Geochemical fingerprinting and source discrimination ...

Chemical Geology 396 (2015) 1–15

Contents lists available at ScienceDirect

Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

Geochemical fingerprinting and source discrimination of agricultural soils at continental scale Philippe Négrel a,⁎, Martiya Sadeghi b, Anna Ladenberger b, Clemens Reimann c, Manfred Birke d, the GEMAS Project Team 1 a

BRGM, Laboratories Division, Orléans, France Geological Survey of Sweden, Uppsala, Sweden Geological Survey of Norway, Trondheim, Norway d Bundesanstalt für Geowissenschaften und Rohstoffe, Stillweg 2, 30655 Hannover, Germany b c

a r t i c l e

i n f o

Article history: Received 5 March 2014 Received in revised form 1 December 2014 Accepted 5 December 2014 Available online 13 December 2014 Editor: Carla M. Koretsky Keywords: Agricultural soil Parent material Weathering Geochemistry Upper continental crust UCC

a b s t r a c t 2108 agricultural soil samples (Ap-horizon, 0–20 cm) were collected in Europe (33 countries, area 5.6 million km2) as part of the recently completed GEMAS (GEochemical Mapping of Agricultural and grazing land Soil) soil mapping project. GEMAS soil data have been used to provide a general view of element origin and mobility with a main focus on source parent material (and source rocks) at the continental scale, either by reference to average crustal abundances or to normalized patterns of element mobility during weathering processes. The survey area covers a large territory with diverse types of soil parent materials, with distinct geological history and a wide range of climate zones, and landscapes. To normalize the chemical composition of European agricultural soil, mean values and standard deviation of the selected elements have been compared to model compositions of the upper continental crust (UCC) and mean European river suspended sediment. Some elements are enriched relative to the UCC (Al, P, Pb, Zr,) whereas others, such as Mg, Na and Sr are depleted. The concept of the UCC extended normalization pattern has been applied to selected elements. The mean values of Rb, K, Y, Ti, Al, Si, Zr, Ce and Fe are very similar to the values from the UCC model, even when standard deviations indicate slight enrichment or depletion. Zirconium has the best fit to the UCC model using both mean value and standard deviation. Lead and Cr are enriched in European soil when compared to the UCC model, but their standard deviation values span a large, particularly towards very low values, which can be interpreted as a lithological effect. GEMAS soil data have been normalized to Al and Na, taking into account the main lithologies of the UCC, in order to discriminate provenance sources. Additionally, sodium normalization highlights variations related to the soluble and insoluble behavior of some elements (e.g., K, Rb versus Ti, Al, Si, V, Y, Zr, Ba, and La, respectively), their reactivity (e.g, Fe, Mn, Zn) and association with carbonates (e.g., Ca and Sr). Maps of Europe showing the spatial distribution of normalized compositions and element ratios reveal difficulties with the use of classical element ratios because of the large lithological differences in compositions of soil parent material. The ratio maps and color composite images extracted from the GEMAS data can help to discriminate the main lithologies in Europe at the regional scale but need to be used with caution due to the complexity of superimposed processes responsible for the soil chemical composition. © 2014 Elsevier B.V. All rights reserved.

⁎ Corresponding author: Tel.:+33 (0)2 38 64 39 69; fax:+33 (0)2 38 64 37 11. E-mail addresses: p.negrel@brgm.fr (P. Négrel), martiya.sadeghi@sgu.se (M. Sadeghi), anna.ladenberger@sgu.se (A. Ladenberger), clemens.reimann@ngu.no (C. Reimann), Manfred.Birke@bgr.de (M. Birke). 1 The GEMAS Project Team: S. Albanese, M. Andersson, R. Baritz, M.J. Batista, A. Bel-lan, D. Cicchella, A. Demetriades, B. De Vivo,W. De Vos, E. Dinelli, M. Ďuriš, A. Dusza-Dobek, O.A. Eggen, M. Eklund, V. Ernstsen, P. Filzmoser, D.M.A. Flight, S. Forrester, M. Fuchs, U. Fügedi, A. Gilucis, M. Gosar, V. Gregorauskiene, W. De Groot, A. Gulan, J. Halamić, E. Haslinger, P. Hayoz, R. Hoffmann, J. Hoogewerff, H. Hrvatovic, S. Husnjak, L. Janik, G. Jordan, M. Kaminari, J. Kirby, J. Kivisilla, V. Klos, F. Krone, P. Kwećko, L. Kuti, A. Lima, J. Locutura, D. P. Lucivjansky, A. Mann, D. Mackovych, M. McLaughlin, B.I. Malyuk, R. Maquil, R.G. Meuli, G. Mol, P. O'Connor, R. K. Oorts, R.T. Ottesen, A. Pasieczna, W. Petersell, S. Pfleiderer, M. Poňavič, S. Pramuka, C. Prazeres, U. Rauch, S. Radusinović, I. Salpeteur, R. Scanlon, A. Schedl, A.J. Scheib, I. Schoeters, P. Šefčik, E Sellersjö, F. Skopljak, I. Slaninka, A. Šorša, R. Srvkota, T. Stafilov, T. Tarvainen, V. Trendavilov, P. Valera, V. Verougstraete, D. Vidojević, A. Zissimos and Z. Zomeni.

http://dx.doi.org/10.1016/j.chemgeo.2014.12.004 0009-2541/© 2014 Elsevier B.V. All rights reserved.

1. Introduction The interaction between water and rocks at the Earth's surface consumes atmospheric CO2, releases the most soluble ions and leads to the formation of soils and sediments (Amiotte Suchet et al., 2003). The mineralogy, particle size and elemental composition of soil and sediment are related to the nature of the parent rock material and the degree to which this material has been weathered (summary in Jain and Sharma, 2001). Chemical weathering of the Earth's surface results in a large loss of the most soluble elements, whereas the least soluble ones essentially remain in situ (Gaillardet et al., 1999). Weathering

2

P. Négrel et al. / Chemical Geology 396 (2015) 1–15

stages of soils and sediments are usually evaluated by comparing their chemistry and mineralogy with the chemistry and mineralogy of their parent material. This may be quantitatively estimated by some weathering indicators such as enrichment factors (Nesbitt, 1979; Chester et al., 1985; Goldstein and Jacobsen, 1988), indices of geochemical maturity, such as the resistate index (corresponding to the ratio of SiO2 over CaO + K2O + Na2O + MgO), the hydrolyzate index (corresponding to the ratio of Al2O3 in clays over Al2O3 in sand) and the oxidate index (corresponding to the ratio of Fe2O3 in clays over Fe2O3 in sand) of maturity (Wakatsuki et al., 1977). The concentration of a resistant component (often Si or Al) relative to the more mobile alkaline (Na and K) and alkaline earth (Ca and Mg) components determines the Chemical Index of Alteration (CIA) (Nesbitt and Young, 1982; Gallet et al., 1998; Schucknecht et al., 2012). The CIA ([Al2O3/(Al2O3 + CaO* + Na2O + K2O)] × 100) is based on the transformation of feldspar into clay during weathering where the major element oxides are given in molecular proportions and CaO* represents the CaO content of silicate minerals only (review in Bahlburg and Dobrzinski, 2011). At river catchment scale, one way to understand the chemical composition of the soils or the sediments developed from the same parent lithology is to normalize each chemical species in sediment or soil by the average value of the same species in the parent material (Dennen and Anderson, 1962; Cullers et al., 1988; Négrel, 2006). This methodology has also been applied to weathering profiles (Nesbitt et al., 1980). Using soil and sediment geochemistry, many researchers have attempted to estimate weathering and denudation rates of continents, recycling of continental crust and anthropogenic influence, (Huanghe: Yang et al., 2004; Ganges–Brahmaputra: Stummeyer et al., 2002; Congo: Gaillardet et al., 1995; Dupré et al., 1996; large river systems: Gaillardet et al., 1999; Viers et al., 2009; loess: Gallet et al., 1998; world soils: de Caritat et al., 2012; Schucknecht et al., 2012: regional soils: Rawlins et al., 2003). A similar approach in sedimentary basin studies provided information about weathering history, composition, and tectonic setting of the continental crust (Roser and Korsch, 1986; Raza et al., 2012). The aim of this paper is to present a combined approach for interpreting the large database on the major and trace element composition of European soils from the continental-scale GEMAS project (Geochemical mapping of agricultural and grazing land soils of Europe, Reimann et al., 2014a,b). Comparison with upper continental crust UCC, UCC extended normalized patterns, Al-normalized enrichment factors, Na normalized ratios, normalized composition maps of weathering indices has been applied to the dataset, which represents highly variable parent materials, climate conditions, and landscapes. We provide a continental scale view of element mobility by a combined approach that helps to verify the validity of widely used models and methods when applied to large-scale datasets. 2. The GEMAS project The GEMAS project was carried out by the Geochemistry Expert Group of EuroGeoSurveys in cooperation with Eurometaux and managed by the Geological Survey of Norway NGU (Reimann et al., 2012a). Soil samples were collected in 33 European countries, covering an area of 5.6 million km2 (Reimann et al., 2011, 2012a). The survey area is shown in Fig. 1 with a grid representing a sample density of 1 site per 2500 km2. Two sample materials have been used in this survey: agricultural soil (Ap-horizon, 0–20 cm depth range) and grazing land soil (i.e., soil under permanent grass cover, 0–10 cm depth range). This study focuses on the Ap-horizon (0–20 cm) of regularly plowed agricultural land. Sample collection followed a strict field manual protocol regarding the landscape and sample site location (EuroGeoSurveys

