Tree stems are a net source of CH4 and N2O in a hemiboreal drained peatland forest during the winter period

Nutrient-rich northern peatlands are often drained to enhance forest productivity, turning peatland soils into sinks of methane (CH4) and sources of nitrous oxide (N2O). However, further attention is needed on CH4 and N2O dynamics during the winter period to fully understand the spatio-temporal variability of fluxes. Besides soil, tree stems can also emit CH4 and N2O. However, stem contribution is not considered in most biogeochemical models. We determined the temporal dynamics of winter-time CH4 and N2O fluxes in a drained peatland forest by simultaneously measuring stem and soil fluxes and exploring the relationships between gas fluxes and soil environmental parameters. During sampling (October 2020–May 2021), gas samples from Downy Birch (Betula pubescens) and Norway Spruce (Picea abies) trees were collected from different tree heights using manual static chambers and analysed using gas chromatography. Soil CH4 and N2O concentrations were measured using an automated dynamic soil chamber system. Tree stems were a net source of CH4 and N2O during the winter period. The origin of stem CH4 emissions was unclear, as stem and soil CH4 fluxes had opposite flux directions, and the irregular vertical stem flux profile did not indicate a connection between stem and soil fluxes. Stem N2O emissions may have originated from the soil, as emissions decreased with increasing stem height and were driven by soil N2O emissions and environmental parameters. Soil was a net sink for CH4, largely determined by changes in soil temperature. Soil N2O dynamics were characterised by hot moments—short periods of high emissions related to changes in soil water content. Tree stem emissions offset the soil CH4 sink by 14% and added 2% to forest floor N2O emissions. Therefore, CH4 and N2O budgets that do not incorporate stem emissions can overestimate the sink strength or underestimate the total emissions of the ecosystem.


Introduction
Peatlands form a globally significant carbon (C) and nitrogen (N) reserve. Unmanaged peatland soils are generally a moderate source of methane (CH 4 ) and a weak source of nitrous oxide (N 2 O), both powerful greenhouse gases (GHGs) [1,2]. However, nutrient-rich northern peatlands are often drained to enhance forest productivity [3][4][5][6]. Drainage of these soils, lowering the soil water table and increasing oxygen availability in the topsoil layers, accelerates peat decomposition. As the soil environment is switched from anaerobic to aerobic, CH 4 production by methanogens is reduced and CH 4 oxidation by methanotrophs is enhanced, turning peatland soils into CH 4 sinks [3][4][5]. The established aerobic soil conditions can also increase N 2 O emissions in N-rich peatlands, as enhanced nitrification processes release N 2 O as a by-product [4,5,7]. Although the general Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. effects of these microbial processes on GHG dynamics in drained peatland soils are relatively well studied, more attention is needed on winter-time activity.
Seasonal variability in weather conditions strongly influences the GHG dynamics of northern peatlands by creating annual variability in the environmental parameters that most influence soil fluxes, such as soil water content (SWC), water table depth (WTD), air and soil temperatures. Although growing-season soil GHG dynamics in boreal and hemiboreal drained peatlands have been thoroughly studied [8][9][10], winter-time fluxes have received little attention, with a lack of high-frequency measurements. Few previous winter-time studies in hemiboreal drained peatlands show potential for significant CH 4 oxidation [11] and substantial N 2 O emissions with peaks during freeze-thaw events [11,12]. These previous results suggest that the dormant period could still be a significant contributor to the annual budgets of CH 4 and N 2 O. However, further data is needed to understand the role of different compartments of the ecosystem and the underlying processes behind CH 4 and N 2 O flux dynamics. In addition, more long-term studies are required to account for potential variability in GHG fluxes across years. This is particularly crucial in the context of climate change and the related effects on water and temperature regimes in northern peatlands [13].
Furthermore, tree stem fluxes' seasonality, particularly the winter-time activity, has not yet received much attention. Previous stem flux studies focus mainly on short measuring periods during the growing season with low sampling frequencies [15,16,18,22,24,25]. Annual studies displaying seasonal patterns are rare [12,19,23] and N-rich drained peatland soils have not been studied. Although the physiological activity of trees in northern regions is reduced in the winter [23], it is essential to examine the specific environmental drivers behind the seasonal variability of CH 4 and N 2 O flux dynamics.
In this study, we look at spatial and temporal patterns of CH 4 and N 2 O fluxes from tree stems and soil in a drained peatland forest during the winter months. We examine the possible origin of tree stem fluxes by simultaneously measuring tree stem and soil fluxes with a high sampling frequency, and exploring the relations between the gas fluxes and soil environmental parameters. We hypothesise that: (I) soil and stems emit N 2 O during the winter period, (II) SWC is the main driver of soil and stem CH 4 and N 2 O fluxes, and (III), stem fluxes decrease with stem height, indicating that fluxes originate from soils.

