Vegetation changes and associated climatic changes in the southern Altai Mountains within China during the Holocene

Abstract

The location of the Altai Mountains at the limits of both the Pacific and Atlantic influences implies that this mountain range is an important climatic boundary. Based on pollen data of 188 samples of a 390-cm core from Narenxia Peat in the southern Altai with a chronologic support of 11 accelerator mass spectrometry (AMS) dates, we reconstructed the Holocene climatic change at Narenxia Peat. The reconstruction revealed five stages of climatic change: a cold and dry latest deglacial (prior to ~11,500 cal. yr BP), a warm and wet early-Holocene (~11,500 to ~7000 cal. yr BP), a considerably cooled and dried middle Holocene (~7000 to ~4000 cal. yr BP), a resumed warm and wet late-Holocene (~4000 to ~1200 cal. yr BP), and a relatively cool and dry latest Holocene (past ~1200 years). The reconstructions of mean annual temperature (MAT) and mean annual precipitation (MAP) from Narenxia Peat well resemble the reconstructions of North Atlantic Oscillations (NAO) and El Niño–Southern Oscillations (ENSO). The resemblance implies that the Holocene millennial-scale changes in MAT and MAP in the Altai might have been causally associated with the variations in NAO and ENSO.

Keywords

Altai Mountains ENSO Holocene climate NAO peat studies pollen analysis

Introduction

The Altai Mountains range is one of the most prominent mountain ranges in central Asia. It merges with the Western Sayan Mountains in the east and its southeastern extension becomes a different range (i.e. Gobi Altai Mountains). The Altai is politically divided into four parts: Russian Altai in the north, Mongolian Altai in the east, Kazakhstan Altai in the west, and Chinese Altai in the south (Figure 1a). The range is important climatologically because it might have been the conjunctional area between the Westerlies-dominated climates from the west and the Asian Monsoon–dominated climates from the east during the Holocene (Blyakharchuk et al., 2004, 2007, 2008; Harrison et al., 1996; Rudaya et al., 2009; Tarasov et al., 2000). The Altai was close to the centre of scientific attention before AD 2000 in terms of Holocene climate change studies (e.g. Dorofeyuk and Tarasov, 1998; Grunert et al., 2000; Kremenetski et al., 1997a, 1997b; Tarasov et al., 1997). Recent rejuvenation of scientific attention on the Altai Holocene climate change (e.g. Blyakharchuk et al., 2004, 2007, 2008; Rudaya et al., 2009) was stimulated by renewed archaeological interests in understanding the relationships between climate change and cultural evolution along the ‘Eurasian Steppe Silk Road’. And, the Altai was reported to be the conjunction between the Asian cultures and European cultures along the ‘Eurasian Steppe Silk Road’ (Blyakharchuk and Chernova, 2013; Rudaya et al., 2012; Van Geel et al., 2004).

Figure 1.

Geomorphic settings (a) in the context of the core area of central Asia and vegetation backgrounds (b) in the context of the southern Altai Mountains (NRX: Narenxia Peat).

Several proposals have been put forward to explain the proxy-recorded Holocene climate change in the Altai. First, according to the meteorologically documented domination of the prevailing westerly airflow over the Altai (Aizen et al., 1995, 2001), the westerlies-domination of the entire Holocene was proposed by demonstrating the temporal consistency between the records from the North Atlantic region and the records from the southern Siberia including the Altai (Chen et al., 2008; Ran and Feng, 2013; Wang and Feng, 2013). Second, the Asian monsoon domination was speculated to have extended to southern Siberia including the entire Altai (Harrison et al., 1996; Tarasov et al., 2000) or at least to northern Xinjiang including the southern Altai (Rhodes et al., 1996) during the Holocene warm intervals. Third, an unconventional proposal was put forward by Bush (2005) that was later corroborated by Prokopenko et al. (2007) and Feng et al. (2013). The proposal states that the combined effect of increased atmospheric CO₂ concentration and increased winter insolation under warmed ocean surface and lessened ice cover conditions after ~7000 cal. yr BP may have strongly modulated the climate changes in the southern Siberia including the Altai (note: cal. yr BP = calendar years before present).

The fourth proposal is the most relevant to this study. It states that different parts of the Altai have been controlled by different climate systems during different time intervals. Specifically, Rudaya et al. (2009) made an attempt, primarily based on published data (Blyakharchuk et al., 2004, 2007, 2008; Dorofeyuk and Tarasov, 1998; Grunert et al., 2000; Kremenetski et al., 1997a, 1997b; Tarasov et al., 1997, 2000), to depict the Holocene climate dynamics of the Altai. They concluded that in the eastern Altai within Mongolia, the first half of the Holocene (~11,000 to ~5000 cal. yr BP) was wet and warm and the second half (~5000 to ~0 cal. yr BP) cool and dry, being consistent with the East Asian monsoon climate pattern. In the western Altai within Kazakhstan, the first half of the Holocene was warm and dry and the second half cool and wet, being consistent with the European westerlies climate pattern. In the northern Altai within Russia, the first half of the Holocene was wet and warm (i.e. being consistent with the East Asian monsoon climate pattern) and the second half cool and wet (i.e. being consistent with the European westerlies climate pattern).

In brief, the Altai mountain range is an important climatic boundary, and the importance calls for more extensive and more in-depth investigations into the past climatic dynamics. This paper attempts to reconstruct the Holocene climatic change in the southern Altai with the hope that our understanding of the climatic change in the entire Altai can be furthered.

Study area

According to the meteorologically documented data, the westerly airflow prevails over the Altai throughout a year and the Siberian high-pressure system dominates the Altai during winters (Aizen et al., 1995, 2001). The westerly airflow impacts on the Altai in two ways: precipitation decreasing eastward, and more orographic precipitation on western sides of the mountains. The Siberian high-pressure system impacts on the Altai in two ways: climatic continentality increasing eastward, and frequent winter snow storms resulting from its interaction with the westerly airflow. The vegetation well reflects the N–S gradients of decreasing moisture and increasing temperature, and also well reflects the W–E gradients of decreasing moisture and increasing climatic continentality. Although the Altai is mostly situated at the latitudes of the zonal forest-steppe, the relief promotes the expansion of mountain forests (Blyakharchuk et al., 2004, 2007; Rudaya et al., 2009). The northern, western and southwestern parts of the mountains are covered by dark-coniferous forests with Abies sibirica, Picea obovata and Pinus sibirica. Further towards southeast, Pinus sibirica and Larix sibirica forests become dominant with decreasing moisture and increasing climatic continentality. Beyond the forest limits further towards the southeastern direction along which climatic continentality and climatic dryness continue to increase, a gradual decline in tundra-steppe taxa is accompanied with a gradual increase in tundra-desert taxa within the alpine meadow zone, the latter (i.e. alpine meadow zone) becoming the most important vegetation zone towards the southeastern direction.

