Rewetting strategies to reduce nitrous oxide emissions from European peatlands

Nitrous oxide (N2O) is approximately 265 times more potent than carbon dioxide (CO2) in atmospheric warming. Degraded peatlands are important sources of N2O. The more a peat soil is degraded, the higher the N2O-N emissions from peat. In this study, soil bulk density was used as a proxy for peat degradation to predict N2O-N emissions. Here we report that the annual N2O-N emissions from European managed peatlands (EU-28) sum up to approximately 145 Gg N year−1. From the viewpoint of greenhouse gas emissions, highly degraded agriculturally used peatlands should be rewetted first to optimally reduce cumulative N2O-N emissions. Compared to a business-as-usual scenario (no peatland rewetting), rewetting of all drained European peatlands until 2050 using the suggested strategy reduces the cumulative N2O-N emissions by 70%. In conclusion, the status of peat degradation should be made a pivotal criterion in prioritising peatlands for restoration. Rewetting agricultural peatlands first is the best strategy for reducing cumulative nitrous oxide emissions from European peatlands, according to an analysis of soil bulk density as a proxy for peat degradation.

A griculture contributes to one-quarter of the worldwide greenhouse gas (GHG) emissions 1,2 . Nitrous oxide (N 2 O) is classified as a long-lived GHG and has a global warming potential of~265 times 3 that of carbon dioxide (CO 2 ). It is also the main driver of stratospheric ozone depletion 4 . Nitrous oxide emissions, accounting for 46% of GHG emissions from agricultural soils, largely come from soil and nutrient management such as tillage and fertilizer application 5,6 .
Peatlands cover only about 3% of the global land surface but store 21% of the global soil C pool 7 and 8-15 Gt N 8,9 . To date, at least 15% of the world's peatlands have been artificially drained for agriculture, forestry, peat extraction, and bioenergy plantations 10 . The drained peatlands are mostly located in Europe and South-east Asia 11 . Peatland drainage causes land subsidence and carbon mineralization leading to soil degradation, GHG emissions (e.g., CO 2 , N 2 O), and dissolved organic carbon (DOC) leaching into downstream water bodies 12,13 . It has been reported that northern peatlands may emit 30-100 Gg N 2 O-N year −[1 14 . The world's drained peatlands cumulatively release 2.3 Gt N 8 .
The N 2 O production in soils originates mainly from nitrification and denitrification processes 15 . The N 2 O emissions from peat are closely linked to peat type, water management, and climate zones 14,16,17 . In natural undisturbed peatlands, the N 2 O emissions are generally low due to the low oxygen and/or nitrogen availability 18,19 . Peatland drainage increases N 2 O emissions by enhancing the oxygen and nitrogen availability 16,18,20 . It has been reported that the well-drained and nitrogen-rich tropical peatlands are global N 2 O emission hotspots 16 . The N 2 O emissions from drained peatlands vary greatly because of nutrient content variations and land management. For instance, lowering the water table had no effect on N 2 O emissions from nutrientpoor but increased those from nutrient-rich peatland 14 . Furthermore, several studies 19,21,22 found that N 2 O emissions from cropland and grassland are generally higher than those from the forest (natural peatlands drained for forestry). However, the conversion of agriculture to forest leads to no significant reduction in N 2 O emissions as compared to peat soils under active agricultural use 23 . Lastly, nitrogen, as well as phosphorus fertilizer applications, may additionally enhance N 2 O emissions from agricultural peatlands 24,25 .
Monitoring N 2 O emissions from peatland is time-consuming and expensive 23,26 . Therefore, numerous simulation models and statistical relationships have been developed to predict N 2 O emissions at multiple spatial scales 21,27,28 . In the national GHG inventories, the published IPCC default emission factors (Tier 1) have been used to estimate N 2 O emissions from peatlands. Several studies reported that vegetation and soil properties (C/N; bulk density, BD) are also good proxies to estimate N 2 O emissions at the field or national scales 18,21,29,30 . The C/N ratio decreases and BD increases along with soil degradation 30 . The more a peatland is degraded, the higher the N 2 O fluxes 30 . Soil BD as a proxy for peat degradation was superior to other parameters (C/N, pH) in estimating annual N 2 O emissions in a previous study 30 because it is an integrating parameter reflecting both physical and biogeochemical transformation processes.
In Europe, <1% of the drained peatland has been rewetted over the past decades 31 . Peatland rewetting is an effective measure to rehabilitate ecosystem functions and can reduce soil subsidence and greenhouse gas emissions (CO 2 and N 2 O) [32][33][34][35] . However, little information is available on how to best prioritize drained peatlands for rewetting, to maximize the reduction of N 2 O emissions. The objectives of this study were to (1) re-estimate the N 2 O emissions from European drained peatlands (EU-28, 2013) using a newly generated soil bulk density map; (2) predict the N 2 O emission for several decades under scenarios with different rewetting priorities.
