INTRODUCTION

Gas exchange with the atmosphere is one of the most important ecosystem functions of soils, the relevance of which is currently undoubted [6, 9]. The soil is actively involved now and participated in the past in the formation of the gas composition of the atmosphere. Soil respiration (SR) is crucial for the global carbon cycle, since, according to available estimates, soils contain two times more carbon than the atmosphere [38, 44]. Annual emissions of CO2 into the atmosphere through SR on a global scale are about ten times greater than emissions from burning fossil fuels [18]. Annually, a tenth of atmospheric CO2 passes through the soil [40]. Thus, the connection of soil with climate change is undoubted, and one of the key questions is the reaction of soil to climate change, namely the magnitude and potential of this feedback [18].

Soil respiration is the total production of CO2 by soil components: plant roots, meso- and microbiota. This concept is not identical to the concept of “emission” of CO2 from the soil surface, since the second is rather a physical process of gas release from the soil into the atmosphere [10], the source of which can be, among other things, nonbiological processes. Along with this, part of the CO2 formed as a result of respiration can be sorbed by the soil and not emit through its surface [13]. In this paper, we will use the terms in accordance with these definitions.

There are many reasons for the absence of accurate estimates of CO2 emission both on regional and global scales and for uncertainty of the predictive modeling of this part of the carbon balance. The abundance of predictors affecting SR is the main reason [37]. They can be combined into several groups. The first one is soil properties: the quality, quantity, and availability of soil organic matter [25, 41, 42]; soil acidity and nutrient availability [46]; composition and activity of soil microbiota [31]; physical properties of soils: bulk density, porosity, particle size distribution [34]. The second group is the nature and state of the vegetation cover, since the root respiration and the nature of the plant material entering the soil depend on it [43, 50]. The third group is hydrothermal parameters. In this case, we can speak both about the global climate, as a predictor of the formation of certain ecosystems, and, narrower, about soil moisture and temperature regimes. Hydrothermal properties of soils are the most actively studied predictors of the production of greenhouse gases in soils [49]. Soil respiration, like any other biochemical process, is temperature dependent. An increase in SR with increasing temperature is exponential and is described with the temperature sensitivity coefficient Q10. According to extensive literature data, it varies over a very wide range, but the vast majority of values are in the range from 1.5 to 3 [53]. The influence of moisture content is described by more complex dependences; according to general ideas, this factor limits CO2 emission both in the low and high ranges of values [22]. It is customary to consider and model SR values as a function of both parameters [47].

Thus, a wide range of factors affecting SR and their complex relationships cause extremely high spatial and temporal variability of SR, which is difficult to model or predict [54]. The annual SR dynamics are most often well described by seasonal changes in hydrothermal parameters and phenological features of vegetation in various biomes, especially in forest ecosystems of the northern hemisphere [19, 26]. There are generalized estimates of CO2 emission for most biomes. At the same time, spatial variability, especially large-scale (local) variability, though widely discussed, remains much more difficult to describe using the usual set of predictors. Recent studies have shown that temperature is not always a factor in the spatial variability of emission [7]. Often, spatial variability at the local level exceeds that at the regional level [45]. Undoubtedly, all these circumstances determine the need for detailed studies of the processes that determine the SR value, the search for patterns of its spatial variation at the level of individual ecosystems, which is repeatedly emphasized in review papers [49]. This is especially true for regions, where there are objectively few such studies. In particular, for tundra ecosystems, it was noted that very few studies of carbon balance included SR measurements.

The north of Western Siberia is a unique region near the southern boundary of discontinuous permafrost with sharply contrasting landscapes in small areas. This region is a natural experimental site for monitoring all processes associated with rapid climate change [11]. The present study has several tasks: (1) assessing the CO2 emission in contrasting landscapes of the study region during the growing seasons of 2019–2022; (2) analysis of the factors that determine the spatial and interannual variability of the emission.

