Plant and Soil 231: 175–185, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.

175

Organic carbon content and distribution in soils under different land uses in tropical and subtropical China Li Zhong1 & Zhao Qiguo Institute of Soil Science, Academia Sinica, P.O. Box 821, Nanjing 210008, China. 1 Current address: Northern Forestry Centre, Canadian Forest Service, 5320-122 Street, Edmonton, Alberta, T6H3S5, Canada∗ Received 26 May 1999. Accepted in revised form 3 August 2000

Key words: C density, C distribution, land use, soil, tropical and subtropical China

Abstract Tropical and subtropical China comprises a land area of about 215 Mha, but reports on its soil C storage and contents are limited. The objective of this study was to investigate the C density, stocks and distribution in soils of this region under different land uses by using soil species data from the Second National Soil Survey and the Vegetation Map of the People’s Republic of China (1:4 M). It was estimated that there is a total of about 28.7±8.2 Pg organic C stored in the upper 1 m of soils of the entire region. Changes of C content (C) with depth (D) were observed following a relationship of C = (c + D)/ (a + bD), or C = 1/(a + bD). Of the various patterns of land uses in the region, soil C density was generally higher in the west than in the east, and while small differences were found in croplands, there were large variations in natural soils. In the west, the C density of meadow and herbaceous swamp soil was the highest (about 40 kg C/m2 ), followed by coniferous and broad-leaf forest soils (19.6 and 19.2 kg C/m2 , respectively). The C density of paddy, bush and coppice forest soils showed a density of 12.6 and 14.6 kg C/m2 , respectively. Upland and grass-savanah soils ranked the lowest (9.4 and 10.5 kg C/m2 , respectively). In the east, meadow and herbaceous swamp soil had the highest C density (25.2 kg C/m2 ), but differences in C density among soils under coniferous forest, broad-leaf forest, bush and coppice forest, and rice were small, varying from 10.2 to 11.4 kg C/m2 . The C density of upland soil (7.2 kg C/m2 ), appeared a little higher than that of grass-savanah soil (6.3 kg C/m2 ). For the various land uses in the region, the C density estimation is accompanied by relatively large variations. Abbreviations: C – carbon; BC – bush and coppice forest; BF – broad-leaf forest; CF – coniferous forest; GS – grassland and savanah; MS – meadow and herbaceous swamp; PF – paddy field; UP – upland Introduction Soil organic C comprises a major pool in the global carbon cycle, and plays an important role in regulating the atmospheric CO2 concentration. Many attempts have been made to estimate the amount of C stored in soils of the world (Eswaran et al., 1993; Post et al., 1982; Sundquist, 1993), and an organic C stock of 1400–1600 Pg in the upper 1 m soils of the world seems to be accepted. Global estimates are often accompanied by quite a large coefficient of variation, and a more reliable estimate must treat many uncer∗ E-mail: [email protected]

tainties such as lack of measured C content, bulk density and stoniness, usage of regionally unbalanced databases and unconfined extrapolation of individual data (Arnold, 1995; Eswaran et al., 1995). Confined by scale, information on organic C drawn from studies on a global scale has quite limited implication locally. Therefore, regional studies on soil C content and storage can help narrow variations of the global estimates, and they are valuable in assessing the role of regional soils in the global C cycle. Globally and in the long run, soil organic C level is dependent on vegetation (Jenny, 1980), which is largely controlled by climate. Inherently, soil char-

176 acteristics play a critical role in soil C stabilization (Martin and Haider, 1986). Land conversion, human disturbances on natural ecosystems and cultivation practices can cause, in the short term, remarkable changes of soil organic C (Detwiler, 1986; Houghton et al., 1983; Tiessen et al., 1998), leading to C emission to the atmosphere or in some cases to C sequestration (e.g. afforestation). Tropical and subtropical soils are less studied than temperate soils in this regard and are of particular concern due to their diverse land uses, sensitivity to human impact and poor agricultural management (Lal and Logan, 1995). This research was intended to assess the organic C level, storage and changes with respect to the various land uses in tropical and subtropical China.

