Soil carbon balance of afforested peatlands in the maritime temperate climatic zone

Drainage and conversion of natural peatlands to forestry increases soil CO2 emissions through decomposition of peat and modifies the quantity and quality of litter inputs and therefore the soil carbon balance. In organic soils, CO2 net emissions and removals are reported using carbon emission factors (EF). The choice of specific default Tier 1 EF values from the IPCC 2013 Wetlands supplement depends on land‐use categories and climate zones. However, Tier 1 EF for afforested peatlands in the temperate maritime climate zone are based on data from eight sites, mainly located in the hemiboreal zone, and the uncertainty associated with these default values is a concern. In addition, moving from Tier 1 to higher‐Tier carbon reporting values is highly desirable when large areas are affected by land‐use changes. In this study, we estimated site‐specific soil carbon balance for the development of Tier 2 soil CO2‐C EFs for afforested peatlands. Soil heterotrophic respiration and aboveground tree litterfall were measured during two years at eight afforested peatland sites in Ireland. In addition, fine‐root turnover rate and site‐specific fine‐root biomass were used to quantify belowground litter inputs. We found that drainage of peatlands and planting them with either Sitka spruce or lodgepole pine, resulted in soils being net carbon sources. The soil carbon balance at multi‐year sites varied between 63 ± 92 and 309 ± 67 g C m−2 year−1. Mean CO2‐C EF for afforested peatlands was 1.68 ± 0.33 t CO2‐C ha−1 year−1. The improved CO2‐C EFs presented here for afforested peatlands are proposed as a basis to update national CO2‐C emissions from this land‐use class in Ireland. Furthermore, new data from these sites will significantly contribute to the development of more reliable IPCC default Tier 1 CO2‐C EFs for afforested peatlands in the maritime temperate climate zone.


