Climatically Driven Holocene Glacier Advances in the Russian Altai Based on Radiocarbon and OSL Dating and Tree Ring Analysis

Abstract

Analysis of new chronological data, including 55 radiocarbon, 1 OSL, and 8 dendrochronological dates, obtained in the upper reaches of trough valleys within the Katun, North Chuya, South Chuya, and Chikhachev ranges, together with the 55 previously published ones, specifies climatically driven glacier dynamic in the Russian Altai. Available data refute the traditional concept of the Russian Altai Holocene glaciations as a consecutive retreat of the Late Pleistocene glaciation. Considerable and prolonged warming in the Early Holocene started no later than 11.3–11.4 cal kBP. It caused significant shrinking or even complete degradation of alpine glaciers and regeneration of forest vegetation 300–400 m above the modern upper timber limit. Stadial advances occurred in the middle of the Holocene (4.9–4.2 cal kBP), during the Historical (2.3–1.7 cal kBP), and the Aktru (LIA thirteenth–nineteenth century) stages. New radiocarbon ages of fossil soils limited glaciers expansion in the Middle Holocene by the size of the Historical moraine. Lesser glacial activity between 5 and 4 cal kBP is also supported by rapid reforestation in the heads of trough valleys. Glaciers advance within the Russian Altai, accompanied by accumulation of the Akkem moraine, could have occurred at the end of the Late Pleistocene.

1. Introduction

Glaciers are one of the natural markers of global climate change. With the observed progressive warming, rapidly shrinking glaciers continue to be one of the main sources of fresh water in the center of Eurasia—the largest continent on Earth. To assess the effects of their complete disappearance and to establish the recurrence of climatic events similar to those observed at the present time, it is necessary to obtain the full advantage from studying dynamics of glaciation in the mountain systems of Central Asia in the recent geological time. The Holocene is the period that replaced the era of global glaciations and continues to the present day. The relatively small (in comparison with the Pleistocene) surface of the Holocene glaciation stipulated a more sensitive reaction of glaciers even to small climate changes. It allows looking for possible analogies of progressive retreat of modern intracontinental glaciers in the Holocene.

Altai is the northernmost mountain uplift within the Central Asian mountain belt (Figure 1). It borders the West Siberian Plain in the north. Altitudes of the ranges exceed 3500–4200 m a.s.l. with the highest summit, Belukha Peak (Katun range), reaching 4506 m. Despite the arid climate, due to its position above the regional snow line, this mountain uplift hosts the modern glaciation that feeds the rivers of the Arctic Ocean basin and the inland drainage basins of Central Asia. Altai stretches northwest for more than 1500 km across the borders of Mongolia, China, Kazakhstan, and Russia. Within the Russian Altai, glaciers occupy the highest ridges in its southeastern part. Downstream the glacier tongues, several complexes of terminal and lateral moraines have been preserved in topography. Moraines are located both next to retreating glaciers and up to several tens of kilometers far from them. They are associated with the Holocene glacial advances and degradation of the much more powerful last Pleistocene glaciation.

Modern Altai glaciers were described for the first time by Gebler in 1835. At the edge of the nineteenth and twentieth centuries, an important step in regional glaciological investigations was made by Sapozhnikov [1], who refuted the widespread opinion at that time about the insignificance of both ancient and modern glaciations in Altai (as well as in Siberia as a whole). He initiated systematic observations of the retreating glaciers in the Russian Altai.

The first attempt to systematize the Holocene glacial dynamics within the Russian Altai was made in 1938 by Vardanyants [2]. He identified eight stages of glacial retreat in the Caucasus and correlated them with the Alpine ones. To expand this scheme to the Russian Altai, he analyzed the data obtained by that time on glaciations of the Katun range [3,4]. As a result, the Holocene stages (named after local lakes and rivers) for Altai glaciation were correlated to the scheme of Würm glacier degradation (

Table 1. Concept of the Holocene glaciation as stages of a successive reduction of the last Pleistocene glaciation according to different authors.

Glacial Stages, after [2]	Altitude of Terminal Moraines of Different Stages for Russian Altai. Glacial Stages Chronology for Some Central Asia Mountains, after [6]		Glaciers Advances and ELA Depressions for Aktru Basin, the North Chuya Range, after [7]		Stadial Terminal Moraines for the Second Megastadial and ELA Depressions for the Right Aktru Glacier, North Chuya Range, after [8]
	Altitude m a.s.l.	Chronological Limits of Glacial Stages *	Phase of Glaciers Advances	ELA Depression m	Moraine Number (from the Glacier)	ELA Depression m
Seventeenth–nineteenth centuries	2200		VII	50–70	1	70
Historical	2300	~0 AD	VI	100	2	175
Akkem	1900–2000	~1900 BC	V	250	3	285
Kochurlinskaya	1700	3700–3800 BC	IV	450–470	4	355
Multinskaya	≈1550	5600–5800 BC	III	600	5	430
Ognevskaya	1350	7400–7600 BC	II	650	6	485
First	1250	9200–9400 BC	I	<700	7	540
Würm maximum	≈1100	11,000–11,300 BC	-		8	625

* Interstadial periods are supposed to be ~1800 years.

). Since then, this concept of the Altai Holocene glaciation as stages of a successive reduction of the last Pleistocene glaciation became traditional and was not questioned for many decades [5,6,7,8]. Due to the weak development of numerical dating techniques in 1960–1980, there were attempts to estimate the age of moraines in trough valleys of the Russian Altai, applying a time difference of 1800–1900 years between subsequent landforms [6,8]. This simplified “dating” technique was based on Shnitnikov’s hypothesis of the eight-fold, 1850 year rhythm of humidity fluctuation in the Northern Hemisphere [6]. Correlations for high mountainous glaciated areas in inner Asia (by the example of the Russian Altai) were made by Snitnikov on the basis of the Vardanyants scheme [2]. However, moraine complexes associated with all eight “expected” Holocene glacial stages cannot be clearly identified in any of the trough valleys of the Russian Altai.

Generally, in the heads of trough valleys of the SE Altai, three moraine complexes approach confident morphological identification. Their typical mutual location is given in Figure 2. Moraines of the Aktru (Little Ice Age–LIA) and Historical stages are the nearest to the glaciers’ tongues. The Aktru moraine often partly covers the Historical moraine. The Akkem moraine is located far from the glacier terminus (about 5–7 km downstream the valley) and is not clearly expressed in all valleys.

A moraine of the Aktru stage is presented in all trough valleys within the study area. Usually, it consists of a stack of different till units that are nested one after another and are located in the immediate vicinity of the modern glaciers. Terminal moraines dam valleys from one slope to another. Even small glaciers formed massive terminal moraines—often over 100 m in height. Initial tree vegetation settles the moraine’s surface, and inchoate top-soil has an insular distribution. The beginning of the Aktru stage can be attributed to the second half of the thirteenth century [9] and references therein, when the larger glaciers started to advance. Several glacier oscillations are identified in topography and confirmed by lichen analysis [10] and dendrochronology [11]. The accumulation of the nearest to the modern glaciers’ morphologically expressed moraines took place in the seventeenth–eighteenth centuries [9,12,13].

The moraine of the Historical stage represents clearly identified, extensive ramparts more or less covered by tree vegetation, which contacts the moraine of the Aktru stage. Usually, it consists of large stone blocks. The Historical moraine of large glaciers often includes two or three main oscillation ramparts, which are complicated by smaller ones, indicating a consecutive reduction in glacial activity. Identifying such structures for the smaller glaciers is difficult. The Historical moraine of the Maly Aktru glacier (North Chuya range) is completely overlapped by the Aktru (LIA) moraine. This is an exception, which indicates glaciers’ advances during the Historical stage to be more powerful in comparison with LIA expansions. The chronological limits of the Historical stage, about 2.3–1.7 cal kBP, have been estimated by radiocarbon dating of warm periods before and after the stage [9]. Direct dating of glacial deposits associated with the Historical moraine as well as analysis of newly available proxy data could clarify the time limits of this stage.

The Akkem terminal moraine complex, for the first time, was described in the Akkem trough valley, Katun range [3]. It is located at 2050–2080 m a.s.l., about 6–7 km far from the front of the retreating eponymous glacier. The Akkem moraine is also described in detail in some trough valleys of the North and South Chuya ranges [7], although this moraine cannot be identified in all glacial valleys of the Russian Altai. Ancient glaciers formed terminal moraines in the form of short ramparts. The external sides of these ramparts form steep slopes, whilst the internal sides confine a gentle, slightly widened part of the valley bottom, which is usually occupied by a lake or covered by pebbles. In some valleys, this moraine has more than one rampart, indicating several glacial advances within a stage. In spite of good preservation and clear-cut morphological distinction, there are no absolute age determinations for the Akkem moraine. Up to the present time, recognizing the chronological limits of its formation is quite a debatable issue. Some researchers supposed the Holocene age of these landforms [2,6,7,8,9,14,15], sometimes citing the Vardanyants scheme as an argument. At the same time, in the Altai stratigraphic scheme [16], the cold Akkem stage is correlated with the end of the Late Pleistocene glaciation. However, none of these views has a reliable geochronological confirmation yet. Thus, today, one of the most debatable issues in regional paleogeography is a problem of the relationship between the Akkem moraine and the Middle Holocene cooling: what is the age of the landform identified as the Akkem moraine, and what landforms correspond to the climate deterioration and possible glaciers advances in the middle of the Holocene?

Before the wide development of numerical dating techniques, age estimation for glacial sediments and landforms’ relative dating techniques were applied based on morphological criteria, absolute altitudes of landforms, the state of their preservation, estimations of equilibrium line altitude (ELA) depression, etc. Problems in applying these data were discussed [9]. The development of numerical dating techniques suitable for glacial and associated deposits resulted in the appearance of absolute dates since the 1970s. Among them are radiocarbon [9,15,17,18,19,20,21,22,23,24,25,26], dendrochronological [12,13,14,27,28,29], and lichenometry [10] dates of glacial landforms and deposits.