Materials and methods Data source and collection The area covered by Map of Soils in the Red Soil and Yellow Soil Region of China (1:4 M) (CGRY, 1985) was defined in this study as tropical and subtropical China, which matches roughly the common concept by Chinese soil scientists. According to this definition, it includes the whole provinces of Fujian (FJ), Jiangxi (JX), Hunan (HN), Guangdong (GD), Hainan (HI), Guangxi (GX), Guizhou (GZ) and Taiwan (TW), the most part of Zhejiang (ZJ), Yunnan (YN) and Sichuan (SC), and a small part of Hubei (HB), Anhui (AH), Jiangsu (JS) and Tibet (TB) (Figure 1). As there is a clear difference in climate, geography, land use and social economy (see next section), we divided the whole area into two sub-regions: East and west. Some islands in the studied region were excluded because of their small areas. The Vegetation Map of the People’s Republic of China (1:4 M) (Hou, 1982) was used as the ‘blueprint’ for landuse classification in this study. Based on the map and with a consideration of the main ecological types, we classified, by merging map units, land uses of the whole region into six categories: coniferous forest (from seven subordinates), broad-leaf forest (from 15 subordinates), bush and coppice forest (from 14 subordinates), grassland and savanah (from five subordinates), meadow and herbaceous swamp (from four subordinates) and agricultural land (including upland and paddy field). The area of each land cover was estimated by digitizing and summing the area of the relevant polygons.

The primary data employed in this study were based on the Second National Soil Survey, collected from provincial records of soil species and Soil Species of China Volume I, III and VI (National Soil Survey Office, 1993, 1994, 1996). Soil species, an important base classification unit in the Chinese Soil Taxonomy (National Soil Survey Committee, 1998), is defined as a soil association composed of a group of soil pedons with the same or similar characteristics in landscape, profile morphology, as well as in definite soil physio-chemical and biological properties. It occurs also as a relatively independent ecological unit with homogeneous vegetation, parent material, utilization pattern and identical hydrothermal conditions (Zhao et al., 1997). This makes soil species a solid base for land use identification and thereby links regional land cover to the Chinese soil survey. As the soil survey principally focused on soil itself rather than its vegetation, no vegetation/land use classification was intentionally made during the survey, and such a grouping in this study was made by authors based on each profile’s location (e.g. elevation), plant coverage, species and growth. Attached to the Vegetation Map of the People’s Republic of China (Hou, 1982), there was a clearcut illustration of how vegetation was classified, which was strictly followed to assign profile data for each category. To ensure accuracy, soil pedons without detailed vegetation information for classification were excluded. For a given soil species, when all the supporting pedons fell within the same pattern of land use, a statistical profile (averaged and provided in the survey) instead of a typical one was used. As a result, a total of 13 767 profiles were collected, of which as many as 11 141 pedons were agricultural soils. Pedogenetically, they fall within the 18 soil groups and 39 subgroups of the region. The Second National Soil Survey did not cover Taiwan province, but soil C density for different land uses in the east was correspondingly assigned to their counterparts in Taiwan to maintain integrality. Calculation of soil C density For an individual soil species (a typical or a statistical profile) with n layers, its organic C density (SOCd , in kg C/m2 ) was calculated as: SOCd =

n
i=1

Ti υi Ci (1 − δi )/10

(1)

177

Figure 1. Area of the study.

where Ti is the thickness (cm) of layer i, υi is the bulk density (g/cm3) of layer i, Ci is the C content (%) in layer i, and δi is the volumetric percentage of fragments > 2 mm. C content was converted from organic matter concentration using a conversion coefficient of 0.58 since organic matter was determined by wet combustion with Cr2 O7 2− (Wen, 1984). For soil species without recorded bulk density (about 70% of the total), the mean value of the corresponding subgroup was used. For those without recorded data on stoniness but with qualitative classification (as slight stony, medium stony or heavy stony), the median of 0–15%, 15–50% or 50–100% was assigned in light of its definition in the survey.

For a given land cover type supported by n profiles, its average C density (SOCL ) was computed by using: SOCL =

n


SOCi f (i)

(2)

i=1

where SOCi is the C density of profile i, and f (i ) is the area fraction of profile i in the total supported by n profiles. The total area of a landuse class obtained by simply summing the surveyed area of each soil species did not match the total obtained by digitizing the vegetation map, but the former was assumed to be a representative sample of the latter.

