lable at ScienceDirect

Quaternary International 240 (2011) 12e21

Contents lists avai

Quaternary International

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

An evaluation of geochemical weathering indices in loessepaleosol studies

Björn Buggle a,*, Bruno Glaser a, Ulrich Hambach b, Natalia Gerasimenko c, Slobodan Markovi�c d

a Soil Physics Department, University of Bayreuth, D-95440 Bayreuth, GermanybChair of Geomorphology, University of Bayreuth, D-95440 Bayreuth, Germanyc Earth Science and Geomorphology Department, Tarasa Shevchenko National University of Kyiv, UkrainedChair of Physical Geography, Faculty of Sciences, University of Novi Sad, Trg Dositeja Obradovi�ca 3, 21000 Novi Sad, Serbia

a r t i c l e i n f o

Article history:Available online 21 July 2010

* Corresponding author. Tel.: þ49 0 921 552174; faE-mail address: Bjoern.Buggle@uni-bayreuth.de (B

1040-6182/$ e see front matter � 2010 Elsevier Ltd adoi:10.1016/j.quaint.2010.07.019

a b s t r a c t

Applying geochemical proxies as measure for the weathering intensity of paleosols and sediments suchas loess, the Quaternary scientist is confronted with various element ratios that have been proposed inliterature. This paper gives an overview on the principle of geochemical weathering indices. Differenttypes of indices are evaluated with respect to the suitability for loessepaleosol sequences, regarding thespecial characteristics of this type of sediments and paleosols. Case examples in this study are keysections in Southeastern and Eastern Europe: the loessepaleosol sequences Batajnica/Stari Slankamen(Serbia), Mircea Voda (Romania) and Stary Kaydaky (Ukraine), which represent archives of the Late andMid-Pleistocene climate change of the region. Considering element behavior during weathering ordiagenesis, the Chemical Proxy of Alteration (CPA) e i.e. the molar ratio Al2O3/(Al2O3 þ Na2O) � 100 e isproposed as the most appropriate index for silicate weathering. The CPA was evaluated againstcommonly used weathering indices including the “Chemical Index of Alteration” (CIA), the “ChemicalIndex of Weathering“ (CIW), the “Plagioclase Index of Alteration“ (PIA), the Index B of Kronberg andNesbitt, and the Ba/Sr and Rb/Sr ratio. Depth profiles of “Sr-type indices” (e.g. Ba/Sr, Rb/Sr) are likely to beinfluenced by the dynamics of secondary carbonate. On the other hand, common “Na-type indices” (e.g.CIA, PIA, CIW) may suffer from uncertainties in separating carbonateeCa from silicateeCa or from biasesdue to K-fixation (illitization). The CPA is insensitive against such effects. Additionally, using the CPA (aswith other Na-type indices) provides the possibility to evaluate the homogeneity of the parent materialregarding the relevant host minerals via the AeCNeK diagram.

� 2010 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

“Loess is a terrestrial clastic sediment, composed predominantlyof silt-sized particles, which is formed essentially by the accumu-lation of wind-blown dust” (Pye, 1995). In Northern- and Mid-latitudes, this dust originated mainly from sparsely vegetatedforeland areas of the ice sheets and the alluvial plains of large riversduring the Pleistocene cold periods (Smalley and Leach, 1978;Buggle et al., 2008; Újvári et al., 2008; Smalley et al., 2009). Also,desert regions may represent important dust source areas (e.g.Smalley and Krinsley, 1978). It was only during interstadial andinterglacial warm periods and more humid periods, respectively,when dust deposition decreased or even ceased, so that environ-mental conditions allowed extensive mineral weathering and thussoil formation. Therefore, sequences of relatively unaltered loess

x: þ49 0 921 552246.. Buggle).

nd INQUA. All rights reserved.

and more or less well-developed paleosols e so calledloessepaleosol sequences (LPSS) e reflect land surface stability andPleistocene climate development.

Under weathering conditions, the element composition ofa given parent material changes. Soluble and mobile elements aredepleted and less soluble and immobile elements are enriched.However, pedogenesis does not only mean weathering of mineralsand loss of elements, but also mineral transformation and forma-tion of new (secondary) minerals such as, for example, clayminerals or iron oxides. Amount and composition of iron oxides(for example) can be reflected in mineral magnetic properties aswell as the color of a soil sample (Schwertmann, 1993; Evans andHeller, 2003). As the type and intensity of such pedogenicfeatures essentially depend on (soil-) environmental conditions,they can be valuable indicators of the (past) climatic characteristics.There are a number of parameters and various proposals by pale-opedologists for proxies enabling a quantification of paleoclimaticmeaningful pedogenic processes (e.g. Derbyshire et al., 1997).Maher et al. (1994) for example introduced a quantitative

B. Buggle et al. / Quaternary International 240 (2011) 12e21 13

relationship between enhancement of the magnetic susceptibilityas proxy for the pedogenic formation of ferromagnetics and meanannual precipitation. However, in several situations this approachhas been shown to be not straightforward (Evans and Heller, 2003;Buggle et al., 2009). Other widely used measures of pedogenesisintensity are geochemically based weathering indices. Althoughrelying on similar concepts, a variety of such indices have beenpublished (e.g. Ding et al., 2001; Smykatz-Kloss, 2003;Schellenberger and Veit, 2006). Especially, non-geochemicallyspecialized Quaternary scientists are often confronted with thequestion: “what is the most appropriate index for a certain LPSSand, respectively, LPSS in general?”. From this evolves the motiva-tion to give an overview on the principle of geochemical weath-ering indices aiming towards an answer for this question. The caseexamples for this purpose use geochemical data of theloessepaleosol sequences Batajnica/Stari Slankamen (Serbia), Mir-cea Voda (Romania) and Stary Kaydaky (Ukraine).

2. Material and methods

The composite loessepaleosol sequence Batajnica and StariSlankamen (44�5502900N, 20�1901100E and 45�705800N, 20�1804400E) issituated in the Vojvodina, the Serbian part of the Pannonian (Car-pathian) Basin. The two individual profiles are about 40 km distantto each other and are located along the banks of the Danube River.Each section is about 40 m thick and comprises at least six majorinterglacial pedocomplexes (Buggle et al., 2009; Markovi�c et al.,2009). Due to a major hiatus in the younger part of the Stari Slan-kamen site and ground water influence at the older part of theBatajnica site, Buggle et al. (2008, 2009) built a composite rockmagnetic and geochemical record from these two sites using theyounger (<MIS 9) part from Batajnica and the older (>MIS 9) fromStari Slankamen. The timespan representedby the composite recordincludes the last 17 Marine Isotope Stages (Buggle et al., 2009).

The Mircea Voda loessepaleosol sequence (44�1901500N,28�1102100E) is located in the Dobrudja region (Romania) betweenthe Danube and the Black Sea coast. This site is formed by a similarsuccession of loess and paleosol units as the Serbian one,comprising also the last 17 MIS.

Both the Serbian and Romanian sections represent LPSS onloesseplateaus, with thick loess layers separating the major inter-glacial pedocomplexes. The main climatic difference between bothsites is the more pronounced dryness at Mircea Voda(w400e450 mmmean annual precipitation) than in the Vojvodinaregion (w600e680 mm mean annual precipitation).

The outcrops of the Stary Kaydaky site (48�2204200N, 35�0703000E)are situated in a system of gullies near to the Dnieper River(Ukraine). The profile comprises the last 15 MIS. However, mostloess layers separating the interglacial pedocomplexes are erodedand thin. While the Serbian and Romanian sites contain carbonatethroughout the profile, parts of the Stary Kaydaky LPSS arecompletely free of carbonates. Therefore, older loess layers andpaleosols are partly influenced by pedogenesis and weatheringduring subsequent interglacials.

