PDF version

DISTRIBUTION OF ELEMENTS IN SUBSOIL AND TOPSOIL

W. De Vos¹, V. Gregorauskiene², K. Marsina³, R. Salminen⁴, I. Salpeteur⁵, T. Tarvainen⁴, P.J. O’Connor⁶, A. Demetriades⁷, S. Pirc⁸, M. J. Batista⁹, M. Bidovec¹⁰ with contributions by A. Bel-lan¹¹, M. Birke¹², N. Breward¹³, B. De Vivo¹⁴, M. Duris¹⁵, J. Halamic¹⁶, P. Klein¹⁷, A. Lima¹⁴, J. Locutura¹¹, J. Lis¹⁸, A. Mazreku¹⁹, R.T. Ottesen²⁰, A. Pasieczna¹⁸, V. Petersell²¹, S. Reeder¹³, U. Siewers¹², I. Slaninka³

¹Geological Survey of Belgium, Brussels, Belgium; ²Geological Survey of Lithuania, Lithuania; ³Geological Survey of Slovak Republic, Slovak Republic; ⁴Geological Survey of Finland, Espoo, Finland; ⁵Geological Survey of France, France; ⁶Geological Survey of Ireland, Dublin, Ireland; ⁷Institute of Geology and Mineral Exploration, Athens, Greece; ⁸Geology Department, University of Ljubljana, Ljubljana, Slovenia; ⁹Geological Survey of Portugal, Portugal; ¹⁰Geological Survey of Slovenia, Ljubljana, Slovenia; ¹¹Geological Survey of Spain, Madrid, Spain; ¹²Bundesanstalt für Geowissenschaften und Rohstoffe, Hannover, Germany; ¹³British Geological Survey, Keyworth, Nottingham, United Kingdom; ¹⁴Dipartimento di Scienze della Terra, Universita' di Napoli "Federico II", Naples, Italy; ¹⁵Czech Geological Survey, Prague, Czech Republic; ¹⁶Institute of Geology, Croatia, Zagreb, Croatia; ¹⁷Geological Survey of Austria, Wien, Austria; ¹⁸Polish Geological Institute, Warsaw, Poland; ¹⁹Centre of Civil Geology, Tirana Albania; ²⁰Geological Survey of Norway, Trondheim, Norway; ²¹Geological Survey of Estonia, Tallinn, Estonia.

The FOREGS soil samples do not represent all soil types of Europe. According to the FOREGS Geochemical Mapping Field Manual (Salminen, Tarvainen et al. 1998), the sample should be taken from residual and sedentary (i.e., undisplaced) soil, which generally reflects the underlying lithology. Residual soil samples were collected from the small, second order drainage basins at suitable sites above their alluvial plain and base of slope, where alluvium and colluvium are respectively deposited. Residual soil may have been developed either on bedrock or on till, as is the case in glaciated terrains. Soil samples were mostly collected from forest soils. However, agriculture or pastureland sometimes constitutes the dominant land use type in the small catchments selected for sampling in southern Europe (Map 1). Photographs from each soil sampling site are available from the electronic version of the Atlas.

The main purpose of the FOREGS geochemical survey is not to document the geochemistry of different soil types, but to discover the large pattern of geochemical signatures on the scale of the continent, and to investigate the different factors influencing this pattern, notably bedrock geology, climate and human influences. Detailed pedogenetic processes fall outside the scope of this investigation. From this point of view, soil geochemistry helps us to understand geology on the continental scale, and not the other way around.

The distribution of most elements in soil shows a pattern related to geology and/or mineralisation. A major geochemical dichotomy appears on the various maps between soil developed on crystalline and Palaeozoic basement rocks, and those on unfolded cover rocks that are naturally impoverished in most trace elements (except for areas of intense agricultural activity). Palaeocene and Eocene climates caused strong argillic and ferralitic weathering of some basement rocks, with a change in mineralogy, and a modification of vertical and lateral distribution patterns of most major elements.

On crystalline bedrock, there is also a systematic difference between the glacial till area encompassing Fennoscandia and Scotland, and the rest of Europe. Till is mostly quite fresh and chemically unweathered detritus, while in other areas the surficial deposits contain old and chemically strongly weathered material. Moreover, weathering of carbonates and silicates shows a different pattern. The anomaly pattern shown on the Na₂O map (higher content over glacial till areas) can be found also for elements such as Ca, Sr, Ba, K and some others. Since these elements are also derived from other sources (carbonate in the case of Ca and Sr, shale in the case of K and Ba) high element values can be found in non-glacial southern Europe as well. The dilution of many element concentrations in the marginal area of the last glaciation is another geological process. The amount of weathered material in till increases towards the ice marginal area, where till contains a large amount of silica.

