Lithologically inherited variation in Pb isotope ratios in sedimentary soils in The Netherlands

Applied Geochemistry 37 (2013) 228–241

Contents lists available at SciVerse ScienceDirect

Applied Geochemistry

journal homepage: www.elsevier .com/ locate /apgeochem

Lithologically inherited variation in Pb isotope ratios in sedimentarysoils in The Netherlands

0883-2927/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.apgeochem.2013.06.006

⇑ Corresponding author.E-mail address: [email protected] (N. Walraven).

N. Walraven a,⇑, P.F.M. van Gaans b, G. van der Veer c, B.J.H. van Os d, G.Th. Klaver e, S.P. Vriend f,J.J. Middelburg f, G.R. Davies g

a GeoConnect, Meester Dekkerstraat 4, 1901 PV Castricum, The Netherlandsb Deltares, Soil and Groundwater Systems/Geological Survey of The Netherlands, P.O. Box 85467, 3508 AL Utrecht, The Netherlandsc Rikilt Institute of Food Safety, P.O. Box 230, 6700 AE Wageningen, The Netherlandsd Rijksdienst voor Archeologie, Cultuurlandschap en Monumenten, P.O. Box 1600, 3800 BP Amersfoort, The Netherlandse BRGM, 3 Avenue Claude-Guillemin, BP 36009, 45060 Orléans Cedex 2, Francef University Utrecht, Faculty of Geosciences, P.O. Box 80021, 3508 TA Utrecht, The Netherlandsg VU University Amsterdam, Faculty of Earth and Life Sciences, Petrology, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands

a r t i c l e i n f o

Article history:Available online 27 June 2013Editorial handling by C. Reimann

a b s t r a c t

Knowledge on the lithologically inherited variation in present day Pb isotope ratios in soils is remarkablylimited. Such information is essential to determine the anthropogenic Pb fraction and anthropogenic Pbsources in Pb-polluted soils. This study presents results of a survey of subsoil samples from approxi-mately 350 rural locations covering the entire Netherlands, for which the bulk geochemical and Pb iso-tope composition was determined. The sample density was approximately 1 site per 70 km2. The aim wasto establish a geochemical reference for the lithologically inherited variation in Pb isotope ratios in Dutchsoils based on the subsoil samples, with which to compare the topsoils (companion paper in this journalissue).

The lithologically inherited variation in Pb isotope ratios of the subsoils in The Netherlands is estab-lished at 1.175–1.221, 2.441–2.494 and 0.478–0.492 for 206Pb/207Pb, 207Pb/208Pb and 206Pb/208Pb respec-tively. The four main lithologies distinguished, sand, clay, peat and loess, have distinct Pb isotopesignatures. No significant difference in isotope signature was found between marine and fluviatile clays.Multiple regression analysis established that the observed variation can be primarily explained by thetextural and mineralogical variation within Dutch subsoils, with Al and Zr content representing usefulpredictors for the observed Pb isotope variability. Clay soils are characterised by a radiogenic Pb isotopesignature that is notably low in 207Pb. Soils with a high Zr content are especially high in 206Pb. Althoughthe vast majority (�90%) of the Pb isotope variation in the subsoils appears to be controlled by litholog-ical inheritance, some subsoils (mainly peats) are suspected of containing a component of non-litholog-ically derived Pb.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Lead has an average crustal abundance of 17 mg/kg (Wedepohl,1995). Because of the malleability of the pure metal and its specificchemical and physical properties, Pb ores have been mined since5000 B.C. (Patterson et al., 1970; Heskel, 1983). As a result of theproduction (including mining and smelting) and the use of the me-tal (batteries, pigments, ceramics, plastics, etc.), combustion of coaland leaded gasoline, and the recycling and disposal of Pb and Pbcontaining products, anthropogenic Pb has entered the environ-ment on a worldwide basis. It is estimated that in the environment,anthropogenic Pb emissions are at least 1–2 orders of magnitudegreater than from natural sources (Komárek et al., 2008).

The World Health Organisation lists Pb as a neurotoxin with noknown biological benefit to humans. It is a cumulative poison thataffects the development and functioning of the brain, especially inyoung children. It can further damage other organs, such as thereproductive system and kidneys, and can cause high blood pres-sure and anaemia. At very high levels, Pb may lead to convulsions,coma and death. Because of its toxicity, the use of Pb has been dis-couraged over the years. However, despite current caution, Pb inthe environment still poses a serious problem. In The Netherlandsalone, several thousand sites, including back yards of familyhomes, are estimated to be polluted with Pb (Hagens et al., 2009;Compendium, 2011).

Information on the occurrence, behaviour and fate of the vari-ous anthropogenic Pb sources in the environment is consideredto be crucial in understanding and abating the Pb problem (e.g.Steele et al., 1990; Ruby et al., 1999; Rieuwerts et al., 2000; Hagens

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 229

et al., 2009). This includes distinguishing historic from recent pol-lution, as the former of course cannot be prevented; identifying thesource-related chemical form of environmental Pb, as this mayinfluence its toxicity; and the need sometimes to identify the originof Pb pollution for legal reasons.

Based on measurements of Pb content alone, it is often difficult,if not impossible, to determine the cause of the elevated Pb contentand to distinguish between natural and anthropogenic origins andamong specific anthropogenic Pb sources. Chow and Johnston(1965) were some of the first researchers to show that clues onPb provenance could be obtained through the study of the four sta-ble Pb isotopes: 204Pb, 206Pb, 207Pb and 208Pb. Of these, 204Pb is non-radiogenic and 206Pb, 207Pb and 208Pb are formed by the radioactivedecay of 238U, 235U and 232Th, respectively. Lead ores, and the prod-ucts derived from them, have characteristic Pb isotope signatures,depending on the initial U/Pb and Th/Pb ratios of the source rocksand the geological age at which Pb, as the daughter, separated fromthe melt (mother) during ore formation. In their use as geochemi-cal tracers, Pb isotopes are commonly utilised as ratios; with the206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios often used in envi-ronmental studies as they can be determined rapidly and preciselywith ICP-MS and generally exhibit sufficient variability betweendifferent sources (Komárek et al., 2008). Sangster et al. (2000) cal-culated that 86% of the discriminative power of Pb isotope signa-tures is due to the 206Pb, 207Pb and 208Pb isotopes.

In many environmental studies, Pb isotope analysis has provedto be a powerful tool. However, Pb isotope measurements can onlybe used to trace anthropogenic Pb sources in polluted soils andsediments if:

1. the Pb isotope composition of anthropogenic Pb differs fromthat of lithologically inherited Pb and

2. the content and isotopic composition of lithologically inheritedPb are known, to allow correction for its presence.

Polluted soils always contain lithologically inherited as well asanthropogenic Pb. If the ratio of anthropogenic Pb versus litholog-ically inherited Pb is high, the influence of the latter on the Pb iso-tope composition is negligible. Conversely, if this ratio is small, thepresence of lithologically inherited Pb dominates the Pb isotopecomposition of the total sample. For this reason, knowledge ofthe lithologically inherited variability and spatial distribution ofthe Pb content and the Pb isotope composition of sediments andsoils is indispensable in environmental tracer studies involving Pb.

To the authors’ knowledge, there are few extensive studies onthe spatial distribution of the Pb isotope composition of soils. Nota-ble exceptions are Reimann et al. (2011, 2012), who determined Pbcontents and Pb isotope ratios in topsoils collected (1) along aneast–west transect across the USA at 40 km resolution and (2) inAp horizons of agricultural areas in Europe at an average densityof 1 site/2500 km2. All other studies of Pb in mineral soils are localin nature and based on a limited number of selected profiles thatare sampled at various depths, to distinguish the hypothesisedanthropogenic Pb isotope signature in the upper (and generallymore organic rich) part of the soil from the lithologically inheritedsignature at depth (Chow, 1970; Gulson et al., 1981; Walravenet al., 1997; Hansmann and Köppel, 2000; Cloquet et al., 2006).Although Reimann et al. (2011, 2012) determined the spatial distri-bution of the Pb isotope composition in soils, they did not distin-guish lithologically inherited from anthropogenic Pb. Their Pbisotope maps consequently represent a mixture of lithologicallyinherited and anthropogenic Pb in topsoils.

In this study, results are presented of a survey of subsoil sam-ples of approximately 350 rural locations covering the entire Neth-erlands, of which the bulk geochemical (Pb content) and Pb isotopecompositions were determined. The aim of this study is to deter-

mine the lithologically inherited variation in Pb isotope ratios insubsoils in The Netherlands. A proper understanding and adequatedescription of this variation will serve as a base for the analysis ofspatial patterns of anthropogenic Pb isotope signatures in the top-soils in a companion article in this issue (Walraven et al., 2013).

