Seasonal deposition processes and chronology of a varved Holocene lake sediment record from Lake Chatyr Kol (Kyrgyz Republic)

A finely laminated lake sediment record with a basal age of 11,619 ± 603 years BP was retrieved from Lake Chatyr Kol (Kyrgyz Republic). Microfacies analysis reveals the presence of seasonal laminae (varves) from the sediment basis to ~360 ± 15 40 years BP. The Chatvd19 floating varve chronology covers the time span from 360 ± 40 years BP to the base and relies on replicate varve counts on overlapping petrographic thin sections with an uncertainty of ± 5 %. The uppermost non-varved interval was chronologically constrained by Pb and Cs γ-spectrometry and interpolation based on varve thickness measurements of adjacent varved intervals with an assumed uncertainty of 10 %. Six varve types were distinguished, are described in detail and show a changing predominance of clastic-organic, clastic-calcitic or -aragonitic, calcitic-clastic, 20 organic-clastic and clastic-diatom varves throughout the Holocene. Variations in varve thickness and the number and composition of seasonal sublayers are attributed to 1) changes in the amount of summer or winter/spring precipitation affecting local runoff and erosion and/or to 2) evaporative conditions during summer. Radiocarbon dating of bulk organic matter, daphnia remains, aquatic plant remains and Ruppia maritima seeds reveal reservoir ages with a clear decreasing trend up core from ~6,150 years in the early Holocene, to ~3,000 years in the mid-Holocene, to ~1,000 years and less in the late Holocene 25 and modern times. In contrast, two radiocarbon dates from terrestrial plant remains are in good agreement with the varve-based chronology. https://doi.org/10.5194/gchron-2019-18 Preprint. Discussion started: 20 January 2020 c © Author(s) 2020. CC BY 4.0 License.

of high-resolution paleo-climate archives and the partly problematic dating of these archives in this area. Information about Holocene climate variability in CA derive from several types of archives, including tree rings (Esper et al., 2003), speleothems (Fohlmeister et al., 2017;Wolff et al., 2017), ice cores (Aizen, 2004), aeolian deposits (Huayu et al., 2010) and lakes (Heinecke et al., 2017;Lauterbach et al., 2014;Mathis et al., 2014;Rasmussen et al., 2000;Ricketts et al., 2001;Schwarz et al., 2017). 35 However, none of these lake records has reported annually laminated sediments. In Kyrgyzstan, varves have been only reported from Lake Sary Chelek for the short time interval from ~1940´s to 2013 (Lauterbach et al., 2019). Other varved records in the wider region are from Lake Telmen in northern Mongolia which includes discontinuous varved intervals during the last ca. 4,390 cal years BP (Peck, 2002) and from Lake Sugan in north western China covering the last ~2,670 years BP (Zhou et al., 2007). Deciphering Holocene climate changes based on limnic records in CA is challenging due to the influences of several 40 factors: 1) chronological uncertainties caused by the scarcity of datable terrestrial plant material at high altitudes and often large 14 C reservoir effects of aquatic organic material (Hou et al., 2012;Lockot et al., 2016;Mischke et al., 2013), 2) human influence (Boomer et al., 2000;Mathis et al., 2014), possibly overprinting the natural climate signals in the archives and 3) variations in the dominance of the mid-latitude Westerlies, the Siberian High and the Asian Monsoon system leading to different spatial and temporal climate effects over CA Herzschuh, 2006;Mischke et al., 2017;Schroeter et 45 al., 2020). All these factors can hamper data comparison and may lead to different paleo-environmental interpretations Hou et al., 2012;Mischke et al., 2017). The investigation of varved lake sediments offers the unique opportunity for independent dating through varve counting. In addition, the description of varve micro-facies has the potential to provide detailed insights into environmental and climate variations at a seasonal scale. The sediment record from Lake Chatyr Kol is the first varved record from CA covering most of the Holocene and the main goal of this study is to establish a robust age 50 model through an integrated dating approach primarily based on varve counting. Varve counting requires an in-depth understanding of seasonal deposition of all varve types occurring in the sediment record. Therefore, we apply continuous microfacies analysis for the entire sediment profile to describe the Holocene evolution of varve formation in detail and discuss fundamental deposition processes.