178 C density estimation of 0–20 cm and 0–100 cm soils In the survey, soil profiles were divided pedogenetically into horizon A, B, C, D. . . , so horizon A was not necessarily at a depth of 20 cm while profiles sometimes failed to reach 100 cm. To estimate C density of the upper 20 cm soil, the depth of the A horizon was used as the criterion to assign characteristics. When A·20 cm, a layer of 20 cm was assumed with its other characteristics assigned directly from A. In case of A<20 cm, a compensatory layer from B was taken with other properties of B being used. The 0–100 cm soil C density (SOC100) was estimated for each landuse type by using: SOC100 = SOCp + SOCp−100

(3)

where SOCp is the C density to depth of the actual profile, and SOCp−100 is the C density from profile depth to 100 cm. SOCp was calculated by using the classic layer-based method. To estimate SOCp−100for each land use, an empirical relationship between soil depth and C concentration was explored so that the pattern of C changes in horizon p-100 can be determined (see section below). And then, with the other characteristics of the lowest horizon (the horizon just above horizon p-100) being extrapolated to layer P100, SOCp−100 was approximated by employing the integral calculus approach. Throughout this study, area-weighted means were used when C density and profile depth needed to be averaged.

Results and discussion Categorized land uses and their distribution The principal feature of the Vegetation map of the People’s Republic of China (1:4 M) (Hou, 1982) was of the vegetation at the time when it was compiled. So, in regions which were sparsely populated as in Qingzang plateau in the west, where vegetation was still kept intact or only slightly impacted, the vegetation as given in the map was dominantly native vegetation. But in some areas of the densely populated east where existing plants were mainly agricultural (crops and fruits) and secondary vegetation, zonal natural vegetation remained only in small fragments in remote regions. Therefore, enlargement of the contours had to be made for some polygons in the east to ensure their appearance on the map. As a whole, the area of the

native vegetation in the east was overestimated while some others were definitely underestimated. Compiled by merging units of the vegetation map mentioned above, Figure 2 is a map showing the spatial distribution of the six categorized land use patterns in tropical and subtropical China. Based on digitizing computation, the total area of the entire region is about 215.2 M ha (tropical area accounts for less than 5%), of which 95.3 and 119.9 M ha are in the east and west subregions, respectively. In the east, coniferous forest, broad-leaf forest, bush and coppice forest, grass and savanah, meadow and herbaceous swamp, and cultivated land cover 18.5%, 9.9%, 47.3%, 0.8%, 0.1% and 22.2%, respectively, while in the west, they occupy correspondingly 17.7%, 15.5%, 45.3%, 2.9%, 0.7% and 17.4%. For cultivated land, paddy field and upland make up 18.6% and 3.6%, respectively, in the east and 9.8% and 7.6%, respectively, in the west (CAAS, 1992). The mean annual temperature and rainfall in tropical and subtropical China vary from 15 to 28 ◦ C and 1200 to 2000 mm, respectively. As the Pacific Sea is to the east and the west is shielded by the Himalayas, they are commonly higher in the east compared to the west, with higher rainfall occurred generally in the mountainous area (up to 3000 mm). Geographically, hills are the basic topography of the east while the west is mostly made up of plateaus and mountains (CGRY, 1985). These basic geographic and climatic conditions account in part for the difference in land uses in the two regions. On the other hand, extensive harvesting and land conversion to agriculture in the past has strongly disturbed a large part of the native vegetation. As a result, the contemporary pattern of land uses in tropical and subtropical China reflects the combined interaction between climate, vegetation and human activities (CGRY, 1985). Vertical distribution of organic C in soils The content of soil organic matter is the balance between plant material input and losses caused mainly by heterotrophic decomposition. In natural ecosystems, litter fallen to the soil surface and turnover of fine roots are the main pathways of organic C input. Because of the high inputs at the soil surface, soil organic C level generally declines with soil depth. Simple power and exponential functions have been proven applicable in describing the vertical distribution of C in profiles for some soils (Arrouays and Pelissier, 1994; Bennema, 1974; Bernoux et al.,

179

Figure 2. Categorized land uses and their distribution in tropical and subtropical China.

