INTRODUCTION
Reconstructing Holocene climate change is important for predicting and understanding future warming and its regional ecological response (Wanner et al., Reference Wanner, Beer, Bütikofer, Crowley, Cubasch, Flückiger, Goosse, Grosjean, Joos and Kaplan2008). In particular, most of the profound climate changes occurred in the Middle Holocene (Steig, Reference Steig1999; Ruddiman, Reference Ruddiman2001). However, knowledge of the Mid-Holocene climate changes remains limited, and more high-resolution paleoclimate records from climate-sensitive regions are especially needed. Arid and semiarid China is a highly climate-sensitive area (Reynolds et al., Reference Reynolds, Smith, Lambin, Turner, Mortimore, Batterbury and Downing2007; Huang et al., Reference Huang, Yu, Guan, Wang and Guo2015).
The Holocene optimum in China was characterized by warm temperatures and abundant precipitation, which occurred in the Middle Holocene (8.5–3.0 cal ka BP) (Shi et al., Reference Shi, Kong, Wang, Tang, Wang, Yao, Zhao, Zhang and Shi1994a, Reference Shi, Kong, Wang, Tang, Wang, Yao, Zhao, Zhang and Shi1994b). Most studies suggest that the warm and wet Mid-Holocene was widely recorded in the arid and semiarid areas of China (e.g. Feng et al., Reference Feng, An and Wang2006; Liu et al., Reference Liu, Lu, Du, Lyu, Wang, Gao and Wu2018; Feng and Yang, Reference Feng and Yang2019; Wang et al., Reference Wang, Liu, Li, Niu, He, Ma and Mensing2019; Gao et al., Reference Gao, Li, Zhu and Wang2020; Liu et al., Reference Liu, Lu, Ding, Lyu, Li and Dong2021; Ning et al., Reference Ning, Wang, Yang, Zhang, Wang and Li2021; Peng et al., Reference Peng, Wang, Yin, Adamiec, Du, Zhao, Kang, Hu and Zheng2022; Chen et al., Reference Chen, Lu, Liu and Ding2023). However, some studies suggest that a dry Mid-Holocene climate has been found in desert regions of arid and semiarid China (such as the Tengger Desert, the Hobq Desert, and the Taklimakan Desert) (see An et al. Reference An, Feng and Barton2006 and references therein). Wang et al. (Reference Wang, Ma, Feng, Narantsetseg, Liu and Zhai2011) further suggest that the Mid-Holocene drought might have prevailed in the central–east Asian arid and hyperarid areas, which may have remained dry for longer periods. Therefore, more work is needed to analyze the variation in Mid-Holocene humidity for the arid and hyperarid areas.
The Tengger Desert, located in arid and semiarid China, is characterized by low precipitation but high evaporation, sparse vegetation, and fragile ecosystems and is sensitive to climate change (Zhu et al., Reference Zhu, Liu, Wu and Di1980). However, limited research suggests that Holocene climate change in the Tengger Desert is complicated (Herzschuh et al., Reference Herzschuh, Tarasov, Wünnemann and Hartmann2004; Chen et al., Reference Chen, Cheng, Zhao, Zhu and Madsen2006; Zhao et al., Reference Zhao, Yu, Chen and Li2008, Reference Zhao and Yu2012a; Li et al., Reference Li, Wang, Cheng, Long and Zhao2009), probably due to the combined effects of the Asian summer monsoon, westerlies, topography, and regional vegetation factors (Li et al., Reference Li, Wang, Li and Zhang2011; Zhao et al., Reference Zhao and Yu2012a). In this study, we present a sediment core (YLH15A) drilled from Moon Lake in the Tengger Desert to reconstruct the vegetation history, lake hydrology, and associated climate change since the Middle Holocene based on pollen, grain size, elemental, and total organic carbon (TOC) information. Furthermore, regional records were used to determine the climatic characteristics and explore possible mechanisms in the Tengger Desert and even the arid and hyperarid areas during the Middle Holocene.
REGIONAL SETTING
Moon Lake, a closed desert lake, is located in the eastern margin of the Tengger Desert, with a lake surface area of ~1.3 km2 and water depths of 2 to ~4 m (Fig. 1a). Previous studies indicated that many residual lakes in the Tengger Desert are supplied by atmospheric precipitation, near-source ground water, and spring water (Wu, Reference Wu2009; Lai et al., Reference Lai, Wang, Huang, Zhang, Zhao and Xu2012). The study area is dominated by a dry and temperate continental climate (Luo and Sun, Reference Luo and Sun2005). The neighboring Toudaohu meteorological station indicates that the regional mean monthly temperature is −9.4℃ in December and 23.5℃ in June, with a mean annual precipitation of 150 mm and a high potential evaporation of 3200 mm (Duan et al., Reference Duan, An, Zhao, Wang, Cao and Zhou2018; Fig. 1b). A field survey found that there are few vegetation types around the lake area, mainly xerophytic shrubs, including Calligonum mongolicum Turcz., Hedysarum fruticosum var. mongolicum, Nitraria sibirica Pall., Haloxylon ammodendron (CA Mey.) Bunge, and Artemisia ordosica Krasch. Large numbers of common reed (Phragmites australis [Cav.] Trin. ex Steud.) grow in the lake (Fig. 1c). Sandy soil with a low organic matter content, which is not conducive to plant growth and results in low regional vegetation cover, is widely distributed in this region (Luo and Sun, Reference Luo and Sun2005).
