REGULAR ARTICLE

Soil organic carbon losses due to land use changein a semiarid grassland

Liping Qiu & Xiaorong Wei & Xingchang Zhang &

Jimin Cheng & William Gale & Chao Guo &

Tao Long

Received: 10 June 2011 /Accepted: 7 December 2011 /Published online: 5 January 2012# Springer Science+Business Media B.V. 2012

AbstractBackground and Aims Knowledge about the effectof land use change on soil organic carbon (OC) insemiarid grassland is essential for understandingC cycles and for forecasting ecosystem C sequestra-tion. Our objectives were (1) to study the effect ofland use change on aggregate size distribution,aggregate-associated OC concentrations, and aggregate-associated stocks in a semiarid grassland area and (2) torelate changes in the aggregate fractions to changes intotal soil OC.Methods Cropland and shrubland plots were estab-lished in a semiarid grassland area in 1982. We col-lected soil samples from adjacent grassland, cropland,and shrubland plots 27 years later and measured OCconcentrations in the macroaggregate (>0.25 mm),microaggregate (0.25–0.053 mm) and silt+clay(<0.053 mm) fractions.

Results Total soil OC concentrations and stocks de-creased significantly after the grassland was convertedto cropland or shrubland. Soil microbial biomass C, rootbiomass, and root C also declined. The proportion of soilin themacroaggregate fraction decreased after conversionto cropland or shrubland. Decreases in macroaggregate-associated OC stocks accounted for more than half of theOC losses that occurred when grassland was converted tocropland. The decreases in macroaggregate-associatedOC stocks were due to declines in bothmacroaggregationand macroaggregate-associated OC concentrations afterconversion to cropland. In contrast, decreases inmicroaggregate-associated OC stocks accounted formore than half of the OC losses when grassland wasconverted to shrubland. The declines in microaggregate-associated OC stocks were primarily due to a decrease inmicroaggregate-associated OC concentrations after con-version to shrubland.Conclusions Land use changed caused significantdecreases in soil OC stocks. Conversion to croplandsoil resulted in large decreases in macroaggregate-associated OC stocks whereas conversion to shrublandresulted in large decreases in microaggregate-associated OC stocks. Any changes in land use insemiarid grasslands could cause the grassland soil tobecome a source of atmospheric CO2; therefore ex-treme caution should be taken to avoid this hazard.

Keywords Aggregates . Cultivation . Semiaridgrassland . Shrub establishment . Soil organic carbon

Plant Soil (2012) 355:299–309DOI 10.1007/s11104-011-1099-x

Responsible Editor: Per Ambus.

L. Qiu :X. Wei (*) :X. Zhang : J. ChengState Key Laboratory of Soil Erosion and Dryland Farmingin the Loess Plateau, Northwest A & F University,Xinong Road, #26,Yangling, Shaanxi Province, China 712100e-mail: xrwei78@163.com

W. Gale : C. Guo : T. LongCollege of Resources and Environment,Northwest A & F University,Yangling, Shaanxi Province 712100, China

Introduction

Semiarid and arid regions cover about 41% of theearth’s land surface and are home to more than 38%of the global population (GLP 2005; MEA 2005).These regions are particularly sensitive to global cli-mate change and human activity (Schlesinger et al.1990; Reynolds et al. 2007). As the predominant eco-system in semiarid and arid regions (Sala et al. 1997),grasslands play an important role in preventing deserti-fication and soil erosion in semiarid regions (Souchereet al. 2003; Kim et al. 2006; Li et al. 2007; Zuo et al.2009). However, as demands for food and other agri-cultural commodities rise, semiarid grasslands are in-creasingly converted to other land uses.

Land use change in semiarid grasslands generallyinvolves either crop cultivation or the establishment ofshrub vegetation (Archer 1994; Wang 2000; John et al.2009). In many parts of northern China, especially inthe semiarid agropastoral zone, natural grasslandshave been widely converted to cropland during thepast century (Wang 2000). This has accelerated soilerosion and land degradation (Zhao et al. 2005). Semi-arid and arid grasslands worldwide have also beenwidely invaded by shrubs over the past century (Archer1994). However, the effects of land use change on soilorganic C (OC) in semiarid grassland regions has notbeen well documented. This information is essential forunderstanding C biogeochemistry and forecasting eco-system C sequestration in these habitats.

The cultivation of grasslands usually leads to soilOC loss (Houghton et al. 1999; Malhi et al. 2003;Wang et al. 2009). The cultivation of forest and grass-land released about 25 Pg C into the atmospherebetween 1700 and 1990 (Houghton et al. 1999). Theeffect of shrub encroachment on OC in grassland soilsis extremely variable (Wessman et al. 2004). Jacksonet al. (2002) showed that invasion by shrubs increasedsoil OC in dry areas, but decreased soil OC in wetareas. Wei et al. (2009) reported that soil OC declinedafter the establishment of a tree-shrub mixture on agrassland. The effects of land use change on soil OCare mainly due to changes in OC input and C miner-alization (Post and Kwon 2000; Guo and Gifford2002; Strickland et al. 2010). The cultivation andafforestation of grasslands can increase C mineraliza-tion and decrease root biomass in semiarid grassland(Elliott et al. 1996; Guo et al. 2007; Wei et al. 2009).Therefore, we hypothesized that converting grassland

to cropland or shrubland would lead to decreases insoil OC.