By the end of the twentieth century, however, the available set of numerical dates was still insufficient to reconstruct the Holocene chronology of glacial advances, and only the time limits of the youngest Aktru stage (LIA) were reasonably constrained [9] and references therein. In the last two decades, glaciers’ retreat allowed the possibility of radiocarbon dating of tree fragments buried by moraines and wood remains washed up from glaciers and scattered over proglacial forefields [9,15,17,21,22,23,26,28,29,30,31]. Additionally, the fluctuation of the upper timber limit, associated with the oscillation of glaciers, was studied, as well as the settling of tree vegetation of the modern glaciation zone during warm interstadial periods [9,27,28]. It was established that no later than 7 ka ago, the preceding Pleistocene glaciation completely degraded, and the reactivation of glaciers—neoglacial— already began in the second half of the Holocene [9,17,27].

Radiocarbon dating of wood fragments melted out from retreating glaciers and found above the modern upper timber limit showed the presence of time gaps in the growth of tree vegetation within the axial parts of the ridges during the Holocene [9,15,17,21,22,23,27]. Since, with the exception of the closest to the glacier LIA moraine, there was no possibility to date moraine directly, these gaps were correlated with the time of formation of moraine complexes located in the heads of trough valleys. This led to the conclusion about the beginning of the neoglacial in Russian Altai at the middle of the Holocene. Supposed climate deterioration at that time caused glaciers to advance and the formation of the Akkem moraine—the third one from the modern glaciers.

Since the time of the last reviews on the topic [9,17,27], we have obtained a large amount of new proxy data on the climatically driven Holocene glaciers dynamics in the Russian Altai. It allows clarifying and partially revising the previous conclusions. This article examines the following main problems: (i) the time of final degradation of the Last Pleistocene glaciation in the northern segment of the Central Asian mountain belt, as well as the validity of views on the Holocene glaciation of the Russian Altai as stages of its degradation; (ii) the chronological benchmarks of the beginning of the neoglacial as well as the magnitude of climatically driven glaciers advances in the Russian Altai in the middle of the Holocene; (iii) the time of formation of the Akkem moraine in trough valleys of the Russian Altai.

2. Study Area

The high-mountain, south-eastern part of the Russian Altai (SE Altai, Figure 1), which is examined in this paper, is characterized by strongly dissected topography—the result of Cenozoic orogenesis [5,32,33]. Summits rise above the floors of intermountain depressions and major valleys up to 2600–3000 m, which is one of the largest values for the whole Altai Mountains system [33].

Modern glaciation within the SE Altai is controlled by a severe ultracontinental climate. Within the Chuya depression, the mean annual temperature is −5.2 °C, and the mean annual precipitation is less than 120 mm [34]. The main moisture transfer is from the west and to a lesser degree from the north. It defines aridity intensification in the southeast direction. The mean annual precipitation near the snow line decreases along the W–E axis from 2000 mm to less than 500 mm [35], and as a result, the altitude of the snow line rises from 2400 m a.s.l. in the western part of Katun range to 3350 m a.s.l. in the Chikhachev range.

Despite the arid climate due to the high altitudes of the ridges, the SE Altai is the center of modern Altai glaciation; about 75% of the modern glaciated area in the Russian Altai, with a total surface of 910 km², is concentrated here [36]. Under conditions of moisture deficit, the main factors controlling glacier cover are absolute altitudes, slope insolation (which is particularly important in the case of the sub-latitudinal trending of the most mountain ranges within the SE Altai), asymmetric structure of mountain ranges (gentle northern slopes are much longer than steep southern ones, which cause a larger surface for the glacier basins with northern exposure), and the concentration of solid precipitation on northern leeward slopes [9,37,38]. The overwhelming majority of the glaciers are located on the northern slopes of the highest ranges of SE Altai: Katun, South Chuya, and North Chuya. The total ice volume on the northern slopes of the North Chuya range is twice as much as on the southern ones, and within the South Chuya range located to the east, this ratio is about 7:1 [37]. Modern glaciers have been steadily retreating since the middle of the nineteenth century [25,30,39,40]. Holocene glacial deposits are widely distributed within the axial parts of the Katun, North, and South Chuya ranges [9,22,23]. The similar distribution of modern, Holocene, and LatePleistocene glaciers indicates the same oroclimatic zoning and moisture deficit in this area from the beginning of the Late Pleistocene [38].

The area is characterized by a cryoarid, permafrost-affected environment: the thickness of permafrost in the Chuya basin is 6–90 m; the active layer is 3–7.0 m deep [41]. Within mountain ranges, unconsolidated sediments are frozen from a depth of about 0.8–1.5 m [42].

The postglacial Holocene period within the study area shows different types of sedimentation, including alluvial, lacustrine, and colluvial ones. Peats are presented in the central parts of the Chuya and Kurai intermountain depressions along the Chuya river and in the trough valleys of surrounding ridges. The soils were formed and buried repeatedly during the Holocene at different altitudes—from the basins’ floor up to the head of trough valleys [43,44,45,46]. All those buried paleosols have mollic horizons, which are well developed and rich in organic carbon. Comparative analysis of the contemporary surface soils and the Holocene buried soils of different ages allowed for the conclusion that the last thousand years were, climatically, the most arid [45].

Forest development within the study area is limited by temperatures during the growing season. Contrary to the widespread opinion about an increase in the upper forest line in the N–S direction [47], it does not change significantly within the SE Altai. Everywhere, areas of mature forest are limited to heights of 2220–2330 m a.s.l., while single trees and new growth under favorable conditions are found up to 2500 m a.s.l. Today, the upper forest line is slightly above or equal to the altitudes of glacier terminuses. In the southern and southeastern parts of the study area, forest development is significantly affected by moisture deficit, the inflow of groundwater (often associated with permafrost), and slope exposure. This leads to the disappearance of Siberian pine (Pinus sibirica Du Tour) from the forest-forming species near the upper forest line, where larch (Larix sibirica Ledeb) becomes widespread on the northern macroslope of the South Chuya range, northern framing of the Ukok plateau, and the Mongun-Taiga massif. At the same time, within the Katun and North Chuya ranges, Siberian pine reaches the upper limits of distribution along with larch. Today, due to severe climate, forest expansion on the glacier forefields has an insular distribution. The settling time for deglaciated areas can reach up to 45–100 years and more [28]. It is minimal at low altitudes and increases when approaching the forest thermal limit, which, according to [48], is 9 °C (average July temperature within the North Chuya range).

3. Materials and Methods

The presented study is focused on studying the Holocene glaciation in the main glacial centers of the Russian Altai: the Katun, North Chuya, South Chuya, and Chikhachev ranges (Figure 1). Collections of wood fragments found in the course of many years of field research above the upper timber limit on trough slopes and within glaciers forefields are analyzed. Detailed geological geomorphological and geochronological investigations were carried out also in the upper part of the Akkol trough valley, South Chuya range, to understand the relationship between glacial and non-glacial landforms and deposits. Due to discovering of fossil soils and peats, this valley serves as a key location for clarifying the magnitude of climatically forced glaciers advances in the middle of the Holocene and estimating the age of the Akkem moraine.

The paleogeographical and paleoglaciological reconstructions in the Akkol valley were based on geomorphological mapping at a scale of 1: 50,000. Interpretation of remote sensing data and field observation, as well as lithological analysis of loose sediments exposed in natural outcrops and pits, formed the basis of such mapping.

The genetic interpretation of landforms composed of coarse-grained material (moraines, landslides, rockfalls, and rock glaciers) was made by analyzing the morphology of accumulative bodies, their position in the geomorphological system, petrographic composition and roundness of composing particles, and sources of debris removal. Landslides and rockfalls downstream the moraine complexes of the Historical and Aktru (LIA) stages were studied for their possible transformation associated with glacier expansion during the previous Akkem stage.

Paleosoils and peats are a valuable source of paleoenvironmental information. Analysis of the hypsometric positions and radiocarbon ages of buried soils and peats was utilized to specify the chronology and magnitude of the climatically driven glacier advances in the middle of the Holocene.

Another valuable source of information on the chronology and magnitude of the Holocene glacier fluctuations are paleotrees discovered in the upper parts of the trough valleys in the modern glaciated zone or above the upper timber limit. They allow distinguishing stable periods of warming with favorable for tree-growing climate conditions or establishing the fact of tree deaths caused by glaciers advances.

Available materials (paleowood, charcoals, fossil soils, peats, and glaciofluvial sands) defined applied dating techniques: ¹⁴C, OSL methods, and dendrochronological analysis.

Most of the samples were dated by the radiocarbon method. Depending on the weight of the sample, LSC or AMS technique was applied. The first one was utilized at the Cenozoic Laboratory, Institute of Geology and Mineralogy SB RAS, Novosibirsk (SOAN), and at the Laboratory of radiocarbon dating and electronic microscopy of the Institute of Geography RAS, Moscow (IGAN). The production of lithium carbide and benzene synthesis was completed by using the standard technique [49,50]. The activity of ¹⁴C was determined by using the Quantulus-1220 liquid scintillation counters. One date was obtained applying accelerator mass-spectrometry (IGAN_AMS). The sample preparation, separation of the datable fraction, graphitization, and pressing on a target were performed in the Laboratory of radiocarbon dating and electronic microscopy of the Institute of Geography RAS, Moscow. The graphitization was performed by using an AGE-3 graphitization system [51]. The AGE-3 uses a Vario Micro Cube elemental analyzer coupled to an Isoprime PrecisION IRMS (Elementar). Graphite ¹⁴C/¹³C ratios were measured by applying the CAIS 0.5 MeV accelerator mass spectrometer at the Center for Applied Isotope Studies, University of Georgia, USA. The sample ratios were compared to the ratio measured from the Oxalic Acid II (NBS SRM 4990C). All ¹⁴C ages were calculated by applying δ¹³C value of −25‰. The conventional radiocarbon ages were calibrated (2σ standard deviation) by applying the CALIB Rev 7.1 program (http://calib.org/calib/, accessed on 9 September 2019), with the IntCal13 calibration data set [52]. The paper presents both the conventional (years BP) and the calibrated (cal BP) ¹⁴C ages.

