Holocene shorelines and lake evolution in Juyanze Basin,southern Mongolian Plateau,revealed by luminescence dating

Abstract

Holocene environment change in the Gobi Desert of Mongolia and northwestern China is of global significance as it is one of the key dust source areas of the world. The Juyanze Basin, located in the central Gobi Desert of southern Mongolian Plateau, is one of the three terminal basins of the Heihe River that flows north from the Qilian Mountain. A series of geomorphic shorelines at different elevations around the basin indicate that large lake levels fluctuated during the past. In this study, we measured the exact elevation of shorelines in the Juyanze Basin using Digital Global Positioning System and found seven shorelines at ~37, ~36, ~34, ~30, ~28, ~24, and ~22 m above modern basin floor (a.m.b.f.). Optically stimulated luminescence dating was employed to date the deposition of beach sand and gravel sequences at these shorelines. Results indicate a paleolake developed in the basin at ~5 ka when lake levels reached ~26 m a.m.b.f.. The lake level then gradually increased to ~29 m a.m.b.f. at ~3.3 ka and reached its highest Holocene level of~37 m a.m.b.f. at ~2.7 ka. The lake environment during 1.1–0.8 ka was characterized by frequent lake-level fluctuations at ~30 m a.m.b.f. The lake disappeared from both East and West Juyanze Basin after ~0.6 ka. This sequence is out-of-phase with other regional Holocene lake records which indicate Holocene high stands occurred during the early to middle Holocene. We suggest that out-of-phase lake high stands at ~5–1 ka in the Juyanze Basin are related to geomorphological shifts of the Heihe River channels across its fan/delta, feeding the three sub-basins of the Ejina at different times. Using paleolake shorelines in this region to reconstruct the climate changes needs to be very careful.

Keywords

DGPS measurement Gobi Desert Holocene Mongolian Plateau OSL dating paleolake shoreline

Introduction

There are more than 1,300,000 km² of gobi (stony) deserts in the Mongolian Plateau of central Asia that consists of wide, shallow basins with smooth rocky bottoms filled with sand, silt/clay, pebbles, or, more often, gravel (Cable and French, 1943; Cooke, 1970; Wang et al., 2012). These gobi deserts, and the sand deserts around them, are believed to be the dominant sources of dust emissions from central Asia (Natsagdorj et al., 2003; Wang et al., 2006, 2008). Understanding Holocene lake evolution in this region provides insight into the hydrological cycles that led to the deposition and exposure of these sediments, as well as to the underlying climatic and environmental history of the area. A section of the ancient Silk Road runs through this region, and it has a long history of agricultural cultivation and intense human influence (Mischke, 2001).

The Ejina Basin lies in the center of this region of gobi deserts and is the closed-basin catchment of the Heihe River (Figure 1), the second largest inland river in China. The large basin is composed of three major sub-basins that are alternatively fed by the river as it shifts back and forth across its delta. These are, from west to east, the Gaxun Nur, Sogo Nur, and Juyanze Basins (Figure 1). Previous shoreline- and drill core–based studies in these sub-basins have shown that the area experienced dramatic environmental changes during the Holocene (Wünnemann et al., 1998). Nearly all of these studies suggest highly variable climate fluctuations in the basin during the Holocene (Chen et al., 2003, 2008; Wünnemann et al., 2007a, 2007b; Yu and Kelts, 2002; Yu et al., 2006; Zhang et al., 2004) related to Asian monsoon dynamics or/and westerly airflow over China that follow the general trend of northern hemispheric insolation patterns (e.g. Chen et al., 2008; Herzschuh, 2006). These authors have interpreted Holocene signals in proxy records from lake sediments as direct responses to climate variability, but with remarkable differences in timing, frequency, and spatial distribution (e.g. Hartmann and Wünnemann, 2009; Herzschuh et al., 2004; Mischke et al., 2002).

Figure 1.

Map of the Heihe River system and its surroundings.

Mischke et al. (2002) reconstructed East Juyanze lake-level changes based on ostracods and stable isotopes and suggested that the highest lake levels (water depth ~10 m) occurred between 5100 and 4100 cal. yr BP, succeeding intermediate and low levels lasted until about 3150 cal. BP, and a relatively high lake-level phase appeared at 3000 cal. BP. Herzschuh et al. (2004) studied the pollen of a sedimentary record from East Juyanze and interpreted high values of Chenopodiaceae, Ephedra fragilis, and other desert taxa to indicate a rather dry climate ~10,700–5400 cal. BP. More favorable conditions are thought to have occurred between 5400 and 3900 cal. BP on the basis of a relative increase in the abundance of Artemisia pollen, with the lake finally desiccating after 1700 cal. BP. However, Hartmann and Wünnemann (2009), using multivariate statistics based on lithological, geochemical, and mineral data from the same core (core G36) in East Juyanze, suggest that the highest lake levels appeared ~10.7–8.9 cal. ka BP and that there was a general decreasing trend toward complete desiccation during the middle-to-late Holocene. Wünnemann and Hartmann (2002; see also Wünnemann et al., 1998) measured elevations of a series of shorelines in northern West Juyanze using a Digital Global Positioning System (DGPS) and reported six shorelines formed at elevations of ~926, ~925, ~921, ~917, ~913, and ~912 m. ¹⁴C dating of mollusks from the ~917- and ~913-m shorelines produced age estimates of ~5.3 and ~2.8 cal. ka, respectively.

Holocene lake evolution in the Juyanze Basin remains a subject of debate partly because Holocene lacustrine deposits have been episodically deflated by strong wind erosion in the region, and it is hard to retrieve a continuous depositional sequence. An additional problem is that nearly all these studies have employed ¹⁴C dating methods to establish a chronology. However, the radiocarbon method can be problematic in dating eolian deposits because of their low organic carbon content (Li et al., 2007), and dates on organic carbon from lacustrine deposits are often overestimated because of a reservoir effect (Björck et al., 1991; Colinan et al., 1996; Long et al., 2011; Wang et al., 2002; Zhou et al., 2009). Alternatively, the ages of eolian and lacustrine deposits can be estimated using optically stimulated luminescence (OSL) dating technology to directly determine quartz or feldspar minerals (Aitken, 1985). With the development of the single-aliquot regenerative (SAR) dose protocol (Murray and Wintle, 2000), quartz OSL dating is now widely applied to the dating of Quaternary sediments (Li et al., 2007; Murray and Olley, 2002; Wintle, 2008; Wintle and Murray, 2006).

In this study, we systematically measured the exact elevation of well-preserved paleoshorelines around the Juyanze Basin using DGPS and then trenched the shorelines in the Juyanze Basin in a number of locations to retrieve sand-and-gravel sequences from paleolake shorelines at different elevations. Quartz OSL dating was employed to date these sand-and-gravel sequences. The reliability of the OSL age estimates was confirmed by internal checks of luminescence characteristics. Using these age estimates, we establish a chronology for depositional sequences at different shoreline elevations, discuss the process of lake-level fluctuations in the Juyanze Basin during the Holocene, and explore possible causal mechanisms of lake evolution.

