Global and Planetary Change 108 (2013) 29–41

Contents lists available at SciVerse ScienceDirect

Global and Planetary Change

j ourna l homepage: www.e lsev ie r .com/ locate /g lop lacha

Assessing the impact of restoration-induced land conversion andmanagement alternatives on net primary productivity inInner Mongolian grassland, China

Shaojie Mu a, Shuangxi Zhou b, Yizhao Chen a, Jianlong Li a,⁎, Weimin Ju c, I.O.A. Odeh d

a School of Life Science, Nanjing University, Nanjing 210093, PR Chinab Department of Biological Sciences, Macquarie University, Sydney NSW 2109, Australiac International Institute of Earth System Science, Nanjing University, Nanjing 210093, PR Chinad Faculty of Agriculture, Food & Natural Resources, The University of Sydney, Sydney NSW 2006, Australia

⁎ Corresponding author at: School of Life Science, Nan22, Nanjing 210093, PR China. Tel.: +86 25 86214644;

E-mail address: jianlongli@gmail.com (J. Li).

0921-8181/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.gloplacha.2013.06.007

a b s t r a c t

a r t i c l e i n f o

Article history:Received 31 December 2012Accepted 13 June 2013Available online 20 June 2013

Keywords:Inner Mongolia grasslandvegetation restoration programsland use and cover changenet primary productivityhuman activities

To address severe grassland degradation problems, China has been implementing a number of national res-toration programs, whose significant environmental effect has attracted the attention of many researchers. Inthis paper, land use and cover change (LUCC) in the Inner Mongolia grassland and the consequent change innet primary productivity (NPP) were studied by combining the land use data of the study area for 2001 and2009 derived from the MODIS global land cover product and the CASA (Carnegie–Ames–Stanford Approach)model driven with MODIS-NDVI data. The results indicate that the area of Inner Mongolia grassland had anet increase of 77,993 km2 during the study period, which was mainly attributed to the conversion fromdesert and cropland. The total NPP of Inner Mongolia grassland increased by 29,432.71 Gg C yr−1 during2001–2009, of which the human activities and climate change were responsible for 80.23% and 19.77%, re-spectively. Land conversion and improved management increased grassland NPP directly, and the ecologicalrestoration conducted by large-scale conservation programs could be the intrinsic driving force for thischange.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Land use activities have transformed one third to one half of ourplanet's land surface in the form of land use and cover change(LUCC), which made profound impacts on ecosystem structure, func-tion and diversity (Foley et al., 2005; Pielke, 2005; Yan et al., 2009). Asgrassland vegetation type is one of the world's most widespread landcover types, it has been deeply influenced by human activities forfood and forage production (Conant et al., 2001). China has 3.93 mil-lion km2 of grassland, accounting for 40% of China's total land areaand 13% of the world's total grassland (Ni, 2002). Concurrent withpopulation growth and socio-economic development, substantialLUCC has occurred in China's grassland over the last half century.However, unsound human activities have led to large-scale land deg-radation across the vast Inner Mongolia grassland, which constitutesthe main grassland region of China and a significant part of theEurope-Asia Steppe. Specifically, overgrazing and land conversionare suggested to be the primary impetus for grassland degradationin this region (Jiang et al., 2006). Historically, the generation of land

jing University, Hankou Roadfax: +86 25 83302728.

rights reserved.

degradation was as a result of intensive migration of Han peopleinto Inner Mongolia since the early 19th century. Since then thehuman population of the grassland area has increased rapidly, caus-ing drastic conversion of grassland to cropland and thus increasinggrazing pressure. This overexploitation of the grassland led to soilstructural decline and depletion of mineral content, therefore, trig-gering growing erosion and land degradation. The consequence ofthis is a significant decrease in land suitable for grazing, for example,between 1949 and 2000, the number of grazing animals has increased18-fold, while the total usable grassland decreased from 88 Mha to63 Mha during the same period (Chuluun and Ojima, 2002).

Land degradation in this region has led to the deterioration ofbiodiversity and ecosystem function and services as well as seriousenvironmental problems such as desertification and carbon sink loss(Liu et al., 2006). Taking Hunshandak sandy land for example, theavailable grasslands declined by some 40% between the 1950s and1990s, while the proportion of shifting sand dunes rose from 2.3% to50% during the same period. This decline in grassland and proportion-ate increase in desert land is generally believed to be a major reasonfor the increased frequency of severe sandstorms in northern China inrecent decades (Liu et al., 2006). Similarly the Horqin sandy land ineastern Inner Mongolia, the total loss of soil organic carbon resultingfrom grassland degradation was 107.53 Mt on land area of 26,393 ha

30 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

during the last century (Su et al., 2010). To mitigate the impacts ofland degradation, China has launched a number of national conserva-tion policies on payments during the late 1990s and early 2000s. Twoof them, the Grain to Green Program (GTGP) and the Grazing With-drawal Program (GWP) have been introduced in different ways inInner Mongolia with particular emphasis on grassland (Dai, 2010;Yin et al., 2010; Qiu et al., 2011). The GTGP, which is usually explainedas “replacing cropping and livestock grazing in fragile areas with treesand grass”, is the most renowned activity because of its ambitiousgoals, massive scales, huge payment and potentially enormous im-pacts (Wang et al., 2007). A GTGP pilot study started in three prov-inces in 1999 which was expanded to 25 provinces in 2002 (Liu etal., 2008). To complement the effort of GTGP, the national governmentinitiated the GWP, another large vegetation restoration program, in2003. Compared with the GTGP, the GWP focuses on alleviating graz-ing pressure in the degraded natural grassland in western China. Theprogramwas aimed to conserve grassland through banning of grazing,rotational grazing or converting grazing land to cultivated pasture(Tong et al., 2004). These two vegetation restoration programsimplemented in Inner Mongolia have induced changes in area cover-age and management practices of grassland, and consequently, thegrassland productivity.

Numerous studies have been published on the ecological effects ofvegetation restoration programs in China's grassland. Huang and Liu(2002) reported an increase in the grassland area ratio and a decreasein the farmland ratio in a catchment following the implementation ofGTGP. This could also effectively reduce surface runoff. A recent studycarried out in Hunshandak sandy land shows that without grazing,the previously degraded grassland was restored to the 1960s levelafter only three years of fencing and plant height (Jiang et al., 2006).As a result, plant cover and above-ground biomass all increased oncegrazingwas stopped under the GWP. In this paper, we focus on regionalecosystemnet primary productivity (NPP), defined as the net amount ofsolar energy converted to chemical energy through photosynthesis.

NPP represents the net carbon input from the atmosphere to theterrestrial vegetation (Imhoff et al., 2004). As an important parameterof ecosystem function and the carbon cycle, NPP can be used to quan-tify the impact of LUCC across a broad spectrum of issues in earth sys-tem science and global change research (Xu et al., 2007). This paperaims to assess the impact of LUCC on NPP change as a consequencefollowing the implementation of government policies on the restora-tion of degraded grassland. Reciprocally, through distinguishing NPPchange due to human activities (land conversion andmanagement al-ternatives) and climate change, we also determined the two drivingfactors to LUCC using NPP as an index for measurement. This is essen-tial to maintain optimal ecosystem functioning and to predict futureglobal carbon cycle trends.

We used satellite remote sensing to monitor LUCC at large scalesand quantifying vegetation productivity, as it is able to detect landcover type and actual vegetation dynamics top-down over largeareas directly. By integrating land use data derived from the MODISglobal land cover product (MCD12Q1) and NPP data, the impacts ofhuman activities on ecosystem productivity from those of climatechange for the period of 2001–2009 were disaggregated. Further-more, the spatial–temporal distribution of grassland productivitycaused by various types of grassland transformation was differentiat-ed in detail, and its implications for regional carbon cycle and sandstorms control were examined.

2. Data and methods

2.1. Study site

Inner Mongolia is located between 37°24′–53°23′N and 97°12′–126°04′E, and is the 3rd largest in China (Fig. 1). Most of InnerMongolia is 1000–2000 mabove sea levelwithwidely varied topography,

comprising mainly of plateaus which are not generally precipitous,extending 2400 km from northeast to southwest. The province is char-acterized by an arid to semi-arid continental climatewith strong climat-ic gradients and varied land use practices. The majority of cropland islocated in the transitional zone where the grassland biome meets theforest biome in northern Inner Mongolia, while a fraction of croplandscatter in the transitional agro-pasture area in the central region ofgrassland biome. There is a large area of desert in Inner Mongolia, in-cluding the Horqin sandy land in Chifeng city, the Hunshandak sandyland in contiguous area of Xilin Gol and Ulanqab league, the Tenggerand Badain Jarian desert in Alax league and the Hobq desert and MuUs sandy land centering on the Ordos Plateau.