Geochemistry Working Group, 2008). As e main objective of the project was to detect and to map the natural element variation at the European scale, the soil samples were never taken at known contaminated sites, in the immediate vicinity of industry or power plants, near a railway line or a major road, directly below high electric power lines or close to pylons. Flat-lying arable land sites were selected wherever possible and landscape depressions were avoided. In addition, sampling was carried out at a time when fertilizers were not being used by landowners. A total of 2108 samples were analyzed and reported on (Reimann et al., 2014a). 3. Materials and methods The methods for GEMAS sampling, sample preparation and analysis for major and trace elements were described in detail in Reimann et al. (2011, 2012a, 2014a) and Reimann and de Caritat (2005). All composite samples were collected as rather large (2–2.5 kg) samples from one large field, sieved to b 2 mm, and milled to less than 63 μm. Depending on LOI (with a threshold of 25%), two different types of fused disks were prepared and analyzed by wavelength dispersive X-ray fluorescence spectrometry (WD-XRF). Project standards and replicates of project samples were used to monitor trueness, repeatability and accuracy, detection limits and QC results and are discussed in detail in Reimann et al. (2011). The classification of European soils is presented here in the form of descriptive elemental ratios or ternary plots, following the chemical classification presented by Garrels and Mackenzie (1971). We have then used the upper continental crust (UCC) model as a reference for investigating element mobility following Taylor et al. (1983), either by comparison with the UCC, or using the UCC extended normalized patterns. We therefore implemented the normalization of element concentrations to a background indicator as used by Nesbitt (1979), i.e. the Al-normalized enrichment factor, or the Na normalized ratios. Finally, the spatial distribution of weathering indices in soils have been displayed in a series of interpolated geochemical maps produced with Arcview GIS software using the inverse distance weighted (IDW) algorithm to generate a regular grid with a 6 × 6 km cell size and a circular window with fixed radius of 150 km (Sadeghi et al., 2013). 4. Results Total concentrations of SiO2, TiO2, Al2O3, Fe2O3, MnO, MgO, CaO, Na2O, K2O, P2O5, Cl, F, As, Ba, Bi, Ce, Co, Cr, Cs, Cu, Ga, Hf, La, Mo, Nb, Ni, Pb, Rb, Sb, Sc, Sn, Sr, Ta, Th, U, V, W, Y, Zn and Zr were determined by wavelength dispersive X-ray fluorescence spectrometry (WD-XRF). The whole GEMAS dataset is provided in a two-volume geochemical atlas (Reimann et al., 2014a, 2014b accompanied by a DVD containing all data and maps) where the XRF results are discussed in detail with respect to their spatial distribution on geochemical maps and their statistical parameters. In several earlier publications, the GEMAS XRF data have been used to discuss the mathematical nature of such datasets and have been compared to other continental-scale geochemical surveys (e.g., Reimann et al., 2012a; de Caritat et al., 2012; Reimann et al., 2012c). The regional distribution of single elements or related groups of elements was presented in e.g. Reimann et al. (2012b — Pb), Ottesen et al. (2013 — Hg), Tarvainen et al. (2013 — As), Sadeghi et al. (2013 Ce, La, Y). In our study, only selected elements are presented with a focus on major elements and associated trace elements. In order to classify the GEMAS soil parent material, we follow the approach from de Caritat et al., 2012 who calculated the relative proportions of five major lithological types. According to the global rock lithology model of Amiotte

Fig. 1 (a) Map of parent materials in Europe showing the distribution of various lithologies across the continent (modified from Günther et al., 2013). (b) Sample locations of the agricultural soil (Ap-samples) from the EuroGeoSurveys GEMAS Project (n = 2108). (c) Generalized geological map of Europe with major lithotectonic units, Variscan and Alpine belts, Transeuropean Suture Zone (TESZ) and the extension of maximum glaciation (modified from Reimann et al., 2012b).


3

4


Suchet et al. (2003), the most common lithologies in Europe are (1) plutonic and metamorphic, (2) shales, and (3) carbonate rocks. Based on these assumptions, de Caritat et al., 2012 defined the source parent material where the plutonic and metamorphic rocks (39%) and shales (37%) dominate; carbonate rocks (14%) and sand-sandstones (9.5%) are significant, whereas acid volcanic rocks and basalts (0.5% each, respectively) play a subordinate role. 5. Discussion 5.1. Classification of European soils The sand class system introduced by Garrels and Mackenzie (1971), based on the geochemical composition of major elements, has been applied to various geologic materials such as loess (Gallet et al., 1998) and stream sediments (e.g. the Amazon Basin, Vital and Stattegger, 2000; the Ganga basin, Lupker et al., 2012). The Garrels and Mackenzie (1971) chemical classification, which has been proven to be useful in studies of provenance and source rock weathering (Kasanzu et al., 2008), can be interpolated from sedimentary rocks to soils, such as the GEMAS database, which allows discrimination of soils by their maturity status. The most used classification parameters are the SiO2/Al2O3 ratio that primarily reflects the abundance of quartz, clay, and feldspar, the Na2O/K2O ratio that defines an index of chemical maturity, and the Fe2O3/K2O ratio that defines an index of mineral stability. These ratios have been applied to the GEMAS dataset (Fig. 2), as previously done for loess (Gallet et al., 1998) or stream sediments (see references in Lupker et al., 2012). In Fig. 2, the log ratios of Na2O/K2O are plotted against the log ratios of SiO2/Al2O3, and the log ratios of Fe2O3/K2O are plotted against the log ratios of SiO2/Al2O3 (Pettijohn et al., 1972; Herron, 1988). Fig. 2a shows that GEMAS soils in geochemical classification fields for terrigeneous sands are bracketed mainly by greywacke (sand-mud-and clay) and litharenite, with a diminishing trend towards sub-litharenite, sub-arkose, arkose and quartz-arenite, reflecting increasing abundances of quartz and feldspar. This parent-material based information can be translated into soil types that contain mainly mud and clay end-members rich in feldspars and micas and their weathering products (e.g., illite). Quartz-dominated (sandy) soils are less common. Apart from the dominant mineral components, this diagram has the potential to be used to show soil maturity and soil aging trends, i.e. quartz-rich soil is generally classified as young and clay-rich soil as mature and older. Although this contradicts traditional sediment classifications, in which greater quantities of quartz indicate greater transport and maturity, this interpretation is supported by the spatial distribution of the GEMAS soils. This is because young quartzrich soils occur in high latitudes, glaciated terrains and mountainous regions and clay-rich soils dominate in the large sedimentary basins. Fig. 2b confirms that potassium-rich components in soils (e.g., mica and feldspar) dominate in the soils with minor enrichment in iron which can be caused by grain coatings composed of iron oxides and hydroxides (e.g., goethite), and there is a large group of non-crystalline soil components, rich both in aluminum and iron (e.g., allophane, Birkeland, 1984). Rather than comparing data on binary X-Y diagrams, we apply ternary plots of parameters that can be used to evaluate chemical weathering trends (e.g., Kasanzu et al., 2008). Fig. 3 presents Ca, Na and Ti contents in molecular proportions in the GEMAS samples together with the main lithological end-members from Parker (1967), e.g. plutonic and metamorphic rocks and shales, carbonate rocks and sandsandstones, and the upper continental crust (UCC) model from Rudnick and Gao (2003). The main spread of data along the Na axis reflects the variation in Na content between sedimentary end-member (carbonate and sandstone) and igneous (low-Ca granite) end-member parent materials. The closeness with which a soil sample approaches the 0 or 100% apex along the Na axis is an indication of similarity to a lithologic end-member. The two arrows denote compositional trends of

