Soil–atmosphere exchange of H2 is controlled by gas diffusion and the microbial production and oxidation activities in soil. Among these parameters, the H2 oxidation activity catalyzed by soil microorganisms harboring high affinity hydrogenase is the most difficult variable to parameterize because it is influenced by many unknown edaphic factors that shape microbial community structure and function. Here we seek to formulate a model combining microbiological and physicochemical variables to predict the H2 oxidation rate (u) in soil. Soil sample replicates collected from a grassland and three forests exhibited different H2 oxidation potentials. We examined the microbial community structure based on ribotyping ... More
Soil–atmosphere exchange of H2 is controlled by gas diffusion and the microbial production and oxidation activities in soil. Among these parameters, the H2 oxidation activity catalyzed by soil microorganisms harboring high affinity hydrogenase is the most difficult variable to parameterize because it is influenced by many unknown edaphic factors that shape microbial community structure and function. Here we seek to formulate a model combining microbiological and physicochemical variables to predict the H2 oxidation rate (u) in soil. Soil sample replicates collected from a grassland and three forests exhibited different H2 oxidation potentials. We examined the microbial community structure based on ribotyping analysis, the relative abundance of high affinity H2-oxidizing bacteria (HOB) estimated by qPCR and soil physicochemical characteristics as predictors for u. A single linear regression parameterized by total carbon content and a multiple linear regression using total carbon content and HOB relative abundance in soil explained 66 and 92% of the variance in u, respectively. Microbial community composition based on 16S rRNA gene pyrosequencing profiles was not a reliable predictor for u. Indeed, we found that HOB are members of the rare biosphere, comprising less than 1% of total bacteria as estimated by qPCR. We confirmed this relationship of u with total carbon content and HOB by an independent soil survey of 14 samples collected from maize monocultures, grasslands, deciduous forests and larch plantations. Observations made from both soil surveys thus were combined to build a predictive model for u parameterized with total carbon content and HOB relative abundance. Our results show that molecular biogeochemistry is a potential approach to improve performance of classical H2 surface flux models which estimate u empirically without considering variation in HOB distribution and activity in soil.