Study site
Lake Chatyr Kol (40°36' N, 75°14' E) ( Fig.1) is located at ~3,530 m above sea level (a.s.l.) in the intramontane Aksai Basin (De Grave et al., 2011;Koppes et al., 2008) in the southern Kyrgyz Republic. In the north, the basin is restricted by the At Bashy Range and in the south by the Torugat Range resulting in a catchment area of about 1,084 km 2 . Geologically, the surrounding mountain ranges belong to Silurian to Carboniferous sedimentary-volcanogenic complexes of marine-continental 60 collision zones, consisting of limestones and dolomites, that crop out directly along the northern lake shore, as well as siliceous rocks, shales and scattered Permian granites that crop out in the south and north-east (Academy of Science of the Kyrgyz SSR, 1987). The modern lake, which has a maximum length of 23 km, a width of 10 km and a maximum depth of 20 m in its western-central part, is endorheic and separated from the neighbouring Arpa river basin in the north-west by a moraine (Shnitnikov, 1978). The moraine originates from glacial advances of unknown age from the western Torugat mountain range. 65 Present day glaciers exist above ~4000 m a.s.l. on the Torugat and At Bashy mountain ranges but only some of the At Bashy glaciers drain into Chatyr Kol via the Kegagyr River. The lake further receives convective rainfall in summer (Aizen et al., 2001). A shallow dam at ~ 3,550 m a.s.l. hinders outflow to the east. Modern climate conditions are generally dry and mainly controlled by the Westerlies and the Siberian Anticyclone Circulation (Aizen et al., 2001;Koppes et al., 2008). Mean annual precipitation is ~275 mm/a as indicated by Aizen´s (2001) evaluation and spatial averaging of annual precipitation of historical 70 records published by Hydrometeo (Reference Book of Climate USSR, Kyrgyz SSR, 1988). This is comparable to long-term instrumental data from nearby (about 50 km away) weather stations at comparable altitudes, where annual precipitation means are 237 mm/a (station "Chatirkul", 75°8´E, 40°6´N, 3,540 m a.s.l, AD 1961AD -1990 and 294 mm/a (station "TienShan", 78°2´E, 41°9´N, 3,614 m a.s.l., AD 1930AD -2000 (Williams and Konovalov, 2008). Monthly mean temperatures range from -26.0 to 8.0 °C (Koppes et al., 2008; Academy of Science of the Kyrgyz SSR, 1987) with means of -5.4 °C (station "Chatirkul") and -7.6 75 °C (station "TienShan") (Williams and Konovalov, 2008). The salinity of the lake water ranges from 1.06-1.15 g/l in the deeper western part of the lake to 0.24 g/l in the shallower eastern part near the inflow (Romanovsky, 2007). Measurements of oxygen concentrations (YSI Pro 6600 V2) during a field trip in July 2012 ranged from ~6 mg/l at the water surface to ~1 mg/l in 19 m depth with a clear oxygen minimum zone below ~11 m depth (Fig. 2). Water surface and bottom water temperatures (YSI Castaway CTD) at 19 m depth reached 13.2 °C and 9.4 °C. Specific conductivity (CTD) ranged from 1,902 80 µS/cm -1 at 19 m to 1,825 µS/cm -1 , pH was ~9 (YSI Pro 6600 V2). Secchi depth was about 4 m. Nowadays, the lake is largely occupied by the amphipod Gammarus alius sp. nov. (Sidorov, 2012) and no fish live in the lake (Shnitnikov, 1978).
The permafrost level is located at a depth of 2.5-3 m in the littoral coast zones and the lake is covered by ice from October to April (Shnitnikov, 1978). Modern permafrost thawing results in instable shores visible at the Maloye lake located < 2 km to the South of Chatyr Kol (Fig. 1 Photo) and the development of small ponds on the shallow south-western shore of this lake 85 and lake Chatyr Kol during summer. Several terraces in the north, south and east of the lake result from Pleistocene-Holocene lake level fluctuations (Romanovsky and Shatravin, 2007;Shnitnikov, 1978). Vegetation around the lake is generally poor and represented by high alpine meadows (Shnitnikov, 1978;Taft et al., 2011).

Coring and Composite profile
Five parallel cores each of 3 to 6 m length have been retrieved in 2012 from the deepest part of the lake (40°36.37' N, 75°14.02' E) by using an UWITEC piston corer (Fig. 1, Tab. 1). All cores were opened, split and photographed at GFZ Potsdam, where they are archived in a cold store at 4°C. A continuous composite profile of 623.5 cm length (CHAT12) was established by correlating the individual, overlapping cores via macroscopically visible marker layers (Fig. 3). Furthermore, seven parallel 95 gravity cores (SC17_1-7) have been retrieved with a UWITEC gravity corer in 2017 (Fig.1, Tab.1) to recover the undisturbed sediment-water interface, from which the best preserved parallel core SC17_7 was used for gamma spectrometric analysis.

Sediment microfacies analysis and varve counting
Continuous 10-cm-long sediment slabs with an overlap of 2 cm were taken from the whole composite profile to prepare largescale petrographic thin sections. Thin section preparation followed the method described by Brauer and Casanova (2001) and 100 included freeze-drying and vacuum impregnation of the sediment slabs with Araldite epoxy resin. Microfacies analysis, including a semi-quantitative evaluation of planktic and periphytic diatoms, aquatic plant remains (e.g. Potamogeton sp., Ruppia maritima), ostracods, daphnia, characeae and chrysophytes, was carried out on a Zeiss Axioplan microscope using different magnifications (25-400 x) and included measurement of varve thicknesses, microfacies/varve type characterization, the definition of varve boundaries and the development of process-related deposition models. A varve quality index (VQI) 105 ranging from 0-5 was given for each varve, comparable to the method from (Żarczyński et al., 2018) and references therein.
• VQI 0 = no varves or strongly disturbed varved sequences, no reliable counting (interpolation) • VQI 1 = very low varve preservation, horizontally discontinuous varve and less well-preserved sublayer boundaries, difficult counting • VQI 2 = low varve preservation, occasional horizontally discontinuous varve and sublayer boundaries, reliable 110 counting • VQI 3= medium varve preservation, horizontally continuous varve and sublayer boundaries, only small disturbances, reliable counting • VQI 4= high varve preservation, clearly distinguishable varve and sublayer boundaries, reliable counting • VQI 5= highest varve preservation, clearly distinguishable varve and sublayer boundaries, no disturbances, reliable 115 counting Varve counting was performed to establish a floating varve chronology. Non-varved intervals (VQI=0) between varved sediment sections were therefore interpolated by using the mean of sedimentation rates derived from about 20 varves above and below the non-varved part. Varves were counted twice by the same author. Counting uncertainty estimates were first 120 assessed by the percentage deviation of the second to the first count within one thin section. The mean of these deviations was used as an overall counting uncertainty estimate and assigned to the entire varved record. The uncertainty estimates were thus also assigned to interpolated sequences.

XRF element mapping
X-ray Fluorescence (XRF) element mapping was performed on two selected Araldite impregnated sediment blocks (ca. 2 x 10 125 cm), which were prepared for thin sections used for the microfacies analyses. XRF element mapping of these two sediment blocks allows linking micro-facies analyses of typical varve types directly with geochemical sediment compositions. Element mapping was performed at 50 µm resolution and covering most of the surface of the sediment block (15 x 100 mm) using a Bruker M4 Tornado at GFZ Potsdam. This scanner is equipped with a Rh X-ray source operated at 50 kV and 600 mA in combination with poly-capillary X-ray optics that irradiate a spot of 20 µm for 50 ms. After measuring and an initial spectrum 130 deconvolution, normalized element intensities are used to visualize relative element abundances as 2D maps.

