Key Laboratory of Continental Collision and Plateau Uplift, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China
Key Laboratory of Western China's Environmental Systems, Ministry of Education of China and College of Resources and Environment, Lanzhou University, Lanzhou, China
Corresponding author: X. Fang, Key Laboratory of Continental Collision and Plateau Uplift, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China. (email@example.com)
 Between 2010 and 2011, this project collected and analyzed forty-nine superficial soil samples, for wax-derivedn-alkaneδD values (δDwax), along a south to north transect of the Tibetan Plateau, from the southern Plateau and Nepalese Himalayas, passing through the Nam Co basin, to Qilian Mountains. Twenty-two paired river water samples were also collected from northeastern Tibet during this period and analyzed forδD (δDRW). The δDRW and δDwaxvalues become progressively more negative northward from ∼27.5°N, and reach a minimum at ∼30.6°N (Nam Co basin). North of the Nam Co basin, to 35°N, isotope values increase, due to increasing contributions from the year-round westerlies and recycled moisture from the Plateau. Relatively high and constantδDwax and δDRW values prevail in areas north of ∼35°N in northeastern Tibet. Results show that these δDwaxvalues vary considerably with location and relate closely to the influences of the summer monsoon and circulation changes. These changes track the spatial variability of isotopes from modern river water and precipitation at large spatial scales. Paleoelevation reconstructions should take into account the impact of mixing between continental and monsoon-derived moisture on the relationships with elevation andδDwax (and linear isotopic lapse rates) for central northeastern Tibet. Based on extended and more sensitive (relative to δ18O) δDwax and δDRW values, we infer that in the past the westerlies reached further south, to the north piedmont of the Nyainqentanglha Range, and that Indian Summer Monsoon moisture pushes across the Tanggula Mountains and approaches the Kunlun Mountains.
 Stable-isotope paleoelevation reconstructions of the Tibetan Plateau rely on empirical data or model-based assumptions of the relationship between the Δδ18O or ΔδD of precipitation and elevation that can be applied to paleoprecipitation isotope proxy localities [Garzione et al., 2000; Rowley et al., 2001; Poage and Chamberlain, 2001; Rowley and Garzione, 2007]. Previous studies have focused mainly on the relationship between soil δDwax and elevations from northwestern and southeastern Tibet and in their nearby regions [Jia et al., 2008; Luo et al., 2011; Pu and Liu, 2011;Bai et al., 2011]. At those locations, the isotopic elevation effect of precipitation has been found to control the δDwax elevation gradients in the mountains, suggesting the potential of n-alkaneδDwax as a paleoelevation proxy. Although Polissar et al. applied this new technique of using isotope-based biomarkers for determining paleoelevations to the Cenozoic Tibetan Plateau (TP), the understanding of the response to elevation in soilδDwax values under the influence of different moisture sources on the TP remains limited [Luo et al., 2011; Bai et al., 2011]. Furthermore, the arid region (which still receives abundant summer precipitation) in the Tianshan Mountains witnesses the increased isotopic lapse rate (−2.62‰/100 m, <3200 m) compared to that found in the West Kunlun Shan, even though precipitation from the Polar Westerlies dominates both these regions [Bai et al., 2011; Luo et al., 2011].
 The isotopic lapse rate (the average change in δD with elevation) depends on the distribution of condensation and is controlled by low-elevation temperatures and relative humidity [Rowley et al., 2001]. Previous studies have indicated that the isotopic variability in areas north of ∼35°N seem to be driven primarily by temperature, while in regions farther south, the amount of precipitation controls the isotopic rates [Araguás-Araguás et al., 1998; Johnson and Ingram, 2004; Yao et al., 1996]. Aridity has been suggested as one of the causes for enriched δ18O and δD in precipitation on the northern TP (relative to equivalent elevations to the south), as well as the reduced isotopic lapse rates on the northern Plateau [Hren et al., 2009; Polissar et al., 2009; Yao et al., 2009]. Consistent with δ18O and δD in precipitation, fractionation factors tend to be smaller for both mid-chain and long chainn-alkanes at sites receiving lower amounts of precipitation on the arid TP [Xia et al., 2008; Aichner et al., 2010; Bai et al., 2011]. Thus, aridity increases δDwax and the isotopic fractionation between δDwax and precipitation, and results in the decreased isotopic lapse rates characteristic of northern Tibet today [Polissar and Freeman, 2010; Bai et al., 2011]. Detailed local investigations are still needed to establish empirical, local calibrations of the stable isotope relationships on the TP.
