Drivers of seasonal and interannual change in the soil microbiome

Zoey R. Werbin, Biology, Boston University, Boston, MA, Michael C. Dietze, Earth and Environment, Boston University, Boston, MA and Jennifer M. Bhatnagar, Department of Biology, Boston University, Boston, MA

Background/Question/Methods: The soil microbiome contributes many different ecosystem services, yet we currently know very little about how the soil microbiome changes over time or what the drivers of that change may be. These basic scientific questions can finally be addressed thanks to the infrastructure of the National Ecological Observatory Network, which has dozens of sites across the United States, each collecting soil microbiome samples multiple times per year. We leverage this infrastructure to understand patterns of change within soil microbiomes from months to years and to tease apart the ecological forces influencing these changes. We focus on fungi and bacteria at multiple taxonomic resolutions, as well as functional groups representing capabilities for nutrient cycling or stress tolerance. In predicting the temporal patterns of microbial groups, we evaluated ecological factors that have long been hypothesized as important controls of the soil microbiome, including: soil moisture and temperature, soil chemistry and micronutrients, long-term climate, plant community, plant phenology, and land cover type. We used Dirichlet regression to model time-series relative abundance data, and reversible-jump Markov-chain Monte Carlo (RJMCMC) to identify covariates with substantial predictive power.

Results/Conclusions: We found high intra-annual variability for microbial communities, with periodic fluctuations in abundances of groups such as animal and plant pathogens, nitrogen cyclers, and decomposers. We found that the predictive power of ecological factors varies by taxonomic resolution, with models fitting the best at lower taxonomic resolutions (phylum and class). Fungi and bacteria showed broad differences in predictor importance, with plant-related predictors significant for the majority of fungal taxa. Our statistical models represent a formalized understanding of how complex ecological factors shape dynamic microbial communities.