Radiocarbon dating
In total, 36 accelerator mass spectrometry (AMS) 14 C measurements were carried out at the Pozńan Radiocarbon Laboratory 135 in Poland. Samples for 14 C measurements comprised two pieces of wood, bulk TOC samples, aquatic plant macro remains, daphnia remains and Ruppia maritima seeds (Tab.2). Additional samples of recent living daphnia and aquatic plants have been collected to assess the modern 14 C reservoir effect. The resulting conventional 14 C ages were calibrated using OxCal 4.3 (Ramsey, 2009) with the IntCal13 calibration curve (Reimer et al., 2013).

Gamma spectrometry dating 140
Gamma spectrometry measurements were performed on 0.5-cm-thick sediment slices that were continuously sampled from the upper 15.0 cm of gravity core SC17_7 (Suppl. Table. 1). The samples were freeze-dried and sieved through a 200 µm mesh for homogenization and removal of larger plant particles. Individual sample-aliquots were filled into gas-tight sealable lowactivity Kryal © tubes at identical fill heights and accurately weighted. After sufficient in-growth-time, the gamma energies of 210 Pb (T 1/2 = 22 a) and 214 Pb (T 1/2 = 26.8 min), which is a daughter nuclide of 222 Rn (T1/2= 3.8 d), were measured at 46.54, 145 295.24 and 351.93 keV. In addition, the gamma energies of 137 Cs (T 1/2 = 30.1 a) were measured at 661.66 keV. For this purpose, the Kryal© tubes were placed into shielded measurement chambers equipped with two well-type germanium detectors G1 and G2 (Canberra Industries) for ~1.5 to 7 days at GFZ Potsdam (Suppl. Tab. 1) (Schettler et al., 2006). Hardware control, data storage, and spectrum analysis were realized with the software Genie 2000 (Canberra Industries). The average counting uncertainty for 210 Pb was 5.9 %, for 214 Pb 7.7 % (295 keV) and 3.7 % (351 keV) and for 137 Cs 5.2 %. Efficiency calibrations 150 were carried out for 210 Pb, 214 Pb and 137 Cs with the same analytical setup using a lab-internal standard and the "Loess Nussloch" standard (Potts et al., 2003). Blank activities for 137 Cs were negligible while average 210 Pb blank activities of 10 mBq/g for detector G2 and 214 Pb blank activities of 9 mBq/g for the detectors G1 and G2 were considered. The activity measurements of 214 Pb were used to quantify the proportion of supported 210 Pb ( 210 Pb supp ) produced by the decay of 226 Ra in the sediment. The activity of unsupported 210 Pb ( 210 Pb unsupp ) in the sediment, which originates from the decay of 222 Rn in the atmosphere and 155 associated aeolian deposition, is quantified by the difference between measured 210 Pb total and 210 Pb supp . We selected sections that showed linear correlations in the semi-logarithmic plot of 210 Pb unsupp versus depth to infer average sedimentation rates using the constant initial concentration (CIC) model (c.f. Appleby, 2002) (Suppl. Tab. 2). Intercalated sediment sections showed nearly uncorrelated ln( 210 Pb unsupp ) vs. depth relationships at 10.25-9.25, 6.25-4.25 and 2.25-1.75 cm depth (Suppl. Fig.2). Therefore, the initial 210 Pb unsupp activities of samples that bridged these sections were used alternatively to determine 160 time intervals between these samples to infer a chronology. To assess possible changes of the sedimentation regime we additionally calculated sedimentation rates of each 0.5-cm-thick sediment slice using the CRS model (constant initial 210 Pb unsupp supply) (c.f. Appleby, 2002;Appleby and Oldfield, 1978) (Suppl. Tab.3).

Lithology 165
The composite profile can be subdivided into six lithological units (Fig. 3). Lithozone (LZ) I from 623.5 to 566.0 cm depth consists of greyish-brownish clastic-calcareous sediments. It shows mm-scale laminations of fine sandy and silty to clayey

Sediment microfacies analysis
Microscopic sediment analysis revealed, that clastic sublayers are present throughout the finely laminated sediments below 63.0 cm depth ( Fig. 4.1). These clastic sublayers are variably intercalated with calcitic, aragonitic and organic sublayers and thus form different types of cyclic successions. In total, we classified six different types of sublayer successions as described 180 below. The name for these types reflects the dominant sublayer for each of the six types. For example, the 'clastic-organic type' is characterized by the dominance of clastic sublayers, while in the organic-clastic type organic sublayers dominate. The names are not related to the order of sublayer succession within each type. Changing dominances of different sublayer successions reflect the lithozones.
These laminae exhibit the general pattern of clastic-organic laminae in LZ I, with a coarse-grained and thick basal detrital sublayer, but the overlying mixed (detrital calcite, mica, fsp, qtz and medium amounts of endogenic calcite) fine-grained sublayer additionally contains idiomorphic aragonite needles that are not found in clastic-organic varves. The sublayer 190 succession ends with an amorphous organic matter sublayer.