Our study site, Narenxia Peat (48°48′N, 86°54′E, 1760 m a.s.l.), is situated within the well-known Kanas Lake Basin. The basin is bioclimatically the southernmost extension of the Siberian taiga forests (Figure 1b), the only taiga-dominated area of China (Chen, 2010). To provide more detailed vegetation background for interpreting the fossil pollen data, we examined the local-scale vegetation zones and the corresponding climates along an S–N transect from the Buerjin in the south to the Friendship Peak in the north (see Figure 1b for the location of the S–N transect). The transect exhibits a dramatic latitudinal differentiation of vegetation within a relatively short distance (Figure 2). Specifically, desert vegetation occupies the low elevations (<700 m a.s.l.) in the south where the mean annual temperature (MAT) is relatively high (>+3°C) and the mean annual precipitation (MAP) is low (<150 mm). Steppe vegetation dominates the medium elevations (700–1400 m a.s.l.) where MAT ranges from +3°C to −1°C and MAP from 150 to 350 mm. Forests and forest-meadows are the predominant vegetation in the middle-high elevations in the north (>1400 m a.s.l.) where MAT is below −1°C and MAP is above 350 mm. The vegetation above the treeline (~2200 m a.s.l.) is of alpine nature (Chen, 2010; Qin, 1957).

Figure 2.

Schematic diagram showing the vertical zones of vegetation distribution around Narenxia Peat (NRX) in the context of the southern Altai Mountains (Chen, 2010; Qin, 1957).

Again, our study site is Narenxia Peat from which a 390-cm long peat–lacustrine core was obtained in 2013. Our dates show that the peat sequence has developed after a lake was desiccated at ~9000 cal. yr BP. The peat-occupying valley is surrounded at three sides by coniferous forests in higher elevations and in shady slopes (i.e. north- and west-facing slopes) of middle-high elevations (see inset photo in Figure 2). Steppes are dominant in lower elevations and sunny slopes of middle-low elevations. Alpine meadow occupies the elevations above tree lines and a narrow tundra zone appears above the alpine meadow zone.

Materials and methods

Stratigraphy and chronology

A 390-cm-long peat–lacustrine core was collected with a Holland-made peat corer in 2013 in the central part of Narenxia Peat. The core was cut longitudinally into two halves and visually described. The core can be stratigraphically divided into four units (Figure 3). Unit 1 (390–350 cm) is an eluvial layer (small and angular gravels mixed with sand and silt). Unit 2 (350–298 cm) is a lacustrine deposit (dark-grey clayey mud with ostracod shells). Unit 3 (298–274 cm) is a transition from lacustrine deposition to peat accumulation (repetitive lake-peat alternations). Unit 4 (274–0 cm) is a poorly decomposed peat layer. Nine peat samples (poorly decomposed) and two lake-mud bulk samples were radiocarbon dated using AMS at the NSF-AMS Facility, University of Arizona (Table 1). Figure 3 shows that all of 11 dates are in perfect order and we attribute this to the ideal dating targets, mostly horizontally grown Sphagnum and Cyperaceae stems. The dates were calibrated to calendar years before present (BP = AD 1950) with the program CALIB 6.0 using the INTCAL 09 calibration dataset (Reimer et al., 2009). The age–depth model was established based on a best-fitting polynomial linear relationship between depths and dates with an assumption that the surface age (0 cm) was AD 2013 (i.e. −63 cal. yr BP). It should be noted that the bottom of the lacustrine unit at depth of 350 cm was dated at 9776 ± 83 ¹⁴C yr BP (11,065–11,394 cal. yr BP) and that there is no chronologic control for the eluvium unit (350–390 cm) for lack of dateable material.

Figure 3.

Stratigraphy and chronology of Narenxia Peat–lacustrine core.

Table 1.

AMS ¹⁴C dating results of Narenxia Peat–lacustrine sequence (AA: laboratory code of NSF-Arizona AMS ¹⁴C Dating Facilities).

LAB code	Depth (cm)	Dated material	δ¹³C (‰)	¹⁴C age (yr BP)	Calibrated age (2δ) (cal. yr BP)
AA102070	39–40	Peat	−25.9	379 ± 39	421–508
AA102071	69–70	Peat	−25.9	1499 ± 40	1306–1422
AA100610	99–100	Peat	−26.7	3214 ± 46	3361–3512
AA102072	129–130	Peat	−26.0	3857 ± 45	4152–4414
AA102073	159–160	Peat	−25.3	4368 ± 46	4842–5050
AA100609	189–190	Peat	−26.7	5530 ± 42	6277–6406
AA102074	219–220	Peat	−26.8	6046 ± 75	6735–7034
AA102075	249–250	Peat	−27.7	6408 ± 51	7259–7425
AA100611	289–290	Peat	−27.0	7448 ± 72	8157–8402
AA102076	319–320	Lake mud	−25.3	8412 ± 61	9294–9530
AA100612	349–350	Lake mud	−25.2	9776 ± 83	11,065–11,394

AMS: accelerator mass spectrometry.

Pollen analysis

In total, 188 pollen samples were obtained at 2-cm intervals from the peat–lacustrine core. For pollen analysis, a tablet of Lycopodium (containing 27,637 ± 563 spores) was first added for calculation of pollen concentration (Moore et al., 1991). The fossil pollen samples were first treated with HCl (5–10%) and HF (36%) and sieved in ultrasonic bath. Pollen types were identified using pollen references (Moore and Webb, 1987; Wang et al., 1995). The treated samples were then mounted and examined with a transmitted light microscope (Olympus BX 51) at 400× magnification. More than 300 pollen grains (not including spores and aquatic pollen) were counted for each one of the samples. The percentage of each pollen type was calculated based on the sum of all counted pollen grains and the pollen diagrams were plotted using Grapher (8.0) software.

Pollen-based biome scores and quantitative climate reconstructions

To semi-quantitatively reconstruct the changes of past vegetation, biomization method was developed based on contemporary knowledge of the ecology and biogeography of modern plants and associated pollen assemblages (Prentice et al., 1992, 1996). The method consists of four steps: (1) assignment of each pollen taxon to one PFT (plant function type) or more PFTs according to known ecology and biogeography, (2) assignment of characteristic PFTs to biomes according to their bioclimatic ranges, (3) construction of a biome-by-taxon matrix, and (4) calculation of the affinity scores for all pollen samples by a simple equation in which the score of a given biome is the sum of the square roots of the percentage (e.g. above 0.5%) of each taxon present in the biome. The method was later modified for suiting the bioclimatic settings in southern Siberia (Tarasov et al., 1998) and in northern China (Yu et al., 2000) (Table 2). To sum up, the premise of biomization is that the relative changes in past vegetation types are represented by the recorded pollen taxa groups corresponding to the modern bioclimatic settings.

Table 2.

Assignment of pollen taxa to biomes (after Tarasov et al., 1998; Yu et al., 2000).

