Multi‐proxy climate and environmental records from a Holocene eutrophic mire, southern taiga subzone, West Siberia

Palaeoenvironmental reconstructions from peat are strongly focused on ombrotrophic mires, but this study demonstrates that eutrophic mires can also be used. A multi‐proxy approach was applied to a eutrophic mire on a floodplain terrace in the southern taiga of West Siberia. The results of the reconstruction were considered in the wide geographic context of the surrounding regions, including Siberia and Central Asia. Different palaeoecological proxies (analysis of plant macrofossils, testate amoebae, oribatid mites, molluscs, peat humification, ash content and spectral characteristics of humic acids) were used in this study. The results of different proxies showed a high level of consistency among themselves, which allowed for a robust interpretation of Holocene mire development. Throughout the ~7800 years history of the mire, there was a high level of surface wetness. The presence of mineral matter in the peat between 7800 and 5100 cal. a BP indicates regular flooding caused by the intensive fluvial activity, apparently resulting from increased precipitation. This was followed by a trend towards a gradual decrease in surface wetness from conditions of high surface moisture (stagnant water) between 5100 and 3000 cal. a BP to present day conditions of moderate surface moisture with a water table slightly below the mire surface. This pattern is consistent with the well‐documented long‐term trend from palaeoecological records throughout the taiga and arctic zones in West Siberia and central arid Asia. Our data further support the idea that the westerlies were the dominant driver of climate for the southern taiga of West Siberia during the Middle to Late Holocene.

Palaeoclimatic information for periods during the Holocene that are warmer and wetter than present day conditions can serve as a climate reference for understanding and predicting extreme meteorological fluctuations in the future (IPCC 2014). As B€ ohner (2006) pointed out, considering the patterns of climatic change in a wide geographical context allows a deeper understanding of their nature and mechanisms of formation in individual regions.
Indeed, data from various reconstructions from different regions in Asia are now available and allow us to study their palaeoclimatic history. In particular, reconstructions with high temporal resolution and covering the entire Holocene have now been carried out for the territory of Central Asia (Chen et al. 2008(Chen et al. , 2019, including reviews of individual regions of Mongolia (Wang & Feng 2013;Klinge & Sauer 2019) and the Altai Mountains (Zhang & Feng 2018) and Siberia (Mikhailova et al. 2021). The experience of synthesizing data from different reconstructions performed using different proxies (pollen, diatoms, ostracods, grain size, stable isotopes, etc.) has shown that to compare them with each other, it is effective to use a semi-quantitative estimate in the form of calculating a regional index for the key palaeoclimate parameter (Chen et al. 2008;Wang & Feng 2013;Zhang & Feng 2018). This approach helps to unify heterogeneous data and identify similar patterns, as well as differences between various reconstructions. Meanwhile, it is important to note that for the taiga and arctic territories of Siberia, the available reviews of palaeoclimatic trends and of vegetation distribution (Velichko et al. 1997;MacDonald et al. 2000;Blyakharchuk 2009;Binney et al. 2017) were carried out mostly 10-20 years ago and include sparse reconstructions with a rather low temporal resolution. In general, for this territory, we lack estimates of regionally averaged moisture indices, as was done for arid regions of Central Asia (Chen et al. 2008), and one of the aims of this study was to provide a regional overview of moisture variations during the Holocene for the southern taiga subzone of West Siberia.
Peatlands are natural archives (Godwin 1981) that provide excellent opportunities for reconstructing palaeoenvironmental and palaeoclimatic conditions. Understanding the response of peatlands to climate fluctuations during the Holocene can provide insight into possible responses of these sensitive ecosystems to ongoing and future climate change. In the southern part of West Siberia, small peatlands are widespreadboth eutrophic and ombrotrophic (Kremenetski et al. 2003). In general, there are fewer palaeoecological studies of minerotrophic peatlands than of ombrotrophic ones (Lamentowicz et al. 2013;Bao et al. 2018). Unlike ombrotrophic bogs, the hydrological state of a minerotrophic mire, in addition to atmospheric precipitation, is affected by more environmental factorssurface runoff from adjacent territories, groundwater, intra-mire hydrological networkof flows, seepage from the mire, the impact of a nearby river through flooding and/or enhanced drainage. Therefore, it is more difficult to distinguish climate-driven changes in the peat of a minerotrophic mire (Barber 1993;Apolinarska & Gałka 2017). In addition, the preservation of bioindicators is often worse in minerotrophic peat than in ombrotrophic peat (Pyavchenko 1972;Barber 1993). Nevertheless, recent studies (Gałka et al. 2018;Dobrowolski et al. 2019;Blaus et al. 2020) demonstrate the responsiveness of minerotrophic peat deposits to palaeoclimate fluctuations and their successful use for reconstructions. We believe that the success of these works can mostly be explained with the use of a multiproxy approach, where different proxies reveal more details in the development of the peatland and help to separate the influences of different environmental factors, highlighting separately the influence of climate. Thus, Pawłowski et al. (2015) showed that geochemical indicators might display a rising groundwater table, river flood and alluvial deposition, while biological indicators better mirror environmental details on the mire surface. Payne (2010) concluded that reconstructions in minerotrophic mires are especially needed in areas where ombrotrophic bogs are rare or absent (arid and/or mountainous regions), and in addition, in those cases when the conditions of the Early Holocene are in focus, the time when modern bogs had not yet become ombrotrophic and were at the stage of a minerotrophic mire (fen and fen-bog transition).
During the development of a mire, the formation of its peat deposits is influenced by both autogenous factors (peat growth and decay, local hydrology, succession of mire vegetation) and allogenous factors (climate, basin topography, regional hydrology, anthropogenic disturbances; Lavoie et al. 2013). It can be difficult to unravel the relative contribution of each factor in the development of the mire. Previous research (Loisel & Garneau 2010) showed that synchronous uniform changes in different peat cores, collected from different sites in one region, were a consequence of allogenous factors (regional climate change). Asynchronous changes, however, were probably the result of influential local autogenous factors (Loisel & Garneau 2010;Lavoie et al. 2013). Separating individual autogenous mire development from regional climatic trends can be achieved by comparative reconstructions using peat cores obtained from different mires in the same region.
The aim of our study was to assess long-term trends of palaeoclimatic changes in humidity in the Holocene in the southern taiga subzone of West Siberia using a multiproxy study of a eutrophic mire in a river valley while considering the wide geographical context of neighbouring regions. The objectives in this study were to: (i) reconstruct the history of a eutrophic mire development based on a multi-proxy approach using analysis of plant macrofossils, testate amoebae, ash content, peat humification, spectral characteristics of humic acids (HA) and mollusc and oribatid mite remains; (ii) examine the influence of autogenous and allogenous factors on the development of the mire in the Holocene by comparing it with other regional records from the southern taiga of West Siberia; and (iii) consider changes in the mire during its development in the context of palaeoclimatic changes in the neighbouring regions of Siberia and Central Asia.

