Introduction

Greenhouse gas (GHG) emissions are playing a major role in the acceleration of climate change effects, and their concentration in the atmosphere has continued to rise rapidly since the pre-industrial era (IPCC 2014). To tackle the climate crisis European Union (EU) has set a target to reduce greenhouse gas emissions by 55% by 2030 and reach climate neutrality by 2050 (European Commission 2011, 2019). To reach climate neutrality, a balance in policy regulations between air and climate, nature, sustainability and well-being and economic sectors has to be found. Forests play an important role in the global carbon budget and in the climate change mitigation targets, as forestry is the only sector that can store and sequester large amounts of carbon into biomass or soil (Gundersen et al. 2021). It is crucial to understand and promote such forest management practices that contribute to long-term climate change mitigation.

In hemiboreal region, the most significant management measure (in terms of carbon storage and sequestration) is forest drainage, by which annual increment of trees can be improved several times (Sikström et al. 2020). The effect of drainage depends on climate and soil organic composition where the ditch system is established. Additionally, soil GHG emissions are dependent on time since the drainage system has been established (Conchedda and Tubiello 2020), and on further land use of the drained soils (forestry or agriculture). The long-term effect of forest drainage can result in positive net ecosystem exchange (NEE) of carbon budget (Minkkinen and Laine 1998). However, tree and stand development plays an important role in the carbon balance of organic soils after drainage in boreal regions (Minkkinen and Laine 1998; Lohila et al. 2011), yet less is known about organic soils in the hemiboreal region. Change in soil hydrology due to drainage from anaerobic to aerobic conditions can trigger rapid peat decomposition and influence dissolved organic carbon (DOC) entering water systems (Freeman et al. 2004). However, the influence of soil organic matter composition on peat decomposition and C loss is not fully studied (Bader et al. 2018).

Old-growth forests are forests in late stages of stand development, characterized by old and large dimension trees, large dimension deadwood, various stand canopy layers, species composition and ecosystem functions (Buchwald 2005). Along with biodiversity strategy of EU, the role of old-growth forests has also been emphasized in the EU Climate policy. Therefore, with increasing interest to promote carbon sequestration and GHG balance through processes related to soil biogeochemistry in old-growth forests, it is necessary to acknowledge also long-term drainage ditching effect, which has been performed during past century in large parts of northern Europe (Tong et al. 2022). Scots pine (Pinus sylvestris L.) is economically one of the most important tree species in hemiboreal forest zone. Pine forests cover 26.3% of the total forest area in Latvia, but old Scots pine forests (age > 120 years) cover 10% of the total pine forest area in Latvia. Pine forests with organic soils cover ca. 29% of the total pine forest area, of which 14% are drained and 15% are periodically waterlogged (without drainage systems) (NFM 2018). Therefore, it is crucial to estimate GHG emissions from old-growth pine forests on organic soils with diverse groundwater levels. Such information is important in the evaluation of the effects of new drainage systems’ establishment or maintenance, and the effect of drainage system dysfunctionality or wetland restoration, which also determines carbon dioxide (CO2) and methane (CH4) flux (Tong et al. 2022). Restoration of wetlands by closing the drainage systems has been practiced to favor biodiversity targets (Craft 2016; Minayeva et al. 2017) and ensure carbon storage in wetland soil organic matter (Temmink et al. 2022), as the majority of sequestrated carbon is stored in the soils rather than plants (Moomaw et al. 2018). Nevertheless, biodiversity is important and has to be balanced with carbon storage and climate change mitigation targets. Old-growth forests might positively contribute to both biodiversity and climate change mitigation; but there is still lack of empirical data on GHG balance in such stands.