Biome	Code	Taxa included
Desert	DESE	Artemisia, Caryphyllaceae, Chenopodiaceae, Ephedra, Polygonum
Steppe	STEP	Apiaceae, Artemisia, Asteraceae, Brassicaceae, Caryophyllaceae, Chenopodiaceae, Elaeagnaceae, Fabaceae, Galium, Humulus, Labiatae, Liliaceae, Moraceae, Plantago, Poaceae, Polygonaceae, Polygonum, Potentilla, Rosaceae, Rumex, Sanguisorba, Thalictrum, Urtica, Valerianaceae
Tundra	TUND	Alnus, Asteraceae, Betula, Cyperaceae, Ericaceae, Poaceae, Polygonaceae, Polygonum, Salix, Saxifragaceae
Taiga	TAIG	Abies, Alnus, Betula, Ericaceae, Larix, Picea, Pinus Diploxylon (sylvestris-type), Pinus Haloxylon (sibirica-type), Populus, Salix
Cold deciduous forest	CLDE	Alnus, Betula, Ericaceae, Larix, Pinus Diploxylon (sylvestris-type), Pinus Haloxylon (sibirica-type), Populus, Salix
Cold mixed forest	CLMX	Abies, Alnus, Betula, Carpinus, Corylus, Ericaceae, Pinus Diploxylon (sylvestris-type), Pinus Haloxylon (sibirica-type), Populus, Quercus, Salix, Tilia, Ulmus
Cool conifer forest	COCO	Abies, Alnus, Betula, Ericaceae, Picea, Pinus Diploxylon (sylvestris-type), Pinus Haloxylon (sibirica-type), Salix
Cool mixed forest	COMX	Abies, Acer, Alnus, Betula, Carpinus, Corylus, Ericaceae, Fraxinus, Picea, Pinus Diploxylon (sylvestris-type), Pinus Haloxylon (sibirica-type), Quercus, Rubia, Salix, Tilia, Ulmus

To quantitatively reconstruct the changes of past climate, we tried three different methods: a transfer function developed along an N-S transect in central-eastern Asia around ~110°E (Wen et al., 2013), a response surface developed in northern China (Sun et al., 1996), and the PFT-MAT (PFT = plant function type; MAT = modern analogue technique) using modern pollen dataset covering a vast area with well-differentiated biogeographic settings (Guiot, 1990; Guiot et al., 1993). All the three methods produced more or less acceptable reconstructions of modern (i.e. surface samples) MAP (~300 mm), but the former two (i.e. transfer function and response surface) generated absolutely unacceptable reconstructions of modern (i.e. surface samples) MAT, for example, the reconstructed MAT is 4–6°C higher than the measured MAT. In contrast, the third method (i.e. PFT-MAT) generated an acceptable reconstruction of modern (surface samples) MAT (below −2°C). We speculate that the reason for the PFT-MAT method being acceptable is because the study area (i.e. southern Altai) is within the PFT-MAT method–used surface pollen data–covered area and that the reason for other two methods being unacceptable is because they were developed in different biogeographic settings than the Altai.

PFT-MAT was developed to calculate a chord distance between the pollen assemblage considered and the modern analogues compared for determining the similarity between the fossil and modern pollen spectra (Davis et al., 2003; Guiot, 1990; Nakagawa et al., 2002). The modern pollen dataset used in this study includes 1855 surface pollen spectra from Europe, Asian parts of Former Soviet Union territory and Mongolia (Guiot, 1990; Guiot et al., 1993; Magny et al., 2001) and 211 surface pollen spectra from northern China (Jiang et al., 2009). To determine the similarity between the fossil and modern pollen spectra, a threshold has to be set to select the ‘best modern analogues’. Specifically, if the chord distance between the fossil and modern pollen spectra is below the predefined threshold, the modern samples are considered as the ‘best modern analogues’ and taken into account in the quantitative reconstructions. The commonly acknowledged three climatic parameters controlling plant growth were calculated using the PPPBASE software package (Guiot and Goeury, 1996) and they are warm-season temperature (Tw), MAT and MAP. It should be particularly noted that cold-season temperature (Tc) was also calculated because it was well-documented to be an important factor determining the presence of certain plant species in Siberia where continental climates dominate (Blyakharchuk and Chernova, 2013).

Two additional points need to be addressed to enhance the validity of the PFT-MAT method. First, to define the threshold, we tried five options of selecting ‘best modern analogues’: top-2 smallest chord distances (SCD), top-4 SCD, top-6 SCD, top-8 SCD and top-10 SCD. Here, we rejected the top-2 and top-4 options because the included ‘best modern analogues’ are too few. We also rejected the top-8 and top-10 options because the error bars of the estimated climatic parameters are significantly larger than those of top-6 option. Therefore, top-6 option was adopted in this study. Estimates of palaeoclimatic parameters are obtained by using a weighted average of the values for all selected six ‘best modern analogues’, where the weights used are the inverses of the chord distances. Second, to further validate the acceptability of the PFT-MAT method, we compared the PFT-MAT inferences with the measured climatic data. Specifically, we randomly selected 100 pollen data points from the dataset (i.e. totally 2066 data points) falling within a predefined area (i.e. 35–60°N and 70–100°E) that includes Narenxia Peat (48°48′N and 86°54′E) and statistically compared the modern pollen-based inferences of the climatic parameters with the corresponding climatic parameters that were either instrument-measured or spatially interpolated. The correlation coefficient (R) and the root mean square error (RMSE) listed in Table 3 indicate that the PFT-MAT is an acceptable method for quantitatively reconstructing the Holocene climate changes in the southern Altai that is within the PFT-MAT method–used surface pollen data–covered area.

Table 3.

Statistical comparison between the modern climate parameters (measured and spatially interpolated) and the climate reconstructions using PFT-MAT. Pollen data used: 100 pollen data points were randomly selected from the area (35–60°N, 70–100°E) that includes Narenxia Peat (48°48′N, 86°54′E).

Climatic parameters	Correlation coefficients (R)	R ²	Root mean square error (RMSE)	Significance level (ρ)
Tc	0.49	0.24	5.83	<0.001
Tw	0.75	0.56	2.77	<0.001
MAT	0.65	0.43	3.63	<0.001
MAP	0.62	0.39	120.25	<0.001

PFT-MAT: plant function type–modern analogue technique; MAT: mean annual temperature; MAP: mean annual precipitation.

Results

Pollen data

The core can be palynologically divided into four pollen assemblage zones using the constrained incremental sum of squares (CONISS) method and more sub-units within the pollen zones C and D were also identified (Figure 4):

Pollen Zone A (390–352 cm; before ~12,240 cal. yr BP) is stratigraphically correspondent with the eluvial layer (390–350 cm). The pollen concentration is extremely low and the pollen assemblage is absolutely of herbaceous origin.

Pollen Zone B (352–304 cm; ~12,240 to ~9530 cal. yr BP) is approximately correspondent with the lacustrine layer (350–298 cm) and still dominated by herbaceous component (mainly including Cyperaceae, Chenopodiaceae, Artemisia and Poaceae). Although the pollen concentration remains low, tree component starts its appearance (mainly including Pinus sibiric, Picea, Platydadus and Betula). Four pollen types are pronouncedly expressed in this zone and they are Taraxacum, Aster, Saussurea and Ephedra.