For detailed information on the regional setting, description ofthe loessepaleosol successions, the chronostratigraphy, samplingstrategy and sample preparation for geochemical analyses pleaserefer to Buggle et al. (2008, 2009). Composition of major and traceelements was determined using a Philips 2404 X-Ray Fluorescence(XRF) Spectrophotometer. Determination of the carbonate contentfollowed the procedure of Hedges and Stern (1984) by calculatingthe difference in C content of the sample material with and withoutvapor fumigation. The measurement was carried out on a Vario ELelement analyzer (Elementar, Hanau, Germany) at the University ofBayreuth. Element ratios are calculated on molar base. The

nomenclature of the lithological units follows the Chinese “SeL”system (Buggle et al., 2009).

3. Chemical weathering indices

3.1. Choosing a chemical proxy of alteration for LPSS? e Principalconsiderations and hypotheses

The concept of geochemical proxies of mineral alteration (i.e.weathering indices) relies on the selective removal of soluble andmobile elements from a weathering profile compared to the rela-tive enrichment of rather immobile and non-soluble elements (e.g.Smykatz-Kloss, 2003; Yang et al., 2004). A number of elementindices, based on this principle, have been published as proxies ofmineral weathering for various kinds of sediments, including LPSS(e.g. Liu et al., 1993; Chen et al., 1999; Ding et al., 2001; Muhs et al.,2001; Yang et al., 2004; Schellenberger and Veit, 2006; Tan et al.,2006; Jeong et al., 2008; Bokhorst et al., 2009). Especially fornon-geochemists, it is often difficult to decide which index is mostappropriate for a given weathering profile. The following sectiongives some background information on element behavior underweathering conditions to deduce the answer of this question in thecase of LPSS.

More specifically, two questions must be considered. (i) What isthe most appropriate soluble and mobile element? (ii) What is themost appropriate immobile non-soluble element?

A first step in answering both questions is the classification ofthe elements according to their ionic potential (IP), i.e., the ratiobetween ionic charge and ionic radius, as shown in Fig. 1. Cationshaving an IP below 3 form only weak bondswith oxygen. Thus, theyare preferentially released from their host minerals duringweathering. In solution, they can be regarded as soluble cations,since they are fully hydrated. If the IP is higher (between 3 and 8),the high density of the positive charge enables strong bonds withoxygen. Thus, these elements form weathering resistant oxides.Furthermore, between an IP of 3 and 10 or 12 (depending on theliterature source), the density of the positive charge is in the properrange to cause a deprotonation of water molecules in the hydrationshell of the cation. As a result, insoluble hydroxides or oxyhydratesare formed to achieve charge neutrality. Elements of this category,when released during weathering, precipitate quickly as insolubleand immobile hydrolyzates. With a further increase of the IP, i.e.charge density of the cation, all protons of the water molecules arerepelled and water soluble anionic complexes are formed(Goldschmidt, 1937; Mason and Moore, 1985; Kutterolf, 2001;Railsback, 2003, 2005; Smykatz-Kloss, 2003). There are also otherfactors that can influence the solubility of ions, such as the activityof further components in solution, pH-value, redox conditions andtemperature. Under near-neutral and oxidizing conditions,however, the classification according to the IP successfully predictsthe behavior of the most common elements of interest for weath-ering indices (e.g. alkali-, earth-alkali elements, elements of the Al-and Ti-group), at least in a general way (Blumer, 1950; McBride,1994; Railsback, 2003; Smykatz-Kloss, 2003).

Concerning question one, all alkali- and earth alkali elements,except for Be, can be categorized as soluble cations according to theionic potential (Fig. 1; Wedepohl, 1978; Railsback, 2003; Reederet al., 2006). However, solubility does not always mean mobility.With increasing radius, the tendency of an ion for being adsorbedon clay minerals is higher and thus, its mobility is reduced (Nesbittet al., 1980; Smykatz-Kloss, 2003). Therefore, weathering indicesrelying on the mobility of the elements Cs, Rb, Ba, K such as Cs/K,Ba/K, Rb/K, K/Zr, K/Ti, K/Al (Harriss and Adams, 1966; Nesbitt et al.,1980; Liu et al., 1993; Muhs et al., 2001) can be expected to be lesssensitive for weak pedogenic alteration.

Fig. 1. Classification of the elements according to the ionic potential (IP). Values for the ionic potential and ionic radius are taken fromMason and Moore (1985). Cations having an IPbelow 3 are generally soluble in water, whereas cations with an IP between 3 and 10 (according to Mason and Moore, 1985) or 12, respectively (Goldschmidt, 1937; Kabata-Pendiasand Pendias, 2001), form insoluble, immobile hydrolyzates under near-neutral conditions. Elements having a higher IP tend to form soluble anionic complexes. The adsorption toclay minerals tends to increase with the radius of the cation (Nesbitt et al., 1980; Smykatz-Kloss, 2003).

B. Buggle et al. / Quaternary International 240 (2011) 12e2114

The earth alkali elements Ca, Mg and Sr, having a smaller ionicradius, are common in silicate minerals such as plagioclase,pyroxene, amphibole and biotite, which are susceptible to weath-ering (Nesbitt et al., 1980; Reeder et al., 2006). As these elementsare highly mobile in the weathering environment, they appear inseveral weathering indices like the Ba/Sr, Rb/Sr, Sr/K, Sr/Zr, Mg/K,Mg/Ti, Ca/K, Ca/Zr and Ca/Ti-ratio (e.g. Nesbitt et al., 1980; Liu et al.,1993; Chen et al., 1999; Muhs et al., 2001; Bokhorst et al., 2009).However, in a parent material containing carbonate e as in mostloess deposits e the mobility of Ca and Mg is predominantlycontrolled by the behavior of calcite and dolomite. This is also truefor Sr, which can substitute Ca in carbonates (Wedepohl, 1978;Reeder et al., 2006). Therefore, indices relying on Ca, Mg or Sr arenot expected to reflect the true weathering and leaching intensityof a paleosol, owing to postpedogenetic formation of secondarycarbonates. Thus, it is advisable to restrict the use of such indices tocarbonate-free parent material to avoid the problem that effects ofsilicate weathering are masked by the dynamics of carbonates(Smykatz-Kloss, 2003).

As stated above, the distribution of some elements might becontrolled by redox conditions, also after burial of the paleosol. Forthis reason, the use of the elements Fe and Mn in weatheringstudies is not recommended.

Having checked all thosementioned criteria, Liþ and Naþ appearto be the most suitable mobile cations for weathering indices inloessepaleosol sequences. As the former is seldomly analyzed, theuse of Naþ is proposed.

