Oasis evolution processes and mechanisms in the lower reaches of Heihe River,Inner Mongolia,China since 1 ka ago

Abstract

Two major oases in the lower reaches of Heihe River, Inner Mongolia of China, the desertified ancient Juyanze oasis (AJO) and the modern Ejina oasis (MEO), experienced dramatic environmental changes and human migrations as documented in historic records. The processes and mechanisms of their evolution are the key for us to understand the environmental changes and human activities in this region, yet are still unclear. This study provides an evolution chronology of these two oases by analyzing the optical dating records from MEO and ¹⁴C dating records from AJO and discusses their evolution mechanisms based on drainage network analysis and field investigations. The optical dating samples collected from aeolian sands below humus layers in Populus euphratica woodlands on MEO indicate that this oasis developed between 1.05 and 0.45 ka ago, while the ¹⁴C dating samples collected from a number of dead P. euphratica trees from AJO consistently suggest that the complete desertification of AJO occurred before ~400 cal. a BP. These new dating results imply that the two oases evolved simultaneously but in opposite directions. That is, AJO desertified while MEO developed gradually in the last 1 ka. Climate change occurring over the last 1 ka in northern China cannot explain the opposite evolution directions of these two adjacent oases. The drainage network analysis based on digital elevation models (DEMs) and field observations and the dates (~0.5 ka) of the top lacustrine layer from GaxuNur Lake north of MEO suggest that the evolution of the two oases was mainly governed by Heihe River migration.

Keywords

ancient Juyanze oasis drainage network modern Ejina oasis oases evolution mechanisms optical dating

Introduction

Alluvial fans in the lower reaches of Heihe River (40°50′–42°30′N, 99°40′–101°30′E with elevation 875–950 m) are located in the southwest of the Mongolian Plateau and adjacent to the north Badain Jaran Desert (Figure 1). This region administratively belongs to the Ejina banner (i.e. county) of western Inner Mongolia, China. Located at the margin of the Asian summer monsoon which provides most of the annual precipitation during the summer season (Chen et al., 2003), the area has an extremely arid continental climate, with a long dry season from September to May and a 3-month rainy season from June to August. Climatic data (1957–2010) recorded at Ejina meteorological station showed that the mean annual precipitation is only ~38 mm, while potential evaporation reaches 3700–4000 mm. The mean annual air temperature is 8.9°C, with an average temperature of 27°C in the warmest month July, and −11°C in the coldest month, January. The prevalent wind direction is from west/northwest, and there are 52 strong-wind (≥17 m/s) days annually including 16 sandstorm days. This region is currently considered as one of the four main dust sources in northern China (Wang et al., 2001). This ecologically fragile region has undergone tremendous environmental changes, which led to frequent human migrations as recorded in historic documents (Du, 2003).

Figure 1.

Study area and site map. Rectangle in the insert map of China indicates the location of study area. Triangles are ancient and modern cities. Dash lines outline main river courses. Dots indicate sampling site locations in this study. HYL-1, 2, 3, 4, and 5 are OSL dating sampling sites on MEO. GXN is one section in GaxuNur Lake (Wang et al., 2012), while the study site Ejin was described in Wen et al. (2006). The ‘Strange Forest’ is the ¹⁴C dating sampling site from AJO.

Oases on the alluvial fans have been important human livelihood areas since ancient times. Human activities were documented in this area as early as the Neolithic Age, and agricultural civilization was highly developed during the Han (2.2–1.8 ka BP), Western Xia (1.0–0.8 ka BP), and Yuan (0.8–0.6 ka BP) dynasties (Local Chronicles Compilation Committee of Ejina Banner, 1998). There are two main oases on the alluvial fans, the desertified ancient Juyanze oasis (AJO), with elevations of ~900 m on its lower part and ~950 m on its upper part, and the modern Ejina oasis (MEO), with the elevation range between ~910 and ~945 m (Figure 1). Fixed and semi-fixed dunes are widely distributed in both oases. Populus euphratica woods are developed widely on the MEO, while dead P. euphratica trees are found extensively on the AJO, which makes this region a famous tourist attraction. Located to the west of Juyanze paleo-lake (Figure 1), AJO has been completely desertified and is presently covered by sand dunes. The abandoned human settlements such as ancient Juyan City, Heicheng City, and farmlands imply previous human activities on the ancient oasis. Large-scale cultivations had been developed in the lower part of this oasis from the time of Han Dynasty to support the war with the Hunnish empire (~2.1 ka BP; Zhu et al., 1983). The AJO with the ancient Juyan City (Figure 1) was gradually abandoned by the Han Dynasty after the victory in this war (~1.9 ka BP). Hundreds of years later, another ancient Heicheng City (Figure 1) and farmland were developed by the Western Xia (~1.0 ka BP) and Yuan (~0.8 ka BP) dynasties on the AJO. This ancient oasis was abandoned at the beginning of the Ming Dynasty (~0.6 ka BP) due to water stress (Du, 2003). Located to the south of the GaxuNur Lake and SuguNur Lake, MEO currently is the location of the modern town of the Ejina banner. The region was named as ‘Ejina’ in the Qing Dynasty (~0.3 ka ago) when a Mongolian tribe was permitted to use this oasis as their pasture (Du, 2003). Any cultivation of farmland on MEO was forbidden in this tribe. However, water consumption in this oasis increased greatly due to rapid agricultural developments after 1949 when the P.R. China was established. GaxuNur Lake (Figure 1) desiccated in the 1960s and then the SuguNur Lake started to dry up occasionally, which led to the rapid shrinkage of the MEO (Zhao et al., 2007b). Following the implementation of the water distribution program in the Heihe River basin at the end of the last century, MEO was restored and SuguNur Lake received sufficient water to become permanent again.