OBJECTS AND METHODS

Objects and environmental conditions of the study area. The intensity of fluxes of climatically active gases in the north of Western Siberia (Yamalo–Nenets Autonomous Okrug) at temporary observation sites was estimated in the vicinity of the city of Nadym. The research site was organized in 2009 by employees of the Faculty of Soil Science of Moscow State University on the interfluve of the Kheigiyakha and Levaya Khetta rivers, 40 km from the city of Nadym (Nadym district, Tyumen oblast) (65°18′52.8″ N, 72°52′54.2″ E). The experimental station is a seasonal tent camp, established annually during the work period. The study area is not disturbed and is not subjected to anthropogenic loads, except for local impact of linear structures (gas pipeline and road along it). Of the natural disturbances, fires are developed, traces of which are found in the soil cover of the entire territory. In areas with permafrost, active thermal erosion and thermokarst are observed. Bare palsa surfaces are locally subjected to wind erosion; blowouts are present on elevations. The study area represents the northern taiga–southern tundra ecotone with small areas of taiga and larger areas of southern tundra and peatlands; it belongs to the discontinuous permafrost zone [11]. On the territory of Western Siberia, continuous permafrost with a thickness of several tens of meters is recorded to the north of the Arctic Circle. To the south, permafrost is divided into two uneven layers, forming a “dovetail” structure in the cross section [15]. The 50-m surface layer occurs discontinuously between 66° and 64° N and is considered to be a result of the Late Holocene cooling. The second, much thicker layer, separated from the surficial permafrost by a 100–200-m-deep talik, extends to the 60th parallel and is considered a relic of the Pleistocene [2].

The climate of the area is moderately continental, with very cold winters. According to the weather station in Nadym (30 km north of the study area), the mean annual air temperature since 2004 is –4.5°С (with variation from –2.4 to –6.8°С); the mean annual precipitation is 550 mm (with variation from 466 to 687 mm) [36]. Two main types of landscapes differing in the presence of permafrost and the degree of hydromorphism are distinguished in the area of the station: automorphic forest landscapes, where permafrost is currently absent, and hydromorphic landscapes represented by a complex of oligotrophic bogs and specific permafrost ecosystems, namely flat and large-mound palsas with permafrost at a depth of 1–2 m. Three main ecosystems differing in geocryological conditions, the nature of plant communities, and soil properties were chosen for the study. All study sites are located close to one another within a small area and represent automorphic forest ecosystems and a complex of palsa mounds developed in hydromorphic and semihydromorphic conditions. The soils described at the key sites reflect the most characteristic conditions of soil formation in the study area (Fig. 1). Within some landscapes, specific structures or plant groups have been identified, which are also included in the study.

Fig. 1.
figure 1

Main landscapes of the study area: (1) forest ecosystems, (2) permafrost-affected palsas, (3) bog ecosystems, and (4) thermokarst depressions.

Full size image

The forest site is represented by a hummocky green moss–pine forest with a slightly pronounced mesorelief (absolute heights about 22 m a.s.l.). The tree layer is composed of Pinus sibirica, Larix sibirica, and Betula sp. The shrub layer includes representatives of the Ericaceae family. Mosses and lichens (Polytrichum strictum, Cladonia rangiferina, Sphagnum sp.) predominate in the ground cover. Parent rocks are sands of lacustrine–alluvial origin. Modern permafrost is absent; however, the formation of these landscapes is influenced by the paleocryogenesis: wedge-shaped pseudomorphs and paleocryoturbation features. These phenomena lead to the formation of a pronounced residual polygonal relief and high heterogeneity of the soil cover pattern and soil properties. The soil cover consists of podzols (Albic Podzols) and podburs (Entic Podzols) [8, 52]. The podzol/podbur profile (O–(E)–BHF–BC) consists of a 5–10-cm peaty litter (O), a bleached sandy albic horizon of varying thickness, sometimes fragmentary or absent (E), and a spodic (BHF) loamy sand horizon of up to 30 cm in thickness gradually turning into sandy rock of heterogeneous color: from gray to yellowish-brown. The reaction of the medium changes from strongly acidic in the organic horizon to acidic in the underlying mineral horizons. In the organic horizon, the total carbon content reaches 46%; in the mineral layer, it does not exceed 1%. Carbon stocks average about 6 kg C/m2 (for the entire studied profile, about 60 cm). Stocks of root biomass range from 1800 to 3000 g/m2. From 80 to 90% of the roots are concentrated in the upper peaty litter horizon. A noticeable number of roots is noted in the lower horizons, 5–15% of the total root biomass is the spodic horizon [4]. The maximum share was noted for roots of medium thickness in the upper horizon and fine roots ones in the lower horizons. On the territory of the experimental station, there are specific forest sites, namely lichen pine forests with absolute dominance of Cladonia rangiferina (L.) Weber ex F.H. Wigg in the ground cover. These ecosystems are fragmentary and occupy no more than 5–10% of the total area of pine forests. The soils are characterized by a developed albic horizon and are classified as tonguing (glossic) podzols. The thickness of the albic horizon can reach 40–50 cm; the peaty litter is very thin. The soils have low carbon stocks and low root biomass.