Calcitic-clastic type
The deposition of calcitic-clastic laminae (6 %) with a dominating endogenic calcite sublayer is restricted to LZ II. This subtype is composed of three sublayers and the mean thickness is 0.41 mm, with a maximum of 2.0 mm. Calcitic-clastic laminae ( Fig.   4.1b lower part, Suppl. Fig. 2b) are usually characterized by a basal detrital sublayer which, however, is not developed in all 195 calcitic-clastic laminae. The overlying sublayer generally exhibits low species abundancies of diatom frustules, chrysophyte cysts, aquatic plant remains, daphnia, ostracods and characeae but massive and fine-grained endogenic calcite, which is not the case in the clastic-calcitic laminae subtype. Endogenic calcite formed in the water column ( Fig. 4.2a) is recognized by its well-developed idiomorphic rhombohedral shapes. Scattered detrital grains occasionally occur within the endogenic calcite matrix. One depositional cycle ends with an amorphous organic matter sublayer. 200

Clastic-diatom type
Clastic-diatom laminae (20 %) occur in LZ II, III and IV. This subtype is composed of three sublayers and the mean thickness vary between 0.28 mm (LZ II), 0.34 mm (LZ III) and 0.35 mm (LZ IV). The depositional cycle starts with a basal detrital sublayer, which is overlain by a finer-grained mixed sublayer (detrital calcite, mica, fsp, qtz) occasionally containing chrysophytes and different diatom taxa. The third sublayer is formed by diatom blooms exclusively consisting of the planktic 205

Clastic-calcitic type
The second most common laminae subtype (23.5 %) are clastic-calcitic laminae ( Clastic-calcitic laminae exhibit a basal detrital sublayer with a sharp lower boundary, which is followed by a bloom layer of chrysophytes and/or diatoms, occurring sporadically after and/or within the detrital sublayer. The third, overlying mixed sublayer contains medium amounts of endogenic as well as fine-grained detrital calcite ( Fig. 4.3a), as well as mica, fsp and qtz grains but low amounts of diatom frustules and chrysophyte cysts. One depositional cycle typically ends with an amorphous 215 organic matter sublayer. In LZ V, these clastic-calcitic laminae occasionally contain a very fine-grained, light greyish, micritic sublayer before the cycle ends with the amorphous organic sublayer.

Organic-clastic type
Horizons of organic-clastic laminae (5.2 %) with dominating organic sublayers are mainly present within LZ IV and V ( Organic-clastic laminae exhibit an often horizontally discontinuous basal detrital sublayer (lens-shaped) in LZ IV, which is overlain by a mixed sublayer that contains detrital calcite, mica, fsp and qtz grains and many aquatic plant remains and periphytic diatoms (Achnanthes brevipes, pers. comm. Anja Schwarz, TU Braunschweig), whose colony chains are often preserved. One deposition cyclic ends with a yellowish amorphous organic matter layer. 225

Clastic-organic type
Clastic-organic laminae are present in all lithozones and most abundant in the record (42.5 % of all observed and measured laminae). This microfacies type is composed of four sublayers of which the most prominent is a basal clastic-detrital sublayer with a sharp lower boundary. The basal detrital sublayer contains mainly detrital calcite, which is distinguished from endogenic calcite by microscopic analyses. Detrital calcite is characterized by irregularly shaped gains and generally larger grain sizes of 230 average of 0.6 mm and up to 1.82 mm in LZ I. Detrital layers further contain siliciclastic minerals as mica, quartz (qtz) and feldspars (fsp). This basal layer is often, but not regularly overlain by chrysophyte and/or diatom blooms and a third, mixed sublayer containing mainly fine-grained detrital calcite, mica, fsp and qtz with low amounts of endogenic calcite and varying amounts of diatom frustules, chrysophytes, characeae, ostracods and daphnia. The deposition cycle ends with a yellowish layer of amorphous organic material. 235 The mean thickness of clastic-organic laminae differs between the lithozones. In LZ I, the mean thickness is 0.59 mm with a maximum thickness of 3.1 mm. In LZ I, the basal detrital sublayer is thick and coarse-grained, rich in pyrite, and contains mainly silt-to fine sand-sized grains and occasionally sand-sized qtz, calcite and fsp grains, whereas diatoms and chrysophytes are rare. In LZ II and III, clastic-organic laminae are less thick with a mean thickness of 0.27 mm and 0.48 mm respectively.
In these lithozones, the basal sublayer contains no sand-sized particles. In LZ IV, mean varve thickness is 0.43 mm and the 240 basal sublayer is often lens-shaped and horizontally discontinuous. In LZ V between 130.0 and 63.0 cm depth thickest clasticorganic laminae occur with a mean thickness of 1.5 mm and a maximum thickness of up to 7.0 mm ( Fig. 4.1e, Suppl. Fig. 2f).
These clastic-organic laminae often include an additional detrital sublayer intercalated in the finer grained mixed sublayer.

Homogenous sediments
The uppermost 41.0 cm of the sediment record consist of homogenous sediments, containing a fine-grained mix of 245 autochthonous and allochthonous calcite, mica, qtz and fsp. The sediments are generally rich in organic remains, such as aquatic plant remains, chrysophytes, diatoms and chlorophytes (Botryococcus). Faint and discontinuous calcite laminae occur in the uppermost centimeter ( Fig. 4.1f).

XRF element mapping
The two selected impregnated sediment blocks from 507 to 497.5 cm (XRF-Map 1 Fig. 4.2) and from 346.5 to 338.5 cm depth 250 (XRF-Map 2 Fig. 4.3) contain calcitic-clastic, clastic-diatom, clastic-calcitic and clastic-organic microfacies types. These sediments are dominated by alternating calcitic and siliciclastic sediments represented by the elements Ca, Sr, Mg and Si, Al respectively ( Fig. 4.2 and 4.3). Color variations of the element maps show that the calcitic and siliciclastic sediments are clearly separated in sample the XRF-Map 1 (Fig. 4.2) but slightly more mixed in the XRF-Map 2 ( Fig. 4.3). In both XRF-Maps 1 and 2, the carbonate sublayers are enriched in Sr (Fig. 4.2