Overview of the study area. (a) Map of China noting locations of the Tengger Desert and Moon Lake (sampling site) in the Tengger Desert. (b) Mean monthly temperature and precipitation from the Toudaohu meteorological station (1962–2012) (Duan et al., Reference Duan, An, Zhao, Wang, Cao and Zhou2018). (c) View of the sampling site (YLH15).
MATERIALS AND METHODS
Sampling and dating
In December 2015, a 278-cm sediment core was retrieved from Moon Lake with a Kullenberg-type piston corer, namely, core YLH15A (38°27.72′N, 105°9.14′ E; 1311 m above sea level) (Fig. 1c). The sediment samples were taken at different intervals for multiproxy analyses (pollen, grain size, TOC, and elements) to explore regional paleoenvironmental changes. In addition, 11 organic samples were selected for accelerator mass spectrometry (AMS) 14C dating at Beta Analytic Inc. (Miami, FL, USA).
Grain size, TOC, and elements
A total of 278 grain-size samples were collected from the YLH15A core at 1 cm intervals for laboratory measurements. The pretreatment procedure followed the description in Peng et al. (Reference Peng, Xiao, Nakamura, Liu and Inouchi2005). The grain-size distribution was measured using a Mastersizer 2000 with a range of 0.02–2000 μm, and the mean measurement error was less than 2%. Based on grain-size distribution data, the grain-size component analysis method (Sun et al., Reference Sun, Bloemendal, Rea, Vandenberghe, Jiang, An and Su2002, Reference Sun, Bloemendal, Rea, An, Vandenberghe, Lu, Su and Liu2004, Reference Sun, Su, Bloemendal and Lu2008) was used to analyze the proportions and origins of different components in the YLH15A core.
Ninety-three samples were collected at 3 cm intervals for TOC content measurements using a TOC analyzer (Elementar, Langenselbold). First, the dried samples were ground and sieved, and then 10% HCl was added into 0.2–0.3 g to remove carbonate. The samples were dried and pressed into pieces for measurement.
The elemental contents of samples at a 1 cm resolution were measured with an Avaatech X-ray fluorescence core scanner. Before core scanning, the split sediment cores were flattened and covered with Ultralene film. The relative variations in the elemental compositions of the sediment core were determined, and then the elemental ratios (e.g., Ca/Ti ratio and Fe/Mn ratio) were calculated. All laboratory work was conducted at the Key Laboratory of Western China's Environmental Systems, Lanzhou University.
Pollen
Sixty-nine samples of approximately 2 g each were selected for pollen analyses at 4 cm intervals. Pollen extraction followed the standard preparation protocol described by Fægri and Iversen (Reference Fægri and Iversen1989). One Lycopodium-spore tablet (27,637 grains/tablet) was added to each sample before the chemical treatment to calculate pollen concentrations (Stockmarr, Reference Stockmarr1971). Pollen identification and counting were performed using pollen atlases of the arid and semiarid areas of China (Xi and Ning, Reference Xi and Ning1994; Wang et al., Reference Wang, Qian, Zhang and Yang1995; Tang et al., Reference Tang, Mao, Li, Shen and Zhou2016) and a Nikon light microscope at 400× magnification. At least 300 grains of terrestrial plant pollen were counted for most samples. A pollen diagram was plotted using Tilia v. 2.02 (Grimm, Reference Grimm2004). CONISS was used for stratigraphically constrained cluster analysis (Grimm, Reference Grimm1987).
RESULTS
Lithology and chronology
The YLH15A sediment core is mainly composed of fine sand and silt, containing almost no clay; in addition, the upper 183 cm of the core also consists of coarse sand (Fig. 2b). The AMS 14C dates of the sediment core were calibrated to calendar years before present (BP) with the Calib v. 7.0 software using the IntCal13 calibration database (Reimer et al., Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey and Buck2013; Table 1). There is an age reversal at depth of 132 cm, which is younger than the upper age. Therefore, the chronology of core YLH15A was established based on 10 AMS 14C dates. Considering that the age of the top of the sediment core is 2015 CE, an age–depth model was generated using a cubic spline interpolator with Bayesian statistical methods (Blaauw and Christen, Reference Blaauw and Christen2011; Fig. 2a). The chronological sequence of core YLH15A was established based on the weighted mean ages modeled using the Bacon package in R software, and the bottom age of the YLH15A core was estimated to be 7.6 cal ka BP.
(a) Age–depth model and (b) median grain-size distribution from sediment core YLH15A.
Accelerator mass spectrometry (AMS) 14C dating results from core YLH15A.