In China, more than 20% of grassland distributed in Inner Mongolia,which is the representative for large areas of the Europe–Asia Steppebelt that stretches from east China to Hungary. Grassland is the domi-nant vegetation type in InnerMongolia. A strong east-to-west precipita-tion gradient results in a decrease in annual precipitation from morethan 500 mm in eastern Inner Mongolia to less than 100 mm inwestern part. With this large precipitation range, three major zonalgrassland types, meadow steppe, typical steppe and desert steppeare distributed along the northeast to southwest in Inner Mongolia.Typical steppe, developed under semi-arid conditions with annualprecipitation from 200 to 400 mm and annual mean temperaturefrom 0 to 8 °C in central Inner Mongolia, is the most widely spreadtype (Piao et al., 2006). Meadow steppe, which is more productivethan typical steppe, is developed in areas with moist fertile soils richin organic matter in northeastern Inner Mongolia, with annual averageprecipitation ranging from 300 to 600 mm and annual mean tempera-ture from2 to 5 °C. The desert steppe found in areaswith annual precip-itation between 150 and 200 mm and annual mean temperaturebetween 5 and 10 °C, has the least biomass (John et al., 2008).

2.2. Data source

2.2.1. NDVI dataWe chose the MODIS data and geo-spatial meteorological data as

input parameters to CASAmodel for calculating NPP in Inner Mongolia.We downloaded theMODIS-derived 16-day composite atmosphericallycorrected vegetation indices (MOD13A1) at 500-m resolution from EOSdata gateway (http://edcimswww.cr.usgs.gov/pub/imswelcome/) for2001 and 2009. MODIS-NDVI data was extracted from MOD13A1using theMODIS re-projection tool (MRT). A 32-day composite productof the maximal values was produced for a time series based on theseNDVI data. Given that each period covers 32 days, one year thus in-cludes about 11 time series of composite product of maximal NDVI.These data were re-projected to the Albers equal area projection andWGS84 datum using MRT and nearest neighbor method resampling.We then calculated NPP by using the MODIS-NDVI data as input tothe CASA model to evaluate the effect of restoration oriented LUCC.

2.2.2. Land cover dataThe MODIS Collection 5 Land Cover Type product (MCD12Q1)

based on IGBP global vegetation classification scheme was used totrack land cover changes and to guide the calculation of NPP inInner Mongolia. The IGBP classification has 17 LCLU classes, including11 natural vegetation classes, 3 developed and mosaiced land classes,and three non-vegetated land classes. In this paper, the land covermaps were reclassified into the following 7 dominant categoriesbased on the IGBP classification scheme: (1) water bodies, (2) forest,(3) grassland, (4) cropland, (5) cropland/natural vegetation, (6) urbanand built-up, and (7) desert. Evergreen needleleaf, deciduous needle-leaf, deciduous broadleaf, mixed forests and closed shrublands wererecoded to forest,whereas open shrublands,woody savannas, savannas,grasslands and permanent wetlands were recoded to grassland.

The MODIS global land cover product was derived from MODIS500-m resolution data using a state of the art supervised classification

Fig. 1. Current distribution of the GTGP and GWP in China. Names of cities and leagues in Inner Mongolia are shown on the location map of study region. The biome boundary wasderived from World Wildlife Funds (WWF) terrestrial eco-region boundaries (http://www.worldwildlife.org/).

31S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

system based on the decision tree classifier (Friedl et al., 2002). The ac-curacy of the IGBP layer of MCD12Q1 was estimated to be 74.8% global-ly, with a 95% confidence interval of 72.3–77.4% (Friedl et al., 2009).Global accuracy estimates for the dominant IGBP classes in InnerMongolia were 66% for grassland, 58% for cropland, 85% for openshrubland, 65% for mixed forest and 74.5% for barren. A separate analysisof the urban land cover class in the MCD12Q1 product, based on Landsatdata, indicates an overall accuracy of 93% (John et al., 2009).

2.2.3. Meteorological dataMeteorological data including daily average temperature, daily total

precipitation and daily total solar radiation for the same period wasobtained from China Meteorological Data Sharing Service System. Toprepare the input parameters to the CASA model, the Meteorologicaldata from 50 stations in Inner Mongolia were used to derive a 32-daycomposite meteorological data, interpolated using Inverse DistanceWeighted (IDW) method to produce raster images. These images areof the same temporal (32-day) and spatial (500 m × 500 m) resolu-tions as the remote-sensing images used for further analysis.

2.3. Calculation of NPP

In this study, we detected the relative role of climate change andhuman activities by comparing the difference between potential NPPand actual NPP. Potential NPP represented the amount of NPP thatwould be available in a grassland ecosystem with no human distur-bance. Actual NPP stood for the amount of NPP which actually remainsin the grassland ecosystem under current management practices.

2.3.1. Calculation of actual NPPNumerous models have been developed to estimate NPP at global

or regional scales, which mainly include statistical models, parametermodels and process-based models (Ruimy and Saugier, 1994). The

CASA (Carnegie–Ames–Stanford Approach) model employed in thisstudy is a process-based model developed on the Resource-balancetheory which postulates that plants regulate their physiological, bio-chemical and morphological characteristics when the resources avail-able to the plants change in response to changes in environmentalconditions (Field et al., 1995). The proportion of photosyntheticallyactive radiation (PAR) that is absorbed by vegetation is both the driv-ing potential for photosynthesis and the availability of whatever re-source limits growth. This implies that NPP can be modeled usingits relationship with the absorbed photosynthetically active radiation(APAR). The effects of environmental factors, such as temperatureand water stress on light-use efficiency, are also useful for estimatingNPP (Potter et al., 1993). As it is possible to make the estimation ofNPP based on satellite data and ground data in large-scale, the useof CASA model is widespread in recent years.

In the CASA model, NPP is the product of two major driving vari-ables which are the absorbed photosynthetically active radiation(APAR) and light-use efficiency (ε). The basic principle of the modelcan be described by the following formula (Yu et al., 2011):

NPP x; tð Þ ¼ APAR x; tð Þ � ε x; tð Þ ð1Þ

where x is spatial location (the pixel number), and t is time. APAR(x, t)(MJ m−2 mon−1) represents the photosynthetically active radiationabsorbed by pixel x in time t while ε(x, t) represents the actual lightuse efficiency (g C MJ−1) of pixel x in time t. APAR (x, t) and ε(x, t)in the equation were calculated from Eqs. (2) and (3) (Wang et al.,2009; Yu et al., 2011):

APAR x; tð Þ ¼ SOL x; tð Þ � FPAR x; tð Þ � 0:5 ð2Þ

where SOL (x, t) is total solar radiation (MJ m−2) of pixel x in time t.FPAR (x, t) is the fraction of PAR absorbed by vegetation canopy, and it

32 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

can be determined by NDVI. 0.5 stands for the fraction of total solarradiation that can be used by vegetation (0.38–0.71 μm); Light-useefficiency is derived from:

ε x; tð Þ ¼ Tε1 x; tð Þ � Tε2 x; tð Þ �Wε x; tð Þ � εmax ð3Þ

where Tε1 (x, t) and Tε2 (x, t) are temperature stress coefficients whichreflect the reduction of light-use efficiency caused by temperaturefactor. Wε (x, t) is the moisture stress coefficient which indicates thereduction of light-use efficiency caused by moisture factor. εmax isthe maximum light-use efficiency under ideal condition. A more de-tailed description of the model has been reported by Yu et al.(2011).

2.3.2. Calculation of potential NPPPotential NPP was calculated by using the model with the same

frame of the CASA except for the calculation of FPAR. We calculatedFPAR from vegetation and meteorological parameters for potentialNPP as follows:

FPAR ¼ 1−e−kLAI ð4Þ

where k = 0.5, and LAI (leaf area index) can be calculated as follows:

LAI ¼ LAImin þ fsw� fst � LAImax−LAIminð Þ ð5Þ

where LAImin and LAImax are maximum and minimum values of LAIand were determined by vegetation types. fsw and fst are coefficientsrepresenting the restriction of soil water and temperature to plantgrowth, which are expressed as

fsw ¼ min 1; max 0;Wsoil=Wmax

Wcrit

� �� �ð6Þ

fst ¼ min 1; max 0;1−0:0016� 298−Tð Þ2� �� �

ð7Þ

where Wsoil and Wmax are soil water content and the maximum soilwater content (m3 m−3), calculated by Soil Water sub model. Wcrit

is a constant (0.25) stands for a fractional water content reflectingthe permanent wilting point. T is the temperature of subsoil (20 cmdepth).