Dendrochronological analysis was utilized for dating paleotrees discovered on the Historical moraine in the Akkol valley. Within the study area, there are good opportunities to obtain reliable dendrochronological ages for Larix sibirica Ledeb up to 3200 years old, and for Pinus sibirica Du Tour, up to 1200 years old [28]. The success of dating practically does not depend on the location of collected samples. The only requirement is the altitude to be no lower than 2100 m a.s.l. Below this mark, the effect of the main limiting factor—temperature—is significantly reduced. Fragments of pine trunks were dated applying the “Aktru” chronology [28], which is 1188 years long (AD 822–2010) and was built on Pinus sibirica Du Tour.

To estimate the age of glaciofluvial deposits one sample for OSL dating was collected by driving a steel tube into the sandy sediment. In the laboratory, the sample was prepared both for gamma spectrometry and luminescence measurements. High-resolution gamma spectrometry using an HPGe detector was used to determine the content of U, Th, and K in the sample. Prior to these measurements, the sample was stored for three weeks to ensure equilibrium between gaseous ²²²Rn and ²²⁶Ra in the ²³⁸U decay chain. The gamma spectrometry measurement lasted for 24 h. The activities of the isotopes present in the sediment were determined by using IAEA standards RGU, RGTh, and RGK after subtracting background values from the detector. The dose rates were calculated by using the conversion factors of Guerin et al. [53]. The method of Prescott and Stephan [54] was used for the cosmic ray beta dose rate calculation. The average water content was assumed to be 15 ± 5%. For OSL measurements, sand-sized grains of quartz (90–125 μm) were extracted, and standard chemical procedures were applied [55]. The sediment sample was treated with 20% hydrochloric acid (HCl) and 20% hydrogen peroxide (H₂O₂) to remove carbonates and organic material. Then, density separation of quartz grains was applied by using sodium polytungstate solutions to pick out grains of densities between 2.62 g·cm⁻³ and 2.75 g·cm⁻³. The selected grains were etched twice in 40% hydroflfluoric (HF) acid for 60 min and 40 min to remove the outer 10 μm layer, which absorbed a dose from alpha radiation and remaining feldspar contamination. An automated Risø TL/OSL DA-20 reader was used for the OSL measurements of multi-grain aliquots, each weighing ~1 mg. The stimulation light source was a blue (470 ± 30 nm) light-emitting diode array delivering 50 mW·cm⁻² at the sample [56]. The detection was through a 7.5 mm-thick Hoya U-340 fifilter. Equivalent doses were determined by using the single-aliquot regenerative-dose protocol [57]. To calculate the luminescence ages, the central age model (CAM) of Galbraith et al. [58] was applied.

To estimate the chronological limits of glacial and nonglacial events, new dating results were accompanied by an analysis of previously reported radiocarbon data.

4. Results

4.1. Holocene Chronology of Glaciers Advances in the Russian Altai. Distribution of Forest Vegetation within the Periglacial Part of the Ridges in the Russian Altai

As a result of long-term field studies, the database of radiocarbon dates of paleotrees found in the upper reaches of the trough valleys beyond the modern upper timber limit was expanded by 24 dates (

Table 2. New radiocarbon dating results for paleotrees fragments buried by moraines, collected from proglacial forefields and above the modern upper timber limit in the heads of trough valleys within the study area.

N	Lab. Code	Location	N° E°	Altitude m a.s.l.	Sample Type	14C Age	Calibrated Age (2σ)	Tree Specie	Comments
Katun range
1	SOAN-8745	Mensu valley, forefield of Mensu glacier (number 4, Figure 4)	N 49°51′35″ E 86°42′31″	2137	Tree fragment washed out from glacier	3315 ± 35	3545 ± 85	Pinus sibirica	Trees grow within modern glacial zone
2	SOAN-8746	Mensu valley, forefield of Mensu glacier (number 4, Figure 4)	N 49°51′19″ E 86°42′23″	2152	Tree fragment washed out from glacier	4625 ± 85	5315 ± 265	Pinus sibirica	Trees grow within modern glacial zone
3	SOAN-8752	Suluajry valley, Iedygem basin (number 1, Figure 4)	N 49°55′30″ E 86°40′20″	2596	Tree trunk buried by landslides	5670 ± 110	6500 ± 220	Pinus sibirica	Warm and humid climate, trees grow above the modern upper timber limit
4	SOAN-8753	Suluajry valley, Iedygem basin (number 1, Figure 4)	N 49°55′30″ E 86°40′19″	2596	Tree trunk buried by landslides	9880 ± 115	11,345 ± 465	Pinus sibirica	Warm and humid climate, trees grow above the modern upper timber limit
5	SOAN-8754	Suluajry valley, Iedygem basin (number 1, Figure 4)	N 49°55′30″ E 86°40′19″	2596	Tree trunk buried by landslides	5305 ± 90	6100 ± 180	Pinus sibirica	Warm and humid climate, trees grow above the modern upper timber limit
6	SOAN-8755	Suluajry valley, Iedygem basin (number 1, Figure 4)	N 49°56′54″ E 86°40′36″	2620	Tree trunk in fluvial deposits in front of moraine	4760 ± 85	5480 ± 165	Pinus sibirica	Trees grow within modern glacial zone
7	SOAN-9110	Suluajry valley, Iedygem basin (number 1, Figure 4)	N 49°55′31″ E 86°40′11″	2651	Tree trunk buried by landslides	6620 ± 70	7515 ± 95	Pinus sibirica	Warm and humid climate, trees grow above the modern upper timber limit
8	SOAN-9111	Tekelu valley, Akkem basin (number 3, Figure 4)	N 49°54′04″ E 86°37′59″	2721	Root fragment in glaciofluvial deposits, inner part of modern moraine	2960 ± 80	3120 ± 230	Juniperus	Trees grow within modern glacial zone
9	SOAN-9439	Mensu right tributary valley (number 4, Figure 4)	N 49°51′39″ E 86°41′20″	2557	Tree fragment washed out from glacier	7440 ± 95	8225 ± 180	Pinus sibirica	Warm and humid climate, trees grow above the modern upper timber limit
10	SOAN-9810	Jarlu, watershed Jarlu-Tekelu (number 2, Figure 4)	N 49°55′48″ E 86°36′34″	2728	Fragment of tree roots	4910 ± 115	5620 ± 290	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
11	SOAN-9811	Jarlu, watershed Jarlu-Tekelu (number 2, Figure 4)	N 49°55′07″ E 86°36′37″	2700	Fragment of tree trunk	7235 ± 140	8070 ± 280	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
12	SOAN-9812	Jarlu, watershed Jarlu-Tekelu (number 2, Figure 4)	N 49°55′30″ E 86°36′45″	2700	Fragment of tree trunk in moraine	8605 ± 130	9730 ± 420	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
13	SOAN-9813	Jarlu, watershed Jarlu-Tekelu (number 2, Figure 4)	N 49°55′14″ E 86°36′42″	2674	Fragment of tree roots in situ	6050 ± 250	6050 ± 250	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
14	SOAN-9888	Mensu valley, forefield of Sapozhnikov glacier (number 4, Figure 4)	N 49°52′00″ E 86°43′11″	1993	Tree fragment in front of Aktru moraine	3215 ± 90	3445 ± 230	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
15	SOAN-9889	Mensu valley, forefield of Sapozhnikov glacier (number 4, Figure 4)	N 49°51′32″ E 86°42′50″	2149	Tree fragment washed out from modern glacier	2925 ± 85	3100 ± 240	Larix sibirica	Warm and humid climate, trees grow above the modern upper timber limit
North Chuya range
16	SOAN-8748	Jan-Karasu valley (number 2, Figure 5)	N 50°06′59″ E 87°44′55″	2337	Tree fragment in rock glacier	4250 ± 75	4785 ± 250	Larix sibirica	Warm and humid climate
17	SOAN-8749	Jan-Karasu valley (number 2, Figure 5)	N 50°06′57″ E 87°44′40″	2384	Tree fragment in rock glacier	4540 ± 65	5210 ± 235	Larix sibirica	Warm and humid climate
18	SOAN-9840	Maashey valley, forefield of Maashey glacier (number 1, Figure 5)	N 50°06′44″ E 87°35′37″	2159	Tree fragment washed out from modern glacier	5085 ± 95	5850 ± 245	Pinus sibirica	Warm and humid climate
19	SOAN-9841	Maashey valley, forefield of Maashey glacier (number 1, Figure 5)	N 50°06′44″ E 87°35′37″	2159	Tree fragment washed out from glacier	5520 ± 95	6260 ± 245	Pinus sibirica	Warm and humid climate
South Chuya range
20	IGAN 5966	Akkol valley (number 2, Figure 6)	N 49°50′19″ E 87°52′40″	2334	Charred fragment of tree trunk buried by aeolian sediments	3630 ± 70	3935 ± 215	Larix sibirica	Warm and humid climate
21	IGAN 6009	Akkol valley (number 2, Figure 6)	N 49°49′12″ E 87°50′51″	2381	Bark fragment on landslide in front of the Historical moraine	3320 ± 80	3600 ± 220	Larix sibirica	Warm and humid climate
22	SOAN-8747	Taltura valley, forefield of Taltura glacier (number 1, Figure 6)	N 49°50′21″ E 87°42′13″	2460	Tree fragment washed out from glacier	3070 ± 35	3240 ± 125	Larix sibirica	Warm and humid climate
23	SOAN-9908	Akkol valley (number 1, Figure 6)	N 49°49′06″ E 87°50′45″	2390	Tree fragment	2865 ± 110	3045 ± 280	Larix sibirica	Warm and humid climate
Chikhachev range
24	IGANAMS 6525	Boguty valley	N 49°41′32″ E 89°31′06″	2473	Fragment of charcoal	6500 ± 25	7400 ± 70	Larix sibirica	Warm and humid climate

, Figure 3 and Figure 4). Wood fragments were found in most of the glacial centers of the Russian Altai—Katun (15 samples), North Chuya (4 samples), South Chuya (4 samples), and Chikhachev (1 sample) ranges. Samples were collected in the upper part of troughs at altitudes from 2730 down to 1990 m a.s.l.: (i) within the watersheds rising above (400–500 m) the modern upper timber limit; (ii) within forefields of modern glaciers, where paleotree fragments are washed out from glaciers by meltwater; and (iii) downstream the Historical moraine. The ¹⁴C dates fall within the time interval of 11.5–3 cal kBP.

