Field experimental evidence that stochastic processes predominate in the initial assembly of bacterial communities.

To assess the relative importance of environmental selection, dispersal and stochastic processes in structuring ecological communities, we conducted a bacterial community assembly experiment using microcosms filled with sterile liquid medium under field conditions in the Inner Mongolian grasslands. Multiple replicate microcosms containing different carbon substrates were placed at nine locations across three spatial scales (10, 300 and 10 000 m distance between locations) in such a way that the environment of microcosms varies independently of the geographical distance. The operational taxonomic units within the experimental communities were assessed via the terminal restriction fragment length polymorphism techniques on the 10th and 17th days after the onset of the experiment. We found no evidence of distance decay in community similarity, and communities within a given location were more similar to each other regardless of environment than communities at other locations within the same spatial scale. Variance partitioning indicated that location explained more compositional variation in microbial communities than environment, particularly on the 17th day, despite that environment and location in combination could only explain less than half of the total variation. These results suggest that bacterial dispersal is not limited by distance in this experiment, and community assembly in microcosms is not environmentally determined but governed by stochastic processes.