SCIENCE ADVANCES | RESEARCH ARTICLE ECOLOGY Copyright © 2021 The Authors, some Large historical carbon emissions from cultivated rights reserved; exclusive licensee northern peatlands American Association for the Advancement of Science. No claim to Chunjing Qiu1,2*, Philippe Ciais1, Dan Zhu1,3, Bertrand Guenet1,4, Shushi Peng5, original U.S. Government Ana Maria Roxana Petrescu6, Ronny Lauerwald7, David Makowski2, Works. Distributed Angela V. Gallego-Sala8, Dan J. Charman8, Simon C. Brewer9 under a Creative Commons Attribution When a peatland is drained and cultivated, it behaves as a notable source of CO2. However, we lack temporally NonCommercial and spatially explicit estimates of carbon losses from cultivated peatlands. Using a process-based land surface License 4.0 (CC BY-NC). model that explicitly includes representation of peatland processes, we estimate that northern peatlands con- verted to croplands emitted 72 Pg C over 850–2010, with 45% of this source having occurred before 1750. This source surpassed the carbon accumulation by high-latitude undisturbed peatlands (36 to 47 Pg C). Carbon losses from the cultivation of northern peatlands are omitted in previous land-use emission assessments. Adding this ignored historical land-use emission implies an 18% larger terrestrial carbon storage since 1750 to close the historical global carbon budget. We also show that carbon emission per unit area decrease with time since drainage, suggesting that time since drainage should be accounted for in inventories to refine land-use emissions from cultivated peatlands. INTRODUCTION because of lack of data (7, 8), and the dynamic global vegetation Land-use and land-cover change (LULCC) was the dominant models (DGVMs) used to simulate the global land carbon budget source of anthropogenic carbon emissions to the atmosphere, do not represent peatland (1). To date, peatland drainage and cul- before being overtaken by fossil fuel burning in the 1940s (1). tivation emissions have been estimated only using empirical ap- However, relative to other terms in the global carbon budget, there proaches: multiplying emission factors (EFs; carbon emissions per are still very large uncertainties in estimates of the emission of car- unit area from cultivated peatland) estimated from local field data bon from LULCC (2). One of the greatest challenges in constrain- (9) by the area of cultivated peatlands, derived from land-use sur- ing these emissions is quantifying carbon emission from human’s veys (3) or from the geospatial overlay of a present-day peatland use of peatland. Peatlands are ecosystems with a surface soil layer distribution map and an agricultural land-cover map (10–14). of plant litter that is only partially decomposed (peat) because of This method known as Intergovernmental Panel on Climate Change waterlogged conditions and has a thickness of at least 30 to 40 cm (IPCC) tier 1 (15) does not take account of EFs decreasing with time (3, 4). While natural peatlands remove CO2 from the atmosphere, after drainage because of the increased recalcitrance of the re- drainage and land-use change could trigger large CO2 emissions maining material (16). This slow time trend of EFs cannot be from these ecosystems. characterized accurately from local CO2 flux monitoring data given Anthropogenic changes to northern peatlands can be traced the lack of long-term observations. On the other hand, historical back to long before the industrial revolution (5): to crop cultivation, land-use history of peatlands is unknown, given that peatland pastures used for grazing, forestry, and peat fuel extraction. The use inventories only extend to the past three decades and cultivated of peatlands for arable cropping is of particular concern because the regions where peat has disappeared (17) cannot be seen from present-­ intensive field drainage aerates the soil and then enhances the mineral- day peatland distribution maps. Thus, a temporally and spatially ization of soil organic matter, hence markedly increasing respiration explicit estimate of carbon emissions from historical cultivation of (6), and the removal of harvested crops reduces plant litter inputs peatlands is still missing. into the soil. However, the contribution of cultivated peatlands to To fill this knowledge gap, we developed the process-based the global historical carbon budget is unknown: Bookkeeping models ORCHIDEE-PEAT (18, 19) land surface model to represent undis- used to reconstruct past and present land-use change emissions have turbed natural peatland and cultivated peatland as two separate soil ignored carbon emissions from peatlands converted to croplands columns and hydrological soil tiles within a model grid cell. Hydro- logical and carbon cycles of each subgrid tile are simulated inde- pendently so that the new version of the model simulates the 1 Laboratoire des Sciences du Climat et de l’Environnement, UMR 8212, CEA-CNRS- conversion of peatland to cropland and completely separate the car- UVSQ F-91191 Gif-sur-Yvette, France. 2UMR MIA 518, Université Paris-Saclay, INRAE, AgroParisTech, 16 rue Claude Bernard, 75231 Paris, France. 3Institut de Ciència i bon balances of undisturbed and cultivated peatland. In this study, Tecnologia Ambientals (ICTA), Universitat Autonoma de Barcelona, 08193 Barcelona, we use the model to simulate the Northern Hemisphere’s (regions Spain. 4Laboratoire de Géologie, UMR 8538, Ecole Normale Supérieure, PSL Re- north of 30°N) peatland carbon and hydrological processes coupled search University, CNRS, Paris, France. 5Sino-French Institute for Earth System Science, College of Urban and Environmental Sciences, Peking University, 100871 Beijing, with land-use change and to produce spatially explicit distributions China. 6Department of Earth Sciences, Vrije Universiteit Amsterdam, 1081 HV of the CO2 sequestrations by undisturbed peatlands and the CO2 Amsterdam, Netherlands. 7Université Paris-Saclay, INRAE, AgroParisTech, UMR emissions from historical conversion of peatlands to croplands over ECOSYS, 78850, Thiverval-Grignon, France. 8Geography Department, College of the past millennium. We first simulate natural peat carbon accumu- Life and Environmental Sciences, University of Exeter, Exeter EX4 4RJ, UK. 9Depart- ment of Geography, University of Utah, Salt Lake City, UT, USA. lation and dynamic peatland area extent from the Early Holocene *Corresponding author. Email: chunjing.qiu@lsce.ipsl.fr to the year 850 CE (from here, all years are CE), and then, by Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 1 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE prescribing to the model annual maps of reconstructed cropland RESULTS areas (20), we model the progressive land-use conversion and Historical evolution of northern peatlands drainage of peatland to cropland and the resulting CO2 emissions The simulated total area of northern peatlands before agricultural con- during the period 850–2010 (Materials and Methods). Past land-use version was 4.5 million km2 with a total soil carbon stock of 544 Pg C decisions about whether a new cropland area was claimed from (fig. S4). Between 850 and 2010, the area of undisturbed peatland peatlands or from other natural ecosystems are poorly constrained, decreased as a result of agricultural conversion by 0.8 million km2 and thus, we construct three contrasting conversion scenarios (fig. in CTL (range, 0.05 million to 1.8 million km2 from the MIN and S1). The MIN scenario is a minimum peatland conversion scenario, MAX scenarios, respectively; Fig. 1). The area of cultivated peat- which assumes that the increase in cropland area in a grid cell is lands showed a slow, gradual increase from 850 to 1750, during given in priority over non-peatland areas, with peatland being which only 0.2 million km2 northern peatlands became cultivated in drained only when no other vegetated area is available to match the CTL (range, 0.01 million to 0.6 million km2 from the MIN and expanding cropland area in each grid cell. The