The evolution processes and mechanisms of the AJO and MEO during the historic period are essential to understand human activities and environmental changes in this region. Despite a large number of studies of this area, a general agreement has not yet been reached. Zhu et al. (1983) found that the degradation of AJO happened first in its lower part, the area close to the Juyanze paleo-lake, and then gradually extended to its upper part during the historic period, and concluded that large-scale human activities were the main reason for desertification of the ancient cities of Juyan and Heicheng. Using the historic ecological and environmental documents of the lower reaches of Heihe River from the Han to the Tang Dynasty and recent documents, Du (2003) similarly concluded that human over-exploitation was the primary reason for the gradually decline of the AJO civilization. However, sedimentary analysis (Pachur et al., 1995) of one 230-m drilling profile from GaxuNur Lake indicates that a large paleo-lake existed covering the entire area of lakes Juyanze, GaxuNur, and SuguNur between 39 and 21 ka BP and that lacustrine sediments reappeared ~13 ka BP. In addition, the lacustrine and aeolian deposits alternating appearance during the Holocene indicate that the long-term drying is a result of climate changes in this region (Pachur et al., 1995). Wen et al. (2006) applied particle-size analysis and ¹⁴C dating to a stabilized sand dune in the MEO and suggested that the widely distributed stabilized/semi-fixed aeolian sand dunes record the occurrence of dust storms in the past 2.5 ka of this region, coinciding with the northern China ‘dust-rain’ frequency (Zhang, 1984) from 1 ka ago. Therefore, Wen et al. (2006) concluded that climate change was the main reason for the occurrence of dust storms in this region, although human activities accelerated the environmental degradation. From one profile ~10 km to the west of GaxuNur Lake and at an elevation ~12 m higher than the modern lakebed, the deposit sequences record alternative switching between desert and lake landscapes since the late Pleistocene in the study area (Yang et al., 2009). The appearance of the lacustrine deposit between 14 and 4 ka ago indicates a lake expansion and water level rising. However, the climate became arid from 4 ka ago and the lake shrank and disappeared (Yang et al., 2009) as indicated by the deposit of aeolian sediments. Xiao et al. (2004) reviewed existing studies and argued that oasis distribution in the lower reaches of Heihe River transferred from AJO to MEO as a result of changes in the course of the Heihe River (Figure 1). The watercourse of the Heihe River migrated from east to west during the last 2 ka. The terminal lake of Heihe River also migrated from the east Juyanze paleo-lake to the west SuguNur Lake and GaxuNur Lake. However, these arguments still need chronological evidence.

The environmental evolution processes and mechanisms during the historic period in the lower reaches of Heihe River still leave important scientific questions unanswered: (1) What are the evolution processes of the AJO and MEO? Were their evolution processes synchronous or out-of-phase? and (2) What are the mechanisms of their evolution? What is the key driving force of the evolution: human activities, climate change, or Heihe River migration?

During field investigations at AJO and MEO, optically stimulated luminescence (OSL) dating samples from aeolian sands under humus were collected in P. euphratica woodlands on MEO, and ¹⁴C dating samples were collected from the dead P. euphratica trees on AJO. Lacustrine sediments were also dated by OSL and ¹⁴C dating methods to determine the last time the Heihe River entered GaxuNur Lake. This paper analyzes the evolution processes of these two oases based on these dating results. Finally, drainage network analysis using a digital elevation model (DEM) is conducted to understand the evolution mechanisms in this region.

Field investigation

P. euphratica woodlands are widely developed on MEO. Five P. euphratica woodlands from north to south at MEO were selected to collect optical dating samples (Figure 1). At woodland HYL-1 (42°12′19″N, 101° 06′38″E; Figure 2a), about 10 km southwest to SuguNur Lake, a 25- to 30-cm humus top layer was found with rich plant residues (P. euphratica leaves and branches) in fine sand, under which a well-sorted yellow-white fine sand layer exists. In this extremely arid and windy environment with low precipitation, active fluvial is rare and sediments are hardly accumulated and stabilized. Only after the P. euphratica woodlands developed, could the dead leaves and branches be accumulated in the aeolian sands to form a humus layer and then the underlying layer of sediments could be fixed. Therefore, the deposition age of the top of the underlying layer provides a minimum age for woodland development. Woodland HYL-2 (42°10′57″N, 101°05′47″E) is under the largest P. euphratica ‘Holy Tree’ on MEO, and its depositional sequence is similar to HYL-1. Woodlands HYL-3 (42°06′29″N, 101°07′43″E) and HYL-4 (42°06′30″N, 101°03′07″E) are located at the middle part of the oasis. Their depositional sequences are also similar to HYL-1. Woodland HYL-5 (42°01′53″N, 101°03′17″E) is located about 10 km north of the town of Ejina banner with a different depositional sequence from the above four woodlands. There is an additional layer of 40-cm silty-clay lacustrine sediment between the top humus layer and the aeolian sand layer. A ponding period could be inferred prior to the establishment of the P. euphratica woodland. We dug ~60-cm-deep pits in these Populus woodlands sample sites (Figure 3) to obtain a clean profile and horizontally hammered a 4-cm-diameter stainless steel pipe into the profile to collect optical dating samples at the top of the aeolian sand layer. Optical dating results of the first four samples represent the initial formation time of the P. euphratica woodlands because the aeolian sands could be fixed only when the woodlands got developed. The age of HYL-5 represents the start of ponding time at this location. Since the top humus layer should be developed after the woods and may introduce a reworking problem, no OSL/¹⁴C samples were collected from this layer.

