Absence of Large-Scale Ice Masses in Central Northeast Siberia During the Late Pleistocene

Ongoing speculation regarding the existence of large Late Pleistocene ice masses in Northeast Eurasia reflects the dearth of age constraints on glaciations across this vast region. Here, we report the first dates from the central part of Northeast Siberia, consisting of 22 cosmogenic 10 Be exposure ages from boulders deriving from a sequence of three moraines in the Chersky Range. The dated moraine sequence indicates progressive contraction of maximum glacier extent from Marine Isotope Stage 6 to the Last Glacial

The growth of ice masses in Northeast Siberia hinges upon two potential moisture sources, west and east.The westerlies provide the main source of precipitation along the western flank of the Verkhoyansk and Suntar-Khayata ranges (Figure 1).The deposits of piedmont glaciers extending west from the Verkhoyansk indicate progressively smaller ice advances through time, as shown by optically stimulated luminescence ages of 140 to 50 ka from intercalated aeolian and fluvial sediments (Stauch & Lehmkuhl, 2010;Stauch et al., 2007).To the east, at the opposite periphery, the Pacific Ocean is the predominant source of precipitation.Cosmogenic 36 Cl exposure ages from the Kuveveem and Tanyurer valleys (Figure 1) indicate small-scale mountain glaciation at the LGM along with evidence of at least one prior, more extensive glacial advance (Brigham-Grette et al., 2003;Gualtieri et al., 2000).While on-land evidence is lacking for a former mountain-centered ice sheet thick enough to behave independently of the underlying topography (Barr & Clark, 2012), glacial landforms discovered on the northern continental shelf suggest an ice sheet up to 1 km thick (Figure 1; Niessen et al. (2013)).Such an ice sheet may have grown repeatedly prior to the LGM, but whether it coalesced with mountain-centered ice masses remains an open question.Geomorphic evidence indicates that the extensive ice sheet on the continental shelf predates the LGM and most likely took place during Marine Isotope Stage (MIS) 6 (Jakobsson et al., 2016).
In between the Verkhoyansk Range to the west, the Pacific Ocean to the east and the Arctic Ocean to the north, there is a ∼2,000 × 1,500 km expanse that includes the Suntar-Khayata Range, the Kolyma Highlands and the Chersky Range (Figure 1).This area remains something of a blank spot in terms of moisture availability and glacial history.Here, we present the first cosmogenic 10 Be exposure ages from the Chersky Range, the highest mountains in Northeast Siberia at just over 3,000 m above sea level (masl).We target a glacial moraine sequence at Lake Malyk with three well-defined ridges formed by an outlet glacier that emanated from a mountain icefield flanked by peaks up to ∼2,300 masl.(Figures 2 and 3).

Methods
The three moraine ridges at Lake Malyk were identified in the regional moraine inventory of Barr and Clark (2012).
We mapped the surrounding mountain topography from the high-resolution digital elevation model ArcticDEM (Morin et al. (2016), http://arcticDEM.org) to gain a complete glacial geomorphological overview of the study area (Figure 2).
Fieldwork was carried out in late summer 2021.The moraine ridges were accessed on foot, while searching for suitable ice-transported boulders to sample for cosmogenic 10 Be exposure dating.Boulders were few and scattered sparsely across the moraines, hence all those that we deemed of suitable size and in good condition were sampled with an electric angle-grinder targeting the uppermost 2-3 cm.A total of 22 moraine boulders were sampled and analyzed; 9 samples from the outer moraine ridge, 6 from the middle moraine, and 7 from the inner moraine (Table 1, Figure 3).All sampled boulders were of granitic composition and none showed evidence of surface spalling suggestive of frost action despite the extremely low winter temperatures characteristic for the region.
To determine the abundance of cosmogenic 10 Be in the boulder samples, and thereby constrain their depositional age, we employed the Aarhus University cosmogenic nuclides laboratory and accelerator mass spectrometry center (see Supporting Information S1 for full details on procedures and AMS results).Exposure ages were derived from the online calculators formerly known as the CRONUS-Earth online calculators, version 3 (Balco et al., 2008), using the time-dependent scaling scheme (LSDn) of Lifton et al. (2014) (Table 1).

