Agricultural soils hold potential for the expansion of carbon sequestration. With this in mind, we investigated changes in the soil organic carbon (SOC) on the basis of an analysis of data sets extracted from 146 publications and further projected the SOC sequestration potential in China's cropland. Our results suggest that a significant increase in the SOC occurred in east and north China, while a decrease appeared in northeast China. As a whole, the organic carbon density in the topsoil to 30 cm depth increased by 3.36 (2.54 to 4.26) Mg/ha between 1980 and 2000. Accordingly, the croplands in China that cover an area of over 130 Mha sequestered 437 (331 to 555) Tg C, with an average rate of 21.9 (16.6 to 27.8) Tg/yr, during this period. The potential of SOC sequestration in China was estimated to be 2–2.5 Pg C, which could be achieved by the 2050s if crop production and field management are improved.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Estimates from the IPCC Second Assessment Report suggested that 400–800 Tg C per year could be sequestered in global agricultural soils, with the finite capacity saturating after 50–100 years [Smith et al., 2007]. Lal [2004a] suggested that carbon sequestration in agrosystems has the potential to offset 5–15% of the global fossil-fuel carbon emissions. Smith  reported the biological potential for carbon storage in European (EU15) cropland to be 90–120 Tg C per year with the available management options including reduced and zero tillage, set-aside(i.e., land that farmers are not allowed to use for any agricultural purpose), and the more efficient use of organic amendments. The annual rate of C sequestration in cropland soils in the United States was estimated to increase by 39% from 1990 to 2006, due largely to an increase in annual cropland enrolled in the Conservation Reserve Program, the intensification of crop production, and the adoption of conservation tillage [U.S. Environmental Protection Agency, 2008]. Sperow et al.  further projected that U.S. cropland has the potential to increase soil C sequestration by an additional 60–70 Tg C/yr over the rate of 17 Tg C/yr in the early 2000s.
 China has more than 20% of the world population, while its arable land is only 7% of the world's total. An improvement in crop productivity is essential to meet the food demand of the increasing population in China, while inappropriate management practices such as crop residue removal and plow tillage will reduce soil carbon sequestration and/or result in carbon loss [West and Post, 2002; Lal, 2004b; Eynard et al., 2005]. Moreover, a low soil organic carbon (SOC) level depresses the stability of crop yields [Pan et al., 2009].
 Over the past 30 years, crop yields in China have approximately doubled as a result of the adoption of modern cultivars, the increased use of fertilizer, improved field management, and expanded irrigation [National Bureau of Statistics of China, 2009]. Enhanced crop production increases the amount of residue and root input into the soil, and thus likely causes an increase in carbon sequestration under appropriate field management conditions. Liao et al.  suggested that some 24 Tg C were sequestered by cropland soils (∼4.7 M ha) in the Jiangsu Province of China over a 20-year period. Xu et al.  reported that organic C in cropland soils increased in north China and decreased significantly in northeast China. The increase in SOC was mainly attributed to an improvement in crop production and residue management in the Jiangsu Province [Liao et al., 2009] and in north China [Xu et al., 2004], but the decrease in SOC was largely due to lower C input in northeast China [Xu et al., 2004]. Huang and Sun  estimated that as a whole, the cropland soils (118 M ha) in China sequestered 310–400 Tg C from 1980 to 2000. Similarly, Xie et al.  suggested a sequestration of 472 Tg C (156 M ha, including orchards) over the same period in China.
 Calculations with the data in the monographs of the China Soil Series Vols. 1–6, edited by the State Soil Survey Service of China [SSSSC, 1993, 1994a, 1994b, 1995a, 1995b, 1996], indicated that the average thickness of the cultivated layer (Ap horizon) of Chinese cropland was 17.5 ± 3.3 (mean ± SD) cm (n = 798) and that the sub-cultivated layer (P horizon) extended a further 11.5 ± 5.1 cm (n = 798). We hypothesized that the changes in the SOC pool occurred mainly in the Ap and P horizons (hereafter referred to as topsoil) in order to capture the majority of the SOC changes. The objectives of this study were to estimate the changes in SOC stocks up to topsoil of ∼30 cm depth in Chinese croplands over the period from 1980 to 2000, and to project the potential SOC sequestration in the future.
2. Materials and Methods
2.1. Database Compilation
 Data were extracted from 146 published papers that reported the SOC dynamics in the cultivated layers (those defined as the Ap horizon in the papers) from field surveys and from long-term field measurements in various regions of China taken in the late 1970s/early 1980s up to the early 2000s. The reported SOC dynamics were measured from more than 85000 soil samples (Data Set S1 and Text S1), representing paddy soil (Inceptisols) and upland soils (Mollisol, Ustochrepts, Aridisols, Alfisols, Udert, and Udalf, etc.).1 Observations from the model plots for high-yield cultivation were not taken into account in this study. All of the data sets met the following criteria: (i) the reference period coincides with the Second National Soil Survey (SNSS) that was completed in the early 1980s; (ii) the methods used to measure the SOC concentration and the bulk density are the same as those used in the SNSS; and (iii) the same sampling sites or surveyed areas are used as in the SNSS.
