Impact of Soil Organic Layer Thickness on Soil-to-Atmosphere GHG Fluxes in Grassland in Latvia

Abstract

Drained organic soils in agricultural land are considered significant contributors to total greenhouse gas (GHG) emissions, although the temporal and spatial variation of GHG emissions is high. Here, we present results of the study on soil-to-atmosphere fluxes of carbon dioxide (CO₂), nitrous oxide (N₂O) and methane (CH₄) from drained organic (fen) soils in grassland. A two-year study (from July 2021 to June 2023) was conducted in three research sites in Latvia (Europe’s hemiboreal zone). Soil total respiration (R_tot), CH₄ and N₂O fluxes were determined using a manual opaque chamber technique in combination with gas chromatography, while soil heterotrophic respiration (R_het) was measured with a portable spectrometer. Among research sites, the thickness of the soil organic layer ranged from 10 to 70 cm and mean groundwater level ranged from 27 to 99 cm below the soil surface. Drained organic soil in all research sites was a net source of CO₂ emissions (mean 3.48 ± 0.33 t CO₂-C ha⁻¹ yr⁻¹). No evidence was obtained that the thickness of the soil organic layer (ranging from 10 to 70 cm) and OC stock in soil can be considered one of the main affecting factors of magnitude of net CO₂ emissions from drained organic soil. Drained organic soil in grassland was mostly a source of N₂O emissions (mean 2.39 ± 0.70 kg N₂O-N ha⁻¹ yr⁻¹), while the soil both emitted and consumed atmospheric CH₄ depending on the thickness of the soil organic layer (ranging from −3.26 ± 1.33 to 0.96 ± 0.10 kg CH₄-C ha⁻¹ yr⁻¹).

1. Introduction

Pristine peatlands are wetland ecosystems characterised by common water-logged conditions and the accumulation of organic matter (peat) at the surface [1]. Although pristine peatlands ensure a wide variation of ecosystem services [2,3], their high water saturation makes them unsuitable for traditional forestry or agricultural practices, including grassland management. In the past 200 years, roughly 15 percent of the world’s peatlands have been altered by drainage [4] and thus, among other management of drained peatlands, agricultural production on peatland has increased significantly in many countries [5]. In boreal and cool temperate climatic zones, the total area of drained organic soils for agriculture is 15.6 million ha (about 60% of total organic soil drained for agriculture worldwide) [6]. Following drainage, hydrological and biogeochemical processes and thus physical and chemical properties of organic soil (peat) have been significantly altered [7,8]. Drainage leads to lower groundwater (GW) level and increases soil aeration, meeting the requirements of land use for agriculture [5]. Simultaneously, drainage increases the aerobic decay (mineralisation) of soil organic matter resulting in carbon (C) and nitrogen (N) losses from peatlands mainly in the form of carbon dioxide (CO₂) and nitrous oxide (N₂O) as well as dissolved and particulate material [9]. Soil-to-atmosphere CO₂ exchange depends on the balance between C sequestration through photosynthesis and subsequent C input to the soil, and C release through soil autotrophic and heterotrophic respiration as well as methane (CH₄) oxidation under aerobic conditions [9]. Soil autotrophic respiration reflects the release of CO₂ from plant roots and the associated rhizosphere (mycorrhizae and rhizosphere microorganisms), while soil heterotrophic respiration (R_het) represents the CO₂ release from the decomposition of soil organic matter, including litter, driven by microorganisms [10,11,12]. Peatlands drained for agriculture are considered hotspots of CO₂ emissions from the agriculture sector [7,8,13]. At the same time, drainage (GW level drawdown) significantly reduces CH₄ emissions from soil to the atmosphere due to change from anaerobic conditions, which are favourable for CH₄ production by methanogens, to aerobic conditions, which limit CH₄ production and enhance CH₄ oxidation to CO₂ by methanotrophs [14,15].