Results
Our mapping reveals an area of hummocky terrain in the forefield of the Lake Malyk moraine sequence, displacing the course of the Berelyokh River to the foot of the lower mountains in the west (see Figure 2).We interpret this area as a degraded moraine from a glaciation that extended well beyond the Lake Malyk moraines.This is in line with the maximum extent of the mountain ice field, identified from the mapped glacial geomorphological record elsewhere around its periphery (Figure 2).
Of the three sampled moraines, the outermost moraine ridge of the Lake Malyk sequence is the largest and seemingly most denuded given its flattish slopes.The samples from the outer moraine were mainly collected from the distal slope and moraine crest (Figure 3), but three samples were also collected from the gentler dipping proximal side of the moraine (red arrow in Figure 3).The middle and inner moraines are of similar size and degree of degradation, and on both moraines, samples were collected close to the crest.The three moraines are sparsely covered by open-canopy mountain taiga with undergrowth of shrubs, lichen and mosses; the only moraine materials exposed are boulders and blocks ranging from a few tens of centimeters to several meters in diameter.
Apparent exposure ages of the nine outer moraine boulders (Figure 3) range widely between ∼7 and 130 ka, including an age cluster consisting of the three individual ages 109 ± 7, 124 ± 8, and 130 ± 9 ka associated with the three large blocks on the moraine's proximal side.The six boulders of the middle moraine yielded ages Note.Geographical coordinates and elevation above sea level were measured with a handheld GPS.Sample thickness and boulder height were estimated in the field with a tape measure.Ages reported are calculated with the online calculator formerly known as the CRONUS-Earth online calculators, version 3 (Balco et al., 2008) with an assumed rock density 2.65 g cm −3 , a boulder erosion of 0 m yr −1 and the time-dependent scaling scheme (LSDn) of Lifton et al. (2014).All ages are corrected for topographic shielding (additional details in Table S1 in Supporting Information S1).
Corrections for cosmic-ray shielding owing to topography, vegetation, and snow cover, were calculated according to Gosse and Phillips (2001).Topographic shielding estimated from field observations is included in our age calculations (Table 1).Shielding by vegetation was deemed negligible due to the sparse plant cover (Figure S1 in Supporting Information S1).Snow shielding was estimated from the nearest meteorological stations in Yakutsk and Magadan; these suggest a 50 cm snow cover for 8 months per year, but given the stations are located in the lowlands and far from our field site, we acknowledge the high uncertainty of these estimates.As a safeguard, we calculate the snow shielding estimate for both 50 and 100 cm snow cover for 8 months per year, which effectively increases our exposure ages by ∼4% and ∼8%, respectively.Such effects exert little influence on our study findings, so instead we suffice by providing the complete shielding corrections in Table S1 in Supporting Information S1.