Figure 2.

Landscape of sampling sites: (a) section HYL-1 on MEO and (b) ‘Strange Forest’ on AJO where¹⁴C samples were collected.

Figure 3.

Profiles of OSL dating sampling sites. HYL-1, 2, 3, 4, and 5 are sections in P. euphratica woodlands on MEO, and GXN is the section from the GaxuNur Lake. Their locations are shown in Figure 1.

P. euphratica trees seldom survive on the modern AJO. Located on the upper part of AJO, the area with the highest elevation, and ~7 km northwest to the Heicheng City ruins (Figure 1), the scenic spot ‘Strange Forest’, with a total area of >10 km², features thousands of huge dead P. euphratica trees and their remains (Figure 2b). Massive dead P. euphratica trees provide reliable ¹⁴C dating samples to determine their death age. We randomly collected major branches from five dead trees for ¹⁴C dating. If these samples have similar ¹⁴C results, then the trees probably died at the same time for the same reason. Since the P. euphratica trees in this region need groundwater supply to survive (Feng et al., 1998), the death of this forest should be synchronous with the desertification of AJO.

The section GXN is located in the dry lakebed of GaxuNur Lake (Figure 1). A detailed description of GXN can be found in Wang et al. (2012). Riverbed gravels were found in the bottom layer of the section, followed by a layer of flood plain muddy silt and then the shallow lacustrine deposits rich in organic matters gradually appeared at ~1 m of the profile. One OSL sample (GXN-108) and one ¹⁴C dating sample (GXN-100) of organic matter were collected in the lacustrine layer at the depth of 106–110 and 100 cm, respectively (Figure 3).

Dating process and results

Optical dating

As a chronological method, optical dating estimates sediment deposition time by measuring the luminescence signals in the sediment. Since the discovery of the signals of OSL (Huntley et al., 1985) and infrared stimulated luminescence (IRSL; Hutt et al., 1988), OSL and IRSL dating methods have been applied widely to age determination for Quaternary deposition. The single-aliquot-regenerative-dose (SAR) protocol technique for quartz has greatly improved the accuracy of the equivalent dose (D_e) determination (Murray and Wintle, 2000; Wintle and Murray, 2006). The SAR technique for K-feldspar has also been used although it may have underestimation issues (Wallinga et al., 2000).

The OSL dating was conducted in the luminescence laboratory of the Cold and Arid Environmental and Engineering Research Institute, Chinese Academy of Sciences. The luminescence signal detection and the D_e determination procedures were conducted on one RisØ DA-15 OSL/TL reader with a single ⁹⁰Sr/⁹⁰Y β source. The irradiation dose rate of the β source is 0.101 Gy/s. All laboratory procedures of sample preparation and luminescence measurement were carried out in a darkroom with subdued red light. The raw sample at the two ends of sampling tubes was removed for the dose rate measurements. All remaining raw samples were treated with HCl and H₂O₂ to remove carbonate and organic matter. The samples were then sieved in water to select sediments with grain size of 90–180 µm. Heavy liquids with series densities were then used to separate each sample respectively to obtain quartz and feldspar fractions. Both quartz and K-feldspar grains were treated with HF to remove the outer layer irradiated by alpha particles. Since OSL signals from quartz grains of HYL-1 are too dim to obtain reasonable D_e values, IRSL signals from K-feldspar fraction was used in dating this sample. The D_e value was obtained by using the improved SAR procedure (Wallinga et al., 2000) for the IRSL signals. OSL dating was successfully applied to quartz fractions for other samples. Their D_e values were obtained by using the routine SAR procedure (Wintle and Murray, 2006).

The environmental dose rate of every sample was obtained from the radioactive elements’ contents in the surrounding sediments and sample grains and also from cosmic ray contributions. The contributions from radioactive elements were determined mainly by the radioactive decay of the uranium (U), thorium (Th), and potassium (K). In this study, the U and Th concentrations and K contents were measured by means of neutron activation analysis (NAA) and then converted to beta and gamma dose rates based on the conversion factors by Aitken (1998). The environmental dose rate from cosmic rays was calculated according to sample’s altitude and burial depth (Prescott and Hutton, 1988, 1994). The internal dose rate from ⁴⁰K in K-feldspar grains was calculated with a K content of 13 ± 1% (Zhao and Li, 2005). Since water content in sediments affects the external dose rate, it was calculated by the ratio of sample weights before and after drying in an oven for lacustrine sample GXN-108. The water contents of samples HYL-1–HYL-5 were determined between 0.3% and 2% by laboratory measurements. The sampling time was in April, so that before the rainy season, the water content of buried aeolian sands could reach 7% after rain (Chen et al., 1998) and may also be affected by groundwater during high flow period of Heihe River. Due to the uncertainty of sediment water content during their burial period, a water content of 5% was selected in calculating ages for the aeolian sand samples.