Palsas are represented by flat-topped mounds and large mounds. Flat-topped palsas are elevated by 0.5–1.5 m above the surrounding bogs and are characterized by flat or slightly sloping tussocky surface. The ground cover vegetation consists of Cladonia rangiferina, C. stellaris, C. sylvatica, Sphagnum sp.; in the shrub layer, there are Betula nana, Rubus chamaemorus, Ledum sp., Vaccinium uliginosum, Vaccinium myrtillus; herbs include representatives of the Cyperaceae family. Parent rocks are sands and loams (less often) of the lacustrine–alluvial origin. Permafrost is found at a depth of about 60 cm and represents frozen sands or loamy sands. The soil cover consists of a group of peaty permafrost-affected soils. Peat cryozems (Histic Oxyaquic Turbic Cryosols (Arenic)) with loamy sandy or sandy loamy texture are the most common. Their profile (O⎯TO–CR⊥) consists of the remains of lichens (O), one–two weakly or moderately decomposed oligotrophic peat horizons with a total thickness of up to 50 cm, and a cryoturbated horizon of sandy loamy texture and heterogeneous color (gray, whitish, brownish) underlain by the frozen sandy rock. The soil reaction varies from strongly acidic in peat horizons to slightly acidic in mineral horizons. The total carbon content in peat horizons is about 43%; in mineral horizons, <1%. The average ash content of peat is 9%. The C stocks in peat cryozems average 26 kg C/m2 (to the permafrost). The root biomass stocks in the soil are from 200 to 300 g/m2, with almost 100% of all roots located in the upper 6–10 cm of the peat horizon. There are no roots larger than 5 mm in the soil, the proportion of fine roots is from 25 to 45% [4]. The large-mound forms of the palsas are dome-shaped and elevated above the surrounding bogs by 3–4 m (absolute heights about 25 m a.s.l.). These peat mounds do not have a continuous vegetation cover: significant areas of bare peat have a cracked, disturbed surface. In the areas of peat mounds, thermal degradation processes are observed sporadically and lead to the development of thermokarst depressions. They are characterized by a sharp deepening of the permafrost table, a significant subsidence of the surface, and a change in plant communities.

Hydromorphic (bog) ecosystems are elongated waterlogged depressions (hollows) between palsas (absolute heights about 20 m a.s.l.). The bog water level is within 0.1–0.2 m. Vegetation is mainly represented by mosses of Sphagnum fuscum and herbaceous vegetation of the Cyperaceae family—Eriophorum vaginatum, Carex sp. Permafrost is absent at least within 2 m from the surface. The soil cover is homogeneous and is represented by oligotrophic peat soil consisting of 1–2 layers of weakly decomposed sphagnum peat with a thickness of 0.5 m and more. The hollow sites with typical oligotrophic peat soils (TO–TT) (Fibric Histosols) were selected as monitoring objects.

