In agricultural studies, estimating and testing differences between treatment regimes in a designed experiment usually determines the subsequent success or failure of inputs or management practices. Fine-scale soil variability of the experimental site can reduce statistical power and should be determined in order to optimize the experiment design. Although randomization is often performed, there is a probability that such randomization does not cover for fine scale soil variability. To determine the effect of this fine scale soil variability, a well characterized test system comprising of biochar's effects on soil microbial communities were studied, in conjunction with Electrical Resisitivity Tomography (ERT) to identify its role on biochar's differential interaction with soil microbial communities. Two main soil profiles were identified by the ERT survey and soil characteristics and soil microbial communities were differently affected by the biochar application with regards to soil profile. Fungal communities (mycorrhization intensity and fungal richness in particular) seemed to be more strongly affected than bacterial communities, however, these effects didn't show a similar trend in between the two studied soil profiles. Consequently, most of the effects of biochar application were masked by the soil variability of the experimental site when analysed in a nested design with randomization. Our results support the relevance of taking fine scale soil variability into account prior to the establishment of a field trial to optimize experimental design. (Résumé d'auteur)