Fig. 1

Pulmonary rehabilitation modulates the microbiota composition of people with COPD, but does not affect alpha diversity. A Global rate of change of beta diversity was estimated by subtraction of baseline values at all timepoints. Blue and grey loess lines were fitted over the data points representing beta-diversity relative to baseline along time in the intervention and control groups, respectively. The blue and grey areas represent the 0.95 confidence intervals. (M0 = immediately prior PR; M1, M2, M3 = 1, 2, 3 months after initiating PR; M4, M5 = 1 and 2 months after finishing PR). Intervention and control group presented different microbiota dynamics over the 5-month follow-up (LMER on Weighted Unifrac distance matrix, Group:Time-Point, p < 0.001). The grey rectangle indicates the time interval spanning PR. B Principal coordinate analysis of Weighted Unifrac distance between groups in each timepoint. (M0, M1, M2, M3, M4 and M5). Significant differences in microbiota composition between groups were observed in M1 (PERMANOVA, F = 2.27, p = 0.038). Blue and grey dots and ellipses represent intervention and control groups, respectively. The biplot represented in M1 show the top ASVs contributing for sample segregation (5608c3e6c9de9ceb79610e7786bd0ac4: Veillonella, d0b698c7298bf04110a6d2f220879bfb: Prevotella melaninogenica, e27680d4009f98f30248d823bc17fb8e: Haemophilus parainfluenzae, a5189f77a2cfeab3bc1602ff5c8ac3e9: Streptococcus). C No pre-post differences were observed in alpha diversity (Phylogenetic index and Shannon diversity index) in the Intervention group (Wilcoxon test: phylogenetic diversity: W = − 85. p = 0.48; Shannon diversity index: W = − 1 p > 0.99). *p < 0.05, **p < 0.01, ***p < 0.001