In recent years, GHG monitoring and assessment of GHG flux trends have been widely studied in relation to improving knowledge on carbon storage, sequestration and substitution of fossil carbon for renewable carbon. The world’s largest carbon sinks are oceans, atmosphere, soil (especially organic soils) and forests (Hamilton and Freiss 2018). Any changes in the carbon dioxide balance of the forest ecosystem can have a strong impact on the CO2 emissions (Bradshaw and Warkentin 2015). The total forest floor ecosystem (soil) respiration (Rt) consists of autotrophic and heterotrophic respiration and comes from both, soil and forest floor. Autotrophic respiration (Ra) includes respiration from plants and originates from plant roots, mycorrhizae and other rhizospheric microorganisms (Bond-Lamberty et al. 2004). Heterotrophic respiration (Rh) is derived from the decomposition of soil organic matter (SOM) and plant residues (Ding et al. 2016). Even though methane is far more radiative forcing and responsive to environmental changes than CO2, it has received less attention in terms of climate change (Hansen et al. 2013; Neubauer and Megonigal 2015). It is known that forested wetlands (under anaerobic conditions) are significant global source of CH4 emissions from the soil (Matthews and Fung 1987), however upland forests on drained soils (under aerobic conditions) are functioning as CH4 sinks (uptake) in the global CH4 budget (Le Mer and Roger 2001). Recent studies have focused on the assessment of hemiboreal forest carbon pools in old-growth forests of Norway spruce (Picea abies (L.) Karst.) and Scots pine (Ķēniņa et al. 2018, 2019) relationship of carbon pools with stand age and elevation (Seedre et al. 2015), and chronosequence of carbon storage dynamics in Scots pine stands (Uri et al. 2022). Additionally, Krasnova et al. (2019) studied the net ecosystem carbon exchange in hemiboreal forest ecosystem in relation to tree species composition. Furthermore, studies of soil GHG fluxes from different management practices, e.g. partial and clear-cutting (Korkiakoski et al. 2019, 2023), soil fertilization (Håkansson et al. 2021), and drainage (Von Arnold et al. 2005), soil preparation, hydrology, fire and biodiversity management (Mäkipää et al. 2023) have been carried out. Also, studies have reported soil GHG fluxes after different natural disturbances, e.g. time since fire (Köster et al. 2016), experimental warming and drying after fire (Song et al. 2018), cleared and un-cleared windthrown stands (Köster et al. 2011), and experimental short-term flooding (Schindler et al. 2020). At the same time the information about soil GHG fluxes from old-growth forests with organic soils is missing.

The study aims to assess the seasonal dynamics of soil heterotrophic and total forest floor ecosystem respiration and CH4 fluxes in old-growth Scots pine stands on organic soils along a moisture gradient. We hypothesize that in old-growth Scots pine stands on organic soils: (1) soil CO2 efflux will increase with groundwater level decrease; (2) soil CH4 efflux will increase with groundwater level rise.

Materials and methods

Study area

The study was carried out in the growing season of 2021. The study area is located in the hemiboreal vegetation zone in Latvia, where the climate is described as temperate, yet strongly affected by the Baltic Sea and the North Atlantic. The mean annual sum of precipitation in the study area is 692 mm, and the mean annual air temperature is 6.4 °C. The coldest month is February, with the average air temperature around − 3.7 °C, and the warmest month is July, when the average air temperature is 17.4 °C (LEGMC 2020). Information about the mean air temperature and precipitation during the measurement season of 2021 has been obtained from the Latvian Environment, Geology and Meteorology (Fig. 1a).