Research methods. All field studies were carried out in August. For this region, this is not the warmest month (July is the warmest), but it is a month with maximum soil temperatures, which is associated with seasonal warming and maximum soil thawing. Measurements of the CO2 emission from the soil surface were carried out for 2–3 weeks daily, excluding days with heavy rainfall. The total number of measurements annually ranged from 200 to 300. Most of the measurements were performed on palsas and in bog ecosystems as the main study objects. The measurements were carried out both at stationary sites located on several palsas and in surrounding bog ecosystems (from 6 to 12 repetitions at each, 5–6 times per season), and at other sites selected annually, using a regular or an irregular grid pattern, or using radial transects of 100–200 m in length with a measurement step of 10–20 m (20–40 repetitions). The CO2 emission from the soil surface was measured with the static closed chamber method [13]. The measurement chamber was a stainless cylinder (12 cm in height, 10 cm in diameter), which was installed either in a plastic base permanently placed into the soil to a depth of 2–4 cm, or directly on the soil surface with a depth of 2 cm. The measurements were carried out on the plots with previously removed vegetation. In dynamics, air samples from the isolated chambers were taken into syringes with a volume of 20 cm3 immediately after the chamber installation and then after 10–20 min of exposure (for sampling, the chambers had openings closed with rubber stoppers). The exposure time depended on the flux intensity. The CO2 concentration was determined in the field with portable gas analyzers with an infrared sensor RMT DX 6210 (accuracy of 0.002%) and LI-830 (accuracy of 0.001%). The CO2 emission was calculated by formula (1) [13] taking into account changes in the gas concentration in the chamber, barometric (atmospheric) pressure, temperature, chamber volume, and exposure time. The results were expressed in mg CO2/(m2 h).

$$Q = \frac{{\Delta CPMh}}{{100RT\Delta t}} = \frac{{{\text{3}}{\text{.18}}\Delta CPh}}{{T\Delta t}},$$
(1)

where Q is CO2 emission, mg CO2/(m2 h); ΔС is change in gas concentration in the chamber, ppm; Р is atmospheric pressure, kPa; h is chamber height, cm; T is temperature on the Kelvin scale, K; Δt is exposure time, min; 3.18 is coefficient taking into account the numerical values of the constants included in the formula (R, M))) and the ratio of dimensions (Pa/kPa, h/min, m/cm, mg/g, %/ppm).

The depth of the seasonally thawed layer was measured by probing with a pointed metal rod 10 mm in a diameter and 1.2 m in length according to GOST 26 262-2014, 2015. Air temperature were controlled during studies (sensors were installed for the entire study period); soil temperature was measured with Thermochron iButton TM data loggers (Dallas Semiconductor Corporation, TX, USA; resolution of 0.5°C, accuracy of ±1°C) and electronic thermometers TP3001 (resolution of 0.1°C, accuracy of ±1°C); volumetric soil moisture content was determined with a field moisture meter Field Scout TDR 100 (resolution of 0.1%, accuracy of ±3.0%) in the layers of 10–12 cm and 15–20 cm for mineral and organic soils, respectively. The temperature of the air, soil surface, and soil at depths of 20, 40, 60 cm was measured year-round using temperature loggers with a measurement frequency of every 4 h. Our microclimatic measurements and data from the database of the All-Russia Research Institute of Hydrometeorological Information—World Data Center—were used for the analysis of climatic conditions and calculation of the soil temperature regimes. The soil temperature regimes have been studied at the experimental site since 2014 [23, 24, 36]. The analyzed year is considered not as the calendar year (from January to December) but as the seasonal (hydrometeorological) year from September to August. This is due both to the time of reprogramming the sensors (in August) and to a more convenient calculation of the regimes. For example, this eases assessment of the impact of the winter period on the functioning of ecosystems in the summer, etc.

Soil temperature regime is the distribution of temperature in the soil profile and continuous changes in this distribution over time [12]. Various parameters are used to characterize it on the basis of temperature data on soil horizons or selected layers measured at different times and spatial intervals. Both standard parameters proposed by Dimo [5] and parameters actively applied abroad in ecology, climatology, botany, and soil science were used. One of them is the N-factor, i.e., the temperature index of the surface, a method of parametrization of the surface energy balance [33]. The summer (Nt) N-factors were calculated through the ratio of the sums of mean daily temperatures above zero on the soil surface to the same sums in the air for the same period. Winter N-factors (Nf) were calculated similarly using the sums of temperatures below zero and the sums of subzero air temperatures. The maximum influence on the winter (Nf) factors is exerted by the snow cover depth (the greater the thickness of the snow cover, the smaller the factor value); summer (Nt) factors are mainly influenced by the ground vegetation cover [29].

