Conditioning treatment effects, via soil, on responding plant communities
Soil legacies that were created by conditioning treatments influenced responding plant communities. The forb:grass ratio in sowing treatments in the conditioning phase resulted in grass and forb covers that differed significantly from each other (Figure 1a, b). This, in turn influenced the relative abundance of grasses and forbs in the responding plant communities. Specifically, grass abundance in the responding communities was lower in plots with a legacy of higher grass abundance (Figure 1c), while forb abundance in theresponding communities was significantly lower in plots with a legacy of higher forb abundance (Figure 1d). The pattern did not depend on plant community identity and was observed in each of the six experimental plant communities (Supplementary Figures 1,2). Furthermore, the observed grass and forb cover in the conditioning phase and the responding phase at the subplot level, showed the same significant relationships (Supplementary Figure 3).
Conditioning time (i.e. 1 or 2 years) affected the total cover of theresponding plant communities, with higher total cover in the plots during the responding phase after a two-year conditioning legacy (mean cover ~80% vs. ~90%, Supplementary Table 2).
There were significant main effects of conditioning plant community, forb:grass ratio and conditioning time on the responding plant community structure (Supplementary Table 3). The effects of forb:grass ratio strongly differed between the six different conditioning plant communities, indicated by a significant interaction between the two (Supplementary Figure 4). The forb:grass ratio significantly affected responding plant community structure in three out of six conditioning communities. In the affected communities (Supplementary Figure 5), responding species of a specific functional type were often negatively associated with their own respective abundance in the conditioning phase.
Conspecific and heterospecific soil legacy effects were assessed and visualized using correlation plots including conditioning andresponding plant species (Supplementary Figure 6). There were only a limited number of (predominantly positive) conspecific effects, and these effects )were not consistent between the six plant communities. For instance, we observed positive conspecific relationships for Rumex acetosella (in community 4),Clinopodium vulgare (in community 5), Taraxacum officinaleand Holcus lanatus (both in community 6). Only one negative conspecific relationship was observed, for Anthoxanthum odoratum(in community 5). Furthermore, there were heterospecific relationships between conditioning plant species and other respondingplant species in each of the experimental communities (Supplementary Figure 6). Finally, there were conditioning plant species that had a strong effect on cover of specific functional types in theresponding phase. For instance, in community 2, cover in theconditioning plant community of Achillea millefoliumpositively - and H. lanatus negatively - affected grass cover in the responding plant communities. In community 3, A. millefolium and Briza media cover in the conditioningplant community negatively affected forb cover. In community 5,Festuca ovina cover in the conditioning plant community, negatively affected responding grass cover (Supplementary Figure 6).