Grain size, TOC, and elemental analyses
All samples of the strata have different sizes and proportions, indicating specific modal size ranges of components with specific origins (Bagnold and Barndorff-Nielsen, Reference Bagnold and Barndorff-Nielsen1980). Figure 3 presents two types of statistical plots of the YLH15A core. The grain-size frequency distribution (Fig. 3a) shows the modal size ranges of components with different origins. Three different groups are identified according to the combination of inflections at 38 and 418 μm on the grain-size frequency distribution curve. The group between 0 and 38 μm is short-term suspended dust with a modal size of ~24 μm (Guo et al., Reference Guo, Sun, Wamg, Li and Li2014; Z.J. Li et al., Reference Li, Wang and Zhang2014), which is the typical modal size of the coarse dust component (Sun et al., Reference Sun, Bloemendal, Rea, Vandenberghe, Jiang, An and Su2002, Reference Sun, Bloemendal, Rea, An, Vandenberghe, Lu, Su and Liu2004, Reference Sun, Su, Bloemendal and Lu2008; Yin et al., Reference Yin, Qin, Wu and Ning2009; Wang et al., Reference Wang, Sun, Chen, Bloemendal, Guo, Li, Zhang, Li and Wang2015). The group between 38 and 418 μm corresponds to the fine sand component with a modal size of ~138 μm, which represents the typical aeolian sand component (e.g., Qiang et al., Reference Qiang, Chen, Wang, Niu and Song2010; Dietze et al., Reference Dietze, Maussion, Ahlborn, Diekmann, Hartmann, Henkel, Kasper, Lockot, Opitz and Haberzettl2014; Z.J. Li et al., Reference Li, Sun, Chen, Wang, Zhang, Guo, Wang and Li2014; Mou et al., Reference Mou, Ma, Wang and Fan2018). The >418 μm group with a modal size of ~957 μm corresponds to the coarse sand component. Moreover, the standard deviation for the grain-size class plot (Fig. 3b) shows two peaks observed at grain sizes of 38–418 μm and >418 μm, which represent the two grain-size components that have high variability throughout the sedimentary sequence (Boulay et al., Reference Boulay, Colin, Trentesaux, Pluquet, Bertaux, Blamart, Buehring and Wang2003).
Statistical plots for sediment component analyses: (a) mean size distribution of volume frequency for grain-size classes and (b) standard deviation of grain-size classes.
Generally, the three grain-size components are characterized by obvious fluctuations above 186 cm, but their variations are relatively stable below 186 cm, except at depths of 243–235 cm. Aeolian sand (38–418 μm) dominates in core YLH15A (mean: 68%; maximum: 95%) and shows the opposite trend relative to coarse dust (<38 μm); the proportion of coarse sand (>418 μm) ranges between 0% and 56% with a mean value of 5% above 186 cm and is accompanied by strong fluctuations, indicating that the sedimentary environment changes greatly. The TOC content is generally low, with a mean value of 1.7. The variations in the Ca/Ti ratio display a similar pattern to TOC, and the trend of changes in the Fe/Mn ratio generally shows an inverse pattern relative to the Ca/Ti ratio. According to the combination of each proxy, core YLH15A could be divided into four stages, labeled A to D from bottom to top (Fig. 4).
Variation in non-pollen proxies of core YLH15A from Moon Lake. TOC, total organic carbon.
Stage A (278–186 cm, ~7.6 to ~2.8 cal ka BP) is characterized by aeolian sand and coarse dust, and the aeolian sand component generally remains high. Based on the combination of each proxy, this stage is divided into three substages. Substage A-1 (278–243 cm, ~7.6 to ~5.4 cal ka BP) consists of aeolian sand mixed with minor coarse dust (mean: 17%). Aeolian sand is the dominant component in this stage (mean: 82%). The TOC content (mean: 0.2%) and Ca/Ti ratio (mean: 6.9) are low, while the Fe/Mn ratio reaches a maximum during this stage and maintains a relatively high value (mean: 61). Substage A-2 (243–235 cm, ~5.4 to ~4.9 cal ka BP) mainly consists of aeolian sand (mean: 56%) and dust components (mean: 44%), and the coarse dust component exhibits a significant increase to a maximum of 76%, while aeolian sand content declines to the lowest value of 24%. TOC content shows a slight increase with a mean value of 3.5%, and the Ca/Ti ratio is relatively low (mean: 24), followed by an obvious increase at 237 cm (~5.0 cal ka BP). The variations in the Fe/Mn ratio show an increase, followed by a significant decline at 239 cm (~5.1 cal ka BP). In Substage A-3 (235–186 cm, ~4.9 to ~2.8 cal ka BP), aeolian sand dominates and maintains a high content (mean: 70%), and the coarse dust component ranges between 22% and 46%, with a mean value of 30%, which is higher than that of Stage A but lower than that of Stage B. Variations in the Ca/Ti ratio and TOC content generally show similar patterns, with low values (means: 45% and 1.1%, respectively), while the Fe/Mn ratio is high (mean: 59).
In Stage B (186–150 cm, ~2.8 to ~1.8 cal ka BP), the proportion of aeolian sand slightly decreases (mean: 63%), but a slight increase is observed in the coarse dust component (mean: 31%). Notably, the coarse sand component starts to appear in this stage and shows strong fluctuations ranging from 0% to 56%, with a mean value of 6%. The Fe/Mn ratio exhibits a significant decrease and remains low (mean: 24). Both the Ca/Ti ratio and TOC content significantly increased.
