Introduction
Northeast Siberia is characterized by some of the most extreme continental climate found in the Northern Hemisphere. The glacial history of high northern latitudes generally follows the global climate trends recorded in the benthic marine δ18O record (Lisiecki and Raymo, Reference Lisiecki and Raymo2005; Stokes et al., Reference Stokes, Tarasov and Dyke2012; Simms et al., Reference Simms, Lisiecki, Gebbie, Whitehouse and Clark2019). However, interior continental settings often display anomalies in this regard, with local glacial expansion asynchronous with global ice-volume trends (Svendsen et al., Reference Svendsen, Alexanderson, Astakhov, Demidov, Dowdeswell, Funder and Gataullin2004; Ward et al., Reference Ward, Bond and Gosse2007; Batchelor et al., Reference Batchelor, Margold, Krapp, Murton, Dalton, Gibbard, Stokes, Murton and Manica2019), so it is possible that the cold, landlocked climate of northeast Siberia creates an environment in which glaciation may not follow global patterns. While most Northern Hemisphere ice masses reached their overall maximum extent over the last glacial cycle during the last glacial maximum (LGM) 26–19 ka ago (Svendsen et al., Reference Svendsen, Alexanderson, Astakhov, Demidov, Dowdeswell, Funder and Gataullin2004; Barr and Clark, Reference Barr and Clark2012a; Batchelor et al., Reference Batchelor, Margold, Krapp, Murton, Dalton, Gibbard, Stokes, Murton and Manica2019; Simms et al., Reference Simms, Lisiecki, Gebbie, Whitehouse and Clark2019; Dalton et al., Reference Dalton, Stokes and Batchelor2022a, Halsted et al., Reference Halsted, Bierman, Shakun, Davis, Corbett, Drebber and Ridge2024), glaciers in northeast Siberia may have reached their maximum well before the global LGM.
The general consensus is that LGM ice formation in northeast Siberia was limited to alpine-style glaciers and mountain ice fields (Gualtieri et al., Reference Gualtieri, Glushkova and Brigham-Grette2000; Glushkova, Reference Glushkova2001; Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003; Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023), with some regions, such as the Verkhoyansk Mountains (Fig. 1), remaining more or less ice free during the LGM (Stauch and Gualtieri, Reference Stauch and Gualtieri2008; Zech et al., Reference Zech, Zech, Zech, Leiber, Dippold, Frechen, Bussert and Andreev2011; Barr and Clark, Reference Barr and Clark2012a). While the maximum ice extent during the last glacial cycle in northeast Siberia is reported to have occurred early (pre-global LGM), an exact timing has not been established for the whole region and may possibly have varied within it (Gualtieri et al., Reference Gualtieri, Glushkova and Brigham-Grette2000; Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003; Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023) due to orographic and climatic variations across the vast mountainous region (Glushkova, Reference Glushkova2001). The glacial history could have been controlled by many local factors such as the region’s cold, continental climate, complex topography, and/or the frigid, periodically ice-covered Arctic Ocean to the north influencing moisture availability. However, previous work has also speculated that the limited extent of glaciation at the time of the global LGM is primarily a result of reduction in moisture supply connected to the growth of the western part of the Eurasian Ice Sheet blocking circulation of the westerlies (Krinner et al., Reference Krinner, Diekmann, Colleoni and Stauch2011; Zech et al., Reference Zech, Zech, Zech, Leiber, Dippold, Frechen, Bussert and Andreev2011; Barr and Clark, Reference Barr and Clark2012a). Going further back in time, several studies have shown that the ice extent in northeast Siberia during Marine Isotope Stage (MIS) 6 surpassed any of the glaciations that occurred during the last glacial cycle (Stauch et al., Reference Stauch, Lehmkuhl and Frechen2007; Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023). Moraines of the penultimate glacial maximum remain the oldest dated landforms in northeast Siberia so far, but studies have reported geomorphological evidence of undated glacial landforms that could belong to older, even more extensive glaciations (Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003; Stauch and Gualtieri, Reference Stauch and Gualtieri2008; Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023), such as the mid-Pleistocene superglaciations tentatively associated with MIS 16 or 12 (Batchelor et al., Reference Batchelor, Margold, Krapp, Murton, Dalton, Gibbard, Stokes, Murton and Manica2019).