The European map of the weathering index (Map 2) shows that weathering in Fennoscandia is generally less advanced than elsewhere in Europe. The weathering index is defined as:

CIA =

100 x [Al2O3 / (Al2O3 + CaO* + Na2O + K2O)],

where CaO* = Ca in silicate fraction (McClennan et al. 1990).

A high CIA value indicates a long weathering history of the soil. It is noted that the underlying bedrock also affects the CIA index value, e.g., Al-rich sedimentary rocks tend to have higher values, whereas Al-poor limestone always has low values (usually <40).

In the section on soil, for the sake of uniformity, and as a necessary simplification in descriptions, the following definitions were adopted with reference to the coloured maps and histograms in Part 1 of the Geochemical Atlas of Europe (Salminen et al. 2005)

Low values group the three lowest shades of blue in the colour scale, corresponding to the range from the minimum value up to the 25th percentile, defined as “very low” and “low background” concentrations in Part 1 (Tarvainen et al. 2005, p.97), and

High values group the three highest shades of red in the colour scale, corresponding to the range of values from the 75th percentile up to the maximum, defined as “high”, “very high” and “highly anomalous” concentrations in Part 1 (Tarvainen et al. 2005, p.97).

Correlation coefficients were calculated with Pearson’s product-moment linear correlation method (Table available in electronic format on website www.gtk.fi/publ/foregsatlas) after deletion of outliers and subsequent pairwise deletion of absent data. For a given element or determinand, outliers were defined here as values exceeding by a factor of 1.5 other nearby results, when all analytical results are ranked. They are generally visible on the histogram accompanying each map in Part 1 of the Geochemical Atlas. A maximum of four outliers were removed in this work for the calculation of linear correlation coefficients. A list of outliers is given separately for subsoil (Table 2) and topsoil (Table 3).

Throughout the text the following notation is used for the correlation coefficients:

• Very strong correlation: > 0.8;
• Strong correlation: between 0.6 and 0.8;
• Good correlation: between 0.4 and 0.6, and
• Weak correlation: between 0.3 and 0.4.

Because of the large number of samples, even the so-called weak correlations are significant at the 0.01 confidence level.

The use of Pearson’s correlation coefficients rather than Spearman’s rank correlation, and the deletion of outliers, is linked to a factor analysis, performed on the dataset (Batista et al. 2006, Annex 5 in this volume).

When interpreting correlation coefficients of major elements, especially Si, Al and Ca, it should be kept in mind that we work in a closed system where the sum total of all elements (or oxides in the case of majors) add up to approximately 100%. This leads to some form of autocorrelation, as observed in the negative correlation between Si and Ca, and the negative correlation Si-Al. This phenomenon also contributes to the significance of the so-called weak correlations mentioned earlier.

Despite the critical objections that can be made against using linear correlation coefficients in this data set covering very dissimilar regions, the coefficients were found to express general endencies of element associations at the continental scale. These tendencies are similar for soil, stream sediment and floodplain sediment (together forming the solid sample media), and are, therefore, taken to be meaningful indicators of geochemical processes. For this reason, and despite the obvious shortcomings, correlations are mentioned throughout the text. The overall pattern will further be discussed in the factor analysis section by Batista et al. 2006 (Annex 5 in this volume).

The ratios topsoil/subsoil were calculated for all elements (Table 4), averaging all individual ratios between pairs of subsoil and topsoil samples, for the whole dataset including the outliers, provided both subsoil and topsoil were sampled at the same site. For calculation of the correlation coefficient between topsoil and subsoil, likewise, outliers were included (Table 4). The only purpose of this correlation coefficient is to evaluate how systematic an increasing (or decreasing) tendency is from subsoil to topsoil: a high coefficient means that enrichment (ratio >1) or leaching (ratio <1) in topsoil occurs in most of Europe, a low coefficient indicates that enrichment or leaching is more erratic or applies only to certain regions.

Acknowledgements

The Geochemical Atlas “Agricultural Soils in northern Europe” by Clemens Reimann et al. (2003) was often consulted for comparison.

PDF version