It is estimated that 11–15% of drained or otherwise degraded peatlands worldwide are responsible for roughly 5% of total greenhouse gas (GHG) emissions of anthropogenic origin [4]. Regarding the climate change mitigation, the relatively large contribution to total anthropogenic GHG emissions from the relatively small area of drained organic soils is the reason for increased interest from policy makers, land owners and managers as well as from scientists both at global and regional level. In 2021, among agricultural land in Latvia, drained organic soils made up a total area of 166.3 kha (6.7% of the total agricultural land) [16]. Specifically in grassland, the total area of drained organic soils was 77.0 kha (8.3% of the total grassland area) in Latvia, and it is estimated that these drained organic soils contributed to CO₂ emissions of 1110.9 kt CO₂ eq. or 97.9% from the total net GHG emissions in the grassland category in Latvia in 2021 [16]. It is important to underline that the area of drained organic soils reported under Latvia`s National GHG Inventory includes both deep and shallow organic soils. A previous study has shown that among organic soils with varying soil organic carbon (SOC) content, there is a trend towards higher soil-specific basal respiration with lower SOC content (the specific basal respiration was the highest at the boundary between mineral and organic soils) [7]. Also, at the global scale, it has been found that SOC content has a negative influence on soil heterotrophic respiration [11]. At the same time, no effect on net ecosystem exchange of CO₂ has been found regarding the SOC content within other studies (e.g., [17]). This underscores the potential underestimation of GHG emissions from organic soils that do not fall under the definition stated by the Intergovernmental Panel on Climate Change (IPCC) [18]. Contrary to the expectations, Leiber-Sauheitl et al. (2014), based on a study in northern Germany, concluded that shallow histic Gleysols (including peat mixed with mineral soil, with organic carbon content ~10%) release a similar quantity of CO₂ as deep peat soils (unmixed peat soil with organic carbon content > 30%) [19]. In addition to meteorological conditions (temperature) and GW level, which have been identified as key limiting factors of the magnitude of GHG emissions from organic soils, soil moisture, nutrient content (including plant-available nitrogen and phosphorus), C/N ratio, soil pH, peat type, degree of decomposition, management activities (such as tillage and fertilization) and other aspects have been reported as factors affecting GHG emissions [5,7,8,13]. The result of complex interactions of multiple GHG affecting factors is a large variation in GHG emissions [13,20]. The studies conducted so far underline the need for further region-specific studies involving long-term measurements that additionally focus on the impact of climate and variations in land use practices (e.g., [21]).

In general, grasslands are ecosystems where plant biomass is produced mostly by perennial grasses, sedges and other herbaceous species and where the constant removal of biomass from the ecosystem occurs by wild animals or human activities (e.g., grazing and hay or silage production) [22]. In Latvia, grassland ecosystems are of secondary origin—semi-natural and cultivated grasslands (10 and 90% from the total grassland area, respectively) [22]. This study focused on grasslands with drained organic soils in Europe’s hemiboreal zone. The main objective of this study was to estimate soil-to-atmosphere CO₂, CH₄ and N₂O fluxes from drained organic soils in grassland in Latvia with additional focus on evaluating how the thickness of soil organic layer and the organic C content in soil influence the magnitude of GHG fluxes. We tested the hypothesis that the thickness of the soil organic layer and the organic C content in soil have a significant impact on the magnitude of GHG emissions from drained organic soils in grassland.

2. Materials and Methods

2.1. Research Sites

A two-year (between July 2021 and June 2023) study investigated three research sites with organic (fen) soil in grassland in Latvia (in Europe’s hemiboreal zone, Figure 1). In total, three subplots in each research site were established (

Table 1. General description of research sites in grassland in Latvia included in the study.

Research Site (RS), the Dominant Plant Functional Group	Subplot	Thickness of Soil Organic Layer, cm	Mean Groundwater Level ± S.E. (Range), cm	Coordinates of Subplot (WGS84)
				X	Y
RS1, graminoid	A	15	87.6 ± 2.4 (47–118)	21.18826	56.21136
	B	20	96.1 ± 2.5 (60–126)	21.18817	56.21148
	C	30	98.8 ± 2.6 (58–127)	21.18812	56.21168
RS2, graminoid	A	20	55.4 ± 3.1 (0–121)	22.84421	56.55879
	B	40	54.8 ± 3.3 (8–125)	22.84415	56.55887
	C	70	27.2 ± 3.8 (0–123)	22.84395	56.55900
RS3, forbs and graminoid	A	10	89.2 ± 4.0 (0–146)	24.75648	56.77243
	B	15	84.2 ± 4.0 (0–144)	24.75663	56.77254
	C	25	75.6 ± 8.8 (16–124)	24.75687	56.77279

). Among the research subplots, the thickness of the soil organic layer ranged from 10 to 70 cm, and the mean GW level ranged from 27.2 ± 3.8 to 98.8 ± 2.6 cm below the soil surface.

In Latvia, the mean long-term (1991–2020) annual air temperature was +6.8 °C and annual precipitation was 685.6 mm. In 2021 and 2022, the mean annual air temperature in Latvia was 7.0 and 7.3 °C, respectively, while the annual precipitation was 676.3 and 685.8 mm, respectively [23].

2.2. GHG Sampling Design and Measurements

Soil GHG fluxes were monitored at least once a month for 24 consecutive months between July 2021 and June 2023. Gas sampling to estimate soil total respiration (R_tot)–CO₂ fluxes including both soil autotrophic and heterotrophic respiration as well as CH₄ and N₂O fluxes was conducted with manual sampling of discrete gas samples taken at intervals from closed non-flow-through and non-steady-state chambers (closed static chambers) [24,25] in the three replicates per subplot, nine replicates per plot. Ground vegetation was preserved intact both during collar installation and throughout the entirety of R_tot measurements. Chambers with a volume of 0.0655 m³ were used. Before sampling, the chambers were flushed and then placed on permanently installed collars (installed in soil at 5 cm depth). During the next 30 min period, after the chamber position on the collar, four consecutive gas samples (100 cm³) were taken in 10-min intervals (i.e., four samples per sampling set). CO₂, CH₄ and N₂O concentrations in gas samples were measured using a gas chromatograph method (Shimadzu Nexis GC-230, Shimadzu USA manufacturing, Inc., Canby, OR, USA, delivered by Shimadzu authorised dealer “Armgate” (Liliju iela 20, Mārupe, Mārupes novads, Latvia); software LabSolutions 5.93) at the Latvian State Forest Research Institute ‘Silava’.