| INTRODUC TI ON
Soil plays a major role in the global carbon (C) balance as it is the largest terrestrial pool of organic C worldwide with 1500-2400 Pg C in the upper 100-200 cm of the soil (Batjes, 2014). Of this, approximately 547 Pg C is stored in northern peatlands and organic soils (Yu et al., 2010) covering about 3% of the Earth's land surface (Parish et al., 2008). In undrained or rewetted conditions, a high water table maintains anaerobic soil conditions, which restrict organic matter decomposition, while continuing plant-litter inputs maintain an accumulating, negative soil C balance. When peatlands are drained for other land uses such as peat extraction, agriculture or forestry, the soil becomes aerobic and the peat progressively decomposes thereby emitting large amounts of carbon dioxide (CO 2 ) through respiration by decomposer organisms (Parish et al., 2008). In addition, drainage alters the physical and chemical properties of peat, while land-use changes modify the quantity and quality of litter inputs and therefore the soil C balance in peatlands (Hargreaves et al., 2003;Laiho & Finér, 1996;Laiho & Pearson, 2016).
From an approximate extent of 50 million ha of natural peatlands that have been drained worldwide (Joosten, 2010), around 15 million ha have been drained for forestry in northern biomes (Paavilainen & Päivänen, 1995). Moreover, it has been estimated that approximately 440,000 ha of peatlands have been drained for afforestation in Ireland (Renou- Wilson & Byrne, 2015) and 439,000 ha in the United Kingdom (Evans et al., 2017). Between 46 and 51% of total afforestation in Ireland during the period 1990-2005 was on organic soils (Black et al., 2008). Drainage for forestry increases soil CO 2 emissions through decomposition of peat, while forest growth accumulates biomass and litter C. Therefore these afforested peatlands can act as sinks or sources of CO 2 depending on the balance between C input and loss. Understanding the soil C balance of these extensive ecosystems has great importance in the global C cycle as CO 2 is the anthropogenic greenhouse gas (GHG) contributing most to climate change (IPCC, 2007). Although the net CH 4 exchange should also be considered in the soil C balance, net CH 4 fluxes in drained peatland forest ecosystems are low (Coles & Yavitt, 2002;Yamulki et al., 2013;Zerva et al., 2005) and they have their own emission factor (EF) reported as CH 4_organic (Evans et al., 2017;IPCC, 2014).
Approximately 26% of total CO 2 emitted (around 555 Pg C) during the industrial era  is derived from land-use change activities (including deforestation, reforestation and afforestation) with the other main contributors being fossil fuel combustion and cement production (Hartmann et al., 2013;Le Quéré et al., 2015).
Moreover, the atmospheric concentration of CO 2 has significantly increased from 317 ppm in 1958 to 412 ppm in 2018 (Keeling et al., 2001). To help tackle the negative effects of increased atmospheric GHG concentrations on the global climate, signatory countries of the Kyoto Protocol (UNFCCC, 1997), and the Paris Agreement (COP21), need to report national GHG inventories biennially under the United Nations Framework Convention on Climate Change. This includes monitoring and reporting soil C stocks (SCS), rates of change and their associated GHG emissions.
In organic soils, CO 2 net emissions and removals are reported using C EFs (in t CO 2 -C ha −1 year −1 ) (IPCC, 2003). The Intergovernmental Panel on Climate Change (IPCC, 2014) established the need to disaggregate CO 2 -C EFs among off-site losses (e.g. leaching of dissolved organic C; CO 2 -C DOC ), anthropogenic peat fires (L fire -CO 2 -C) and on-site emissions (CO 2 -C on-site ) -the EF considered in this study. In drained peatland forests, CO 2 -C on-site EF represents the net decomposition loss, which can be estimated as the difference between soil CO 2 emissions originating from the decomposition of peat plus litter (heterotrophic respiration; R H ), and CO 2 removals from the atmosphere through litter accumulation (aboveand belowground inputs) (Evans et al., 2017;IPCC, 2014;Minkkinen et al., 2018;Ojanen et al., 2012). Byrne and Farrell (2005) reported that soil respiration in afforested blanket peat in Ireland varies between 1.0 and 2.7 t C ha −1 year −1 . The same authors concluded that it was likely that the soda-lime method used to measure soil respiration may have underestimated soil CO 2 fluxes because this method is not as reliable as infrared-absorption methods. Reported soil respiration values for afforested peatlands in oceanic climates and forestry-drained peatlands in boreal regions vary between 4.5 and 11.1 t C ha −1 year −1 (Jovani-Sancho et al., 2018;Mäkiranta et al., 2008;Minkkinen et al., 2018;Ojanen et al., 2010;Yamulki et al., 2013). Soil R H is commonly measured in partitioning studies of soil respiration using trenching techniques and portable soil respiration chambers (Byrne & Kiely, 2006;Ferréa et al., 2012;Jovani-Sancho et al., 2018;Mäkiranta et al., 2008;Saiz et al., 2006). Although more accurate methods with less soil disturbance effects exist to separate R H from total soil respiration, such as C isotope techniques, these methods are expensive and difficult to implement in the field or over large areas (Kuzyakov, 2006). In addition, trenching experiments and chamber methods are recommended for the estimation of the soil C balance over large areas (Ojanen et al., 2012). Nevertheless, it is important to acknowledge that trenching of roots generates some artefacts that may lead to overestimation or underestimation of R H . These are: changes in soil moisture conditions, decomposition of roots, elimination of root exudates and their soil-priming effects (Comstedt et al., 2011;Díaz-Pinés et al., 2010;Heinemeyer et al., 2012;Savage et al., 2018).
Litterfall production can be estimated by direct measurements of tree litterfall using litterfall collectors (McShane et al., 1983;Saarsalmi et al., 2007) and indirect estimation of annual root inputs by applying root turnover rates to measured root biomass (Laiho et al., 2008). Furthermore, litter from ground vegetation (shrubs, herbaceous vegetation and mosses) should also be considered when appropriate (Laiho et al., 2008(Laiho et al., , 2011Ojanen et al., 2014).
The choice of specific default EF values to estimate net emissions-Tier 1 on the IPCC three-Tier scale of methodological approaches of increasing analytical complexity-depends on land-use categories and climate zones (IPCC, 2014). In temperate drained afforested organic soils default EFs range between 2.0 and 3.3 t CO 2 -C ha −1 year −1 (IPCC, 2014). However, these EFs are based on data from eight sites only and the uncertainty associated with these default values is a concern. Moreover, seven out of the eight study sites were located in the hemiboreal zone ( Figure 1). In addition, moving from Tier 1 to region-specific and higher-Tier GHG reporting values is highly desirable when large areas are affected by land-use changes (IPCC, 2006;Wilson et al., 2015) such as drainage and afforestation of natural peatlands. The same authors also recommend developing strong Tier 2 values for different land-use categories and climate regions. The EF for afforested peatland currently used in Ireland's GHG National Inventory Report (NIR) for the period 1990-2019 is 0.59 t CO 2 -C ha −1 year −1 (Duffy et al., 2021). This EF is much lower than the Tier 1 default value range of 2.0 and 3.3 t CO 2 -C ha −1 year −1 for the temperate zone (IPCC, 2014). Duffy et al. (2021) attributed their lower EF to an interpretation of the soda-lime-derived soil respiration rates reported by Byrne and Farrell (2005), though more reliable measures, and documented calculation, are now needed.
Although the United Kingdom uses a model-based Tier 3 method to report CO 2 emissions from drained forest land, Evans et al. (2017) developed a Tier 2 EF of 2.0 t CO 2 -C ha −1 year −1 for this land use and climatic zone. However, the same authors concluded that more studies were needed to reduce the large uncertainty of this EF because of: the limited number of studies used to develop the EFs; the estimation of R H fluxes based on total soil respiration measurement (e.g. UK Tier 2 EF assumes that R H represents 50% of total soil respiration); and the incomplete quantification of litter inputs into the soil (for both IPCC Tier 1 and UK Tier 2 values) (Evans et al., 2017).
Similarly, net CO 2 emissions and removals for all C pools in the forest land category (including afforested peatlands) in Ireland are reported using a Tier 3 model (i.e. CFS-CBM-Carbon Budget Model of the Canadian Forest Sector). However, this process-based model used in the NIR uses a static Tier 2 value to report soil CO 2 emissions from afforested peatlands (Duffy et al., 2021). Although a more dynamic EF based on site characteristics, forest age, temperature and ground water level would allow more-accurate simulation of soil CO 2 emissions from peatland forests, a static Tier 2 EF could be used to cross-validate CO 2 estimates from Tier 3 models (Evans et al., 2017).
To understand whether soils under this land use are sinks or sources of C, and also to scale up results to regional and countrywide level, studies in which all the components of the soil C balance are quantified are necessary. Although several studies have been conducted to assess the soil C balance in forestry-drained peatlands in boreal and hemiboreal climates (Bechtold et al., 2018;Meyer et al., 2013;Minkkinen et al., 2018;Ojanen et al., 2013Ojanen et al., , 2014 and in afforested organo-mineral soils in temperate climates (Friggens et al., 2020;Zerva et al., 2005) the soil C balance of afforested peatlands in oceanic climates has received little attention. While undrained blanket peatlands in temperate maritime conditions are soil C sinks (excluding DOC losses and CH 4 fluxes) of around −0.56 ± 0.19 t C ha −1 year −1 (Koehler et al., 2011;McVeigh et al., 2014) the soil C balance of afforested peatlands is still uncertain. To the best of our knowledge, no work has been conducted in afforested peatlands in this climatic zone using the soil chamber method and accounting for  Glenn et al. (1993). In each frame, scale bar represents 1000 km estimates of tree and root litter inputs. The aim of this study was to estimate site-specific soil C balance for the development of Tier 2 countrywide soil CO 2 -C EFs for afforested peatlands in temperate maritime climate conditions. Improved CO 2 -C EFs presented here for afforested organic soils in Ireland are proposed as a basis to update national CO 2 -C emissions from this land-use class. It was hypothesized that in these climatic conditions, soil CO 2 emissions from decomposition of peat and litter in drained afforested peatlands cannot be compensated by the C incorporation from litter inputs.