The OC in soil physical fractions responds morequickly to land use change than the OC in bulk soils(John et al. 2005; Hoyos and Comerford 2005;Ashagrie et al. 2007; Paul et al. 2008). Understandingthe response of aggregate-associated OC is thereforefundamental to assessing the effect of land use changeon soil OC. Ashagrie et al. (2005) reported significantchanges in aggregate-associated OC concentrations,but not bulk soil OC concentrations, after changes inland use and management. Li and Pang (2010)reported substantial changes in the OC concentrationof different aggregate fractions following land usechange in the southern Loess Plateau. They observedthat the aggregate-associated OC concentration at the0–20 cm depth decreased following conversionof grassland to cropland, but increased following con-version from cropland to forests and orchards.Ashagrie et al. (2007) reported that the cultivation ofnative forest areas in Ethiopia had a more pronouncedeffect on macroaggregate-associated OC than onmicroaggregate-associated OC. Steffens et al. (2009)suggested that greater inputs of organic matter led togreater macroaggregate-associated OC concentrations.Ashagrie et al. (2005) also observed that land use andmanagement changes affected macroaggregate-associated OC more than microaggregate-associatedOC. All these studies suggest that macroaggregate-and macroaggregate-associated OC are sensitive toland use change. We therefore hypothesized that con-verting semiarid grassland to cropland or shrublandwould cause a decrease in total OC due to a decline inmacroaggregation and a loss of macroaggregate-associated OC.

In order to test our hypotheses, we investigatedsoil OC distributions under three land uses in asemiarid region: grassland, cropland, and shrubland.The crop and shrublands were established on exist-ing grassland 27 years ago. We measured OC con-centrations in macroaggregates, microaggregates,and silt+clay fractions, and then calculated OCstocks in each aggregate fraction to determine theeffect of land use change on soil OC. The objec-tives of the study were (1) to quantify OC concen-trations and stocks in each aggregate fraction afterland use change in semiarid grassland and (2) torelate changes in the aggregate fractions to changesin whole soil OC.

300 Plant Soil (2012) 355:299–309

Materials and methods

Study sites

This study was conducted in the Yunwushan naturalgrassland protection zone (36°13′–36°19′N; 106°24′–106°28′E) at Guyuan City, Ningxia Hui AutonomousRegion, China. The site is located in the center of theLoess Plateau. The grassland protection zone was setup in 1984 and has an area of 4000 km2 and anelevation ranging from 1800 to 2148 m. The studyarea has a continental monsoon climate. The meantemperature is 6.9°C. The maximum and minimumtemperatures occur in July (24°C) and January (−14°C), respectively. The frost-free period is 124 d, nor-mally beginning in mid-April and ending in late Sep-tember. The mean annual precipitation is 425 mm. Thesoil in the study area is a mountain grey-cinnamon soilclassified as a Calci-Orthic Aridisol according to theChinese Taxonomy system, equivalent to a HaplicCalcisol in the FAO/Unesco system.

Field investigation and sampling

A long-term study was set-up within the grasslandprotection zone in 1982. Three adjacent areas withsimilar topography and slope were selected. One areawas kept as grassland. The vegetation was primarilybunge needlegrass (Stipa bungeana) and Russianwormwood (Artemisia sacrorum Ledeb.). A secondarea was converted to shrubland. The predominantvegetation in the shrubland was korshinsk peashrub(Caragana korshinskii Kom). Bunge needlegrass grewbetween the shrubs. The third area was converted tocropland. The cropland was primarily used for maize(Zea mays L.) and potato (Solanum tuberosum) pro-duction. All the land use types have the same soil typeand similar physiographic conditions and slope gra-dients. We therefore assume that the soils in the threeland use types had similar initial conditions. The con-centrations and stocks of OC, total N, and total P in thegrassland soil before land use change in 1982 areshown in Table 1.

Five pseudo-replicated plots (30 m×30 m) wererandomly established within each land use treatmentin 2009. True replication was not possible in thisstudy. For the grassland plots, five 2 m×2 m subplotswere established in each plot to measure the canopycover and aboveground biomass of the grasses. The

average canopy cover was 52%. The average above-ground biomass was 1.36 Mg ha−1 for bunge needle-grass and 0.55 Mg ha−1 for Russian wormwood. Forthe shrubland plots, five korshinsk peashrub plantswere randomly selected in each plot. The averageheight of the peashrub plants was 1.2 m. The averagestem diameter at ground level was 1.3 cm. The aver-age canopy width in two cardinal directions was106 cm×89 cm. For the cropland soil, there was nocrop on the field when the samples were collected. Theprevious crop (i.e., tomato, Lycopersicurn esculentumMill) was harvested one month before sampling.

Soil bulk density was measured in each plot at the0–10 and 10–20 cm depths using a 5.0 cm high by5.0 cm diameter stainless steel cutting ring. The soilcores were dried at 105°C for 24 h. Three representa-tive soil samples were randomly collected in each plotfor the measurement of microbial biomass C, aggre-gate size distribution, and soil OC. The samples werecollected at the 0–10 and 10–20 cm depths with a5.0 cm diameter tube auger. Visible pieces of organicmaterial were removed, and the moist field soil sam-ples were brought to the laboratory. A subsample wasstored at 4°C and microbial biomass C (MBC) meas-urements were made within two days of collection.The remaining soil was air-dried.

Three additional samples were collected from eachplot for the measurement of root biomass. The samples(0–10 and 10–20 cm depths) were collected near to thesampling locations described in the previous para-graph using a 9.0 cm diameter tube auger. Roots wereseparated from the soil samples, washed with deion-ized water, and dried at 65°C for 48 h. The dry rootswere weighed and then ground to pass through a0.25 mm sieve for C measurement.

Laboratory analysis and data analysis

Soil MBC was measured using the fumigation-extraction method (Brookes et al. 1985; Wu et al.