Figure 1 shows the spatial distribution of the soil sampling sites. In these field surveys, the intervals between the final and initial measurements are 6–25 years (Data Set S1 and Text S1). The intervals of 6–9 years, 10–15 years, 16–20 years, and 21–25 years accounted for 4.3%, 30.6%, 48.9%, and 16.1%, respectively, of the total measurements reported.
 All of the measurements of SOM contents reported in these papers were made using the potassium dichromate volumetric method [ISSAS, 1978; Schulz, 2002]. The SOM was then converted to the SOC by multiplying a constant of 0.58 [Evrendilek et al., 2004]. The soil bulk density was determined as a ratio of oven-dried mass to the field volume of the soil sample [ISSAS, 1978; Lu, 2000].
2.2. Estimates of the Changes in the SOC Stock
2.2.1. Calculation of the SOC Density
 The SOC density in a given horizon (DOC, Mg/ha) was calculated as [Pan et al., 2004]:
where SOC and BD are the soil organic C concentration (g/kg) and bulk density (g/cm3), respectively. A negative exponential relationship of BD = 1.377 × e−0.0048×SOC [Song et al., 2005] was used to estimate the values of BD when they were not reported in the literature (Data Set S1 and Text S1). H is the soil thickness (cm). Table 1 shows the soil thickness of croplands in different regions of China, which were calculated with the data from the SSSSC monographs [1993, 1994a, 1994b, 1995a, 1995b, 1996]. The δ2mm is the fractional percentage (%) of >2 mm gravel in the soils. Using the data from the China Soil Series Vols. 1–6 [SSSSC, 1993, 1994a, 1994b, 1995a, 1995b, 1996], δ2 mm was estimated to be 5.9 ± 0.7.
Table 1. Soil Thickness (cm) of the Cropland in Different Regions of Chinaa
 Since approximately 72% of the measurements (Data Set S1 and Text S1) were taken in the late 1970s/early 1980s and the late 1990s/early 2000s, we assumed that the changes in the SOC mainly occurred in the period of 1980–2000.
220.127.116.11. Calculation of the Changes in the SOC Density of the Ap Horizon
 According to equation (1), changes in the Ap horizon organic C density (ΔDOC_A, Mg/ha) over the twenty-year period were determined by:
where SOC, BD, H, and δ2mm have the same meaning as in equation (1). The subscripts F and R denote the measurements in the final and reference year, respectively. t is the time interval between the final and the reference year when the field measurements were taken (Data Set S1 and Text S1). The constant 20 with a unit in year is used to normalize to the 1980–2000 period, adjusting for different sampling periods at different sites (Data Set S1 and Text S1).
18.104.22.168. Calculation of Changes in the Topsoil Organic C Density
 Although we obtained a considerable number of data sets in the Ap horizon (Data Set S1 and Text S1), the data sets of the C changes in the P horizon were limited. Moreover, very few papers could be found that recorded the changes in the SOC in both the Ap and P horizons over the same time scale. Accordingly, we took three steps to determine the changes in the topsoil organic C density.
 First, we adopted the ‘space-for-time substitution’ method [Sun et al., 2009] to estimate the changes in SOC density in both the Ap and P horizons from the spatial data in the monographs of the China Soil Series Vols. 1–6 [SSSSC, 1993, 1994a, 1994b, 1995a, 1995b, 1996], which recorded detailed information on the SOC content, the bulk density, and the thickness of the Ap and P horizons taken at different sites across China. To use the ‘space-for-time substitution’ [e.g., Sparling et al., 2003; Zaitsev et al., 2006], soils of different types and stages of development at separate locations (‘space’) were identified to obtain a chronosequence of ages (‘time’). According to Sun et al. , changes in the SOC density were calculated by:
where ΔDOC (Mg/ha) refers to the SOC density in either the Ap or the P horizon, and OC is the average value of for total recorded spatial sites in a given area or for a given soil usage. was computed from the spatial data [SSSSC, 1993, 1994a, 1994b, 1995a, 1995b, 1996] using equation (1).
 Second, we determined the relationship between the SOC density changes that occurred in the topsoil and in the Ap horizon as suggested by Sun et al. :
where ΔDOC_T and ΔDOC_A (Mg/ha) are the sum of the SOC density changes in the Ap and P horizons and in the Ap horizon, respectively, and are estimated by using equation (3). k is a regression coefficient.
