Similar Processes but Different Environmental Filters for Soil Bacterial and Fungal Community Composition Turnover on a Broad Spatial Scale

International audienceSpatial scaling of microorganisms has been demonstrated over the last decade. However, the processes and environmentalfilters shaping soil microbial community structure on a broad spatial scale still need to be refined and ranked. Here, wecompared bacterial and fungal community composition turnovers through a biogeographical approach on the same soilsampling design at a broad spatial scale (area range: 13300 to 31000 km2): i) to examine their spatial structuring; ii) toinvestigate the relative importance of environmental selection and spatial autocorrelation in determining their communitycomposition turnover; and iii) to identify and rank the relevant environmental filters and scales involved in their spatialvariations. Molecular fingerprinting of soil bacterial and fungal communities was performed on 413 soils from four Frenchregions of contrasting environmental heterogeneity (Landes,Burgundy#Brittany,,South-East) using the systematic gridof French Soil Quality Monitoring Network to evaluate the communities’ composition turnovers. The relative importance ofprocesses and filters was assessed by distance-based redundancy analysis. This study demonstrates significant communitycomposition turnover rates for soil bacteria and fungi, which were dependent on the region. Bacterial and fungalcommunity composition turnovers were mainly driven by environmental selection explaining from 10% to 20% ofcommunity composition variations, but spatial variables also explained 3% to 9% of total variance. These variableshighlighted significant spatial autocorrelation of both communities unexplained by the environmental variables measuredand could partly be explained by dispersal limitations. Although the identified filters and their hierarchy were dependent onthe region and organism, selection was systematically based on a common group of environmental variables: pH, trophicresources, texture and land use. Spatial autocorrelation was also important at coarse (80 to 120 km radius) and/or medium(40 to 65 km radius) spatial scales, suggesting dispersal limitations at these scales