Table 1 The concentrations and stocks of OC, total N, and totalP in the grassland soil (0–20 cm depth) before land use changein 1982 (YNGPZB 2001)

Organic C Total N Total P

Concentration (g kg−1) 11.8 0.9 0.3

Stocks (Mg ha−1) 28.2 2.2 0.8

Plant Soil (2012) 355:299–309 301

1990). The extracts were analyzed with a Phoenix-8000 TOC Analyzer (Tekmar-Dohrmann, US). TheMBC values were calculated as follows (Wu et al.1990):

MBC mg kg�1� � ¼ OC fumigated soilð Þ � OCunfumigated soilð Þ½ �=Kc

where the conversion factor Kc00.45 (Wu et al. 1990).Aggregate size classes were separated by wet

sieving through 0.25 and 0.053 mm sieves follow-ing the procedures described by Cambardella andElliott (1993). A 100 g air-dried and unground soilsample was spread on top of a 0.25 mm sievesubmerged in deionized water. Soil samples wereleft immersed in the water for 10 min and thensieved by moving the sieve 3 cm vertically 50times over a period of 2 min. The material remain-ing on the 0.25 mm sieve was transferred to a glasspan. The soil plus water that passed through thesieve was poured onto the 0.053 mm sieve and theprocess repeated. The macroaggregate (>0.25 mm),microaggregate (0.25–0.053 mm), and silt+clay(<0.053 mm) fractions were dried in an oven at50°C for 24 h and then weighed.

A sub-sample of air-dried, undisturbed soil fromeach site was ground to pass through a 0.25 mm sieveto measure whole soil OC concentration. The OCconcentrations of the aggregate fractions, whole soils,and root samples were analyzed using a VARIO EL IIICHON analyzer (Elementar, Germany) at the Testingand Analysis Center of Northwest University, China.

Stocks of soil OC (Mg C ha−1) were calculated asfollows:

Stocks of OC ¼ D� BD� OC=10

where D is the thickness (cm) of the soil layer, BD isthe bulk density (g cm−3), and OC is the OC concen-tration (g kg−1) of the 0–10 and 10–20 cm soil depths.

Stocks of OC (Mg C ha−1) in each size fractionwere calculated as follows:

Stocks of OCi ¼ D� BD�Mi � OCi=10000

where Mi is the aggregate mass in the ith size fraction(g kg−1), and OCi is the OC concentration of the ith sizefraction (g kg−1 aggregate). The conversion factor be-tween m2 and ha is 10000. Changes in the OC stocks ofeach aggregate fraction contributed to the overallchange in whole soil OC stocks. The contribution of

each aggregate faction to the total change in soil OCstock was calculated on a percentage basis.

The OC stocks of the grassland soil in 2009(28.2 Mg ha−1) were the same as those when theland use changes were implemented in 1982(28.1 Mg ha−1). Therefore we assumed that therewas no change in the aggregate size distribution orthe aggregate associated OC of the grassland soilduring the experiment. We used the aggregate sizedistribution and aggregate associated OC of thegrassland in 2009 to represent the soil conditionsof the cropland and shrubland plots in 1982. Weassumed that changes in OC stock within anyparticular aggregate fraction were caused both bychanges in OC concentration in the fraction (F1)as well as by changes in the mass of the fraction(F2). We also assumed that the mass that wasgained or lost from an aggregate fraction due toland use change had the same OC concentration asthe rest of that fraction after land use change. Wetherefore calculated the contribution of F1 and F2to the total change in OC stock within an aggre-gate fraction as follows:

F1 ¼ M�ΔC=1000F2 ¼ ΔM� C=1000

where 1) F1 is the change in the OC stock (Mg Cha−1) within an aggregate fraction due to changesin aggregate-associated OC concentration and F2 isthe change in the OC stock (Mg C ha−1) within anaggregate fraction due to changes in aggregatemass; 2) ΔM is the change in the mass of acertain fraction (Mg ha−1); 3) M is the initial massof the aggregate fraction (Mg ha−1) before landuse change; 4) C is the final OC concentrationof the aggregate fraction (g kg−1) after land usechange; 5) ΔC is the change in OC concentrationin this aggregate fraction (g kg−1) due to land usechange; and 6) the conversion factor betweengrams and kilograms is 1000.

Two-way analysis of variance (ANOVA) was con-ducted using SAS version 8 to test the effects of landuse and soil depth on 1) soil OC concentrations andstocks; 2) MBC; 3) root biomass and root C; 4) ag-gregate size distribution; and 5) OC concentrationsand stocks associated with each aggregate fraction.The least significant difference (LSD) between anytwo means was calculated using the Student’s t testat the 5 percent level.

302 Plant Soil (2012) 355:299–309

Results

Decreases in soil OC due to land use change

Soil OC concentrations and stocks decreased by morethan half after the conversion of grassland to croplandor shrubland (Fig. 1, Table 2). The decreases weresimilar at both depths (0–10 and 10–20 cm). Specifi-cally, the OC concentrations and stocks of both thecropland and the shrubland decreased by 57% in the0–10 cm depth and by 61% in the 10–20 cm depth.

Soil MBC in 0–10 cm depth was significantlylower in crop and shrubland compared to grassland(Fig. 1, Table 2). The MBC was greater in the 0–10 cmdepth than in the 10–20 cm depth in both grasslandand shrubland soil. However, in cropland soil, theMBC of the 0–10 cm depth was less than that of the10–20 cm depth. The decreases in MBC indicate thatthe conversion of grassland to cropland or shrublanddecreased the availability of soil OC and degraded thesoil microbial environment.