Geological setting

The Ejina Basin lies between the Qilian Mountains to the south and the Gobi-Tienshan mountain complex to the north. The basin is part of the Alashan Plateau that lies in the desert belt north of the Tibetan Plateau (TP). This area is known generically as the Gobi Desert. The Heihe River originates in the ice-covered peaks of the Qilian Mountains and terminates in the Ejina Basin, alternately feeding the Gaxun Nur, Sogo Nur, and Juyanze terminal lakes (Figure 1). The Juyanze Basin is heart-like in shape and is divided into two parts separated by a distinct south-north striking cliff-like structure that acts as a hydro-morphological threshold between the two sub-basins (Figure 2a). The Juyanze Basin is ~3500 km² in area and is surrounded by a pediment margin and Paleozoic and older crystalline rocks. Marine meta-sediments occur on the north and northeast, and massive ~100 m high dunes of the Badain Jaran Desert are on the south. The Gaxun Nur Basin is now completely dry, the Sogo Nur Basin fills ephemerally, and the Juyanze Basin contains small shallow water bodies fed by groundwater and sporadic river discharge. The Juyanze Basin consists of East Juyanze with a basin floor elevation of ~894 m and West Juyanze with a basin floor elevation of ~890 m. These two sub-basins are divided by a ridge from north to south, and the threshold elevation between East and West Juyanze is about ~914 m (Google earth elevation). Lake Tia E Hu is a small seasonal lake located at the lowest point in the central West Juyanze Basin at ~890 m. As the Heihe River shifts back and forth across its fan/delta, it alternately fills its Gaxun Nur, Sogo Nur, and Juyanze Basins, with some flow probably reaching all the sub-basins simultaneously via both surface and groundwater flow. Extensive land use and irrigation in the upper valley of the Heihe River has reduced modern flow to these Ejina Basin terminal lakes (Mischke et al., 2002). This portion of the Heihe River lies within the densely populated ‘Hexi Corridor’, a part of the ancient Silk Road connecting eastern China with the Middle East and Europe, and has a long history of agricultural cultivation (Haude, 1931).

Figure 2.

(a) Shoreline distributions in the Juyanze Basin (inset squares are the locations of b and c). The dashed line is the elevation of a transect from West Juyanze to East Juyanze as shown in Figure 7. (b) Remote sensing image of shorelines in the East Juyanze Basin. ‘A’ is the locality of section ET-7 (927 m), ‘B’ is ET-6 (926 m), ‘C’ is ET-5 (924 m), and ‘D’ is ET-2 (918 m). (c) Remote sensing image of shorelines in the West Juyanze Basin. ‘E’ is the locality of section WT-4 (920 m), ‘F’ is WT-2 (918 m), and ‘G’ is WT-1 (912 m).

The Asian monsoon penetrates the interior of the continent in the summer, carrying moisture from the South China Sea and Pacific Ocean. During the winter season, in contrast, northwestern China is under the influence of cold and dry air related to a high-pressure cell over Siberia and Mongolia. The interplay of both air masses with opposite wind directions results in a seasonal monsoon climate with very pronounced summer rainfall. The annual precipitation in the upper reaches of the Heihe River amounts to about 400 mm but less than 50 mm/yr falls in the region around the Ejina Basin terminal lakes (Provincial Atlas of Gansu, Lanzhou, 1975). The amount of potential evaporation is in the range of 2000–2500 mm/yr for the plains and foothills.

Material and method

Juyanze Basin paleolakeshore distributions, DGPS measurements, and OSL sampling

As shown in the satellite image in Figure 2a, a series of paleoshorelines are clearly visible in the East Juyanze Basin (Figure 2b) and on both slopes of the watershed between East and West Juyanze (Figure 2c). We identified eight shoreline barrier bar features between the basin margin and the center of East Juyanze and term these shorelines ET-7, ET-6, ET-5, ET-4, ET-3, ET-2, ET-1, and ET-0. We identified six shorelines in West Juyanze near the watershed area and named them WT-6, WT-5, WT-4, WT-3, WT-2, and WT-1.

The exact elevations of shorelines in the Juyanze Basin were measured using DGPS. Following the measurement techniques, processing, and error estimations described by Yang et al. (2008), we used two second-level national survey control points on the eastern margin of East Juyanze (C1: 41°51′11.42″N, 101°59′19.55″E, 944.0 m) and at the watershed area between the East and West Juyanze Basins (C2: 41°54′06.85″N, 101°45′27.68″E, 944.3 m) as the benchmarks for elevation control. The exact altitudes for each shoreline were retrieved using two Trimble GPS receivers in a real-time kinematic (RTK) mode, one at a control point and the other one at a shoreline. The measured DGPS data for shorelines in the Juyanze Basin are shown in Table 1.

Table 1.

DGPS measurement results for Juyanze Basin shorelines.

East Juyanze shoreline sites	Altitude	Latitude	Elevation (m)	West Juyanze shoreline sites	Altitude	Latitude	Elevation (m)	Shoreline elevation (m)
ET-7	41°51′16.37″	101°58′07.93″	927.05	WT-6	41°54′13.87″	101°45′37.70″	926.87	~927
ET-6	41°51′15.76″	101°58′05.28″	926.37	WT-6	41°54′13.87″	101°45′37.70″	926.87	~927
ET-5	41°51′18.71″	101°57′51.87″	923.85	WT-5	41°54′13.21″	101°45′32.75″	923.55	~924
ET-4	41°51′19.14″	101°57′48.98″	923.48	WT-5	41°54′13.21″	101°45′32.75″	923.55	~924
ET-3	41°51′12.00″	101°57′24.11″	920.22	WT-4	41°54′15.66″	101°45′25.17″	920.69	~920
ET-2	41°51′23.50″	101°57′0.13″	917.71	WT-3	41°54′18.13″	101°45′17.99″	918.04	~918
ET-2	41°51′23.50″	101°57′0.13″	917.71	WT-2	41°54′18.22″	101°45′15.32″	917.63	~918
ET-1	41°51′26.09″	101°56′45.97″	913.98	WT-1	41°54′19.03″	101°45′0.02″	912.00	~914
ET-0	41°50′38.85″	101°55′47.83″	911.88	WT-1	41°54′19.03″	101°45′0.02″	912.00	~912

DGPS: Digital Global Positioning System.

In the East Juyanze Basin, a bulldozer was used to dig trenches across the ~924- and ~918-m shoreline features, penetrating through uncompact beach sands and gravels to bedrock. The sand/gravel sequence at the ET-5 section at the ~924-m shoreline reached a depth of 550 cm, while that at the ET-2 section reached 210 cm. OSL samples were collected at ET-5 from near-shore sands at 534 cm, from shoreline sandy gravel at depths of 100 and 80 cm, and from near-shore eolian sand at depths of 435 and 232 cm. Two samples were collected at the depths of 205 and 60 cm from the ET-2 section at the ~918-m shoreline. Two samples were also collected at depths of 140 and 90 cm from the ET-6 section at the ~926-m altitude shoreline, and two samples were collected from depths of 70 and 120 cm from the ET-7 section at the 927-m shoreline. In the West Juyanze Basin, two OSL samples were collected at depths of 130 and 120 cm from the 918- and 920-m shorelines, respectively. One sample was collected from a depth of 60 cm from the WT-1 section at the 912-m shoreline. All OSL samples were obtained by hammering 4 cm diameter stainless tubes into sediments in the exposed sections. The tubes were sealed inside black plastic bags with tape immediately after they were taken from the section.

