The application of unsupervised machine learning for the elucidation of the burial effect of lignin in peatland: Case of TMAH thermochemolysis

Abstract

Lignin is one of the major components of organic macromolecules in peatland. Analysis of lignin in sediments is considered as a valuable tool for understanding peat vegetation as well as carbon cycle. It provides valuable perspectives on different sides such as occurrence of various vegetation types and the extent of decay in terrestrial organic material. Thermochemolysis technique using tetramethyl ammonium hydroxide (TMAH) was used for lignin analysis, in an ombrotrophic peatland. It specifically breaks of O-aryl bonds followed by methylation of the lignin components yielding phenolic subunits. Several discrepancies exist between TMAH thermochemolysis findings and those obtained from previous investigations with CuO oxidation. This could be explained by the specific pool of aryl ether, specifically targeted by the thermochemolysis, with discarding of C-C bonds. Hence, this would show the capacity of TMAH thermochemolysis to cleave oxidised lignin leaving lignin of fresh and preserved organic matter intact. The TMAH thermochemolysis method's efficiency in molecular characterization and lignin degradation was evaluated using principal component analysis (PCA). PCA reduced dimensionality and redundancy of component contributions, with the first two principal components accounting for 70.33 % of the total variance. PCA has reinforced the selectively of the thermochemolytic approach in targeting O-aryl bonds while maintaining C-C bonding of polyphenolic sub-units. Hence, two distinct oxidation phases: recent (acrotelm and mesotelm) and older (catotelm) have been revealed. PCA was applied to diagenetic and source vegetation proxies, revealing strong agreement among variables which is supported by the use of molecular ratios as indicators of organic matter source and dynamics in a peat core.