SEARCH

SEARCH BY CITATION

Keywords:

  • Soil organic carbon;
  • grain productivity;
  • land evaluation;
  • China

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

In this paper, we present an assessment of the content and effects of cropland soil organic carbon (SOC) on grain productivity at the national scale in China using a Web-based Land Evaluation System. Homogeneous 5 km × 5 km grid data sets of climate, crop, soil and management parameters were created and grain production in 2005 was simulated. Attempts were made to incorporate SOC into the land evaluation procedure and to quantify the potential effects of SOC deficiency on grain productivity. Results were statistically analysed and the modelling approach was validated. National cropland SOC maps were generated. At the national scale, the cropland SOC content averaged 1.20, 0.58, 0.41, 0.31 and 0.26% for the five 20-cm sections consecutively from the surface downwards. At the regional scale it tended to decline slightly from northeast (1.63%) to southwest (1.11%). On average, 64% of grain yield was lost due to SOC deficiency for the humid provinces and 7% for the arid and sub-arid ones. Soil management options are suggested based on the simulation results.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

In 2005, in China grain crops of rice, wheat, maize, millet, sorghum, etc. occupied 78% of the sown area of all food crops (i.e. grain plus pulse, root and tuber crops) and produced 88% of the total food production. The average unit yield of grain crops was 5187 kg ha−1 which was 566 kg ha−1 higher than that of food crops. The grains are the primary source of food and play the most important role in food security in China.

Climate, soil, crop performance and management practices are the four most important factors that influence grain productivity. Soil organic carbon (SOC) is among a range of soil characteristics essential for crop growth. Many aspects, especially geographical patterns and storage of SOC in China, as well as correlation to grain productivity, have been studied at the national (Wu et al., 2003b; c; Zhou et al., 2003; Tang et al., 2006) or regional scale (Liu et al., 2006). Although the SOC–grain productivity relationship was reported in many cases as positive (Fan et al., 2005), negative observations (Cai & Qin, 2006) were not negligible. We attempt to reveal the rationale by which the effect of SOC on grain productivity should be assessed in a comprehensive approach. In particular, this paper focuses on (i) the content of the cropland SOC in China and its spatial patterns, (ii) the role of SOC in grain productivity assessments and the way to integrate it into a broader process of land evaluation based on limitations and (iii) the effects of SOC on the productivity of grain crops, individually or collectively.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

Inventory of croplands

The ‘farmland’ and ‘mosaic cropping’ classes were extracted from the GLC2000 land cover map (Wu et al., 2003a) and put into the target map of cropland (Figure 1a). The latter was then projected from WGS84 to Lambert Azimuthal Equal Area coordinate system, and its spatial resolution was downscaled to homogeneous 5 km × 5 km. The spatial distribution of popular cultivars of rice (Oryza sativa), wheat (Triticum aestivum) and maize (Zea mays) is shown in Figure 1b.

image

Figure 1.  Map of cropland in China in 2000 (a) and spatial distribution of grain crops and cultivars (b).

Download figure to PowerPoint

Data manipulation

The study area was divided into 808 rows by 963 columns of grid cells, each of 25 km2. A unique serial number, running from 1 at the upper left to 778 104 at the lower right corner of the coverage, was assigned to every cell. This cell number was used as an index to store the climatic, soil, crop and management data in a SQL Server 2000 database for easy and fast access. The dataset collected and required by the Web-based Land Evaluation System (WLES) for model run is summarized in Table 1. The crop, soil and management parameters were first geo-referenced, projected and then rasterized to Arc/Info GRID format with 5 km × 5 km as the resolution. The climatic data set was derived from 374 stations (IAP & CDIAC, 1991; ISSAS & ISRIC, 1994; WMO, 1996) to provide a representative cover for the study area (Figure 2a). Due to the multi-source, multi-density nature of the point datasets, climatic parameters were interpolated to generate continuous surfaces of 5 km × 5 km cells using ordinary kriging (ESRI, 2001). Figure 2b shows the annual reference evapotranspiration (ETo), summed from monthly subdivisions that were calculated from these generated climatic surfaces. Spatial patterns of high-low ETo values were found to match climatic, ecological and topographical conditions.