OSL dating sample preparation and measurement methods

In all, 14 OSL samples collected from paleolake shorelines at different Juyanze Basin elevations were measured to establish the chronology of lake-level evolution. OSL sample preparation followed the process described by Aitken (1998). All laboratory processing, sample preparations, and luminescence measurements were carried out in a darkroom with subdued red light. All raw samples were treated with 10% HCl and 20% H₂O₂ to remove carbonate and organic matter. The samples were then sieved in water to select sediments with the specific grain sizes of 90–125 and 150–180 µm. Heavy liquids with densities of 2.62 and 2.75 g/cm³ were then used to separate quartz fraction of each sample. Quartz grains were treated with 40% HF for 60 min to remove the outer layer irradiated by alpha particles and any remaining feldspar grains. Then, all samples were treated with 1 mol/L HCl for 10 min to remove fluorides created during the HF etching.

OSL signals were measured using an automated Risø TL/OSL-DA-20 reader. Laboratory irradiation was carried out using ⁹⁰Sr/⁹⁰Y sources mounted within the reader, with a dose rate of 0.1486 Gy/s. The quartz OSL signal was detected through two 3-mm-thick Hoya U-340 filters, and the purity of quartz extracts was verified using an IR depletion ratio (Duller, 2003). In order to eliminate any influence from feldspar inclusions, measurement of the post-IR blue OSL signal has been proposed (Banerjee et al., 2001). The post-IR SAR protocol (Murray and Wintle, 2003) was employed to measure D_e values from quartz extracts of all OSL samples.

In order to retrieve detectable luminescence signals, a 5-mm aliquot size was applied for these quartz samples. The suitability of the quartz SAR sequence for D_e determination was checked with a ‘dose recovery test’ (Murray and Wintle, 2003) for ET-7-122 from section ET-7. This test examines the combined function of all the conditions of the procedure, such as preheat temperature and size of test dose. The sample was bleached using two room temperature optical stimulations with the blue LEDs switched on for 40 s, with 10 ks room temperature storage between the two stimulations (Wintle and Murray, 2006). Then a dose of 14.86 Gy was given to six groups with three aliquots in each group. Preheat temperatures from 160°C to 280°C at 20°C intervals for 10 s were tested and the cut-heat was 160–240°C for 10 s, using a heating rate of 5°C/s. The calculated measured/ given dose, recuperation, and recycling ratios for sample ET-7-122 are plotted in Figure 3, and the measured/ given dose ratio is ~1.01 at a temperature of 180°C and between 1.01 and 0.92 at different temperatures from 180°C to 240°C, indicating no dependence of preheat temperature to given dose measurement. The recycling ratio for different measurement temperatures is between 0.96 and 1.07. The recuperations for different preheat temperatures are less than 2%, ranging from −0.23% to 1.05%. Based on these results, the SAR sequence with a preheat temperature of 180°C and a cut-heat of 160°C was selected to measure the D_e of the quartz fraction from each Holocene sample.

Figure 3.

Dose recovery test results for quartz fractions of sample ET-7-120.

As shown in Figure 4, the OSL natural signal, test dose signal, and regeneration dose signal for typical sand samples ET-7-122 and ET-2-205 all decreased rapidly during the first second of stimulation (Figure 4a and c), indicating that the OSL signal is dominated by a fast component. The growth curve can be readily fitted using a single saturation exponential (Figure 4b and d). For each sample, at least 12 aliquots were measured. The growth curve has been forced to cross the origin in the D_e calculation considering that a recuperation as high as 10% appeared. The following rejection criteria were employed for D_e determination: (a) a recycling ratio outside 0.9–1.1, (b) a signal < 3σ above background, and (c) a D_e error > 10%. A central age model (Galbraith et al., 1999) was used to calculate the D_es for the quartz samples. The over-dispersion of the D_es for each sample varies between 11% and 32% (see Figure 5). The D_es and the calculated quartz OSL ages for each sample are listed in Table 2 and illustrated in Figure 5.

Figure 4.

Luminescence characteristics of samples ET-7-122 and ET-2-205: quartz OSL shine-down curves for the natural signal and regeneration doses of 14.86 and 0 Gy of samples (a) ET-7-122 and (c) ET-2-205. The curve of 0 Gy shows that recuperation is negligible. Best fit growth curves for samples (b) ET-7-122 and (d) ET-2-205 using an exponential function.

Figure 5.

Distribution of D_e values in radial plots for all 14 samples in this study.

Table 2.

Quart OSL dating result for samples from shorelines in the Juyanze Basin.

Shoreline elevation (m)	Sample no.	Depth (m)	Grain size (µm)	Aliquots	D_e (Gy)	U (ppm)	Th (ppm)	K (%)	Cosmic ray (Gy/ka)	Dose rate (Gy/ka)	Age (ka)
918	ET-2-060	0.60	90–125	12	8.95 ± 0.51	4.51 ± 0.15	1.56 ± 0.08	1.88 ± 0.06	0.23	2.74 ± 0.13	3.27 ± 0.17
918	ET-2-205	2.05	90–125	16	10.98 ± 0.17	1.77 ± 0.08	4.85 ± 0.17	1.84 ± 0.06	0.19	2.39 ± 0.12	4.59 ± 0.25
924	ET-5-080	0.80	90–125	10	2.47 ± 0.05	1.9 ± 0.08	3.89 ± 0.14	1.79 ± 0.06	0.22	2.35 ± 0.11	1.05 ± 0.06
924	ET-5-100	1.00	90–125	8	2.24 ± 0.09	2.21 ± 0.08	4.73 ± 0.16	1.69 ± 0.06	0.21	2.36 ± 0.11	0.95 ± 0.06
924	ET-5-232	2.32	90–125	12	2.12 ± 0.04	1.85 ± 0.08	3.68 ± 0.14	1.67 ± 0.06	0.18	2.18 ± 0.11	0.97 ± 0.05
924	ET-5-435	4.35	90–125	13	4.79 ± 0.10	2.67 ± 0.10	4.54 ± 0.16	1.77 ± 0.06	0.14	2.42 ± 0.12	1.98 ± 0.11
924	ET-5-534	5.34	90–125	18	6.39 ± 0.12	1.44 ± 0.07	3.49 ± 0.14	1.9 ± 0.06	0.13	2.24 ± 0.11	2.85 ± 0.16
926	ET-6-090	0.90	90–125	10	3.24 ± 0.06	2.32 ± 0.09	3.61 ± 0.14	1.79 ± 0.06	0.22	2.40 ± 0.12	1.35 ± 0.07
926	ET-6-140	1.40	90–125	10	2.23 ± 0.22	3.2 ± 0.19	2.32 ± 0.09	1.72 ± 0.06	0.20	2.39 ± 0.12	0.93 ± 0.10
927	ET-7-070	0.70	150–180	7	3.46 ± 0.27	2.22 ± 0.09	4.03 ± 0.14	1.63 ± 0.06	0.23	2.25 ± 0.11	1.54 ± 0.24
927	ET-7-122*	1.22	90–125	21	6.98 ± 0.08	3.43 ± 0.12	4.28 ± 0.16	1.81 ± 0.06	0.21	2.40 ± 0.12	2.91 ± 0.16
912	WT-1-060	0.6	90–125	6	0.76 ± 0.11	2.62 ± 0.10	4.07 ± 0.15	1.75 ± 0.06	0.23	2.46 ± 0.12	0.31 ± 0.05
918	WT-2-120	1.20	90–125	6	2.24 ± 0.26	1.5 ± 0.07	2.9 ± 0.12	1.93 ± 0.06	0.21	2.32 ± 0.11	0.96 ± 0.12
920	WT-4-130	1.30	90–125	11	2.26 ± 0.06	2.04 ± 0.08	5.47 ± 0.18	1.63 ± 0.06	0.20	2.32 ± 0.11	0.98 ± 0.06

OSL: optically stimulated luminescence.

