Introduction and objectives Agricultural practices (e.g. crop rotation, tillage) lead to profound changes in soil properties, ecosystem structure (e.g. biodiversity) and functioning (e.g. ecosystem services). Whereas this has been very often characterized in the medium and long terms, little is known so far about how fast soil properties respond to changing practices at the time scale (year to several years) in which farmers take their decision in the management of their crops. In agricultural systems, increasing consideration is given to soil biodiversity, whose role has long been overlooked by agronomists, but whose preservation in now recognized as key for maintaining soil functioning capabilities, such as litter decomposition and nutrient cycling. In this context, the SOFIA project addresses the short-term impact of agricultural practices on the soil chemical and physical properties (e.g. soil C and N, pH), the taxonomic and functional diversity of soil communities and the consequences on some soil functions related to organic matter cycling (N mineralization, N2O production, C storage). Materials & methods The SOFIA experiment was held on the SOERE experimental platform of Estrées-Mons (Northern France) dedicated to the study of long-term anthropogenic disturbance on biogeochemical cycles and biodiversity. This experiment, started in 2010, is characterized by a series of treatments comprising different crop rotations (annual, perennial and energy-based crop rotations), rate of N fertilization, deep and reduced soil tillage. These practices are known to affect soil organisms by modulating the amount, chemical nature and location of their food resources and by modifying their habitat. During four years (2010-2014), the induced-differentiation of the agrosystem was characterized every year for plant biomass production and soil physical, chemical and microbiological properties (bacterial and fungal communities, nitrifying and denitrifying communities). Concomitantly the taxonomic and functional diversity of the earthworm, macroinvertebrates, microfauna and bacterial and fungal communities were determined. The CO2 and N2O emissions were continuously measured using automatic chambers. We used the Random Forest method to identify physicochemical (11 variables) and biological (56 variables) parameters which are were able to predict soil functions (organic matter decomposition, greenhouse gas emissions, primary production) as a first step toward the development of indicators in cultivated soils. General trends of global soil biodiversity were assessed based on the number (richness) and assemblage of all sampled families. Results and discussion We observed an early differentiation of the experimental treatments, notably a significant stratification of organic carbon, microbial biomass and enzymatic activities in treatments with reduced tillage compared to conventional tillage. Soil communities’ differentiation was less affected by residues management and not affected by N fertilization rate after 4 years. The N2O fluxes were low (max 2.5 kg N-N2O over 3 years), with peaks mainly after N fertilization in spring. Differentiation in N2O emissions was however mainly related to N input, with 2 to 3 times lower emissions with the reduced N fertilization. Our results showed that functional groups of nematode (omnivores and predators), the microbial biomass carbon (C) and the soil soluble carbon C contents best explained organic matter mineralization (C-CO2 and gross N mineralization ) in the top first five centimeters of soil which was the most affected by practices. We observed a slow but effective differentiation of soil invertebrate with globally higher diversity values in reduce-tillage systems. Discussion Significant changes in the soil chemical and biological properties were observed already two years after shift in agricultural practices. The main response was observed with the reduction of tillage, which led to significant modification in the stratification of soil total C and N, soil microbial C, enzymatic activities, which affected gross N fluxes, N2O emissions and functional groups of nematodes in the upper 20cm of soil. At the scale of the experiment (4 year-differentiation), no significant effect of plant residues removal (vs. recycled) and of reduced N fertilization were observed, except on N2O emissions. Soil organisms varied in their response to agricultural practices and our study highlighted that trophic interactions likely contributed to the observed dynamics. Most of the changes in soil pools and soil communities were significantly correlated to changes in soil C.