Fig. 2: Precipitation legacy effects on the soil microbiota are resilient to short-term perturbations.

a, Box plots showing phenotype distributions of T. dactyloides plants grown during the conditioning phase, in which they experienced 5 months of drought (+D) or well-watered (+W) conditions and were grown in soils with either low-precipitation legacies (drought legacy (DL); brown) or high-precipitation legacies (PL; blue) (for additional phenotypes, see Extended Data Fig. 4b,d). PL and DL indicate the baseline soils before the initiation of conditioning-phase treatments. Box edges represent the first and third quartiles, whiskers indicate the range of data points that fall within 1.5× the interquartile range of the first and third quartiles, and the centre lines indicate the medians. The letters indicate pairwise group differences based on ANOVA (N = 93 plants) followed by a two-tailed Tukey post hoc test with FDR correction (q < 0.05). b, Left: constrained ordination of soil metagenome taxonomic composition following the conditioning-phase treatments. Differences in β-diversity are shown across soil precipitation legacy groups. Group differences were assessed using PERMANOVA with 9,999 permutations, with R² and corresponding P values indicated on the plot. The bar on the left represents the percentage of variance explained by the experimental variables. Right: enrichment patterns of precipitation biomarker taxa in response to the different treatments, relative to the pre-conditioning baseline. Rectangles outlined in black indicate bacterial markers that were significantly enriched (red) or depleted (blue) (q < 0.1). c, Constrained ordination of soil metatranscriptome profiles after the conditioning-phase treatments. Bray–Curtis dissimilarity matrices were calculated from RNA-based bacterial counts to assess transcriptional differences across soil precipitation legacy groups. Differences among groups were tested using PERMANOVA with 9,999 permutations; R² and P values are shown on the plot. The bar on the left indicates the proportion of variance explained by the experimental variables. d, Principal coordinates analysis of standardized pairwise genetic distances calculated from SNPs in the genomes of the precipitation biomarker taxa. Genetic variation between groups was assessed using analysis of molecular variance, with significance tested by permutation (Φ and P values are shown on the plot). To quantify genetic differentiation among groups, fixation index (F_ST) values were also calculated and are displayed in the plot. For b–d, note that even after 5 months of experimental perturbation, there is a clear separation of the samples on the first axis based on the precipitation legacy. H, host; Rep, biological replicate.