1998). However, more complex patterns of C accumulation have been identified in other soils (Elzein and Balesdent, 1995; Stone et al., 1993). Regression analyses of soils under different land covers in the two subregions show that the relationship between soil organic C content and soil depth can be described by using: C = (c + D)/(a + bD)

(4)

C = 1/(a + bD)

(5)

or:

where C is soil organic C content (%), and D is soil depth (cm). Figure 3 gives some fitted results by using both Equation (4) and Equation (5). With one more parameter, Equation (4) demonstrated more flexibility and stronger capacity, and in a number of cases when Equation (4) was supported by relatively large constants of a, b and c, results obtained were similar to Equation (5). This is because the variable D in this study rarely has values larger than 100 cm, and in the case of a relatively large parameter c (e.g. c

>>100), the D in the numerator of Equation (4) could be ignored, converting Equation (4) to Equation (5). Figures 4 and 5 present the fitted curves of organic C with depth in the two subregions (MS was not included) by using one of the two equations which gives the better fit. Table 1 lists values of the related parameters, standard errors (SE) and R 2 . In the east, correlation coefficients (R 2 ) varied from 0.26 to 0.58, averaged 0.37, and in the western part, the fitting for MS was supported by a relatively low value of R 2 , with that of the others ranging from 0.18 to 0.49, averaged 0.32. Judging from the R 2 and p values, we can see that in most cases, the two equations are acceptable in depicting the vertical distribution of organic C in soils of the region, but R 2 values are generally low. The low values of R 2 may reflect to some extent the existence of large spatial variations of C content even among soils under the same type of land use. The level of C in soils is controlled by a number of factors, rather than vegetation or land use type alone, so when climate, soil and vegetation coverage are known to be

180

Figure 3. Simulated distribution of organic C in some soils by using both Equation (4) and Equation (5) (a) bush and coppice forest, western (b) coniferous forest, eastern (c) upland, eastern (d) broad-leaf forest, western.

Figure 4. Fitted curves of soil organic C in the eastern part of the region.

181

Figure 5. Fitted curves of soil organic C in the western part of the region. Table 1. Values of a, b, c and R 2 of the fitted curves Land

a

b

R2

P valuec

– – – – – 15.119 6.352

0.44 0.58 0.26 0.26 0.46 0.34 0.32

<0.001 <0.001 <0.01 <0.01 <0.01 <0.001 <0.001

– – 2.182 4.775 16.921 – –

0.28 0.49 0.36 0.36 0.05 0.22 0.18

<0.01 <0.001 <0.001 <0.001 ns <0.05 ns

c

Estimated

SEb

Estimated

SE

Estimated

SE

East region CF 359 BF 117 BC 237 GS 259 MS 45 PF 893 UP 402

0.307 0.237 0.349 0.791 0.091 219.138 115.490

0.020 0.026 0.038 0.069 0.021 7.652 6.126

0.023 0.019 0.021 0.035 0.007 8.693 5.197

0.002 0.002 0.004 0.006 0.002 0.530 0.357

– – – – – 445.669 130.998

West region CF 318 BF 139 BC 248 GS 253 MS 106 PF 407 UP 709

0.210 0.174 0.073 4.265 9.914 0.430 0.463

0.021 0.020 0.319 1.484 1.913 0.029 0.027

0.014 0.015 1.225 3.111 0.373 0.013 0.015

0.002 0.002 0.081 0.262 0.118 0.002 0.002

– – 17.906 34.246 92.741 – –

use

na

a n: the number of data involved. b SE: standard error. c ns: no significant difference (p>0.05).

comparable, it is possible to establish a more satisfactory quantitative relationship, supported by much larger R 2 values between soil depth and C concentration. (Arrouays and Pelissier, 1994; Bernoux et al., 1998). Therefore, caution should be exercised when the regional empirical equations obtained here are to be used somewhere else.