Pollen Zone C (304–104 cm; ~9530 to ~3200 cal. yr BP) includes the entire lacustrine–peat transition (298–274 cm) and the lower portion of the peat layer (274–0 cm). It is generally characterized by drastic increases both in pollen concentration and in pollen diversity. Overall, tree and shrub component is greatly increased at the expense of herbaceous component compared with previous two zones (i.e. Zone A and Zone B). This zone can be further divided into three sub-zones: C1, C2 and C3. The middle sub-zone (i.e. C2) is distinguishable from the underlying sub-zone (i.e. C1) and also from the overlying sub-zone (i.e. C3) by following four features: (1) a higher pollen concentration, (2) a lower percentage of Artemisia, (3) a lower percentage of Chenopodiaceae and (4) a higher percentage of Cyperaceae.

Pollen Zone D (104–0 cm; ~3200 to ~0 cal. yr BP) occurs in the upper portion of the peat layer (274–0 cm) and is typified by a drastic increase in tree pollen percentage. The Poaceae and Artemisia percentages also increase at the expenses of Chenopodiaceae and Cyperaceae percentages. This zone can be further divided into two sub-zones (D1 and D2) and the upper sub-zone has higher percentages of Pinus sibiric and Chenopodiaceae and lower percentages of Poaceae and Cyperaceae than the low sub-zone.

Figure 4.

Pollen data of Narenxia Peat–lacustrine core and CONISS-based divisions of pollen assemblages.

Biome scores and vegetation changes

Our following discussion is primarily based on the biome scores of pollen zones (Figure 5). It should be particularly noted that there is no category of alpine meadow in the biomization scheme (see Table 2). Instead, the following herbaceous plant types within alpine meadow zone were classified as tundra: Alnus, Asteraceae, Betula, Cyperaceae, Ericaceae, Poaceae, Polygonaceae, Polygonum, Salix and Saxifragaceae. It means that an increase in tundra biome score of a pollen spectrum actually suggests an expansion of alpine meadow if pollen concentration is not extremely low. Another point that deserves mentioning is Cyperaceae-related. Cyperaceae is a very important component of alpine meadow and is thus taken as alpine meadow component in biomization and also in PFT-MAT climatic inference. But, Cyperaceae is also very important component of wetland vegetation, thus undermining the validity of biomization and also of PFT-MAT climatic inference. However, the geomorphic constraint of Narenxia Peat is here interpreted to enhance the validity of Cyperaceae as an alpine meadow component. That is, Narenxia Peat is actually a terminal moraine-dammed wetland and the limited areal extent of the wetland has been constrained by the height of the terminal moraine since its formation at ~9000 cal. yr BP. In other words, the areal extent of the wetland has not changed significantly since its formation and thus the contribution of the wetland to the Cyperaceae pollen can be approximated as a low constant:

Vegetation Stage 1 (i.e. Pollen Zone A; before ~12,240 cal. yr BP) is correspondent with the eluvial layer (390–350 cm). The extremely low pollen concentration and the absolutely herbaceous origin of pollen suggest that the vegetation was most likely a real tundra or a cold desert-steppe.

Vegetation Stage 2 (i.e. Pollen Zone B; ~12,240 to ~9530 cal. yr BP) is approximately correspondent with the lacustrine layer (350–298 cm). Figure 5 shows that the biome scores of tundra and steppe reach the highest of the entire core. A relatively high biome score of taiga indicates the appearance of taiga forests, and a low pollen concentration suggests a low coverage of vegetation. Our interpretation is that the vegetation was most likely dominated by real tundra and alpine meadows with the appearance of taiga forests probably in lower elevations and in sunny slopes.

Vegetation Stage 3 (i.e. Pollen Zone C; ~9530 to ~3200 cal. yr BP) includes the entire lacustrine–peat transition (298–274 cm) and the lower portion (274–104) of the peat layer (274–0 cm). Overall, tree and shrub component is greatly increased at the expense of herbaceous component with a dramatic increase in pollen concentration. Corresponding to the three pollen sub-zones (i.e. D1, D2 and D3) are three vegetation sub-stages: 3–1, 3–2 and 3–3.

Figure 5.

Calculated biome scores and inferred vegetation variations. The used biomization scheme (Prentice et al., 1992, 1996) was a modified one (Table 2) for suiting the bioclimatic settings in southern Siberia (Tarasov et al., 1998) and in northern China (Yu et al., 2000).

The sub-stage 3–1 (~9530 to ~8500 cal. yr BP) is characterized by an increasing trend of pollen concentration, very high biome score of taiga, and moderate biome scores of tundra and steppe. The sub-stage 3–2 (~8500 to ~7000 cal. yr BP) has the highest pollen concentration of the entire core. A relatively high biome score of taiga and a very high biome score of tundra are accompanied with considerably lowered biome scores of steppe and desert. The sub-stage 3–3 (~7000 to ~3200 cal. yr BP) has dramatically lowered biome scores of taiga and tundra and significantly increased biome scores of steppe and desert.

Following is our interpretation of the vegetation history during the vegetation stage 3. Taiga forest started its appearance in lower elevations and in sunny slopes before the stage 3 (i.e. before ~9530 cal. yr BP) and reached its climax between ~9530 and ~7000 cal. yr BP (i.e. sub-stages 3–1 and 3–2). A dramatic increase in tundra biome score and a climax level of taiga biome score between ~8500 and ~7000 cal. yr BP (i.e. sub-stage 3–2), accompanied with decreases in the biome scores of steppe and desert, might be associated with significant expansions of alpine meadows and taiga forests resulting from maximal thawing of permafrost under an optimal climate (warm and wet). A considerably lowered biome score of taiga and significantly increased biome scores of steppe and desert between ~7000 and ~3200 cal. yr BP (i.e. sub-stage 3–3) suggest a sizable shrinkage of taiga forests and accompanied expansions of steppes and deserts.

Vegetation Stage 4 (i.e. Pollen Zone D; ~3200 to ~0 cal. yr BP) occurs in the upper portion (104–0 cm) of the peat layer (274–0 cm) and is typified by a dramatic increase in tree pollen percentage (mainly including Pinus sibiric, Picea, Abies and Oleaceae). Corresponding with two pollen sub-zones (i.e. D1 and D2) are two vegetation sub-stages: 4–1 and 4–2. In comparison with the underlying sub-stage 3–3 (~7000 to ~3200 cal. yr BP), sub-stage 4–1 (~3200 to ~800 cal. yr BP) is characterized by a dramatic increase in taiga biome score and also by a significant increase in steppe biome score. Sub-stage 4–2 (past ~800 years) is expressed by the highest taiga biome score of the entire core and also by slightly lowered biome score of tundra. In terms of vegetation variations, the sub-stage 4–1 (~3200 to ~800 cal. yr BP) experienced a dramatic expansion of taiga forests and also a significant expansion of steppes. The sub-stage 4–2 (past ~800 years) experienced a further expansion of taiga forests and a slight shrinkage of tundra.

