Conditioning treatment effects on soil communities and abiotic parameters
When the cover of forbs was experimentally reduced, the relative abundance of soil pathogenic fungi decreased concomitantly (Figure 2a, Supplementary Table 4). Especially grass pathogens, dominated bySlopeiomyces cylindrosporus showed similar patterns as observed for total relative abundance of pathogens (Figures 2b,c, Supplementary Table 4). Forb-specific pathogens had a significantly higher relative abundance in one than in two-year legacies (Figure 2d, Supplementary Table 4) while the relative abundance of saprotrophs was higher in plots with two-year than one-year legacies (Figure 2e, Supplementary Table 4). Arbuscular mycorrhizal fungi were not affected by conditioning treatments (data not shown, Supplementary Table 4).
After the conditioning phase, the soil bacterial community structure was significantly affected by conditioning plant community and conditioning time (Supplementary Table 5, Supplementary Figure 7). The soil fungal community structure was significantly affected by conditioning plant community, conditioning time and by forb-grass ratio, but the effect of forb-grass ratio differed between conditioning plant communities (Supplementary Table 5, Supplementary Figure 8). The structure of soil abiotic parameters was affected by conditioning community and conditioning time (Supplementary Table 5).
Soil-mediated pathways between conditioning and responding plant communities
The composition of the conditioning plant communities significantly explained the composition of the responding plant communities (Figure 3). The fact that these two plant communities were separated in time (visualized by a dashed arrow) indicates that the effects of conditioning plant community on the responding plant communities must be mediated via the soil. The conditioning plant community significantly explained community composition of soil fungi and bacteria in the conditioning phase, but did not affect the composition of soil abiotic parameters. Importantly, microbial and soil abiotic parameters measured at the end of the conditioning phase, significantly explained these parameters measured again three months after the transition to responding phase. The soil fungal composition (but not bacteria or abiotic parameters) measured in theresponding phase correlated with the composition of theresponding plant communities.