Combining N2:Ar and Bayesian Methods to Quantify Underestimation and Uncertainty of Sediment Denitrification Determined by the Acetylene Inhibition Method

Abstract

Accurately determining denitrification rates is essential for estimating the nitrogen budget in flooded ecosystems. The acetylene inhibition method (AIT) is a simple and common technique for measuring sediment denitrification. However, incomplete inhibition of nitrous oxide reduction or concurrent inhibition of nitrification may lead to underestimations of denitrification rates, whereas the N2:Ar ratio method with membrane inlet mass spectrometry (MIMS) provides an accurately way quantifying denitrification rates. The objective of this study was to compare sediment denitrification rates measured by AIT and MIMS, correspondingly quantify the underestimation and the uncertainty of AIT-measured denitrification by combining a Bayesian hierarchical method. A multilevel Bayesian framework was constructed to predict the true values of sediment denitrification based on the relationships between MIMS and AIT-measured denitrification. A Markov chain Monte Carlo (MCMC) algorithm was applied to obtain the targeted parameters and their associated uncertainties in the Bayesian. Results showed that both AIT and MIMS detected similar trends in denitrification responses to increasing NO3¯-N concentrations, but the values of denitrification measured by AIT were considerably lower than those of MIMS. These differences decreased rapidly with increasing NO3¯-N concentrations, then stabilized when NO3¯-N was greater than 2 mg N L−1 and AIT estimations of denitrification were about 21 times less than those of MIMS. The results were used to obtain a multilevel Bayesian model that quantified the underestimation and uncertainty of the AIT-based denitrification. We concluded that the AIT can be used to accurately estimate soil denitrification after quantification of system error and correction coefficients.