Late Quaternary glaciations in the Taniantaweng Mountains

Abstract

Constraining the timing and extent of Quaternary glaciations in the Tibetan Plateau (TP) is significant for the reconstruction of paleoclimatic environment and understanding the interrelationships among climate, tectonics, and glacial systems. We investigated the late Quaternary glacial history of the Qinggulong and Juequ valleys in the Taniantaweng Mountains, southeastern TP, using cosmogenic ¹⁰Be surface exposure dating. Four major glacial events were identified based on 26 ¹⁰Be ages. The exposure ages of the oldest late Quaternary glaciation correspond to Marine Oxygen Isotope Stage (MIS) 6. The maximum glacial extent was dated to 48.5–41.1 ka (MIS 3), during the last glaciation, and was more advanced than that of the last glacial maximum (LGM). Geochronology and geomorphological evidence indicate that multiple glacial fluctuations occurred in the study area during the Early–Middle Holocene. These glacial fluctuations likely were driven by the North Atlantic climate oscillations, summer solar insolation variability, Asian summer monsoon intensity, and CO₂ concentration.

INTRODUCTION

Glaciers have shaped the Tibetan Plateau (TP) land surface, as well as its surrounding mountains. The paleoclimatic information preserved in glacial deposits can provide key evidence for exploring the climatic change of the TP since the Quaternary, and the scientific basis for predicting the effects of glacial changes on human environment under future climatic conditions (Immerzeel et al., 2010; Yao et al., 2012).

In recent decades, researchers studied the glacial chronology using in-situ terrestrial cosmogenic nuclide (TCN) exposure dating, which revealed the complex pattern of Quaternary glacial development on the TP (Dortch et al., 2013; Murari et al., 2014). Results show that the Marine Isotope Stage (MIS) 6 represents one of the oldest glacial periods in the semi-arid areas in the central and western TP (Schäfer et al., 2002; Owen et al., 2005), such as the Tanggula Mountains and the more humid southeastern TP (Zhou et al., 2007; Fu et al., 2013; Chevalier and Replumaz, 2019). However, many moraines from this period are poorly preserved, so it remains difficult to explore their actual spatial distribution. After MIS 6, glacial advances of the central and western TP appear to have been most common and widely distributed during MIS 3 (Li et al., 2014; Xu and Glasser, 2015; Blomdin et al., 2016); these glaciers were no smaller than, or may have been even larger than, those of the last glacial maximum (LGM). The glaciers in the humid areas of the southeastern TP reached their greatest extent during the LGM (Owen and Dortch, 2014), although there is a lack of chronological evidence for any glacial development during MIS 3 (Chevalier and Replumaz, 2019). It is thus clear that there are significant regional differences in glacial changes of the TP and its surrounding mountains during the LGM and MIS 3. In addition, recent studies showed that frequent climatic fluctuations during the Holocene triggered multiple glacial advances and retreats in most parts of the TP (Solomina et al., 2015; Saha et al., 2018), although little attention has been paid to the exposure chronology of Holocene glaciers in the southeastern TP.

This article focuses on the central section of the Taniantaweng Mountains, which is located in the western segment of the Hengduan Mountains. This area marks the transitional zone between the southeastern TP and the Yunnan Guizhou Plateau (YGP) (Fig. 1). Using electron spin resonance (ESR) and optically stimulated luminescence (OSL) dating methods, we explored the distribution, evolution, and development of Quaternary glacial landforms on both sides of Yuqu River (Zhang and Chai, 2016; Zhang et al., 2019; Chai et al., 2022). The results constrained at least four glacial stages since the late Quaternary that correspond to MIS 6, MIS 3, MIS 2, and MIS 1. After MIS 6, glaciers gradually decrease. The MIS 6 glaciers were the largest, ranging 25–35 km in length, generally reaching the mouth of glacial valley, and extending down towards the Yuqu River valley. The lengths of glaciers were 12–15 km during MIS 3 and were 7–8 km during MIS 2. Multiple MIS 1 terminal moraines are well preserved within 1–5 km of the termini of modern glaciers. However, ESR dating of glacial deposits is controversial owing to the potential incomplete reset of ESR signals (Yi et al., 2016). In previous studies, a major source of OSL dating uncertainty was whether mineral particles were exposed to sufficient daylight to reset the luminescent signals (Duller, 2006). TCN dating not only can measure the exposure ages of glacial landforms, but also determine the burial ages of moraines, greatly promoting the development of Quaternary glacial chronology in the TP (Heyman et al., 2011; Heyman, 2014; Chen et al., 2015). This study used the TCN dating method to explore the Quaternary glacial sequence of the study area, which provides important clues for understanding the dynamic processes and mechanisms driving the development of mountain glaciers in monsoonal regions.

Figure 1. Map showing the study area. (a) Shaded relief map of the Tibetan Plateau (TP). (b) Relief map of the southeast TP illustrating available ¹⁰Be data across the area; locations of Figures 2 and 4 also shown. (1) Basongcuo (Hu et al., 2017); (2) Boduizangbu valley (Zhou et al., 2007); (3) Zhuqin (ZDG) (Chevalier et al., 2018); (4) Ganzi (GZ) (Chevalier et al., 2018); (5) Xinlong Plateau (Graf et al., 2008); (6) Cuopu (Chevalier et al., 2016); (7) Litang County (Schäfer et al., 2002); (8) Gemuxiang (Chevalier and Replumaz, 2019); (9) Daocheng (Fu et al., 2013); (10) Selaha (SLH) (Bai et al., 2018); (11) Yangjiagou (YJG) (Bai et al., 2018); (12) Kandding(KD) (Strasky et al., 2009); (13) Gongga Mountain (Owen et al., 2005; Wang et al., 2023); (14) Siguniang Mountain (Wang et al., 2023); (15) Xuebaoding (Liu et al., 2018); (16) Yulong Mountain (Kong et al., 2009).