Conversion from grassland to cropland and shrublandresulted in significant reductions in root biomass and rootC (Fig. 2, Table 2). Root biomass in the 0–20 depth of thegrassland was 4.43 Mg ha−1 and root C was 1.74 Mg Cha−1. Root biomass in the 0–20 cm depth of the shrublandwas 54% less than that of the grassland and root C was61% less. Root biomass and root C in croplandwere evenless than in shrubland, with values approaching zero (rootbiomass, 0.02 Mg ha−1; root C, 0.01 Mg C ha−1).

Changes in soil OC associated with aggregates

The conversion of grassland to cropland significantlyreduced the proportion of soil in the macroaggregatesize class and increased the proportion of soil in themicroaggregate size class at both the 0–10 and 10–20 cm depths (Fig. 3, Table 2). The conversion ofgrassland to cropland reduced the macroaggregatecontent by 78 to 87%, but increased the microaggre-gate content by 38 to 44%. Less dramatic changeswere observed when grassland was converted toshrubland. Specifically, the conversion of grasslandto shrubland reduced the macroaggregate content by15 to 22%, but increased the microaggregate contentby 5 to 12%. Conversion to either cropland or shrub-land had no significant effect on the proportion of soilin the silt+clay fraction. The results imply that theconversion of grassland to cropland has greater effect

on macroaggregation than the conversion of grasslandto shrubland.

Land use change caused significant decreases inaggregate-associated OC concentrations (Fig. 3, Ta-ble 2). The greatest decreases were for macroaggregatesin the 0–10 cm soil depth. Macroaggregate-associated

0

4

8

12

160-10cm 10-20cm

0

4

8

12

16

0

200

400

600

800

1000

OC

con

cent

rati

on (

g kg

-1)

OC

sto

ck (

Mg

C h

a-1)

MB

C (

mg

kg-1

)

Grassland Cropland Shrubland

a

b

cc

c

c

a

a

bb

b

b

a

bb

b

c

bc

Fig. 1 The effects of land use change on soil organic carbon(OC) concentrations, OC stocks, and soil microbial biomass C(MBC). Bars with different letters within each panel are differ-ent at P<0.05

Plant Soil (2012) 355:299–309 303

OC concentrations in the 0–10 cm depth of cropland andshrubland soil were 8.7 to 11.5 gkg−1 less than that ofgrassland soil. In comparison, conversion to cropland orshrubland reduced microaggregate-associated OC con-centrations in the 0–10 cm depth by 8.3 to 8.6 gkg−1 andsilt+clay-associated OC concentrations by 6.9 to 7.2 gkg−1. The decrease in macroaggregate-associated OCconcentration was larger when grassland was convertedto cropland than when grassland was converted toshrubland. The decreases in microaggregate- and silt+clay-associated OC concentrations were similar forboth cropland and shrubland (Fig. 3).

Land use change also caused significant decreasesin aggregate-associated OC stocks, but the amount ofdecrease varied among aggregate fractions and landuse types. Losses in macroaggregate-associated OCstocks were greater when grassland was converted tocropland than when grassland was converted to shrub-land. The macroaggregate-associated OC stocks ofcropland soil were 6.1 Mg C ha−1 less than those ofgrassland soils at the 0–10 cm depth and 4.6 Mg Cha−1 less than those of grassland soils at the 10–20 cmdepth. For shrubland soil, macroaggregate-associatedOC stocks were 4.1 Mg C ha−1 less than those ofgrassland soils at the 0–10 cm depth and 2.5 Mg Cha−1 less than those of grassland soils at the 10–20 cmdepth. In contrast to the above findings, the conversionof grassland to shrubland resulted in significant losses

of microaggregate- and silt+clay-associated OCstocks. The microaggregate OC stocks of shrublandsoil were 5.5 Mg C ha−1 less than those of grasslandsoil at the 0–10 cm depth and 5.0 Mg C ha−1 less thanthose of grassland soil at the 10–20 cm depth. Silt+clay-associated OC stocks in shrubland soil declinedby 0.5 Mg C ha−1 at the 0–10 cm depth and by0.3 Mg C ha−1 at the 10–20 depth. For cropland soil,microaggregate-associated OC stocks were 4.5 Mg Cha−1 less than those of grassland soil at the 0–10 cmdepth and 3.9 Mg C ha−1 less than those of grasslandsoil at the 10–20 cm depth. Silt+clay-associated OCstocks of cropland soil decreased by 0.3 Mg C ha−1 atthe 0–10 cm depth and by 0.2 Mg C ha−1 at the 10–20 cm depth. These results suggest that land usechange results in substantial loss of OC from soilaggregates. The losses in OC from the 0–10 cm depthwere greater than those from the 10–20 cm depth.Furthermore, the macroaggregate fraction lost themost OC when grassland was converted to croplandwhereas the microaggregate fraction losses were great-est when grassland was converted to shrubland.

Discussion

Our results demonstrated that whole soil OC concen-trations and stocks were significantly less in cropland