Regarding the choice of the immobile element, ions of interme-diate ionic potential, i.e. ions that tend to form insoluble hydroly-zates (Fig. 1) are generally employed. Rb, Ba and K, i.e. ions that canbe immobilized by adsorption on clay minerals due to their large

ionic radius, are often used as immobile references, for example inthe Rb/Sr, Ba/Sr or Na/K ratio (e.g. Liu et al., 1993; Gallet et al., 1996;Chen et al., 1999; Smykatz-Kloss, 2003; Tan et al., 2006.) However,under intense weathering conditions, significant losses of these“large ionic radius elements” can occur during the transformation ofmicas, feldspars and other host minerals into secondary clayminerals (e.g. Gallet et al.,1996;Muhset al., 2001). Gallet et al. (1996)for example, observed a significant loss of Rb in the strongly devel-oped Chinese paleosol S5. Thus, the focus should be on elements ofthe “insoluble hydrolyzate” category, especially on Al, Si, Ti and Zr,which aremost frequently used inweathering proxies. As a criterionfor a suitable reference element, themobile and immobile elementsform “a homogeneous mineralogical weathering system”. Thismeans that the immobile element also is hosted by the same,mobileelement bearingminerals and theirweathering residues. The reasonfor such a criterion is to minimize possible effects of down-profilevariations in the parent material composition. This is especiallyimportant in LPSS, where it is not always possible to determine thecomposition of an unaltered parent material for each pedomember.Thin stadial loess layers for example are often pedogeneticallyinfluenced by succeeding periods of soil formation. In that case, thecomposition of the parentmaterial has to be estimated from thickerunmodified loess layers, assuming a uniform composition of thedifferent loess units. According to the proposed criterion, Na and Alis a suitable pair of elements (in non-saline soils). The main hostmineral group of Na and Al in unweathered loess protoliths is thefeldspar group.Weathering products of this group are claymineralsand as end product under extensiveweathering conditions kaolinite(Al2Si2O5(OH4)) or gibbsite (Al(OH)3), both aluminous residues(Taylor et al., 1983; Taylor andMcLennan,1985; Reeder et al., 2006).In contrast, quartz is an important host mineral for Si. Therefore,

B. Buggle et al. / Quaternary International 240 (2011) 12e21 15

a weathering index based on Na and Si would be sensitive forinhomogenities in the quartz content (Smykatz-Kloss, 2003). Thesame is true for Zr and Ti, which reside in substantial proportions indiscrete weathering resistant minerals such as zircon, anatase,rutile, beidellite and ilmenite, respectively. These minerals arepossibly present in variable amounts within the profile sequencedue to temporally changing heavy mineral enrichment duringtransport processes (Wedepohl, 1978; Reeder et al., 2006). Toconclude, usingNaas themobile elementof aweathering index, Al isproposed as the immobile counterpart to minimize biases due tovariable mineralogical composition of the loess parent material.Accordingly, simple Al/Na ratios have been used already in earlierstudies to characterize the weathering intensity of a material or soilhorizon (Gallet et al., 1998; Ding et al., 2001; Smykatz-Kloss, 2003).Instead of a simple Al/Na ratio, the use of molar Al/(Na þ Al) ratiotimes 100 is suggested to restrict the index to values between 0 and100. This avoids out-of-scale variations and values if Na contents arelow. This ratio (Formula 1) was formerly applied by Cullers (2000)for carbonate rich shales, siltstones and sandstones and intro-duced as Chemical Index of Weathering (CIW’). Cullers used theapostrophe to indicate thathis ratio is amodified (Ca-free) versionofthe classical CIW, published by Harnois (1988), Table 1. To theauthors’ knowledge, the CIW’ was never applied to loess depositspreviously. However, this index is the most appropriate geochemi-cally based weathering proxy for most LPSS. Regarding the mainhost minerals of Na and Al in loess protoliths, it should indicatefeldspar, especially plagioclase weathering. In the following, theterm Chemical Proxy of Alteration (CPA) is used instead of CIW’ toavoid any confusion with the classical CIW of Harnois (1988).

CPA¼ 100�Al2O3=ðAl2O3þNa2OÞðinmolarproportionsÞ (1)

3.2. Overview on widely used indices of feldspar weathering

To test the before mentioned hypotheses, the results derived bythe proposed CPA were compared with the Rb/Sr and the Ba/Srratios, which are widely employed to characterize weatheringintensity in LPSS (e.g. Liu et al., 1993; Gallet et al., 1996; Chen et al.,1999; Ding et al., 2001; Tan et al., 2006; Bokhorst et al., 2009), oftennotwithstanding the above-mentioned problems concerning theseratios. Furthermore, CPAwas compared to the common establishedindices for feldspar or plagioclase weathering (Table 1): theChemical Index of Alteration (CIA; Nesbitt and Young, 1982), theChemical Index of Weathering (CIW; Harnois, 1988), the PlagioclaseIndex of Alteration (PIA, Fedo et al., 1995) and Index B of Kronbergand Nesbitt (1981) (see also Guggenberger et al., 1998).

The rationale of the CIA is to give a quantitative measure offeldspar weathering by relating Al, which is enriched in theweathering residues, toNa, Ca andK,which shouldbe removed froma soil profile in the course of plagioclase and K-feldspar weathering(Nesbitt and Young, 1982). Index B of Kronberg and Nesbitt (1981)

Table 1Weathering indices (molecular proportions). Note, CaO* refers to silicatic Ca.

CPAa ¼ [Al2O3/(Al2O3 þ Na2O)] � 100CIAb ¼ [Al2O3/(Al2O3 þ Na2O þ CaO* þ K2O)] � 100Index Bc ¼ (CaO* þ Na2O þ K2O)/(Al2O3 þ CaO* þ Na2O þ K2O)CIWd ¼ [Al2O3/(Al2O3 þ Na2O þ CaO*)] � 100PIAe ¼ [(Al2O3 e K2O)/(Al2O3 þ CaO* þ Na2O e K2O)] � 100

a Chemical proxy of alteration (this paper).b Chemical index of alteration (Nesbitt and Young, 1982).c Index B (Kronberg and Nesbitt, 1981).d Chemical index of weathering (Harnois, 1988).e Plagioclase Index of Alteration (Fedo et al., 1995).

(Guggenberger et al., 1998) is based on the same considerations. In1988, Harnoismodified the CIA. He emphasized that K should not beused in weathering indices, since it shows no consistent behaviorduring weathering, being either enriched in the residue, if weath-ering is weak, or depleted under more intense weathering condi-tions. Thus, Kwas eliminated from the CIA and the resulting indexoffeldspar weathering was reported as CIW (Harnois, 1988) or K-freeCIA (Maynard,1992). Fedoet al. (1995) introduced a correctionof theCIW for the Al content in K-feldspar, otherwise rocks rich in K-feldspar would be characterized by misleadingly high CIW values.This modified version of the CIW is reported as PIA, indicatingplagioclaseweathering. The CIA, Index B, CIWand PIA index, requireall the content of silicatic Ca (¼CaO*). This value frommeasured CaOwas obtained according to the procedure described by McLennan(1993), who assumed that the molar CaO/Na2O ratio of carbonate-free silicatic material does not exceed 1.

4. Results

Raw data of the geochemical analyses have been already pre-sented in Buggle et al. (2008). This contribution presents the depthprofiles of the applied weathering indices and of the carbonatecontent (Fig. 2aec). In the following, “Na-type” weathering indicesrefer to the CIA, Index B, CIW, PIA, and the CPA, whereas Rb/Sr andBa/Sr will be regarded as “Sr-type” weathering indices.

At Batajnica/Stari Slankamen, all indices of the Na-type showa similar trend of more intense weathering in the older loess units.Regarding the pedocomplexes, strongest weathering intensity isrecorded in the older unitswith amaximum in the S5. Fromthe S5 tothe recent soil, S0, the peak values of the interglacial pedocomplexesgenerally decreased, except for the S1,which exhibits again strongerfeldspar weathering than the next older interglacial soil formation(S2). The intensity of feldspar weathering in the S6 is lower than inthe S5 and comparable to the S4, as shownbyall Na-typeweatheringproxies. Also, with respect to the detailed patterns, the Na-typefeldspar weathering indices resemble each other closely and reflectsensitively different phases of pedogenesis within a pedocomplex,mostly consistent with patterns of the magnetic susceptibilityrecord. The depth profile of the Sr-type indices shows similarities tothe magnetic susceptibility record. However, these patterns alsostrongly follow the calcium carbonate content and exhibit somedifferences to the “Na-type” indices (Fig. 2a).