Stage C (150–73 cm, ~1.8 to ~1.1 cal ka BP) is characterized by aeolian sand and coarse dust mixed with coarse sand deposits, which are accompanied by high variability. Based on the combination of each proxy, this stage is divided into two substages. In Substage C-1 (150–107 cm, ~1.8 to ~1.3 cal ka BP), the coarse dust components are between 9% and 48%, with a mean value of 29%; the aeolian sand component varies from 36% to 84%, with a mean value of 65%; and coarse sand ranges between 0% and 47%, with a mean value of 6%. The TOC content is generally low (with a mean of 1.7%). The Ca/Ti ratio shows a significant decrease (mean: 28), while the Fe/Mn ratio displays a slight increase (mean: 33). In Substage C-2 (107–73 cm, ~1.3 to ~1.1 cal ka BP), aeolian sand and the Fe/Mn ratio slightly decrease, while coarse dust and the Ca/Ti ratio generally increase. The TOC content is low but increases to a mean value of 2.8%. Variations in the coarse sand component are more frequent than those in the previous substage.
Stage D (73–0 cm, since 1.1 cal ka BP) still consists of aeolian sand, coarse dust, and coarse sand, but an obvious change occurs at a depth of 50 cm, and the elemental ratios and TOC content also show similar changes. Therefore, this stage is divided into two substages. In Substage D-1 (73–50 cm, ~1.1 to ~0.5 cal ka BP), the coarse sand component almost disappears, and aeolian sand (mean: 65%) and coarse dust (mean: 32%) dominate the deposits. The Ca/Ti ratios perceptibly increase from 47 to 76, while the Fe/Mn ratios decrease slightly from 28 to 25. The TOC content maintains a low value but increases to a maximum of 3.7. In Substage D-2 (50–0 cm, since 0.5 cal ka BP), aeolian sand and coarse dust are mixed with coarse sand deposits with high variability. The proportion of coarse dust decreases (mean: 21%), but aeolian sand increases (mean: 69%). The Ca/Ti ratio and TOC content generally decrease, showing an inverse change relative to the Fe/Mn ratio.
Pollen analysis
In the YLH15A core, 20 pollen taxa were identified from 69 samples, including 18 terrestrial pollen taxa and two aquatic pollen taxa. The pollen assemblages are dominated by herb pollen taxa, ranging from 84.7% to 98.7% in abundance, with a mean of 94.5%. The herb pollen types mainly consisted of Artemisia, Amaranthaceae, and Poaceae. The abundance of shrubs ranges from 0.8% to 11.7%, with a mean of 3.3%, and mainly consists of Nitraria and Ephedra. Arboreal pollen accounts for only approximately 2.2% of the pollen assemblages (ranging from 0.2% to 8.2%). The terrestrial pollen concentrations range from 0 to 181 grains/g, with a mean of 42 grains/g. Some aquatic pollen types are seen, such as Typha and Pediastrum. The pollen assemblages are divided into four main zones based on CONISS analysis, and detailed descriptions are as follows (Fig. 5).
Pollen diagram of the percentage of main pollen taxa and changes in non-pollen proxies (including the total organic carbon [TOC] content, Ca/Ti ratio, and aeolian sand proportion (38–418 μm)) from core YLH15A in Moon Lake.
In Stage A (278–186 cm, ~7.6 to ~2.8 cal ka BP), the study area generally experienced an extremely harsh desert environment with no vegetation cover, which is consistent with the record of no pollen sample found. However, during Substage A-2 (243–235 cm, ~5.4 to ~4.9 cal ka BP), the pollen assemblages are dominated by Artemisia (mean: 45.0%) and Amaranthaceae (mean: 19.3%), with some Poaceae (approximately 5.3%). The pollen concentration reaches a relatively high value during this stage, with a mean of 51 grains/g. The Artemisia/Chenopodiaceae (A/C) pollen ratio is approximately 1.8.
In Stage B (186–150 cm, ~2.8 to ~1.8 cal ka BP), the pollen assemblage is characterized by Artemisia, Asteraceae (excluding Artemisia), and Amaranthaceae. A notable feature is the increase in Asteraceae and tree pollen, which reach their maximum values of 8.2% and 40.3%, respectively. However, this stage has an extremely low pollen concentration (mean: 1 grains/g) and insufficient pollen counts.
In Stage C (150–73 cm, ~1.8 to ~1.1 cal ka BP), shrub and herb pollen percentages are controlled by Artemisia (mean: 51.6%), Amaranthaceae (mean: 27.1%), and Poaceae (mean: 11.9%). In Substage C-1 (150–107 cm, ~1.8 to ~1.3 cal ka BP), the Amaranthaceae percentage reaches a maximum of 60.3% (mean: 32.3%), and the Nitraria percentages are at the highest level for the whole period (up to 10.8%). The A/C ratio is generally low (mean: 1.6) and reaches a minimum of 0.5 at a depth of 134 cm. The pollen concentration starts to increase (mean: 61.6 grains/g). In Substage C-2 (107–73 cm, ~1.3 to ~1.1 cal ka BP), the Artemisia percentage increases (mean: 56.0%), accompanied by an obvious drop in Amaranthaceae (mean: 20.6%). The A/C ratios increase to 2.8. Poaceae pollen significantly increases (up to 21.8%) and remains at a relatively stable value (mean: 14.9%). The pollen concentration increases noticeably (up to 181 grains/g). In addition, the Typha percentage increases and accounts for approximately 4.4% during this stage.