| Study sites
Field studies were conducted at eight drained and afforested peatland sites located on the Mullaghareirk Mountains, in southern Ireland (Figure 1). The study sites were located in a maritime temperate climate zone-Cfb of the Köppen-Geiger climate classification by Kottek et al. (2006), characterized by abundant annual rainfall (1326-1716 mm year −1 ) and mild mean annual air temperatures (8.9-10.9°C) ( Table 1, Rockchapel, rainfall and Mount Russell, temperature, weather stations for the years 2010-2020 and 1993-2020 respectively, Met Éireann, Irish Meteorological Service). The high precipitation, with nearly 200 days with 1 mm or more rainfall, in addition to the low annual potential evapotranspiration (about 500 mm year −1 ) conditions prevailing in the area leads to persistently wet soils (Collins & Cummins, 1996). Seven of the sites had plantations of Sitka spruce (Picea sitchensis (Bong.) Carr.), between 18 and 44 years old ( Table 2). The other site was a 23-year-old lodgepole pine (Pinus contorta Dougl.) plantation (site P23). All sites except S18 were first rotation plantations. All sites were established on poorly drained Dystric Histosols (IUSS Working Group WRB, 2015). All sites had closed canopy and the older sites were mature and ready for harvesting. In all sites, ground vegetation consisted of thin and patchy mosses over about half the ground area (Hylocomium splendens, Pleurozium schreberi and Polytrichum sp.) and forest lichens.