2. Background information

2.1. General geology of The Netherlands

The Netherlands is located in north-western Europe and is bor-dered by Germany, Belgium and the North Sea (Fig. 1). It is part ofthe subsiding North Sea Basin that is enclosed by the Brabant Mas-sif in the South and the Rhenisch Massif in the East. As a result ofextensive subsidence and sedimentary infill of the North Sea Basinduring the Cenozoic, the subsurface of The Netherlands largelyconsists of a 200–500 m thick layer of unconsolidated sedimentsthat are of Pleistocene age in the eastern and southern part andof Holocene age in the northern and western part. (see e.g., Zagwijnand Van Staalduinen, 1975; Westerhoff et al., 2003).

The Pleistocene sediments are made up of coarse river depositsand various glaciogenic deposits, which are overlain by a substan-tial layer of aeolian cover sand. The coarse river deposits have botha southern provenance (drainage basins of Rhine, Meuse andScheldt) as well as a partly eastern provenance (drainage basin ofBaltic rivers). The glaciogenic deposits, which are confined to thenorthern part of The Netherlands, are largely derived from Scandi-navian massifs. As a result, the overlying Pleistocene aeolian sandshave a diverse origin in terms of sediment provenance. Yet, litho-logical and mineralogical differences within the aeolian depositsare small. Other Pleistocene aeolian sediments include loess depos-its, which are locally found in the south-eastern parts of TheNetherlands.

The Holocene sediments consist of more silty to clayey marineand fluviatile sediments alternating with extensive peat layers.Apart from these locally formed peat layers, the provenance ofthe Holocene deposits is restricted to the drainage basins of theRhine, Meuse and Scheldt.

2.2. General pedology of The Netherlands

For classification purposes, soil in The Netherlands is arbitrarilydefined as the first 120 cm of the profile below the litter layer(Locher and de Bakker, 1987). The depth of actual soil formation,if even present, is often much less. Due to intensive sediment depo-sition, soil profiles often contain various sediment layers withdifferent lithological characteristics (Van der Veer, 2006). Leadmineralisation and soil formation on hard bedrock (only in theWinterswijk area and most southern part of Limburg) does notoccur in The Netherlands.

The different types of parent material in The Netherlands areclassically grouped into five districts: sand, loess, peat, fluviatileand marine clay. Whereas the largely Pleistocene sand and loessdistricts are mainly found in the elevated northern, eastern andsouthern parts of The Netherlands, the Holocene clay depositsare mainly confined to the low lying coastal areas in the SW, westand north (Fig. 1). Peat is found in various areas in The Nether-lands, but a substantial fraction has been excavated for fuel overthe last centuries (Bazelmans et al., 2011).

Time is an important soil forming factor. Soil formation in TheNetherlands started only after ice receded. Therefore, soils in TheNetherlands are relatively young: 12 ka for loess to <50 a for mar-ine clay (for details see Van der Veer, 2006). As a result of the re-stricted age of the parent material and a temperate climate, soilsin The Netherlands are poorly developed compared to those in

Fig. 1. Left: Overview of north-western Europe showing The Netherlands (darker grey) and surrounding countries (basemap data from www.esri.com). Right: lithologicalmap of the soils in The Netherlands (Dutch coordinate system, Rijksdriehoeks-projection). (Source: Van der Veer, 2006; with permission)

230 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

many other parts of the world (Van der Veer, 2006). Thereby, thehuman influence on the soil and soil profile (deep-ploughing, mix-ing, fertilisation) has been so extensive in The Netherlands that itshould be considered as a important soil forming factor in itself(Van der Veer, 2006).

2.3. Natural Pb sources

Lead occurs naturally in all soils (de Treville, 1964). Its terres-trial abundance ranges from 1 to 200 mg/kg with most valuesbeing in the range of 5 to 25 mg/kg (Swaine, 1955; USGS, 1976).Several researchers have determined baseline levels for Pb in TheNetherlands by sampling subsoils in rural areas at depths generally>0.5 m below surface (Huisman, 1998; Lamé et al., 2004; Spijker,2005; Van der Veer, 2006). According to these studies the naturalPb content of Dutch soils ranges from <0.05 to 55 mg/kg.

The dominant mineral groups in Dutch soils and sediments arequartz, feldspars, micas, chlorites, clay minerals, heavy mineralsand secondary minerals (e.g., CaCO3 and Fe oxides). All mineralscontain ‘natural’ trace contents of Pb. The Pb content of some dom-

Table 1The Pb content of some dominant minerals that (can) occur in sedimentary soils.

Group Mineral Pb

Me

Quartz Quartz <1.Feldspars K feldspar, pegmatitic 98Feldspars K fedspar, granitic 53Feldspars Plagioclase feldspars 19.Micas Biotites 21Micas Muscovites 26Clay minerals Fluviatile clays, type of clay unknown 24Clay minerals Marine clays 16Heavy minerals Zircon (lamproite) 6.6Heavy minerals Zircon (carbonatite) 7.7Heavy minerals Zircon (kimberlite) 0.2Heavy minerals Apatite 102Secondary minerals Pyrites, sedimentary 62Secondary minerals Pyrites, sedimentary –

inant minerals that can occur in sedimentary soils is shown in Ta-ble 1. High Pb contents, up to 10,000 mg/kg, can be encountered inK feldspars. In addition, heavy minerals can contain significantamounts of Pb. For example, the Pb content of apatites, pyritesand zircons can vary from 5–2000 mg/kg, 0.3–1000 mg/kg and<0.03 to 69 mg/kg, respectively (Table 1). In addition, Pb, like manyother metals, has a high affinity for Fe and Mn oxides. The clay frac-tion of soils contains the residual Pb of weathered rocks fromwhich they have formed. The average Pb content of fluviatile claysand marine clays is 24 and 16 mg/kg, respectively (Table 1).

3. Methods

3.1. General approach and assumptions

Subsoil samples collected by Van der Veer (2006) were used inthis study. Van der Veer (2006) performed a geochemical soil sur-vey of The Netherlands based on a top–bottom sampling approach.One of the aims of the survey of Van der Veer (2006) was to gain an

(mg/kg) n References

an Range

0 0–5 15 Wedepohl and Doe (1974)0.9–10,000 219 Wedepohl and Doe (1974)3–700 419 Wedepohl and Doe (1974)

5 1–75 61 Wedepohl and Doe (1974)4–95 259 Wedepohl and Doe (1974)5–77 32 Wedepohl and Doe (1974)6–53 28 Van der Veer (2006)6–34 115 Van der Veer (2006)<1–29 34 Belousova et al. (2002)0.5–18 37 Belousova et al. (2002)

9 <0.03–69 284 Belousova et al. (2002)5–2000 58 Wedepohl and Doe (1974)2–1000 297 Wedepohl and Doe (1974)0.3–666 8 Huisman (1998)

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 231

accurate overview of the natural range of the soil composition inThe Netherlands. Since the topsoil layer, both in forests and naturalareas, has been affected by anthropogenic input (e.g., Blaser et al.,2000; Baize and Sterckeman, 2001; Steinnes, 2001; Anderson et al.,2002; Hernandez et al., 2003), so topsoil is not useful as a proxy forthe natural composition of the soil. Van der Veer (2006), therefore,used the soil parent material, or C-horizon (called subsoil in thisstudy), as a proxy for the natural composition of soils in The Neth-erlands. This approach has been used in many regional geochemi-cal surveys in The Netherlands (Mol, 2002; Lamé et al., 2004;Spijker, 2005) as well as in other countries (e.g., Salminen and Tar-vainen, 1995; Ódor et al., 1997; Reimann et al., 2000; Sterckemanet al., 2004, 2006a,b). This approach assumes that the effects of dif-fuse anthropogenic input of elements on average has had a verylimited effect on the composition of the deeper soil layer (subsoil).Additionally, Al was used as indicator for textural variation. Vander Veer (2006) and others (e.g. Huisman, 1998; Spijker, 2005)have observed a linear relationship between the contents of Aland various (trace) elements, including Pb, in the soil parent mate-rial (subsoils) in The Netherlands. This correlation is in part due tothe joint occurrence of Al and trace metals in alumino-silicates(a.o., feldspars and clay minerals), partly to the presence of high-metal-containing minerals predominantly in the clay fraction andthe higher adsorption capacity of clayey soils in comparison to san-dy soils. Subsoil samples with a deviating Pb/Al ratio might containadditional, ‘anomalous’ Pb not representing lithologically inheritedPb. Van der Veer (2006) used the relationship between Al and Pb(i.e., Pb/Al ratio) to calculate the enrichment/depletion of Al in top-soils in the Netherlands (see also Walraven et al.,2013). In thisstudy deviating Pb/Al ratios in the subsoil are used to identify out-liers in the subsoil (see Section 3.5.2).

