Dissolved organic carbon and major and trace elements in peat porewater of sporadic , discontinuous , and continuous permafrost zones of western Siberia

Mobilization of dissolved organic carbon (DOC) and related trace elements (TEs) from the frozen peat to surface waters in the permafrost zone is expected to enhance under ongoing permafrost thaw and active layer thickness (ALT) deepening in high-latitude regions. The interstitial soil solutions are efficient tracers of ongoing bio-geochemical processes in the critical zone and can help to decipher the intensity of carbon and metals migration from the soil to the rivers and further to the ocean. To this end, we collected, across a 640 km latitudinal transect of the sporadic to continuous permafrost zone of western Siberia peatlands, soil porewaters from 30 cm depth using suction cups and we analyzed DOC, dissolved inorganic carbon (DIC), and 40 major elements and TEs in 0.45 μm filtered fraction of 80 soil porewaters. Despite an expected decrease in the intensity of DOC and TE mobilization from the soil and vegetation litter to the interstitial fluids with the increase in the permafrost coverage and a decrease in the annual temperature and ALT, the DOC and many major and trace elements did not exhibit any distinct decrease in concentration along the latitudinal transect from 62.2 to 67.4 N. The DOC demonstrated a maximum of concentration at 66 N, on the border of the discontinuous/continuous permafrost zone, whereas the DOC concentration in peat soil solutions from the continuous permafrost zone was equal to or higher than that in the sporadic/discontinuous permafrost zone. Moreover, a number of major (Ca, Mg) and trace (Al, Ti, Sr, Ga, rare earth elements (REEs), Zr, Hf, Th) elements exhibited an increasing, not decreasing, northward concentration trend. We hypothesize that the effects of temperature and thickness of the ALT are of secondary importance relative to the leaching capacity of peat, which is in turn controlled by the water saturation of the peat core. The water residence time in peat pores also plays a role in enriching the fluids in some elements: the DOC, V, Cu, Pb, REEs, and Th were a factor of 1.5 to 2.0 higher in mounds relative to hollows. As such, it is possible that the time of reaction between the peat and downward infiltrating waters essentially controls the degree of peat porewater enrichments in DOC and other solutes. A 2 northward shift in the position of the permafrost boundaries may bring about a factor of 1.3± 0.2 decrease in Ca, Mg, Sr, Al, Fe, Ti, Mn, Ni, Co, V, Zr, Hf, Th, and REE porewater concentration in continuous and discontinuous permafrost zones, and a possible decrease in DOC, specific ultraviolet absorbency (SUVA), Ca, Mg, Fe, and Sr will not exceed 20 % of their current values. The projected increase in ALT and vegetation density, northward migration of the permafrost boundary, or the change of hydrological regime is unlikely to modify chemical composition of peat porewater fluids larger than their natural variations within different micro-landscapes, i.e., within a factor of 2. The decrease in DOC and metal delivery to small rivers and lakes by peat soil leachate may also decrease the overall export of dissolved components from the continuous permafrost zone to the Arctic Ocean. This challenges Published by Copernicus Publications on behalf of the European Geosciences Union. 3562 T. V. Raudina et al.: Dissolved organic carbon and elements in peat porewater of western Siberia the current paradigm on the increase in DOC export from the land to the ocean under climate warming in high latitudes.

the current paradigm on the increase in DOC export from the land to the ocean under climate warming in high latitudes.

Introduction
Boreal and subarctic regions of the Northern Hemisphere are among the most vulnerable areas to ongoing climate warming (Natali et al., 2011(Natali et al., , 2015;;Schuur et al., 2015;Vonk et al., 2015b;Pries et al., 2016).Because of sizeable carbon storage in frozen soils of Siberia (Botch et al., 1995;Krementski et al., 2003;Frey and Smith, 2007;Beilman et al., 2009;Tarnocai et al., 2009;Gentsch et al., 2015), the warming in this region is especially important for global projections of the carbon balance on the planet (Smith et al., 2004;Frey and Smith, 2005;Feng et al., 2013).In this regard, permafrostbearing part of Western Siberia Lowland (WSL) is highly sensitive to soil warming, due to (i) the dominance of discontinuous, sporadic, and intermittent permafrost coverage compared to continuous and discontinuous permafrost of central and eastern Siberia and Canada High Arctic; (ii) the surface layer temperature of the WSL permafrost is often between 0 and −2 • C, which is warmer than in other regions of the world (Romanovsky et al., 2010); (iii) essentially flat area of the WSL and high impact of flooding and thermokarst development; and, most importantly, (iv) high stock of ancient and recent organic carbon in the form of partially frozen peat deposits of 1 to 4 m thickness.
Mobilization of dissolved organic and inorganic carbon (DOC and DIC, respectively) and related trace elements (TEs) including metal contaminants and micronutrients from the frozen peat to surface waters and further to the Arctic Ocean is one the major consequences of ongoing permafrost thaw (Tank et al., 2012a(Tank et al., , b, 2016;;Striegl et al., 2005;Rember and Trefry, 2004;Prokushkin et al., 2011;Mann et al., 2012;Grosse et al., 2016;Holmes et al., 2013).The impact of warming on Arctic and subarctic soil is primarily through the active layer thickness (ALT) rise (Zhang et al., 2005;Akerman and Johannson, 2008) although a number of other phenomena (plant productivity, drainage and hydrological regime change, ground fires, etc.) may be even more important in changing the biogeochemical cycle of carbon and metals in permafrost-affected soils (Jorgenson et al., 2013).For these reasons, the peat land zones have received significant attention (Haapalehto et al., 2011;Olefeldt and Roulet, 2012;Charman et al., 2013;Quinton and Baltzer, 2013;Muller et al., 2015;Morison et al., 2017), notably via natural manipulation experiments in order to assess the responses of peat carbon to simulated warming and oxidizing (Dielemann et al., 2016;Liu et al., 2016), water table manipulation (Blodau and Moore, 2003;Strack et al., 2008;Goldberg et al., 2010), and drought (Clark et al., 2012).
In this work we sampled, across a 640 km latitudinal transect of sporadic to continuous permafrost, the interstitial soil solutions of the largest peatland of the world.Our main goal was to quantify the distribution of DOC, major elements, and TEs in porewaters along a permafrost gradient of similar micro-landscapes.Within the upper unfrozen peat horizon, we hypothesize a trend of diminishing DOC and metal concentration northward, due to the decrease in mean annual temperature, vegetation density, and active layer thickness.We aimed at quantifying the latitudinal trend of peat porewater concentration of DOC, major elements, and TEs and testing the difference in solute concentration sampled from various micro-landscape such as mound, hollow, depression, and polygon.Implying an approach substituting space for time, developed for surface waters of western Siberia, (i.e., Frey et al., 2007a, b;Frey and Smith, 2005), the obtained results should allow a straightforward empirical provisions of soil water chemistry change during northward migration of the permafrost boundary.Because the main source to inland waters in this vast territory (over 1 million km 2 ) occurs as suprapermafrost flow over the impermeable frozen peat horizon (Novikov et al., 2009), and due to the fact that the western Siberian peatlands contain the largest soil water and ice resources in the northern hemisphere (Smith et al., 2012), the assessment of soil peat water chemical composition should The inserts represent aerial (drone-made) photos of main sites with the position of mound/polygon (M/P), hollow (H), frost crack (FC), and permafrost subsidence (Ps).On the Kogalym site, a hollow (H) -ridge (R) -lake complex is a dominating landscape type.The numbers on the legend represent the following: 1, tundra; 2, forest-tundra; 3, northern taiga; 4, middle taiga; 5, borders between natural biomes; 6, borders between permafrost zones; 7, continuous permafrost; 8, discontinuous permafrost; 9, sporadic permafrost; 10, isolated permafrost; 11, key study sites with mean annual temperature in the parentheses.
help predicting the possible change of DOC and metal transport of permafrost-bearing Siberian rivers and lakes under climate warming scenarios.