Wsoil was calculated as a state variable with the potential to accu-mulate moisture over several months (Saxton et al., 1986):

Wsoil x; tð Þ ¼ Wsoil x; t−1ð Þ− PET x; tð Þ−PPT x; tð Þ½ �⋅RDR⋅ For PPT x; tð ÞbPET x; tð Þ ð8aÞ

Wsoil x; tð Þ ¼ Wsoil x; t−1ð Þ þ PPT x; tð Þ−PET x; tð Þ½ �⋅ For PPT x; tð Þ≥PET x; tð Þ ð8bÞ

where PPT is mean precipitation at month t, PET is potential evapo-transpiration at month t, and RDR is a relative drying rate scalar forpotential water extraction as a function of soil water. For monthswhen temperature is less than or equal to 0 °C, PET and PPT are setequal to zero and there is no net change in Wsoil. During thesemonths, precipitation accumulates as snow in a state variable pack,which is added to PPT in the first month that monthly average airtemperature (Tair) > 0 °C. PET is calculated with the method ofThornthwaite (1948). RDR was calculated by using a transformationof the relationship between soil water potential and volumetricwater content presented by Saxton et al. (1986):

RDR ¼ 1þ að Þ= 1þ aθb� �

ð9Þ

where a and b are texture-dependent empirical coefficient and θ isthe volumetric moisture content.

Wmax was the product of field capacity (FC) and soil rooting depth.FC for each vegetation types were given by Potter et al. (1993). Thesoil rooting depth for forests is set to 2.0 m. The other vegetationtypes were assigned a rooting depth of 1.0 m (Potter et al., 1993).

Temperatures of the soil were simulated using the implicit solu-tion of Fick's law of heat diffusion:

Cihi Tsi;tþ1

h i86400Δt

¼ 21−Bð Þ Tsi−1;t−Tsi;t

di−1=λi−1 þ di=λi−

Tsi;t−Tsiþ1;t

di=λi þ diþ1=λiþ1

� �

þBTsi−1;tþ1−Tsi;tþ1

di−1=λi−1 þ di=λi−

Tsi;tþ1−Tsiþ1;tþ1

di=λi þ diþ1=λiþ1

� �8>><>>:

9>>=>>;ð10Þ

where C is the apparent volumetric heat capacity (J m−3 K), d is thedepth of layer (m), Ts represents soil temperature (°C),Δt is the numberof days permonth and λ is the thermal conductivity (W m−1 K−1); andsubscripts i and t denote the soil layer and the time step, respectively. Bis theweight given to implicit formulation (B = 0 for the purely explicitsolution, B = 0.5 for the Crank–Nicolson solution, and B = 1.0 for thepurely implicit solution). A more detailed description of soil tempera-tures simulation can be find in our previously published work (Ju etal., 2010).

The related vegetationparameters, such as LAImax and LAImin,werede-termined based on values defined in the Land Ecosystem-AtmosphereFeedback model (LEAF) of the Regional Atmospheric Modeling System(RAMS 6.0) and the Biosphere-Atmosphere Transfer Scheme (BATS)(Dickinson et al., 1993; Walko and Tremback, 2005).

2.3.3. Validation of NPP modelValidation was conducted by comparing the independent observa-

tion data for grassland vegetation under study in July and August2009 with the estimated data. The observation data contains informa-tion on the geographic coordinates, elevation, soil type, vegetationtype and net primary productivity of seven ecosystem sites acrossInner Mongolia (Fig. 1, Table 1). Fig. 2A presents the results of the cor-relation analysis between the observed NPP data and actual NPP. Thecorrelation was significant (R2 = 0.61, P b 0.001, n = 63), which in-dicates that the model's estimation accuracy is satisfactory and thatthe CASA model can be used to support research on changes in actualNPP in Inner Mongolia grassland.

Three of the seven sites were fenced from grazing 3 years beforethe field sampling. During the enclosure period, all of the sites wereconsidered to be in excellent condition and the representative ofundisturbed, natural communities of grassland ecosystems. Fig. 2Bpresents the results of the correlation analysis between the observedNPP data and potential NPP, and indicates a significant correlation be-tween the observed NPP and the modeled potential NPP (R2 = 0.53,P b 0.001, n = 30). Furthermore, the modeled actual NPP and poten-tial NPP (Fig. 2C) also presents a significant correlation (R2 = 0.66,P b 0.001, n = 30). Xu et al. (2009, 2010) has used these two modelsto estimate the actual NPP and potential NPP in Ordos region of InnerMongolia and the validation of the CASA model suggested that themodel's estimation accuracy (R2 = 0. 663, P = 0. 01) is satisfactoryto estimate the actual NPP in Inner Mongolia grassland.

2.4. The relative contribution of climate change and human activitiesto LUCC

Climate change and human activities were the primary drivingforce both to LUCC and NPP change. Many researchers used NPP asan index to determine the relative contributions of climate changeand human activities to LUCC (Zika and Erb, 2007; Xu et al., 2009,2011). In this study, the relative contribution of the two factorswere determined through comparing the change of potential NPPand actual NPP in the Inner Mongolia grassland during 2001–2009.The potential NPP was used to estimate the climate induced NPPchange, while the difference between potential NPP and actual NPPcan reflect the human activities induced NPP change. Furthermore,the human activities here mainly include land conversion and themanagement alternatives, and the relative role of the two factors

Table 1Abiotic and biotic characteristics of the seven observation sites in the Inner Mongolia grassland.

Site Latitude (E) Longitude (N) Elevation(m)

Vegetable type Soil type Community type Management

A 119.61° 50.17° 585 Meadow steppe Typical chernozem soil S. baicalensis–C. pediformis EnclosureB 119.23° 48.95° 702 Meadow steppe Dark chestnut soil S. grandiS–L. chinensis GrazingC 116.70° 44.58° 967 Typical steppe Light chernozem soil S. grandis GrazingD 114.50° 44.18° 1050 Typical steppe Light chernozem soil L. chinensis EnclosureE 115.24° 43.68° 993 Typical steppe Light chernozem soil S. grandis GrazingF 110.32° 41.50° 1514 Desert steppe Typical brown soil S. klemenzii GrazingG 108.40° 39.40° 1411 Desert steppe Eolian sand soil A. ordosica–C. komarovii Enclosure

a S. baicalensis, Stipa baicalensis; C. pediformis, Carex pediformis; L. chinensis, Leymus chinensis; S. grandis, Stipa grandis; S. klemenzii, Stipa klemenzii; A. ordosica, Artemisia ordosica;C. komarovii, Cynanchum komarovii.

33S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

were distinguished by comparing the difference between potentialNPP and actual NPP in unchanged grassland and new developedgrassland, separately. A more detail of the method was describedspecifically based on the estimated result in Section 3.3.

2.5. Calculation of fractional vegetation cover

Fractional vegetation cover (fv), which is defined as the projectedarea of vegetation per unit ground area, is an important ecologicalparameter. It can be used as an indicator of the quantitative charac-teristics of vegetation and an evaluation index of land degradation,salinization and desertification. According to Gutman and Ignatov(1998), fv is calculated from NDVI by using a linear mixture modelwith two endmembers representing fully vegetated land surfaceand bare ground. The equation is illustrated as follow:

f v ¼NDVI−NDVImin

NDVImax−NDVIminð11Þ

where NDVImin is the minimum NDVI corresponding to 0% vegetationcover or bare soil and NDVImax is the maximum one 100% vegetationcover. The annual mean fractional vegetation cover for each pixel canbe calculated from this method. This method has been proved to be avalid method to estimate large area fractional vegetation cover whichsatisfies the demand of large scale ecological and climatic models. Inthis study, we estimated the fractional vegetation cover of unchangedgrassland in 2001 and 2009 to qualify the influence of managementalternatives on growth status of plants.