Study area and fieldwork
The research was performed in the southern taiga subzone of West Siberia, in the interfluve area between the Ob River and the Irtysh River, where mires cover approximately 36% of the territory (Alekseeva et al. 2015), a fifth of which are eutrophic terrace mires. The wide distribution of minerotrophic mires in this region is a result of the high mineralization of the groundwater and the alkalinity of the carbonates in the underlying rocks (Ufimtseva 1974). Minerotrophic mires are primarily located in floodplains and on river terraces. The study region is characterized by a continental climate (Kremenetski et al. 2003) with a mean annual temperature of À0.4°C. The mean temperatures of the coldest month (January) and the warmest month (July) are À19.3 and +18.1°C, respectively. The mean annual precipitation is around 465 mm a À1 (weather station 29328, 1934Bulygina et al. 2014Bulygina et al. , 2015. The eutrophic mire investigated in this study is located on the first floodplain terrace, along the gently sloping left bank of the Bakchar River (56.92°N, 82.51°E, elevation 105 m a.s.l.). It is an open mire, covering an area of about 120 ha and occupied by shrubs, brown mosses and sedges. In spring, meltwater from the upper terraces flows down the surface of the mire into the river, along a naturally formed system of mire streams.
The peat core (named S-2) was collected 100 m from the edge of the mire (56.9236°N, 82.5111°E; Fig. 1) closest to the river, to track the periods of river floods affecting the mire surface. The peat thickness at this place was 350 cm, overlying a layer of carbonate-rich loam and clay.
The mire at the coring location was occupied by a plant community dominated by sedges and herbs (Table 1). Mosses included mainly brown mosses and Sphagnum mosses. Higher areas of surface microtopography (up to 30 cm elevation) were occupied with shrubs, and occasionally birch trees (height 2-7 m) were observed.
Peat was cored with a Russian peat corer (chamber length 50 cm, diameter 5 cm) in September 2013. Core S-2 was cut into 35 continuous 10-cm-long sections. During processing, these sections were divided into subsamples for different analyses. Owing to the sampling resolution of 10 cm, results for radiocarbon, plant macrofossils, ash content, humic acids, testate amoebae, molluscs and oribatids should be considered with an uncertainty of AE5 cm from the centre of each slice. First, we took subsamples for radiocarbon dating and for microanalyses (ash content, peat humification, testate amoebae) that require less peat, precisely from the middle of each 10-cm slice; then we distributed the remaining peat for other analyses (plant macrofossils, humic acids, oribatids, molluscs) that require more peat. Subsamples were placed in plastic bags and stored in a refrigerator until examination.

Material and methods
The radiocarbon ( 14 C) dating of bulk peat samples, each consisting of 20 g of dry peat, was carried out via liquid scintillation counting using a Quantulus 1220 (Table 2). Dates were calibrated using rbacon version 2.5.3 Fig. 1. Geographical location of the study area, core S-2 marked with a star. A= West Siberia; B = southern taiga; C = satellite image of the studied mire. (Blaauw & Christen 2011) and an IntCal20 calibration curve (Reimer et al. 2020), and an age model was created using linear Bayesian modelling. In this work, samples are represented as calibrated age BP where present is defined as 1950 (Table 2).

BOREAS
Peat subsamples (2-3 g) for ash content analysis were placed in a porcelain crucible, dried at 105°C overnight and weighed. Then they were ignited in a muffle furnace at 800°C for 2 h and subsequently weighed. The ash content is the weight of the subsample after ignition at 800°C divided by the dry weight after heating at 105°C and represented as the percentage dry weight. Ash content is typically analysed in Russian palaeoenvironmental peat records (Katz 1941;Nikonov 1955;Lishtvan & Korol 1975) instead of loss-on-ignition at 550°C, which is commonly used in other regions. To allow comparisons with other Russian records, the ash content was analysed in this study. Total ash content generally represents the conditions of the water-mineral nutrition regime on the mire surface during its development and peat formation (Lishtvan 2010). We used ash content as an indicator of alluvial deposition of mineral matter in the peat. Earlier investigations indicate that the ash content of eutrophic peat derived exclusively from mire plants does not exceed 15% (Katz 1941;Nikonov 1955;Lishtvan & Korol 1975), and hence we interpret values >15% as indicative of river flooding.
Subsamples for the analysis of peat humification were prepared following the standard technique (Chambers et al. 2011). Colorimetric measurements at 540 nm of the extracted subsamples were made using a UV-1601 (PC) Shimadzu spectrophotometer. Each subsample was measured in triplicate and reported as a mean. Obtained values are represented as the percentage of light transmission, where distilled water is 100% transmission.
Ultra-violet (UV) and Fourier transform infrared spectroscopy (FT-IR) were used to assess the amounts of the aromatic and aliphatic parts of humic acid (HA), which can be related to environmental conditions during HA formation. The humic acid spectral characteristics have been used in studies of mineral soils (Dergacheva & Zykina 1988;Milori et al. 2002), but have not been used extensively in peat sequences as indicators of palaeoenvironmental change. Humic acids produced in dry and warm conditions have a well-formed aromatic component and a small aliphatic part. In cold and humid conditions, HA is formed with a larger proportion of the peripheral aliphatic fraction (Dergacheva & Zykina 1988;Milori et al. 2002). The proportion of aromatic and aliphatic parts in HA molecules is expressed as the E 4 /E 6 ratio, with a low ratio indicating a higher condensation of aromatic fragments and their abundant formation, while a high ratio indicates more aliphatic structures in the HA molecule and a low condensation of the aromatic core (Chen et al. 1977;Milori et al. 2002). In addition, we measured absorption bands at A 2920 for aliphatic C-H bonds (C alif ) and at A 1610 for aromatic C-C bonds (C ar ). Low values of E 4 /E 6 and C alif /C ar are expected in dry and/or warm conditions and higher values of these ratios are expected in wet and/or cold conditions (Zaccone et al. 2011(Zaccone et al. , 2018. Humic acids were isolated from peat subsamples with 0.1 M NaOH at room temperature, then precipitated out from the extract with 10% HCl solution. The sample was washed with distilled water until it reached a neutral reaction and dried at room temperature. The UV-visible spectra of HA extractions were obtained with a UVIKON 943 UV-vis-spectrophotometer with a wavelength range of 190-700 nm. The E 4 /E 6 ratio of absorbance at 465 nm (E 4 ) and 665 nm (E 6 ) was recorded according to Chen et al. (1977). The FT-IRspectra of HAwere estimated for a wavenumber range of