Fig. 1
figure 1

Precipitation (mm) and the mean air temperature (°C); DOY—day of year; grey lines separate measurement cycles (a), the mean groundwater level (GWL) depth (b), and soil temperature (T_10) at 10 cm depth (c), during the measurement period in 2021 (data: LEGMC, 2020). Whiskers denote ± 95% confidence interval

Full size image

Study stands were selected based on soil type (organic soil) with different groundwater levels (with and without drainage systems). The old-growth forest stands selected for investigation within this study were checked in the field for the occurrence of target forest type, age group (> 130 years), the dominance of Scots Pine (> 60% from the basal area), old trees as dominant forest element and no signs of recent human intervention. Selected stands correspond to E. Buchwald (2005) old-growth forest definition’s ‘n6’ category. Altogether six Scots pine dominated old-growth forest stands (130–180 years old) on deep peat soils with different soil moisture regimes and conditions were selected (Table 1). The selected study stands belong to two different forest site types—Myrtillosa turf. mel. and Caricoso-phragmitosa. The first forest site type –Myrtillosa turf. mel. described as fertile, very productive forests on mesotrophic drained peat soil. The organic layer is at least 20 cm thick, composed of cottongrass (Eriophorum spp.), tree biomass and small amounts of Sphagnum peat. Typically, the canopy is dominated by Scots pine, birch (Betula pendula Roth. and Betula pubescens Ehrh..), Norway spruce (Picea abies (L.) Karst.) with an admixture of black alder (Alnus glutinosa (L.) Gaertn.). Ground vegetation is characterized by a dense and diverse herb, dwarf shrub and moss layer, dominated by European blueberry (Vaccinium myrtillus L.), lingonberry (Vaccinium vitis-idaea L.), purple Moor-grass (Molinia caerulea (L.) Moench.), Pleurozium schreberi, Hylocomium splendens, Dicranum spp. The shrub layer is dominated by alder buckthorn (Frangula alnus), willows (Salix spp.), common juniper (Juniperus communis) and rowan (Sorbus aucuparia). The second forest site type—Caricoso-phragmitosa described as forests of moderate productivity on mesotrophic peat soil. Soil consists of moderately decomposed acidic peat, at least 30 cm thick, composed mostly of sedge (Carex spp.), tree and sphagnum biomass. Typically, the canopy is dominated by Scots pine with an admixture of birch, black alder and Norway spruce. The ground vegetation is characterized by a dense herb and moss layer, dominated by sedges, reedgrass (Calamagrostis spp.), common reed (Phragmites australis), Sphagnum, Pleurozium schreberi, Hylocomium splendens, Dicranum spp. The shrub layer is dominated by common juniper, alder buckthorn (Frangula alnus), and willows. Increased paludification in such sites causes the forest to change to a pine swamp (Bušs 1997, cited from Jansons et al. 2019). Stands at different forest types differ in soil moisture and groundwater levels due to drainage ditches established in Myrtillosa turf. mel. sites and no ditches in Caricoso-phragmitosa sites.

Table 1 Taxation indices of old-growth Scots pine forests (average values ± 95% confidence interval)
Full size table

Sample plot design

The drainage system was established around the year 1960 (ca. 50–60 years ago)—therefore, the peat soil conditions have already changed/adapted as a result of drainage. In each of the selected stands six sample plots (R = 12.62 m, 500 m2) were established to perform measurements and collection of the main tree variables (tree species, tree height, diameter at breast height, age) to determine the taxation indices (Table 1).

In each selected stand, on three of the six sample plots soil GHG measurements were carried out. Plots were prepared in the winter and early spring seasons (at least one month before measurements) to minimize the effect of root trenching and human intervention during plot establishment on GHG fluxes. The required time for trenched root decomposition is uncertain, however literature reports that decomposition of recently trenched roots occurs in the first two months (Chin et al. 2023). Plot establishment included installation of wooden pathway near the center of each sample plot, to reduce human movement factors on the GHG measurements (Fig. S1 in supplementary material). On one side of each pathway two collars (six collars per stand) for measurements of total forest floor ecosystem respiration were installed (D = 20.1 cm, H = 10 cm) and compacted with sand from the outside to prevent gas leakage. Vegetation inside the collars remained intact. On the other side of each pathway, a root trenching was performed to the depth of 45 cm, water permeable root isolation fabric material (geotextile) was placed and vegetation was removed (once before measurement season) to ensure that emissions were coming only from the organic matter of decomposition. In each prepared (trenched) plot, three heterotrophic soil respiration collars (nine collars per stand) were installed in the soil ca. 8 cm deep. The size of trenched plot was 180 × 40 cm, and the distance between collar and root isolation fabric was ca, 10 cm from top and bottom, and ca. 30 cm distance between collars and sides of trenched plot. In each sample plot two groundwater pipes (six pipes per stand) were installed—one for manual water level monitoring and the other one for water sampling (Fig. 1b). To continuously monitor the changes in soil temperature and their influence on soil GHG fluxes, automatic temperature sensors with two depth measurement points (10 cm and 40 cm depth) were installed in the soil in the center of each plot (three sensors with two measurement depths per stand). Measurements were performed at least 1 month after sample plot establishment to minimize trenching artefact influence on GHG fluxes.