3. Results

3.1. Land use/land cover change during the process ofgrassland restoration

According to the land use change analysis, the Inner Mongoliagrassland area had a net increase of 77,993 km2 between 2001 and2009, as the newly restored grassland area was 128,188 km2. Mean-while, 50,195 km2 of grassland were converted for other uses —

Fig. 2. Correlations between actual NPP and observed NPP (A), potential NPP and observ

mainly for cropland. As presented in Fig. 3A, there were mainlythree land transformation types, which were the conversion fromdesert to grassland, from cropland to grassland, and from grasslandto cropland. Fig. 3B presents the grassland transformation in InnerMongolia between 2001 and 2009. The increase of grassland areaderived mainly from the transformation from desert and cropland,accounting for 47.6% and 41.1% of the newly developed grassland,respectively. The conversion from desert to grassland was the mostsignificant landscape transformation during the 9 years interval,contributing 61,052 km2 to grassland expansion, and the reverse con-version was limited and almost non-existent. In the whole study area,69,857 km2 of cropland was converted to other classes between 2001and 2009, of which about 75.4%was converted into grassland. However,the opposite conversion led to the most loss of original grassland,which was 36,136 km2, accounting for 72% of the total loss. Mean-while, 8359 km2 of cropland/natural vegetation were converted tograssland while 3936 km2 of grassland were converted in reverse. Theland conversions from forest land to grassland and from grassland toforest land nearly reached near equilibrium during the 9 years period.Water body and urban land occupied a small percentage of the totalarea, and changed minimally.

LUCC of Inner Mongolia grassland between 2001 and 2009showed a significant spatial difference in Fig. 3A. The significant geo-graphical variations are mainly characterized by the transformationfrom desert to grassland in the west, from cropland to grassland inthe east and central south and grassland reclamation in the northeast.Spatially, transformation from desert to grassland occurred mainly inthe transitional zone where the grasslandmet the desert in the south-west of Inner Mongolia, including the west fringe of Hunshandaksandy land, the Hobq desert and Mu Us sandy land (Figs. 1 and 3A).The newly developed grassland from the abandoned cropland wasgenerally concentrated in farm land and pasture interleaving region,which had suffered serious degradation and were considered as theprimary sand storms source area. The land converted to grasslandalso included the cropland patches in Horqin sandy land and thescattered cropland patches in the east fringe of Hunshandak sandyland and that in the northeast of Xilin Gol League. As shown in Fig. 3A,the spatial distribution of grassland conversion to cropland was

ed NPP (B), potential NPP and actual NPP (C) for grassland in July and August 2009.

Fig. 3. Grassland transformation in Inner Mongolia between 2001 and 2009. A is spatialdistribution of grassland transformation in Inner Mongolia between 2001 and 2009.B presents the areas of main grassland transformation in Inner Mongolia. The corre-sponding conversion types' names are presented in Table 2.

34 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

fragmented. The grassland reclamation occurred mainly in the north-east of Inner Mongolia, where the cropland had sprawled andencroached on the original grassland on both sides of the GreaterHinggan Mountain in Hulunbuir league. Meanwhile, a small fraction ofgrassland were also converted to cropland on Hetao plain and sur-rounding developed cities including Baotou, Hohhot and Chifeng.

3.2. The effect of grassland restoration on net primary productivity

The NPP variation of Inner Mongolia grassland during 2001–2009is shown in Table 3. Total NPP of Inner Mongolia grassland was137,289.36 Gg C yr−1 (1Gg = 109 g) in 2001and 166,722.05 Gg C yr−1

in 2009, showing a significant increase of 29,432.69 Gg C yr−1, or21.44% of that in 2001 during the 9 years. The mean NPP of unchangedgrassland increased by 30.80 g C m−2 yr−1 from 2001 to 2009,resulting in the increase total NPP by 14,954.04 Gg C yr−1. The trans-formation of Inner Mongolia grassland led to the increase of total NPP

Table 2Codes for the conversions between grassland and other land use and cover types from 200

Water to grassland (13) Unchanged grassland (Forest land to grassland (23) Grassland to cropland (Grassland to water (31) Grassland to crop/natuGrassland to forest land (32) Grassland to desert (37

by 14,478.67 Gg C yr−1. During the 9 years period, the newly devel-oped grassland had an increase of 34,499.47 Gg C yr−1 in total NPP,to which the most significant contributions were from the conversionof cropland to grassland (increased 21,094.06 Gg C yr−1) and that ofdesert to grassland (increased 8142.89 Gg C yr−1). Meanwhile, theconversion of Inner Mongolia grassland into other land classes(converted-out grassland) had caused a loss of 20,020.80 Gg C yr−1

in total NPP, in which the grassland reclamation caused a loss of14,529.92 Gg C yr−1 and played a dominant role. Therefore, the newlydeveloped grassland in Inner Mongolia could sufficiently compensatefor the loss of productivity caused by the transformation from grasslandto other land cover types.

3.3. Contributions of human activities on grassland NPP contrasting toclimate change

NPP change can be driven particularly by two factors — climatechange and human disturbance. In order to differentiate the human-induced NPP change from the climate-induced NPP change, thepotential NPP was used to estimate the climate-induced NPP changein the newly developed and unchanged grassland during 2001–2009.Fig. 4 showed the relative contributions of human activities and climatechange to total NPP change of Inner Mongolia grassland. The total NPPin 2001 (NPP2001) consisted of total NPP of both unchanged grassland(NPPuch01) and converted-out grassland (NPPloss, which was grasslandin 2001 but converted to other classes during 2001–2009):

NPP2001 ¼ NPPuch01 þ NPPloss: ð12Þ

The total NPP in 2009 (NPP2009) consisted of total NPP of unchangedgrassland (NPPuch09) and newly developed grassland (NPP new):

NPP2009 ¼ NPPuch09 þ NPPnew: ð13Þ

The net increment of total NPP during 2001–2009 (NPPinc) was theaggregate effect of changes in NPP resulting from human activities(NPPhum) and climate change. NPPhum measures the combined effectsof land conversion (NPPlc) and management alternatives (NPPma) ontotal NPP:

NPPhum ¼ NPPlc þ NPPma: ð14Þ

NPPlc was calculated by subtracting NPPnew with both the NPP lossfrom converted-out grassland (NPPloss) and the climate-induced NPPchange in the newly developed grassland (NPPcnew) during 2001–2009:

NPPlc ¼ NPPnew−NPPloss−NPPcnew: ð15Þ

On the other hand,NPPmawas calculated by subtractingNPPuch09 withthe amount of total NPP of unchanged grassland in 2001 (NPPuch01) andthe climate-induced NPP change in the unchanged grassland (NPPcuch):

NPPma ¼ NPPuch09−NPPuch01−NPPcuch: ð16Þ

Human activities caused an increase of 23,612.19 Gg C yr−1

(80.23% of the total increment), within which 12,381.98 Gg C yr−1

was attributed to land conversion, and 11,230.21 Gg C yr−1 was at-tributed to improved grassland management in the unchanged grass-land (Table 4). The simulation results also showed that the climate

1 to 2009.

33) Cropland to grassland (43)34) Crop/natural veg to grassland (53)ral veg (35) Urban and built-up to grassland (63)) Desert to grassland (73)

Table 3Estimated value of mean and total NPP of different forms of grassland conversion in Inner Mongolia between 2001 and 2009.

2001 2009

Codes Mean NPP(g C m−2 yr−1)

Area(km2)

Total NPP(Gg C yr−1)

Codes Mean NPP(g C m−2 yr−1)

Area(km2)

Total NPP(Gg C yr−1)

Unchanged grassland Unchanged grassland33 241.51 485,568.00 117,268.56 33 272.31 485,568.00 132,222.59

Converted-out grassland during 2001–2009 Newly developed grassland during 2001–200931 243.94 80.9263 19.74 13 104.16 513.68 53.5132 504.25 6483.77 3269.45 23 431.02 5586.49 2407.8934 402.09 36,135.9 14,529.92 43 400.45 52,675.9 21,094.0635 486.13 3936.41 1913.59 53 335.08 8359.24 2800.9836 0 0 0 63 626.00 0.21 0.1337 80.98 3557.75 288.10 73 133.38 61,052.6 8142.89Sum 535,762.76 137,289.36 613,756.12 166,722.05

35S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

change caused an increase of 5820.5 Gg C yr−1 in the grassland areaduring the 9 years. Land conversion induced by human activitiesturned out to be the most influencing driving force for grasslandNPP increase. In the newly developed grassland, the NPP increase in-duced by land conversion (12,381.98 Gg C yr−1) was almost six timesof that induced by climate change (2096.68 Gg C yr−1)when subtractingthe NPP loss resulting from converted-out grassland. After filtering theclimate induced NPP change, the total NPP gain resulting from develop-ment of new grassland was as high as 32,402.78 Gg C yr−1, to whichthe transformation from cropland and desert contributed mostly, being18,926.34 Gg C yr−1 and 8604.39 Gg C yr−1, respectively. This sig-nificant gain can sufficiently counteract the NPP loss fromconverted-out grassland (NPPloss, 20,020.80 Gg C yr−1). For the netincrease of total NPP in unchanged grassland, 24.90% derived from cli-mate change, while the rest was possibly due to improved grasslandmanagement such as reducing grazing intensity and adjusting thegrassland use pattern.