Geographical setting and local micro-landscapes
The Western Siberia Lowland (WSL) includes the watershed of the Ob, Pur, Nadym, Poluy, and Taz rivers that drain Pleistocene sands and clays, covered by thick (1 to 3 m) peat.All three major zones of the boreal biome, taiga, forest-tundra and tundra, can be found in this region.The territory investigated in this work includes three main permafrost zones: sporadic, discontinuous, and continuous (Fig. 1).Quaternary clays, sands, and alevrolites underlying the surface peat deposits range in thickness from several meters to 200-250 m and have fluvio-glacial and lake-glacial origin in the north of 60 • N. The climate is humid semi-continental with mean annual temperature (MAT) ranging from −2.8 • C in the south of the cryolithozone (Syrgut region) to −9.1 • C in the north (Tazovsky).The annual precipitation ranges from 600 mm in Kogalym to 360 mm in Tazovsky.Along the gradient of discontinuous to sporadic to continuous permafrost zones, we selected five main test sites whose physico-geographical characteristics are given in Table 1.
A typical feature of the WSL is the presence of positive and negative forms of relief -micro-landscapes.The initial bog with weakly pronounced micro-relief was subjected to freezing during the Subboreal period (∼ 4500 years ago).During the Subatlantic period (2500 years ago) and the increase in temperature and precipitation, the thermokarst started.The hollows received sufficient water and they started to thaw, whereas the mounds were rising due to ice  -Dystric Ombric Fibric Histosols wedges underneath (Panova et al., 2010;Ponomareva et al., 2012;Pastukhov et al., 2016).The positive forms include ridges in the permafrost-free and sporadic permafrost zone, mounds in discontinuous permafrost zones, and polygons in the subarctic tundra of continuous permafrost.The negative forms comprise hollows (abundant across all zones), permafrost subsidences in discontinuous and continuous permafrost zones, and frost cracks of the polygonal tundra biome.In each of five major sites, several micro-landscapes corresponding to one positive and two negative form of relief were selected as specified in Table 1 and shown as aerial views in Fig. 1.The cross sections of dominant microlandscapes with corresponding soil specifications are represented in Fig. 2 and include (i) peat mounds in the four southern sites of flat mound peat bog, and corresponding polygon in the most northern (Tazovsky) site of polygonal tundra; (ii) hollows in all five sites; and (iii) permafrost subsidences in four southern sites and corresponding frost crack in Tazovsky.Typical soil profiles of studied sites are illustrated in Fig. S1 in the Supplement.

Soil porewater sampling
Altogether, 80 soil porous waters in five main sampling sites were collected in the end of July-beginning of August 2015.In this study, suction cup lysimeters were used.The chemical composition of interstitial soil solution is known to depend on the extraction method (e.g., Geibe et al., 2006;Schlotter et al., 2012).Detailed comparison between suction cup and press technique is described in methodological work of our group (Raudina et al., 2016).In the peat profile of each micro-landscape, the PTFE suction cup lysimeters (95 mm long and 21 mm diameter, 2 µm pore size) of SDEC (France) were installed at the depth of 30 ± 15 cm below the moss layer (Fig. S2 in the Supplement).The choice of the sampling depth was determined by the position of the permafrost table: typically, the cup was installed at 10 cm from the peat outcrop vertical surface, 5-10 cm above the bottom of the active layer but not deeper than 40-50 cm from the moss layer.In all sites, the cups were installed exclusively in soils that belonged to the Histosols group (according to WRB 2014, i.e., having a thickness of peat > 60 cm).The cups were connected via PTFE tubing to polypropylene 1 L container maintained at 75 to 50 kPa via a Mityvac MV8255 PVC hand pump or a portable electric vacuum pump (KNF Neuberger W/VAC.5.5 L).Before each installation, the suction cups were cleaned by flushing with Milli-Q water (∼ 250 mL), followed by 3 % ultrapure HNO3 (∼ 250 mL), and finally Milli-Q water (∼ 750 mL).Each cup was soaked in Milli-Q water for at least 1 day before the experiment and was used only once.The porewater was collected in two steps.The first portion (100-200 mL) was collected during 24 h and the fluid was discarded, allowing for the saturation of the tubing and the recipient bottle surface.The second portion (100-300 mL) was collected during the next 24 h of deployment or, The numbers on the legend represent the following: 1, moss-lichensedge peat of medium degree of decomposition (Hemic); 2, permanently frozen peat; 3, moss-based peat of low degree of decomposition; 4, illuvial-Fe-humic (spodic) horizon; 5, permanently frozen spodic horizon; 6, sand and silt deposits; 7, frozen sand and silts; 8, heavy clay deposits; 9, frozen clays; 10, the level of suprapermafrost waters in August; 11, ice wedges; 12, cryoturbation features in soil; 13, the direction of soil water transport, typically along the permafrost boundary; 14, small crack on the polygonal bog.
in case of dryer conditions, over 48 h and used for analyses.The vacuum in the recipient bottle decreased from 75 kPa to atmospheric pressure over 24 h, and the first portion of the fluid appeared at 45 to 50 kPa.

Analyses
Collected waters were immediately filtered in pre-washed 30 mL PP Nalgene ® flacons through single-use Minisart filter units (Sartorius, acetate cellulose filter) having a diameter of 25 mm and a pore size of 0.45 µm.The first 20 mL of filtrate was discarded.Filtered solutions for cation analyses were acidified (pH ∼ 2) with ultrapure double-distilled HNO 3 and stored in pre-washed HDPE bottles.The preparation of bottles for sample storage was performed in a clean bench room (ISO A 10 000).Blanks were performed to control the level of pollution induced by sampling and filtration.The DOC blanks of Milli-Q filtrate never exceeded 0.1 mg L −1 , which is quite low for the organic-rich porewaters sampled in this study (i.e., 10-100 mg L −1 DOC).pH was measured in the field using a combined electrode with un uncertainty of ±0.02 pH units.DOC and DIC were analyzed using a carbon total analyzer (Shimadzu TOC VSCN) with an uncertainty better than 3 %.The instrument was calibrated for analysis of both form of dissolved carbon in organic-rich, DIC-poor waters (e.g., Prokushkin et al., 2011).The UV absorbance of the filtered samples was measured at 280 nm using a quartz 10 mm cuvette on a Cary-50 spectrophotometer.The specific UV absorbency (SUVA 280 , L mg −1 m −1 ) is used as a proxy for aromatic C, molecular weight, and source of dissolved organic matter (DOM) (Uyguner and Bekbolet, 2005;Weishaar et al., 2003;Ilina et al., 2014, and references therein).The SUVA 280 in the present study was used for consistency with previous measurements of lakes and rivers in western Siberia (Shirokova et al., 2013;Manasypov et al., 2015Manasypov et al., , 2017;;Pokrovsky et al., 2015) and permafrost-draining rivers in central Siberia (Prokushkin et al., 2011).
Major anion (Cl − , SO 2− 4 ) concentrations were measured by ion chromatography (HPLC, Dionex ICS 2000) with an uncertainty of 2 %.Major cations (Ca, Mg, Na, K), Si, and ∼ 40 TEs were determined with an ICP-MS Agilent ce 7500 with In and Re as internal standards and three various external standards, placed each 10 samples in a series of river water.Details of TE analyses in DOC-rich waters of western Siberia are given elsewhere (Pokrovsky et al., 2016a, b).The SLRS-5 (Riverine Water Reference Material for Trace Metals certified by the National Research Council of Canada) was used to check the accuracy and reproducibility of each analysis (Yeghicheyan et al., 2013).Only the elements that exhibited good agreement between replicated measurements of SLRS-5 and the certified values (relative difference < 15 %) are reported in this study.

