Oxygen isotope signatures of quartz from major Asian dust sources: Implications for changes in the provenance of Chinese loess

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 139 (2014) 399–410

Oxygen isotope signatures of quartz from major Asiandust sources: Implications for changes in the provenance

of Chinese loess

Yan Yan a,b,⇑, Youbin Sun a,c,⇑, Hongyun Chen d, Long Ma a,b

a State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi’an 710075, Chinab University of Chinese Academy of Sciences, Beijing 100049, China

c Institute of Global Environmental Change, Xi’an Jiaotong University, Xi’an 710049, Chinad The Institute of Hydrogeology and Environmental Geology, Chinese Academy of Geological Sciences, Shijiazhuang 050803, China

Received 30 August 2013; accepted in revised form 25 April 2014; Available online 9 May 2014

Abstract

We present a systematic investigation of the oxygen isotopic composition of quartz in both fine and coarse fractions (<16and 16–63 lm) from major dust source regions in East Asia, including the Mongolian Gobi, the northern Chinese deserts, theTaklimakan desert, and the Qaidam Basin. The results demonstrate that the quartz oxygen isotope ratios of the Taklimakandesert and the Mongolian Gobi are more heterogeneous compared with the other areas. The quartz d18O values of both thefine and coarse fractions from the various sources are overlapped to varying degrees, thus making it difficult to differentiatethem. Nevertheless, the quartz d18O values of both fractions exhibit an increasing trend from the Mongolian Gobi, to thenorthern Chinese deserts, and then to the Taklimakan desert. This implies that the geological settings of the source areasare different, which in turn results in differing contributions of high-temperature igneous rocks. The combination of quartzd18O results with other quartz-based provenance tracers can clearly differentiate the three major source areas, i.e., theTaklimakan desert, the Mongolian Gobi, and the northern Chinese deserts. In addition, comparison of our results with pre-vious d18O measurements of fine-grained quartz from the Luochuan loess sequence suggests the likely glacial–interglacial fluc-tuations in dust provenance. Finally, we suggest that the combination of quartz d18O signatures and other dust provenancetracers can potentially improve the recognition of long-term fluctuations in the provenance of Chinese loess-red clay deposits.� 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

The extensive loess deposit on the Chinese Loess Plateau(CLP) provides a valuable geological archive comprisingmillions of years of environmental history (Liu, 1985; An,2000; Guo et al., 2002; Qiang et al., 2011). Based on variousloess-based proxies, such as grain size, sedimentation rate,

http://dx.doi.org/10.1016/j.gca.2014.04.043

0016-7037/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding authors. Address: 10 Fenghui South Road,High-tech zone, Xi’an 710075, China. Tel.: +86 29 88321862.

E-mail addresses: [email protected] (Y. Yan), [email protected](Y. Sun).

magnetic susceptibility, carbonate concentration, elementalratios, and isotopic composition, the paleoclimatic andinland aridification history of East Asia since the late Oligo-cene have been intensively reconstructed (e.g., Liu andDing, 1998; An, 2000; Guo et al., 2002; Sun and An,2002). In addition, dynamical connections between EastAsian monsoon variation and solar radiation, global icevolume (An et al., 1991; Ding et al., 1995), Tibetan Plateauuplift (An et al., 2001), and Atlantic meridional overturningcirculation (Sun et al., 2012), have also been investigatedbased on comparison of environmental proxies. Based onthe interpretation of the causes of temporal variations of

400 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

these proxies, the evolution of past atmospheric circulationis often considered to be the primary controlling factor onloess accumulation and post-depositional alteration.However, variations in the sources of the loess depositscan be a complicating factor influencing the variability ofcertain geochemical and isotopic proxies (Gu et al., 1999;Xiong et al., 2001; Hao et al., 2010; Xiao et al., 2012). Prov-enance studies of Chinese loess are therefore important tounravel the complex environmental records preserved inChinese loess and red clay sequences.

Dust originates mainly from arid inland drainage basins(Middleton et al., 1986; Prospero et al., 2002; Washingtonet al., 2003). The vast deserts occupying inland drainagebasins in northwestern China and southern Mongolia areconsidered as the main dust sources of Chinese loess depos-its (Liu, 1985). Provenance studies of modern dust, includ-ing monitoring and numerical modeling, indicate that thereare three major source areas: the Taklimakan desert, thedeserts in northern China, including the Badain Juranand Tengger deserts (subsequently referred to as the north-ern Chinese deserts), and the Gobi desert in southern Mon-golia (Sun et al., 2001; Zhang et al., 2003). These majorsource areas also have relatively distinct mineralogical, ele-mental, and isotopic characteristics (Zhang et al., 1997;Derbyshire et al., 1998; Jahn et al., 2001; Sun, 2002; Chenet al., 2007; Li et al., 2007; Ferrat et al., 2011; Pullenet al., 2011; Li et al., 2013), which facilitate additional prov-enance studies of Chinese loess.

According to major and trace element and Nd–Sr iso-tope characteristics of loess deposits on the CLP, it was sug-gested that loess originates from a rather homogeneous andstable source region (Gallet et al., 1996; Jahn et al., 2001;Wang et al., 2007; Li et al., 2009). Comparison of elementaldata, Nd–Sr isotopic evidence, Electronic Spin Resonance(ESR) signal intensity of the E01 center and crystallinityindex (CI) of quartz, and the U–Pb age spectrum of zirconfrom major dust sources and loess sections suggests thatloess provenance varies on tectonic and orbital timescales(Zhang et al.,1997; Sun, 2002, 2005; Sun et al., 2008; Liet al., 2011; Pullen et al., 2011; Xiao et al., 2012; Chenand Li, 2013). In particular, the ESR signal intensity andCI characteristics of fine quartz suggest that since the lastinterglacial the loess deposits were mainly derived fromthe northern Chinese deserts and Mongolian Gobi withdistinctive glacial–interglacial source fluctuations (Sunet al., 2008). Based on U–Pb evidence, however, Pullenet al. (2011) suggested that the loess was largely derivedfrom the Qaidam Basin and the northern Tibetan Plateau,especially during glacial periods, in contrast to insignificantdifferences in zircon ages that had been observed forChinese loess (Che and Li, 2013). Thus more systematicevidence is needed in order to better understand whetheror not loess provenance fluctuated on glacial–interglacialtimescales.

The oxygen isotopic composition of quartz has beensuggested to be a good source tracer (Sridhar et al., 1975;Clayton et al., 1978; Aleon et al., 2002). Since the fraction-ation factor of oxygen isotopes is mainly controlled by tem-perature (Chacko et al., 2001), the d18O of quartz varieswidely in different types of host rocks according to their

formation temperature. For example, quartz d18O in igneousrocks varies between 6.4& and 12.5&, and is usually lessthan 10& (Taylor and Epstein, 1962; Taylor, 1968).Precipitated quartz, such as cherts and crystalline quartzformed at low temperature, has a markedly higher d18Orange (14.9–44&), with an average value of at least over20& (Syers et al., 1969; Clayton et al., 1972; Knauth andEpstein, 1976). In contrast, metamorphic rocks, usuallyformed from around 200 to 900 �C, are characterized by ad18O range decreasing from 20& to 10& according toincreasing temperature (Taylor and Coleman, 1968;Clayton et al., 1972). Without recrystallization, the oxygenisotopic composition of quartz remains stable during diage-netic and weathering processes, and thus retains informationabout its host rocks (Sridhar et al., 1975; Clayton et al.,1978).