Soil GHG flux monitoring

Soil GHG (CO2, CH4) flux measurements on each stand was performed at a three-week interval starting from the beginning of the growing season in April 2021 and lasted till November 2021. Overall, the soil GHG measurements were performed in 9 measurement cycles, whereas in each cycle soil GHG fluxes were recorded at different times of the day (morning, noon, afternoon). Morning time corresponds to the time from 9:00 to 11:30, noon time from 11:30 to 13:00, and afternoon time from 13:00 to 17:00.

The total forest floor ecosystem respiration (total CO2 flux, RtCO2) and CH4 fluxes (RtCH4) were measured as a CO2 and CH4 concentration change in the chamber headspace and was registered for a 5 min chamber deployment time at 2 s interval between data loggings. We used a closed darkened cylindrical chamber (D = 20 cm, H = 30 cm) connected to a cavity ring-down spectroscopy gas concentration analyzer Picarro (Gas Scouter G4301, Picarro, Santa Clara, CA) (Korkiakoski et al. 2019; Villa et al. 2020). Next to the collars, soil temperature was measured (Fig. 1c) at four depths (10, 20, 30, 40 cm) with four temperature sensors connected to a data logger (Comet U0141, COMET SYSTEM s.r.o., Czech Republic).

The heterotrophic soil CO2 flux (RhCO2) of the trenched plots was measured with Vaisala non-dispersive infrared CO2 probe (GMP-343, HMP76, MI70, Vaisala Oyj, Finland) using a closed chamber system consisting of a darkened cylindrical chamber (D = 20 cm, H = 40 cm) with inserted CO2 analyzer, and relative humidity and temperature sensors connected to a data logger (Zhu et al. 2020). The time for each measurement was 5 min with a 5 s interval between data loggings. Additionally, soil moisture (%) and soil temperature (℃) were measured next to each collar inside the trenched plot with a digital thermometer at 5 cm depth (WET-2, HH2 moisture meter, Delta-T Devices Ltd., Cambridge, UK), connected to a data logger.

Additional sampling and measurements

Additional measurements were performed on the same sample plots as GHG measurements. Groundwater level and parameters (water temperature and pH) were measured in each cycle during the GHG measurements. Groundwater level was measured with a tape measure as a distance from the ground surface to the water level. Groundwater temperature and pH were measured with a YSI water quality meter (ProDss, YSI inc., Ohio, USA).

To perform soil chemical analysis, soil nutrient content and pH level, we collected soil samples using an auger (100 cm3) with two replications per plot (Table 2). The soil samples were collected up to the depth of 40 cm and afterwards divided into 10 cm segments (0–10; 10–20; 20–30; 30–40 cm). Soil analysis was carried out in the certified laboratory of Latvian State Forest Research Institute ‘Silava’ using ISO standard methods.

