Open Geosci.

2015; 7: 234243

Research Article

Open Access

Yanmin Yao, Liming Ye*, Huajun Tang*, Pengqin Tang, Deying Wang, Haiqing Si, Wenjun Hu,
and Eric Van Ranst

Cropland soil organic matter content change in

Northeast China, 1985-2005
DOI 10.1515/geo-2015-0034
Received October 29, 2014; accepted April 20, 2015

Abstract: Soil organic matter (SOM) content is one of the

most important indicators of soil quality and hence the
productive capacity of soils. Northeast China (NEC) is the
most important region in grain production in China. In
this study, we assessed the spatiotemporal change of cropland SOM content in NEC using sampling data of 2005 and
survey data of 1985. We also analysed the driving forces
behind the SOM content change. Our results showed that
SOM content decreased in 39% of all the cropland in NEC,
while increase in SOM content was only detected on 16%
of the cropland. SOM remained unchanged in nearly half
(i.e. 45%) of the cropland. Our results also revealed that
cropping intensity and fertilizer application were the two
most important factors driving SOM change. Overall, results from this research provided novel details of the spatiotemporal patterns of cropland SOM content change in
NEC which was not revealed in earlier assessments. The
datasets presented here can be used not only as baselines
for the calibration of process-based carbon budget models, but also to identify regional soil quality hotspots and
to guide spatial-explicit soil management practices.
Keywords: Agriculture, Geostatistics, Kriging, Soil quality,
Organic carbon, Change attribution

*Corresponding Author: Liming Ye: Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural
Sciences, Beijing 100081, China; Chinese Academy of Agricultural
Sciences-Ghent University Joint Laboratory of Global Change and
Food Security, Beijing 100081, China, Department of Geology and
Soil Science (WE13), Ghent University; 9000 Gent, Belgium, E-mail:
[Link]@[Link]
*Corresponding Author: Huajun Tang: Chinese Academy
of Agricultural Sciences-Ghent University Joint Laboratory of
Global Change and Food Security, Beijing 100081; China, Chinese
Academy of Agricultural Sciences, Beijing 100081, China, E-mail:
tanghuajun@[Link]
Yanmin Yao, Pengqin Tang, Deying Wang, Haiqing Si, Wenjun
Hu: Institute of Agricultural Resources and Regional Planning,

1 Introduction
Soil is a vital non-renewable natural resource which plays
many essential roles in terrestrial ecosystems. The productive capacity of soils in particular, underlies the foundation of the well-being of humanity on Earth. Soil organic
matter (SOM) content is one of the most important indicators of soil quality and hence the productive capacity of
soils [13]. Therefore assessment of SOM content change
in space and time is of great importance for food security
decision-making both globally and in populous countries
like China [48].
Northeast China (NEC) is one of the most important regions in grain production in China. In 2009, it produced
15% of the countrys total grain output on 17% of the countrys cropland, resulting in a much higher regional level
of per capita grain of 620 kg than the national average
level of 370 kg. However, soil fertility in NEC has been observed declining after long-time cultivation of agricultural
crops [9, 10]. Assessments suggested that the magnitude
of this decline could be as high as 30-40% and 60% after
20 and 100 years of agricultural land use respectively [11],
posing risks to the stable supply of food either for the region or for the country as a whole. Moreover, crop production has been increasingly intensified in NEC since the
1980s alongside the general warming trend observed in
the region. The cropping intensity, or the ratio between
sown area and cropland area, increased from 66% in 1985
to 87% in 2005 [12] at an average rate of 1% yr1 , casting
uncertainties on current SOM levels in regional cropland.
During the past decade, research efforts have increasingly
focused on improving our knowledge about soil C en-

Chinese Academy of Agricultural Sciences, Beijing 100081, China;

Chinese Academy of Agricultural Sciences-Ghent University Joint
Laboratory of Global Change and Food Security, Beijing 100081,
China
Eric Van Ranst: Chinese Academy of Agricultural Sciences-Ghent
University Joint Laboratory of Global Change and Food Security,
Beijing 100081, China; Department of Geology and Soil Science
(WE13), Ghent University, 9000 Gent, Belgium

2015 Yanmin Yao et al., licensee De Gruyter Open.

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

Cropland SOM change in NE China

| 235

hancement and its potential to offset atmospheric CO2

concentrations and safeguard national food security in
China. The majority of the research employed processedbased C budget models to characterize historical patterns
and predict future trends, such as DNDC [13], RothC [14]
and Agro-C [15]. However, empirical studies revealing spatial changes of soil C since the 2nd National Soil Survey
(NSS-2), which was conducted in the 1980s, are still lacking, especially at regional scale. The objectives of this paper are therefore to (1) assess the spatiotemporal change
of SOM content in NEC using data from NSS-2 in 1985
and from the grid sampling of SOM in 2005 under the
Cropland Fertility Monitoring Network (CFMN) program in
China [16]; (2) characterize spatial and temporal patterns
in SOM content change at the regional scale; and (3) reveal
major factors driving SOM change in Northeast China over
space and time.
Figure 1: SOM sampling sites in Northeast China in 2005. A sampling site is represented by a black dot on map. The underlying
digital terrain is extracted from the USGS 30 arc-second global elevation dataset [52].