In Stage D (73–0 cm, since 1.1 cal ka BP), the pollen assemblages resemble those of Stage C, except for the maximized Artemisia percentages (up to 74.4%). The A/C ratios are generally high (mean: 4.0). Notably, Poaceae percentages and pollen concentrations generally show a similar fluctuation from high to low since 0.5 cal ka BP (50–0 cm, Substage D-2).
DISCUSSION
Interpretation of the climatic proxies
Grain-size components
The coarse dust component (<38 μm) is aeolian sediment reworked by flowing water, and a higher content indicates that the sediment is closer to the center of the lake and vice versa (Yin et al., Reference Yin, Qin, Wu and Ning2008, Reference Yin, Qin, Wu and Ning2009). Therefore, the grain-size component is assumed to reflect fluctuations in the lake level. The aeolian sand component (38–418 μm) is transported by saltation or rolling when strong surface wind occurs (e.g., Qiang et al., Reference Qiang, Chen, Wang, Niu and Song2010; Dietze et al., Reference Dietze, Maussion, Ahlborn, Diekmann, Hartmann, Henkel, Kasper, Lockot, Opitz and Haberzettl2014; Z.J. Li et al., Reference Li, Sun, Chen, Wang, Zhang, Guo, Wang and Li2014; Mou et al., Reference Mou, Ma, Wang and Fan2018; Liu et al., Reference Liu, Lu, Ding, Lyu, Li and Dong2021). Moreover, this transport of aeolian sand on frozen lakes is associated with strong winds that particularly occur during winter and early spring storms (Sun et al., Reference Sun, Bloemendal, Rea, Vandenberghe, Jiang, An and Su2002). Therefore, this component indicates the wind intensity or aeolian activities. The coarse sand component (>418 μm) is alluvial–diluvial sand from the edge of the Tengger Desert and is transported to the lake area by temporary floods triggered by heavy rainfall in mountainous areas (Tsoar and Pye, Reference Tsoar and Pye1987; Huang et al., Reference Huang, Chen, Xiao, Lv, Chen and Zhou2008; Yin et al., Reference Yin, Qin, Wu and Ning2009; Dietze et al., Reference Dietze, Maussion, Ahlborn, Diekmann, Hartmann, Henkel, Kasper, Lockot, Opitz and Haberzettl2014; Z.J. Li et al., Reference Li, Sun, Chen, Wang, Zhang, Guo, Wang and Li2014; Jiang et al., Reference Jiang, Nie, Liu, Wang, Zhao, Yang and Shen2021). This situation is common in extremely arid desert-edge areas, where temporary floods caused by heavy rainfall from mountainous areas are an important source of groundwater recharge (Dahan et al., Reference Dahan, Tatarsky, Enzel, Kulls, Seely and Benito2008; Greenbaum et al., Reference Greenbaum, Schwartz, Benito, Porat and Cloete2014; Jiang et al., Reference Jiang, Nie, Liu, Wang, Zhao, Yang and Shen2021). Therefore, during the period of frequent temporary floods, the increase in groundwater recharge facilitates the development of lakes in desert areas.
The Fe/Mn ratio, Ca/Ti ratio, and TOC
The Fe/Mn ratio can be used to reconstruct bottom-water oxygenation in lakes; and the ratio increases from the maximum lake depth toward shallower water depths (Naeher et al., Reference Naeher, Gilli, North, Hamann and Schubert2013). This is likely due to geochemical focusing, a process in which redox-sensitive trace metals are transferred and enriched at greater lake water depths (Schaller and Wehrli, Reference Schaller and Wehrli1996). The Ca/Ti ratio is used to indirectly reflect the proportion of authigenic and terrigenous lake sediments, which is controlled mainly by precipitation/evaporation and the density of vegetation cover in the watershed (Çağatay et al., Reference Çağatay, Öğretmen, Damcı, Stockhecke, Sancar, Eriş and Özeren2014). Warm and wet conditions may increase the density of vegetation cover in the watershed, which would reduce erosion rates and increase authigenic carbonate precipitation in the lake, resulting in high Ca/Ti values. This is consistent with the high TOC content attributed to increased lake productivity (Kylander et al., Reference Kylander, Klaminder, Wohlfarth and Löwemark2013).
