For the targeted selection of microbial communities that provide cellulose degradation, soil samples containing cellulolytic microorganisms and specific plant residues as a substrate can be used. The details of this process have not been studied: in particular, whether the use of different soils determines the varying efficiency of communities; whether these established cellulolytic communities will have substrate specificity, and other factors. To answer these questions, four soil microbial communities with different cellulolytic activity (Podzol and the soil of Chernevaya taiga) and substrates (oat straw and hemp shives) with different levels of cellulose availability were used, followed by trained communities that were tested on botrooth substrates (in all possible combinations). Based on the analysis of the taxonomic structure of all communities and their efficiency across all substrates (decomposition level, carbon, and nitrogen content), it was shown that the most important taxa of all trained microbial cellulolytic communities are recruited from secondary soil taxa. The original soil does not affect the efficiency of cellulose decomposition: both soils produce equally active communities. Unexpectedly, the resulting communities trained on oats were more effective on hemp than the communities trained on hemp. In general, the usage of pre-trained microbial communities increases the efficiency of decomposition.