Research on biodiversity and ecosystem functioning has built strong disciplinary links between community ecology and ecosystem science. However, the consequences of changes in biodiversity for ecosystem functioning are still a matter of debate, especially for microbial communities. We do not know yet how variations in the structure of microbial diversity affect the productivity in a network of trophic interactions, e.g. anaerobic digesters. More efficient use of resources may result from the selection of individual performing microorganisms or from the selection of interactions within the complex network. These competing selection processes may be observed when a wastewater treatment plant community is confronted with wastewater (with its own source community) entering the plant. In bioaugmentation, specific strains are typically isolated outside their environmental context and then mixed with a naturally assembled community. It may be a more successful strategy to assemble artificial communities by mixing already naturally occurring networks instead of adding individuals. Following this idea, we here present a novel interpretation of the classic common garden experiment where we expose mixtures of different source communities to standardized environmental conditions. The source communities with their own set of well-established interactions originate from functionally redundant anaerobic digesters but differ in community structures. Before mixing, the communities were maintained in anaerobic chemostats for three months. We tested how initial diversity of the mixtures affects ecosystem performance, i.e. production of biogas from a radiation-sterilized standardized complex organic substrate.