Discussions

Climatic changes associated with vegetation changes

Two notes deserve mentioning. First, the reconstructed ‘modern’ temperatures (e.g. Tc = about −24°C; Tw = about 16°C; MAT = about −3°C) and the reconstructed ‘modern’ precipitation (MAP = about 330 mm) for the topmost portion of Narenxia Peat–lacustrine core are within the acceptable ranges with references to the instrument-recorded climatic parameters, considering the complexity of pollen production in the topographically complicated Kanas Lake Basin where our study site is situated. Second, although the error bars of our reconstructions appear undesirably large (Figure 6) because of the complexity of pollen production and also because of the inherited uncertainties in the PFT-MAT method, the reconstructions should be more or less trustworthy because the PFT-MAT used in this study was extensively tested in modern bioclimatic settings (Guiot et al., 1993; Prentice et al., 1992, 1996) and also because some of these modern bioclimatic settings (e.g. those in southern Siberia) are quite similar to those in the southern Altai. The climate changes retrieved from Narenxia core can be visually divided into five stages (Figure 6):

Climatic Stage 1 (prior to ~11,500 cal. yr BP). A barren terrain–dominated tundra landscape is suggested by following four features: (1) an extremely low pollen concentration (~100 grains/g), (2) complete absence of tree pollen, (3) a relatively high biome score of tundra and (4) very high biome scores of steppe and desert. It should be noted that high biome score of steppe was probably a by-product of high biome scores of desert and tundra because the three (steppe, desert and tundra) share many herb species in the biomization scheme (see Table 2). This paper purposely ignores the quantitative reconstructions of this stage for two reasons: (1) an extremely low pollen concentration may severely undermine the acceptability of the quantitative reconstructions, and (2) the focus of this paper is on the Holocene (i.e. past ~11,500 cal. yr BP).

Climatic Stage 2 (~11,500 to ~7000 cal. yr BP). A dramatic expansion of taiga forests is suggested by high values in taiga biome score, AP/NAP ratio and pollen concentration. The reconstructed temperature curves (i.e. Tc, Tw and MAT) and precipitation curve (i.e. MAP) run more or less in parallel with each other, and they are averagely higher (i.e. warmer and wetter) than those in the succeeding stage 3 (i.e. ~7000 to ~4000 cal. yr BP). On average, Tc was above −25°C, Tw was above +14°C, MAT was above −4°C, and MAP was above 330 mm.

Climatic Stage 3 (~7000 to ~4000 cal. yr BP). A noticeable feature is the significantly lowered values in taiga biome score, AP/NAP ratio and pollen concentration, suggesting that the taiga forests shrunk considerably in comparison with the preceding stage 2. On average, the reconstructed temperatures (i.e. Tc, Tw and MAT) and precipitation (i.e. MAP) are considerably lower (i.e. colder and drier) not only than those in the preceding stage 2 (i.e. ~11,500 to ~7000 cal. yr BP) but also than those in the succeeding stage 4 (i.e. ~4000 to ~1200 cal. yr BP). On average, Tc was below −25°C, Tw was below +14°C, MAT was below −4°C, and MAP was below 330 mm.

Climatic Stage 4 (~4000 to ~1200 cal. yr BP). The reconstructed temperatures (i.e. Tc, Tw and MAT) and precipitation (i.e. MAP) suggest that the climate during stage 4 (~4000 to ~1200 cal. yr BP) was considerably warmer and wetter than the preceding stage 3 and also than the succeeding stage 5 (i.e. ~1200 to ~0 cal. yr BP). On average, Tc was above −25°C, Tw was above +14°C, MAT was above −4°C, and MAP was above 330 mm.

Climatic Stage 5 (~1200 to ~0 cal. yr BP). A cool-dry climate seems to have prevailed during the past ~1200 years. The coolness was primarily indicated by an increase in Pinus sibiric percentage and the dryness by an increase in Chenopodiaceae percentage (see Figure 4). The increase in Pinus sibiric percentage is also the reason for having high values in taiga biome score, AP/NAP ratio and pollen concentration. On average, Tc was below −25°C, Tw was below +14°C, MAT was below −4°C, and MAP was below 330 mm.

Figure 6.

Pollen-based estimates of temperatures (Tc, Tw and MAT) and precipitation (MAP). Note 1: AP/NAP is the ratio of arboreal over non-arboreal percentages; note 2: green bars in Tc, Tw, MAT and MAP are the variation ranges inferred from the selected six ‘best modern analogues’; note 3: Habahe MAT and Habahe MAP are the GDGTs-based reconstructions of MAT and MAP at nearby Habahe Peat.

To our surprise, our reconstructions are reasonably well corroborated by the reconstruction at a nearby site. Specifically, Tang (2014) in his master’s thesis produced GDGT (Glycerol Dialkyl Glycerol Tetraethers)-based reconstructions of MAT and MAP at nearby Habahe Peat (48°49′N, 86°57′E, 1763 m a.s.l.) that well resemble our pollen-based reconstructions of MAT and MAP at Narenxia Peat (48°48′N, 86°54′E, 1760 m a.s.l.). It should be particularly noted that the quantitative reconstructions at Habahe Peat (Tang, 2014) were based on the empirical relationships between climatic parameters (i.e. MAT and MAP) and GDGTs parameters (i.e. MBT and CBT) developed by Weijers et al. (2007). It is quite notable from Figure 6 that the GDGTs-based MAT appears too high, for example, the GDGTs-based MAT for the surface samples is 5–6°C higher than the actual modern MAT. The reason for the GDGTs-based MAT being too high is most likely because the adopted relationship between MAT and GDGTs parameters (i.e. MAT = 50 × MBT − 9.35 × CBT − 6.1) was not locally specified. Assuming that the trends and shapes of the GDGTs-based MAT and MAP curves from Habahe Peat remain more or less similar to the pollen-based MAT and MAP curves from Narenxia Peat after the adopted relationship between MAT and GDGTs parameters is locally specified, the resemblance between these two sets of curves further boosts our confidence in accepting our pollen-based reconstructions.

