Increased microbial anabolism contributes to soil carbon sequestration by mineral fertilization in temperate grasslands

Ecosystem responses to nitrogen (N) additions are manifold and complex, and also affect the carbon (C) cycle. It has been suggested that increased microbial carbon use efficiency (CUE), i.e. growth per C uptake, due to higher N availability potentially increases the stabilization rates of organic inputs to the soil. However, evidence for a direct link between altered microbial anabolism and soil organic C (SOC) stocks is lacking. In this study, unfertilized (control) and NPK-fertilized (NPK) treatments of seven temperate grassland experiments were used to test the hypothesis that fertilizer-induced differences in SOC stocks (ΔSOC) cannot be explained by differences in C input alone, but that microbial anabolism plays an important role in C sequestration. At two experimental sites, microbial CUE and related metabolic parameters was determined using an 18O labeling approach at two different incubation temperatures (10 °C and 20 °C). Fertilization effects on the abundance of Bacteria, Archaea and Fungi were also determined using quantitative PCR targeting the respective rRNA genes. Due to the availability of yield and belowground biomass data, the introductory carbon balance model (ICBM) could be used for all seven sites to estimate the contribution of C input to ΔSOC. A significantly higher microbial growth (+102 ± 6%), lower specific respiration (−16 ± 7%) and thus significantly higher CUE (+53 ± 21%) was found for the NPK treatments, which was consistent across experiments and incubation temperatures and correlated with measured root C:N ratios. Growth (+49 ± 5%) and respiration (+70 ± 9%) were increased by a higher incubation temperature, but this was not the case for CUE. The fungi to bacteria ratio changed significantly from 0.18 ± 0.02 (control) to 0.09 ± 0.02 (NPK). On average, only 77% (51% when excluding one extreme site) of observed ΔSOC was explained by C inputs. The optimized humification coefficient h of the model used to fit the observed ΔSOC was strongly correlated to differences in the root C:N ratio between the control and NPK treatments (R2 = 0.71), thus confirming a link between microbial anabolism and substrate C:N ratio. Furthermore, varying h directly by observed differences in CUE improved the model fit at the two sites investigated. This study provides direct evidence that CUE of soil microbial communities is relevant for SOC sequestration, and its dependency on soil N availability or substrate C:N ratio might allow for its inclusion in models without explicit microbial C pools.