REGIONAL SETTING

The central section of the Taniantaweng Mountains (30°40′–30°11′N; 96°39′–97°16′E) is located in the western segment of the Hengduan Mountains. The mountains stretch in a NW-SE direction, with an altitude of ~4000–5500 meters above sea level (m asl). Overall, the Hengduan Mountains are higher in the north than the south, and higher in the west than in the east. The ground surface of this mountain range fluctuates, with an altitudinal difference between its highest and lowest points of <1000 m asl (Su and Pu, 1996). The watersheds of both the Nujiang and Lancang rivers are found in the study area. The main river is the Yuqu River, which is a primary tributary of the Nujiang River. In terms of geological structure, the study area is located within the Changdu Paraplatform, which is bounded by the Nujiang Fault Zone to the west, the Lancang River Fault Zone to the east, and the Wahe–Bangda–Chawalong Fault Zone in its central section. The Yuqu River system developed along a NW-SE fault zone. The exposed strata in the area are mainly composed of granites, limestones, sandstones, and ultrabasic rocks of Devonian, Triassic, and Jurassic periods (Zhang and Chai, 2016).

The area is situated along the South Asian Monsoon (SAM) pathway and has an alpine, continental climate. The advance and retreat of glaciers are closely correlated with fluctuations of the SAM. The mean annual precipitation (MAP) is 474.2 mm, and the mean annual temperature (MAT) is 7.6°C. This area has clear divisions between dry and wet seasons. Modern glaciers in the study area belong to the transitional zone between marine and subcontinental glaciers and include cirque and hanging glaciers (Su and Pu, 1996). According to the Randolph Glacier Inventory 6.0, which was released in 2017, there are 88 modern glaciers in the study area, covering a total area of 12.955 km² (Preffer et al., 2014). The sizes of the glaciers, which are mainly distributed on the western bank of the Yuqu River, are generally small; their termini extend down to 5100–5200 m above sea level (asl). The modern snow line is located at ~5400 m asl.

Quaternary glacial remnants are preserved above 4200 m asl in this area. The distribution of Quaternary glacial erosional and depositional landforms indicates that large-scale piedmont glaciers, valley glaciers, and a large number of cirque glaciers existed in the past.

SAMPLING AND ANALYZING METHODS

Geomorphological mapping and sample collection

Google Earth high-resolution images and a 30-m-resolution digital elevation model (DEM) from the Shuttle Radar Topography Mission (SRTM, http://www.gscloud.cn/sources) were used to identify and map glacial features along the Yuqu River (Zhang et al., 2019), focusing on the Qinggulong and the Juequ valleys (Fig. 1). The morphostratigraphic relationships that we investigated focused on the shape, weathering, and distribution of glacial landforms (Hughes et al., 2005). The relative ages of moraines were established in the field, and compared with other moraine sequences using absolute chronological data from the southeastern TP. The moraines in the Qinggulong and Juequ valleys were named QM4–QM1 and JM5–JM1, respectively. Glacial boulders on the tops of moraine ridges were sampled for ¹⁰Be exposure dating. The ideal boulder samples are ones that were not exposed before deposition of moraines, but have been exposed on the tops of moraine ridges since deposition without rolling or surface erosion. This means that the exposure time of boulders closely corresponds to the formation age of the specific moraine ridge (Lal, 1991; Gosse and Phillips, 2001). We used the morphostratigraphic relationships between glacial and associated landforms to determine our sampling strategy. The size and lithology of each boulder were photographed and recorded. The sampling latitude, longitude, and elevation were measured using a handheld global positioning system (GPS) instrument in the field (Zhang et al., 2016). Twenty-six samples were collected from the Qinggulong and Juequ valleys in 2017 and 2018 for ¹⁰Be exposure dating, two of which were from bedrock surfaces.

Sample preparation and measurement

All samples were prepared in the Key Laboratory of Tibetan Environment Change and Land Surface Processes, Chinese Academy of Sciences (CAS), Lhasa, followed the methods of Kohl and Nishiizumi (1992). After physical and chemical separation and purification, 20–60 g of pure quartz were dissolved with the addition of ~0.3 mg ⁹Be carrier. The solution was purified using defluorination and transesterification. Next, ammonium hydroxide was added to the Be solution to prepare hydroxide and dehydrated to extract BeO. Finally, the BeO sample was mixed with Nb powder to create an accelerator mass spectrometry (AMS) measurement target. The ¹⁰Be/⁹Be ratios were measured at the Xi'an Accelerator Mass Spectrometry Centre (Xi'an-AMS Centre), using a certified ratio of 2.851×10⁻¹² (Nishiizumi et al., 2007), corresponding to the 07KNSTD standard. The measured isotope ratios were corrected using the values of two blank samples (4.57 ± 2.04 ×10⁻¹⁵ and 1.17 ± 0.68 × 10⁻¹⁵).

Exposure age calculation

All ¹⁰Be exposure ages were calculated using the updated CRONUS-Earth V3 calculator (Balco et al., 2008; http://hess.ess.washington.edu/math) and St time-independent (Lal, 1991; Stone, 2000), Lm time-dependent (Balco et. al., 2008), and LSDn production rate (Lifton et al., 2014) scaling. We applied a default ¹⁰Be production rate derived from the primary calibration dataset of Borchers et al. (2016). The ¹⁰Be ages and other parameters for the samples are shown in Table 1. The calculation was based on the assumption that the samples collected were not affected by previous exposure or incomplete exposure. The surface erosion rate was assumed to be zero. Because precipitation in the study area may affect the erosion rate of the boulders, we also calculated the dating results under different the erosion rates: 1 mm/ka, 2 mm/ka, and 3 mm/ka. The rock density was set to 2.65 g/cm³. The topographic shielding factor was calculated using a Python tool and the 30-m ASTER DEM (Li, 2013). Because of dry climatic conditions and sparse vegetation, the snow/vegetation cover at the sampling site was not corrected, potentially underestimating exposure ages. Therefore, each exposure age calculated in this study represents the minimum age. Detailed sample information is provided in Table 1.

