Late Quaternary sedimentation dynamics in the Beenchime-Salaatinsky Crater, Northern Yakutia

The Beenchime-Salaatinsky Crater (BSC) is located west of the Olenyok River in Northern Yakutia, ~ 260 km south-west of Tiksi and the Lena Delta. The age and origin (volcanic versus meteoritic) of this crater is poorly understood. The key scientific interest in re-visiting the BSC is the reappraisal of the Quaternary sedimentation dynamics for a better understanding of the sediment history and thickness in the basin. This aides for an assessment, if the site is prospective for a deeper drilling of a Quaternary (or Cenozoic) sediment archive. Soil pits and auger cores from slopes and lowland terrain in the basin were sampled and studied to infer sediment ages and transport dynamics. This also included a thermokarst lake placed in the centre of the basin. Studied properties include grain-size distribution, organic carbon and nitrogen contents (TOC and TN), heavy mineral compositions, δ13C of organic carbon, 14C ages from sediment, δ18O and δD from ground ice and waters, and lake bathymetry from GPR profiling, in addition. We conclude that the crater floor in the BSC is underlain by fluvial/alluvial sediments from the MIS 3 period. Thermokarst lake formation took place during the Holocene Thermal Maximum between 7600 and 6100 cal yr BP. The lake has been shrinking hereafter. Fluvial/alluvial sedimentation along the drainage pattern was active again between 5700 to 1500 cal yr BP, and it was flanked by the accumulation of peaty and organic-rich sediments and the formation of ice-wedge polygons.


Introduction
A geologic and geophysical sampling project in northern Yakutia is designed for tracing back Arctic permafrost inception and Cenozoic environmental changes on a yet unknown time scale-potentially several million years. The locality is the Beenchime-Salaatinsky Crater (BSC) in northern Yakutia, which presumably arose around 40 million years ago and has been acting as a sediment trap for a yet unknown period. If the given ages are true, the BSC may contain the oldest environmental archive of the terrestrial Arctic. A multimillion-year old continental record from the BSC could become a counterpart to the prominent marine ACEX core, which was collected over the North Pole and stretches back to the Palaeocene/ Eocene (ca. 55 million years ago, [36]). If the core consists of lake sediments to a great part, it possibly can be compared with the palaeoclimate and palaeoenvironmental history contained in the 3.5 million years old record from El'gygytgyn Impact Crater Lake [10,30]. In any case, a long BSC environmental record at least can serve for studying periglacial and permafrost processes on a Quaternary time scale in an area that today has an estimated permafrost thickness of > 500 m [13]. Even though BSC has been a closed basin for a long time, a multidisciplinary research would benefit from analysing a big variety of scientific fields such as sedimentology and permafrost hydrology, geomorphology, near surface, and borehole geophysics. The study scope would also include mineralogy, geochemistry, and isotope analyses along with studying vegetation diversity, macrofossil (e.g., ostracods, diatoms, and gastropods) and microbial diversity, biomarker detection, and drilling technology.
Little is known about the beginning of permafrost formation and the associated environmental and climate conditions, this may help to better interpret long-term permafrost dynamics. It is suspected that permafrost had its origin at the beginning of the Pleistocene era 2.6 million years ago, but this should not imply that more ancient and intervening thermal conditions were without significance [27]. The late Pleistocene environment of North Yakutia favoured accumulation of ground ice and formation of syngenetic permafrost. Severely cold climate and polygenetic sedimentation (i.e., aeolian, fluvial, alluvial, and nival processes) resulted in building up frozen sediment sequences that are several dozens of meters thick [47,49,54]. In fact, extremely ice-rich and silt-dominated permafrost deposits (termed "Yedoma Ice Complex") were formed and remain one of the most prominent features of the periglacial environment in the terrestrial Arctic. The dry and cold conditions over much of the Pleistocene in Northeast Siberia [19] ensured that depositional basins in the region have remained ice free, which thus may have accumulated a near continuous record of environmental and climate proxies in this area. An enclosed basin such as the BSC allows for studying a variety of multiproxy data (biotic and mineralogic indicators) to reconstructing the environmental conditions framing the permafrost evolution in the area.
Various detailed palaeoenvironmental reconstructions have been made from frozen depositional sequences including Yedoma Ice Complex and they cover a time axis back to marine isotope stage (MIS) 6 [60]. Understanding the nature and distribution of ice-rich permafrost is of concern, because it is widespread in Arctic and sub-Arctic east Siberia, and it provides the foundation for diverse Arctic and boreal ecosystems that are exposed to permafrost thawing, if permafrost warming continues [8].
The BSC, located west of the Olenyok River in Northern Yakutia (Fig. 1), is an 8-km-wide basin inside a ring structure that has neither been verified for its Quaternary sediment age and thickness, nor for its potential to allow a palaeoenvironmental reconstruction in the area, and it is placed in a corridor that has never been glaciated during the Pleistocene [15]. It has been a prospective site, although the crater is virtually unknown in terms of its age. A multi-million-year old formation is likely due to its geomorphological down worn crater rim, the BSC is listed as 65 M years old [35], < 65 M years old [18], or 40 ± 20 M years old (data from [14]). The formation of the crater has been debated differently. Pinchuk [43] and Khain [23] describe the BSC as a kimberlitic pipe of uncommon size, comparable to the kimberlitic pipe Mwadui in Tanzania, whereas Mikhailov et al. [34] and Masaitis [29] preferred a formation due to a meteoritic impact event.
Scientific key interests for re-visiting BSC were (i) assessing sediment processes and ages of the Quaternary Schematic map with inset and satellite view of the study region (credits: PlanetScope [40], I. Nitze under "Personal Research License"). The red star marks the location of the BSC (Beenchime-Salaatinsky Crater). The circular shape, which has 8 km in diameter, as seen on satellite imagery strata and (ii) determining the potential of the basin to preserve a long-lasting Quaternary sedimentary archive. Field work focused on sampling of near-surface deposits from frozen ground (down to 2 m depth) and from a shallow thermokarst lake (water depth 5 m at maximum), which is placed in the crater's centre.