MAX scenario as- MAX scenarios, respectively). After 1750, a rapid expansion of agri- sumes a maximum peatland conversion, with any increase of crop- culture on peatlands took place. The area of cultivated peatlands land area in a grid cell being preferentially taken from peatland. reached a maximum in the second half of the 20th century: 0.5 mil- The control (CTL) scenario assumes that the increase in cropland lion km2 in CTL (range, 0.05 million to 1.6 million km2 in the MIN area is supplied proportionally from peatland and other vegetated and MAX scenarios, respectively). During the period 1990 to 2010, areas present in each grid cell where cropland expands. When the cropland area in the Northern Hemisphere decreased as a result of conversion of a natural peatland occurs in the fraction of a model regional abandonment (fig. S5). We simulated a reduction in culti- grid cell, the water table drops as water is allowed to drain freely, vated northern peatland area of 0.1 million km2 under our CTL sce- which fosters the aerobic decomposition of organic carbon in the nario (range, 0 to 0.1 million km2 for the MIN and MAX scenarios, upper soil horizons (Materials and Methods). Crop harvests are respectively). Because of the lack of large-scale observations, it is modeled as a direct removal and consumption of a fraction of hard to specify whether the decrease in cultivated peatland area was a aboveground biomass by humans (21), with residues being laid on result of management practices (i.e., rewetting of abandoned peatland) the ground. The historical evolution of crop productivity and of or land-use conversions (cultivated peatlands were converted into harvest index, that is, the ratio of harvested grain to aboveground another land-use type). For simplicity, all the reduction in the area biomass, is parameterized following Han et al. (22) and calibrated of cultivated peatlands was considered to have been reclaimed by to reproduce the historical increase in crop yields in northern coun- natural forest and/or grassland, and CO2 emissions after the re- tries (figs. S2 and S3). Subsequent carbon input from crop residues claim were constrained by soil moisture and temperature of a after drainage is generally insufficient to compensate for increased specific subgrid soil tile for natural forest/grassland. After a cul- peat decomposition losses, resulting in legacy carbon emissions tivated peatland is reclaimed by natural forest/grassland, it is ex- to the atmosphere long after the initial conversion of peatland cluded from our analysis because the model cannot explicitly track to cropland. changes in its carbon balance. By the year 2000, simulated total Fig. 1. Simulated area and cumulative net biome production of northern peatland in different scenarios. (A) Undisturbed peatland area. (B) Cultivated peatland area. (C) Cumulative net biome production (NBP) of undisturbed and cultivated peatland. Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 2 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE (undisturbed + cultivated) area of northern peatlands was 4.2 mil- Historical carbon emissions from northern peatlands lion km2 in CTL (range, 4.2 million to 4.5 million km2 for the MAX Simulated undisturbed peatlands was a small atmospheric carbon and MIN scenarios, respectively) with a total soil carbon stock of sink of 0.03 Pg C year−1 (under all three scenarios) over 850–1850, 500 Pg C (range, 394 to 581 Pg C for the MAX and MIN scenarios, and this sink has intensified from 1850 in response to rising atmo- respectively). Both values are at the upper end of previous estimates spheric CO2 and climate change, reaching 0.2, 0.3, and 0.1 Pg C for present-day northern peatlands (3.4 million to 4.0 million km2 year−1 in the CTL, MIN, and MAX scenarios, respectively, over and 270 to 540 Pg C, respectively) (23, 24). 1990–2000. Previous studies suggested a northern peatland carbon We compiled previously published estimates of contempo- sink of ~0.09 Pg C year−1 for the past 1000 years (35) and of 0.10 Pg rary peatland area for 12 northern countries to evaluate our sim- C year−1 for 850–1850 (36). However, both estimates were based on ulation of peatland area (3, 11, 24–34). The total peatland area carbon accumulation rates derived from age-dated peat cores and of these countries shown in Fig. 2 accounts for ~90% of peatlands in did not take into account the decay of deeper peat (peat accumulated the Northern Hemisphere. Our simulated area of cultivated peatlands before 850). After considering the decay of previously accumulated generally matches with observations, but total and consequently peat, northern natural peatlands were a net carbon sink of only cultivated peatland areas are overestimated in the United States, 0.04 Pg C year−1 over the past 1000 years (35), which compares well Ukraine, and the United Kingdom. Undisturbed peatland area in with our estimate of 0.04 Pg C year−1 over 850–2010. Finland and Sweden are underestimated by the model, while the Carbon emission in the context of this study is defined as net area of cultivated peatlands in both countries is well reproduced. biome production (NBP), which includes both the net CO2 exchange Fig. 2. Simulated total (green transparent boxes) and cultivated (orange transparent boxes) peatland area under three scenarios (MIN, CTL, and MAX), com- pared with previously published estimates for 12 countries. *Drained area for agriculture, not only for crop cultivation. †Following the 2006 IPCC Guidelines (38), land that has not undergone any land-use conversion for a period of at least 20 years is reported in a “remaining” category to UNFCCC National Greenhouse Gas Inventory (Submissions 2019) (39), while land affected by land-use conversion is reported in a “conversion” category for 20 years and then will be transferred to the remaining cat- egory. Estimates of United Nations Framework Convention on Climate Change (UNFCCC) are the sum of the category reported as “Cropland remaining Cropland on organic soils” and the category reported as “Wetland converted to Cropland on organic soils” (common reporting format table 4.B). Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 3 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE between the ecosystem and atmosphere and carbon losses due to Change of EFs of cultivated northern peatlands with time harvest of crop biomass. Positive NBP represents net carbon loss since drainage from the ecosystem, whereas negative NBP represents net car- The simulated carbon emission rates of cultivated peatlands from bon uptake by the ecosystem. We estimated an average carbon our three conversion scenarios in the previous section depend both emission from cultivated northern peatlands of 0.15 Pg C year−1 on changes in the area of converted peatland and on changes in over the 1990–2000 decade (range, 0.02 to 0.25 Pg C year−1 from the carbon per unit area, equivalent to EFs (Materials and Methods). In MIN and MAX scenarios, respectively), larger than the estimate by contrast, in situ observed EFs are the rate of carbon emissions per Carlson et al. (11) (0.08 Pg C year−1) and comparable to the esti- unit area of cultivated peatland, independent of the area of converted mates by Joosten (3) (0.11 Pg C year−1) and Frolking et al. (4) peatland. To obtain simulated EFs that can be compared with field (0.15 Pg C year−1) for northern peatlands under drained agricultur- measurements, we performed additional model sensitivity experi- al land use. The difference between those estimates is attributable to ments in which all the natural peatlands were converted at a certain the area of cultivated peatlands and the EFs used by each study. We date and then the cultivated peatland area is held constant after that found that the area of cultivated peatlands ranked as this study > first conversion. In these model experiments, the conversion of Frolking et al. > Joosten > Carlson et al., whereas the EFs ranked as northern peatlands to croplands is prescribed to occur