Boulder Exposure Ages at Lake Malyk
Two geological sources contribute to the often-observed scatter in boulder exposure ages in moraines: (a) post-depositional shielding linked to the movement of boulders on the moraine or their exhumation over time; and (b) pre-depositional exposure, the persistence of nuclides inherited from exposure prior to moraine deposition (Applegate et al., 2010;Fabel & Harbor, 1999;Heyman et al., 2011).Post-depositional shielding yields exposure ages that underestimate moraine age, whereas inheritance gives rise to ages that are too old.Such factors are important to consider at Lake Malyk given that the exposure ages scatter so widely that the moraine ages all apparently overlap (Figure 4).And yet, the three moraine ridges are assumed a priori to be progressively older from inner to outer, as we expect the glacier to remove or at least smooth any existing ridges that it overwrites on re-advancement.How to address this issue?
First, all moraine ridges are subject to gradual surface lowering and reshaping via diffusional processes, especially those like at Lake Malyk that can, due to the climate of the region, be assumed to be partially ice-cored.The degree of reshaping can be expected to increase with time, thereby enhancing post-depositional shielding effects on boulders in the moraine and skewing the exposure age distribution toward too-young ages (Heyman  , 2011;Putkonen & Swanson, 2003).We can recognize this effect in the notable increase in age scatter from the inner, younger ridge to the outer, older and more intensely denuded ridge (ranging ∼ 29 and 123 Kyr, respectively) (Figure 4).Second, nuclide inheritance is also a common feature.For instance, it is identified in 27% (n = 590) of ice-transported, exposure dated boulders in Fennoscandia (Jansen et al., 2019).While nuclide inheritance in moraine boulders is largely stochastic, it exerts a disproportionate effect on younger moraines because any inheritance present will comprise a larger fraction of the apparent age.
The notion that age scatter at the outer moraine is the result of post-depositional shielding is consistent with the positive relationship we observe between exposure age and boulder height above ground on the outer ridge (Table 1, Figure S2 in Supporting Information S1, Heyman et al. (2016)).That same trend is also present (albeit weaker) at the middle moraine but it is largely absent at the inner moraine, reflecting the difference in post-depositional modifications between moraines.Accordingly, we hinge our interpretation of the outer moraine age on the three oldest, well clustered samples (MS-M1,5B-1, MS-M1,5B-2, MS-M1,5B-3), which indicate deposition toward the end of MIS 6 (Figure 4).The apparent stability of these three boulders sets them apart in our data set.They all stand >130 cm above ground-level, which is more than two-fold taller than the other outer moraine boulders (Table 1), and they rest on the flattish proximal side of the moraine (Figure 3), which presumably protected them from any post-depositional slope activity.
For the middle moraine, we also invoke post-depositional shielding as the main driver of the age scatter (∼89-32 ka); hence, we interpret the oldest ages to be most representative of the moraine age (Applegate et al., 2012;Putkonen & Swanson, 2003).The oldest sample, M2B-8, spans MIS 4 and both glacial sub-stages of MIS 5 (Figure 4) making it hard to distinguish between these.Nevertheless, in accordance with our a priori model of the relative ages, the middle moraine likely arose during one of these periods, but we cannot rule out that it arose during MIS 3.
The seven boulder ages of the inner moraine are generally younger and closer in agreement with each other than for the outer and middle moraines.Following our model noted above, here we consider that the age scatter derives mainly from nuclide inheritance via pre-depositional exposure.We envisage the likely scenario in which boulders transported by the earlier ice advances are incorporated into the inner moraine.This would result in some of the boulder exposure ages being older than the moraine itself (Heyman et al., 2011); namely samples MS-M3B-7 and MS-M3B-11.The remaining ages place the deposition of the inner moraine to after stabilization at the end of the global LGM.

Siberian and Northern Hemisphere Glacial Contraction Since Middle Pleistocene
Our results represent the first robust glacial chronology for the central portion of Northeast Siberia that extends beyond MIS 2. Combined with previous findings from peripheral areas, we identify a regional pattern of glaciation despite the putative diversity of moisture sources.
An emergent picture of contracting ice extents, both for ice sheet and mountain glaciation, during the Middle and Late Pleistocene is shared by other regions with continental climates extending geographically over both Eurasia and North America.It applies to the eastern margin of the Eurasian Ice Sheet (Astakhov et al., 2016;Svendsen et al., 2004), the Cordilleran Ice Sheet margin in the Yukon (Hidy et al., 2013;Ward et al., 2017), as well as to the mountain ice masses of Alaska (Briner & Kaufman, 2008;Kaufman & Manley, 2004), northern High Asia (Blomdin et al., 2016) and southern-central Siberia (Margold et al., 2016).Like others before us (Barr & Clark, 2011;Stauch & Gualtieri, 2008;Stauch et al., 2007), we speculate that these contracting continental ice masses reflect the moisture-starved conditions caused by the massive bulk of the Laurentide and Eurasian ice sheets, which effectively divert atmospheric circulation patterns during glacial maxima (Krinner et al., 2011;Löfverström et al., 2014;Yanase & Abe-Ouchi, 2010).However, we cannot completely exclude the possibility that the climatic conditions at Lake Malyk stayed relatively constant during periods of glacier build-up, and that the erosional feedback and geomorphological changes from previous advancements were responsible for the shortening of the glacier (Anderson et al., 2012;Rowan et al., 2022).
In most of the continental regions noted above, the timing of Pleistocene glacial maxima remains unresolved or problematic.Considering the reconstructed ice extents relative to the global ice volumes implied by the δ 18 O record (Lisiecki & Raymo, 2005), we confirm that continental Northeast Siberia did not host the "missing LGM ice sheet" (Simms et al., 2019).Indeed, if a large ice sheet did develop somewhere in Northeast Siberia during the LGM, it seems unlikely that it coincided with minimal ice extent in the western periphery (the Verkhoyansk Range, Stauch and Lehmkuhl ( 2010)), the eastern periphery (the Pacific Ocean, (Brigham-Grette et al., 2003;Gualtieri et al., 2000)), and the central part of Northeast Siberia (the Chersky Range (this study)).Additionally, proxy-based temperature reconstructions indicate that the temperature decrease was minimal in Northeast Siberia and Beringia during the LGM (Osman et al., 2021).Given the reasonably limited ice extent in the Chersky Range and the Verkhoyansk Range toward the end of MIS 6, some ∼130 Kyr ago, a large ice sheet connecting the various mountain icefields in the Northeast Siberia also seems unlikely during the penultimate glacial period.With regard to the yet unconstrained Pleistocene maximum glacier advance in this region, we favor one of the super-glacials of the Middle Pleistocene (MIS 16 or 12), or the more recent MIS 10.An Early Pleistocene maximum is also plausible, given the first major advance of the Cordilleran Ice Sheet at ∼ 2.8-2.4Ma (Hidy et al., 2013).However, this would exacerbate the difficulty of matching the low global ice volumes of the Early Pleistocene (Lisiecki & Raymo, 2005) with regionally extensive ice masses (Balco & Rovey, 2010;Hidy et al., 2013).