Study site
BSC is placed in the north-eastern part of the Siberian Craton ( [29], Fig. 1) and named after the nearby passing rivers Beenchime (south) and Salaata (west). Formed by collisions of several microcontinents of different ages, the basement of the Siberian Craton consists of Lower Precambrian rocks (cropping out primarily in the Anabar and Aldan shields). In the study area, the crystalline bedrock is overlain by Upper Precambrian and Lower-to-Middle Cambrian platform sediments including carbon-bearing formations and argillocarbonates [26,44]. Next to this, marine sedimentary rocks of Upper Cambrian-to-Lower Triassic age are described for this area [26].
The geology of the crater is dominated by Cambrian limestones and dolomites partially mixed with siliceous sandstones, conglomerates, and Permian sandstones [39]. To the north and south-west of the BSC, Upper Palaeozoic allogenic breccia has been identified.
BSC forms a circular basin with a fairly gentle topographic relief partially filled by Quaternary sediments dipping slightly to the west [34]. The inner part has a seasonally active drainage pattern exiting the crater rim to the west. A second break of the rim is located in the southern segment of the rim (Fig. 2).
Based on literature data, the following scenario is summarised with respect to the BSC basin history: (i) platform sedimentation during Vendian, Cambrian, and Permian time [39], (ii) (Post-) Permian impact event, Mesozoic (Lower Triassic) overburdening [26], (iii) subsequent erosion in the course of the Mesozoic-Cenozoic Olenyok Uplift [26] with exposure, erosion, and mobilisation of Paleozoic rock.
Inside the crater, the distribution of Quaternary sediments can be attributed to two morphological levels; the lower one (Holocene) is dominated by the seasonally active river network and holds polygonal patterned ground. Partly, this area is subaerially degrading, and partly, it is boggy. It is vegetated by forest tundra and shrubs. The upper level (late Pleistocene to Holocene) lacks those physiographic features and is covered by forested tundra with Larix stands (Fig. 2).
The area has sub-Arctic climate. Data from the closest weather station in Olenek (241,250, N68.5000, E112.4331) provided by NOAA show an average annual precipitation of 285 mm and an average temperature of −12.4 °C with a maximum of 20.3 °C and a minimum of −42.7 °C (period: 1961-1990).