Chronology of the non-varved uppermost sediments 280
The uppermost 63.0 cm of the sediment profile are not varved and thus require alternative dating approaches including 210 Pb dating, activity profiles of 137 Cs and sedimentation-rate based interpolation. First, we measured 210 Pb activity concentrations of the uppermost 15 cm of short core SC17_7 and applied the CIC and CRS models (Fig. 5c & 6, Suppl. Tab. 3). SC17_7 is correlated to the composite profile through macroscopically visible facies change at 1.0 cm composite depth and through a laminated section from 45.0-41.0 cm composite depth (Suppl. Fig.1). The CIC and CRS-model based chronologies are broadly 285 consistent and particularly date sediments at 8.75 cm (SC17_7) or 7.5 cm composite depth to AD 1945/46 ( Fig. 5c and Fig.   6b, Suppl. Table 3,). This coincides with the onset of increased 137 Cs activity concentrations (Fig. 5c, 6c) marking the onset of nuclear weapon testing in AD 1945(Ferm, 2000Kudo et al., 1998;Norris and Arkin, 1998). Therefore, we applied the date

Radiocarbon dating
In total, we dated 36 samples of bulk organic carbon, daphnia remains, aquatic plant remains and Ruppia martima seeds. Only 300 two samples were terrestrial plant remains (wood fragments) and sufficiently large to be used for AMS 14 C dating (Tab. 2, Fig.   5 & 7). Except the two ages from terrestrial plant remains (Poz-54302 with 9988 ± 203 and Poz-63307 with 6140 ± 137 cal yr BP), all other ages deviate from the varve chronology between 155 years at 0.0 cm depth and 6,150 years at 585.0 cm depth (Fig. 7). We observe a general trend of decreasing deviations up core with the maximum deviation of ~6,150 years at 585.0 cm depth in LZ I. Looking at more detail, the deviations between radiocarbon and varve ages exhibit a prominent step-wise 305 increase particularly at the boundary between lithozone LZ IV and LZ V when it abruptly decreases from ~3,000 years to ~1,000 years. Modern aquatic plants collected during the field campaign in 2012 showed large modern reservoir ages of 330 ± 30 and 2425 ± 25 14 C years and living daphnia yielded ages of 225 ± 30 14 C years.

Interpretation of fine laminations as varves 310
The construction of varve chronologies relies on the proof of seasonal origin of fine laminations (Brauer et al., 2014;Ojala et al., 2012;Zolitschka et al., 2015). Laminations are absent in the upper 63.0 cm of the Lake Chatyr Kol sediment core and cyclic successions of mixed-clastic laminations are only observed below this depth. Therefore, the seasonal origin of the Chatyr Kol sediments cannot be proofed through modern observation in sediment traps because no varves are formed at present day.
Instead, we applied process-related deposition models (Fig. 4.1) based on detailed micro-facies analyses obtained from 315 petrographic thin sections and compared our observations with varve types described in literature (Brauer, 2004;Zolitschka et al., 2015). We associate the observed successions of different types of mixed-clastic laminations with the formation of different seasonal sublayers that are known from lakes with carbonaceous catchments (Brauer and Casanova, 2001;Kelts and Hsü, 1978;Lauterbach et al., 2011;Lauterbach et al., 2019) and high-altitude glacial environments (Guyard et al., 2007;Leemann and Niessen, 1994). The observed succession of sublayers are interpreted as mixed varve types (clastic, -organic and -320 endogenic) as defined by Zolitschka et al. (2015).
Varve formation at Lake Chatyr Kol is related to the high seasonality of the local climate with an ice cover during winter as well as strong annual variations of the temperature and precipitation affecting productivity, endogenic carbonate formation and local runoff. Varve preservation is promoted by the unique morphology of the deep western lake basin, where anoxic bottom water conditions can be maintained even under relatively low lake levels (Fig.2). 325 We interpret the annual sedimentary cycle to always start with the deposition of a basal detrital sublayer with a sharp lower boundary which results from winter/spring snow and/or glacial melt (Guyard et al., 2007;Leemann and Niessen, 1994;Zolitschka et al., 2015) after the ice break-up in ~April (Shnitnikov, 1978). Runoff with suspended sediment load is then likely directed through the Kegagyr River in the east but may also be the result of surface runoff through the activation of several widely distributed smaller tributaries in the catchment (Fig. 1). 330 Basal detrital sublayers are generally overlain by blooms of chrysophytes and/or diatoms within clastic-organic, organic-clastic, clastic-diatom and clastic-calcitic varve types. Chrysophytes and/or diatom blooms develop in consequence of available nutrients provided by runoff and spring overturn in combination with rising temperatures during the summer season (Zolitschka et al., 2015). The productive phase in calcitic-clastic varves is however reflected by calcite precipitation, which is the main carbonate phase (endogenic, detrital and resuspended) in the Chatyr Kol sediments. The formation of endogenic 335 calcite in Lake Chatyr Kol is controlled by: 1) photosynthesis, when high aquatic productivity lowers the concentrations of CO 2 , increases the pH of the lake water and leads to a reduced solubility of CO 3 2- (Hodell et al., 1998;Kelts and Hsü, 1978;Zolitschka et al., 2015), 2) evaporation leading to an oversaturation of carbonate ions, 3) sufficient supply of dissolved cations either through surface runoff or groundwater inflow (Shapley et al., 2005). Changes in weathering and hydrological conditions can lead to variations in the supply of Ca 2+ and Mg 2+ ions and subsequently change the Mg/Ca ratio of the lake water (Müller 340 et al., 1972). The formation of aragonite requires high lake water Mg/Ca ratios (>12), whereas magnesium-calcite forms at lower Mg/Ca ratios (Kelts and Hsü, 1978;Müller et al., 1972). XRF element intensity maps do not provide quantitative results but do indicate, that Mg is abundant in the XRF map 1 from LZ II which results in the formation of endogenic Sr-and Mgrich calcites (Fig. 4.2 a).
Aragonite precipitates, related to an evaporative concentration in summer, were only microscopically observed in the 345 intervals between 600.0-605.0 and 609.0-616.0 cm composite depth suggesting that Mg/Ca ratios probably remained above Mg/Ca ratios >12.
After the spring to early summer lake productivity, the deposition of a mixed sublayer consisting of silt-to clay-sized detrital grains and low to medium amounts of endogenic calcite is observed in all lamination types (Fig. 4.1, Suppl. Fig. 2). In clastic-calcitic varves, the mixed sublayers appear different and include especially resuspended calcite as evidenced also in Ca 350 and Sr intensities (Fig. 4.1c, Fig. 4.3a, Suppl. Fig. 2d). The mixed sublayer indicates resuspension of shore material (littoral calcite) to the core location due to e.g. wind induced wave activity and weak runoff during the ice-cover free season from ~April to October (Shnitnikov, 1978).
The intercalation of discrete detrital layers within the mixed sublayer ( Fig. 4.1e, Suppl. Fig. 2e), as observed in clasticorganic laminae in LZ V, indicates pulses of runoff of suspended material which may be caused by late rainfall events in 355 summer (Aizen et al., 2001;Shnitnikov, 1978).
One annual depositional cycle usually ends with the deposition of a thin sublayer of very fine amorphous organic matter which is deposited under quiet water conditions when the lake was ice covered (Fig. 4.1). In lithozone V, an additional micritic sublayer is deposited before the amorphous organic sublayer at the end of the seasonal cycle in clastic-calcitic laminae when water turbulence is low. 360