Although the IRSL signal from K-feldspar has the disadvantage of anomalous fading for OSL dating (Wallinga et al., 2001), it still has the advantages of high intensity and low measurement error. For samples deposited during the Holocene, the IRSL signal from K-feldspar could be used to get reliable optical dating results when the OSL signal from quartz could not (Li et al., 2011). We applied IRSL dating procedures to the K-feldspar fraction for HYL-1, but no dominant peak was found in the D_e distributions. The frequency distribution of the D_e values of 35 effective aliquots is shown in Figure 4a. The D_e values are distributed evenly in the range of 3.5–15.5 Gy. The reason for the scattered D_e values should be related to poor bleaching before deposition. Watercourses are widely distributed in the MEO; a considerable proportion of aeolian sand grains should come from the nearby riverbeds, which may lead to the poor bleaching problem of these samples. The D_e values of HYL-1 show a nearly linear correlation with the natural IRSL intensities (Figure 4b). This is a typical behavior of poorly bleached samples (Li, 1994; Zhang et al., 2001). It has been well established (Olley et al., 1998) that the minimum D_e values should be selected for the age calculation for this kind of samples. Therefore, the minimum D_e value of the lower four aliquots in Figure 4a and b was selected for age calculation for HYL-1.

Figure 4.

D_e results of two typical samples: (a and b) D_e value distribution of K-feldspar fraction for sample HYL-1 and quartz fraction for sample HYL-5. (c and d) The correlations of the D_e values of the two samples with their natural IRSL/OSL intensities, which are corrected by their sensitivity changes. For the HYL-1 K-feldspar fraction, the D_e values are distributed evenly in the range of 3.5–15.5 Gy and increase linearly with the IRSL intensity. For the HYL-5 quartz fraction, although individual aliquots show their D_e values increase linearly, most of the D_e values are distributed close to 2.5 Gy.

The OSL signals from quartz fractions of other samples are all bright enough to obtain reliable dating results. The frequency distribution of D_e values and their correlation with natural OSL intensities of one typical sample HYL-5 are shown in Figure 4c and d. The depositional environment of samples HYL-2, 3, 4, and 5 was similar to that of HYL-1, while the sample GXN-108 is lacustrine sediment. Only a few aliquots in all the samples were poorly bleached as their D_e values increased linearly with their OSL intensity (Figure 4d), while most D_e values are concentrated in one narrow region, forming a dominant peak, and these aliquots were well bleached (Figure 4b and d). The peak D_e values were selected to calculate these samples’ ages. Table 1 shows all the OSL dating results.

Table 1.

OSL dating results.

Sample	D_e (Gy)	U (ppm)	Th (ppm)	K (%)	Water content (%)^a	Cosmic ray (Gy/ka)	Dose rate (Gy/ka)	Age (ka)
HYL-1	3.68 ± 0.11	2.30 ± 0.10	8.52 ± 0.29	1.60 ± 0.04	5	0.22	3.50 ± 0.13^b	1.05 ± 0.10
HYL-2	2.38 ± 0.24	2.41 ± 0.10	7.92 ± 0.19	1.95 ± 0.05	5	0.22	2.94 ± 0.11	0.81 ± 0.09
HYL-3	1.40 ± 0.21	2.28 ± 0.10	8.92 ± 0.30	1.85 ± 0.04	5	0.22	2.86 ± 0.10	0.48 ± 0.10
HYL-4	1.37 ± 0.12	2.20 ± 0.10	9.10 ± 0.31	1.89 ± 0.05	5	0.22	2.95 ± 0.11	0.46 ± 0.05
HYL-5	2.47 ± 0.22	2.25 ± 0.10	8.10 ± 0.25	2.15 ± 0.10	5	0.22	3.08 ± 0.10	0.78 ± 0.08
GXN-108	1.41 ± 0.14	5.79 ± 0.10	9.24 ± 0.22	1.92 ± 0.05	20	0.22	3.27 ± 0.12	0.43 ± 0.13

OSL: optically stimulated luminescence.

The water content of HYL-1–HYL-5 were set artificially as 5%.

The dose rate of HYL-1 has included its internal dose rate contribution from ⁴⁰K (Zhao and Li, 2005).

¹⁴C dating

As a widely used chronological method, ¹⁴C dating obtains the sample’s age since the sample stopped carbon exchange with atmosphere. Major branches of dead P. euphratica are ideal ¹⁴C dating materials. The ¹⁴C dating in this study was conducted in Key Laboratory of Western China’s Environmental System (Ministry of Education), Lanzhou University. All ¹⁴C dating samples were processed using the standard methods (¹⁴C Laboratory, Institute of Geochemistry, CAS, 1977; Chou, 1990). They were pre-treated using the acid-alkali-acid (HCl 2M, NaOH 0.5%) procedures. Benzene was synthesized using the calcium method after distillation. HIS-3 scintillation fluid was formulated with the prepared benzene and stood for 7 days before counting. Counting was carried out using a 1220 low-background liquid scintillation counter (Finland), with a measurement time of 2000 min. The obtained ¹⁴C conventional age (a BP) was calibrated to calendar year (cal. a BP) using the CALIB 7.02 program (Reimer et al., 2013) with the IntCal13 northern hemisphere terrestrial calibration mode with 2-sigma error range. Table 2 shows ¹⁴C dating results of five collected dead P. euphratica branches (HHY-1, 2, 3, 4, and 5) from the ‘Strange Forest’ on AJO and of an organic matter sample (GXN-100) from the lacustrine sediment in GaxuNur Lake.

Table 2.

¹⁴C dating results.