2 Methods
2.1 SOM sampling and interpolation
NEC occupies a land mass of 788,000 km2 and consists
of three provinces: Heilongjiang, Jilin and Liaoning. Cropland is accounted for 21% of the land mass or 16.4 million ha. In 2005, a regular grid of 2020 km was established on cropland in NEC with a total of 750 composite
sampling sites (Figure 1). At each site, three to five points
were randomly selected within a radius of 100 m where
the topsoil (0-20 cm) was sampled. These samples were
pooled for laboratory analysis. Organic carbon content
was determined using the Walkley-Black wet combustion
method [17], the same method as in 1985 for NSS-2. The soil
organic carbon (SOC) content was measured directly by
chemical oxidation with a mixture of dichromate and sulphuric acid solutions. Heat was applied at 170 C to accelerate the reaction. A correction factor of 1.1 was adopted to
compensate the incomplete oxidation. The obtained SOC
content was then converted to SOM content by using the
inverse of the van Bemmelen index of 0.58 [18, 19]. The
SOM measurements were statistically analysed to reject
outliers and tested for normal distribution. Values that lie
more than 1.5 times of the inter-quartile range from the first
or the third quartile were considered as outliers [20] and
rejected. Normality was achieved by logarithmic transformation of the outlier-free SOM data, and confirmed by the
Shapiro-Wilk test [21]. This statistically processed point
dataset of SOM observations in 2005 was finally interpo-

lated into a continuous surface of 1 km spatial resolution

using ordinary kriging [19, 22].
The soil map of NEC at the scale of 1:4 million was obtained from the National Soil Survey Office [23] based on
soil survey conducted in NEC in 1985 as part of the NSS2. In total, 314,813 samples were taken from the top soil
in NEC at an average sampling density of 1 sample per
67 ha of cropland [23]. Vector-format soil maps were produced from these samples at the scale of 1:10,000 in the
county level, and upscaled to 1:1 million at the provincial
level and finally to 1:4 million at the national level. The
SOM attribute associated with this 1985 soil map was given
in value ranges or classes (Table 1) derived from multiple
samples within a single soil mapping unit. It was a common practice to report SOM classes in NSS-2, as stipulated
by the National SOM Grading System of China [23, 24].
Although the soil map in 1985 was produced based on a
dense sampling scheme, these samples were not georeferenced due to technological limitations. Consequently, a revisit to the exact 1985-sampling localities in 2005 was unfeasible. In order to compare the SOM maps in 1985 and in
2005, therefore, the SOM attribute was extracted from the
vector soil map in 1985 and converted to a grid map at 1 km
spatial resolution using the centre-point value assignment
method implemented with the Spatial Analyst extension of
the ArcGIS 9 software package.

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

236 | Yanmin Yao et al.

Table 1: Cropland area change per SOM class in Northeast China, 1985-2005.
SOM
class
I
II
III
IV
V
VI
VII
IIII
VVII

SOM
range
(w/w %)
< 0.6
0.61.0
1.02.0
2.03.0
3.04.0
4.08.0
> 8.0
<2
>3

1985
percent

million
ha
0.27
1.32
5.11
4.22
6.79
9.66
0.93
6.70
17.39

million
ha
0.51
2.03
7.95
3.92
3.99
9.90
0.00
10.50
13.89

0.9
4.7
18.0
14.9
24.0
34.1
3.3
23.7
61.4

2.2 Geostatistical model estimation and

validation

(1)

where is the location-irrelevant trend term and (S) is the

spatially correlated stochastic error term. The SOM con^ 0 ), is estimated
tent, Z, at an unsampled location S0 , Z(S
as the weighted average of n neighboring samples of S0 :

Z (S0 ) =

{ i Z (S i )},

(2)

i=1

where i is the kriging weight assigned to sampling site S i .

The weighting factor i is estimated by using the semivariances derived from neighboring samples:
(h) =

1
2
[Z (S i ) Z (S i + h)] ,
2

Area
change
million ha
0.25
0.71
2.85
-0.30
-2.80
0.24
-0.93
3.80
-3.50

1.8
7.2
28.1
13.8
14.1
35.0
0.0
37.1
49.1

Percentage
change
0.9
2.5
10.1
-1.1
-9.9
0.9
-3.3
13.4
-12.4

Z (S i ) Z (S i ),
N
i=i

Z
S

Z
S

(
)
(
)

i
i
1
100 ,
MAPE =
N
Z (S i )

MAE =

Ordinary kriging (OK) was employed to model the spatial

variability of measured SOM samples and to predict SOM
values at unsampled locations. The prediction is based on
the following model:
Z (S) = + (S) ,

2005
percent

(3)

where h is the distance between two neighboring samples. A set of n point observations yield 0.5n(n 1) sample pairs. A mathematical model (e.g., spherical, exponential or Gaussian model) is then fitted to the semivariancedistance plot, which is called a variogram, to minimize the
variance of errors and the parameters of this variogram
model are used to assign weights in OK interpolation.
The goodness of a fitted variogram model was evaluated using the leave-one-out cross validation (LOOCV)
method [19] based on the following three measurements of
errors: the root-mean-squared error (RMSE), the mean absolute error (MAE) and the mean absolute percentage error
(MAPE):

N (
)2
1

RMSE =
Z (S i ) Z (S i ) ,
(4)
N

(5)

(6)

i=i

^ i ) are measured and predicted SOM

where Z(S i ) and Z(S
levels, respectively, at location S i , and N is the number of
validation sites.