Table 2 Analysis of variance results for all the variables

Land use Soil depth Land use × Soil depth

F P F P F P

OC concentration 49.5 <.0001 9.0 0.0110 0.34 0.7184

OC stock 48.3 <.0001 6.1 0.0298 0.25 0.7832

MBC 11.7 0.0015 3.8 0.0753 6.49 0.0123

Root biomass 46.0 <.0001 25.6 0.0003 6.43 0.0126

Root C 47.3 <.0001 17.8 0.0012 4.53 0.0342

Macroaggregate mass 33.3 <.0001 0.3 0.6149 0.59 0.5702

Microaggregate mass 31.4 <.0001 1.5 0.2441 0.38 0.6902

Silt+clay mass 3.4 0.0665 0.0 0.8835 1.03 0.3860

Macroaggregate-associated OC concentration 84.2 <.0001 16.7 0.0015 8.67 0.0047

Microaggregate-associated OC concentration 87.7 <.0001 7.5 0.0179 0.70 0.5149

Silt+clay-associated OC concentration 33.8 <.0001 0.3 0.6003 0.13 0.8836

Macroaggregate-associated OC stock 128.7 <.0001 6.7 0.0235 3.06 0.0846

Microaggregate-associated OC stock 36.6 <.0001 2.8 0.1220 0.15 0.8644

Silt+clay-associated OC stock 29.1 <.0001 1.9 0.1938 0.47 0.6339

304 Plant Soil (2012) 355:299–309

and shrubland than in grassland (Table 2), supportingour hypothesis that the conversion of semiarid grass-lands to cropland or shrubland decreases soil OC.Losses in OC due to cultivation were mainly causedby declines in macroaggregate-associated OC, where-as losses in OC due to conversion to shrubland weremainly due to declines in microaggregate-associatedOC. Conversion of grassland to cropland resulted in a58% decline in soil OC stocks at the 0–20 cm depth,whereas conversion to shrubland resulted in a 60%decline in soil OC stocks. These declines, which cor-responded to losses of 0.60 Mg C ha−1 yr−1 in crop-land and 0.63 Mg C ha−1 yr−1 in shrubland, weregreater than those observed after land use change ina more arid region but less than those observed afterland use change in a more humid region (Guo et al.2007; Malhi et al. 2003; Wang et al. 2009; Wei et al.2009). Declines in OC stocks due to land use change

have been attributed to decreases in OC inputs (Guoand Gifford 2002; Guo et al. 2007; Wei et al. 2009)and increases in OC mineralization (Skjemstad et al.2008; Nyamadzawo et al. 2009).

Declines in root C accounted for only about 10% ofthe OC stock losses in cropland soil and 7% of the OCstock losses in shrubland soil. This indicates thatdecreases in root C contributed little to overall lossesin OC after conversion of grassland to cropland orshrubland. Therefore, the decrease in soil OC is mostlikely due to enhanced OC mineralization, which var-ied among aggregate size classes.

In agreement with previous reports, OC concentra-tions and OC stock decreased significantly in all aggre-gate size fractions after grassland was converted tocropland or shrubland (Ashagrie et al. 2005; Ashagrieet al. 2007; Steffens et al. 2009; Li and Pang 2010).Losses in macroaggregate-associated OC stockaccounted for than half of the OC loss after grasslandwas converted to cropland. The losses in macroaggre-gate OC stock resulted from declines in both macroag-gregate content and macroaggregate-associated OCconcentration. In the 0–10 cm layer, declines inmacroaggregate-associated OC stocks accounted for56% of the total loss in soil OC stocks after conversionto cropland, declines in microaggregate-associated OCstocks accounted for 42% of the total loss, and declinesin silt+clay-associated OC stocks accounted for 3% ofthe total loss. In the 10–20 cm layer, declines inmacroaggregate-associated OC stocks accounted for53% of the total loss in soil OC stocks after convertingto cropland, declines in microaggregate-associated OCstocks accounted for 45% of the total loss, and declinesin silt+clay-associated OC stocks accounted for 3% ofthe total loss. These results were consistent with otherstudies that showed that decreases in soil OC stocksafter land use change were mainly due to declines inmacroaggregate-associated OC (Ashagrie et al. 2005,2007; Steffens et al. 2009),

Changes in the microaggregate fraction were themajor contributor to OC loss following the conversionof grassland to shrubland. In the 0–10 cm layer, declinesin macroaggregate-associated OC stocks accounted for40% of the total loss in soil OC stocks after conversionto cropland, declines in microaggregate-associated OCstocks accounted for 55% of the total loss, and declinesin silt+clay-associated OC stocks accounted for 5% ofthe total loss. In the 10–20 cm layer, declines inmacroaggregate-associated OC stocks accounted for

0

1

2

3

4

0-10cm 10-20cm

0

1

2

Roo

t bio

mas

s (M

g ha

-1)

Roo

t C (

Mg

C h

a-1)

Grassland Cropland Shrubland

a

b

b

cc c

a

b b

cc c

Fig. 2 The effects of land use change on root biomass and rootC in the 0–20 cm soil depth. Bars with different letters withineach panel are different at P<0.05

Plant Soil (2012) 355:299–309 305

32% of the total loss in soil OC stocks after convertingto shrubland, declines in microaggregate-associated OCstocks accounted for 64% of the total loss, and declinesin silt+clay-associated OC stocks accounted for 4% ofthe total loss.

The changes in aggregate-associated OC stock areattributable to two factors: (1) changes in the soil masswithin a given aggregate fraction and (2) changes inaggregate-associated OC concentrations within the

fraction. Although the effects of land use change onaggregate size distribution and aggregate-associatedOC concentrations have been reported (John et al.2005; Hoyos and Comerford 2005; Ashagrie et al.2007; Paul et al. 2008), the relative contribution ofthese two factors to changes in soil OC has rarely beenquantified (Noellemeyer et al. 2008). This is essentialfor understanding how soil OC responds to land usechange.

