Soil microbial communities play a major role in soil fertility and are important as bioindicators to evaluate the impacts of agricultural practices on soils. The challenge is to promote their use in order to support / accelerate innovation by proposing to farmers the tools necessary to assess the impacts of practice modifications on the soil microbial composition before implementing and validating them. The goal of this thesis is to characterize spatial and temporal dynamics of soil microbial communities at the scale of an agricultural landscape in order to create models and propose a Decision Support Instrument (DSI) for stakeholders in the agricultural communities. This will use a multi-disciplinary approach combining microbial ecology, environment assessment, Computer analysis and modeling, associated to a participatory approach which effectively mobilize stakeholders’ knowledge (scientists, agricultural advisors and growers). This study will be based on data collected in Fénay agricultural landscape (1200 ha, Dijon, 21, France) where soil microbial communities were characterized in 2011 and 2016 and farming practices are known. The genericity of the model will be evaluated on the experimental network and scientific watch data which covers a large pedological diversity with the same methodology that was used for Fénay landscape.