Fig. 1: Distributions of fitness effects during mutualism and competition.

A Interactions between S. enterica (S), E. coli (E), and M. extorquens (M) can be switched from mutualism (left) to competition (right) by changing the media. Central dotted inset–S. enterica population is composed of a transposon insertion (gene knockout) library. The circle within each cell represents a DNA chromosome–each ‘X’ represents a transposon insertion site (knockout). The library was tested in monoculture, 2-species co-culture, and 3-species co-culture (rightmost box). Plate color indicates carbon source—yellow = galactose; blue = lactose; orange = succinate. B Histograms of gene-level fitness effects. The fitness effect of each gene is the average across five replicates within a given treatment. Fitness effect, as described in the Materials and methods, is the normalized fitness where a value of zero indicates a neutral knockout. The y-axis scaling is log₂-transformed. C Means of the fitness effects over all genes within a replicate. Each point is the mean of one replicate’s fitness effects. The bold horizontal line is the mean of these means. The mean fitness effect of knockouts in S monoculture was significantly (p = 2e-10) negative. There was a significant increase in the mean fitness effect when S depended on E (p = 2e-7) or M (p = 1e-5). Dependence on both E and M did not additively increase mean fitness effect, and instead lowered the mean below that for dependence on either species (p = 1.8e-6). The mean fitness effect of knockouts in competitive treatments was significantly below zero (intercept p < 2e-16) and the community members had no effect (smallest p = 0.144).