as a single this study < Frolking et al. < Joosten < Carlson et al. (table S1). Note cohort so that interference from further changes in the area of con- that any former cultivated peatland that has been converted into a verted peatland is excluded. The simulated EFs deduced from these mineral soil because of peat carbon loss cannot be identified by model experiments are diagnosed in each model grid cell as the observations of present soil properties but has been included in our NBP of cultivated peatlands per unit area. model estimates. There is observational evidence that extensive Measurements of carbon emissions from cultivated peatlands areas of former peatlands in northern countries, such as the are scarce. For instance, the default tier 1 EFs of the 2013 Wetland Netherlands (17), Belarus (25), and Denmark (37), have been re- Supplement (9) to the 2006 IPCC Guidelines for the boreal and classified into other soil classes because of peat subsidence, the loss temperate zone were based on measurements from only 39 European of soil organic matter, and erosion following human activities (28), sites (table S2). On the basis of eddy covariance/chamber and/or in although it is likely that crop cultivation was not the only human situ peat subsidence measurements (3, 4, 9, 39–44), carbon emis- activity in history that may have caused degradation of peatlands. sions from temperate and boreal cultivated peatlands were estimated We estimated cumulative emissions from cultivated peatlands to range between 0.8 and 12.6 tonnes C ha−1 year−1 (Fig. 3A). These under the CTL, MIN, and MAX scenarios to be 31, 1, and 84 Pg C, site-level estimates, however, were derived from peatlands that have respectively, over 850–1750 and 40, 3, and 84 Pg C, respectively, been drained for decades or centuries (table S2); thus, we cannot over 1750–2010 (Fig. 1C). Cumulative sequestrations by undisturbed differentiate the initial pulse of carbon emissions after conversion northern peatlands are 43, 47, and 36 Pg C over 850–2010 and 21, from the long-term tail of emissions after peatland has been drained 25, and 14 Pg C for 1750–2010 in the CTL, MIN, and MAX scenarios, and cultivated for decades (9). To gain insight into how EFs change respectively, thus compensating only half of carbon emissions from with time after conversion, the first conversion of northern peatlands cultivated northern peatlands. to croplands is prescribed to occur in 1900, 1950, 1960, 1970, 1980, Fig. 3. Observed and simulated EFs. (A) Simulated present-day (averaged over 1990–2010) EFs when the conversion of peatlands to croplands occurred in 1900, 1950, 1960, 1970, 1980, and 1990, compared to observations. In each violin, the black mini-box shows the first and the third percentile of simulated EFs, the white dot shows the median, and whiskers show first quartile − 1.5 × IQR and third quartile + 1.5 × IQR. Note that EFs from (40) are the average and the ranges, whereas EFs from (9, 41, 42) are the average and the 95% confidence interval of the measurement. For (39, 43), we show the average and ranges of EFs used by the same 12 countries as in Fig. 2. EFs for the category reported as Wetland converted to Cropland on organic soils to UNFCCC considered carbon stock change in living biomass and are different from EFs used for the category reported as Cropland remaining Cropland on organic soils; thus, the EFs for (39) in this figure represent area-weighted averages of these two categories. (B) Simulated EFs after peatlands have been converted for 7 years, 8 to 14 years, 15 to 21 years, 22 to 28 years, and 29 to 35 years (simulated drainage and conversion of peatlands to croplands occurred in 1960) compared to measured SOC loss rates of a peatland used for row crop rotation from a long-term continuous experiment (44). Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 4 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE and 1990, respectively, in each sensitivity experiment (Materials From our two extreme conversion scenarios (MIN and MAX), and Methods). Those simulated EFs in Fig. 3A clearly show that the we estimate that a range of 3 to 84 Pg C was emitted by cultivated earlier peatlands were drained, the smaller the predicted present-­ northern peatlands over 1750–2010. At site scale, observed carbon day (average of 1990–2010) EFs. For instance, the median EF of emissions from cultivated peatlands depend on local context, i.e., peat peatlands that were converted in 1900 is 1.1 t C ha −1 year −1 , type, land-use history, crop type, and management practices such as which is 92% smaller than the one of peatlands converted in 1990. water table level controls, fertilization, cultivation intensity, and tillage The interquartile range (IQR) of present-day EFs goes down from (32, 48–51). These local conditions and management history are 13.6 t C ha−1 year−1 (if peatlands were converted in 1990) to 1.9 t C not known from gridded datasets in the Northern Hemisphere and ha−1 year−1 (if peatlands were converted in 1900), indicating that further cannot be reproduced by a large-scale land surface model with the spatial variation in present-day cultivated peatlands EFs also de- generic parameters, which hinders our capacity to understand the creases with time since conversion. This implies that EFs should full uncertainty range in cultivated peatland carbon emissions. depend on the time since conversion, with high values shortly after The model sensitivity experiments show a sizeable decrease of EFs conversion, which then decrease over time. We found only one with time since conversion, suggesting that inventories should col- study (44) that measured peat soil carbon before cultivation and lect EF data for a range of cultivated peatlands of different ages, and continuously followed the evolution of the drained peat soil after include the time of peatland agriculture inception in their estimate. drainage for more than four decades, the observed decreasing soil Alternatively, a calibrated model such as ORCHIDEE-PEAT could be organic carbon (SOC) loss rates with time since conversion match used to simulate how EFs vary with time after land-use change. well with our results (Fig. 3B). Carbon loss by peatland is irreversible on centennial time scales; In Fig. 3A, the best agreement between simulated EFs and this previously overlooked anthropogenic carbon emission should be available observed EFs is found when the conversion of northern accounted for in assessments of the historical carbon budget and in peatlands is set to the mid-20th century. This is consistent with the estimates of the dwindling carbon budget for assessing future warm- fact that the few peatland sites measured to define the tier 1 CO2 EFs ing targets. Future work should focus on improving the reconstruction in the 2013 Wetland Supplement (9) were subject to extensive of land-use change involving peatland, including practices such as the peatland conversion and/or deepening of the artificial peatland extraction of peat as a fuel, the conversion and improvement of peatland drainage in the mid-20th century (table S2), although no precise for pasture and forestry, as well as the restoration of peatlands date of conversion was reported. (10, 52). The second priority is to extend estimates of LULCC carbon emissions from northern peatlands to the tropics, where deforestation and fires, oil palm and rubber plantations, and the encroachment of DISCUSSION urban areas into coastal peat ecosystems appear to have caused a Our study accounts for the full spatial and temporal development of reduction of peat carbon stocks in many regions (53, 54) and are historical peatland cultivation and pertaining carbon emissions. It likely to be major sources of future emissions from peatlands. reveals large historical carbon emissions from the cultivation of northern peatlands. Land-use emissions due to the cultivation of peatlands have not otherwise been accounted for either by book- MATERIALS AND METHODS keeping models or DGVMs. These models