Soil heterotrophic respiration (R_het) was measured by the non-steady-state through-flow chamber (closed dynamic chamber) method at least once a month during the vegetation periods between July 2021 and June 2023 with an EGM-5 portable CO₂ gas analyser (PP Systems, Amesbury, MA, USA) using a manual opaque chamber with a volume of 0.023 m³. Each R_het measurement area was prepared by removing the vegetation and trenching using geotextiles to avoid root distribution into the measurement area. Before soil R_het measurements, chambers were flushed. The duration of each measurement was three minutes, and measurements in three replicates were conducted in each subplot.

All soil GHG fluxes were calculated using the equation of ideal gas law and slope of linear regression constructed based on obtained data on GHG concentrations in gas samples reflecting changes of GHG concentration over the measurement time. The GHG sampling design and measurement procedure were the same at all study subplots. Details on the basic design of subplots can be found in Purvina et al. (2023) [26].

2.3. Soil Sampling and Analyses

In June 2021, soil was sampled at each subplot using a soil sample probe (stainless-steel, 100 cm³ core) from the certain layers: 0–10 cm, 10–20 cm, 20–30 cm, 30–40 cm, 40–50 cm, 50–75 cm and 75–100 cm. Pretreatment of soil samples for physico-chemical analysis was completed as stated in LVS ISO 11464:2006 [27]. Soil reaction (pH in KCl) was determined as stated in LVS EN ISO 10390:2022 [28]. The content of the HNO₃-extractable potassium (K) and phosphorus (P) was measured with the inductively coupled plasma-optical emission spectrometry (ICP-OES) method (the microwave mineralization was used). Total carbon (TC) and total nitrogen (TN) content was determined by dry combustion (elementary analysis) as stated in LVS ISO 10694:2006 [29] and LVS ISO 13878:1998 [30], respectively. Inorganic carbon (carbonate) content was determined by the volumetric method as stated in LVS ISO 10693:2014 [31]. The difference between TC and inorganic C content is equal to organic C (OC) content. In addition, the ratio of soil OC and TN content (C/N ratio) was calculated.

2.4. Measurements of Environmental Parameters

At the each GHG flux measurement date, at each subplot, air and soil temperature at 10 cm depth was determined with a Comet data logger equipped with temperature probes. The soil moisture was determined with a ProCheck meter equipped with a moisture sensor. The GW level was measured by measuring tape inside a GW well installed vertically down to a depth of 1.5 m; the GW temperature and dissolved oxygen (DO) content were measured with a YSI ProDSS water quality meter. The GW wells made of PVC pipes were covered between field surveys to minimize the impact of ambient air.

In addition, at the each GHG flux measurement date, GW was sampled from PVC pipes, and the following variables were determined at the laboratory: pH (reaction) was determined according to LVS ISO 10523:2012 [32]; electrical conductivity (Cond.) was determined according to LVS EN 27888:1993 [33]; total N (TN) and dissolved organic caron (DOC) content was determined as stated in LVS EN ISO 20236:2022 [34] and LVS EN 1484:2000 [35]; potassium (K) content was determined by the flame emission spectrometry method, as stated in LVS ISO 9964-3:2000 [36].

2.5. Sampling and Analyses of Above- and Below-Ground Parts of Vegetation

In each subplot, samples of vegetation (both above- and below-ground parts) were taken at four repetitions in August 2021 (area of each sample plot was 625 cm²). First, the above-ground biomass was collected by cutting all vegetation in the sample plot. Then, the below-ground biomass (roots) was collected by excavating to a depth of 0.25 m and removing soil particles. All samples were transported to the laboratory at the Latvian State Forest Research Institute ‘Silava’. To determine the dry mass of above- and below-ground parts of vegetation, samples were cleaned of remained soil particles by wet sieving (samples of below-ground part of vegetation) and dried at 70 °C. The C content in samples was detected by dry combustion (elementary analysis) as stated in LVS ISO 10694:2006 [29].