The sampling approach of Van der Veer (2006) was restricted bytime and costs (some 800 samples could be analysed). Sinceemphasis was put on the spatial variation of the soil composition,sampling density was maximised. Therefore, from each soil profileonly two samples were taken; one from the parent material (sub-soil) and one from the topsoil layer. In general, the sampling ap-proach and procedures of the survey of Van der Veer (2006)largely conform to international recommendations set for regionalgeochemical baseline surveys, such as IGCP 259 and FOREGS(Darnley et al., 1995; Plant et al., 1997). The major differences withrespect to these recommendations are: (i) no compositing (singlespot sample), (ii) a variable depth of the deeper soil sample and(iii) a more narrowly defined target area. ICGP 259 and FOREGSfurthermore recommend sampling a wide variety of sample media,such as soil, overbank, litter material and (surface) water. Van derVeer (2006) focused on soil and soil parent material.

3.2. Soil sampling

During an extensive soil sampling campaign in The Nether-lands, Van der Veer (2006) collected soil samples using a prede-fined scheme for optimal unbiased sampling. This selectionconsists of soils used for agricultural and semi-natural purposes,excluding the potentially contaminated soils in urbanised andindustrial areas. The predefined area covered by this selection,amounts to about 25,600 km2, comprising �75% of the Dutch landsurface (sampling density of �1 sample per 70 km2).

The predefined sampling area was cross-stratified according tothe soil parent material districts in Fig. 1 (sand, loess, peat, marineclay and fluviatile clay) as well as to land use type (agricultural andsemi-natural). Within these strata, 400 sample locations were ran-domly allocated, of which finally 358 were sampled (for details seeVan der Veer, 2006).

At each sample location, a subsoil sample was taken from thesoil parent material (BC, or C horizon, 100–120 cm by default) as

well as from the topsoil (A-horizon, 0–20 cm). This study centreson the results of 342 subsoil samples that were analysed for Pb iso-topes as well as major and trace element composition.

3.3. Sample preparation

Prior to analysis, soil samples were dried at 40 �C to constantweight, disaggregated and sieved (<2 mm). To create representa-tive subsamples for laboratory analysis as well as a sample archive,the material was split into 20–30 g aliquots using a mechanicalsplitter. At this stage, sample duplicates and a standard samplewere also inserted for quality evaluation purposes. The subsamplesfor analysis were thereafter ground (<15 lm) using an automatedtungsten-carbide mill (Herzog HSM-HTP). The grinding procedureensured homogeneous samples, of which representative subsam-ples could be taken for each analytical method (for details seeVan der Veer, 2006).

3.4. Analysis

The subsoil samples were analysed for major and trace ele-ments using X-ray Fluorescence (XRF) spectrometry. XRF analysiswas conducted on pressed powder tablets using an ARL9400 spec-trometer with a Rh-tube and a full matrix correction for the majorelements using over 100 geological reference material samples. Fortrace elements the Rh-compton peak was used to correct for ma-trix differences. In addition, the organic matter (OM) contents weredetermined by Thermo-Gravimetric Analysis (LECO 604 TGA). Allcontents are either expressed as wt% or as mg/kg, all based on40 �C dry sample weight.

Although a broad range of elements was analysed (Van derVeer, 2006), only Pb, Al, Zr and OM content are discussed here.The OM content is used to recalculate the Pb, Zr and Al contentto an organic-free basis (see Section 3.4.1). Zirconium and Al are in-cluded because they are independent proxies for lithologicallyinherited Pb (see Section 3.4.3). For the elements considered in thisstudy, Pb, Al, Zr, and OM-content, the relative precision rangesfrom 3.2% to 9.6%, twice the relative standard deviation (RSD)(based on sample duplicates with values >3� limit of quantifica-tion (LOQ)). The accuracy for Pb, Al, Zr, OM-content, as determinedby soil standard ISE 921 (www.wepal.nl) – is 1.9%, 14%, 4.2% and0.2%, respectively (relative bias). Precision and accuracy are basedon the entire analytical procedure starting with the sample splits.See Van der Veer (2006) for details.

Lead isotope analysis was performed after HF-based sampledestruction. Details of the HF-based sample destruction procedurecan be found in Van der Veer (2006). After destruction, sampleswere measured for Pb isotopes using an Agilent 7500 inductivelycoupled plasma mass spectrometer (ICP-MS) with a low uptakenebulizer. In addition, total Pb contents were determined by ICP-MS and compared to those derived from XRF determination, to val-idate that Pb was totally extracted by the HF-based destructionprocedure. Lead contents measured by ICP-MS match those deter-mined by XRF (logPbicp-ms = 0.97 � logPbxrf + 0.009; n = 336;R2 = 0.91). Lead content determined by XRF is henceforth used inthis study.

To determine Pb isotope ratios accurately using ICP-MS, a meth-od similar to that described by Krachler et al. (2004) was used. Todiminish mass bias due to concentration differences, the HFdestruction solutions as obtained for trace element analysis werediluted with 1 M HNO3 to a Pb concentration of about 50 lg/kgin solution prior to being introduced into the mass spectrometer.Data were taken in the peak jumping mode with three data pointsacquired across each peak at masses m/z 206, 207 and 208; dwelltime for these masses was 20 ms/channel. Ten runs were used foreach sample and the average of the ten ratios for 206Pb/207Pb and

232 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

208Pb/207Pb and 206Pb/208Pb determined. For external calibrationand drift, a series of six samples was bracketed by NIST SRM 981.Ratios were corrected independently (for mass bias), i.e. a correc-tion factor was calculated according to the reference value in thestandard for each ratio, separately. Ratios were corrected assuminga linear drift relative to their measurement position from bothbracketing standard samples. The precision based on 37 duplicatepairs was at 2 RSD < 0.22% for 206Pb/207Pb, 2 RSD < 0.17% for208Pb/207Pb, and 2 RSD < 0.44% for 206Pb/208Pb. The average mea-sured value and absolute precision (2 SD) of the measured ISE921 sample (n = 45) was 1.166 ± 0.003 for 206Pb/207Pb,2.444 ± 0.004 for 208Pb/207Pb, and 0.477 ± 0.002 for 206Pb/208Pb.Accuracy and precision are based on the entire analytical proce-dure starting with the sample splits. Blanks were also measuredand were found to contain negligible amounts of Pb (<20 ng kg�1).The analytical results of 6 samples out of the total 342 did not passthe final quality control criteria and were, therefore, not includedin this study.

3.5. Data analysis

3.5.1. Correction for the organic fractionUltimately, the topsoil, which has undergone soil formation and

anthropogenic influence, is to be compared with the parent mate-rial (companion article by Walraven et al., 2013). To adequatelycompare trace element contents in topsoil with those in the parentmaterial, the dilution effect of the organic fraction should be cor-rected for in some way. Here, contents are, therefore, recalculatedto an organic free basis:

½C�recalc ¼ ½C� � 100=ð100� OM%Þ ð1Þ

with [C] the content of any constituent (in% or mg kg�1) on a dry to-tal mass basis, and [C]recalc the content on a dry organic free massbasis. OM% is the content of organic matter on a dry total mass ba-sis. This correction for the organic fraction has no influence on thePb isotope ratios themselves, as within one sample the correctionfactor is the same for all analytes.

In principle, recalculation could also have been performed to acarbonate-free basis. This was not done because the correctionswould be very minor (on average less than about 1/10th of thatfor OM; Van der Veer, 2006), and would require detailed knowl-edge on the carbonate mineralogy (presence of primary and sec-ondary carbonate and non-unique attribution of Ca to CaCO3).

3.5.2. Identifying outliersWhile the set of subsoil samples is assumed to primarily repre-

sent the lithologically derived variation in Pb content and Pb iso-tope composition, some of the samples might contain additional,‘anomalous’ Pb. For example, subsoils could be polluted with Pbfrom anthropogenic sources derived from the topsoil, both natu-rally and during sampling. Walraven et al. (1997, 2001a, b) showedthat Dutch rural soils and sediments can be polluted to depths of100–120 cm (default subsoil sampling depth in this study) due tohuman activities. In addition, subsoils might contain Pb that isnot representative of the general lithologically inherited variationas a result of precipitation/adsorption of Pb transported by upwardseeping groundwater. Since Pb mineralisation does not occur in theshallow subsurface of The Netherlands (<20 m below surface),anomalous Pb isotope compositions in subsoils will not derivefrom such mineralisation.

