Functional Redundancy Secures Resilience of Chain Elongation Communities upon pH Shifts in Closed Bioreactor Ecosystems

For anaerobic mixed cultures performing microbial chain elongation, it is unclear how pH alterations affect the abundance of key players, microbial interactions, and community functioning in terms of medium-chain carboxylate yields. We explored pH effects on mixed cultures enriched in continuous anaerobic bioreactors representing closed model ecosystems. Gradual pH increase from 5.5 to 6.5 induced dramatic shifts in community composition, whereas product range and yields returned to previous states after transient fluctuations. To understand community responses to pH perturbations over long-term reactor operation, we applied Aitchison PCA clustering, linear mixed-effects models, and random forest classification on 16S rRNA gene amplicon sequencing and process data. Different pH preferences of two key chain elongation species—one Clostridium IV species related to Ruminococcaceae bacterium CPB6 and one Clostridium sensu stricto species related to Clostridium luticellarii—were determined. Network analysis revealed positive correlations of Clostridium IV with lactic acid bacteria, which switched from Olsenella to Lactobacillus along the pH increase, illustrating the plasticity of the food web in chain elongation communities. Despite long-term cultivation in closed systems over the pH shift experiment, the communities retained functional redundancy in fermentation pathways, reflected by the emergence of rare species and concomitant recovery of chain elongation functions.


S17
Table S1.Linear mixed-effects model results for diversity of order one ( 1 D).We considered time and pH as the fixed effects, and bioreactor as the random effect.

S18
Table S3.Linear mixed-effects model results for richness.We considered time and pH as the fixed effects, and bioreactor as the random effect.

FigureFigure S6 .
Figure S1.α-Diversity rarefaction curves.ASVs of all samples were rarefied to an equal sequencing depth of 21,389 reads.Colors represent the different samples.

Figure S10 .
Figure S10.Core time-dependent taxa of individual pH levels.Using relative abundance data of ASVs of both bioreactors, a Microbial Temporal Variability Linear Mixed Model (MTV-LMM) was applied to identify time-dependent taxa of each individual pH level, whose abundance can be predicted based on the previous microbial community composition.As described, the time-explainability is denoted as the temporal variance explained by the microbial community in the previous time points.The time-explainability P-values: P*** < 0.001 < ** < 0.01 < * < 0.05.

Figure S11 .
Figure S11.Co-occurrence network for the entire period of reactor operation.Edges indicate the significant (P < 0.05) correlations.Edge thickness reflects the strength of the correlation.Size of each ASV node is proportional to the mean relative abundance over the whole period.ASV nodes are colored and grouped by family."Others" include the ASVs belonging to families Eubacteriaceae (ASV015),

Table S2 . Linear mixed-effects model results for evenness of order one ( 1
E).We considered time and pH as the fixed effects, and bioreactor as the random effect.a Variance of pH b Variance of bioreactor [treatment of bioreactor B] c Covariance of pH and bioreactor (random intercept)

Table S4 . Linear mixed-effects model results for the relative abundance of Clostridium IV ASV008 at the different pH levels.
We considered time and pH as the fixed effects, and bioreactor as the random effect.
a Variance of pH b Variance of bioreactor [treatment of bioreactor B] S19 c Covariance of pH and bioreactor (random intercept)

Table S5 . Linear mixed-effects model results for the relative abundance of Clostridium sensu stricto ASV009 at the different pH levels.
We considered time and pH as the fixed effects, and bioreactor as the random effect.
a Variance of pH b Variance of bioreactor [treatment of bioreactor B] c Covariance of pH and bioreactor (random intercept) S20

Table S6 . Linear mixed-effects model results for microbial community composition that is represented by the PC1 from the Aitchison distance-based principal component analysis.
We considered time and pH as the fixed effects, and bioreactor as the random effect.
a Variance of pH b Variance of bioreactor [treatment of bioreactor B] c Covariance of pH and bioreactor (random intercept) S21

Table S7 . Linear mixed-effects model results for microbial community composition that is represented by the PC1 from the Bray-Curtis distance-based principal coordinate analysis.
We considered time and pH as the fixed effects, and bioreactor as the random effect.
a Variance of pH b Variance of bioreactor [treatment of bioreactor B] c Covariance of pH and bioreactor (random intercept)

Table S9 . Metagenome-assembled genomes (MAGs) with the same taxonomy as ASVs.
a Taxonomy refers to the Genome Taxonomy Database (GTDB) phylogenomic classification b Accession numbers refer to the European Nucleotide Archive (ENA)