 Finally, the changes in the SOC density in the topsoil were computed by using the k values (equation (4)) and the detected SOC density changes in the Ap horizon (equation (2)).
2.2.3. Determination of the Lower and Upper Estimates of the ΔDOC
 Considering the wide variation in the thickness of the soil (Table 1), the potential lower and upper estimates of the ΔDOC_A from equation (2) were computed with H (average thickness) − SD (standard deviation of H) and H + SD (Table 1), respectively.
 Using k values and the ΔDOC_A estimated with equation (2), we calculated the changes in the topsoil SOC density (ΔDOC_T) under four scenarios including the lower estimates of the ΔDOC_A with k − SD (SD is the standard deviation of k), the lower estimates of the ΔDOC_A with k + SD, the upper estimates of the ΔDOC_A with k − SD, and the upper estimates of the ΔDOC_A with k + SD. The average amount of ΔDOC_T was then determined from the four calculations. The calculations of the lower estimates of ΔDOC_A with k − SD and the upper estimates of ΔDOC_A with k + SD correspond to the lower and upper estimates of ΔDOC_T, respectively.
2.2.4. Calculation of Changes in the SOC Stocks
 Changes in the SOC stocks were estimated by:
where is the acreage-weighted mean of in a given sector j. Sj is the acreage of the given sector. Replacing the acreage-weighted mean with the lower and upper values of we obtained the corresponding lower and upper estimates of ΔPOC.
 To achieve a level of confidence in the values obtained for ΔPOC, the was calculated in light of two bases, i.e., the administrative regions and the soil usage (paddy and upland soils in terms of soil order). The Sj was then accordingly expressed as the area covered by the sample in the given administrative region and the area of the given soil order.
 Regional means for the ΔDOC_A on a provincial scale were estimated by weighting the surveyed acreage in a given province (Data Set S1 and Text S1). Due to a lack of data on the SOC changes in the Hainan Province, the Tibet Autonomous Region, and the Yunnan Province, we used the average values of the ΔDOC_A in their adjacent regions to make estimates.
 According to the SSSSC , there are 62 soil orders for agricultural utilization in China, one for paddy soil and 61 for upland soils. Data from the field survey represent 23 upland soil orders (Data Set S1 and Text S1). We selected 16 soil orders to estimate the ΔDOC_A from the 23 upland soils. The area of the 16 upland soil orders accounts for 78.2% of the national total upland soil area [SSSSC, 1998]. The ΔDOC_A of the remainder of the 45 upland soils was assumed to have the same value as the acreage-weighted mean ΔDOC_A from the 16 upland soils.
2.3. Estimation of SOC Sequestration Potential
 We used a simple approach to estimate the SOC sequestration potential with time-dependent scenarios of C input and no-till (NT) practices by:
where ΔSOCP is an increment of SOC stocks in a given year. Cm, Cr, and P × Cs represent the annual C input from manure, roots, and the aboveground residues of crops, respectively. Cs refers to the quantity of annual aboveground residues and P is the proportion of Cs that is retained in the soil. F is the fraction of the area with NT practices. R1 and R2 are the conversion efficiencies of the input C into the SOC sequestration with NT and conventional tillage (CT) practices, respectively. Based on the calculations in this study (See section 4.2 for details) and on the works by Duiker and Lal  and Chen et al. , R1 and R2 were given the values of 0.1 and 0.08, respectively. The SOC sequestration potential was estimated by summing up the ΔSOCP over a significant time span.
3.1. Changes in the SOC Stocks in the Ap Horizon
3.1.1. Observed Changes in the SOC
 Approximately 79% of the total records showed an increase in the organic C concentration in the Ap horizon (Data Set S1 and Text S1). Compared with the reference year, the records of SOC concentrations lower than 15 g/kg, particularly those lower than 5 g/kg, decreased in the final sampling year (Figure 2). Moreover, the records of SOC concentrations higher than 15 g/kg increased (Figure 2), suggesting that the SOC concentration tends to increase as a whole. The changes in the SOC concentration (ΔSOC) follow a normal distribution with an arithmetic mean of 1.73 g/kg (Figure 3). Approximately 50% of the records with increasing SOC are centralized in the range of 1–5 g/kg.
3.1.2. Estimated Changes in SOC Stocks Based on the Administrative Region
 Using equation (2) and the corresponding sample area, the acreage-weighted mean ΔDOC_A was estimated to increase by 2.54 Mg/ha, with a range of 2.11 to 2.95 Mg/ha (Table 2). By extrapolating ΔDOC_A to the total area of the corresponding regions in China, the net increase in the SOC stock was estimated to be 329 Tg, with a range of 275 to 384 Tg in the Ap horizon (Table 2). A significant increase in the SOC stock occurred in east China (109–147 Tg), while a considerable C loss (45–67 Tg) was seen in northeast China (Table 2).