Site description and study design
The study was conducted in a drained peatland forest site (58°17′N, 27°17′E; 38 m.a.s.l.) in the eastern part of Estonia, which belongs to the hemiboreal vegetation zone [28]. The stand was drained in the early 1970s using an open-ditch network drainage system [29] and belongs to the Oxalis site type. The soil was classified as Drainic Eutric Histosol [30] with low dry bulk density, high organic C content, and low pH [6,29]. The area is covered mainly by Downy birch (Betula pubescens Ehrh.) and Norway spruce (Picea abies (L.) H. Karst.) trees. The longterm average precipitation in the area is 650 mm, the average temperature is 17°C in July and −6.7°C in January, and the growing season lasts 175-180 days [31]. The tree stand characteristics are presented in Supplementary table 1.

CH 4 and N 2 O flux measurements
Twelve representative monitoring points were selected in the study area. Each of the first six points consists of one Downy Birch and one Norway Spruce tree with installed stem chambers and one automatic dynamic soil chamber. The remaining six monitoring points are set pairs of one birch tree and one soil chamber.

Tree stem gas fluxes
Gas samples from the stems of Downy birch (n = 12) and Norway spruce (n = 6) trees were collected manually using static stem chamber systems during weekly measurement campaigns (29 campaigns in total) from 22 October 2020 to 3 May 2021. A chamber system consisted of two chambers per height profile, placed randomly across 180°(total area of 0.0108 m 2 stem surface, total volume of 0.00119 m 3 ). Stem chambers were made of transparent rectangular plastic containers and removable airtight lids (Lock & Lock, Seoul, South Korea), with the bottom removed and hot-glued with a neoprene band [32]. Chambers were attached to the smoothed stem surface with non-acid silicone and tested for air-tightness. Chambers were installed at stem heights of 0.1, 0.8 and 1.7 m above the ground to measure the stem flux's vertical profile, except for six birch trees with chambers only at the lowest height.
During sampling between 09:00 and 13:00, four mixed 25 ml gas samples were collected in pre-evacuated (0.3 bar) gas-tight vials (LabCo International, Lampeter, United Kingdom) from each chamber system in 60 min intervals (0/60/120/180 min sequence). Gas samples were analysed within two weeks using gas chromatography (GC-2014, Shimadzu, Kyoto, Japan), equipped with a flame ionisation detector for CH 4 and an electron capture detector for detection of carbon dioxide (CO 2 used for data quality check) and N 2 O concentrations.

Soil gas fluxes
Soil gas fluxes were measured from 4 December 2020-3 May 2021 using twelve automated dynamic chambers (area of 0.16 m 2 soil surface, volume of 0.032 m 3 ), one on each monitoring point. The closing time of the chamber during the measurement was nine minutes, followed by a one-minute flushing period with ambient air for the whole system. Air was sampled from the closed chamber's headspace and analysed with a gas analyser (G2508, Picarro Inc., Santa Clara, California, United States), which uses cavity ring-down spectroscopy technology to measure CO 2 , CH 4 and N 2 O concentrations.

