Coupling of Belowground Carbon Cycling and Stoichiometry from Organisms to Ecosystems along a Soil C Gradient Under Rice Cultivation

Abstract:

Ecological stoichiometry is a framework linking biogeochemical cycles to organism functional traits that has been widely applied in aquatic ecosystems, animals and plants, but is poorly explored in soil microbes. We evaluated relationships among soil stoichiometry, carbon (C) cycling, and microbial community structure and function along a soil gradient spanning ~5-25% C in cultivated rice fields with experimental nitrogen (N) amendments. We found rates of soil C turnover were associated with nutrient stoichiometry and phosphorus (P) availability at ecosystem, community, and organism scales. At the ecosystem scale, soil C turnover was highest in mineral soils with lower C content and N:P ratios, and was positively correlated with soil inorganic P. Effects of N fertilization on soil C cycling also appeared to be mediated by soil P availability, while microbial community composition (by 16S rRNA sequencing) was not altered by N addition. Microbial communities varied along the soil C gradient, corresponding with highly covariant soil %C, N:P ratios, C quality, and carbon turnover. In contrast, we observed unambiguous shifts in microbial community function, imputed from taxonomy and directly assessed by shotgun sequenced metagenomes. The abundance of genes for carbohydrate utilization decreased with increasing soil C (and declining C turnover), while genes for aromatic C uptake, N fixation and P scavenging increased along with potential incorporation of C into biomass pools. Ecosystem and community-scale associations between C and nutrient substrate availability were also reflected in patterns of resource allocation among individual genomes (imputed and assembled). Microbes associated with higher rates of soil C turnover harbored more genes for carbohydrate utilization, fewer genes for obtaining energetically costly forms of C, N and P, more ribosomal RNA gene copies, and potentially lower C use efficiency. We suggest genome clustering by functional gene suites might yield simplified guilds related to biogeochemical cycling, even when function is imputed directly from taxonomy. Our findings in a controlled model wetland ecosystem bolster evidence for the role of P in influencing soil C cycling, and our approach could be leveraged to reduce complex microbial data for trait-based modeling of soil C cycling.