2.3 Detection of SOM content change

To compare with the SOM classes in 1985, the kriginginterpolated SOM content values in 2005 were grouped
into SOM classes as given in Table 1. The map of cropland
SOM content change between 1985 and 2005 was obtained
by deducting the SOM content map in 1985 from the map in
2005. A SOM content change was only detected when the
SOM content in 2005 was one class higher or lower than
the SOM content in 1985.

2.4 Factor analyses

Single- and multi-factor regression analyses were employed to reveal the cause-effect relationships between
SOM content change and crop management practices. Two
indicators, namely cropping intensity and fertilizer application, were used to represent crop management practices. Cropping intensity was characterized by the total
sown area of all agricultural crops per grid cell. The grid
dataset of crop areas in 1985 and 2005 was obtained from
the Chinese Academy of Agricultural Sciences [25]. The
dataset was produced using a plausible spatial disaggregation method, cross-entropy, on the basis of multiple data

i=1

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

Cropland SOM change in NE China

sources including crop area and yield census, satellite imagery of crop distribution and irrigation, crop suitability
map, etc. [25]. The grid dataset of the application rate of
mineral fertilizers and manures in crop production in circa
2000 was obtained from the Center for International Earth
Science Information Network [26]. The R statistical software package [27] was used to conduct the regression analysis.

3 Results
3.1 Sampling results
The results of the laboratory analysis showed that the SOM
content of the soil samples taken from cropland in NEC in
2005 was averaged at 2.7%, with a standard deviation of
1.9% (Table 2). This corresponds to a SOC content level of
1.460.98%.

3.2 Kriging interpolation and validation

Three types of variographic models were fitted to the semivariogram calculated from the SOM sampling dataset in
2005. These included the spherical, exponential and Gaussian models (Table 3). The prediction performance of each
fitted semivariogram model (Figure 2) was tested using
the LOOCV method [19] and given in Table 3. The results
showed that the obtained exponential model had the highest performance in predicting SOM contents at unsampled
locations.

3.3 Cropland SOM contents in 1985 and

2005
The spatial distribution of cropland SOM content in terms
of SOM classes [23], based on NSS-2 in 1985 and CFMN
in 2005, is given in Figures 3a and 3b. These two maps
showed that cropland in the northeastern NEC had considerably higher SOM content, indicated by the dominant
greenish colours, than in the southwestern NEC where the
dominant colours were brownish. This downward gradient of SOM along the northeastern-southwestern axis was
largely persisted during the 1985-2005 period, supporting
earlier findings either at regional scale in NEC [28] or at
national scale [5, 29]. Area statistics (Table 1) showed that
Class VI (i.e. SOM content ranging at 4-8%) covered an area
of 9.66 million ha or 34.1% of all the cropland in NEC in
1985. In 2005, Class VI still took the largest area share of

| 237

35.0%, a small increase of 0.9% since 1985. However, the

biggest increase in cropland area was attributed to Class
III which expanded from 5.11 million ha in 1985 to 7.95 million ha in 2005 by a margin of 10.1%. To the contrary, area
of Class V was observed to decrease substantially by 9.9%,
from 6.79 million ha in 1985 down to 3.99 million ha in
2005 (Table 1). Overall, cropland with SOM < 2% including
Classes I through III occupied an area of 10.50 million ha
in 2005, a 13.4% increase since 1985 (see the expansion of
reddish/brownish-colour regions marked by annotations
A and B on maps in Figure 3). Area statistics also showed
that cropland of SOM classes IV, V and VII occupied an
area of 11.94 million ha or 42.2% in 1985; but in 2005 the
area collectively taken by these three classes shrank to a
mere of 7.91 million ha or 27.9%.

3.4 SOM content change between 1985 and

2005
The SOM content change between 1985 and 2005 in terms
of changes in SOM classes from 1985 to 2005 is shown in
Figure 3c. Spatial statistics revealed that cropland with
decreasing SOM classes was accounted for 11.05 million
ha or 39% of total cropland area in NEC (e.g. annotations
A and B in Figure 3c). In contrast, the area of cropland
whose SOM classes were increased during the same period
was much smaller. An increase in SOM content was identified only on 4.43 million ha or 15.7% of all the cropland
in NEC (e.g. annotation C in Figure 3c). Spatial statistics
also revealed that SOM in nearly half (i.e., 45.31%) of the
cropland in NEC remained unchanged during the last 20
years. Cropland with decreasing SOM was mainly located
in southwestern Heilongjiang, the bulk of Jilin and eastern
Liaoning province. Large areas in Heilongjiang province,
especially in the Three-River Plain in the northeast of the
province, had stable or improved SOM levels.