At Mircea Voda, the records of the Na-type indices, on the onehand, are very similar to each other and on the other hand, the Sr-type indices resemble each other closely, showing parallels to theCaCO3 record (Fig. 2b). The Na-type indices reveal most intensefeldspar weathering in the basal pedocomplexes S6 and S5, withthe maximum in the lowermost pedocomplex. The feldsparweathering intensity of the S4 and S3 is less. Nevertheless Na-typeindices exhibit still a pronounced enhancement in weatheringintensity in this paleosols compared to the loess. An even lowerdegree of feldspar weathering is revealed for the S2, S1 and therecent soil S0. In contrast to these findings, no comparable trend isshown by the Ba/Sr and Rb/Sr ratios. Comparing loess units, no cleartrend can be recognized in the “Na-type” weathering record, dis-regarding the high values for the thin loess unit L6, which arepossibly caused by the influence of pedogenic alteration during theformation of the S5.

At Stary Kaydaky (Fig. 2c), the succession of loess layers andpaleosols is especially in the lower part of the profile (below S2)hardly reflected by the weathering indices. There, loess units areonly thin and loess as well as paleosol units show multiple pedo-genetic overprinting and exhibit enhanced mineral weathering. Inaccordance with these findings, the profile sequence is almostcarbonate-free, except for some parts of the L1, L3 and L4. These

Fig. 3. Correlation of the Rb/Sr (a) and Ba/Sr ratio (b) with the CaCO3-content for allstudied profiles sequences. The asterisk indicates that the correlation is significant atp < 0.05 (t-test, Statistica 6 software package, Statsoft, Inc., 2001).

B. Buggle et al. / Quaternary International 240 (2011) 12e21 17

carbonate peaks are also clearly reflected by the Ba/Sr and Rb/Srrecord. In the S1, Sr-type indices show an upward decreasing trend,which is neither reflected by the Na-type indices, nor by thecarbonate content or by the magnetic susceptibility record. Thispattern therefore can be best explained by changing composition ofthe Rb/Sr and Ba/Sr ratio of the parent material.

A paleoenvironmental interpretation of the presented weath-ering records and a discussion of the dataset with respect to loessprovenance is beyond the scope of this study and will be givenelsewhere.

5. Discussion

5.1. Evaluation of the geochemical weathering indices

5.1.1. Sr type vs. Na type indicesWithin the last 15 years, “Sr-type“ indices including Rb/Sr or Ba/

Sr gained increasing popularity as weathering proxies, also in LPSS(e.g. Gallet et al., 1996; Chen et al., 1999; Ding et al., 2001; Tan et al.,2006; Bokhorst et al., 2009). The rationale behind this practice isthe fact that Sr can substitute for Ca in minerals and also shows ananalogous behavior to Ca in the weathering profile. Accordingly, Sris easily released into solution and mobilized in the course ofweathering, whereas Rb or Ba can be regarded as relativelyimmobile under moderate weathering conditions due to strongadsorption to clay minerals (Dasch, 1969; Nesbitt et al., 1980; Liuet al., 1993; McLennan et al., 1993; Reeder et al., 2006). For theLPSS Mircea Voda, Batajnica/Stari Slankamen and Stary Kaydaky,the depth profiles of “Sr-type“ indices were compared with those of“Na-type“ indices. The latter rely on the principle that Na is easilyreleased from minerals and mobilized during weathering, whereasAl is retained in the profile, forming secondary clayminerals and/orAl-oxides (see Section 3.2.). As revealed by the depth profiles of theapplied weathering indices (Fig. 2), all “Na-type” indices and all “Sr-type” indices resemble each other closely, but between these twotypes of weathering indices distinct differences can be observed. Acomparison between the depth profiles of the Sr-type indices andthe distribution of CaCO3 suggests that in most cases low and highBa/Sr and Rb/Sr ratios are connected with high and low carbonatecontents, respectively. This observation is confirmed by a signifi-cant correlation (p < 0.05) between the “Sr-type” indices and thecarbonate content (Fig. 3) in all sections and gives evidence fora significant substitution of carbonateeCa by Sr. Therefore, theinitial Rb/Sr and Ba/Sr ratios are supposed to be at least partly post-pedogenetically masked by the dynamics of carbonate-bound Sr.Hence, it is expected that paleosols altered by the precipitation ofsecondary carbonate, which is leached from overlying loess orpaleosols, would exhibit misleadingly low Rb/Sr and Ba/Sr ratios.This would cause an underestimation of the weathering intensityand would also bias the interpretation of these ratios as proxies ofthe leaching intensity, i.e. paleoprecipitation, as also stated byRetallack and Germán-Heins (1994), Retallack (1997) and Tan et al.(2006). To conclude, it is recommended to restrict the use of “Sr-type“ weathering indices such as Rb/Sr and Ba/Sr to carbonate-freematerial, where they should reflect the intensity of silicateweathering (Dasch, 1969; Nesbitt et al., 1980), keeping in mind thatunder extreme weathering conditions also Rb and eventually evenBa may undergo mobilization (Nesbitt et al., 1980; Marques et al.,2004; Reeder et al., 2006). For LPSS, mostly characterized by

Fig. 2. The CPA, CIA, Index B, CIW, PIA, Ba/Sr, Rb/Sr record of the Batajnica/Stari SlankameKaydaky section in Ukraine (Fig. 2c). See Section 3.2 for a more detailed description of thesebetter comparison with the weathering proxy records. The magnetic susceptibility record (2008; Buggle et al., 2009; Markovi�c et al., 2009).

strong carbonate dynamics, it is proposed to employ weatheringindices of the “Na-type”, such as the CPA.

5.1.2. The “classical“ Na-type weathering indices e uncertaintiesdue to calcium carbonate

The most prominent “Na-type“ indices of feldspar weatheringare the Index B of Kronberg and Nesbitt (1981), the CIA (Nesbitt andYoung, 1982) and especially for plagioclase weathering the CIW(Harnois, 1988) and the PIA (Fedo et al., 1995; see Section 3.2 andTable 1). All of these indices employ themolar content of the silicaticbound Ca, given usually as CaO*. For this, one has to know thecontribution of the carbonate Ca to themeasured CaO,whendealingwith calcareous material. The most widely used methods to deter-mine the carbonate content rely on the “selective removal tech-nique”, i.e. eitheron the selective removal of theorganic carbon fromthe inorganic carbon during low temperature combustion or theselective removal of carbonate during acid treatment (Hedges andStern, 1984; Bisutti et al., 2004, 2007). However, due to an imper-fect selectivity of these methods (Froelich, 1980; Hedges and Stern,1984; Bisutti et al., 2004), one may derive erroneous carbonatecontents, which consequently bias also the calculation of CaO* fromtotal CaO and carbonate-bound CaO.