Statistical treatment
The concentrations of carbon and major elements in soil porewaters were treated using the least squares method and Pearson correlation (SigmaPlot version 11.0, Systat Software, Inc.).Regressions and power functions were used to examine the relationships between the elemental concentrations and the latitude of sampling.The normality of data distribution was checked using the criterion of Kolmogorov-Smirnov, separately for each site and for the full set of the data.The significance value was < 0.01 and thus nonparametric criteria for data comparison were used.First, major elements and TEs concentrations in soil porewaters of (1) five main sampling sites and (2) four main micro-relief landscapes (polygon, permafrost/subsidence, frost crack and hollow) were processed using the non-parametric H-criterion Kruskal-Wallis test.This test is suitable for evaluation of difference of each component among several samplings simultaneously.It is considered statistically significant at p < 0.05.In case of significant differences, a comparison of DOC, major elements, and TE concentration between soil porewaters sampled in three main pair micro-landscapes (moundhollow, mound-subsidence, and hollow-subsidence) of each 5 major sampling site was conducted using non-parametric pair Wilcoxon-Mann Whitney test.All graphics were performed using MS Excel 2010 and the GS Grapher 11 package.Principal component analysis (PCA) was used for the full set of sampled soil porewaters across the microlandscapes and permafrost zones.In this treatment, the main numerical variables were the geographic latitude of the sampling site, the depth of peat horizon, ALT, specific conductivity, pH, DOC, DIC, Cl, SO 4 , Si, all major cations, and 43 TE concentrations.
The PCA allowed for testing the influence of various parameters, notably the latitude and the ALT on the soil porewater DOC and element variability.All the variables were normalized as necessary in standard package of STATISTICA-7 (http://www.statsoft.com)given that the units of measurements of various components are different.The identification of factors was performed using the method of Raw Data and the extraction method was principal component.The scree test involved plotting the eigenvalues in descending order of their magnitude against their factor numbers and determining where they level off.The PCA values demonstrated significant decrease in the value between F2 and F3 suggesting therefore that at least two factors are interpretable.

PCA and correlations between elements
The PCA of all micro-landscapes and geographical zones yielded two possible factors contributing to observed variations in element concentration (i.e., 20 and 9 %, Fig. S3A, B in the Supplement).Such a relatively low proportion of the variance explained by PCA is consistent with previous treatments of the WSL river water, conducted on a much larger dataset (Pokrovsky et al., 2016a).Because the standard STATISTICA-7 package used in this work does not allow realization of the Kaiser-Meyer-Olkin (KMO) criterion, we computed this criterion using Excel ® .The KMO value was equal to 0.533, which suggests rather low adequacy: the analysis does not make sense at KMO < 0.5.Note that the removal of a part of the data series and conducting separate PCA for major elements, TEs, various forms of micro-relief, and various geographical sites did not yield any better de-scription of the variance mainly because of insufficient size of the dataset.
The first factor explains a greater variance in heavy element hydrolysates such as REEs, Cr, Nb, Zr, Hf, Th, and U, whereas the second factor was pronounced for soluble and biogenic elements (Mn, Co, Ni, V, Si, Ca, Mg, Sr), pH, and latitude but also included Al and Fe, presumably due to organic complexation (see Sect. 4.2 below).The correlation matrix (Table S1 in the Supplement) and respective dendrogram of a hierarchical cluster for scaled porewater score variation (Fig. S3C) demonstrated pronounced link of Si with REEs, Zr, Nb, Fe, Cr, V, and Li, probably corresponding to the source of these elements from silicate matrix of the peat profile.There was positive correlation between Mn and Ca and Sr and Ca, reflecting the biological impact or soluble carbonate minerals as it is established for riverwater of the region (Pokrovsky et al., 2016a).Note that the correlations of latitude, specific conductivity, pH, and DOC with all major elements and TEs were poorly pronounced (R < 0.5), whereas Fe and Al correlated with Si, Ti, V, Cr, Co, Ni, As, Zr, heavy REEs, and Hf.

Effect of micro-landscape
The mean values with SD of all major elements and TEs in soil porewaters of main micro-landscapes in each site are listed in Table 2.The mean values for the whole WSL territory for two dominant micro-landscapes, mound and hollow, are given in the last two columns of this table.Results of the application of Wilcoxon-Mann Whitney test for assessing the differences of DOC and several major element and TE mean values between the dominant micro-landscapes in each site are listed in Table S2 in the Supplement.According to the chosen statistical criteria, only a few elements (DOC, Al, Fe, Si, Mn, Cu, Cd, Pb, Hf, U) showed significant differences in their concentration between different micro-landscapes.The DOC was approximately twice higher (p = 0.023 to 0.043) in mounds (or polygons) compared to hollows in all four sites except Pangody, where the difference was only a factor of 1.1, which is not significant (p = 0.082).In Khanymey, Urengoy, and Tazovsky, the order of DOC concentration in various micro-landscapes was (mound or polygon) ≥ (permafrost subsidence or frost crack) > hollow.Cu and, sometimes, Zn followed this order.Concentrations of Al, Si, Fe, Sr did not demonstrate any systematic difference between positive and negative forms of relief for each site, without distinct preferential enrichment of one micro-landscape versus another in the north or in the south.The minimal contrast in DOC and element concentration between micro-landscapes was observed in Pangody and the maximal variability was in Khanymey.
Within the standard deviation of the mean values, there was no difference in DIC, Si, Ca, and Mg concentration between different micro-landscapes in all studied sites.The exception was Khanymey, where the hollows demonstrated a factor of 1.5-2.8higher Mg, Si, and Ca concentration compared to mounds and Urengoy, where the mounds contained less Mg and Si than the hollows.However, in the latter case, at p = 0.041 to 0.048, this difference was within the variation of the average (Table S2).The mean concentrations of DIC, Cl, K, Si, Ca, Mg, Al, Fe, Ti, Sr, Ba, Zn, Mn, Ni, and TEs over the full WSL territory are quite similar (±20 %) between positive and negative forms of relief (compare the last two columns of Table 2).The DOC, B, Na, V, Ga, Cu, Cs, Pb, REEs, and Th exhibited a factor of 1.5 ± 0.2 (significant at p < 0.05) higher WSL-mean concentrations in mounds/polygons compared to hollows.

