Natural and anthropogenic patterns of covariance and spatial variability of minor and trace elements in agricultural topsoil

The aim of this study is to analyze the sources of variation in soil geochemical composition related to spatial scale and sampling procedure. Apart from an intrinsic interest, research was motivated by practical questions regarding sampling strategies for regional soil surveys. An unbalanced nested sampling design, based on single samples, was implemented at Walcheren/Zuid-Beveland, an agricultural area in the south-west of the Netherlands. This polder landscape consists of Holocene marine deposits. The samples were digested with aqua regia and analyzed by ICP-MS. The resulting data for 40 elements were examined through principal component analysis in combination with varimax rotation (PCA-V), unbalanced analysis of variance (UANOVA) and a fuzzy c-means clustering of the UANOVA results. The increase in noise when using single samples was further assessed through comparison with results of a previous study in the area based on composite samples. The three factors of the PCA-V are interpreted as being the results of natural, human, and geohydrological processes. The UANOVA analysis makes a different grouping into three spatial variance patterns. Two groups show a gradual increase of variance with distance and contain respectively clay/feldspar related elements and heavy rare earth related elements. The third group encompasses all anthropogenically and geohydrologically influenced elements with a larger part of the variance at small scales. In general regional variance predominates over local variance and the extra effort of compositing is of limited value for a regional survey. However, environmental priority metals like Cd, Cu, Pb, Sb, Sn and Zn, were found to have larger local variance components. For these elements, compositing is relatively more beneficial.