Fig. 4: Effects of mixed planting on herbivore damage in silico and in situ.

a Effect size estimates for pairwise mixed planting among 199 A. thaliana genotypes. Positive and negative values indicate associational resistance and susceptibility to herbivore damage, respectively. b Estimated herbivore damage under virtual mixture conditions plotted against that under monoculture conditions. A single dot corresponds to a single genotype. The y-axis shows the average of estimated effect sizes among the 198 counterpart genotypes for each focal genotype. The blue line and gray area indicate linear trends with standard errors alongside Spearman’s \(\rho\) (see Results for the exact \(p\)-value). c Simulated damage plotted against the number of randomly selected genotypes. Circles and bars indicate mean \(\pm \,\)SD among 9999 iterations of the simulations (see “Methods”). d Herbivore damage by flea beetles on the three pairs of genotypes under monoculture (white) or mixture (gray) conditions at the Zurich field site. The y-axis represents the number of leaf holes divided by the initial plant size (no. /cm). Asterisks indicate significant differences in marginal means between the monoculture and mixture conditions: * \(p\) < 0.05 and ** \(p\) < 0.01 (see Supplementary Table 4 for the list of test statistics and \(p\)-values). No. of plants: \(n\) = 224 for Bg-2, Uod-1, Vastervik, and Jm-0 under the monoculture; \(n\) = 112 for Bg-2, Uod-1, Vastervik, and Jm-0 under the mixture; \(n\) = 223 for Bla-1 and Bro1-6 under the monoculture; and \(n\) = 111 for Bla-1 and Bro1-6 under the mixture. Box plots visualize the median with upper and lower quartiles, with whiskers extending to 1.5 \(\times\) inter-quartile range. The estimated effect sizes of the three pairs tested in panel d are labeled above the histogram of panel a.