Introduction

The welfare of the large human population of East Asia is inextricably linked with the East Asian monsoon precipitation. The ecosystems, economy, and human civilizations in the East Asian monsoon region have depended heavily on the intensity and distribution of monsoon precipitation1,2. For example, Neolithic populations migrated across the Hu Line to the wetter west in response to changes in precipitation3, and the collapse of civilizations has been linked extreme drought events4,5,6. Therefore, an improved understanding of the hydrological evolution of the East Asian monsoon region, especially the spatiotemporal pattern of drought events and their origin, is important for ecosystem maintenance and socioeconomic development in this region.

The hydrological evolution of the Asian summer monsoon (ASM) region on several timescales and at different times has been comprehensively addressed in several studies. This research has revealed that bipolar and tri-polar patterns of hydrological evolution developed in the middle Holocene7; that similar hydrological modes occurring during the Little Ice Age (LIA) and the last deglaciation8,9; and that the Qinling Mountains were the boundary between the northern and southern hydrological regimes in eastern China during the LIA10. However, it is unclear whether antiphase hydrological modes existed during the early Holocene; and even the hydrological evolution of northern China during the early Holocene remains controversial, since research has tended to focus on determining whether the wettest interval in northern China occurred during the middle Holocene2,11,12, or the early Holocene13,14,15. However, a δ13C record from Diaojiaohaizi Lake in the Daqing Mountains has identified a prolonged dry period in the early Holocene16, This finding agrees with a paleohydrological record from Daihai Lake, which revealed low precipitation in the early Holocene, and abundant precipitation in the middle Holocene17. On orbital timescales, the variation of the Asian summer monsoon (ASM) intensity is consistent with that of insolation at 65°N18. Various mechanisms have been proposed to explain extreme hydrological events in the East Asian monsoon region, including the sapping of groundwater from Hunshandake Sandy Land by the Xilamulun River, which led to irreversible drought conditions and the disruption of the Hongshan culture4. Additionally, precipitation anomalies induced by the high-frequency variability of the ASM, modulated by centennial- to interannual-scale mechanisms such as the Pacific Interdecadal Oscillation and the El Niño-Southern Oscillation (ENSO), are proposed to have contributed to the collapse of Neolithic culture5. While the impact of extreme drought events on ancient societies and civilizations in China has attracted much research attention, this has imposed temporal and spatial limitations on the interpretation of extreme drought events, resulting in uncertainty regarding the influence of the climate system on regional drought events.

Precipitation is the most important hydrological recharge mechanism of closed lakes, and thus the levels of such lakes are closely related to precipitation, while closed alpine lakes in particular act as rain gauges10; however, in some lakes the effects of surface runoff and evapotranspiration can bias the precipitation estimates. With chronological controls, the elevation of former lake shorelines can be used as a proxy to quantify paleo-lake levels13,19,20; however, such records are often fragmentary due to factors like hydrodynamic perturbations or the failure of lake shorelines to develop21. Recently, modern process studies have validated the biomarkers of %OH-GDGTs and %Cren as a paleo-lake-level proxy10,22,23, making it possible to quantitatively reconstruct paleolake levels at a high resolution10,24. In this study, we used sedimentary records of GDGTs (%OH-GDGTs) and δ2H to reconstruct lake level changes in Daihai Lake (Fig. 1), in northern China, since the last deglaciation. Combined with other records from this region, we use the results from Dahai Lake to characterize the hydrological evolution of northern China, including abrupt drought events, with the aim of providing new insights into the climatic evolution of the ASM region and its driving mechanisms.

Fig. 1: Location of Daihai Lake and other sites mentioned in the text.
figure 1

The broken red line represents the modern northern limit of the Asian summer monsoon region.