Flux calculations and data quality check
Stem and soil CH 4 and N 2 O flux rates were calculated based on the linear regression approach of chamber headspace gas concentration change over time using the following equation (equation (1)): Where M = molecular mass of the gas, P = air pressure, V = chamber volume, v s = the slope of linear regression gas concentration change in chamber headspace during the sampling time, R = gas constant (8.314), T = temperature, t = sampling time, and A = soil surface area covered by the chamber. The quality of the chamber measurement session was validated using the the adjusted R 2 value of the linear regression for the CO 2 measurements. Flux values were accepted only if the R 2 value exceeded 0.9. To compare the contribution of soil and stem CH 4 and N 2 O fluxes, stem fluxes averaged across the three heights were upscaled to a hectare of ground area, calculated based on tree stand characteristics brought out in Supplementary table 1, assuming a cylindric shape of the tree stems [17,33].

Ancillary measurements
Soil and air temperature, SWC and WTD were continuously measured during the study period. Soil temperature (107, CAMPBELL SCIENTIFIC. INC, Logan, Utah, USA) and SWC sensors (ML3 ThetaProbe, Delta-T Devices, Cambridge, United Kingdom) were placed vertically at 0.1 m soil depth inside the soil chambers. In groundwater wells, the WTD was observed using automatic data loggers (Hobo U20L-04, Onset Computer Corporation, Bourne, Massachusetts, USA).

Statistical analysis
Statistical analysis was performed using R version 4.0.3 [34]. The normality of data distribution was examined using the Kolmogorov-Smirnov, Lilliefors and Shapiro-Wilks tests. As flux data were not normally distributed, non-parametric tests were used. Kruskal-Wallis one-way analysis of variance was used to determine the significance of temporal variability of gas fluxes, and differences between stem fluxes at different heights. Dunn's multiple comparison, corrected with the Bonferroni method, was conducted as a post hoc test to determine which groups differed. Spearman's rank correlation was used to analyse relationships between soil and stem CH 4 and N 2 O fluxes and meteorological parameters. The significance level was defined at p < 0.05 for all tests.

Stem CH 4 and N 2 O fluxes
Stems were a net source of CH 4 and N 2 O during the winter period. On average, birch stems emitted more CH 4 . Fluxes were statistically different in time and between tree species (p < 0.05).
Temporal dynamics of both birch and spruce stem CH 4 emissions were characterised by a notable peak in November (daily mean reached 1.78 ± 0.47 μg C m −2 h −1 for birch and 1.96 ± 0.49 μg C m −2 h −1 for spruce), fluctuations between small emissions and uptake in the winter, and a slight increase in emissions in spring ( figure 1(B)). Birch CH 4 fluxes had the highest correlation with SWC, while spruce CH 4 fluxes did not have statistically significant correlations with any meteorological parameters (Supplementary table 2).
Birch trees emitted N 2 O throughout the study period, with the highest emissions in October (up to 11.43 ± 3.89 μg N m −2 h −1 ), after which emissions dropped to near-zero in the winter months. Fluxes increased from the end of February to the end of March but dropped again in April and May. Spruce stem fluxes fluctuated around zero between small emissions and uptake throughout the study period ( figure 1(C)). Both birch and spruce N 2 O fluxes were mostly driven by changes in WTD (Supplementary table 2 2(B)).