treat cultivated peatlands Input data as croplands on mineral soils. They assess Northern Hemisphere Simulations were performed using the ORCHIDEE-PEAT land emissions from all LULCC activities to be about 60 Pg C over the surface model (18, 19), driven by meteorological forcing data, histor- period 1750–2018 (fig. S6) (1, 45). Adding land-use emissions from ical land-cover maps, and historical data of atmospheric CO2 concen- northern peatland conversion from the CTL scenario of our study, trations. For the meteorological forcing, we used CRUNCEP version 8 therefore, implies a major upward reappraisal of total historical (http://forge.ipsl.jussieu.fr/orchidee/wiki/Documentation/Forcings) LULCC emissions between 1750 and 2018 by 67% for the Northern at 2°-by-2° spatial resolution. This dataset covers the years 1901 to Hemisphere [calculated as (40/60 Pg C) × 100%]. As the historical 2010. Historical land-cover maps at 0.25°-by-0.25° spatial resolu- (1750–2018) cumulative human-caused carbon emissions from fos- tion and covering the period from 850 to 2010 were reconstructed by sil fuels and cement (440 ± 20 Pg C) (46), the cumulative carbon merging the European Space Agency (ESA) Climate Change Initiative uptake by oceans (170 ± 20 Pg C) (1), and the cumulative increase (CCI) dataset (www.esa-landcover-cci.org/) with the LUH2v2 data- in atmospheric CO2 (275 ± 5 Pg C) (47) are relatively well con- set (http://luh.umd.edu/). In those maps, vegetation is discretized strained, the global land biosphere is constrained to be near neutral into 13 plant functional types (PFTs), including one for bare soil over 1750–2018. Our upward revision of northern LULCC emis- type, eight for trees, and one each for C3 natural grass, C4 natural sions after including emissions from cultivated northern peat- grass, C3 crops, and C4 crops. A detailed description of the recon- lands implies that the cumulative carbon uptake by terrestrial struction of the land-cover maps can be found at https://orchidas. ecosystems has been underestimated because the carbon emitted lsce.ipsl.fr/dev/lccci/. For simulation years before 1861, atmospher- by cultivated northern peatlands needs to be balanced by a corre- ic CO2 concentration was fixed at the preindustrial level (286 parts sponding increase in the terrestrial carbon uptake fluxes (2). The per million). For the period 1861–2010, observation-­based historical most recent assessment of the global carbon budget (1) estimated CO2 concentrations were used. that the cumulative carbon uptake by terrestrial ecosystems over 1750–2018 is 220 ± 50 Pg C. We show that this number needs to ORCHIDEE-PEAT land surface model be increased by 18% [calculated as (40/220 Pg C) × 100%] if the The starting point for ORCHIDEE-PEAT is ORCHIDEE-MICT re- budget takes emissions from cultivated northern peatlands into vision 4229 (55). The model groups vegetation species with similar account. characteristics into functional types (PFTs), and the areal extent of Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 5 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE each PFT in a grid cell is described as its fractional area in the grid in non-peatland soil tiles. In contrast, nondisturbed peatlands cell (56). Carbon exchanges between the atmosphere and plants and remain on soil tiles that are not drained from the bottom and thus soils are simulated independently for each PFT. keep water in the peat column. Switching on the deep drainage pro- In this study, fractional areas of 13 PFTs growing on non-peatland cess after conversion causes previously water-saturated “catotelm” were predefined from the land-cover map described above. An ad- deep layers to lose water and triggers a reduction of the soil mois- ditional, peatland-specific PFT (the 14th PFT) was added to repre- ture content in the upper part of the peat column (see example for sent natural peatlands (18, 19). The extent of this peatland-specific one grid cell in fig. S9). Consequently to this drying of the upper PFT was dynamically simulated from the Early Holocene to 850. peat column, an enhancement of aerobic decomposition of peat For each grid cell, the inundation extent was first simulated by a occurs. Like for undisturbed peatlands, the decomposition of SOC cost-efficient version of TOPMODEL according to subgrid-scale in a cultivated peatland is modeled to depend on soil temperature topographic information, simulated soil temperature, and soil and moisture for each layer. Aerobic decomposition occurs in any moisture of the grid cell (19, 57). An inundation area could then unsaturated layer. Simulated peat depth in grid cells where part of develop into peatland if the peatland development criteria below natural peatlands has been converted to croplands is generally were met: (i) Over a long period, the area is frequently inundated (at greater than 1.5 m, and about 30% of those grid cells have a peat least Num months during 30 years, with Num being the number of depth greater than 5.1 m (fig. S10). However, because both the soil growing season months in the 30 years). (ii) The grid cell has a pos- moisture content and the turnover time of SOC of cultivated peat- itive summer water balance (total summer precipitation − potential land increased with depth, the largest decomposition rates and, evaporation ≥ 60 mm). (iii) The peatland PFT has accumulated soil thus, the largest CO2 emissions from drainage occurred in the top carbon of more than 50 kg C m−2 (this last condition only con- 1.5 m (>20% loss of SOC 20 years after the conversion of peatlands strains the expansion of peatland; the inception of peatland in the to croplands; fig. S10). grid cell is not limited by it). Subsequently, the simulated areal frac- Crops planted on cultivated peatlands are represented by a C3 tion of the peatland-specific PFT was subtracted from the pre- crop PFT (the 15th PFT) and a C4 crop PFT (the 16th PFT). The defined fraction of natural PFTs from the land-cover map. 15th and 16th PFTs follow the same suite of equations and parame- The model resolves vertical soil water flow and heat transfer ters as the C3 (the 12th) and C4 crop (the 13th) PFTs on non-peatland within the soil column by multilayer physically based schemes at a areas, respectively (56). Their areal fraction was calculated and half-hourly time step (55). The natural peatland PFT was represent- updated annually by a grid cell–by–grid cell overlay of the simulated ed as a subgrid soil tile characterized by a large porosity (0.9 m3 m−3) peatland area with the cropland area from the land-cover map and a large saturated water conductivity (2120 mm day−1); thus, the (see the next section for three conversion scenarios). The areal frac- downward water percolation of this peatland tile is larger than tion of the 15th and 16th PFT was then subtracted from the pre- non-peatland tiles (18). Furthermore, the natural peatland soil tile defined fraction of the 12th and 13th PFT from the land-cover map, has an impermeable bottom and receives water input from surface respectively. runoff from non-peatland tiles in the model grid cell (in addition to Harvest of crops is represented as a direct removal and con- water from precipitation) (18); both parameterizations contribute sumption of a fraction of aboveground biomass by humans (21). to simulated shallow water tables of the natural peatland soil tile Tillage of cultivated peatlands is represented by a 20% increase of (fig. S7). soil carbon decomposition rates (21). To evaluate the simulated Two litter (metabolic and structural) and three soil (active, slow, decomposability of SOC of cultivated peatlands, we created an and passive) carbon pools are represented by the model; all are ver- emulator to model incubation experiments using the structure and tically discretized into 32 layers (38 m in total, with exponentially parameters of ORCHIDEE-PEAT. More specifically, we simulated coarser resolution as depth increases). Aboveground plant residues CO2 emission rates from each soil