One-way analysis of variance (ANOVA) was used to test for differences in the CO2 emission, temperature, and soil moisture between ecosystems. Significance of differences was determined using LSD test at 95% and 99% probability levels (p < 0.05 and <0.1). Correlations between CO2 emission and environmental factors were assessed with Pearson’s correlation analysis. The calculations were performed using the StatSoft Statistica 8 program.

RESULTS

Climatic conditions during the study period and hydrothermal regime of soils. Of the four years, in which the studies were carried out, two years—2018/2019 and 2020/2021—were relatively cold according to mean annual air temperatures (Table 1). The other two years—2019/2020 and 2021/2022—were relatively warm, with both very warm winters and warm growing seasons. The year of 2018/2019 was distinguished by a very thick snow cover and abundant precipitation during the warm period. The rest of the years were similar in terms of rainfall. According to the annual temperature data for palsas and forest soils, which were obtained in the year-round monitoring, the last season stands out as the maximum soil warming. It should be noted that it was in 2022 that the greatest depths of seasonal thawing of soils were recorded both at the studied sites and at monitoring sites of the CALM program also located in the study area [55]. Obviously, this is a consequence of a warm winter, weak soil freezing (practically its absence), and a warm growing season. Note that, according to the Nadym weather station, the snow cover in the winter of 2021/2022 was not deep. However, the Nf values for this winter were the lowest, which indicates a significant isolation of soils from freezing and, probably, attests to a deeper snow cover in comparison with that measured at the weather station. The temperature of the soil surface is another important parameter that gives an idea of the amount of heat entering the soil. The maximum temperature was recorded for the last season during the period under review.

Table 1.   Annual temperature parameters of air and soil for the period 2018–2022 according to field micrometeorological observations and data from the Nadym weather station
Full size table

Data on the temperature of the upper soil horizons obtained simultaneously with the emission measurements for all sampling points in all years of observations are presented in Fig. 2a. The palsa soils with a shallow permafrost are the coldest soils. The temperature of the upper soil layer in palsa soil did not differ significantly in August of all the years of observation; a slight tendency for its increase from 5.9 to 6.6°С can be noted. The temperature of the upper horizons of forest soils did not differ significantly in 2019, 2021, and 2022 (the data obtained in 2020 data are scarce for statistical analysis) and varied from 10.2 to 11.7°C. Bog soils were the warmest in all the years of observation with the highest temperature in 2019 (an average of 15.9°C) and slightly lower temperature in 2020 (14.6°C). In 2021 and 2022, the temperatures of the upper horizons of bog soils did not differ significantly and averaged 13°С. Naturally, the bog soils are the wettest (Fig. 2b); the moisture content exceeds 60%, and varies little over the years. The soils of forest sites are characterized by the lowest moisture content of the upper horizons: 8–12%. The moisture content of the palsa soils also varies little over the years, its values fluctuate around 40%. The maximum spatial variability of soil temperature was observed for palsas (coefficient of variation (CV) was 30–50%), low variability was noted for soils of bog and forest ecosystems (CV 8–12%). Similar patterns are typical of the moisture content variability: the maximum variability in palsas, and the minimum one in bog soils. In 2021 and 2022, we studied thermokarst depressions, and hydrothermal conditions in them were close to those in bog ecosystems. The seasonal thawing depth at the standard observation sites (palsas) approximately doubled, from 47 to 85 cm on average (Fig. 2c). It should be noted that only points with thawing depth less than 1 m were included in the statistical processing. The number of points with deep thawing increased significantly over the years of observations, especially in 2022.