Fig. 5 showed the spatial distribution of the total NPP change andits three fractions (NPP change induced by land conversion, manage-ment alternatives and climate change) across Inner Mongolia be-tween 2001 and 2009. Although climate change was responsible forthe NPP increase in northeast Inner Mongolia and the decrease insouthwest Inner Mongolia, its effects on the total NPP change werespatially scattered due to human activities. The negative effects of cli-mate change on NPP in most of the southwestern region may havebeen alleviated or offset by the newly developed grassland and im-proved management in the unchanged part. Fig. 5 also showed thatthe NPP increase induced by land-conversion had occurred largely

Fig. 4. Total NPP of Inner Mongolia grassland in 2001 and 2009 and contributions ofhuman activities and climate fluctuation on NPP change.

in the transitional zone between the grassland and desert in thesouthwestern region and in some farms juxtaposed with pasturelandin the southeastern region. However, in the northeastern part of InnerMongolia, land conversion from grassland to other classes had led to asignificant decrease in the total NPP.

4. Discussion

4.1. Possible effects of climate change on grassland expansion

Climate change could be an important driving factor on grasslandexpansion, the variation of annual mean temperature and precipita-tion was shown in Fig. 6. We can find that the annual average temper-ature in 2009 (4.82 °C) was lower than that in 2001 (4.89 °C), and theannual precipitation in 2009 (239.33 mm) was higher than that in2001 (215.99 mm). It indicated there was lower surface evaporationpotential and higher soil moisture in 2009 compared with that of2001in Inner Mongolia. When we focused on the year 2001 and2009, climate factors may be partially responsible for the grasslandexpansion as shown in Fig. 4.

As for the region of newly developed grassland (the land cover typewas not grassland in 2001 and was grassland in 2009, Fig. 7), the totalNPP in 2001 was 24,765.60 Gg C yr−1, and was 34,499.46 Gg C yr−1

in 2009 (NPPnew2009). The total NPP of this region increased9733.89 Gg C yr−1 over the 9-year period (NPPnewinc), of which cli-mate change induced increase was 2096.68 Gg C yr−1 (Table 4). Ittherefore implied that climate change contributed 21.54% to grasslandexpansion.

From 2001 to 2009, however, the annual average temperature ex-hibits an insignificant increasing trend and the annual precipitationexhibits an insignificant decreasing trend. That was in agreementwith the analysis results of Lu et al. (2009) who suggests that InnerMongolia, as a region, has changed to a warmer and drier environ-ment over the past 50 years. The climate change trend during the9-year period does not facilitate the grassland expansion in the re-gional scale. It then implied that human activities has played a moreimportant role in the expansion of Inner Mongolia grassland during2001–2009, which may be a long, slow process.

4.2. Role of vegetation restoration programs in the process ofgrassland rehabilitation

4.2.1. Effects of vegetation restoration programs on land conversionNPPwaswidely used as an index to determine the relative contribu-

tions of human activities and climate factors to LUCC (Zika and Erb,2007; Xu et al., 2009, 2011). This study's findings have shown that dur-ing the 2001 to 2009 period, land conversion led to an increase in totalNPP of 12,381.98 Gg C yr−1 (42.1% of the total increase), after filteringout the minor influence of climate change. It implied that human

Table 4Effects of human activities and climate fluctuation on total NPP for different forms of grassland transformation in Inner Mongolia between 2001 and 2009 (Unit: Gg C yr−1).

Driving forces Codes Total NPP (2009) Climate induced change NPP0a Total NPP (2001) NPP hum

Land conversion Newly developed (gain) Converted-out (loss)13 53.51 31.57 21.9423 2407.89 274.55 2133.3443 21,094.06 2167.72 18,926.3453 2800.98 84.34 2716.6463 0.13 −0.01 0.1473 8142.89 −461.50 8604.39Sum 34,499.46 2096.68 32,402.78 20,020.80 12,381.98

Management alternatives Unchanged grassland33 132,222.59 3723.82 117,268.56 11,230.21

a NPP0 presents the change of total NPP induced by human activities in the newly developed grassland in 2009, which was calculated by subtracting the climate induced changefrom the total NPP of newly developed grassland in 2009.

36 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

activity played a key role in grassland restoration. According to datafrom the Inner Mongolia Statistical Yearbook (1986–2010), large-scalegrass planting (including aircraft sowing and artificial planting) acceler-ated in 1999 (Fig. 8A), so that the annualmean area of planted grasslandincreased from 62.26 Mha from 1986 to 1998 to 132.49 Mha from 1999to 2010. Meanwhile, the surviving area of planted grassland alsoshowed an increase after 1999, reaching 438.41 Mha by the end of theyear 2010 (Fig. 8B). It was, therefore, implied that the implementationof vegetation restoration programs was a possible intrinsic cause forgrassland expansion.

Fig. 5. Effects of land conversion, management alternatives and climate on grassland NPP iB: land conversion induced change, C: management alternatives induced change, and D: cl

The GTGP aims to increase China's vegetation areas by 32 million haby 2010, of which 14.7 million ha of cropland on steep slopes wereplanned to be converted to grassland or forest land for ecological pur-poses in areas where food security is basically guaranteed (Chen et al.,2009). During the nine years of this study, returning cropland to grass-land has caused an increase of 52,676.00 km2 in grassland area, whilereclamation reduced grassland areas by 36,135.90 km2. These resultsare in accordance with the study by Dong et al. (2011) conducted in atypical agro-pastoral transitional zone in themiddle-eastern Inner Mon-golia, and showing that the speed of grassland reclamationhas decreased

n Inner Mongolia between 2001 and 2009 (Unit: g C m−2 yr−1), A: total NPP change,imate induced change.

Fig. 6. The inter-annual variations of annual average temperature and precipitation inInner Mongolia during 2001 and 2009.

37S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

while the opposite conversion from cropland to grassland has increasedsince the implementation of GTGP in 1999. The GTGP also led tolarge-scale cropland abandonment and revegetation in other regions ofChina. For example, in Zigui county of the Three Gorges Reservoir region,Hubei province, 3085 ha of cropland (8.1% of the total cropland in Ziguicounty) was converted to forest in 2000 (Liu et al., 2008). In Yunnanprovince, 335,556 ha of eligible cropland was planted with trees be-tween the years 2001 and 2007 (Chen et al., 2009). In comparison withthese twoworks, although the present study reaches similar conclusionsthat the GTGP plays a dominant role in land conversion, however, theseresults differ slightly in that the cropland in Inner Mongolia was mainlyconverted into grassland rather than forest land. On a national scale,trees (such as Populus tremula and Pinus armandi) were selected as re-vegetation species more often in the GTGP (Andersson et al., 2011).However, planted trees in arid and semi-arid zones are oftenunsustainable because their water demand exceeds supply, which maylead to new degradation (Cao et al., 2010). The greater part of InnerMongolia receives less than 400 mm in annual rainfall, thus, plantingdrought-tolerant grasses and shrubs would be a more effective methodfor sustainable revegetation (Wang et al., 2007).

Under the GTGP, 17.33 Mha of decertified barren land, particularlyin sandy land in the edge of oasis,was planned to be planted nationwidebetween 1999 and 2010 (Chen et al., 2009). According to this presentstudy, 61,253.8 km2 of desert was converted to other classes between2001 and 2009 in InnerMongolia, of which about 99.67%was convertedinto grassland. These findings are generally in accordance with those of

Fig. 7. Total NPP change of the new developed grassland where the land cover type wasnot grassland in 2001 and was grassland in 2009.

