J. Mt. Sci. (2014) 11(3): 774-781 e-mail: jms@imde.ac.cn http://jms.imde.ac.cn DOI: 10.1007/s11629-013-2843-x

774

Abstract: Land cover type is critical for soil organic

carbon (SOC) stocks in territorial ecosystems.

However, impacts of land cover on SOC stocks in a

karst landscape are not fully understood due to

discontinuous soil distribution. In this study,

considering soil distribution, SOC content and density

were investigated along positive successional stages

(cropland, plantation, grassland, scrubland,

secondary forest, and primary forest) to determine

the effects of land cover type on SOC stocks in a

subtropical karst area. The proportion of continuous

soil on the ground surface under different land cover

types ranged between 0.0% and 79.8%. As land cover

types changed across the positive successional stages,

SOC content in both the 0–20 cm and 20–50 cm soil

layers increased significantly. SOC density (SOCD)

within 0–100 cm soil depth ranged from 1.45 to 8.72

kg m-2, and increased from secondary forest to

primary forest, plantation, grassland, scrubland, and

cropland, due to discontinuous soil distribution.

Discontinuous soil distribution had a negative effect

on SOC stocks, highlighting the necessity for accurate

determination of soil distribution in karst areas.

Generally, ecological restoration had positive impacts

on SOC accumulation in karst areas, but this is a slow

process. In the short term, the conversion of cropland

to grassland was found to be the most efficient way

for SOC sequestration.

Keywords: Soil organic carbon (SOC); Karst area;

Discontinuous soil distribution; Land cover type;

Carbon sequestration potential

Introduction

Globally, the amount of organic carbon stored

in soil is twice that stored in the atmosphere

(Krogh et al. 2003). Consequently, even slight

changes in soil organic carbon (SOC) stocks can

result in large variation in the atmospheric CO2

fluxes. SOC is directly influenced by land cover

type (Karchegani et al. 2012). In Southeast Asia,

the SOC stock (soil type: Hapludults) within a

depth of 0–100 cm was found to have increased by

23% after the conversion of primary forest

(Eusideroxylon zwageri, lowland tropical rain

forest) to grassland (Imperata cylindrica), over a

period of approximately 10 years (Yonekura et al.

2010). In Hawaii, the conversion of sugarcane

plantation to forest sequestered more SOC than

conversion of the plantation to pasture (on silty

CHEN Xiang-bi1,2, ZHENG Hua3, ZHANG Wei1,2, HE Xun-yang1, 2, LI Lei1,2, WU Jin-shui1, HUANG Dao-you1, SU Yi-rong1,2*

1 Key Laboratory of Agro-ecological Processes in Subtropical Region, Institute of Subtropical Agriculture, Chinese Academy of Sciences, Changsha 410125, China

2 Huanjiang Observation and Research Station for Karst Eco-systems, Huanjiang 547100, China

3 Guangxi Institute of Subtropical Crops, Nanning 530001, China

*Corresponding author, e-mail: yrsu@isa.ac.cn; Tel.: +86-731-84615222; Fax: +86-731-84612685; First author, e-mail: xbchen@isa.ac.cn

Citation: Chen XB, Zheng H, Zhang W, et al. (2014) Effects of land cover on soil organic carbon stock in a karst landscape with discontinuous soil distribution. Journal of Mountain Science 11(3). DOI: 10.1007/s11629-013-2843-x

© Science Press and Institute of Mountain Hazards and Environment, CAS and Springer-Verlag Berlin Heidelberg 2014

Effects of Land Cover on Soil Organic Carbon Stock in a

Karst Landscape with Discontinuous Soil Distribution

Received: 25 July 2013 Accepted: 20 November 2013

J. Mt. Sci. (2014) 11(3): 774-781

775

clay loam soils) (Li and Mathews 2010). Conversely,

at a national scale, SOC decreased as land cover

changed from pasture to plantation in New Zealand

(Scott et al. 1999). SOC also decreased with a shift

from native forest to plantation, secondary forest,

orchards, or arable land, within mid-subtropical

mountainous areas (red soil region) in southern

China (Yang et al. 2009). Generally, SOC was

found to accumulate when agricultural lands were

converted to natural or planted with perennial

vegetation (Post and Kwon 2000; Laganière et al.