Soil CH 4 and N 2 O fluxes
Soil was a net sink for CH 4 (−2.00 ± 0.12 μg C m −2 h −1 ) and a source of N 2 O (50.46 ± 2.77 μg N m −2 h −1 ) across the study period. Fluxes were statistically different in time (p < 0.05). Methane uptake was higher during winter onset, with the largest consumption at the beginning of January (up to −8.13 μg C m −2 h −1 ). From late January to early April, fluxes hovered around zero, with occasional CH 4 release. From April through May, soils again turned into a CH 4 sink ( figure 1(B)). Soil CH 4 fluxes had the most significant relationship with soil temperature, but also SWC and WTD. No correlation was found between soil and stem CH 4 fluxes (Supplementary table 2).
Hot moments-short periods of notably higher daily average emissions compared to the period averagecontributed 64.2% of the whole study period N 2 O emissions, while accounting for 31.8% of the time in days  (Supplementary table 3).
During the whole study period, soil N 2 O fluxes correlated positively with air temperature, SWC, soil temperature, and WTD. Soil N 2 O fluxes correlated with stem N 2 O fluxes from birch, but not from spruce (Supplementary table 2). Relationships between soil fluxes and SWC had higher statistical significance during hot moments (II) (p < 0.00001, r = 0.88) and (III) (p < 0.00001, r = −0.88). In addition, N 2 O flux correlated positively with air temperature during hot moment (I) (p < 0.01, r = 0.83) and with WTD during hot moment (II) (p < 0.00001, r = 0.86).
Stem and soil fluxes, when upscaled to hectare of forest ground, showed that birch and spruce stem CH 4 emissions offset the soil CH 4 sink by 10.4% and 3.5%, respectively ( figure 3(A)). Stem emissions accounted for 1.9% of the total soil and stem N 2 O emissions, almost entirely related to birch fluxes ( figure 3(B)).

Tree stems as a net source of CH 4
Birch and spruce stem CH 4 fluxes in drained peatlands have not been previously studied and cold-period measurements from other ecosystems are limited. Thus, comparison with other studies is challenging. Compared to our results of average birch stem emissions of 0.18 μg C m −2 h −1 , previous studies on cold-period emissions from various deciduous tree species showed much higher emissions. For example, emissions around 10 μg C m −2 h −1 have been reported from alder stems in a hemiboreal riparian forest [19], whereas average stem emissions reached around 50 μg C m −2 h −1 from birch stems and 110 μg C m −2 h −1 from alder stems during winter months in a temperate forested peatland [21]. In addition, we found spruce stem CH 4 emissions to be lower than birch emissions, similarly to boreal trees measured by Vainio et al [35]. Spatial and species-level variability in stem fluxes can been linked to hydrologic conditions in the soil [14,17], aspects of stem morphology and tree physiology, such as wood density, lenticel abundance, transpiration, sap flow rates, and wood vessel structure [22,26,27,36,37], as well as differences in microbiology in stems and soils [17,21].
Stem CH 4 emissions peaked in November, with fluctuations between small emissions and uptake in the winter and a slight increase in emissions in spring ( figure 1(B)). Similar seasonal patterns were observed in a hemiboreal riparian forest [19] and in a temperate forested peatland [21]. Several previous studies with smaller sampling frequencies [20] and only growing season measurements [22,37] have not found clear seasonal trends in CH 4 stem fluxes. Moreover, Barba et al [15] found seasonality only with automated high-frequency measurements, but not with manual measurements, emphasising the need for increased high-frequency stem CH 4 and N 2 O measurements to better understand the temporal and spatial variations.
Birch and spruce stem CH 4 emissions combined offset the soil sink by 14%. A wide range of estimates have been proposed previously, from stem flux contributions of 3.5% [37] and 1%-6% [26] to 63% [38] in temperate forests. In addition, in ecosystems where soil is a net CH 4 source, stem emissions can add up to 27% to the total soil source in a temperate wetland [21] and up to 83% in a riparian forest [19]. Therefore, CH 4 budgets that do not incorporate stem emissions can significantly overestimate the sink strength or underestimate the total emissions of the ecosystem [38].