2.6. Estimation of Soil Annual GHG Emissions

Soil annual GHG emissions were calculated for each study subplot as the sum of monthly mean GHG emissions expressed as t CO₂-C ha⁻¹ month⁻¹ for CO₂ emissions, kg CH₄-C ha⁻¹ month⁻¹ for CH₄ emissions and kg N₂O-N ha⁻¹ month⁻¹ for N₂O emissions (Equation (1)). Monthly mean GHG emissions were calculated using measurement results from two consecutive years covering all calendar months and, thus, seasons. It was assumed that the monthly mean value of measurement results of instantaneous soil GHG emissions were equal to the mean GHG emissions in the relevant month and subplot. Annual net CO₂ emissions were calculated as the difference between annual R_het (expressed as t CO₂-C ha⁻¹ yr⁻¹) and C input with above- and below-ground parts of vegetation (expressed as t C ha⁻¹ yr⁻¹). We did not directly utilize the data of the R_het measurements; instead, R_het was derived using the observed relationship in the study, indicating that R_het is on average 70% of R_tot. Such an approach was used to justify soil C stock change calculation as a sum of estimated R_het and soil C input. Additional assumptions of the decomposition rate of litter would be needed to estimate soil C stock changes by using data of direct R_het and litter biomass measurement data. The annual C input from above-ground parts of vegetation was considered the same as C stock in above-ground parts of vegetation in the end of the vegetation season. To calculate the annual C input with below-ground parts of vegetation, it was assumed that the root turnover rate is 0.5 based on Gill and Jackson (2000) [37].

$${GHG}_{annual}=\sum {GHG}_{monthly-mean}(Jan\dots Dec)$$

(1)

where GHG_annual—soil annual GHG emissions expressed as t CO₂-C ha⁻¹ yr⁻¹ for CO₂ emissions, kg CH₄-C ha⁻¹ yr⁻¹ for CH₄ emissions and kg N₂O-N ha⁻¹ yr⁻¹ for N₂O emissions; GHG_{monthly−mean}(Jan…Dec)—soil monthly mean GHG emissions covering all calendar months (from January to December) expressed as t CO₂-C ha⁻¹ month⁻¹ for CO₂ emissions, kg CH₄-C ha⁻¹ month⁻¹ for CH₄ emissions and kg N₂O-N ha⁻¹ month⁻¹ for N₂O emissions.

2.7. Statistical Analysis

Software environment R (version 3.4.3) was used for all statistical analyses and graphics [38]. The data sets of instantaneous GHG fluxes did not fit a normal distribution (Shapiro–Wilk normality test was used, p < 0.001). Differences in GHG fluxes values grouped by, for instance, research sites or thickness of soil organic layer were evaluated using pairwise comparisons using Wilcoxon rank sum test with continuity correction. Multivariate regression method was used to analyse the relationship between several independent variables (for instance, environmental parameters) and a dependent variable (GHG fluxes).

Based on data quality check, the results of soil R_het measurements from 4 subplots (subplot B at research site 1 and subplots A, B and C in research site 3) were included in further data analysis. Data quality check included exclusion of measurement results obtained in subplots where measurement results of R_het exceeds the results of R_tot from further analysis to avoid risks of overestimating C losses due to soil R_het.

3. Results and Discussion

3.1. Soil Total Respiration (Instantaneous)

In terrestrial ecosystems, soil R_tot (sum of an autotrophic and heterotrophic component) is one of the key components in the C balance and nutrient cycling [39]. Results of our study conducted in grassland with organic soils (thickness of soil organic layer ≤ 70 cm) indicated that the monthly mean R_tot among different subplots ranged from −19.0 ± 11.8 mg C m⁻² h⁻¹ in December to 435.6 ± 62.0 mg C m⁻² h⁻¹ in June, while the annual mean R_tot calculated as the average from monthly means among subplots varied from 84.2 ± 28.8 to 114.6 ± 33.7 mg C m⁻² h⁻¹. No statistically significant difference in R_tot values between subplots with different thickness of soil organic layer was found (r = −0.47, p > 0.41, Figure 2A). Although there is a slight tendency (trend is not significant) towards higher soil R_tot with lower soil organic C content, no significant impact of C content in soil (0–20 cm layer) on the mean R_tot was found (r = −0.39, p = 0.103, Figure 2B) and between mean R_tot and C stock in soil at the 0–20 cm layer (r = −0.38, p = 0.307) and other variables of soil chemistry (

Table 2. Soil variables (0–20, 20–40 and 40–100 cm soil layer) at the research sites (all research sites are pooled, mean ± S.E., minimum and maximum values are provided; n_subplots—number of subplots, n_{soil samples}—number of soil samples).

Values	Soil Physico-Chemical Variables (0–20, 20–40 and 40–100 cm Soil Layer)
	OC Stock, t ha−1	TN Stock, t ha−1	P Stock, t ha−1	K Stock, t ha−1	C/N Ratio	pH (KCl)
0–20 cm soil layer (nsubplots = 9, nsoil samples = 18)
Mean ± S.E.	132.9 ± 12.2	10.7 ± 1.1	1.04 ± 0.12	3.04 ± 0.69	12.8 ± 0.8	5.9 ± 0.3
Minimum	68.5	5.5	0.65	0.98	10.6	5.0
Maximum	174.7	14.9	1.89	6.73	18.8	7.5
20–40 cm soil layer (nsubplots = 9, nsoil samples = 18)
Mean ± S.E.	67.7 ± 19.0	4.7 ± 1.5	0.68 ± 0.09	2.91 ± 0.47	12.3 ± 1.2	6.0 ± 0.3
Minimum	2.3	0.4	0.33	1.63	6.5	5.1
Maximum	167.5	11.2	1.09	5.22	18.8	8.3
40–100 cm soil layer (nsubplots = 9, nsoil samples = 27)
Mean ± S.E.	143.7 ± 37.9	3.9 ± 1.7	2.14 ± 0.32	25.1 ± 11.6	-	6.6 ± 0.3
Minimum	9.9	1.1	0.62	4.58	-	5.6
Maximum	317.9	17.4	3.50	93.5	-	8.2