Kuchelmeister and Meyer (2007) and Normile (2007), who suggestedthat artificially expanding oases at the desert's edge under the GTGPcould successfully stop the process of grassland desertification and pro-vide a chance to regenerate vegetation. Bagan et al. (2010) also reportedthat in Horqin sandy land, the implementation of GTGP successfullysuppressed the sand area and increased the sparse-grass area between1999 and 2007. During the restoration process in decertified regions,vegetation patches formed by shrubs could create “islands of fertility”with a less harsh micro-environment and facilitate the establishmentand growth of a herbaceous community. As the age of shrub plantationsincreased, herbaceous vegetation patches tended to expand (Zhao et al.,2007). However, vegetation with over-high coverage could form a big“water pump” and aggravate the consumption of the soil moisture(Wang et al., 2010). Therefore, during the restoration process in thedecertified land, water balance at community level or even regionallevel should be considered to establish appropriate configuration ofvegetation, and vegetation restoration programs should be adjusted toadapt to local conditions so as to bring lasting benefits over a longperiod.

4.2.2. Effects of vegetation restoration programs onmanagement alternatives

Land use and management practices largely govern the sustain-ability of a given land (Foley et al., 2005), and grazing intensity andimproved management have profound influences on grassland eco-system. Recent surveys have shown that nearly 90% of the naturalgrassland in China is degraded to varying degrees, which is attribut-able mainly to overgrazing (Liu et al., 2004). With the launch of theGTGP and its successor, the GWP, various efforts aimed at alleviatinggrazing pressure were promoted to conserve and restore the degrad-ed grassland. Grassland with different degrees of degradation requiredifferent measures for restoration. For heavily degraded grassland,grazing forbidden by fencing so as to allow native vegetation tore-establish naturally may be necessary. In this process, the pioneerspecies, such as colonizing plant that can provide soil stability, shouldfirst occupy the degraded areas, and then gradually be replaced byherbs and shrubs that offer plant community structure (Liu et al.,2004). Undisturbed mature grassland ecosystems seem to culminatein high biodiversity, productivity and ecosystem stability concurrent-ly (Bai et al., 2004). For example, in a degraded Alxa desert steppe,with grazing exclusion for six years, the grass biomass increased by56%, and the ground cover increased 1.5 times as compared withthe grazed area. Meanwhile, the number of species recorded was 19for the enclosure site, but only 10 for the freely grazed site (Pei etal., 2008). As for slightly and moderately degraded steppes, grazingrotation and seasonal enclosures should be considered. For instance,grazing should be banned temporarily during April to June eachyear when the grass just turns green or to partition the grasslandinto different zones and each zone is grazed in turn (Gao and Liu,2010). Furthermore, parts of high productivity grassland have beenconverted into cultivated pasture using intensive agronomic proce-dures such as plowing, sowing, irrigating and fertilizing. By plantingaridity- and cold-resistant forage grasses such as Medicago sativaand Astragalus adsurgens for breeding animals, vegetation productiv-ity in these regions has clearly increased (Tong et al., 2004).

According to the data from the Inner Mongolia Statistical Yearbook(1986–2010), the area of fenced grassland in Inner Mongolia in-creased significantly since the implementation of vegetation restora-tion programs in the late 1990s and 2000s, and had increased nearly 3times between 2001 and 2010 (Fig. 8C). Grazing rotation and season-al enclosures were also widely implemented in Inner Mongolia. In thegrassland of Xilin Gol League, 88% of the total area were fenced off toallow the grass to germinate in the spring during 2002 and 2004 (Zhuet al., 2008). Meanwhile, a significant increase in area of animal shedsand corrals (Fig. 8D) implied that the use pattern of grassland wasshifting from freely grazing to indoor feeding. This present studyfinds that between 2001 and 2009, the total NPP of unchanged

Fig. 8. Annual area of planted grassland (A), surviving area of planted grassland (B), area of fenced grassland (C), area of animal sheds and corrals (D) and livestock number (E) inInner Mongolia from 1986 to 2010. Data sources: Inner Mongolia Statistical Yearbook (1986–2010).

38 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

grassland increased by 14,954.04 Gg C yr−1, accounting for 50.81% thetotal increment. There is even an increase of 11,230.21 Gg C yr−1 afterfiltering out the influence of climate factors, and the vegetation produc-tivity in these regions has clearly increased. We also analyzed theinter-annual change of mean NPP in the unchanged grassland from2001 to 2009 (Fig. 9), which exhibiting an increase trend with a rateof 40.6 g C m−2 yr−1 per decade. In addition, the mean annual frac-tional vegetation cover (fv) of unchanged grassland was 30.75% in2001 and 33.16% in 2009, indicating that vegetation productivity inthese regions had clearly increased. As shown in Fig. 10A, there weretwo peaks, both a major and a minor crest, of fv in 2001, and thatmost of the fv values were distributed between 15% and 40%. From thehistogram shown in Fig. 10B, one clear crest of fv is shown in 2009,and most of the fv values were between 20% and 40%, being relativelyconcentrated around 30%. One tendency was that moderate/low cover-age grassland was converted to high coverage grassland between 2001and 2009, with the improved grassland management practicesconducted by the GTGP and GWP as the possible dominant drivingforce.

However, the implementation of the GWP mainly focused on seri-ously degraded grassland and the sandstorm source areas, such as

Mu Us sandy land, Hobq desert, Hunshandak sandy land and Horqinsandy land (Jiang et al., 2006; Dai, 2010; Zhang et al., 2012). Indeed,the grassland in these ecologically fragile regions experienced obvi-ous an restoration process between 2001 and 2009. Over the sameperiod, the quantity of livestock in Inner Mongolia increased signifi-cantly from 73.35 million head in 2001 to 107.99 million head in2010, with the largest quantity of 110.51 million head in 2006(Fig. 8E). Production of livestock in Inner Mongolia traditionally de-pends on the natural grass in Inner Mongolia, with little feed importfrom outside. The increase in the number of livestock thereforemeans an increased feed or consumption of plant biomass from thegrassland in Inner Mongolia. Thus, in relation with the increasingarea of fenced grassland in the ecologically fragile regions and the in-creasing livestock number in the whole region, the grazing intensitywas drastically increased in the original non-degraded or lightly de-graded grassland which was without enclosure during the nine yearperiod. Throughout the region, grazing pressure was not alleviated,due to increased human activity, but shifted and became more con-centrated, triggering new degradation. As shown in Fig. 5C, the grass-land in central Xinlin Gol League and western Hulun Biur League wasdegraded in 2009 as compared with in 2001.

Fig. 10. Vegetation coverage histogram of unchanged grassland in 2001(A) and 2009 (B).

Fig. 9. The inter-annual change ofmean NPP in the unchanged grassland from 2001 to 2009.

39S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

4.3. Implication of grassland rehabilitation on carbon cycle and sandstorms control

Conversion of cropland and desert to grassland and improvedgrassland management could positively promote enhanced soil car-bon sequestration and reduce carbon losses in grassland whichwere probably due to higher NPP and lower organic matter decompo-sition. With vegetation regeneration on previously degraded land,biomass returning to the soil can be increased and the physical andchemical properties of the soil may be improved due to alleviativehuman intervention, leading to an increase in carbon sequestrationpotential and a decrease in soil respiration (Gao and Liu, 2010). Insemi-arid Horqin sandy land, Su and Zhao (2003) reported that theorganic carbon concentration found in the soil at a depth of 5 cm in-creased 13.3 times and 20.5 times respectively after the introductionof Caragana microphylla to shifting sand dunes for periods of 5 and28 years. According the results of Wang et al. (2011), the achieve-ment of the national objective to exclude grazing livestock from150 million ha of China's grasslands and to establish 30 million haof cultivated pasture by 2020 would sequester over 0.24 Pg C yr−1,which is equivalent to about 16% of fossil fuel CO2 emissions inChina in 2006. Soils with coarse texture and a low initial C contentmay have greater C sequestration potential compared with finely tex-tured soils (Su, 2007), indicating that vegetation regeneration in thedegraded land is an effective way to sequester C and reverse the spiraleffect of continuous degradation.