Samples	Type	¹⁴C age (a BP)	cal. a BP (2 sigma)
HHY-1	Tree branches	265 ± 41	405 ± 58
HHY-2	Tree branches	361 ± 60	407 ± 102
HHY-3	Tree branches	636 ± 41	609 ± 90
HHY-4	Tree branches	355 ± 42	406 ± 92
HHY-5	Tree branches	1922 ± 50	1843 ± 109
GXN-100	Organic matter	562 ± 55	584 ± 70

Discussion

Oases evolution process

MEO development in the last 1 ka

The development of the MEO would have been accompanied by the flourishing of P. euphratica forests. The OSL age (1.05 ± 0.10 ka) of the aeolian sands under the humus layer in P. euphratica woodland HYL-1 in the lower part of MEO suggests the starting time of the woods’ establishment and MEO development. This age is also supported by previous studies. Wen et al. (2006) obtained the ¹⁴C age of a fixed aeolian dune at MEO and concluded that 1020–1460 a BP was a sandstorm-active period with major brown coarse sand (>100 µm) deposition. They also found the period of 0.3–1.0 ka BP was mainly dominated by fine material (<50 µm) deposition. Our OSL results suggest that the environment of MEO was substantially improved from ~1.0 ka BP.

According to the local Scenic Introduction brochure, the biggest P. euphratica tree on the northern part of MEO called the ‘Holy Tree’ has an age of 880 years. This age is unavailable in any academic publications. However, based on the correlation of the growth height with tree-ring data of P. euphratica from the lower reaches of Heihe River (Li et al., 2010), the age of the 27-m-high ‘Holy Tree’ could be estimated as 300–600 years. According to one Mongolian tribe’s immigration history to the MEO documented in the Qing Dynasty (~0.3 ka BP; Du, 2003), the age of the ‘Holy Tree’ should be much older than 300 years. The OSL age of HYL-2 under the ‘Holy Tree’ is 0.81 ± 0.09 ka, indicating that there was plenty of water to support this tree since ~800 years ago. Samples HYL-3 and HYL-4 were collected ~10 km apart at the middle part of MEO (Figure 1). They have similar OSL ages of 0.48 ± 0.10 ka and 0.46 ± 0.05 ka, respectively, implying that P. euphratica woods were widely established in the middle part of MEO 400–500 years ago. The age of aeolian sands at HYL-5 in the upper part of MEO is 0.78 ± 0.08 ka. The silty-clay layer above the aeolian sands indicates a ponding environment afterward. The dating results suggest that this location was still in an aeolian landscape ~800 years ago before being covered by water bodies and that P. euphratica woods developed after the water retreated. The sampling site HYL-5 (Figure 1) is close (<1 km) to a lowland basin that is occasionally flooded today.

All the above OSL dating results suggest a gradual development process of MEO from 1.05 ka ago in the north (i.e. lower parts) to 0.45 ka in the south (i.e. middle parts). Several dendrochronological studies (Peng et al., 2013; Zhang et al., 2012) in the MEO found some P. euphratica trees to have ages of about 300 years, and the growth of this kind of tree depends strongly on stream runoff of the Heihe River in this extremely arid area. These studies also indicate that there was plenty of groundwater to support the P. euphratica growth and MEO development at least before 300 a BP.

AJO desertification

The ¹⁴C dating results from five dead P. euphratica trees in the ‘Strange Forest’ on AJO are generally similar (Table 2). Only one tree died at about 1800 cal. a BP, another one at ~600 a BP, whereas the other three trees all died at ~400 a BP, implying that most P. euphratica trees in this forest suffered a sudden death about 400 years ago. P. euphratica trees grow well when the groundwater level is above 4 m and become stressed with leaf blight when the groundwater level drops to 4–6 m (Zhang et al., 2004). Most cannot survive when the groundwater level is between 6 and 10 m, and none can survive when the groundwater level falls below 10 m (Zhang et al., 2004). The dating results suggest that the groundwater level of this P. euphratica forest decreased significantly during the period 600–400 years ago, leading to the death of the forest. Since this P. euphratica forest was located at the upper part of AJO, the timing of the death of this forest should be synchronous with the desertification period of AJO.

The ‘Strange Forest’ and the ruin of the Heicheng City are about 5 km apart. The death of this forest (ca. 400–600 a BP) happened close to the historically documented abandonment (~600 a BP) of Heicheng City. It can be inferred that the main reason for the abandonment of Heicheng City was similarly due to the water deficiency. The ancient city was likely abandoned when the groundwater level fell too far to support agriculture on AJO. Two ¹⁴C ages (525 ± 40 cal. a BP, 490 ± 60 cal. a BP) were obtained from the bottom of an aeolian sand dune located at the Wuta temple in Heicheng City (Xiao et al., 2004). These two ages are similar to the age of the sand dune at the same location estimated from the forest floor of the Tamarisk (Zhu et al., 1983). It can be inferred that the aeolian sand dunes developed widely at ~500 years ago in Heicheng City and that AJO desertified completely after that time.

From the above dating results in the two oases, we can conclude that AJO and MEO evolved simultaneously during the last ~1 ka but in opposite directions, with MEO developing and AJO desiccating. Agriculture was well developed during the Han Dynasty (~2000 a BP) at the lower AJO. Farming lands were mainly located in the upper oasis during the Western Xia (~800 a BP) and Yuan (~700 a BP) dynasties. The desertification was also extended from the lower to the upper AJO from the Han to the Yuan dynasties (Zhu et al., 1983). At about 400 years ago, the entire AJO was desertified as suggested by the death of the P. euphratica forest, human abandonment of the ancient city, and migration away from the oasis. In contrast, the lower part of the MEO started to develop ~1 ka ago, and there was sufficient water to support the ‘Holy Tree’ ~800 years ago. MEO developed to its middle part (with an elevation of ca. 920 m) about 500 years ago with thriving P. euphratica woods. However, humans did not migrate from AJO to MEO directly 400–500 a BP for agriculture, and husbandry on MEO was developed ~300 a BP.