). Similarly, a study conducted at the Ruukki research station in Finland found that the thickness of the peat layer (20–80 cm) did not significantly influence CO₂ emissions in agricultural land [40]. In Figure 2B, which shows the regression of mean R_tot depending on the C content in soil at the 0–20 cm layer, this study’s results are supplemented by data from a previous study in Latvia conducted in drained grasslands with deep peat soils [41] to obtain a wider data range that includes research sites with higher C content in soil for estimating the relationship.

The instantaneous R_tot was positively correlated with the air temperature (r = 0.71, p < 0.001) and soil temperature at 10 cm depth (r = 0.73, p < 0.001), while it was either not correlated or only weakly correlated with other monitored environmental parameters (r < |0.50|, Figure 3), including the soil moisture, GW level below the soil surface, GW temperature and other water physico-chemical variables (

Table 3. Groundwater physico-chemical variables at the research sites (all research sites are pooled, mean ± S.E., minimum and maximum values are provided).

Values	Groundwater Physico-Chemical Variables
	pH	TN, mg L−1	DOC, mg L−1	K, mg L−1	Cond., μS cm−1	DO, mg L−1
Mean ± S.E.	7.3 ± 0.1	4.48 ± 1.51	19.8 ± 2.1	2.25 ± 0.74	357.6 ± 83.8	7.75 ± 0.93
Minimum	6.9	1.54	9.4	0.41	99.5	5.40
Maximum	7.7	15.16	27.0	7.10	723.0	14.4

). Hence, the environmental variable that best explained the variation in R_tot was soil temperature at 10 cm depth (R² of the linear model with one independent variable was 0.54). The inclusion of other variables into the model such as the thickness of the soil organic layer, C content in soil and other environmental variables did not increase the adjusted R-squared value of the model. Similarly, a study on Danish agricultural peat soils under cool temperate conditions [42] as well as other studies in different climatic regions of the world (e.g., [43]) indicated that temperature (especially soil temperature at 5–10 cm depth), rather than GW level, was the main driver of ecosystem respiration [42].

Within this study, the lowest variation in soil R_tot was observed in a research site where the GW level was constantly below 40 cm during the entire study period, while in other research sites, periodically, the GW level rose and even reached the soil surface (thus, the fluctuation in GW level was higher). Also, the mean R_tot in this object was lower than in other research sites, but the difference was not significant. Thus, short-term soil saturation does not have a significant impact on the magnitude of soil R_tot.

3.2. Soil Heterotrophic Respiration (Instantaneous)

In April–November, the monthly mean R_het among different sublots varied between 13.9 ± 0.5 mg C m⁻² h⁻¹ in November and 274.6 ± 82.7 mg C m⁻² h⁻¹ in July, while the mean R_het calculated as the average from monthly means in April–November, which varied between 74.2 ± 17.8 and 150.1 ± 29.9 mg C m⁻² h⁻¹. The mean contribution of soil R_het to soil total respiration is 70.3%, while the plant-derived CO₂ emissions, calculated as the difference between simultaneously detected R_tot (total CO₂ fluxes from plots with vegetation cover) and R_het (CO₂ fluxes from bare soil) was 29.7%. The estimated contribution of plant-derived CO₂ fluxes is similar to results (27–63%) obtained in central and southern Sweden (where the climate is similar to that of Latvia) and reported by Berglund et al. (2011, 2021) [44,45] and Norberg et al. (2016) [46], thus validating the methodologic approach of R_het‘s measurement results interpretation in the study.

The instantaneous R_het was positively correlated with the air temperature (among the subplots, r ranged up to 0.72, p < 0.001) and soil temperature at 10 cm depth (among the subplots, r ranged up to 0.68, p < 0.001), while these were either not correlated or only weakly correlated with other monitored environmental parameters (r < |0.50|). Thus, based on our dataset, the variable that best reflected the variation in R_het was air temperature (R² of linear model with one independent variable was 0.20, R² of polynomial model was 0.21). However, there was a highly research subplot-specific dependency of soil R_het response to the monitored environmental parameters.

Although no correlation between R_het and soil moisture was found, there is a slight trend that at high air temperatures and low soil moisture, R_het does not follow an increasing regression between R_het and air temperature (Figure 4). Thus, at high air temperature and simultaneously dry conditions, the R_het intensity may stop increasing or even begin lowering. Also, previous studies have concluded that soil moisture has an impact on CO₂ fluxes—a parabolic dependence of CO₂ fluxes on soil moisture was observed [8].

