Résumé
Summary Soil aggregates are critical to soil functionality, but there remain many uncertainties with respect to the role of biotic factors in forming aggregates. Understanding the interacting effects of soil, land use type, vegetation and microbial communities is a major challenge that needs assessment in both field and controlled laboratory conditions, as well as in bulk and rhizosphere soils. To address these effects and their feedbacks, we first examined the influence of soil, root and litter characteristics along a land use gradient (ancient woodland, secondary woodland, grassland, pasture and arable land) on microbial community structure (in both bulk and rhizosphere soil), as well as on aggregate stability. Then, we performed an inoculation experiment where we extracted soil columns from the arable and secondary woodland and used a third unstructured loamy soil as a control. We sterilized these three soils to remove microbial communities, and then either inoculated the tops of sterilized soil columns with soil from the secondary woodland or the arable field sites. Control columns of all soil types were not inoculated. In a fully-crossed design, we planted two species possessing distinct root system morphological traits: Brachypodium sylvaticum (fibrous system with many thin and fine roots) and Urtica dioica (taproot system with few fine roots). After four months, microbial communities (in bulk and rhizospheric soil) and aggregate stability were measured, along with root traits. In both the field and laboratory experiments, bacterial (16S) and fungal (ITS) biodiversity was determined using high throughput sequencing. In the field study we found that: i) there were strong relationships between aggregate stability and microbial community composition that were driven by land use, ii) the relationship between aggregate stability along the land use gradient and the trophic nature of bacterial communities was not significant, but that certain soil, root and litter parameters shaped bacterial phyla, with oligotrophic bacteria conditioned by the rhizosphere niche, and copiotrophic phyla more dependent on bulk soil conditions, iii) land use gradient (from woodland to arable), reduced the relative abundance of saprotrophic and ectomycorrhizal fungi with an increase in the relative abundance of Ascomycota and a reduction in the relative abundance of Basidiomycota. In the laboratory experiment we found that: i) the inoculation of sterilized soils with soils from the field significantly increased aggregate stability in control soil that was initially poorly structured, ii) the effects of inoculation on aggregate stability were similar when either secondary woodland or arable soils were used as inoculums and iii) these effects were affected significantly by root length density. Our results show that microbial communities influence soil structure and that bacterial communities are intimately associated to rhizospheric conditions and root traits (of which root length density was the most pertinent).