Shrubs
Betula nana L., Salix cinerea L., S. rosmarinifolia L., S. pentandra L. Trees Betula pubescens Ehrh. 4000-500 cm À1 using a Nikolet 5700 FT-IRspectrometer. The C alif /C ar ratio of the optical densities at A 2920 (C alif ) and A 1610 (C ar ) was recorded. Before the FT-IR spectrometry, the powdered mixtures of HA extractions and KBr in the ratio of 1:100, respectively, were compressed into pellet form.
Peat subsamples for plant macrofossil analysis were prepared using the standard technique (Birks 2001). Subsamples were sieved under a gentle stream of water (mesh size 0.25 mm). The residue was identified under a stereomicroscope with 10-4009 magnification, using the key by Katz et al. (1977).
Subsamples of raw peat (2-3 cm 3 ) for testate amoeba analysis were washed through a 0.355 mm sieve into beakers, leaving a filtrate volume of 10 mL (Mazei et al. 2011). Prepared samples were analysed under light microscopy at 200-4009 magnification. Testate amoebae were counted up to 250-300 specimens, with the exception of six subsamples from the depths 350, 320, 310, 180, 100 and 70 cm, where the concentration of tests was low (from 27 to 235). Identification was made following multiple established taxonomic keys (Geltzer et al. 1985;Mazei & Tsyganov 2006). The density of tests was calculated in relation to the air-dried weight of peat.
The results of the plant macrofossil and testate amoeba analyses were visualized with the Tilia graphics programme (Grimm 2004). Different periods of mire development were indicated by the CONISS method of cluster analaysis, using plant macrofossil data and the statistical significance of these periods was estimated using a broken stick model (MacArthur 1957).
Oribatid mites were analysed selectively in three samples of peat, which were formed under contrasting environmental conditions, from varying depth ranges (290-280, 230-220 and 140-130 cm). Approximately 30 mL of raw peat samples were collected and oribatid remains were extracted using the paraffin flotation technique (Krivolutsky & Sidorchuk 2005). Oribatid mite specimens were then placed on slides with Berlese mounting medium for further identification.
Mollusc shells were abundant in the lower layers of the peat deposit at depths of 330-280 cm. Their analysis was carried out in a single peat sample at a depth of 290-280 cm. Mollusc remains were picked out manually and washed from peat with water. Mollusc shells were identified under a stereomicroscope using standard identification keys (Likharev & Rammelmeyer 1952;Schileyko 1978Schileyko , 1984Schileyko & Likharev 1986;Starobogatov et al. 2004). Occasionally, a modified comparator technique was applied (Starobogatov & Tolstikova 1986), providing a comparison of the contours of the identified shells with the contours of the selected pattern specimens used as a standard.
The values of a reconstructed depth to water-table (DWT) based on testate amoeba data were calculated using rioja package (Juggins 2020) in R 4.0.4 (R Core Team 2020). Bootstrapping (n = 1000 cycles) errors of inferred DWT values were estimated. The calculations were performed with the transfer function model (Kurina & Li 2019), developed earlier specifically for testate amoeba taxa that inhabit minerotrophic mires from the southern taiga of West Siberia. The weighted averaging function with tolerance downweighting and inverse deshrinking (WA-tol-inv) was applied with the following performance: RMSEP = 6.6 cm, R 2 = 0.82, max. bias = 20.3 cm (Kurina & Li 2019). Several diagnostic statistics of the DWT reconstruction were estimated using rioja (Juggins 2020), vegan (Oksanen et al. 2012) and analogue (Simpson & Oksanen 2020) packages in R.
Firstly, we used chi square distances between fossil assemblages and assemblages in the modern calibration data set larger than the second and the fifth percentiles of all chi square distances in the modern data as indicators of the fossil assemblage having 'no close' and 'no good' analogue, respectively (Birks et al. 1990;Heiri et al. 2003). Secondly, according to the goodness-of-fit measures, fossil samples with a residual distance to the first CCA axis larger than the 90th and 95th percentiles of the residual distances of all modern samples were considered as samples with 'poor fit' and 'very poor fit' with water table depth, respectively (Birks et al. 1990). Thirdly, testate amoeba taxa with a Hill's N2 (Hill 1973) below 5 in the calibration data set were considered to be rare (Heiri et al. 2003). Fourthly, we calculated the percentage of fossil taxa in a sample that were absent from the calibration data set. See also Fig. S1.
Regional averaged moisture indices were calculated for the reconstructions covering more than 4000 years from one region with a similar palaeohydrological trend in the Holocene in accordance with the principles set out in Chen et al. (2008) and Wang & Feng (2013). Moisture fluctuations were coded on a three-point scale moisture index (0 = dry, 1 = intermediate, 2 = wet conditions) at individual reconstruction sites. These fluctuations were averaged for each 500-year interval during the Holocene. Finally, the summarized regional averaged index was assessed using the algorithm proposed by Chen et al. (2008). For those reconstructions where only uncalibrated ages were reported in the original publications, the radiocarbon dates were calibrated based on Bayesian modelling and new age models produced using rbacon package version 2.5.3 (Blaauw & Christen 2011), to be able to compare them with other calibrated reconstructions.
We assume that the accumulation rate was low in the first stage of mire development (350-280 cm, 7800 to 5500 cal. a BP) as a result of the regular flow of the water on the mire surface, under the influence of a nearby river, as previous research (Ivanov 1975) showed that the presence of flowing water on the mire surface usually leads to reduced peat accumulation. Afterwards, an attenuation of fluvial activity and/or agradual increase in the thickness of the peat deposit above the flood level led to a reduction of flooding, causing an increase in the peat accumulation rate between 5500 and 4100 cal. a BP (280-180 cm). The following slowdown of peat accumulation between 180 and 40 cm (4100 to 600 cal. a BP) might be linked to cooler and/or drier climate conditions during this period, or have a different cause. The current results of our study do not permit a more exact conclusion.
The results of calibration of the liquid scintillation radiocarbon dates from the S-2 core (Table 2) showed large chronological uncertainties (Fig. 2). Additional age uncertainty is caused by a depth range of 5 cm above/below the central point of each 10-cm sediment slice. The estimated range of ages for each 10-cm sample is therefore between 442 and 1876 years (970AE449 years, averageAEstandard deviation). We acknowledge this large uncertainty in age and are cautious when interpreting the reported absolute ages, but it is important to stress the focus on long-term trends in this study, rather than specific events.