The fitted curves indicate that under different land uses, C levels differ mainly in the upper horizon, especially in the upper 40 cm soils, and with increasing soil depth, this difference will gradually decline but will remain notable even to a depth of 100 cm. For instance, the soils under both coniferous and broadleaf forests have a higher carbon content than soils

182 associated with other land uses. On the other hand, soils under grass and savanah, a seriously degraded landuse pattern in the region, generally have the lowest carbon level. The cultivated soils are characterized by a slow C decrease with depth starting from a relatively low level at the topsoil. In contrast, the natural soils exhibit a sharp C decrease, especially in the upper 40 cm. Soil organic C in the upper 100 cm, especially in the upper 50 cm of soils, has drawn much attention because of its role in global C cycling (Batjes, 1992; Bouwman, 1990; IPCC, 1992). Carbon dating results show that subsoil C has usually a greater age than that in the surface soil both in the temperature and tropical zones (Jenkinson, 1975; Scharpenseel, 1993), giving strong evidence that the dynamics of subsoil C differs from that of its surface counterpart. But, much less attention has been focused on C stored in the subsoil, although a large amount of C has been identified in both mineral and organic soils (Batjes, 1996; Sombroek et al., 1993; Tarnocai, 1994). In tropical and subtropical China, quite a large part of the so-called ‘grassland and savanah’ was derived from forested land induced by intensive human disturbance (CGRY, 1985). Their C distribution pattern along with that of uplands show that strong human impacts can cause C losses not only in the upper 20 cm soil but in soils far below as well. It is suggested that more attention be given to deeper soil C in the context of global change. Organic C density of soils under different land uses Soil organic C content (%) as discussed in the previous section is a soil mass-based unit while soil C density is volume-based. The natural pedon of soil has structure and always includes non-soil components such as rocks and pebbles, especially in the lower horizon, so C density can more realistically reflect the amount of C stored in natural soils. Figure 6 presents C density of the 0–20 cm soils. It indicates that in both subregions, upland soils have the lowest C density (2.1 and 2.9 kg C/m2 , respectively, for the east and the west). On contrast, soils under meadow and herbaceous swamp rank the highest (9.4 and 9.0 kg C/m2 , respectively, for the east and west), with soil from other land covers ranging in between. It is also notable that soils under a given pattern of landuse appear generally higher in C density in the west, apart from meadow and herbaceous swamp soil, with absolute difference occurring significantly large in coniferous-forest, broad-leaf forest and grass-

savanah soils but relatively small in cultivated lands. As the depth of the A horizon occurs shallower than 20 cm in most soils and a compensatory layer from horizon B had to be taken, the mass of C in the upper 20 cm soils, on average, may have been underestimated as long as the methodology we used is concerned. The C density to 100 cm depth (Figure 7) is consistently higher in the west than the east. The extent of the difference depends on vegetation type. Small differences were again found in croplands while soils under natural vegetation reflected large variations. In the west, the C density of meadow and herbaceous swamp soil is the highest (about 40 kg C/m2 ), followed by that of soils under coniferous and broad-leaf forests (19.6 and 19.2 kg C/m2 , respectively), and subsequently the paddy soils and bush and coppice forest soils (12.6 and 14.6 kg C/m2 , respectively). Upland and grass-savanah soils rank the lowest (9.4 and 10.5 kg C/m2 , respectively). In the east, meadow and herbaceous swamp soil has the highest C density (25.2 kg C/m2 ), but differences among soils under coniferous forest, broad-leaf forest, bush and coppice forest and rice are quite small, varying from 10.2 to 11.4 kg C/m2 . The C density of upland soil (7.2 kg C/m2 ) in this region appears a little higher than that of the grass-savanah soil (6.3 kg C/m2 ) but no significant difference was found. Based on our study, the C density of coniferous and broad-leaf forests in the east is much less than that of the tropical wet forest (19.1 kg C/m2 ) reported by Post et al. (1982), whose tropical wet forest included tropical and subtropical rain forests as well as tropical and subtropical wet forests. However, our estimate is comparable with their estimate for tropical and subtropical moist forest (11.4 kg C/m2 ) and a little higher than their result for tropical and subtropical dry forest (9.9 kg C/m2 ). The C density of our western coniferous and broad-leaf forests roughly match the result of tropical wet forest by Post et al. (1982), but occur much higher than their estimates for the other tropical and subtropical forests. The C density given for tropical and subtropical woodland and savanna (5.4 kg C/m2 ) by Post et al. (1982) is slightly less than that of its counterpart in our study in the east but nearly 50% less than ours in the west. As Post et al. (1982) did not separate bush and coppice forest from the forest category, the soil C density of ‘forest’ might be underestimated if such a separation is thought to be able to provide more reasonable/detailed estimation. Schlesinger (1984) reported a C density of 7.9 kg C/m2 for upland soil, which lies within the range of ours for the