Full size image

Results and discussion

Interpretation and validation of OH-GDGTs as a paleohydrological proxy

Group I.1a Thaumarchaeota inhabit subsurface waters in order to reduce ammonia-oxygen competition25,26,27, and during the rainy season, Thaumarchaeota are more active in surface waters compared to other archaea28. Soil Thaumarchaeota are dominated by Group 1.1b, which can produce abundant crenarchaeol (Cren) and crenarchaeol regioisomer (Cren’)29,30; whereas aquatic Group 1.1a Thaumarchaeota, in addition to the production of Cren and Cren’, can also produce OH-GDGTs. These observations indicate that OH-GDGTs are produced in situ in lakes31. The distribution patterns of isoGDGTs in soils and surface sediments are similar, but OH-GDGTs were only detected in the surface sediments (Supplementary Fig. 1), which is consistent with results from the Lugu Lake basin32. Therefore, the contribution of soil-derived Cren and Cren’ from inputs of terrestrial source Group 1.1b Thaumarchaeota may influence the assessment of the source of Cren and Cren’ in sediments, which may lead to errors in paleohydrological reconstructions22. Habitat selection in subsurface water bodies is such that aquatic Group I.1a Thaumarchaeota produce OH-GDGTs in response to changes in water level30,33. This finding has been verified by abundant data, which confirm the relationship between %OH-GDGTs and water depth22,24 and it enables the reconstruction of past lake levels22,23.

We combined the previous lake shoreline data from Daihai Lake19 with %OH-GDGTs data in the corresponding dated core-sediments to establish a calibration function for water depth vs. GDGTs. This revealed a good positive correlation between water depth and GDGTs (Supplementary Fig. 2), showing the potential of %OH-GDGTs for quantitatively reconstructing paleo-water depths. However, this calibration function lacks control data from shallow water-depths (<25 m), and thus it may be subject to a large bias in the reconstruction of low water-levels.

Hydrological patterns and synchronous drought events in the EASM region since the last deglaciation

An inverse hydrological pattern (i.e., “southern floods (drought)–northern drought (floods)” occurred between northern and southern China during the last deglaciation9, the middle Holocene7,34, and the Little Ice Age8. In northern China a wetness maximum in the middle Holocene has been proposed, which is supported by pollen-based precipitation reconstructions from Gonghai Lake2 and Daihai Lake17, and by the increased frequency of paleosol development on the Chinese Loess Plateau, which also indicate a middle Holocene wetness maximum11. Lake level changes, indicating changes in environmental humidity, are the integrated effects of precipitation, evaporation, and surface runoff recharge. High lake-levels in the early Holocene are demonstrated by studies of lake shorelines at several lakes, including Dali Lake13, Daihai Lake19, and Angulinao Lake15, and these high levels were maintained from the early through middle Holocene. Additionally, Hunshandake Sandy Land was an extremely wet environment during the early Holocene14. A nutrient-poor freshwater environment at this time also indicated a maximum in lake volume31,35. An early Holocene lake level maximum was also determined from the lakeshore embankments of Chagan Nuur21 and Angulinao Lake15. This interpretation of a wet early Holocene in northern China is controversial and two alternative hypotheses have been proposed to explain it: strong monsoon precipitation13,14, and increased snow/ice melt runoff recharge12,36,37

The timing of the reconstructed precipitation maximum (the upper limit of estimates), based on pollen assemblages, falls consistently within the early–middle Holocene38, which indicates that we cannot exclude the role of extreme precipitation in contributing to the observed high lake-levels. Depleted δ2H29 values in the sediments of Daihai Lake during the early–middle Holocene suggest the gradual enhancement of the Asian summer monsoon (Fig. 2). However, this averaged-state enhancement cannot fully explain the overall high water-level in the early–middle Holocene based on the water depth responses indicated by %OH-GDGTs, despite the good positive correlation between %OH-GDGTs and δ2H29 (Supplementary Fig. 3). This indicates that the hydrological evolution of Daihai Lake on a long timescale is modulated by the Asian summer monsoon, and that the early Holocene high water-levels may be related to recharge by non-monsoon precipitation.