Quartz is the most stable and abundant mineral inChinese loess (Liu, 1985), thus enabling the use of quartzd18O as a reliable source tracer for Asian dust (Gu et al.,1987; Ishii et al., 1995). Studies of three deserts and sandylands in northern China found that d18O signatures of quartzare unique in each desert or sandy land (Fu and Yang, 2004;Yang et al., 2008). Ishii et al. (1995) investigated the d18O offine-grained quartz in desert areas in Xinjiang and InnerMongolia provinces, and concluded that large scale homog-enization occurs. Gu et al. (1987) conducted an extensivestudy of quartz oxygen isotope ratios of Chinese loess forthe first time, demonstrating significant glacial–interglacialoscillations in the Luochuan loess sequence. Hou et al.(2003) completed an oxygen isotopic investigation of theLingtai red clay-loess section, revealing subtle variations ofthe d18O over the past 7 Ma. Unfortunately, the limitedavailability of quartz d18O data from the potential sourceareas, together with inconsistent isotopic results betweentwo loess sections, have greatly hampered the applicationof quartz d18O as a tracer for loess provenance.

In the present paper we present a systematic investigationof the oxygen isotopic composition of quartz from the majordust sources in East Asia. A total of 34 surface soil sampleswere collected from the three major source areas, i.e., theTaklimakan desert, the Mongolian Gobi, and the northernChinese deserts, and two additional sources, including theMu Us desert and Qaidam Basin, which were also suggestedas possible sources of Chinese loess deposits (Bowler et al.,1987; Ding et al., 1999; Pullen et al., 2011) (Fig. 1).The spa-tial variability of quartz oxygen isotope ratios both betweenand within these regions is investigated in order to assess itspotential for tracing the provenance of Asian dust. Our dataare also compared with previous quartz oxygen isotoperesults in order to evaluate the possible provenance varia-tion of loess on the CLP.

2. SAMPLING AND METHODOLOGY

2.1. Geographic setting and sampling

The major dust sources of East Asia are situated at geo-graphic lows within five inland drainage basins (Fig. 1),including the Mongolian Gobi, Tarim Basin, Alxa Plateau,Ordos Plateau, and Qaidam Basin. These drainage basins,

Fig. 1. Major dust sources in East Asia. Distribution of deserts and Gobi deserts comes from the 1:200,000 Desert Distribution Datasetprovided by the Environmental and Ecological Science Data Center for West China, National Natural Science Foundation of China (http://westdc.westgis.ac.cn). Distribution of the Mongolian Gobi desert is figurative. Boundaries of inland drainage basins in China and Mongoliaare adapted from Shi (2011) and Davaa et al. (2007), respectively. Loess distribution is adapted from Lu et al. (2008). Near-surface winddirections are adapted from Chen and Li (2011) and Wang et al. (2004). TK-Taklimakan desert, QB-Qaidam Basin, BJ-Badain Juran desert,TG-Tengger desert, MU-Mu Us desert, MG-Mongolian Gobi, and CLP-Chinese Loess Plateau.

Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410 401

located either in topographic lows or on flat tablelands, arethe inferred major dust sources in East Asia (Sun et al.,2001; Zhang et al., 2003; Wang et al., 2007; Bullard et al.,2011). The Taklimakan desert (TK) belongs to the Tariminterior drainage region (hereafter referred to as Taklima-kan). It has an average elevation of around 1000 m, startingfrom over 1000 m in the west and dropping to around800 m in the east; and is bound by mountains with averageheights around 4000–6000 m, including the TianshanMountains on the north, the West Kunlun Mountains onthe southwest, and the Altun Mountains on the southeast.Qaidam Basin (QB) belongs to the Qaidam Interior Region;its elevation averages around 2600–3000 m and dips east-wards. It is surrounded by the East Kunlun Mountainson the south, the Qilian Mountains on the north and theAltun Mountains on the northwest; their average elevationsexceed 4000 m. The Badain Juran (BJ) and Tengger (TG)deserts are situated inside the Hexi Corridor-Alxa InteriorRegion (Alxa Plateau), which is bound by theQilian Mountains (4000–6000 m) on the south, the HelanMountains (2000–2500 m) on the east, the GobiAltai Mountains (1500–2500 m) on the north, and theBeishan Mountains (around 2000 m) on the west. BJ(1200–1700 m) and TG (1200–1400 m) are separated bythe Yabulai Mountains (1600–1800 m). The Mu Us desert(MU) has an average elevation of around 1100–1300 m,and is located mainly in the Ordos Interior Region (OrdosPlateau). MU is surrounded by the slightly higher OrdosPlateau to the north and by the Loess Plateau to the south.The Mongolian Gobi (MG) belongs to the Asian InternalBasin of Mongolia (Mongolian Gobi). It has an average

elevation of over 1000 m, and is surrounded by the HenteynMountains (around 2000 m) to the north, the HangaynMountains (around 3000 m) to the northwest, the GobiAltai Mountains to the southwest, and the Inner Mongo-lian Plateau (1000–1400 m) to the southeast.

Sampling expeditions were conducted during 2007–2009. We took samples from the upper 30 cm of theground, most of which consist of mud crust or sandy soil.Further details can be found in Sun et al. (2013). In thisstudy, 34 samples were selected at roughly a 100–200 kminterval to cover each source region, including 10 samplesfrom the Taklimakan desert, 6 samples from the Qaidambasin, 4 samples from the Tengger desert, 4 samples fromthe Badain Juran desert, 7 samples from the MongolianGobi, and 3 samples from the Mu Us desert. Locationsof the 34 surface samples are presented in Fig. 1 andTable 1.

2.2. Methodology

The coarse fractions of dust are mainly transported bysaltation or rolling, while suspension is the primary trans-portation mode for the fine fraction (Pye, 1987; Zhanget al., 1999). Dust particles of less than 16 lm diameter con-stitute about 70% of the total atmospheric loading (Zhanget al., 2003), and can be transported over long distances insuspension (Rea, 1994; Prospero, 1999). Considering thepossible different transportation dynamics and transportranges of fine and coarse particles, we separated the bulksamples into two fractions, <16 and 16–63 lm, for furtheranalyses. Since the presence of authigenic quartz, either

Table 1Location and quartz d18O of fine and coarse fractions of selected samplesa.