The A/C ratio
Artemisia (A) and Amaranthaceae (formerly named Chenopodiaceae) (C) are the dominant plant taxa in arid and semiarid regions, and the former requires more water than the latter during the growing season. Previous studies suggest that the A/C ratio can be used to distinguish steppe from desert and to indicate moisture changes in arid and semiarid regions across the world (e.g., El-Moslimany, Reference El-Moslimany1990; Yan, Reference Yan1991; Cour et al., Reference Cour, Zheng, Duzer, Calleja and Yao1999; Herzschuh et al., Reference Herzschuh, Tarasov, Wünnemann and Hartmann2004; Li et al., Reference Li, Xu, Zhao, Yang, Xiao, Chen and Lu2005; Zhao et al., Reference Zhao, Yu, Chen, Ito and Zhao2007). Zhao et al. (Reference Zhao, Liu, Li, Huang, Sun, Zhao, Herzschuh and Tang2012b) further proposed that the A/C ratio can be used only in regions with precipitation <450–500 mm and in steppe, steppe desert and desert areas, but the threshold A/C values for different vegetation types may vary in different regions. In the Alashan Plateau, the A/C ratios are mostly <1 in desert communities, while in the nearby steppes, the A/C ratios are >2.3 (Herzschuh et al., Reference Herzschuh, Tarasov, Wünnemann and Hartmann2004). Therefore, the A/C ratio from Moon Lake can be used to indicate the past vegetation and moisture conditions; high (low) A/C ratios are associated with an increase (decrease) in steppe pollen production, which corresponds to high (low) relative humidity in this study region (Fig. 5).
Paleoclimate changes recorded by Moon Lake
Paleoclimatic evidence from lake hydrology
Based on grain size, elemental ratios, and TOC analysis, the hydrological changes in Moon Lake since the Middle Holocene can be divided into four main stages.
From 7.6 to 2.8 cal ka BP, there may have been a long-term dried-up lake environment. This inference is supported by the higher content of aeolian sand and higher Fe/Mn ratios (Fig. 4), reflecting strong aeolian activities and extremely shallow water depths in the lake region. Therefore, this period is likely characterized by extremely arid climate. With a notable exception, the temporary lake, pond, or wetland environment occurred in the interval of 5.4–4.9 cal ka BP, which is supported by the high content of coarse dust, reflecting that the sediment was deposited closer to the center of the lake and thus reveals an elevated lake level. This is also supported by the high Ca/Ti ratio and high TOC content (Fig. 4), which further suggest relatively wet conditions during this interval.
The lake environment has existed since 2.8 cal yr BP, but has exhibited instability. From 2.8 to 1.8 cal ka BP, slightly decreased aeolian sand and increased TOC content indicate an improved lake environment (Fig. 4) compared with that in the previous period. During the same period, the temporary flood indicated by the coarse sand component caused an increase in groundwater recharge and thus facilitated the development of the lake, which was also inferred from relatively low Fe/Mn ratios and increased coarse dust. Moreover, high Ca/Ti ratios indicate an increase in authigenic carbonate precipitation in the lake, probably resulting from increased regional precipitation.
From 1.8 to 1.1 cal ka BP, a shallow lake environment occurred, as inferred from relatively high Fe/Mn ratios (Fig. 4). However, the lake environment has improved since 1.3 cal ka BP, which is supported by weakened aeolian activities, increased lake productivity, and increased authigenic carbonate precipitation in the lake. Furthermore, the frequent occurrence of temporary floods and the resultant increased groundwater recharge further suggest a relatively humid lake environment.
From 1.1 to 0.5 cal ka BP, the lake expanded under warm and wet conditions, which is supported by significantly increased Ca/Ti ratios and TOC content. However, relatively high Fe/Mn ratios and decreased coarse dust components indicate an obvious decrease in the lake level since 0.5 cal ka BP (Fig. 4). Moreover, stronger aeolian activities revealed by higher aeolian sand proportions have a more significant impact on the declining lake levels, which further suggests a drying climate since 0.5 cal ka BP.
In summary, the lake hydrology and associated climate change in the lake area mainly experienced two stages: before 2.8 cal ka BP, a dried-up lake environment accompanied by stronger aeolian activities reflected generally dry conditions in this region; since 2.8 cal ka BP, lake development and the increased lake level probably resulted from increased groundwater recharge, which originated from regional precipitation and frequent temporary floods from adjacent mountain areas.
Paleoclimatic evidence from the vegetation history
According to our pollen analysis, desert and desert steppe vegetation has dominated the landscape of the lake region since the Middle Holocene. In general, most pollen represents plants growing in the lake region, but tree pollen and Asteraceae pollen were probably growing in the surrounding mountains at some distance. Detailed descriptions are presented in the following paragraphs.
From 7.6 to 5.4 cal ka BP, desert environment with no vegetation cover prevailed, which is also evidenced by an extremely high aeolian sand deposit content. From 5.4 to 4.9 cal ka BP, this lake region was covered by steppe desert vegetation, which was dominated by Artemisia, Amaranthaceae, and Poaceae. The relatively high A/C ratios suggest a moderate moisture availability (Zhao et al., Reference Zhao, Liu, Li, Huang, Sun, Zhao, Herzschuh and Tang2012b), which increases vegetation cover. However, the desert landscape prevailed again from 4.9 to 2.8 cal ka BP, similar to the initial environment.