2010; Chen et al. 2007).

The karst area in southwestern China

(approximately 0.55 million km2) is considered to

be a typically ecologically fragile zone, because

their soils are usually thin, coarse-textured, and

very susceptible to erode and degrade. In this area,

land cover types have changed over time through

ecological restoration initiatives, with the

implementation of a national project entitled Grain

for Green. Thus, evaluation of the effects of land

cover types on SOC stocks in this area is feasible

based on a space-for-time substitution

(“Chronosequence”) method (Pickett 1989).

However, a few such studies have been carried out

in karst areas to date, as the discontinuous soil

distributed in rock crevices and fissures results in

high variability (Zhang et al. 2007). Classical

methods used to estimate SOC stocks are based on

the assumption that soil distribution is continuous

(Zheng et al. 2012). As note, this assumption is not

valid for karst areas.

Our objectives in this study were to: (1)

investigate changes in SOC content, SOC density,

and potential for SOC sequestration, considering

soil distribution along advancing seral stages in a

karst area, and (2) clarify the effects of soil

distribution on SOC content and SOC sequestration

potential.

1 Materials and Methods

1.1 Study sites and sampling

A karst ecological restoration (KER, 24°55′ N,

107°57′ E) and a karst natural reserve sites (KNR,

25°9′ N, 108°1′ E; Mulun National Nature Reserve

Park) within the Huanjiang County in southwest

China were selected for the study (Figure 1).

Detailed information on these study sites has been

described by Zheng et al. (2012). Briefly, xeric

croplands with KER were situated in depressions

or on slopes with the gradient less than 25°. Some

of the sloping croplands were abandoned and

subsequently converted to grassland, scrubland, or

plantation (Chinese chestnut), following 5–30, 50,

and 10 years, respectively. The time was obtained

by interview investigation and literature research.

However, some croplands in depressions or on

slopes were retained. Secondary forests were

located on slopes with gradient more than 25°.

Primary forests exist solely within KNR. Figure 2

shows the characteristics of the studied ecosystems

under different land cover types.

Based on the space-for-time substitution

method, we regarded cropland as the initial stage

of ecological recovery, while primary or secondary

forests were considered to represent the ultimate

stage of the successional process. Both grassland

and scrubland were viewed as intermediate stages

of natural vegetation restoration, and plantation

was considered an intermediate stage of artificial

ecological restoration. Thus, positive vegetation

succession followed the order: (1) cropland, (2)

Figure 1 The karst study sites in China (modified from Zheng et al. 2012). KNR - karst natural reserve; KER - karst ecological restoration.

J. Mt. Sci. (2014) 11(3): 774-781

776

plantation/grassland, (3) scrubland, and (4)

secondary/primary forest.

Soil distribution was examined using eight line

transects, with 2 m × 2 m quadrats at 10 m interval

along each transect. Three transects were

established in primary forest, and two in secondary

forest, two in combined scrub, plantation, grass

and cropland on slopes (because all were

previously sloping cropland), and one in cropland

in a depression. The number of quadrats in each

line transect is given in Table 1. Four frames (1 m ×

1 m each) with grids (20 cm × 20 cm each) were

horizontally laid in each quadrat. We noted the

number of grids on the ground surface that were

covered by bare rock, continuous soil, rocky soil,

soil in rock cracks and fissures, or soil on rock

Figure 2 Characteristics of karst ecosystems under different land cover types.

J. Mt. Sci. (2014) 11(3): 774-781

777

surfaces. Soil depth was measured in the quadrat

grids containing soil on rock surfaces or rocky soil,

by pressing iron sticks marked at 10-cm interval to

maximum depth (no more than 100 cm). Soil depth

on rock surfaces and in rock cracks and fissures

was estimated as the average value of the collected

soil profiles.