Field work
During fieldwork in July 2016 [50], six sites from the basin lowland have been sampled (Fig. 2). Sample depths ranged between 0.80 m and 1.85 m. Samples were collected by digging the active layer and drilling into the upper permafrost using an SIPRE Coring Auger Set (diameter 70 mm). Three sections are located on polygonal patterned ground in the central part (BSC-PS-02, BSC-PS-05A, BSC-PS-05B). Others (BSC-PS-01, BSC-PS-03, and BSC-PS-04) are located in areas, which are interpreted to be remains of ancient fluvial terraces. BSC-PS-02 is located in the centre of a degrading polygon. BSC-PS-05A and BSC-PS-05B are located in a low-centre polygon near the shoreline of the biggest lake in the crater's centre (informally named "Big Lake"; labels A and B result from relocating the equipment at close distance of 50 cm. Sediment field description included colour, grain-size composition, occurrence of organic material, and ground ice fabric. Gravimetric ice contents were measured already in the field and are expressed as weight percentage (wt%). For comparison with sediment and ground ice properties, we also sampled lake water from the Big Lake (BSC-W) and water and surface sediment from the nearby Beenchime-Salaatinsky River (BSC-FS).
Three short cores have been collected from the Big Lake. The cores had lengths of 0.22 m, 0.30 m, and 0.51 m, and were extracted using a UWITEC gravity corer. One of them is further analysed for this study (BSC-LS-03). Information on bathymetry and lake bed geomorphology was acquired with a 50 MHz GPR antenna (ground-penetrating radar from Mala/Geoscience). GPR profiling was conducted from a rubber boat along two traverse lines running perpendicular to each other (Fig. 3). The water depth has been repeatedly verified by manual measurements using a rope with a weight fixed to it. GPR data were recorded in the single-offset mode with a GPS receiver connected to the system. Good penetration of the electromagnetic waves was permitted through the water column, since the lake water has a low conductivity (~ 160 μS/cm). The data were imported into standard processing software (ReflexW) and an optimal processing flow determined included the following standard procedures: (i) subtract DC shift, (ii) static correction, (iii) bandpass lower 12.5 MHz, upper 70 MHz, (iv) auto interpolation, and (v) travel time-todepth conversion with a depth axis v = 0.033 m/ns for water [51].

Laboratory work
After transport to the lab, the sediment samples were analysed for the following properties. Mass-specific magnetic susceptibility (MS) was measured on freeze-dried subsamples with a Bartington MS2 magnetic susceptibility system using the low-frequency option of a MS2B Sensor. The data are expressed in SI units (according to Le Système International d'Unités, 10 -8 m 3 kg −1 ). For analysing grainsize distribution (GSD), a Malvern Mastersizer 3000 was used. For total carbon (TC) and total nitrogen (TN) content, samples were measured on a Vario EL III element analyser with reference to three lab internal standards (IVA 2150, STSD-4, and GSD-4). Total organic carbon (TOC) content was measured on a VarioMax C using glutamine-quartz mixtures of different ratios as references. Values for inorganic carbon (TIC) were calculated by subtracting TOC from TC. The composition of stable organic carbon isotopes (δ 13 C) was analysed on a Thermo Fisher Delta V Advantage mass spectrometer with USGS24 tryptone and polyethylene as standards. For the analysis of heavy mineral (HM) composition, selected samples were separated by density solution after the 63-125 µm fraction was extracted by sieving. For inspections under the microscope, particles were attached on slides with Meltmount 1.68™ and around 200 grains were counted following procedures outlined in Boenigk [9]. In addition, material found in Mange and Maurer [28] was used as reference for identification. Radiocarbon dating was done by MICADAS (MIni radioCArbon DAting System, AWI Bremerhaven) and by CologneAMS (University of Cologne). For calibration of 14 C ages into calendar years, the routine of Stuiver et al. [56] was performed using the IntCal13-curve and the CALIB 7.1 Internet program.
Analysis of pH and electrical conductivity (EC) from ground ice and water samples was done already in the field with a WTW system using pH-Electrode Sentix 81 and Tet-raCon 325. Stable water isotopic composition (δD, δ 18 O) was measured on a Finnigan MAT Delta-S mass spectrometer. The values are given as per mil (δ, ‰) difference from the Vienna Standard Mean Ocean Water (VSMOW) standard, with internal 1σ errors of better than 0.8‰ for δD and 0.1‰ for δ 18 O. The deuterium excess (d-excess) is calculated according to Dansgaard [11] as d = δD − 8δ 18 O.