During 2.8–1.8 cal ka BP, a dry desert/steppe desert landscape developed with scarce vegetation cover, which was dominated by Artemisia, Asteraceae, and Amaranthaceae. Asteraceae pollen at the YLH15A core likely comes from Ajania fruticulosa based on our vegetation investigation in the field; it is widespread in gravel valleys in desert and steppe desert regions (Wang, Reference Wang1988). Therefore, the abrupt increase in Asteraceae pollen during this period might suggest the expansion of A. fruticulosa in the floodplain after flooding from the mountains, with flooding also supported by an increase in tree pollen (Fig. 5). A similar situation also occurs in Qingtu Lake (Zhao et al., Reference Zhao, Yu, Chen and Li2008). Moreover, the extremely low vegetation cover was probably related to the joint effects of an arid climate and enhanced human activities. The number of archaeological sites from the Alashan Plateau, particularly sites dating to the Qin and Han dynasties (corresponding to ~2.2 to ~1.8 cal ka BP), has increased dramatically (Guo, Reference Guo2003; An, Reference An2022), supporting the speculation about low vegetation cover caused by strong human activities.
From 1.8 to 1.1 cal ka BP, desert steppe and steppe desert dominated by Artemisia, Amaranthaceae, and Poaceae developed successively in the lake region. The significant increase in Amaranthaceae between 1.7 and 1.6 cal ka BP may be related to human overgrazing activities (Mercuri, Reference Mercuri2008; Wen et al., Reference Wen, Xiao, Chang, Zhai, Xu, Li, Itoh and Lomtatidze2010), supported by historical evidence from the states founded by the northern nomadic people (such as Xianbei in the Sixteen States period). Between 1.6 and 1.1 cal ka BP, a high Poaceae content indicates an expansion of steppe over desert and a relatively humid climate (e.g., Li, Reference Li1998; Herzschuh et al., Reference Herzschuh, Kürschner and Ma2003; Li et al., Reference Li, Xu, Zhao, Yang, Xiao, Chen and Lu2005; Zhao et al., Reference Zhao, Yu, Chen, Ito and Zhao2007). Moreover, agricultural policies promoted by the northern Wei dynasty may also be responsible for increases in Poaceae plants. Overall, the lake environment improved, as indicated by the significant increase in vegetation cover during this period (Fig. 5).
From 1.1 cal ka BP to the present, the vegetation landscape was similar to that in the last period. Between 1.1 and 0.5 cal ka BP, higher A/C ratios suggest that the effective moisture availability was favorable for the growth of moisture-demanding plants. However, the significant decreases in the Poaceae percentages and pollen concentrations afterward collectively indicate a drying climate, especially since 0.2 cal ka BP. Moreover, enhanced aeolian activities further reveal the deterioration of the regional environment (Fig. 5). Furthermore, the A/C ratios are still high, which might be caused by increasing water availability to xerophytic plants in the lake area from the aforementioned temporary flood.
In general, the vegetation history and associated climate change in the lake area mainly experienced two stages: before 2.8 cal ka BP, a desert environment with no vegetation cover prevailed, which is also evidenced by a dried-up lake and strong aeolian activities; therefore, the climate of this period was rather arid overall. Since 2.8 cal ka BP, the expansion of desert steppe/steppe desert and increased vegetation cover suggest relatively wet conditions, accompanied by increased lake levels.
Prolonged dry Mid-Holocene in the central–east Asian arid and hyperarid areas
The hydrological changes and vegetation history inferred from Moon Lake suggest the following climatic sequence since the Middle Holocene. The climate was generally dry from 7.6 to 2.8 cal ka BP, as inferred from a dried-up lake and strong aeolian activities, except for a short-term wet episode between 5.4 and 4.9 cal ka BP. The arid climate shifted after 2.8 cal ka BP, as suggested by expanded desert steppe/steppe desert and increased vegetation cover. Meanwhile, it was accompanied by a variable environment, as indicated by changes of pollen types and unstable lake levels during the Late Holocene.
A prolonged dry Mid-Holocene at Moon Lake in the Tengger Desert is also recorded at Lake Zhuyeze (Chen et al., Reference Chen, Cheng, Zhao, Zhu and Madsen2006), Lake Juyanze (Hartmann and Wünnemann, Reference Hartmann and Wünnemann2009) in the Badain Jaran Desert, the Midiwan peat section in the Mu Us Desert (Li et al., Reference Li, Zhou, An and Dong2000, Reference Li, Zhou, An and Dodson2003), and Yanhaizi Lake in the Hobq Desert (Chen et al., Reference Chen, Lan, Lou and Chen2003; Fig. 6). An et al. (Reference An, Feng and Barton2006) further proposed that a prolonged dry Mid-Holocene occurred only in desert regions, but this record cannot be extended to the entire Inner Mongolia Plateau. Recently, the prolonged Mid-Holocene drought has also been extensively documented in central–northern Mongolia (Peck et al., Reference Peck, Khosbayar, Fowell, Pearce, Ariunbileg, Hansen and Soninkhishig2002; Fowell et al., Reference Fowell, Hansen, Peck, Khosbayar and Ganbold2003; Feng et al., Reference Feng, Zhai, Ma, Huang, Wang, Zhang, Khosbayar, Narantsetseg, Liu and Rutter2007; Prokopenko et al., Reference Prokopenko, Khursevich, Bezrukova, Kuzmin, Boes, Williams, Fedenya, Kulagina, Letunova and Abzaeva2007; An et al., Reference An, Chen and Barton2008; Wang et al., Reference Wang, Ma, Feng, Narantsetseg, Liu and Zhai2011; Zhang et al., Reference Zhang, Zhang, Feng, Mischke, Gao, Gao and Sun2012; Fig. 6). These studies suggest that a prolonged drought might have prevailed extensively in the central–east Asian arid and hyperarid areas during the Middle Holocene (Wang et al., Reference Wang, Ma, Feng, Narantsetseg, Liu and Zhai2011), possibly as a result of the well-documented large-scale temperature rise (Shi et al., Reference Shi, Kong and Wang1993). Evaporation triggered by high temperatures might have exceeded the precipitation increase in arid and hyperarid areas, resulting in the increased aridity during the Middle Holocene (Chen et al., Reference Chen, Lan, Lou and Chen2003; An et al., Reference An, Feng and Barton2006; Wang et al., Reference Wang, Ma, Feng, Narantsetseg, Liu and Zhai2011). However, the beginning and end of the dry interval could differ in different places due to the different geographic settings (Y. Li et al., Reference Li, Sun, Chen, Wang, Zhang, Guo, Wang and Li2014; Fig. 6).