Table 1. Cosmogenic¹⁰Be surface exposure dating sample details and¹⁰Be exposure ages from the Taniantaweng Mountains

¹Samples JQB-02, 03, 05, 07, QGLG-01–11, with blank of 797133.253 ± 356477.991 at g⁻¹ carrier (0.2521 g, 1148.95 ppm Be, 4.57 ± 2.04 × 10^{−15 10}Be/⁹Be)

²Samples JQB-08–10, 14, 16, 17, 22, 24, 25, 27-28, with blank of 259107.985 ± 149608.950 at g⁻¹ carrier (0.3352 g, 1097.1 ppm Be, 1.17 ± 0.68 × 10^−15.10Be/⁹Be)

This study focused on the calculation of ages based on the LSDn scaling model because the LSDn model predicts scaling behavior more consistently at high-altitude and low-latitude locations (Lifton et al., 2014). To compare with previously published data from the southeastern TP, we recalculated these data using the same scaling model. Both internal and external uncertainties for each ¹⁰Be exposure age are reported in Table 2.

Table 2. Apparent ¹⁰Be exposure-ages calculated using three scaling schemes.

Moraine age determination

The age determination of moraines using the TCN dating method is complicated owing geological uncertainties. However, the application of appropriate statistical methods can effectively improve the accuracy of moraine chronologies. We used the Probabilistic Cosmogenic Age Analysis Tool (P-CAAT) (Dortch et al., 2022) to quantitatively exclude outliers in the dataset and to explain the uncertainties in geomorphic processes. The P-CAAT incorporates both sample age and geological uncertainties and uses Monte Carlo simulations to isolate component normal distributions from a cumulative probability density estimate for chronological datasets with three or more samples, using the following formula:

$${\rm t} = \sigma + P\ast \left({\displaystyle{{E-I} \over A}} \right) $$

where t is the total landform-age uncertainty, σ is the one-sigma bound of the Gaussian curve, P is the peak of the Gaussian curve, and E, I, and A represent the external uncertainty, internal uncertainty, and ¹⁰Be exposure ages, respectively. The cluster degree of the samples was decided by the STD/IQR bandwidth estimator in P-CAAT. The internal uncertainty was used to compare the ¹⁰Be exposure ages within the study area, and the external uncertainty was used to compare the exposure ages between different sites.

To ensure an effective comparison with other glacial regions in the southeastern TP, the same P-CCAT method was applied to previously published data, considering only well- and moderately clustered groups with n ≥ 3 ¹⁰Be samples (Dortch et al., 2022). The ¹⁰Be ages from some regions failed to meet the above conditions and were excluded (e.g., those of Wang et al., 2006, for which ¹⁰Be concentrations and AMS test standards were not available). The chronological results for moraines with two- or single-sample and dispersed boulder-sample groups also were excluded (e.g., Schäfer et al., 2002, Lit 7 sample; Graf et al., 2008, TSO-1 sample; Kong et al., 2009, Renhe valley, Ganheba valley, and Baishui valley; Strasky et al., 2009, K101 sample; Fu et al., 2013, A, F, G, I, K L, O, and R moraines; Zhang et al., 2015, in which there was only one sample per group of moraines; Chevalier et al., 2016, b and f moraines; Liu et al., 2018, M1 moraine in the Yanjin valley, Gr moraine in the Rijiu valley; Chevalier and Replumaz, 2019, inner moraine).

Equilibrium line altitude (ELA) reconstruction

Based on the map of the moraines (Figs. 2, 4), we used the accumulation area ratio (AAR) and area-altitude balance ratio (AABR) methods to calculate the equilibrium-line altitude (ELA) of paleo- and modern glaciers. The AAR method assumes that when the glacier is in a stable state, the ratio of the glacier accumulation area to the total glacier area is fixed. The ratios are typically between 0.50 and 0.80—for middle latitude glaciers, they are between 0.55 and 0.65 (Owen and Benn, 2005), and for mountain glaciers, they are 0.65 ± 0.05 (Porter, 1975). An AAR value of 0.60 ± 0.05 was used in this study.

Figure 2. Google Earth image of the study area, showing the locations of moraines in the Qinggulong valley, sampling sites, and ¹⁰Be exposure ages.

Figure 3. Main glacial landforms and photographs showing areas sampled in the Qinggulong valley. (a) Cirque in the Qinggulong Valley; the pink line represents the QM2 moraine; the green line represents the QM1 moraine; (b) view of bedrock sample number QLGB-01; (c–f, h) views of boulder samples; (g) geomorphology of the Qinggulong Valley; the blue line represents the QM3 moraine; (i, j) photographs showing the geomorphic features of moraine QM4.

Figure 4. Google Earth images of the study area, showing the locations of moraines in Juequ valley, sampling sites, and ¹⁰Be exposure ages.

The AABR method is based on the principle that the mass balance far from the ELA exceeds that of the nearby mass balance. This method considers both the reconstructed glacier surface height and mass balance gradient. The results show that the average balance ratio (BR) of the 65 glaciers was 1.75 ± 0.71 (Rea, 2009). In this study, we used this ratio to calculate the former ELA. The calculations were performed using the ArcGIS toolbox developed by Pellitero et al. (2015, 2016).