Radiocarbon ages
When grouping the samples according to their ages ( Fig. 4; Table 1), it becomes obvious that sediments filling the crater lowland have been deposited during the late Pleistocene Kargin Interstadial (~ MIS 3, [59]) and during the Holocene (~ MIS 1). Kargin sediments are found in sections BSC-PS-01, -03, and -04, whereas Holocene deposits are associated with organic-rich and peaty sections of BSC-PS-02 and -05 and lake sediments from core BSC-LS-03 (Fig. 5).
The material used for radiocarbon dating is different in nature. Sites BSC-PS-01, -03, and -04 are dominated by sandy to silty sediments and dating of these sections mostly relies on bulk detritus. BSC-PS-03 and -04 allowed for only one date in each of the sections (Table 1) in the absence of other datable material. BSC-PS-01 has late Pleistocene deposits overlain by late Holocene sediments. BSC-PS-03 and -04 layers point to a late Pleistocene age in the absence of more datable material, whereby BSC-PS-03 provided only an individual piece of charcoal for dating. In contrast, peaty sections BSC-PS-02 and -05 allowed for picking rootlets and wood fragments (Table 1) and lake sediments from short core BSC-LS-03 are rich in macroscopic plant remains and organic detritus. This allowed for a better temporal resolution of the peaty and lake sediments when compared with the soil pit deposits mentioned above. Age inversions occur in all studied depositional settings; in sandy sediments (BSC-PS-01), in the peaty section (BSC-PS-05), and in the lake sediments (BSC-LS-03).

Bathymetry
Below first arrivals of air wave and direct wave, the GPR profiles image the first subsurface reflector below shallow water in the Big Lake (Fig. 3). The reflector plunges fairly This first subaquatic reflector is interpreted to represent the lake bed. According to the travel time-to-depth conversion the water depth is greatest at about 5 m as can be seen in the SW-NE profile. Below the first reflector, the energy is quickly absorbed and any more subground information is masked.
BSC-PS-03, drilled in the same terrace as BSC-PS-01, was sampled down to 1.1 m depth, is made of sand and contains patches of dark, organic-rich material between 0.5 and 0.26 m depth. The sandy sediments peak in the finesand spectrum (Fig. 6). At a depth of 0.40 m, sediments were dated to an age of 40,067 cal yr BP. The MS shows little variation with overall low values between 9 and 24 SI similar to BSC-PS-01 (Fig. 5). The ice content from just below the permafrost table at 1.0 m amounts to 19 wt%.  top of a fine-sand layer. The grain-size distribution peaks in the medium and very fine-sand spectrum (Fig. 6). Excess ground ice was absent and stable water isotope measurements were discarded for this profile. Samples from 0.83 m depth were dated to 32,090 cal yr BP using the bulk fraction and to 46,513 ± 640 year BP (uncalibrated) using a charcoal fragment (Table 1). MS values show little variation around 9 SI (Fig. 5). TN values are close to the detection limit, except at a depth of 0.82 m (0.1%) and at the surface (0.3%). Similarly, TOC values are often below the detection limit but peak to 2.7% at 0.82 m depth and 3.6% at the surface. TIC is low below 0.60 m depth (below 0.8%) and higher   Opaque  67  68  64  62  48  69  41  Garnet  19  18  13  21  18  16  3  Zircon  6  5  6  5  16  5  5  Rutile  1  0  1  0  3  2  1  Tourmaline  1  0  0  0  0  0  0  Epidote  1  2  5  2  5  3  9  Zoisite  1  0  2  2  4  1  5  Clinozoisite  0  1  1  1  1  1  2  Orthopyroxene 1 (Table 2). BSC-PS-02 has been drilled down to 1.8 m depth and is located within the seasonally active river network. It consists of ice-rich peat (pore ice and ice veins). The section is underlain by massive ground ice, which was visible at the core bottom. The mineral fraction of the organic deposits was negligible and a grain-size determination was discarded. The ground ice stable isotope composition is heavier than in BSC-PS-01, and δ 18   BSC-PS-05A/-05B, located at the southern margin of the Big Lake and sampled down to 1.0 m, is also dominated by peaty deposits. It has similar ground ice structures as seen in BSC-PS-02, and similarly, massive ground ice is underlying the sampled section. δ 18 O varies between −26.68‰ and −18.36‰ and δD varies between −203.47‰ and −139.09‰, and has the lightest values at the bottom. Remarkably, all hydrochemical values from this section show a distinct offset occurring at 0.70 m depth when the massive ice is encountered, e.g., δ 18 O becomes heavier by about 7‰ and δD for more than 50‰. This offset is also mirrored in the d-excess trend, which ranges between 11.0 and 4.3 ( Table 3).
Samples of BSC-PS-05A have a minor portion of mineral content, which peak in the medium silt spectrum (Φ 5, Fig. 5). Radiocarbon dating reveals ages of 2312 cal yr BP (0.55 m depth) and 2432 cal yr BP (0.35 m depth) ( Table 1). From the bottom to the top, MS increase from 14 to 78 SI and TOC is moderate at the bottom (12%) and high at the top (43%). TN contents increase from 0.7% to 0.9% with an individual maximum at 0.12 m (1.7%) and TIC contents increase from 0.4% to 2.3%. δ 13 C ranges fairly constant around −29‰. The ice content accounts for 84 wt% (0.56 m) and 54 wt% (0.36 m). Section BSC-PS-05B, sampled in half a meter distance to BSC-PS-05A, overall has the same sediment characteristics as described above (Fig. 5).