4.1.1. Katun Range

Most of the finds (15) were made in the Katun range—the highest one within the Altai uplift. An equal number of samples (five) was collected in each of the Akkem, Iedygem, and Mensu basins (Figure 5), which drain the northern slope of the range located in Russia.

In the upper reaches of the Akkem river (Tekelu—the right tributary of the Akkem river), a fragment of a coniferous (juniper?) root shoot, of about 3 ka (SOAN-9111, Table 2), was found in fluvioglacial deposits within the moraine of the Tekelu glacier at an altitude of 2720 m a.s.l. Today, modern tree vegetation, including juniper, does not grow in the valley at this altitude.

On the narrow watershed of the Akkem river tributaries—Tekkelu and Yarlu—at an altitude of 2670–2730 m a.s.l., as a result of erosion, a horizon with numerous root fragments (Larix sibirica) buried by slope deposits was uncovered. This horizon is developed over the Pleistocene (?) moraine overlying the bedrocks. Radiocarbon ages (SOAN-9810, 9813, Table 2) of roots fragments in situ indicate the growth of the larch forest here about 5.6–6 ka ago. Together with the dates (SOAN-9615, 9617, 9640, 9637, 9641, Table S1) obtained earlier from this horizon in other locations, this information indicates a forest distribution in the Katun range at an altitude of 2670–2730 m a.s.l. for a long period (at least 2.5–3 thousand years long), from about 6.8 until 4.3 ka ago.

In addition to the root horizon, numerous finds of fragments of larch trunks (sometimes more than 2 m long and 30 cm in diameter) were made in this location. They are exposed to sliding moraine deposits for almost half a kilometer along the watershed. Obtained radiocarbon ages (SOAN-9728, 8071, Table 2; SOAN-9590, 9591, 9593, 9638, 9811, 9812, Table S1) indicate an even earlier settling of the periglacial zone of the Katun range by forest vegetation. It occupied this place, which is much higher above the modern upper forest line, already 9.7–8 cal kBP.

In the upper reaches of the Iedygem basin, at the headwaters of the Suluayry river (Figure 5), numerous finds of large fragments of tree trunks were made near the watershed at an altitude of 2600–2650 m a.s.l. The nearest forest in this place is located more than 4 km downstream of the valley. A section of buried forest is exposed in a large erosion funnel as a result of erosion of moraine, slope, and proluvial deposits. The discovered trunk fragments are up to 180 cm in length and 40 cm in diameter. The four radiocarbon ages (SOAN-8752, 8753, 8754, 9110, Table 2) indicate a wide scattering of dates—from 6.1 to 11.3 cal kBP. Despite such a significant difference in the lifetime of trees, all the trunks fragments were well preserved. There are no rot and traces of decaying wood, which indicates their rapid burial. At present, the slope angle of the erosion funnel exceeds 10°. On its surface, there are widespread sliding and creeping terraces with a height of frontal scarps from 1 to 3 m and length from several tens to the first hundreds of meters.

Based on the scatter of dates, the forest had been growing here for a long time—at least four thousand years; however, the repeated death of trees during this period was most likely caused by the sliding of loose waterlogged soil. The tree-ring analysis of discovered tree fragments shows the presence of only young trees no older than 233 years. The youthfulness of the trees also testifies to the activity of slope processes in this part of the watershed during the entire period of the forest’s existence. Out of 21 cuts, 14 belong to pine (Pinus sibirica Du Tour) and 7 to larch (Larix sibirica Ledeb). All samples have a pronounced age trend, which disappears only after 100–150 years. An attempt to build a tree-ring chronology with these samples failed, which indirectly confirms the significant scattering of the tree lifetimes.

Siberian pine is more sensitive (in comparison with larch) to temperature and humidity. Its presence above the modern upper forest line is also evidence of a much warmer and more humid climate during the entire period of their existence. Intensification of soil creep and land sliding incompatible with reforestation was the most likely reason for its final death.

Fragments of tree trunks were also found in the headwaters of the Suluayry river 40–50 m downstream the front of the modern moraine at an altitude of 2620 m a.s.l. The radiocarbon age of one of the fragments (SOAN-8755, Table 2) indicates favorable climate conditions for the growth of forest vegetation in the modern glacial zone of the Katun range of about 5.5 cal kBP.

In the upper reaches of the Mensu valley, two samples (SOAN-8745, 8746, Table 2) were collected in the immediate vicinity of the Mensu glacier. Tree fragments were washed out by meltwater from the glacier tongue. Two more samples were collected in deposits of a high floodplain nearest to the glacier Aktru (LIA) moraine (SOAN-9889, 9888, Table 2). Three of these four samples, including two pines, have the same age: 3.1–3.5 cal kBP. The closest to the glacier tree fragment is older: 5.3 cal kBP. Fragments of trunk and chips on a rocky bar of the hanging valley in the Mensu tributary at about 2560 m a.s.l. are evidence for even earlier (about 8.2 cal kBP SOAN-9439, Table 2) settling of the upper reaches of the Mensu valley by forest vegetation.

4.1.2. North Chuya Range

In the upper reaches of the Maashey river (Figure 6), two fragments of pine trunks (Pinus sibirica Du Tour) were found. During one summer season, they were washed out from the glacier tongue to a distance of more than 1 km. The age of these samples is about 5.8–6.3 cal kBP (SOAN-9840, 9841, Table 2).

In the upper reaches of the Yan-Karasu river, two large trunks of Larix sibirica of excellent preservation were found in rock glacier deposits at 2340–2380 m a.s.l. Lateral moraines, which were accumulated during one of the glacier advances, served as a source of clastic material for rock glaciers. Tree trunks were exposed as a result of the debris movement and erosion of the rock glacier front. Radiocarbon ages of these trees (SOAN-8748, 8749, Table 2) indicate favorable climate conditions for growing tree vegetation in the upper part of the trough valley at the edge of the fifth and sixth millennia. At the same time, the 95% confidence interval of the SOAN-8748 sample partially falls within the interval of 4.2–4.9 cal kBP. This period was assumed to be extremely cold and humid and could potentially provide conditions for accumulating the most distant Akkem moraine [9,17].

4.1.3. South Chuya Range

In the upper reaches of the Taltura river (Figure 7), a fragment of the trunk (Larix sibirica) with an age of 3.2 cal kBP (SOAN-8747, Table 2) was found on the forefields of the Taltura glacier.

In the upper reaches of the Akkol river, small fragments of wood and bark (Larix sibirica, SOAN-9908, IGAN 6009, Table 2) with a similar age (3.0–3.6 cal kBP) were found. They were buried by a landslide. The landslide body, covered by glaciofluvial sands, directly contacts with the moraine of the Historical stage.

Charred fragments of a tree trunk buried by aeolian sands are about 3.9 cal kBP (IGAN 5966, Table 2) and mark a buried ancient-day surface. Deposits are represented by glaciofluvial sands (possibly, sediments of a flowing lake) and are located 3.6 km downstream from the Historical moraine at the foot of the left slope of the valley at 2343 m a.s.l. The sample was collected upstream of the moraine, which can be compared with the moraine of the Akkem stage in the Katun range.

Even taking into account the possibility of the growth of a tree upon the valley slope (from the place of its burning), its lifetime casts doubt on such a distant glacier advance during the period 4.2–4.9 cal kBP (the estimated time of the Akkem moraine formation). This period also includes the date (4590 ± 235 cal BP, IGAN 3694, Table S1) of another trunk fragment, previously found on the forefields of the Sofiysky glacier in the upper reaches of the Akkol valley. The significant width—up to 4 mm—of the tree rings in this sample is not typical for the region and could be an anomaly.

4.1.4. Chikhachev Range

In the Boguty river valley, buried charcoals of Larix sibirica of about 7.4 cal kBP (IGAN_AMS 6525, Table 2) were found near the Upper Boguty lake at 2473 m a.s.l. Earlier, at the same altitude, the charcoals of larches aged 9.3–10 and 8.3–8.6 cal kBP (Table S1) were repeatedly found 10 km downstream the valley, within the mountain framing of Low Boguty lake [24,45,59]. Today, the Boguty valley, like other valleys of the Chikhachev and Sailugem ranges, is completely treeless. This eastern periphery of the Chuya basin is characterized by the most arid climate in the Russian Altai.