Table 2. Administrative Region Based Estimation of Changes in the Carbon Density (ΔDOC_A) and Carbon Stocks (ΔPOC_A) in the Ap Horizon of Chinese Croplands Over a 20 Year Perioda
Administrative Region of China
Acreage (× 104 ha)
The mean, lower, and upper estimates of ΔDOC_A were computed using equation (2) with H (mean of thickness), H − SD, and H + SD, respectively. The mean and standard deviation of H are given in Table 1. The mean, lower, and upper estimates of ΔPOC_A were computed as the product of total acreage and corresponding ΔDOC_A.
Acreage-weighted mean calculated with total acreage in different regions.
3.1.3. Estimated Changes in the SOC Stocks Based on the Soil Usage
 The acreages of paddy soils and upland soils specified by the surveyed region (Data Set S1 and Text S1) were 10.3 M ha and 16.8 M ha and accounted for approximately 35% and 17% of the total paddy and upland soils in China, respectively (Table 3).
Table 3. Soil Usage Based Estimation of Changes in the Carbon Density (ΔDOC_A) and Carbon Stocks (ΔPOC_A) in the Ap Horizon of Chinese Croplands over a 20 Year Period
 The acreage-weighted mean of the ΔDOC_A was estimated to increase by 2.28 Mg/ha with a range of 1.89 to 2.67 Mg/ha (Table 3). On average, the SOC density in the paddy soils increased more than that in the upland soils (Table 3). By extrapolating the ΔDOC_A to the corresponding area of soil usage for each soil order in China (Table 3), the SOC stock was estimated to increase by 296 Tg with a range of 245 to 347 Tg in the Ap horizon (Table 3). Significant increases in the SOC stock occurred in the paddy soils (68–96 Tg) and in the Fluvo-aquic soils (62–86 Tg), while black soils showed a considerable decrease in the SOC (22–33 Tg).
3.2. Estimated Changes in the Topsoil Organic C Stocks
 To estimate the changes in the topsoil organic C density (ΔDOC_T), the k values in equation (4) were determined using a linear regression. Table 4 shows the estimated k values in different regions and for the different soil usages.
Table 4. Parameters of the ΔDOC_T Correlated to the ΔDOC_A in Different Regions or Soil Usages in Chinaa
When estimated on the basis of administrative region, both the paddy and the upland soils were included. When estimated on the basis of soil usage, all of the paddy soils and the upland soils in croplands of China were included.
Central and south
 Using the k values (Table 4) and the ΔDOC_A (Table 2), the SOC density in Chinese croplands within the topsoil up to a depth of 30 cm was estimated to increase by 3.54 (2.85 to 4.26) Mg/ha when calculated on the basis of administrative region. Accordingly, the SOC stock was estimated to increase by 460 (371 to 555) Tg (Table 5). Similar to the changes in the Ap horizon C (Table 2), significant C sequestration occurred in east China (179 Tg) where the paddy soils account for 31% of the national total paddy, while approximately 69 Tg of carbon might have been lost in northeast China where black soils are dominant.
Table 5. Estimated Changes in the Soil Organic Carbon Density and Stocks of Chinese Croplands Between 1980 and 2000
Changes in SOC Density (Mg/ha)
Changes in SOC Stock (Tg)
Determined from four calculations of lower and upper estimates on the basis of administrative region and soil usage (see section 2.2.3 for details).
Determined from eight calculations on the basis of administrative region and soil usage (see section 2.2.3 for details).
 Using the k values (Table 4) and the ΔDOC_A in Table 3, the SOC density in the topsoil to a depth of 30 cm was estimated to increase by 3.18 (2.54 to 3.85) Mg/ha on the basis of soil usage, and the SOC stock increased accordingly by 414 (331 to 501) Tg (Table 5). The SOC density in the top 30 cm of the soil was estimated to increase by 2.98 Mg/ha in the upland soils and by 3.82 Mg/ha in the paddy soils. Accordingly, the SOC stock was estimated to increase by 299 Tg in the upland soils and by 114 Tg in the rice paddies.
 By combining the estimates based on the administrative region and the soil usage, the sequestration of organic C by croplands in China was calculated to be about 437 Tg from 1980 to 2000. The lower and upper estimates are approximately 331 Tg and 555 Tg (Table 5), respectively. According to the synthesized mean values of the ΔPOC_A and the ΔPOC_T in Table 5, about 70% of the sequestration occurred in the Ap horizon.