Lake core
The lake sediments are composed of silty sediments rich in organic detritus. Remarkably, the lower lake sediments (0.52-0.44 m) are slightly coarser (mean mode at 25 µm) when compared with the overlying lake sediments (mean mode at 15 µm) and are shifted by 10 µm with their mean peak from the medium to the fine silt spectrum.
TOC values in the lake sediments increase from the bottom to the top from 4.2 to 25.1%. MS values increase from 28 to 58 SI including a distinct peak of 99.5 SI (0.21 m), where the boundary is between more consolidated sediment at the bottom and more loose sediment overlying them (Fig. 5). TN increases upwards from 0.5% to 1.8% and TIC decreases upwards from 0.8% to almost 0%. The TOC/TN ratio is relatively constant around 12. δ 13

Organic source identification using TOC/TN and δ 13 C
Plotting TOC/TN versus δ 13 C is used to illustrate the proportional mixing of terrestrial and subaquatic sedimentary organic matter according to Meyers [33]. The sandy sections (BSC-PS-01, -03, -04) group mostly in the transitional field between the two fields ( Fig. 7a) with some individual samples indicating a pure origin from either "C3 land plants" or "lacustrine algae". Plotting the peat locations (BSC-PS-02, -05A/B) shows higher TOC values when compared with the sandy sections and δ 13 C becomes distinctly lighter (Fig. 7b). Much of the samples from the two peaty sites group in the field "C3 land plants". Organic matter from the lake (BSC-LS-03) shows data points that all group in the transitional field between the two fields "lacustrine algae" and "C3 land plants". Samples with lightest δ 13

Surface water electrical conductivity, pH, and stable water isotopes
Three measurements from different depths in the Big Lake have EC values between 82 and 162 μS/cm ( Table 3) Table 3). The Beenchime River, for comparison, had a pH value of 9.2. The composition of stable isotopes from the lake water has mean values of δ 18 O of −18.94‰ and δD of −146.54‰ (Fig. 8) and a d-excess between 5 and 5.1. The stable isotope composition of Beenchime River is slightly heavier with δ 18 O = −17.90‰ and δD = −138.71‰ and a d-excess of 4.5.