The presence of even small quantities of anomalous Pb can havea large influence on the overall Pb isotope composition. To mini-mise the influence of Pb from other origins on the statistical assess-ment of the lithologically inherited variation in Pb isotope ratios insubsoils, two selection criteria have been formulated to identifysuch outliers in the dataset. The first criterion is based on total

Pb content and the common relationship between Pb and Al inunpolluted soils and sediments (see Section 3.1), here expressedas Pbrecalc/Alrecalc, or its equivalent Pb/Al. Sonke et al. (2002) sug-gested that Al can be enriched at extremely polluted sites and thatcare should be taken in using Al as a lithogenic proxy to identify Pbpollution. However, the selected sites in this study are all ruralsites, not located close to large industrial sites. Subsoils withPbrecalc/Alrecalc values smaller than the first quartile or larger thanthe third quartile plus several times the interquartile range are la-belled as outliers (Eqs. (2) and (3)).

Pbrecalc=Alrecalc < P25 � f � ðP75 � P25Þ ð2Þ

Pbrecalc=Alrecalc > P75 þ f � ðP75 � P25Þ ð3Þ

This procedure is equivalent to calculating the fences for boxand whiskers plots according to Tukey (1977). In this formula Pdenotes a percentile and Pbrecalc/Alrecalc is calculated based onthe OM-free Pb and Al content in the subsoils (Eq. (1)). The factorf is set to 1.5 as is commonly done for inner fences (Spijker,2005). The second criterion is based on identifying deviating Pbisotope ratios in the dataset. The same procedure as in criterion1 is followed (see Eq. (2) and (3)), but with 206Pb/207Pb,208Pb/207Pb and 206Pb/208Pb values instead of Pbrecalc/Alrecalc

values. All subsoils that passed criterion 1 were tested in criterion2.

Outliers were not completely removed from the dataset butflagged as such so they could be treated separately. Only the sub-soils that passed both criteria were used to determine the generalstatistics of the lithologically inherited Pb isotope composition ofsubsoils in The Netherlands.

3.5.3. Regression model for lithology dependent variation of Pb isotoperatios

Lithology (soil parent material) will probably influence thevariation in Pb isotope ratios, but to the authors’ knowledge, nosystematic studies have been undertaken on this issue. An at-tempt was made to identify which geochemical proxies or predic-tors could be used to predict Pb isotope ratios. A power relation(log–log regression) was used rather than a linear relation, asfor most trace elements it better describes the relative variabilityin content. For the Pb isotope ratios, using a log–log relation hasthe added advantage that the way the ratios are expressed (A/B orB/A) has no influence on the regression results, since the log valueof a ratio and its inverse differ only in sign but not in number. Toeliminate the effects of outliers, not only of those already identi-fied as being anomalous for Pb or Pb isotope composition, butalso of samples being anomalous for the selected predictors, a ro-bust trimmed least squares regression was used (Montgomeryand Peck, 1982; Rousseeuw and Van Driessen, 2002; Van Gaanset al., 2007).

Mass bias correction was applied independently for the threePb-isotope ratios: 206Pb/207Pb, 208Pb/207Pb, and 206Pb/208Pb (seeabove). This means that the ratio of 206Pb/207Pb and 208Pb/207Pbmay differ slightly from the given value for 206Pb/208Pb (and simi-larly for other combinations) as a consequence of measurement er-ror. A model that predicts the isotope ratios based on bulksediment chemistry of course should observe the actual depen-dency. Therefore, a combined regression model is formulated asfollows:

logð206Pb=207PbÞ ¼ a1 þ b1 logðXrecalcÞ þ c1 logðYrecalcÞ þ d1

� logðZrecalcÞ þ � � � ð4aÞ

logð208Pb=207PbÞ ¼ a2 þ b2 logðXrecalcÞ þ c2 logðYrecalcÞ þ d2

� logðZrecalcÞ þ � � � ð4bÞ

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 233

logð206Pb=208PbÞ ¼ ða1 � a2Þ þ ðb1 � b2Þ logðXrecalcÞ þ ðc1

� c2Þ logðYrecalcÞ þ � � � ð4cÞ

with Xrecalc, Yrecalc, etc. = measured bulk element chemistry basedon organic matter-free basis in mg/kg, and 206Pb/207Pb, 208Pb/207Pb,206Pb/208Pb = measured Pb isotope ratios. All ratio data are com-bined to calculate the robust optimal values for a1, a2, b1, b2, etc.

3.5.4. Graphical representationThe geographical distribution of the Pb isotopes (and relevant

bulk geochemical parameters) is presented in the form of bubbleplot maps. The Pb isotope ratios are plotted in triangular plots aswell as in more common bivariate plots of the Pb isotope ratios(so-called 3-isotope plots with a common isotope denominator).Both bivariate plots with 207Pb and 208Pb as denominator are pre-sented, as 207Pb is commonly used as denominator in environmen-tal Pb isotope studies and plots with 208Pb as denominator give thehighest resolution.

4. Results

4.1. Analytical results

The results of the outlier identification, as described in Sec-tion 3.5.2, are visualised in Fig. 2. This figure shows that the major-ity (n = 303) of the subsoil samples have Pbrecalc/Alrecalc ratios in a

Fig. 2. Left: Box-whisker plots of Pbrecalc/Alrecalc, 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pbsample depth is 100–120 cm below surface). Subsoil samples that have Pbrecalc/Alrecalc

The Netherlands. d Subsoil samples that have deviating Pbrecalc/Alrecalc ratios accordinghave deviating Pb isotope ratios according to criterion 2.

Fig. 3. Measured 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios versus Pbrecalc/Alrecalc. Sthe lithologically inherited variation in The Netherlands. d Subsoil samples that havanthropogenic Pb). � Subsoils samples that have deviating Pb isotope ratios according t

distinct range between 2.5 and 7.7 that represents the lithologi-cally inherited variation in subsoils. A total of 33 samples - 2 fluvi-atile clays, 13 peats and 18 sands - have deviating Pbrecalc/Alrecalc

ratios or Pb isotope ratios (Fig. 2). The samples with enhancedPbrecalc/Alrecalc ratios (n = 27) have relatively low 206Pb/207Pb,208Pb/207Pb and 206Pb/208Pb ratios (Fig. 3), all being indicators forthe presence of additional, probably anthropogenic Pb. In addition,the Curecalc/Alrecalc ratios (<limit of detection (LOD) to 30 a medianof 4) and Znrecalc/Alrecalc ratios (<LOD to 626 a median of 21) ofthese samples are elevated, which is also indicative of the presenceof anthropogenic metal sources (database Van der Veer, 2006). TheCurecalc/Alrecalc and Znrecalc/Alrecalc ratios in the other subsoils(n = 303) vary from <LOD to 9 with a median of 1 and <LOD to52 with a median of 5 respectively. The subsoil samples with devi-ating Pb isotope ratios have similar Pbrecalc/Alrecalc ratios (n = 6) butshow no evidence of added Pb. Most likely the relatively high206Pb/207Pb and 206Pb/208Pb ratios, and relatively low 208Pb/207Pbratios in these samples are caused by analytical errors or otheranomalies.

For the subsoil data that passed the two criteria, the regressionbetween Alrecalc and Pbrecalc in the studied subsoils (see Fig. 2) is:

Pbrecalc ¼ 3:61� Alrecalc þ 2:05 ð5Þ

with Pbrecalc the measured Pb content on organic matter-free basisin mg/kg, and Alrecalc the measured Al content on organic

. Right: Pbrecalc (mg/kg) versus Alrecalc (wt%) content for all subsoil samples (defaultratios and Pb isotope ratios characteristic for the lithologically inherited variation into criterion 1 (most likely non lithologically inherited Pb). � Subsoils samples that

ubsoil samples that have Pbrecalc/Alrecalc ratios and Pb isotope ratios characteristic fore deviating Pbrecalc/Alrecalc ratios according to criterion 1 (most likely containingo criterion 2.

234 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

matter-free basis in wt% (n = 303; R2 = 0.88; Standard error of esti-mate is 3.5 mg/kg).

