Peatlands cover approximately 3% of the global land area and store about 679 Gt of carbon (C) (Leifeld and Menichetti, 2018; Xu et al., 2018). Peat is a C deposit from decayed vegetation formed under anaerobic conditions during the postglacial period when C uptake exceeded C losses from peatland ecosystems. Most pristine peatlands act as carbon dioxide (CO₂) sinks, emit small amounts of nitrous oxide (N₂O; Minkkinen et al., 2020) and are sources of methane (CH₄; Frolking and Roulet, 2007). The high global warming potential (GWP) of CH₄ (27 times more efficient than CO₂ over a 100-year period, GWP₁₀₀, Foster et al., 2021) makes it that some peatlands have a net warming effect on the climate (Frolking et al., 2006; Koehler et al., 2011). However, due to the relatively short lifetime of CH₄ in the atmosphere, most pristine peatlands act as overall net C sinks, sequestering approximately 13–22 g C m⁻² yr⁻¹ (Yu et al., 2010). Furthermore, peatland’s CO₂ uptake exceeds the emitted CH₄ in terms of CO₂ equivalents (CO_2EQ) (Lai, 2009; Gallego-Sala et al., 2018). When drained for agricultural use, the decomposition of organic matter (peat) is accelerated in the aerated peat and the ecosystem begins to lose C in the form of CO₂. The rate of CO₂ lost to the atmosphere is related to temperature and the water table depth, but at the same time, the aerated peat exhibits reduced CH₄ emissions or turns even into a CH₄ sink (Maljanen et al., 2010; 2004; Ballantyne et al., 2014; Knox et al., 2015; Evans et al., 2021). Agricultural peatlands, especially when ploughed and fertilized, can further accelerate decomposition rates and increase CO₂ emissions due to the increased oxygen and nitrogen (N) availability that may lead to a higher soil microbial activity (Maljanen et al., 2007; 2004; Lund et al., 2009). Nitrogen fertilization increases N availability not only for plants but also for microorganisms. Under anaerobic conditions, denitrifiers can reduce nitrate via nitrite and nitric oxide to N₂O during denitrification, serving as the main source of atmospheric N₂O (Ravishankara et al., 2009; Kuypers et al., 2018). This process often takes place in oxygen-deficient aggregates in the soil, referred to as “hot-spots” for N₂O production and at sporadic “hot-moments” in time when the conditions are favourable (Kuzyakov and Blagodatskaya, 2015). Ammonia-oxidizing microorganisms can perform a similar process under aerobic conditions, which is known as “nitrifier denitrification” (Kuypers et al., 2018). Thus, the use of N fertilizer can increases N₂O emissions from agricultural ecosystems (Davidson, 2009).