Varve counting and chronology construction
The interpretation of different types of fine laminations allowed varve counting as a main tool for constructing the Chatyr Kol chronology largely based on incremental methods. Around 80% of the varves in the sediment record are double-counted in petrographic thin sections while the remaining part of ca 20% had to be interpolated based on sedimentation-rate estimates due to poor varve preservation. The resulting chronology comprises 11,259 years and is anchored to the absolute time scale at 63.0 365 cm sediment depth supported by a combination of lead-210 dating and occasional sedimentation rate measurements as described below. The resulting age-depth model is within uncertainties in good agreement with two calibrated AMS 14 C dates of wood pieces at 380.5 cm depth (6,140 ± 137 cal years BP; Poz-63307) and at 528.0 cm depth (9,988 ± 203 cal years BP; Poz-54302) (Fig. 5b, Tab. 2). The corresponding varve-based ages are 5,905 ± 320 years BP and 9,611 ± 505 years BP, respectively. As for all chronologies, uncertainties are inherent also to varve chronologies, which are commonly assessed via 370 replicate counts (Brauer and Casanova, 2001;Lamoureux, 2001;Lotter and Lemcke, 1999;Ojala et al., 2012;Żarczyński et al., 2018;Zolitschka et al., 2015). However, there is no standard procedure on how to calculate and present the uncertainties (Ojala et al., 2012;Zolitschka et al., 2015). Commonly, mean values of replicate count differences, the difference of maximum and minimum counts or their standard deviation are reported (Brauer et al., 2014;Ojala et al., 2012;Żarczyński et al., 2018;Zolitschka et al., 2015). Despite the inevitable increase of cumulative uncertainties with age or depth, systematic uncertainties 375 arise and are caused by changes in varve preservation, strongly and abruptly varying sedimentation rates and the challenging differentiation of varve types with complex structures (Ojala et al., 2012;Żarczyński et al., 2018;Zolitschka et al., 2015). The overall very small difference between the two counts of the Chatyr Kol varved record of only -71 varves is due to the compensating effect between over-and underestimations of varve counts throughout the record. For the floating varve chronology, we therefore compare the results for each individual thin section comprising between a maximum of 324 (506.8-380 497.6 cm) and a minimum of 13 (varves) (65.4-63.0 cm) (Fig. 5a, Fig. 8).
Counting uncertainties for individual thin sections are reported as their percentage deviation from the first count used for the chronology and range between 0 and 23.7 % (Fig. 5a, Fig. 8 8). For the total uncertainty estimate for the floating varve chronology we use the mean of ± 5 % calculated from the uncertainties for each 10 cm interval. This conservative estimate is more realistic than the very low difference in the two repeated varve counts. An uncertainty of 5 % is in the range of elsewhere reported varve chronologies (Ojala et al., 2012).
Since the upmost 63.0 cm of the sediment profile are largely homogeneous, the varve chronology is floating and needs to be 395 anchored to an absolute chronology at this point. The interpolation with a mean SR derived from the combination of the consistent CIC and CRS 210 Pb marker (AD 1945), the SR derived from discontinuously varved sequences between 41.0 and 63.0 cm depth and from 100 measured varves below 63.0 cm depth seems to be the best approach for constraining the uppermost age-depth relationship within the homogenous sediments, where further chronological markers are lacking. We are aware, that the interpolation-based "floating" anchor point at 63.0 cm depth is prone to additional uncertainties. We considered 400 this by assuming a higher uncertainty of 10 % for this interval, than that of ± 5 % for the floating varve chronology.