MORAINE SEQUENCE

As illustrated in Figure 2, four sets of moraines are preserved in the Qinggulong valley. The QM4 moraine ridge is preserved at ~4300 m asl on the northeastern side of Qinggulong Village (Figs. 3i, j), and 12 km from the terminal of the glacier. It extends ~50 m, close to Highway 214, 20–25 m above the bed of the Yuqu River. No boulders were found at the top of the moraine ridge, therefore no ¹⁰Be samples were collected here.

Moraine QM3 covers an altitudinal range of 4600–5250 m asl and is ~7 km long. The most discernible moraines are the two parallel and symmetrical lateral moraines, and the downstream end moraine ridges (Fig. 3g). The exposed profiles show that the lithology is principally composed of granite and phyllite, with little weathering. Seven boulder samples (QGLB-05, 06, 07, 08, 09, 10, 11) were collected from QM3. A large, polished surface is preserved on the right-hand side of the upper reaches of the QM3 moraine ridge, near the roche moutonnée. The lithology is granite; the top surface (5275 m asl) gently undulates and is 80 m from the cirque floor. Bedrock on the roche moutonnée was sampled (QGLB-01) to date the retreat of the main valley glacier (Fig. 3b).

Terminal moraine QM2 is preserved between 5175 and 5290 m asl downstream from the cirque glacier, 60–75 m above the valley floor. It is a grayish yellow color and spans the cirque mouth (Fig. 3a, c). Based on an analysis of the exposed section of the moraine ridge, the moraine is mainly composed of granite, with fresh gravels and a low degree of weathering. We collected samples from three QM2 boulders for ¹⁰Be exposure dating (QGLB-02, 03, 04).

Terminal moraine QM1 is located at the end of the modern glacier, 80 m above the valley floor (Fig. 3a), at an altitudinal range of 5270–5360 m asl. A large quantity of gravels and small granite boulders are preserved at the top of QM1. No samples were collected from this moraine because there were no suitably large boulders.

The five sets of moraines in the Juequ valley become progressively more weathered and extensive down-valley (Fig. 4). JM5, which is preserved on both sides of the mouth of the Juequ valley (Fig. 5a), indicating the largest glacial extent in this area, is ~25 km from the cirque and 200–300 m higher than the riverbed. Low shrubs (0.5~1 m) grow on top of the moraine, and many granite boulders are scattered on its top. The surfaces of some boulders are extensively weathered. We collected samples from four boulders for ¹⁰Be exposure dating (JQB-07, 14, 16, 17).

Figure 5. Views of moraines. (a) View of the JM5, JM4, and JM3 moraines (tributary valley of the Juequ valley); (b) view of the JM3 Moraine (main valley of the Juequ valley).

JM4 is preserved on both sides of the river at the bottom of the Juequ valley mouth (Fig. 5a), reaching 4400–4600 m asl. We collected samples from four of the multiple scattered granite boulders on JM4 for ¹⁰Be exposure dating (JQB-05, 08, 09, 10).

JM3 is a terminal moraine (4600–4770 m asl) at the intersection of the main Juequ valley and a branch valley, 13 km from the terminal of the glacier (Fig. 5b). Granite boulders are scattered on the moraine, and low shrubs grow on its top. We collected samples from three boulders for ¹⁰Be exposure dating (JQB-02, 03, 22).

JM2 moraine (4850–5080 m asl) extends downstream to Xionggacuo, across a roche moutonnée, at the cirque mouth (Fig. 6). We collected samples from three JM2 boulders for ¹⁰Be exposure dating (JQB-24, 25, 28), and one sample from bedrock on the roche moutonnée (JQB-27).

Figure 6. Views of the JM2 moraine and sample localities.

JM1 moraine is preserved in the form of an end moraine within 2 km from the terminus of the modern glacier. The granite boulders on top of the moraine were thoroughly intermixed with weathered bedrock, therefore no suitable ¹⁰Be samples were found.

RESULTS AND INTERPRETATIONS

The 10 boulder and one polished-surface samples collected from the Qinggulong valley yielded apparent ¹⁰Be exposure ages ranging from 44.9 ± 0.3 ka to 6.5 ± 0.2 ka. The 14 glacial boulder and one polished-surface samples from the Juequ valley yielded apparent ¹⁰Be exposure ages ranging from 107.6 ± 0.6 ka to 3.4 ± 0.2 ka. The dating results were consistent with the geomorphologic stratigraphy observed in the field. All ¹⁰Be ages are shown in Table 2 and Figures 7 and 8.

Figure 7. Gauss peak ¹⁰Be ages and their uncertainties were analyzed by P-CAAT in two glaciated valleys of the Taniantaweng Mountains. PDE = probability density estimate. The three samples from JM3 yielded ages of 27.6 ± 0.6 ka, 20.9 ± 0.6 ka, and 12.5 ± 0.05 ka. The P-CAAT analysis results indicate that the age of 12.5 ± 3.7 ka (n = 2) may represent the time of deglaciation. However, the most well-preserved TP glacier landform with a true formation age between 8 ka and 44 ka is the moraine ridge during the LGM period (Zhou et al., 2007; Fu et al., 2013; Chevalier et al., 2018). Therefore, JM3 was possibly formed during the LGM, and another peak value of 24.8 ka from P-CAAT analysis may represent the true formation age of JM3.

Figure 8. ¹⁰Be ages obtained using different rates of erosion for the LSDn production rate model (a–d). (a) Age with 0 mm/ka erosion rate; (b) age with 1 mm/ka erosion rate; (c) age with 2 mm/ka erosion rate; and (d) age with 3 mm/ka erosion rate. Triangles, circles, stars, and squares represent the ages of boulders on moraines JM5, JM4, JM3, and JM2, respectively. Inverted triangles and diamonds represent the ages of boulders on moraines QM3 and QM2, respectively. (e) The blue line is the δ¹⁸O record of the global stack (Lisiecki and Raymo, 2005).