Source area Sample ID Longitude (�) Latitude (�) Elevation (m) d18O (&) <16 lm d18O (&) 16–63 lm

Mongolian Gobi desert MG02 109.20 45.89 1062 15.8 14.3MG09 109.04 43.17 1128 16.2 12.6MG14 105.87 43.31 1349 15.9 15.1MG17 104.57 43.61 1390 17.3 13.8MG20 102.66 43.36 1436 17.1 14.3MG23 102.40 44.12 1215 16.7 14.1MG26 102.26 44.96 1341 15.1 13.7

Av. ± SD 16.3 ± 0.78 14.0 ± 0.76

Badain Juran desert BJ43 102.58 41.96 1113 16.7 14.9BJ47 100.54 41.80 970 16.6 15.2BJ52 99.39 40.89 1380 16.9 15.3BJ27 100.59 39.65 1505 16.8 14.8

Av. ± SD 16.8 ± 0.13 15.1 ± 0.24

Tengger desert TG12 104.46 37.46 1710 17.3 14.9TG15 103.06 37.76 1601 16.9 15.1TG18 103.57 39.01 1304 17.5 14.6TG22 103.02 39.54 1244 17.7 14.0

Av. ± SD 17.4 ± 0.34 14.7 ± 0.48Mu Us desert MU06 109.01 38.69 1347 16.1 15.1

MU01 108.44 37.52 1374 17.3 14.6MU03 107.29 37.84 1433 17.7 15.0

Av. ± SD 17.0 ± 0.83 14.9 ± 0.26

Qaidam basin QB08 97.57 37.16 2852 17.5 15.5QB17 95.45 37.09 2720 17.2 15.7QB21 95.52 36.37 2822 17.1 15.3QB23 96.64 36.35 2783 16.2 15.7QB25 97.65 36.03 3038 17.4 14.8QB30 98.46 36.44 3484 17.6 15.5

Av. ± SD 17.2 ± 0.51 15.4 ± 0.34

Taklimakan desert TK69 86.02 41.83 907 18.2 16.7TK73 83.48 41.91 1001 18.5 16.5TK77 80.98 41.42 1197 17.5 16.8TK83 78.29 39.93 1131 17.4 17.0TK92 77.58 37.84 1354 17.6 15.6TK96 79.53 37.14 1377 18.0 16.3TK103 82.27 36.80 1720 16.9 15.9TK108 84.04 37.72 1366 17.8 16.4TK111 85.76 38.48 1133 17.3 16.1TK116 87.51 38.76 981 16.8 15.6

Av. ± SD 17.6 ± 0.54 16.3 ± 0.49

a The average values (Av.) and standard deviations (SD) of d18O of each source are also provided.

402 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

precipitated or converted during illitization in surface sedi-ments, would result in the increased d18O of the very finequartz fraction (Blatt, 1987; Hou et al., 2003), the clay frac-tion (<4 lm) of two additional desert surface samples (SS1and SS2) was removed from the fine fraction (<16 lm) inorder to evaluate this effect. The >63 lm fraction of thesamples was removed by wet-sieving, while the fine fraction(<16 lm) was separated using the pipette method based onStokes’ Law (Lerman et al., 1974). The residue, the fractionfrom 16 to 63 lm, is regarded as the coarse fraction.

Quartz was extracted from the two separated fractionsusing the K2S2O7 fusing – H2SiF6 soaking method (Xiaoet al., 1995; Sun et al., 2000; Hou et al., 2003). X-ray diffrac-tion measurements of the pretreated samples indicate thatthe purity of the extracted quartz is around 98%, whichguarantees precise measurement of quartz oxygen isotope

ratios. The extracted quartz was reacted with BrF5 at550 �C to liberate O2. Oxygen was converted to CO2 byreacting with a graphite rod at 700 �C. The oxygen isotopecomposition of the CO2 was then measured using a Finni-gan Mat 251 Mass Spectrometer with an analytical preci-sion of 0.2&, in the Institute of Mineral Resources,Chinese Academy of Geological Sciences. The results arereported using the usual d notation as per mil deviationsfrom Standard Mean Ocean Water (SMOW):

d18OSA�SMOW ¼ ðd18OSA�R þ 103Þ � ðd18OST�SMOW

þ 103Þ=ðd18OST�R þ 103Þ � 103

where d18OSA–SMOW is the d18O of the sample relative toSMOW, d18OSA–R is the d18O of the sample relative tothe referential CO2, d18OST–SMOW is the d18O of the

Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410 403

standard sample relative to SMOW, and d18OST–R is thed18O of the standard sample relative to the referential CO2.Measurements of standard samples indicated that the stan-dard deviation of the measurements is 0.07& (n = 12).

3. RESULTS

The d18O values of the fine-grained (<16 lm) quartz are15.5& and 16.7& for SS1 and SS2, respectively; and thosefor the 4–16 lm fraction of SS1 and SS2 are 15.7& and16.6&, respectively. Removal of the <4 lm fraction fromthe <16 lm fraction of SS1 and SS2 yielded differences of0.2& and 0.1&, respectively, which are within the analyti-cal precision of the instrument. This suggests that no authi-genic quartz formed in the arid source areas and thereforethat the d18O composition of fine-grained quartz can beregarded as a reliable tracer for the fine dust deposited onthe land and in the ocean.

The range of d18O values in the fine fraction is the great-est in the Mongolian Gobi (15.1–17.3&); the lowest in theBadain Juran desert (16.6–16.9&); and moderate in theTaklimakan desert (16.8–18.5&), the Mu Us desert (16.1–17.7&), Qaidam Basin (16.2–17.6&), and the Tengger des-ert (16.9–17.7&) (Table 1 and Fig. 2). Coarse-grainedquartz has systematically lower d18O values than that offine-grained quartz, which is consistent with former studies(Clayton et al., 1972; Hou et al., 2003; Fu and Yang, 2004).The range of d18O values in the coarse fraction is also thelargest in the Mongolian Gobi (12.6–15.1&), and the low-est in both the Badain Juran desert (14.8–15.3&) and theMu Us desert (14.6–15.1&). The spatial variability ismoderate in the Taklimakan desert (15.6–17.0&), in theTengger desert (14.0–15.1&), and in the Qaidam Basin(14.8–15.7&).

The coarse and fine quartz in the Mongolian Gobiclearly has lower average d18O values than those of theother regions (14.0& and 16.3&, respectively). The averaged18O of coarse quartz in the Taklimakan desert (16.3&) ismarkedly higher than that in the other regions, while theaverage d18O of fine quartz (17.6&) is slightly higher thanthat of the other regions. The average d18O values of coarseand fine quartz in the Mu Us (14.9& and 17.0& respec-tively), Tengger (14.7& and 17.4&), Badain Juran

Fig. 2. Quartz d18O of fine and coarse fractions in the dust sourceareas. Short bars indicate standard deviations, and long bars arethe average values.

(15.1& and 16.8&), and Qaidam (15.4& and 17.2&), arevery close and intermediate between those of the Taklima-kan desert and Mongolian Gobi.

4. DISCUSSION

4.1. Spatial characteristics of quartz d18O among the major

dust sources

The oxygen isotopic composition of fine quartz of theinvestigated sources varies from 15.1& to 18.5&, while thatof coarse quartz is systematically lower, varying from12.6& to 17&. Consistent with previous studies (Claytonet al., 1972, 1978; Aleon et al., 2002), quartz d18O valuesof both fractions are intermediate between those of rocksformed at low and high temperatures, suggesting a mixedorigin of the detritus from both low- and high-temperaturehost rocks. More low-temperature quartz in the fine frac-tion is indicated by the systematically higher values of thisfraction (Clayton et al., 1972, 1978; Aleon et al., 2002).Quartz d18O values of both fractions show roughly thesame increasing trend from the Mongolian Gobi, to thenorthern Chinese deserts and Qaidam Basin, and then tothe Taklimakan desert (Fig. 2). Previous studies of surfacesoil and sand samples from Xinjiang, Inner Mongolia, theBadain Juran desert, the Hulun Buir and Otingdag sandylands, revealed a similarly increasing trend of quartz d18Ofrom the north to the south (Ishii et al., 1995; Fu andYang, 2004; Yang et al., 2008). The spatial variation ofthe quartz d18O suggests that the Mongolian Gobi is situ-ated in a geological setting with more high-temperaturerocks, while the Taklimakan desert lies in a geologicalsetting of lower-temperature rocks. All of the sources aresituated within different inland drainage basins (Fig. 3),where the quartz detritus should be derived from the sur-rounding high-altitude terrains by weathering and grindingprocesses (Sun, 2002).