 Traditionally, the maximum northward transport of summer monsoonal moisture was thought to extend only as far north as the Tanggula Mountains (∼33°N), which separate the southern TP from Qinghai Province [Ahrens, 1991; Wang, 2006; Ding et al., 2009]. A number of other δ18O studies of precipitation and snow also conclude that the region of the Tanggula Mountain range, extending from west to east between the Nagqu and Tuotuohe stations, forms an important climatic divide [Yao et al., 1991; Tian et al., 2001a, 2003; Yu et al., 2006, 2008], but some of these studies were based on a limited number of samples from south of 35°N latitude [Tian et al., 2001a; Yu et al., 2008].
 This study extended the collection of surface soils to forty-nine from the southern to central and northeastern TP, in order to investigate spatial patterns of soilδDwax at regional scales (see Figure 1). Analyses of these samples have provided new findings: 1) that the westerlies reached further south in the past, to the north piedmont of the Nyainqentanglha Range, and that Indian Summer Monsoon moisture pushes across the Tanggula Mountains and approaches the Kunlun Mountains; and 2) that additional constraints apply to the application of δDwaxvalues for paleoelevation reconstructions in this region. We compared these data with published hydrogen stable isotope ratios for precipitation, stream water, leaf wax-derivedn-alkanes [e.g.,Xia et al., 2008; Hren et al., 2009; Ding et al., 2009; Aichner et al., 2010; Duan et al., 2011], and with relational data from the IAEA Global Network for Isotopes in Precipitation (GNIP) or the China Network for Isotopes in Precipitation (CHNIP). These comparisons will help to lead to a comprehensive understanding of basic δDwax features.
2. Study Area, Sampling and Experimental Procedures
2.1. Geographic and Climatic Characteristics of the Study Region
 On the TP, vapor producing rainfall comes mainly from western sources, low-latitude oceans and the inner Plateau, and is influenced by local factors, esp. topography. Across Tibet rainfall decreases westward and northward from, for example, 1300 mm/yr at Darlag in easternmost Tibet, to 436 mm/yr in Lhasa in the south-central part of the country, to 200 mm/yr at Qilian Mountains lie at the north edge of the TP.
 The northwest to southeast trending Himalayan mountain range dominates Nepal's topography. From south to north, the subtropical plains of the Terai and the Siwalik highland region increase in elevation. While the intermediate elevation ranges (1000–2500 m) have subtropical valley bottoms, they have warm temperate valley sides and cool temperate regions on the higher ridges (which experience occasional snowfall).
 The Himalayas consist of a 2000 km long mountain belt, which significantly shapes the climate of Southeast Asia and the northern Hemisphere [Clemens et al., 1991]. The Indian Summer Monsoon (ISM) dominates the atmospheric circulation [Bookhagen et al., 2005; Anders et al., 2006; Bookhagen and Burbank, 2006]. Mean annual precipitation varies considerably across the Himalayas and TP [Hren et al., 2009, and references therein]. The contribution of summer precipitation (May to October) to the annual total ranges from ∼80% south of the range, to greater than 95% on the central Plateau. Tian et al. [2001a, 2003] and Yao et al. even found that rivers in southeast Tibet were mainly fed by monsoon precipitation, which supplied up to 85% of the annual precipitation there. Also, the snow/icemelt from high-elevation ranges at the orographic barrier has a large contribution in Himalayan catchments [Bookhagen and Burbank, 2010].
 The Qilian Mountains create a strong rain shadow effect for monsoon precipitation coming from the southeast (a region that includes parallel mountains chains and river basins, with elevations between 2000 m and 6000 m in the cold and semi-arid region). Distributed forests lie between 2500 m and 3500 m; while permafrost, permanent snow, and modern glaciers extend above 3600 m, 4400 m, and 5000 m, respectively. Annual precipitation decreases from east to west (ranging from 600 to 200 mm), but increases along with the increase in elevation. Precipitation takes place mainly in summer (60–70% of the total annual amount). The main supply of mountain runoff was rainfall during the rainy season at the headwaters of the Heihe River basin, with a low contribution from glacial and snowmelt water [Zhao et al., 2011].
