Knowledge of microbial mechanisms is critical to understanding Earth’s biogeochemical cycle under climate and environmental changes. However, large uncertainties remain in model simulations and predictions due to the lack of explicit parameterization with microbial data. The availability of high throughput sequencing and associated metagenomic technologies is a revolution in microbial ecology, but it is challenging to integrate microbial community information into ecosystem models to improve their predictive ability. Here we address three questions: (i) how can we incorporate microbial information into ecosystem modeling? (ii) what are the costs and benefits to assimilate microbial data into ecosystem modeling? and (iii) if the inclusion of dynamic microbial data (e.g., microbial biomass, gene abundances or enzyme densities/activities) in model parameterization could reduce model uncertainties and therefore improve confidence in model simulations? We explicitly represent microbial and enzymatic functions in an ecosystem model with flexible stoichiometry. We apply the model to multiple laboratory and field datasets from different ecosystems. Model performance and uncertainties are evaluated by parameterizing the model with or without microbial information. Long-term model simulations beyond the calibration period are also conducted to examine the difference in model projections of soil C dynamics with or without microbial information.
Results/Conclusions
We show that explicit representation of microbial physiology and functions in ecosystem model can improve our mechanistic understanding of microbially-driven soil biogeochemical processes, such as the priming effect and carbon-nutrient interactions. Microbially-enabled ecosystem models has the capability to achieve better model performance in soil respiration simulations compared with non-microbial models. Unlike the forcing data (e.g., soil temperature and moisture), microbial data may not be directly used to drive model simulations due to low-frequent measurements. Microbial data, in addition to traditional measurements (e.g., respiration and soil carbon), can be used as additional observations to constrain the model. Using microbial data as additional constraints or objective functions might not improve the goodness-of-fit of soil respiration. However, model parameters could be better constrained with microbial information, which means reduced uncertainty in parameter estimates and more accurate and reasonable modeling of carbon and nutrient dynamics. We advocate that the inclusion of dynamic microbial data in model parameterization could reduce model uncertainties and therefore increase confidence in model projections of C cycle in response to climate and environmental changes.