Figure 1. Overview map of the northeast Siberian region. Localities from studies mentioned in the text are marked with red dots (Gualtieri et al., Reference Gualtieri, Glushkova and Brigham-Grette2000; Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003; Melles et al., Reference Melles, Brigham-Grette, Glushkova, Minyuk, Nowaczyk and Hubberten2007; Stauch and Gualtieri, Reference Stauch and Gualtieri2008; Zech et al., Reference Zech, Zech, Zech, Leiber, Dippold, Frechen, Bussert and Andreev2011; Diekmann et al., Reference Diekmann, Pestryakova, Nazarova, Subetto, Tarasov, Stauch and Thiemann2017; Baumer et al., Reference Baumer, Wagner, Meyer, Leicher, Lenz, Fedorov, Pestryakova and Melles2020; Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023). Major mountain chains, rivers, and settlements are also labeled. The bottom right insert is adapted from Ehlers and Gibbard (Reference Ehlers and Gibbard2007) and shows the Northern Hemisphere ice extent at the LGM, including the Laurentide Ice Sheet (LIS) and the Eurasian Ice Sheet (EIS). The digital elevation map is based on “General Bathymetric Chart of the Oceans” (https://gebco.net).
It remains difficult to ascertain clear spatial and temporal patterns in the glacial history of northeast Siberia with only a handful of chronological studies to rely on in an extensive mountain landscape spanning over 1 million km2 of formerly glaciated terrain (Barr and Clark, Reference Barr and Clark2012b). Thus, a more extensive coverage of age constraints is needed to improve our understanding of spatial variations in this region’s complex glacial history. Specifically, the vast mountainous landscape at the heart of the region, which includes the Chersky Range and the Suntar-Khayata Mountains, remains understudied and represents a large knowledge gap between the eastern and western parts of the region (Fig. 1). Here, we present cosmogenic exposure age results from a valley in the Tas-Kystabyt Range of the Chersky Mountains close to the settlement of Ust-Nera and far from any existing studies of glacial chronology, revealing a so-far undocumented deglaciation around 46 ka ago in the area.
Study area and fieldwork
The Yuryuyete River valley
In the field, we investigated the U-shaped valley of the Ambar-Yuryuyete (Амбар-Юрюете) River that lies within the Tas-Kystabyt Range of the Chersky Mountains. It is hereafter referred to simply as the Ambar-Yuryuyete Valley. It is located at approximately 64.5°N, 143.3°E by the confluence of the large Indigirka River and the smaller Nera River (Fig. 1). The present climate of the nearby settlement Ust-Nera (Fig. 2A), located approximately 6 km away and 500 m lower than the sampling site, is mild during summer, with daily temperatures averaging ∼12°C. Winters are very cold and dry, with mean temperatures in January as low as −44°C (https://www.timeanddate.com/weather/russia/ust-nera/climate, accessed 25 March 2024). Subzero temperatures persist from mid-September to early May. Precipitation outside the main summer period (mid-May to mid-September) is low and totals ∼50 mm/yr based on weather reports from 2005 to 2015 (Sakha, Russia, 64.57°N, 143.24°E, 492 m above sea level [m asl]) making for a light snow cover (<1 m) typically lasting until early June. The surrounding mountains reach elevations up to 2000 m, while the elevation of the valley floor and the top of the valley headwall are approximately 650 and 1400 m asl, respectively. The valley faces north, with the ridges on both sides of the valley striking north-south. The elevation gain from the bottom of the valley to the tor-studded ridge crests is approximately 200 m. The valley is sparsely covered in coniferous trees, bushes, and scrub, with thinning vegetation cover toward the upper reaches of the valley. Within the valley is a lateral end moraine and a well-developed cirque with steep headwalls in the uppermost part of the valley (Fig. 2B).
Figure 2. (A) Overview map of geomorphological features in the Tas-Kystabyt Range. Black lines mark the valley distances from toe to headwall and numbers in black are the corresponding equilibrium line altitude (ELA) estimates. (B) Close-up of the geomorphology of the valley at Ust-Nera. Black dots show the locations of the sampled boulders, red dot shows the location of the sampled bedrock. Numbers indicate exposure ages (in ka). The elevation model is ArcticDEM (Porter et al., Reference Porter, Morin, Howat, Noh, Bates, Peterman and Keesey2018).
Field conditions and sample collection
Fieldwork in the Tas-Kystabyt Range in August 2021 was limited to the Ambar-Yuryuyete Valley due to logistical reasons. We traversed the Ambar-Yuryuyete Valley on foot in search of boulders and bedrock outcrops suitable for exposure dating. Most of the terrain was overgrown with vegetation, and the moraine was not easily distinguishable in the field, except for scattered moraine boulders up to a few meters in height. The boulders and bedrock exposed in the valley, as well as the tors on the ridge crests, all had a granitic composition, suggesting that the boulders were derived from local bedrock. In total, we collected seven samples: (1) five samples from boulders on the moraine in the valley center and (2) two samples from bedrock outcrops on the western side of the valley ∼10 m above the valley base where the Ambar-Yuryuyete River runs.