0

300

600

900

1200

1500

0-10cm 10-20cm 0-10cm 10-20cm 0-10cm 10-20cm

510.0<52.0-510.052.0>

grassland

crop land

shrub land

0

5

10

15

20

0-10cm 10-20cm 0-10cm 10-20cm 0-10cm 10-20cm

510.0<52.0-510.052.0>

0

2

4

6

8

10

12

0-10cm 10-20cm 0-10cm 10-20cm 0-10cm 10-20cm

C&SAFAC

Agg

rega

te a

mou

nt (

Mg

ha-1

)O

C c

once

ntra

tion

(g

kg-1

)O

C s

tock

(M

g C

ha-1

)

Grassland

Cropland

Shrubland

ac

bd cd

abac

ab

a

dd

cd

c

b

a

d d

cc

b

b

aa

bb b

a

c ccc

b

a

bb b

b

a

ab a ab abb ab

ab

c

bc bcc

a

a b bc cca

Macroaggregate Microaggregate Silt+Clay

Fig. 3 The effects of landuse change on organic C(OC) concentrations and OCstocks in soil aggregatefractions. Bars with differentletters within each aggregatesize are different at P<0.05

Table 3 Changes in the OC stocks associated with each aggre-gate fraction after the conversion of grassland to cropland orshrubland. F1 represents the change in OC stock that wasattributable to changes in the aggregate-associated OC

concentration of the fraction after land conversion. F2 representsthe change in OC stock that was attributable to changes in themass of the aggregate fraction after conversion

Land use change Macroaggregate Microaggregate Silt+clay

0–10 cm 10–20 cm 0–10 cm 10–20 cm 0–10 cm 10–20 cm

Grass to crop F1 (Δ Mg C ha−1) −4.23 −2.26 −5.99 −5.27 −0.47 −0.39F2 (Δ Mg C ha−1) −1.82 −2.31 +1.51 +1.41 +0.19 +0.15

Grass to shrub F1 (Δ Mg C ha−1) −3.18 −2.01 −5.75 −5.10 −0.47 −0.45F2 (Δ Mg C ha−1) −0.87 −0.51 +0.24 +0.08 −0.02 +0.11

306 Plant Soil (2012) 355:299–309

Our calculations showed that changes in OC stockwithin the macroaggregate fraction were attributableboth to changes in the soil mass within the fractionand to changes in the aggregate-associated OC concen-tration of the fraction. Specifically, the decrease in mac-roaggregate mass after conversion to cropland reducedmacroaggregate-associated OC stocks by 30% in the 0–10 cm depth and by 51% in the 10–20 cm depth. Thedecrease inOC concentrations reducedmacroaggregate-associated OC stock by 70% in the 0–10 cm depth andby 49% in the 10–20 cm depth. For shrubland soil, thedecrease in macroaggregate mass after conversion toshrubland reduced macroaggregate-associated OCstocks by 21% in the 0–10 cm depth and 20% in the10–20 cm depth. The decrease in OC concentrations dueto shrub establishment reduced macroaggregate-associated OC stock by 79% in the 0–10 cm depth andby 80% in the 10–20 cm depth. We therefore concludethat the decrease in macroaggregate-associated OCstocks after conversion to cropland was due to declinesin both macroaggregate mass and macroaggregate-associated OC concentrations, whereas the decrease inmacroaggregate-associated OC stocks after conversionto shrublands was primarily due to declines inmacroaggregate-associated OC concentrations.

The proportion of soil in the microaggregate andsilt+clay fractions increased after conversion to crop-land or shrubland; however, the OC concentrations inthese two fractions decreased. Overall, there was a netdecline in both microaggregate- and silt+clay-associ-ated OC stocks. (Table 3). We conclude that the de-crease in microaggregate- and silt+clay-associated OCstocks after conversion to cropland or shrubland wereprimarily due to decreases in microaggregate- and silt+clay-associated OC concentrations.

The conversion of grassland to cropland or shrub-land caused changes in both aggregate size distribu-tion and aggregate-associated OC concentrations. Wefound that declines in whole soil OC stocks after theconversion to cropland were primarily due to losses inmacroaggregate OC stocks as previously reported byothers (Ashagrie et al. 2005, 2007; Steffens et al.2009). In contrast, the declines in whole soil OCstocks after conversion to shrubland were primarilyattributable to losses in microaggregate OC stocks.One explanation is that the cropland soil was cultivat-ed annually, whereas the shrubland soil was cultivatedonly once, when the shrubs were planted in 1982. Ithas been reported that repeated tillage significantly

decreased the proportion of macroaggregates, decreasedthe accumulation of plant-derived particulate OC withinthe aggregate structure, and increased the mineralizationof aggregate-associated OC (Cambardella and Elliott1993; Shi et al. 2010). This resulted in a decline inmacroaggregate-associated OC.

The potential of grasslands to sequester C has beenwell documented (Scurlock and Hall 1998; Leahy etal. 2004; Don et al. 2009; Ciais et al. 2010; Wolkovichet al. 2010). The OC stocks in the 0–20 cm depth ofthe grassland soil at our semiarid study site were28.2 Mg ha−1 in 1982, and 28.1 Mg ha−1 in 2009.We assume that soil OC would remain at a stable level.However, the conversion of grassland to cropland orshrubland could result in losses of 0.60 to 0.63 Mg Cha−1 yr−1, thus becoming a source of atmospheric CO2

in semiarid regions. Additionally, the disruption of soilmacroaggregates due to land use changes might leadto degradation of soil quality (Six et al. 2000; Snyderand Vázquez 2004) and make soils more prone toerosion (Barthès and Roose 2002; Valmis et al.2005). Decreases in soil OC and macroaggregationwould both lead to reduced ecosystem function insemiarid grassland ecosystems and this is an undesir-able result. Any land use changes in semiarid grass-land could reduce soil OC and degrade soil quality;therefore these changes should only be undertakenwith extreme caution.