183

Figure 6. Soil organic C density in the upper 20 cm soils under different land uses.

two subregions. Kern (1994) estimated a value of 10.7 kg C/m2 for North American tropical and subtropical moist broad-leaf forest, which is slightly lower than that of our broad-leaf forest in the east, but obviously less than ours in the west. However, Kern’s result for the seasonal tropical dry woodland (11.2 kg C/m2 ) appeared higher than that of ours. Our results for forests in the east are lower than the estimates given by Brown et al. (1992) for Latin American tropical forests (13.6 and 14.9 kg C/m2 , respectively), but our west values are much higher. Because the vegetation classification criteria used differs between studies, it is difficult to conduct a detailed comparison of the results, yet we still find that most of the C density reported elsewhere fall in the range of ours between the east and the west. Spatially, and in the long run, soil C level is controlled by natural factors particularly climatic conditions (Adams, et al., 1990; Jenny, 1980), while human disturbances can cause considerable short term changes (Lal, 1976; Mann, 1986; Schlesinger, 1986). The difference in C density between the east and the west reflects the comprehensive effects of both natural and anthropogenic factors on soil C in tropical and subtropical China. Organic C storage Multiplied by the area occupied by a pattern of land use, the C density of a horizon or for a given depth can give the corresponding stored C mass. Table 2 shows that in soil of meadow and herbaceous swamp

soils, about 24% of the 0–100 cm soil C is held in the upper 20 cm, relatively lower compared to that of the others. Cultivated land (both upland and paddy soil) also occurrs low (30–32%) in that percentage. Soils under other land uses in the region show a fraction of 36–43% C stored in the upper 20 cm layer. Based on this study, in the upper 100 cm of soils in tropical and subtropical China, there is a total of about 28.7±8.2 Pg of stored organic C, of which 44% is distributed in the soils of bush and coppice forests, 38% in coniferous and broad-leaf forest soils and nearly 16% in cultivated lands as a whole. Various authors have estimated the amount of organic C in the soil of tropics. Kimble et al. (1990) estimated a value of 496 Pg C in the upper 100 cm soil, which is comparable to the estimate of 506 Pg C by Eswaran et al. (1993), however, Batjes (1996) reported a relatively lower result of 384–403 Pg C. The amount of organic C stored in the soil of tropical and subtropical China amounts to 5.7–7.5% of the world’s tropical total.

Conclusions Soil organic C concentration changes with land cover and is subject to human disturbances. Knowledge of soil C density under different land uses may aid in estimation of recent and future release or sequestration of soil C. Integrated with soil and vegetation maps, soil survey data provide a solid base to estimate regional C

184

Figure 7. Soil organic C density in the upper 100 cm soils under different land uses. Table 2. Soil organic C storage and allocation in the region Land use

CF BF BC GS MS PF UP Total

No. Of Profiles

(M ha)

1034 367 643 373 209 6750 4391 13767

38.89 28.02 99.43 4.30 0.84 29.50 12.62 213.59

Area % of the total 18.21 13.12 46.55 2.01 0.39 13.81 5.91 100.00

storage and distribution, but the lack of soil bulk density, insufficient depth of soil profiles (<1 m) and the limited number of soil pedons on natural soils (as compared to cultivated soils) are the main constraints for a more accurate estimation in tropical and subtropical China. Alternatively, the establishment of a mathematical relationship between soil C content and depth can help extend our estimation of C density to 1 m depth but the relationship is weakened by the large spatial variations of the soil itself. The pattern of the vertical distribution of C in this study showed that for different land uses, soil C content differences mainly occurred in the upper 40 cm layer, but in some cases a difference remained to 100 cm. Based on this study, the C density of soils under different land covers in tropical and subtropical China

0–20 cm (A)

C storage (Pg) 0–100 cm (B)

% of A in B

2.43±0.85 1.90±0.60 4.61±1.74 0.18±0.07 0.08±0.02 1.13±0.28 0.34±0.11 10.66±3.05

6.11±2.18 4.64±1.43 12.56±4.80 0.42±0.15 0.33±0.09 3.51±0.99 1.11±0.42 28.69±8.23

39.8 40.9 36.7 42.9 24.2 32.2 30.6 37.2

is comparable to its counterparts reported by others, and a regional total of about 28.7 Pg C storage was estimated. But as long as the methodology used is concerned, both the C density and storage to 1 m depth were overestimated to some extent.