Methods
Geomorphological mapping and snow-line approximation
We mapped the glacial geomorphology of the Tas-Kystabyt Range using hill-shaded imagery derived from the ArcticDEM v. 3 mosaic with 2 m spatial resolution (Porter et al., Reference Porter, Morin, Howat, Noh, Bates, Peterman and Keesey2018) in the QGIS Geographic Information System (Fig. 2). Moraines were the principal landform of interest. Other mapped landforms include meltwater channels and hummocky terrain. Landforms that were identified with some uncertainty were included in the mapping category “speculative” (moraines or meltwater channels). For approximation of the equilibrium line altitude (ELA), we used toe-to-headwall altitude ratios (Benn and Lemkuhl, Reference Benn and Lehmkuhl2000) of 0.4 and assumed the ice-surface altitude to be 50 m below the headwall crests. Both altitudes were derived from the ArcticDEM maps once the features had been mapped.
Cosmogenic nuclide measurements
Samples were collected for cosmogenic nuclide exposure dating using an angle grinder, hammer, and chisel, targeting the uppermost 2–3 cm of the rock surface. Quartz separation was conducted at the Aarhus University Cosmogenic Nuclide Laboratory, following standard procedures (Kohl and Nishiizumi, Reference Kohl and Nishiizumi1992). We verified quartz purity by inductively coupled plasma optical emission spectrometry (ICP-OES) and performed additional acid etching on a subset of the samples as needed to reduce the concentration of cations, including Al, Ca, Fe, K, Mg, Na, and Ti.
We extracted beryllium (n = 7) and aluminum (n = 7) from the samples in the NSF/UVM Community Cosmogenic Facility using methods described in Corbett et al. (Reference Corbett, Bierman and Rood2016) and 20.64–21.47 g of quartz per sample. The batch included two full processing blanks and a reference material aliquot (UVM-A; Corbett et al., Reference Corbett, Bierman, Caffee, Hidy and Woodruff2024). We spiked the samples with ∼250 µg of 9Be using a carrier termed UVM-SPEX, created from a dilution of SPEX 10,000 ppm Be standard, with a resulting Be concentration of 304 µg/mL (Supplementary Table S1). We spiked samples with Al as needed to ensure at least 1500 µg of total Al, based on the native Al contained in the quartz, using a SPEX commercial Al standard with a concentration of 1000 ppm.
We quantified total Al in the samples (Supplementary Table S2) by ICP-OES immediately following sample digestion. Following standard procedures (Corbett et al., Reference Corbett, Bierman and Rood2016), we removed replicate aliquots from the samples by mass (representing ∼2% and ∼4% of the sample, respectively), added 25 μL H2SO4 to each, evaporated the HF, then diluted the residual H2SO4 droplets by mass with a 0.25% H2SO4 solution containing 10 ppm of yttrium to serve as an internal standard.
Accelerator mass spectrometry analysis (AMS) was performed at the Aarhus AMS Centre, Department of Physics and Astronomy, Aarhus University. For 10Be/9Be, measured ratios were normalized to primary standard ICN-01-5-4 with a 10Be/Be ratio of 2.85 × 10−12 (Nishiizumi et al., Reference Nishiizumi, Imamura, Caffee, Southon, Finkel and McAninch2007). For 26Al/27Al, analyses were normalized to primary standard ICN-01-5-2 with a 26Al/Al ratio of 1.82 × 10−12 (Nishiizumi, Reference Nishiizumi2004). We corrected samples for backgrounds using the average and standard deviation of blanks associated with the samples (4.65 ± 1.26) × 10−15 for 10Be/9Be and (2.05 ± 4.64) × 10−15 for 26Al/27Al derived from a single blank (Supplementary Table S3). We propagated the blank uncertainties in quadrature. For 10Be/9Be, background-corrected sample ratios range from 5.96 × 10−13 to 1.40 × 10−12, with the blank correction representing 0.3–0.8% of the sample ratios. For 26Al/27Al, background-corrected sample ratios range from 5.53 × 10−13 to 1.39 × 10−12, with the blank correction representing 0.1–0.4% of the sample ratios.
Exposure ages and outlier detection were calculated using the online calculators formerly known as the CRONUS-Earth online calculators (Balco et al., Reference Balco, Stone, Lifton and Dunai2008), applying the LSDn time-dependent scaling scheme of Lifton et al. (Reference Lifton, Sato and Dunai2014) and global production calibration (Borchers et al., Reference Borchers, Marrero, Balco, Caffee, Goehring, Lifton, Nishiizumi, Phillips, Schaefer and Stone2016). In these calculations, we have assumed a sample density of 2.65 g/cm3 typical of granitic rocks. Topographic shielding and snow shielding are calculated following the methods outlined in Gosse and Phillips (Reference Gosse and Phillips2001).