layer when it is exposed to a given are added to the top layer of litter pools, while belowground plant incubation temperature and optimal moisture condition (no mois- residues are added to active layers above the permafrost, assuming ture limitation). Our simulated CO2 release rates fall well within an exponential distribution of vegetation roots. To take the vertical observed rates from laboratory aerobic incubation experiments for heterogeneity in soil temperature, soil freezing, and soil moisture arable peatland: simulated 0.09 to 0.18 versus measured 0.07 to into account, the decomposition of peat is also parameterized as a 0.31 mg C g−1 C day−1 incubated at 10° to 20°C from seven arable multilayer scheme. For each model layer, different residence times peatland sites in Switzerland (59), simulated 0.29 to 0.40 versus are prescribed for each carbon pool; these are then modified by measured 0.38 to 0.65 mg C g−1 C day−1 incubated at 22°C from two temperature and moisture of the layer at a daily time step (19). The arable peatland sites in Canada (60), and simulated 0.10 to 0.21 ver- carbon flow between pools is also calculated for each layer, follow- sus 0.12 to 0.26 mg C g−1 C day−1 incubated at 23°C from 50 sites in ing the CENTURY model (fig. S8) (58). More detailed descriptions Germany (61) [cultivated sites were not distinguished from other of the treatment of natural peatland in ORCHIDEE-PEAT can be agricultural uses (61)]. found in (18, 19). To represent peatlands reclaimed for crop cultivation, we added Model experiments and simulation protocol another independent subgrid soil tile. It has the same porosity and Simulation protocol saturated water conductivity as natural peatlands. In reality, the wa- Simulations were conducted in four steps. The first two steps (Early ter tables of cultivated peatlands are managed by humans, i.e., water Holocene and Later Holocene to 850) were used to simulate carbon table is lowered by drainage ditches. Detailed peat conversion prac- accumulation and dynamic area expansion of natural peatlands tices cannot be explicitly modeled. We assumed that each time a using TOPMODEL and the peatland development criteria. In the peat fraction is converted, it is allocated to a new soil tile from which third and fourth steps (850 to 1900 and 1901 to 2010), the conver- water in the bottom of the soil column is allowed to drain freely as sion of natural peatland to cropland was simulated. According to Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 6 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE HYDE 3.2 (20), total cropland area in northern regions (>30°N) was If the area of cultivated peatland in a grid cell increases, cultivated small (~0.7 million km2) before 850. We therefore assumed that all peatland will encroach on natural peatland, and then, the areal these croplands were on mineral soils and that natural peatlands fraction of both natural (Eq. 1) and cultivated peatland (Eq. 2) and have been reclaimed for agricultural use only since 850. The first the SOC density of the cultivated peatland in the grid cell (Eq. 3) step of the simulation aimed at reaching a steady state for natural will be updated, whereas the SOC density of the natural peatland peatland area and soil carbon stocks in the Early Holocene. This remains unchanged (Eq. 4) step included 180 years of full model runs to reach steady state for t+1 biomass, followed by a 12,000-year spin-up of the soil carbon sub- ​​f​natural ​t ​​ = ​fnatural​​  −  ​f​ cultivated​​​ (1) model to reach steady state for soil carbon pools, looping over the meteorological forcing data for 1960–1990 to approximate the t+1  ​ = ​ ​​f​cultivated t ​ f​cultivated​​  +  ​f​ cultivated​​​ (2) warmer-than-preindustrial climate of the Early Holocene (62). The second step consisted of running the full model over another 100 years without peatland conversion to croplands, before the model was run ​​ in the third step with changing land cover, and the consequent con- ​SOC​tcultivated​​ × ​f​cultivated t ​​  + ​SOC​tnatural​ ​×  ​f​ cultivated​​ SOC​t+1 ​ = ​ cultivated​ ──────────────────────────          t ​​ (3) version of natural to cultivated peatlands from 850 to 1900. For the ​fcultivated ​ ​​  +  ​f​ cultivated​​ second and third steps, we looped over the meteorological forcing data for 1901–1920 to approximate the cooler climate of the late ​​SOC​t+1 ​ ​ = ​SOC​tnatural​​ ​ natural (4) Holocene. In the fourth step, fully transient runs for 1901–2010 were conducted using the corresponding historical data for both the where fcultivated is the change in the areal fraction of cultivated peat- meteorological forcing and land cover. land in the grid cell, SOCcultivated and fcultivated are the SOC density Three scenarios of peatland conversion and the areal fraction of cultivated peatland in the grid cell, respec- To estimate historical conversion of peatland from 850 to 2010, for tively, SOCnatural and fnatural are the SOC density and the areal frac- each year, a reconstructed cropland cover map was first predefined tion of natural peatland in the grid cell, respectively, superscript in the model, and then, the model performed a grid cell–by–grid trepresents the state before the expansion of cultivated peatland cell overlay of the cropland map with the simulated peatland distri- area, and superscript t + 1 represents the state after the expansion of bution. The time of peatland agriculture inception and its expan- cultivated peatland area. sion trajectory varies greatly between different regions (5); for If the area of cultivated peatland in a grid cell decreases, then example, the use of peatlands in some regions is inhibited by limited abandoned cultivated peatland area will be reclaimed by a forest accessibility, while some other peatlands were extensively used for and/or grassland PFT and its SOC will be allocated to forest and/or crops (63). Lacking information on historical changes in the spatial grassland SOC stock. As a result, only the areal fraction of cultivated distribution of cultivated peatlands, we defined two scenarios to peatland will be updated (Eq. 5) span a conservative uncertainty range. One scenario describes a t+1 minimum estimate for peatland conversion (MIN): It assumes that ​​f​cultivated ​t ​​ = ​fcultivated​​  −  ​f​ cultivated​​​ (5) the increase of cropland area in a grid cell is first allocated to non-peatland area and then to peatland if no more non-peatland Therefore, the SOC density of a cultivated peatland in a grid cell area is available. The other scenario describes a maximum estimate does not change if its area decreases. In contrast, when the area of of peatland conversion (MAX): It assumes that the increase of crop- cultivated peatland in a grid cell increases, the cultivated peatland land area in a grid cell is first allocated to peatland and then to ecosystem will “take over” carbon from natural peatland, resulting non-peatland area if no more peatland is available. The true area of in an increase in its SOC density and, subsequently, an increase in cultivated peatlands is somewhere within the range of MIN and carbon emission rates. We thus performed model sensitivity exper- MAX. In addition, we followed a third scenario (the control scenario, iments with a fixed area of cultivated peatland to reproduce the evo- CTL) as the most plausible: It assumes that the increase in cropland lution of emission rates as a function of time since drainage and area of a grid cell is supplied proportionally from peatland and thus to obtain simulated EFs values that can be compared with field non-peatland ecosystems present in the grid cell (fig. S1) (10–13). measurements. Thus, simulated changes in the extent of cultivated peatland depend Sensitivity experiments only on cropland area dynamics (from reconstructed cropland cov- The first conversion of peatlands to croplands for providing simu- er maps) and on our scenarios of different peatland fractions being lated EFs values is prescribed to occur in 1900, 