A total of 887 surface soil samples (at depths

of 0–20 cm) were collected, air dried, and weighed.

Gravel (> 2 mm) was removed. 52 and 22 random

soil profiles were plotted from the KER and KNR

sites, respectively. Abnormal profiles were rejected.

The number of soil profiles used for calculations in

each ecosystem is given in Table 2. A total of 431

samples were collected from the soil profiles at

seven depth increments (0–10, 10–20, 20–30, 30–

50, 50–70, 70–90, and 90–100 cm).

The SOC content was determined by wet

digestion with potassium dichromate. The soil bulk

density was determined using cylinder rings

(volume of 100 cm3, length of 5 cm).

1.2 Estimation of SOC density

SOC density (SOCD) was calculated taking into

consideration of soil distribution. Virtual SOCD

was used to calculate the effects of soil distribution

on SOCD, which was calculated without

considering the gravel content or soil distribution.

Virtual SOCD and SOCD were estimated as:

100

)(iii dc

iVP××

= (1)

k

iVPDVL

k

n

m

i∑∑== =1 1

)()( (2)

∑ ××==

q

s

is -iVPpiP1

)%1()()( (3)

k

iPDL

k

n

m

i∑∑== =1 1

)()( (4)

where VP(i) is the virtual SOCD calculated in layer

Table 1 Characteristics of soil distribution under different land cover types

Land cover type

N(1) N(2) m/n Proportion of soil and rock surface to ground surface (%)

Soil depth

Rock Soil RS S(1) S(2) N M

Cropland (D) 1 101 1/101 20.2 79.8 0.0 0.0 0.0 10 99±3Cropland (S) 2 7+11 2/18 43.8 50.4 0 5.7 0 - -Plantation 2 12+0 2/12 49.2 50.8 0 0 0 7 58±9

Grass 2 8+9 2/17 37.8 45.2

10.0

6.4

0.6

30 18

26±1179±17

Scrub 2 11+5 2/16 35.2 55.3

0.1

9.4

0 15 9

25±1257±15

Secondary forest

2 31+23 2/54 91.1 2.3 0.0 0 6.6 11 55±12

Primary forest

3 21+24+21 3/66 57.8 0.0 0 40.0

2.1

596 20

16±882±22

Notes: (D)= land cover types in depressions; (S)=land cover types on slopes; N=the number of total samples for each land cover type; N(1)= Number of line transects; N(2)= Number of quadrats in each line transect; RS= Rocky soil; S(1)= Soil on rock; S(2)= Soil in rock cracks and fissure; M= Mean ± Std. (cm).

Table 2 SOC content under different land cover types

Land cover type

Nos. of surface soil samples

Nos. of soil profiles

SOC (g·kg-1) 0–20 cm 20–50 cm 50–90 cm

Cropland (D) 371 10 13.86±3.90 f 8.50±4.26 b 4.57±1.78 bCropland (S) 123 - 16.69±3.93 e - - Plantation 93 7 18.87±5.28 e 13.39±5.30 ab - Grass 45 15 26.50±6.09 d 13.13±4.10 ab 11.28±3.39 aScrub 48 9 37.48±12.65 c 16.57±5.85 ab - Secondary forest

18 11 48.27±16.78 b 18.35±5.23 a -

Primary forest 20 20 64.33±30.24 a 24.71±11.83 a 12.91±6.48 a

Notes: (S), (D) - see Table 1; SOC data indicate mean ± SD; Different lowercase letters refer to significant differences (P < 0.05).