Results N 2 O-N emissions from managed European peatlands. The managed European peatlands are characterized by a wide range of topsoil (0-30 cm) bulk density (BD) from 0.1 to 0.9 g cm −3 (Supplemental Fig. 1). The average (10th percentile, 90th percentile) topsoil BD of cropland, grassland, and forest was 0.7 (0.5, 0.8), 0.6 (0.4, 0.8), and 0.5 (0.2, 0.8) g cm −3 , respectively. This finding indicates that peatlands under cropland are more severely degraded than those under forest, which is most likely related to the deep drainage of cropland-peat-systems. The estimated average N 2 O-N emission factors using BD for cropland, grassland, and forest were 19.3, 17.4, and 3.4 kg N ha −1 year −1 , respectively (Table 1), which are greater than the default values from IPCC (Supplemental Table 1). The N 2 O-N emission factor for a forest is significantly lower than those for agricultural peatlands. The 95% confidence intervals (Table 1) suggest that there is no significant difference in N 2 O-N emission factors between cropland and grassland.
The estimated N 2 O-N emissions from European managed peatlands under different land uses were calculated to be 47.1, 61.4, and 36.6 Gg N year −1 , for cropland, grassland, and forest, respectively (Table 1). In this study, the overall N 2 O-N emissions from European (EU-28) managed peatlands were estimated to be 145.1 Gg N, which is twice the number estimated by the IPCC approach. The N 2 O-N emission hotspots (15-21 kg N ha −1 year −1 ) are located in Ireland, Sweden, Poland, Germany, and The Netherlands ( Fig. 1), where the organic soils are extensively drained.  The parentheses indicate 95% confidence intervals.
95% confidence intervals (Fig. 2a, b) indicate no distinct differences in N 2 O-N emissions between scenarios 1 and 2 (no change scenario). Under both scenarios, the accumulated N 2 O-N emissions over 30 years will add up to~4500 Gg (95% confidence interval from 3000 to 7000 Gg). If all drained peatlands were rewetted over the coming 30 years with an annual water table at the ground surface, the annual N 2 O-N emission will decrease to 0 kg N ha −1 year −1 . However, the accumulated N 2 O-N emissions under scenario 4 (starting to rewet the highly degraded peatlands and rewetting agricultural peatlands first) will be substantially lower (1229 Gg N) than those under scenario 3 (3267 Gg N; starting to rewet from low to high degradation status and rewetting the forested peatlands first). Compared to the nochange scenario, cumulative N 2 O-N emissions will decrease by 30% under scenario 3. However, the reduction will be 70% under scenario 4 after 30 years.

Discussion
It has been reported that IPCC underestimated the average N 2 O-N emission factors for managed peatlands 19,36 . In this study, the estimated N 2 O-N emission factor for croplands is similar to the values reported by previous studies ranging from 16  organic soils are strongly disturbed and extensively drained. The estimates as presented here take soil degradation (soil BD) explicitly into account and are, thus, considered superior to IPCC default emission factors.
Here, we estimated the effect of different peatland management scenarios on future N 2 O-N emissions. It is surprisingly found that if the area of artificially drained peatland is expanded by 30%, the N 2 O-N emissions will not significantly increase in the near future (<30 years). The estimated N 2 O-N emissions from these newly drained peatlands are comparable to natural and undrained peatlands (0.01 to 1.6 kg N ha −1 year −1 26,40 ). One possible reason is that the carbon mineralization rate is lower at the early stage of peatland drainage 41 , and a relatively large soil C/N ratio is maintained, which constrains N 2 O emissions from peat 18,42 .
The effect of rewetting measures on N 2 O-N emissions is related to the groundwater table height 30,43 . Despite a huge variability in N 2 O-N emissions, it is very clear that the emissions are approaching 0, if the average annual water table is near or above the ground surface (Fig. 3). For rewetted and degraded peatlands with an annual water table of 10-30 cm below ground surface, the topsoils suffer both aerobic and anaerobic conditions, allowing N 2 O-N emissions from both nitrification and denitrification processes 30 . Therefore, if the peatlands are not properly rewetted, the accumulated N 2 O-N emissions from European managed peatlands remain comparable to scenarios 1 and 2.  It is necessary to estimate the cumulative GHG emissions of CO 2 , CH 4 , and N 2 O, if peatland restoration strategies are to be evaluated because the gases factually accumulate in the atmosphere 37,44 . Figure 2 suggests that peatland rewetting starting from highly to lowly degraded peatlands and from agricultural to forested peatland is the most effective strategy to reduce the cumulative N 2 O-N emissions. This strategy is likewise supposed to be effective in reducing cumulative CO 2 emissions because the strongly disturbed and extensively drained agricultural peatlands, recognized as highly degraded peatlands, are also emitting CO 2 at high rates 25,45 . Several studies reported that rewetting of highly degraded peatlands is a major challenge especially for the biodiversity targets and suggest lightly degraded peatlands should be prioritized for rewetting from an economic and biological viewpoint 32,46 . However, most managed peatlands in Europe are in a high degradation stage and they emit most of the GHG of all peatlands. Postponing rewetting of these highly degraded soils may increase the long-term warming effect through continued GHG emissions. The most effective restoration strategy should be further evaluated in future studies in an interdisciplinary approach.