The quartz d18O characteristic of each source should bedetermined by the petrology of the surrounding mountain-ous areas within the drainage basins. Since quartz detritusis mostly derived from igneous and metamorphic rocks(Blatt, 1987), the variable d18O values are likely to resultfrom the mixing ratio of igneous and metamorphic rocks.In the five inland drainage basins, aside from igneous rocks,mountainous areas are dominated by rock formationsmetamorphosed to varying degrees (Zhang and Wu, 2002;Badarch, 2005; Institute of Geology, 2006). The proportionof quartz from igneous rocks is therefore a majorcontrolling factor of the d18O of the derived sediments.We calculated the coverage ratios of igneous rocks for themountainous areas in each inland drainage basin and theresults are presented in Fig. 3 and Table 2. To obtain cov-erage ratios, we used the “Reclass” function of ArcMap(software developed by the Environment System ResearchInstitute, Inc.) to group areas of different elevation range.Based on the topographic map of the People’s Republicof China (Zhou and Shi, 2003), the lowest and most wide-spread elevation ranges are the lower flatlands of the basins.For each individual basin, the area of mountainous terrainwas estimated by subtracting the area of the lower flatlands

Fig. 3. Distribution of igneous rock formations in different inland drainage basins. Boundaries of igneous rock formations are adapted fromthe 1:2,500,000 Geological map of western China and adjacent areas (Institute of Geology, 2006). The gray levels correspond to the elevations.

Table 2Coverage of igneous rock formations in different source areas.

Interior region Mountainous regions (m2) Igneous rock formations (m2) Coverage ratio (%)

Mongolian Gobi 5.37E+11 1.48E+11 27.6Taklimakan 4.07E+11 5.08E+10 12.5Alxa Plateau 2.56E+11 4.79E+10 18.7Qaidam Basin 2.12E+11 3.84E+10 18.1

404 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

from that of the entire basin, and subsequently the coverageratio was calculated by dividing the area of igneous rockformation by the area of mountainous terrain.

Clearly, the Mongolian Gobi has the largest percentageof igneous rock coverage (27.6%), while the Taklimakanhas the lowest coverage of around 12.5%. The QaidamBasin has a rather similar percentage to the Alxa Plateau(18.1–18.7%), since the Qilian Mountains probably contrib-ute to most of the deposited sediments in these two basins.The Ordos Plateau, in contrast, has no igneous rock forma-tion. The sediments inside this region are mainly sourcedfrom surrounding/underlying Jurassic to Cretaceous sand-stones and from the bank of the Yellow River (Sun et al.,2007). This distribution pattern of igneous rock coverageof these basins provides a reasonable interpretation forthe trend of spatially decreasing d18O values from the Mon-golian Gobi, to the northern Chinese deserts, and then tothe Taklimakan desert. Notably, the close location andoverlapping value ranges of quartz d18O in the Mu Us des-ert compared to the Badain Juran and Tengger deserts sug-gest that the sediments in the Ordos Plateau might have aprovenance linkage to the Alxa Plateau or to the YellowRiver, as suggested by Che and Li (2013) and Stevenset al. (2013). Consequently the Mu Us desert is hereaftergrouped with the northern Chinese deserts based on thequartz oxygen isotope ratios.

4.2. Differentiating major dust sources

Previous studies have proposed several proxies for dif-ferentiating various dust sources in East Asia. Using ele-mental ratios including Fe/Al, Mg/Al, and Sc/Al, Zhanget al. (1997) were able to distinguish three regions, includingthe Taklimakan desert, the Tengger and Badain Jurandeserts, and the Mu Us desert. Utilizing Nd–Sr isotope ofsilicates, Chen et al. (2007) were able to differentiate theTaklimakan desert and the Mu Us desert from the mixedsignals of the Qaidam Basin, the Tengger desert, and theBadain Juran desert. Note that the Mongolian Gobi wasalso not included in these provenance studies. While U–Pb age spectra of zircon grains show distinct patterns in dif-ferent sources, overlapping of the age spectra still occurs(Stevens et al., 2010; Pullen et al., 2011). Although fine-grained dust from the major Asian dust sources can beeffectively distinguished using ESR–CI results, this tracingapproach does not seem to be effective in the case ofcoarse-grained quartz (Sun et al., 2013). Overall, previousstudies have not yet reached a consensus on loess prove-nance variations utilizing these proxies, either because theyhave focused on proxies derived from different grain sizefractions (e.g., fractions of <16 and 16–63 lm in Sunet al. (2008), <5 and <75 lm in Chen et al. (2007),coarse-grained zircon in Stevens et al. (2010) and Pullen

Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410 405

et al. (2011)), and/or they did not consider all of the poten-tial dust sources. Our study aims to provide an additionalprovenance tracer (quartz d18O) that encompasses bothfine- and coarse-grained dust from the major dust sourcesin East Asia; and in addition to shed new light on the prov-enance study of Asian dust, especially in combination withother source tracers.

d18O variability in both fractions within each source islarger than the analytical precision of the instrument(0.2&), implying that these sources are somewhat heteroge-neous (Fig. 2). The overall higher ranges in the MongolianGobi and Taklimakan desert suggest that these sources aremore heterogeneous compared with the others, which isconsistent with the heterogeneous nature inferred fromquartz ESR–CI results (Sun et al., 2013). The large variabil-ity of the quartz d18O values in both fine and coarse frac-tions from each desert makes it difficult to distinguishthese source areas spatially. For example, the d18O valuesof fine quartz are largely overlapped among these sources,given the average values of the three major sources areslightly different. For coarse quartz, the Taklimakan desertand Mongolian Gobi have negligible overlap with the othersources, and thus can be effectively distinguished. Combin-ing the d18O of the two fractions, however, permits a betterdifferentiation of these sources (Fig. 4). The MongolianGobi can be easily differentiated from the Taklimakan des-ert by the relatively higher values in both size fractions. Thenorthern Chinese deserts can also be distinguished as anintermediate group. The Qaidam Basin, however, is isotopi-cally located in between the Taklimakan desert and thenorthern Chinese deserts.