To avoid the time consuming step of carbonate determination,the CaO* is often estimated following the procedure of McLennan(1993) (e.g. Gallet et al., 1998; Yadav and Rajamani, 2004;Schellenberger and Veit, 2006; Lacka et al., 2007). He suggestedto first correct the measured molar CaO content for Ca in apatite, ascalculated from the P2O5 content, and then to compare the result-ing value (here termed CaOcorr) with the molar Na2O content. If the

n section in Serbia (Fig. 2a), the Mircea Voda section in Romania (Fig. 2b), the Staryweathering indices. Note that the carbonate content is given with an inverse scale for

Buggle et al., 2009) is shown to facilitate correlation to previous studies (Buggle et al.,

B. Buggle et al. / Quaternary International 240 (2011) 12e2118

latter is smaller than CaOcorr, a molar CaO*/Na2O ratio of one isassumed. In the other case, the molar CaO* is set equal to the molarCaOcorr. However, this estimation procedure may cause uncer-tainties in calculating CaO*-based weathering indices. Table 2exhibits this for a 1/7 mixing ratio of the plagioclase end-members anorthite/albite, which can be a realistic value for loess(Dultz and Graf von Reichenbach, 1995). Three scenarios werecalculated. The first one, which assumes a pure mixture of 80 gplagioclase and 20 g calcium carbonate, shows that the CIWcalculated from estimated CaO* values is underestimating the“real” CIW by 11.2 units, i.e. 22.4% deviation from the real value.This is also the case for the second scenario applying only 10 gCaCO3. Thus, in both cases the CIW calculated by following theprocedure of McLennan (1993) is lower than the theoretical valueof the unaltered plagioclase, i.e. the “real” CIW. Also for the otherindices an underestimation of the weathering intensity wasobtained. Regarding the formula of the CIW, PIA and CIA, the actualerror due to CaO* estimation should decrease with increasing Alcontent. Therefore, the third scenario also accounts for the pres-ence of other aluminous phases in loess such as K-feldspar orsecondary clay minerals (see caption of Table 2 for a detaileddescription of this scenario). However, also this variant, regardinga composition more realistic for loess deposits, reveals an under-estimation of the weathering intensity by more than 10%. Conse-quently, the interpreter of these weathering proxy records, wouldfor example overestimate the weathering enhancement of

Table 2Sensitivity analysis for the CIW, PIA, CIA and Index B (see Table 1) and the obtainederror due to the estimation of silicate bound Ca (CaO*) following the procedure ofMcLennan (1993). Three scenarios were calculated. Scenarios 1 and 2 assumea mixture of 20 and 10 g calcite, respectively, with 80 g plagioclase. For theplagioclase composition, an anorthite/albite mixing ratio of 1/7 is assumed e

a realistic value for loess deposits (Dultz and Graf von Reichenbach,1995). Scenario 3takes also account of other Al phases, such as K-feldspar and secondary Al minerals.To achieve realistic element ratios we choose a K-feldspar content of 54.3 g and anadditional Al2O3 content (Al2O3-sec) of 34.8 g. The Al2O3-sec can be regarded as Al ofsecondary clay minerals or Al-oxides. These preset values correspond to an Na2O/K2O ([%]/[%]) ratio of 0.9 and an Al2O3/Na2O ([%]/[%]) ratio of 7.5. These values are inbetween the observed range for most loesses in various parts of the world, i.e.1.3e0.5 for Na2O/K2O and 6e9 for Al2O3/Na2O (Taylor et al., 1983; Gallet et al., 1996,1998; Buggle et al., 2008; Újvári et al., 2008). The subscript “tot” refers to the totalcontent of the oxide, as calculated from the preset mineralogical composition. Thereal content of CaO*, as calculated for each scenario, is given as “CaO*real”, whereasCaO*estimated terms the estimated CaO* following McLennan (1993). Accordingly, theresults present “real” and “estimated” weathering indices. Difference between bothis given in percent of the real value.

Scenariopresetting

Scenario 1 Scenario 2 Scenario 3Anorthite/AlbiteMixing ratio [g/g]

10/70 10/70 10/70

CaCO3 [g] 20 10 10K-feldspar [g] 0 0 54.3A12O3-sec [g] 0 0 34.8

Interim valuesfor indexcalculation

A12O3tot [mmol] 169.4 169.4 608.5Na2Otot [mmol] 133.5 133.5 133.5K2Otot [mmol] 0 0 97.6CaOtot [mmol] 235.8 135.9 135.9CaO*

estimated [mmol] 133.5 133.5 133.5CaO*

real [mmol] 35.9 35.9 35.9Results CIWestimated 38.8 38.8 69.5

CIWreal 50 50 78.2Difference [%] 22.4 22.4 11.1PIAestimated 38.8 38.8 65.7PIAreal 50 50 75.1Difference [%] 22.4 22.4 12.5CIAestimated 38.8 38.8 62.5CIAreal 50 50 69.5Difference [%] 22.4 22.4 10.1Index Bestimated 61.2 61.2 37.5Index Breal 50 50 30.4Difference [%] 22.4 22.4 23.4

a paleosol compared to the underlying loess parent material, if thepaleosol is carbonate-free and the loess is not.

To avoid such uncertainties, it is recommended for LPSS studiesto apply a “Na-type” weathering index as CPA, which does notemploy Ca. This is in line with Jeong et al. (2008), who proposed toomit CaO from weathering indices in LPSS due to the presence ofsecondary carbonates.

5.1.3. The chemical proxy of alteration (CPA) e an evaluationAs hypothesized in Section 3.2, the CPA should be a suitable

weathering index for LPSS, indicating especially plagioclaseweathering. Indeed, the good correspondence to the “classical” Na-type plagioclase weathering indices, i.e. the CIWand the PIA (Fig. 2)confirms the proposed interpretation of the CPA as proxy of theplagioclase weathering intensity. However, in contrast to the“classical” indices, the CPA does not involve CaO*. Therefore, it isfree of the CaO* related uncertainties. Though these uncertaintiesare apparently small in the Serbian, Romanian and Ukrainian sites,they could be remarkable on other loess sites depending onmineralogical composition, as shown in Section 5.1.2.

Fig. 2 compared the CPA and other plagioclase weatheringindices to the CIA and Index B, which have been previouslyproposed as silicate weathering proxies. The objective of theseindices is to quantify also weathering of K-feldspar and mica byemploying K. However, these indices show close similarity to theCIA and CIW (Fig. 2). Thus, as long as plagioclase weathering doesnot reach saturation, K-free indices are also a good proxy for theintensity of silicate weathering in general. This is supported by theAl2O3/Na2O and Al2O3/K2O depth profiles of the studied sections,showing that K variations mimic the Na variations, however, withsmaller amplitude (Fig. 4).

These results are in accordance with observations in otherweathering studies and theoretical considerations of the elementbehavior, suggesting that K release is small compared to the Narelease. This is due to stronger weathering resistance of K phasessuch as K-feldspar and due to the fixation of K on clay minerals(Nesbitt and Young, 1984, 1989; Blum, 1994; Smykatz-Kloss, 2003;Yang et al., 2004; Reeder et al., 2006). Though weathering profilesof the proposed CPA and K-free Na-type indices are consistent forthe studied sections, this might not be true in other sites withstrong K fixation and illitization. Harnois (1988) has pointed outthat K-fixation can cause an inconsistent behavior of this element inthe weathering environment and thus, he recommended not to useK inweathering indices. The CPA being a K-free index takes accountof this recommendation. Furthermore, it avoids uncertainties dueto determination of CaO* and thus the CPA can be applied moregeneral on loessepaleosol sequences.