4.2. Magnitude of the Holocene Glaciers Advances in the Russian Altai. Geological, Geomorphological, and Geochronological Study of the Upper Reaches of the Akkol River

A unique geological situation for understanding the magnitude of glaciers advances in the second half of the Holocene in Russian Altai has developed in the Akkol valley, South Chuya range. In the upper reaches of this trough valley, 17 sections were studied; 34 radiocarbon and 1 OSL dates were obtained (Figure 8 and Figure 9;

Table 3. Radiocarbon dating results for samples of different origins collected in the upper part of the Akkol valley, South Chuya range.

N	Lab Code	Section	N° E°	Altitude m a.s.l.	Sample Type	Dating Material	14C Age	Calibrated Age (2σ)
1	SOAN-9902	1	N 49°48′54″ E 87°50′32″	2438	Soil on the frontal part of the Historical moraine	Soil	Modern
2	IGANAMS 8772	3	N 49°49′02″ E 87°50′37″	2406	Soil on the frontal part of the Historical moraine	Soil	Modern
3	SOAN-9901	3	N 49°49′02″ E 87°50′37″	2406	Soil on the frontal part of the Historical moraine	Soil	Modern
4	SOAN-9909	3	N 49°49′02″ E 87°50′37″	2406	Soil on the frontal part of the Historical moraine	Soil	Modern
5	SOAN-9683	4	N 49°49′05″ E 87°50′26″	2413	Cryo-humus horizon of polygenetic soil profile in slope sediments	Soil	340 ± 35	395 ± 90
6	IGAN 6246	4	N 49°49′05″ E 87°50′26″	2413	Gray-humus horizon of polygenetic soil profile in slope sediments	Soil	1220 ± 70	1135 ± 150
7	SOAN-9682	4	N 49°49′05″ E 87°50′26″	2413	Dark humus horizon of polygenetic soil profile in slope sediments	Soil	1645 ± 95	1575 ± 230
8	SOAN-9910	5	N 49°49′03″ E 87°50′38″	2396	Paleosoil in landslide body	Soil	930 ± 80	830 ± 140
9	SOAN-9903	5	N 49°49′03″ E 87°50′38″	2396	Paleosoil on the landslide body	Soil	1570 ± 80	1495 ± 190
10	SOAN-9908	6	N 49°49′06″ E 87°50′45″	2392	Tree fragment in paleosoil	Wood	2865 ± 110	3045 ± 280
11	SOAN-9894	6	N 49°49′06″ E 87°50′45″	2392	Paleosoil in front of Historical moraine	Soil	4420 ± 105	5130 ± 305
12	SOAN-9893	7	N 49°49′07″ E 87°50′45″	2391	Humus layer of aeoline deposits above glacio-fluvial sands	Soil	1695 ± 80	1610 ± 200
13	SOAN-9897	8	N 49°49′06″ E 87°50′46″	2391	Paleosoil above glaciofluvial deposits	Soil	615 ± 50	600 ± 65
14	SOAN-9896	8	N 49°49′06″ E 87°50′46″	2391	Paleosoil above glaciofluvial deposits	Soil	1145 ± 65	1085 ± 150
15	SOAN-9895	8	N 49°49′06″ E 87°50′46″	2391	Paleosoil above glaciofluvial deposits	Soil	2980 ± 120	3150 ± 295
16	IGAN 6247	9	N 49°49′12″ E 87°50′51″	2381	Paleosoil on landslide body	Soil	140 ± 50	225 ± 60
17	IGAN 6248	9	N 49°49′12″ E 87°50′51″	2381	Paleosoil on landslide body	Soil	1110 ± 70	1075 ± 160
18	IGAN 6009	9	N 49°49′12″ E 87°50′51″	2381	Fragment of larch bark on landslide	Wood	3320 ± 80	3600 ± 220
19	IGANAMS 8771	10	N 49°49′10″ E 87°50′55″	2373	Paleosoil on the landslide body	Soil	330 ± 20	385 ± 75
20	SOAN-9898	10	N 49°49′10″ E 87°50′55″	2373	Paleosoil on the landslide body	Soil	2225 ± 95	2210 ± 255
21	SOAN-9684	11	N 49°49′24″ E 87°51′28″	2354	Paleosoil in the river terrace	Soil	1490 ± 90	1430 ± 165
22	SOAN-9686	11	N 49°49′24″ E 87°51′28″	2354	Lens of humus material with charcoal in the lower river terrace	Soil	4470 ± 105	5145 ± 300
23	SOAN-9685	11	N 49°49′24″ E 87°51′28″	2354	Lens of humus material with a large amount of root detritus in the lower river terrace	Detritus	5090 ± 95	5885 ± 285
24	SOAN-9687	12	N 49°49′40″ E 87°52′08″	2348	Peat with detritus in the scarp of the lower river terrace	Detritus	1730 ± 60	1675 ± 140
25	IGAN 6100	13	N 49°49′12″ E 87°50′51″	2339	Peat bog on the lower river terrace	Peat	Modern
26	IGAN 6099	13	N 49°49′12″ E 87°50′51″	2339	Peat bog on the lower river terrace	Peat	200 ± 50	330 ± 90
27	IGAN 6098	13	N 49°49′12″ E 87°50′51″	2339	Peat bog on the lower river terrace	Peat	1160 ± 60	1105 ± 150
28	IGAN 6097	13	N 49°49′12″ E 87°50′51″	2339	Peat bog on the lower river terrace	Peat	1760 ± 60	1685 ± 135
29	SOAN-9904	14	N 49°49′41″ E 87°52′16″	2347	Peat in an erosion scarp, high floodplain	Peat	Modern
30	SOAN-9907	15	N 49°50′09″ E 87°52′28″	2334	Peat in the first terrace above the floodplain	Peat	320 ± 40	390 ± 90
31	SOAN-9906	15	N 49°50′09″ E 87°52′28″	2334	Peat in the first terrace above the floodplain	Peat	1350 ± 85	1235 ± 175
32	SOAN-9905	15	N 49°50′09″ E 87°52′28″	2334	Peat in the first terrace above the floodplain	Peat	1440 ± 80	1355 ± 170
33	IGAN 5966	16	N 49°50′19″ E 87°52′40″	2334	Charred fragment of tree trunk buried by aeolian sediments	Charcoal	3630 ± 70	3935 ± 215
34	SOAN-9899	17	N 49°51′01″ E 87°53′20″	2311	Peat in lacustrine deposits of the river terrace	Peat	805 ± 55	780 ± 120

and

Table 4. OSL age (using CAM, [58]) for glaciofluvial sands collected in the upper reaches of the Akkol valley.

Sample ID	Water Content (%)	U (Bq/kg)	Th (Bq/kg)	K (Bq/kg)	Dose Rate (Gy/ka)	Equivalent Dose (Gy)	OSL Age (ka)
GdTL-3905	10 ± 5	26.9 ± 1.4	30.8 ± 1.8	547 ± 46	2.77 ± 0.15	8.12 ± 0.14	2.93 ± 0.16

).

The Akkol river belongs to the Chagan-Uzun river basin—the largest tributary of the Chuya river. It drains the northern macroslope of the South Chuya range. The Sofiysky glacier, being one of the largest glaciers in the Russian Altai, feeds the Akkol river. The axial part of the South Chuya range, actively eroded by glaciers, is mainly composed of gneisses (of chlorite, biotite, cordierite, garnet–sillimanite facies of metamorphism), as well as granites. At the same time, the macroslopes of the range within the upper reaches of the rivers are composed of crystalline quartz–chlorite cordierite schists. The different petrographic composition of rocks makes it possible to confidently determine the source area of debris material presented in the accumulative landforms within the upper part of the valleys. A huge amount of mica, washed out from the glaciers, paints the water of the river in the Chagan-Uzun basin milky white (“Akkol” is a white river in Turkic). Above the mouth of the Chagan-Uzun river, the Chuya waters are clear.

4.2.1. The Upper Reaches of the Akkol Valley: Modern Moraines of the Sofiysky Glacier and Moraine Complexes of the Aktru and Historical Stages Range

Like other glaciers of the Russian Altai, the Sofiysky glacier was rapidly retreating (Figure 10) at an average rate of about 18 m/a over the twentieth century [25,60]. In 1898, during the first visit to the Sofiysky glacier by Sapozhnikov, the glacier front was already separated from the moraines by a lake, and there were icebergs calving [1]. The size of the lake was smaller at that time. From AD 1898 to 2000, it retreated to 1935 m, and the glacier’s terminus was located at 2484.5 m a.s.l. [25]. Since AD 2000, the glacier has continued to retreat. Its surface has noticeably decreased and darkened. Deep sinkholes and subsidence appeared on the tongue. Subglacial meltwater runoff escapes from the glacier tongue with a noisy, muddy stream. A huge amount of micaceous suspension paints it white. Downstream, it cuts closest to the modern glacier moraines, which contain blocks of gneisses up to 6 m, as well as a vast (about 1 km long) outwash plain composed of small debris up to fine sands.

The formation of the outwash plain in front of the retreating glacier was first documented in 1939 [61]. Taking into account the average rate of glacier retreat (18.3 m/a for 1939–1963), two moraine ramparts, documented in 1963, were attributed to the time interval 1944–1951 [62]. By the year 1997, they were practically leveled back. At that time, the right tributary of the glacier was separated from the main glacier [25]. Young moraines, which are located between the glacier terminus and the outwash plain, are well expressed in topography. They consist of unsorted debris and loose deposits and have a complex topography—there are both flattened sandy areas, as well as areas with a large number of boulders and blocks of gneisses and, less often, granites up to 6–7 m.

Steep (up to 30°) lateral moraines stretch along the slopes of the valley above the surface of the glacier. At a distance of about 1.9 km from the glacier’s tongue, they pass into the terminal moraines of the Aktru (LIA) stage, which dam up the Sofiysky Lake. The lake is 860 × 600 m in size; the depth is up to 32 m [63]. Water runoff from the lake runs along the foot of the left slope of the valley. Since 1952 (the time of aerial photographic surveying), the shoreline of the lake has significantly changed as a result of thermal abrasion. No later than the year 2009, a short (about 80 m) bridge between the Sofiysky lake and a small thermokarst lake in the right part of the moraine complex collapsed [63]. The Thermokarst lake was merged, forming a bay. After 2011, the basin of another small lake became a part of this bay. At the present time, the size and the shape of the lakes are still intensively changing.