Table 1.   Parameters required by the WLES
CategoryParameters
ClimaticTmax, Tmin, RH, P and frequency, daily sunshine duration, wind speed
CropName and cultivar, leguminity, photosynthetic group, leaf area index at maximum growth rate, harvest index, optimal rooting depth, sowing date, lengths of the crop cycle and phenological subdivisions
SoilParent material and type, soil structure, sand class, root-limiting layer, % clay-silt-sand, % coarse fragments, bulk density, pF-curve, % CaCO3, gypsum, CEC-soil or -clay, concentrations Ca2+, Mg2+, K+ and Na+, pH-H2O and pH-KCl, % SOC, EC and ESP
Factor inputsFertilizers and chemicals, agro-electrical power consumption, effective irrigation rate
image

Figure 2.  Distribution of weather stations (a) and grid surface of ETo (b).

Download figure to PowerPoint

The soil dataset was extracted and converted from the ISRIC-WISE database of 5′ × 5′ (Batjes, 1997, 2002a) based on an inventory of the FAO soil units in China (FAO, 1999) correlated to the revised Legend (FAO-UNESCO-ISRIC, 1990). WISE is intrinsically profile-based. A locally representative soil profile was linked at the very beginning by native experts, with each FAO soil unit within the original 0.5° × 0.5° grid cell of the Digital Soil Map of the World, on which the unique soil mapping unit number (SNUM) was based. The type and relative area of the component soil units, including the dominant, associated and included units, occurring at the centre of each 5 km × 5 km grid cell were identified and characterized using FAO composing rules FAO (1999). Priority had been given to local experts on estimating missing values. The median of all soil units within the same grouping was used as the last resort in filling parameter gaps where local knowledge was unavailable. The WISE data set was considered appropriate for SOC inventories at the national (Batjes, 2004, 2005, 2006), continental (Batjes, 2002b) and global (Batjes, 1996) scales. A grid cell was viewed as a virtual soil profile of five equal sections of 20 cm down to 1 m. Crop physiological and phenological data were arranged according to cropping system and agro-ecological zones. The crop ID was used to link crop parameters to crop rotations and the zone ID to attribute geographical coordinates to cropping systems.

Assessment of grain productivity

The quantitative assessment of grain productivity was conducted using WLES (http://weble.ugent.be) (Ye & Van Ranst, 2004; Ye et al., 2004) as the evaluation engine which uses a three-step, hierarchical, deterministic land evaluation model (Ye & Van Ranst, 2002), based for specific crops on the radiation regime (FAO, 1984), and the water-limited and land production potentials (Sys et al., 1991; Tang et al., 1992). The evaluation process was looped to iterate all the 778 104 grid cells with data retrieved from and results saved in the database. The potential productivities were then analysed and mapped.

The average productivities of rainfed (Yrain) and irrigated (Yirri) grains were assessed using the following equations, respectively:

  • image(1)
  • image(2)

where Bn is the net biomass, HI harvest index, fW yield reduction coefficient due to water stress, Sy soil index and My management index. To calculate Sy, crop-specific soil requirement tables (Sys et al., 1993) were used to obtain rating values of (i) CEC-soil, SOC and the more limiting one between pH-H2O and exchangeable Σ(Ca2++Mg2++K+) for regions of LGP ≥ 120 days (FAO, 1996), or (ii) CaCO3, gypsum and the more limiting one between EC and ESP for regions of LGP < 120 days where LGP is length of growing period. The soil requirements for summer maize, for example, are given in Table 2. The Sy was achieved as the product of the rating values:

  • image(3)

where R is the rating of a soil parameter with a value 0–100.

Table 2.   Soil requirements of summer maize
 10095856040250
  1. Parent materials: (1) kaolinitic; (2) non-kaolinitic, non-calcareous; (3) calcareous. acmol(+) kg−1 clay. bBase saturation. ccmol(+) kg−1 soil.

CaCO3 (%)0615  2535>55
Gypsum (%)024  1020>20
CEC-claya>242416<16(–)<16(+)
BSb (%)>808050  3520<20
(Ca + Mg + K)c>8853.5 2<2
pH-H2O6.66.25.85.5 5.2<5.2
6.67.07.88.2 8.5>8.5
SOC (%)
(1)>2.02.01.20.8<0.8
(2)<1.21.20.80.5<0.5
(3)>0.80.80.4<0.4
EC (dS m−1)0246 812>12
ESP (%)08152025>25

The My was assigned to a crop in relation to a particular level of factor inputs (Table 3) which was defined on the basis of the overall scores from correlation with factor inputs (Table 4). The actual grain productivity (Y) was therefore simulated by combining equations (1) and (2) using the effective irrigation rate, r:

  • image(4)

where r is expressed as a percentage (%) of the area of cropland with permanent irrigating infrastructures over the total area of cropland.