Qinggulong valley

Seven exposure ages were obtained from Moraine QM3, ranging from 44.9 ± 0.3 ka to 12.3 ± 0.3 ka. The oldest exposure age, 44.9 ± 0.3 ka (QGLB-09), is significantly different from other ages. The sampling location of QGLB-09 was at the front end of the moraine, which might have recycled some boulders exposed prior to the glacial advance. We infer that the QGLB-09 boulder may be related to the glacial event corresponding to the QM4 Moraine, with a much older exposure age. The P-CAAT analysis constrains the age of this moraine to 15.3 ± 2.8 ka (n = 7) (Fig. 7). This age is consistent with previous ages determined using ESR and OSL dating techniques (Fig. 2) (Zhang and Chai, 2016; Zhang et al., 2019), reflecting that a glaciation event occurred in the Qinggulong valley during the late glacial period. The exposure time of a glacially polished surface generally represents the time of glacier disappearance/retreat, if the glacier eroded 2–3 m of bedrock to completely remove the inherited nuclide (Dielforder and Hetzel, 2014). The exposure age of sample QGLB-01 from the ridge of the roche moutonnée suggests that the glacier in the Qinggulong valley retreated or disappeared from this ridge at ca. 14.9 ± 0.2 ka (Fig. 2).

The three samples from QM2 (QGLB-02, 03, 04) yielded ages of 6.5 ± 0.2 ka, 8.7 ± 0.2 ka, and 8.4 ± 0.2 ka, respectively. P-CAAT analysis constrained the age of this moraine to 8.5 ± 0.9 ka (n = 2) (Fig. 7). This indicates that a glacial event occurred during the Early Holocene.

Juequ valley

The four samples from JM5 yielded ages ranging from 107.6 ± 0.6 ka to 14.2 ± 0.9 ka. The P-CAAT analysis showed that the age of this moraine is 96.1 ± 64.0 ka (n = 3) (Fig. 7). Moraine degradation may have led to the overturning or tilting of granite boulders, leading to much younger apparent exposure ages. In addition, field observation indicated that the granite boulders on the surface have undergone obvious weathering and denudation, and some granite boulders have been weathered into detritus. The location, sedimentary profile, and weathered boulders of the moraine all support that this moraine was formed earlier than the oldest sample age obtained. Combining these results with the other chronological and geomorphological characteristics, we infer that this glaciation event may have occurred during MIS 6.

The four samples from JM4 yielded ages ranging from 87.3 ± 2.4 ka to 41.2 ± 0.8 ka. The P-CAAT analysis constrained the age of this moraine to 44.4 ± 10.0 ka (n = 3) (Fig. 7), which suggests that the JM4 moraine recorded a MIS 3 glacial advance. The age 87.3 ± 2.4 (JQB-10) is a potential outlier.

The three samples from JM3 yielded ages ranging from 27.6 ± 0.6 ka to 12.5 ± 0.05 ka. P-CAAT results indicate that the age of 24.8 ± 7.8 ka (n = 2) may represent the time of deglaciation (Fig. 7), suggesting that the JM3 moraine represents a LGM and transition glaciation event.

The four samples from JM2 yielded ages ranging between 8.7 ± 0.3 ka and 3.4 ± 0.2 ka. The youngest exposure age of 3.4 ± 0.2 ka (JQB-25) is a potential outlier. It was collected from the top of the moraine ridge; the selected boulder may have been affected by a late degradation process, resulting in incomplete exposure. The P-CAAT analysis constrains the age of this moraine to 7.8 ± 1.6 ka (n = 3) (Fig. 7), reflecting an Early Holocene glacial event in the Juequ valley.

In summary, several major glaciation events might have occurred during the late Quaternary in the central Taniantaweng Mountains, corresponding to MIS 6, 3, 2, and 1. Moraine ridges preserved at the termini of modern glaciers (JM1 and QM1) did not yield any usable dates, but investigation of their relative geomorphological characteristics suggests that these moraines may have formed during the Late Holocene. The moraines (JM2 and QM2) are dated to 7.8 ± 1.6 ka and 8.5 ± 0.9 ka, respectively, indicating that a glacial advance occurred during the early Middle Holocene. The dating results for moraines JM3 and QM3 for the last glacial period were 24.8 ± 7.8 ka and 15.3 ± 2.8 ka, respectively, reflecting glacial events during MIS 2. The age extracted from the JM4 moraine ridge (44.4 ± 10.0 ka) indicates a MIS 3 glacial event. Although the application of the ESR dating to Quaternary glacial dating remains controversial, our previous ESR chronology results (54 ± 9 ka and 55 ± 8 ka) for the QM4 moraine in the Qinggulong valley (Zhang and Chai, 2016) were consistent with those of the JM4 moraine in the Juequ valley, indicating that a large-scale glacial advance may have occurred in the central Taniantaweng Mountains during MIS 3. The oldest results for late Quaternary glaciers in the study area may be as old as 96.1 ± 63.6 ka, indicating that there was a large-scale glacial advance in the central Taniantaweng Mountains before the last glacial period that corresponds to MIS 6.

Distribution of ELAs

The paleoglacial ELAs reconstructed using the AAR and AABR methods are listed in Table 3. The ELAs of the modern glaciers in the two glacier valleys were 5392 m and 5530 m. The results showed that the ΔELA value was 614 m during MIS 6. During MIS 3, the ΔELA was 530 m. The ELA for the study area decreased by 368 m and 365 m during MIS 2. During the early Middle Holocene, the ΔELA values were 158 m and 187 m.