3.5 Attribution of SOM content change

Results of the single- and multi-factor regression analyses
(Table 4) showed that the SOM class change between 1985
and 2005 (Figure 3c) was closely related to cropping intensity change (Figure 4a) and to the application of mineral fertilizers and manures (Figure 4b) during the same
period. Linear relationships were successfully established
taking SOM class change as the response variable and
cropping intensity change (Figure 4c, R2 = 0.82) or fertilizer application (Figure 4d, R2 = 0.89) as the explainatory variable. Results also showed that variations in crop-

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

238 | Yanmin Yao et al.

Table 2: Descriptive statistics of the SOM sampling dataset in 2005, %.
Data
set
N = 750

Variation
Min
0.1

Max
9.62

Mean
2.7

Quartiles
Standard deviation Q1
1.9
1.0

Median
2.3

Q3
3.95

Table 3: Semivariogram models fitted to the SOM sampling data in Northeast China in 2005 and their prediction performances. The prediction performance is measured both by the coeflcients of a linear trend in the form of y = a + bx fitted to the predicted versus measured
data, and by the prediction error resulting from the cross validation of the models.
Model
type

Model fitting
Range
Sill
(m)
Spherical 371280
0.2664
Exponential 442821
0.3368
Gaussian 319339
0.2268

Nugget
0.2492
0.1992
0.2903

Lag
(m)
44618
44618
44618

Cross validation
a
b

R2

RMSE

MAE

0.65
0.45
0.70

0.64
0.73
0.62

1.04
1.03
1.08

0.78
0.78
0.81

0.80
0.85
0.78

MAPE
(%)
52.31
51.27
54.45

Figure 2: Semivariogram fitted with an exponential model (a) and cross validation of the obtained variogram model (b) for the interpolation
of SOM content in 2005. The obtained linear regression equation in (b) is y = 0.85x + 0.45 (R2 = 0.73).

Figure 3: Cropland SOM content in 1985 (a), 2005 (b) and SOM change in terms of content classes between 1985 and 2005 in Northeast
China (c). Value 1 in (c) means SOM has increased by one content class from 1985 to 2005, while value -1 means decrease by one class.
Dashed ovals annotated by A and B indicate regions where SOM content decreased, while dashed oval C indicates where SOM content
increased during the period.

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

Cropland SOM change in NE China

| 239

Table 4: Single- and multi-factor regression models in the form of Y = a+bX1 +cX2 where Y is SOM class change (dimensionless) from 1985
to 2005, X1 is percent change of cropping intensity (%) from 1985 to 2005 and X2 is the application rate of fertilizers and manures (kg/ha)
in circa 2000.
Predictor
Model obtained
Cropping intensity change (X1 )
Y = 0.04X1
Fertilizer and manure (X2 )
Y = 4.51 + 0.11X2
X1 and X2
Y = 2.63 + 0.02X1 + 0.07X2
Source: Center for International Earth Science Information Network [26]

ping intensity change and in fertilizer application simultaneously could explain 93% of the variations in SOM class
change (Table 4). This suggested that both cropping intensity change and fertilizer application were significant predictors of SOM change over time.
The cause-effect relationship characterized in this paper
is in accordance with earlier findings that proper application of mineral fertilizers in combination with manures
tends to increase the SOM content [30] and that long-term
crop cultivation tends to decrease the SOM content [10]. A
fundamental difference is that earlier findings were mostly
based on controlled experiments, while this finding was
established using much larger datasets in space and time.
It was not only more robust, but also extended the validity
of earlier findings to larger tempospatial scales.

4 Discussion
4.1 Comparison with previous assessments
Comprehensive analyses of spatial change of cropland
SOM content in NEC are relatively rare, compared to other
regions of the country [18]. Using NSS-2 data from 11 counties in NEC, Yu et al. [29] reported a decrease in topsoil OC
stock between 1980 and 2000, contrasting the general upward trend in SOC for China as a whole based on the same
data. At national scale, a similar trend was also observed
by Yan et al. [18], who compared soil profile data from the
periods of 1979-1982 and 2007-2008 and found that SOC
slightly increased from 1.20% in 1980s to 1.27% in 2000s
for the top layer of Chinese cropland, despite increasing intensity in crop production. However, at the regional scale
in NEC, most studies showed a general downward trend in
cropland SOM content during the past 20 years. For example, Cheng et al. [31] reported that SOC content declined
slightly from 1.70% in 1988 to 1.67% in 2007 according to
data from long-term monitoring and experimental stations
in NEC. More importantly, this downward trend in cropland SOM content was confirmed by meta-analyses. Using
frequency change as an indicator, Pan et al. [32] found that