This spatial pattern of quartz d18O suggests a possibleprovenance linkage of the Qaidam Basin to the Taklimakandesert and the northern Chinese deserts to the MongolianGobi. This is a plausible conclusion because two adjacentsources may share the same watersheds, and dust originat-ing from the Taklimakan desert could be transported to theQaidam Basin. However, since coarse dust is mainly trans-ported by near-surface winds (Pye, 1987; Prins et al., 2007),it is impossible for the coarse dust from the Taklimakandesert to be transported eastwards to the Qaidam Basindue to its predominant southwesterly near-surface wind

Fig. 4. Bivariate plot of quartz d18O of the fine fraction versus thatof the coarse fraction. NCD stands for the northern Chinesedeserts.

trajectory (Fig. 1). Similarly, fine dust from the Taklimakandesert cannot be readily deposited in proximal regions (e.g.,Qaidam Basin) because it is usually entrained to an eleva-tion of 5000 m and transported by the westerly jet stream(Sun, 2002).

Although Asian dust dominantly originates from threemajor sources in East Asia (e.g., Sun et al., 2001; Zhanget al., 2003), our quartz d18O results cannot effectively dis-tinguish these three sources. The d18O results can differenti-ate statistically two major sources (i.e., MG and TK),whereas the d18O values of NCD samples fall in betweenthose of TK and MG. Differentiating the dust contributionof the northern Chinese deserts from a mixed contributionof Taklimakan desert and Mongolian Gobi desert dust isimpossible based on d18O values alone. Climatologically,dust from the Taklimakan desert is unlikely to be depositedin the northern Chinese deserts. However, dust from theMongolian Gobi can be transported to the northern Chi-nese deserts by northwesterly surface winds. Consequently,we now compare the quartz d18O values with the ESR–CIresults (Sun et al., 2013). Plotting quartz d18O versus ESRand CI achieves a clearer separation of the Taklimakan des-ert and Mongolian Gobi, particularly in the case of the finefraction (Fig. 5). The CI values of fine-grained quartz in thenorthern Chinese deserts are higher than those of both theTaklimakan desert and Mongolian Gobi. This excludes thepossibility of dust from the northern Chinese deserts com-prising a mixture of dust from the Mongolian Gobi andTaklimakan desert, and also rules out the possibility of con-fusing a dust contribution of the northern Chinese desertswith a mixed dust contribution of the Taklimakan desertand the Mongolian Gobi. In the case of the coarse fraction,the northern Chinese deserts exhibit characteristics similarto the Mongolian Gobi, but are significantly different fromthe Taklimakan desert.

4.3. Glacial–interglacial provenance variations within the

loess-paleosol sequence

Comparison of the quartz d18O values of desert surfacesediments and loess samples can provide new insights intochanges in loess provenance. Two loess sections have beenstudied using quartz d18O (Gu et al., 1987; Hou et al.,2003), but unfortunately the d18O results are inconsistentand the interpretations are rather controversial. Gu et al.(1987) first reported glacial–interglacial variations of thed18O of fine quartz (2–10 lm) in the Luochuan section dur-ing the past 0.8 Ma (Fig. 6), and interpreted the results interms of monsoonal variation associated with the influenceof meteoric water during pedogenesis. In contrast, Houet al. (2003) investigated the d18O variation of fine quartz(4–16 lm) in the Lingtai red clay-loess section, and sug-gested that it reflects major changes in the dust source areasover the past 7 Ma.

To examine the loess provenance oscillations, we plottedthe average quartz d18O of the three major sources (i.e.,17.6& in the Taklimakan desert, 16.3& in the MongolianGobi, and 17% in the northern Chinese deserts) againstquartz d18O variations in the Luochuan and Lingtaisections (Fig. 6). It is clear that quartz d18O of the Lingtai

Fig. 5. Quartz d18O versus ESR and CI of both fine and coarse fractions. Data of quartz ESR–CI are from Sun et al. (2013).

406 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

section appears to have a large systematic deviation bothfrom that of the sources and from the Luochuan section.In addition, although the average d18O of the Lingtai sec-tion is higher in paleosols (19.4&) than that in loess(19.0&), d18O does not show consistent variations betweenglacial loess and interglacial paleosols (lower values in L1,S2, S3, and L9, and higher ones in S1, L3, and S5). Thusthe quartz d18O values of the Lingtai section cannot becompared either with those of the Luochuan section or withthose of the sources, and thus are excluded from furtherdiscussion. The reasons for this discrepancy clearly needfurther investigation. In contrast, the d18O values of thesource areas are rather comparable with those of theLuochuan loess sequence.

In the Luochuan section, quartz d18O of loess layers hasa mean value of 17.4& with a range of 17.2–17.75&, whilethat of paleosols is consistently higher, averaging 17.8&

with a range of 17.6–18.1&. The relatively high d18O valuesin the paleosol layers were previously attributed to the for-mation of authigenic quartz under increased temperatureand d18O of precipitation by Gu et al. (1987). However,quartz d18O values of the <2 and 2–10 lm fractions of fiveloess samples on the CLP exhibit an average difference of0.26& (Gu et al., 1987). In particular, in the Malan loessof Luochuan, the quartz d18O values of the <2 and2–10 lm fractions are 17.1& and 17.0&, respectively, indi-cating no detectable effect of authigenic quartz on the d18Oresults. Meanwhile scanning electronic micrographs of thequartz in Chinese loess reveal that the authigenic quartzcomponent is rather limited (Xiao et al., 1995) and is almostentirely removed during chemical pretreatment (Sun et al.,2000). Meanwhile, the glacial-interglacial fluctuations may

also be attributable to grain-size sorting effects, since thed18O values are usually higher in the fine fraction comparedto the coarse fractions (Gu et al., 1987; Hou et al., 2003). Itis noteworthy that the quartz d18O differences between the<2, 2–10 and 10–20 lm fractions are between 0.24& and0.36& (Gu et al., 1987), which is lower than the �0.9&

difference between loess and paleosol layers (Fig. 6).Therefore, the glacial–interglacial oscillations of quartzd18O of the 2–10 lm fraction in the Luochuan sectioncannot be attributed to grain-size sorting effects.

Since the quartz d18O of well-mixed fine dust has agreater statistical probability of approaching the averagevalue of its source, by the comparison of the d18O of fine-grained quartz in the Luochuan loess section with theaverage d18O values of the three major sources (Fig. 6),we tentatively conclude that the dust sources likely shiftedfrom the northern Chinese deserts and/or Mongolian Gobiduring glacial periods to the Taklimakan desert in intergla-cials, which is contrary to previous inference of loess prov-enance changes (Sun, 2002, 2005; Sun et al., 2008). It isnoteworthy that the 2–10 lm quartz fraction used in thestudy of the Luochuan section is finer than the fine fractionused in the present study (<16 lm), suggesting that quartzd18O of the <16 lm fraction in the section should be sys-tematically lower than that of 2–10 lm fraction. Assumingthat the Luochuan d18O results reflect provenance shiftsbetween the Mongolian Gobi and the northern Chinesedeserts, to be consistent with our previous work (Sunet al., 2008), then an offset of �0.8& should be considered(Fig. 6). This possible offset, however, needs to beconstrained in the future by measuring the d18O of the samesize fraction (<16 lm) for loess and paleosol samples. This

Fig. 6. Quartz d18O variations in the Luochuan (2–10 lm) and Lingtai sections (4–16 lm) during the last 0.8 Ma, compared with that of the<16 lm of the three major dust sources. Stratigraphy and the corresponding age of the Luochuan section are modified from Gu et al. (1987).Pink and purple curves correspond to data from the Luochuan section (Gu et al., 1987) and Lingtai section (Hou et al., 2003), respectively.Dashed line denotes the d18O data of the Luochuan section offset by �0.8& (see text for explanation). Averaged d18O values of fine quartz(<16 lm) of the three major sources are indicated by the solid straight lines. (For interpretation of the references to colour in this figure legend,the reader is referred to the web version of this article.)

Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410 407

may also help to resolve the issue of the lack of comparabil-ity between the quartz d18O variations of the Luochuan andLingtai sections.

Coarse-grained dust can be transported by differentpathways compared to fine-grained dust (Sun et al., 2003;Prins et al., 2007). Currently we have no availablecoarse-quartz d18O datasets from loess sections for evalua-tion of possible provenance shifts of coarse-grained dust.However, we have demonstrated that by utilizing pairedquartz d18O measurements of fine and coarse fractions,and/or by combining quartz d18O with ESR–CI results ofdifferent size fractions, dust originating from the threemajor sources can be effectively distinguished. We suggestthat in the future combining quartz d18O of both fineand coarse fractions, and other source tracers, such asquartz ESR–CI (Sun et al., 2013), Nd–Sr isotopes (Chenet al., 2007), U–Pb age spectrum of zircon (Pullen et al.,2011; Che and Li, 2013), will provide an improved under-standing of loess provenance fluctuations at tectonic andorbital timescales.

5. CONCLUSION

We systematically studied the oxygen isotopic composi-tion of both fine- and coarse-grained quartz from major

dust sources in East Asia. The d18O results of both fractionsexhibit a consistently increasing trend from the MongolianGobi to the northern Chinese deserts and Qaidam Basin,and then to the Taklimakan desert. The spatial pattern ofd18O variability suggests that the geological setting of theMongolian Gobi desert comprises the greatest proportionof high-temperature rocks compared to the other areas,while the Taklimakan desert has the least. This inferenceis confirmed by spatial changes in the coverage ratios ofigneous rock formations which decrease from the Mongo-lian Gobi, to the northern Chinese deserts and QaidamBasin, and finally to the Taklimakan desert.

d18O variability within each desert indicates that the sur-face sediments are heterogeneous to a certain degree, espe-cially in the case of the Mongolian Gobi and Taklimakandesert. Spatially, the quartz d18O values of both fine andcoarse fractions overlap to varying degrees, making it diffi-cult to discriminate the dust from individual sources. How-ever, bivariate plots of quartz d18O of both fractions candifferentiate the three major source areas, i.e. the Mongo-lian Gobi, the Taklimakan desert, and the northern Chinesedeserts. In addition, the combination of quartz d18O mea-surements with measurements of ESR and CI can provideunequivocal discrimination of fine-grained dust from thethree different sources.

408 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

Comparison of our d18O results with those from loesssections reveals that quartz d18O variations of fine dustare probably related to glacial–interglacial changes in dustprovenance. However, due to the large discrepancy betweenthe existing d18O datasets, it is difficult to ascertain reliablyhow the provenance of loess varies on glacial–interglacialtimescales. In order to resolve current disagreements con-cerning variations in loess provenance, we intend to inves-tigate the quartz d18O of both fine and coarse fractions,and then to combine the results with other proxies gener-ated from the Chinese loess-red clay sequences.

ACKNOWLEDGMENTS

We thank Dominik Weiss and three anonymous reviewers fortheir valuable comments, and Defang Wan for oxygen isotope mea-surements. Jan Bloemendal and Susan Clemens are thanked forEnglish editing. This work was supported by Key Innovation Pro-ject (KZCX-EW-114) and International Partnership Program(KZZD-EW-TZ-03) from the Chinese Academy of Sciences, andfunding from the Natural Science Foundation of China (No.41072272).

REFERENCES

Aleon J., Chaussidon M., Marty B., Schutz L. and Jaenicke R.(2002) Oxygen isotopes in single micrometer-sized quartzgrains: tracing the source of Saharan dust over long-distanceatmospheric transport. Geochim. Cosmochim. Acta 66, 3351–3365.

An Z. S. (2000) The history and variability of the East Asianpaleomonsoon climate. Quat. Sci. Rev. 19, 171–187.

An Z. S., Kukla G. J., Porter S. C. and Xiao J. L. (1991) Magneticsusceptibility evidence of monsoon variation on the LoessPlateau of central China during the last 130,000 years. Quat.

Res. 36, 29–36.An Z. S., Kutzbach J. E., Prell W. L. and Porter S. C. (2001)

Evolution of Asian monsoons and phased uplift of theHimalaya–Tibetan Plateau since Late Miocene time. Nature

411, 62–66.Badarch G. (2005) Tectonic overview of Mongolia. Mongolian

Geoscientist 27, 1–7.Blatt H. (1987) Perspectives: oxygen isotopes and the origin of

quartz. J. Sediment. Petrol. 57, 373–377.Bowler J. M., Chen K. Z. and Yuan B. Y. (1987) Systematic

variations in loess source areas: evidence from Qaidam andQinghai basins, western China. In Aspects of Loess Research

(ed. T. S. Liu). China Ocean Press, pp. 39–51.Bullard J. E., Harrison S. P., Baddock M. C., Drake N., Gill T. E.,

McTainsh G. and Sun Y. (2011) Preferential dust sources: ageomorphological classification designed for use in global dust-cycle models. J. Geophys. Res. Earth Surf. (2003–2012)

116(F4).Chacko T., Cole D. R. and Horita J. (2001) Equilibrium oxygen,

hydrogen and carbon isotope fractionation factors applicable togeologic systems. Rev. Mineral. Geochem. 43, 1–81.

Che X. D. and Li G. J. (2013) Binary sources of loess on theChinese Loess Plateau revealed by U–Pb ages of zircon. Quat.

Res. 80, 545–551.Chen J. and Li G. J. (2011) Geochemical studies on the source

region of Asian dust. Sci. China Earth Sci. 54, 1279–1301.Chen Z. and Li G. J. (2013) Evolving sources of eolian detritus on

the Chinese Loess Plateau since early Miocene: tectonic andclimatic controls. Earth Planet. Sci. Lett. 371–372, 220–225.

Chen J., Li G. J., Yang J. D., Rao W. B., Lu H. Y., Balsam W.,Sun Y. B. and Ji J. F. (2007) Nd and Sr isotopic characteristicsof Chinese deserts: implications for the provenances of Asiandust. Geochim. Cosmochim. Acta 71, 3904–3914.

Clayton R. N., Rex R. W., Syers J. K. and Jackson M. L. (1972)Oxygen isotopic abundance of quartz in Pacific pelagicsediments. J. Geophys. Res. 77, 3907–3915.

Clayton R. N., Jackson M. L. and Sridhar K. (1978) Resistance ofquartz silt to isotopic exchange under burial and intenseweathering conditions. Geochim. Cosmochim. Acta 42, 1517–1522.

Davaa G., Oyunbaatar D. and Sugita M. (2007) Surface Water ofMongolia. Environmental Book of Mongolia 2006, 55–82

Derbyshire E., Meng X. M. and Kemp R. A. (1998) Provenance,transport and characteristics of modern aeolian dust in westernGansu province, China, and interpretation of the quaternaryloess record. J. Arid Environ. 39, 497–516.