Discussion of late Quaternary palaeoenvironments in the BSC
Based on dating results, we have grouped the data into two main periods; the Kargin Interstadial (50-30 cal kyr BP) and the Holocene (< 11.5 cal kyr BP) that is subdivided into the Holocene thermal optimum and the late Holocene. The discussion combines various sediment properties within one particular period.

Kargin interstadial (50-30 cal kyr BP)
Dating information is overall sparse for this time period and also includes an age reversal. The relevant sandy sections (BSC-PS-01, -03, and -04, Fig. 4a) have distinct sediment layering and partly show erosional contacts. It suggests that sediments have been relocated under different transport regimes and sediment reworking may play a role. The peak in the fine-sand spectrum remains fairly stationary in all inspected samples; only the silt portion varies to some extent (Fig. 6). We interpret that varying fluvial-to-alluvial transport energy mobilised the material along drainage pathways or as slope wash presumably during melt water periods and after rain events. The properties of detrital organic components scatter in terms of δ 13 C and TOC/TN composition and suggest that biomass remains in the sediments stem from both, lacustrine and soil-related sources (Fig. 7). If the material is not re-worked, this would either argue for wet soil conditions in ice-wedge polygonal ponds providing a habitat to algae growth (lacustrine algae samples), or in the other case eroded fragments from land plants that have been incorporated into the sandy sediments (C3 land plants samples). This mixture from lacustrine and terrestrial detrital material is also known from other thermokarst terrain or polygonal ground in Northern Siberia [20,45,55].
Heavy mineral assemblages of BSC's late Quaternary detritus have pronounced occurrence of garnet, epidote, or zoisite. These minerals are likely connected to the Amphibolite or Greenschist Facies, which result from moderate pressure and temperature metamorphosis (e.g., [61]). Since bedrock around BSC is formed by Paleozoic sedimentary rock, these mineral signals may point to the Pre-Cambrian terrane orogeny of the eastern (active) margin of the Siberian Craton, where acid volcanics and carbonaceous silicites have been transformed to greenschist metamorphics and subsequently have been intruded by granites [44].
According to palaeoenvironmental reconstructions for NE Siberia, summer climates varied spatially during the Kargin Interstadial. Warm periods-when summer climates were probably as warm or nearly as warm as those at present-are determined between 39-33 cal kyr BP and 30-26 cal kyr BP [1]. This also includes a phase of thermokarst activity with thermokarst lake (alas) development. The climatic optimum of the Kargin Interstadial in the Laptev Sea region is placed between 40 and 32 cal kyr BP based on various bioindicators [2]. In this interval, it is likely that sediment mobility was increased under varying stream power especially during meltwater and summer seasons when compared with cooler periods. A lake record from the Verkhoyansk Mountains (Lake Billyakh) demonstrates that the Kargin Interstadial in Central Yakutia was comparably wet with high lake levels and increased sedimentation rates [12]. Similar results are known from Lake Kyutunda on the western Lena side [7].
There are no dates available from the BSC for the period 33-14.5 kyear BP (~ MIS 2). This might be due to the fact that the relevant deposits have been overlooked in the field, or they have been removed by subsequent erosion during MIS 1, or they are simply absent in the basin. Finally, a lack of dates and datable material may also explain this large temporal gap in the sedimentary record. By now, this issue must be left open. Sedimentation rates generally are expected to be low during the last glacial maximum (LGM), because the LGM is commonly known as being arid in the region. This is known from both, summer and winter proxy data; i.e., pollen records from lake sediments [37,57] and ground ice stable isotopes from ice wedge formations [38].