| Soil carbon balance
Site-specific soil C balance (ΔC soil ) was calculated as the difference between the (negative) C inputs and (positive) C outputs of the soil (Ojanen et al., 2012), net accumulation of soil C being a negative balance (i.e. atmospheric view). Soil C inputs were dependent on plant production and consisted of the C incorporation from aboveground litterfall, belowground root litter and moss-layer litter production.
Sitka spruce and lodgepole pine were the only vascular plants growing in the study sites (except site P23 which had some scattered plants of Vaccinium myrtillus), therefore, aboveground vascular-plant litterfall consisted of tree litterfall only. Given the low abundance of ground vegetation including mosses (see site description), soil C inputs from moss turnover were not included in the study. By contrast, C outputs consisted of CO 2 emissions produced as a consequence of the combined aerobic decomposition of litter and peat. Therefore, and after omitting C outputs as CH 4 fluxes and as off-site C losses in water, and C inputs from moss litter, the soil C balance was calculated as: where ΔC soil is the soil C balance, Litter AG is the aboveground tree litterfall and Litter BG is the belowground fine-root litter. All components in Equation (1) were expressed in g C m −2 year −1 . In addition, at multiyear sites, single calculated mean values of each component were used in each site balance. Soil C EFs were reported in t CO 2 -C ha −1 year −1 .

| Soil heterotrophic respiration
Between January and February 2014, seven measurement points were identified within each of the eight study sites. A stratification    (2) and (3) were the best models to simulate hourly R H effluxes at site S38 and at all the other sites respectively, where y is the measured soil CO 2 efflux rate, T is the measured soil temperature at 10 cm depth, a i and b i are fitted parameters greater than 0 obtained by nonlinear regression analysis, WTD is the measured water table level, c i , d i and WTD i are specific-fitted parameters determined using least squares nonlinear regression (Table S1)

| Aboveground litterfall
At each site, seven circular litterfall traps (0.08 m 2 ), installed next to each soil respiration subsite, were used to collect Litter AG . Litterfall traps consisted of PVC containers (27 cm high) perforated at the bottom to allow free drainage of water. To prevent accumulation of water TA B L E 2 General features of the Sitka spruce (S) and lodgepole pine (P) sites. Description includes stand basal area (ba) and soil carbon stocks (SCS) Results were scaled up to site level and monthly and annual Litter AG were calculated for each site. Based on (Reidy & Bolger, 2013), the C content of aboveground litterfall was assumed to be 47.57%.