3.3. Projected SOC Sequestration Potential
 In order to estimate the future C sequestration in Chinese croplands (equation (6)), we assumed (1) that the increase rate of crop NPP would be 6 Tg C/yr from 2000 to 2050, equal to about half of that from 1961 to 1999 [Huang et al., 2007]; (2) that the proportion of aboveground residue retained in the field would increase from 40% in 2000 [Gao et al., 2002] to 90% in 2050, and the manure input would be maintained at a similar level (110 Tg C/yr) as in 2000 [Li et al., 2003]; and (3) that the NT practice would be extended to 50% of the national cropland in 2050. Detailed descriptions of these assumptions are given in section 4.3. Table 6 shows the scenario of C input and NT practice for 10-year intervals.
Table 6. Scenario of C Input and NT Practice in Cropland Soils of China
 Using equation (6), the annual rate of SOC sequestration was computed with the scenario in Table 6. The scenario in each year was estimated by linearly interpolating the value between two adjacent 10 years in Table 6. Changes in the topsoil (0–30 cm) C density were calculated by dividing the accumulated C sequestration by the total cropland area (130 Mha).
 The annual rate of soil C sequestration was estimated to range from 22.6 Tg in 2000 to 54.4 Tg in 2050 (Figure 4a), when the increase in organic C inputs and the proportions of residue retention and no-tillage occur (Table 6). The C sequestration rate increased nonlinearly with the expansion of the NT area, accounting for 56% of the annual C sequestration in 2050 (Figure 4a). Summing up the annual rate of C sequestration from 2000 to 2050, the C sequestration in the top 30 cm of agricultural soils of China was estimated to reach 1.86 Pg by 2050, when the organic C density will approach 49.3 Mg/ha (Figure 4b). Extrapolating the NPP scenario in Table 6 to the year 2060, but keeping the proportion of residue retention and NT practice at the same level as in 2050, the organic C density in the top 30 cm of the soil was estimated to be 53.6 Mg/ha, approaching the world mean of 53 Mg/ha [Batjes, 1999]. Accordingly, the total amount of C sequestration would be 2.42 Pg by 2060.
4.1. Uncertainties in the Estimation of the SOC Stock
 We extracted data from the literature (Data Set S1 and Text S1) to estimate changes in the SOC stock. Precise estimation of these changes relies on the spatial representation of the data and the thickness measured for the given soil horizons when the SOC concentration is determined.
 The data in Table 2 indicate that the surveyed areas reported in the literature (Data Set S1 and Text S1) account for 40% of the total cropland in north and northeast China, 56% of the total in east China, 51% of the total in central and south China, 48% of the total in northwest China, and only 26% of the total in southwest China. Due to the insufficient data for 45 of the upland soils, which account for 21.8% of the national total upland soils [SSSSC, 1998], we assumed that the changes in the SOC density in these soils were the same as the acreage-weighted mean ΔDOC_A, which covers 78.2% of the national upland soils. The spatial unevenness in the surveyed area and the assumed changes in the SOC density in the 45 upland soils would inevitably generate a large uncertainty in the SOC estimates, while quantifying this uncertainty appears difficult at present. It is expected that the uncertainty could be reduced when field measurements are conducted across a wider domain.
 Most of the literature (Data Set S1 and Text S1) did not report the thickness of the soil sampling, but claimed that the sampling was conducted in the cultivated layers. We took the scenarios of H (average thickness) − SD and H + SD to obtain the lower and upper estimates of ΔDOC_A (equation (2)). The SD was calculated from a large number of soil samples taken in Chinese agricultural soils [SSSSC, 1993, 1994a, 1994b, 1995a, 1995b, 1996]. Although the spatial variations of the soil thickness for the given soil horizons will result in uncertainty in the SOC stock estimates, the uncertainty can be appropriately captured using the lower and upper estimates of ΔDOC_A. With respect to the determination of the changes in the topsoil organic carbon stock, we used the lower estimates of ΔDOC_A with the lower k value (k − SD) and the upper estimates of ΔDOC_A with the upper k value (k + SD) to quantify the corresponding estimates of ΔDOC_T (see section 2.2.3 for details). In this case, the uncertainties in the estimates of the organic carbon stock in the topsoil might also be appropriately captured.
 The great majority of the changes in the SOC were thought to appear in the top 30 cm [Batjes, 1999; Smith et al., 2000a], although some changes have been noted to occur below 30 cm [Sá et al., 2001; Olson et al., 2005; Rudrappa et al., 2006]. The present estimates were dedicated to the top 30 cm of soil, and we do not know of the changes in the SOC below these depths. An overall count of the SOC up to a significant soil depth will be helpful in better understanding the changes in the organic carbon stocks of China's agricultural soils.