R2
0.82***
0.89***
0.93***

SOC decreased on 49.2% of cropland area in NEC between

1985 and 2006 based on 62 observations published. Similarly, a decrease in SOC content between 1993 and 2006
was found in 74.4% of all the soil samples published in 132
references in NEC [10].
The results obtained in this paper are consistent with these
earlier assessments for the detection of a downward trend
in cropland SOM in NEC during the past 20 years. But our
assessment on the extent of this downward trend in terms
of cropland area differs from previous studies [10, 32].
Firstly, we adopted a systematic sampling strategy and collected 750 samples on cropland of NEC, resulting in a considerably fair spatial coverage (Figure 1). The previous assessments, such as those by Huang and Sun [10] or Pan
et al. [32], used much more limited data in a somewhat
compromised spatial coverage of cropland in NEC. Secondly, we used SOM classes (Table 1) to detect the temporal SOM content change between 1985 and 2005. This
approach may be less sensitive than using the absolute
difference to represent SOM content change as adopted by
e.g. Pan et al. [32]. However, it is more robust to use class
changes because the results can be more easily verified
in field surveys. Moreover, management decisions made
based on SOM class changes are more relevant to fertility
levels in field and thus more reliable to implement in practice. This is because the adoption of the SOM class change
approach is in effect the adoption of a higher threshold
in SOM change detection, which consequently lowers the
likelihood of false identification of carbon hotspots in soil
management.
Overall, the cropland SOM class maps (e.g. Figure 3) resulting from this research provide fairly detailed spatial
information, which was absent from earlier assessments,
and can be used to assist eco-environmental and agroeconomic decision-making at regional scale.

4.2 Spatial change of cropland SOM

Cropland SOM in NEC is observed to decrease along a NESW gradient (Figure 3), providing regional evidence to an

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

240 | Yanmin Yao et al.

Figure 4: Percent change of cropping intensity between 1985 and 2005 (a) and application rate of mineral fertilizers and manures in circa
2000 (b), and their relationships to SOM class change (c and d), respectively, in Northeast China. The red, solid line in (c) and (d) is the
linear trend fitted between the xaxis and yaxis variables; R2 = 0.82 for (c) and 0.89 for (d). Spatial resolution of maps in (a) and (b) is 40
km. Dashed ovals annotated by A, B and C have the same meaning as in Figure 3.

already observed pattern at the national scale. While the

SOM content in NEC is estimated to be the highest among
regions in China, surface SOM contents in croplands of the
North China Plain [33], the Loess Plateau [34], the Southwest Highlands [35], etc. are found decreasing gradually
from the northeast to the southwest of China. The major driving forces for this NE-SW gradient are, among others, climate [36], land use change [37] and soil management [38].
Firstly, temperature is a key factor controlling the rate of
decomposition of organic materials in soil [39]. Overall,
the SOM mineralization rate on the Southwest Highlands
is two times higher than in NEC because of the difference in
annual temperatures. The annual temperature varies from
10 C to 21 C in the southwest, but in the northeast the temperature varies at a much lower range, i.e. from -2 C to
10 C. It has been observed in experiments that the mineralization rate doubled for an increase in mean annual
temperature by 8-9 C [40]. The observed SOM gradient correlates well with the temperature gradient along the NESW axis in China. On the other hand, surface temperature
in China has risen 0.8 C in the twentieth century, being
consistent with the general global trend [41]. The warming
has been observed stronger in NEC and weaker in southwest China. The average annual temperature in NEC has

risen 1.0 C in the past 50 years. However, the magnitude

of this temperature change is obviously less influential on
SOM mineralization rate compared to the temperature difference between northeast and southwest China. Evidence
from Belgium, which is in a similar climate zone, also suggested that climate trends have little effect on observed
SOC change [42].
Secondly, dynamic changes have occurred to cropland in
NEC since the early 1980s. Fertile cropland has been lost to
other uses such as infrastructure and urbanization. Large
areas of cropland expansion have been observed too. NEC
is one of the few regions in China that still have a net
gain in cropland area [43]. According to She et al. [44] NEC
gained 3 million ha of cropland during 1993-2003, either
by converting natural grassland and forestland into cropland (Figure 3, annotation C or northern NEC in general) or
by re-cultivating marginal lands which were previously set
aside (Figure 3, annotation B or southeastern Jilin). Meanwhile, some 670,000 ha of cropland was lost during the
same period. This loss was mainly caused by rapid urbanization during the past few decades. It has been observed
that urbanization in NEC followed a reverse NE-SW pattern [45], [Figure 14], meaning that a higher urbanization
rate in the southwestern part of NEC took more cropland

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

Cropland SOM change in NE China |

out of agricultural use than in northeast NEC where the urbanization rate was lower.
Thirdly, management plays an important role in maintaining SOM levels in agro-ecosystems [42]. Although soiluse duration and intensity have a strong impact on SOM
content [37, 46], proper management can substantially
improve cropland SOM content under intensive farming.
Long-term experiments conducted in Gongzhuling [30] in
Jilin province and Hailun [28] in Heilongjiang province
show that the dual goal of optimal crop yield and higher
SOM content can be simultaneously met by fine-tuned
application of mineral fertilizers, farmyard manures and
crop residues according to nutrient requirements. At the
end of the experiments, surface SOM contents were found
20-50% higher under soybean-maize rotation in Jilin and
maize mono-cropping in Heilongjiang respectively. This
suggests that in addition to climate and land conversion,
soil and crop management is another important driver responsible for SOM content variability across NEC.