Results
Remote sensing of glacial landforms
Geomorphological mapping based on satellite imagery (ArcticDEM v. 3) reveals sizable (kilometer-scale) moraines in most main valleys draining the Tas-Kystabyt Range (Fig. 2A). One to two terminal or recessional moraines typically occur upvalley from the terminal moraines associated with the most extensive glaciations. The mapped moraines indicate past glaciation that took the form of an extensive mountain ice field southeast of the Ambar-Yuryuyete Valley, possibly during several distinct periods, with the main outlet glaciers reaching several tens of kilometers from the ice divide to the ice front.
In contrast, the northwestern parts of the range, where the Ambar-Yuryuyete Valley is located, contain little evidence of former glaciation besides isolated and solitary moraines in a few of the valleys (Fig. 2A). In most of the other short, steep-sloping valleys that constitute this part of the range, we did not map any moraines. The Ambar-Yuryuyete Valley, where we collected samples, stands out as one of the few valleys with a moraine in this northernmost part of the Tas-Kystabyt Range. The valley moraine is a 2.5-km-long slope-parallel lateral moraine that transitions into a narrow terminal moraine downstream (250-m-long segment oblique to the thalweg), tracing the snout of the former valley glacier (Fig. 2B). Further upvalley, an area of hummocky ground occurs at the foot of the cirque headwall that might relate to the former presence of a small cirque glacier or, alternatively, represents a small rock glacier generated from debris falling off the steep headwall. Consequently, there is no clear geomorphological evidence for more limited glacial advances younger than the solitary moraine we sampled, nor is there evidence for larger glaciations lower down the Ambar-Yuryuyete Valley in either the valley itself or in the Indigirka and Nera valleys beyond the small-valley mouth (Fig. 2A). Shapefiles of all mapped features can be found in the Supplementary Material (filename: Supp.Fig.2_shp).
Reconstruction of paleo-ELAs (12 in total) was carried out where both the toe and headwall could be reliably identified in the valleys and yielded an ELA of 1182 m asl for the Ambar-Yuryuyete Valley and 1210 m asl for the neighboring valley on the east side. In the valleys to the southwest of the range, the many identified moraines are associated with different ELAs that range from 1217 m asl to 1644 m asl. The ELA for the maximum glacial extent in the largest west-draining valley is 1217 m asl, but other mapped moraines farther up the valley yield ELAs of 1417 m asl 33 km upvalley and 1514 m asl 39 km upvalley, respectively. The average ELA for the minimum glacial extent in the western part is 1466 m asl.
Ambar-Yuryuyete Valley exposure ages
Calculated 10Be exposure ages corrected only for topographic shielding range from 41.6 ± 3.1 to 96.5 ± 6.1 ka, while calculated 26Al ages range from 41.0 ± 3.8 to 99.0 ± 9.5 ka, with the two bedrock samples UN-VB-3 and UN-VB-4 providing the youngest ages (Fig. 3C, Table 1). The five boulder samples show considerable variation in calculated ages, with 10Be ages ranging from 45.7 ± 3.4 to 96.5 ± 6.1 ka and 26Al ages ranging from 46.8 ± 4.5 to 99.0 ± 9.5 ka. The boulder ages do not indicate any immediate correlation between sampling location and sample ages (Fig. 2B). For all samples, the 10Be and 26Al ages overlap within 1σ uncertainty regardless of sample type and location. Because the bedrock samples were at the ground surface level, it is more likely that they were continuously covered by the annual snow cover of roughly 50 cm (based on the ∼50 mm/yr winter precipitation) compared with the moraine boulders that each stood >1.5 m above ground (Fig. 3, Table 1). The boulders possibly experienced significantly less snow cover due to the wind remobilizing the freshly fallen snow. Accounting for snow shielding (see Supplementary Material) of the bedrock samples UN-VB-3 and UN-VB-4 yields calculated 10Be ages of 43.3 and 45.2 ka and 26Al ages of 42.7 and 45.6 ka for these two samples, respectively (Table 1).
Figure 3. (A-F) Photos of sampled boulders and bedrock. Superscript a and b denote apparent exposure ages of 26Al and 10Be respectively, which are corrected for topographic shielding but not potential snow shielding.
Table 1. Cosmogenic nuclide data. Coordinates and elevations are measured using handheld GPS and cross-checked against the ArcticDEM elevation model (Porter et al., Reference Porter, Morin, Howat, Noh, Bates, Peterman and Keesey2018).a
a All uncertainties are reported as 1σ external errors.
b Boulder height was estimated as the distance from the ground to the highest point on the boulder.
* An asterisk () denotes boulder height above terrain as opposed to full height estimate.