Globally, GHG emissions from drained peatlands are high and the potential for anthropogenic emission reduction by peatland restoration (e.g., re-wetting) is estimated to 1.9 Gt CO_2EQ (Leifeld and Menichetti, 2018). Even though only a small fraction of Nordic agriculture is located on drained peatland, this fraction strongly contributes to national GHG emissions. About 10% of the agricultural land in Sweden is located on peat and gyttja soils (gyttja: mineral matrix enriched by humic deposits, see Holstad and Degago, 2021) and is responsible for about 6%–8% of the national GHG emissions (Berglund and Berglund, 2010). In Finland, this area is 15% of the total agricultural land, which leads to about 8% of national GHG emissions (Kasimir-Klemedtsson et al., 1997; Maljanen et al., 2004). Politicians are therefore well advised to incentivize practices for reducing GHG emissions from agricultural peatlands if they endeavour to meet the goals defined in the Paris Agreements. However, since many drained peatlands are productive agricultural soils (e.g., Richardson and Smith, 1977; Kramer and Shabman, 1993), a balanced compromise between the degree of environmental protection, or the livelihood of future generations, and short-term economic profitability needs to be made. Until today, incentives on EU and member state level are in place for supporting peatland drainage, e.g., direct payments to farmers for setting aside drained peatlands as grassland (Chen et al., 2023; EU Regulation No 1307/2013, 2013). Some may favour an intensification of agricultural production on a limited land area to enable setting-aside farmland (i.e., land-sparing) for environmental protection purposes and eventually reducing the GHG emissions from agriculture (e.g., Burney et al., 2010; Tilman et al., 2011; Wang et al., 2018). In a Canadian study, Wang et al. (2018) found that C sequestration of a pristine bog was lower than of an adjacent abandoned pasture on the same peatland complex in Newfoundland and relate this difference to contrasting above ground biomass. The authors tentatively suggested that abandoning peatland pastures could be a method to increase C sequestration in peatland ecosystems. However, this may not be applicable to many peatlands. Several studies have investigated how different management intensities including setting-aside agricultural land influence GHG fluxes from Nordic peatlands, with variable results (Maljanen et al., 2010; 2007; Hadden and Grelle, 2017; Wang et al., 2018; Berglund et al., 2021; Evans et al., 2021). Berglund et al. (2021) and Hadden and Grelle (2017), however, showed that a set-aside grassland on drained peat had higher respiratory net CO₂ losses than an adjacent cultivated peat in central Sweden. Hadden and Grelle (2017) measured CO₂ fluxes from a cultivated cereal cropping system and an adjacent set-aside grassland on shallow peat over a period of 4 years. They concluded that the set-aside grassland acted as a small net CO₂ source, whereas the cropland acted as a small net sink. Berglund et al. (2021) measured CO₂, CH₄, and N₂O fluxes at the same site using manual and automatic chambers and found that the set-aside grassland acted as a stronger GHG source than the cropland. Maljanen et al. (2010) concluded in their review that annual CO₂ and N₂O fluxes from abandoned croplands on peat were similarly high as those from cultivated peatlands in the Nordic countries. Maljanen et al. (2007) came to a similar conclusion when comparing the GHG fluxes from several sites of cultivated and abandoned peatlands in Finland. Evans et al. (2021) found that the effect of peatland management was negligible compared to the importance of the effective water table depth for C emissions from peatland ecosystems.

In this study, we build upon the work of Hadden and Grelle (2017) and Berglund et al. (2021) with a more extensive data-set including more than 6 years of eddy covariance (EC) CO₂ flux data and combine it with 3 years of manual chamber measurements of CO₂, N₂O, and CH₄ fluxes and investigate the GHG flux dynamics at two adjacent sites, a cultivated (CL) peatland and a set-aside (SA) grassland on peat. We apply modern EC data processing protocols, comparing our results with those of Hadden and Grelle (2017) and extend on their work by comparing EC and chamber-based CO₂ flux measurements. We hypothesize that (i) different management has no effect on the cumulative CO₂ fluxes from either site, (ii) the CL site is a larger source of N₂O and (iii) a stronger sink for CH₄.

