Consistent predictors of microbial community composition across spatial scales in grasslands reveal low context-dependency

Environmental circumstances shaping soil microbial communities have been studied extensively. However, due to disparate study designs, it has been difficult to resolve whether a globally consistent set of predictors exists, or context-dependency prevails. Here, we used a network of 18 grassland sites (11 of those containing regional plant productivity gradients) to examine i) if similar abiotic or biotic factors predict both large-scale (across sites) and regional-scale (within sites) patterns in bacterial and fungal community composition, and ii) if microbial community composition differs consistently at two levels of regional plant productivity (low vs high). We found that bacteria were consistently associated with certain soil properties and both bacteria and fungi were consistently associated with plant community composition within different sites. Moreover, there was a microbial community signal that clearly distinguished high and low-productivity soils that was shared across different grasslands independent of their location in the world. In this study, we show that there is high congruence between predictors of bacterial and fungal community composition at different spatial scales and that regional productivity differences are typified by characteristic soil microbial communities across the grassland biome. These results suggest that it might be feasible to predict the overall effects of global changes on soil microbial community composition in different grasslands, as well as to discriminate fertile from infertile systems using generally applicable microbial indicators.Funding provided by: Fonds Wetenschappelijk OnderzoekCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100003130Award Number: Funding provided by: European Research CouncilCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100000781Award Number: ERC-SyG-610028 IMBALANCE-PFunding provided by: Methusalem funding *Crossref Funder Registry ID: Award Number: Funding provided by: GINOP project*Crossref Funder Registry ID: Award Number: 2.3.2-15-2016-00019Data were collected from 18 Herbaceous Diversity Network (HerbDivNet) grassland sites located in 12 countries. Each of the 18 sites contained between two and six 8 × 8 m plots: 11 sites contained six plots, one site contained four plots, one site three plots and five sites contained two plots; giving a total of 83 plots. Most sites were chosen to represent a site-specific gradient in productivity based on their plant biomass production; with six plots (two replicates of low, medium and high productivity) located within the same region with little to no variation in climatic conditions. However, some sites contained fewer plots and did not show a prominent productivity gradient. A clear gradient in biomass production was captured in 11 sites. For each plot within a site, five soil subsamples to a depth of 10 cm were taken from four corners and the centre of the plot. Subsamples for microbial analyses were analysed separately, and soil properties were analysed from one composite sample per plot (n=83). Plant species presence and total aboveground biomass were measured from each m2 of each 64 m2 plot in a single event at the peak of the growing season. Mean annual precipitation (MAP) and temperature (MAT) were derived from the CHELSA database on the geographical position (latitude and longitude) of each plot. Data on total inorganic nitrogen deposition [kg/ha/yr] were derived from Ackerman et al. (2018). We analysed 14 soil properties: soil organic matter (SOM), total nitrogen (N), total carbon (C), total phosphorus (P), available P, base saturation (BS), cation exchange capacity (CEC), pH, soil texture (sand, clay, silt), extractable Ca, Mg and K. The bacterial 16S V4 region was amplified using the 515F-806R primer pair and the fungal ITS1 region was amplified using general fungal primers ITS1 and ITS2. The abundance of bacterial and fungal gene copies per sample was quantified using qPCR targeting the 16s V4 region for bacteria and the 18s region for fungi