Humic acid spectral characteristics
There is a general trend of a gradual increase in the amount of HA upward from the S-2 core (Fig. 5). Also, the HA amount is negatively correlated with the ash content of the peat following combustion (R 2 Spearman = À0.52, p < 0.01). During the periods of ash content exceeding the threshold of 15%, the decrease in the HA amount is observed and vice versa (Fig. 5). The ratios E 4 /E 6 and C alif /C ar show a similar pattern (Fig. 5), giving confidence in the observed trends. A general trend towards lower values (warmer/drier conditions) is observed in our record, although increased values of the ratios (colder/wetter conditions) were documented in the periods 340-330 cm (7400 to 7000 cal. a BP), 290-250 cm (5700 to 5100 cal. a BP), 160-80 cm (3700 to 1900 cal. a BP) and 40-10 cm (c. 600 to 0 cal. a BP) (Fig. 5).
These characteristics were completed by the notable presence of leaf litter (25%) and single presence of wood (Betula, Salix and Picea). In Stage 3 (230-140 cm, 4800 to 3200 cal. a BP), there was an increase in Menyanthes (up to 30%), Calla (up to 20%) and brown mosses with a domination of Drepanocladus (up to 30-65%). Remains of Carex lasiocarpa and wood were also observed. In Stage 4 (140-0 cm, 3200 cal. a BP to present) there was a decrease in brown mosses and a subsequent increase in sedges. In that stage, C. chordorhiza and next C. lasiocarpa dominated among sedges followed by C. vesicaria, C. canescens, C. riparia, C. rostrata and C. pseudocyperus. This was combined with a diverse herb distribution: Menyanthes (up to 10-30%), Calla, Comarum, Calamagrostis, Filipendula, Eriophorum angustifolium, Naumburgia, Scirpus, Typha and Triglochin. Wood (Betula) was also recorded.
Initially Phragmites and Equisetum plants dominated the mire (Stage 1). High abundances of these plants might indicate conditions rich in carbonates with weakly flowing water on the mire surface (Khramov & Valutsky 1977;Liss et al. 2001). Then sedges and herbs prevailed, suggesting high levels of surface wetness, indicative of near-surface water tables (Stages 2 and 4). Between 230-140 cm (4800 to 3200 cal. a BP) brown mosses were particularly abundant (Stage 3), indicating stagnant oxygen-depleted water in the mire and, apparently, a near-surface water table.

Testate amoebae
A total of 96 taxawere revealed in peat core S-2. The most abundant taxa were Centropyxis aculeata, Tracheleocorythion pulchellum, Cryptodifflugia sp., Euglypha rotunda, Trinema enchelys and T. lineare (Fig. 4). The highest frequency occurred for the tests from genera Centropyxis, Euglypha, Trinema and Tracheleocorythion. The density of tests varied along the peat core (69AE52 9 10 3 tests per 1 g of air-dry peat, averageAEstandard error, SE). However, the density was relatively high, even down to the basal layers of peat. There was also a notable high level of species richness (19AE6 taxa per peat sample, averageAESE).
Three statistically significant stages of the succession of testate amoeba assemblages were distinguished (Fig. 4). In Stage am1 (350-300 cm, 7800 to 6000 cal. a BP), Trinema lineare absolutely dominated in assemblages. The mean DWT was relatively high (13AE4, averageAESE) indicating low wetness on the mire surface. In Stage am2 (300-180 cm, 6000 to 4100 cal. a BP), the percentage of T. lineare decreased, while the percentages of other dry-related (Schoenbornia humicola, Cryptodifflugia sp., E. rotunda and E. cristata) and wet-related taxa (T. enchelys) increased. According to DWT values (11AE3 cm), the wetness on the mire surface slightly increased. Next, in Stage am3 (180-10 cm, 4100 to 0 cal. a BP), the percentage of wet-related taxa notably increased. Mainly C. aculeata, T. enchelys, T. lineare and E. rotunda prevailed in assemblages. The mean DWT became smaller (3AE5 cm), indicating increased wetness on the mire surface.
Our results confirm that amoeba tests can be preserved in large quantities, not only in Sphagnum peat, but also in peat consisting of sedges, herbs and brown mosses. It is important to note the extraordinary preservation of the tests from the genera Trinema and Euglypha, and even their domination in our study, because shells from these genera have the weakest robustness in ombrotrophic peat (Mitchell et al. 2008a). Until now, testate amoeba analysis was applied mainly to peat from ombrotrophic bogs (Mitchell et al. 2008b). However, the high values of test density and species richness of testate amoebae in minerotrophic peat, revealed in this study, expand the performance of testate amoeba analysis to palaeoecological reconstruction in minerotrophic mires as well.
To evaluate the reliability of the testate amoeba-based reconstruction of DWT we estimated four diagnostic statistics. Analysis of chi square distances between fossil and modern assemblages revealed that 28 out of a total of 34 fossil testate amoeba assemblages had 'no close' modern analogue, except for three samples which had 'no good' analogue (240-220, 210-200 cm; 5000 to 4700 and 4600 to 4400 cal. a BP, respectively), while another three samples had a 'good' analogue (250-240, 170-160, 40-30 cm; 5100 to 5000, 3900 to 3700 and 600 to 400 cal. a BP, respectively; Fig. S1A). When estimating goodnessof-fit, only one sample (100-90 cm, 2400 to 2200 cal. a BP) in the fossil record had a 'good fit' and all the other samples were considered to have a 'very poor fit' (Fig. S1B). Although the percentage of rare taxa with Hill's N2 < 5 in the peat core does not exceed 7% (Fig. S1C), the percentage of taxa absent in modern calibration data set reaches 35-36% in three samples from the peat core (220-200 and 130-120 cm; 4700 to 4400 and 3000 to 2800 cal. a BP, respectively; Fig. S1D). We suggest that the small size of the modern training set (43 samples; Kurina & Li 2019) might be one of the causes leading to the low consistency between fossil and modern assemblages, which can explain the poor analogue and low goodness-of-fit statistics. This indicates that the absolute reconstructed DWT values should be interpreted with caution, although the trends in the DWT reconstructions seem to be correct.