To our knowledge, this is the first time that a bulk density map for soils was generated and used to estimate the N 2 O-N emissions from managed European peatlands. Our work is in line with studies showing that the drained peatlands in Europe are in a stage of severe degradation and suggests that they emit considerable amounts of N 2 O-N. Less than 6 × 10 6 ha of agriculturally used peatlands (cropland and grassland) emit over 100 Gg N 2 O-N year −1 , equivalent to 25 % of soil N 2 O emissions from the entire agriculture land of EU-28 47 . In this study, the estimated N 2 O-N emissions from European managed peatlands are twice the value of those calculated using IPCC default emission factors. The results suggest that the best restoration strategy to reduce cumulative N 2 O-N emissions is to rewet highly degraded peatlands first before attending to those with a low degradation status. Such a strategy would result in a cumulative reduction of N 2 O-N emissions from European peatlands by 70% over the next 30 years as compared to business as usual. Only a 30% reduction will be achieved if rewetting is initiated at less degraded peatlands first. This study provides a new perspective on how to prioritize peatland restoration strategies. However, the quality of the BD map needs to be improved and more data is required on N 2 O-N emissions from highly degraded peatlands to reduce prediction uncertainty.  Table 2). The topsoil OM map of Croatia was derived from the topsoil organic carbon map 51 by applying a factor of 1.72.

Methods
The N 2 O-N emissions from managed European peatlands were estimated based on the derived BD map for peatlands and the relationship between soil BD and annual N 2 O-N emissions (Fig. 4 30 ). The data set from Liu et al. 30 shows that most of the high N 2 O-N emission values (>10 kg N ha −1 year −1 ) for forested peatlands originate from afforested formerly agriculturally used peatland 23 . For forested peatlands without any cultivation history, the N 2 O-N emissions are generally <7 kg N ha −1 year −1 . Therefore, we apply different BD-N 2 O-N functions for agricultural and forested peatlands (excluding afforested agricultural peatlands). The N 2 O-N emissions from afforested peatlands on former agricultural sites are comparable to those from agriculture before conversion; this rare situation was excluded from the study. The relation between soil BD and annual N 2 O-N emissions did not differ between grassland and cropland. Therefore, we assume that N 2 O-N emissions from grassland and cropland follow the same function with soil BD. For peat soils with a BD of 0.6 g cm −3 , N 2 O-N emissions from cropland, grassland, and forested peatland were 20.8, 20.8, and 5.1 kg ha −1 year −1 , respectively (Fig. 4). Little information is available on N 2 O-N emissions from peat soils with BD > 0.6 g cm -3 , therefore, for these soils we set a fixed value of 20.9 (14.3, 29.5), 20.99 (14.3, 29.5), and 5.1 (2.8, 9.4) kg N ha −1 year −1 for cropland, grassland, and forested peatland, respectively. We used the standard errors of the coefficient estimates from the statistical models to calculate the 95% confidence intervals.
Default emission factors from IPCC 1 were also applied to compare with the results using BD functions. The average N 2 O-N emission factor (95% confidence interval) for boreal and temperate cropland was set to 13.0 (8.2, 18) kg N ha −1 year −1 . The average N 2 O-N emission factor (95% confidence interval) for boreal and temperate grassland on peatland was set to 9.5 (4.6, 14) and 8.2 (4.9, 11) kg N ha −1 year −1 , respectively. The average N 2 O-N emission factor (95% confidence interval) for boreal and temperate forested peatlands was set to 0.22 (0.15, 0.28) and 2.8 (−0.57, 6.1) kg N ha −1 year −1 , respectively. Nutrient and drainage conditions are not available in a spatially explicit way; therefore, we assumed nutrient-poor conditions for boreal forests and nutrient-rich conditions for temperate forests. We also assumed deep drainage conditions for temperate grasslands 19 .  degraded to those with a high degradation status and from forested peatlands to agricultural peatlands (3.3% of drained peatland or 554,300 ha year −1 ); (4) rewetting of all drained peatland until 2050 in the order of highest to lowest degradation stages and from agricultural peatlands to forested peatlands (3.3% of drained peatland or 554,300 ha year −1 ). For scenario (1), the topsoil BD of newly drained peatlands was estimated using a function between BD and peatland drainage years 53 . After 30 years of drainage, the soil BD is expected to increase from 0.1 g cm −3 to 0.15 and 0.26 for forested and agricultural peatland, respectively. We estimate that with these BD values, the N 2 O-N emissions from forested and agricultural peatland were 0.6 (0.5, 0.9) and 2.7 (2.0, 3.5) kg N ha −1 year −1 (after 30 years). We assume that the status of already degraded peat soils does not change within the 30 years for scenarios (1) and (2). For scenarios (3) and (4), we considered that the world has to reach zero GHG emissions by 2050 54 .