3.3. Soil-to-Atmosphere CH₄ and N₂O Fluxes (Instantaneous)

CH₄ fluxes remained predominantly low, with minimal removal or zero emissions observed during most of the study period. CH₄ fluxes above 0.50 mg CH₄-C m⁻² h⁻¹ were seldom observed. The monthly mean CH₄ fluxes among different sublots ranged from −0.099 ± 0.005 mg C m⁻² h⁻¹ in July to 0.207 ± 0.114 mg C m⁻² h⁻¹ in April, while the annual mean instantaneous soil-to-atmosphere CH₄ fluxes calculated as the average from monthly means among different subplots ranged from −0.057 ± 0.009 to 0.012 ± 0.010 mg C m⁻² h⁻¹. The thickness of the soil organic layer and C content in soil as well as C stock had a strong impact on CH₄ fluxes, and emissions increased with the increasing thickness of the organic soil layer (Figure 5A) and C content in soil (Figure 5B). In general, CH₄ production may occur under anaerobic conditions through methanogenesis in the absence of electron acceptors, while under aerobic conditions, both soil and atmospheric CH₄ can be oxidized [47]. The observed increase in magnitude of CH₄ fluxes with an increase in the thickness of the soil organic layer and C content in soil could be explained by the larger soil organic C availability in deeper soil layer in combination with more anaerobic conditions and, thus, a greater thickness of the potential CH₄ production zone [48]. Furthermore, among our research sites, the thickness of the organic soil layer negatively correlated with the mean GW level below the soil surface (r = −0.79, p = 0.012).

It is well known that temperature is a significant influencing factor of CH₄ fluxes, since both CH₄ production and consumption are microorganism-driven processes [47]. Simultaneously, the GW level and soil moisture are among the key influencing factors of CH₄ emissions; furthermore, it is underlined that significant CH₄ emissions in drained areas occur only when the mean GW level is near the surface for a sufficiently long period [47,49]. Within this study, CH₄ fluxes were largely similar throughout the year despite variations in air temperature, soil temperature, soil moisture and GW level (Figure 6); no strong correlations were found between instantaneous CH₄ fluxes and different environmental parameters (r < |0.50|). Similar results were observed, for instance, in a study focused on organic soils in western Denmark managed by agriculture [50].

The majority of N₂O released from organic soils is the result (by-product) of both denitrification and nitrification processes [51]. In general, we observed low N₂O fluxes with occasional peaks reaching less than 1.0 mg N₂O-N m⁻² h⁻¹. The monthly mean N₂O fluxes among different sublots varied between −0.027 ± 0.004 mg N m⁻² h⁻¹ in December and 0.550 ± 0.076 mg N m⁻² h⁻¹ in June, while the annual mean instantaneous soil-to-atmosphere N₂O fluxes, calculated as the average from monthly means among different subplots, varied between −0.001 ± 0.004 and 0.072 ± 0.030 mg N m⁻² h⁻¹.

No significant difference in N₂O fluxes between subplots with different soil organic layer thicknesses was found (r = −0.18, p = 0.68, Figure 7A). Also, no significant dependence of mean N₂O fluxes on C content in soil at the 0–20 cm layer was found (r = −0.02, p = 0.94, Figure 7B). The mean N₂O fluxes correlated positively with C stock in soil at the 0–20 cm layer (r = 0.46) and N content in soil at the 0–40 cm layer (r = 0.59), although the correlations were not statistically significant (p = 0.215 and p = 0.094, respectively).

The magnitude of N₂O emission is mainly controlled by a number of factors such as climatic variables (especially temperature), electron donor availability, mineral N concentrations, oxygen status and soil carbon to nitrogen ratio and pH [50,51,52]. Within this study, no significant dependence of instantaneous N₂O fluxes on different environmental parameters (r < |0.50|, p > 0.05, Figure 8) was found.

3.4. Annual GHG Fluxes

A summary of the annual R_tot, R_het, C input with above- and below-ground parts of vegetation, as well as the soil-to-atmosphere annual CH₄ and N₂O fluxes is shown in

Table 4. Summary of annual soil-to-atmosphere GHG fluxes released from drained organic soils in grassland in Latvia. Comparison with results obtained in drained grasslands with deep peat soil within previous study in Latvia [41] and with IPCC default emission factors [53] is provided.