Fig. 2. Plots of GEMAS soils (black crosses) on the geochemical classification diagrams for (a) terrigenous sands; log ratios of Na2O/K2O versus SiO2/Al2O3 (Pettijohn et al., 1972), and for (b) terrigenous sands and shales; log ratios of Fe2O3/K2O versus SiO2/Al2O3 (Herron, 1988).

weathering of silicate and carbonate rock types. Increased weathering results in Ca loss and Ti enrichment. Both indices are a measure of the geochemical maturity of the soil material with respect to the weathering of underlying bedrock. 5.2. Comparison with the chemical composition of the continental crust The major elements of the Earth's continental crust are generally divided into three subgroups (Goldschmidt, 1954): immobile elements (e.g., Si, Fe, Al), mobile elements (e.g., Ca, Mg) and very mobile elements (e.g., K, Na). Elements in each group are often associated with a trace element with similar properties and geochemical behavior, e.g. Ca with Sr, K with Rb, Si with immobile elements such as Zr and Ti. de Caritat et al., 2012 compared the soil median element concentrations of continental scale surveys (GEMAS soil data and Australian survey) with published world soil average compositions, upper continental crust composition and two other continental surveys in the USA and China. They noted substantial differences among these datasets and argued that only results obtained for large areas can provide real estimates of average soil composition. The authors concluded that neither


5

Fig. 3. Ca–Na–Ti (in molecular proportions) distribution in GEMAS soils (black crosses), plotted together with bedrock and Upper Continental Crust UCC compositions (yellow circles) according to Parker (1967) and Rudnick and Gao (2003). Arrows denote compositional weathering trends of silicate and carbonate rock types towards Ca loss and Ti enrichment. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

average crustal values, world soil average values nor the spatial distribution of lithological units provided good estimates of soil chemistry. On the other hand, weak relationships between geological (e.g. lithological) information and soil chemistry are mainly due to the fact that lithological maps of sufficient precision are not available at the continental scale (Cicchella et al., 2005; de Caritat et al., 2012). In our study we used the map of parent materials in Europe (Fig. 1) which shows the distribution of various lithologies across the continent as modified from Günther et al. (2013). The representation in Fig. 1 is quite different from that of Dürr et al. (2005), demonstrating the difficulties of producing a reliable lithological map as only the use of a regional-scale information is effective in comparison to large scale (e.g. continental) lithological information due to dilution effect. To examine the major difficulties in constraining values for soil, we compared mean and median values for compositions from the literature and values calculated for the GEMAS dataset. In Fig. 4, the mean versus the median for the 2108 GEMAS soil samples was plotted to check the spread in GEMAS data in relation to the extreme values. Most major and trace elements (represented as [X]) turned out to have a strong, near x = y linear correlation at the logarithmic scale. Exceptions are Ca and Sr and to a lesser extent Mg, Na, Cr and Pb, which all have a positive shift towards higher mean values (mean N median). No major shift off the x = y line is observed for the majority of the elements under consideration, particularly those resistant to weathering e.g., Zr, Ti, La and Ce (Gaillardet et al., 1995; Dupré et al., 1996). Subsequently, in Fig. 5, the mean values and their variation (expressed as standard deviation) in GEMAS soils have been compared with the average content of respective elements in the upper continental crust (UCC, Taylor and McLennan, 1995; Rudnick and Gao, 2003). The UCC has been used previously as a reference for investigating element mobility for loess deposits in America, China, New Zealand and Europe by Taylor et al. (1983). Their study showed enrichment in Si, Zr and Hf, and depletion in several major elements (Ca, Na, Fe, Mg, K) but concentrations of most trace elements such as U, Pb, Cr, V and Ba were

comparable to UCC values. Recently, UCC values were also used in the study of sediments and soils by Ahrens and McLennan (2013). The UCC model has also been used as a reference to investigate the elemental composition of soil parent material, weathering processes and the influence of anthropogenic practices (Li et al., 2009 in Tibetean Plateau; Takeda et al., 2004 in Japan). Application of the UCC normalization has been executed at various scales: in the vicinity of industrial sites (Cloquet et al., 2006) and in agricultural topsoil of urban regions (Romic and Romic, 2003; Cicchella et al., 2005), at the regional scale

Fig. 4. Binary plot of the major and trace element mean and median values for the GEMAS soil samples (n = 2108).

6


Fig. 6. Extended patterns normalized to the upper crust for the GEMAS soils (plotted as mean and standard deviation), according to Gaillardet et al. (1995) and Dupré et al. (1996). The order of the elements is a monotonic decrease of crustal abundances normalized to primitive mantle concentrations ([C]PM, Hofmann, 1988), using the upper crust model of Taylor and McLennan (1995) and Rudnick and Gao (2003). Fig. 5. Comparison of selected major and trace element mean values from the GEMAS soil dataset (including means and standard deviations) with the average composition of the upper continental crust UCC (Taylor and McLennan, 1995; Rudnick and Gao, 2003).

(e.g. in northern France, Sterckeman et al., 2006) and the global scale (e.g. terrestrial surface, Hartmann et al., 2012). In the GEMAS soils, several elements (Al, P, Zr, Pb, Y) show enrichment compared to the UCC, whereas Mg, Na, Ba and Sr are depleted and plot on the left side of the reference line (Fig. 5). For a few elements (Ca, Mg, Mn, and Sr), the major depletion pattern is mainly due to the large concentration range of these elements in the soils database. On the other hand, Al and P are visibly enriched in comparison to the UCC values, which can be interpreted as both a natural and anthropogenic effect (e.g. from fertilizers). It is worth noting that relatively immobile elements (Ti and Zr) show similar mean values accompanied by a low standard deviation, suggesting that the concentration of heavy minerals has a negligible impact on the chemical composition as suggested by Key et al. (2012). 5.3. Major and trace elements: upper crust extended normalized pattern representation The concept of extended patterns normalized to the upper crust for selected elements has been adopted from the previous studies of stream sediments (Gaillardet et al., 1995; Dupré et al., 1996) and loess (Gallet et al., 1998), and has been applied to the GEMAS data. Element concentrations from the upper continental crust model (UCC) of Taylor and McLennan (1995) and Rudnick and Gao (2003) have been normalized to the primitive mantle of the Earth ([C]PM) according to the method of Hofmann (1988). The order of elements on the diagram, from left to right, is designed to show a monotonic decrease of normalized abundances and also corresponds to their more compatible behavior during magma differentiation, which form a soil perspective points to the chemical character of parent material. For each soil sample, the element concentrations have been normalized to the primitive mantle values ([C]PM, Hofmann, 1988), with mean and standard deviation being calculated for the entire dataset (Fig. 6). The mean values of most elements (Rb, K, Y, Ti, Al, Si, Zr, Ce and Fe) nearly overlap with the UCC model normalized concentrations with a few exceptions for elements with high variations within the GEMAS dataset (large standard deviation), indicating either enrichment (e.g., Zr, Al, Pb) or depletion (e.g., Fe, Ba, Sr, Na). Lead and Cr are enriched when compared to the normalized UCC model concentrations (Fig. 6), but both elements (in addition to Ca and Sr) are characterized by very large variation, particularly towards very low values. Such patterns can be related to