Table 2 Description of soil characteristics (average of 0–40 cm depth) of the studied stands (average values ± 95% confidence interval)
Full size table

Data analysis

Data processing and analysis were performed using the statistical software R 4.1.2. (R Core Team 2021). The linear regression between the measurement time and GHG concentration change within the chamber was used to calculate the flux rates of CO2 (mg CO2 m−2 s−1) and CH4 (mg CH4 m−2 s−1). Model estimates, chamber volume, ground area, air pressure and chamber temperature were used in the flux calculations for each measurement and compiled in a dataset. Model fit of each measurement was assessed with a linear regression model, if the coefficient of determination was lower than R2 = 0.80 then the measurement was omitted and classified as an outlier. The gas flux rates of all measurements were expressed as mg m−2 s−1. Seasonal variability was estimated for RtCO2, RhCO2, and RtCH4 flux, the average flux and 95% confidence interval per measurement cycle in each stand were calculated. The average RtCO2 and RhCO2 fluxes of sample plots were subtracted from each other to assess the respiration from roots and ground vegetation (RaCO2). Normal distribution of data was checked with the Shapiro–Wilk normality test. A T-test was used to assess differences in the mean values of parametric variables (RtCO2 and RhCO2). Mann–Whitney U test was used to assess significant differences between the non-parametric variables (RtCH4).

Linear regression was used to assess single factor relationship with RtCO2 and RhCO2 fluxes. To examine the environmental factors affecting RtCO2 and RhCO2 fluxes, linear mixed effect models were developed using the R package “lmerTest” (Kuznetsova et al. 2017). Groundwater level (GWL), soil temperature (T_soil) were used as fixed effects, but stand, sample plot and measurement cycle were used as random effects. Model validation was performed by plotting model residuals versus fitted values, and using the Akaike information criterion (AIC) test. Models with normally distributed residuals and with the lowest AIC values were selected as the best. Significant variation (p ≤ 0.05) in fixed effects was examined with ANOVA type III test (sum of squares).

Results

CO2 respiration

The soil CO2 flux rates were expressed as total forest floor ecosystem respiration (RtCO2) and heterotrophic respiration (RhCO2) from the soil (Fig. 2). In both cases, a clear seasonal trend was visible, as fluxes increased from spring to summer, with the highest values during the peak of the growing season (July and August) and decreased in the autumn. RtCO2 flux ranged from 0.04 to 0.46 mg m−2 s−1. The season’s average RtCO2 flux was 0.17 ± 0.01 mg m−2 s−1 (average ± 95% confidence interval). Average RtCO2 flux in drained and undrained sites was 0.18 ± 0.018 mg m−2 s−1 and 0.17 ± 0.015 mg m−2 s−1, respectively. RhCO2 flux ranged from 0.007 to 0.33 mg m−2 s−1. The season’s average RhCO2 flux was 0.07 ± 0.005 mg m−2 s−1. Average RhCO2 flux in drained and undrained sites was similar, 0.08 ± 0.007 mg m−2 s−1 and 0.07 ± 0.007 mg m−2 s−1, respectively. The RhCO2 flux in different measurement cycles ranged from 37 to 49%, and on average accounted for 41% of the total forest floor ecosystem CO2 respiration. Furthermore, the soil autotrophic CO2 (RaCO2) flux ranged from 51 to 63%, and on average accounted for 59% of the total forest floor ecosystem respiration.

Fig. 2
figure 2

The total forest floor ecosystem respiration (RtCO2, mg m−2 s−1) (a) and the heterotrophic respiration (RhCO2, mg m−2 s−1) (b) from the soil per measurement cycle. Whiskers denote a 95% confidence interval

Full size image

The linear relationship between CO2 fluxes and groundwater level in analyzed stands was positive and significant (p < 0.001), but R2 was very low for RtCO2 and RhCO2—0.06 and 0.08, respectively. Furthermore, a close significant positive relationship (p < 0.001) was observed between the RtCO2 and T_soil, where the R2 was 0.61 and 0.41, for RtCO2 and RhCO2, respectively (Fig. 3).

