Fig. 3: Fungal functional and taxonomic components of species richness and composition significantly linked to tree growth.

Generalized additive model coefficients showing the slope of the linear relationship between tree growth and fungal richness or fungal composition that is independent of all other co-variables (a). The standardized model coefficient (bars) and standard error (error bars) are shown so each predictor is on the same scale but note that the effect size of richness and composition are on different scales and should not be directly compared. Separate models were made for each fungal functional group. ‘Pure’ refers to analyses with fungi only identified to one versus multiple functional categories and asterisks (*) indicates a significant effect (*P ≤ 0.05, **P < 0.01; see Supplementary Table 4 for full statistical model summaries, including exact P-values). Correlation between endophyte fungal richness and tree growth rate (b; n = 112). The plotted line shows the linear correlation, shaded area around the line shows the 95% confidence interval, and the r value is the Pearson correlation coefficient. The top biotrophic fungal indicator species of variation in tree growth rates (c). The top five positively and negatively correlated ectomycorrhizal fungi and all endophytic fungal indicators are visualized. We only show the top ten ectomycorrhizal fungi because there were too many ectomycorrhizal indicator species to fit in one graphic (see Supplementary Data 2 for a complete list). And we only show ectomycorrhizal fungi and endophytes because both are the major biotrophic fungal group in European forest soils, and their compositions both predicted tree growth rates. Bars with a genus level designation are OTUs that could not be identified at the species level. For example, there are two Russula OTUs with opposite directional effect sizes distinguished as species (spp) 1 and 2. Indicator species were identified as those having significant differential relative abundances based on the negative binomial distribution. Values are reported on a logarithmic scale to base 2 and represent changes in relative abundance for a unit change in tree growth (bars) and their standard error (error bars). Volcano plot showing the strength of all endophytic and ectomycorrhizal fungal OTUs significantly correlated to tree growth (d). Values less and more than 0 indicate negative and positive correlations with tree growth, respectively.