Oasis evolution mechanisms

Climate background in northern China

From a sandstorm frequency analysis for northern China (Zhang, 1984), there was a low occurrence period around 1050–1450 a BP. Climate fluctuations during the past 2000 years based on Chinese historical documentary records also show a humid climate between 950 and 1200 a BP in east China (Zhang and Lu, 2007). Tree-ring widths (Liu et al., 2005) and the δ¹⁸O values recorded in Dunde ice core (Yao and Thompson, 1992) from Qilian Mountain recovered temperature variation in the northeast Tibetan Plateau over the last millennium. They recorded a warm period between 900 and 1200 a BP (Yang et al., 2003), and the tree-ring data (Kang et al., 2002) also reconstructed that the Heihe River in the mountain area had a high runoff period between 1050 and 1250 a BP. The climate was warm/humid across the northern China and in the area of NE Tibetan Plateau ~1 ka BP, and the Heihe River had high-level runoff. During the warm/humid period across much of northern China, the Heihe River also had high-level runoff. The agriculture on AJO was also well developed during this period, whereas MEO presented a desert landscape with aeolian sand dune deposits.

A sandstorm prevalent period between 300 and 1000 a BP in northern China (Zhang, 1984) indicates a major drought event occurred starting 1 ka ago. A dry climate condition between 750 and 1050 a BP in east and north China was also recorded by the Chinese historical documents (Wen et al., 2011; Zhang and Lu, 2007). This drought event is also recorded by the loess deposit in Guanzhong, Shaanxi (Zhao et al., 2007a). The tree-ring and ice core records from Qilian mountains (Liu et al., 2005; Yao and Thompson, 1992) indicate a lower temperature period between 200 and 900 a BP in Northeast Tibetan Plateau (Yang et al., 2003). The Heihe River runoff was at a low level during 250–1052 a BP as recorded by tree rings in the Qilian mountains (Kang et al., 2002). On one hand, AJO gradually desertified between 1000 and 400 years ago, while on the other hand, MEO began to develop during this period with silty clay deposited and the P. euphratica woods widely established. Therefore, the desertification of AJO and the development of MEO occurred simultaneously under the same background climate. The climate changes during the last 1 ka of northern China are not the dominant driving force for the observed oasis evolution. However, under the drought climate condition between 300 and 1000 a BP, MEO could not support agriculture and was used only for animal husbandry.

Heihe River migration effects and evidence

From the above discussion, the oasis evolution in the lower reaches of the Heihe River was not driven by the regional climate. The river networks on the alluvial fan are sensitive to geomorphological changes and erosion, which could have substantially changed water allocation and distribution in this area at century scales. Therefore, we analyzed the hydrogeomorphic characteristics of this area (Figure 5) using DEMs (acquired in February 2002, 90-m resolution; Rodriguez et al., 2006; Sheng, 2009).

Figure 5.

Drainage network and geomorphology of the study area: (a) extracted drainage network (white watercourse) displayed on top of the DEM image. Light blue represents water body locations. The black dash line interpreted from the remotely sensed image indicates ancient watercourses. (b) Close-up view of ancient watercourses in the box of (a). (c) Landscape photograph of an ancient watercourse taken in the field.

The drainage network was extracted using the DEM data of the study area (Shen and Sheng, 2012). Comparing with a high-resolution remote sensing image (Figure 1), the extracted drainage network (AB and CD) can adequately reproduce the modern watercourses (dash lines in Figure 1) in the lower reaches of Heihe River (Figure 5a) and identifies the main watercourses into the terminal lake basins. Observed from the remote sensing image and in field explorations, the extracted watercourse EF currently is a desertified ancient watercourse. From the DEM data and the location of Heicheng City, EF should have been the main watercourse of the ancient Heihe River that flowed into the Juyanze paleo-lake. The ancient watercourse EF should have once intersected with the watercourse AB. The terrain analysis of EF and AB shows that they have almost same slope change. The AB course does not seem to have any advantage over EF for water flow. The only possibility for river diversion is that EF was blocked by the dunes. When the Heihe River was diverted northward to GaxuNur Lake and SuguNur Lake, EF was abandoned. EF and AB are currently about 10 km apart, and the present geomorphology does not allow water to flow freely between them. Desertified ancient watercourses can be clearly identified on the remote sensing image and in field explorations (dash lines in Figure 5a and b) in this 10-km zone. Active and fixed dunes are widely distributed, and the P. euphratica trees can be occasionally found in the ancient riverbeds (Figure 5c) although seldom observed in the area. The watercourse diversion should be responsible for the desiccation of Juyanze paleo-lake and AJO desertification. Moreover, the river diversion should also be the main reason for the emergence of the GaxuNur Lake and SuguNur Lake and MEO.

The dating results of the GXN section (Figure 3) from GaxuNur Lake support the process of drainage network diversion. The OSL age (0.43 ka) of the lacustrine sediments and the ¹⁴C age (580 cal. a BP) of the organic matter from the same layer are consistent within their uncertainty ranges. The dating results suggest that the most recent lake period in this area was about 500 a BP and imply that the Heihe River was diverted in the direction of GaxuNur Lake during that period.