Effect of latitude and permafrost zone on peat porewater concentrations of DOC and metals
In order to examine the latitudinal trend of element concentration in the porewater, first we ran the Kruskal-Wallis and then the Wilcoxon-Mann Whitney pair test for overall differences.After that we assessed which micro-landscape exhibited the largest difference between sites.Results include the p value of the difference between one given site and other sites located northward (Table S3 in the Supplement).The difference between sites was tested for mounds/polygons and hollows for all 5 sites and for permafrost subsidence/frost crack for three most northern sites (Khanymey, Urengoy, and Tazovskiy).The DOC and major elements (Ca, K, Al, Si, Fe) exhibited clear difference (p <0.05) between different geographic zones.The most pronounced difference between pair sites was observed for hollows.Thus, the porewaters from hollows in most southern site (Kogalym, of the sporadic permafrost) demonstrated statistically significant differences in DOC, Ca, K, Al, Si, Ni, Cu, Sr, and Rb concentrations from hollows of Khanymey, Pangody, Urengoy, and Tazovskiy.Among the elements listed in Table 2, DOC, Ca, Fe, and Sr were found to be most sensitive to the latitude of the sampling site regardless of the type of micro-landscape.
The general latitudinal trend in element concentration together with mean values in each micro-landscape as a function of latitude was examined for all major elements and TEs.The latitudinal trend was approximated by a linear regression using all micro-landscapes and individually for hollows and mound/polygons: where A and B are the element-specific empirical coefficients.Parameters of equation for each element are listed in Table 3.For most major components including DOC there was no systematic trend of increasing or decreasing of average concentration across the 640 km latitudinal profile.There was a local maximum of DOC concentrations in porewaters of peat mounds sampled at the Khanymey-Urengoy sites.
Overall, three patterns of concentration-latitude dependence could be distinguished, shown in Figs.     1. Specific conductivity, pH, DIC, DOC, K, Na, SO 4 , Si, B, Li, Fe, Ti, Cr, Ba, Mo, As, light REEs (La, Ce), W, and U did not exhibit any statistically significant trend (R 2 < 0.5) or this trend was within the uncertainties, as illustrated in Figs.3a-h and S4E-K.
2. A clear trend of steady increasing concentration northward was observed for SUVA 280 , Mg, Ca, Al, Cu, V, Mn, Ni, Sr, heavy REEs, Zr, Hf, and Th (0.45 < R 2 < 0.62, p < 0.05).The overall increase from the sporadic to continuous permafrost zone ranged from a factor of 2 to a factor of 5, illustrated in Figs.4a-h and Fig. S5A-F.

