Figure 3 The linear regression of stability and standard richness under different proportions of network positive interactions. The shaded area indicates the 0.95 confidence interval. Fig. 3A and 3B depict the linear relationship of standard richness to resistance and resilience under higher proportions of positive interactions, respectively, while 3C and 3D were under a balanced proportion of positive interactions. P% indicate the positive proportion for all interactions. The dark dots indicate the means of bacterial or fungal stability for different ecosystems.
In order to investigate the effect of interactions on RSRs, we chose the richness and stability components (resistance and resilience) for regression analysis. However, we did not find any direct linear relationship between standard richness and stability (Fig. S5), and the R2 of regression to resistance and resilience were 0.03 and 0.05, respectively, (P > 0.05). This indicated potential drivers were governing their relationship. Next, we adopted network analysis to obtain a detailed overview of the microbial community interactions. Interestingly, the results showed that when the proportion of positive interactions was high and unbalanced (P% > 0.52), richness decreased resistance (Fig. 3A, slope=-0.22, R2=0.52), but increased resilience (Fig. 3B, slope=0.32, R2=0.76). When the proportion of positive interactions were balanced (0.48 < P% < 0.52), richness increased resistance (Fig. 3C, slope=0.92, R2=0.95) and decreased resilience (Fig. 3D, slope=-0.97, R2=0.65). These results showed that the relationship between richness and stability was governed by the balance between positive and negative interactions, which highlighted that the organizational pattern of community members plays an important role in influencing the relationship between the size of species pool and system stability.