Fig. 2: Hydrological records of the ASM region since the last deglaciation.
figure 2

a %OH-GDGTs (blue circles) and δ2H (cyan crosses) from Daihai Lake. b Lakeshore elevation (orange squares19) and a pollen-based annual precipitation reconstruction from Daihai Lake (blue line17). c RIK37-based lake salinity record from Dali Lake (green circles and line35), and a precipitation record (the upper limit of estimates) from Bayanchagan Lake (dark blue circles and lines38). d Pollen-based annual precipitation reconstruction from Gonghai Lake (blue line2), and a synthesis of palaeosol development in the Chinese Loess Plateau (red open squares11). e Hydroclimatic record of water-table depth from the Dajiuhu Peat based on hopanoid flux (orange circle39) and CIA from Nanyi Lake (red line34): f tropical + subtropical arboreal taxa (blue line60) and TOM (green line67) from Huguangyan Maar Lake.

Full size image

Due to the influence of the Meiyu belt and ENSO, the hydrological pattern of southern China is the opposite to that in northern China9,34, except during the early Holocene when a common pattern of hydrological evolution occurred—a gradual increase in precipitation and high environmental humidity. The high concentration of hopanoids in Dajiuhu peatland in the middle Holocene suggests an arid environment39, which agrees with the trend of δ2H40. In these two records, extreme drought events can be clearly recognized, and in addition, pollen records from the middle and lower reaches of the Yangtze River indicate abrupt decreases in precipitation at ~7–6 cal kyr BP, 7.7–7 cal kyr BP, and ~4.9–4.2 cal kyr BP41,42 Similarly, in central China, low values of the environmental magnetic parameter IRMsoft-flux, which was linked to ENSO-related storms, suggest an arid environment during the middle Holocene43.

Although the dipolar hydrological pattern between northern and southern China developed in the middle Holocene, rapid drought events occurred almost synchronously throughout the regional influenced by the ASM. This is demonstrated by low precipitation, low environmental humidity, and the salinization of lakes due to falling water levels. (Fig. 2). Here, we identify at least seven well-defined drought events: during 11.7–11.3 cal kyr BP, 10.6–10.2 cal kyr BP, 9.6–9 cal kyr BP, 7.6–7.3 cal kyr BP, 7–6 cal kyr BP, 4.7–4.2 cal kyr BP, and 3.9–3.4 cal kyr BP.

Freshwater forcing led to independent early Holocene drought events in northern China

The variability of the ASM and its precipitation has been explained by regional-scale differences in ocean–land thermal gradients, influenced on orbital timescales by insolation and high-latitude Northern Hemisphere cryospheric forcing, and these variations show pronounced precession and obliquity cycles18,44,45. Ice-sheet forcing impeded the northward extension of the early Holocene ASM precipitation, which was proposed as the cause of the mismatch between monsoon precipitation and monsoon circulation2. Meltwater inputs from melting ice sheets or glaciers weakened the thermohaline circulation in the North Atlantic, and subsequently the summer monsoon precipitation decreased, most notably during the early Holocene intervals of 11.7–11.3 cal kyr BP and 9.6–9.0 cal kyr BP46,47. The Pre-Boreal Oscillation (PBO) has occasionally been detected outside the Atlantic region48,49. This event is documented in sedimentary records from mid-latitude East Asia, including Jeju Island in Korea, and in Hunshandake Sandy Land and Daihai Lake13,14,50. However, a weakening of monsoon activity during the PBO is not evident in the oxygen isotope record of Hulu cave51, and precipitation in southern China does not appear to have been affected by meltwater forcing in the early Holocene (Fig. 3)

Fig. 3: Links between rapid drought events in the ASM region and the ocean–atmosphere circulation.
figure 3

a %OH-GDGTs (blue circles) and δ2H (light blue cross) from Daihai Lake. b δ18O record from Dongge cave (blue line68) and integrated ISM proxy based on four cave δ18O records (green line69). c Global sea-level70. d Ti concentration from the Cariaco Basin71. e ENSO variability (1.5–2.7 years)72. f ΔSST of the western73 and eastern74 tropical Pacific Ocean. g ΔT of the Northern and Southern Hemispheres (ΔT = TNH − TSH, solid line75, dashed line76).