J. Mt. Sci. (2014) 11(3): 774-781

778

i in a profile (kg m-2); ci is the SOC content (g kg-1);

di is the soil thickness (cm); and ρi is the soil bulk

density (g cm-3). VL(D) is the virtual SOCD in the

0–D cm depth range (kg m-2); n is profile number;

k is the number of profiles for a given land cover

type; and m is the maximum number of layers

included in profile n. For example, SOCD for 0–50

cm is calculated as the sum of SOCD in the 0–10,

10–20, 20–30, and 30–50 cm layers with m equal

to 4. P(i) is the SOCD in layer i in a single profile

within a given land cover type (kg m-2); s is the type

of soil surface (i.e. continuous soil, rocky soil, soil

in rock cracks and fissures, and soil on rock

surfaces); q is the number of soil surface types; ps is

the average proportion of surface s to the ground;

δ% is the gravel content (%); and L(D) is the SOCD

in the 0–D cm depth range (kg m-2).

The change in SOCD (SOC sequestration

potential) associated with the land cover type was

estimated as:

)(-)()( ba iVPiVPiVP =Δ (5)

[ ]∑ ×Δ+×=Δ=

q

s

is -iVPiVPpiP1

)%1()()()( (6)

k

iPDL

k

n

m

j∑∑Δ=Δ = =1 1

)()( (7)

where ΔVP(i) (kg m-2) is the change in virtual

SOCD in layer i, when the land cover type changes

from land cover a to land cover b; ΔP(i) (kg m-2) is

the change of SOCD in layer i; and ΔL(D) (kg m-2)

is the change in SOCD (kg m-2) in the 0–D cm

depth range.

2 Results and Discussion

2.1 Characteristics of soil distribution

Discontinuous soil distribution is one of the

most important features of karst ecosystems. The

proportion of soil on the ground surface in the

tested ecosystems ranged between 0.0% and 79.8%

(Table 1). More than half of the ground surface was

covered by rock in the primary and secondary

forest areas, especially in the latter. In cropland on

slopes, plantation, and scrubland, about half of the

ground surface was covered by continuous soil.

Cropland in depressions showed the highest

proportion of continuous soil (79.8%) and the

thickest soil depths (99±3 cm). Average soil depths

in other ecosystems were less than 1 m (Table 1).

The heterogeneity of soil distribution has

historically led to the differentiation of land use

types in different karstic topographies (Yang et al.

2010). In this study, primary and secondary forests

located on slopes with gradient more than 25° were

mainly covered with rocks and thus were not

suitable for cultivation. In contrast, plantation,

grassland, and scrubland areas converted from

croplands were distributed on slopes with gradient

less than 25° and 45.2–55.3% of the surface was

covered by continuous soil (Table 1). Moreover,

karst soil is generally shallow for its slow pedogenic

processes (Li et al. 1991; Wang et al. 2004), and the

differential weathering of carbonate rocks results

in variability in soil depth (Sun et al. 2002).

2.2 Effect of land cover type and soil

distribution on SOC content

The SOC content in the 0–20 cm and 20–50

cm soil layers, on average, showed a similar trend

with apparently higher values in the surface soil

(Table 2). Soil distribution, land cover type, and

human disturbance had important influences on

the surface SOC content. In this study, surface SOC

content increased with positive vegetation

succession in the following order: cropland with

the highest percentage of soil distribution <

intermediate seral stages of plantation, grassland,

and scrubland, mainly covered by continuous soil

distributions < secondary and primary forests

scarcely covered by continuous soil (Table 1 and 2).

The highest SOC level in primary and secondary

forests may come from the significant

concentration of forest litter. An additional factor is

that human disturbance often leads to SOC loss

(Baker et al. 2007; Jackson et al. 2002). In the

course of positive vegetation succession, the

intensity of disturbance gradually weakened, thus

benefiting for C accumulation.

Surface SOC increased by 13% (P > 0.05) over

10 years when with the sloping cropland was

converted to plantation (Chinese chestnut), and

increased by 58.8% (P < 0.05) after the cropland

was converted to grassland (Table 2). Furthermore,

the SOC content in different soil depths also varied

J. Mt. Sci. (2014) 11(3): 774-781

779

between land cover types. The conversion of

abandoned land to grassland is considered to be an

effective way to improve the SOC content in deep

soil (Baisden and Parfitt 2007; Baisden et al. 2002).