Historic documents recorded that the Heihe River diversion occurred during the war of the Ming replacing the Yuan Dynasty (~600 a BP; Du, 2003). According to the evolution process of the two oases, however, some water from the Heihe River could have flowed into SuguNur Lake since ~1 ka ago and the MEO began to develop at that time. At ~600 a BP, a substantial amount of water from the Heihe River flowed into the SuguNur Lake and GaxuNur Lake, and the ancient Heicheng City had to be abandoned, and the AJO desertified completely before ~400 a BP. The MEO developed comprehensively during 1.05–0.45 ka ago, and the GaxuNur became a lake at ~500 a BP. Heihe River’s water diversion occurred gradually over the last 1 ka, and there is no direct evidence of human-induced river diversion in this period. Heihe River was at low-level runoff during 250–1052 a BP as recorded by tree rings (Kang et al., 2002). The river diversion was likely due to the reduced river discharge and the blockage by sands in the dry periods. In field investigation, the modern watercourse is about 1–2 m lower than the ancient watercourse which was full of sand dunes (Figure 5c). Although human cultivation could accelerate the erosion of the land surface and increase sediment loads in watercourse, the human activities still might not be the main force driving the river diversion given the cultivation capability in ancient time.

Conclusion

This paper applies OSL and ¹⁴C dating methods to aeolian sediments and dead P. euphratica woods, respectively, to understand oases evolution processes and mechanisms in the lower reaches of the Heihe River. The optical dating results of aeolian sands in the P. euphratica woods show that the MEO developed gradually during last 1.05–0.46 ka. The ¹⁴C dating results of dead P. euphratica trees in the AJO show that the oasis desertified completely about 400 years ago. These dating results suggest that the two oases evolved simultaneously but in opposite directions in the last 1 ka, with AJO drying out and MEO developing gradually. The drought event started 1 ka ago in northern China cannot explain the opposite evolution processes of the two oases. The DEM-based drainage network analysis, remote sensing, and field investigations all suggest that the oasis revolution was governed by the watercourse diversion in the lower reaches of the Heihe River. The age (~500 a BP) of the top lacustrine layer from GaxuNur Lake also supports the interpretation that the Heihe River swung in the direction of GaxuNur Lake and SuguNur Lake at the same time of AJO desertification and MEO development. There is no direct evidence of the human activities to induce the evolution of the two oases, and all the evidences derived from this study suggest that natural processes played a crucial role in the oasis evolution.

Footnotes

Acknowledgements

We wish to thank the anonymous referees and Professor John Matthews for helpful comments on the manuscript. We appreciate Dr Min Jin’s assistance in the fieldtrip.

Funding

This study was financially supported by National Natural Science Foundation of China (grant no. 41272190), National Basic Research Program of China (973 Program; grant no. 2009CB421306), and 100 talents project of CAS and USGS Landsat Science Team Program (grant no. G12PC00071).

References

Aitken

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

¹⁴C Laboratory, Institute of Geochemistry, CAS (1977) The Measurements and Methodology of ¹⁴C Dating. Beijing: Science Press, pp. 31–55. (In Chinese).

Chen

Holmes

. (2003) A mid-Holocene drought interval as evidenced by lake desiccation in the Alashan Plateau, Inner Mongolia, China. Chinese Science Bulletin 48(14): 1401–1410.

Chen

. (1998) Aeolian Sandy Soils in China. Beijing: Science Press, pp. 32–57. (In Chinese).

Chou

(1990) ¹⁴C Chronology Studies in China. Beijing: Science Press. (In Chinese).

(2003) The history and environmental changes of Juyan in past 2 ka. Collections of Essays on Chinese Historical Geography 18(1): 123–130. (In Chinese, with English abstract).

Feng

Chen

Tan

(1998) The studies of the desert- oasis vegetation growth and ecological groundwater level. Journal of Desert Research 18(add 1): 106–109. (In Chinese, with English abstract).

Huntley

Godfrey-Smith

Thewalt

(1985) Optical dating of sediments. Nature 313: 105–107.

Hutt

Jaek

Tchonka

(1988) Optical dating-feldspar optical response stimulation spectra. Quaternary Science Reviews 7: 381–385.

10.

Kang

Chen

Kang

. (2002) Reconstruct the runoff at mountain exit of Heihe River based on tree ring data in last 1 thousand years. Science in China (Series D) 32(8): 675–685. (In Chinese, with English abstract).

11.

Zhao

Chen

(2011) Comparison of three K-feldspar luminescence dating methods for Holocene samples. Geochronometria 38(1): 14–22.

12.

Feng

Zhai

(2010) Study of the height growth dynamic based on tree-ring data in Populus euphratica from the lower reach of the Heihe River, China. Dendrochronologia 28: 49–64.

13.

(1994) Optical dating: Insufficiently bleached sediments. Radiation Measurements 23: 563–568.

14.

Liu

Qin

Shao

. (2005) Temperature variations recovered from tree-rings in the middle Qilian Mountain over the last millennium. Science in China Series D: Earth Sciences 48(4): 521–529.

15.

Local Chronicles Compilation Committee of Ejina Banner (1998) Local Chronicles of Ejina Banner. Beijing: Chorography Press, pp. 1–45. (In Chinese).

16.

Murray

Wintle

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

17.

Olley

Caitcheon

Murray

(1998) The distribution of apparent dose as determined by optically stimulated luminescence in small aliquots of fluvial quartz: Implications for dating young sediments. Quaternary Science Reviews 17(11): 1033–1040.

18.

Pachur

Wuennemann

Zhang

(1995) Lake evolution in the Tengger Desert, Northwest China during the last 40 000 years. Quaternary Research 44: 171–180.