A proposal for underlying mechanisms

Significance of NAO and ENSO

The Holocene climates in the Altai should have been dominated either by the prevailing westerly climates (Chen et al., 2008; Ran and Feng, 2013; Wang and Feng, 2013) or by the competing interactions among different climate systems (e.g. Atlantic-related westerlies, high latitude–associated winter monsoon, and low latitude–associated summer monsoon) (Blyakharchuk et al., 2004; Rudaya et al., 2009; Tarasov et al., 2000). The strength of westerlies and the interactions among different climate systems should have been primarily driven by four major forcing factors: (1) winter insolation, (2) summer insolation, (3) atmospheric CO₂ concentration and (4) the extent of the Northern Hemisphere ice cover (Ruddiman, 2008). However, our Holocene climatic reconstructions from Narenxia Peat can be simply explained neither by the Holocene variations in the four major forcing factors nor by the reported Holocene variations in the upstream westerly climates in North Atlantic (Bond et al., 2001) and in western Europe (Davis et al., 2003). Our attention was thus turned to the climatically influential NAO (North Atlantic Oscillations) and ENSO (El Niño–Southern Oscillations). Many studies have shown that NAO controls the winter climatic variability of the entire Northern Hemisphere (Hurrell, 1996; Polonsky et al., 2004; Xu et al., 2016) and that ENSO is of global significance in modulating precipitation patterns (Bronnimann et al., 2007; Ropelewski and Halpert, 1996; Wu and Lin, 2012). The significance of NAO and ENSO in shaping the climates of central Asia was demonstrated by Syed et al. (2006). They showed that a low-pressure trough (between sea-level pressure and 500 hPa) over the Central-Southwest Asia (CSWA) region (including northern Pakistan, southern Uzbekistan, Afghanistan, Tajikistan and southern Kazakhstan) was formed during positive NAO phases or/and during warm ENSO phases (i.e. El Niño phases). The NAO-related trough was proposed to be associated with deepening of the Icelandic Low (Figure 7a), while the ENSO-related trough was proposed to be associated with weakening of the Siberian High (Figure 7b). Both observed data (Syed et al., 2006) and modelling results (Syed et al., 2010) showed that the NAO- or/and ENSO-resulted low-pressure trough over CSWA region greatly enhanced non-summer (i.e. winter, autumn, spring) precipitation over the CSWA region through intensifying the western disturbances (i.e. extratropical storms) that originated in the Mediterranean Sea and picked up additional moisture from the Arabian and Caspian Seas.

Figure 7.

Proposed teleconnections of (a) NAO and (b) ENSO to the climate over the Central-Southwest Asia (Syed et al., 2006, 2010) and beyond.

Encouraged by the works of Syed et al. (2006, 2010), we are here proposing that the Holocene variations in NAO and ENSO might have dictated the millennial-scale climatic changes in the southern Siberia including the Altai. Figure 8a is the reconstruction of NAO-index variations of the past ~5200 years (Olsen et al., 2012; Trouet et al., 2009) and Figure 8b is the reconstruction of ENSO-strength variations of the past ~12,000 years (Rein et al., 2005). The approximate synchronism on centennial scales between NAO index and ENSO strength during the overlapping period (i.e. past ~5200 years) seems to lend further support to the proposed causal linkage between NAO and ENSO for the past ~100 years (Polonsky et al., 2004; Raible et al., 2004; Syed et al., 2006; Wu and Lin, 2012). Assuming that the approximate synchronism between NAO index and ENSO strength during the overlapping period (i.e. past ~5200 years) also holds for the early-Holocene (from ~5000 to ~11,500 cal. yr BP), it is justifiable to use the reconstructed ENSO strength as a proxy of the combined effect of ENSO and NAO for the entire Holocene.

Figure 8.

Evidence corroborative to the proposed teleconnections of NAO and ENSO to the climate over Central-Southwest Asia and beyond. Corroborative data are from (1) Baikal Lake in southern Siberia (Tarasov et al., 2007), (2) Accesa Lake of Italy data (Peyron et al., 2011) and (3) Guliya Ice Core of western Tibetan Plateau (Thompson et al., 1997).

Comparing our reconstructions of precipitation (Figure 8c) and temperatures (Figure 8d–f) from Narenxia Peat with the Holocene ENSO-strength reconstruction (i.e. Figure 8b) indicates that the climate at Narenxia Peat has changed more or less synchronously with ENSO on millennial scales, implying a causal linkage between ENSO (and also NAO) and the climate in the Altai. The approximate synchronism between ENSO (and also NAO) and MAP of Narenxia Peat can be explained by provoking ENSO- or/and NAO-intensified western disturbances (i.e. extratropical storms) that brought more water vapour to CSWA region and probably beyond reaching the Altai. In addition, ENSO- or/and NAO-resulted low-pressure trough over CSWA region might have further enhanced the precipitation through stimulating regional air convection.

What could then be the reason(s) behind the approximate synchronism between ENSO (and also NAO) and temperatures of Narenxia Peat? The observed data show that positively phased NAO yielded enhanced advection of warm air over extratropical Eurasia north of about 45°N and that the positively phased NAO was thus an important cause of the observed winter half-year warming in southern Siberia (Hurrell and Van Loon, 1997; Xu et al., 2016). Furthermore, warm-phased ENSO was suggested to be teleconnected with weakening of the Siberian High and the weakening was also a cause of the observed winter half-year warming in southern Siberia (Polonsky et al., 2004; Syed et al., 2006).

Now, the remaining question is, how to explain the approximate synchronism between ENSO (and also NAO) and warm-season temperature? The multi-decadal ENSO-NAO teleconnection pattern and the associated coherence between the extratropical atmospheric circulation anomalies and the sea-surface temperature anomalies over the North Pacific during summers were observed to be similar to those during winters (Wu and Lin, 2012). It implies that the Holocene warm-season temperature in southern Siberia including the Altai might have also been modulated by ENSO and NAO, although the detailed processes need in-depth investigations.

Corroborative evidence

As mentioned above, Rudaya et al. (2009) concluded that in the eastern Altai the first half of the Holocene (~11,000 to ~5000 cal. yr BP) was wet and warm and the second half (~5000 to ~0 cal. yr BP) cool and dry. In the western Altai, the first half of the Holocene was warm and dry and the second half cool and wet. In the northern Altai the first half of the Holocene was wet and warm and the second half cool and wet. However, our quantitative reconstructions from Narenxia Peat in the southern Altai are dramatically different from the qualitative depiction by Rudaya et al. (2009). The most pronounced difference is a well-expressed temperature maximum of the late-Holocene (~4000 to ~1200 cal. yr BP) in Narenxia Peat within the southern Altai that absolutely lacks counterparts from other parts of the Altai.

Although there are no published data to our knowledge showing millennial-scale correlation between the Holocene ENSO (and also NAO) variations and the Holocene climate changes in southern Siberia including the Altai, three well-accepted quantitative temperature reconstructions from three key locations are somewhat boosting our confidence for advocating the significance of ENSO and NAO in modulating the Holocene climate changes in the Altai. The first reconstruction is from Baikal Lake of Russia situated in the core of Siberian High (Tarasov et al., 2007). The reconstructed cold-season temperature (Tc) and warm-season temperature (Tw) exhibit four stages of variation (Figure 8g and h): (1) a low temperature from ~11,500 to ~9400 cal. yr BP, (2) a high temperature from ~9400 to ~6800 cal. yr BP, (3) a low temperature from ~6800 to ~4000 cal. yr BP and (4) a high temperature from ~4000 to ~200 cal. yr BP. The second reconstruction is from Guiliya Ice Core situated in the western Tibetan Plateau adjacent to CSWA region (Thompson et al., 1997). The δ¹⁸O-suggested temperature displays three stages of variation (Figure 8i): (1) a high temperature from ~11,500 to ~7200 cal. yr BP, (2) a low temperature from ~7200 to ~2500 cal. yr BP and (3) a high temperature from ~2500 to ~0 cal. yr BP. It should be noted that a minor peak of temperature at ~4000 cal. yr BP divides the low temperature interval from ~7200 to ~2500 cal. yr BP into two sub-intervals with the earlier sub-interval (~7200 to ~4000 cal. yr BP) being considerably colder than the later one (~4000 to ~2500 cal. yr BP). The third reconstruction is from Accessa Lake of Italy situated on the pathway of the western disturbances (Peyron et al., 2011). The approximate synchronism between the reconstructed temperatures (Tc and Tw) of Accessa of Italy (Figure 8j and k) and the reconstructed temperatures (Tc and Tw) of Narenxia of the southern Altai may suggest a causal linkage between the ENSO- or/and NAO-intensified western disturbances and the Holocene climate changes in southern Siberia including the Altai.