The relationship between Al and Pb in soils and sediments hasalso been examined by others (e.g. Huisman; 1998; Spijker,2005; Van der Veer, 2006). In these studies, the Al and Pb contentwere not calculated on an organic matter-free basis. Nevertheless,

Table 2Summary of the analytical results of the subsoil samples. MIN = minimum; MED = median

n Statistic 206Pb/207Pb 208Pb/207Pb 206Pb/20

All 336 MIN#1 1.154 2.428 0.473303 MIN 1.175 2.441 0.478303 MED 1.197 2.468 0.485303 MAX 1.221 2.494 0.492336 MAX#1 1.230 2.494 0.495

Sand 202 MIN#1 1.168 2.428 0.477184 MIN 1.175 2.441 0.479184 MED 1.193 2.463 0.485184 MAX 1.217 2.493 0.492202 MAX#1 1.230 2.493 0.495

Fluviatile clay 24 MIN#1 1.188 2.462 0.48222 MIN 1.188 2.462 0.48222 MED 1.199 2.472 0.48522 MAX 1.209 2.480 0.48824 MAX#1 1.209 2.480 0.488

Marine clay 81 MIN#1 1.186 2.451 0.47881 MIN 1.186 2.451 0.47881 MED 1.202 2.475 0.48581 MAX 1.213 2.494 0.49081 MAX#1 1.213 2.494 0.490

Peat 25 MIN#1 1.154 2.438 0.47312 MIN 1.180 2.459 0.47912 MED 1.190 2.471 0.48212 MAX 1.204 2.477 0.48825 MAX#1 1.218 2.477 0.492

Loess 4 MIN#1 1.219 2.484 0.4904 MIN 1.219 2.484 0.4904 MED 1.220 2.488 0.4914 MAX 1.221 2.489 0.4914 MAX#1 1.221 2.489 0.491

#1 Includes all samples defined as outliers.

Fig. 4. Box-whisker plots of Pbrecalc content, 206Pb/207Pb, 208Pb/20

the inter-element relationships in these studies are comparablewith the results obtained in this study.

The variation in 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb,Pbrecalc, OM, Alrecalc and Zrrecalc content in the sedimentary subsoilsof The Netherlands is summarised in Table 2 and visualized in box-whisker plots in Fig. 4 for the five lithologies distinguished. The

; MAX = maximum; OM = organic matter.

8Pb Pbrecalc (mg/kg) OM (wt%) Alrecalc (wt%) Zrrecalc (mg/kg)

3 0.24 0.18 373 0.24 0.58 399 1.28 1.94 247

69 88.24 12.92 719288 95.27 12.92 762

3 0.24 0.18 373 0.24 0.58 397 0.69 1.35 241

22 18.82 4.05 71950 18.82 4.98 762

13 1.19 1.19 12913 1.19 3.21 12926 5.17 6.53 23269 23.63 11.56 35169 23.63 11.56 351

9 1.13 2.78 1629 1.13 2.78 162

20 3.92 5.11 28337 18.42 8.32 65337 18.42 8.32 653

6 24.20 0.88 8313 24.20 2.72 8330 63.81 8.17 13750 88.24 12.92 234

288 95.27 12.92 460

19 2.01 5.02 52519 2.01 5.02 52519 2.23 5.34 56020 2.32 5.63 60520 2.32 5.63 605

7Pb and 206Pb/208Pb in Dutch subsoils classified by lithology.

Fig. 5. Spatial distribution of the Zrrecalc, Alrecalc and Pbrecalc content, and the Pb isotope ratios of subsoils in The Netherlands.

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 235

Fig. 6. Variance explained in the ratio data subsets as a function of the size of thecore data set (from 100% to 50%). The difference between the total explainedvariance and the sum of variance explained (log–log basis) by Zr and Al separatelyfor the same dataset is shown as well.

236 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

spatial distribution of the Zrrecalc, Alrecalc and Pbrecalc content andthe Pb isotope composition in the subsoils in The Netherlands isvisualised in bubble maps in Fig. 5. Sand samples (n = 184) arecharacterised by a low Pbrecalc content and relatively low206Pb/207Pb, 208P/207Pb and 207Pb/208Pb ratios (Fig. 4, Table 2). Thefluviatile (n = 22) and marine (n = 81) clays have a relatively highPbrecalc content and relatively high 206Pb/207Pb, 208P/207Pb and207Pb/208Pb ratios (Fig. 4, Table 2). No difference in isotope signa-ture was found between the marine and fluviatile clays. They are,therefore, combined as one group, called clays. The Pbrecalc contentand the Pb isotope composition of peat (n = 12) overlaps with thoseof sand and clay (Fig. 4, Table 2). The peat samples more often con-tain an enhanced Pbrecalc content than the other samples (Table 2).Only 4 loess samples were analysed. These samples are character-ised by an intermediate Pbrecalc content, in between sand and clay,and the highest 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios ofall samples (Fig. 4, Table 2).

4.2. Pb isotope regression results

The geochemical proxies or predictors that could be used in aregression model for Pb isotope ratios were first identified. Preli-minary investigation showed that for the parent material dataset, the Pb isotope ratios did show a significant dependence onAlrecalc content (a proxy for clay content), as well as on the Zrrecalc

content (which could be a proxy for U/Th bearing minerals like zir-con). Neither Alrecalc nor Zrrecalc is subject to major anthropogenicinfluence, which makes them suitable as predictors for the litho-logically inherited Pb isotope composition, including anthropogen-ically influenced situations. The probability of other potentialmineralogical relationships (with e.g., apatite, feldspar or pyrite)could not be established, as most inorganic constituents, includingphosphate (all recalculated on an organic free basis), show a veryhigh correlation with Alrecalc, and thus have no added value in aregression model.

The regression results yield the following relationships for theisotope ratios with Zrrecalc and Alrecalc as independent variables:

logð206Pb=207PbÞ ¼ 0:0590þ 0:00738 logðZrrecalc=mg kg�1Þþ 0:003325 logðAlrecalc=%Þ ð6:1Þ

logð208Pb=207PbÞ ¼ 0:3876þ 0:00146 logðZrrecalc=mg kg�1Þþ 0:003325 logðAlrecalc=%Þ ð6:2Þ

logð206Pb=208PbÞ ¼ �0:3286þ 0:00591

� logðZrrecalc=mg kg�1Þ ð6:3Þ

The various ratio data are not assigned proportionally to thecore dataset in the combined robust regression model (ntotal = 720;206Pb/207Pb, n = 201; 208Pb/207Pb, n = 289; 206Pb/208Pb, n = 230).This makes it less straightforward to quantify the variance ex-plained by the set of regression equations, as R2 obviously varieswith the chosen size of the core dataset. Fig. 6 presents the vari-ance as a function of the size of the core dataset for the subsetsof ratio data. Explained variance for the overall dataset (n = 336)is around 30–40%; for core datasets of n = 240 for the individual ra-tios the explained variance is 71.3% (206Pb/207Pb), 70.5%(208Pb/207Pb), and 63.7% (206Pb/208Pb). The total explained variancefor 206Pb/207Pb and 208Pb/207Pb is only about 10% less than the sumof the explained variances by Zr and Al individually, showing that,as expected, these two proxies are largely independent predictors.

Plots of predicted versus measured isotope ratios are shown inFig. 7. To clarify the relationship of the Pb isotope ratios withAlrecalc and Zrrecalc, bivariate plots versus these predictors are alsogiven. In general, Zrrecalc is the better predictor, although less sofor 208Pb/207Pb. The prediction power of Alrecalc is highest for208Pb/207Pb and absent for 206Pb/208Pb. The latter was identifiedbefore setting up the final regression model, which, therefore,included only one regression coefficient for Alrecalc (c1 = c2 in Eq.(4)). within this dataset. However, as a group they clearly havehigher 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios thanpredicted. A substantial umber of the peat samples (�50%) appearto contain significant non-lithologically inherited Pb. The deviatingpredicted Pb isotope ratios of these samples can be related to thepresence of this additional Pb.

The size of the core dataset in the robust regression procedurewas chosen more or less arbitrarily at around 70% of the wholedataset. This ensures a robust regression equation, but does notimply that the 30% non-core data are all outliers. Equations similarto Eqs. (2) and (3) were used to determine what samples would beconsidered outliers based on the regression results. It emerges that8 samples, all sands, that are not identified as outliers by the twocriteria defined in Section 3.5.2, would be identified as outliersbased on the regression results. These samples have unexpectedPb isotope ratios in view of their composition in terms of Al andZr. Put otherwise, for these samples the regression model failsand one or two of the isotope ratios are not predicted well. How-ever, for 13 samples identified as outliers based on criterion 1the isotope ratios are still reasonably well predicted. These sam-ples have relatively high Pb contents without this having a notableeffect on the Pb isotope ratios. All samples identified as outliersbased on criterion 2 (extreme isotope ratio) are also consideredanomalous based on the regression results. Although the loesssamples individually do not appear as statistical outliers, their Pbisotope ratios differ from the other lithologies (Fig. 7). They clearlyform a distinct group, with higher measured 206Pb/207Pb,208Pb/207Pb and 207Pb/208Pb ratios than would be predicted basedon their Alrecalc and Zrrecalc contents.

Fig. 7. Measured 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios in the subsoil versus (a) Alrecalc, (b) Zrrecalc, and (c) predicted 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios.Closed symbols represent the data of the combined 70% core dataset, open symbols the 30% non-core data. Lines represent the log–log regression line of the core data.