Discussion
Calculated exposure ages and inherited nuclides
The agreement between the 10Be and 26Al ages suggests that all samples experienced relatively simple exposure histories, which in this context means one or few periods of exposure within recent times and insufficient shielding of the samples to produce significant decay of either nuclides. However, we observe a scatter in the exposure ages among the five boulder samples. Scatter in exposure ages is typically explained by either (1) inheritance of nuclides accumulated in the rocks before moraine formation, rendering the apparent ages of some of the boulder samples too old (Fabel and Harbor, Reference Fabel and Harbor1999; Applegate et al., Reference Applegate, Urban, Laabs, Keller and Alley2010; Heyman et al., Reference Heyman, Stroeven, Harbor and Caffee2011); or (2) postglacial shielding or erosion, which could render some of the exposure ages too young (Fabel and Harbor, Reference Fabel and Harbor1999; Applegate et al., Reference Applegate, Urban, Laabs, Keller and Alley2010; Heyman et al., Reference Heyman, Stroeven, Harbor and Caffee2011). Here, we favor the first option for the following reasons: (1) the skewness of the age distribution, including multiple young samples with overlapping uncertainties and a tail of older ages (Fig. 4); and (2) outlier detection analysis on the sample set suggests the three oldest samples are outliers. Outlier detection was done using the CRONUS-Earth online calculator, which performs a χ2-based pruning of the data (Bevington and Robinson, Reference Bevington and Robinson1992). Hence, the three boulders that likely contain inherited nuclides are UN-VB-1, UN-VB-7, and UN-VB-9, based on their 10Be exposure ages (91.4 ± 5.7, 96.5 ± 6.1, and 65.7 ± 5.4 ka) and 26Al exposure ages (84.4 ± 7.9, 99.0 ± 9.5, and 62.3 ± 5.9 ka), respectively. These three boulders were likely picked up by an advancing ice and emplaced at their current location in such a way that the upward-facing surface we sampled contained a significant concentration of nuclides produced during previous periods of surface exposure. It is not possible to determine whether the inherited concentrations relate to orientation and rock-depth during previous exposure or the duration of that exposure.
Figure 4. Gaussian distributions of the calculated exposure ages with internal errors as standard deviation. UN-VB-3 and UN-VB-4 ages are corrected for 50 cm snow shielding to help illustrate the clustering. Detected outliers have been indicated with dashed lines. The black line is the cumulative age distribution.
In contrast, the two remaining boulders were likely emplaced in such a way that the sampled surfaces contained an insignificant concentration of inherited nuclides from prior exposure. This could be either because these boulders were sourced from considerable depth (requiring deep erosion) or because they were emplaced upside down compared with their previous orientation before glacial transport. The two bedrock samples were found in the low-central part of the valley, where glacial erosion and scouring would have been most efficient, implying that the nuclide inventories of these bedrock samples likely would have been reset during the last glacial event. The fact that two boulders yield 10Be and 26Al ages that confidently overlap the 10Be and 26Al bedrock ages suggests that these four samples belong to the same population and that their nuclide inventories reflect a simple exposure history that constrains the last deglaciation event of the Ambar-Yuryuyete Valley. This presents the simplest scenario that accounts for the nuclide inventories in both bedrock and boulder samples. In any other conceivable scenario, it is unlikely that the ages derived from the two youngest boulder samples would overlap the ages associated with the bedrock samples.
Nevertheless, given the relative spread among the apparent exposure ages, we cannot completely rule out the possibility that deglaciation occurred before MIS 3, perhaps even as early as MIS 6. It is possible that some or all samples were covered by sediments for extensive periods of time after deglaciation, resulting in incomplete exposure after the last deglaciation event. It has been shown in earlier publications that postdepositional shielding is often the most important source of scatter among exposure ages of glacial boulders and that the maximum apparent exposure age should be viewed as a minimum deglaciation age for boulder groups with a wide age scatter (Heyman et al., Reference Heyman, Stroeven, Harbor and Caffee2011). However, we do not think that this is the case for the Ambar-Yuryuyete Valley samples. First, a group of three closely spaced boulders (UN-VB-7, UN-VB-8, and UN-VB-9) of similar height (∼2 m) yield widely different apparent 10Be ages (96.5 ± 6.1, 46.9 ± 3.0, and 65.7 ± 5.4 ka), while two other boulders (UN-VB-1 and UN-VB-2) located some ∼500 m away also yield different 10Be ages (91.4 ± 5.7, and 45.7 ± 3.4 ka). Second, the younger ages in both groups of boulders are consistent with the ages of bedrock samples collected next to the moraine. It is difficult to explain this age pattern with postdepositional shielding, as the shielding would have to vary locally among closely spaced boulders. It is, on the other hand, easier to reason that the apparent exposure age shared by the two bedrock samples and the two youngest boulders reflects a deglaciation event at 45.6 ± 3.4 ka and that the three remaining boulders contain an inherited nuclide inventory. The large inherited nuclide inventory observed in two of the samples (UN-VB-1 and UN-VB-7) likely reflects the cold and exceptionally dry climate in the region, implying that erosion may have been limited because the glaciers were frozen to the ground during large parts of the time. Such patterns have also been observed in boulder and bedrock samples from areas in Greenland occupied by cold-based glaciers (e.g., Søndergaard et al., Reference Søndergaard, Larsen, Steinemann, Olsen, Funder, Egholm and Kjær2020).