Tree stems as a net source of N 2 O
To the best of our knowledge, our study is among the first to focus on winter-time N 2 O stem fluxes. Stems were a net source of N 2 O with birch stems emitting significantly more than spruce. Conversely, spruce stems in a boreal forest emitted significantly more N 2 O anually, as well as during the dormant season, than birch stems [23].
Inter-species variability of N 2 O fluxes could be explained by spatial variability of N 2 O concentration in the soil or differences in trees' physiological properties, such as projected leaf area per tree [18,23]. Further examination of these variables in different ecosystems is needed to explain the discrepancy in species-level fluxes in different ecosystems. Seasonality of stem N 2 O fluxes is generally driven by ecosystem activity and physiological activity of trees [23]. We observed birch N 2 O emissions throughout the study period, but spruce fluxes were negligible. Birch emissions were highest in late autumn and early spring, and fluctuated near-zero in the winter. Early autumn and early spring also contribute most to the cumulative stem N 2 O emissions from Alnus incana in a riparian forest [12]. Boreal trees also emit N 2 O during dormancy, agreeing with our results of prevailing emissions during winter months [23].
Stem fluxes added about 2% to forest floor N 2 O emissions, almost entirely related to birch fluxes. Previous studies also report low contributions on annual scales, such as Machacova et al [23] (birch 0.75% and spruce 2.5%), and Mander et al [12] (alder 0.8%). However, there are no studies to compare stem fluxes' contributions solely during the winter months. Although the contribution of stem N 2 O fluxes to total emissions is relatively insignificant, even minor emissions from trees can accumulate to a substantial addition to the annual balance of forest N 2 O emissions.

Origin and drivers of stem fluxes
The origin and drivers of CH 4 and N 2 O stem fluxes are contested. Stems have been suggested to act as conduits for CH 4 [16-18, 26, 36, 39] and N 2 O [23,24] produced in the soil. However, CH 4 can also be produced on the stem bark [40] or inside the tree stem by methanogenic microbes [15,38,41,42]. The origin of tree stem CH 4 emissions is not clear based on our results. Stems constituted a CH 4 source, while the soil was a net sink, which could be evidence for methanogenic activity inside the woody tissue [27,42]. Birch emissions were driven by changes in SWC (figures 1(A), (B); Supplementary table 2), but the vertical stem flux profile was irregular (figure 2(A)). Therefore, it is possible that the CH 4 that is taken up by the tree roots originates from deeper soil layers where roots are more abundant, whereas methanotrophic activity leading to net soil CH 4 uptake dominates over methanogenesis in the shallower layers of the soil. This could also explain higher emissions from birch stems than from spruce, as the fine root density of Norway spruce is generally higher closer to the soil surface [43]. Thus, only birch roots may reach the deeper soil layers to facilitate CH 4 transport [35]. To further examine the origin of tree stem CH 4 emissions, information about CH 4 concentrations in the soil vertical profile and identification of methanotrophic and methanogenic bacteria in different soil layers, as well as in the tree heartwood, is needed.
Our results suggest that N 2 O fluxes may originate from the soil. Birch emissions decreased with increasing stem height (figure 2(B)) and were driven by soil N 2 O emissions, changes in WTD and soil temperature (Supplementary table 2). In agreement with previous studies reporting declining N 2 O stem fluxes with tree height [12,20,33], this supports that N 2 O is predominantly produced in the soil.

Soil CH 4 dynamics
Soil in the drained peatland forest was a net CH 4 sink during the winter with higher uptake during winter onset ( figure 1(B)). Fluxes stayed nearly dormant in February, coinciding with constant low soil temperatures. Snow cover may be a barrier for gas exchange and the frozen topsoil layer could restrict gas diffusion due to ice in soil pores [44]. Spring onset presented fluxes fluctuating between emissions and uptake, agreeing with results from an adjacent drained peatland location [11]. Methane stored in frozen soils can be released to the atmosphere through changes in pressure during freeze-thaw periods [45]. Soils turned into a CH 4 sink in mid-April, together with increasing soil temperatures. Soil temperature plays a role in determining CH 4 dynamics by increasing the metabolic activities of microorganisms [46,47]. In addition to soil temperature, soil CH 4 seasonality was also determined by SWC and WTD (Supplementary table 2). Previous studies consider SWC as the primary factor determining CH 4 dynamics in forest ecosystems [3,19,46,48]. Methanogenesis is facilitated by anaerobic conditions induced by higher SWC [46,49], whereas methanotrophic CH 4 consumption is increased under drier aerobic conditions [44,46].

