Fig. 5: The response to mixed food patches shows that effective density is additive.

a Relative number of worms predicted by the sigmoidal model for patches of DA1885 (dark blue) and CR266 (light blue), as a function of bacterial density (measured in OD). Our experiment took place at OD = 0.5 (black dashed line), where DA1885 is about 10 times more attractive than CR266 (circles). The null model for a 50:50 mixture of both strains goes from the geometric mean to the arithmetic mean of the two pure patches (black errorbar). b Relative number of worms in a food patch, as a function of its effective bacterial density (\({D/D}_{{{\mbox{attract}}}}\)). Circles: Effective density for CR266 and DA1885 at OD = 0.5. Red cross: Effective bacterial density for a 50:50 mixture of CR266 and DA1885 at OD = 0.5 (the effective density of the 50:50 mixture is the arithmetic mean of both effective densities, which in a logarithmic scale is located closer to the highest one). c Experimental scheme: Worms were exposed to 5 food patches with different fractions of CR266 and 1885, always at OD = 0.5. Worms at each patch were counted after 2 h. d Number of worms at each food patch, as a function of the fraction of DA1885 in the food patch. Dots: Experimental results (errorbars show the 95% confidence interval, computed via bootstrapping). Red line: Prediction from the sigmoidal model. Gray patch: Prediction of the null model. See Supplementary Table 1 for sample sizes and Supplementary Data 1 for the data and computer code that generate this figure.