certain lithological effects. The Pb and Cr enrichments in Fig. 6 suggest that the normalized UCC model underestimates the contribution of source rock types such as mafic and ultramafic rocks in European soils and sediments or it is an indication of anthropogenic contamination. The group of rather mobile elements, comprising Ba, Na, Sr and Ca has values generally lower than the normalized UCC model concentrations and show negative anomalies relative to their neighboring elements. However, their normalized CsPM values (including standard deviation) overlap with the normalized UCC model values with Ca and Sr being most depleted within the group. This trend can be explained by high mobility of Ca and Sr in carbonate-dominated terrains which cover large areas of southern Europe. 5.4. Al normalization and the concept of enrichment factors A way to explore element variability is to look at the variations using indicator elements. The normalization of element concentrations to a background indicator element is often used in geochemistry (Nesbitt, 1979; Chester et al., 1985; Goldstein and Jacobsen, 1988). Enrichment factors (EF), defined as the ratio of an element to an indicator element, are often used in soil studies, or to describe weathering profiles or particulate matter transported by rivers (e.g., Ansari et al., 2000). The EF can be defined as EF = [X]/[Y]sample / [X]/[Y]upper crust where [X] represents the concentration of element X, and [Y] is the concentration of the reference element. The reference element is generally chosen such that it can be solely assigned to a terrigenous origin. Thus, a ratio close to 1 means that the concentration of X can be explained by a terrigenous origin alone, while a ratio N 1 implies an enrichment relative to the reference, either through anthropogenic input or due to an incorrect choice of geological reference bedrock. A ratio below 1 indicates a depletion of the element that can be ascribed to the mobility of the element due to weathering processes or bioavailability of the element. However, EFs can be of limited use because the chosen reference element can be ‘unreactive’ during weathering processes while the contaminant elements of interest are ‘reactive’ (e.g. being adsorbed onto clays, with a high degree of bioavailability, etc.). Sucharovà et al. (2012) demonstrated this limitation of EFs using five different reference elements (Al, Ce, Fe, Li and Si) in a study of soils in the Czech Republic. Their study showed that high EF values can be due to high variation in the reference element rather than indicating true enrichment in the element of interest. Moreover, Reimann and de Caritat (2005) concluded that, close to contamination sources, EFs are efficient in


7

Fig. 7. Plot of the enrichment factors (EFs) for Ca versus Sr (a) and K versus Rb (b), (each referred to as XN, X being the chemical element) together with some bedrock EFs, UCC and mean European river suspended particulate matter (Parker, 1967; Rudnick and Gao, 2003; Viers et al., 2009) for the GEMAS soils. The enrichment factor is defined as EF = [X]/[Y]sample / [X]/[Y]upper crust where [X] represents the concentration of the element X, and [Y] the concentration of the reference element (all EFs normalized to Al).

providing evidence of contamination, but because EFs can also be strongly influenced by other processes, EFs are not a robust method to characterize the anthropogenic impact on the environment at a global (or continental) scale. It is worth noting, however, that the concept of EFs (as well as other normalized ratios; see next section) can be used at a global scale to compare the potential sources of parent materials (Viers et al., 2009). In order to characterize more precisely the different parent sources in the GEMAS samples, EFs for Ca versus Sr and K versus Rb (normalized to Al) were plotted together with bedrock EFs, the model UCC composition EF, and mean European river suspended sediment concentration EF (Fig. 7, Parker, 1967; Rudnick and Gao, 2003; Viers et al., 2009). The UCC value is represented by the intersection at 1 to 1 in each plot. The spread

of the data in the CaN versus SrN plot (Fig. 7a) is highlighted by a trend (blue line) between a carbonate end-member (highest CaN and SrN) and a silicate end-member (granodiorite and low-Ca granite) with a subordinate sandstone component. Depletion in Ca in comparison to Sr towards the carbonate end-member (soil enrichment) may indicate that in soils calcium is more mobile than strontium. The plot of K versus Rb (Fig. 7b) shows that the data with all bedrock values cluster around value 1 and then spread towards mafic endmembers (blue line). Some of the KN and RbN values are higher than any available rock component, but they seem to be correlated together with positive co-variations as indicated by the red arrow. This trend may reflect the weathering processes resulting in K and Rb enrichment, and inferred to be a consequence of an increase in the illite and kaolinite

Fig. 8. The CIA index (see text for definition) versus (a) Ti/Na and (b) Ca/Na ratios together with selected bedrock and UCC average compositions (Parker, 1967; Rudnick and Gao, 2003) for the GEMAS soils.

8


Fig. 9. Si (mg/kg) content versus Ca/Na ratios (a) together with bedrock and UCC compositions (Parker, 1967; Rudnick and Gao, 2003; Viers et al., 2009) for the GEMAS soils. A trend line (blue line) is drawn from low-Ca granite to the carbonate end-members. Ca/Na vs. K/Rb ratios (b) for GEMAS soils with reference values for average rock compositions, UCC compositions, and the mean European river suspended matter (Parker, 1967; Rudnick and Gao, 2003; Viers et al., 2009).

content (Nesbitt et al., 1980) and/or higher contribution of potassium feldspar and muscovite. 5.5. Na normalization and the mobility of the element during weathering The chemical compositions of sediments and soils represent to a large extent the primary mineralogy of the source bedrock, the effects of pre- and post-depositional chemical weathering, formation of secondary products such as clays, and element mobility, either by leaching or mineral sorting. Therefore, in soils, elements such as Al, K, Mg, and Rb are often retained in weathering profiles while Na, Ca and Sr are rapidly flushed out of the system as dissolved ions. In river systems, it is well known that depletion of Na in the suspended sediments relative to the mean upper crust complements its enrichment in the dissolved phase. Sodium is mainly derived from weathering of alumino-silicate phases and evaporites (halite); a minor part comes from carbonate rock dissolution. Sodium originates mainly from the dissolution of plagioclase crystals and it is conserved during weathering processes like saprolitization (Braun et al., 2012) or transport by river systems (Dupré et al., 1996). The fraction of Na approaches zero when plagioclase crystals are completely broken down and Na can thus be leached away either to form clays or to become dissolved ions in the water cycle. Therefore, sodium has been used as a reference element in studies of weathering processes (Négrel et al., 1993, 2013; Gaillardet et al., 2003; Yang et al., 2004), and for defining weathering indices in suspended sediments (e.g., mobility; Gaillardet et al., 1999). In suspended sediments of large rivers, the Na weathering index is always greater than unity indicating that suspended sediments are depleted with respect to UCC, whereas the weathering indices of Ca and Mg usually exhibit values less than one, reflecting the presence of carbonate in the suspended sediment. Following the methodologies used in studies of sedimentary basins (Roser and Korsch, 1986; Rawlins et al., 2003; Raza et al., 2012), we have used Na as a normalizing element to study the intensity of weathering processes in relation to the provenance of parent materials for soils formed from the lithologies of the UCC (Fig. 8). The weathering history and possible parental rock sources of GEMAS soils have been evaluated by the use of the Chemical Index of Alteration CIA = [Al2O3/(Al2O3 + CaO* + Na2O + K2O)] × 100 as defined in Nesbitt and Young (1982). The CIA is widely used as an indicator of the degree of weathering of source areas (e.g. Kasanzu et al., 2008;

Bahlburg and Dobrzinski, 2011). Three fields with variable weathering status have been defined by Servaraj and Arthur (2006): a low weathering field, with CIA in the range 50–60, an intermediate weathering one with CIA in the range 60–80 and a field corresponding to intense weathering with high CIA values N 80. The calculated CIA values show that most of the GEMAS soil samples fall within the intermediate weathering field (68% of the samples with CIA values between 60 and 80), a minor percentage plot within the low weathering field (20% samples with CIA values from 40 to 60), less than 10% of the samples — in the intense weathering field (CIA values between 80 to 100) and 1% of soils have low CIA values (b 50, see Fig. 8). This shows clearly that GEMAS soils represent highly weathered materials, as minerogenic Ca, Na and Mg are removed from the soil profile during soil development.

