European Journal of Soil Science vol:60 issue:6 pages:956-965
Most methods to fractionate natural dissolved organic matter (DOM) rely on sorption of acidified DOM samples onto XAD-8 or DAX-8 resin. Procedural differences among methods are large and their interpretation is limited because there is a lack of calibration with DOM model molecules. An automated column-based DOM fractionation method was set up for 10-ml DOM samples, dividing DOM into hydrophilic (HPI), hydrophobic acid (HPOA) and hydrophobic neutral (HPON) fractions. Fifteen DOM model components were tested in isolation and in combination. Three reference DOM samples of the International Humic Substances Society were included to facilitate comparison with other methods. Aliphatic low-molecular-weight acids (LMWAs) and carbohydrates were classified as HPI DOM, but some LMWAs showed also a partial HPO character. Aromatic LMWAs and polyphenols partitioned in the HPOA fraction, menadione (quinone) and geraniol (terpenoid) in HPON DOM. Molecules with log K-ow > 0.5 had negligible HPI fractions. The HPO molecules except geraniol had specific UV absorbance (SUVA, measure for aromaticity) > 3 litres g-1 cm-1 while HPI molecules had SUVA values < 3 litres g-1 cm-1. Distributions of DOM from eight soils ranged from 31 to 72% HPI, 25 to 46% HPOA and 2 to 28% HPON of total dissolved organic carbon. The SUVA of the HPI DOM was consistently smaller compared with the HPOA DOM. The SUVA of the natural DOM samples was not explained statistically by fractionation and the variation coefficient of SUVA among samples was not reduced by fractionation. Hence, fractionation did not reduce the variability in this DOM property, which casts some doubts on the practical role of DOM fractionation in predicting DOM properties.