Fig. 4

Reconstructions of metabolic networks during bacterial succession. A Principal component analysis (PCA) of the metabolic capabilities of the recolonization samples. Each sample contains the metabolic pathway abundances that were derived from inferred metabolic pathways combined with 16S rRNA gene abundance data. Colors indicate the time point of the sample. B Pathways associated with early (2d, 7d) and late (14d, 28d) time points were summarized to subsystems. The filling indicates the early vs. late colonizers mean log2 fold change of the pathway abundances at early and late time points. C Carbohydrate degradation pathways separating early vs. late colonizers from random forest feature selection. The log2 fold change of mean pathway abundances at early (2d,7d) and late (14d, 28d) time points is shown together with the importance score from random forest feature selection (Boruta). D Time series of chitin degradation-associated pathway abundances. The pathway abundance indicates the distribution of pathways among colonizing bacteria (based on 16S relative abundances). Time series of pathway abundances for E nitrogen and F sulfur cycle-associated pathways