4.2. Contribution of Cropland Management to the Changes in the SOC
 Our results suggested that the average rate of SOC sequestration from 1980 to 2000 was 21.8 Tg C/yr with a range of 16.5 to 27.7 Tg C/yr in Chinese cropland. This result is comparable with Young's work, which estimated an average rate of 17 Tg C/yr from 1982 to 1997 [Young, 2003]. The SOC sequestration may be mainly attributed to an increase in crop production, and residue and manure management.
 More recent estimates suggested that the crop net primary production (NPP) in China increased from 312.3 ± 66.5 Tg C/yr in 1980 to 547.0 ± 118.8 Tg C/yr in 1999 [Huang et al., 2007]. The total NPP amounted to 9.07 Pg C over the two decades. Assuming that the ratio of the aboveground residue production to the economic yield was 1.5, and that the ratio of root to shoot was 0.1 [Huang et al., 2007], the amounts of aboveground residue and root mass were 4.95 Pg C and 0.82 Pg C, respectively. According to Gao et al. , about 40% of the aboveground residue was retained in the field, which corresponds to 1.98 Pg C. Thus, the aboveground residue and the root C input into the soils was about 2.8 Pg over these two decades. Li et al.  estimated that the amount of manure amendment into croplands was 110 Tg C/yr in China, which corresponds to 2.2 Pg over the 20-year period. The total input of organic C was thus about 5 Pg from 1980 to 2000. The 437 (331–555) Tg SOC sequestration that was detected (Table 5) accounts for 8.7% (6.6–11.1%) of the incorporated C, which is comparable with the work by Angela et al. , who suggested a conversion efficiency of 7.6% for residue-C into SOC.
 It is noteworthy that the cropland soils in northeast China have lost organic C over the 20-year period (Table 2), particularly in the Heilongjiang province (Data Set S1 and Text S1). A recognized reason for this loss is that the initial SOC concentration in this province is higher than that in other regions (Data Set S1 and Text S1), and that the lower crop residue input could not compensate for the SOC losses via respiration. Due to a long frost period [Chinese Academy of Sciences, 1990], the cropping system in the Heilongjiang province is single crop cultivation with one harvest per year. The average amount of crop NPP was 2.6 Mg C/yr from 1980 to 1999, approximately 37% of that in southeast China where two or three harvests over one year course are normal [Huang et al., 2007] thanks to a longer non-frost period [Chinese Academy of Sciences, 1990]. Runoff and soil erosion may also cause SOC losses in northeast China [Wang et al., 2002; Ren, 2004; Xu et al., 2004]. According to remote sensing images, the area of runoff and soil erosion was about 4.47 × 104 km2 in 1986 [Yue et al., 1999], while this value expanded to 7.43 × 104 km2 in 1999 [Wang et al., 2002].
4.3. SOC Sequestration Potential
 The present projection of the future carbon uptake (Figure 4) is based on the assumptions of C input and NT implementation (Table 6). Any inappropriate assumption would either exaggerate or minimize the projection.
4.3.1. NPP Scenario and C Input
 The crop NPP in China increased from 1960 to 1999 at a rate of approximately 12 Tg C/yr [Huang et al., 2007]. The increase was mainly attributed to the adoption of modern varieties and to the improvement of field management [Qiao et al., 1996; Zhou et al., 2007]. The annual genetic gain was reported to range from 0.48% to 1.23% for wheat yield [Zhou et al., 2007] and 1% for rice yield [Peng et al., 2000] between the 1960s and 1990s. The later-released maize hybrids in north China during the past 50 years produced higher grain yield compared to the earlier-released ones, mainly because the later-released hybrids could remain photosynthetically active when the older hybrids aged during grain-filling period [Ding et al., 2005]. Long et al.  estimated that the increase in solar radiation conversion efficiency could improve the yield potential of grain crops by ∼50%. The improvement of field management, such as increasing fertilizer application [Huang et al., 2007], matching temporal and spatial N supply with plant demand [Ju et al., 2009], and the expansion of irrigation systems [Wang et al., 2009] also promoted crop production in China.
 The increase in crop production in China has continued over the last several years. Using the data from National Bureau of Statistics of China , a linear regression analysis of grain yield (y) versus year (t) suggested a growth rate of 9.14 Tg/yr between 2000 and 2008 (y = 9.14t + 430.4, r2 = 0.70, n = 9, data not shown). According to the National Grain Production Development Plan: 2006–2020 [Ministry of Agriculture of China, 2008], grain yield in China is expected to potentially increase by 100 Tg/yr by 2020 relative to 2005, yielding an average rate of 6.67 Tg/yr. Using the dry matter fraction of the crop grain, the ratio of residue to grain yield, the ratio of root to shoot, and the C fractions of the grain yield and residues suggested by Huang et al. , an increase of 6.67 Tg/yr in grain yield would increase the crop NPP to 6.74 Tg C/yr, which is close to our present assumption (6 Tg C/yr; see Table 6).