The environmental dose rate was calculated from the measurements of radioactive element contents in the sample, with a small contribution from cosmic rays. For all samples, the uranium (U) and thorium (Th) concentrations and potassium (K) content were determined by means of neutron activation analysis (NAA). All results were converted to beta and gamma dose rates according to the conversion factors of Guérin et al. (2012). The dose rate from cosmic rays was calculated according to sample burial depth and the altitude of the section (Prescott and Hutton, 1994). A summary of the dosimetry data for all samples is listed in Table 2. Water content may change drastically after burial for those samples that were collected from lacustrine sediments in an arid area. Due to the uncertainty of sediment water content in the burial period, a water content of 20 ± 5% was introduced to calculate ages for paleolake shoreline sediments. As a change of 1% water content can cause a 1% difference in age, using a reduced water content of 15% changes the age estimates by only ~5%, making them slightly younger.

Results

DGPS measurement results

The DGPS measurements were taken at the center highpoint of each shoreline barrier bar, and results are shown in Table 1. As illustrated in Figure 7, the elevations of shorelines ET-7, ET-6, ET-5, ET-4, ET-3, ET-2, ET-1, and ET-0 at East Juyanze are ~927.1, ~926.4, ~923.8, ~923.5, ~920.3, ~917.7, ~914.0, and ~911.9 m, respectively. The exact shoreline elevations of WT-6, WT-5, WT-4, WT-3, WT-2, and WT-1 in West Juyanze are ~926.9, ~923.6, ~920.7, ~918.0, ~917.6, and ~912.0 m, respectively. These elevational measurements may be overly precise, as the elevation of the same constructional shoreline features can vary as much as 3 m depending on factors such as length of fetch, prevailing wind direction, and local geomorphic features (Reheis et al., 2014). In small basins, such as the Juyanze Basin, however, variation in the elevation of constructional features may be much less than 1 m, with the stable water level located less than 1 m below the barrier bar crests. Both the East and West Juyanze Basins are relatively small, flat, and featureless, and Hörner and Chen (1935), the first to identify and map these shorelines, used a spirit level and found that the shoreline features were ‘exceedingly well preserved’ and ‘horizontal’.

The elevations of shorelines in the West Juyanze Basin are consistent with those in East Juyanze. The elevation of 926.9 m at WT-6 in West Juyanze is similar to the elevations of ET-7 of 927.1 m and ET-6 926.4 m in East Juyanze, and the 923.6-m shoreline at WT-5 in West Juyanze shoreline is consistent with the 923.9- and 923.4-m shorelines at ET-5 and ET-4 in East Juyanze. The 920.7-m elevation shoreline at WT-4 in West Juyanze corresponds to the 920.2-m shoreline at ET-3 in East Juyanze. The 918.0- and 917.6-m shorelines at WT-2 and WT-3 are similar to the 917.7-m shoreline at ET-2. The lowest shoreline elevation is also consistent between both Juyanze Basins (e.g. 912.0 m in West Juyanze and 911.9 m in East Juyanze). The consistency of the paleolake shoreline elevations in East and West Juyanze is quite good considering measurement error, although the number of shoreline is different (e.g. eight lines of shorelines in East Juyanze and six lines of shorelines in West Juyanze). However, because a single-lake shoreline may be represented by multiple shoreline features at slightly different elevations at different locations, we only consider the dominate shoreline here. In short, in combining the shorelines in East and West Juyanze, we consider the dominate shorelines in Juyanze Basin to be at ~927, ~924, ~920, ~918, ~914, and ~912 m a.s.l. These are marked in Figure 2b and c for shorelines in East Juyanze and West Juyanze, respectively.

Chrono-stratigraphy

In order to reconstruct the ages of shorelines around the Juyanze Basin, we trenched sections ET-7, ET-5, and ET-2 in East Juyanze using power equipment. These trenches were cut across shoreline features at 927, 924, and 918 m, and all three trenched sections reached bedrock (sections ET-5 and ET-2 are shown in Figure 6). Smaller, hand-dug, trenches were excavated at section ET-6 on the 926-m East Juyanze shoreline and at sections WT-4, WT-2, and WT-1 on the 920-, 918-, and 912-m elevation shorelines in West Juyanze. The ET-7 section (41°51′16.37″N, 101°58′07.93″E, 927.05 m) in East Juyanze is ~6.1 m in depth and consists of near-shore eolian sand and sand/gravel cycles. Only the top 1.7 m of the well-sorted uncompacted sand and gravels are Holocene in age, and they rest atop an unconformity separating them from much older deposits (Figure 7). Deposits below 1.7 m formed during MIS 5 (Li et al., under review) are not emphasized in this study. The initial Holocene sediments consist of shoreline sands and gravels in a barrier bar. Eolian sand was deposited on the lee side of this bar during a period of stability or minor regression, with subsequent shoreline sands and gravel, and a second layer of eolian sand deposited above this minor lake fluctuation. Snail shells are widely distributed at 90–100 cm depths in the section. Quartz OSL ages for near-shore eolian sand at depths of 70 and 122 cm are 1.5 ± 0.2 and 2.9 ± 0.2 ka, respectively (Figure 7).

Figure 6.

Lithology and OSL ages for sections ET-5 and ET-2 in the East Juyanze Basin: (a) ET-5 section at 984 m and (b) ET-2 section at 918 m.

Figure 7.

Shoreline distribution in transect from West to East Juyanze, and lithology and chronology of all sections from shorelines in the Juyanze Basin.

As shown in Figure 7, the ET-6 section (41°51′15.76″N, 101°58′05.28″E, 926.37 m) in East Juyanze consists of 1.7 m of sand and gravel, with a thick beach sand layer occurring between 1.4 and 0.9 m. Quartz OSL ages for samples from shoreline sands at depths of 90 and 140 cm are 1.4 ± 0.1 and 0.9 ± 0.1, respectively.

The ET-5 section (41°51′18.71″N, 101°57′51.87″E, 923.9 m) in East Juyanze is ~550 cm deep and reached sandstone bedrock. Carbonate-cemented yellowish sand and gray near-shore eolian sand occur at 425–550 cm in the section (Figures 6 and 7). The upper part of this unit (425–500 cm) is white carbonate compacted medium coarse sand, with a 2- to 3-cm-long scalloped wave structure suggesting a near-shore deposition. The lower part (500–550 cm) consists of fine blue-gray clayey sand at 500–520 cm and yellowish medium coarse sand. At 425–310 cm in the section, numerous lenses of horizontally bedded medium to coarse eolian sand are interlayered with shoreline gravels and sandy gravels suggesting rapid short-term lake fluctuation. Sandy beach gravel with foreset bedding at a 10–20° orientation occurs at 310–270 cm. Horizontally interbedded eolian sand and poorly sorted sandy gravel of 2–3 cm diameter are at 270–210 cm. The top 210 cm of the section consists of well-sorted sandy gravel. Quartz OSL ages for samples collected at depths of 80, 100, 232, 435, and 534 cm are 1.1 ± 0.1, 1.0 ± 0.1, 1.0 ± 0.1, 2.0 ± 0.1, 2.9 ± 0.2 ka, respectively.