Dissolved organic carbon transport in peat soils
The first unexpected result of this study was the lack of significant decrease in DOC concentration in peat porewaters northward, from the sporadic to discontinuous and continuous permafrost zone (Fig. 3c).The character of the DOM also remained highly constant across the latitudinal/permafrost gradient as the SUVA 280 ranged from 2.4 to 3.5 L mg −1 m −1 in all sites regardless of the micro-landscape, with a weak increase northward (Fig. 4a).These values of SUVA 280 are consistent with those of the lakes (2 to 4 L mg −1 m −1 ; Manasypov et al., 2015) and rivers (2 to 3.5 L mg −1 m −1 ; Pokrovsky et al., 2015) of the region during summer period.The previously published values of SUVA 280 in WSL surface waters were similar across a large scale of lake size (from 50 to 500 000 m 2 ) and latitudinal position of the river watershed (from 57 to 66 • N).This strongly suggests highly uniform feeding of Siberian inland waters by allochthonous DOM originating from peat leaching within the soil profile.The DOC transport to the river and lake presumably occurs via suprapermafrost flow over the frozen peat layers at the depth ranging between 20 and 80 cm depending on the season, the latitude, and the micro-landscape context (see Fig. 2).Given the similarity of SUVA 280 values across significant geographical transect on positive forms of micro-relief (Fig. 4a, Table 3), we hypothesize the similarity of the nature of watersoluble OM that constitutes the peat on mounds.At the same time, sizeable increase in the SUVA 280 nm northward may indicate a higher aromaticity of soil porewater DOM in the continuous permafrost zone relative to the discontinuous and sporadic permafrost zone (Fig. 4a).The change of SUVA from 2.4 to 3.4 in hollows demonstrates a significant shift in the composition of the DOM and may have a pronounced effect upon the biogeochemical processing of DOM upon ex-port, as has been recently shown in eastern Siberia (Frey et al., 2016).This contradicts the conclusion reached in recent studies of surface waters and soil leachates that the DOM leached from the permafrost soil layer has a consistently lower concentration of aromatic carbon (i.e., lower SUVA 254 values; Mann et al., 2012;Cory et al., 2013Cory et al., , 2014;;Abbott et al., 2014;Ward and Cory, 2015), compared to DOM draining from the active, organic surface layer.However, the majority of previous studies dealt with non-peat permafrost environment.In the case of the WSL peatland, the contribution of UV-transparent microbial exometabolites and plant exudates including low-molecular-weight organic acids (i.e., Giesler et al., 2006) is certainly much higher in the southern forest-tundra and taiga zone compared to northern sites of the polygonal tundra.In the present study, statistically significant increase in SUVA 280 northward in hollows (R 2 = 0.599; see Table 3) may also indicate the lower rates of DOM processing in soils in the north, linked to either shorter residence time of soil fluids or weaker processes of photo-and biodegradation in the continuous permafrost zone compared to sporadic and discontinuous zone.
Generally higher DOC concentration in porewaters of mounds compared to that of hollows (Table 2) has two possible explanations.The soluble DOC retainment by clay horizon that underlays the peat in the WSL was hypothesized as the main regulator of the DOC level in rivers of large latitudinal transect of WSL, from the permafrost-free to continuous permafrost zone (Pokrovsky et al., 2015).The gradient consisted in increasing the DOC concentration northward of 64 • N (Pokrovsky et al., 2015) because the DOC-adsorbing clay horizon that underlays the peat may be frozen in the north (Kawahigashi et al., 2004).The latter authors suggested that the DOC in northern, permafrost-affected tributaries of the Yenisey River was less biodegradable (and thus better preserved during its transport from the soil to the river) than that in southern tributaries.If true, the lower DOC concentrations in hollows and subsidence relative to the mounds observed in the present study is due to DOC adsorption on unfrozen mineral layers (silt, clays) located below the peat horizon in depressions and hollows, which have much deeper position of the ALT than the mounds (see Table 1 and Fig. 2).At the same time, if soil porewaters are affected by the presence of minerals, then it should impact primarily the lithogenic elements (Ca, Mg, Sr, Si, Ti, Al, Zr, etc.) whose concentration should be higher in negative forms of relief relative to that in the positive ones.This hypothesis is not supported by the concentration pattern of inorganic constituents of porewaters as shown in the next section.Note also that, because the mounds thaw later than hollows, the period of unfrozen exchange of constituents in the soil with porewater is shorter in mounds compared to hollows.However, this does not go in line with the observed difference of higher DOC and metal concentration in porewater of mounds relative to hollows.The second explanation of the elevated DOC concentration in mounds compared to hollows across the whole permafrost gradient is related to the time of reaction between the peat and the pore fluids.From detailed hydrological studies on frozen peatbog of western Siberia, the water residence time in peat mound is a factor of 14 higher than that in hollows and depressions (Novikov et al., 2009).The latter have much higher hydrological connectivity to surrounding streams and temporary water channels and as such offer shorter contact time and pathways of vertically infiltrating and laterally migrating water.During the summer baseflow period, up to 70-80 % of watershed covered by mounds in frozen peatland of western Siberia may remain disconnected from the hydrological network (Batuev, 2012).The mounds and polygons are therefore essentially controlled by water evaporation, leading to evaporative concentration of DOC and other solutes within the soil profile.The available data on water infiltration parameters of hollows and permafrost subsidences located in the discontinuous permafrost zone of the WSL demonstrate an order of magnitude faster water migration in various depressions (hollows, subsidences) compared to mounds (Novikov et al., 2009, and unpublished data of the authors on NaCl tracer migration in frozen polygons and palsa peatbogs of the WSL).The density of the peat in the mounds and polygons is a factor of 2 to 10 higher than that in the hollows and depressions (Ivanov and Novikov, 1976).Thus, an analogy of ground surface and deep peat can be used for comparison between negative and positive forms of microrelief, respectively.In the peatland-dominated zone of discontinuous permafrost, the total porosity was reported to drop by about 10 % between the ground surface and 35 cm depth; however, the active porosity decreased by as much as 40 % over the same distance (Quinton et al., 2000).The saturated hydraulic conductivity of peat decreases rapidly with depth (Quinton et al., 2009).It thus can be hypothesized that, in the dense peat on mounds and polygons, the pores are significantly smaller with less interconnection, which leads to more restricted flow and greater tortuosity (Rezanezhad et al., 2009(Rezanezhad et al., , 2010(Rezanezhad et al., , 2016)).All these factors should increase the water residence time in pores of peat in mounds relative to hollows and allow for efficient enrichment of peat porewater by DOC in the former.
The DOC porewater concentration invariance across the latitudinal gradient of the WSL is consistent with the lack of peat thickness and thermal regime effect on porewater chemistry.First, the peat thickness did not exert a direct impact on the degree of porewater enrichment in DOC among various micro-landscapes: there was no dependence between the DOC concentration in porewater and the total thickness of the peat (R 2 < 0.01, not shown).Second, the thermal regime of soil porewater is responsible neither for the difference between mounds and hollows nor for latitudinal dependence of DOC concentration.The effect of temperature on peat leaching in aqueous solution is not known, but by analogy with surface-controlled dissolution reaction of minerals (i.e., Schott et al., 2009) it can be by a factor of 2 to 3 for each 10 • C rise.Such a large difference in 10 • C between different adjacent micro-landscape seems highly unlikely.This is confirmed by both our field measurements in Tazovsky (mean annual temperature of peat at 5 cm depth is equal to −1.9 • C in mound and +1.9 • C in hollow), and the observations of other researchers in the WSL.In the Nadym region (discontinuous permafrost zone), the mean annual temperature of mounds and hollows is 1.0 and 1.6 • C, respectively (Bobrik et al., 2015).At the latitude of Urengoy-Tazovsky and Khanymey, the average difference between mound and hollow of summertime temperature at 20 cm depth is 2.9 and 3.4 • C, respectively (Novikov et al., 2009).A similar difference of peat temperature between mounds and depressions at 20 cm depth (<4 • C) was reported for the Noyabrsk region (discontinuous permafrost zone; Makhatkov and Ermolov, 2015).Globally, the temperature of soil porewater across the latitudinal gradient does not exceed 10 • C (Novikov et al., 2009), which is not sufficient to exert any pronounced control on DOC concentrations.
To summarize, we hypothesize that (i) the DOC concentration should be controlled by the DOC residence time and travel pathway through the organic topsoil and (ii) the enrichment in DOM of the interstitial soil solution occurs via lichens, moss, litter, and peat leaching.Although the runoff is known to exert the primary control on stream DOC export from the boreal peatland catchments (Olefeldt et al., 2013;Leach et al., 2016), the existing hydrological modeling of subsurface transport of dissolved carbon in a discontinuous permafrost zone suggests that both concentration and load of DOC are water-flow-independent (Jantze et al., 2013).As such, it is the time of reaction between the peat and downward infiltrating waters that essentially controls the degree of peat porewater enrichments in DOC.This time is presumably similar across significant permafrost and climate gradients.