19.

Peng

Xiao

(2013) Preliminary dendrochronological studies on Populus euphratica in the lower reaches of the Heihe River basin in northwest China. Dendrochronologia 31: 242–249.

20.

Prescott

Hutton

(1988) Cosmic ray and gamma ray dosimetry for TL and ESR. Radiation Measurements 14: 223–227.

21.

Prescott

Hutton

(1994) Cosmic ray contributions to dose rates for luminescence and ESR dating: Large depths and long-term time variations. Radiation Measurements 23(2–3): 497–500.

22.

Reimer

Bard

Bayliss

. (2013) IntCal13 and MARINE13 radiocarbon age calibration curves 0-50000 years cal BP. Radiocarbon 55(4). DOI: 10.2458/azu_js_rc.55.16947.

23.

Rodriguez

Morris

Belz

(2006) A global assessment of the SRTM performance. Photogrammetric Engineering & Remote Sensing 72(3): 249–260.

24.

Shen

Sheng

(2012) Area partitioning for channel network extraction using digital elevation models and remote sensing. IEEE Geoscience and Remote Sensing Letters 9(2): 194–198.

25.

Sheng

(2009) PaleoLakeR: A semiautomated tool for regional-scale Paleolake recovery using geospatial information technologies. IEEE Geoscience and Remote Sensing Letters 6: 797–801.

26.

Wallinga

Murray

Wintle

(2000) The single-aliquot regenerative-dose (SAR) protocol applied to coarse-grain feldspar. Radiation Measurements 32(5–6): 529–533.

27.

Wallinga

Murray

Duller

. (2001) Testing optically stimulated luminescence dating of sand-sized quartz and feldspar from fluvial deposits. Earth and Planetary Science Letters 193(3–4): 617–630.

28.

Wang

Deng

. (2012) Paleoenvironmental significance of magnetic susceptibility and grain size of lake sediments from GaxuNur, Inner Mongolia, China. Journal of Desert Research 32(3): 661–668. (In Chinese, with English abstract).

29.

Wang

Chen

Qian

. (2001) Situation of Sand-dust storms and countermeasures in north China. Journal of Desert Research 21(4): 322–327. (In Chinese, with English abstract).

30.

Wen

Wang

. (2006) Deposition of sandstorms in a vegetation-covered sand dune in Ejin Oasis and its characteristics. Journal of Geographical Sciences 16(4): 502–508.

31.

Wen

Zhao

(2011) Drought/Flood change in the Guanzhong region between the 8th and the 13th Century. In: Remote Sensing, Environment and Transportation Engineering. DOI: 10.1109/RSETE.2011.5964476.

32.

Wintle

Murray

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

33.

Xiao

Song

. (2004) Water-land Resources utilization and environmental evolution in middle to the lower reaches of Heihe River Basin during the past 2000 years. Journal of Desert Research 24(4): 405–408. (In Chinese, with English abstract).

34.

Yang

Braeuning

Shi

. (2003) Temperature variations on the Tibetan Plateau over the last two millennia. Chinese Science Bulletin 48(14): 1446–1450.

35.

Yang

Zhang

Lei

. (2009) Compositions of the organic matter and carbonate in Xiaohushan section from Ejina basin, western Inner Mongolia, China and their paleo environmental significance. Quaternary Sciences 29(2): 256–267. (In Chinese, with English abstract).

36.

Yao

Thompson

(1992) Trends and features of climatic changes in the past 5000 years recorded by the Dunde ice core. Annals of Glaciology 16: 21–24.

37.

Zhang

(1984) The synoptic climatology preliminary analysis of the dust fall during the historical period in China. Science in China Series B 14(3): 278–288.

38.

Zhang

(2007) Anti-correlation of summer/winter monsoons? Nature 450: E7–E8.

39.

Zhang

Tso

WMY

(2001) Assessment of bleaching of K-feldspar grains. Radiation Measurements 33: 103–108.

40.

Zhang

Dong

Huang

(2004) Modeling on relation between major plants growth and Groundwater depth in arid area. Journal of Desert Research 24(1): 110–113. (In Chinese, with English abstract).

41.

Zhang

Liu

. (2012) On the vulnerability of oasis forest to changing environmental conditions: Perspectives from tree rings. Landscape Ecology 27: 343–353.

42.

Zhao

(2005) Internal dose rate to K-feldspar grains from radioactive elements other than potassium. Radiation Measurements 40(1): 84–93.

43.

Zhao

Chen

. (2007a) A record of Holocene climate change in the Guanzhong Basin, China. The Holocene 17: 1015–1022.

44.

Zhao

Chang

. (2007b) Study on vegetation ecological water requirement in Ejina Oasis. Science in China Series D: Earth Sciences 50(1): 121–129.

45.

Zhu

Liu

Gao

. (1983) The environmental changes and desertification processes in historical period in the areas of ancient Juyan-Heicheng region in western Inner Mongolia. Journal of Desert Research 3(2): 1–8. (In Chinese, with English abstract).

Oasis evolution processes and mechanisms in the lower reaches of Heihe River,Inner Mongolia,China since 1 ka ago

Abstract

Keywords

Introduction

Field investigation

Dating process and results

Optical dating

14C dating

Discussion

Oases evolution process

MEO development in the last 1 ka

AJO desertification

Oasis evolution mechanisms

Climate background in northern China

Heihe River migration effects and evidence

Conclusion

Footnotes

Acknowledgements

Funding

References

¹⁴C dating