Thickness of Organic Soil Layer, cm	Research Site (RS), Subplot	CH4, kg C ha−1 yr−1	N2O, kg N ha−1 yr−1	Rtot, t C ha−1 yr−1	Rhet *, t C ha−1 yr−1	Cinput **, t C ha−1 yr−1	Rhet − Cinput, t C ha−1 yr−1
<20 cm	RS1, A	−0.66	4.63	8.78	6.17	1.28	4.90
	RS3, A	−5.03	1.27	10.09	7.09	2.01	5.08
	RS3, B	−4.09	1.81	8.85	6.22	2.62	3.60
20–40 cm	RS1, B	0.59	1.23	7.76	5.46	1.67	3.78
	RS1, C	0.26	1.15	8.47	5.95	2.72	3.23
	RS2, A	−0.82	6.29	9.52	6.69	4.01	2.69
	RS3, C	−2.73	−0.06	8.04	5.65	2.87	2.78
>40 cm	RS2, B	0.86	4.12	8.03	5.64	3.58	2.06
	RS2, C	1.07	1.03	7.42	5.21	2.01	3.21
<20 cm	Mean	−3.26 ± 1.33	2.39 ± 0.70	8.55 ± 0.29	6.01 ± 0.20	2.53 ± 0.30	3.48 ± 0.33
20–40 cm	Mean	−0.68 ± 0.75
>40 cm	Mean	0.96 ± 0.10
Results from previous study in Latvia conducted in drained grasslands with deep peat (>40 cm) soils *** [41]		57.8 ± 44.3	0.26 ± 0.25	-	-	-	4.39 ± 0.87
IPCC default emission factors for deep-drained, nutrient-rich organic soils in grassland in temperate zone [53]		12.0 (95% confidence interval 1.8–21.7)	8.2 (95% confidence interval 4.9–11)	-	-	-	6.1 (95% confidence interval 5.0–7.3)

* R_het was calculated as 70% from R_tot based on mean observation (R_tot and R_het field measurements). ** Annul C input with above- and below-ground parts of vegetation (Table A1). *** Organic soil with OC content > 190 g kg⁻¹ at 0–20 cm soil depth [41].

. The annual C input into soil with above- and below-ground parts of vegetation in grassland (mean 2.53 ± 0.30 t C ha⁻¹ yr⁻¹,

Table A1. Biomass of vegetation of both above- and below-ground parts (AGB and BGB, respectively) in research sites in grassland in Latvia (sampled in August 2021) and estimated C input to soil with above- and below-ground parts of vegetation (mean ± S.E. values are provided).

Research Site	Biomass, t DM ha−1		C Stock, t C ha−1		Annual C Input, t C ha−1 yr−1
	AGB	BGB	AGB	BGB	Total
RS1	3.88 ± 1.13	0.69 ± 0.10	1.75 ± 0.52	0.29 ± 0.04	1.89 ± 0.43
RS2	6.50 ± 0.92	0.98 ± 0.21	3.00 ± 0.43	0.40 ± 0.09	3.20 ± 0.61
RS3	2.75 ± 0.38	6.26 ± 0.49	1.26 ± 0.18	2.49 ± 0.18	2.50 ± 0.25
All research sites pooled	4.37 ± 0.55	2.65 ± 0.47	2.00 ± 0.26	1.06 ± 0.18	2.53 ± 0.30

) does not compensate for losses of soil C caused by the mineralization of soil organic matter (mean R_het 6.01 ± 0.20 t C ha⁻¹ yr⁻¹). Thus, all soils in the studied research sites were net sources of CO₂ emissions. The annual net CO₂ emissions from studied soils in grassland were calculated as the difference between R_het and C input with the above- and below-ground parts of vegetation, which ranged from 2.06 to 5.08 t CO₂-C ha⁻¹ yr⁻¹ with a mean value of 3.48 ± 0.33 t CO₂-C ha⁻¹ yr⁻¹. The estimated mean annual net CO₂ emissions are lower than the IPCC default emission factors [53] for drained grasslands in temperate zones with deep-drained, nutrient-rich organic soils (6.1 t CO₂-C ha⁻¹ yr⁻¹) and nutrient-poor organic soils (5.3 t CO₂-C ha⁻¹ yr⁻¹), while they were similar to those provided for shallow-drained, nutrient-rich organic soils (3.6 t CO₂-C ha⁻¹ yr⁻¹) and reported in previous studies—for instance, in Finland (3.95 t CO₂-C ha⁻¹ yr⁻¹) [54]. The annual net CO₂ emission factor estimated for drained grassland with deep peat soil in a previous study in Latvia [41] is slightly higher (4.4 t CO₂-C ha⁻¹ yr⁻¹) than our estimates within this study. Among our research sites, no significant correlations between annual net CO₂ emissions and thickness of the soil organic layer or SOC stock were found (r < 0.50, p > 0.05), while the annual net CO₂ emissions correlated positively with mean GW level (r = 0.52), although the correlation was not statistically significant (p = 0.150).