Factors controlling major and trace element concentration in peat soil porewaters
Organic and organo-Fe, Al colloids dominate the speciation of most cations (including alkaline-earth metals) and TEs in humic surface waters of permafrost-affected WSL territory (Pokrovsky et al., 2016b), similar to other boreal catchments (Köhler et al., 2014).As a result, the behavior of many major elements and TEs in peat porewater is likely to follow that of DOC, Fe, and Al as main colloidal carriers.The importance of colloidal Fe and Al as primary carriers of TEs in peat soils is confirmed by results of this study: in porewaters, none of the TEs correlated with DOC (R < 0.5), whereas Fe and Al concentrations correlated with many TEs such as Ti, V, Cr, Co, Ni, As, Sr, Zr, Nb, heavy REEs, and Hf.This is consistent with decoupling of TE 3+ and DOC during size separation procedure as two independent colloidal pools (high-molecular-weight Fe, Al-rich and lowmolecular-weight C org -rich), already demonstrated for Eurowww.biogeosciences.net/14/3561/2017/Biogeosciences, 14, 3561-3584, 2017 T. V. Raudina et al.: Dissolved organic carbon and elements in peat porewater of western Siberia pean boreal rivers (Neubauer et al., 2013;Vasyukova et al., 2010) and other Siberian rivers and WSL thermokarst lakes (Pokrovsky et al., 2006(Pokrovsky et al., , 2011(Pokrovsky et al., , 2016b)).At the same time, although organo-ferric and organo-aluminum colloids are certainly important factors of insoluble element transport in peat soil, the source of TEs may become more limiting for overall concentration of TEs in soil porewater than their speciation.There are two possible sources of "lithophile" elements in the peat and peat porewaters: atmospheric dust deposition at the moss and lichen surface and upward migration of soil fluids that carry mineral particles from underlying loam horizons.The loam horizons are rich in silicate clay minerals (e.g., Ovchinnikov et al., 1973;Golovleva et al., 2017) that contain insoluble elements.The geochemical analysis of TE distribution in WSL peat cores across the studied permafrost gradient allowed to distinguish several categories of TEs depending on their source such as soluble atmospheric aerosols, atmospheric dust, underlying mineral layers, plant biomass, and surface water flooding (Stepanova et al., 2015).The atmospheric deposition of lithogenic elements in the form of soluble aerosols on the moss surfaces followed by incorporation into the peat is expected to be low as shown by thorough snow analyses across the large WSL gradient (Shevchenko et al., 2016).Therefore, atmospheric dust seems to be the main source of insoluble metals in WSL peat as it is also known from other northern bogs (Shotyk et al., 2016a).Regardless of the origin of lithophile elements, we hypothesize that the leaching of insoluble trivalent and tetravalent hydrolysates (TE 3+ , TE 4+ ) from solid phase to interstitial soil solution may be restricted by the availability of silicate clay minerals within the peat core.
Based on results of the PCA treatment (Fig. S3A, B), the dendrogram of a hierarchical cluster (Fig. S3C) and the correlations between elements (Table S1) we hypothesize that the source of Cr, V, Al, REEs, Nb, Zr, Hf, Th, and U but also of Mg and Li is silicate minerals dispersed within the peat matrix.These elements exhibit the highest correlation with Si in porewaters and appear to be linked to the first factor (F1) of the PCA.The silicate minerals may originate from both atmospheric dust and underlying clay/silt horizons.The lack of correlation of K, Rb, Mn, Ba, Mo, W, Zn, Pb, Cd, Cs, and Sb with DOC, Fe, or Al in peat porewaters of WSL (Table S1) can be explained by specificity of these elements.In particular, K, Rb, Mn, Cu, Ba are biotically controlled by moss growth and thus unlikely to be linked to any mineral source (Stepanova et al., 2015).It seems also plausible that indifferent oxyanions (Mo, Sb, W) or disperse pollutants delivered by atmospheric deposition on moss surface followed by incorporation into peat (Zn, Cd, Pb, Sb, Tl) do not exhibit significant correlation with main colloidal components.
One can expect that dissolved element decreases its concentration in the peat porewater northward regardless of the micro-landscape due to (i) decrease in the thickness of peat deposits in total and the active soil (peat) layer in particular (Beilman et al., 2009;Novikov et al., 2009: Stepanova et al., 2015), which decreases the amount of peat interacting with downward-penetrating fluids; (ii) decrease in plant biomass (Frey and Smith, 2007), which diminishes the amount of plant litter that can release the elements (Pokrovsky et al., 2006;Fraysse et al., 2010) and also decreases the plant's ability to weather minerals within the soil profile (Moulton et al., 2000); (iii) shortening the unfrozen period of the year leading to the decrease in the residence time of water in soil pores; and (iv) overall decrease in the intensity of chemical weathering, CO 2 consumption, and riverine fluxes with mean annual temperature decrease (Dessert et al., 2003).However, an unexpected result of this study was that the overwhelming number of major elements and TEs did not exhibit any statistically significant decreasing trend of concentration with latitude.Instead, we observed a measurable northward increase in concentration of a number of lithogenic elements, whose presence is known to mark the intensity of mineral weathering.These are Mg, Al, Ti, V, Sr, REEs, Zr, Hf, and Th, originated from silicate minerals of the soil profile.For example, Al, Ba, Fe, and Mn were reported to reflect the mineral weathering as they exhibited elevated concentrations in Alaskan rivers during the late fall, which correlated with the maximal depth of the thawed active layer (Barker et al., 2014).The mechanism related to enhanced mobilization of low-soluble elements during deepening of the ALT is penetration of DOM-rich surface fluids to deeper soil horizon and leaching of lithogenic elements from underlying mineral substances, in the form of strong organic complexes (chelates).This mechanism can be tested via comparison of lithogenic element concentration in contrasting micro-landscapes.Thus, Sr, which is considered as an indicator of mineral sources in surface waters of the permafrost zone (Keller et al., 2010;Bagard et al., 2013), was highly similar between mound and hollow or even higher in mounds than in hollows or subsidences (Table 2).Given that the negative forms of relief in the WSL exhibit higher proximity of thawed layer to the mineral horizon because of lower thickness of peat and deeper ALT (Tyrtikov, 1973;Lupachev et al., 2016), the lack of link between Sr concentration and ALT position within the peat-silt/clay profile suggests that the underlying minerals do not participate in feeding the soil solutions by lithogenic elements.Rather, aeolian (long-range) dust deposits throughout the territory may lead to incorporation of solid atmospheric particles into the moss biomass.Subsequently, it is the dissolution of agglutinated minerals that enriches the peat porewater in lithogenic elements, including Si.Moreover, the concentration of elements likely originated from silicate matrix (Al, Si, Fe) in hollows and subsidences did not exceed that in mounds.Taken into account that the position of the permafrost boundary is much closer to the mineral substrate in negative forms of relief compared to mounds (see Table 1 and Fig. 2), this strongly suggests the lack of element leaching from the underlain mineral matrix.As such, the observed trends of element concentration with latitude reflect the leaching of essentially peat constituents with associated silicate particles without interferences with massive deposits of underlying sand, clay, and silt in various micro-landscapes.Following the same reasoning, the lack of DIC, Mg, and Ca variation among the micro landscapes suggests a negligible role of silicate and carbonate mineral weathering within the peat profile.
In addition to evaporative concentration mechanism and the greater residence time of solutes in mound compared to hollows, identified for DOC pattern in Sect.4.1, the peat chemical composition may be different between negative and positive forms of relief and thus it can contribute to porewater enrichment in major elements and TEs.Indeed, the degree of peat decomposition and elementary content of peat on mounds is higher than that on hollows and depressions (Stepanova et al., 2015): a comparison of peat elementary composition at 15 cm depth at the Pangody site demonstrated a factor of 1.5 to 3.5 higher concentration in mounds compared to hollows of major (Ca, K, Na, Fe) and ∼ 40 TEs except Mg, Zn, Sb, and Pb (a factor of 1.3 to 3 richer in hollows than in mounds).
The lack of increase in Cl, SO 4 , and Na in peat porewaters from the most northern site (Tazovskiy) compared to the intermediate sites (Urengoy, Pandogy) dismisses the possibility of element leaching from frozen saline sediments abundant in the Russian Arctic Coast (e.g., Brouchkov, 2002).Presumably, these saline sediments are not in contact with soil and suprapermafrost waters even at the time of maximal ALT, as also inferred from riverwater geochemistry in the permafrost-affected region of WSL (Pokrovsky et al., 2015).The elements originated from marine aerosols such as Na, Cl, SO 4 , B, Li, Rb, and Cs exhibited a decreasing or indifferent, but not increasing, trend of concentration northward.This precludes a strong influence of marine atmospheric deposition on surface water chemistry, unlike it was suggested in earlier works in this region (Syso, 2007;Smolyakov, 2000).