1950, 1960, 1970, converted when the area of cropland changes in a grid cell. Changes 1980, and 1990, respectively, in each model experiment. The area of in the extent of cultivated peat are independent from the carbon cultivated peatlands was kept fixed after the year of conversion so balance of cultivated peatland per unit area after a conversion, i.e., that cultivated peatland in the experiment is represented as a single cultivated peatlands that have become CO2 sinks because of deple- cohort. This procedure excludes additional carbon losses, e.g., from tion of peat or the recharge of soils from cropland input were not subsequent changes in the area of cultivated peatland and allows to excluded from our estimates. simulate EFs that mimic the observation conditions. Figure S11 As a large-scale land surface model, ORCHIDEE-PEAT cannot shows changes in the area-weighted mean carbon emission rates track individual cohorts of cultivated peatlands explicitly. Changes from cultivated northern peatlands. in the extent of a cultivated peatland in a grid cell are represented as an increase/decrease in its areal fraction in the grid cell. As a result, SUPPLEMENTARY MATERIALS changes in the area of cultivated peatland in a grid cell interfere with Supplementary material for this article is available at http://advances.sciencemag.org/cgi/ the simulated SOC density of cultivated peatland. content/full/7/23/eabf1332/DC1 Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 7 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE REFERENCES AND NOTES 21. S. Piao, P. Ciais, P. Friedlingstein, N. de Noblet-Ducoudré, P. Cadule, N. Viovy, T. Wang, 1. P. Friedlingstein, M. Jones, M. O’Sullivan, R. Andrew, J. Hauck, G. Peters, W. Peters, Spatiotemporal patterns of terrestrial carbon cycle during the 20th century. J. Pongratz, S. Sitch, C. Le Quéré, D. C. E. Bakker, J. G. Canadell, P. Ciais, R. B. Jackson, Global Biogeochem. Cycles 23, GB4026 (2009). P. Anthoni, L. Barbero, A. Bastos, V. Bastrikov, M. Becker, L. Bopp, E. Buitenhuis, 22. P. Han, N. Zeng, F. Zhao, X. Lin, Estimating global cropland production from 1961 to 2010. N. Chandra, F. Chevallier, L. P. Chini, K. I. Currie, R. A. Feely, M. Gehlen, D. Gilfillan, Earth Syst. Dynam. 8, 875–887 (2017). T. Gkritzalis, D. S. Goll, N. Gruber, S. Gutekunst, I. Harris, V. Haverd, R. A. Houghton, 23. Z. C. Yu, Northern peatland carbon stocks and dynamics: A review. Biogeosciences 9, G. Hurtt, T. Ilyina, A. K. Jain, E. Joetzjer, J. O. Kaplan, E. Kato, K. K. Goldewijk, 4071–4085 (2012). J. I. Korsbakken, P. Landschützer, S. K. Lauvset, N. Lefèvre, A. Lenton, S. Lienert, 24. J. Xu, P. J. Morris, J. Liu, J. Holden, PEATMAP: Refining estimates of global peatland D. Lombardozzi, G. Marland, P. C. McGuire, J. R. Melton, N. Metzl, D. R. Munro, distribution based on a meta-analysis. Catena 160, 134–140 (2018). J. E. M. S. Nabel, S.-I. Nakaoka, C. Neill, A. M. Omar, T. Ono, A. Peregon, D. Pierrot, 25. A. Kozulin, N. Tanovitskaya, N. Minchenko, Developing a national strategy B. Poulter, G. Rehder, L. Resplandy, E. Robertson, C. Rödenbeck, R. Séférian, J. Schwinger, for the conservation and sustainable use of peatlands in the Republic of Belarus. Mires Peat. N. Smith, P. P. Tans, H. Tian, B. Tilbrook, F. N. Tubiello, G. R. van der Werf, A. J. Wiltshire, 21, 1–17 (2018). S. Zaehle, Global carbon budget 2019. Earth Syst. Sci. Data 11, 1783–1838 (2019). 26. T. Lapveteläinen, K. Regina, P. Perälä, Peat-based emissions in Finland’s national 2. A. Arneth, S. Sitch, J. Pongratz, B. D. Stocker, P. Ciais, B. Poulter, A. D. Bayer, A. Bondeau, greenhouse gas inventory. Boreal Environ. Res. 12, 225–236 (2007). L. Calle, L. P. Chini, T. Gasser, M. Fader, P. Friedlingstein, E. Kato, W. Li, M. Lindeskog, 27. A. Grønlund, A. Hauge, A. Hovde, D. P. Rasse, Carbon loss estimates from cultivated peat J. E. M. S. Nabel, T. A. M. Pugh, E. Robertson, N. Viovy, C. Yue, S. Zaehle, Historical carbon soils in Norway: A comparison of three methods. Nutr. Cycl. Agroecosyst. 81, 157–167 (2008). dioxide emissions caused by land-use changes are possibly larger than assumed. 28. H. Joosten, D. Clarke, Wise Use of Mires and Peatlands: Background and Principles Including Nat. Geosci. 10, 79–84 (2017). A Framework for Decision-Making (International Mire Conservation Group and 3. H. Joosten, The Global Peatland CO2 Picture: Peatland Status and Drainage Related International Peat Society, 2002). Emissions in all Countries of the World (Wetlands International, 2010). 29. N. H. Batjes, Harmonized soil property values for broad-scale modelling (WISE30sec) 4. S. Frolking, J. Talbot, M. C. Jones, C. C. Treat, J. B. Kauffman, E.-S. Tuittila, N. T. Roulet, with estimates of global soil carbon stocks. Geoderma 269, 61–68 (2016). Peatlands in the Earth’s 21st century climate system. Environ. Rev. 19, 371–396 (2011). 30. F. Tanneberger, C. Tegetmeyer, S. Busse, A. Barthelmes, S. Shumka, A. M. Mariné, 5. J. Holden, P. J. Chapman, J. C. Labadz, Artificial drainage of peatlands: Hydrological K. Jenderedjian, G. M. Steiner, F. Essl, J. Etzold, C. Mendes, A. Kozulin, P. Frankard, and hydrochemical process and wetland restoration. Prog. Phys. Geogr. 28, 95–123 Đ. Milanović, A. Ganeva, I. Apostolova, A. Alegro, P. Delipetrou, J. Navrátilová, M. Risager, (2004). A. Leivits, A. M. Fosaa, S. Tuominen, F. Muller, T. Bakuradze, M. Sommer, K. Christanis, 6. H. E. Taft, P. A. Cross, G. Edwards-Jones, E. R. Moorhouse, D. L. Jones, Greenhouse gas E. Szurdoki, H. Oskarsson, S. H. Brink, J. Connolly, L. Bragazza, G. Martinelli, O. Aleksāns, emissions from intensively managed peat soils in an arable production system. A. Priede, D. Sungaila, L. Melovski, T. Belous, D. Saveljić, F. De Vries, A. Moen, W. Dembek, Agric. Ecosyst. Environ. 237, 162–172 (2017). J. Mateus, J. Hanganu, A. Sirin, A. Markina, M. Napreenko, P. Lazarević, V. Š. Stanová, 7. R. A. Houghton, A. A. Nassikas, Global and regional fluxes of carbon from land use P. Skoberne, P. H. Pérez, X. Pontevedra-Pombal, J. Lonnstad, M. Küchler, C. Wüst-Galley, and land cover change 1850–2015. Global Biogeochem. Cycles 31, 456–472 (2017). S. Kirca, O. Mykytiuk, R. Lindsay, H. Joosten, The peatland map of Europe. Mires Peat. 19, 8. E. Hansis, S. J. Davis, J. Pongratz, Relevance of methodological choices for accounting 1–17 (2017). of land use change carbon fluxes. Global Biogeochem. Cycles 29, 1230–1246 (2015). 31. E. Lappalainen, Global Peat Resources (International Peat Society, 1996). 9. IPCC, 2013 Supplement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories: 32. Ö. Berglund, K. Berglund, Distribution and cultivation intensity of agricultural peat Wetlands, T. Hiraishi, T. Krug, K. Tanabe, N. Srivastava, J. Baasansuren, M. Fukuda, and gyttja soils in Sweden and estimation of greenhouse gas emissions from cultivated T. G. Troxler, Eds. (IPCC, 2014). peat soils. Geoderma 154, 173–180 (2010). 10. J. Leifeld, L. Menichetti, The underappreciated potential of peatlands in global climate 33. N. Roeder, B. Osterburg, The impact of map and data resolution on the determination change mitigation strategies. Nat. Commun. 9, 1071 (2018). of the agricultural utilisation of organic soils in Germany. Environ. Manag. 49, 1150–1162 11. K. M. Carlson, J. S. Gerber, N. D. Mueller, M. Herrero, G. K. MacDonald, K. A. Brauman, (2012). P. Havlik, C. S. O’Connell, J. A. Johnson, S. Saatchi, P. C. West, Greenhouse gas emissions 34. H. Kekkonen, H. Ojanen, M. Haakana, A. Latukka, K. Regina, Mapping of cultivated intensity of global croplands. Nat. Clim. Chang. 7, 63–68 (2017). organic soils for targeting greenhouse gas mitigation. Carbon Manag. 10, 115–126 12. J. Leifeld, C. Wüst-Galley, S. Page, Intact and managed peatland soils as a source and sink (2019). of GHGs from 1850 to 2100. Nat. Clim. Chang. 9, 945–947 (2019). 