Fig. 10. Mn/Na ratios vs. Zn contents (mg/kg) for GEMAS soils with reference values for average rock composition, UCC compositions, and the mean European river suspended sediment composition (Parker, 1967; Rudnick and Gao, 2003; Viers et al., 2009); symbols as in Fig. 9.


In Fig. 8a and b, the CIA index has been compared with Ti/Na and Ca/Na ratios, with average values for the main lithological endmembers (Parker, 1967) as quantified by de Caritat et al., 2012 and based on the UCC model composition from Rudnick and Gao (2003). In Fig. 8a, the GEMAS soils define a trend between felsic rock compositions and clay and shale compositions, suggesting removal of mobile elements (e.g. increasing Ti/Na ratios). Soils with intense CIA weathering values would be the most mature. On the other hand, the carbonate component has geochemical characteristics not well defined by the CIA index, therefore, soils originating in carbonate-rich areas plot in the low-weathering field (Fig. 8b), and their evolution towards high Ca and low Na composition is decoupled from the main trend for the CIA index. A better link between the carbonate-derived and siliciclastic sediments and soil developed on limestone and silicate-rich bedrock is observed in Fig. 9a (Ca/Na versus Si). The majority of samples plot along a calculated trend line (blue line) between the main lithological silicate end-members (both intrusive and sedimentary, with the exception of shale) and the carbonate-end-member which has been calculated according to the binary mixing equation given by Langmuir et al. (1978). The shale-end-member is an outlier due to its lower Si

9

and Ca contents, and deviations from the general trend (mainly for soil samples) are caused by the presence of shale in the parent material. Comparison of Ca/Na to K/Rb ratios (Fig. 9b) allows discrimination among siliciclastic components due to the enrichment in K in intrusive rocks and the enrichment in Rb in clay-rich sedimentary rocks. However, the strong coherence between potassium and rubidium in many geochemical processes and the strong partitioning of Rb into potassium minerals (e.g. feldspar and micas), makes these discriminations rather difficult in practice and values tend to overlap. In general, soil compositions which are shifted towards lower K/Rb ratios may indicate Rb removal (due to soil forming processes, plant uptake and/or mineral effects). On the other hand, the higher Ca/Na ratios suggest the dominance of a carbonate end-member, while lower Ca/Na ratios correspond to the silicate end-members that encompass a wide range of lithologies. The plot of Mn/Na versus Zn (Fig. 10) illustrates well the exceptional tendency of Fe–Mn oxy-hydroxides and particularly Mn oxides to scavenge certain trace elements (e.g. Zn, Pb, Cu, Ni) from solution to form coatings on grains in soil particles (e.g. Carpenter et al., 1975; Sposito, 1989). However, the very weak correlation coefficient for GEMAS soils (R2 ~ 0.20) between Mn/Na ratios and Zn contents suggests that their origin in soil is unrelated and their concentrations are

Arctic Ocean

0

250

500 km

Russia

0.02 - 2.5 2.6 - 4.4 4.5 - 6.9 7 - 10.5 10.6 - 15.6 15.7 - 24.1 24.2 - 41.8

North Sea Baltic Sea Belarus

M d ol ov a

Canary Islands

Romania Black Sea

Albania

Turkey

Mediterranean Sea

Morocco

Algeria

Tunisia

Malta

Mediterranean Sea

Fig. 11. The Ce/Ni ratio map of Europe for GEMAS soils representing dominant lithologies (parent material) in Europe. The map was produced using an interpolated distance weighted method and cell size 10 × 10 km.

10


governed by different processes in soil. The locations of source endmembers define three trends; first (red line) – defined by the clay (located close to the UCC) and carbonate components (high Mn/Na ratios), second (green line) – defined by the mean European rivers suspended material (towards a general enrichment in Zn and Mn) and third (blue line) with low ratios assigned to felsic silicatedominated compositions. The GEMAS samples show a wide range of Mn/Na ratios, from very low values, typical of felsic end-members to higher ratios located somewhat between a carbonate end-member and a river-related component. 5.6. Spatial variability: the link with geology Ternary and elemental ratio plots and maps have been used in the interpretation of geochemical soil data for mapping and exploration (e.g., Brand, 1999). These types of maps can assist in distinguishing lithology of parent material (e.g., Garrett et al., 2008), for environmental purposes and risk assessments (e.g., Stanley and Noble, 2008), for urban soil classification and evaluation (e.g., Cicchella et al., 2008), and for special-case scenarios such as mapping of rare earth element distributions in soil (e.g., Sadeghi et al., 2013) related to lithology. Here, we use the GEMAS soil dataset to produce maps of element ratios reflecting

contrasts in chemical composition of major parent material lithologies in Europe as shown in Fig. 1. The Ce/Ni ratio map (Fig. 11) illustrates the felsic and mafic components in bedrock parent material, and also young sandy and clayey substrates. For example, in northern Europe, the general pattern in Ce/Ni ratios depicts the presence of mafic and ultramafic (ophiolite) lithologies in the Scandinavian Caledonides and the presence of Paleoproterozoic greenstone belts that are not really visible due to the scale of the geological map of Europe (Fig. 1a). Cerium enrichment in soils occurs in areas with felsic intrusive rocks covered by young sandy glacial deposits such as in northern Germany and Poland. In southern Europe, mafic volcanics and ultramafic rocks (e.g., Sicily, Cyprus) as well as carbonates can be linked with low Ce/Ni soil anomalies (Fig. 11). The soil map of the Ca/Sr versus K/Na ratios (Fig. 12) outlines not only the distribution of different granitoids (e.g., Alps) and carbonates but also highlights the old metamorphic rocks in the northern part of Sweden and Finland and western part of Norway. The high ratio values can also be attributed to the presence of karst (Mediterranean region) and carbonates (limestone, marl, dolomite and marble) as well as mafic and ultramafic rocks (e.g., ophiolite in Cyprus, Scandinavian Caledonides and Paleoproterozoic greenstone belts; Fig. 1a). Areas with unconsolidated sandy and clayey soils typically have low Ca/Sr and K/Na ratios (Fig. 12).

Arctic Ocean

0

250

500 km

Russia

2 - 152 153 - 327 328 - 602 603 - 1 080 1 090 - 2 080 2 090 - 4 200 4 210 - 6 350


M d ol ov a

Canary Islands

Romania Black Sea

Albania

Turkey

Mediterranean Sea

Morocco

Algeria

Tunisia

Malta

Mediterranean Sea

Fig. 12. Map of Europe showing combined element ratios Ca/Sr versus K/Na for GEMAS soils. The map was produced using an interpolated distance weighted method and cell size 10 × 10 km.


Combining single element or ratio maps into color composite maps is a useful tool to extract or visualize geochemical features related to dominant lithology. Fig. 13 displays a color composite image of a Ca-Ti-Na ternary representation, which helps to discriminate, at the continental scale, among parent material lithologies, particularly between ultramafic and maficrocks, -granite, -carbonate and shale, by comparison with Figs. 1 and 3. The red color in the map represents lithologies with high Ca concentrations (e.g., carbonate rocks and/or ultramafic rocks) such as in the eastern part of Spain, the southern and northern part of France, the central and southern part of Italy, and southern Greece and Cyprus. The blue area in the color composite map represents high Na concentrations that can be correlated with granitic rocks, e.g. in Scandinavian countries, Scotland, northern Sardinia and Corsica. The green color in the map represents the area with high Ti concentrations which can be observed over central and southern Europe, and which indicates the more advanced weathering of older soils than those in northern Europe, north of the maximum glaciation boundary. The mixture of green (Ti) and blue (Na) north of the limit of the last glaciation (Scandinavia, United Kingdom, and Ireland) shows young soils developed on glacial till and outwash sediments that still retain feldspar.