Acknowledgment We thank Pengyuan Li, Zizhuang Liu, JiChen, Yanjun Hu, Le Wang and Shuai Yuan for their help infield and laboratory experiments and the reviewers for theircomments in improving the quality of this paper. This researchwas supported by National Natural Science Foundation of China(40901145, 40801111), Special Foundation for State MajorBasic Research Program of China (SQ2012FY4910023), andthe Program from Northwest A & F University (QN2011147).

References

Archer S (1994) Woody plant encroachment into southwesterngrasslands and savannas: rates, patterns and proximatecauses. In: Vavra M, Laycock WA, Pieper RD (eds) Eco-logical implications of livestock herbivory in the West.Society for Range Management, Denver, pp 13–68

Ashagrie Y, Zech W, Guggenberger G (2005) Transformation ofa Podocarpus falcatus dominated natural forest into amonoculture Eucalyptus globulus plantation at Munesa,Ethiopia: soil organic C, N and S dynamics in primaryparticle and aggregate-size fractions. Agric Ecosyst Envi-ron 106:89–98

Plant Soil (2012) 355:299–309 307

Ashagrie Y, Zech W, Guggenberger G, Mamo T (2007) Soilaggregation, and total and particulate organic matter fol-lowing conversion of native forests to continuous cultiva-tion in Ethiopia. Soil Till Res 94:101–108

Barthès B, Roose E (2002) Aggregate stability as an indicator ofsoil susceptibility to runoff and erosion; validation at sev-eral levels. Catena 47:133–149

Brookes PC, Landman A, Pruden G, Jenkinson DS (1985)Chloroform fumigation and the release of soil nitrogen, arapid direct extraction method to measure microbial bio-mass nitrogen in soil. Soil Biol Biochem 17:837–842

Cambardella CA, Elliott ET (1993) Carbon and nitrogen distri-butions in aggregates from cultivated and grassland soils.Soil Sci Soc Am J 57:1071–1076

Ciais P, Soussana JF, Vuichard N, Luyssaert S, Don A, JanssensIA, Piao SL, Dechow R, Lathlére J, Maignan F, WattenbachM, Smith P, Ammann C, Freibauer A, Schulze ED (2010)The greenhouse gas balance of European grasslands. Bio-geosciences Discussions 7:5997–6050

Don A, Rebmann C, Kolle O, Scherer-Lorenzen M, Schulze ED(2009) Impact of afforestation-associated managementchanges on the carbon balance of grassland. GlobalChange Biol 15:1990–2002

Elliott LF, Lynch JM, Papendick RI (1996) The microbial com-ponent of soil quality. In: Stotzky G, Bollag JM (eds) Soilbiochemistry. Marcel Dekker Inc, New York, pp 1–20

GLP (Global Land Project) (2005) Science Plan and Implemen-tation Strategy. IGBP Report No. 53/IHDP Report No. 19.IGBP Secretariat, Stockholm. pp 64.

Guo LB, Gifford RM (2002) Soil carbon stocks and land usechange: a meta analysis. Global Change Biol 8:345–360

Guo LB, Wang MB, Gifford RM (2007) The change of soilcarbon stocks and fine root dynamics after land use changefrom a native pasture to a pine plantation. Plant Soil299:251–262

Houghton RA, Hackler JL, Lawrence KT (1999) The U.S.carbon budget: contributions from land-use change. Sci-ence 285:574–578

Hoyos N, Comerford NB (2005) Land use and landscape effectson aggregate stability and total carbon of Andisols from theColombian Andes. Geoderma 129:268–278

Jackson RB, Banner JL, Jobbágy EG, William T, Pockman WT,Wall DH (2002) Ecosystem carbon loss with woody plantinvasion of grasslands. Nature 418:623–626

John B, Yamashita T, Ludwig B, Flessa H (2005) Storage oforganic carbon in aggregate and density fractions of siltysoils under different types of land use. Geoderma 128:63–79

John R, Chen J, Lu N, Wilske B (2009) Land cover/land usechange in semi-arid Inner Mongolia: 1992–2004. EnvironRes Lett 4:045010

Kim ES, Park DK, Zhao XY, Hong SK, Koh KS, Suh MH, KimYS (2006) Sustainable management of grassland ecosys-tems for controlling Asian dusts and desertification inAsian continent and a suggestion of Eco-Village study inChina. Ecol Res 21:907–911

Leahy P, Kiely G, Scanlon TM (2004) Managed grasslands: agreenhouse gas sink or source? Geophys Res Lett 31:L20507

Li GL, Pang XM (2010) Effect of land-use conversion on C andN distribution in aggregate fractions of soils in the southernLoess Plateau, China. Land Use Policy 27:706–712

Li J, Okin GS, Alvarez L, Epstein H (2007) Quantitative effectsof vegetation cover on wind erosion and soil nutrient lossin a desert grassland of southern New Mexico, USA.Biogeochemistry 85:317–332

Malhi SS, Brandt S, Gill KS (2003) Cultivation and grasslandtype effects on light fraction and total organic C and N in aDark Brown Chernozemic soil. Can J Soil Sci 83:145–153

MEA (Millennium Ecosystem Assessment) (2005) Ecosystemsand human well-being: desertification synthesis. WorldResources Institute, Washington

Noellemeyer E, Frank F, Alvarez C, Morazzo G, Quiroga A(2008) Carbon contents and aggregation related to soilphysical and biological properties under a land-use se-quence in the semiarid region of Central Argentina. SoilTill Res 99:179–190

Nyamadzawo G, Nyamangara J, Nyamugafata P, Muzulu A(2009) Soil microbial biomass and mineralization of aggre-gate protected carbon in fallow-maize systems under con-ventional and no-tillage in Central Zimbabwe. Soil Till Res102:151–157