Radiocarbon reservoir effects
Compared to the floating varve chronology, including two terrestrial (wood) AMS 14 C dates, we observed a general trend of decreasing reservoir effects of dated aquatic material up core with the maximum deviation of ~6,150 years at 585.0 cm depth (10,930 ± 570 years BP) in LZ I (Fig. 7). The step-wise decrease of deviations between radiocarbon and varve ages is most 405 pronounced at the boundary between lithozone LZ IV and LZ V, when it abruptly decreases from ~3,000 years to ~1,000 years.
The reservoir effect generally depends on the rate of atmospheric CO 2 exchange between the water column and the air, internal mixing dynamics and the input of 14 C depleted carbonaceous material (Ascough et al., 2010;Jull et al., 2013;Keaveney and Reimer, 2012;Lockot et al., 2016;MacDonald et al., 1991). The catchment of Lake Chatyr Kol exhibits several sources that could be responsible for a 14 C-depletion of dissolved carbon species in the lake water. Highest reservoir ages in the early 410 Holocene are likely the result of the combined influence of these sources: 1) the input of old, 14 C-depleted CO 2 with glacial meltwater (c.f. Hall and Henderson, 2001) at the onset of a warming Holocene and 2) the weathering and erosion of the northern outcropping limestones, which led to the release and input of dissolved bicarbonate to the lake (c.f. Abbott and Stafford, 1996;Hutchinson et al., 2004). Both processes lead to a 14 C-depleted CO 2 and HCO 3uptake during photosynthesis by e.g. submerged aquatic plants like Ruppia martima at 585.0 cm depth (Fig.7) and by phytoplankton, on which daphnia feed and which therefore 415 also show similar reservoir effects. A high detrital input and thus a potentially high input of dissolved bicarbonate is supported by increased varve thicknesses during the early Holocene (Sect. 5.4.1, Fig. 9). Furthermore, 3) thawing of permafrost since the beginning of a warming Holocene might have released dissolved 14 C-depleted organic material and thus affect the 14 C TOC bulk measurements. Our fieldtrip observations and observations by Shnitnikov (1978) of modern permafrost reduction and the development of thermokarst in the southern part of the catchment around the neighbouring Lake Maloye (Fig. 1) support this 420 assumption. The cause of a step-wise reservoir effect reduction is therefore likely also related to the combined effect of a generally decreasing glacial influence and a decreasing input of bicarbonate until ~AD 1150 at the boundary between LZ IV and LZ V (Fig.7, 9). The abrupt decrease of the reservoir effect after ~AD 1150, despite an increase in detrital carbonate supply (Sect. 5.4.5, Fig. 9) might be related to the silting up of the basin leading to a shallower water depth, which is more susceptible to water circulation and an enhanced atmospheric CO 2 exchange (c.f. Geyh et al., 1997). 425

Holocene variations in varve microfacies
The Lake Chatyr Kol sediment profile comprises six different varve types (Sect. 4.2, Fig. 4.1, Suppl. Fig. 2), which occurrences showed varying dominances in the different lithozones that are described below. The individual lithozones always comprised more than one varve type ( Fig. 9) with a maximum of five different varve types occurring in LZ III and II to three varve types in LZ V. were observed in all varve types in this LZ, are indicative for intense runoff by winter/spring snow meltwater and/or by glacial thawing during summer (Shnitnikov, 1978) caused by highest insolation (Berger and Loutre, 1991;Chen et al., 2008;Jin et al., 2011;Li and Morrill, 2010) at the onset of a warming early Holocene. Glaciers of the inner Tian Shan started to retreat between ~12-8 ka years BP (Bondarev, 1997;Shnitnikov, 1978) causing enhanced detrital input into the lake. The low species abundancies of aquatic plants (Ruppia Maritima or Potamogeton sp.), daphnia and characeae reflect a littoral community and 440 indicate a low aquatic productivity and a relative low lake level during this time. Clastic-calcitic varves appear at the base of the composite profile and towards the end of LZ I, whereas clastic-aragonitic varves dominate in the period from ~11,500 to 11,000 years BP. Most likely, idiomorphic aragonite formed due to a combination of Mg-rich water supply to the lake and strong evaporative conditions causing lake water Mg/Ca ratios of >12 (Kelts and Hsü, 1978;Müller et al., 1972). In this lithozone calcitic-clastic varves constitute about 21 % of the observed varves and clastic-calcitic varves ~22 %, while clastic-organic varves make up ~42 % and clastic-diatom varves ~15% (Fig. 9 LZ II). This lithozone is characterized by intercalations of calcitic varve types (calcitic-clastic & clastic-calcitic) with clastic-organic and clastic-diatom varves ( Fig.   4.2). The variations of these varve types might be related to external (climatic) forcing or lake-internal or sedimentation 450 variability (Turner et al., 2016 and reference herein). XRF element maps show endogenic calcite sublayers that are enriched in Sr and Mg alternating with clastic (Si and Al) or diatom (Si) layers ( Fig. 4.2) suggesting Sr-and Mg-rich calcite in this lithozone which indicates evaporative concentration (Müller et al., 1972). The shift of the dominant endogenic carbonate type from aragonite in LZ I to calcite in LZ II around ~10,730 years BP ( Fig. 9 LZ II) coincides with an increase in biological and photosynthetic activity, as inferred from the establishment of a diverse lake fauna seen in high abundancies of chrysophytes, 455 planktic (Cyclotella choctawhatcheeana) and periphytic diatoms (Achnanthes brevipes) as well as of aquatic plants, ostracods, characeae and few daphnia. Identifying the main drivers controlling endogenic carbonate formation thus remains speculative.
Species assemblages and associated biological activity during the summer season indicates favourable warm summers and sufficient nutrient supply through e.g. runoff. Because the species assemblages show mixed littoral and pelagic species abundancies, these are interpreted as an indication for a low lake level. The deposition of clastic-calcitic varves comprise ~38 % of the observed varves in this lithozone, while clastic-organic varves make up ~27 % and clastic-diatom varves ~34 % ( Fig. 9 LZ III). Clastic-calcitic varves are generally thicker than the other varve types of this LZ mainly because of exceptionally thick summer sublayers. These summer layers consist of endogenic 465 calcite mixed with resuspended calcites and fine-grained detrital grains (Fig. 4.1c, Fig. 4.3 a). This is confirmed by elevated Sr values of the XRF element mapping, indicating the presence of Sr-rich carbonates. These varves likely reflect increased resuspension of carbonates from the littoral zone due to wind induced wave activity. As in LZ II, alternations of clastic-calcitic (Ca, Sr) (Fig. 4.3 a) and clastic-diatom, clastic-organic (Si, Al) (Fig. 4.3 b) varves are characteristic for this lithozone as well.
At ~8,040 years BP the deposition of calcitic-clastic varves ceased and is replaced by the deposition of clastic-calcitic, clastic-470 organic and clastic-diatom varves probably due to decreasing summer insolation and lower summer temperatures (Berger and Loutre, 1991;Chen et al., 2008;Jin et al., 2011;Li and Morrill, 2010). In addition, higher lake levels since that time are indicated by the dominance of planktic diatoms and the occurrence of lake deposits at the eastern and southern shore which have been dated from 6,688 ± 473 to 4,621 ± 594 cal years BP ( 14 C ages published by Shnitnikov (1978) calibrated with OxCal4.3 & IntCal13). During our field work we found lake sediments on a shallow terrace ~7 m above the current lake level 475 east of the lake which also revealed a mid-Holocene age of 5,786 ± 122 cal years BP (Poz-109830 Tab.2). Higher lake levels at that time have been explained by the preceding early Holocene glacier retreat in the catchment (Bondarev, 1997;Shnitnikov, 1978). However, more recently even minor glacial advances in the Aksai Basin east of the Chatyr Kol catchment 10 Be exposure dated between 7.5 and ~4.5 ka were reported (Koppes et al., 2008). Thicker summer sublayers result from thicker mixed sublayers rich in algae remains (Botryococcus, chrysophytes, diatoms) 500 and additional late summer detrital sublayers (Fig. 4.1.e, Suppl. Fig. 2f). Hence, the increase in summer layer thickness suggest both, higher lacustrine productivity and an increase in summer runoff events. However, the reasons for these changes remain elusive and a relation to known climatic periods like the Medieval Climate Anomaly and the Little Ice Age is not found. One might speculate that the frequent occurrence of late summer runoff layers either reflects convective rainfall events due to recycling of local moisture sources (Aizen et al., 2001) or changing atmospheric circulation regimes. Changes in boundary 505 conditions in the catchment of the lake are unlikely since microfacies analyses does not show pronounced changes in grain size distribution of the detrital material. Human impact cannot fully be excluded but low indices of human and livestock fecal biomarkers (Schroeter et al., 2020) are an argument against major human impact. The presence of lake deposits at the northern and southern shores ca 1.5 -1 m above present day lake level dated at AD 1420 ± 204, AD 1044 ± 160 and AD 858± 166 (Shnitnikov, 1978) suggests that increased summer runoff events might have resulted in a more positive water budget and lake 510 level rise. At around AD 1730 ± 30 varve formation and/or preservation ceased and sediments became predominantly homogeneous. The cessation of varves might be related to enhanced mixing of the water column resulting in a loss of the oxygen minimum zone 515 ( Fig. 2a) caused by decreasing water depth due to silting-up, which accelerated with the abrupt increase in sedimentation rate at AD 1150 and/or due to strengthening of the wind conditions and wave activity.