| Root litter inputs
In July 2015, seven soil peat cores, including the litter (recognizable leaves, twigs and small branches; L) and fermentation (a mix of partially decomposed organic matter, roots and fungi; F) layers were taken within each study site. Samples were taken following the same stratification design used for the R H measurements.
Therefore, three cores were taken in the flat area, two in the furrow microtopography and two in the tree-planting lines. Continuous volumetric soil cores (internal dimensions of 7.0 × 7.5 × 80 cm) were collected, using a modification of the volumetric peat sampler proposed by Jeglum et al. (1991). Prior to sampling, the sampler was gently pushed into the soil. Then, using a sharp knife, soil and roots were pre-cut through the top 20 cm. Thereafter, the sampler was completely pushed into the soil. Using a lever and a tripod, the sampler was extracted from the soil, and the soil core (containing the roots) was then divided into ten segments at the following depths: 0.0-2.5, 2.5-5.0, 5.0-7.5, 7.5-10.0, 10.0-12.5, 12.5-15.0, 15-20, 20-30, 30-50 and 50-80 cm. Each soil segment was sealed in a plastic bag and kept at −18°C until processed. Living roots were manually sorted and fine roots <2 mm were washed to remove soil. Roots were then oven-dried for 72 hours at 65°C and weighed to calculate fine-root dry biomass. Finally, results were scaled up to site level.
Belowground litter inputs were estimated from the measured root biomass and a mean root turnover rate of 0.33 (range of 0.20-0.65). This turnover rate was derived from a Sitka spruce chronosequence (five sites of ages varying between 12 and 41 years old) growing on organo-mineral soils within the same area as the present study (Lane, 2016). Lane's study, a split tube sampler of 4.8 cm diameter was used to collect 20 cm long soil cores to measure fine-root biomass (< 2 mm). In addition, and based on the method in the furrow and 14 in the ribbon locations). After digging out the meshes (between 12 and 18 months after inserting them), all roots that had grown through the mesh were trimmed to 1 cm (on each side of the mesh) and all fine roots ≤2 mm were collected to quantify fine-root production. The root turnover rate was calculated as the ratio between annual fine-root production in one year and total root biomass. Although the soil types differ, this turnover ratio constitutes the best available estimate for Sitka spruce under the same climatic conditions. The same root turnover rate was used to estimate root litter inputs at the lodgepole pine site. Finally, total belowground C inputs were calculated for each site by multiplying the root litter production by a C content of 46.58% (Olajuyigbe et al., 2012). Site S39 was clearfelled in June 2015 before the collection of the peat cores. Therefore, an average of fine-root biomass from the other six Sitka spruce sites was used in the soil C balance.

| Statistics and uncertainty
It was assumed that all component fluxes were independent variables.
Therefore, the 95% confidence intervals and the uncertainty of the

| Soil heterotrophic respiration
Heterotrophic respiration was closely related to soil temperature and followed a seasonal pattern with minima in late winter and emission rates increasing gradually into late summer (Figure 2). Similarly, the lowest and highest CO 2 emissions were observed during the late winter and late summer months respectively (Figure 3). Across the (4 Figure S1). In addition, among the Sitka spruce sites, mean hourly R H was lowest at S43 (p < 0.001) and highest in S27 but it was not significantly different than hourly R H rates in sites S18, S27 and S39 ( Figure S1). Although S18 and S28 did not show any difference in R H across the three microtopographies, this effect was significantly different in other sites (p < 0.001). Heterotrophic respiration was highest in the furrow microtopography in sites S24 and S43 ( Figure   S2). However, in sites S27, S39, S44 and P23, R H effluxes from the furrow microtopography were not different from those measured in the plough ribbon. Lowest mean R H efflux was measured in the flat microtopography in site P23. Simulated annual R H across all sites ranged between 377 and 679 g C m −2 year −1 (Table 4). Mean annual R H at multi-year sites ranged between 416 ± 14 g C m −2 year −1 , at site S43, and 607 ± 53 g C m −2 year −1 , at site S18. Sites S27 and S39 were clearfelled in June 2015 and therefore annual R H was calculated for a single year only, and was 627 and 580 g C m −2 year −1 respectively.

| Aboveground litterfall
Across the seven Sitka spruce sites, mean monthly litterfall varied between 5 and 139 g m −2 month −1 (Figure 4 and Table 4). These were Therefore, litterfall from those months was included in the consecutive month of sampling and the 2-month collection amount was split in half. Mean annual litterfall at the Sitka spruce sites varied between 160 ± 31 g C m −2 year −1 (at site S43) and 320 ± 38 g C m −2 year −1 (S28). Mean annual litterfall at the lodgepole pine site was 245 ± 16 g C m −2 year −1 .

| Root biomass
Total dry fine-root biomass at the Sitka spruce sites varied between 600.0 ± 73.4 g m −2 at the youngest site, S18, and 1181.0 ± 223.1 g m −2 in site S27. Fine-root biomass at the lodgepole pine site was 496.2 ± 55.4 g m −2 . All study sites had similar vertical fine-root distributions, with over 98%, on average, in the top 20 cm of soil ( Figure 5). All Sitka spruce sites had 100% of the fine roots located in the top 30 cm of the soil profile. The lodgepole pine site had less than 1% of the fine-root biomass in the 30-50-cm depth interval. Root turnover varied between 92 ± 11 g C m −2 year −1 (at site S18) and 182 ± 34 g C m −2 year −1 (at site S27). Root turnover at the pine site was 76 ± 9 g C m −2 year −1 (Table 4).