Fig. 3
figure 3

Relationship between the total forest floor ecosystem respiration (RtCO2, mg m−2 s−1) and heterotrophic respiration (RhCO2, mg m−2 s−1) and soil temperature (°C). The grey area denotes a 95% confidence interval

Full size image

Linear mixed effect model for the RtCO2 flux from the soil indicated a significance of soil temperature (T_soil) at 10 cm depth (p < 0.001) and GWL (p < 0.01). Overall, the model’s conditional R2 was high—0.78 (Table 3 in supplementary material). Linear mixed effect model for the RhCO2 flux from the soil also indicated a significance of soil temperature (T_soil) at 10 cm depth (p < 0.001) and GWL (p < 0.01). Overall, the model’s conditional R2 was moderate—0.56 (Table 4 in supplementary material).

CH4 flux

The soil CH4 flux had an opposite seasonal trend compared to the soil CO2 flux. Highest CH4 emissions from soil were observed at the beginning and at the end of the growing season, but CH4 uptake by soil was observed in the summer (Fig. 4). The soil CH4 flux ranged from − 1.77e-06 to 2.90e-06 mg m−2 s−1. The annual (measurement season) average CH4 flux was − 2.03e-08 ± 1.18e-07 mg m−2 s−1. Moreover, in June, October and November high variability of CH4 flux rates can be observed, as indicated by notably higher confidence intervals. Season average CH4 flux rates in drained sites was 1.47e-06 ± 1.24e-06 mg m−2 s−1, but in undrained sites 3.60e-06 ± 2.97e-06 mg m−2 s−1 (p > 0.001).

Fig. 4
figure 4

The soil CH4 flux (RtCH4, mg m−2 s−1) per measurement cycle. Whiskers denote a 95% confidence interval

Full size image

The relationship of RtCH4 flux and groundwater level was negative (R2 = 0.39, p < 0.001, Fig. 5) Higher groundwater level promotes CH4 emissions from the soil, but lower groundwater levels result in CH4 uptake by the soils.

Fig. 5
figure 5

Relationship between the soil CH4 flux (RtCH4, mg m−2 s−1) and groundwater level (cm). The grey area denotes a 95% confidence interval

Full size image

Linear mixed effect model for the RtCH4 emissions rates could not ensure a normally distributed residual linear fitting, thus the influence of additional factors affecting RtCH4 flux was strong and the model tended to underestimate high CH4 flux. A good model fit could be obtained only for RtCH4 uptake (RtCH4 < 0; Table 5 in supplementary material), but not for RtCH4 emission flux (Table 6 in supplementary material). The determined significant factor affecting the RtCH4 uptake was groundwater level (p < 0.001) and T_soil (p < 0.001) with overall conditional R2 = 0.67.

Discussion

In hemiboreal Latvia 14% of the total pine forests on organic soils are drained and 15% are periodically waterlogged (without drainage systems) (NFM 2018), and from climate’s perspective it is important to adapt management practices to benefit climate change mitigation goals. The results from our study suggest that the existence of drainage systems alone is not sufficient to characterize the actual soil gas exchange from such areas as no differences between drained and not drained forest stands were observed, possibly due to overdue ditch maintenance at our selected stands. Moreover, the long-term impact of the ditch system has resulted in significantly higher tree carbon stock (Table 1), and significantly higher soil carbon and nitrogen stock over the upper 40 cm for drained (Mt) stands (Table 2). We are aware that our study excludes N2O emissions and lacks flux data from winter periods (air temperature < 0 °C), which indicates a potential bias, however previous studies suggest low fluxes during winter periods, usually contributing less than 10% of the annual flux (Korkiakoski et al. 2017). Our study deals with one season measurements, thus the results are preliminary.