Besides this, with the formation of grass-root layers and the increasein soil carbon content, formation of soil aggregates can be enhanced sothat land surface resistance to wind erosion may be significantlyincreased (Xu, 2006). Consequently, the intensity of wind erosiondecreases markedly and the frequency of sand storms on-site and indownwind areaswill be reduced. Reciprocally, the improved ecosystemfacilitates colonization, reproduction and development of herbaceousspecies in the restoration process (Bronick and Lal, 2005; Su et al.,2010). A previous study carried out in Ordos Plateau has shown thatthe relative contribution of land conversion to the variation in the annu-al number of days when sand stormswere recorded has been estimatedas 59.7%, suggesting that reducing the area of cropland and restoring thenatural steppe vegetation in ecologically fragile areas, that is, in arid andsemi-arid climates,may effectively reduce the frequency of sand storms(Xu, 2006).

According to the reaction equations of photosynthesis and respira-tion, vegetation absorbs 1.62 g CO2 to produce 1 g carbon of drymatter.And 1.2 g O2 is released in the process. Between 2001 and 2009, thetotal NPP of InnerMongolia grassland increased by 29,432.71 Gg of car-bon. This is equivalent to increasing absorption of CO2 by 47,680.59 Ggand an increase in the release of O2 by 35,318.73 Gg. During the period

2001 to 2009, conversion from cropland and desert to grasslandwith anarea of 113,728.5 km2 in Inner Mongolia, which mainly occurred in thesandstorm source area (such as Hunshandak and Mu Us sandy land),may have exerted positive effects on sand storm control. Therefore,due to the extensive coverage area and increased productivity of thegrassland, restoration-oriented land conversion and improvedmanage-ment may provide important insights regarding sequestration of atmo-spheric carbon as well as in the elimination of sand storms, at presentand in future, both inside China and around the world.

5. Conclusion

Land use and land cover had changed greatly during the process ofvegetation restoration in Inner Mongolia grassland between 2001 and2009. The land area occupied by grassland had a net increase of77,993 km2 during the study period. Transformation of desert and crop-land to grassland played a dominant role in grassland expansion, respec-tively contributing 47.6% and 41.1% to the total area of newly developedgrassland, while the conversion of grassland to cropland was the biggestloss of original grassland. The estimated total NPP for Inner Mongoliagrassland was 137,289.36 Gg C yr−1 in 2001 and 166,722.05 Gg C yr−1

in 2009, respectively. Human activities had caused 80.23% of the totalNPP increase (29,432.71 Gg C yr−1) of Inner Mongolia grasslandduring 2001–2009. The newly developed grassland transformed fromcropland and desert had respectively increased the total NPP by18,926.34 Gg C yr−1 and 8604.39 Gg C yr−1 after filtering the climateinduced NPP change, while grassland reclamation had reduced the

40 S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

total NPP of grassland by 14,529.92 Gg C yr−1. In the unchangedgrassland, the improved grassland management led to the NPP in-crease of 11,230.21 Gg C yr−1.

Our study highlighted that it is human activity, not climatechange, which was the main cause of grassland NPP increase inInner Mongolia between 2001 and 2009, and that vegetation restora-tion programs could remarkably influence LUCC. The land conversionsuch as returning cropland to grassland and converting desert tograssland are the dominant driving forces for the development ofgrassland in Inner Mongolia, and the improved grassland manage-ment such as grazing banning, rotational grazing and convertinggrazing land to cultivated pasture are mainly responsible for the in-crease of the vegetation coverage. However, the restoration projectsshould be further adjusted to cope with the ever-increasing grazingpressure on the original non-degraded or lightly degraded grassland.If implemented adequately and sustainably, the GFGP and GWP willundoubtedly alleviate the grassland degradation and improve China'scarbon sequestration potential as well as sand storm controlling.

Acknowledgments

We would like to thank Yongliang You from China AgriculturalUniversity and Qiang Li from Inner Mongolia University for their helpduring the field investigation. This work was supported by the key pro-ject of Chinese national programs for fundamental research and devel-opment (973 program, 2010CB950702), the China's high-tech specialprojects (863 plan, No. 2007AA10Z231) and APN project (ARCP2011-06CMY-Li). The constructive comments and suggestions from anony-mous reviewers are also highly appreciated.

References

Andersson, E., Brogaard, S., Olsson, L., 2011. The political ecology of land degradation.Annual Review of Environment and Resources 36, 295–319.

Bagan, H., Takeuchi, W., Kinoshita, T., Bao, Y., Yamagata, Y., 2010. Land cover classifica-tion and change analysis in the Horqin Sandy Land from 1975 to 2007. IEEE Journalof Selected Topics in Applied Earth Observations and Remote Sensing 3 (2),168–177.

Bai, Y.F., Han, X.G.,Wu, J.G., Chen, Z.Z., Li, L.H., 2004. Ecosystem stability and compensatoryeffects in the Inner Mongolia grassland. Nature 431, 181–184.

Bronick, C.J., Lal, R., 2005. Soil structure and management: a review. Geoderma 124,3–22.

Cao, S.X., Tian, T., Chen, L., Dong, X.B., Yu, X.X., Wang, G.S., 2010. Damage caused to theenvironment by reforestation policies in arid and semi-arid areas of China. Ambio39, 279–283.

Chen, X.G., Zhang, X.Q., Zhang, Y.P., Wan, C.B., 2009. Carbon sequestration potential ofthe stands under the Grain for Green Program in Yunnan Province, China. ForestEcology and Management 258, 199–206.

Chuluun, T., Ojima, D., 2002. Land use change and carbon cycle in arid and semi-aridlands of East and Central Asia. Science in China Series C 45, 48–54.

Conant, R.T., Paustian, K., Elliott, E.T., 2001. Grassland management and conversion intograssland: effects on soil carbon. Ecological Applications 11, 343–355.

Dai, Z.G., 2010. Intensive agropastoralism: dryland degradation, the Grain-to-GreenProgram and islands of sustainability in the Mu Us Sandy Land of China. Agriculture,Ecosystems & Environment 188 (3), 249–256.

Dickinson, R.E., Henderson-Sellers, A., Kennedy, P.J., 1993. Biosphere-AtmosphereTransfer Scheme (BATS) Version 1e as Coupled to the NCAR Community ClimateModel.

Dong, J.W., Liu, J.Y., Yan, H.M., Tao, F.L., Kuang, W.H., 2011. Spatio-temporal pattern andrationality of land reclamation and cropland abandonment in mid-eastern InnerMongolia of China in 1990–2005. Environmental Monitoring and Assessment179, 137–153.

Field, C.B., Randerson, J.T., Malmstrom, C.M., 1995. Global net primary production: com-bining ecology and remote sensing. Remote Sensing of Environment 51, 74–88.

Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin, F.S.,Coe, M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A.,Kucharik, C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N., Snyder, P.K.,2005. Global consequences of land use. Science 309, 570–574.

Friedl, M.A., McIver, D.K., Hodges, J.C., Zhang, X.Y., Muchoney, D., Strahler, A.H.,Woodcock, C.E., Gopal, S., Schneider, A., Cooper, A., Bacinni, A., Gao, F., Schaaf, C.,2002. Global land cover mapping from MODIS: algorithms and early results.Remote Sensing of Environment 83, 287–302.

Friedl, M.A., Sulla-Menashe, D., Tan, B., Schneider, A., Ramankutty, N., Sibley, A., Huang,X.M., 2009. MODIS Collection 5 Global Land Cover: algorithm refinements andcharacterization of new datasets. Remote Sensing of Environment 114, 168–182.

Gao, J., Liu, Y.S., 2010. Determination of land degradation causes in Tongyu County,Northeast China via land cover change detection. International Journal of AppliedEarth Observation and Geoinformation 12, 9–16.

Gutman, G., Ignatov, A., 1998. The derivation of the green vegetation fraction fromNOAA/AVHRR data for use in numerical weather prediction models. InternationalJournal of Remote Sensing 19, 1533–1543.

Huang, M.B., Liu, X.Z., 2002. Regulation effect of forest vegetation on watershed runoffin the Loess Plateau. Chinese Journal of Applied Ecology 13, 1057–1060.

Imhoff, M.L., Bounoua, L., DeFries, R., Lawrence, W.T., Stutzer, D., Tucker, C.J., Ricketts,T., 2004. The consequences of urban land transformation on net primary produc-tivity in the United States. Remote Sensing of Environment 89, 434–443.

Jiang, G.M., Han, X.G., Wu, J.G., 2006. Restoration and management of the Inner MongoliaGrassland require a sustainable strategy. Ambio 35, 269–270.

John, R., Chen, J.Q., Lu, N., Guo, K., Liang, C.Z., Wei, Y.F., Noormets, A., Ma, K.P., Han, X.G.,2008. Predicting plant diversity based on remote sensing products in the semi-aridregion of Inner Mongolia. Remote Sensing of Environment 112, 2018–2032.

