this phylum. Also, by analysing the potential metabolism of 888 soil microbial genomes, we show that CPR/Patescibacteria representatives share similar functional profiles, but different from other microbial phyla and are frequently neglected in the soil microbial surveys. Finally, we argue that the use of MAGs may be a better choice over SAGs to expand the soil microbial databases, like RefSoil.