Conclusion
We have presented 22 new cosmogenic 10 Be exposure ages from glacially-transported boulders in the Chersky Range, the highest mountains of Northeast Siberia, where the glacial history was so far undated.The ages from our three distinct end-moraines are strongly influenced by geological processes, causing partial cosmic-ray shielding and/or inheritance of nuclides.Taking into account these complications, the tripartite-moraine system is interpreted to comprise a MIS 6 outer moraine, a LGM inner moraine, and an intermediate moraine that likely was deposited during MIS 4 or 5. Our findings reveal a progressively smaller glacier footprint in Northeast Siberia since MIS 6.The shrinking trend likely reflects the effects of extreme continentality and/or the development of moisture-blocking ice masses in western Eurasia and North America.Akin to studies reporting from the periphery of Northeast Siberia (Brigham-Grette et al., 2003;Gualtieri et al., 2000;Stauch & Lehmkuhl, 2010;Stauch et al., 2007), we show that Late Pleistocene glaciations in the Chersky Range were limited to alpine-type valley glaciers.A large coalescent ice sheet, or "missing LGM ice sheet" (Simms et al., 2019) did not develop here during the Late Pleistocene.We mapped traces of the local Pleistocene maximum glacial extent that we tentatively link to one of the super-glacials MIS 16 or 12, or the more recent MIS 10.However, the question of whether the mountain-centered ice masses coalesced into a larger ice sheet conjoined with the glaciation on the continental shelf (cf.Niessen et al., 2013) requires further investigation.

Figure 1 .
Figure 1.Northeast Siberia, showing the regional setting of the study area (red box) in the Chersky Range and other locations and studies mentioned in the text.The bottom right insert shows the Northern Hemisphere maximum ice extent during the Last glacial cycle adapted from Ehlers and Gibbard (2007), Hughes et al. (2013) with the Laurentide Ice Sheet (LIS) and Eurasian Ice Sheet (EIS) marked.

Figure 2 .
Figure 2. Geomorphological mapping of the area surrounding the study site.The overview indicates that the dated moraines are not representative of the largest glaciations in the Chersky Range.The maximum extent of glaciation (light blue outline) remains undated.

Figure 3 .
Figure 3. Top panel shows the Lake Malyk valley, including the three transverse end-moraines and sampling locations (orange dots).Boxes show the exposure ages within the sample clusters.The lower panel shows a topographic profile through the transect in the upper panel (A-A'), highlighting the differing moraine morphologies (sample elevations are projected onto the profile).The red arrow on the outer moraine points to the key samples MS-M1,5B-1, MS-M1,5B-2, and MS-M1,5B-3 in a position less vulnerable to post-depositional instability.

Figure 4 .
Figure 4. Boulder exposure ages for the outer, middle and inner moraines shown with 1-σ (box) and 2-σ (whisker) uncertainties.Fading colors in the background indicate interpreted moraine ages (green = outer moraine, orange = middle, blue = inner).The lowermost panel is a visualization of the inferred age of the glaciations at Lake Malyk.
• Be-10 exposure ages from moraines in the Chersky Range show that glaciers in central NE Siberia contracted over the last two glacial cycles • Northeast Siberia hosted mountain-centered icefields during the Late Pleistocene, but no large-scale continental ice masses • Largest ice extent in the Chersky Range predates the penultimate glacial cycle and likely occurred during the Mid-Pleistocene super-glacials

Table 1
Summary of Cosmogenic Nuclide Data