As a conclusion, the CPA seems to be the most promisingweathering proxy for LPSS. However, as with other weatheringindices, it requires also certain prerequisites to be fulfilled. Dealingwith a “Na-type” weathering index, the studied material has to befree of Na salts, which would lead to an underestimation of theweathering intensity. Within mid-latitudinal loess deposits,significant amounts of these salts are only expected in exceptionalsettings as near to the seashore or in locations with warm-(semi-)arid climate and groundwater near to surface, either in the past orin the present time. For the studied sections, an influence of Na-salts is not likely due to the plateau situation of the loess, the lack ofa soil structure characteristic of a natric horizon (IUSS WorkingGroup WRB, 2006), and the geochemical composition (Buggleet al., 2008). The latter does not indicate a relationship of the CPAto the dynamics of other salts such as gypsum, but rather to themagnetic susceptibility as independent pedogenesis proxy (Fig. 2).However, it has to be evaluated by further studies whether theremarkable minimum of the CPA at the lower boundary of the L5 in

Fig. 4. Molar Al2O3/K2O and Al2O3/Na2O ratios of the studied profiles. Subscript “n” indicates that the data are normalized to the lowest value of each section in order to comparethe relative changes of the ratios.

B. Buggle et al. / Quaternary International 240 (2011) 12e21 19

Mircea Voda indeed reflects low weathering intensity due to coldand/or dry paleoclimatic conditions. A possible connection to therecent gypsum formation at the front face of the exposure wall inthe respective depth cannot be excluded. A record of the chlorinecontent would be useful to clarify such inconclusive situations infuture studies.

A second prerequisite for all Na-type indices is the absence ofmineral or grain size sorting in the sampled material. This can betested by using an Al2O3eCaO* þ Na2OeK2O ternary plot e alsoknown as AeCNeK diagram (Nesbitt and Young, 1984). Thisdiagram informs about weathering and sorting effects of alumi-nosilicates, as well as the initial composition of the unweatheredmaterial (e.g. Nesbitt and Young, 1989; McLennan et al., 1993;Nesbitt et al., 1996; Fig. 5). A sorting effect, i.e. a selective enrich-ment of coarser (finer), more feldspathic (more clayey and alumi-nous material), as revealed for the Stary Kaydaky section (Fig. 5b,d), would cause a decrease (increase) of the Al/Na ratio and the CPA.

The third prerequisite is common for all types of weatheringindices: the homogeneity of the parent material. With respect tothe CPA, a relatively homogeneous composition of the unweath-ered material regarding the most abundant aluminous Na phase,i.e. albite, in relation to the aluminous K phases, i.e. mostly K-feldspar and mica, and to the Ca phase, i.e. anorthite, is important.For example, an increasing K-feldspar/albite ratio of the parentmaterial would cause a higher Al/Na ratio (Fedo et al., 1995). Thiswould result in a misleading increase of the CPA. Also thisprerequisite can be tested using the AeCNeK diagram. Variations inthe K-feldspar or mica to plagioclase ratio of the unweatheredparent material would be revealed by a scatter of the data pointsparallel to the CNeK axis (Fig. 5a). On the other hand, a singleweathering line would indicate parent material with an invariablecomposition of aluminosilicates, as it is the case for the data pointsof the sections Mircea Voda and Batajnica/Stari Slankamen (Fig. 5).The congruence between the CIA, Index B, PIA and K-free indices asthe CIW and CPA gives further reason to assume homogeneity ofthe unweathered loess parent material at the investigated sections,

regarding the (K-feldspar þ mica)/albite ratio. For the Stary Kay-daky site, it is not possible to exclude variations in the (K-feldsparþmica)/albite ratio due to the scatter along the sorting linebeing parallel to the CNeK axis (Fig. 5d).

The invariance of the albite/anorthite mixing ratio of plagioclasecan be assessed indirectly, assuming that it is controlled by the typeand composition of the igneous source rocks of the loess parentmaterial. This assumption seems plausible, since many loessdeposits around the world have been identified as recycled sedi-mentary material (Taylor et al., 1983; Gallet et al., 1998; Buggleet al., 2008), essentially originating from igneous protoliths.Accordingly, an increasing felsic (mafic) character of the idealizedprotolith would cause higher (lower) albite/anorthite ratios. As K isenriched in felsic rocks, an invariant K/Na ratio of the protolith, ascan be inferred for the Batajnica/Stari Slankamen and Mircea Vodasections, should also indicate a relatively stable albite/anorthiteratio of the unweathered protolith.

If the composition of the parent material changes down-profile,the CPA still could be reasonably applied to LPSS using DCPA values.DCPA values can be obtained by relating CPA values of a weatheringhorizon or paleosol to the CPA value of the respective parentmaterial (background CPA), i.e. the loess layer from which eachpaleosol developed. These DCPA values can be interpreted in termsof weathering enhancement.

Therefore, the application of the Al/(Na þ Al) � 100 ratio (CPA,CWI’ according to Cullers, 2000) is proposed not only for calcareousmarine sediments, but also for loessepaleosol sequences asa measure of silicate weathering intensity. As with other weath-ering indices, a homogeneous parent material (regarding the rele-vant host minerals) is required to obtain a continuous weatheringrecord. However, using the CPA, this prerequisite can be easilyevaluated via the AeCNeK diagram not needing UCC normalizedplots of trace elements and REE. Furthermore, diagenetic effectsdue to dynamics of secondary carbonate or K-fixation (illitization)are no issue in contrast to other indices (Sr-type indices, Na-typeindices involving CaO* and K2O).

Fig. 5. The AeCNeK (Al2O3eCaO* þ Na2OeK2O) e ternary diagram according to Nesbitt and Young (1984). The characteristic position of the upper continental crust (UCC), basalt,granite and the minerals plagioclase (Pl.), K-feldspar (Ks.), biotite (Bi.), muscovite (Mu.), illite (IL.), smectite (Sm.), kaolinite (Ka.), and gibbsite (Gi.) is given for orientation. Note thatonly the upper part of the ternary diagrams is shown, which is of interest for the present study. In Fig. 5aed a typical weathering line is presented, emerging from loess sourcematerial with UCC-like composition, as found to be true for many loess deposits around the world and also the Southeastern European loesses (e.g. Taylor and McLennan, 1985;Gallet et al., 1998; Buggle et al., 2008). The first part of the weathering line is (sub-) parallel to the AeCN join, representing prevailing Ca and Na removal due to plagioclaseweathering. With plagioclase weathering being in saturation, i.e. approaching to the AeK join, the second part of the weathering line is redirected to the Al2O3-apex as a result ofpredominantly loss of K by weathering of K-rich phases like K-feldspar (Nesbitt and Young, 1984). In Fig. 5a, it is shown how biases due to a changing composition of the parentmaterial would appear in the AeCNeK diagram. Variations in the K-feldspar/plagioclase ratio (Ks./Pl. ratio) of the parent material would cause a shift parallel to the CNeK join andthe datapoints would not plot on the same weathering line. In Fig. 5b, the sorting effect is demonstrated. A sample enriched in fine and more clayey material due to grain size andmineral sorting plots closer to the Al2O3-apex and a sample enriched in coarse and less clayey material plots vice versa. Fig. 5c shows the effect of errors in the CaO* content, forexample due to the estimation procedure of McLennan (1993). An overestimation of the CaO* would cause a shift from the original weathering line towards the CN apex, anunderestimation vice versa. Since this line of “CaO* uncertainty” is close to the original weathering line, an erroneous CaO* would hardly affect the identification of mineral/grainsize sorting and of a variable Ks./Pl. ratio of the source material. Fig. 5d. Datapoints for loess and paleosol samples from the Batajnica/Stari Slankamen, Mircea Voda and StaryKaydaky sections are shown (modified after Buggle et al., 2008). The samples from the Batajnica/Stari Slankamen and Mircea Voda sections plot on a plagioclase weathering lineoriginating from the UCC, not indicating a variable Ks./Pl. ratio. Samples from Stary Kaydaky plot on a sorting line, which is possibly modified by variable Ks./Pl. ratios. See Buggleet al. (2008) for a more detailed discussion of these features with respect to loess provenance.