 Wheat, maize, and rice are the main crops grown in China, accounting altogether for 89% of the total grain yield [National Bureau of Statistics of China, 2009]. We assumed that the crop NPP would increase by ∼55% from 2000 to 2050 (Table 6) for estimating the SOC accumulation. The grain yield of wheat, maize, and rice in China was 3.7 Mg/ha, 4.6 Mg/ha, and 6.3 Mg/ha in 2000 [National Bureau of Statistics of China, 2009]. In contrast, the current grain yield of wheat in Western Europe has reached 7.4 Mg/ha, and that of maize in Northern America has reached 9.6 Mg/ha (FAO, http://faostat.fao.org/), which is approximately twice that of the yield in China in 2000. In 1996, a program of super-hybrid rice breeding was launched in China. The grain yield of hybrid rice in field experiments has reached 11–16 Mg/ha in northeast, east, south, and southwest China with the implementation of this program [Cheng and Li, 2007]. It must also be pointed out that the harvest areas of grain crops accounts for 68% of the total harvest area in China [National Bureau of Statistics of China, 2009]. The remainder (32%) of the area that is planted in other crops (e.g., cash crops, oil-yielding crops, and vegetables) would further contribute to crop production, and thus residue retention. In this case, an addition 55% increase in the NPP (Table 6) would be possible by the year 2050.
 It has been well recognized that the increase in organic C input promotes SOC sequestration [e.g., Paustian et al., 2000; VandenBygaart et al., 2004]. However, the proportion of residue retained in the field was only approximately 40% in 2000 [Gao et al., 2002]. We assumed a 1% increase per year in the proportion of residue retention to estimate the C input from crop residues (Table 6). This proportion would reach 90% by 2050 (Table 6). According to Liu et al.  and the Agricultural Technology Extension Center of China [Zeng et al., 2001], it is technically possible to retain all of the crop straw in the soils.
4.3.2. NT Practice
Duiker and Lal  reported that the carbon conversion efficiencies of the residue C into the SOC stocks were 8% per year for plow till and 10% per year for no-till, suggesting that the conversion efficiency with no-till is 25% higher than that with plow till. Similarly, an eight-year field experiment conducted on a single cropping system by Chen et al.  indicated that no-till practices increased the SOC by 27% compared with conventional tillage. In addition, no-till with residue cover was proven to significantly reduce runoff and soil erosion [Jin et al., 2008; Wang et al., 2008], which will no doubt reduce the SOC loss.
 The conventional plow till has been dominant in China over the last half century. The no-till (NT) area in China was only 4.12 million ha in 2002 [Editorial Board of China Agricultural Machinery Yearbook, 2003], accounting for approximately 3% of the total cropland area. According to the Ministry of Agriculture of China, about 60% of the croplands in China are suitable for NT practices [China Agricultural Technology Extension, 2007]. These croplands are mainly distributed in southern China and the Yangtze River Basin with the double-cropping systems of rice–rice or rice–upland rotations, in the Huang-Huai-Hai Region with double-cropping systems of wheat–maize or wheat–soybean rotations, and in northwest and northeast China with single-cropping systems of maize, soybean, or rice [Xia, 2006].
 Recognizing the role of NT practices in crop production, as well as in soil and water conservation, the central government of China has made it a policy to increase compensation to farmers for agricultural machinery including NT planting machines and straw crushers [Ministry of Agriculture of China, 2008]. Available data (Ministry of Agriculture of China, http://www.agri.gov.cn/sjzl/nongyety.htm) have indicated that the NT planting machine and straw crusher numbers increased from 18.7 × 104 and 31.7 × 104 sets in 2001 to 33.4 × 104 and 54.6 × 104 sets in 2006, respectively. The Ministry of Agriculture of China projected that the NT practice would be extended to ∼17% of the total cropland by 2010 [China Agricultural Technology Extension, 2007]. Practicing NT on 50% of the total cropland by 2050 could be expected.
4.3.3. Comparison With Other Projections
 Using a coupled remote sensing- and process-based ecosystem model of CEVSA, Yan et al.  projected that practicing NT on 50% of the arable lands and retaining 50% of the crop residue on the soil would lead to an annual soil C sequestration rate of 32.5 Tg C/yr. With the same assumptions of the NT practice and residue management as above, the present projections suggest a SOC sequestration potential from 27.9 Tg C/yr in 2000 to 37.7 Tg C/yr in 2050 using equation (6) with the NPP scenario (Table 6), which is comparable with Yan et al. .