Table 3.   Management index in relation to input levels
Crop groupMyExample crop
Ha-inputIa-inputLa-input
  1. aInput levels: H = high; I = intermediate; L = low.

I1.000.650.45Maize
II1.000.550.30Cotton
III1.000.700.50Wheat
IV1.000.750.60Rice
Table 4.   Definition of input levels based on factor input scores of economic-development belts
Belt ProvincesaFertilizers and chemicalsMachinery and electricityIrrigation and infrastructureOverall scoreInput level
  1. aRefer to Table 5 for province names.

East1–3, 6, 9–11, 13, 15, 19–21, 31–331.415.441.648.00H
Central4–5, 7–8, 12, 14, 16–182.201.361.325.00I
West22–301.001.001.003.00L

Results and discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

Model validation: simulated versus observed yields

The province-specific averages of yields of grain crops were calculated after the grain yields had been simulated on a per cell basis. The weighted mean of grain productivity (Table 5) was then estimated by applying the actual sown structure of grain crops and cultivars in 2005.

Table 5.   Province-specific observed and simulated yields of grain, individually and collectively (kg ha−1)
IDProvinceRice(Oa)Rice(Sb)Wheat(O)Wheat(S)Maize(O)Maize(S)Grain(O)Grain(S)
  1. aO = Observed. bS = Simulated. cNaN = Not a number.

 1Beijing6250NaNc517921194652659247795264
 2Tianjin8102NaN478532115059618350585093
 3Hebei566512 626487345614401870743576916
 4Shanxi4231NaN365412085614393142222969
 5Nei Monggol67375955263912075658299845692684
 6Liaoning73788207432027506753717166537105
 7Jilin72926718298322236238569463595807
 8Heilongjiang71176326325518264311431452474860
 9Shanghai80057644365361616191NaN70907349
10Jiangsu7919942642956805556710 54061858485
11Zhejiang66815723319337134128NaN63145586
12Anhui60678779383657264844897549577480
13Fujian55396118306533793360720454296131
14Jiangxi52136871151851593333NaN51776858
15Shandong7283NaN533855126107939757047270
16Henan704412 31251097065433910 24549818385
17Hubei75487462292443585010675762356714
18Hunan61497803191659134579995859237899
19Guangdong52515795283324454068617451695801
20Guangxi47685345162423463002410143915087
21Hainan43983102NaN15723431340843563114
22Sichuan72137156321623894808526053875308
23Guizhou665738021789 8934726276146732695
24Yunnan588747382240 8483831155341332558
25Xizang5455NaN6429NaN4849NaN5299NaN
26Shannxi59676853356012483886255137272170
27Gansu795950112917 557502415643410 998
28QinghaiNaNNaN3630 748750013883560 867
29Ningxia8152NaN288212156264141842371323
30Xinjiang5883NaN513831346979420858643569
31Taiwan49747832NaN3994NaN770949747832

Yield difference analysis.  A visual check of the mean observed and simulated yields, shown as box-and-whisker plots (a, c, e, g) in Figure 3, and the goodness of fit between them, represented by scattered plots against the 1:1 line (b, d, f, h), reveals that the productivities of rice (a, b) and wheat (c, d) are closely simulated whereas the productivity of maize (e, f) is overestimated. The fitted line (f) suggests that the overestimation probably occurred in the croplands with a yield of 5 t ha−1 or higher. Nevertheless, the observed-simulated differences in productivity of all grain crops (g) are still acceptable as the underestimation counterbalances the overestimation (h) to a great deal.

image

Figure 3.  Comparability of the means and goodness of fit between simulated and observed yields.

Download figure to PowerPoint

Statistical analyses.  The equality of the means of the simulated and observed yields of grain was statistically tested using the paired t-test (R Development Core Team, 2006) with the yield data grouped by province. Results (Table 6) confirm that the simulated yields have the same means as the observed ones, either individually (as rice, wheat, maize) or collectively (as grain). In other words, results from the simulation proved to be statistically representative of reality and vice versa.