Paul S, Flessa H, Veldkamp E, Lopez-Ulloa M (2008) Stabili-zation of recent soil carbon in the humid tropics followingland use changes: evidence from aggregate fractionationand stable isotope analyses. Biogeochemistry 87:247–263

Post WM, Kwon KC (2000) Soil carbon sequestration and land-use change: processes and potential. Global Change Biol6:317–328

Reynolds JF, Smith DMS, Lambin EF, Turner BL, MortimoreM, Batterbury SPJ, Downing TE, Dowlatabadi H, FernándezRJ, Herrick JE, Huber-Sannwald E, Jiang H, Leemans R,Lynam T, Maestre FT, Ayarza M, Walker B (2007) Globaldesertification: building a science for dryland development.Science 316:847–851

Sala OE, Lauenroth WK, Golluscio RA (1997) Plant functionaltypes in temperate semi-arid regions. In: Smith TM, ShugartHH, Woodward FI (eds) Plant functional types: their rele-vance to ecosystem properties and global change. CambridgeUniversity Press, New York, pp 217–233

Schlesinger WH, Reynolds JF, Cunningham GL, Huenneke LF,Jarrell WM, Virginia RA, Whitford WG, Walter G (1990)Biological feedbacks in global desertification. Science47:1043–1048

Scurlock JMO, Hall DO (1998) The global carbon sink: agrassland perspective. Global Change Biol 4:229–233

Shi XM, Li XG, Long RJ, Singh BP, Li ZT, Li FM (2010)Dynamics of soil organic carbon and nitrogen associatedwith physically separated fractions in a grassland-cultivationsequence in the Qinghai-Tibetan plateau. Biol Fertil Soils46:103–111

Six J, Elliott ET, Paustian K (2000) Soil macroaggregate turn-over and microaggregate formation: a mechanism for Csequestration under no-tillage agriculture. Soil Biol Bio-chem 32:2099–2103

Skjemstad JO, Krull ES, Swift RS, Szarvas S (2008) Mecha-nisms of protection of soil organic matter under pasturefollowing clearing of rainforest on an Oxisol. Geoderma143:231–242

Snyder VA, Vázquez MA (2004) Structure. In: Hillel D, HatfieldJH, Powlson DS, Rosenzweig C, Scow KM, Singer MJ,Sparks DL (eds) Encyclopedia of soils in the environment.Academic Press, pp. 54–58.

308 Plant Soil (2012) 355:299–309

Souchere V, King C, Dubreuil N, Lecomte-Morel V, LeBissonnais Y, Chalat M (2003) Grassland and crop trends:role of the European Union Common Agricultural Pol-icy and consequences for runoff and soil erosion. Environ SciPol 6:7–16

Steffens M, Kölbl A, Kögel-Knabner I (2009) Alteration of soilorganic matter pools and aggregation in semi-arid steppetopsoils as driven by organic matter input. Eur J Soil Sci60:198–212

StricklandMS, CallahamMA Jr, Davies CA, Lauber CL, RamirezK, Richter DD Jr, Fierer N, Bradford MA (2010) Rates of insitu carbon mineralization in relation to land-use, microbialcommunity and edaphic characteristics. Soil Biol Biochem42:260–269

Valmis S, Dimoyiannis D, Danalatos NG (2005) Assessinginterrill erosion rate from soil aggregate instability index,rainfall intensity and slope angle on cultivated soils incentral Greece. Soil Till Res 80:139–147

Wang T (2000) Land use and sandy desertification in northChina. J Des Res 20:103–108

Wang Q, Zhang L, Li L, Bai Y, Cao J, Han X (2009) Changes incarbon and nitrogen of Chernozem soil along a cultivationchronosequence in a semi-arid grassland. Eur J Soil Sci60:916–923

Wei X, Shao M, Fu X, Horton R, Li Y, Zhang X (2009)Distribution of soil organic C, N and P in three adjacentland use patterns in the northern Loess Plateau, China.Biogeochemistry 96:149–162

Wessman C, Archer S, Johnson L, Asner G (2004) Woodlandexpansion in US grasslands: assessing land-cover changeand biogeochemical impacts. In: Gutman G,Janetos AC, Justice CO, Moran EF, Mustard JF, RindfussRR, Skole D, Turner BL II, Cochrane MA (eds) Landchange science: observing, monitoring and understandingtrajectories of change on the earth’s surface. Kluwer Aca-demic Publishers, Dordrecht, pp 185–208

Wolkovich EM, Lipson DA, Virginia RA, Cottingham KL,Bolger DT (2010) Grass invasion causes rapid increasesin ecosystem carbon and nitrogen storage in a semiaridshrubland. Global Change Biol 16:1351–1365

Wu J, Joergensen RG, Pommerening B, Chaussod R, BrookesPC (1990) Measurement of soil microbial biomass C byfumigation-extraction-an automated procedure. Soil BiolBiochem 22:1167–1169

Yunwushan Natural Grassland Protection Zone Bureau (YNGPZB)(2001) A collected works on scientific investigation and man-agement in Ningxia Yunwushan Reservation. NingxiaPeople’s Press, Yinchuan

Zhao WZ, Xiao HL, Liu ZM, Li J (2005) Soil degradation andrestoration as affected by land use change in the semiaridBashang area, northern China. Catena 59:173–186

Zuo XA, Zhao HL, Zhao XY, Guo YR, Yun JY, Wang SK,Miyasaka T (2009) Vegetation pattern variation, soil deg-radation and their relationship along a grassland desertifi-cation gradient in Horqin Sandy Land, northern China.Environ Geol 58:1227–1237

Plant Soil (2012) 355:299–309 309