B. Buggle et al. / Quaternary International 240 (2011) 12e2120

6. Conclusions

Commonly appliedweathering indices involving Ti, Zr, and Si arerelatively sensitive for changes in parent material composition.Otherwidely usedweathering indices relying on Al as the immobileelement suchas theCIA (Nesbitt andYoung,1982), theCIW(Harnois,1988), the PIA (Fedo et al., 1995) and the Index B (Kronberg andNesbitt, 1981) involve uncertainties due to diagenetic effects (illiti-zation). Estimation of silicate Ca in calcareous material, as commonin most loesses, may lead to biased weathering records using theseindices. Furthermore, carbonate-free element ratios incorporatingSr, such as the Ba/Sr and Rb/Sr ratio (e.g. Liu et al., 1993; Gallet et al.,1996; Bokhorst et al., 2009), can be problematic due to interferencesof the carbonate and Sr dynamics. To overcome such uncertainties,the Chemical Proxy of Alteration CPA (the molar ratio Al2O3/(Al2O3 þ Na2O) � 100) e also known as CIW’ (Cullers, 2000) e is

proposed as a more appropriate geochemical proxy of silicateweathering for LPSS. Homogeneity of the parent material can bechecked for this index via the AeCNeK diagram.

Acknowledgements

We thank Tivadar Gaudenyi and Mladjen Jovanovi�c for assis-tance during fieldwork in Serbia, Ana Malagodi for support insample preparation, and J. Eidam (University of Greifswald) for XRFanalyses. This study was financially supported by the GermanResearch Foundation DFG (GL 327/8-2).

References

Bisutti, I., Hilke, I., Raessler, M., 2004. Determination of total organic carbon e anoverviewof currentmethods. Trends inAnalyticalChemistry23 (10e11), 716e726.

B. Buggle et al. / Quaternary International 240 (2011) 12e21 21

Bisutti, I., Hilke, I., Schumacher, J., Raessler, M., 2007. A novel single-run dualtemperature combustion (SRDTC) method fort he determination of organic, in-organic and total carbon in soil samples. Talanta 71, 521e528.

Blum, A.E., 1994. Feldspars in weathering. In: Parsons, I. (Ed.), Feldspars and TheirReactions. NATO ASI Series, Series C: Mathematical and Physical Sciences, vol.421. Kluwer Academic Publishers, Dordrecht, pp. 595e630.

Blumer, M., 1950. Die Existenzgrenzen anorganischer Ionen bei der Bildung vonSedimentgesteinen. Geochemische Untersuchungen IV. Helvetica Chimica Acta33 (6), 1468e1581.

Bokhorst, M.P., Beets, C.J., Markovic, S.B., Gerasimenko, N.P., Matviishina, Z.N.,Frechen, M., 2009. Pedo-chemical climate proxies in Late Pleistocene Ser-bianeUkrainian loess sequences. Quaternary International 198, 113e123.

Buggle,B.,Glaser, B., Zöller, L.,Hambach,U.,Markovi�c, S.,Glaser, I., Gerasimenko,N., 2008.Geochemical characterization and origin of Southeastern and eastern Europeanloesses (Serbia, Romania, Ukraine). Quaternary Science Reviews 27, 1058e1075.

Buggle, B., Hambach, U., Glaser, B., Gerasimenko, N., Markovi�c, S.B., Glaser, I.,Zöller, L., 2009. Stratigraphy and spatial and temporal paleoclimatic trends inEast European loess paleosol sequences. Quaternary International 196, 86e106.

Chen, J., An, Z., Head, J., 1999. Variation of the Rb/Sr ratios in the loessepaleosolsequences of Central China during the last 130,000 years and their implicationsfor monsoon paleoclimatology. Quaternary Research 51, 215e219.

Cullers, R.L., 2000. The geochemistry of shales, siltstones and sandstones of Penn-sylvanianePermian age, Colorado, USA: implications for provenance andmetamorphic studies. Lithos 51, 181e203.

Dasch, E.J., 1969. Strontium isotopes in weathering profiles, deep-sea sediments,and sedimentary rocks. Geochimica et Cosmochimica Acta 33 (12), 1521e1552.

Derbyshire, E., Kemp, R.A., Meng, X., 1997. Climate change, loess and palaeosols:proxy measures and resolution in North China. Journal of the GeologicalSociety, London 154, 793e805.

Ding, Z.L., Sun, J.M., Yang, S.L., Liu, T.S., 2001. Geochemistry of the Pliocene red clayformation in the Chinese Loess Plateau and implications for its origin, sourceprovenance and paleoclimatic change. Geochimica et Cosmochimica Acta 65(6), 901e913.

Dultz, S., Graf von Reichenbach, H., 1995. Quantitative Mineralbestimmung in derSchlufffraktion von Böden auf der Grundlage der chemischen Analyse undunter Anwendung der Karl-Fischer-Titration. II. Ergebnisse bei Böden ausGeschiebedecksand, Geschiebemergel und Löß. Journal of Plant Nutrition andSoil Science 158, 459e464.

Evans, M.E., Heller, F., 2003. Environmental Magnetism e Principles and Applica-tions of Enviromagnetics. Academic Press, Amsterdam.

Fedo, C.M., Nesbitt, H.W., Young, G.M., 1995. Unraveling the effects of potassiummetasomatism in sedimentary rocks and paleosols, with implications forpaleoweathering conditions and provenance. Geology 23 (10), 921e924.

Froelich, P.N., 1980. Analysis of organic carbon in marine sediments. Limnology andOceanography 25 (3), 564e572.

Gallet, S., Jahn, B., Torii, M., 1996. Geochemical characterization of the Luochuanloess paleosol sequence, China, and paleoclimatic implications. ChemicalGeology 133, 67e88.

Gallet, S., Jahn, B., Van Vliet Lanoë, B., Dia, A., Rossello, E., 1998. Loess geochemistryand its implications for the particle origin and composition of the uppercontinental crust. Earth and Planetary Science Letters 156, 157e172.

Goldschmidt, V.M., 1937. The principles of distribution of chemical elements inminerals and rocks. Journal of the Chemical Society, 655e673.

Guggenberger, G., Bäumler, R., Zech,W.,1998.Weathering of soils developed in eolianmaterial overlyingglacial deposits ineasternNepal. Soil Science163 (4), 325e337.

Harnois, L., 1988. The CIW index: a new chemical index of weathering. SedimentaryGeology 55, 319e322.

Harriss, R.C., Adams, J.A.S., 1966. Geochemical and mineralogical studies on theweathering of granitic rocks. American Journal of Science 264 (2), 146e178.

Hedges, J.I., Stern, J.H., 1984. Carbon and nitrogen determinations of carbonate-containing solids. Limnology and Oceanography 29 (3), 657e663.

IUSS Working Group WRB, 2006. World Reference Base for Soil Resources 2006. In:World Soil Resources No. 103. FAO, Rome.

Jeong, G.Y., Hillier, S., Kemp, R.A., 2008. Quantitative bulk and single particlemineralogy of a thick Chinese loessepaleosol section: implications for loessprovenance and weathering. Quaternary International 27, 1271e1287.

Kabata-Pendias, A., Pendias, H., 2001. Trace Elements in Soils and Plants. CRC Press,Boca Raton.

Kronberg, B.I., Nesbitt, H.W., 1981. Quantification of weathering, soil geochemistryand soil fertility. Journal of Soil Science 32, 453e459.