4.3 Temporal change of cropland SOM

content
As observed earlier, cropland SOM tends to increase in
northern NEC and decrease in southern NEC during the
1985-2005 period (Figure 3c). In the main part of the Heilongjiang province, for example, SOM content either increased by 1-2 classes or remained unchanged in 2005,
compared to 1985. However, different processes are found
behind this largely optimistic temporal change of SOM
content in the province. In central Heilongjiang province,
for example, changes in cropping intensity were hardly
observed (Figure 4a); continuous fertilization at nearoptimal rates [28] maintained at least a stable SOM level
during 1985-2005. Taking the Three-River Plain in northeastern Heilongjiang province (Figure 3c, annotation C) as
another example, a 50% decrease in cropping intensity
(Figure 4a) and a relatively low fertilizer application rate
of 30-40 kg/ha (Figure 4b) were simultaneously observed.
The collective effects of these two factors (Table 4) may
lead to an increase of SOM by one content class, which is
lower than the observed increase of 1-2 classes (Figure 3c).
The uncharacterized, residual magnitude of increase in
SOM content can be explained by the conversion of grassland or forestland to cropland [43], a phenomenon which
is usual in this region but rare in other regions.
A general pattern on cropland SOM content change between 1985 and 2005 is that cropland with lower-thanaverage SOM levels was gaining area from cropland with
higher SOM levels, meaning that cropland SOM content

241

tends to decrease when the initial SOM content is high. As

shown in Table 1, the area of cropland with SOM > 3% decreased by a margin of 3.5 million ha or 12.4% from 1985
to 2005, while the area of cropland with SOM < 2% increased by 3.8 million ha. This finding of a negative relationship between the spatial extent of temporal change
and the magnitude of initial SOM content supports earlier
findings of the dependence of relative annual change in
SOM content on the baseline SOM value using monitoring
data from China [18, 32] or inventory data from the UK [47].
This characterized temporal pattern of SOM content
change in NEC has important implications on carbon
management in practice. Firstly, attention is needed on
the regions with high SOM values, because these regions are highly vulnerable to carbon losses. Management
measures should therefore be tailored to specific natural and socio-economic conditions of these regions. Secondly, investments in soil management in SOM-poor regions will more likely generate higher returns than in other
regions [24]. Ideally, soil management in these regions
should be guided through extension programs for smallholder farmers as in NEC. Thirdly, although the concept of
soil management is not new, the importance of it has not
been fully recognized in agricultural practice in the past
two decades in NEC [38], as evidenced by the general declining trend in cropland SOM and by the spatial extent of
cropland with decreasing SOM classes (Figure 3).

4.4 Uncertainties
Major sources of uncertainty in this paper include the rasterization of polygon-based soil map, kriging interpolation of SOM samples and scaling of the grid datasets. Although the errors associated with the vector-to-raster or
raster-to-vector conversion methodologies were well documented [48], much less has been known on how these errors propagate to the assessment results or whether these
errors are large enough to impact the assessment results.
High-resolution soil datasets are useful to decrease uncertainties, but such datasets are either not always available
or too expensive to collect. In trying to search for an optimal resolution, Yu et al. [49] recently evaluated the effect of varying raster grid sizes on the estimated SOC stock
in the Taihu Lake region in China, using vector-converted
soil maps. Such efforts deserve more attention. Attention
should also be given to researches on error quantification,
propagation, uncertainty control [2, 50], and so forth.
Cause-effect analyses used coarse grid datasets at a resolution of 40 km to attribute SOM changes to cropping intensity and/or fertilization in NEC between 1985 and 2005.

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

242 | Yanmin Yao et al.

Although such census-based, spatially disaggregated grid
maps are the best available datasets, they may not be sufficiently detailed to isolate the effects of individual factors
by statistical analyses [51]. Further research is needed, not
only to fine-tune the spatial disaggregation method itself
but also to validate its result maps against denser samples.

5 Conclusions
Based on systematic sampling of cropland soils in NEC in
2005 and the Second National Soil Survey in 1985, an overall decrease in cropland SOM was detected in NEC for the
last 20 years. Although this downward trend is mostly consistent with earlier findings, the spatial extent ( 40% in
area terms) of it is substantially smaller than in previous
assessments ( 50-70% of cropland). We suggest that previous assessments overestimated the area extent of the declining trend of cropland SOM due to limited spatial coverage of soil data used. We also find that SOM-rich croplands tend to lose carbon when SOM is higher than 2%.
Moreover, our results reveal that cropping intensity and
fertilizer application are two most important factors driving SOM change in NEC. These findings indicate that better soil management strategies need to be urgently established through research so that the dual goal of optimal
crop yield and higher SOM content can be met.
Acknowledgement: This research is funded by the National Nonprofit Institute Research Grant of CAAS (IARRP2015-28) and the National Basic Research Program of
China (Grant no. 2010CB951501 and 2010CB951504). Special thanks are given to H.L. for language improvements.

