Data Analyses
For all further analyses, we used R Version 4.0.3 (R Development Core Team; http://www.r-project.org). In general, models were compared by sequentially deleting terms and comparing model fits using F-tests or χ2-tests (where appropriate), after which pairwise contrasts were computed using the emmeans’ packages , with α < 0.05. We checked residual behaviour using the ‘DHARMa’ package . All plots were produced using the ‘ggplot2’ package .
To test for the effect of evolution treatment on temporal changes in gas production, we used a linear mixed model (LMM) with treatment × time as fixed explanatory categorical variables, as well as their interaction. To account for non-independency of observations over time, we fitted random intercepts for each reactor. Based on the obtained simulation-based residual plots, we included a dispersion parameter for levels of treatment and time, using the ‘glmmTMB’ function in the‘glmmTMB’ package . Based on the full model, we calculated pairwise treatment contrasts for each week, adjusted for multiple testing using the ‘tukey’ method in the emmeans package.
To look at the impact of the 1% transfer on ecological changes, we looked at (1) dissimilarity of linked pre-adapted versus adapted communities (i.e. those that received 1% enrichment from a linked pre-adapted community, n = 12 samples), (2) dissimilarity of pre-adapted and adapted communities that were not directly linked through 1% transfer (n = 132 non-linked samples) and (3) dissimilarity of the pre-adapted versus control communities (n = 144 samples).
We also simulated potential impact of the 1% enrichment by in silico adding a 1%-rarefied pre-adaptation samples to the control treatments and comparing their composition with the original control samples via PERMANOVA (adonis).
To determine which taxa differed in abundance in the control versus evolution treatments, we fitted a negative binomial GLM to the sequencing data using the ‘DESeq’ function in the R package ‘DESeq2 ’ . Focussing on the 100 most common taxa, we calculated significant differences in the abundance of taxa using Wald tests and corrected P –values for multiple testing using the ‘frd’ method.