Ding Z. L., Liu T. S., Rutter R. W., Yu Z. W., Guo Z. T. andZhu R. X. (1995) Ice-volume forcing of east Asian wintermonsoon variations in the past 800,000 years. Quat. Res. 44,149–159.

Ding Z. L., Sun J. M., Rutter N. W., Rokosh D. and Liu T. S.(1999) Changes in sand content of loess deposits along a north-south transect of the Chinese Loess Plateau and the implica-tions for desert variations. Quat. Res. 52, 56–62.

Ferrat M., Weiss D. J., Strekopytov S., Dong S., Chen H., NajorkaJ., Sun Y., Gupta S., Tada R. and Sinha R. (2011) Improvedprovenance tracing of Asian dust sources using rare earthelements and selected trace elements for palaeomonsoon studieson the eastern Tibetan Plateau. Geochim. Cosmochim. Acta 75,6374–6399.

Fu X. D. and Yang X. P. (2004) Preliminary measurement andanalysis of quartz d18O of northern deserts in China. Quat. Sci.

24, 243 (In Chinese).Gallet S., Jahn B. and Torii M. (1996) Geochemical characteriza-

tion of the Luochuan loess paleosol sequence, China, andpaleoclimatic implications. Chem. Geol. 133, 67–88.

Gu Z. Y., Liu T. S. and Zheng S. H. (1987) A preliminary study onquartz oxygen isotope in Chinese loess and soils. In Aspects of

Loess Research (ed. T. S. Liu). China Ocean Press, pp. 291–301.Gu Z. Y., Ding Z. L., Xiong S. F. and Liu T. S. (1999) A seven

million geochemical record from Chinese red-clay and loess-paleosol sequence: weathering and erosion in northwesternChina. Quat. Sci. 4, 357–365 (In Chinese).

Guo Z. T., Ruddiman W. F., Hao Q. Z., Wu H. B., Qiao Y. S.,Zhu R. X., Peng S. Z., Wei J. J., Yuan B. Y. and Liu T. S.(2002) Onset of Asian desertification by 22 Myr ago inferredfrom loess deposits in China. Nature 416, 159–163.

Hao Q. Z., Guo Z. T., Qiao Y. S., Xu B. and Oldfield F. (2010)Geochemical evidence for the provenance of middle Pleistoceneloess deposits in southern China. Quat. Sci. Rev. 29, 3317–3326.

Hou S. S., Yang S. L., Sun J. M. and Ding Z. L. (2003) Oxygenisotope compositions of quartz grains (4–16 mm) from Chineseeolian deposits and their implications for provenance. Sci.

China (Ser. D) 46, 1003–1011.Institute of Geology, Chinese Academy of Geological Sciences

(2006) Geological Map of Western China and Adjacent Areas.Geological Publishing House (In Chinese).

Ishii T., Isobe I., Mizuno K., Kanai Y., Matsuhisa Y., Mizota C.,Qian Y. B., Terashima S. and Okumura K. (1995) Study of theformation processes and sedimentary environment of surfacegeological features in desertic areas of China, with specialreference to the characteristics and origin of eolian sediments.Bull. Geol. Surv. Jpn. 46, 651–685.

Jahn B. M., Gallet S. and Han J. M. (2001) Geochemistry of theXining, Xifeng and Jixian sections, Loess Plateau of China:

Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410 409

eolian dust provenance and paleosol evolution during the last140 ka. Chem. Geol. 178, 71–94.

Knauth L. P. and Epstein S. (1976) Hydrogen and oxygen isotoperatios in nodular and bedded cherts. Geochim. Cosmochim. Acta

40, 1095–1108.Lerman A., Lal D. and Dacey M. F. (1974) Stokes’ settling and

chemical reactivity of suspended particles in natural waters. InSuspended Solids in Water (ed. R. J. Gibbs). Plenum Press, NewYork, pp. 17–47.

Li G. J., Chen J., Chen Y., Yang J. D., Ji J. F. and Liu L. (2007)Dolomite as a tracer for the source regions of Asian dust. J.

Geophys. Res. 112, D17201. http://dx.doi.org/10.1029/2007jd008676.

Li G. J., Chen J., Ji J. F., Yang J. D. and Conway T. M. (2009)Natural and anthropogenic sources of East Asian dust. Geology

37, 727–730. http://dx.doi.org/10.1130/G30031A.1.Li G. J., Pettke T. and Chen J. (2011) Increasing Nd isotopic ratio

of Asian dust indicates progressive uplift of the north TibetanPlateau since the middle Miocene. Geology 39, 199–202.

Li G. J., Chen J. and Chen Y. (2013) Primary and secondarycarbonate in Chinese loess discriminated by trace elementcomposition. Geochim. Cosmochim. Acta 103, 26–35.

Liu T. S. (1985) Loess and the Environment. China Ocean Press, pp.189. (In Chinese).

Liu T. S. and Ding Z. L. (1998) Chinese loess and the paleomon-soon. Annu. Rev. Earth Planet. Sci. 26, 111–145.

Lu H. Y., Wang X. Y. and Li L. P. (2008) Aeolian dust recordsindicate the linkage of global cooling and Asian drying in latecenozoic. Quat. Sci. 28, 949–956 (In Chinese).

Middleton N. J., Goudie A. S. and Wells G. L. (1986) Thefrequency and source areas of dust storms. In Aeolian

Geomorphology (ed. W. G. Nickling). Allen & Unwin, Boston,pp. 237–259.

Prins M. A., Vriend M., Nugteren G., Vandenberghe J., Lu H.,Zheng H. and Jan Weltje. G. (2007) Late quaternary aeoliandust input variability on the Chinese Loess Plateau: inferencesfrom unmixing of loess grain-size records. Quat. Sci. Rev. 26,230–242.

Prospero J. M. (1999) Long-range transport of mineral dust in theglobal atmosphere: impact of African dust on the environmentof the southeastern United States. Proc. Natl. Acad. Sci. 96,3396–3403.

Prospero J. M., Ginoux P., Torres O., Nicholson S. E. and Gill T. E.(2002) Environmental characterization of global sources ofatmospheric soil dust identified with the Nimbus 7 Total OzoneMapping Spectrometer (TOMS) absorbing aerosol product. Rev.

Geophys. 40(1002). http://dx.doi.org/10.1029/2000 RG000095.Pullen A., Kapp P., McCallister A. T., Chang H., Gehrels G. E.,

Garzione C. N., Heermance R. V. and Ding L. (2011) QaidamBasin and northern Tibetan Plateau as dust sources for theChinese Loess Plateau and paleoclimatic implications. Geology

39, 1031–1034.Pye K. (1987) Aeolian Dust and Dust Deposits. Academic Press,

London, 68.Qiang X., An Z., Song Y., Chang H., Sun Y., Liu W., Ao H., Dong

J., Fu C., Wu F., Lu F., Cai Y., Zhou W., Cao J., Xu X. and AiL. (2011) New eolian red clay sequence on the western ChineseLoess Plateau linked to onset of Asian desertification about25 Ma ago. Sci. China Earth Sci. 54, 136–144.

