Simultaneous measurement of soil organic and inorganic carbon: evaluation of a thermal gradient analysis

Abstract

The best method for determining soil organic carbon (SOC) in carbonate-containing samples is still open to debate. The objective of this work was to evaluate a thermal gradient method (ThG), which can determine simultaneously inorganic carbon (SIC) and SOC in a wide range of soil samples. The determination of SOC by ThG (SOCThG) was compared to the following widespread standard methods: (1) acidification (ACI) as pretreatment and subsequent dry combustion (SOCACI) and (2) volumetric quantification of SIC by a calcimeter (CALC) and subtraction of the total carbon content as determined by dry combustion (SOCCALC). Precision (F test) and bias (t test) were tested on a subset of seven samples (n = 3). Comparison of the ThG and CALC methods was performed by regression analysis (n = 76) on samples representing a wide range of SOC (5.5 to 212.0 g kg−1) and SIC (0 to 59.2 g kg−1) contents. Tests on the replicated subset showed that the precision of ThG was not significantly different from ACI or CALC (F values < 39, n = 3) for SOC and SIC measurements. However, SOCACI and SOCCALC contents were systematically and significantly lower compared to SOCThG contents. The positive bias for SOCThG relative to SOCCALC contents appeared also in the regression analysis (given numbers ± standard errors) of the whole data set (y = (4.67 ± 0.70) + (0.99 ± 0.01)x, R 2 = 0.99, n = 76). When performing a regression with carbonate-free samples, the bias between the methods was negative (−2.90 ± 0.63, n = 29) but was positive in the set with carbonate-containing samples (3.95 ± 1.41, n = 47). This observation corroborated the suspicion that the use of acid for carbonate decomposition can lead to an underestimation of SOC. All methods were suitable for differentiation between SIC and SOC, but the use of acid resulted in lower estimates of SOC contents. When comparing soil samples with different carbonate concentrations, the use of the ThG method is more reliable.