Fig. 2.
figure 2

Parameters of soil functioning in the study area: (a) temperature of the upper horizon, (b) moisture content of the upper horizon, (c) depth of the seasonally thawing layer, and (d) CO2 emission from the soil surface. Blue circles are palsas, green rhombuses are forest ecosystems, red squares are bog ecosystems, purple triangles are thermokarst depressions. Mean values and standard error of the mean are shown.

Full size image

CO2 emission from the soil surface. The maximum CO2 emission from the soil surface in the years of observations were recorded for zonal automorphic soils of forest ecosystems—podzols and podburs. They significantly differed from those for other ecosystems. According to the analysis of variance (p < 0.05), there were no interannual differences for forest ecosystems, the average values fluctuated in a narrow range: 540 ± 173; 490 ± 291, 483 ± 241, 502 ± 139 mg CO2/(m2 h) in 2019, 2020, 2021, and 2022 respectively (Fig. 2d). The CO2 emission from the surface of flat-topped palsas (mainly, Histic Cryosols ) and bog soils differed significantly in 2019 and 2020; no differences were found in 2021 and 2022. On average, the CO2 emission for the palsa–bog complex was 2–2.5 times lower than for forest ecosystems. In the first two years of observation on palsas, it did not differ significantly and amounted to 186 ± 116 and 182 ± 106 mg CO2/(m2 h); in the next two years, it was somewhat lower: 150 ± 90 and 153 ± 67 mg CO2/(m2 h). For bog ecosystems, CO2 emission in 2019 and 2020 reached 251 ± 140 and 262 ± 134 mg CO2/(m2 h). In in the next two years, it did not differ significantly from that in palsas and amounted to 156 ± 90 and 141 ± 104 mg CO2/(m2 h). Annual (seasonal) CVs were high for all ecosystems: 30–60% in forests and 50–75% in palsas and bogs. The CO2 emission from young thermokarst depressions directly on the studied palsas, did not differ from that for palsas in 2019–2021 and was significantly higher (370 ± 79 mg CO2/(m2 h) on average) in 2022.

Influence of environmental factors on the CO2 emission. If the entire four-year data set is considered, the CO2 emission has a very weak positive correlation with soil temperature and a weak negative correlation with moisture content. Obviously, the significance of the correlation is achieved by a large amount of data. At the same time, moisture content and temperature taken separately explain only a few percent of the total variance of CO2 emission; when both parameters are included in the model, the explained variance increases only up to 10%. If the interannual variability is excluded and dependences of CO2 emission on temperature and moisture content are considered separately for each year, the trends generally remain the same: a weak positive correlation with soil temperature and a weak negative correlation with moisture content are observed in all years. The resulting data set makes it possible to consider the factors of spatial variability for each ecosystem over the years (Table 2). In 2019, 2020, 2021 a correlation between CO2 emission and soil temperatures in palsas is also found. Moisture content in palsas over three years played a significant role in spatial variability; in 2021 and 2022, the correlation was positive, and in 2019 it was negative. For bog ecosystems, the correlation with moisture content was not revealed in any year of observation; a significant positive correlation with temperature was observed in 2021 and 2022. A significant positive correlation with the seasonal thawing depth was found in 2020 and 2021. The lack of a significant correlation in other years may be due to insufficient data.

Table 2.   Correlation coefficients between temperature, moisture content of the upper soil horizon, depth of seasonally thawing layer, and CO2 emission (only significant correlations are shown, * and ** indicate significant correlation at p < 0.05 and <0.1, respectively)
Full size table

DISCUSSION

Climatic trends and changes in the temperature regimes of soils in the study area. Since the monitoring of the temperature regime of soils in key landscapes and microclimatic features of the study area is carried out in a long-term year-round mode, we can try to assess trends in their change over the past 8–10 years and compare values for 2018–2022 with those for 2011–2015 according to previous studies [3]. Comparing the two periods, it should be noted that there were no changes in the mean annual air temperature; years with relatively low and relatively high annual temperatures alternated. With regard to seasonal parameters, a significant mitigation of winters is observed. As for the temperature of soils, there is a clear trend towards its rise: the mean annual temperatures of soils of the palsa and the forest increased both at a depth of 20 cm and at the surface. So, the mean annual soil temperature in palsas at a depth of 20 cm became steadily positive; at the surface, it increased by almost 1°C. The obvious reasons for this have not been identified; probably, the main reason of the increasing soil temperature is a combination of factors in particular years and seasons, as shown earlier [3].