| Soil carbon balance
Across all sites, the annual soil C balance varied between −86 ± 28 and 422 ± 12.6 g C m −2 year −1 ( Figure 6 and Table 4). In addition, the annual soil C balance at multi-year Sitka spruce sites (not S27 and S39, where only 1 year was used) varied between 63 ± 92 and 309 ± 67 g C m −2 year −1 . The annual soil C balance at the lodgepole pine site was 107 ± 40 g C m −2 year −1 (Figure 7). A mean soil C balance for afforested peatland in maritime temperate conditions was derived from these measurements in Sitka spruce plantations. This soil C balance is equivalent to the soil C EF. Therefore, the calculated EF for this land-use category and tree species was 1.77 ± 0.34 t CO 2 -C ha −1 year −1 . The mean C EF for afforested peatlands, independent of tree species, was 1.68 ± 0.33 t CO 2 -C ha −1 year −1 (Table 5).

| DISCUSS ION
This study represents one of the few soil C balance studies in afforested peatlands in maritime temperate conditions. The combination of soil respiration measurements using chamber methods with the quantification of C inputs into the soil aims to be a practical approach to assess whether these soils are acting as C sources or sinks.
The main challenge of this method is the necessity to estimate all the C components accurately (Ojanen et al., 2012). Although soil C inputs from ground vegetation were not measured at the field sites, the potential implications of such inputs are included here.

| Soil carbon balance
Drainage of blanket peatlands, and planting them with either Sitka spruce or lodgepole pine, resulted in soils being net C sources.
That is, these forested peatland soils emitted more C to the atmosphere than they accumulated. One site had inconclusive results, due to the large uncertainty associated with this site. The 95% confidence interval of the multi-year soil C balance in S24 varied between -117 and 242 g C m −2 year −1 . The annual soil C balance varied across sites and also between the two studied years. F I G U R E 5 Measured (symbols) and simulated (line) cumulative fine-root biomass distribution with depth at the Sitka spruce (S) and lodgepole pine (P) sites. Shaded area represents the 95% confidence interval for the simulated cumulative fine-root distribution (n = 7). Sampling was conducted at the different soil microtopographies (3 samples in the flat and undisturbed area, 2 samples in the furrows and 2 samples in the ribbon locations) Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 6 Annual soil carbon balance (SCB) and its carbon (C) flux components: aboveground litterfall (Litter AG ), fine-root litter (Litter BG ) and heterotrophic respiration (R H ) at the Sitka spruce (S) and lodgepole pine (P) sites. Error bars are standard error of the means. Negative and positive C fluxes represents C input and outputs into and from the soil, respectively. Therefore, a negative soil C balance means that the site is a C sink while a positive balance means that the site is a C source Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 7 Summary of carbon (C) fluxes in drained peatland forests. Soil carbon balance (ΔC soil ), used to develop soil C emission factors (EF), was estimated as the difference between soil CO 2 emissions originating from heterotrophic respiration (R H ), and CO 2 removals from above and belowground litter accumulation (Litter AG and Litter BG , respectively). Range values for all components are expressed in g C m −2 year −1 . Positive values indicate a loss of C to the atmosphere and negative values indicate a C accumulation into the soil. Dissolved organic carbon (DOC) and CH 4 fluxes are reported using their own EFs and therefore, they were not considered in the present study Colour figure can be viewed at wileyonlinelibrary.com] in the peat) and that this CO 2 efflux was maximum when WTD was −66 cm. This would support the present results.
Although no other soil C balance estimates for afforested peatland in the same climate zone exist in the literature, the soil C balance estimates reported in the present study contrast with some studies from boreal peatland forests. Bjarnadottir et al. (2021), using eddy covariance techniques, found that a black cottonwood (Populus balsamifera ssp. trichocarpa) plantation on a shallowdrained peatland was a soil C sink of −55 g C m −2 year −1 . Similarly, Minkkinen et al. (2018), also using eddy covariance methods, reported that a boreal drained peatland forest acted as a soil C sink of −60 g C m −2 year −1 . However, another study at the same site in Finland concluded that, if the same chamber method as the one used in the present study was used instead of the eddy covariance method, the soil C balance would vary between −16 ± 44 and 106 ± 44 g C m −2 year −1 (Ojanen et al., 2012), illustrating the uncertainties in the chamber method. Differences between the boreal and the Irish sites are likely due to differences in mean annual tem- and 810 g C m −2 year −1 depending on the method used (eddy covariance techniques vs. chamber and trenching methods) (Meyer et al., 2013). The same authors attributed this large net soil C efflux to the very high R H emissions measured in this nutrientrich and former agricultural land (i.e. 1300 g C m −2 year −1 ) and to the accumulated uncertainties associated with the chamber method. Similarly, another chronosequence study in a forestrydrained peatland with natural regeneration of pure downy birch (Betula pubescens) found that these sites acted as continuous soil C sources (Uri et al., 2017). This study, based on chamber and trenching methods, reported that the soil C balance varied between 161 and 293 g C m −2 year −1 .
Another reason for the rather contradictory results, between these reported soil C balances could be accounting for the presence of mosses (and other herbaceous vegetation) and the incorporation of this plant litter into the soil C balance. Uri et al. (2017) reported that in their study site, the herbaceous litter input varied between 5 and 19 g C m −2 year −1 . In addition, Laiho et al. (2011) found that, in drained peatland forests, mosses may incorporate significant amounts of C into the soil. Furthermore, Badorek et al. (2011) reported that ground vegetation represented 20-30% of the total gross primary production in a drained peatland forest and Minkkinen et al. (2018) found that, in the same study site, moss litter input represented over 20% of the total litter production (i.e. 90 g  have not been significant as reported by Mäkiranta et al. (2008) in a similar trenching experiment in Finland-although soil moisture content was not measure in this study. Another limitation of the chamber method is that the proposed R H models require continuous measurements of soil T and WTD to simulate annual CO 2 fluxes accurately. While soil temperature is usually recorded at weather stations and it may be estimated accurately over large areas using remote sensing techniques (Xu et al., 2020), continuous WTD data on peatland forests is scarce and the use of remote sensing techniques to simulate WTD time series accurately needs further refinement (Bechtold et al., 2018;Burdun et al., 2020).
Therefore, if continuous WTD is not available or if the uncertainty of simulating WTD gaps between consecutive measurements by linear interpolation is a concern, it is suggested that temperature exponential models are used, instead of combined model of soil temperature and WTD, to scale up R H effluxes at a regional level (Jovani-Sancho et al., 2018).