Holocene thermal optimum (9000-6800 cal yr BP)
Lake sediment ages belong to the late early to mid-Holocene (Fig. 4b). The age distribution suggests that the time of deposition was confined to only about 1500 years between 7600 and 6100 cal yr BP with no detrital sedimentation after about 6000 cal yr BP. However, it must be noted that deeper coring may extend the datable lake sediment record though. There is one age inversion occurring, which indicates that older drift material has admixed into a sediment layer at 0.40-0.42 m depth ( Table 1).
The GPR-based bathymetry shows a stepwise plunging of the slopes into the lake basin and a bumpy lake bed in the lake centre. The shape of the lake bed resembles those, which evolve from thermokarst dynamics, where ice-rich permafrost thaws out and where the depression is subsequently filled with water. Similar lakes following a thermokarst basin evolution have been described elsewhere [5][6][7]17].
There is a lack of deeper reflectors in the GPR profiles, which suggests that the thickness of lake sediments amounts only to a few decimetres, unless the GPR penetration suffers from the attenuation of the electromagnetic waves. In this case, the bottom of BSC-LS-03 is seen nearly as the lake sediment bottom and it would date the onset of lake sediment deposition around 7600 cal yr BP. Deeper coring may yield a greater ages for the onset of lake formation though. Nevertheless, this time frame is similar to other thermokarst lakes found in Northern Siberia, for example Lake Nikolay in the Lena Delta, which started to form at a 7900 cal yr BP [51].
A distinct change in grain-size distributions occurs at 0.40 m core depth where overlying sediments have mean grain sizes finer for about 10 µm in the medium to fine silt fraction (Fig. 6). According to the age distribution the coarser layers at the bottom are deposited around 7600 to 7500 cal yr BP, whereas the overlying finer portion has been deposited between 7500 and 6100 cal yr BP. This argues for a loss of transportation energy into the basin after about 6100 cal yr BP. Shorelines may have retreated since then and the area, where site BSC-PS-05 has been sampled, emerged to become subaerial. Greatest age for the emerged ancient shore is 3000 cal yr BP at 0.8 m depth (Fig. 5), but might be extended with deeper coring. A regional cooling trend during the late Holocene is known from other sites in the area; a decrease of temperatures by about ~ 2 °C from warmerthan-present mid-Holocene summer temperatures until the establishment of modern conditions around 3000 years ago is reconstructed based on a regional pollen-climate transfer function [24,25]. In BSC, this might be reflected in a shrinking lake area as is also suggested by satellite image interpretation (Fig. 6). Other reasons such as decreased precipitation and site-specific geomorphic change (i.e., changing run-off pathways) still may also play a role.
When combining the interpretations of age (mid-Holocene) and origin (thermokarst induced subaquatic morphology), the lake history matches those from e.g., Lake Nikolay [3], or Lake Kyutyunda [7], or other lakes described in Biskaborn et al. [6] and Klemm et al. [24]. Increased thermokarst activity leading to lake formation during this time is known also from the southern Laptev Sea margin (e.g., [3]). Thermokarst basins developed after the increased warming during the regional early Holocene thermal maximum [21], when summer air temperatures in Northern Yakutia were ≤ 4 °C higher than present, and precipitation was higher than present [2].