11

Weathering indices, in addition to the CIA discussed above, have been widely used in studies evaluating soil fertility and development and demonstrating the impact of climate on bedrock weathering (Delvaux et al., 1989; Bahlburg and Dobrzinski, 2011). Furthermore, weathering indices are also used to interpret the weathering history of sediments (see Price and Velbel, 2003 and references therein; Bahlburg and Dobrzinski, 2011). We tested various weathering indices for the GEMAS dataset, and the WIP (Weathering Index of Parker; Parker, 1970) defined as 100 ∗ [(2Na2O/0.35) + (MgO/0.09) + (2K2O/ 0.25) + (CaO/0.7)] and the CIA (Chemical Index of Alteration, as defined by Nesbitt and Young, 1982; see above) were found to best illustrate general trends of weathering in Europe. Both the WIP and CIA can be affected by several factors including the climate, the lithology of bedrock units, and the homogeneity of the parent bedrock. Therefore, in interpreting the weathering indices at the continental scale, all of the above mentioned factors should be considered. In the WIP calculation, only highly mobile alkali and alkaline earth metals (e.g., Na K, Mg, and Ca) that can be lost during the early stages of weathering have been used. For the CIA, an immobile element (Al) has been used in the calculation. By combining mobile and resistant components, we propose that parallel interpretation of CIA and WIP is most effective in

Ca Predominantly Precambrian felsic plutonic and volcanic rocks

Ti

Na Glacial sandy sediments

Maximum glaciation line

Calcareous rock

Fig. 13. Color composite map of Europe plotted with values from the ternary Ca–Ti–Na diagram (see Fig. 3 for comparison). Red = Ca, Green = Ti, Blue = Na. The map was produced using an interpolated distance weighted method and cell size 10 × 10 km. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

12


quantifying the effects of chemical weathering relative to bedrock (Fig. 1a), climate and topography. As shown in Fig. 14, a major contrast exists between the northern, southern and central parts of Europe. A break occurs along the southern border of the maximum extent of glaciation (Fig. 1a), with young soils occurring in the north and old mature soils dominating in southern Europe. In Scandinavia, the WIP shows surprisingly high values, given the young soils, while the CIA shows relatively low values; and these latter are clearly related to the early stage of soil weathering. An interesting feature is that in the central part of Scandinavia (at the Norwegian/Swedish border) both WIP and CIA show low values which can only be explained by the presence of quartz-rich rocks (e.g., quartzite, sandstone). The very low value of WIP in soil developed on top of the thick glacial sediments in northern and central Europe depicts the Si enrichment in addition to the fact that most alkali and alkaline metals are depleted in this region. At the continental scale, the patterns displayed by both ratios are thus strongly driven by the type of parent material, climatic conditions (cold climate in Scandinavia versus a warmer climate in northern Germany and Poland) and the relative proportions of chemical and physical weathering (related to

latitude and elevation). All these factors may have a modifying effect on the characteristics of weathering indices. 6. Conclusions The GEMAS agricultural soil dataset of Europe provides an excellent base for studying chemical variations in soil composition at the continental scale related to factors such as lithology, climate, topography and soil texture. Using available classification methods, GEMAS data reveal important trends in the spatial distribution of element patterns and as a result, the major source rock (parent material) can be identified. Certain elements are diagnostic for specific rock types while others, especially mobile elements, can move from one environment to another, changing their primary chemical signature. The comparison of soil mean values and their standard deviation with the average Upper Continental Crust composition shows enrichment and depletion patterns pointing to the importance of soil forming processes, element mobility, physical sorting, and variable maturity of the sample media in relation to the climate and glacial history. Element enrichment (e.g. for Al, P, Zr, Pb) and depletion (e.g. Mg, Na, Sr and Pb)

a Arctic Ocean

WIP (Weathering index of Parker)

0

250

500 km

256 - 2 330 2 340 - 3 630 3 640 - 4 750 4 760 - 5 880 5 890 - 7 700 7 710 - 11 600 11 700 - 22 300 North Sea Baltic Sea Belarus

M d ol ov a

Canary Islands

Romania Black Sea

Albania

Turkey

Mediterranean Sea

Morocco

Algeria

Tunisia

Malta

Mediterranean Sea

Fig. 14. Maps of Europe showing calculated weathering indices (a) the WIP (Weathering Index of Parker) (Parker (1970): 100 ∗ [(2Na2O/0.35) + (MgO/0.09) + (2K2O/0.25) + (CaO/0.7)] and (b) the CIA (Chemical Index of Alteration of Nesbitt and Young, 1982), 100 ∗ [(Al2O3/(Al2O3 + CaO* + Na2O + K2O))]. The maps were produced using an interpolated distance weighted method and cell size 10 × 10 km.


13

b Arctic Ocean

CIA (Chemical Index of Alteration)

0

250

500 km

Russia

8.1 - 40.5 40.6 - 52.7 52.8 - 58.4 58.5 - 63.3 63.4 - 68.6 68.7 - 74.9 75 - 92.4


M d ol ov a

Canary Islands

Romania Black Sea

Albania

Turkey

Mediterranean Sea

Morocco

Algeria

Tunisia

Malta

Mediterranean Sea

Fig. 14 (continued).

in comparison to the UCC are closely related to the variation represented by the standard deviation. The UCC extended normalized patterns applied to selected elements (Rb, K, Y, Ti, Al, Si, Zr, Ce and Fe) demonstrate their good correlation with the UCC model even when standard deviation indicates minor enrichment (e.g., Pb and Cr) or depletion (e.g., Ca and Sr). By using Al-normalized enrichment factors, the source parent material can be defined and Na-normalization allows discrimination between insoluble (Ti, Al, Si, V, Y, Zr, Ba, and La), reactive (Fe, Mn, Zn) and soluble elements (K, Rb). Additionally, elements classically associated with carbonates (Ca and Sr) and phosphates (P and Ce) can be recognized. The main conclusion emerging from the comparison of soil compositions with the mean upper continental crust composition is that element fractionation does occur during weathering, sorting, and soil formation processes. Application of weathering indices to soil chemical composition is controversial. Although well-established classifications (e.g., WIP, CIA) for surficial deposits show regional trends in element distribution patterns in Europe, there is a major problem with their interpretation because of large variations in the chemical composition of European parent material (or source lithologies), climate zones, and the chemical

decoupling of carbonate-rich and (alumino)silicate-rich soils and sediments. Acknowledgments The GEMAS project is a cooperative project of the EuroGeoSurveys Geochemistry Expert Group with a number of outside organizations (e.g., Alterra, The Netherlands; Norwegian Forest and Landscape Institute; Research Group Swiss Soil Monitoring Network, Swiss Research Station Agroscope Reckenholz-Tänikon, several Ministries of the Environment and University Departments of Geosciences, Chemistry and Mathematics in a number of European countries and New Zealand; ARCHE Consulting in Belgium; CSIRO Land and Water in Adelaide, Australia). The analytical work was co-financed by the following industry organizations: Eurometaux, European Borates Association, European Copper Institute, European Precious Metals Federation, International Antimony Association, International Lead AssociationEurope, International Manganese Institute, International Molybdenum Association, International Tin Research Institute, International Zinc Association, The Cobalt Development Institute, The Nickel Institute, The (REACH) Selenium and Tellurium Consortium and The (REACH)