John, R., Chen, J.Q., Lu, N., Wilske, B., 2009. Land cover/land use change in semi-aridInner Mongolia: 1992–2004. Environmental Research Letters 4, 045010.

Ju, W.M., Chen, J.M., Black, T.A., Barr, A.G., McCaughey, H., 2010. Spatially simulatingchanges of soil water content and their effects on carbon sequestration in Canada'sforests and wetlands. Tellus Series B: Chemical and Physical Meteorology 62,140–159.

Kuchelmeister, G., Meyer, N., 2007. Desertification control in China—a formula forsuccess? Agriculture and Rural Development 1, 16–18.

Liu, M.Z., Jiang, G.M., Li, L.H., Li, Y.G., Gao, L.M., Niu, S.L., 2004. Control of sandstorms inInner Mongolia, China. Environmental Conservation 31, 269–273.

Liu, L.S., Zhang, Y.L., Bai, W.Q., Yan, J.Z., Ding, M.J., Shen, Z.X., Li, S.C., Zheng, D., 2006.Characteristics of grassland degradation and driving forces in the source region ofthe Yellow River from 1985 to 2000. Journal of Geographical Sciences 16, 131–142.

Liu, J.G., Li, S.X., Ouyang, Z.Y., Tam, C., Chen, X.D., 2008. Ecological and socieoeconomiceffects of China's policies for ecosystem services. Proceedings of the National Acad-emy of Sciences 105, 9477–9482.

Lu, N., Wilske, B., Ni, J., John, R., Chen, J., 2009. Climate change in Inner Mongolia from1955 to 2005 - trends at regional, biome and local scales. Environmental ResearchLetters 4, 045006.

Ni, J., 2002. Carbon storage in grasslands of China. Journal of Arid Environments 50,205–218.

Normile, D., 2007. Getting at the roots of killer dust storms. Science 317, 314–316.Pei, S.F., Fu, H., Wan, C.G., 2008. Changes in soil properties and vegetation following

exclosure and grazing in degraded Alxa desert steppe of Inner Mongolia, China.Agriculture, Ecosystems & Environment 124, 33–39.

Piao, S.L., Mohammata, A., Fang, J.Y., Cai, Q., Feng, J.M., 2006. NDVI-based increase ingrowth of temperate grasslands and its responses to climate changes in China.Global Environmental Change 16, 340–348.

Pielke, R.A., 2005. Land use and climate change. Science 310, 1625–1626.Potter, C.S., Randerson, J.T., Field, C.B., Matson, P.A., Vitousek, P.M., Mooney, H.A.,

Klooster, S.A., 1993. Terrestrial ecosystem production: a process model based onglobal satellite and surface data. Global Biogeochemical Cycles 7 (4), 811–841.

Qiu, G.Y., Xie, F., Feng, C.Y., Tian, F., 2011. Experimental studies on the effects of the“Conversion of Cropland to Grassland Program” on the water budget and evapo-transpiration in a semi-arid steppe in Inner Mongolia, China. Journal of Hydrology411, 120–129.

Ruimy, A., Saugier, B., 1994. Methodology for the estimation of terrestrial net primary pro-duction from remotely sensed data. Journal of Geophysical Research 97, 18515–18521.

Saxton, K.E., Rawls, W.J., Romberger, J.S., Papendick, R.I., 1986. Estimating generalizedsoil water characteristics from texture. Soil Science Society of America Journal50, 1031–1036.

Su, Y.Z., 2007. Soil carbon and nitrogen sequestration following the conversion of croplandto alfalfa forage land in northwest China. Soil & Tillage Research 92, 181–189.

Su, Y.Z., Zhao, H.L., 2003. Soil properties and plant species in an age sequence ofCaragana microphylla plantations in the Horqin Sandy Land, north China. EcologicalEngineering 20, 223–235.

Su, Z.Y., Wang, X.F., Yang, R., Lee, J., 2010. Effects of sandy desertified land rehabilitationon soil carbon sequestration and aggregation in an arid region in China. Journal ofEnvironmental Management 9, 2109–2116.

Thornthwaite, C.W., 1948. An approach toward rational classification of climate. Geo-graphic Review 38, 55–94.

Tong, C., Wu, J., Yong, S., Yang, J., Yong, W., 2004. A landscape-scale assessment of steppedegradation in the Xilin River Basin, Inner Mongolia, China. Journal of Arid Environ-ments 59, 133–149.

Walko, R.L., Tremback, C.J., 2005. ATMET Technical note. Modifications for the transitionfrom LEAF-2 to LEAF, 3, pp. 1–13.

Wang, X.H., Lu, C.H., Fang, J.F., Shen, Y.C., 2007. Implications for development of grain-for-green policy based on cropland suitability evaluation in desertification-affectednorth China. Land Use Policy 24, 417–424.

Wang, H., Li, X.B., Long, H.L., Gai, Y.Q., Wei, D.D., 2009. Monitoring the effects of landuse and cover changes on net primary production: a case study in China'sYongding River basin. Forest Ecology and Management 258, 2654–2665.

Wang, X., Chen, F., Zhang, J., Yang, Y., Li, J., Hasi, E., Zhang, C.X., Xia, D.S., 2010. Climate,desertification, and the rise and collapse of China's historical dynasties. HumanEcology 38, 157–172.

Wang, S.P., Wilkes, A., Zhang, Z.C., Chang, X.F., Lang, R., Wang, Y.F., Niu, H.S., 2011. Man-agement and land use change effects on soil carbon in northern China's grasslands:a synthesis. Agriculture, Ecosystems & Environment 142, 329–340.

Xu, J.X., 2006. Sand-dust storms in and around the Ordos Plateau of China as influencedby land use change and desertification. Catena 65, 279–284.

41S. Mu et al. / Global and Planetary Change 108 (2013) 29–41

Xu, C., Liu, M., An, S., Chen, J.M., Yan, P., 2007. Assessing the impact of urbanization onregional net primary productivity in Jiangyin County, China. Journal of Environ-mental Management 85, 597–606.

Xu, D.Y., Kang, X.W., Liu, Z.L., Zhuang, D.F., Pan, J.J., 2009. Assessing the relative role ofclimate change and human activities in sandy desertification of Ordos region,China. Science in China Series D 52, 855–868.

Xu, D.Y., Kang, X.W., Zhuang, D.F., Pan, J.J., 2010. Multi-scale quantitative assessmentof the relative roles of climate change and human activities in desertification—a case study of the Ordos Plateau, China. Journal of Arid Environments 74,498–507.

Xu, D.Y., Li, C.L., Zhuang, D.F., Pan, J.J., 2011. Assessment of the relative role of climateChange and human activities in desertification: a review. Journal of GeographicalSciences 21 (5), 926–936.

Yan, H.M., Liu, J.Y., Huang, H.Q., Tao, B., Cao, M.K., 2009. Assessing the consequence ofland use change on agricultural productivity in China. Global and Planetary Change67, 13–19.

Yin, R.S., Yin, G.P., Li, L.Y., 2010. Assessing China's ecological vegetation restorationprograms: what's been done and what remains to be done? EnvironmentalManagement 45, 442–453.

Yu, D.Y., Shi, P.J., Han, G.Y., Zhu, W.Q., Du, S.Q., Xun, B., 2011. Forest ecosystem restorationdue to a national conservation plan in China. Ecological Engineering 37, 1387–1397.

Zhang, G.L., Dong, J.W., Xiao, X.M., Hu, Z.M., Sheldon, S., 2012. Effectiveness of ecologicalrestoration projects in Horqin Sandy Land, China based on SPOT-VGT NDVI data.Ecological Engineering 38, 20–29.

Zhao, H.L., Zhou, R.L., Su, Y.Z., Zhang, H., Zhao, L.Y., Drakec, S., 2007. Shrub facilitation ofdesert land restoration in the Horqin Sand Land of Inner Mongolia. EcologicalEngineering 31, 1–8.

Zhu, L.B., Zeng, Z.H., Zhao, B.P., Wang, X., Hu, Y.G., Hai, T., 2008. Effect of spring rest-grazing on steppe vegetation. Acta Agrestia Sinica 16 (3), 278–282 (in Chinesewith English abstract).

Zika, M., Erb, K., 2007. Net primary production losses due to human-induced desertifi-cation. Second International Conference on Earth System Modelling (ICESM).