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 237

5. Discussion

5.1. Lithologically inherited Pb isotope signatures

Dutch sediments mainly originate from geological units in cen-tral Western Europe. The oldest geological unit in the Dutch hinter-land is Variscan granite with 206Pb/207Pb ratios varying between1.18 and 1.20 (Michard-Vitrac et al., 1981; Steinmann and Stille,1997). Most post-Variscan sediments in central Western Europehave 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios ranging from1.18–1.22, 2.46–2.48 and 0.481–0.487, respectively (Elbaz-Pouli-

chet et al., 1986; Monna et al., 1995; Steinmann and Stille, 1997).The range of Pb isotope ratios of the subsoils in The Netherlandsencompasses this range and is established at 1.175–1.221, 2.441–2.494 and 0.478–0.492 for 206Pb/207Pb, 207Pb/208Pb and 206Pb/208Pb,respectively (Table 2). The median Pb isotope composition of Dutchsubsoils is 1.197, 2.468 and 0.485 for 206Pb/207Pb, 208Pb/207Pb and206Pb/208Pb, respectively (Table 2). These values almost exactlymatch the Pb isotope composition of average common Pb (AC) asgiven by Stacey and Kramers (1975) and these values are still com-monly used in geochemical models, e.g., Teixeira et al. (2011):(206Pb/207Pb)AC = 1.197; (208Pb/207Pb)AC = 2.472 and (206Pb/208Pb)

1.26

1.24

1.22

1.20

1.18

1.16

1.14

1.12

1.10

1.08

2.572.542.512.482.452.422.392.362.332.30

0.455

0.460

0.465

0.470

0.475

0.480

0.485

0.490

0.495

0.500sandclaypeatloessAC Pb

208Pb =54.5%207Pb =23%

1%

206Pb/207Pb

208Pb/207Pb

206

208Pb denom.

206Pb denom.

207Pb denom.

208

206

207

206Pb =26.5%

Pb/208Pb

Fig. 8. Triangular plot of the measured 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratios in subsoils in The Netherlands. The small inset top left shows the full triangle withsingle-isotope compositions at the apexes. Solid grid lines are lines of equal percentage (apex compositions indicated). Dashed gridlines are lines of equal isotope ratio asindicated along the sides. The grey shaded rhomboid areas show that, as 208Pb dominates here over both 206Pb and 207Pb, minimal distortion in data presentation will beobtained with a bivariate ratio-plot with 208Pb as the common denominator (near-square rhomboid), whereas in 206Pb and 207Pb common denominator plots the data cloudwill be stretched in the direction of the smallest diagonal of the respective rhomboids, which may lead to spurious correlations. AC–Pb = Pb isotope composition of averagecommon Pb (Stacey and Kramers, 1975).

238 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

AC = 0.484. This implies that the Dutch sediments are derived froma mixture of rocks/minerals from the Dutch hinterland that to-gether represent average common Pb. The process of rock weath-ering, transport, size fractionation, deposition and soil formationapparently did not result in a deviation of the Pb isotope composi-tion from average common Pb. Van der Veer (2006) compared themajor and trace element composition of Dutch soils with the aver-age upper continental crust (UCC). According to this study, Dutchsoils are on average more sandy and enriched in As and Se, buton the whole are comparable with UCC, especially the clayey soils.This study shows that an average UCC origin also accounts for themedian Pb isotope composition of Dutch soils.

The 4 distinguished lithologies of the Dutch subsoils, sand, clay,peat and loess, have distinct chemical compositions (Table 2). Toillustrate the lithologically inherited variation in Pb isotope com-position of the 4 lithologies, the data are shown in a triangular plot(Fig. 8). Additionally 208Pb/207Pb is plotted versus 206Pb/207Pb and206Pb/208Pb versus 207Pb/208Pb in 3-isotope plots (Fig. 9). Fig. 8and 9 clearly show higher average 206Pb/207Pb and 208Pb/207Pb ra-tios for the clays than for the sands. Peat samples have Pb isotoperatios comparable to both sands and clays (Fig. 8 and 9). Lead inpeat can be present both in minerals in the peat (e.g., clay minerals,quartz, pyrite, etc.) or bound to organic matter. Loess shows themost distinct Pb isotope ratios (Fig. 8 and 9). It should be noted,however, that the loess group only comprises 4 samples.

5.2. Relationship of Pb isotope signature to mineralogy

Sedimentary soils contain a variety of minerals and to fullyunderstand the controls of the observed lithologically inheritedvariation on the Pb isotope composition of sedimentary soils inThe Netherlands, the mineralogy and isotope composition of theindividual minerals should be known. In the absence of such de-tailed information, however, some general observations can bemade.

First, it is necessary to distinguish between minerals with andwithout U and Th as major constituents in the mineral lattice.The Pb isotope composition of minerals that contain U and Th,for example zircons, continuously changes in time due to the buildup of radiogenic Pb through the decay of 235U, 238U and 232Th. Theirpresent day Pb isotope composition is the result of this cumulativebuild up since the time they were formed, and further reflects theirinitial U/Th ratio. The Pb isotope compositions of minerals withessentially no U and Th, for example galena and K-feldspar, haveisotopic compositions that do not change with time. Their Pb iso-tope composition reflects the initial (magmatic) whole rock Pb iso-tope ratios, which reflect the time integrated build up of Pb untilthe they were formed, which in turn depends on the U/Pb andTh/Pb ratios in the source rocks. These minerals are thus compar-atively less radiogenic (containing ‘‘old’’ Pb), while the U and Thcontaining minerals are on average more radiogenic (containing‘‘young’’ Pb).

Second, secondary minerals such as clay minerals that areformed during weathering should be distinguished from primaryminerals. Secondary minerals weather from primary Pb containingminerals and thus will have a similar Pb isotope signature. How-ever, they may additionally include radiogenic Pb that is contem-poraneously released, because of lattice disruption, from U andTh containing minerals (e.g., zircons). Secondary minerals, there-fore, are expected to be enriched in Pb specifically low in 207Pb, be-cause of the notably shorter half life of 235U.

These considerations explain the observed and modelled Pb iso-tope signatures (Eqs. (6.1)-(6.3)). The regression derived depen-dency of the Pb isotope ratios on Al and Zr is visualised inFig. 10. Arrows indicate the relative increase in the two plotted ra-tios, for independent increments in either Al or Zr. Higher Zrcontents in soils indicate a higher proportion of U and Th contain-ing minerals and hence higher (more radiogenic) 206Pb/207Pb,207Pb/208Pb and 206Pb/208Pb ratios (Fig. 10). Furthermore, theU/Th ratio of Zr containing minerals is generally higher than that

Fig. 9. Measured 208Pb/207Pb versus 206Pb/207Pb ratios and 206Pb/208Pb ratios versus 207Pb/208Pb ratios in subsoils in The Netherlands. Ellipses represent the lithologicallyinherited Pb isotope composition of the 4 distinguished lithologies (NL sand = Dutch sands; NL clay = Dutch clays; NL peat = Dutch peat; NL loess = Dutch loess). Outliersmarked with the symbol ‘l’ or outliers according to criterion 2.

Fig. 10. Pb isotope ratio plots for the subsoils, as modelled from the measured Alrecalc and Zrrecalc content (70% core data). The arrows indicate the direction and magnitude ofchange with increase of either Zrrecalc or Alrecalc over the order of magnitude observed in the dataset.

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 239

in bulk rock (Hoskin and Schaltegger, 2003; Fig. 10). Aluminiumcontent is known to be indicative of the amount of clay mineralsin soils. Higher Al contents in soils represent a higher proportion

of secondary minerals, and hence a shift towards lower 207Pb(Fig. 10). The lower 206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb ratiosof the sands compared to clays can be partly attributed to the

240 N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241

lower proportion of radiogenic Pb from secondary minerals (clays)as well as to a broader range towards lower Zr contents (Table 1,Fig. 5).

The regression model for lithology dependent variation of Pbisotope ratios works well for the sands and clays, and for the peatswithout additional non-background Pb. It is noted, however, thatthe loess samples have a more radiogenic Pb isotope signature thanpredicted from their Alrecalc and Zrrecalc content. The regressionmodel could be improved if the mineralogy and isotope composi-tion of the individual minerals in the subsoil are known.