Soil N 2 O dynamics
Soils were a net source of N 2 O during the winter period. Hot moments and hot spots predominantly determine spatio-temporal dynamics of N 2 O emissions from soils [12]. We identified three hot moments, during which daily average N 2 O emissions were notably higher than the period average, accounting for 64.2% of the total release of N 2 O (figure 1(C); Supplementary table 3). Hot moments are primarily controlled by changes in SWC and they can be induced by drying-rewetting or freeze-thaw cycles [12,50,51].
Hot moments (I) and (II) were likely the result of soil freeze-thaw events. The increases in daily average N 2 O emissions coincided with the rising WTD and air temperature (figures 1(A), (C)). During hot moment (II), SWC also increased simultaneously with N 2 O fluxes, reaching 0.88 m 3 m −3 , indicating a high SWC optimum for N 2 O release. Various explanations have been proposed to describe the effects of freeze-thaw on N 2 O emissions. Disruption of soil aggregates can lead to rapid mineralisation of previously physically protected organic matter [52]. Dead cells of microorganisms, fine roots, and mycorrhiza in the soil, destroyed by freezing, can rapidly decompose during thawing [50,52]. Death of fine roots can also leave more nitrate available for microorganisms for N 2 O production [12,51]. As snow cover keeps deeper soil layers unforozen, soil microorganisms can remain active, stimulating organic matter decomposition and N mineralisation, prompting N 2 O production and release [52,53]. In addition, rapid increase in N 2 O fluxes at high SWC during the freeze-thaw hot moments can be explained by increased O 2 content in snow melt water which can inhibit the full denitrification pathway from N 2 O to N 2 [54]. In low temperature zones (below 10°C), nitrifier denitrification is likely to dominate total denitrification in fluctuating aerobic-anaerobic conditions [55].
Hot moment (III) showed the highest emissions of N 2 O and is related to fluctuations in SWC. An increase in SWC happened directly prior to the hot moment, followed by a drop in SWC at the beginning of the hot moment (figures 1(A), (B)). Therefore, before the hot moment, SWC may have exceeded the optimum, and the water level has decreased to reach the optimum SWC for N 2 O release to create the peak in emissions. Furthermore, the initial increase in N 2 O emissions concurred with rising air and soil temperatures, indicating that temperature increase may be driving the N 2 O emissions rise by stimulating microbial activity in the soil [12].

Conclusions
We studied the spatial and temporal dynamics of CH 4 and N 2 O fluxes from tree stems and soil in a drained peatland forest during the winter months, narrowing the knowledge gap on seasonal gas exchange patterns. We observed a substantial amount of N 2 O being emitted from tree stems and soils during the winter. The availability of water in the soil was the primary factor influencing both CH 4 and N 2 O from soil and tree stems. Birch stems played a much greater role in the winter CH 4 and N 2 O dynamics of the forest than did spruce stems. The vertical profile of stem fluxes allowed us to observe the possible origin of stem fluxes, showing that while the source of stem CH 4 is unclear, N 2 O is likely produced in the top soil layers. Stem emissions offset the soil CH 4 sink by 14% and added about 2% to forest floor N 2 O emissions, indicating that CH 4 and N 2 O budgets that do not incorporate stem emissions can overestimate the sink strength or underestimate the total emissions of the ecosystem. Thus, stem fluxes both during dormant and growing seasons must not be neglected in annual forest GHG budgets and the potential contribution of these fluxes to GHG balances of different ecosystems must be investigated further. Longer multi-year studies are needed on GHG dynamics from different forest compartments under different environmental conditions to address uncertainties related to inter-annual variability of GHG fluxes, as these fluxes can vary considerably across years. Furthermore, our results demonstrate that more research is needed into the origin of stem GHG fluxes and how different environmental parameters influence these fluxes in various tree species. Identifying biogeochemical pathways and microbial processes associated with CH 4 and N 2 O dynamics in tree stems could help further our understanding of the origin and drivers of stem fluxes.