Water-table depth reconstruction and peat humification
Generally, the values of DWT and peat humification agree well (R 2 Spearman = À0.67 at p < 0.001; Fig. 5), with the exception of peat layers at a depth of 330-310 and 300-270 cm (7000 to 6300 and 6000 to 5300 cal. a BP, respectively), when there was an opposite course of these indicators.
On the whole, the reconstructed increased DWT and low peat humification values display decreasing mire surface wetness at a depth of 350-190 cm (7800 to 4300 cal. a BP), then an interval with variable surface wetness, according to the DWT, is recorded at 190-70 cm (4300 to 1600 cal. a BP), while peat humification values remain at a constant high level at this depth. For the upper part of peat deposits a slightly increasing trend of mire surface wetness from 70 to 0 cm (1600 cal. a BP to present) is observed (Fig. 5).

Oribatid mites
In the three samples from 290-280, 230-220 and 140-130 cm depths analysed for oribatid mites, 240 specimens were examined, belonging to 20 taxa (Table S1). There was decrease in the number of oribatid remains with peat depth.
The first sample (290-280 cm, 5700 to 5500 cal. a BP) contained only five identified oribatid carapaces. Hydrozetes was the most abundant in that sample. This oribatid mite generally prefers high mire surface wetness and was often observed on shore vegetation and even on the surface of open water, yet avoiding stagnant water (Krivolutsky et al. 1990). This taxon requires water saturated with oxygen, therefore Hydrozetes might be considered as an indirect indicator of flowing water on the mire surface. Two other identified oribatid mites, Tectocepheus and Scapheremaeus, inhabit emergent aquatic vegetation. We assume, therefore, that the combination of these three taxa in one fossil assemblage indicates conditions of weakly flowing, shallow water on the mire surface among clumps of reeds sticking out of the water.
The second sample (230-220 cm, 4800 to 4700 cal. a BP) was rich in oribatid mites (52 carapaces) representing different environments, such as the hydrophilic Hydrozetes and Limnozetes, and mesophilic to xerophilic Liebstadia and Lepidozetes. Inhabitants of soil surfaces and unspecialized forms of eurybiont oribatids were also strongly represented in the second sample, although Hydrozetes was the dominant taxon similarly with the first sample (290-280 cm, 5700 to 5500 cal. a BP). We assume that this assemblage of oribatid mites indicates very wet conditions in the mire and the water table positioned near the surface. Also, surface moisture might vary during the growing season from high moisture levels in spring during snowmelt to drier conditions in late summer/early autumn.
The greatest number of 183 oribatid carapaces characterized the third sample (140-130 cm; 3200 to 3000 cal. a BP), yet only six taxa were represented. Limnozetes was completely dominant, indicating a high level of surface wetness, probably indicating that the water table was close to the mire surface, a condition which is unfavourable for most oribatid taxa.

Molluscs
A total of 19 taxa (aquatic and terrestrial) were found in the interval 290-280 cm (5700 to 5500 cal. a BP) with aquatic taxa prevailing. The different aquatic species represent both permanent and semi-permanent floodplain lakes of various sizes and depths that are flooded annually by river waters (Dolgin et al. 2014). All taxa strictly indicate conditions of standing water (Table S2); only two of them (Musculium compressum and Amesoda asiatica) can inhabit occasionally silty sediment in slow flowing rivers (Dolgin 2003). The species of terrestrial molluscs identified here prefer habitats with high environmental moisture (Table S3) and are common in modern minerotrophic mires (Schenkov a et al. 2014). Based on the ecology of the species, all molluscs identified in this study apparently lived their entire lives in the mire and were not introduced with river flooding.