Late Holocene
Other ages acquired from soil pit sampling belong to the late Holocene (< 5000 cal yr BP) (Fig. 4a). When encountered in one section, a clear boundary between late Pleistocene and overlying late Holocene layers is not obvious though (e.g., BSC-PS-01); either because of the limited temporal resolution within the stratigraphy, or because of the absence of well-defined contrasts in measured sediment properties (Fig. 5). Grain-size compositions do not differ markedly from late Pleistocene deposits and, thus, similar transport processes are interpreted; namely snowmelt run-off and slope wash from rain events. In contrast to the late Pleistocene layers, the Holocene layers at least partly contain detrital organic components with lighter δ 13 C values, namely in BSC-PS-02 at 0.30 m depth (i.e., < 27). Light δ 13 C values have been found indicative for Holocene lake sediments or for Holocene peat deposits elsewhere in Northern Siberia (e.g., [52,58]). Warmer-than-modern summer air temperatures in NE Siberia are reconstructed for the period between 3700 and 3300 cal yr BP as inferred from high values of Betula nana in pollen records and the continuance of Larix north of its present range [4]. In BSC especially peat (BSC-PS-02) and other organic-rich sediments (BSC-PS-05) are formed during this time. The organogenic deposits accumulated in boggy lowland and along water-logged drainage pathways and their accumulation areas are shaped by polygonal ice wedge growth. The peat formation period during late Holocene time is similar as described from other sites in Siberia north of the treeline [41]. Hereafter (< 3300 cal yr BP), environmental conditions are reconstructed to be similar to those at present [2].
Ground ice samples taken from sand and peat sections plot relatively well on the GMWL, which suggests a low impact of secondary evaporation or refreezing on its formation (Fig. 8). Only two samples from BSC-PS-01 are found above the GMWL indicative for a more evaporated vapour source (e.g., [42]). EC in one of these two samples (sample depth 1.40-1.60 m) peaks at 2168 μS/cm (Table 3). Possibly, this EC peak evolved from an active layer thawing and refreezing deeper than modern, where two freezing fronts from below and above have caused an enrichment of the light soluble fraction between the fronts before final refreezing. This refreezing process might also have caused the depletion of stable water isotopes along the freezing front in the same part of the section. Similar freezing front dynamics have been observed elsewhere [16]. The deeper than modern active layer dynamics might be related to local factors or to palaeoclimate events such as the Holocene thermal optimum in the area (see [31]).
The stable isotope composition with δ 18 O values around −18‰ for many of the samples is interpreted to result from syngenetic freezing during late Holocene time. The stable isotopes from the intrasedimental ground ice (mixture from rain and snow, [53] is clearly heavier than few individual samples related to buried ice wedge ice, which has been hit at the bottom of section BSC-PS-05 (Fig. 8). Here, close to the lake shore ice-wedge polygons have developed in organic-rich sediments. Whereas intrasedimental ground ice results from a mixture of snowmelt and rain [53], massive ground ice, including ice wedges, is fed by snowmelt only and has a lighter stable isotope composition. Values of massive ground ice in BSC-PS-05 match those described for the Holocene from Northern Siberia [46,48] or are lighter than those described for Holocene samples from the Lena Delta region [32], but ice wedge formation in BSC may take place under more continental climate conditions in the Yakutian hinterland than those placed around the Laptev Sea margin. Nevertheless, δ 18 O values around −26 ‰ from BSC ice wedges are similar to those described from late Holocene ice wedges sampled at the Batagay mega slump; whereby the Batagay site at the lower reaches of the Yana River has a more continental setting with δ 18 O values for late Holocene ice wedges of −29 ‰ [38].
The core portion above the massive ground ice (< 0.7 m depth) shows a similar isotopic signature like lake water. Presumably, the same source, a mixture of snowmelt and rain, is assumed to feed the lake and temporarily infiltrated the shore lines during Holocene time.
Water isotope signatures of the three Big Lake samples from different water depths (Fig. 8) reveal a well-mixed summer water body with similar isotopic composition and d-excess. A d-excess around 5‰ indicates an evaporative character of the lake system [22]. For comparison, the isotopic ratio of the river water is ∼ 7.5‰ higher for δD and ∼ 1‰ higher for δ 18 O with a lower d-excess (∼ 0.5‰ lower). This difference suggests a higher input from snowmelt to the lake when compared with the river water.

Conclusions
The BSC basin is partly covered by fluvial/alluvial sediments from the MIS 3 and the late MIS 1 periods and from lake sediments that have been deposited between 7600 and 6100 cal yr BP. The late Quaternary sedimentation history is summarised in three stages: (i) the mobilisation of fluvial/alluvial sediments along the lowland drainage pattern between 50 and 30 cal kyear BP; (ii) a thermokarst lake formation during the Holocene thermal maximum (between 7600 and 6100 cal yr BP), a first shrinkage of the thermokarst lake (Big Lake) after 7500 cal yr BP (expressed in a grain-size shift), and a second shrinkage after 6100 cal yr BP (absence of detrital sedimentation after 6100 cal yr BP); (iii) a re-mobilisation of fluvial/alluvial sediments along the lowland drainage pattern, an accumulation of peaty and organic-rich sediments, and the formation of ice-wedge polygons between 5700 and 1500 cal yr BP.
The sedimentation during the late Quaternary in BSC likely results from weathering and erosion of the underlying Palaeozoic bedrock. The thickness of Quaternary deposits has yet to be defined. This requires deeper geophysical investigations and deeper drilling. The basin nowadays acts as an intermediate sediment storage place before the detrital sediments are exported to the west and south where the crater rim is eroded.

3
frame of the German-Russian Expedition "Lena 2016" including services by the Hydrobase Tiksi. Ingmar Nitze kindly provided satellite imagery.

Conflict of interest There is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.