Kutterolf S., 2001. Die klastischen Sedimente der karbonen Hochwipfel e undAuernig Formation der Ostkarawanken (Österreich/Slowenien). Dissertation,University of Stuttgart, URL: http://elib.uni-stuttgart.de/opus/volltexte/2002/994/, URN: urn:nbn:de:bsz:93-opus-9948; downloaded at 02.06.2010.

Lacka, B., Lanczont, M., Madeyska, T., Boguckyj, A., 2007. Geochemical composition ofVistulial loess andmicromorphologyof interstadial palaeosols at theKolodiiv site(East Carpathian Foreland, Ukraine). Geological Quarterly 51 (2), 127e146. .l.

Liu, C.-Q., Masuda, A., Okada, A., Yabuki, S., Zhang, J., Fan Z.-, L., 1993. A geochemicalstudy of loess and desert sand in northern China: implications for continentalcrust weathering and composition. Chemical Geology 106, 359e374.

Maher, B.A., Thompson, R., Zhou, L.P., 1994. Spatial and temporal reconstructions ofchanges in the Asian palaeomonsoon: a new mineral magnetic approach. Earthand Planetary Science Letters 125, 461e471.

Markovi�c, S.B., Hambach, U., Catto, N., Jovanovi�c,M., Buggle, B.,Machalett, B., Zöller, L.,Glaser, B., Frechen, M., 2009. Middle and Late Pleistocene loess sequences atBatajnica, Vojvodina, Serbia. Quaternary International 198, 255e266.

Marques, J.J., Schulze, D.G., Curi, N., Mertzman, S.A., 2004. Trace elementgeochemistry in Brazilian Cerrado soil. Geoderma 121 (1e2), 31e43.

Mason, B., Moore, C.B., 1985. Grundzüge der Geochemie. Enke-Verlag, Stuttgart.Maynard, J.B., 1992. Chemistry of modern soils as a guide to interpreting Precam-

brian paleosols. The Journal of Geology 100 (3), 279e289.McBride, M.B., 1994. Environmental Chemistry of Soils. Oxford University Press,

New York.McLennan, S.M., 1993. Weathering and global denudation. The Journal of Geology

101, 295e303.McLennan, S.M., Hemming, S., McDaniel, D.K., Hanson, G.N., 1993. Geochemical

approaches to sedimentation, provenance, and tectonics. In: Johnsson, M.J.,Basu, A. (Eds.), Processes Controlling the Composition of Clastic Sediments,Special Paper. Geological Society of America, vol. 284, pp. 21e40.

Muhs, D.R., Bettis, E.A., Been, J., McGeehin, J.P., 2001. Impact of climate and parentmaterial on chemical weathering in loess-derived soils of the Mississippi Rivervalley. Soil Science Society of America Journal 65 (6), 1761e1777.

Nesbitt, H.W., Young, G.M., 1982. Early Proterozoic climates and plate motionsinferred from major element chemistry of lutites. Nature 299, 715e717.

Nesbitt, H.W., Young, G.M., 1984. Prediction of some weathering trends of plutonicand volcanic rocks based on thermodynamic and kinetic considerations. Geo-chimica et Cosmochimica Acta 48, 1523e1534.

Nesbitt, H.W., Young, G.M., 1989. Formation and diagenesis of weathering profiles.The Journal of Geology 97, 129e147.

Nesbitt, H.W., Markovics, G., Price, R.C., 1980. Chemical processes affecting alkalisand alkaline earths during continental weathering. Geochimica et Cosmochi-mica Acta 44, 1659e1666.

Nesbitt, H.W., Young, G.M., McLennan, S.M., Keays, R.R., 1996. Effects of chemicalweathering and sorting on the petrogenesis of siliciclastic sediments, withimplications for provenance studies. The Journal of Geology 104, 525e542.

Pye, K., 1995. The nature, origin and accumulation of loess. Quaternary ScienceReviews 14, 653e667.

Railsback, L.B., 2003. An earth scientist’s periodic table of the elements and theirions. Geology 31 (9), 737e740.

Railsback, L.B., 2005. A synthesis of systematic mineralogy. American Mineralogist90 (7), 1033e1041.

Reeder, S., Taylor, H., Shaw, R.A., Demetriades, A., 2006. Introduction to the chem-istry and geochemistry of the elements. In: Tarvainen, T., de Vos, M. (Eds.),Geochemical Atlas of Europe. Part 2. Interpretation of Geochemical Maps,Additional Tables, Figures, Maps, and Related Publications. Geological Survey ofFinland, Espoo, pp. 48e429.

Retallack, G.J., 1997. A Colour Guide to Paleosols. Wiley, Chichester.Retallack, G.J., Germán-Heins, J., 1994. Evidence from paleosols for the geological

antiquity of rain-forest. Science 265 (5171), 499e502.Schellenberger, A., Veit, H., 2006. Pedostratigraphy and pedological and geochem-

ical characterization of Las Carreras loessepaleosol sequence, Valle de Tafi, NW-Argentina. Quaternary Science Reviews 25, 811e831.

Schwertmann, U., 1993. Relations between iron oxides, soil color, and soil forma-tion. In: Bigham, J.M., Ciolkosz, E.J., Luxmoore, R.J. (Eds.), Soil Color SSSA SpecialPublication, vol. 31, pp. 51e70.

Smalley, I., Leach, J.A., 1978. The origin and distribution of the loess in the DanubeBasin and associated regions of East-Central Europe - a review. SedimentaryGeology 21, 1e26.

Smalley, I., Krinsley,D.H.,1978. Loessdeposits associatedwithdeserts. Catena5,53e66.Smalley, I., O’Hara-Dhand, K.,Machalett, B., Jary, Z., Jefferson, I., 2009. Rivers and loess:

the significance of long river transportation in the complex event-sequenceapproach to loess deposit formation. Quaternary International 198 (1e2), 7e18.

Smykatz-Kloss B., 2003. Die Lößvorkommes des Pleiser Hügellandes bei Bonn undvon Neustadt/Wied sowie der Picardie: Mineralogisch-geochemische undgeomorphologische Charakterisierung, Verwitterungs-Beeinflussung und Her-kunft der Lösse. Dissertation, University of Bonn. URL: http://hss.ulb.uni-Bonn.de/diss_online/math_nat_fak/2003/smykatz-kloss_bettina, downloaded at02.06. 2008, URN: urn:nbn:de:hbz:5n-03082.

Tan, H., Ma, H., Zhang, X., Lu, H., Wang, J., 2006. Typical geochemical elements inloess deposits in the Northeastern Tibetan Plateau and its paleoclimaticimplications. Acta Geologica Sinica 80 (1), 110e117.

Taylor, S.R., McLennan, S.M., 1985. The Continental Crust: Its Composition andEvolution. Blackwell, London.

Taylor, S.R., McLennan, S.M., McCulloch, M.T., 1983. Geochemistry of loess, conti-nental crustal composition and crustal model ages. Geochimica et Cosmochi-mica Acta 47, 1897e1905.

Újvári, G., Varga, A., Balogh-Brunstad, Z., 2008. Origin, weathering and geochemicalcomposition of loess in southwestern Hungary. Quaternary Research 69 (3),421e437.

Wedepohl, K.H., 1978. Handbook of Geochemistry. Springer Verlag, Berlin.Yadav, S., Rajamani, V., 2004. Geochemistry of aerosols of northwestern part of India

adjoining the Thar Desert. Geochimica et Cosmochimica Acta 68 (9),1975e1988.

Yang, S.Y., Li, C.X., Yang, D.Y., Li, X.S., 2004. Chemical weathering of the loessdeposits in the lower Changjiang Valley, China, and paleoclimatic implications.Quaternary International 117, 27e34.