Comparison of peat porewaters with rivers and thermokarst lakes
The peat soil porewaters sampled above the position of the permafrost table can serve as representative sources of water and solutes prior to export to the thermokarst lakes and rivers (Fig. 2).Therefore, a first-order comparison of concentrations between these aquatic systems allows evaluation of the role of peat (shallow surface) versus mineral (deep subsurface and underground waters) feeding of Siberian inland waters.This comparison was based on mean values of DOC and TE concentration in porewaters for the whole permafrost-affected WSL territory (Table 2) and those previously published for lakes and rivers of the same latitudinal gradient (Manasypov et al., 2014;Pokrovsky et al., 2015Pokrovsky et al., , 2016a)).The dissolved components measured in rivers and lakes during summer period can be classified into three categories: (1) rivers or lakes exceed soil porewaters by a factor of 3 to 10, (2) river or lakes are similar to porewaters within a factor of 2, and (3) rivers or lakes are significantly lower (more than a factor of 3) than the porewaters.The elements of the first category are DIC, Ca, Mg, Si, B, Al, Mn, Na for rivers and only Si for lakes.The second category comprises DOC, Li, K, Rb, Fe, Ni, Co, Cr, As, Sr, and U for rivers and Li, B, Na, K, Rb, Cs, Ca, Mg, Ti, V, Mn, Ni, Cu, Zn, Co, Cd, Sr, Mo, As, and Sb for lakes.The third category includes Ti, Cu, Pb, Cd, Mo, and REEs for rivers and DOC, Al, Fe, Ga, Y, Zr, Ba, W, REEs, Th, and U for lakes.This first-order comparison demonstrates that the soil porewaters alone are sufficient to provide the concentrations of all major elements and TEs in lakes.In other words, the transport of soil porewaters along the permafrost boundary in the form of suprapermafrost flow may be the sole source of incoming solutes to thermokarst lakes of western Siberia, across all three permafrost zones.This hypothesis is fully consistent with the lack of any underground feeding of WSL thermokarst lakes, demonstrated in earlier studies (Manasypov et al., 2015).
In contrast to lakes that can be fully supplied by solutes from surrounding peat porewaters, the rivers require some "mineral" influx in addition to surface and shallow subsurface "organic" influx in order to explain the elevated concentrations of DIC, Ca, Mg, Na, Si, and Al in the riverwater relative to the peat porewater.This influx, mostly pronounced during summer baseflow period, may include the groundwater seeping via taliks on the river bed and shallow subsurface flow over clays and silt deposits.This process is fairly well known for other, non-peatland permafrost setting (MacLean et al., 1999;Bagard et al., 2011;Barker et al., 2014;Tank et al., 2016).
The latitudinal dependences of element concentration in the peat porewater revealed in this study can be compared to the latitudinal dependences of DOC and element concentration in adjacent thermokarst lakes and rivers.The elementary trends in the inland waters of western Siberia were associated with the influence of marine aerosols or long-range atmospheric transport of industrial pollutants in lakes (Manasypov et al., 2014) and the evolution of chemical composition of the peat and underlying mineral deposits in rivers (Pokrovsky et al., 2015(Pokrovsky et al., , 2016a)).However, the possible links are not straightforward and valid only for a small number of elements.Thus, increasing concentrations of Ca, Ni, and Sr (Fig. 4c, g, h, respectively) and decreasing concentration of Sb and Pb (Fig. 5d and e, respectively) northward are consistent with the trend in thermokarst lakes of western Siberia from 63 to 71 • N (Manasypov et al., 2014).However, the other elements exhibiting a clear increasing (K, Cu, Mo) or decreasing (V, Ba) latitudinal trend in lakes (Manasypov et al., 2014) do not show such a trend in peat porewaters sampled in this study.Presumably, variable and simultaneously acting processes control the delivery of element from the peat core to the adjacent lakes over the permafrost gradient.
Because the leaching of peat constituents by downwardpenetrating fluids is very fast and weakly depends on temperature and local hydrological pathway within the peat pores, one can expect that the global hydrological setting will primarily control the peat weathering intensity.As such, it is the amount of water that passes through the peat soil column before being evacuated to the river that defines the overall export fluxes of elements from the peatland to the hydrological network.This prediction is consistent with reported higher riverine fluxes of DOC, Si, and cations in the northern region of the WSL (66.5 to 67.5 • N) relative to the southern region (62-65 • N) of this territory corresponding to higher surface runoff in the north (Pokrovsky et al., 2015).
The fluxes of Ca, Mg, and HCO − 3 ions carried by rivers are used for calculation the CO 2 uptake flux due to chemical weathering, i.e., reaction of atmospheric CO 2 with Ca, Mgbearing silicate minerals (Dessert et al., 2003;Beaulieu et al., 2012).Not more than 10 % of total riverine flux of Ca, Mg, and HCO 3 is considered to be due to atmospheric input.An important consequence of our obtained results on soil porewaters in the WSL is that the intensity of chemical weathering and associated CO 2 consumption in the permafrost regions (i.e., Beaulieu et al., 2012) by small rivers without pronounced underground feeding in peatlands could be overestimated relative to the regions with shallow organic soil horizons.As a result, the flux of DIC and major cations in the peatland-draining rivers should be corrected for the input of these elements via peat porewater discharge to the river main stream.For a number of small rivers (S watershed < 1000 km 2 ) in the permafrost zone of the WSL that are fed by shallow surface runoff through the peat horizon, this correction can range from 20 to 80 % of total riverine DIC, Ca, and Mg flux.The global consequence of this correction is that the continental-weathering CO 2 sink in northern peatland regions might be a factor of 2 to 4 smaller than that currently deduced from the fluxes of large rivers.