The ET-2 section (41°51′23.50″N, 101°57′0.13″E, 917.7 m) in the East Juyanze is 410 cm deep and penetrated bedrock. Holocene sediments in the top 210 cm above bedrock consist of near-shore eolian sand and sheet-washed gravels at 210–175 cm, foreset sandy gravels at 175–130 cm (but thinning to the east), and alternating well-rounded sandy gravel from 130 to 0 cm (Figure 6). Quartz OSL ages for samples collected from shore eolian sand at 205 cm and from sandy shoreline gravels at 60 cm are 4.6 ± 0.3 and 3.3 ± 0.2 ka, respectively (Figure 7).

In West Juyanze, the WT-4 section (41°54′15.66″N, 101°45′25.17″E, 920.7 m) is located near the threshold between East and West Juyanze and is ~1.5 m deep. It consists of sandy gravels in the top 30 cm of the section and pure gravels in the middle and bottom of the section (Figure 7). An OSL sample was collected from a 20-cm-thick shoreline sand layer at 120–140 cm and dated to 1.0 ± 0.1 ka. The WT-2 section (41°54′18.22″N, 101°45′15.32″E, 917.63) in West Juyanze consists of 1.5 m of shoreline sand and gravels. The gravels are well sorted with an uncompacted structure. Fresh snail shells occur at different depths in the section (Figure 7). An OSL sample was collected from a predominately sand unit at 120 cm and dated to 1.0 ± 0.1 ka. The WT-1 section (41°54′19.03″N, 101°45′0.02″E, 912 m) is on the lowest shoreline that can be found in both East and West Juyanze and consists of 1 m of sand and gravels (Figure 7). The base of this section was not reached due to the uncompacted nature of the beach gravels. Fresh snail shells occur at different depths in the section. The quartz OSL age for sample collected at 60 cm is 0.3 ± 0.1 ka.

Discussion

Paleolake evolution in the Juyanze Basin during the Holocene

As shown in Table 1, the DGPS measurement results for shorelines in East Juyanze indicate the elevations of shorelines are 927.1, 926.4, 923.9, 923.5, 920.2, 917.7, 914.0, and 911.9 m from the lake margin to the center of the basin, with the lowest of these shorelines at least ~12 m above the modern basin floor of the West Juyanze sub-basin (~900 m a.s.l.) and ~22 m above the lowest modern basin floor in the Tia E Hu lake area (~890 m a.s.l.). The shoreline features at these elevations indicate that a late Holocene paleolake fluctuated frequently between 912 and 927 m.

The 918-m shoreline is well preserved around East Juyanze and can be clearly seen in remote sensing images (Figure 2a). The OSL ages of beach sand at 0.6 m and near-shore eolian sand interbedded within shoreline gravels at 2.05 m in the ET-2 section are 3.3 ± 0.2 and 4.6 ± 0.2 ka, respectively (Figure 7), indicating the lake reached at least ~918 m (corresponding to a water depth of 28 m) at around ~4.6 ka and that a lacustrine environment with water fluctuating around that depth existed between ~4.6 and 3.3 ka.

The quartz ages of near-shore eolian sand, bracketing sandy gravel beach sediments at 0.7 and 1.22 m at the ET-7 section on the 927-m shoreline, are 1.5 ± 0.2 and 2.9 ± 0.2 ka, indicating the lake rose to that level shortly after ~2.9 ka and again after ~1.5 ka. The quartz OSL ages at 0.9 and 1.4 m of the ET-6 section on the ~926-m shoreline are 1.4 ± 0.1 and 0.9 ± 0.1 ka, respectively. Although the two age estimates are slightly reversed in relation to depth, these two ages can be considered to be identical considering OSL dating system errors for samples less than 2 ka. They suggest the lake fell slightly to ~926 m at ~0.9 ka. Reliable quartz OSL ages at depths of 534, 435, 232, 100, and 80 cm for samples from the ET-5 section on the 923.9-m shoreline are 2.9 ± 0.2, 2.0 ± 0.1, 1.0 ± 0.1, 1.0 ± 0.1, and 1.1 ± 0.1 ka, respectively (Figure 7). Although the ages of ~1.0 and ~1.1 ka are also reversed in relation to depth, they too can be considered to be identical. These ages for the ~924-m shoreline indicate that a lake reached that elevation at ~2.9 ka. It may have reached an even higher elevation (possibly ~927 m at site ET-7) after ~2.9 ka because the deep-water deposition is between 2.9 and 2.0 ka. The lake regressed after ~2.0 ka and reached a level of ~923 m at ~1.0 ka (Figure 7). Based on all these age estimates, we conclude that lake levels in East Juyanze fluctuated frequently between 924 and 927 m (corresponding to a lake depth of ~24–27 m) during the ~2.9- to 1.0-ka period.

The 927.05-m elevation shoreline in both East Juyanze and West Juyanze is obvious in remote sensing images (Figure 2b). The shoreline at 923.6 m extends more than 5 km in East Juyanze, and a shoreline with a similar elevation of ~923.6 m can be found in West Juyanze near the threshold between the basins, indicating a paleolake at ~924 m occurred in both basins, with a possible connection in the southern Juyanze Basin. We did not identify any morphologically recent, Holocene age, shorelines above 924 m in the Sogo Nur or Gaxun Nur Basin despite intensive surface examination of the basin margins. The absence of relatively recent shoreline features at these elevations in the western sub-basins suggests that the Holocene high stand was restricted to the Juyanze Basin, with perhaps some overflow to lower elevations in the Sogo Nur Basin. More study is needed to confirm this.

Lake Tia E Hu is a small seasonal lake located at the lowest point in the central West Juyanze Basin at ~890 m. DGPS measurement results indicate that paleolake shorelines in the basin are at ~926.9, ~923.6, ~920.7, ~918.0, ~917.6, and ~912.0 m (Table 1), suggesting late Holocene lake depths in West Juyanze were between ~22 and 37 m (Figure 7). The OSL ages of sand at ~1.3 m in the WT-4 section on the 920.7-m shoreline and at ~1.2 m in the WT-2 section from the 917.6-m shoreline are 1.0 ± 0.1 and 1.0 ± 0.1 ka, respectively. The OSL age of sand at a depth of ~60 cm in the lowest 912 m shoreline in West Juyanze is 0.3 ± 0.2 ka. These OSL ages indicate lake levels in West Juyanze fluctuated between 912 and 921 m (corresponding to lake depths of ~22–31 m) during ~1.1–0.3 ka. The threshold elevation between East and West Juyanze is about ~914 m (Google earth elevation), and shorelines below that elevation cannot be traced between the two basins. This suggests that lakes in the two basins at the lowest shoreline elevations were not connected and that one or both basins may have been fed by groundwater rather than by river flow below that elevation. According to the historical record, Juyanze dried during the Yuan (AD 1271–1368) to middle Ming (AD 1368–1644) dynasties (Liu, 1992).

