The microbial landscape: soil microbiome properties predict plant species distributions

Plant species distributions are shaped by interactions with the abiotic and biotic environment. Despite the known importance of soil microbiomes in shaping plant diversity and function, no study has explicitly determined the ability of the soil microbiome to predict plant species distributions at large scales. We employed paired above- and belowground surveys of plant occurrence and soil microbial taxa and functions across habitat patches (n = 676), applying machine-learning-based distribution modeling to identify the relative influence of the soil microbiome and environmental attributes in predicting plant species (n = 50) distributions across the landscape. We discovered that while abiotic gradients of known importance in this ecosystem were the strongest predictors of many plant species’ distributions, microbial predictors could have similar or greater influence. Microbiome predictors were collectively more important than abiotic environmental variables for 38% of plant species in this study and explained >70% of the predicted distribution for one species. We identified four microbiome attributes of landscape-scale importance for predicting plant species distributions including prokaryotic richness, fungal richness, the abundance of fungal pathogens, and the abundance of prokaryotic phosphate transport genes in soil. Our findings reveal a previously underappreciated role of the soil microbiome in shaping plant species distributions at a landscape scale, with implications for plant community structure in the context of both ecosystem restoration and future global change.THESE DATA are the files used in the data analysis for the manuscript, including species distribution model output with variable importance measures per plant species, as well as metadata used in statistical analysis, and community data files for 16S (bacteria/archaea) and ITS (fungi) with taxonomy ("gap_..._table_taxonomy"). The R script for data analysis and figure creation is also published here.