Discussion
Relevance of minerotrophic mires to the reconstruction of palaeoclimate changes As highlighted above, there is a great potential to use minerotrophic mires in palaeoenvironmental reconstructions, particularly to reconstruct changes in local hydrology and (indirectly) climate, especially in areas where other climate archives are absent or where they do not provide continuous Holocene records. In this regard, reconstructions of minerotrophic mires are needed in the region of the southern taiga subzone of West Siberia where, in contrast to the central part of the West Siberian Plain, the area of peatlands decreases to 25-30% (Liss et al. 2001;Kremenetski et al. 2003;Vompersky et al. 2011), although the distribution of minerotrophic peatlands, often small in area, increases, and the proportion of ombrotrophic bogs decreases (Vompersky et al. 2011). The time of formation of ombrotrophic bogs in the region refers to the Early Holocene, while they passed to the ombrotrophic stage no earlier than the Middle and Late Holocene (Liss et al. 2001). Palaeoenvironmental reconstructions in peatlands in the southern part of the West Siberian Plain (southern taiga, subtaiga, forest steppe) were carried out both on ombrotrophic and on minerotrophic mires; in most cases they are based on pollen data (Velichko et al. 1997;Borisova et al. 2011;Blyakharchuk 2012). Comprehensive studies with a multi-proxy approach are rare (Krivonogov et al. 2012;Ryabogina et al. 2019).
In addition, we should call attention to the fact that ombrotrophic bogs do not always clearly reflect palaeo-climatic changes. Different studies (Aaby 1976;Barber 1993;Swindles et al. 2012) highlight that endogenous processes occurring on the mire surface can obscure the effects of the palaeoclimate. Furthermore, the stage of the fen-bog transition is a serious endogenous rearrangement of the peat deposit, which itself can screen out palaeoclimatic fluctuations, although some studies (V€ aliranta et al. 2017;Loisel & Bunsen 2020) indicate that the initiation of the fen-bog transition may be climate driven. In this perspective, fens with a homogenous peat deposit have an advantage when reconstructing the palaeoclimate, because they lack periods of considerable endogenous rearrangements.
Apparently, the suitability of a peatland for the reconstruction of palaeoclimate changes depends not only on the ombrotrophic or minerotrophic state, but also on other individual properties of the deposit, as well as the choice of indicator proxies. In a recent paper, Greiser & Joosten (2018) developed a quantitative assessment of the appropriateness of peatland for palaeoenvironmental reconstruction. It follows from their work that the indicator value of deposits directly depends on their thickness and age, and the degree of homogeneity of peat stratigraphy. Thus, heterogeneous sequences, including many thin, sharp layers of peat of various types, potentially reflect local intraregional changes in the palaeoenvironment rather than regional and interregional ones. In this regard, the peat record from our mire has a number of advantages for palaeoclimatic reconstruction: rather homogeneous composition, great age and thick deposits. In addition, we believe that the presence of various bioindicators (testate amoebae, oribatids, molluscs) in peat and frequent fluctuations of small amplitude in their relative abundances, combined with fluctuations in ash content, properties of humic acids, humification and reconstructed DWTmeans that in the future with more precise and accurate dating, the peatland under study can be a good object not only for the reconstruction of longterm climate trends, as shown in this work, but also for a more detailed reconstruction of short-term events on the centennial scale.

Reconstructed mire formation history against the background of palaeoclimate changes
We created a reconstruction of the palaeoenvironment in a minerotrophic mire taking into account combinations of data from different proxies, biological and geochemical. In this work, the most difficult task was to compose the responses of different indicators into a single plausible picture of the palaeoenvironment, because their combinations with each other in some cases turned out to be inconsistent and questionable, although we believe that taken together they balance each other. We had in mind that these responses highlight different aspects of the same palaeoenvironment and they do not directly reflect climatic changes, but only indirectly through the conditions on the mire surface. For example, the response of testate amoebae, taken by itself in the studied mire, showed a long-term trend of a gradual increase in the surface moisture, with maximum moisture at depths of 180-70 cm (4100 to 1600 cal. a BP; Fig. 4), while the responses of all indicators used, taken together, indicated another trend of gradual decrease in humidity during the Holocene with maximum humidity at depths of 350-250 cm (7800 to 5100 cal. a BP; Fig. 5). This example confirms the need to use a multi-proxy approach in reconstruction in minerotrophic mire and to avoid interpretation based on single indicator. Further, we compared the obtained comprehensive history of the palaeoenvironment in the minerotrophic mire with the already existing palaeorecords for the region. According to the consistency between them, we judged the responsiveness of the studied mire to changes in the regional palaeoclimate.
The development of the mire at the coring site began 7800 cal. a BP in a period of increased precipitation (Glebov et al. 1996;Andreev et al. 2003;Blyakharchuk 2009). Earlier research (L'vov 1976Glebov 1988;Liss et al. 2001) concluded that mire formation in the river valleys in the south of West Siberia resulted from flooding of forested areas. In our study, the mire was initially forested and characterized by the presence of different herbs and hydrophilic plants. The wetness of the mire surface was high according to the low value of DWT and the high value of humification (Fig. 5).
Three long-term periods can be distinguished in the development of the mire (Fig. 6). The first period, between 7800 and 5100 cal. a BP, is marked by peat ash content exceeding the 15% threshold, indicating alluvial deposition on the mire surface; we suggest that this was caused by intensive fluvial activity. The abundance of molluscs on the mire surface in shallow water habitats and the development of thickets of Phragmites and Equisetum plants, which prefer the conditions of weak flow in shallow water, also support this idea of regular flooding. This contrasts with the reconstructed DWT values, indicating low wetness on the mire surface. A possible explanation for this discrepancy is that testate amoebae inhabited the mire surface when the river water seasonally receded, but not during periods of flooding as this was unfavourable for them. This hypothesis is supported by the prevalence of small testate amoebae taxa (Trinema, Cryptodifflugia, Schoenbornia; Fig. 4), which can develop quicker, in contrast to the larger taxa (C. aculeata), which require favourable conditions for a longer time. Increased frequency of river flooding was also registered in other research from the south taiga of West Siberia at 11000 to 5000 cal. a BP (Borisova et al. 2011), in the middle taiga of West Siberia at 8100 to 6800 cal. a BP , in the south of East Siberia at 6800 to 6600 cal. a BP (Sidorchuk et al. 2003 andVorobyeva et al. 1992 therein) and in the north of East Siberia at 9000 to 4000 cal. a BP (Biskaborn et al. 2016) (Fig. 6). The synchronicity of floods over a wide area suggests that enhanced precipitation in the region may have been a common cause. This period corresponded to the Holocene thermal optimum, when warm and humid conditions were observed by pollen data in the southern taiga of West Siberia at c. 6800 to 5700 cal. a BP (corresponding to 6000-5000 14 C a BP in Borisova et al. 2011) and c. 6800 to 6200 cal. a BP (corresponding to 6000-5500 14 C a BP in Blyakharchuk 2009), in the middle taiga of West Siberia at 7000 to 5000 cal. a BP (Pitk€ anen et al. 2002) and on the east slope of the Middle Ural mountains at 7800 to 4200 cal. a BP (Panova & Antipina 2016). In our research on the period 7800 to 5100 cal. a BP, the values of the ratios E 4 /E 6 and C alif /C ar remained relatively low, indicating tentatively that warmer conditions rather than drier conditions prevailed (Fig. 5).
The next period in the mire development, between 5100 and 3000 cal. a BP, is mainly distinguished by a notable decrease in ash content in peat below or near the threshold of 15%, which indicates the absence of major flood events. The values of the ratios E 4 /E 6 and C alif /C ar become higher, suggesting wetter and/or colder conditions at that time. Between 5000 and 4700 cal. a BP the increased values of DWTand the decreased humification values indicate drier conditions on the mire surface. Moreover, slight drying is confirmed by the increase in taxonomical diversity of oribatid mites and the presence of mesophilic to xerophilic taxa although, from 4700 cal. a BP, the abundance of brown mosses and hydrophilic herbs (Menyanthes and Calla) in the mire gradually increased, indicating a return to wetter conditions. Between 4700 and 3000 cal. a BP, the humificationvalues increased and DWT values were smaller, indicating increased surface wetness (Fig. 5). In addition, oribatid assemblage changed to a predominance of Limnozetes taxon (140-130 cm; 3200 to 3000 cal. a BP), which reflected stagnant moisture on the mire surface.
We suppose that the cessation of flooding of the mire surface since 5100 cal. a BP can tentatively be explained by drier and colder climate conditions. The absence of flowing water on the mire surface is evidenced by various indicators for stagnant water at that time. Similar trends of decreasing precipitation and cooling that followed the Holocene thermal optimum were documented in other peatlands and lake sediments across the taiga and arctic zones of West Siberia (Velichko et al. 1997;Andreev et al. 2003;Blyakharchuk 2009).
During the third period in the mire development, from 3000 cal. a BP to present, a gradual change in plant assemblages towards a decrease in the proportion of brown mosses and increase in the proportion of sedges and herbs was observed. In addition, greater DWT values and reduced humification values indicate drier conditions on the mire surface compared with the previous period, which is supported by the ratios E 4 /E 6 and C alif /C ar . We tentatively interpret progressive drying of the mire surface as a consequence of a decrease in precipitation continuing after the Holocene thermal optimum. Dry climate conditions in the Late Holocene have been recorded more widely in peatlands and lake sediments of the taiga and arctic zones in West Siberia (Velichko et al. 1997;Andreev et al. 2003;Blyakharchuk 2009;Amon et al. 2020;Fig. 6).