Conclusion

Climatic reconstructions

Narenxia Peat experienced five stages of bioclimatic evolution. Stage 1 (prior to ~11,500 cal. yr BP) was dominated by a barren terrain under a cold and dry climate and stage 2 (~11,500 to ~7000 cal. yr BP) was a period of dramatic expansions of taiga forests under a generally warm and wet climate. Stage 3 (~7000 to ~4000 cal. yr BP) was an interval of taiga forest shrinkage under a considerably cooled and dried climate and stage 4 (~4000 to ~1200 cal. yr BP) witnessed another expansion of taiga forests under a resumed warm and wet climate. The past ~1200 years (i.e. stage 5) have been relatively cool and dry.

Proposed mechanisms

The temperature and precipitation reconstructions from Narenxia Peat resemble the ENSO-strength reconstruction of past ~10,000 years, and the ENSO-strength curve is approximately parallel with the NAO-index curve for the overlapping period (i.e. past ~5200 years). The resemblance implies that the Holocene climate in the Altai might have been causally associated with ENSO and NAO. In terms of precipitation, ENSO- or/and NAO-intensified western disturbances might have brought more water vapour to CSWA region and probably beyond reaching the Altai. As for temperature, NAO-enhanced advection of warm air over extratropical Eurasia north of about 45°N and ENSO-resulted weakening of the Siberian High might have resulted in warming in southern Siberia. Here, we propose that the Holocene variations in NAO and ENSO might have dictated the millennial-scale climatic changes in the southern Siberia including the Altai.

Footnotes

Funding

This research was financially supported by CAS International Cooperation Program (grant number: GJHZ201315) and also by Director’s Fund of Xinjiang Institute of Ecology and Geography. This research was also indirectly benefited from two NSFC projects (No. U1203821L08, No. 40930102) and one NSF project (BCS-06-23478).

References

Aizen

Melack

(1995) Climate, snow, glaciers and runoff in the Tien Shan, Central Asia. Water Resources Bulletin 30: 1113–1129.

Aizen

Melack

(2001) Precipitation and atmospheric circulation patterns at mid-latitudes of Asia. International Journal of Climatology 21: 535–556.

Blyakharchuk

Chernova

(2013) Vegetation and climate in the Western Sayan Mts according to pollen data from Lugovoe Mire as a background for prehistoric cultural change in southern Middle Siberia. Quaternary Science Reviews 75: 22–42.

Blyakharchuk

Wright

Borodavko

(2004) Late Glacial and Holocene vegetational changes on the Ulagan high-mountain plateau, Altai Mountains, southern Siberia. Palaeogeography, Palaeoclimatology, Palaeoecology 209: 259–279.

Blyakharchuk

Wright

Borodavko

(2007) Late Glacial and Holocene vegetational history of the Altai Mountains (southwestern Tuva Republic, Siberia). Palaeogeography, Palaeoclimatology, Palaeoecology 245: 518–534.

Blyakharchuk

Wright

Borodavko

(2008) The role of pingos in the development of the Dzhangyskol lake-pingo complex, central Altai Mountains, southern Siberia. Palaeogeography, Palaeoclimatology, Palaeoecology 257: 404–420.

Bond

Kromer

Beer

(2001) Persistent solar influence on north Atlantic climate during the Holocene. Science 294: 2130–2136.

Bronnimann

Xoplaki

Casty

(2007) ENSO influence on Europe during the last centuries. Climate Dynamics 28: 181–197.

Bush

ABG

(2005) CO₂/H₂O and orbitally driven climate variability over central Asia through the Holocene. Quaternary International 136: 15–23.

10.

Chen

Yang

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

11.

Chen

(2010) Physical Geography of China’s Arid Zones. Beijing: Science Press (a book in Chinese).

12.

Davis

BAS

Brewer

Stervenson

(2003) The temperature of Europe during the Holocene reconstructed from pollen data. Quaternary Science Reviews 22: 1701–1716.

13.

Dorofeyuk

Tarasov

(1998) Vegetation and lake levels in Northern Mongolia in the last 12,500 years as indicated by data of pollen and diatom analyses. Stratigraphy and Geological Correlation 6: 70–83.

14.

Feng

Z-D

Zhang

(2013) Holocene climate variations retrieved from Gun Nuur core in the Northern Mongolian Plateau. The Holocene 23: 1719–1728.

15.

Grunert

Lehmkuhl

Walther

(2000) Paleoclimatic evolution of the Uvs Nuur basin and adjacent areas (Western Mongolia). Quaternary International 65–66: 171–192.

16.

Guiot

(1990) Methodology of the last climatic cycle reconstruction in France from pollen data. Palaeogeography, Palaeoclimatology, Palaeoecology 80: 49–69.

17.

Guiot

Goeury

(1996) PPPBASE, a software for statistical analysis of paleoecological and paleoclimatological data. Dendrochronologia 14: 295–300.

18.

Guiot

Harrison

Prentice

(1993) Reconstruction of Holocene patterns of moisture in Europe using pollen and lake-level data. Quaternary Research 40: 39–49.

19.

Harrison

Tarasov

(1996) Late Quaternary lake-level record from Northern Eurasia. Quaternary Research 45: 138–159.

20.

Hurrell

(1996) Influence of variations in extratropical wintertime teleconnections on Northern Hemisphere temperature. Geophysics Research Letters 23: 665–668.

21.

Hurrell

Van Loon

(1997) Decadal variations in climate associated with the North Atlantic oscillation. Climatic Change 36: 301–326.

22.

Jiang

Guiot

Chu

(2009) An improved methodology of the modern analogues technique for palaeoclimate reconstruction in arid and semi-arid regions. Boreas 39: 145–153.

23.

Kremenetski

Tarasov

Cherkinsky

(1997a) Postglacial development of Kazakhstan pine forests. Geographie Physique et Quaternaire 51: 391–404.

24.

Kremenetski

Tarasov

Cherkinsky

(1997b) The latest Pleistocene in southwestern Siberia and Kazakhstan. Quaternary International 41: 125–134.

25.