Our research sites acted as both a small CH₄ sink and source with a mean value of −1.17 ± 0.75 kg CH₄-C ha⁻¹ yr⁻¹. Thus, the contribution of CH₄ fluxes to the total GHG balance of studied organic soils in grassland in Latvia was generally insignificant. Monitoring sites in grasslands with organic soil with CH₄ sink profiles or that are neutral with respect to CH₄ fluxes were found also in previous studies in Denmark [50,55], Germany [17] and Nordic countries (e.g., [54,56]). The default CH₄ emission factors stated by the IPCC for drained grassland in temperate zone are significantly higher than our estimates and range from 1.8 kg CH₄ ha⁻¹ yr⁻¹ for nutrient-poor areas to 39 kg CH₄ ha⁻¹ yr⁻¹ for shallow-drained, nutrient-rich areas [53]. Also, the CH₄ emission factor estimated for drained grassland with deep peat (>40 cm) soil in a previous study in Latvia [41] is significantly higher (57.8 kg CH₄-C ha⁻¹ yr⁻¹) than our estimates within this study. All of our research sites were deep-drained with a mean GW level > 50 cm below the soil surface (Table 1) excluding one subplot with a mean GW level of 27.2 ± 3.8 cm and organic soil layer thickness of 70 cm, where the highest CH₄ emissions were detected (1.07 kg CH₄-C ha⁻¹ yr⁻¹). In general, annual CH₄ emissions correlate positively with soil organic layer thickness (r = 0.63, p = 0.067), soil organic C and total N stock in soil (r = 0.86, p = 0.003 and r = 0.62, p = 0.075, respectively). Thus, the thickness of the organic soil layer and soil organic C stock could be one of the reasons for the differences in CH₄ emission factors elaborated previously in Latvia [41] and within this study. No correlation between annual CH₄ emissions and mean GW level was found. Previous studies have emphasized that CH₄ emissions and their variability increased with GW level with higher emissions starting at a GW level of around 20 cm below the soil surface [13]. In our research sites, the mean GW level did not exceed 27 cm below the soil surface (Table 1); thus, the conditions in research sites favour methanotrophy over methanogenesis.

Organic soils in managed grasslands can be a significant source of N₂O emissions [57]. Among our research sites, annual N₂O fluxes ranged from −0.06 to 6.29 kg N₂O-N ha⁻¹ yr⁻¹ with a mean value of 2.39 ± 0.70 kg N₂O-N ha⁻¹ yr⁻¹. The N₂O emission factor provided by IPCC for drained grassland in temperate zone ranges from 1.6 kg N₂O-N ha⁻¹ yr⁻¹ for shallow-drained, nutrient-rich areas to 8.2 kg N₂O-N ha⁻¹ yr⁻¹ for deep-drained, nutrient-rich areas [53]. Our research sites correspond to the deep-drained, nutrient-rich areas; thus, the elaborated annual N₂O emission factor is lower than that provided by the IPCC. The mean N₂O emission factor estimated for drained grassland with deep peat (> 40 cm) soil in a previous study in Latvia [41] is significantly lower (0.3 kg N₂O–N ha⁻¹ yr⁻¹) than our estimates within this study. However, there was a clear spatial variation in the annual N₂O similar to those found from organic agricultural soil, for instance, in Finland [54] and Germany [13]. Studies often emphasize the impact of N fertilizer application, GW level and winter temperature on annual N₂O emissions from organic soils in grassland [20,57]. Among our research sites, the annual N₂O emissions correlated positively with N stock in soil at 0–40 cm depth (r = 0.59, p = 0.095) and GW electrical conductivity (r = 0.65, p = 0.056) and negatively with GW DOC concentration (r = −0.80, p = 0.010).

4. Conclusions

The studied drained organic soils in grassland acted as a source of net CO₂ emissions, releasing a mean of 3.48 ± 0.33 t CO₂-C ha⁻¹ yr⁻¹; furthermore, the net soil C losses made the largest contribution to total soil GHG emissions. The annual C input into soil with above- and below-ground parts of vegetation in grassland does not compensate for losses of soil C caused by the mineralization of soil organic matter. No evidence was obtained that the thickness of the soil organic layer (ranged from 10 to 70 cm) and OC stock in soil can be considered some of the main affecting factors of the magnitude of net CO₂ emissions from drained organic soil.

The studied organic soils were mostly sources of N₂O emissions, releasing a mean of 2.39 ± 0.70 kg N₂O-N ha⁻¹ yr⁻¹, while atmospheric CH₄ exchange ranged from removals in research sites where the thickness of the soil organic layer is <20 cm (mean −3.26 ± 1.33 kg CH₄-C ha⁻¹ yr⁻¹) to emissions in research sites where the thickness of the soil organic layer is >40 cm (mean 0.96 ± 0.10 kg CH₄-C ha⁻¹ yr⁻¹).

Compared to the relevant annual soil-to-atmosphere GHG fluxes expressed by the default emission factors for deep-drained, nutrient-rich organic soils in temperate zones stated by the IPCC, the studied drained organic soils in grassland showed lower GHG fluxes generally. Additionally, both net CO₂ and CH₄ emissions were also lower than previous estimates for deep peat (>40 cm) soils in drained grassland in Latvia.

Future research should focus on continuing to improve estimates of GHG emissions from drained organic soils with different soil organic layer thicknesses distributed across agricultural land in the region, and studies should also be conducted on various additional impacting factors including soil compaction, type of grassland vegetation and species composition as well as nutrient availability.