Prospective for climate change in western Siberia
In accordance with a common scenario of the climate change in the subarctic, a shift of the permafrost boundary further north and the increase in the active layer thickness are anticipated in the WSL (Pavlov and Moskalenko, 2002;Frey and McClelland, 2009;Moskalenko, 2009;Romanovsky et al., 2010;Vasiliev et al., 2011;Anisimov et al., 2013).This agrees with large-scale permafrost shifts consisting in southern boundaries moving northward (see Walvoord and Kurylyk, 2016, for a review).Assuming a scenario substitution space for time, and upscaling the data of peat porewaters obtained in this study, we predict that the shift of the permafrost boundary northward even by 2 • latitude will not affect the concentrations of most major elements and TEs in peat porewaters.The concentrations of DOC, DIC, Ca, Mg, K, Al, Fe, and trace metals in the continuous permafrost zone may remain constant or decrease by a factor of 1.5 to 2, which is often within the natural variation between different microlandscapes, soil depths, and seasons.
The ALT is projected to deepen more than 30 % during this century in the Northern Hemisphere (Anisimov et al., 2002;Stendel and Christensen, 2002;Dankers et al., 2011).As a general scenario in frozen peatlands of the subarctic, this increase will bring about the involvement of mineral horizons into the water infiltration zone down the soil profile (Walvoord and Kurylyk, 2016).The degradation of peat mounds and polygons will be accompanied by the spreading of hollows and depressions (Pastukhov and Kaverin, 2016).As a result, the water coverage of the watershed will increase thus enhancing the anaerobic conditions.On the one hand, this will increase the fraction of hollows and depressions containing less concentrated interstitial soil solutions and thus the stock of DOC, major elements and trace metals in soil fluids will decrease.On the other hand, the increasing anaerobic conditions may preferentially mobilize redox-sensitive elements (Fe, Mn, Cr, V, etc.) from the peat to the porewaters.Overall, the share of spring runoff from the mounds to the rivers and lakes will decrease, whereas during the summer baseflow, the input from the hollows and depressions to the hydrological network will increase.
The concept "substituting space for time" allows foreseeing the consequences of soil warming in the continuous permafrost zone of the WSL peatlands on the adjacent river chemistry and export of carbon and metals from the watersheds.This prediction can be made only for small rivers of the WSL (e.g., watershed area < 10 000 km 2 ) which drain the adjacent peatlands, have no underground feeding, and flow essentially during the unfrozen period of the year (see Pokrovsky et al., 2015Pokrovsky et al., , 2016a)).For this, two basic scenarios can be considered: (i) a constant latitudinal pattern of permafrost distribution (no boundary migration) but complete disappearance of peat mounds and their replacement by hollows and depressions and (ii) a shift of the permafrost boundary to the north and transformation of the continuous permafrost zone into the discontinuous and transformation of the discontinuous permafrost into the sporadic without changing the micro-landscape distribution.We assume as a first approximation that there is no change in precipitation, evapotranspiration, and riverine runoff in the northern part of WSL (60-68 • N), given that the drying trend will be pronounced only in the regions located to the south of 60 • N (Alexandrov et al., 2016).
The first scenario yields a decrease in the concentrations of DOC, DIC, major cations, and trace metals in porewaters of the continuous permafrost zone by not more than 30 %.This estimation stems from the maximal difference in element concentration between mounds and hollows (Table 2) and typical proportion of mounds in the terrestrial landscape of the WSL (35 ± 15 %; Novikov et al., 2009, and authors' unpublished data).The second scenario is based on the latitudinal patterns of element concentration in the peat porewaters S4,S5).For this, a linear dependence of element concentration in all micro-landscapes on the latitude given in Figs. 3 to 5 can be used.A 2 • northward shift in the position of the permafrost boundaries will bring about a factor of 1.3 ± 0.2 decrease in Ca, Mg, Sr, Al, Fe, Ti, Mn, Ni, Co, V, Zr, Hf, Th, and REEs concentration in continuous and discontinuous permafrost zones.Note that a possible decrease in DOC, SUVA 280 , Ca, Mg, Fe, and Sr will not exceed 20 % of their actual values.Finally, there may be an increase in Cl, Na, K, Rb, Cs, Zn, P, and Sb concentration by 30 ± 10 %.In both scenarios of permafrost thawing in the WSL peatlands we do not expect any sizeable increase in soil porewater concentration in DOC and metal and enhancement of the export of solutes by small-size rivers which are not connected to the underground reservoirs.This contradicts the dominating paradigm of the increase in DOC, DIC, major cations, and metal discharge from the land to the ocean upon the ongoing climate warming in other permafrost regions.Combining both scenarios of permafrost thaw (northward permafrost boundary shift and extending the hollows over mounds) suggests that, over the first decades, relatively fast permafrost coverage shift will not be accompanied by the change of micro-landscapes and thus the overall decrease in DOC and metal concentration in peat porewaters will be around 20 to 30 %.The average rate of peat formation in Siberian flat-mound bogs is 0.24 mm y −1 (Inisheva et al., 2013).Thus, taking into account the climate warming and accelerated peat growth, after 500 to 1000 years which are necessary to form the new ca.20 cm peat layer, the second scenario will take over and thus up to 2-fold cumulative element concentration decrease in soil fluids of the continuous permafrost zone may occur.Assuming a dominant feeding of small rivers by soil porewaters transported along the permafrost boundary, a slight decrease (i.e., < 30 %) of riverine transport of DOC, DIC, Fe, Al, Ca, Mg from the northern part of the WSL territory to the Arctic Ocean is anticipated.This decrease will be mostly pronounced for small rivers such as those of the Arctic coastal zone.

Conclusions
A snapshot of peat soil water chemistry allowed for quantifying the distribution of DOC and major and trace elements in peat porewaters at the end of the active period across a sizeable gradient of permafrost.We did not confirm a trend of diminishing DOC and metal concentration in peat porewaters northward, despite a decrease in mean annual temperature, vegetation density, and the active layer thickness.DOC, DIC, and most major elements and TEs did not exhibit any statistically significant trend of concentration with the latitude.A clear trend of increasing concentrations of Mg, Ca, Al, Ti, V, Ni, Sr, heavy REEs, Zr, Hf, and Th marked the increase in the influence of silicate mineral weathering.Concentrations of DOC, SO 2− 4 , B, V, Cs, and Th in porewaters in the peat mounds usually exceeded those in hollows and permafrost subsidences.The water residence time in peat of various densities and the peat chemical composition were hypothesized to be the main factors controlling the degree of element leaching from the peat column to the pore fluids.Applying a "substituting space for time" approach for the climate warming scenario in the WSL, we predict that the northward migration of permafrost boundary and the replacement of thawing frozen peat mounds and polygons by hollows, depressions, and subsidences will decrease the concentrations of DOC, DIC, major cations, and trace metals in porewater of the continuous permafrost zone by a factor of 1.3 ± 0.2.This in turn will decrease the feeding of small rivers and lakes by peat soil leachates and the overall export of DOC and metals from the WSL territory to the Arctic Ocean may decrease.As such, the dominating paradigm of the increase in DOC, DIC, major cation, and metal export fluxes upon the ongoing climate warming in boreal and subarctic regions should be revised for the case of frozen peatlands.
Data availability.The full data set of major and trace element concentration in porewaters (< 0.45 mum) across the latitudinal profile of Western Siberia Lowland is available at Research Gate, https://www.researchgate.net/publication/313058330_Element_concentrations_in_peat_soil_solutions_across_the_ micro-landscapes_and_permafrost_zones_of_western_Siberia_ peatlands (Pokrovsky, 2017).
Competing interests.The authors declare that they have no conflict of interest.Special issue statement.This article is part of the special issue "Changing Permafrost in the Arctic and its Global Effects in the 21st Century (PAGE21) (BG/ESSD/GMD/TC inter-journal SI)".It is not associated with a conference.

Figure 3 .
Figure 3. Mean values of specific conductivity (a), pH (b), DOC (c), DIC (d), SO 2− 4 (e), Si (f), Fe (g), and Ti (h) concentration in peat porewaters of the WSL as a function of latitude for mound and polygons (solid diamonds), hollow (open diamonds), frost crack (grey triangles), and permafrost subsidence/depression (hatched circles).The solid line is a linear fit to all data with the regression equation given on each graph.

Figure 4 .
Figure 4. Mean values of SUVA 280 (a), Mg (b), Ca (c), Al (d), Cu (e), V (f), Ni (g), and Sr (h) concentration in peat porewaters of the WSL as a function of latitude for mound and polygons (solid diamonds), hollow (open diamonds), frost crack (grey triangles), and permafrost subsidence/depression (hatched circles).The solid line is a linear fit to all data with the regression equation given on each graph.

Figure 5 .
Figure 5. Mean concentrations of Cl (a), Zn (b), Cd (c), Pb (d), Sb (e), and Rb (f) in peat porewaters of the WSL as a function of latitude for mound and polygons (solid diamonds), hollow (open diamonds), frost crack (grey triangles), and permafrost subsidence/depression (hatched circles).The solid line is a linear fit to all data with the regression equation given on each graph.

Table 1 .
Physico-geographical, permafrost, and soil parameters of five study sites.

Table 3 .
Latitudinal trends of average element concentration in two main habitats persisting in all five study sites.L is for latitude ( • N) and R 2 is a linear regression coefficient (Eq.1).SC is specific conductivity.