Magny

Guiot

Schoellammer

(2001) Quantitative reconstruction of Younger Dryas to Mid-Holocene paleoclimates at Le Locle, Swiss Jura, using pollen and lake-level data. Quaternary Research 56: 170–180.

26.

Moore

Webb

(1987) Palynology Analytical Manual (trans. Li

Xiao

Liu

). Nanning: Guangxi People’s Publishing House (a book in Chinese).

27.

Moore

Webb

Collinson

(1991) Pollen Analysis. 2nd Edition. Oxford: Blackwell Scientific Publication.

28.

Nakagawa

Tarasov

Nishida

(2002) Quaternary pollen-based climate reconstruction in central Japan: Application to surface and Late Quaternary spectra. Quaternary Science Reviews 21: 2099–2113.

29.

Olsen

Anderson

Knudsen

(2012) Variability of the North Atlantic Oscillation over the past 5200 years. Nature Geoscience 5: 808–812.

30.

Peyron

Goring

Dormoy

(2011) Holocene seasonality changes in the central Mediterranean region reconstructed from the pollen sequences of Lake Accesa (Italy) and Tenaghi Philippon (Greece). The Holocene 21: 131–146.

31.

Polonsky

Basharin

Voskresenskaya

(2004) Relationship between the North Atlantic Oscillation, Eurasian climate anomalies and Pacific Variability. Pacific Oceanography 2: 52–66.

32.

Prentice

Cramer

Harrison

(1992) A global biome model based on plant physiology and dominance, soil properties and climate. Journal of Biogeography 19: 117–134.

33.

Prentice

Guiot

Huntley

(1996) Reconstructing biomes from palaeoecological data: A general method and its application to European pollen data at 0 and 6 ka. Climate Dynamics 12: 185–194.

34.

Prokopenko

Khursevich

Bezrukova

(2007) Paleoenvironmental proxy records from lake Hovsgol, Mongolia, and a synthesis of Holocene climate change in the Lake Baikal Watershed. Quaternary Research 68: 2–17.

35.

Qin

(1957) The Flora, vegetation types and plant resources in Xinjiang Altay. Chinese Science Bulletin 3: 114–115 (in Chinese).

36.

Raible

Luksch

Fraedrich

(2004) Precipitation and Northern Hemisphere regimes. Atmospheric Science Letters 5: 43–55.

37.

Ran

Feng

Z-D

(2013) Holocene moisture variations across China and driving mechanisms: A synthesis of climatic records. Quaternary International 313–314: 179–193.

38.

Reimer

Baillie

MGL

Bard

(2009) IntCal09 and Marine09 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 51: 1111–1150.

39.

Rein

Lückge

Reinhardt

(2005) El Niño variability off Peru during the last 20,000 years. Paleoceanography 20: PA4003.

40.

Rhodes

Gasse

Lin

(1996) A late Pleistocene-Holocene lacustrine record from Lake Manas, Zunggar (northern Xinjiang, western China). Palaeogeography, Palaeoclimatology, Palaeoecology 120: 105–121.

41.

Ropelewski

Halpert

(1996) Quantifying southern oscillation precipitation relationships. Journal of Climate 9: 1043–1059.

42.

Rudaya

Nazarova

Nourgaliev

(2012) Mid-late Holocene environmental history of Kulunda, southern West Siberia: Vegetation, climate and humans. Quaternary Science Reviews 48: 32–42.

43.

Rudaya

Tarasov

Dorofeyuk

(2009) Holocene environments and climate in the Mongolian Altai reconstructed from the Hoton-Nur pollen and diatom records: A step towards better understanding climate dynamics in Central Asia. Quaternary Science Reviews 28: 540–554.

44.

Ruddiman

(2008) Earth’s Climate: Past and Future. New York: Freeman and Company.

45.

Sun

Wang

Song

(1996) Pollen-climate response surface analysis of some pollen taxa in northern China. Sciences in China D 26(5): 431–436.

46.

Syed

Giorgi

Pal

(2006) Effect of remote forcings on the winter precipitation of central southwest Asia part 1: Observations. Theoretical and Applied Climatology 86: 147–160.

47.

Syed

Giorgi

Pal

(2010) Regional climate model simulation of winter climate over Central-Southwest Asia with emphasis on NAO and ENSO effects. International Journal of Climatology 30: 220–235.

48.

Tang

(2014) Holocene environmental changes inferred from GDGTs from peatlands in the Altai Mountains, Xinjiang of China. Master’s Thesis, Lanzhou University of China (in Chinese).

49.

Tarasov

Bezrukova

Karabanov

(2007) Vegetation and climate dynamics during the Holocene and Eemian interglacials derived from Lake Baikal pollen records. Palaeogeography, Palaeoclimatology, Palaeoecology 252: 440–457.

50.

Tarasov

Dorofeyuk

Metltseva

(2000) Holocene vegetation and climate changes in Hoton-Nur basin, northwest Mongolia. Boreas 29: 117–126.

51.

Tarasov

Jolly

Kaplan

(1997) A continuous late glacial and Holocene record of vegetation changes in Kazakhstan. Palaeogeography, Palaeoclimatology, Palaeoecology 136: 281–292.

52.

Tarasov

Webb

Andreev

(1998) Present-day and mid-Holocene biomes reconstructed from pollen and plant macrofossil data from the former Soviet Union and Mongolia. Journal of Biogeography 25: 1029–1053.

53.

Thompson

Yao

Davis

(1997) Tropical climate instability: The last glacial cycles from a Qinghai-Tibetan Plateau ice core. Science 276: 1821–1825.

54.

Trouet

Esper

Graham

(2009) Persistent positive North Atlantic Oscillation mode dominated the medieval climate anomaly. Science 324: 77–80.

55.

Van Geel

Bokovenko

Burova

(2004) Climate change and the expansion of the Scythian culture after 850 BC: A hypothesis. Journal of Archaeological Sciences 31: 1735–1742.

56.

Wang

Qian

Zhang

(1995) Pollen Flora of China. 2nd Edition. Beijing: Science Press (a book in Chinese).

57.

Wang

Feng

Z-D

(2013) Holocene moisture evolution across the Mongolian Plateau and its surrounding areas: A synthesis of climatic records. Earth-Science Reviews 122: 38–57.

58.

Weijers

Schouten

van den Donker

(2007) Environmental controls on bacterial tetraether membrane lipid distribution in soils. Geochim Cosmochim Acta 71: 703–713.

59.

Wen

Xiao

(2013) Pollen-climate transfer functions intended for temperate eastern Asia. Quaternary International 311: 3–11.

60.

Lin

(2012) Interdecadal variability of the ENSO-NAO connection in boreal summer. Quarterly Journal of Royal Meteorological Society 138: 1668–1675.

61.

Shi

Wang

(2016) Nonstationary impact of the winter North Atlantic Oscillation and the response of mid-latitude Eurasian climate. Theoretical Applied Climatology 124: 1–14.

62.

Chen

(2000) Palaeovegetation of China: A pollen data-based synthesis for the mid-Holocene and last glacial maximum. Journal of Biogeography 27: 635–664.