14


Vanadium Consortium. The Directors of the European Geological Surveys, and the additional participating organizations, are thanked for making sampling of almost all of Europe in a tight time schedule possible. The Federal Institute for Geosciences and Natural Resourced (BGR), the Geological Survey of Norway and SGS (Canada) are thanked for special analytical input to the project. The thoughtful and thorough remarks of three anonymous reviewers and Editor Carla Koretsky were much appreciated. We are grateful to Dr Patrick O'Connor for proofreading and editing the text. References Ahrens, T.J., McLennan, S.M., 2013. Sediments and soils: chemistry and abundances. Rock Physics & Phase Relations: A Handbook of Physical Constants. American Geophysical Union http://dx.doi.org/10.1029/RF003p0008. Amiotte Suchet, P., Probst, J.-L., Ludwig, W., 2003. Worldwide distribution of continental rock lithology: implications for the atmospheric/soil CO2 uptake by continental weathering and alkalinity river transport to the oceans. Glob. Biogeochem. Cycles 17, 1038–1051. Ansari, A.A., Singh, I.B., Tobschall, H.J., 2000. Importance of geomorphology and sedimentation processes for metal dispersion in sediments and soils of the Ganga Plain: identification of geochemical domains. Chem. Geol. 162, 245–266. Bahlburg, H., Dobrzinski, N., 2011. A review of the Chemical Index of Alteration (CIA) and its application to the study of Neoproterozoic glacial deposits and climate transitions. In: Arnaud, E., Halverson, G.P., Shields, G.A. (Eds.), The Geological Record of Neoproterozoic Glaciations. Geological Society, London, Memoir 36, pp. 81–92. Birkeland, P.W., 1984. Soils and Geomorphology. Oxford University Press, New York (372pp.). Brand, N.W., 1999. Element ratios in nickel sulphide exploration: vectoring towards ore environments. J. Geochem. Explor. 67, 145–165. Braun, J.-J., Marechal, J.-C., Riotte, J., Boeglin, J.-L., Bedimo Bedimo, J.-P., Ndam Ngoupayou, J.R., Nyeck, B., Robain, H., Sekhar, M., Audry, S., Viers, J., 2012. Elemental weathering fluxes and saprolite production rate in a Central African lateritic terrain (Nsimi, South Cameroon). Geochim. Cosmochim. Acta 99, 243–270. Carpenter, R.H., Pope, T.A., Smith, R.L., 1975. Fe–Mn oxide coatings in stream sediment geochemical surveys. J. Geochem. Explor. 4, 349–363. Chester, R., Kudoja, W.M., Thomas, A., Towner, J., 1985. Pollution reconnaissance in stream sediments using non-residual trace metals. Environ. Pollut. 10, 213–238. Cicchella, D., De Vivo, B., Lima, A., 2005. Background and baseline concentration values of elements harmful to human health in the volcanic soils of the metropolitan and provincial areas of Napoli (Italy). Geochem. Explor. Environ. Anal. 5, 29–40. Cicchella, D., de Vivo, B., Lima, A., Albanese, S., Fedele, L., 2008. Urban geochemical mapping in the Campania region (Italy). Geochem. Explor. Environ. Anal. 8, 19–29. Cloquet, C., Carignan, J., Libourel, G., Sterckeman, T., Perdrix, E., 2006. Tracing source pollution in soils using cadmium and lead isotopes. Environ. Sci. Technol. 40, 2525–2530. Cullers, R.L., Basu, A., Suttner, L.J., 1988. Geochemical signature of provenance in sand-size material in soils, stream sediments near the Tobacco Root batholith, Montana, U.S.A. Chem. Geol. 70, 335–348. de Caritat, P., Reimann, C., the NGSA Team, GEMAS team, 2012. Comparing results from two continental geochemical surveys to world soil composition and deriving Predicted Empirical Global Soil (PEGS2) reference values. Earth Planet. Sci. Lett. 319–320, 269–276. Delvaux, B., Herbillon, A.J., Vielvoye, L., 1989. Characterization of a weathering sequence of soils derived from volcanic ash in Cameroon. Taxonomic, mineralogical and agronomic implications. Geoderma 45, 375–388. Dennen, W.H., Anderson, P.J., 1962. Chemical changes in incipient rock weathering. Geol. Soc. Am. Bull. 73, 375–384. Dupré, B., Gaillardet, J., Rousseau, D., Allègre, C.J., 1996. Major and trace elements of river-borne material: the Congo Basin. Geochim. Cosmochim. Acta 60, 1301–1321. Dürr, H.H., Meybeck, M., Dürr, S.H., 2005. Lithologic composition of the Earth's continental surfaces derived from a new digital map emphasizing riverine material transfer. Glob. Biogeochem. Cycles 19, 1–22. EuroGeoSurveys Geochemistry Working Group, 2008. EuroGeoSurveys Geochemical mapping of agricultural and grazing land soil of Europe (GEMAS) - Field manual. Norges Geologiske Undersøkelse Report 2008.038 (46pp. http://www.ngu.no/ upload/Publikasjoner/Rapporter/2008/2008_038.pdf). Gaillardet, J., Dupré, B., Allègre, C.J., 1995. A global geochemical mass budget applied to the Congo Basin rivers: erosion rates and continental crust composition. Geochim. Cosmochim. Acta 59, 3469–3485. Gaillardet, J., Dupré, B., Allègre, C.J., 1999. Geochemistry of large river suspended sediments: silicate weathering or recycling tracer? Geochim. Cosmochim. Acta 63, 4037–4051. Gaillardet, J., Millot, R., Dupré, B., 2003. Chemical denudation rates of the western Canadian orogenic belt: the Stikine terrane. Chem. Geol. 201, 257–279. Gallet, S., Jahn, B.-M., Van Vliet Lanoë, B., Dia, A., Rossello, E., 1998. Loess geochemistry and its implications for particle origin and composition of the upper continental crust. Earth Planet. Sci. Lett. 156, 157–172. Garrels, R.M., Mackenzie, F.T., 1971. Evolution of Sedimentary Rocks, Norton, New York (397 pp.). Garrett, R.G., Thorleifson, L.H., Matile, G., Adcock, S.W., 2008. Till geochemistry, mineralogy and lithology, and soil geochemistry—data from the 1991–1992 prairie kimberlite study. Geological Survey of Canada Open File 5582, 1 CD-ROM. Goldschmidt, V.M., 1954. Geochemistry. Oxford University, London (730 pp.).