Conclusion
We present the first varved lake sediment record in arid Central Asia that covers almost the entire Holocene. The established floating varve chronology provides an independent dating for a setting with scarce material for radiocarbon dating. 520 In particular, our varve chronology allows a quantification of changes in radiocarbon reservoir ages throughout the Holocene.
The largest reservoir effect of ~6150 years in the early Holocene is likely caused by glacial melt and enhanced local erosion resulting in a surplus of dead carbon. Lowest reservoir ages of ~1,000 years and less in the late Holocene might be related to enhanced atmospheric CO 2 exchange when the lake was shallower due to silting-up of the lake basin and/or increased windiness inducing increased water column mixing and CO 2 exchange with the atmosphere. The construction of the varve-525 based chronology was only possible through detailed micro-facies analyses of the entire sediment sequence in overlapping thin sections that allowed the development of seasonal deposition models for all observed types of fine laminations. Based on these models and their comparison with published varve micro-facies data, we interpret all six Chatyr Kol lamination types as varves. Compared to many other varved lake sediment records, the Chatyr Kol varves are very heterogeneous and a complex pattern of six different micro-facies types developed throughout the Holocene. All varve types are predominantly clastic and 530 comprise variations of their summer sublayers with changing dominances of organic, diatom, calcitic, aragonitic and additional detrital sublayers. Varve thickness changed accordingly with the varve micro-facies types, whereby the most conspicuous increase of varve thickness occurred at AD 1150 which is caused by increased erosion and runoff. The increase in detrital input into the lake further caused an acceleration of the silting-up processes.
XRF element mapping results support our microfacies analysis and provide additional information on the composition 535 of carbonate sublayers and detrital carbonate. Microfacies analysis and XRF element mapping show major variations between partly Mg and Sr rich sublayers in calcitic-clastic and clastic-calcitic varves with Al and Si rich sediments in clastic-organic and clastic diatom varves. Nevertheless, the complex succession and variations of varve types throughout the Holocene including major change points still requires further detailed investigations and interpretation together with other proxy data.
Y. Beutlich are thanked for help with the geochemical analyses. S. Orunbaev, M. Daiyrov, S. Kalmuratov, G. Omurova and K, Jusupova are acknowledged for their support during field trips. We further thank T Goslar for AMS 14 C dating and D. 555 Berger, G. Arnold and B. Brademann for thin section preparation and G. Schettler for his help with lead-210 dating. We also want to thank Rik Tjallingii for conducting µXRF element mapping and his help with the revision of the manuscript. This paper is a contribution to Topic 8 'Rapid Climate Change from Proxy data' within the climate initiative REKLIM of the Helmholtz-Association.

Funding 560
This study was conducted in the framework of CAHOL (Central Asian HOLocene), a subproject of the joint project CAME