Full size image

Linkages between abrupt drought events and the weakening of inter-hemispheric temperature gradients

A “Jet transition hypothesis” of seasonal monsoon change has been proposed, which suggests that East Asian summer monsoon precipitation variability is influenced by the timing and duration of the transition to the Meiyu season52,53. Additionally, at ~6 cal kyr BP, the rapid increase in the meridional temperature gradient caused by the rapid expansion of Atlantic sea ice delayed the northward advance of the seasonal ASM, leading to the rapid shift to drought conditions in lakes and the ending of the Holocene humid period in the Indo–East Asian monsoon region19. Although a precipitation decrease or drought during 5.9–5.5 cal kyr BP has also been observed in the East Asian summer monsoon region (Fig. 2), it does not appear to have been an extreme event. A more extreme drought event occurred in the East Asian monsoon region during 7–6 cal kyr BP, which appears to have been spatially discontinuous based on a summary of lake shoreline studies19, possibly related to very low water levels resulting from a megadrought. Therefore, the limited high-latitude sea-ice extension events cannot fully explain the extreme drought events that occurred in the East Asian monsoon region during the middle and late Holocene, especially when the sea level was relatively close to the present-day level.

The inter-hemispheric temperature gradient is important in regulating the latitudinal position of the planetary atmospheric circulation54; and changes in this gradient are driven by insolation, which controls the position of the Hadley circulation terminus, the Hadley circulation, and the ITCZ55,56,57. The ITCZ, as an important feature of the global climate system, regulates the global monsoon, and its position is regulated by the inter-hemispheric temperature gradient (ΔTN-S)58,59. When the inter-hemispheric temperature gradient decreases, the ITCZ moves southward and the rainbelt is unable to penetrate deeper into the interior of East Asia, thus weakening the intensity of the monsoon rainfall. The role of the ITCZ together with sea level may explain why a marine sedimentary phytolith record failed to track the variations in the tropical Pacific meridional land-ocean climate system during the Holocene60. Warming of the tropical western Pacific strengthens the Hadley circulation61, and thus the enhancement of the lower tropospheric low-latitude return flow (northeasterly winds) reduces the ability of the EASM to move northward62. During abrupt drought events, changes in the latitudinal climate system can lead to the onset of weakened monsoon precipitation (Figs. 3 and 4).

Fig. 4
figure 4

Schematic diagram showing the mechanisms of  extreme drought events in East Asia: the weakened inter-hemispheric temperature gradient leads to the southward shift of the ITCZ and the suppression of the ASM by the Hadley circulation.

Full size image

Seasonal drought events in tropical East Asia are controlled by changes in the zonal SST gradient in the tropical Pacific Ocean, and by the westward shift of the Western Pacific sub-tropical high (WPSH), due to changes in the Hadley circulation60. At the onset of drought events in the middle and late Holocene, the Walker circulation strengthened due to the intensified zonal SST gradient (ΔTW-E) in the tropical Pacific Ocean, in response to warming SSTs in the western Pacific (Fig. 3). At the same time, the westward advance of the WPSH expanded its region of influence in East Asia, leading to zonal synoptic drought in the ASM region. This phenomenon cannot be explained by the movement of the rainbelt controlled by north–south shifts of the WPSH, which may be due to the blocking effect of the enhanced Hadley circulation on the transport of water vapor to middle and high latitudes62. Our conclusions indicate that extreme hydrological events in the EASM region during the middle and late Holocene were caused by changes in latitudinal and zonal SST gradients in the tropical Pacific. This is based on the rise of sea level to a relatively stable level, as a falling sea level would expand the land area and thus reduce the role of water vapor as a recharge mechanism for inland lakes; this would be coupled with a weakening of the East Asian monsoon precipitation caused by changes in latitudinal or zonal climatic regimes.

