Extended Data Fig. 2: Parameter estimates and diagnostics plots for the most parsimonious LMM testing the effect of co-culturing on relative mutator fitness (v) through time (Fig. 1a main text).
(a) Table with estimated regression parameters ± 95% confidence intervals (CI), t-statistics and p values. Parameter estimates are on log2-scale. The minimal adequate model was arrived at by sequentially deleting terms and comparing model fits using χ2-tests. The most parsimonious models could be best described by a LMM with co-culturing as sole fixed explanatory variable (treatment: χ2(1) = 5.28, p = 0.02, transfer: χ2(3) = 3.94, p = 0.27 and 2-way interaction: χ2(3) = 1.61, p = 0.66) and random intercepts fitted for each population (estimated value for σPopulation is 1.20). Because we computed a single pairwise contrast (monoculture versus co-culture), we did not adjust p values for multiple testing. (b-c) DHARMa simulation-based residual plots indicate no misspecification in the fitted LMM: (b) QQ-plot showing no significant overall deviations from the expected distribution, (c) Boxplot demonstrating homoscedastic dispersion of residuals across community treatments. Boxplots depict the median, the lower and upper quartiles (the 25th and 75th percentiles), with whiskers extending to 1.5x the interquartile range.