Table 6.   Results of Shapiro–Wilk, F- and paired t-tests on simulated and observed yields
CropShapiro–Wilk test (H0 = normal distribution)F-test (H0 = same variance)Paired t-test (H0 = same mean)
ObservedSimulated
W (p)H0W (p)H0F (p)H0| t | (p)H0
  1. Significance levels: (.) <0.1; (*) ≤0.05; (**) ≤0.01; (***) ≤0.001; T, true; F, false.

Rice0.95 (.)T0.99 (.)T0.49 (.)T0.27 (.)T
Wheat0.93 (.)T0.92 (.)T0.50 (.)T0.90 (.)T
Maize0.98 (.)T0.94 (.)T0.15 (***)F0.79 (.)T
Grain0.98 (.)T0.94 (.)T0.16 (***)F0.11 (.)T

Spatial patterns of cropland SOC

Cropland SOC was mapped for the topsoil (0–20 cm) and the four subsoil sections (20–100 cm) to demonstrate its vertical and lateral distributions (Figure 4). At the national scale, cropland SOC content averaged 1.20, 0.58, 0.41, 0.31 and 0.26% for the 0–20, 20–40, 40–60, 60–80 and 80–100 cm sections, respectively. The top-section contained twice more SOC than the first sub-section and 3 times more than the average of all sub-sections. The decreasing rate of SOC values over depth was 3.1, 0.9, 0.5 and 0.3% per meter between two adjacent depth ranges, counted consecutively from surface downwards. On average, SOC decreased at a rate of 1.2% per meter over 1 m depth.

image

Figure 4.  Spatial distribution of cropland SOC (%) in China.

Download figure to PowerPoint

At the provincial scale, the SOC of the topsoil varied greatly from one province to another, with values ranging from 0.7% in the loess dominant province of Gansu in the northwest to 1.8% in the ‘black soil’ province of Hei Longjiang in the northeast. In general, high SOC values of the uppermost depth range were found in north-eastern provinces and lower ones in south-western ones. This is similar to patterns previously reported (Wu et al., 2003b; Tang et al., 2006). Although north-western provinces had the lowest values (0.8%), they hardly influenced the national pattern due to the scarcity of cropland in the region. The cropland SOC content tended to decline slightly from the northeast (1.63%) to the southwest (1.11%). The subsoil followed the same patterns as the topsoil (Figure 4). Such spatial patterns correlated with climatic patterns (Wu et al., 2003c). In the southwest soil carbon decomposed at a much higher rate than it accumulated.

Potential effects of SOC on grain productivity

Relationships between a range of selected soil characteristics (CaCO3, gypsum, CEC, BS, pH, EC and ESP) and crop productivity were cross-analysed in an attempt to identify the potential effects of cropland SOC on grain productivity.

Soil characteristics–grain productivity regression analyses.  Single factor linear regression models were applied to reveal the relationships between a soil property, for example SOC, as the factor and the grain productivity as the dependent. Results show that variations in yields are barely explained by SOC contents alone. In the best case, only 37% of the variation in grain yield is explained by variation in subsoil SOC values. The goodness of fit is not better for any other soil characteristic. It is clear that soil interacts with crops as a whole and no single soil characteristic alone correlates well with crop yields.

Soil characteristics–grain productivity ANOVA.  The variances of the productivities of grain crops were analysed against the variances of the soil characteristics in trying to relate the variations in yield to soil values. Analysis of variance (ANOVA) was used for this. Results show that the variation in cropland SOC values indeed explains the variation in the overall yield of grain. However, it does not explain yield variations for any individual crop. This applied to all soil properties. But all soil characteristics taken together are capable of explaining variations in yields of any crop, individually or collectively. This suggests that SOC should not be separated from other soil characteristics in the soil limitations evaluation (equation 3). Instead, the limiting soil characteristics should be treated as a whole.

