The compositions of the humic acid (HAs) components from the surface (0-20 cm) of Amazonian Dark earth (ADE) soils under forest (ASF) were compared with those of cultivated (CAS) soils, and with HAs of adjacent non-ADE soils under forest (NAS) using thermogravimetric analysis (TGA), elemental analyses, and solid state ¹³C NMR spectroscopy (using a VARIAN INOVA spectrometer at ¹³C and ¹H frequencies of 100.5 and 400.0 MHz, respectively). Principal Component Analysis (PCA) was carried out utilising NMR data. TGA allows determinations of thermogravimetric index (TGI) data, or the ratios of the areas under the 378-623 exothermic peak (from thermal decomposition of polysaccharides and other labile structures, and under the 623-923 peak (from the combustion of aromatic structures and cleavages of C-C bonds). The TGI of HAs extracted from anthropogenic soils (ASF and CAS) was larger than those for the control soils indicating that the HAs from anthropogenic samples were more aromatic than those in the controls, and suggesting the occurrence of polycyclic aromatic nuclei which have greater thermal resistance. The HAs from the CAS were more enriched in C and depleted in H, indicating that the cultivation of these soils had led to the selective degradation of aliphatic structures and a possible relative enrichment in hydrogen-deficient, condensed aromatic structures from the older black carbon (BC) HA precursors. The HAs from the anthropogenic soils had lower N contents and higher atomic C/N ratios; however, these ratios were in the range for biologically stabilized materials. Additionally, these HAs had a lower atomic H/C ratio indicating high aromaticity and/or aromatic ring condensation. The O/C atomic ratio values were variable, and there was not a clear tendency for these for the different sample groups. The HA samples from anthropogenic soils presented higher aryl C contents, both total and substituted-C. On the other hand, the O-aryl C and carboxyl C contents were similar among the samples. The contribution of methoxyl C from lignin for the NAS HA samples was greater, indicating that the aromatic region (aryl and O-aryl C) of NAS samples have larger contributions from lignin residues. Therefore, a larger proportion of O-aryl C of NAS samples is not from phenolic groups, and consequently, is not contributing to the total acidity. The same conclusion can be drawn in relation to the carboxyl signal; that is, the NAS HAs have greater contribution from amide groups, indicated by the smaller C/N atomic ratio and the greater N-alkyl composition. The ionizable oxygenated functional groups signals that could contribute to CEC (carboxyl minus amide and O-aryl minus methoxyl C) were larger for the anthropogenic samples. The high carboxyl C content in BC derived HAs is well-known. However, artifact formation from the fractionation method cannot be discarded. The contribution of carbohydrates (O-alkyl and di-O-alkyl C) was larger for NAS. The high correlation between aryl C and TGI (R = 0.82) and aryl C and atomic H/C ratio (R = -0.89) indicate that NMR data are, at least, semi-quantitative. In addition, TGI presented negative correlations with labile groups, such as: alkyl (-0.73); N-alkyl/methoxyl (-0.82) and carbohydrates (-0.86), indicating that this index is a good tool to estimate the degree of humification.

The first principal component (PC1 – 61% of total variance) from PCA data, was characterized by positive loadings for the BC aryl signal (broad featureless signal at 130 ppm) and negative loadings for the signals of carbohydrates, methoxyl, N-alkyl, alkyl (0-100 ppm) and O-aryl (151 ppm). The positive loadings at 169 ppm can be attributed to carboxyl groups attached to BC aromatic rings, and the negative loadings at 174 ppm to amide from proteins/peptides. This PC separated the ADE samples with greater scores, emphasizing the major contributions of BC structures to HAs compositions from ADE soils, and these structures are characterized by recalcitrant (aryl) and reactive groups (carboxyl). The PC1 scores had positive correlations with TGI (0.82), O/H (0.82), and C/N (0.88) atomic ratios, and a negative correlation with the H/C (-0.90) atomic ratio. These data confirm, by independent methods, that HAs from anthropogenic soils can be characterized with regard to high stability in terms of both structural (NMR and elemental composition) and thermal properties.

The second principal component (PC2 – 21% of total variance) was characterized mainly by a broad signal at 126 ppm and other at 170 ppm. These resonances are typical for acetylated analogues of lignin units. The scores of this PC were smaller for NAS and varied within the anthropogenic soils groups.