Deglaciation timing of the Tas-Kystabyt Range
Disregarding the samples affected by inheritance, we calculate a weighted mean age of 45.6 ± 3.4 ka using both 10Be and 26Al from the two bedrock samples with snow shielding (UN-VB-3 and UN-VB-4) and the two youngest boulders without snow shielding (UN-VB-2 and UN-VB-8). We consider this mean exposure age to be the best possible estimate for the most recent deglaciation of the valley. However, given the low elevation of the weather station (∼500 m asl) relative to the site (998–1095 m asl), the 50 cm snow cover might be seen as a minimum. Excess snow above the 50 cm already considered would push the age estimate toward older ages. As an example, adding another 1 m of snow cover during the nine winter months to all four samples considered yields a calculated mean of 48.6 ± 4.0 ka. Which is still in agreement with our initial estimate within uncertainty. Consequently, we arrive at a deglaciation age for this region that in any case corresponds to the global MIS 3.
Relating the deglaciation age to our regional geomorphological mapping, we suggest that the 45.6 ± 3.4 ka deglaciation event is linked to the outer of the two generations of moraines we were able to identify in several valleys in the southeastern parts of the Tas-Kystabyt Range using remote sensing (Fig. 2). This suggestion is based mainly on two weak clusters of the ELAs one around ∼1200 m and another around ∼1500 m. Given the relatively small extent of the Tas-Kystabyt Range, we find it reasonable to assume synchronous growth and retreat of ice masses across the range. Thus, we suggest that the lower ELAs could be associated with the deglaciation event at 45.6 ± 3.4 ka, and the higher ELAs linked to a younger, smaller deglaciation event that possibly could represent ice retreat after the LGM, although we acknowledge that these end-moraine features typically are indicative of equilibrium positions, while the exposure ages are indicative of ice retreat.
Regional comparison and lack of LGM glaciation
Geomorphological mapping did not reveal any other glacial landscape features in the Ambar-Yuryuyete Valley that could represent significant additional ice advances, either above or below the sampled moraine, other than an area of hummocky terrain immediately below the valley headwall, possibly indicating the existence of a small cirque glacier. The lack of geomorphological evidence for other advances, the preservation of the moraine in this valley, and our deglaciation age of 45.6 ± 3.4 ka thus largely rule out the possibility of an LGM ice advance occurring inside this valley, because even a cold-based, nonerosive reglaciation could be expected to leave behind a moraine from the debris falling off the steep slopes above the glacier and transported to the glacier’s snout. Additionally, this is also supported by our calculated ages showing no indication of prolonged shielding. If an LGM glaciation did occur in the Tas-Kystabyt Range, we suggest that it did not reach the Ambar-Yuryuyete Valley and only covered the southeastern part of the range (Fig. 2). Alternatively, the LGM could have produced the aforementioned small cirque glacier, which would have been a solitary feature at the time.
An absence of evidence for LGM glaciation in the Ambar-Yuryuyete Valley is at odds with global trends but aligns with the current interpretation of the glacial history of the Verkhoyansk Mountains (Fig. 1), where previous work suggests that glaciers were largely absent during the LGM (Stauch and Lehmkuhl, Reference Stauch and Lehmkuhl2010). The latest moraine deposit in the Verkhoyansk Mountains slightly predates the moraine age in the Ambar-Yuryuyete Valley at around 50 ka following optically stimulated luminescence (OSL) dating by Stauch and Lehmkuhl (Reference Stauch and Lehmkuhl2010); however, this is largely similar to our deglaciation timing within error. Comparable findings have also been reported elsewhere in northeast Siberia (Fig. 5). Notably, Gualtieri et al. (Reference Gualtieri, Glushkova and Brigham-Grette2000) concluded from 36Cl measurements on bedrock in the Anadyr Lowlands that the area has been ice free since ∼53 ka, while Brigham-Grette et al. (Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003) dated several erratics in the age range 36 to 69 ka using 36Cl in the Tanyurer River valley, even if the lack of consistency between these ages obscured any more precise age estimate. In Kamchatka, Bäumler and Zech (Reference Bäumler and Zech2000) reported deglaciation concluded before 40 ka, as indicated by a dated tephra overlying glacial till.