SOC suitability to grain production

Figure 5a shows the weighted average of the soil indices of grain crops (Figure 1b) in China in 2005 and Figure 5b shows the suitability classes of cropland SOC to grain production in 2005. The maps can be compared to show, for example, that soils with high indices, as indicated by dark shading in Figure 5a, are more productive than soils with lower indices. It is thus possible to ascribe agricultural importance at the national scale to regions such as the northeast, north and Sichuan Basin in terms of grain production. Similar patterns were also observed in Figure 5b. SOC appeared to be suitable (S1) to moderately suitable (S2) in most of the croplands in China. The dominance of the marginally suitable (S3) and the actually unsuitable but potentially suitable (N1) classes in the southeast reflect the decreasing agricultural importance of the region although it was one of the originating localities for China’s traditional agriculture (Chang et al., 2002). Anthropogenic influences on soil quality, especially the impact of short-term human activities in this economically booming region, may account for these unfavourable SOC classes (Chen, 2003).

image

Figure 5.  Cropland SOC suitability classes for grain production (a) and average soil index (b).

Download figure to PowerPoint

Relative yield loss due to SOC deficiency

In general SOC was found more limiting in provinces with a humid climate (northeast and southeast) and less limiting in arid and sub-arid climates (northwest); on average 64% of grain yield was lost due to SOC deficiency for the humid provinces and 7% for the arid and semi-arid ones. The loss applies to 30% of the whole country (Table 7).

Table 7.   Relative yield loss (%) due to SOC deficiency
ProvinceaYield loss (%)
  1. aRefer to Table 5 for province names.

 10.52
 25.20
 31.39
 40.09
 51.59
 63.31
 71.07
 80.01
 967.14
1046.39
1162.70
1242.65
1364.00
1460.25
152.01
1612.60
1755.50
1864.48
1969.83
2068.17
2168.50
2258.47
2362.96
2448.04
25
264.40
273.21
2844.09
298.13
304.72
3174.63

Uncertainties

Scalability and compatibility of data sets are the main sources of uncertainty in this study. Significant improvements have been made on the quality of the WISE database by adopting additional locally representative soil profiles (Batjes, 2002a); for example, the number of profiles increased by 14% between WISE-1 and -2 and the inclusion of more recently created reference profiles is crucial in coping with uncertainty in derived soil parameters. Efforts are needed in applying procedures similar to SOTER (FAO, 1995) for creating parameter and horizon homogeneous datasets out of the results of the second national soil survey in China (NSSO, 1998). However, the reliability, consistency and spatial configuration (Figure 2a) of the source points contribute more to the quality of the resulting datasets—as confirmed by the generated climatic surfaces (Figure 2b)—than the quality of the source points above a certain threshold.

Results of SOC inventories are mostly static in space and time. Monitoring changing trends is the key to controlling uncertainties as well as stimulating appropriate management practices. The relatively high SOC in the northeast, for instance, may hide the fact that SOC there is being lost from soils at a very high rate (Wu et al., 2003b; Tang et al., 2006), probably due to anthropogenic factors (Chen, 2003), especially land use (Wu et al., 2003c). Uncertainty also comes from downscaling of cropping system and management parameters from regional to field scales (Anderson et al., 2003). With downscaling it is common and pragmatic to assume parametric homogeneity and to aggregate simulated yields back to regional scale for validation (Hatfield, 2001). In practice this is due at the regional scale to the lack of homogeneous cover of field-scale parameters, such as effective irrigation rate. Confidence in the overall reliability of the modelling approach was largely based on statistical tests on the final results. Confidence in intermediate parameters such as the potential effects of SOC on productivity was assumed to fluctuate but this needs data support. Further research is needed to verify this under typical agro-ecological conditions as present in the Chinese croplands; this could possibly lead to the removal of the word ‘potential’ in this context.

Soil management options

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

The following management options are suggested based on the SOC inventory made in this study:

  • 1
    Balanced fertilizer application. China is currently the biggest producer, importer and consumer of manufactured fertilizers in the world. Fifty per cent of increased grain productivity was attributed to fertilizers (National Soil Survey Office (NSSO), 1998). Balanced application of organic and manufactured fertilizers is currently the most feasible means of addressing food security and soil quality issues (Fan et al., 2005; Cai & Qin, 2006).
  • 2
    Promotion of cover crops and use of crop residues. Removal of crop residues was found to deplete SOC over the long-term (Chen, 2003), particularly in the arid and semi-arid north and northwest where the accumulation rate is relatively low (Wu et al., 2003c). This is the first but most important step to return the organic matter to soil.
  • 3
      Adoption of reduced tillage and systematic control of soil and water erosion, especially in the northwest.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References