2.2.1. Sample Collection and Preparation, n-Alkane Extraction and Quantification
 During 2010 and 2011, this study removed the surface litter layer and collected a total of 49 superficial soil samples from a maximum depth of 5 cm. Sample sites ranged in elevation from 160 m in Nepal, to 5201 m on the Central TP, and included a number from between 2815 and 4126 m in northeastern Tibet (Table 1). Soil samples from northeastern Tibet were collected over a 2-week period in October 2010, while samples from the central TP area and Nepal were gathered during July and August 2011. During field trips to Qilian Mountains, 22 paired river water samples were collected synchronously, after the peak monsoonal precipitation (Table 2).
Table 1. Superficial Soils δDwax Values Along a South to North Transect of the Tibetan Plateaua
Samples show published δDwax values selected from recent lake surface sediments (sample depth below 0.5 m) on Tibetan Plateau and their nearby areas. The isotopic fractionation between δDwax and river water or precipitation: εwax/p = 1000*((δDwax + 1000)/(δDp + 1000) − 1).
Table 2. Analytical Results of Isotopes for River Water Samples (δDRW) From Qilian Shan Mountains and Published Hydrogen Stable Isotope Ratios for Precipitation (δDP) and Stream Water, and Relational Data From GNIP and CHNIP
 Soil samples were freeze-dried and ground to a powder (80 mesh). Aliquots (of ∼20 g) were used for Soxhlet extraction with CH2Cl2/MeOH (10:1, for 24 h). The extract volume was dried using a gentle stream of pure N2. The dry extracts were re-dissolved in 1 ml of CH2Cl2 and fractionated using column chromatography with a column of neutral alumina over activated silica gel (100–200 mesh), into aliphatic, aromatic and polar, and acidic fractions. Saturated fractions were eluted with hexane (50 ml), aromatic fractions with CH2Cl2(50 ml), and non-hydrocarbon fractions with MeOH (20 ml).
2.2.2. Analysis of δD Values of n-Alkanes andδDRW Values
 Hydrogen isotopic analysis of individual n-alkanes was performed using GC-thermal conversion-isotope ratio mass spectrometry (GC/TC/IRMS). Analyses used a Thermo Trace Ultra gas chromatograph, along with a high temperature pyrolysis unit connected online to a Thermo Delta V Advantage isotope ratio mass spectrometer. Individual compounds were pyrolized to convert organic H to H2 at 1450°C, with the H2 then introduced into the mass spectrometer. We used the same temperature program and GC column as used for the GC analysis. The H3 factor was determined once a day and remained very close to 4.12 (SD 0.03) during sample analysis, ensuring stable ion source conditions. We evaluated reproducibility and accuracy using standards containing six n-alkanes (C21, C23, C25, C27, C29, and C30) between every six measurements. The 1σ precision for the six laboratory standards was <±6‰ throughout the entire process. The 1σ precision for the triplicate analysis of the C29n-alkane from all analyzed samples was <±6‰. TheδDRWvalues of river waters were measured in triplicate with a Picarro L1102-I Isotopic Water Analyzer, with a 2σ analytical precision better than 0.5‰.
 As in our previous work [Bai et al., 2011], we used the abundance-weighted averageδD values of the C29 and C31n-alkanes, which probably provide more suitable indicators of the precipitation signal ofδD values of n-alkanes from soil leaf waxes (δDwax) and better represent the mean isotope signal of local precipitation to compare with elevation in the following discussion.
3. Results and Discussion
3.1. Spatial Patterns of Superficial Soil δDwax Values
 Our δDwax values become progressively more negative from about 27.5°N to 30.6°N (the north piedmont of the Nyainqentanglha Range), where we found the lowest δDwax values, around −244‰ (see Figure 2b). The measured spatial distribution of δDRW shows a similar strong relationship to latitude (Figure 2c). Stable isotopes in precipitation show an apparent precipitation “amount effect” on the southern TP [Araguás-Araguás et al., 1998; Tian et al., 2001a, 2001b, 2003]. The data show a relationship between latitude and δDwax of −23.2‰/°latitude (R2= 0.89). The north piedmont of the Nyainqentanglha Range derives nearly all its moisture from the summer monsoon, and monsoon-dominated isotopic lapse rates are very steep.