Goldstein, S.J., Jacobsen, S.B., 1988. REE in the Great Whale River estuary, Northwest Quebec. Earth Planet. Sci. Lett. 88, 241–252. Günther, A., Reichenbach, P., Malet, J.-P., Van Den Eeckhaut, M., Hervás, J., Dashwood, C., Guzzetti, F., 2013. Tier-based approaches for landslide susceptibility assessment in Europe. Landslides 10, 529–546. Hartmann, J., Dürr, H.H., Moosdorf, N., Meybeck, M., Kempe, S., 2012. The geochemical composition of the terrestrial surface (without soils) and comparison with the upper continental crust. Int. J. Earth Sci. 101, 365–376. Herron, M.M., 1988. Geochemical classification of terrigenous sands and shales from core or log data. J. Sediment. Petrol. 58, 820–829. Hofmann, A.W., 1988. Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust. Earth Planet. Sci. Lett. 90, 297–314. Jain, C.K., Sharma, M.K., 2001. Distribution of trace metals in the Hindon River system, India. J. Hydrol. 253, 81–90. Kasanzu, C., Maboko, M.A.H., Manya, S., 2008. Geochemistry of fine-grained clastic sedimentary rocks of the Neoproterozoic Ikorongo Group, NE Tanzania: implications for provenance and source rock weathering. Precambrian Res. 164, 201–213. Key, R.M., Johnson, C.C., Horstwood, M.S.A., Lapworth, D.J., Knights, K.V., Kemp, S.J., Watts, M., Gillespie, M., Adekanmi, M., Arisekola, T., 2012. Investigating high zircon concentrations in the fine fraction of stream sediments draining the Pan-African Dahomeyan Terrane in Nigeria. Appl. Geochem. 27, 1525–1539. Langmuir, C.H., Vocke Jr., R.D., Hanson, G.N., Hart, S.R., 1978. A general mixing equation with applications to Icelandic basalts. Earth Planet. Sci. Lett. 37, 380–392. Li, C., Kang, S., Zhang, Q., 2009. Elemental composition of Tibetan Plateau top soils and its effect on evaluating atmospheric pollution transport. Environ. Pollut. 157, 2261–2265. Lupker, M., France-Lanord, C., Galy, V., Lavé, J., Gaillardet, J., Gajurel, A.P., Guilmette, C., Rahman, M., Singh, S.K., Sinha, R., 2012. Predominant floodplain over mountain weathering of Himalayan sediments (Ganga basin). Geochim. Cosmochim. Acta 84, 410–432. Négrel, Ph., 2006. Water–granite interaction: clues from strontium, neodymium and rare earth elements in soil and waters. Appl. Geochem. 21, 1432–1454. Négrel, Ph., Allègre, C.J., Dupré, B., Lewin, E., 1993. Erosion sources determined from inversion of major, trace element ratios and strontium isotopic ratios in riverwater: the Congo Basin case. Earth Planet. Sci. Lett. 120, 59–76. Négrel, Ph., Millot, R., Petelet-Giraud, E., Malcuit, E., Brenot, A., 2013. Impact of rock weathering on the chemical composition of groundwater determined by inverse modeling in large sedimentary basins. Procedia Earth Planet. Sci. 7, 615–619. Nesbitt, H.W., 1979. Mobility and fractionation of rare earth elements during weathering of a granodiorite. Nature 279, 206–210. Nesbitt, H.W., Young, G.M., 1982. Early Proterozoic climates and plate motion inferred from major element chemistry of lutites. Nature 299, 715–717. Nesbitt, H.W., Markovics, G., Price, R.C., 1980. Chemical processes affecting alkalis and alkaline earths during continental weathering. Geochim. Cosmochim. Acta 44, 1659–1666. Ottesen, R.T., Birke, M., Finne, T.E., Gosar, M., Locutura, J., Reimann, C., Tarvainen, T., the GEMAS Project Team 1, 2013. Mercury in European agricultural and grazing land soils. Appl. Geochem. 33, 1–12. Parker, R.L., 1967. Composition of the Earth's crust. U.S. Geological Survey Professional Paper 440-D (17 pp.). Parker, A., 1970. An index of weathering for silicate rocks. Geol. Mag. 107, 501–504. Pettijohn, F.J., Potter, P.E., Siever, R., 1972. Sand and Sandstones. Springer–Verlag, New York (618 pp.). Price, J.R., Velbel, M.A., 2003. Chemical weathering indices applied to weathering profiles developed on heterogeneous felsic metamorphic parent rocks. Chem. Geol. 202, 397–416. Rawlins, B.G., Webster, R., Lister, T.R., 2003. The influence of parent material on topsoil geochemistry in eastern England. Earth Surf. Process. Landf. 28, 1389–1409. Raza, M., Khan, A., Bhardwaj, V.R., Rais, S., 2012. Geochemistry of Mesoproterozoic sedimentary rocks of upper Vindhyan Group, southeastern Rajasthan and implications for weathering history, composition and tectonic setting of continental crust in the northern part of Indian shield. J. Asian Earth Sci. 48, 160–172. Reimann, C., de Caritat, P., 2005. Distinguishing between natural and anthropogenic sources for elements in the environment: regional geochemical surveys versus enrichment factors. Sci. Total Environ. 337, 91–107. Reimann, C., Demetriades, A., Eggen, O.A., Filzmoser, P., the EuroGeoSurveys Geochemistrv Expert Group, 2011. The EuroGeoSurveys GEochemical Mapping of Agricultural and grazing land Soils project (GEMAS) Evaluation of quality control results of total C and S, total organic carbon (TOC), cation exchange capacity (CEC), XRF, pH, and particle size distribution (PSD) analysis. NGU Report No. 2011.043 (90pp.). Reimann, C., Filzmoser, P., Fabian, K., Hron, K., Birke, M., Demetriades, A., Dinelli, E., Ladenberger, A., the GEMAS Project Team 1, 2012a. The concept of compositional data analysis in practice — total major element concentrations in agricultural and grazing land soils of Europe. Sci. Total Environ. 426, 196–210. Reimann, C., Flem, B., Fabian, K., Birke, M., Ladenberger, A., Négrel, Ph., Hoogewerff, J., 2012b. Lead and lead isotopes in agricultural soils of Europe — the continental perspective. Appl. Geochem. 27, 532–542. Reimann, C., de Caritat, P., GEMAS Project Team 1, NGSA Project Team 2, 2012c. New soil composition data for Europe and Australia: demonstrating comparability, identifying continental-scale processes and learning lessons for global geochemical mapping. Sci. Total Environ. 416, 239–252. Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), 2014a. Chemistry of Europe's Agricultural Soils — Part A: Methodology and Interpretation of the GEMAS Dataset. Geologisches Jahrbuch (Reihe B), Schweizerbarth, Stuttgart (528pp.). Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), 2014b. Chemistry of Europe's Agricultural Soils — Part B: General Background Information and Further

P. Négrel et al. / Chemical Geology 396 (2015) 1–15 Analysis of the GEMAS Dataset. Geologisches Jahrbuch (Reihe B), Schweizerbarth, Stuttgart (352pp.). Romic, M., Romic, D., 2003. Heavy metals distribution in agricultural topsoils in urban area. Environ. Geol. 43, 795–805. Roser, B.P., Korsch, R.J., 1986. Determination of tectonic setting of sandstone–mudstone suites using SiO2 content and K2O/Na2O ratio. J. Geol. 94, 635–650. Rudnick, R.L., Gao, S., 2003. Composition of the continental crust. In: Holland, H.D., Turekian, K.K., Rudnick, R.L. (Eds.), Treatise on Geochemistry. The Crust vol. 3. Elsevier-Pergamon, Oxford, pp. 1–64. Sadeghi, M., Petrosino, P., Ladenberger, A., Albanese, S., Andersson, M., Morris, G., Lima, A.M., de Vivo, B., the GEMAS Project Team 1, 2013. Ce, La and Y concentrations in agricultural and grazing-land soils of Europe. J. Geochem. Explor. 133, 202–213. Schucknecht, A., Matschullat, J., de Caritat, P., da Silva, J., Melo Jr., G., Pleßow, A., Mello, J.W.V., 2012. Pedogeochemistry in NE-Brazil — compared to Australia and Europe. Sci. Total Environ. 438, 342–356. Servaraj, K., Arthur, C.C., 2006. Moderate chemical weathering of subtropical Taiwan: constraints from solid-phase geochemistry of sediments and sedimentary rocks. J. Geol. 114, 101–116. Sposito, G., 1989. The Chemistry of Soils. Oxford University Press, New York (277pp.). Stanley, C.R., Noble, R.P., 2008. Quantitative assessment of the success of geochemical exploration techniques using minimum probability methods. Geochem. Explor. Environ. Anal. 8, 115–127. Sterckeman, T., Douay, F., Baize, D., Fourrier, H., Proix, N., Schvartz, C., 2006. Trace elements in soils developed in sedimentary materials from Northern France. Geoderma 136, 912–929. Stummeyer, J., Marchig, V., Knabe, W., 2002. The composition of suspended matter from Ganges–Brahmaputra sediment dispersal system during low sediment transport season. Chem. Geol. 185, 125–147.

15

Sucharovà, J., Suchara, I., Hola, M., Marikova, S., Reimann, C., Boyd, R., Filzmoser, P., Englmaier, P., 2012. Top-/bottom-soil ratios and enrichment factors: what do they really show? Appl. Geochem. 27, 138–145. Takeda, A., Kimura, K., Yamasaki, S., 2004. Analysis of 57 elements in Japanese soils, with special reference to soil group and agricultural use. Geoderma 119, 291–307. Tarvainen, T., Albanese, S., Birke, M., Ponavic, M., Reimann, C., the GEMAS Project Team 1., 2013. Arsenic in agricultural and grazing land soils of Europe. Appl. Geochem. 28, 2–10. Taylor, S.R., McLennan, S.M., 1995. The geochemical evolution of the continental crust. Rev. Geophys. 33, 241–265. Taylor, S.R., McLennan, S.M., McCulloch, M.T., 1983. Geochemistry of loess, continental crustal composition and crustal model ages. Geochim. Cosmochim. Acta 47, 1897–1905. Viers, J., Dupré, B., Gaillardet, J., 2009. Chemical composition of suspended sediments in World Rivers: new insights from a new database. Sci. Total Environ. 407, 853–868. Vital, H., Stattegger, K., 2000. Major and trace elements of stream sediments from the lowermost Amazon River. Chem. Geol. 168, 151–168. Wakatsuki, T., Furukawa, H., Kyuma, K., 1977. Geochemical study of the redistribution of elements in soil-I. Evaluation of degree of weathering of transported soil materials by distribution of major elements among the particle size fractions and soil extract. Geochim. Cosmochim. Acta 41, 891–902. Yang, S., Jung, H.S., Li, C., 2004. Two unique weathering regimes in the Changjiang and Huanghe drainage basins: geochemical evidence from river sediments. Sediment. Geol. 164, 19–34.