The methodologies applied in this large-scale quantitative assessment of grain productivities in China proved to be successful and efficient. Statistical validation procedures suggested a close match between simulated and observed yields of grain crops. Regression and variance analysis of grain yields against soil characteristics revealed that yields of grain crops were not significantly correlated with cropland SOC alone. Data on limiting soil characteristics should be collectively evaluated in order to derive a single soil index. The uncertainty analysis showed that the quality of large climatic and soil data sets is largely dependent on the reliability, consistency and spatial configuration of the source points once a threshold number of input source points is available. The croplands of the north-eastern provinces have higher SOC values than those in the southwest. The relative loss in grain yield due to SOC deficiency varied greatly from 7% for the arid and semi-arid regions to 64% for the humid regions, with an average of 30% for the country. Proper soil management practices with special emphasis on cropland SOC are of great importance with respect to conservation and use of the soil resources in China, especially in the SOC-depleted southeast and in the erosion-prone northwest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Soil management options
  7. Conclusions
  8. References
  • Anderson, M.C., Kustas, W.P. & Norman, J.M. 2003. Upscaling and downscaling – a regional view of the soil–plant–atmosphere continuum. Agronomy Journal, 95, 14081423.
  • Batjes, N.H. 1996. Total carbon and nitrogen in the soils of the world. European Journal of Soil Science, 47, 151163. DOI:10.1111/j.1365-2389.1996.tb01386.x.
  • Batjes, N.H. 1997. A world dataset of derived soil properties by FAO-UNESCO soil unit for global modelling. Soil Use and Management, 13, 916. DOI:10.1111/j.1475-2743.1997.tb00550.x.
  • Batjes, N.H. 2002a. Revised soil parameter estimates for the soil types of the world. Soil Use and Management, 18, 232235. DOI:10.1111/j.1475-2743.2002.tb00244.x.
  • Batjes, N.H. 2002b. Carbon and nitrogen stocks in the soils of Central and Eastern Europe. Soil Use and Management, 18, 324329. DOI:10.1111/j.1475-2743.2002.tb00248.x.
  • Batjes, N.H. 2004. Soil carbon stocks and projected changes according to land use and management: a case study for Kenya. Soil Use and Management, 20, 350356. DOI:10.1111/j.1475-2743.2004.tb00380.x.
  • Batjes, N.H. 2005. Organic carbon stocks in the soils of Brazil. Soil Use and Management, 22, 2224. DOI:10.1111/j.1475-2743.2005.tb00102.x.
  • Batjes, N.H. 2006. Soil carbon stocks of Jordan and projected changes upon improved management of croplands. Geoderma, 132, 361371. DOI:10.1016/j.geoderma.2005.05.013.
  • Cai, C. & Qin, S. 2006. Dynamics of crop yields and soil organic carbon in a long-term fertilization experiment in the Huang-Huai-Hai Plain of China. Geoderma, 136, 708715. DOI:10.1016/j.geoderma.2006.05.008.
  • Chang, K.-C., Xu, P., Lu, L., Shao, W., Wang, Y., Xu, H., Yan, W. & Zhang, Z. 2002. The formation of Chinese civilization: an archaeological perspective. Yale University Press, New Haven, CT.
  • Chen, J. 2003. Anthropogenic soils and soil quality change under intensive management in China. Geoderma, 115, 12. DOI:10.1016/S0016-7061(03)00070-3.
  • ESRI. 2001. Using ArcGIS geostatistical analyst. ESRI, Redlands, CA.
  • Fan, T., Stewartb, B.A., Wang, Y., Luo, J. & Zhou, G. 2005. Long-term fertilization effects on grain yield, water-use efficiency and soil fertility in the dryland of Loess Plateau in China. Agriculture, Ecosystems and Environment, 106, 313329. DOI:10.1016/j.agee.2004.09.003.
  • FAO. 1984. Guidelines: land evaluation for rainfed agriculture. World Soil Resources Report 52. FAO, Rome.
  • FAO. 1995. Global and national soils and terrain digital databases (SOTER): procedures manual. World Soil Resources Report 71 Rev. 1. FAO, Rome.
  • FAO. 1996. Agro-ecological zoning guidelines. FAO Soil Bulletin 73. FAO, Rome.