| Litterfall inputs and uncertainty
In this study, fine-root biomass was greatest in the top 10 cm of soil, decreasing exponentially with soil depth ( Figure 5). This was also reported by Steele et al. (1997)  Sitka spruce stand growing on a peaty gley soil (Zerva et al., 2005), and 264 g m −2 in a 31 year old Sitka spruce stand growing on a mineral gley soil (Saiz et al., 2006). Previously reported fine root (<1 mm) biomass estimates for Norway spruce are in the range 100-1090 g m −2 (Santantonio et al., 1977). The measured fine-root values in our sites are at the higher end of previously reported values for each species. We believe that the large amount of fine roots found in our sites is an adaptation of the tree species that maximizes the uptake of nutrients and oxygen from these waterlogged nutrient-poor soils. The low oxygen at depth, and low fertility of these soils led to shallow root systems, growing almost exclusively in the F and L layers where most of the nutrients are located and recycled. This suggestion would be supported by the previously reported and new values for Sitka spruce, where fineroot biomass increased in the order: blanket peat soils > peaty Gleys > mineral Gleys (Saiz et al., 2007;Zerva et al., 2005).
One of the main limitations of the present study was the lack of site-specific fine-root turnover rates. The fine-root turnover rate of 0.33 (range of 0.20-0.65) reported by Lane (2016) is similar to the fine-root turnover rate (roots <2-mm diameter) reported for Scots pine (Pinus sylvestris L.) in southern Finland (between 0.34 and 0.75) (Finér & Laine, 1998) but lower than fine-roots turnover for spruce and pine forest in Finland (i.e. 0.81 and 0.87, respectively) reported by (Liski et al., 2006). The fine-root turnover rate developed by Lane (2016) is, to the best of our knowledge, the only value for Sitka spruce that exist in the literature. Moreover, this fine-root turnover rate was developed in the same study area, under the same environmental conditions and using a similar chronosequence and experimental design approach (similar ages, species, root diameters and sampling strategy). Therefore, with the current existing data, using this fine-root turnover rate is the best way to estimate belowground litter inputs in the present soil C balance study. Notwithstanding, previous studies in boreal peatlands determined that the selection of lower or higher fine-root turnover rates (measured or reported in the literature) had a significant effect on the overall soil C balance, determining whether the site was a C sink or source (Ojanen et al., 2012(Ojanen et al., , 2014. Literature values of fine-root turnover rates for different species are very scarce but Ojanen et al. (2012) concluded that using the higher turnover rates resulted in the most accurate values when compared with eddy-covariance-estimated soil C balance. In addition, the sampling method used to measure fine roots is susceptible to errors (Byrne & Farrell, 2005) and it is possible that the values reported in this new study could have been overestimated due to misidentification of live roots. Further studies should aim to reduce the total fine-root biomass uncertainty and also to develop turnover rates specifically for the most common tree species (i.e. Sitka spruce and lodgepole pine) growing on afforested peatlands.
This will help reducing the overall uncertainty of the soil C balance estimate and improve the accuracy of the soil EF factors used to report CO 2 emissions at the national and international level from this land-use category.