35. Z. Yu, Holocene carbon flux histories of the world’s peatlands: Global carbon-cycle 13. F. Tubiello, R. Biancalani, M. Salvatore, S. Rossi, G. Conchedda, A worldwide implications. Holocene 21, 761–774 (2011). assessment of greenhouse gas emissions from drained organic soils. Sustainability 8, 36. A. V. Gallego-Sala, D. J. Charman, S. Brewer, S. E. Page, I. C. Prentice, P. Friedlingstein, 371 (2016). S. Moreton, M. J. Amesbury, D. W. Beilman, S. Björck, T. Blyakharchuk, C. Bochicchio, 14. G. Conchedda, F. N. Tubiello, Drainage of organic soils and GHG emissions: Validation R. K. Booth, J. Bunbury, P. Camill, D. Carless, R. A. Chimner, M. Clifford, E. Cressey, with country data. Earth Syst. Sci. Data 12, 3113–3137 (2020). C. Courtney-Mustaphi, F. De Vleeschouwer, R. de Jong, B. Fialkiewicz-Koziel, 15. IPCC, 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas S. A. Finkelstein, M. Garneau, E. Githumbi, J. Hribjlan, J. Holmquist, P. D. M. Hughes, Inventories (2019); www.ipcc.ch/report/2019-refinement-to-the-2006-ipcc-guidelines- C. Jones, M. C. Jones, E. Karofeld, E. S. Klein, U. Kokfelt, A. Korhola, T. Lacourse, G. Le Roux, for-national-greenhouse-gas-inventories/ [accessed 11 March 2021]. M. Lamentowicz, D. Large, M. Lavoie, J. Loisel, H. Mackay, G. M. MacDonald, M. Makila, 16. S. Rojstaczer, S. J. Deverel, Time dependence in atmospheric carbon inputs from drainage G. Magnan, R. Marchant, K. Marcisz, A. Martínez Cortizas, C. Massa, P. Mathijssen, of organic soils. Geophys. Res. Lett. 20, 1383–1386 (1993). D. Mauquoy, T. Mighall, F. J. G. Mitchell, P. Moss, J. Nichols, P. O. Oksanen, L. Orme, 17. B. Kempen, D. J. Brus, G. B. M. Heuvelink, J. J. Stoorvogel, Updating the 1:50,000 Dutch M. S. Packalen, S. Robinson, T. P. Roland, N. K. Sanderson, A. B. K. Sannel, N. Silva-Sánchez, soil map using legacy soil data: A multinomial logistic regression approach. Geoderma N. Steinberg, G. T. Swindles, T. E. Turner, J. Uglow, M. Väliranta, S. van Bellen, 151, 311–326 (2009). M. van der Linden, B. van Geel, G. Wang, Z. Yu, J. Zaragoza-Castells, Y. Zhao, Latitudinal 18. C. Qiu, D. Zhu, P. Ciais, B. Guenet, G. Krinner, S. Peng, M. Aurela, C. Bernhofer, C. Brümmer, limits to the predicted increase of the peatland carbon sink with warming. Nat. Clim. S. Bret-Harte, H. Chu, J. Chen, A. R. Desai, J. Dušek, E. S. Euskirchen, K. Fortuniak, Chang. 8, 907–913 (2018). L. B. Flanagan, T. Friborg, M. Grygoruk, S. Gogo, T. Grünwald, B. U. Hansen, D. Holl, 37. K. Regina, A. Budiman, M. H. Greve, A. Grønlund, Å. Kasimir, H. Lehtonen, S. O. Petersen, E. Humphreys, M. Hurkuck, G. Kiely, J. Klatt, L. Kutzbach, C. Largeron, F. Laggoun-Défarge, P. Smith, H. Wösten, GHG mitigation of agricultural peatlands requires coherent policies. M. Lund, P. M. Lafleur, X. Li, I. Mammarella, L. Merbold, M. B. Nilsson, J. Olejnik, Clim. Policy 16, 522–541 (2016). M. Ottosson-Löfvenius, W. Oechel, F.-J. W. Parmentier, M. Peichl, N. Pirk, O. Peltola, 38. IPCC, Agriculture, forestry and other land uses, in 2006 IPCC Guidelines for National W. Pawlak, D. Rasse, J. Rinne, G. Shaver, H. P. Schmid, M. Sottocornola, R. Steinbrecher, Greenhouse Gas Inventories, National Greenhouse Gas Inventories Programme, T. Sachs, M. Urbaniak, D. Zona, K. Ziemblinska, ORCHIDEE-PEAT (revision 4596), a model H. S. Eggleston, L. Buendia, K. Miwa, T. Ngara, K. Tanabe, Eds, (IGES, 2006). for northern peatland CO2, water, and energy fluxes on daily to annual scales. Geosci. 39. UNFCCC, National Inventory Submissions 2019, Common Reporting Format (CRFs) tables Model Dev. 11, 497–519 (2018). (2019); https://unfccc.int/process-and-meetings/transparency-and-reporting/ 19. C. Qiu, D. Zhu, P. Ciais, B. Guenet, S. Peng, G. Krinner, A. Tootchi, A. Ducharne, A. Hastie, reporting-and-review-under-the-convention/greenhouse-gas-inventories-annex-i- Modelling northern peatland area and carbon dynamics since the Holocene parties/national-inventory-submissions-2019 [accessed 11 March 2021]. with the ORCHIDEE-PEAT land surface model (SVN r5488). Geosci. Model Dev. 12, 40. J. Alm, N. J. Shurpali, K. Minkkinen, L. Aro, J. Hytönen, T. Laurila, A. Lohila, M. Maljanen, 2961–2982 (2019). P. J. Martikainen, P. Mäkiranta, T. Penttilä, S. Saarnio, N. Silvan, E.-S. Tuittila, J. Laine, 20. K. Klein Goldewijk, A. Beusen, J. Doelman, E. Stehfest, New anthropogenic land use Emission factors and their uncertainty for the exchange of CO2, CH4 and N2O in Finnish estimates for the Holocene; HYDE 3.2. Earth Syst. Sci. Data 9, 927–953 (2017). managed peatlands. Boreal Environ. Res. 12, 191–209 (2007). Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 8 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE 41. B. Tiemeyer, A. Freibauer, E. Albiac, J. Augustin, M. Bechtold, S. Beetz, C. Beyer, M. Ebli, 63. R. W. Tiner Jr., Wetlands of the United States: Current Status and Recent Trends (United T. Eickenscheidt, S. Fiedler, C. Förster, A. Gensior, M. Giebels, S. Glatzel, J. Heinichen, States Fish and Wildlife Service, 1984). M. Hoffmann, H. Höper, G. Jurasinski, A. Laggner, K. Leiber-sauheitl, M. Peichl-brak, 64. J. Wolf, T. O. West, Y. L. Le Page, G. Kyle, X. Zhang, G. J. Collatz, M. L. Imhoff, CMS: Carbon M. Drösler, A new methodology for organic soils in national greenhouse gas inventories: Fluxes from Global Agricultural Production and Consumption, 2005–2011. ORNL DAAC, Data synthesis, derivation and application. Ecol. Indic. 109, 105838 (2020). Oak Ridge, Tennessee, USA (2015). 42. C. Evans, R. Morrison, A. Burden, J. Williamson, A. Baird, N. Callaghan, P. Chapman, 65. C. Le Quéré, R. M. Andrew, P. Friedlingstein, S. Sitch, J. Pongratz, A. C. Manning, A. Cumming, H. Dean, G. Dooling, J. Evans, V. Gauci, R. Grayson, Y. He, K. Heppell, J. Ivar Korsbakken, G. P. Peters, J. G. Canadell, R. B. Jackson, T. A. Boden, P. P. Tans, J. Holden, S. Hughes, J. Kaduk, R. Matthews, N. Menichino, T. Misselbrook, S. Page, O. D. Andrews, V. K. Arora, D. C. E. Bakker, L. Barbero, M. Becker, R. A. Betts, L. Bopp, G. Pan, M. Peacock, M. Rayment, L. Ridley, I. Robinson, D. Rylett, K. Stanley, F. Worrall, F. Chevallier, L. P. Chini, P. Ciais, C. E. Cosca, J. Cross, K. Currie, T. Gasser, I. Harris, Final report on project SP1210: Lowland peatland systems in England and Wale— J. Hauck, V. Haverd, R. A. Houghton, C. W. Hunt, G. Hurtt, T. Ilyina, A. K. Jain, E. Kato, Evaluating greenhouse gas fluxes and carbon balances (Centre for Ecology and M. Kautz, R. F. Keeling, K. Klein Goldewijk, A. Körtzinger, P. Landschützer, N. Lefèvre, Hydrology, 2016). A. Lenton, S. Lienert, I. Lima, D. Lombardozzi, N. Metzl, F. Millero, P. M. S. Monteiro, 43. FAO, FAOSTAT emissions database (2020); www.fao.org/faostat/en/#data/GC [accessed D. R. Munro, J. E. M. S. Nabel, S. I. Nakaoka, Y. Nojiri, X. Antonio Padin, A. Peregon, 11 March 2021]. B. Pfeil, D. Pierrot, B. Poulter, G. Rehder, J. Reimer, C. Rödenbeck, J. Schwinger, 44. R. S. Truskavetskii, Carbon budget of drained peat bogs in Ukrainian Polesie. Eurasian Soil R. Séférian, I. Skjelvan, B. D. Stocker, H. Tian, B. Tilbrook, F. N. Tubiello, I. T. V. Laan-Luijkx, Sci. 47, 687–693 (2014). G. R. V. Werf, S. Van Heuven, N. Viovy, N. Vuichard, A. P. Walker, A. J. Watson, 45. T. Gasser, L. Crepin, Y. Quilcaille, R. A. Houghton, P. Ciais, M. Obersteiner, Historical CO2 A. J. Wiltshire, S. Zaehle, D. Zhu, Global carbon budget 2017. Earth Syst. Sci. Data 10, emissions from land-use and land-cover change and their uncertainty. Biogeosciences 17, 405–448 (2018). 4075–4101 (2020). 66. FAO, Global Forest Resources Assessment 2015, Desk Reference (2015); 46. D. Gilfillan, G. Marland, T. Boden, R. Andres, Global, Regional, and National Fossil-Fuel CO2 www.fao.org/3/a-i4808e.pdf [accessed 11 March 2021]]. Emissions (2021); https://energy.appstate.edu/CDIAC [accessed 11 March 2021]. 67. J. Couwenberg, Emission Factors for Managed Peat Soils: An Analysis of IPCC Default Values 47. E. Dlugokencky, P. Tans, Trends in atmospheric carbon dioxide, National Oceanic & (Wetlands International, 2009). Atmospheric Administration, Earth System Research Laboratory (NOAA/ESRL) (2021); 68. S. O. Petersen, C. C. Hoffmann, C.