Do PSF-causing microorganisms respond more to vegetation than to soil?
Indicator OTUs causing PSF were not more strongly associated with plant community structure than with soil properties, as compared to the rest of the soil microbiome (Fig. 3). In fact, procrustean analyses revealed a very strong correlation between the predictors of microbial community structure when comparing indicator vs non-indicator prokaryotes (r = 0.86, P = 0.002). A similar result was found for fungi (r = 0.88, P < 0.001). For both prokaryotic and fungal communities, the strongest predictors were soil pH, moisture and N levels, as well as two plant community structure ordination axes associated mainly with the abundances of Bromusinermis , Nassella viridula and Carex stenophylla(plant ordination axes 2 and 3, jointly explaining ~ 34% of plant community structure, Fig. 3, Table S1). Although we had found that high-pH, high-moisture and low-fertility soils were conducive to negative PSF in these soils (Chagnon et al. , 2018), we did not find strong associations between these soil properties and pathogenic indicator OTUs in our canonical ordinations (Fig. S6). Likewise, we did not find that the proportion of fungal sequence reads belonging to pathogenic OTUs was significantly influenced by soil properties (assessed through generalized additive mixed models), except for a very weak unimodal trend with soil moisture (Fig. S7). However, this trend explained less than 5% of the variance. Collectively, these results further confirm that pathogens were not strongly associated with the decrease in plant growth observed with inocula collected from low-nutrients, high-pH and high-moisture soils (Chagnon et al. , 2018).