Table 3. ELA and ELA depression (ΔELA) for Juequ and Qinggulong valleys (all data are in meters).

DISCUSSION

We summarized 225 previously published ¹⁰Be ages of the 40 moraines in southeastern TP to compare and analyze the similarities and differences in the glacial history between the study area and the whole southeastern TP (Fig. 9 and supplementary materials).

Figure 9. Comparison of ¹⁰Be exposure ages with different rates of erosion for the LSDn production rate model in the southeast TP; moraine ages obtained using P-CAAT. (a) Age with 0 mm/ka erosion rate; (b) age with 1 mm/ka erosion rate; (c) age with 2 mm/ka erosion rate; (d) age with 3 mm/ka erosion rate; (e) δ¹⁸O record of the global stack (Lisiecki and Raymo, 2005). See text for details.

Erosion-corrected exposure ages

Table 2 presents minimum exposure ages, assuming an erosion rate of 0 mm/ka in the Taniantaweng Mountains. However, the erosion rate in the Tibetan Plateau cannot be zero. Wang et al. (2006) estimated the average erosion rate of the bedrock as ~1 mm/ka in the Haizishan area. Therefore, we recalculated the exposure ages of the boulders using erosion rates of 0 mm/ka, 1 mm/ka, 2 mm/ka, and 3 mm/ka for across the southeast TP. The results show that surface weathering and erosion have significant effects on the exposure age of older samples (e.g., the ages of the JM5 moraine are 96.1 ± 63.6 ka, 104 ± 76.6 ka, 114 ± 88 ka, and 129.9 ± 107.6 ka, respectively) (Fig. 8), but have little effect on the exposure age of relatively young boulders (Fig. 9 and supplementary materials). Thus, the exposure ages discussed below are calculated based on an erosion rate of 1 mm/ka.

Timing of glaciations

MIS 6 glacial stage (102.6–199.6 ka)

The exposure ages of JM5 show that the largest glacial extent occurred in the Taniantaweng Mountains around 104.0 ± 76.6 ka (n = 3). Due to the great uncertainty in surface erosion and snow cover, we suggest that the maximum glacial period in the Taniantaweng Mountains corresponds to MIS 6. Similar ages were found in the Shaluli Mountains (Fu et al., 2013; Chevalier and Replumaz, 2019; Wang et al., 2023). In eastern Nyainqentanglha, the exposure age recorded for the Guxiang glacial period was 160.7 ± 86.6 ka (n = 4, GX-A). Based on the ages of the above moraines, the maximum glacial period in the southeastern TP occurred during MIS 6 (102–199.6 ka).

MIS 3b glacial stage (46.3–47.8 ka)

During MIS 3b (46.3 ± 12.8 ka; n = 3), an important glacial advance occurred in the Juequ valley of the southeastern TP, consistent with the Yulong Mountain results (47.8 ± 11.7 ka, n = 5; Kong et al., 2009). With its abundant precipitation, the MIS 3 glaciation is the most common and extensive glacial stage in High Mountain Asia (HMA) (Chevalier et al., 2011, 2022a; Dong et al., 2014). However, little evidence of a MIS 3b glaciation was found in the 225 exposure ages from the southeastern TP. Xu et al. (2010) obtained three OSL geochronological samples (M2) of 49 ± 4 ka, 58 ± 5 ka, and 45 ± 4 ka from the outer Yingpu valley moraine in the Queer Mountains, indicating that a glacial advance occurred in the region during MIS 3b. Some Chinese scholars have also obtained a small number of chronological results using ESR dating that would appear to indicate a MIS 3b glacial advance on the southeastern TP (Zhao et al., 2007; Wang, 2010).

MIS 2 glacial stage (12.9–25.3 ka)

The recalculated ¹⁰Be exposure chronological data show that glacial advances occurred widely between 12.9 ka and 25.3 ka at the southeastern TP (Figs. 1, 9, and supplementary materials), such as eastern Nyainqentanglha, Taniantaweng Mountains, Shaluli Mountains, Daxue Mountains, Siguniang Mountain, and Gongga Mountain. Among the glacial advances at the southeastern TP, most of them were in response to the LGM, while some of them occurred during the late stage of MIS 2.

Younger Dryas (YD) (12.7 ka)

The Younger Dryas event (12.9–11.6 ka) was a global millennium-scale extreme event that occurred during the last glacial–Holocene transition period (Cheng et al., 2020; Gupta et al., 2021). However, in the southeastern TP, only the recalculated ¹⁰Be ages of moraines on Xuebaoding Mountain (SCM 2) (Liu et al., 2018) are identified to be corresponding to the YD event. Although researchers think that the differences in scaling models and production rates are one possible reason for the above phenomenon (Wang et al., 2023), the detailed chronology of YD event glaciation in southeastern TP needs to be studied further.

Holocene

In the southeastern TP, glacial advances during the early Middle Holocene and Neoglacial were identified by a small number of ¹⁰Be ages and assigned to the Holocene. Exposure age groups JM2 and QM2 show that a glacial event occurred in the Taniantaweng Mountains during the Early Holocene and ended in the early Middle Holocene. In addition to the Taniantaweng Mountains, glacial events from the Early Holocene to Middle Holocene also were identified at Gongga Mountain and Xuebaoding Mountain. This implies that deglaciation occurred in the southeast TP during the Early–Middle Holocene. Exposure age group CS-A (1.1 ± 0.8 ka, n = 7) shows that a glaciation co-occurred at Gongga Mountain, and RJM1 (1.5 ± 0.2 ka, n = 2) at Xuebaoding Mountain, which is consistent with deglaciation during the Neoglacial. This implies that deglaciation occurred in the southeast TP during the Neoglacial.