5.3. Outliers

Based on the Pbrecalc/Alrecalc ratios (criterion 1), approximately8% of the subsoils are flagged as being outliers (Fig. 2). The Pb iso-tope composition of these samples is visualised in Pb-isotope plots(Fig. 9). Fig. 9 demonstrates that these outliers are predominantlypeat and sand (and not clay and loess) samples. The majority ofthese subsoils only have a slightly enhanced Pb content (<10 mg/kg additional Pb) that does not result in a significant shift in Pb iso-tope ratios (Fig. 9). Most peat samples and three sand samples,however, do show a significant shift in Pb isotope ratios (Fig. 9).The 206Pb/207Pb and 206Pb/208Pb ratios of these samples are lowerthat those of the majority of the subsoils, which indicates ananthropogenic or otherwise less radiogenic Pb source (Fig. 9). Thedeviating Pb isotope signatures could for example be caused byman-caused soil disturbance, physical migration or leaching ofanthropogenic Pb from the topsoil to the subsoil and subsequentimmobilization, or by scavenging of less radiogenic Pb from up-ward seeping groundwater. The origin of these outliers will be fur-ther discussed in the companion article on anthropogenic Pb(Walraven et al., 2013).

6. Conclusions

The lithologically inherited variation in Pb isotope ratios in soilsin The Netherlands was determined by means of Pb isotope analy-sis of 342 subsoil samples (default sample depth 100–120 cm). Theestablished Pb isotope signatures of subsoils in The Netherlandsare 1.175–1.221, 2.441–2.494 and 0.478–0.492 for 206Pb/207Pb,207Pb/208Pb and 206Pb/208Pb, respectively.

The overall average Pb isotope composition of the Dutch sub-soils equals present day common Pb, but the four main lithologiesof the Dutch subsoils, sand, clay, peat and loess, have distinctchemical and isotopic compositions. Clays have higher average206Pb/207Pb and 208Pb/207Pb ratios than sands. Some of the peatsamples show Pb isotope ratios that match both sands and claysdue to the presence of Pb containing sand and clay minerals inpeat. About half of the peat samples, however, show less radiogenicPb isotope signatures, that indicate the presence of additional, pos-sibly anthropogenic Pb sources. Loess shows the highest average206Pb/207Pb, 208Pb/207Pb and 206Pb/208Pb values of all samples andhas very distinct Pb isotope ratios. It should be noted, however,that the loess group only comprises a limited number of samples.

Multiple regression analysis demonstrates that the lithologi-cally inherited variation in the Pb isotope composition of Dutchsubsoils can be partly explained by the Al and Zr content of thesubsoils. Al and Zr are mutually independent and not expected tobe subjected to anthropogenic influence. The Zr content representsthe proportion of U and/or Th containing primary minerals that onaverage have more radiogenic Pb isotope compositions. The U/Thratio of zircons is generally much higher than that in the bulkrocks, which explains the trend in Dutch subsoils of lower contri-butions of 208Pb with higher Zr content. The Al content representsthe proportion of secondary clay minerals that incorporate the

more recently formed radiogenic Pb that is relatively low in207Pb compared to 206Pb and 208Pb. The different, apparently older,isotopic signature of the sands compared to the clays, can thus beattributed to the lower proportion of radiogenic Pb from secondaryminerals (clays) as well as a broader range towards lower Zr con-tents. The more radiogenic signature of the loess samples, com-pared to the sands and clays, is only partly explained by theirrelatively high Al and Zr content.

Approximately 8% of the subsoils, mainly peats, contain addi-tional Pb that is assumed to be non-lithologically inherited, possi-bly anthropogenic. Pollution of these subsoils might be explainedby human-derived soil disturbance, by physical migration or leach-ing of anthropogenic Pb from the topsoil to the subsoil. Alterna-tively, it could be the result of scavenging of less radiogenic Pbfrom upward seeping groundwater.

Acknowledgments

This research was sponsored by TNO and Deltares. We thankErik van Vilsteren of Deltares for determining the Pb isotope ratiosof the samples. Rob van Galen of Deltares is acknowledged for per-forming the major part of the XRF analysis. We thank the threeanonymous reviewers for their thorough and constructive reviewsof the manuscript.

References

Anderson, M.K., Rauland-Rasmussen, K., Hansen, H.C.B., Strobel, B.W., 2002.Distribution and fractionation of heavy metals in pairs of arable andafforested soils in Denmark. Eur. J. Soil Sci. 53, 491–502.

Baize, D., Sterckeman, T., 2001. Of the necessity of knowledge of the natural pedo-geochemical background content in the evaluation of the contamination of soilsby trace elements. Sci. Total Environ. 264, 127–139.

Bazelmans, J., Van der Meulen, M., Vos, P.C., Weerts, H.J.T., 2011. Atlas vanNederland in het Holoceen: landschap en bewoning vanaf de laatste ijstijd totnu. Bakker, Amsterdam, The Netherlands.

Belousova, E.A., Griffin, W.L., O’Reilly, S.Y., Fisher, N.I., 2002. Igneous zircon: traceelement composition as an indicator of source rock type. Contrib. Mineral.Petrol. 141, 602–622.

Blaser, P., Zimmerman, S., Luster, J., Shotyk, W., 2000. Critical examinations of traceelement enrichments and depletions in soils: As, Cr, Cu, Ni, Pb, and Zn in Swissforest soils. Sci. Total Environ. 249, 257–280.

Chow, T.J., 1970. Pb accumulation in roadside soils and grass. Nature 225, 295–296.Chow, T.J., Johnston, M.S., 1965. Lead isotopes in gasoline and aerosols of Los

Angeles basin, California. Science 147, 502–503.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.

Compendium, 2011. <http://www.compendiumvoordeleefomgeving.nl/indicatoren/nl0258-Inventarisatie-van-aantal-locaties-met-bodemverontreiniging.html?i=3-13>(in Dutch) (accessed November 2011).

Darnley, A.G., Borjklund, A., Bolviken, B., Gustavsson, N., Koval, P.V., Plant, J.A.,Steenfelt, A., Tauchid, M., Xuejing, X., Garret, R.G., Hall, G.E.M., 1995. A GlobalGeochemical Database for Environmental and Resource Management:Recommendations for International Geochemical Mapping, second ed. FinalReport of IGCP Project 259, Paris, UNESCO.

de Treville, R.T.P., 1964. Natural occurrence of lead. Arch. Environ. Health 8, 212–221.

Elbaz-Poulichet, F., Holliger, P., Martin, J.M., Petit, D., 1986. Stable lead isotopesratios in major French rivers and estuaries. Sci. Total Environ. 54, 61–76.

Gulson, B.L., Tiller, K.G., Mizon, K.J., Merry, R.H., 1981. Use of Pb isotope ratios insoils to identify the source of Pb contamination near Adelaide, South Australia.Environ. Sci. Technol. 15, 691–696.

Hagens, W.I., Walraven, N., Minekus, M., Havenaar, R., Lijzen, J.P.A., Oomen, A.G.,2009. Relative Oral Bioavailability of Lead from Dutch Made Grounds. RIVMReport 711701086/2009, Bilthoven, The Netherlands.

Hansmann, W., Köppel, V., 2000. Lead-isotopes as tracers of pollutants in soils.Chem. Geol. 171, 123–144.

Hernandez, L., Probst, A., Probst, J.-L., Ulrich, E., 2003. Heavy metal distribution insome French forest soils: evidence for atmospheric contamination. Sci. TotalEnviron. 312, 195–219.

Heskel, D.L., 1983. A model for the adoption of metallurgy in the ancient MiddleEast. Curr. Anthropol. 24, 362–366.

Hoskin, P.W.O., Schaltegger, U., 2003. The composition of zircon and igneous andmetamorphic petrogenesis. In: Hanchar, J.M., Hoskins, P.W.O. (Eds.), Zircon:Reviews in Mineralogy and Geochemistry 53, 27–62.

N. Walraven et al. / Applied Geochemistry 37 (2013) 228–241 241

Huisman, D.J., 1998. Geochemical Characterization of Subsurface Sediments in TheNetherlands. Netherlands Institute for Applied Geosciences, Utrecht, TheNetherlands.

Komárek, M., Ettler, V., Chrastny, V., Mihaljevic, M., 2008. Lead isotopes inenvironmental sciences: a review. Environ. Int. 34, 562–577.

Krachler, M., Le Roux, G., Kober, B., Shotyk, W., 2004. Optimising accuracy andprecision of lead isotope measurements (206Pb, 207Pb, 208Pb) in acid digests ofpeat with ICP-SMS using individual mass discrimination correction. J. Anal.Atom. Spectrom. 19, 354–361.

Lamé, F.P.J., Brus, D.J., Nieuwenhuis, R.H., 2004. Achtergrondwaarden 2000,hoofdrapport AW2000 fase 1. NITG 04-242-A, Netherlands Institute ofApplied Geosciences (NITG-TNO), Utrecht, The Netherlands.

Locher, W.P., de Bakker, H., 1987. Bodemkunde van Nederland, deel 1: algemenebodemkunde. Stichting voor Bodemkartering (Stiboka), Malmberg, Den Bosch,The Netherlands.

Michard-Vitrac, A., Albarède, F., Allègre, C.J., 1981. Isotopic composition ofHercynian granitic K-feldspars constrains continental genesis. Nature 291,460–464.