Acknowledgements We thank Dr. Z. C. Cai and Dr. C. F. Xi for helpful suggestions to this study and two anonymous reviewers for many critical comments. This research was jointly founded by the National Natural Sciences Foundation of China (Grant no. 49631010), NKBRDF (G1999011805) and the President Foundation of the Chinese Academy of Sciences.

185 References Adams J M, Faure H, Faure-Denard L, McGlade J M and Woodward F I 1990 Increases in terrestrial carbon storage from the Last Glacial maximum to the present. Nature 348, 711–714. Arnold R W 1995 Role of soil survey in obtaining a global carbon budget. In Soils and Global Change. Eds. R Lal, J Kimble, E Levine and B A Stewart. pp 257–263. CRC/Lewis Publishers, Boca Raton, FL, USA. Arrouays D and Pelissier P 1994 Modeling carbon storage profiles in temperate forest humic loamy soils of France. Soil Sci. 157, 185–192. Batjes N H 1992 Organic matter and carbon dioxide. In A Review of Soil Factors and Processes that Control Fluxes of Heat, Moisture and Greenhouses Gases. Technical Paper 23. Eds. NH Batjes and EM Bridges. pp 97–148. International Soil Reference and Information Centre, Wageningen. Batjes N H 1996 Total carbon and nitrogen in the soils of the world. Eur. J. Soil Sci. 47, 151–163. Bennema J 1974 Organic carbon profiles in Oxisols. Pedologie XXIV, 2, 119–146. Bernoux M, Arrouays D, Cerri C C and Bourennane H 1998 Modelling vertical distribution of carbon in Oxisols of the western Brazilian Amazon (Rondonia). Soil Sci. 163, 941–951. Bouwman A F 1990 Exchange of greenhouse gases between terrestrial ecosystems and the atmosphere. In Soils and the Greenhouse Effects. Ed. AF Bouwman. pp 61–127. John Wiley and Sons, New York. Brown S, Lugo A E and Iverson L R 1992 Processes and lands for sequestering carbon in tropical landscape. Wat. Air Soil Pollut. 64, 139–156. CAAS (Institute of National Resources and Agricultural Regional Planning, Chinese Academy of Agricultural Sciences, and National Soil and Fertilizer Service, Ministry of Agriculture) 1992 Cultivated Land Resource Development and Utilization in China (in Chinese). Surveying and Mapping Publishing House, Beijing. 450 p. CGRY (Cooperative Group for Soil Improvement and Utilization Division in Red and Yellow Soil Regions) 1985 Soil Utilization and Amelioration Division in Red and Yellow Soil Regions (in Chinese). Agriculture Press, Beijing. 192 p. Detwiler R P 1986 Land use change and the global carbon cycle: the role of tropical soils. Biogeochemistry 2, 67–93. Elzein A and Balesdent J 1995 Mechanistic simulation of vertical distribution of carbon concentrations and residence times in soils. Soil Sci. Soc. Am. J. 59, 1328–1335. Eswaran H, Van den Berg E and Reich P 1993 Organic carbon in soils of the world. Soil Sci. Soc. Am. J. 57, 192–194. Eswaran H, Van den Berg E, Reich P and Kimble J 1995 Global Soil Carbon Resources. In Soils and Global Change. Eds. R Lal, J Kimble, E Levine and BA Stewart. Pp 27–43. CRC/Lewis Publishers, Boca Raton, FL, USA. Hou X 1982 Vegetation Map of the People’s Republic of China (1:4 m). Map Press, Beijing. Houghton R A, Hobbie J E, Melillo J M, Moore B, Peterson B J, Shaver G R and Woodwell G M 1983 Changes in the carbon content of terrestrial biota and soils between 1860 and 1980: A net release of CO2 to the atmosphere. Ecol. Monogr. 53, 235– 262. IPCC 1992 Climatic Change 1992: In The Supplementary Report to the IPCC Scientific Assessment. Eds. JT Houghton, BA Callander and SK Varney. Cambridge University Press, Cambridge.