 However, δDwax values become progressively less negative north of about 30.6°N latitude, to 35°N, from −244‰ to −158‰ with increasing distance northward, despite relatively constant elevations above 4 km (Figure 2a and Table 1). For these sites, we found a relationship between δDwax and latitude of 14.0‰/°latitude (R2 = 0.76), comparable to the 15.2‰/°latitude (R2 = 0.54) value for δDRW determined by Hren et al. . Summer monsoon precipitation decreases from >1 m south of the Himalayas, to <0.1 m across much of the TP, with little variability in total rainfall across much of the high elevation landscape [Hren et al., 2009]. Constant summer precipitation and increasing δDwax in these soils and δDRW in stream waters (Figures 2b and 2c) reflect decreasing contributions of moisture derived from more southern areas, where the westerlies control the year-round Central Asian climate and where recycled moisture contributes to the winter precipitation. These effects gradually increase from 30.6 to 35°N.
 Measured δDwax values in areas north of about 35°N in northeastern Tibet remain relatively constant and high (average −161‰, ranging from −172.6‰ to −149.5‰, at elevations above ∼2800 m). Table 1 and Figure 2c show these values for sites ranging from 2815 to 4126 m with an average elevation of 3272 m. Exceptions exist where lower δDwaxvalues were found at some high-elevation sites (Figure 2b; e.g., −171‰ at 4677 m, −180‰ at 4498 m, −173.5‰ at 4750 m, and −168.8‰ at 4556 m). The δD values at these sites seem to be controlled mainly by the year-round westerlies and recycled moisture, and driven by temperature in a manner consistent withδDRW in stream waters [Araguás-Araguás et al., 1998; Johnson and Ingram, 2004; Hren et al., 2009; Hough et al., 2011]. Ding et al.  study the regional variations of the δ18O values of river water from different vapor sources and recycling patterns on the TP, although river water δ18Ow values also change from an increasing trend south of about 35°N to relatively constant values to the north in areas north of about 35°N on the central and northern TP (Figure 3), their paper does not discuss their significance because of the less sensitive δ18O values. The Mongolian High controls the moisture of Qilian Mountains during the winter, spring, and autumn [Ding and Kang, 1985], and the moisture originates from continental recycling or rapid evaporation over relatively warm water bodies (such as the Black, Caspian, and Aral Seas) when dry westerly traveling air masses pass over. Hence, very high δ18Ow in precipitation result [Zhou et al., 2007]. Accordingly, the variability of δDwax in soils tracks the spatial variability of modern isotopes δDWater in precipitation/water. This suggests that δDWater in precipitation/water is of importance for the δD values of n-alkanes.
3.2. The Isotopic Fractionation Between δDwax and River Water (εwax/RW) on the Tibetan Plateau
 Our analyses show εwax/RW values to be fairly constant and relatively low (Table 1; −108.9‰ to −137‰; with a mean value of −118.2‰, n = 17) in northeastern Tibet (annual precipitation (Pann) ranging from 200 to 600 mm). These δDwax values are close to a net fractionation of ∼−90‰ between source waters and terrestrial lipids at sites on the northeastern and central TP and from the Qaidam Basin (Pann 43 mm), with a negative moisture balance (Table 1) [Xia et al., 2008; Aichner et al., 2010]. While surface soil samples from the Nam Co basin (Pann of 282 mm) contain the most negative δDwax values, the εwax/RW value was found to be −141.8‰ (δDRW = −119.1‰ [from Xu et al., 2011]). Our results support the premise that aridity or water stress in the growing season acts as the primary factor controlling εwax-p; the more arid, the more positive the εwax-p values [Luo et al., 2011].
3.3. Boundaries of the Various Air Masses and the Monsoon Intensity
 Based on the spatial variations of δDwax values in soils on the TP, this study finds that major changes occur on the north piedmont of the Nyainqentanglha Range (∼30.6°N) and at 35°N (Figure 2b).
 Extended river water δ18Ow and the more sensitive δDRW values both show steadily increasing in areas between ∼30°N to ∼35°N and relatively constant values north of about 35°N on the central and northern TP (Figures 3 and 2c). Similarly, we find that our δDwax values steadily increase north of about 30.6°N, to about 35°N (Figure 2b). The change from an increasing trend south of about 35°N to more constant values to the north might indicate that westerly winds transport water vapor from the Black, Caspian, and/or Aral Seas, past Qilian Mountains to the north piedmont of the Nyainqentanglha Range. Alternatively, summer monsoonal moisture could cross the Tanggula Mountain and extend close to the Kunlun Mountains.
