Nitrous oxide is a potent greenhouse gas and its production is mediated by the soil microbial processes of denitrification and nitrification. A thorough understanding of denitrification drivers is necessary to accurately predict and manage nitrous oxide emissions. However, studies disagree on the utility of quantifying the denitrifier community to predict denitrification rates. This study examines the influence of nitrite reductase gene (nirK and nirS) abundance on denitrification rates across a topographically sloping region. The study was carried out over different seasons (autumn, winter, spring, summer) and topographic positions (shoulder and backslope) within a field cropped to spring wheat. A footslope cropped to winter wheat was included in certain analyses to introduce additional variation at a broader scale. We measured denitrification enzyme activity (DEA) and basal denitrification (BD), nirK and nirS abundance, bacterial and archaeal ammonia monooxygenase gene (amoA) abundance, and soil environmental and chemical characteristics which included gravimetric water content, pH, electrical conductivity, and concentrations of nitrate (NO3-N), ammonium (NH4-N), total nitrogen, total soluble nitrogen, total carbon and soluble non-purgeable organic carbon (NPOC). Stepwise multivariate regression (SMR) models of DEA and BD were performed using the measured soil characteristics and denitrifier and nitrifier abundance as explanatory factors to determine whether the microbial community size influenced prediction of denitrification activity. Two SMR models were generated: one SMR model was that with the greatest R2 from utilizing 1–3 significant variables (P < 0.05) and another model was developed solely from a few proposed primary drivers of denitrification, which we described as soluble NPOC, total soluble N, and denitrifier abundance. The SMR models were also generated at two different landscape scales (with and without footslope). We found nirS abundance to be a significant explanatory variable (P < 0.05) of DEA rates when footslope samples were included in the analysis, explaining 9% and 16% of variation depending upon whether soil gravimetric water content (greatest R2 model) or soluble NPOC (proposed drivers model) was accepted as the primary explanatory variable, respectively. This was however lower than the variance explained by environmental variables (53% and 37%, respectively). When only shoulder and backslope samples were included in the analysis, nirS abundance was only a significant explanatory variable of DEA rates when soluble NPOC (proposed drivers model) was the primary explanatory variable. The proposed drivers models that included soluble NPOC, soluble total N, and nirS population size consistently produced robust predictive models of DEA at the two different field scales (R2 = 0.53 and 0.70 compared to our greatest-R2 models of R2 = 0.61 and 0.79, with and without footslope inclusion, respectively). The nirK abundance was insignificant in the stepwise multivariate regression models of DEA, although it was significantly correlated to DEA (r = 0.35, P = 0.003) when footslope samples were included in the analysis. The nirK + nirS abundance was found to be a significant explanatory variable of BD rates (explaining 5% of variance) with footslope inclusion, but total soluble nitrogen concentration explained much more variance (43%). Although we found that denitrifier abundance can be a significant predictor of denitrification rates at the field scale, we conclude that measurements of soil moisture, carbon and nitrogen availability are more useful. Additionally, our study gives evidence that greater landscape variation results in a stronger correlation between denitrification rates and denitrifier abundance, indicating that measurements of the denitrifier community are likely important and useful predictors of denitrification rates at larger scales.
Read full abstract
Dec 14, 2019
Kukuh Murtilaksono + 7
Open Access