Jenkinson D S 1975 The turnover of organic matter in agricultural soils. Reports Welsh Soils Discussion Group 16, 91–105. Jenny H 1980 The Soil Resource. Springer Verlag, New York. Kern J S 1994 Spatial pattern of soil organic carbon in the contiguous United States. Soil Sci. Soc. Am. J. 58, 439–455. Kimble J M, Eswaran H and Cook T 1990 Organic carbon on a volume basis in tropical and temperate soils. Trans. 14th Int. Congr. Soil Sci. 5, 248–253. Lal R 1976 No-tillage effects on soil properties under different crops in western Nigeria. Soil Sci. Soc. Am. Proc. 40, 762–768. Lal R and Logan T J 1995 Agricultural activities and greenhouse gas emissions from soils of the Tropics. In Soils and Global Change. Eds. R Lal, J Kimble, E Levine and B Stewart. pp 293–307. CRC/Lewis Publishers, Boca Raton, FL, USA. Mann L K 1986 Changes in soil carbon storage after cultivation. Soil Sci. 142, 279–288. Martin J P and Haider K 1986 Influence of mineral colloids on turnover rates of soil organic carbon. In Interactions of Soil Minerals with Natural Organics and Microbes. Eds. PM Huang and M Schnitzer. pp 283–304. SSSA, Madison, WI. National Soil Survey Committee 1998 Soils of China (in Chinese). China Agricultural Press, Beijing, p. National Soil Survey Office 1993 Soil Species of China Vol. I (in Chinese). China Agricultural Press, Beijing. 924 p. National Soil Survey Office 1994 Soil Species of China Vol. III (in Chinese). China Agricultural Press, Beijing. 744 p. National Soil Survey Office 1996 Soil Species of China Vol. VI (in Chinese). China Agricultural Press, Beijing. 880 p. Post W M, Emanuel W R, Zinke P J and Stangenberger A G 1982 Soil Carbon Pools and World Life Zones. Nature 298, 156–159. Scharpenseel H W 1993 Sustainable land use in the light of resilience to soil organic matter fluctuations. In Soil Resilience and Sustainable Land Use. Eds. DJ Greenland and I Szabolcs. CAB-International, Wallingford, UK. Schlesinger W H 1984 Soil organic matter: A source of atmospheric CO2 . In The Role of Terrestrial Vegetation in the Global Carbon Cycle. SCOPE 23. Ed. GW Woodwell. pp 111–127. Wiley and sons, New York. Schlesinger W H 1986 Changes in soil carbon storage and associated properties with disturbance and recovery. In The Changing Carbon Cycle, A Global Analysis. Eds. JR Trabalka and DE Reichle. pp 194–220. Springer Verlag. Sombroek W G, Nachtergaele F O and Hebel A 1993 Amounts, dynamics and sequestrations of carbon in tropical and subtropical soils. Ambio 22, 417–426. Stone E L, Harris W G, Brown R B and Kuehl R J 1993 Carbon storage in Florida spodosols. Soil Sci. Soc. Am. J. 57, 179–182. Sundquist E 1993 The global carbon dioxide budget. Science 259, 934–941. Tarnocai C 1994 Amount of organic carbon in Canadian soils. Trans. 15th Int. Congr. Soil Sci. 6a, 67–82. Tiessen H, Feller C, Sampaio E V S B and Garin P 1998 Carbon sequestration and turnover in semiarid savannas and dry forest. Climatic Change 40, 105–117. Wen Q X 1984 Study Methods of Soil Organic Matter (in Chinese). Agricultural press, Beijing. 318 p. Zhao Q G, Li Z and Xia Y F 1997 Organic carbon storage in soils of southeast China. Nutr. Cycl. Agroecosyst. 49, 229–234. Section editor: S. Recous