-M. Schäfer, G. Blicher-Mathiesen, L. Elsgaard, www.esrl.noaa.gov/gmd/ccgg/trends/ [accessed 11 March 2021]. K. Kristensen, S. E. Larsen, S. B. Torp, M. H. Greve, Annual emissions of CH4 and N2O, 48. L. Elsgaard, C. Görres, C. Christian, G. Blicher-mathiesen, K. Schelde, S. O. Petersen, Net and ecosystem respiration, from eight organic soils in Western Denmark managed by ecosystem exchange of CO2 and carbon balance for eight temperate organic soils under agriculture. Biogeosciences 9, 403–422 (2012). agricultural management. Agric. Ecosyst. Environ. 162, 52–67 (2012). 69. M. Drösler, W. Adelmann, J. Augustin, L. Bergmann, C. Beyer, B. Chojnicki, C. Förster, 49. T. P. Kandel, P. E. Lærke, L. Elsgaard, Annual emissions of CO2, CH4 and N2O A. Freibauer, M. Giebels, S. Görlitz, Klimaschutz durch Moorschutz. Schlussbericht des from a temperate peat bog: Comparison of an undrained and four drained sites under Vorhabens" Klimaschutz-Moornutzungsstrategien" 2006–2010. Tech. Univ. München, V. permanent grass and arable crop rotations with cereals and potato. Agric. For. Meteorol. (Ed.), München, 201 (2013). 256–257, 470–481 (2018). 70. C. Beyer, H. Liebersbach, H. Höper, Multiyear greenhouse gas flux measurements 50. Y. Wen, H. Zang, Q. Ma, B. Freeman, D. R. Chadwick, C. D. Evans, D. L. Jones, Impact on a temperate fen soil used for cropland or grassland. J. Plant Nutr. Soil Sci. 178, 99–111 of water table levels and winter cover crops on greenhouse gas emissions from cultivated (2015). peat soils. Sci. Total Environ. 719, 135130 (2019). 71. S. Koszinski, B. A. Miller, W. Hierold, H. Haelbich, M. Sommer, Spatial modeling of organic carbon in degraded peatland soils of northeast Germany. Soil Sci. Soc. Am. J. 79, 51. M. Matysek, J. Leake, S. Banwart, I. Johnson, S. Page, J. Kaduk, A. Smalley, A. Cumming, 1496–1508 (2015). D. Zona, Impact of fertiliser, water table, and warming on celery yield and CO2 and CH4 72. M. Hahn-Schöfl, thesis, Gottfried Wilhelm Leibniz Universität Hannover (2015). emissions from fenland agricultural peat. Sci. Total Environ. 667, 179–190 (2019). 73. Å. Kasimir-Klemedtsson, L. Klemdtsson, K. Berglund, P. Martikainen, J. Silvola, O. Oenema, 52. P. Mäkiranta, J. Hytönen, L. Aro, M. Maljanen, M. Pihlatie, H. Potila, N. J. Shurpali, Greenhouse gas emissions from farmed organic soils: A review. Soil Use Manage. 13, J. Laine, A. Lohila, P. J. Martikainen, K. Minkkinen, Soil greenhouse gas emissions 245–250 (1997). from afforested organic soil croplands and cutaway peatlands. Boreal Environ. Res. 12, 74. K. Berglund, Ytsänkning på mosstorvjord: Sammanställning av material från Lidhult, 159–175 (2007). Jönköpings län (Sveriges lantbruksuniversitet, 1989). 53. H. V. Cooper, S. Evers, P. Aplin, N. Crout, M. Puat, B. Dahalan, S. Sjogersten, Greenhouse 75. J. Leifeld, M. Müller, J. Fuhrer, Peatland subsidence and carbon loss from drained gas emissions resulting from conversion of peat swamp forest to oil palm plantation. temperate fens. Soil Use Manage. 27, 170–176 (2011). Nat. Commun. 11, 407 (2020). 76. M. Maljanen, P. J. Martikainen, J. Walden, J. Silvola, CO2 exchange in an organic 54. J. Miettinen, A. Hooijer, R. Vernimmen, S. C. Liew, S. E. Page, From carbon sink to carbon field growing barley or grass in eastern Finland. Glob. Chang. Biol. 7, 679–692 source: Extensive peat oxidation in insular Southeast Asia since 1990. Environ. Res. Lett. (2001). 12, 024014 (2017). 77. M. Maljanen, A. Liikanen, J. Silvola, P. J. Martikainen, Methane fluxes on agricultural 55. M. Guimberteau, D. Zhu, F. Maignan, Y. Huang, Y. Chao, S. Dantec-Nédélec, C. Ottlé, and forested boreal organic soils. Soil Use Manage. 19, 73–79 (2003). A. Jornet-Puig, A. Bastos, P. Laurent, D. Goll, S. Bowring, J. Chang, B. Guenet, M. Tifafi, 78. M. Maljanen, V.-M. Komulainen, J. Hytönen, P. J. Martikainen, J. Laine, Carbon dioxide, S. Peng, G. Krinner, A. Ducharne, F. Wang, T. Wang, ORCHIDEE-MICT (v8.4.1), a land nitrous oxide and methane dynamics in boreal organic agricultural soils with different surface model for the high latitudes: Model description and validation. Geosci. Model Dev. soil characteristics. Soil Biol. Biochem. 36, 1801–1808 (2004). 11, 121–163 (2018). 79. M. Maljanen, J. Hytönen, P. Mäkiranta, J. Alm, K. Minkkinen, J. Laine, P. J. Martikainen, 56. G. Krinner, N. Viovy, N. de Noblet-Ducoudré, J. Ogée, J. Polcher, P. Friedlingstein, P. Ciais, Greenhouse gas emissions from cultivated and abandoned organic croplands in Finland. S. Sitch, I. C. Prentice, I. Colin, A dynamic global vegetation model for studies Boreal Environ. Res. 12, 133–140 (2007). of the coupled atmosphere-biosphere system. Global Biogeochem. Cycles 19, GB1015 80. A. Lohila, M. Aurela, J. Tuovinen, T. Laurila, Annual CO2 exchange of a peat field (2005). growing spring barley or perennial forage grass. J. Geophys. Res. Atmos. 109, D18116 57. B. D. Stocker, R. Spahni, F. Joos, DYPTOP: A cost-efficient TOPMODEL implementation (2004). to simulate sub-grid spatio-temporal dynamics of global wetlands and peatlands. 81. R. Morrison, A. M. J. Cumming, H. E. Taft, J. Kaduk, S. E. Page, D. L. Jones, R. J. Harding, Geosci. Model Dev. 7, 3089–3110 (2014). H. Balzer, Carbon dioxide fluxes at an intensively cultivated temperate lowland peatland 58. W. J. Parton, J. W. B. Stewart, C. V. Cole, Dynamics of C, N, P and S in grassland soils: in the East Anglian Fens, UK. Biogeosciences 10, 4193–4223 (2013). A model. Biogeochemistry. 5, 109–131 (1988). 59. C. Bader, M. Müller, R. Schulin, J. Leifeld, Peat decomposability in managed organic soils in relation to land use, organic matter composition and temperature. Biogeosciences 15, Acknowledgments: We would like to thank L. Elsgaard, S. O. Petersen, E. M. Veenendaal, 703–719 (2018). H. E. Taft, and D. L. Jones for providing the information about the history of agricultural used 60. S. Glenn, A. Heyes, T. Moore, Carbon dioxide and methane fluxes from drained peat soils, peatland sites. Funding: This work benefited from the French state aid managed by the ANR southern Quebec. Global Biogeochem. Cycles 7, 247–257 (1993). under the “Investissements d’avenir” programme with the reference ANR-16-CONV-0003 and 61. A. Säurich, B. Tiemeyer, A. Don, S. Fiedler, M. Bechtold, W. Amelung, A. Freibauer, Drained from the European Research Council Synergy project SyG-2013-610028 IMBALANCE-P. organic soils under agriculture—The more degraded the soil the higher the specific basal Simulations were performed using HPC resources from GENCI-TGCC (grant 2020- respiration. Geoderma 355, 113911 (2019). A0070106328). The data contributed by A.V.G.-S. and D.J.C were obtained with funds from 62. S. a. Marcott, J. D. Shakun, P. U. Clark, A. C. Mix, A reconstruction of regional and global the Natural Environment Research Council (NERC standard grant number NE/I012915/1). temperature for the past 11,300 years. Science 339, 1198–1201 (2013). Author contributions: P.C. and C.Q. designed the study. C.Q. developed the ORCHIDEE-PEAT Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 9 of 10 SCIENCE ADVANCES | RESEARCH ARTICLE model and performed the simulations. S.P., A.M.R.P., A.V.G.-S., D.J.C., and S.C.B. contributed the Submitted 18 October 2020 data. C.Q. and P.C. drafted the manuscript, and all authors contributed to the interpretation of Accepted 21 April 2021 the results and the draft revision. Competing interests: The authors declare that they have no Published 4 June 2021 competing interests. Data and materials availability: All data needed to evaluate the 10.1126/sciadv.abf1332 conclusions in the paper are present in the paper and/or the Supplementary Materials. The outputs from the model simulations are freely available for download at https://sharebox.lsce. Citation: C. Qiu, P. Ciais, D. Zhu, B. Guenet, S. Peng, A. M. R. Petrescu, R. Lauerwald, D. Makowski, ipsl.fr/index.php/s/pcdrF8sA250OAwU. Additional data related to this paper may be requested A. V. Gallego-Sala, D. J. Charman, S. C. Brewer, Large historical carbon emissions from cultivated from the authors. northern peatlands. Sci. Adv. 7, eabf1332 (2021). Qiu et al., Sci. Adv. 2021; 7 : eabf1332 4 June 2021 10 of 10