The Aktru (LIA) moraine has a morphologically fresh surface: it is not turfed (with the exception of rare microgroups of plants), and the soil cover is absent; gneisses and granites are not weathered, which resulted in a distinctive bright gray color of the surface; the terminal and lateral ramparts are settling by sparse undergrowth of Larix sibirica, Picea obovata, and Betula pendula, rejuvenating towards the glacier; mosses and lichens are quite rare, and areas of their distribution are small.

The front of the Aktru moraine, at an altitude of 2460 m a.s.l., overlaps more ancient deposits, which, according to morphological criteria, we refer to the Historical moraine (Figure 11). The moraine complex of the Historical stage stretches downstream the valley for about 750 m until 2390 m a.s.l. (the altitude of the valley bottom), where the moraine front rises up to 35 m. Three or four oscillation ramparts are presented in the structure of the Historical moraine complex. A large number of lakes with water of different colors occupy numerous thermokarst depressions on the moraine’s surface. At the present time, most of the small lakes have become shallow and swampy. Along with this, there is an expansion and an increase in the area of large lakes. This is especially noticeable when compared with the 1952 aerial photographs.

The surface of the moraine has a soil cover and is sodded. This distinguishes it from the Aktru moraine in color despite the similar petrographic composition of the substrate. The thickness of the humus horizon was determined in three pits in the frontal, more ancient, part of the moraine—at 2406, 2430, and 2438 m a.s.l. (Sections 1–3, Figure 8). The soil profile on the lowest moraine rampart (2406 m a.s.l., Section 3, Figure 11) is the most developed. The thickness of horizon A of the soil reaches 19–20 cm. It is formed by mica sands, which serve as a matrix between gneiss and granite boulders and pebbles. Lenses of low-humus sand were also found at a depth of 27–34 cm, which indicates a relatively long pedogenesis. Nevertheless, the radiocarbon dating of all three samples from the depth 5–11, 14–19, and 27–34 cm (SOAN-9909, 9901, IGAN_AMS 8772, Table 3) returned modern ages. Possible rejuvenation of weakly humified sand from a depth of 27–34 cm (IGAN_AMS 8772) can be associated with active modern pedogenesis caused by warming and retreat of glaciers after LIA, as well as mixing of organic material as a result of cryoturbation and permafrost degradation. At the foot and on the crest of the next moraine ridge (at 2430 and 2438 m a.s.l., respectively, Sections 1 and 2), the thickness of horizon A reaches 15 cm. Radiocarbon dating of the lower 10 cm of horizon A on the crest of the rampart (SOAN-9902, Table 3) also returned the modern age.

The moraine surface is settled by shrub (different species of Salicaceae, Juniperus, Lonicera, Betula rotundifolia) and tree (Larix sibirica Ledeb, Picea obovata) vegetation. There are young trees with trunks up to 30 cm in diameter and larch undergrowth. The largest larches grow in the wind shade in the depressions on the frontal part of the Historic moraine. Paleotrees fragments (Pinus sibirica Du Tour), including trunks and trunks with butts, were found on the moraine surface at 2400–2440 m a.s.l. At present, this species, more sensitive to temperature and moisture in comparison to Larix sibirica, does not grow within the South Chuya range. Dendrochronological dating based on Aktru tree ring chronology (AD 822–2010, built on Pinus sibirica Du Tour [28]) revealed the lifetime of these paleotrees was within the thirteenth–seventeenth centuries (AD 1218–1670, Figure 12). The most likely cause of death for the trees was the vicinity of the glacier during the Aktru stage. New growth of tree vegetation (but already the Larix sibirica Ledeb) appeared at the proximal part of the Historical moraine complex in 1897 (according to dendrochronological dating of living trees [64]) when the retreating glacier was about 1 km away from them. So, the lifetime of paleotrees of Siberian pine and the age of new growth of larches, along with different states of soil development, vegetation cover, and topography transformation, confirms the different ages for moraine complexes closest to the Sofiysky glacier. It should be noted that in 1898, Sapozhnikov described the presence of ancient moraines in front of the lake [1]. Nevertheless, later, both ancient and young moraines of the Sofiysky glacier were considered as a single complex, attributed to the LIA [25,26,43,60,62,63].

Near the moraine complexes of the Aktru and Historical stages, the mouth of the right tributary opens into the Akkol valley. It originates from the left branch of the Udachny glacier, which quite recently belonged to transection-type glaciers. Analysis of satellite images indicates that even during the maximal Holocene glacier advances, the left branch of the Udachny glacier did not reach the mouth of its short hanging valley. Its terminal moraines are located on a rocky bar 135 m above the moraine complexes of the Sofiysky glacier. Thus, it could be stated that the asynchrony of moraines formation in different parts of the Akkol valley is a result of the similarity in magnitude of advances of the Sofiysky glacier during the Historical and Aktru stages. The ages of Siberian pines on the Historical moraines indicate the formation of these moraines before the beginning of the thirteenth century, i.e., before the first cooling associated with the Aktru stage [9].

To establish the chronological benchmarks of soil formation in the upper reaches of the Akkol valley and to date the landforms, which contact the moraine complex of the Sofiysky glacier, we studied the soil profile on the surface of the landslide at the foot of the left slope of the valley (Section 4). The landslide occupies the same position as the Historical moraine and is separated from it by the deeply incised Akkol riverbed. Both the landslide and the large non-turfed talus resting on it are composed of blocks of metamorphic quartz–chlorite schists, which compose the slope of the valley in this area. Yellow–brown silty sandy loam covers coarse–clastic slope sediments. It is possible that landslides may partially overlap the Historical moraine. Radiocarbon dates (SOAN-9683, 9682, IGAN 6246, Table 3) indicate that the polygenetic soil profile in the depth interval 12–33 cm (gray-humus, dark humus, and cryo-humus horizons) was formed at about 0.4–1.6 cal kBP. Thus, pedogenesis on the landslide near the Historical moraine began in the warm interstage period (1.7 cal kBP—twelfth century AD according to [9]) and, most likely continued during the Aktru stage. The landslide was triggered earlier than 1580 years ago; however, available data do not allow correlating the time of its formation with the Akkem stage.

4.2.2. Middle Part of the Akkol Valley: Genesis and Age of Landforms Downstream the Moraine Complexes of the Aktru and Historical Stages

Moraines, which conventionally can be correlated with the Akkem moraine in the Katun range, are located about 5 km away from the front of the Historical moraine opposite the mouth of the Upper Turaouk—the left tributary of the Akkol river. To estimate the magnitude of the Sofiysky glacier advance during the cold period 4.9–4.2 ka BP, which we previously associated with the Akkem stage [9,17], buried soils, peats, and wood fragments in sediments of landslides and river terraces were dated.

Downstream, the Historic moraine contacts landslide bodies, which are composed of unrounded large (up to 4–5 m) blocks of crystalline quartz–chlorite schists (these rocks compose the valley slopes in this location). Many blocks preserved the glacial striation and polish—the effect of the Pleistocene glaciers. Voids between the blocks are often not filled with fine earth. Most likely, the conglomeration of debris in front of the Historical moraine is a result of one large landslide. When a fragment of the left slope collapsed, landslide mass disintegrated into a number of bodies that blocked the valley almost completely (the width of the valley does not exceed 300–350 m in this location). The petrographic composition, which is different from the rocks of the axial part of the South Chuya range, the lack of rounded fragments, and absence of substrate between them, do not allow considering these landforms as moraines (as, for example, was supposed by Egli et al. [43]. At the same time, the landslide overlaps eroded moraine deposits composed of rock material transported from the axial parts of the range. The difference in the structure of the Historical moraine and the landslide is particularly evident in the place of their direct contact, where the moraine front is sliding on the nearest landslide rampart.

The surface of the landslide is almost completely covered by glaciofluvial micaceous sands with parallel bedding (inclined downstream of the valley). Non-rounded rock fragments are evidence of a quick accumulation of sands. Later, these glaciofluvial sands were affected by pedogenesis. In the highest part of the landslide, single boulders rise above the flattened turf surface. Larches settled the landslide are the oldest in the upper reaches of the Akkol valley: the diameter of the tree trunks on the eastern leeward side of the hills reaches 40–50 cm. Highly raised paleo-incisions, now dry, have been preserved at the contact between the Historical moraine and the closest landslide body, as well as between two other bodies almost in the center of the valley. Landslide bodies were not affected by a glacier.

Estimating the landslide age allows to post-date the last Sofiysky glacier advance, which was more distant than expansion during the Historical stage (Figure 11). For this purpose, we have tested soil-sedimentation series on three landslide bodies and in erosional paleo-incision between the bodies (Sections 5–10). The oldest radiocarbon date (5130 ± 305 cal BP, SOAN-9894, Table 3) was obtained for the soil buried under the sandy glaciofluvial deposits. This soil (light, dark brown, highly humified peaty loam, about 15 cm thick, Section 6) formed on the slope of the landslide body and was fragmentarily preserved at the base of a single block of quartz–chlorite shales in the shadow of the meltwater stream. Taking into account the long-term formation of the soil and probable time gap between the triggering of the landslide and the beginning of pedogenesis, this date indicates a time of slope failure earlier than 5 cal kBP.