3.4. Implications for Paleoelevation Reconstructions
 As suggested by Hren et al. , changes in the δ18O or δD values in paleoprecipitation on the central and southwestern TP may reflect local topographic uplift along moisture transport pathways or changes in the penetration of monsoonal-derived moisture, rather than regional uplift histories. The variability of ourδDwax values in soils corresponds with the spatial variability of modern isotopes in precipitation, and the δDwax values and altitudinal lapse rate from the northeastern and central TP may be influenced by complicated effects of southerly derived moisture and local convection, or evolution of the westerlies.
 Our results demonstrate the response of δDwax and δDRW compositions to factors other than elevation changes, especially in central Tibet, which has relatively high and constant elevations above 4 km (Figure 2a and Table 1; north of Nam Co basin, to 35°N). These results also highlight the large uncertainties associated with paleoelevation reconstructions based solely on shifts in δDwax paleorecords. For example, Xu et al.  estimated that the average contribution of evaporation from Lake Nam Co to the local atmospheric vapor has varied from 28.4 to 31.1% during the summer monsoon season in recent years. Based on the deuterium excess evaluated for water vapor sources of total precipitation in the Heihe River basin, Zhang and Wu estimated that the percentage of precipitation contributed from recycled continental water evaporation to be at least 31%. Paleoelevation reconstructions should take into account the impact of mixing between continental and monsoon-derived moisture onδDwax values and the altitudinal lapse rate of δDwax values for central northeastern Tibet. These suggest that δDwaxvalues may also vary temporally with monsoon shifts and intensification. It follows that the development of the paleo-monsoons over the TP region should be considered when the altitudinal lapse rates ofδDwax values are applied to paleoelevation reconstructions.
 Furthermore, the variable εwax/RW decreased from a mean of −118.2‰ in northeastern Tibet, to −141.8‰ in the Nam Co basin, which exhibits the most negative δDwax values (Table 1). This pattern is consistent with a surface sediment study of Lake Qiangyong Glacier, Lake Kongmu Co, Lake Nam Co, Lake Keluke, and Lake Xiao Qaidam along the S-N transect from the central and northern TP (Pann ranging from 95 to 379 mm). The isotopic fractionations εwax/p and εC31/p were found to be relatively small (−64.5‰ to −93‰, and −66‰ to −91‰) (Table 1) [Xia et al., 2008]. While samples from lake surface sediment in the humid regions (southeastern Tibetan Plateau, Pann 1073 mm) resulted in εwax/p and εC31/p values of −114‰ and −129‰, close to reported values for aquatic and terrestrial biomarkers in Europe, North America and eastern China [Sachse et al., 2004; Huang et al., 2004; Rao et al., 2009]. These values become smaller (from −119‰ to −77‰ and −122‰ to −73‰) at sites on the northeastern and central TP and in the Qaidam Basin (Pann 43 mm) (Table 1) [Aichner et al., 2010]. As a result, when using climate-specificεwax-pvalues to reconstruct paleo-precipitationδD, or further for calculating theoretical paleoelevations, εwax-p values that represent study sites should be used, rather than averages of an apparent fractionation factor from similar type modern vegetation [e.g., Polissar et al., 2009].
 Thus, we find it essential to evaluate the influence of atmospheric circulation and water source changes on δDRW, in conjunction with δDwax data, for paleoelevation reconstructions. A holistic approach must be taken to distinguish between changes in δDwax values resulting from regional surface uplift versus changes of moisture. Hence, caution must be used when interpreting isotopic data in absence of standard isotopic distillation models for the specific region under study.
 We would like to thank Jimin Sun, Fuli Wu, Jinpo Zhan, Yi Chen, Weilin Zhang, Yunfa Miao and Xiaohui Fang for help with sampling. We also thank Xiaoming Liu for GC analyses and Zhen Zhang (Institute of State Key Laboratory of Loess and Quaternary Geology, CAS, Xian) for measurement of δD values. We are very grateful for the suggestions and comments by Editor Renyi Zhang, the anonymous reviewer, Bodo Bookhagen and C. Page Chamberlain, and Bill Isherwood for editorial suggestions on the manuscript. This work was co-supported by the NSFC grants (grants 189086 and 41071003).