¹⁰Be, OSL, and ESR age comparisons

The ¹⁰Be exposure ages of seven samples collected from the QM3 moraine in the Qinggulong valley ranged between 12.3 ± 0.3 ka and 20.3 ± 0.4 ka (Table 2). The three OSL ages were 17.3 ± 1.3 ka (BDO-4), 31.4 ± 3.4 ka (BDO-5), and 25.8 ± 2 ka (BDO-6) (Zhang et al., 2019). The four ESR ages were 25 ± 1 ka (BD-13), 38 ± 6 ka (BD-14), 26 ± 4 ka (BD-15), and 31 ± 6 ka (BD-16) (Fig. 2). The ESR deposition ages of the QM4 moraine were 54 ± 9 ka (BD-5) and 55 ± 8 ka (BD-4) (Zhang and Chai, 2016). Comparing the results of the ¹⁰Be, ESR, and OSL dating methods for the same moraine provides new support for applying cross-dating methods to constrain the ages of glaciation in other areas.

ESR dating measures the number of paramagnetic electrons, which are generated by ionizing radiation (primarily α-, β-, and γ-radiation) of naturally occurring radioactive elements (e.g., U, Th, K) or cosmic ray irradiation trapped in crystal defects. The glacier formation time is determined based on the last zeroing time of the ESR signal, and then used to infer the time of moraine deposition. Accumulated signals produced by paramagnetic Ge-centers in quartz grains have been used commonly for ESR dating of glacial deposits in Central Asia (Yi et al., 2002; Zhou et al., 2002; Zhao et al., 2010). The ESR dating of moraines assumes that moraine sediments experienced glacial grinding or exposure to sunlight, causing their ESR signals to reset. The efficiency of ESR signal resetting has been tested using sunlight bleaching and manual grinding under laboratory conditions. The results showed a 38% loss in the Ge-center ESR signal of crushed quartz after 1 min of manual grinding (Haldorsen, 1981). The study also suggested that the Ge-center signal of quartz for a mixed mineral sample with a thickness >5 mm could not be bleached under natural sunlight within a week, but reached a stable value (Yi et al., 2016). Therefore, the ESR ages of glacial deposits usually constrain the maximum age of glacial termination (Li et al., 2011; Gribenski et al., 2017). In addition, experimental research has shown that the high silt content in moraine sediments results from glacial crushing and abrasion. It can be assumed that moraine sediments have undergone glacial crushing and abrasion, which may reset the quartz Ge-center ESR signal or lead to the signal reaching a stable residual value (Bi et al., 2016). The four ESR samples from QM3 in the Qinggulong valley all had a fine-grained sandy or silty matrix and were collected at depths >50 cm. However, the ages obtained based on the Ge-center signals were significantly older than the TCN ¹⁰Be exposure ages, potentially because the Ge-center ESR signals were not completely reset.

The OSL dating technique, which calculates the total amount of ionizing radiation received by detrital minerals, including quartz and feldspar, from radioactive elements, such as U, Th, and ⁴⁰K, in the surrounding environment since their last burial (i.e., the equivalent dose) and divides it by the amount of ionizing radiation received by the minerals per unit time (i.e., the annual dose) to obtain the burial time of these detrital minerals, has been widely utilized on glacial sediments (Richards, 2000; Owen et al., 2009; Ou et al., 2015). However, the OSL technique has certain limitations in its application on glacial deposits, specifically from inadequate light exposure or incomplete sunlight bleaching (i.e., the luminescence signals were not zeroed before the last deposition, leaving a residual dose in the samples, which may lead to an overestimation of the OSL age) (Duller, 2006; Bateman et al., 2018). Among the three samples for OSL dating, BDO-4 and BDO-6 were taken from the crest of the moraine ridge, where samples were deposited predominantly on the glacier surface, with greater light exposure, thus the luminescence signals of these deposits were more likely to have been zeroed before burial. In addition, the larger-scale glaciation and the long-distance transport of glacial debris also may have increased light exposure. BDO-5 was collected from the middle and lower parts of the moraine ridge, from more complex debris sources. These sediments may not have received adequate light exposure; thus, their dating results are significantly older.

In conclusion, the geomorphic features of the QM3 moraine in the Qingulong valley are clear, showing little influence of post-glacial processes. Compared with the QM4 moraine, the degree of sediment weathering is less. The results of the three dating techniques varied significantly. In general, the ¹⁰Be exposure ages were the youngest, followed by the OSL ages, and the ESR ages, which were the oldest. One possible interpretation for this phenomenon is that the results of different dating methods have indicate different processes. The OSL and ESR ages indicate the time of moraine deposition, while the ¹⁰Be exposure age indicates the time at which the boulders were exposed on the moraine crests. There may also be a certain time lag between these two events, however, the exact time span is difficult to estimate (Li et al., 2011; Zeng et al., 2019). Another possibility is that the boulders have been susceptible to several other geomorphic processes, such as tumbling, denudation, and burial with sediments, resulting in a younger ¹⁰Be exposure age, while the OSL and ESR ages may have been affected by the presence of residual dose in the samples due to incomplete sunlight bleaching.

Possible driving forces of glacial fluctuations

Glacier fluctuations integrate temperature and precipitation signals. Previous studies have shown that North Atlantic climate oscillations (Seong et al., 2009), summer solar insolation variability (Owen et al., 2005), Asian summer monsoon intensity (Owen, 2009), and CO₂ concentration (Schneider et al., 2013) may be the main driving forces of glacier evolution in the TP.