Local patterns of mire development in the Bakchar River valley
The location of mire in a river valley, first of all, implies a direct or indirect impact of the river on mire development. With regard to the direct impact, we found traces of flooding of the mire surface through an increase in the ash content of the peat (Fig. 5). This impact was not constant, but only in certain periods, which we interpret as phases of increased fluvial activity. Mineral input from the river was characterized by fine clay particles deposited in weakly flowing water; no coarser sand particles indicating running water (Jurasz & Amoros 1991) were found in core S-2. In addition to fluvial inputs, the mire could also be fed and flooded by groundwater or runoff from overlying terraces during snow melt.
The lithological structure ofour record is homogenous and consists entirely of peat without interlayers of gyttja or silt. We cautiously assume that this indicates the absence of abrupt changes in the topography of the river channel during the development of the mire. This is further supported by the smooth (not sharp) changes in the composition of plant macrofossils between the four Fig. 6. Long-term hydrological periods of mire development from core S-2 compared with palaeoclimatic events revealed from published records providing regional context and regional averaged moisture indices: 1 = S-2 core (this study); 2 = peat bog Entarnoe (Bukreeva et al. 1995;Velichko et al. 1997); 3 = peatland Zhukovka (Velichko et al. 1997); 4 = two lakes (Andreev et al. 2003); 5 = two peat bogs from the middle and south taiga, Petropavlovka and Bugristoye, respectively (Blyakharchuk 2009(Blyakharchuk , 2012 identified zones (Fig. 3) of mire development. For 7800 years the mire deposit was formed from mineralrich peat. Today, a shrub-sedge-brown moss community grows on the surface of the mire at the S-2 coring site, indicating its mineral-rich state. We expect that this minerotrophic mire has not changed into an ombrotrophic bog yet because it is regularly fed by highly mineralized groundwater from the overlying terrace. However, Sphagnum hummocks were noted in the centre of the mire, indicative of the initial stages of the fen-bog transition process. Therefore, the continued growth of the peat deposit will probably lead to the gradual formation of an ombrotrophic bog.
Regional patterns in hydroclimatic trends in the taiga zone against the wider geographical context of Siberia and Central Asia during the Holocene The record of mire development in S-2 has strong similarities in mire development to other records in the taiga and arctic zones of West Siberia that cover the Middle and Late Holocene (Velichko et al. 1997;Andreev et al. 2003;Blyakharchuk 2009;Amon et al. 2020;Fig. 6), indicating that the record has been influenced predominantly by (regional) climatic fluctuations, and that local, autogenous processes play a minor role. In addition, this record is consistent with previous studies (Velichko et al. 1997;Bezrukova et al. 2011), which highlighted the dominant influence of westerly winds throughout Eurasia; westerlies gradually weaken eastwards and can be blocked by the Siberian High over parts of Siberia.
The regional average moisture index calculated by us for the taiga and the arctic zones of Siberia (see numbers 1-11 in Figs 6 and 7) showed a long-term trend from intermediate wet to wet and then dry conditions with the period of the most moisture between 8500 and 6000 cal. a BP in the Holocene. This is consistent with the study by Chen et al. (2008), which presented a similar wet Holocene thermal optimum for the westerly dominated area of central arid Asia with a period of maximum moisture between 8000 and 4000 cal. a BP in the Holocene. The series of reconstructions across the Siberian taiga belt (between latitudes 60°and 68°N) reported in the study by Velichko et al. (1997) supported that the wettest period occurred in the Middle Holocene, while the Early Holocene was drier. The explanation for this pattern is that the westerlies brought little precipitation to Asia, since the remnants of ice sheets at high latitudes continued to melt, the surface temperature of the North Atlantic Ocean has been decreased and evaporation has been reduced (Chen et al. 2008;Jiang et al. 2015). Then, in Eurasia, an increase in humidity was observed at 8000 to 4000 cal. a BP, which is associated with the final disappearance of ice sheets at high latitudes, warming of the waters of the North Atlantic Ocean, an increase in evaporated moisture and its transfer to Asia. In the Late Holocene, there was a trend of a gradual and slight decrease in moisture in westerly dominated Asia (Chen et al. 2008) and a similar trend of a gradual drying across the Siberian taiga belt (Velichko et al. 1997).
At the same time, the trend from intermediate wet to wet and then dry conditions in the Holocene (see Fig. 7. Geographical location of core S-2 (marked with star) compared with other palaeoecological records in the taiga and arctic zones (black circles) and in the forest-steppe and steppe zones (half-black circles). Numbers of records are the same as in Fig. 6. The dotted line surrounds the region with reconstructions from Magny et al. (2003). Black squares indicate reconstructions having a long-term trend from dry to wet and then intermediate wet conditions in the Holocene and taken from Chen et al. (2008); half-black squares indicates reconstructions having two periods of increased moisture in the Middle and Late Holocene and taken from Chen et al. (2008). BOREAS numbers 1-11 in Figs 6, 7) in the Siberian taiga and arctic zones differs from that in the region in the south of West Siberia, for which the regional trend is from dry to intermediate wet and then wet conditions (numbers 12-18 in Figs 6, 7), with a period of maximum moisture in the Late Holocene between 4000 and 1000 cal. a BP. A similar trend from dry to intermediate wet and then wet conditions was documented for the regions of southern Siberia, western Mongolia and northern Xinjiang (subregions b, d, e in the review by Zhang & Feng 2018). A few individual records from a synthesis across central arid Asia (sites 3-7 in Chen et al. 2008) recognize one period with increased moisture in the Middle Holocene (c. 8000 to 4000 cal. a BP) and a less clearly defined second period of increased moisture in the Late Holocene between c. 3500 and 500 cal. a BP (halfblack squares in Fig. 7). The mechanism behind this increased moisture in the second half of the Holocene for the given reconstructions (numbers 12-18 and half-black squares in Fig. 7) is apparently the influence of the main mid-latitude jet stream of the westerlies, which extends from the Mediterranean region. A similar pattern was shown by Magny et al. (2003) for the Mediterranean region itself, where during the 8.2 ka event in the Holocene, an increase in precipitation was revealed for the territory between 43°and 50°N latitudes (including France, southern Germany, Switzerland and northern Italy), influenced directly by the main westerly jet, whereas a decrease in moisture was observed for areas to the north and south of this (Fig. 7). Periodically, the main jet stream of the westerlies shifts in a south-north direction and thus brings precipitation to different regions of Eurasia (Shnitnikov 1957;Abrosov 1962;Hong & Lu 2016;Chen et al. 2021). We suggest that in the Late Holocene this jet stream displaced mainly to the south of West Siberia and adjacent regions (Zakh et al. 2010;Blyakharchuk et al. 2019).

