COS 97-4
A hierarchical approach to estimating microbial diversity effects in mechanistic models of decay

Thursday, August 14, 2014: 9:00 AM
Regency Blrm D, Hyatt Regency Hotel
Brad Oberle, Biological Sciences, The George Washington University, Washington, DC
Mariya Shcheglovitova, Biological Sciences, The George Washington University, Washington, DC
Amy Zanne, Biological Sciences, The George Washington University, Washington, DC
Background/Question/Methods

Microorganisms mediate key ecosystem processes.   Mechanistic models typically represent them using fixed parameters and simple functional relationships.  In models of the carbon cycle, this approach reduces the complexity of microbial decomposition to the kinetics of idealized enzymes.  Contrary to this simplified view, environmental sequencing is revealing tremendous diversity and dynamism of communities that decompose organic matter.  Moreover, recent theory and experiments have identified how microbial diversity might produce systematic deviations from widely applied models.  To integrate functional consequences of microbial diversity into carbon cycle models, we present a conceptual framework for identifying microbial diversity effects.  We evaluate support for diversity effects using a hierarchical Bayesian model that we validate against simulated and observational data.  Finally, we test support for microbial diversity effects in wood decomposition using five years of decay data from 21 plant species species coupled with substrate quality and microbial community structure of both fungi and bacteria.

Results/Conclusions

In contrast to simplified systems based on single fixed parameters without biodiversity effects, systems with non-linear dynamics and stochastic parameters exhibit a decrease in apparent aggregate decay rate and a corresponding increase in substrate half-life.  The decrease in the apparent decay rate is related to the variance of the underlying decay rate distribution, which can be additively partitioned into substrate, environment and microbial diversity effects.   Hierarchical Bayesian models readily recover the parameters and dependencies of decay rate distributions in simulated data that include small to moderate measurement error.  They also fit long-term experimental mass loss data better than simpler models.  Among wood samples from 21 woody species in a common garden rot-plot experiment, 16 showed evidence for a decrease in aggregate decay rates, consistent with biodiversity effects.   Species that exhibited decelerating decay rates tended to have more complex anatomy but the connections to variation in fungal and bacterial richness were less clear.  While our hierarchical Bayesian approach paves a path towards incorporating some diversity effects into ecosystem process models for decay, further refinements are necessary to adequately represent more complex community dynamics such as complementarity and facilitation.