Wünnemann and Hartmann (2002) measured a series of shorelines on the northern margin of West Juyanze by DGPS and report six shorelines occurring at ~926, ~925, ~921, ~917, ~913, and ~912 m. These elevations are consistent with our own DGPS measurements, considering the same shoreline may quite commonly have a 1- to 2-m difference at around lake. They report a ¹⁴C age estimate on Corbicula fluminea shells from the 917-m shoreline is ~5300 cal. BP (4590 ± 420 a BP). This is similar to OSL ages of ~4.6 and ~3.3 ka at the 918 shorelines at East Juyanze; however, it is more than 4000 years earlier than the age of ~1.0 ka for the OSL samples we obtained from the 917.63-m shoreline in West Juyanze. The ¹⁴C age of Corbicula fluminea shells from the 913-m shoreline in West Juyanze is ~2862 cal. BP (Wünnemann et al., 1998; Wünnemann and Hartmann, 2002), which is also different from the OSL age of ~1.0 ka reported here. These inconsistencies between the ¹⁴C ages of shells and the OSL ages of shoreline deposits may be the result of a reservoir effect of ¹⁴C dating for shells (Björck et al., 1991; Colinan et al., 1996; Wang et al., 2002). Also, the systematic errors of OSL dating for younger samples may contribute to this offset between radiocarbon dating and OSL dating ages.

Figure 8 shows the ages of Juyanze Basin shorelines and their corresponding elevations, as well as the reconstructed general lake-level fluctuation history during middle and late Holocene. These age/ elevation data suggest a stable lacustrine environment appeared in the Juyanze Basin ~4.6 ka and that lake levels stabilized at ~918 m (~28 m lake depth) at ~4.6–3.3 ka. The lake level rapidly increased to the ~926-m shoreline (~36 m lake depth) from ~3.3 to 2.7 ka and then decreased to the 920-m shoreline (~30 m lake depth). The paleolake water level gradually increased subsequently and reached a high lake level of 926–927 m (~36–37 m lake depth) at ~1.5–1.0 ka, with lake-level fluctuations occurring between ~924 and 927 m (~34–37 m lake depth) from ~1.5 to 1.0 ka. The lake then regressed rapidly in the Juyanze Basin from 1.0 to 0.3 ka, with lake levels ranging from ~921 down to ~912 m (~22–31 m lake depth). The absence of shoreline features dating to after ~0.3 ka implies the paleolake dried up after 0.3 ka, with only seasonal lakes remaining in the basin, as confirmed by the historical record. The reconstructed lake margins, based on modern a Digital Elevation Model (DEM), for water levels of 918, 924, and 927 m are shown in Figure 9. Lake-level fluctuations during the Holocene are also recorded in the lithology of drill core JYZ11A (Li et al., 2014). The OSL dating chronology and lithology of the 27-m drill core suggest that an arid environment characterized by alluvial and eolian sand prevailed in the Juyanze Basin during the Last Glaciation and Early Holocene. The chronology, together with a proxy index of grain size for gray silty clay alternating with fine-to-medium eolian sand in the top 8.9 m of the core, indicates that a shallow lake environment appeared only after ~6.7 ka and then alternated at least in East Juyanze with frequent periods of desiccation and eolian sand deposition until ~3 ka. A stable lake environment occurred in the region after ~3 ka.

Figure 8.

Reconstructed lake-level fluctuation history in the Juyanze Basin during the Holocene. The elevation of the plotted ages is the absolute elevation of the OSL samples. OSL age in the figure is shown with 1σ error bars.

Figure 9.

Reconstructed lake margins in the Juyanze Basin at water levels of 918, 924, and 927 m based on modern DEM data.

Possible forcing mechanisms of lake-level change in the Ejina Basin

The lake environment which appeared at ~4.6 ka in the Juyanze Basin may have been influenced by regional warm and moisture climatic conditions during the middle-to-late Holocene. Pollen from an East Juyanze lake sedimentary record indicates that the highest values of Chenopodiaceae, Ephedra fragilis, and other desert indicating taxa occur at ~10,700–5400 cal. BP, suggesting a relatively arid climate during the early Holocene (Herzschuh et al., 2004). More favorable conditions are interpreted to have occurred between 5400 and 3900 cal. BP on the basis of a relative increase in the abundance of Artemisia pollen (Herzschuh et al., 2004), suggesting the middle-to-late Holocene lake in the Juyanze Basin may be related to more moderate climatic conditions. The OSL chronology of an eolian sand and lacustrine sequence from the 27-m JYZ11A drill core in East Juyanze also suggests an arid environment, characterized by eolian and alluvial sand deposition, prevailed during the early Holocene and that a lacustrine environment occurred sometime after ~6.7 ka (Li et al., 2014). This study has shown paleolake levels increased during the last half of the middle Holocene and reached their highest levels at ~2.9–1.0 ka during the late Holocene when several lake-level fluctuations occurred.

The Juyanze paleolake was filled by Heihe River. This river originates in the Qilian Mountains on the northern TP margin, and lake formation and evolution in the Juyanze Basin are determined by precipitation in the mountains and by hydrologic conditions along the Heihe River. Temperature change–induced evaporation fluctuations in the Ejina Basin also have a direct impact on lake-level fluctuations. Due to the complex geomorphology of the Ejina Basin, using shoreline and/or drill core data as proxy records for climate change is problematic. As the Heihe River shifts back and forth across its fan/delta, it alternately fills its Gaxun Nur, Sogo Nur, and Juyanze Basins, with some flow probably reaching all the sub-basins simultaneously via both surface and groundwater flow. These sub-basins all have different basin floor elevations and surface areas, and the same amount of river flow combined with the same evaporation rate will fill the sub-basins to different levels. For example, using a DEM, we calculated the surface area of the Gaxun Nur/Sogo Nur Basin and the Juyanze Basin at 927 m a.s.l. to be ~3073 and ~1863 km², respectively. As a result, a water budget that would fill the Juyanze Basin to overflowing would not come close to filling the Gaxun Nur/Sogo Nur joint basin to the same level. This means that under similar climatic conditions, when the Heihe River is flowing into the Juyanze Basin, it will fill and overflow much more quickly than the other way around. Using DGPS, we measured a transect across the threshold separating the Gaxun Nur/Sogo Nur Basin from the Juyanze Basin and found the low point of that threshold to be ~924 m.

We think it likely that when the Holocene lake in the Juyanze Basin created the 927-m shoreline barrier bar, it may have had a still-water lake level close to this ~924 m threshold level, especially considering that the threshold is covered by an extensive dune field. In closed-basin lakes, the surfaces of barrier bars are often 1–3 m above their associated still-water lake levels (Reheis et al., 2014). Assuming the 924-m threshold has not changed significantly during the Holocene, it is possible that lake high stands may have reached this threshold on a number of occasions, with storm-driven waves reworking older material each time. The 926- to 927-m shoreline feature dating to 1–3 ka may, thus, only be the most recent of these events. When the water level of Juyanze Lake is less than the 924-m threshold (e.g. varying between 916 and 924 m), the water depth of lake in Juyanze Basin is between 26 and 34 m. At these levels, conditions must have been very different compared to Gobi Desert conditions found in the Juyanze Basin at present, and an enhancement of Heihe River runoff may have been responsible for these differences. Interpreted at face value, Holocene high stands in the Juyanze Basin at 1–3 ka are substantially out-of-phase with numerous other lake records from the northeastern TP margin, in particular terminal lakes draining from the Qilian Mountains (e.g. Qinghai Lake (14.1–6.5 cal. ka BP), Shen et al., 2005; Ji et al., 2005; Chaka Salt Lake (11.4–7.2 cal. ka BP), Liu et al., 2008; Dalianhai Lake (9.4–3.9 ka), Cheng et al., 2013; Naleng Lake (10.7–4.4 cal. ka BP), Kramer et al., 2010; and Tengger lakes (8.0–5.0 ka), Long et al., 2012; Figure 10a–e). The high stands period of the lake in the Juyanze Basin is also inconsistent with a decrease in Asia summer monsoon (Figure 10f and g) and Westerlies during late Holocene (Figure 10h). Climatic interpretations of Holocene drill core and shoreline records from the Gaxun Nur, Sogo Nur, and Juyanze Basins also differ markedly from one another (e.g. Chen et al., 2003; Hartmann and Wünnemann, 2009; Herzschuh et al., 2004; Mischke et al., 2002; Wünnemann et al., 1998). This apparent lack of interpretive uniformity is almost to be expected given the interplay between the shifting Heihe River and the complex geomorphology of Ejina Basin at its lower levels. Barring dramatic tectonic changes in the basin floor at levels above the ~924-m threshold, these interpretive difficulties are resolved, and the Ejina Basin lake history becomes a useful climatic proxy record. Below that level, however, climatic interpretations are fraught with difficulties.

