Bridging Rock-Eval® thermal signature and Van Soest chemical composition of litters

Assessing the chemical composition of soil organic matter (OM) inputs is essential to understand and simulate soil OM decomposition dynamics and thus better evaluate the soil contribution to the global C balance. The Van Soest (VS) chemical extraction distinguishes four fractions of the OM namely “soluble”, “hemicellulose-”, “cellulose-” and “lignin-”like compounds. However, this procedure is time-consuming, costly, and requires acids, solvents and detergents. The Rock-Eval® (RE) thermal analysis consists of pyrolysis of the sample followed by oxidation of the residue. Since the 2000s, it has been increasingly used in soil science to quantify soil organic carbon (SOC) via the TOC parameter and to characterize SOC thermal stability, a proxy of biological stability, through various indices. Although several studies have investigated the relationship between SOC thermal stability and the chemical composition of soil OM, it has to be consolidated in order to identify and quantify the main components of OM, especially for OM of litters. In this study, we further explored this relationship by comparing the chemical composition (VS fractions) and the RE-derived parameters determined on various agricultural and forestry litters. The RE signals obtained before (bulk sample) and after the VS extraction (lignin-like compound residue) were compared. The effect of the VS extraction was reflected on the RE signals of the lignin-like compound residue. The thermolabile compounds emitted during the pyrolysis were lost and the CO and CO₂ signals obtained during the oxidation resembled those of pure lignin. Correlation matrix between the RE parameters and the VS fractions were performed. The cellulose and hemicellulose proportions were positively correlated to the hydrocarbon compounds (HC) emitted below 340 °C (named A1, Spearman coefficient = 0.75, p-value < 0.05) and the CO₂ emitted during pyrolysis (named S3CO₂, Spearman coefficient = 0.76, p-value < 0.05), respectively. These results confirmed the previous assumptions that the A1 proportion and the S3CO₂ signal are tightly related to the amount of carbohydrates in OM. The lignin proportion was positively correlated to the temperature at which 50 % of the total HC signal is emitted (Spearman coefficient = 0.75, p-value < 0.05). Additional insights about these relationships will be provided by Fourier-Transform Infra-Red spectroscopy (FTIR). A multivariate modelling approach will be developed to explore the predictive performance of RE data for estimating the chemical composition of OM assessed with VS fractions and with FTIR.