In the 20–50 cm layer, the SOC content in

grassland was not significantly different from that

in the cropland or plantations, but the SOC content

in the 50–90 cm layer in grassland was comparable

to that in primary forest. Thus, at the early stage

the natural restoration (cropland converted to

grassland) in the karst area was beneficial for SOC

sequestration not only in surface soil but also in

deep soil.

2.3 Effect of land cover

type and soil

distribution on

SOC density

In the previous studies

in the karst limestone area

in China, it was estimated

that the average SOC

density (SOCD) within 0–

100 cm soil depth was

13.05 kg m-2 when it was

assumed that the soil is

continuous (Yu et al. 2005).

In consideration of the

discontinuous soil

distribution, the SOCD in

the 0–20 cm and 0–100

cm depths ranged from

0.74 to 4.76 kg m-2 and

from 1.45 to 8.72 kg m-2,

respectively, with the

lowest values occurring in

secondary forest (Table 3).

In the 0–100 cm range,

SOCD values were 4.35 and

5.42 kg m-2 for primary

forest and plantation,

respectively, values which

were lower than those

obtained for grassland,

scrubland on slopes, and

cropland in depressions

(8.21–8.72).

The reality of

discontinuous soil cover in

karst areas cannot be ignored. A high rock outcrop

ratio decreased the SOCD and increased the SOC

content by reducing the surface area covered by

soil (Table 3). Virtual SOCD (without consideration

of the soil distribution) under different land cover

types ranged from 3.33 to 11.49 kg m-2 in the 0–20

cm and from 10.67 to 26.38 kg m-2 in the 0–100 cm

soil depth. SOCD: virtual SOCD ratios showed that

virtual SOCD in the tested karst area was always

higher than actual SOCD (Table 3, the ratios of

SOCD: virtual SOCD were lower than 80%). Thus,

discontinuous soil distribution negatively impacted

on SOC storage within the karst area.

Table 3 Virtual and essential SOC density for each land cover type

Land cover type

Virtual SOCD

(kg·m-2) SOCD (kg·m-2)

SOCD/virtual SOCD (%)

0–20 cm

0-100 cm

0–20 cm

0-100 cm

0–20 cm

0-100 cm

Cropland (D) 3.33 10.96 2.66 8.72 79.8 79.6Cropland (S) 3.70 - 2.08 - 56.2 Plantation 4.60 10.67 2.34 5.42 50.8 50.8Grass 6.04 15.37 3.75 8.21 62.2 53.4Scrub 7.33 13.99 4.76 8.52 64.8 60.9Secondary forest 8.37 16.31 0.74 1.45 8.9 8.9Primary forest 11.49 26.38 4.04 4.35 35.2 16.5

Notes: (D), (S) - see Table 1.

Table 4 Potential for SOC sequestration resulting from assumed land cover conversion (all land cover types in this table are located on slopes)

Land cover conversion Assumed time (yrs)

Soil depth (cm)

SOC sequestration

Potential (kg·m-2)

Rate (g·m-2·y-1) from to

Cropland

Plantation 10 0–20 0.50 50.0

Grass 5–30 0–20 1.31 43.7–262

Scrub >50 0–20 2.04 <40.8

Plantation

Scrub >50 0–20 1.39 <27.8

0–100 2.17 <43.4

S_forest >100 0–20 1.91 <19.1

0–100 3.59 <35.9

Grass

Scrub >50 0–20 0.81 <16.2

0–100 1.08 <21.6

S_forest >100 0–20 1.45 <14.5

0–100 2.12 <21.2

Scrub

S_forest >100 0–20 0.67 <6.7

0–100 1.18 <11.8

P_forest >100 0–20 2.69 <26.9

0–100 4.31 <43.1

S_forest P_forest >100 0–20 0.28 <2.8

0–100 0.44 <4.4

Notes: S_forest=Secondary forest; P_forest=Primary forest.

