OOS 12-5 - Quantifying stochastic variation of microbial composition and functioning in the field

Tuesday, August 8, 2017: 9:20 AM
E145, Oregon Convention Center
Michaeline Nelson Albright and Jennifer B. H. Martiny, Ecology and Evolutionary Biology, University of California, Irvine, Irvine, CA

It is increasingly recognized that both stochastic and deterministic processes drive microbial community composition. However, quantifying the role of stochastic processes in shaping microbial communities remains a challenge. Here we used a field experiment to address the following questions: 1) Is there stochastic variation in bacterial community composition even after accounting for measurement error? 2) Does stochastic variation at the community level result in stochastic variation in ecosystem functioning? To address these questions, we minimized environmental heterogeneity in a field experiment involving plant litter microbial communities. We homogenized plant litter substrate and irradiated it in replicate litterbags of the two mesh sizes (open vs. closed to microbial dispersal). We inoculated the litterbags with homogenized microbial communities and placed them in the field in close proximity to limit environmental variation. We next manipulated an environmental (deterministic) parameter, water availability (ambient vs. added water). Our sampling design allowed us to estimate how variance in measures of bacterial community composition (16S data) and functioning (extracellular enzyme activity) was distributed among the dispersal and water treatments, measurement error (within bag replicates), and residual variation (stochastic effects).


Treatment effects (water environment, dispersal, and environment-by-dispersal interactions) combined to explain only 9.6% of the estimated variation in community composition. Measurement error, or within bag variability, also contributed significantly to estimated variation, accounting for 22%. This difference between litterbag subsamples captures microscale environmental variability and measurement error during processing. Without this quantification, this variation would be mistakenly attributed to stochasticity. Even so, after accounting for treatment effects and measurement error, 68.4% of the estimated variation in community composition remained unexplained. This suggests that stochastic processes played an important role in shaping community composition in this system. In contrast to community composition, using variance partitioning we explained a larger amount of variation in ecosystem functioning. For extracellular enzyme potential, treatment effects explained an average of 72.9% of the estimated variation, while measurement error explained 12% and residual variation, 15.1%. These results suggest that stochastic variation of our metric of ecosystem functioning was attenuated compared to that of community composition. This study is the first field experiment that disentangles the role of stochastic processes, environmental selection, and dispersal on microbial community composition and functioning. We further provide the first quantification of stochastic variation in a microbial community that accounts for measurement error.