Figure 10.

The fluctuation of lake levels around the Ejina Basin during Holocene: (a) Juyanze Basin (this study), (b) Baijian Lake in the Tengger Desert (Long et al., 2012), (c) Huahai Lake in the Qilian Mountains (Wang et al., 2013), (d) Chaka Salt lake in the Qaidam Basin (Liu et al., 2008), (e) Qinghai Lake on the Tibetan Plateau (Liu et al., 2015), (f) summer monsoon index derived from a Qinghai Lake sedimentological drill core record (An et al., 2012), (g) Dongge cave stalagmite record (Wang et al., 2005), and (h) moisture index for the Westerlies dominated arid central Asia (Chen et al., 2008).

According to historical records, the lakes in the Juyanze Basin were dry from the end of Yuan Dynasty (AD 1271–1368) to beginning of the Ming Dynasty (AD 1368–1644; Liu, 1992). The lake disappearance and the formation of the modern Gobi Desert environment may have occurred in response to hydrological changes (e.g. groundwater replaced Heihe River flow as the dominant input during a regressive stage of East Juyanze Lake; Mischke et al., 2005). It may also be related to human activity along the upper and middle Heihe River valley since there has been extensive agriculture development and military construction along the Hexi Corridor beginning ~1 ka and continuing to the present (Zhang et al., 2010).

Conclusion

Geographic investigations and DGPS measurements were applied to seven well-preserved shorelines in East Juyanze and six shorelines in West Juyanze. The elevations of shorelines in both East Juyanze Basin and West Juyanze Basin vary from 927 to 912 m (the basin floor is at ~890 m elevation). OSL dating was successfully applied to beach sand samples and near-shore eolian sand samples from sections along these shoreline elevations. The dose recovery test and internal checks for a typical sample show that the SAR protocol was reliable for samples in this study, and the reliability of the OSL dating chronology is further confirmed by the age–depth relationship.

A ~26-m deep lake covered Juyanze Basin beginning at ~4.6 ka, with lake levels gradually increasing to ~33 m deep. Lake levels peaked at the beginning of the late Holocene, ~2.9 ka, with fluctuations occurring at ~2.0 ka. The lake continued at higher water levels (~32 m deep) during the late Holocene, with a lake environment lasting until ~1 ka in the Juyanze Basin. A lake environment disappeared in the Juyanze Basin after ~0.3 ka, which is generally consistent with historical record. The geomorphological shifts of the Heihe River channels across its fan/delta, feeding the three sub-basins of the Ejina at different times, may be responsible for lake formation and water-level fluctuations during ~4.6–1 ka in the Juyanze Basin. The effects of climatic changes on lake evolution in the Juyanze Basin need further study, but it is clear that the use of local shorelines to indicate climate or moisture changes in complex lake basins with multiple large sub-basins requires an understanding of the evolution of the entire lake basin.

Footnotes

Funding

This study was supported by NSFC grant nos 41372180, 41302143, and 41371220 and Fundamental Research Funds for the Central Universities (0009-2014G2270012).

References

Aitken

(1985) Thermoluminescence Dating. London: Academic Press, 359 pp.

Aitken

(1998) An Introduction to Optical Dating. Oxford: Oxford University Press.

Colman

Zhou

(2012) Interplay between the Westerlies and Asian monsoon recorded in Lake Qinghai sediments since 32 ka. Scientific Reports 2: 619.

Banerjee

Murray

Bøter-Jensen

(2001) Equivalent dose estimation using a single aliquot of polymineral fine grains. Radiation Measurements 33: 73–94.

Björck

Hjoft

Ingolfsson

(1991) Radiocarbon dates from the Antarctic Peninsula region: Problems and potential. Quaternary Proceedings 1: 55–65.

Cable

French

(1943) The Gobi Desert. London: Hodder & Stoughton.

Chen

Holmes

(2003) A mid-Holocene drought interval as evidenced by lake desiccation in the Alashan Plateau, Inner Mongolia China. Chinese Science Bulletin 48: 1401–1410.

Chen

Yang

(2008) Holocene moisture evolution in arid central Asia and its out-of-phase relationship with Asian monsoon history. Quaternary Science Reviews 27: 351–364.

Cheng

Chen

Zhang

(2013) Palaeovegetational and palaeoenvironmental changes since the last deglacial in Gonghe Basin, northeast Tibetan Plateau. Journal of Geographical Sciences 23: 136–146.

10.

Colinan

Jones

Rubin

(1996) AMS radiocarbon analyses from Lake Baikal, Siberia: Challenges of dating sediments from a large, oligotrophic lake. Quaternary Geochronology 15: 669–684.

11.

Cooke

(1970) Stone pavement in deserts. Annals of the Association of American Geographers 60: 560–577.

12.

Duller

GAT

(2003) Distinguishing quartz and feldspar in single grain luminescence measurements. Radiation Measurements 37: 161–165.

13.

Galbraith

Roberts

Laslett

(1999) Optical dating of single and multiple grains of quartz from Jinmium Rock Shelter, Northern Australia: Part 1, experimental design and statistical models. Archaeometry 41: 339–364.

14.

Guérin

Mercier

Nathan

(2012) On the use of the infinite matrix assumption and associated concepts: A critical review. Radiation Measurements 47: 778–785.

15.

Hartmann

Wünnemann

(2009) Hydrological changes and Holocene climate variations in NW China, inferred from lake sediments of Juyanze palaeolake by factor analyses. Quaternary International 194: 28–44.

16.

Haude

(1931) Siedlungsmöglichkeiten in Zentral-und Westchina unter besonderer Berücksichtigung der Wasserverhältnisse des Edsingols. Zeitschrift der Gesellschaft für Erdkunde zu Berlin 192–202.

17.

Herzschuh

(2006) Palaeo-moisture evolution in monsoonal Central Asia during the last 50,000 years. Quaternary Science Reviews 25: 163–178.

18.

Herzschuh

Tarasov

Wünnemann

(2004) Holocene vegetation and climate of the Alashan Plateau, NW China, reconstructed from pollen data. Palaeogeography, Palaeoclimatology, Palaeoecology 211: 1–17.

19.

Hörner

Chen

(1935) Alternating Lakes, Some River Changes and Lake Displacements in Central Asia (Hyllningsskrift Tillagnad Sven Hedin, Geografiska Annaler). Stockholm: Svenska sallskapet for antropologi och geografi.

20.

Shen

Balsam

(2005) Asian monsoon oscillations in the northeastern Qinghai–Tibet Plateau since the late glacial as interpreted from visible reflectance of Qinghai Lake sediments. Earth Planet Science Letter 233: 61–70.