| Soil C emissions factors and implications for National Inventory
Although the IPCC (2014) (Table 5). Nevertheless, this new EF is almost three times greater than the current CO 2 -C EF used in the Ireland National Inventory Reporting system (0.59 t CO 2 -C ha −1 year −1 ) (Duffy et al., 2021). The CO 2 -C EF used by Duffy et al. (2021) was estimated from annual total soil respiration (which includes root respiration) values that were measured by the soda-lime method (Byrne & Farrell, 2005). It is known that the soda-lime and alkali traps methods may underestimate high soil CO 2 effluxes and their accuracy and performance remain uncertain (Janssens et al., 2000;Rochette et al., 1992;Rochette & Hutchinson, 2005 GHG National Inventory Report, we estimate that on-site CO 2 -C emissions from afforested organic soils in Ireland would be between 1.8 and 3.9 times greater (i.e. 271,159-604,894 t CO 2 -C) than the current reported emissions but also around 35 and 16% lower than CO 2 -C emissions estimated with the IPCC (2014) default Tier 1 and UK Tier 2 values (Table 6).

| Implications for sustainable land management
These plantation forests on drained blanket peatland are net soil C emission sources, in this maturing phase. With assumed greater emissions in establishment and post-harvest phases (Hargreaves et al., 2003), the soil-based assessment of C balance used under the UNFCCC for these managed ecosystems is consistently one of loss of C to the atmosphere as CO 2 . Because of losses from decomposing peat due to oxygen entry following drainage, above-and below-ground C input is too small to give net soil C sequestration. This status demonstrated here suggests that afforestation of blanket peat, with necessary installation of drains, is unsuited as a strategy to fix atmospheric C. When considering the C balance of these forests, soil C storage is crucial to maintaining and increasing any C storage potential. While the tree layer will sequester C, the merchantable fraction of stemwood will be removed and converted into wood products with the harvest residues remaining on site where they will decompose and return accumulated CO 2 to the atmosphere. Root litter and moss litter (coarse-root and moss litters not evaluated here) will contribute a proportion of organic C to soil, and are contributions to C balance requiring further quantification.
Design of measures to reduce emissions, such as rewetting, should not be limited to site level consideration but should also include TA B L E 6 Annual CO 2 emissions from afforested peatlands in Ireland (IE), the United Kingdom (UK) and each UK administration calculated using IPCC Tier 1 emission factors from the IPCC (2014)  potential impacts on other ecosystem services at landscape and regional scale.

| CON CLUS ION
This study confirms that drained afforested blanket peatland soils are net sources of CO 2 emissions from oxidation of soil C. In addition, the new data from eight sites will significantly contribute to the development of more reliable, and less-uncertain, IPCC Tier 1 CO 2 -C default values for afforested peatlands in the maritime temperate climate zone. However, further studies are necessary to reduce the uncertainty associated with these findings such as belowground C allocation and interannual variation over several years. In addition studies should be extended to include regional differences in forested blanket peatland, different peat (nutrient poor vs. nutrient rich) and forest types, as well the soil C balance in second and subsequent rotation sites. Studies are also required to assess the feasibility and C balance of alternative management systems such as rewetting and low impact silvicultural systems and to consider their impact on ecosystem services at site to landscape scale.

ACK N OWLED G EM ENTS
We