The studied old-growth Scots pine forests had a similar seasonal trend for RtCO2 and RhCO2 (Fig. 2), as the soil CO2 flux is lower in spring, then increases during summer and decreases in autumn. The RtCO2 follows similar seasonal trends at all analyzed stands with the main proportion (more than half: 59%) of soil CO2 efflux is produced by the ground vegetation and tree roots (RaCO2), and 41% originating from the RhCO2 due to decomposition of soil organic matter. A similar proportion of RhCO2 (40–43%) and RaCO2 (57–60%) has been reported in studies of boreal region’s afforested peatlands (Berglund et al. 2011; Hermans et al. 2022). Main components of RaCO2 other than tree roots are dwarf shrubs and ground vegetation other than shrubs (e.g. grasses, mosses and herbs), and ericoid and ectomycorrhizal fungal hyphae which has been estimated as a proportion of RtCO2 as 10%, 8%, and 4%, respectively (Ryhti et al. 2021). The required time for trenched root decomposition is uncertain, however literature reports that decomposition of recently trenched roots occurs in the first two months (Chin et al. 2023). Moreover, soil trenching and geotextile insertion might influence soil moisture conditions, which may affect decomposition of soil organic matter (Savage et al. 2018), therefore might result in RhCO2 underestimation in these areas (Chin et al. 2023), as direct comparison of soil trenching influence on soil moisture between trenched and un-trenched plots is not possible due to experimental setup.

Moreover, the seasonal variability of the soil CO2 fluxes (both RtCO2 and RhCO2) are driven by soil temperature (T_soil) and groundwater level. In our study the highest soil CO2 flux was observed in July (Fig. 2), when the average air and soil temperatures were the highest (Fig. 1a, c). Overall, the peak of soil RtCO2 flux during summer in old-growth stands was of similar magnitude compared to control plots of a study carried out in Finland (boreal region) in Scots pine stands on organic soil (Korkiakoski et al. 2019). In August, the air temperature was already decreasing, but soil temperature and consequently the total forest floor ecosystem respiration was still high. Such lack of clear link between air temperature and soil CO2 flux can be explained by the lag effect, also mentioned in other studies, where soil CO2 fluxes had delayed response to air temperature change (Kukumägi et al. 2017; Uri et al. 2022). The observed significant positive relationship between groundwater level and the soil CO2 flux (both RtCO2 and RhCO2) indicates that soil CO2 fluxes can be expected to decrease if the groundwater level increases. However, coefficient of determination was very low in our study, indicating stronger influence of other factors on soil CO2 fluxes. Therefore, our proposed hypothesis that soil CO2 flux decreases with a rise of groundwater level was partly confirmed, and similar relationship have been reported also in other studies (Von Arnold et al. 2005; Jungkunst et al. 2008). However, some studies have also reported a weak impact of groundwater level rise on soil CO2 flux (Olefeldt et al. 2017; Minkkinen et al. 2018) which is also partly confirmed in our study by low values of coefficient of determination. Studies report that soil moisture content regulates CO2 fluxes, at high soil moisture, aerobic respiration is limited by diffusive supply of oxygen, but at low soil moisture CO2 production is favored, yet limited by a diffusion of organic C substrate, thus at intermediate soil moisture CO2 flux peaks (Movano et al. 2012; Fairbairn et al. 2023). Furthermore, some widely acknowledged models use soil temperature as a single predicting variable (Reichstein et al. 2005), however results of linear mixed effect models for RtCO2 and RhCO2 (Tables 3, 4 in supplementary material) highlights the significance and importance of soil temperature and groundwater level, thus soil CO2 fluxes can be predicted with high certainty with T_soil and GWL as predictors.