The annual CO₂ emissions reported in this study are of similar magnitude to other shallow boreal drained peatlands. Lohila et al. (2004) measured CO₂ fluxes over the course of 2 years by the EC method above an agricultural peatland in Finland. They found annual CO₂ losses of 0.77 and 0.29 kg CO₂ m⁻² yr⁻¹ for the first year when barley was grown and the second year when grass was cultivated, respectively. Thus, the cumulative annual emissions were above the ones found for the CL and SA sites (0.16 and 0.41 kg CO₂ m^-2 yr⁻¹, respectively). However, they originated from a slightly thicker peat (40–60 cm) and therefore, a larger volume of peat was exposed to oxidative conditions and by that may have influenced the flux magnitudes. Studies investigating CO₂ emissions from Nordic peatlands with thicker peat layers usually report larger CO₂ emissions (e.g., Maljanen et al., 2001; Maljanen et al., 2004; Maljanen et al., 2007; Maljanen et al., 2010). Above a peatland in Finland with a shallow peat layer (15–30 cm), Gerin et al. (2023) measured GHG fluxes by EC for two consecutive years and estimated the annual CO₂ emissions as NEE to 0.32 and 0.74 kg CO₂ m⁻² yr⁻¹, for the first and second year, respectively. This site was used for grass-silage production interchangeably with cereal cultivation and the management included bi-annual fertilisation and grass cuts, as well as the application of herbicides. Thus, the difference in management may explain the larger emissions compared to our sites. However, Evans et al. (2021) found a strong effect of the effective water table (the aerated peat thickness) on the CO₂ emissions from 16 peatland sites under different management intensities in the United Kingdom and Ireland, with a negligible effect of management intensity. The aerated peat thickness at the CL and SA site are defined by the local peat thickness, because the water table lies well below the peat layer during the growing season (<150 cm). Thus, the slightly shallower peat at the CL (25 cm) compared to the SA site (34 cm) may have contributed to the observed differences between the CO₂ flux magnitudes. Management intensity may play a minor role in controlling CO₂ emissions from drained agricultural peatlands. Maljanen et al. (2007) found by comparing GHG fluxes from five Finnish sites on drained peat, that the mean annual CO₂ emissions of abandoned organic agricultural soils were similar to those of active croplands. However, some other studies have not found such a clear relationship between peat depth and CO₂ emissions and assign a higher importance to the influence of management activities (e.g., Elsgaard et al., 2012; Tiemeyer et al., 2016; Wang et al., 2018). Under the assumption that aerated peat thickness is the sole driver of CO₂ emissions, we estimated the effect of the different peat thickness at our sites on the difference in CO₂ emissions. This was done by using the linear relationship between aerated peat thickness and CO₂ emissions that was found by Evans et al. (2021) and estimated that the 9 cm thicker peat at the SA site may lead to higher CO₂ emissions by about 0.31 kg CO₂ m⁻² yr⁻¹.

Hadden and Grelle (2017) investigated the NEE of the same site from July 2012 to November 2016 and found that the CL site acted as a small CO₂ sink (−200 g m⁻² CO₂) and the SA site as a CO₂ source (200 g m⁻² CO₂) over the measurement period. Whereas, this study found that both sites were sources of CO₂ with the SA emitting more CO₂ annually (0.41 kg CO₂ m⁻² yr⁻¹) than the CL site (0.16 kg CO₂ m^-2 yr⁻¹). This difference originated mainly from the different data post-processing procedures, in particular the exclusion of unrealistic data, the ustar filtering, and gap-filling. Hadden and Grelle (2017) excluded data that originated from periods of known instrument failures or rime or ice formation on the sonic anemometers and removed unrealistic outliers. In this study, we used more rigorous outlier detection (i.e., absolute deviation about the median), ustar filtering (Papale et al., 2006) and excluded negative night-time fluxes. For estimating the cumulative CO₂ fluxes, the data gaps by Hadden and Grelle (2017) were filled by linear interpolation (30-min gaps) or by the mean diurnal course method (Falge et al., 2001). Whereas, in this study, the marginal distribution sampling method by Reichstein et al. (2005) was used. Additionally, some differences between our findings and those of Hadden and Grelle (2017) may be attributed to the difference in the measurement period length (Hadden and Grelle (2017): mid-2012 to end-2016; this study: beginning-2013 to mid-2019) and a change in management in the years after 2016. As at the CL site, ploughing was substituted by carrier disk management, and at the SA grassland, a change in management occurred in 2015 already, when the farmer started to cut the grass once a year.