References
[1] Carter M.R., Organic matter and sustainability. In: Rees R.M.,
Ball B.C., Campbell C.D., Watson C.A. (Eds.), Sustainable Management of Soil Organic Matter. CABI Publishing, Oxon and New
York, 2001, 9-41
[2] Ye L., Van Ranst E., Production scenarios and the effect of soil
degradation on long-term food security in China. Global Environ.
Chang., 2009, 19, 464-481
[3] Duval M.E., Galantini J.A., Iglesias J.O., Canelo S., Martinez J.M.,
Wall L., Analysis of organic fractions as indicators of soil quality
under natural and cultivated systems. Soil Till. Res., 2013, 131,
11-19
[4] Lal R., Soil carbon sequestration impacts on global climate
change and food security. Science, 2004, 304, 1623-1627
[5] Ye L., Tang H., Zhu J., Verdoodt A., Van Ranst E., Spatial patterns
and effects of soil organic carbon on grain productivity assess-

ment in China. Soil Use Manage., 2008, 24, 80-91

[6] Ye L., Xiong W., Li Z., Yang P., Wu W., Yang G. et al., Climate
change impact on China food security in 2050. Agron. Sustain.
Dev., 2013, 33, 363-374
[7] Ye L., Tang H., Wu W., Yang P., Nelson G.C., Mason-DCroz D. et al.,
Chinese food security and climate change: Agriculture futures.
Economics-Kiel, 2014, 8 (2014-1), DOI: 10.5018/[Link].2014-1.
[8] Powlson D.S., Gregory P.J., Whalley W.R., Quinton J.N., Hopkins
D.W., Whitmore A.P. et al., Soil management in relation to sustainable agriculture and ecosystem services. Food Policy, 2011,
36, S72-S87
[9] Fang H., Yang X., Zhang X., Organic carbon stock of black soils in
Northeast China and its contribution to atmospheric CO2. J. Soil
Water Conserv., 2003, 17, 9-12
[10] Huang Y., Sun W.J., Variation trend of surface soil organic carbon
in China for nearly two decades. Chinese Sci. Bull., 2006, 51, 750763
[11] Wang J., Wang T., Zhang X., Guan L., Wang Q., Hu H. et al.,
Black soil quality change I, changes in reclamation duration. J.
Shenyang Agric. Univ., 2002, 33, 43-47
[12] National Bureau of Statistics of China, China Compendium of
Statistics 1949-2008. China Statistical Press, Beijing, 2010
[13] Tang H., Qiu J., Van Ranst E., Li C., Estimations of soil organic
carbon storage in cropland of China based on DNDC model. Geoderma, 2006, 134, 200-206
[14] Wang J., Lu C., Xu M., Zhu P., Huang S., Zhang W. et al., Soil organic carbon sequestration under different fertilizer regimes in
north and northeast China: RothC simulation. Soil Use Manage.,
2013, 29, 182-190
[15] Yu Y., Huang Y., Zhang W., Projected changes in soil organic carbon stocks of Chinas croplands under different agricultural managements, 2011-2050. Agric. Ecosyst. Environ, 2013, 178, 109120
[16] Zhong Y., Collection of Cropland Soil Fertility Monitoring Results
in China. China Agricultural Press, Beijing, 2003, 3-28 (in Chinese)
[17] Bisutti I., Hilke I., Raessler M., Determination of total organic
carbon - an overview of current methods. TRAC-Trend Anal.
Chem., 2004, 23, 716-726
[18] Yan X., Cai Z., Wang S., Smith P., Direct measurement of soil organic carbon content change in the croplands of China. Global
Change Biol., 2011, 17, 1487-1496
[19] Piccini C., Marchetti A., Francaviglia R., Estimation of soil organic matter by geostatistical methods: Use of auxiliary information in agricultural and environmental assessment. Ecol. Indic.,
2014, 36, 301-314
[20] Maindonald J., Braun W.J., Data Analysis and Graphics Using
R - an Example-Based Approach, 3rd Ed. Cambridge University
Press, Cambridge, 2010
[21] Ye L., Verdoodt A., Moussadek R., Tang H., Van Ranst E., Assessment of Chinas food producing capacities using a Web-based
land evaluation engine and a grid-based GIS. Adv. Geoecol.,
2008, 39, 699-718
[22] Mishra U., Lal R., Slater B., Calhoun F., Liu D., Van Meirvenne M.,
Predicting soil organic carbon stock using profile depth distribution functions and ordinary kriging. Soil Sci. Soc. Am. J., 2009,
73, 614-621
[23] National Soil Survey Oflce (NSSO), The Soils of China. China
Agricultural Press, Beijing, 1998 (in Chinese)