The most important factor affecting the soil CH4 flux (RtCH4) is groundwater level—if it is close to ground surface, the soil emits CH4, and if it is low, the organic soil can function as a CH4 sink, however, our results indicate high variability of CH4 fluxes (Fig. 4). The negative linear relationship between the RtCH4 and groundwater level was significant (Fig. 5, p < 0.001). Previous studies have reported that the lowering of groundwater level promotes soil oxygenation, thus enhances the process of mineralization of organic matter as decomposition under aerobic conditions is faster (Clymo 1984). Oxygen pressure regulates the production of CH4, therefore groundwater level change to lower (stand drainage) has the potential to reduce the CH4 efflux from the soils (Elberling et al. 2011), and in the analyzed old-growth Scots pine forests even ensures CH4 uptake. Studies from boreal and hemi-boreal regions have reported that maintenance of GWL 20–30 cm below ground surface results in CH4 uptake (Butlers et al. 2023; Tong et al. 2022), however relatively high GWL (0–20 cm) and soil close to saturated conditions may facilitate CH4 production (Korkiakoski et al. 2019), and such coherence has been observed also in our study. Thus, our second proposed hypothesis that soil CH4 efflux increase with groundwater level rise was confirmed. Study results indicate that RtCH4 uptake can be predicted with GWL and T_soil as variables with relatively high certainty (Table 5 in supplementary material); however mixed effect model of RtCH4 emissions (release) did not show any significance of analyzed environmental parameters, indicating strong other factor influence on CH4 release (Table 6 in supplementary material).

Old-growth forests on organic soils with naturally low GWL, or regulated with ditch systems to reduce excess water, has the potential to ensure CH4 uptake and increase tree carbon stock (Fig. 5, Table 1), thus having a significant long-term contribution to climate targets. Observed insignificance of drainage system (forest type) indicates a ditch system malfunction and overdue ditch system maintenance, thus in the long-term increasing groundwater level and promoting soil moisture saturation. Ditch system maintenance is recommended 1 to 2 times during the stand rotation in order to keep favorable drainage conditions for tree growth, but due to additional costs, it can be avoided on stands where it contributes little to stand productivity (Sarkkola et al. 2013). However, the legacy effect of uncleaned ditches might result in CH4 efflux from soil and create local hotspots for high CH4 flux in the waterlogged areas in the forest or near the ditch (Tong et al. 2022).

From the carbon storage and climate change mitigation perspective fertile drained old-growth forests can store more C in tree biomass compared to not drained stands (Ķēniņa et al. 2022), and ensure CH4 uptake (Le Mer and Roger 2001; Von Arnold et al. 2005), while CO2 respiration can be elevated in stands with low groundwater levels (Freeman et al. 2004; Von Arnold et al. 2005), but possibly compensated with more C stored in biomass. Moreover, old-growth forests are commonly associated with structural complexity and species diversity at the latter stages of development (Buchwald 2005). Typically, a substantial quantity of woody debris is present in various decay stages of decomposition and this creates a habitat for multiple organisms (Herrmann et al. 2015). Old-growth forests on organic soils require further studies to assess other (especially broadleaved) dominant tree species, that may have different effects on ground vegetation and soil, thus also GHG fluxes. Moreover, forest management practices that contribute to long-term climate change mitigation may significantly impact the biodiversity measures in old-growth forests, which was not within the scope of this study.

Conclusions

The study reveals novel information about old-growth Scots pine forest organic soil GHG fluxes. The main source of total forest floor ecosystem respiration is autotrophic respiration (ground vegetation and roots) contributing more than half (59%) to the total respiration, while the heterotrophic respiration contributes slightly less (41%) to the total forest floor ecosystem respiration. The soil CO2 fluxes demonstrate a clear seasonal trend and are closely linked to soil temperature indicating the potential increase of CO2 efflux as the climate is warming, and significantly but weaker linked to groundwater level, indicating CO2 flux decrease if groundwater level rises. The dynamics of CH4 fluxes have high variability, however largely driven by groundwater level fluctuations: if the groundwater level is low, the organic soil functions as a CH4 sink. A drainage system that reduces the excess groundwater level ensures a long-term positive (from a climate change mitigation perspective) effect on soil CH4 uptake, however can result in elevated CO2 efflux.