At about 3.6–3 cal kBP, larches grew in this location: fragments of the bark of Larix sibirica (ca. 3.6 cal kBP, IGAN 6009, Table 3) were found in the voids between the blocks in the landslide at the base of the soil-sedimentation series (Section 9); at the top of the buried soil horizon in Section 6, there is a layer with numerous fragments of the bark and wood of Larix sibirica (ca. 3 cal kBP SOAN-9908, Table 3). The death of trees could have been caused by abrupt watering, probably associated with active ice melting in the Sofiysky glacier and/or the possible draining of periglacial paleolakes at about 3 cal kBP: in both sections (6 and 9), wood fragments were buried by micaceous fluvial sands. The OSL date 2.93 ± 0.16 ka (GdTL-3905, Table 4, Figure 11, Section 7) of sands filling the paleochannel between the two landslides at a depth of 2.1 m confirms this conclusion.

Pedogenesis on the glaciofluvial deposits above the landslide body occurred in several stages and recorded the attenuation of the intense runoff. Sections 5 and 7–10 contain up to three buried soil horizons separated by sandy interlayers. Based on the radiocarbon dates of buried soils—SOAN-9903, 9910 (Section 5, Table 3), SOAN-9893 (Section 7, Table 3), SOAN-9895, 9896, 9897 (Section 8, Table 3), IGAN 6248, 6247 (Section 9, Table 3), SOAN-9898 (Section 10, Table 3), the following stages of soil formation can be established: about ~3 cal kBP, 1.6–1.5 cal kBP, ~1.0 cal kBP, and 0.6–0.8 cal kBP. Soil formation was interrupted by impulses of glaciofluvial activity and aeolian reworking of the sands. Pedogenesis at about 3.23.0 cal kBP was accompanied by the growth of tree vegetation. Like trees, the soil has been affected by fluvial processes. As a result, it was only fragmentarily preserved, mainly in the highest geomorphic position, on the top of one of the hills (Section 8).

The dating of fossil and contemporary soils at the top of the fluvial deposits reflects the presence of differently aged organic matter (OM) in the same soil horizon. This is a result of highly dissected topography, intensive erosion and repeated redeposition of sediments, the open character of soil systems, thinness of soil horizons, and relatively short duration of their formation, and, to some extent, it depends on the dating technique. Thus, within the same landslide body with a size of only 200 × 150 m, significantly different dates were obtained for soils developed on glaciofluvial sands—both for the northern and southern slopes of this landform (Sections 9 and 10) and for the same soil horizon. The beginning of sand accumulation was recorded by the LSC date 3600 ± 220 cal BP IGAN 6009 of the Larix sibirica bark at the base of the soil-sedimentation series (northern slope of the landslide body, Section 9). The LSC dates on humic acids for two buried soils at the top of this section agree with this: 1075 ± 160 IGAN 6248 (22–33 cm) and 225 ± 60 IGAN 6247 (13–15 cm). Earlier, ¹⁴C AMS dates (2σ) 5393 ± 76 cal BP (UZN 6104) and 3516 ± 63 cal BP (UZN 6103) for 15–30 and 0–15 cm, respectively, were presented for surface soil on the southern slope of the landslide body [43]. Dates were obtained on stable OM. Like in Section 9, soil in this location was formed on fluvial sands. At the same time, it includes fragments of gneisses—the eroded moraine substrate that composes the Akkol floodplain, buried by a landslide. In this location (Section 10), we have conducted repeated dating of buried soil at a depth of 23–30 cm, which is separated from the surface soil by a thin sandy layer. Two dates are ¹⁴C LSC 2210 ± 255 cal BP (SOAN-9898, made on humic acids) and ¹⁴C AMS 385 ± 75 cal BP (IGAN_AMS 8771, made on total organic carbon). It is possible that ancient OM (revealed in Section 6 on the opposite side of the landslide body) indicates the similar age (5393 ± 76 cal BP UZN 6104) of the beginning of this soil formation. The near-modern age 385 ± 75 cal BP IGAN_AMS 8771 is most likely a result of young OM influx into the open soil system during pedogenesis.

Downstream the landslide, the valley bottom represents a floodplain for practically its entire width. Due to increasing water runoff (observed during the last 20 years), the bottom sediments have been eroded and thermal abrasion affected coastal terraces, where icy permafrost is widely distributed. As a result, the high floodplain and the first above-floodplain river terrace are only partly preserved in the middle part of the valley. There are fragments of terraces about 3–5 m high at the foot of the slopes (at the base of large talus fans and between the rocky protrusions polished by the glacier and partly covered by the Pleistocene moraine deposits).

The structure of such a terrace (2355 m a.s.l. Section 11) 1.2 km away from the front of the Historical moraine was studied to determine the magnitude and chronology of the last major advance of the Sofiysky glacier. Terrace formation started at about 5.1–5.9 cal kBP (SOAN-9685, 9686, Table 3, Figure 11). Material for dating was collected from a lens of humified loam with a large amount of root detritus on the top of boulder gravel (washed moraine), at the base of the section (depth of 90–110 cm), as well as from a lens of humified material with inclusions of charcoal at a depth of 55–65 cm. Like the landslide, the boulder gravel is covered by fluvial deposits—thinly interbedded sandy loams and clay interlayers with cryotextures. An increase in water runoff and a rise in the water level in the middle part of the Akkol valley occurred between 5.9 and 5.1 cal kBP. At the top of the section, there are two buried soil horizons. The age of the lower one (SOAN-9684, Table 3) is evidence for the synchronicity of soil formation within the landslide and the terrace in the period between the Historical and Aktru stages, 1.6–1.4 cal kBP.

The dating of landslides and a river terrace indicate that the advance of the Sofiysky glacier, which exceeded its expansion during the Historical stage, took place much earlier than 6–5 cal kBP. A fragment of charred larch trunk was found only 1.7 km upstream from the supposed Akkem moraine (Section 16). Its radiocarbon age of about 3.9 cal kBP (IGAN 5966, Table 3) also supports the conclusion about the impossibility of significant glacier advances at 4.9–4.2 cal kBP. At that time, the Sofiysky glacier could advance no further downstream the valley than the front of the Historical moraine.

Dating results for peats from the river terraces and the upper floodplains downstream the valley (down to the Upper Turaoyuk mouth, Sections 12–15 and 17) indicate that meltwater runoff from glacier controlled sedimentation patterns in this part of the valley during the last 1700 years. Since that time (after the Historical stage), peats have been accumulated at an altitude of 2355 m a.s.l. (Sections 12–14, IGAN 6097, 6098, 6099, 6100, SOAN 9687, Table 3).

Downstream the valley at 2335 m a.s.l. (Section 15), peat accumulation, which continued intermittently from 1360 to 390 BP, was interrupted by floods. The change in plant communities in the organo-mineral layers indicates a gradual decrease in moisture up to the moment of a sharp increase in meltwater runoff from the Sofiysky glacier about 400 years ago. Lowland sedge bog with birch existed about 1360 years ago (SOAN-9905, Table 3). It was replaced by a lowland herbaceous (sedge) bog about 1240 years ago (SOAN-9906, Table 3), and about 400 years ago (SOAN-9907, Table 3), it was transformed into a grass meadow. As a result of the floods, the organo-mineral deposits were covered by sands with an apparent thickness of at least 1.5 m.

Meltwater runoff from glaciers located in the valleys of the Akkol left tributaries affected the hydrological system downstream the mouth of the Upper Turaoyuk. This influence became especially pronounced after the draining of the moraine-dammed lake, the relics of which are the modern lakes Akkul and Karakul. This moraine-dammed lake existed for a long time. It was formed no later than 3.2 ka ago [65]. Changes in the water supply near the mouth of the Upper Turaoyuk are recorded by the layers of an organo-mineral substrate with herbal remains (Section 17), including the date 780 cal BP (SOAN-9899, Table 3).

5. Discussion

In contrast to the traditional concept of permanent degradation of the Late Pleistocene glaciation, various proxy data indicate significant warming and deglaciation at the end of the Late Pleistocene–Early Holocene (Figure 13). Thus, by 15–14 cal kBP, glaciers within the mountain framing of the Chuya and Kurai basins significantly degraded, and tree vegetation settled at the foot of the North Chuya Range at 1750 m a.s.l. [66], as well as the Chikhachev Range at 2500 m a.s.l. [67]. Glaciers within the eastern periphery of the Chuya basin at that time completely disappeared or retreated above 2500 m a.s.l. [59]. The Late Dryas cooling of about 13–12 ka BP is reported for neighboring areas of the SW Tuva [68], but it is not clear what moraines in the trough valleys of the SE Altai could be associated with the possible glacier advances at that time. Nevertheless, it could be stated that forest vegetation still existed at that time at the foot of the North Chuya Range.

Numerous Early Holocene radiocarbon ages of tree fragments discovered at the heads of trough valleys indicate prolonged and steady climatic improvement favorable for the forest growth in modern glaciated areas. Tree vegetation settled the Mongun-Taiga massif, Katun and Chikhachev ranges above today’s upper timber limit of 12–10 cal kBP, and by 7 cal kBP, even alpine glaciers within the axial parts of the highest ridges of the Russian Altai either completely degraded or did not extend beyond their modern size [9,23,28,56]. These facts persuasively show that modern climate warming, generally associated with the anthropogenic factor, is not the first nor maximum in the postglacial period.

Finds of a thermophilic Siberian pine ca. 11.3 ka old within the axial part of the Katun range 500 m above the modern upper timber limit do not confirm strong cooling at the beginning of the Holocene (this paper and [41]. Based on analysis of δ¹⁸O record of an ice core from the Belukha peak, Aizen et al. [72] reconstructed an extreme decrease in air temperature for the period 13.0–9.1 ka. Our data suggest a significantly shorter period of possible climate deterioration—the time gap in lifespan of paleotrees falls within the time interval of about 11.3–10 cal kBP (Figure 3).