Figure 5. A compilation of chronological studies of glacial features in northeast Siberia sorted from west to east (left to right: Gualtieri et al., Reference Gualtieri, Glushkova and Brigham-Grette2000; Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003; Melles et al., Reference Melles, Brigham-Grette, Glushkova, Minyuk, Nowaczyk and Hubberten2007; Stauch and Gualtieri, Reference Stauch and Gualtieri2008; Zech et al., Reference Zech, Zech, Zech, Leiber, Dippold, Frechen, Bussert and Andreev2011; Diekmann et al., Reference Diekmann, Pestryakova, Nazarova, Subetto, Tarasov, Stauch and Thiemann2017; Baumer et al., Reference Baumer, Wagner, Meyer, Leicher, Lenz, Fedorov, Pestryakova and Melles2020; Gowan et al., Reference Gowan, Zhang, Khosravi, Rovere, Stocchi, Hughes, Gyllencreutz, Mangerud, Svendsen and Lohmann2021); Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023). Boxes indicate observations. Brown colors are exposure dates, which are typically indicative of glacial retreat; heights of the boxes indicate uncertainty on measurements/spreads of ages at the site. Green colors are infrared-stimulated luminescence (IRSL) bracketing ages of moraines; heights of the boxes indicate upper and lower limits for the moraine age. Dashed outlines and paler colors indicate greater than analytical uncertainty with regard to timing, that is, lack of numerical constraint or ambiguous ages. Gray colors are climatic conditions; heights of the boxes indicate duration. The last two panels show the estimated ice volumes of the Eurasian and North American ice sheets during the last 80 ka (Gowan et al., Reference Gowan, Zhang, Khosravi, Rovere, Stocchi, Hughes, Gyllencreutz, Mangerud, Svendsen and Lohmann2021).
At all these locations, various other glacial features identified higher up in the mountains were interpreted to indicate a very limited ice extent during the LGM. At Lake Malyk, the study site closest to the Ambar-Yuryuyete Valley, an intermediate moraine ridge, sandwiched between an LGM and an MIS 6 moraine, has been dated with cosmogenic 10Be to MIS 4 or MIS 5, but the possibility that this middle moraine was deposited during MIS 3 cannot be excluded (Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023). Two sediment cores from Lake Billyakh and Lake Emanda of the Verkhoyansk Mountains do not reach as far back as MIS 4, but both cores terminate in what could be glacial deposits toward the end of MIS 4 or the onset of MIS 3 (Diekmann et al., Reference Diekmann, Pestryakova, Nazarova, Subetto, Tarasov, Stauch and Thiemann2017; Baumer et al., Reference Baumer, Wagner, Meyer, Leicher, Lenz, Fedorov, Pestryakova and Melles2020). We caution, though, that these two timings (51 ka and 57 ka) are based on radiocarbon ages very close to the limits of the method; hence, they could very well represent minimum ages.
It is possible that many of the abovementioned observations reflect the same glaciation period of northeast Siberia, with variable timings potentially being explained by the different dating methods applied (and their inherent uncertainties) as well as variations in local climatic and orographic conditions. On the other hand, the single 2.5 km lateral end moraine occupying the Ambar-Yuryuyete Valley is not a typical feature compared with the rest of the Tas-Kystabyt Range, leaving it unclear whether our own age represents a local oddity.
Global evidence of MIS 3 glaciations
A maximum ice extent within the last glacial cycle during MIS 3, as indicated by our results, is uncommon in a global context (Batchelor et al., Reference Batchelor, Margold, Krapp, Murton, Dalton, Gibbard, Stokes, Murton and Manica2019; Dalton et al., Reference Dalton, Pico, Gowan, Clague, Forman, McMartin, Sarala and Helmens2022b). Evidence of significant ice retreat during MIS 3 remains scarce, partly because most evidence would have been destroyed by the subsequent LGM ice advance. However, ice advances and retreats have been associated with MIS 3 in several other studies and areas (Hughes et al., Reference Hughes, Gibbard and Ehlers2013), including western Finnish Lapland (Salonen et al., Reference Salonen, Moreau, Hyttinen and Eskola2014), the west coast of Norway (Mangerud et al., Reference Mangerud, Gyllencreutz, Lohne, Svendsen, Ehlers, Gibbard and Hughes2011), the Yukon Territory (Ward et al., Reference Ward, Bond and Gosse2007), the Davis Strait (Seidenkrantz et al., Reference Seidenkrantz, Kuijpers, Olsen, Pearce, Lindblom, Ploug, Przybyło and Snowball2019), two advances of the Laurentide Ice Sheet into the Upper Mississippi River Basin (Kerr et al., Reference Kerr, Tassier-Surine, Kilgore, Bettis, Dorale and Cramer2021), and Central Asia (Zhao et al., Reference Zhao, Zhou, He, Ye and Liu2006, Reference Zhao, Yin, Harbor, Lai, Liu and Li2013; Gillespie et al., Reference Gillespie, Burke, Komatsu and Bayasgalan2008; Xu et al., Reference Xu, Yang, Dong, Wang and Miller2009). However, we also note that some of these results have been challenged in subsequent literature (Gribenski et al., Reference Gribenski, Jansson, Preusser, Harbor, Stroeven, Trauerstein and Blomdin2018; Miller and Andrews, Reference Miller and Andrews2019; Yan et al., Reference Yan, Owen, Guo, Zhang, Zhang and Wang2023), and the notion of significant MIS 3 ice extent anywhere in the world is still debated (Dalton et al., Reference Dalton, Pico, Gowan, Clague, Forman, McMartin, Sarala and Helmens2022b).