21.

Kramer

Herzschuh

Mischke

(2010) Holocene treeline shifts and monsoon variability in the Hengduan Mountains (southeastern Tibetan Plateau), implications from palynological investigations. Palaeogeography, Palaeoclimatology, Palaeoecology 286: 23–41.

22.

Wei

Fan

(2014) The evolution of Juyanze Lake in Ejina Basin since Late Pleistocene. Quaternary Sciences 35: 169–180 (in Chinese with English abstract).

23.

Jin

Duan

(under review) Quartz and K-feldspar luminescence dating revealed Marine Isotope Stage 5 megalake in Juyanze Basin, central Gobi Desert, China. Palaeogeography, Palaeoclimatology, Palaeoecology.

24.

Chen

(2007) OSL dating of sediments from deserts in northern China. Quaternary Geochronology 2: 23–28.

25.

Liu

Lai

Madsen

(2015) Last deglacial and Holocene lake level variations of Qinghai Lake, north-eastern Qinghai–Tibetan Plateau. Journal of Quaternary Science 30: 245–257.

26.

Liu

Dong

Rech

(2008) Evolution of Chaka Salt Lake in NW China in response to climatic change during the Latest Pleistocene–Holocene. Quaternary Science Reviews 27: 867–879.

27.

Liu

(1992) Evolution of the Juyan Lake. Journal of Arid Land Resources and Environment 6: 9–18 (in Chinese with English abstract).

28.

Long

Lai

Fuchs

(2012) Timing of Late Quaternary palaeolake evolution in Tengger Desert of northern China and its possible forcing mechanisms. Global and Planetary Change 92–93: 119–129.

29.

Long

Lai

Wang

(2011) A combined luminescence and radiocarbon dating study of Holocene lacustrine sediments from arid northern China. Quaternary Geochronology 6: 1–9.

30.

Mischke

(2001) Mid and Late Holocene Palaeoenvironment of the Lakes Eastern Juyanze and Sogu Nur in NW China, Based on Ostracod Species Assemblages and Shell Chemistry. Berlin: Selbstverlag Fachbereich Geowissenschaften, FU Berlin.

31.

Mischke

Demske

Wunnernann

(2005) Groundwater discharge to a Gobi desert lake during Mid and Late Holocene dry periods. Palaeogeography, Palaeoclimatology, Palaeoecology 225: 157–172.

32.

Mischke

Fuchs

Riedel

(2002) Mid to Late Holocene palaeoenvironment of Lake Eastern Juyanze (north-western China) based on ostracods and stable isotopes. Geobios 35: 99–110.

33.

Murray

Olley

(2002) Precision and accuracy in the optically stimulated luminescence dating of sedimentary quartz: A status review. Geochronometria 21: 1–16.

34.

Murray

Wintle

(2000) Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol. Radiation Measurements 32: 57–73.

35.

Murray

Wintle

(2003) The single aliquot regenerative dose protocol: Potential for improvements in reliability. Radiation Measurements 37: 377–381.

36.

Natsagdorj

Jugder

Chung

(2003) Analysis of dust storms observed in Mongolia during 1937–1999. Atmospheric Environment 37: 1401–1411.

37.

Prescott

Hutton

(1994) Cosmic ray contributions to dose rates for luminescence and ESR dating: Large depths and long-term time variations. Radiation Measurements 23: 497–500.

38.

Provincial Atlas of Gansu, Lanzhou (1975) 85 maps (in Chinese).

39.

Reheis

Adams

Oviatt

(2014) Pluvial lakes in the Great Basin of the western United States–a view from the outcrop. Quaternary Science Reviews 97: 33–57.

40.

Shen

Liu

Wang

(2005) Palaeoclimatic changes in the Qinghai Lake area during the last 18,000 years. Quaternary International 136: 131–140.

41.

Wang

(2013) Millennial-scale environmental changes in the Asian monsoon margin during the Holocene, implicated by the lake evolution of Huahai Lake in the Hexi Corridor of northwest China. Quaternary International 313–314: 100–109.

42.

Wang

Scarpitta

Zhang

(2002) Later Pleistocene/Holocene climate conditions of Qinghai–Xizang Plateau (Tibet) based on carbon and oxygen stable isotopes of Zabuye Lake sediments. Earth Planetary Science Letters 203: 461–477.

43.

Wang

Zhou

Dong

(2006) Control of dust emissions by geomorphic conditions, wind environments and land use in northern China: An examination based on dust storm frequency from 1960 to 2003. Geomorphology 29: 292–308.

44.

Wang

Hua

Zhang

(2012) Aeolian salts in Gobi deserts of the western region of Inner Mongolia: Gone with the dust aerosols. Atmospheric Research 118: 1–9.

45.

Wang

Xia

Wang

(2008) Dust sources in arid and semiarid China and southern Mongolia: Impacts of geomorphologic setting and surface materials. Geomorphology 97: 583–600.

46.

Wang

Cheng

Edwards

(2005) The Holocene Asian monsoon: Links to solar changes and North Atlantic climate. Science 308: 854–857.

47.

Wintle

(2008) Luminescence dating: Where it has been and where it is going. Boreas 37(4): 471–482.

48.

Wintle

Murray

(2006) A review of quartz optically stimulated luminescence characteristics and their relevance in single-aliquot regeneration dating protocols. Radiation Measurements 41: 369–391.

49.

Wünnemann

Hartmann

(2002) Morphodynamics and Paleohydrography of the Gaxun Nur Basin, Inner Mongolia, China. Zeitschrift fur geomorphologie supplement band 126: 147–168.

50.

Wünnemann

Hartmann

Altmann

(2007a) Interglacial and glacial fingerprints from lake deposits in the Gobi Desert, NW China. Developments in Quaternary Sciences 7: 323–347.

51.

Wünnemann

Hartmann

Janssen

(2007b) Responses of Chinese desert lakes to climate instability during the past 45,000 years. Developments in Quaternary Sciences 9: 11–24.

52.

Wünnemann

Pachur

(1998) The chronology of Pleistocene and Holocene lake level fluctuations at Gaxun Nur/Sogu Nur and Baijian Hu in Inner Mongolia, China. Petermanns Geographische Mitteilungen 142: 191–206.

53.

Yang

Chen

Chun

(2008) The Jilantai Salt Lake shorelines in northwestern arid China revealed by remote sensing images. Journal of Arid Environments 72: 861–866.

54.

Kelts

(2002) Abrupt changes in climatic conditions across the late-glacial/Holocene transition on the N.E. Tibet–Qinghai Plateau: Evidence from Lake Qinghai, China. Journal of Palaeolimnology 28: 195–206.

55.

Yang

(2006) Millennial-scale Holocene climate variability in the NW China drylands and links to the tropical Pacific and the North Atlantic. Palaeogeography, Palaeoclimatology, Palaeoecology 233: 149–162.

56.

Zhang

Peng

(2004) Late quaternary palaeolake levels in Tengger Desert, NW China. Palaeogeography, Palaeoclimatology, Palaeoecology 211: 45–58.

57.

Zhang

Guo

Wang

(2010) Research on the lakes change in Ejin alluvial fan from long time-series landsat images. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Honolulu, HI, 25–30 July, pp. 374–377. New York: IEEE.

58.

Zhou

Chen

Wang

(2009) Temporal change of radiocarbon reservoir effect in Sugan Lake, northwest China during the late Holocene. Radiocarbon 51(2): 529–535.