It should be noted that paleotrees discovered in a modern glaciated zone or beyond the upper timber limit are the most valuable sources of paleoenvironment information. Because direct dating of moraines is an extremely difficult task, the ages of such trees give chronological benchmarks of moraine formation associated with climatically driven glaciers advances. Reconstructions presented in this paper are based on the hypothesis that the date of a tree’s death by itself indicates favorable conditions for living before this date [9]. This approach is more substantiated than the idea that such a date indicates the death of a tree from cooling or from burial by the advancing glacier [21]. Indeed, at the head of trough valleys, besides glacier activity, other widespread, nature processes can cause the tree’s death, including rock falls, landslides, debris flows, snow avalanches, flooding of floodplains, riverbank erosion, thermokarst, strong winds, etc., which could be observed during a period of rapid glacier degradation. During the periods of glaciers activity, trees that died and were buried in the interstage periods can be redeposited in moraines along with trees killed by advanced glaciers. This is confirmed by the broad range of radiocarbon dates of tree fragments that were washed out from the glaciers. Thus, radiocarbon dates of paleotrees are evidence for favorable conditions for their living before this date rather than for adverse conditions leading to their death.

Pollen records, in general, indicate increased temperature and humidity (in comparison with the modern ones) within the study area in the Early Holocene. This resulted in the spreading of coniferous forests to postglacial open landscapes within the higher peripheral parts of intermountain depressions [66,69,70,73,74]. Tree vegetation widely covered the northern side of the Chikhachev Range up to 2400 m a.s.l. since 10 cal kBP [70]. Probably, at that time, mountain forests formed a tree belt along the slopes of the Chikhachev and Sailugem ranges, and the Ukok high-mountain plateau united with the forests in northwestern Mongolia, where they were widespread until the middle of the Holocene [74]. The fossil forest soils of about 11–8 cal kBP and Larix sibirica charcoal fragments of about 9–7.8 cal kBP discovered within the eastern mountain periphery of the Chuya Basin also indicate a prolonged period of forestation of the now treeless landscapes [45,59].

Pollen proxies also show a trend of general cooling and aridization (especially in the southeastern part) of the Russian Altai during the second half of the Holocene. Pronounced cooling at about 5.3 cal kBP affected taiga vegetation, but its final disappearance is attributed to about 3.4 cal kBP [70]. The major expansion of Pinus sibirica forests on the northern side of the North Chuya Range was reconstructed at about 6.5 cal ka BP with the further increasing and prevailing of Larix sibirica as a result of increasing continentality and cooling [66]. Blyakharchuk et al. [70] also report an abrupt decrease in forests in the neighboring areas of Tuva after 6 cal kBP in response to climate cooling and drying. There is also an increase in the steppe elements along with alpine herbs in the region, especially after 2 cal kBP, as a result of progressive aridity intensification. Geographically, pollen data suggest that in the second half of the Holocene taiga forests occupy northern and northwestern parts of the Russian Altai and are less distributed in its eastern and southeastern parts, as well as in the neighboring areas of Tuva and Mongolia.

Investigation of the soil-sedimentary sequences revealed several stages of pedogenesis in the region [45]. The oldest soils have ages of 11.5–6.4 cal kBP. The soil-sedimentary sequences are distributed in a wide altitude range up to 2475 m a.s.l. (the Boguty valley, Chikhachev range). Fossil soils indicate warm and humid climate conditions in the first half of the Holocene, which changed to a cold, humid phase in the mountain tundra and/or alpine pedogenesis during the last 4 ka. This general trend included smaller climate fluctuations, which are recorded in pedo-sedimentary archives. During interstadial warmings, paleosoils formed within vast areas in the basins and even in heads of troughs. It should be noted that within the area of investigation, all studied surface soils develop under cold, ultra-continental, and water-deficit conditions, which are the most severe during the last 2 ka [44,45]. In contrast to modern soils, all buried soils formed in wetter conditions. The presence of cryoturbation features indicates a higher water availability rather than a more severe temperature regime [45]. Better water supply in the region during the first part of the Holocene is also indicated by pollen [66,70], chironomid [68], and diatom [71] proxy data.

Generally, a more humid and warm climate dominated the Russian Altai in the first half of the Holocene, when alpine glaciers significantly retreated or even completely disappeared. The climate in the second half of the Holocene shows trends of cooling and aridity intensification. It controlled the presence of alpine glaciers in the axial parts of the highest ranges and their repeated advances.

The presented results indicate three periods of climate deterioration in the second half of the Holocene. These periods are correlated with the gaps in the lifetime of paleotrees discovered in the upper reaches of trough valleys. At the same time, the previously assumed extensive advance of glaciers in the middle of the Holocene was not confirmed. The ages of buried soils in the upper reaches of the Akkol valley indicate that during the period of climate deterioration between 5 and 4 ka ago, the Sofiysky glacier advanced no further downstream the valley than the front of the Historical moraine. This conclusion is consistent with the finds of paleotrees of a similar age in the upper reaches of the Akkol valley and in some other trough valleys of the North Chuya range.

Were there any climatic prerequisites for the advance of glaciers in the middle of the Holocene at all? Analysis of the ice cores of a flat-topped glacier in the Tsambagarav massif revealed that glaciation in the eastern part of the Mongolian Altai began about 6 ka ago [75]. Traces of glaciers advances about 5 ka ago were also identified in the Mongun-Taiga massif [31]. Interruption in soil formation at the edge of the fifth and fourth millennia was also established for the Mogen-Buren depression (adjusted areas of the SW Tuva) and in the Chuya valley between the Chuya and Kurai basins (SE Altai). Thus, climate deterioration in the region in the middle of the Holocene is recorded in various natural archives. Nevertheless, according to new data, the duration and intensity of these changes were less significant than previously assumed [9,17].

6. Conclusions

Direct dating of glacial landforms is among the most comprehensive tasks and is often associated with fortunate finds suitable for dating material. Glaciers retreat and ice melting in moraines provide an opportunity to collect samples in a modern glaciated zone. In recent decades, numerous tree fragments have been collected within the trough slopes, glaciers forefields, and in moraines in the upper reaches of glacial valleys of the Russian Altai. Samples collected beyond the modern upper forest line and on the surface of moraines closest to the glaciers are of particular importance. In the first case, the age of paleotrees allows distinguishing stable periods of warming with temperatures higher than modern ones. In the second, additionally, such finds can post-date glacier advance and particular moraine formation associated with climate deterioration.

The analysis of all available dates for samples, collected in the upper reaches of trough valleys presented in this paper and published earlier, allows us to draw several important conclusions.

1. There was significant degradation of the last Pleistocene glaciation (Sartan—according to the stratigraphic scheme of Western Siberia, or MIS-2 according to the SPECMAP) within the glacial centers of the Russian Altai by the beginning of the Holocene. About 11.3–11.4 cal kBP, not only Larix sibirica but also more thermophilic Pinus sibirica settled in the Katun range—the highest one within the Altai Mountains at altitudes of 2600–2700 m a.s.l. This is 100–200 m above the modern upper timber limit and 300–500 m above the modern upper forest line. It indicates significantly more warm and humid climatic conditions. The presence of forest vegetation above the modern upper timber limit about 9.3–10 cal kBP is recorded in three ranges—the Katun and Chikhachev ranges as well as the Mongun-Taiga massif. Today, all of them are completely forestless. In the Chikhachev range, the forest grew up to 7.4 cal kBP. Small forests even occupied the modern glaciation zone in some ranges, the SE Altai. All these data refute the traditional concept of the Russian Altai Holocene glaciations as a consecutive retreat of the Late Pleistocene glaciation [5,8]. Considerable and prolonged warming in the Early Holocene was stronger than the modern one. It caused a significant shrinking of glaciers at the head of trough valleys of the Russian Altai. They were smaller than modern, rapidly retreated glaciers, or were even completely degraded. Forest vegetation settled previously glaciated areas. It should be noted that such deep Early Holocene warming, in contrast to the modern one, was not affected by anthropogenic influence. It was exclusively caused by natural factors.

2. New radiocarbon ages of fossil soils discovered downstream the Historical moraine limited glaciers’ expansion in the Middle Holocene by the size of the Historical moraine. Dating of landforms without traces of glacial influence indicates that the advance of the Sofiysky glacier, which exceeded its expansion during the Historical stage, took place much earlier than 6–5 cal kBP. Thus, the accumulation of the Akkem moraine could have occurred at the end of the Late Pleistocene.

3. Despite the huge number of paleowood fragments found in the upper parts of the trough valleys of the Russian Altai, only one date (IGAN 3694) completely falls into the time interval 4.2–4.9 cal kBP. Earlier, we have associated this gap in the paleotrees’ lifespan with the Middle Holocene cooling and glaciers’ advance during the Akkem stage [9,17]. Nevertheless, presently, five more dates (SOAN-6923, 7383, 8748, 9637, 9640) partly enter this interval with their 95% confidence interval (2σ calibration). Thus, a possible hiatus in reforestation (only a few centuries) within the time interval of 4.2–4.9 cal kBP is not enough for such distant glacier advance (about 5–6 km far from modern glaciers) and the formation of the Akkem moraine. At the same time, no finding of paleotrees, the age of which would fall within the Historical stage, has been made. This fact indicates a lesser glacial activity between 5 and 4 cal kBP in comparison with the Historical stage, when moraines much closer to modern glaciers were formed.

4. In general, available data suggest that the climate of the Russian Altai in the first half of the Holocene was warmer and more humid in comparison with the modern one. The climate of the SE Altai in the second half of the Holocene shows trends of cooling accompanied by aridity intensification, which became especially pronounced during the last 2–1.5 ka. It controlled the presence of alpine glaciers in the axial parts of the highest ranges and their repeated advances. The fact that the thermal minimum in the middle of the nineteenth century did not positively influence the mass balance of the Altai glaciers supports this conclusion.