Climatic records of mid-Holocene drought from arid and semi/hyperarid China and Mongolia: 1, Moon Lake (this study); 2, Lake Zhuyeze (Chen et al., Reference Chen, Cheng, Zhao, Zhu and Madsen2006); 3, Lake Juyanze (Hartmann and Wünnemann, Reference Hartmann and Wünnemann2009); 4, Lake Yanhaizi (Chen et al., Reference Chen, Lan, Lou and Chen2003); 5, Midiwan section (Li et al., Reference Li, Zhou, An and Dong2000, Reference Li, Zhou, An and Dodson2003); 6, Ugii Nuur (Wang et al., Reference Wang, Ma, Feng, Narantsetseg, Liu and Zhai2011); 7, Lake Telmen (Fowell et al., Reference Fowell, Hansen, Peck, Khosbayar and Ganbold2003); 8, Lake Hovsgol (Prokopenko et al., Reference Prokopenko, Khursevich, Bezrukova, Kuzmin, Boes, Williams, Fedenya, Kulagina, Letunova and Abzaeva2007); 9, Sharmmar section (Feng et al., Reference Feng, Zhai, Ma, Huang, Wang, Zhang, Khosbayar, Narantsetseg, Liu and Rutter2007); 10, Gun Nuur (Zhang et al., Reference Zhang, Zhang, Feng, Mischke, Gao, Gao and Sun2012).
During the Late Holocene (since 2.8 cal ka BP), the climate inferred from Moon Lake was relatively wet compared with that in the Middle Holocene, but variations in pollen, grain size, and TOC content indicate an alternation of dry and wet climates (Fig. 5). Similar climatic conditions have also been recorded in the SJC section (since 3.8 cal ka BP) (Chen et al., Reference Chen, Cheng, Zhao, Zhu and Madsen2006), Gun Nuur (since 2.5 cal ka BP) (Zhang et al., Reference Zhang, Zhang, Feng, Mischke, Gao, Gao and Sun2012), Ugii Nuur (since 2.8 cal ka BP) (Schwanghart et al., Reference Schwanghart, Schütt and Walther2008), the Sharmmar section (since 3.0 cal ka BP) (Feng et al., Reference Feng, Zhai, Ma, Huang, Wang, Zhang, Khosbayar, Narantsetseg, Liu and Rutter2007), and Lake Telmen (since 4.5 cal ka BP) (Fowell et al., Reference Fowell, Hansen, Peck, Khosbayar and Ganbold2003). However, further evidence to support a relatively stable and wet Late Holocene is only rarely seen in the central–east Asian arid and hyperarid areas. The uncertain pattern of climate change during the Late Holocene may partly arise from the fact that lakes experienced different degrees of human impact and responded in a nonlinear way due to specific catchment characteristics (Zhang et al., Reference Zhang, Zhang, Feng, Mischke, Gao, Gao and Sun2012).
CONCLUSIONS
Multiproxy records from Moon Lake in the Tengger Desert revealed lake hydrology, vegetation history, and associated climate change since the Middle Holocene. From 7.6 to 2.8 cal ka BP, this region was characterized by strong aeolian activities, desert landscapes, and a resultant dried-up lake under an extremely arid climate, and higher evaporation could be primarily responsible for the dried-up lake during this period. Since 2.8 cal ka BP, relatively wet conditions were evidenced by vegetation expansion and lake development; furthermore, regional precipitation and temporary floods from adjacent mountain areas recharged the ground water and thus facilitated lake development in desert areas.
These results reveal that Moon Lake experienced a prolonged dry Mid-Holocene and relatively wet Late Holocene that are basically consistent with climatic records from the central–east Asian arid and hyperarid areas, but it is different from either monsoon- or westerly dominated regions and from the East Asian monsoon margin region. More and well-dated Holocene climate records are needed from the core areas of the arid and hyperarid areas to further validate (or contradict) this conclusion.
Acknowledgments
This study was supported by the Natural Science Foundation of Jiangsu Province (no. BK20210656 and no. BK20210658), the National Natural Science Foundation of China (no. 42220104001 and no. 42201169), and the Start-up Foundation for Introducing Talent of NUIST (no. 2021r136).