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM

Cropland SOM change in NE China |

[24] Xu Y., Zhang F., Hao X., Wang J., Wang R., Kong X., Influence
of management practices on soil organic matter changes in the
Northern China Plain and Northeastern China. Soil Till. Res.,
2006, 86, 230-236
[25] Tan J., Yang P., Liu Z., Wu W., Zhang L., Li Z. et al., Spatiotemporal dynamics of maize cropping system in Northeast China
between 1980 and 2010 by using spatial production allocation
model. J. Geogr. Sci., 2014, 24, 397-410
[26] Potter P., Ramankutty N., Bennett E.M., Donner S.D., Characterizing the spatial patterns of global fertilizer application and manure production. Earth Interact., 2010, 14, 1-22
[27] R Core Team, R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, 2013,
[Link]
[28] Ding X., Han X., Liang Y., Qiao Y., Li L., Li N., Changes in soil organic carbon pools after 10 years of continuous manuring combined with chemical fertilizer in a Mollisol in China. Soil Till. Res.,
2012, 122, 36-41
[29] Yu Y., Guo Z., Wu H., Kahmann J.A., Oldfield F., Spatial changes
in soil organic carbon density and storage of cultivated soils in
China from 1980 to 2000, Global Biogeochem. Cy., 2009, 23,
GB2021
[30] Zhang W., Wang X., Xu M., Huang S., Liu H., Peng C., Soil organic
carbon dynamics under long-term fertilizations in arable land of
northern China. Biogeosciences, 2010, 7, 409-425
[31] Cheng K., Pan G., Tian Y., Li L., Changes in topsoil organic carbon
of Chinas cropland evidenced from the National Soil Monitoring
Network. J. Agro-Environ. Sci., 2009, 28, 2476-2481
[32] Pan P., Xu X., Smith P., Pan W., Lal R., An increase in topsoil SOC
stock of Chinas croplands between 1985 and 2006 revealed by
soil monitoring. Agr. Ecosyst. Environ., 2010, 136, 133-138
[33] Zhu L., Yang M., Chen C., Effect of improved management practices on soil organic carbon sequestration in wheat-maize double
cropping system in North China, J. Agric. Sci., 2012, 4, 114-125
[34] Li Z., Li X., Li M., Yang J., Turner N., Wang X. et al., Countyscale changes in soil organic carbon of croplands in southeastern
Gansu province of China from the 1980s to the mid-2000s. Soil
Sci. Soc. Am. J., 2013, 77, 2111-2121
[35] Wang S., Zhou C., Li K., Zhu S., Huang F., Analysis on spatial distribution characteristics of soil organic carbon reservoir in China.
Acta Geogr. Sinica, 2000, 55, 533-544 (in Chinese)
[36] Wu H., Guo Z., Peng C., Distribution and storage of soil organic
carbon in China. Global Biogeochem. Cy., 2003, 17, 1048
[37] Acn-Carrera M., Jos Marques M., Carral P., lvarez A.M., Lpez
C., Martn-Lpez B. et al., Impacts of land-use intensity on soil organic carbon content, soil structure and water-holding capacity.
Soil Use Manage., 2013, 29, 547-556
[38] Liu X., Herbert S.J., Hashemi A.M., Zhang X., Ding G., Effects
of agricultural management on soil organic matter and carbon
transformation - a review. Plant Soil Environ., 2006, 52, 531-543

243

[39] Baritz R., Seufert G., Montanarella L., Van Ranst E., Carbon
concentrations and stocks in forest soils of Europe. Forest Ecol.
Manag., 2010, 260, 262-277
[40] Bot A., Benites J., The Importance of Soil Organic Matter: Key to
drought-resistant soil and sustained food and production. FAO
Soils Bulletin 80, FAO, Rome, 2005
[41] Ye L., Yang G., Van Ranst E., Tang H., Time-series modeling and
prediction of global monthly absolute temperature for environmental decision making. Adv. Atmos. Sci., 2013, 30, 382-396.
[42] van Wesemael B., Paustian K., Meersmans J., Goidts E., Barancikova G., Easter M., Agricultural management explains historic
changes in regional soil carbon stocks. P. Natl. Acad. Sci. USA,
2010, 107, 14926-14930
[43] Liu J., Liu M., Tian H., Zhuang D., Zhang Z., Zhang W., et al., Spatial and temporal patterns of Chinas cropland during 1990-2000:
An analysis based on Landsat TM data. Remote Sens. Environ.,
2005, 98, 442-456
[44] She G., Chen Y., Yao Y., Chen Y., The changes and macro-control
of cropland soil in Northeast China in the last decade. Econ. Geogr., 2005, 25, 391-396
[45] Yue T., Wang Y., Liu J., Chen S., Qiu D., Deng X. et al., Surface modelling of human population distribution in China. Ecol.
Model., 2005, 181, 461-478
[46] Sleutel S., De Neve S., Hofman G., Estimates of carbon stock
changes in Belgian cropland. Soil Use Manage., 2003, 19, 166171
[47] Bellamy P.H., Loveland P.J., Bradley R.I., Lark R.M., Kirk G.J.D.
Carbon losses from all soils across England and Wales, 19782003. Nature, 2005, 437, 245-248
[48] Wade T.G., Wickham J.D., Nash M.S., Neale A.C., Riitters K.H.,
Jones K.B., A Comparison of vector and raster GIS methods for
calculating landscape metrics used in environmental assessments. Photogramm. Eng. Rem. S., 2003, 69, 1399-1405
[49] Yu D.S., Ni Y.L., Shi X.Z., Wang N., Warner E.D., Liu Y. et al., Optimal soil raster unit resolutions in estimation of soil organic carbon pool at different map scales. Soil Sci. Soc. Am. J. 2014, 78,
1079-1086
[50] Kempen B., Heuvelink G.B.M., Brus D.J., Stoorvogel J.J., Pedometric mapping of soil organic matter using a soil map with quantified uncertainty. Eur. J. Soil Sci., 2010, 61, 333-347
[51] Taghizadeh-Toosia A., Olesen J.E., Kristensen K., Elsgaard L.,
stergaard H.S., Lgdsmand M. et al., Changes in carbon stocks
of Danish agricultural mineral soils between 1986 and 2009. Eur.
J. Soil Sci., 2014, 65, 730-740
[52] USGS, USGS global 30 arc-second elevation data set (GTOPO30
DEM), 1996, [Link]

Brought to you by | University of Ghent

Authenticated
Download Date | 8/28/15 11:06 AM