Overcoming carbonate interference on labile soil organic matter peaks for midDRIFTS analysis

Mid-infrared spectroscopy has been used with various degrees of success in quantitative and qualitative analysis of soil organic matter (SOM). This study evaluated the spectral interference caused by soil carbonates on the feasibility of using the specific peak area approach obtained by diffuse reflectance infrared Fourier transform spectroscopy in mid-range (midDRIFTS) as an indicator of SOM quality and composition. A midDRIFTS peak area corresponding to more labile SOM compounds (2930 cm−1) was related to microbial biomass carbon and nitrogen (Cmic, Nmic) and hot water extractable carbon and nitrogen (CHWE, NHWE) across two contrasting agroecological regions (Kraichgau (K) and Swabian Alb (SA), Germany) with soils containing carbonates up to 41 g kg−1 soil. Although, the 2930 cm−1 peak showed its potential use as an index for representing labile soil carbon, a poor regression performance between the relative peak area at 2930 cm−1 (rPAnt 2910 cm−1) and labile SOM fractions (R2 = 0.31–0.45) was attributed to inorganic carbon (IC) potentially interfering with this particular peak. To avoid carbonate interference, a spectral-based coefficient was developed using a multiple linear regression model consisting of TOC and TIC as predictors of the difference between peak areas at 2930 cm−1 of non-acid and acid treated spectra (ΔPA 2930 cm−1) of bulk soils. Peak areas at 2930 cm−1 of non treated spectra (PAnt 2930 cm−1) were corrected using a coefficient of 0.21 for carbonates (PAcorr 2930 cm−1), which greatly improved relationships with Cmic, Nmic, CHWE and NHWE (R2 = 0.68–0.75; P < 0.0001). We further developed a regression equation to correct PAnt 2930 cm−1 for future application using the relationship between PAnt 713 cm−1 (assigned for carbonates (mainly calcite)) and PA at 2930 cm−1 (ΔPATIC 2930 cm−1) (R2 = 0.98). We concluded that PA 2930 cm−1 has a high potential to be used as an indicator of SOM composition once carbonate interference is corrected for in carbonated containing soils. The proposed approach is free of multivariate calibration and the spectral index developed (PAcorr 2930 cm−1) is suited to predict accurately labile SOM without the need of additional laboratory measurements. For general applicability of the approach, it should be tested on additional soil types of varying TOC and TIC contents.