To estimate the C exports from the CL site, annual potential yield of 0.2 kg m⁻² yr⁻¹ were assumed for all years (Jan-Erik Olsson, pers. comm.) except of 2016 when the field was fallowed and 2018, when barley was grown and harvested, yielding 0.37 kg m⁻². Totalling to a potential yield of 1.17 kg m⁻² for the years 2013–2019. During autumn and winter, wild boars, deer, moose, and birds were foraging on the crop and reducing the estimated potential yields by approximately 80% for all years except 2016 and 2018. Assuming a water content of 15% (w/w) and roughly estimating that about half of the biomass that was consumed by wildlife remained on site with excreta and by applying a biomass-to-C conversion ratio of 0.45 (Hicke et al., 2004) we come to a total C export by harvest and herbivory (C_herb/harv) of 0.26 kg C m⁻² for the years 2013–2019. Thus, in terms of CO₂ an annual export of about 0.16 kg CO₂ m⁻² yr⁻¹. At the SA site, the estimate of C exports by Hadden and Grelle (2017) of 0.03 kg C m⁻² yr⁻¹ was used. In 2018, 0.123 kg C m^-2 was exported with the harvest. Therefore, the C export from the SA site summed up to an annual C loss in terms of CO₂ of 0.167 kg CO₂ m⁻² yr⁻¹. The annual CO₂ loss as the sum of NEE and C_herb/harv from the CL site resulted in 0.32 kg CO₂ m⁻² yr⁻¹ and for the SA site (C_herb/harv = 0), to 0.58 kg CO₂ m⁻² yr⁻¹. By considering the effect of C_herb/harv the CO₂ balance of the CL site remains lower than that of the SA site. This can only be used as an indication for the sites C losses, since the C_herb/harv in this study is not based on measured data of C exports, but on assumptions.

The CO₂ fluxes measured by the manual chambers support our findings by EC of larger fluxes from the SA site compared to the CL site. Other studies on drained boreal peatlands found similar results. In accordance with Berglund et al. (2021), we observed the highest average CO₂ uptake, the highest R_ECO and R_SOIL on the SA site and Maljanen et al. (2004) reported larger NEE fluxes from grass-covered plots than from cereal-covered plots on two drained boreal peatlands. The often observed relatively high N₂O emissions from agricultural peatlands (Maljanen et al., 2003b; Maljanen et al., 2004; Gernin et al., 2023), were not confirmed at our study site. This can be explained by low nitrogen fertiliser input at the CL site and the complete lack of nitrogen fertilisation at the SA site. Both sites were small sinks for CH₄, as is commonly reported for drained agricultural peatlands (e.g., Maljanen et al., 2003a; Maljanen et al., 2004; Hadden and Grelle, 2017; Berglund et al., 2021). A higher CH₄ uptake at the SA site was observed. The same pattern was previously found by Maljanen et al. (2007) on abandoned peatlands in Finland and by Berglund et al. (2021) at the same study site that is subject to this article. The GHG balance for the CL and the bare peat at the SA site are comparable to those by Berglund et al. (2021). However, due to the higher emissions from our vegetated base rings of the SA site our GHG budget from the opaque and vegetated chambers exceeded those reported by Berglund et al. (2021).

For a complete estimate of the local GHG balance, continuous and year round measurements of all three GHGs, systematic measurements of C removed with biomass, estimates of N leached to the ground water, and early season and late season drainage ditch emissions should be considered. Moreover, investigating the impact of management practices such as rewetting or paludiculture on the local GHG balance is essential for understanding the GHG emission reduction potential of the sites.

On the basis of our measurements and in accordance with other recent literature (e.g., Maljanen et al., 2007; Hadden and Grelle, 2017; Berglund et al., 2021; Evans et al., 2021), we found no evidence that setting-aside agricultural peatlands will reduce CO₂ emissions or even turn the peatlands into a C sink. Thus, the political debate about setting-aside agricultural peatlands is in need of reassessment, if the aim is to reduce GHG emissions and agricultural policies that support setting-aside or fallowing agricultural peatlands without rewetting (EU Regulation No 1307/2013, 2013) are likely ineffective means for climate change mitigation.

Based on more than 6 years of cumulative eddy covariance CO₂ flux data, we found no evidence that setting-aside cropland reduced greenhouse gas emissions. The set-aside grassland was a larger source of CO₂ (0.41 kg CO₂ m⁻² yr⁻¹) than the adjacent cropland (0.16 kg CO₂ m⁻² yr⁻¹), both located on a shallow drained peat in central Sweden. Respiratory CO₂ losses, photosynthetic CO₂ assimilation and CH₄ uptake were higher at the set-aside grassland than at the cropland. Nitrous oxide emissions from the cropland exceeded those from the grassland, but this had a minor effect on the sites global warming potential as estimated in this study.