Multivariate statistical assessment of functional relationships between soil physical descriptors and structural features of soil organic matter in Mediterranean ecosystems.

Abstract: Current environmental research is paying increasing attention to reliable analytical surrogates of soil quality. In this work a series of molecular features of soil organic matter were studied in different soil types from Central Spain with the purpose of identifying the soil functions most closely correlated with specific pools of soil organic matter and their structural characteristics. Soil physical variables—including bulk density, total porosity, aggregate stability, available water capacity and water infiltration parameters (Kostiakov's equation coefficients)—were determined. The major soil organic fractions (lipids, particulate free organic matter, fulvic acids, humic acids and humin) were quantified using standard procedures and the soil organic matter was characterised by spectroscopic techniques. Statistical data treatments including simple regression, canonical correlation models and multidimensional scaling suggested two well-defined groups of physical properties in the studied soils: (i) those associated with organic matter of predominantly aromatic character (e.g., water infiltration descriptors), and (ii) soil physical variables related to organic matter with marked aliphatic character and comparatively low degree of humification (e.g., porosity, aggregate stability, available water capacity and field air capacity). From the practical viewpoint, the results support the idea that the detailed structural study of the soil organic matter is useful for accurately monitoring soil physical status. The only determination of total soil organic carbon ought to be complemented with qualitative analyses of the organic matter fractions, at least at the spectroscopic level, which to large extent help to explain the origin of the variability in soil physical properties and can be used for the early diagnosis of possible degradation processes. [Copyright &y& Elsevier]