J. Mt. Sci. (2014) 11(3): 774-781

780

2.4 Potential for SOC sequestration with

the conversion of land cover type

Based on the space-for-time substitution

method (Pickett 1989), rates of SOC accumulation

related to the conversion of the studied ecosystems

ranged from 2.8 to 50.0 and from 4.4 to 43.4 g m-2

y-1 in the 0–20 and 0–100 cm depth range,

respectively. The estimated accumulation rate

decreases in the positive successional sequence:

cropland > plantation > grassland > scrubland >

secondary forest (based on minimum time for

conversion of ecosystems, Table 4). This suggests

that the SOC accumulation rate decreases with

advancing seral stage. Although plantation and

scrubland ecosystems had higher potential for SOC

sequestration, longer period was required for these

ecosystems to transit to the next successional stage.

Our findings suggest that the transition from

cropland to grassland has the highest potential for

SOC sequestration within a relatively short period

(a decade or two). Compared with those of non-

karst ecosystems in subtropical areas (Post and

Kown 2000), the SOC accumulation rates are

generally lower since more time is needed for

ecological restoration in the case of karst

ecosystems (Table 4).

In this study, the method for investigating the

soil distribution was imprecise because the soil

distribution condition at the surface did not well

represent that below the surface. Ground-

penetrating radar (GPR) technology has been

successfully used in karst areas for archaeological

and hydrological studies (Al-fares et al. 2002;

Porsani et al. 2010). Our future work will thus

include GPR technology to determine the soil and

rock distribution in the karst areas.

3 Conclusions

Within karst areas, SOC stocks are often

overestimated due to failure to take into

consideration of the discontinuous soil distribution.

In this study, a new approach accounting this

factor was used to calculate SOC content. The SOC

content was found to increase with long-term

ecological recovery and the land cover type

transited from cropland to grassland, scrubland,

secondary forest, and primary forest. When

following this successional order, SOCD within the

0–100 cm depth range remained low and even

declined, due to the unique soil distribution of

karst areas. Increasing SOC after the degradation

of karst ecosystems is a slow process. As a whole,

the early stage of natural restoration (conversion of

cropland to grassland) was found to be efficient for

SOC sequestration.

Acknowledgement

This study was supported by the Strategic

Priority Research Program of the Chinese Academy

of Sciences (Grant No. XDA05070403), the

National Natural Science Foundation of China

(Grant Nos. 41171246, 41301273), and the National

Science-technology Support Plan Projects (Grant

No. 2012BAD05B03-6).

References

Al-fares W, Bakalowicz M, Guéin R (2002) Analysis of karst aquifer structure of the Lamalou area (Hérault, France) with ground penetrating radar. Journal of Applied Geophysics 51:97-106. DOI: 10.1016/S0926-9851(02)00215-X. Baker JM, Ochsner TE, Venterea RT, et al. (2007) Tillage and soil carbon sequestration—What do we really know? Agriculture, Ecosystems and Environment, 118:1-5. DOI: 10.1016/j.agee.2006.05.014. Baisden WT, Amundson R, Brenner DL, et al. (2002) A multiisotope C and N modeling analysis of soil organic matter turnover and transport as a function of soil depth in a California annual grassland soil chronosequence. Global Biogeochemal Cycles 16: 1135. DOI: 10.1029/2001GB001823. Baisden WT, Parfitt RL (2007) Bomb 14C enrichment indicates

decadal C pool in deep soil? Biogeochemistry 85:59-68. DOI: 10.1007/s10533-007-9101-7.

Chen L, Gong J, Fu B, et al. (2007) Effect of land use conversion on soil organic carbon sequestration in the loess hilly area, loess plateau of China. Ecological Research 22:641-648. DOI: 10.1007/s11284-006-0065-1.

Jackson RB, Banner JL, Jobbágy EG, et al. (2002) Ecosystem carbon loss with woody plant invasion of grasslands. Nature 418:623-626. DOI:10.1038/nature00910.

Karchegani PM, Ayoubi S, Mosaddeghi MR, et al. (2012) Soil organic carbon pools in particle-size fractions as affected by slope gradient and land use change in hilly regions, western Iran. Journal of Mountain Science 9:87-95. DOI: 10.1007/ s11629-012-2211-2.