The ¹⁰Be results indicate that the southeastern TP experienced multiple glacial fluctuations since MIS 6. The glaciers reached their maximum extent during MIS 6 for the Juequ valley in the Taniantaweng Mountains. Moreover, the morphostratigraphic and geochronological evidence in the Shaluli Mountains shows that the largest range of late Quaternary glaciation in the southeastern TP occurred in MIS 6, and most of the geochronological results are consistent with the periodic variation of summer solar radiation at 60°N in the northern hemisphere (Berger and Loutre, 1991; Dong et al., 2022; Fig. 10), and correspondingly high global ice volumes (Lisiecki and Raymo, 2005; Fig. 10). Research shows that the MIS 6 and MIS 2 glacial periods are the coldest periods in Northern Hemisphere (Lisiecki and Raymo, 2005), with an average temperature of 4–9°C lower than today (Shi, 2002), but the MIS 6 glacial moraine on the TP is more extensive than the MIS 2 glacial moraine, possibly because precipitation during MIS 6 was greater than during MIS 2 (Chevalier et al., 2022b). This is consistent with the conclusion that the Asian monsoon reached its peak during MIS 6 using data from Dongge Cave (Cheng et al., 2016; Fig. 10).

Figure 10. (a) All moraine deglaciation ages with 1 mm/ka erosion rate and their uncertainties across the southeast TP; (b) selected deglaciation ages from robustly dated moraines in comparison to the speleothem δ¹⁸O records of Dongge Cave (Cheng et al., 2016), the July insolation at 60°N (Berger and Loutre, 1991), the benthic δ¹⁸O record global stack (Lisiecki and Raymo, 2005), and the CO₂ concentration (Bereiter et al., 2015).

Deglaciation ages of the JM4 moraine in the Juequ valley occurred in MIS 3b. The extent of MIS 3b glacier was no less than that during the LGM. Studies have shown that MIS 3b MATs were lower than present by 5°C (Shi, 2002). In Dalijia Shan, Wang et al. (2015) modeling showed that MAP was 50–100% of present, a temperature depression of 4.2–5.6°C, which allowed the formation of abundant snowfall in high-altitude areas (Finkel et al., 2003), producing conditions that were more conducive to glacier development during MIS 3b. A relatively cold and humid climate facilitated glacier development, indicating that a coupled temperature and precipitation control affected glacier activity during this period.

In addition, glaciers advanced across most of the southeastern TP during the LGM. Compared with the conditions necessary for development of MIS 3b glaciers, the intensity of high-latitude summer insolation during the LGM was weaker (Wang et al., 2008; Fig. 10), the effective humidity was reduced (Herzschuh, 2006), global temperatures were 6–9°C lower than present, and MAP was 30–70% less than today (Shi, 2002). A comparative study of pollen data from Rencuo Lake, which is close to Qinggulong, and records of surrounding regions show annual rainfall of only 250 mm in southeastern Tibet, ~40% of the present-day value. The climate in southeastern TP was also cold, with January temperatures 7–10°C and July temperatures 2–5°C lower than present-day values (Tang et al., 2000, 2004). Dry, cold climate conditions may have been the principal factor in development of LGM glaciers on the southeastern TP.

Studies have shown that the development of glaciers on the Tibetan Plateau may have been controlled by the northward movement of the tropical convergence zone and the enhanced monsoonal and increased precipitation during the Early Holocene (Saha et al., 2018, 2019). Pollen data from Rencuo Lake suggest that the SAM became stronger during the Early Holocene, the mean annual temperature was between 0°C and 1°C, and precipitation was higher than present. During the local Early Holocene glacial stage in the Juequ valley, increases in precipitation transported to the area by the SAM were possibly beneficial to the accumulation of glacial ice (Tang et al., 2000; Owen et al., 2005; Thompson et al., 2005; Fig. 10).

Numerous studies have shown that glacial fluctuations during the late Quaternary were related to North Atlantic cooling in the TP, and correlated with the Heinrich Stadials (Hong et al., 2003) and Bond cycles (Wang et al., 2005; Wanner et al., 2011; Dortch et al., 2013; Murari et al., 2014; Solomina et al., 2015). It is noteworthy that the variation trend of atmospheric CO₂ concentration was consistent with glacier advance (Fig. 10) during the late Quaternary in the southeastern TP. Atmospheric CO₂ changes were key components of past climate variability during the last glacial and deglacial periods (Sigman et al., 2010). Antarctic ice core records have shown that changes in CO₂ concentrations on millennium and centennial scales are consistent with abrupt climate changes during the Heinrich and Dansgaard–Oeschger events in the Northern Hemisphere (Bauska et al., 2018, 2021). Researchers have reasoned that CO₂ also may have been a potential factor affecting glacial development (Marcott et al., 2014; Bereiter et al., 2015; Dong et al., 2020). Thus, fluctuation of late Quaternary glaciers in the southeast TP might be the result of several factors, including insolation changes, Asian summer monsoon intensity, North Atlantic cooling event, and/or changes in CO₂ concentration.

CONCLUSIONS

Twenty-six ¹⁰Be exposure ages from the Juequ and Qinggulong valleys in the central Taniantaweng Mountains constrained four glacial events in the late Quaternary, corresponding to MIS 6, MIS 3, MIS 2, and MIS 1. These results supplement the chronological database of late Quaternary glaciations in the southeastern TP. The oldest glaciation in the study was formed at MIS 6. The MIS 3 glacier was much more extensive than that during LGM glacial events in our study area. In addition, both geochronological and geomorphological evidence indicate that glacial fluctuations occurred during the Early to Middle Holocene (8.5–7.7 ka). North Atlantic climate oscillations, summer solar insolation variability, Asian summer monsoon intensity, and CO₂ concentration are possible influence factors for late Quaternary glacier fluctuations in the study area.