Fig. 1: Relationships between the soil microbiome and forest abiotic and biotic conditions.

Soil was collected from 285 forest monitoring plots across 15 European countries participating in the International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP Forests) network (a). Forests were classified as either broadleaves or conifers based on the dominant tree types at each site (≥50% cover). Fungal (b, d) and bacterial (c, e) community compositions in the organic (b; n = 209; c; n = 255) and mineral (d; n = 195, e; n = 266) soil horizons. Sample size variations result from samples unable to be amplified for fungal and/or bacterial profiling or those not meeting quality control standards of sequencing depth. Vectors show correlations with forest abiotic and biotic variables with principal coordinate analysis (PCoA) axes 1 and 2 where each environmental variable is predicted by PCoA axes 1 and 2 using multiple regression. Solid lines with black labels show significant correlations (P ≤ 0.05) while dashed arrows with gray text show non-significant correlations. The significance of fitted vectors and factors was tested using 999 permutations of environmental variables. Squared correlation coefficients for each variable with respect to PCoA axes 1 and 2 for fungi (panel with left-orientation bars) and bacteria (panel with right-orientation bars) (f). N dep. nitrogen deposition, MAT mean annual temperature, MAP mean annual precipitation, C stocks soil carbon stocks.