Conclusions
The results of our multi-proxy study fit well with previously obtained palaeoenvironmental reconstructions across the wide territory of Siberia, including particularly the taiga and arctic zones, support the idea of a prevailing influence of the westerlies on the regional climate pattern during the Holocene in West Siberia and follow the key long-term climatic trends documented in different regions throughout the taiga zone of West Siberia and also throughout Central Asia over the Middle to Late Holocene.
The eutrophic mire investigated in this study is responsive to the climatic changes throughout the Holocene, including a notable increase in precipitation during the Holocene thermal optimum between 8500 and 6000 cal. a BP and a subsequent downward trend in precipitation. The agreement of these results with existing long-term regional palaeoclimate dynamics provides additional support that eutrophic mires, like ombrotrophic ones, can be used for palaeoenvironmental reconstruction. We believe that palaeoecological studies on minerotrophic peatlands will crucially detail the history of the palaeoenvironment in the southern part of the West Siberian Plain, where ombrotrophic bogs are rare and their bog stage is rather young, and make it possible to extend it in time until the Early Holocene.
Combined, the different bioindicators (plant macrofossils, molluscs, oribatid mites and testate amoebae) and geochemical indicators (ash content, peat humification and spectral characteristics of humic acids) provide a detailed and comprehensive understanding of the palaeoenvironmental change and allow the influence of palaeoclimate, among other variables, to be highlighted. Fig. S1. Estimation of the reliability of testate amoebainferred water table depth reconstruction in the S-2 core: A= minimal chi-squared distance between fossil and modern testate amoeba assemblages (quantile at 2% marked with dashed line and at 5% with dotted line); B = goodness-of-fit measures between fossil and modern assemblages (quantile at 90% marked with dashed line and at 95% with dotted line); C = percentage of rare taxa in fossil assemblages; D = percentage of taxa in fossil assemblages, which are absent in the calibration data set. Table S1. Species composition and occurrence of oribatid mite carapaces in the peat samples from the core S-2. Analysts: L.V. Salisch and M.L. Egorova. Table S2. Ecological preferences of aquatic molluscs from the peat core sample in accordance with Dolgin (2003) and Dolgin et al. (2014). Analyst: V.N. Dolgin. Table S3. Ecological preferences of terrestrial molluscs from the peat core sample. Analyst: A.V. Udaloi.