General tendencies of differences-similarity between ecosystems and annual trends. The average CO2 emission obtained in our study for the northern taiga ecosystems and palsas are consistent with the data presented in literature. The values obtained for bogs are higher than those reported for analogous ecosystems [21, 27, 35, 51]. In fact, data on the CO2 emission for tundra ecosystems, especially for ecotone landscapes, are limited. As also noted in review paper [49], most of the studies in tundra landscapes are related to the assessment of ecosystem respiration and net ecosystem production, i.e., the contribution of mineralization of soil organic matter to the total ecosystem carbon flux is not estimated. Such studies are mainly carried out in laboratory.

Soils of pine forests are specified by significantly higher CO2 emission values in comparison with other ecosystems. This fact can be explained by a more favorable temperature regime of forest soils as compared with cryogenic soils of palsas and by significant stocks of root biomass in forest ecosystems. The temperature of forest soils at a depth of 10 cm in August was 10–12°С, which is 4–6°С (sometimes more) higher than in the soils of palsas. The root biomass stocks in forest soils were approximately ten times greater than in palsa soils, and the contribution of root respiration reached 80% [4]. The reason for the high emission can also be the optimal water–air regime due to the sandy texture, which prevents moisture stagnation and ensures active diffusion of gases.

An interesting fact revealed during the research is the high CO2 emission from the surface of the bogs, which were either similar to those of palsas, or exceeded them by an average of 50 mg CO2/(m2 h), depending on the year of observation. Such estimates have not been found in the literature; the main trend is that CO2 emission from the palsa soils is approximately twice larger as that from the surrounding bogs [21]. According to the literature data of carbon balance of such ecosystems, palsas mainly act as CO2 sources, and bogs are CO2 sink and methane source [16, 30, 39]. There are several explanations for the results obtained. In all the years of observation, bog soils were characterized by maximum temperatures compared to other ecosystems, which reached 12–16°C, which obviously contributed both to the intensification of biochemical processes and active physical degassing of bog waters. The second circumstance is the hydrological transfer of dissolved carbon dioxide from palsas to bog ecosystems. At the upper boundary of the permafrost, which occurs at a depth of 30 cm to 1 m and deeper on palsas, there are suprapermafrost waters (formed during ice melting in the seasonally thawing layer) with low temperatures (1–2°C), in which CO2 can dissolve and accumulate or be transferred to the surrounding hollows—bog ecosystems. This is confirmed by increased (by a factor of 2–4) values of the emission in bog areas immediately adjacent to palsas, as well as in small bogs surrounded on all sides by palsas [14].

Despite a slight increase in the seasonal temperatures of palsa soils, no upward trend of CO2 emission was found, and a slight decrease was observed in the last two years of observations (Fig. 2d). The significant increase in the seasonal thawing depth did not affect the CO2 emission. For bog ecosystems, a different picture is observed: the average CO2 emission in some years clearly positively correlated with average soil temperatures and summer air temperatures but not with annual temperature trends. Therefore, it can be stated that the soils of bog ecosystems are more sensitive in terms of CO2 emission to the weather conditions of the growing season. CO2 emission in forest ecosystems practically did not change over four years of observation, as well as temperature and soil moisture. Interesting trends are noted for soils of thermokarst depressions. For two years of observation, CO2 emission differed very significantly. It tripled in the second year and was almost two times higher than that in the surrounding palsa. This may be due to an increase in the rate of peat decomposition with the development of thermokarst over time, as well as to unusually warm conditions in the summer of 2022, which led to the heating of the upper soil horizon of thermokarst depressions to 15°C. These are the maximum temperatures among all studied objects.