Mol, G., 2002. Soil Acidification Monitoring in The Netherlands. PhD-Thesis UtrechtUniv., Geologica Ultraiectina 216, Utrecht, The Netherlands.

Monna, F., Ben Othman, D., Luck, J.M., 1995. Pb isotopes and Pb, Zn and Cdconcentrations in the rivers feeding a coastal pond (Thau, southern France):constraints on the origin(s) and flux(es) of metals. Sci. Total Environ. 166, 19–34.

Montgomery, D.C., Peck, E.A., 1982. Introduction to Linear Regression Analysis. JohnWiley & Sons, New York.

Ódor, L., Horváth, I., Fügedi, U., 1997. Low-density geochemical mapping inHungary. J. Geochem. Explor. 60, 55–66.

Patterson, C.C., Chow, T.J., Murozumi, M., 1970. The possibility of measuringvariations in the intensity of world-wide lead smelting during medieval andancient times using lead aerosol deposits in polar snow strata. In: Berger, R.(Ed.), Scientific Methods in Medieval Archeology. Univiserity of California Press,Berkely, pp. 339–350.

Plant, J.A., Klaver, G., Locutura, J., Salminen, R., Vrana, K., Fordyce, F.M., 1997. Theforum of European geological surveys. Geochemistry task group inventory1994–1996. J. Geochem. Explor. 59, 123–146.

Reimann, C., Siewers, U., Tarvinen, T., Bityokova, L., Eriksson, J., Gilucis, A.,Gregorauskiene, V., Lukashev, V., Matinian, N.N., Pasieczna, A., 2000. Balticsoil survey: total concentrations of major selected trace elements in arable soilsfrom 10 countries around the Baltic Sea. Sci. Total Environ. 257, 155–170.

Reimann, C., Smith, D.B., Woodruff, L.G., Flem, B., 2011. Pb-concentrations and Pb-isotope ratios in soils collected along an east-west transect across the UnitedStates. Appl. Geochem. 26, 1623–1631.

Reimann, C., Flem, B., Fabian, K., Birke, M., Ladenberger, A., Négrel, P., Demetriades,A., Hoogewerff, J.The GEMAS Project Team, 2012. Lead and lead isotopes inagricultural soils of Europe – The continental perspective. Appl. Geochem. 27,532–542.

Rieuwerts, J.S., Farago, M.E., Cikrt, M., Bencko, V., 2000. Differences in leadbioavailability between a smelting and a mining area. Water Air Soil Pollut.122, 203–229.

Rousseeuw, P.J., Van Driessen, K., 2002. Computing LTS regression for large datasets. Estadistica 54, 163–190.

Ruby, M.V., Schoof, R., Brattin, W., Goldade, M., Post, G., Harnois, M., Mosby, D.E.,Casteel, S.W., Berti, W., Carpenter, M., Edwards, D., Cragin, D., Chappell, W.,1999. Advances in evaluating the oral bioavailability of inorganics in soil for usein human health risk assessment. Environ. Sci. Technol. 33, 3697–3705.

Salminen, R., Tarvainen, T., 1995. Geochemical mapping and databases at GeologicalSurvey of Finland. J. Geochem. Explor. 55, 321.

Sangster, D.F., Outridge, P.M., Davis, W.J., 2000. Stable lead isotope characteristics oflead ore deposits of environmental significance. NCR-CNRC. Environ. Rev. 8,115–147.

Sonke, J.E., Hoogewerff, J.A., van der Laan, S.R., Vangronsveld, J., 2002. A chemicaland mineralogical reconstruction of Zn-smelter emissions in the Kempen region

(Belgium), based on organic pool sediment cores. Sci. Total Environ. 292, 101–119.

Spijker, J., 2005. Geochemical Patterns in the Soils of Zeeland. Natural VariabilityVersus Anthropogenic Impact. NGS 330, Nederlandse Geografische Studies,Utrecht, The Netherlands.

Stacey, J.S., Kramers, J.D., 1975. Approximation of terrestrial lead isotope evolutionby a two-stage model. Earth Planet. Sci. Lett. 26, 207–221.

Steele, M.J., Beck, B.D., Murphy, B.L., Strauss, H.S., 1990. Assessing the contributionfrom lead in mining wastes to blood lead. Regul. Toxicol. Pharmacol. 11, 158–190.

Steinmann, M., Stille, P., 1997. Rare earth element behavior and Pb, Sr, Nd isotopesystematics in a heavy metal contaminated soil. Appl. Geochem. 12, 607–623.

Steinnes, E., 2001. Metal contamination of the natural environment in Norway fromlong range atmospheric transport. Water Air Soil Pollut. Focus 1, 449–460.

Sterckeman, T., Douay, F., Baize, D., Fourrier, H., Proix, N., Schvartz, C., 2004. Factorsaffecting trace element concentrations in soils developed on recent marinedeposits from northern France. Appl. Geochem. 19, 89–103.

Sterckeman, T., Douay, F., Baize, D., Fourrier, H., Proix, N., Schvarz, C., Carignan, J.,2006a. Trace element distributions in soils developed in loess deposits fromnorthern France. Eur. J. Soil Sci. 57, 392–410.

Sterckeman, T., Douay, F., Baize, D., Fourrier, H., Proix, N., Schvartz, C., 2006b. Traceelements in soils developed in sedimentary materials from Northern France.Geoderma 136, 912–929.

Swaine, D.J., 1955. The Trace Element Content of Soils. Commonwealth Bur. Soil Sci.Tech. Comm. No. 48. Herald Printing, York, England.

Teixeira, R.J.S., Neiva, A.M.R., Silva, P.B., Gomes, M.E.P., Andersen, T., Ramos, J.M.F.,2011. Combined U–Pb geochronology and Lu–Hf isotope systematics by LAM–ICPMS of zircons from granites and metasedimentary rocks of Carrazeda deAnsiães and Sabugal areas, Portugal, to constrain granite sources. Lithos 125,321–334.

Tukey, J.W., 1977. Exploratory Data Analysis. Addison-Wesley, Philippines.USGS, 1976. Lead in the Environment. U.S.Geol. Surv. Prof. Paper 957.Van der Veer, G., 2006. Geochemical soil survey of the Netherlands. Atlas of major

and trace elements in topsoil and parent material; assessment of natural andanthropogenic enrichment factors. Netherl. Geograph. Stud. 347, 1–245.<http://igitur-archive.library.uu.nl/dissertations/2006-1011-200742/full.pdf>(accessed February 2013).

Van Gaans, P.F.M., Spijker, J., Vriend, S.P., de Jong, J.N., 2007. Patterns in soil quality:natural geochemical variability versus anthropogenic impact in soils of Zeeland,the Netherlands. Int. J. Geogr. Inf. Sci. 21, 569–687.

Walraven, N., van Os, B.J.H., Klaver, G.Th., Baker, J.H., Vriend, S.P., 1997. Traceelement concentrations and stable lead isotopes in soils as tracers of leadpollution in Graft-De Rijp, the Netherlands. J. Geochem. Explor. 59, 47–58.

Walraven, N., van Os, B.J.H., Vriend, S.P., 2001a. Herkomstbepaling van antropogeenlood in de Utrechtse bodem. TNO-rapport NITG 01-029-B, Utrecht, TheNetherlands.

Walraven, N., van Os, B.J.H., Vriend, S.P., 2001b. Herkomstbepaling van antropogeenlood in de bodem van Fijnaart. TNO-rapport NITG 01-117-A, Utrecht, TheNetherlands.

Walraven, N., van Gaans, P.F.M., van der Veer, G., van Os, B.J.H., Klaver, G.Th., Vriend,S.P., Middelburg, J.J., Davies, G.R., 2013. Tracing diffuse anthropogenic Pbsources in soils by means of Pb isotope analysis. Appl. Geochem. (in press).

Wedepohl, K.H., 1995. The composition of the continental crust. Geochim.Cosmochim. Acta 59, 1217–1232.

Wedepohl, K.H., Doe, B.R., 1974. In: Wedepohl, K.H. (Ed.), Data of Geochemistry.Springer-Verlag, New York, II-4 (Chapter 82).

Westerhoff, W.E., Wong, T.E., De Mulder, E.F.J., 2003. De opbouw van deondergrond. In: De Mulder, E.F.J., Geluk, M.C., Ritsema, I.I., Westerhoff, W.E.,Wong, T.E. (Eds.), De ondergrond van Nederland. Wolters-Noordhoff PublishingB.V, Groningen/Houten, pp. 246–352.

Zagwijn, W.H., Van Staalduinen, C.J., 1975. Toelichting bij de GeologischeOverzichtskaarten van Nederland. Geological Survey of The Netherlands,Haarlem.