J. Mt. Sci. (2014) 11(3): 774-781

781

Krogh L, Noergaard A, Hermansen M, et al. (2003) Preliminary estimates of contemporary soil organic carbon stocks in Denmark using multiple datasets and four scaling-up methods. Agriculture, Ecosystems and Environment 96:19-28. DOI: 10.1016/S0167-8809(03)00016-1. Laganière J, Angers DA, Paré D (2010) Carbon accumulation in agricultural soils after afforestation: a meta-analysis. Global Change Biology 16:439-453. DOI: 10.1111/j.1365-2486.2009. 01930.x. Li J, Wang C, Fan T (1991) Weathering crust of carbonate rocks and process of karst earth formation. Carsologica Sinica 10: 2938. (In Chinese) Li YQ, Mathews BW (2010) Effect of conversion of sugarcane plantation to forest and pasture on soil carbon in Hawaii. Plant and Soil 335:245-253. DOI: 10.1007/s11104-010-0412-4. Pickett STA (1989) Space-for-time substitution as an alternative to long-term studies. Long-Term Studies in Ecology 110-135. Porsani JL, Jangelme GM, Kipnis R (2010) GPR survey at Lapa do Santo archaeological site, Lagoa Santa karstic region, Minas Gerais state, Brazil. Journal of Archaeological Science 37:1141-1148. DOI: org/10.1016/j.jas.2009.12.028. Post WM, Kown KC (2000) Soil carbon sequestration and land-use change: processes and potential. Global Change Biology 6:317-327. DOI: 10.1046/j.1365-2486.2000.00308.x. Scott NA, Tate KR, Ford-robertson J, et al. (1999) Soil carbon storage in plantation forests and pastures: land-use change implications. Tellus B 51:326-335. DOI: 10.1034/j.1600-0889. 1999.00015.x. Sun CX, Wang SJ, Zhou DQ, et al. (2002) Differential weathering and pedogenetic characteristics of carbonate rocks and their effect on the development of rock desertification in karst regions. Acta Mineralogica Sinica 22:308-314. (In

Chinese) Wang SJ, Liu QM, Zhang DF (2004) Karst rocky desertification in southwestern China: geomorphology, landuse, impact and rehabilitation. Land Degradation and Development 15:115-121. DOI: 10.1002/ldr.592.

Yang YS, Xie JS, Sheng H, et al. (2009) The impact of land use/cover change on storage and quality of soil organic carbon in midsubtropical mountainous area of southern China. Journal of Geographical Sciences 19:49-57. DOI: 10.1007/s11442-009-0049-5.

Yang Z, Yang L, Zhang B (2010) Soil erosion and its basic characteristics at karst rocky-desertified land consolidation area: A case study at Muzhe Village of Xichou County in Southeast Yunnan, China. Journal of Mountain Science 7:5572. DOI: 10.1007/s11629-010-1047-x.

Yonekura Y, Ohta S, Kiyono Y, et al. (2010) Changes in soil carbon stock after deforestation and subsequent establishment of "Imperata" grassland in the Asian humid tropics. Plant and Soil 329:495507. DOI: 10.1007/s11104-009-0175-y.

Yu DS, Shi XZ, Sun WX, et al. (2005) Estimation of China soil organic carbon storage and density based on 1:1 000 000 soil database. Chinese Journal of Applied Ecology 16:2279-2283. (In Chinese)

Zhang W, Chen H, Wang K, et al. (2007) The heterogeneity and its influencing factors of soil nutrients in peak-cluster depression areas of karst region. Agricultural Sciences in China 6:322-329. DOI: 10.1016/S1671-2927(07)60052-2.

Zheng H, Su Y, He X, et al. (2012) Modified method for estimating the organic carbon density of discontinuous soils in peak-karst regions in southwest China. Environmental Earth Sciences 67:1743-1755.DOI:10.1007/s12665-012-1618-y.