A dry LGM
The early last deglaciation age we report here for a region in northeast Siberia could be linked to a lack of precipitation during the LGM compared with the preceding MIS 4/3. Indeed, lake cores at Lake Emanda and Lake El’gygytgyn both have facies typical of cold and dry climate around the peak of the global LGM around ∼20 ka ago (Melles et al., Reference Melles, Brigham-Grette, Glushkova, Minyuk, Nowaczyk and Hubberten2007; Baumer et al., Reference Baumer, Wagner, Meyer, Leicher, Lenz, Fedorov, Pestryakova and Melles2020). In addition, the El’gygytgyn core shows evidence that a period of cold and moist climate took place around the onset of MIS 4 (Melles et al., Reference Melles, Brigham-Grette, Glushkova, Minyuk, Nowaczyk and Hubberten2007). This provides support for the idea that northeast Siberia was more arid during the LGM than during MIS 3/4. LGM aridification is likely linked to the growth of the western portion of the Eurasian Ice Sheet. Some suggest that the configuration of the Eurasian Ice Sheet during LGM was blocking Atlantic moisture from traveling east (Krinner et al., Reference Krinner, Diekmann, Colleoni and Stauch2011; Barr and Clark, Reference Barr and Clark2012a). Others suggests that the configuration of the Kara and Barents Ice Sheets allowed for vast proglacial lake formation in central parts of northwest Siberia (Lambeck et al., Reference Lambeck, Purcell, Funder, KjaeR, Larsen and Moller2006; Larsen et al., Reference Larsen, KjaeR, Demidov, Funder, Grøsfjeld, Houmark-Nielsen, Jensen, Linge and Lysa2006; Möller et al., Reference Möller, Alexanderson, Funder and Hjort2015), acting as an additional ephemeral moisture source for northeast Siberia during MIS 4 and 6 but not during the LGM (Krinner et al., Reference Krinner, Diekmann, Colleoni and Stauch2011). However, the full nature of the suggested LGM aridification is not yet fully understood. Although we did not find an LGM advance in the Ambar-Yuryuyete Valley, we note that a few northeast Siberian records suggest an LGM extent nearly as large as during MIS 6–MIS 3 (Nørgaard et al., Reference Nørgaard, Margold, Jansen, Kurbanov, Szuman, Andersen, Olsen and Faurschou Knudsen2023), which is also partly supported by our speculative conclusions from the remote sensing. Additional dating of the glacial record in the Chersky Mountains and more broadly in northeast Siberia is still needed to further explore this question.
Conclusion
Using 10Be and 26Al exposure dating, we investigated the glacial history of a small valley in the Tas-Kystabyt Range of the Chersky Mountains in the central part of northeast Siberia. Based on data from two boulder and two bedrock samples, we infer that the valley was deglaciated 45.6 ± 3.4 ka ago, during the peak of MIS 3. The discrepancy between our estimate and other deglaciation ages reported throughout northeast Siberia may be explained by local variations in orography and climate, but it remains unclear whether our age is representative for the wider region or simply represents a local oddity, as we identified only one moraine within the studied valley. Our investigation provides no direct evidence for LGM ice, which is consistent with previous studies from northeast Siberia (Gualtieri et al., Reference Gualtieri, Glushkova and Brigham-Grette2000; Glushkova, Reference Glushkova2001; Brigham-Grette et al., Reference Brigham-Grette, Gualtieri, Glushkova, Hamilton, Mostoller and Kotov2003) although out of phase with global ice volumes that peaked at the LGM, 26–19 ka ago. Using remote sensing, we identify two generations of moraines in the Tas-Kystabyt Range toward the southwest. One generation can speculatively be linked to the dated moraine through ELA estimates, while the other generation indicating a smaller glacier remains undated but could represent the LGM ice extent.
Supplementary material
The supplementary material for this article can be found at https://doi.org/.10.1017/qua.2025.20.
Acknowledgments
We are grateful for financial support from the Independent Research Fund Denmark–Natural Sciences (grant 9040-00199B). We also thank Nastya Arzhannikova, Sergey Arzhannikov, Vladimir Tumskoy, Nikolay Torgovkin, and Daria Semikolennykh for providing valuable help during fieldwork in northeast Siberia in August–September 2021.
Open access