The following fact attracts attention: at low variability of conditions (temperature and soil moisture) in bog ecosystems, in comparison with palsas, the variability of emission is the same in two ecosystems. The high variability of CO2 emission in bog ecosystems can be associated both with “gas cavities” in peat soils, where air with a high content of biogenic gases is accumulated [13], and with the “edge” effect, which was described above.

Factors of spatial variability of CO2 emission. In general, for soils of the palsa–bog complex, soil temperature was a factor of spatial variability of CO2 emission, but a significant correlation was not always found. Obviously, the influence of temperature on CO2 emission can be increased or decreased due to the influence of other factors, such as moisture content. It was shown in [20] that for tundra ecosystems, soil temperature is significant in terms of spatial variability of CO2 emission only when there is sufficient moisture content. In arctic ecosystems with insufficient moisture content, the latter plays a key role in spatial variability of CO2 emission [17]. Many authors note that temperature plays a subordinate role in the regulation of spatial variability of CO2 emission [7, 32] but most of the studies are related to forest ecosystems. The obtained correlation values, which are quite high in some years, especially for palsas, indicate that the heat supply of soils under these conditions is a limiting and determining factor. This is especially true for the coldest soils. It is obvious that the heat supply of palsa soils, among other factors, is determined by the variability of geocryological conditions (depth of seasonally thawing layer), which follows from the results obtained (Fig. 3). It can be assumed that the absence of a correlation between soil temperature and CO2 emission in 2022 is associated with a sharp increase in the depth of seasonally thawing layer in this year and, as a result, a decrease in the effect of permafrost on the temperature regime of soils and CO2 production. The temperature sensitivity coefficient Q10 was calculated from field data for 2019–2021 for palsas based on the exponential dependence of CO2 emission on soil temperature [28, 48]. It ranged from 1.1 to 3.1 depending on the year of observations.

Fig. 3.
figure 3

Correlation of CO2 emission with (a) soil temperature (CO2 emission = 73.294e0.1129T, R 2 = 0.19) and (b) active layer thickness in August 2020 in palsas.

Full size image

Since the studied frozen palsas are semihydromorphic and sometimes automorphic ecosystems, during the growing season, the moisture content can also be a limiting factor, which is expressed in its significant positive correlation with for some years of observations.

CONCLUSIONS

Despite the fact that no significant trends of climatic parameters over the past 10 years were found for the north of Western Siberia, a gradual increase in the mean annual temperatures of the soils of forest and tundra ecosystems and in the seasonal thawing depth should be noted. Analysis of the database on the soil surface temperatures at the peak of the growing season for the main ecosystems of the study area (more than 1000 measurements) made it possible to conclude that no significant increase was observed. In this regard, it can be assumed that with a change in climatic parameters (warming), the increase in the emitting role of the studied ecosystems will not be due to an increase in the specific CO2 emission (per unit area per unit time) but due to an increase in the duration of the growing season.

The combination of modern and paleocryogenic processes determines the specifics of the formation and functioning of the ecosystems in the study area, which is expressed in their high heterogeneity and dynamism, including dynamics of the carbon cycle. CO2 emission from automorphic soils of forest ecosystems is 2–2.5 times higher than that from soils of the palsa–bog complex, which is determined by a favorable hydrothermal regime, high root biomass stocks, and good water-physical properties. CO2 emission from semihydromorphic soils from permafrost-affected palsas and from bog soils is practically the same, despite the significant difference in hydrothermal conditions. The reasons for this are both the mutual influence of the factors and the redistribution of CO2 fluxes due to CO2 dissolution in suprapermafrost water of palsas, its transfer with water flows to bog ecosystems, and gas release from the latter. Thus, not all CO2 produced by the soils of palsas is released from their surface, it is partially transported by suprapermafrost waters and released from the surface of bog soils.

Soil temperature was a significant predictor of spatial variability of CO2 emission in the soils of the palsa–bog complex; the highest correlation was found for permafrost-affected palsas, where the temperature regime is regulated by seasonal thawing. Although moisture content (positive correlation) increases the explained variance of CO2 emission, the factors not taken into account in the study account for more than 70% of the total variability.