Fig. 3: S. Tm growth in drug-disrupted communities is modulated by the effect of treatment on the close niche competitor E. coli ED1α.

a, Growth of S. Tm in Com20 compared with E. coli-containing Com21 across 240 drug–concentration combinations. The regression line is indicated in red. Conditions with a large difference in the growth of S. Tm between communities (log₂[fold change] of ≥3.5 or ≤−3.5) are highlighted. Conditions followed up by 16S rRNA sequencing are shown in gold and pink. Numbers following the text labels indicate drug concentrations in µM. The Spearman’s correlation coefficient is between the log₂[fold change] in S. Tm luminescence of Com20 and Com21. b, Relative abundance of E. coli ED1α in Com21 scaled by the total community biomass (OD). Pink points correspond to treatments with lower S. Tm growth in Com20 than in Com21 and gold points indicate treatments with higher S. Tm growth in Com20 than in Com21 from a; grey points correspond to untreated Com21. Red lines represent the mean ± 1 s.e.m. P values from two-sided t-tests. c, Spearman’s correlation coefficient of the growth of S. Tm in stool-derived communities from eight donors (1–8) compared with Com20 or Com21 across multiple treatments. Red lines represent the mean ± 1 s.e.m. across all stool-derived communities. d, Growth of a fumarate respiration-impaired ΔfrdD S. Tm mutant relative to WT S. Tm in an in vitro challenge assay, involving exposure to only E. coli ED1α. Red lines represent the mean ± 1 s.e.m. P values are from one-tailed Wilcoxon test. e, Growth of S. Tm ΔfrdD relative to WT in Com20, Com21 or a human donor-derived community (stool-derived) after treatment with E. coli-targeting streptozotocin and E. coli-sparing tiratricol. Red lines represent the mean ± 1 s.e.m. Adjusted P values with Benjamini–Hochberg correction from one-tailed Wilcoxon tests.

Source data