 By synthesizing data from field experiments, Lu et al.  estimated that the carbon sequestration potential in Chinese croplands planted with rice, wheat, and maize could reach 34.4 Tg C/yr when all of the crop straws are retained in the soil under the CT practice. Using equation (6) with the NPP scenario in Table 6, our projection suggests that the SOC sequestration potential would range from 36.8 Tg C/yr in 2000 to 52.1 Tg C/yr in 2050 when all crop straws are retained in the soil with full CT practice, which is higher than that of Lu et al. . This is not surprising because the harvest area of rice, wheat, and maize together account for only 54% of the total harvest area in China [National Bureau of Statistics of China, 2009].
 A model projection by Yan et al.  suggested that the NT practice with an increasing C input could not only promote the C sequestration rate but could also increase the time of C sequestration. Under the scenario of 50% crop residue retention and NT practice in 50% of cropland, the average rate of C sequestration was estimated to be 32.5 Tg/yr for 73 years, resulting in a SOC sequestration potential of 2.37 Pg [Yan et al., 2007]. When crop residues were completely retained in the soils and NT was practiced on all cropland, the C sequestration rate averaged 71.7 Tg/yr for 84 years, resulting in a SOC sequestration potential of 6.02 Pg [Yan et al., 2007]. Pan et al.  estimated an easily attainable SOC sequestration potential of 0.7 Pg under the present conditions, with an upper potential of 3.0 Pg in the paddy soils of China. Paddy soils make up 26% of China's total croplands.
 China's agricultural soils have been cultivated for a long period of time (5000 years in some areas of China) and may have lost from 30 to 50% or more of the antecedent soil organic C pool [Lal, 2004b]. The SOC density in the 0–30 cm topsoil layer of Chinese cropland is currently about 35 Mg/ha [Song et al., 2005], which is lower than other countries or other regions (e.g., Thailand [Pibumrung et al., 2008] and Europe [Smith et al., 2000b]). The SOC density in Chinese topsoil is also lower than the world mean of 53 Mg/ha (Area-weighted mean of ten agro-ecological zones in the world. Data source: Batjes ). Although the environmental factors controlling SOC vary in different regions of the world, the relatively lower SOC in Chinese cropland suggests a potential for SOC sequestration.
Lal [2004b] suggested that some of the depleted SOC pool in China can be re-sequestered through the restoration of degraded soils and through the adoption of the recommended management practices. He estimated the soil C sequestration potential in China to be 25–37 Tg C/yr within the 124.1 M ha of croplands up to 50 years [Lal, 2004b]. In light of our estimates, the average amount of soil C sequestration tends to be 36.8 Tg C/yr within 130 M ha croplands from 2000 to 2050 (Figure 4a), with an average rate of 0.28 Mg C/ha per year, which is lower than the rate of 0.36 Mg C/ha per year in non-Annex I countries [IPCC, 2000]. Furthermore, the carbon density of 49 Mg/ha in 2050 (Figure 4b) is still lower than the world mean of 53 Mg/ha. Referring to the world mean of SOC density and the SOC sequestration potential reported by Lal [2004b] and IPCC , the C sequestration potential of 2–2.5 Pg within China's cropland is most likely if substantial improvements are made in promotion of crop production and in management, including the reduction of residue removal, the expansion of no-till practices, and the control of runoff and soil erosion.
 A further analysis of the data sets extracted from Data Set S1 and Text S1 indicated that 131 of 161 records with initial SOC density lower than 40 Mg/ha showed increases in SOC, while only 10 of 18 records showed an increase when the initial SOC density was higher than 40 Mg/ha (Figure 5). This result further suggests that the soils with lower organic C trended to sequester C in general, but the risk of net C loss may increase with increasing SOC density when the sequestration of input C does not balance the carbon losses via soil respiration or other events such as runoff and soil erosion. In light of present estimates, the soil C density would exceed 40 Mg/ha in 2030 (Figure 4b). In this case, the adoption of the recommended management practices such as increasing the organic C input and practicing NT should be essential to maintaining a certain rate of C sequestration.
 The SOC stocks in China's croplands have changed over the period of 1980–2000. An increasing trend was observed in general, while a decrease appeared in northeast China. As a whole, an estimated 437 (331 to 555) Tg of organic carbon was sequestered in the top 30 cm of the soil in 130 M ha croplands over this period. Although the increase in SOC appears significant, the SOC density is currently much lower than that measured in neighboring countries, as well as that of the world mean. When crop production is improved and the recommended management practices such as reduction of crop residue removal and extension of no-till practice are adopted, a potential SOC sequestration by the 2050s of 2–2.5 Pg C was projected for Chinese cropland.
 This work was jointly supported by the Chinese Academy of Sciences (grant KZCX2-YW-Q1-15) and the Ministry of Science and Technology of China (grant 2008BAD95B12). We thank two anonymous reviewers and James Randerson who provided helpful comments that led to the improvement of this paper.