Rea D. K. (1994) The Paleoclimatic record provided by eoliandeposition in the deep-sea: the geologic history of wind. Rev.

Geophys. 32, 159–195.Shi P. J. (2011) Atlas of Natural Disaster Risk of China. Science

Press, pp. 244.Sridhar K., Jackson M. L. and Clayton R. N. (1975) Quartz

oxygen isotopic stability in relation to isolation from sediments

and diversity of source. Soil Sci. Soc. Am. Proc. 39, 1209–1213.

Stevens T., Palk C., Carter A., Lu H. and Clift P. (2010) Assessingthe provenance of loess and desert sediments in northern Chinausing U–Pb dating and morphology of detrital zircons. GSA

Bull. 122, 1331–1344.Stevens T., Carter A., Watson T. P., Vermeesch P., Ando S., Bird

A. F., Lu H., Garzanti E., Cottam M. A. and Sevastjanova I.(2013) Genetic linkage between the Yellow River, the Mu Usdesert and the Chinese Loess Plateau. Quat. Sci. Rev. 78, 355–368.

Sun J. M. (2002) Provenance of loess material and formation ofloess deposits on the Chinese Loess Plateau. Earth Planet. Sci.

Lett. 203, 845–859.Sun J. M. (2005) Nd and Sr isotopic variations in Chinese eolian

deposits during the past 8 Ma: implications for provenancechange. Earth Planet. Sci. Lett. 240, 454–466.

Sun Y. B. and An Z. S. (2002) History and variability of the Asianinterior aridity recorded by eolian flux in the Chinese LoessPlateau during the past 7 million years. Sci. China (Ser. D) 45,420–429.

Sun Y. B., Lu H. Y. and An Z. S. (2000) Grain size distribution ofquartz isolated from Chinese loess/paleosol. Chin. Sci. Bull. 45,2296–2298.

Sun J. M., Zhang M. Y. and Liu T. S. (2001) Spatial and temporalcharacteristics of dust storms in China and its surroundingregions, 1960–1999: relations to source area and climate. J.

Geophys. Res. 106, 18325–18333.Sun D., An Z., Su R., Lu H. and Sun Y. (2003) Eolian sedimentary

records for the evolution of monsoon and westerly circulationsof northern China in the last 2.6 Ma. Sci. China Ser. D Earth

Sci. 46, 1049–1059.Sun Y. B., Tada R., Chen J., Chen H. Z., Toyoda S., Tani A.,

Isozaki Y., Nagashima K., Hasegawa H. and Ji J. F. (2007)Distinguishing the sources of Asian dust based on electron spinresonance signal intensity and crystallinity of quartz. Atmos.

Environ. 41, 8537–8548.Sun Y., Tada R., Chen J., Liu Q., Toyoda S., Tani A., Ji J. and

Isozaki Y. (2008) Tracing the provenance of fine-grained dustdeposited on the central Chinese Loess Plateau. Geophys. Res.

Lett. 35. http://dx.doi.org/10.1029/2007GL031672L01804.Sun Y., Clemens S. C., Morrill C., Lin X., Wang X. and An Z.

(2012) Influence of Atlantic meridional overturning circulationon the East Asian winter monsoon. Nat. Geosci. 5, 46–49.

Sun Y., Chen H., Tada R., Weiss D., Lin M., Toyoda S., Yan Y.and Isozaki Y. (2013) ESR signal intensity and crystallinity ofquartz from Gobi and sandy deserts in East Asia andimplication for tracing Asian dust provenance. Geochem.

Geophys. Geosyst. 14. http://dx.doi.org/10.1002/ggge.20162.Syers J. K., Jackson M. L., Berkheiser V. E., Clayton R. N. and

Rex R. W. (1969) Eolian sediment influence on pedogenesisduring the Quternary. Soil Sci. 107, 421–427.

Taylor H. P. (1968) The oxygen isotope geochemistry of igneousrocks. Contrib. Mineral. Petrol. 19, 1–71.

Taylor H. P. and Coleman R. G. (1968) O18/O16 ratios ofcoexisting minerals in glaucophane-bearing metamorphic rocks.Geol. Soc. Am. Bull. 79, 1727–1756.

Taylor H. P. and Epstein S. (1962) Relationships between O18/O16

ratios in coexisting minerals of igneous and metamorphic rocks,part I: principles and experimental results. Am. Geol. Soc. Bull.

73, 461–480.Wang X., Dong Z., Zhang J. and Liu L. (2004) Modern dust

storms in China: an overview. J. Arid Environ. 58, 559–574.Wang Y. X., Yang J. D., Chen J., Zhang K. J. and Rao W. B.

(2007) The Sr and Nd isotopic variations of the Chinese LoessPlateau during the past 7 Ma: implications for the east Asian

410 Y. Yan et al. / Geochimica et Cosmochimica Acta 139 (2014) 399–410

winter monsoon and source areas of loess. Palaeogeogr.

Palaeoclimatol. Palaeoecol. 249, 351–361.Washington R., Todd M., Middleton N. J. and Goudie A. S. (2003)

Dust-storm source areas determined by the total ozone monitoringspectrometer and surface observations. Ann. Assoc. Am. Geogr. 93,297–313. http://dx.doi.org/10.1111/1467- 8306.9302003.

Xiao J. L., Porter S. C., An Z. S., Kumai H. and Yoshikawa S.(1995) Grain size of quartz as an indicator of winter monsoonstrength on the Loess Plateau of central China during the last130,000 yr. Quat. Res. 43, 22–29.

Xiao G. Q., Zong K. Q., Li G. J., Hu Z. H., Dupont-Nivet G.,Peng S. Z. and Zhang K. X. (2012) Spatial and glacial–interglacial variations in provenance of the Chinese loessplateau. Geophys. Res. Lett. 39, L20715. http://dx.doi.org/10.1029/2012GL053304.

Xiong S. F., Ding Z. L. and Liu T. S. (2001) Climatic implicationsof loess deposits from the Beijing region. J. Quat. Sci. 16, 575–582.

Yang X., Zhang F., Fu X. and Wang X. (2008) Oxygen isotopiccompositions of quartz in the sand seas and sandy lands of

northern China and their implications for understanding theprovenances of aeolian sands. Geomorphology 102, 278–285.

Zhang S. G. and Wu J. S. (2002) Chinese Metamorphic Map.Geological Publishing House, Beijing (In Chinese).

Zhang X. Y., Arimoto R., An Z. S. and Zhi S. (1997) Dustemission from Chinese desert sources linked to variations inatmospheric circulation. J. Geophys. Res. 102, 28041–28047.

Zhang X. Y., Arimoto R. and An Z. S. (1999) Glacial andinterglacial patterns for Asian dust transport. Quat. Sci. Rev.

18, 811–819.Zhang X. Y., Gong S. L., Zhao T. L., Arimoto R., Wang Y. Q. and

Zhou Z. J. (2003) Sources of Asian dust and role of climatechange versus desertification in Asian dust emission. Geophys.

Res. Lett. 30, 2272. http://dx.doi.org/10.1029/2003GL018206.Zhou B. Y. and Shi J. X. (2003) Topographic Map of the People’s

Republic of China. Sinomaps Press, Beijing (In Chinese).

Associate editor: Dominik Weiss