  • FAO. 1999. Soil and physiographic database for North and Central Eurasia at 1:5 million scale. Land and Water Digital Media Series No. 7. CDROM Version 1.0 for Windows. FAO, Rome.
  • FAO-UNESCO-ISRIC. 1990. Soil map of the world: revised legend. World Soil Resources Report 60 (1988, reprint 1990). FAO-UNESCO-ISRIC, Rome.
  • Hatfield, J.L. 2001. Upscaling and downscaling methods for environmental research (book review). Journal of Environmental Quality, 30, 1100.
  • IAP & CDIAC. 1991. Two long-term instrumental climatic data bases of the People’s Republic of China. Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing; and Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge.
  • ISSAS & ISRIC. 1994. Reference soil profiles of the People’s Republic of China. Field and analytical data. Country Report 2. Institute of Soil Science, Chinese Academy of Sciences, Nanjing; and International Soil Reference and Information Center, Wageningen.
  • Liu, Q.H., Shi, X.Z., Weindorf, D.C., Yu, D.S., Zhao, Y.C., Sun, W.X. & Wang, H.J. 2006. Soil organic carbon storage of paddy soils in China using the 1:1,000,000 soil database and their implications for C sequestration. Global Biogeochemical Cycles, 20, GB3024.
  • National Soil Survey Office (NSSO). 1998. Soils of China. Agricultural Press, Beijing. ISBN 7-109-04354-1.
  • R Development Core Team. 2006. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. ISBN 3-900051-07-0. Available at: http://www.R-project.org (accessed 10/12/2006).
  • Sys, C., Van Ranst, E. & Debaveye, J. 1991. Land evaluation. Part I: principles in land evaluation and crop production calculations. Agricultural Publications No. 7. General Administration for Development Cooperation, Brussels.
  • Sys, C., Van Ranst, E., Debaveye, J. & Beernaert, F. 1993. Land evaluation. Part III: crop requirements. Agricultural Publications No. 7. General Administration for Development Cooperation, Brussels.
  • Tang, H., Van Ranst, E. & Sys, C. 1992. An approach to predict land production potential for irrigated and rainfed winter wheat in Pinan County, China. Soil Technology, 5, 213224. DOI:10.1016/0933-3630(92)90023-T.
  • Tang, H., Qiu, J., Van Ranst, E. & Li, C. 2006. Estimations of soil organic carbon storage in cropland of China based on DNDC model. Geoderma, 134, 200206.
  • WMO. 1996. Climatological normals (CLINO) for the period 1961–1990. WMO-No. 847. WMO, Geneva. Available at: ftp://ftp.ncdc.noaa.gov/pub/data/normals/ (accessed 10/12/2006).
  • Wu, B., Xu, W., Huang, H. & Yan, C. 2003a. The land cover map for China in the year 2000. GLC2000 database. European Commission Joint Research Centre. Available at: http://www.gvm.jrc.it/glc2000 (accessed 20/10/2006).
  • Wu, H., Guo, Z. & Peng, C. 2003b. Distribution and storage of soil organic carbon in China. Global Biochemical Cycles, 17, 1048. DOI:10.1029/2001GB001844.
  • Wu, H., Guo, Z. & Peng, C. 2003c. Land use induced changes of organic carbon storage in soils of China. Global Change Biology, 9, 305315.
  • Ye, L. & Van Ranst, E. 2002. Population carrying capacity and sustainable agricultural use of land resources in Caoxian County (North China). Journal of Sustainable Agriculture, 19, 7594.
  • Ye, L. & Van Ranst, E. 2004. Development of a Web-based land evaluation system and its application to population carrying capacity assessment using NET technology. In: Proceedings of the AFITA/WCCA 2004 Joint Congress on IT in Agriculture (eds F.Zazueta, S.Ninomiya & R.Chitradon), pp. 409414. National Science and Technology Development Agency, Thailand.
  • Ye, L., Van Ranst, E. & Verdoodt, A. 2004. Design and implementation of a Web-based land evaluation system and its application to land value tax assessment using NET technology. In: Innovative technology in soil survey: developing the foundation for a new generation of soil resource inventories and their utilization (eds H.Eswaran, P.Vijarnsorn, T.Vearasilp & E.Padmanabhan), pp. 183196. Land Development Department, Bangkok, Thailand.
  • Zhou, C., Zhou, Q. & Wang, S. 2003. Estimating and analyzing the spatial distribution of soil organic carbon in China. Ambio, 32, 612.