Microbial Functional Diversity Enhances Predictive Models Linking Environmental Parameters to Ecosystem Properties

Microorganisms drive biogeochemical processes, but linking these processes to real changes in microbial communities under field conditions is not trivial. Here, we present a model-based approach to estimate independent contributions of microbial community shifts to ecosystem properties. The approach was tested empirically, using denitrification potential as our model process, in a spatial survey of arable land encompassing a range of edaphic conditions and two agricultural production systems. Soil nitrate was the most important single predictor of denitrification potential (the change in Akaike's information criterion, corrected for sample size, ΔAIC(c) = 20.29); however, the inclusion of biotic variables (particularly the evenness and size of denitrifier communities [ΔAIC(c) = 12.02], and the abundance of one denitrifier genotype [ΔAIC(c) = 18.04]) had a substantial effect on model precision, comparable to the inclusion of abiotic variables (biotic R2 = 0.28, abiotic R2 = 0.50, biotic + abiotic R2 = 0.76). This approach provides a valuable tool for explicitly linking microbial communities to ecosystem functioning. By making this link, we have demonstrated that including aspects of microbial community structure and diversity in biogeochemical models can improve predictions of nutrient cycling in ecosystems and enhance our understanding of ecosystem functionality.