Connecting microbial interactions at molecular scale to ecosystem processes using model ecosystems: How do molecular interactions between plants, mycorrhizas and microbial decomposers alter emergent ecosystem functions? The balance of energy and mass within an ecosystem is fundamentally regulated by the activity of primary producers and decomposers. These organisms do not operate in isolation - there are strong interactions between these groups, many of which are mediated by plant mycorrhizal symbionts. To understand the mechanistic nature of these interactions myself and collaborators have taken a reductionist approach, constructing model ecosystems in the greenhouse. Our team grows Pinus taeda seedlings in symbiosis with different members of the model ectomycorrhizal genus Suillus. We monitor interactions between plants, mycorrhizal fungi, and free-living microbial decomposers using metagenomics, metatranscriptomics, metabolite and redox profiling. We then trace the effect of these interactions on emergent ecosystem carbon cycle processes using 13C stable isotope tracing. This combination of approaches allows us to connect interactions within microbial communities at the molecular scale to emergent carbon cycle processes at the ecosystem scale. This work requires a diverse team with mycological experts at Duke University, mycological and molecular biological expertise at the University of Florida, and technological collaboration with the Department of Energy Joint Genome Institute and the Pacific Northwest National Labs Environmental and Molecular Science Laboratory.
Leveraging forest inventory data to understand and forecast the biogeography of mycorrhizal symbiosis: There are millions of observations of tree distributions and demographic process rates contained within forest inventory databases all over the world. Because many trees associate with specific functional subsets of mycorrhizal fungi, these forest inventory data provide an opportunity to understand the biogeography of major functional types of mycorrhizal symbionts at spatial and temporal scales not possible with molecular methods. I currently use the U.S. Forest Service's Forest Inventory and Analysis database and Bayesian statistical modeling to understand the biogeography of arbusular and ectomycorrhizal forests, as well as how their distributions are changing in response to major global change forcings. Given the spatial extent of forest inventory data, this may represent a tractable way to inject critical aspects of the forest microbiome into Earth scale carbon cycle models.
Building microbial ecology into soil and ecosystem models: Most current ecosystem models conceptualize soil microbial life as a homogenous pool of biomass, with limited ability to adjust physiology or resource allocation patterns in response to environmental change. I incorporate process-level understanding into mathematical models of decomposition to improve C and N cycling predictions. To date, I have integrated microbial community structure (Waring, Averill and Hawkes 2013, Ecology Letters), allocation strategies (Averill 2014, Ecology Letters), plant-decomposer interactions (Averill et al. 2015, Biogeochemistry) and nutrient-physiology feedbacks (Averill and Waring in press, Global Change Biology) into soil decomposition models. These analyses have repeatedly demonstrated that embedding coarse-level ecological constraints into decomposition models increases their predictive power, generating qualitatively different predictions than traditional ecosystem models. Ongoing work in this area is focused on microbial trait based ecology, ecological tradeoffs, and evolutionary game theory.
Near Term Ecological Forecasting of the Soil Microbiome: Most ecological theory is built on observational analysis and short-term experiments, making inferences based on past observations. This knowledge is then used to develop models that make ecological forecasts on decadal and centennial time scales. As a result, very few ecologists ever observe how well their forecasts perform. Near-term forecasts (seasonal to annual) inherently provide one of the strongest tests of any scientific theory – to make a prediction about what will happen before it happens, and then observe the result. Across widely disparate disciplines, research has shown that this process of gaining feedback, building experience, and correcting models and methods is critical for building a forecast capacity. Furthermore, one of the hallmarks of prediction on shorter time scales is the ability to iteratively update a prediction in light of new evidence. Near-term forecasts provide the opportunity to repeatedly cycle between making forecasts and performing analyses to learn from those forecasts, applying the Forecast-Analysis Cycle. As part of a recently funded NSF Macrosystems Biology team, I am developing Bayesian forecasting models to predict key aspects of metagenomic and microbial amplicon data generated by the National Ecological Observation Network. This represents one of the first efforts to forecast microbial composition and functional profiles in real time.