Methods

Regional setting and sample collection

Daihai Lake (40°29′–40°37′N, 112°33′–112°46′E, 1218 m a.s.l) is a hydrologically closed lake in the arid–semiarid region of northern China, located in Liangcheng County, Inner Mongolia, China. Daihai Lake has an area of 68.67 km2 and maximum water depth of ~7 m. The lake is mainly recharged by precipitation and by six rivers (the Gongba, Wuhao, Buliang, Tiancheng, Muhua, and Suodai rivers). The area has a typical East Asian summer monsoon, with summer temperature and precipitation maxima. The average annual temperature is 5.9 °C and the average annual precipitation is 406.4 mm, of which 80% occurs in summer; however, the average annual evaporation exceeds 1100 mm, which is almost three times the annual precipitation63,64.

In October 2020 we collected a 1235-cm-long continuous sediment core (DH20B, 40.57858°N, 112.672831°E) at the water depth of 6 m, using a UWITEC piston corer. This core was subsampled at 1-cm intervals in the laboratory and the samples were then freeze-dried prior to lipids extraction.

Chronology

Sixteen samples from core DH20B were used for AMS 14C dating and assessment of the carbon reservoir. They included four samples of plant remains, eight bulk sediment samples, and four samples of pollen concentrates. The radiocarbon ages of the plant remains and bulk samples were determined by Beta Analytic (Florida, USA), and those of the pollen concentrates were determined in the Key Laboratory of Western China’s Environmental Systems, Lanzhou University. An age-depth model for core DH20B was constructed using the Bacon program implemented in “R” software with the IntCal 20 calibration curve65, and the basal age of the core was estimated by extrapolation (Supplementary Fig. 4).

Lipids extraction and GDGTs analysis

A total of 211 samples from core DH20B, 5 soil samples and 4 surface sediment samples were used for lipids extraction and GDGTs analysis. The samples were homogenized and ultrasonically agitated four times using dichloromethane: methanol (v:v = 9:1), and then concentrated to dryness with N2 gas. The total lipid extracts were redissolved with hexane, analyzed using silica gel column chromatography, and then sequentially eluted with hexane, dichloromethane, and methanol. The polar fraction (containing GDGTs) was eluted with methanol and then filtered using 0.22 μm PTFE filters prior to LC–MS/MS analysis.

GDGTs analyses were performed using an Agilent 1290 II series ultra-performance liquid chromatography-atmospheric pressure chemical ionization-6465B triple quadruple mass spectrometer (UPLC-APCI-MS/MS). A flow rate of 0.3 ml/min was used, and an aliquot of 10 μl was injected and separated using two Hypersil Gold Silica columns in sequence (each 150 mm × 2.1 mm × 1.9 μm; Thermo Fisher Scientific; Lithuania), maintained at 40 °C. GDGTs were eluted isostatically with 84% A and 16% B for the first 5 min, where A = n-hexane and B = Ethyl acetate, followed by a linear gradient change to 82% A and 18% B from 5 min to 65 min, changed linearly to 100% B for 10 min to separate OH-GDGTs (including OH-GDGT-0, OH-GDGT-1, and OH-GDGT-2), then with 100% B maintained for 5 min to wash the system, and then back to 84% A and 16% B to equilibrate the pressure. Analyses were performed with the selective ion monitoring (SIM) mode to track m/z 1302, 1300, 1298, 1296, 1292, 1050, 1048, 1046, 1036, 1034, 1032, 1022, 1020, 1018, 744 (C46). The extraction and analysis of OH-GDGTs and brGDGTs from core DH20B used the same set of experiments and analytical systems. GDGTs were extracted and analyzed at the Key Laboratory of West China’s Environmental System (Ministry of Education), Lanzhou University.

Compound-specific hydrogen isotopes analysis

The GC injector was set at 280 °C, and the oven heating program was as follows: hold at 50 °C for 1 min, heat to 210 °C at 10 °C/min (hold 2 min), ramp (heat) to 300 °C at 4 °C/min, and finally to 310 °C at the rate of 10 °C/min. The internal standard (Squalane, δ2H = −167‰) was used to calibrate the compound-specific δ2H values. Results are reported in the delta notation (‰) relative to the Vienna Standard Mean Ocean Water standard66. δ2Hwax was determined using a Trace GC coupled with a Delta V advantage isotope ratio mass spectrometer in the State Key Laboratory of Biogeology and Environmental Geology, China University of Geosciences, Wuhan.