Spatial variability of N2O, CH4 and CO2 fluxes within the Xilin River catchment of Inner Mongolia, China: a soil core study

REGULAR ARTICLE

Spatial variability of N2O, CH4 and CO2 fluxeswithin the Xilin River catchment of Inner Mongolia, China:a soil core study

Zhisheng Yao & Benjamin Wolf & Weiwei Chen & Klaus Butterbach-Bahl &Nicolas Brüggemann & Martin Wiesmeier & Michael Dannenmann &

Benjamin Blank & Xunhua Zheng

Received: 31 August 2009 /Accepted: 7 December 2009# Springer Science+Business Media B.V. 2009

Abstract In order to identify the effects of land-use/cover types, soil types and soil properties on the soil-atmosphere exchange of greenhouse gases (GHG) insemiarid grasslands as well as provide a reliableestimate of the midsummer GHG budget, nitrousoxide (N2O), methane (CH4) and carbon dioxide(CO2) fluxes of soil cores from 30 representativesites were determined in the upper Xilin Rivercatchment in Inner Mongolia. The soil N2O emissionsacross all of the investigated sites ranged from 0.18 to

21.8 μg N m-2 h-1, with a mean of 3.4 μg N m-2 h-1

and a coefficient of variation (CV, which is given as apercentage ratio of one standard deviation to themean) as large as 130%. CH4 fluxes ranged from-88.6 to 2,782.8 μg C m-2 h-1 (with a CV of 849%).Net CH4 emissions were only observed from corestaken from a marshland site, whereas all of the other29 investigated sites showed net CH4 uptake (mean:-33.3 μg C m-2 h-1). CO2 emissions from all sitesranged from 3.6 to 109.3 mg C m-2 h-1, with a meanvalue of 37.4 mg C m-2 h-1 and a CV of 66%. Soilmoisture primarily and positively regulated the spatialvariability in N2O and CO2 emissions (R2=0.15–0.28,P<0.05). The spatial variation of N2O emissions wasalso influenced by soil inorganic N contents (P<0.05). By simply up-scaling the site measurements bythe various land-use/cover types to the entire catch-ment area (3,900 km2), the fluxes of N2O, CH4 andCO2 at the time of sampling (mid-summer 2007) wereestimated at 29 t CO2-C-eq d-1, -26 t CO2-C-eq d-1

and 3,223 t C d-1, respectively. This suggests that, interms of assessing the spatial variability of total GHGfluxes from the soils at a semiarid catchment/region,intensive studies may focus on CO2 exchange, whichis dominating the global warming potential ofmidsummer soil-atmosphere GHG fluxes. In addition,average GHG fluxes in midsummer, weighted by theareal extent of these land-use/cover types in theregion, were approximately -30.0 μg C m-2 h-1 forCH4, 2.4 μg N m-2 h-1 for N2O and 34.5 mg C m-2 h-1

for CO2.

Plant SoilDOI 10.1007/s11104-009-0257-x

Responsible Editor: Per Ambus.

Z. Yao :W. Chen :X. Zheng (*)State Key Laboratory of Atmospheric Boundary LayerPhysics and Atmospheric Chemistry, Institute ofAtmospheric Physics, Chinese Academy of Sciences,Beijing 100029, Chinae-mail: [email protected]

Z. Yao : B. Wolf :K. Butterbach-Bahl :N. Brüggemann :M. DannenmannInstitute for Meteorology and Climate Research,Atmospheric Environmental Research,Karlsruhe Research Center,Garmisch-Partenkirchen 82467, Germany

M. WiesmeierCenter of Life Science, Department of Ecology,Technical University of Munich,Freising-Weihenstephan 85350, Germany

B. BlankInstitute of Landscape Ecology and ResourcesManagement, Justus-Liebig-University Giessen,Giessen 35392, Germany

Keywords GHG fluxes . Land-use/cover .

Semi-arid grassland . Xilin River catchment

Introduction

Terrestrial ecosystems are dominating sources andimportant sinks for the greenhouse gases (GHG)carbon dioxide (CO2), methane (CH4) and nitrousoxide (N2O) (Chapius-Lardy et al. 2007; IPCC 2007).Grasslands are one of the most important globalterrestrial biome types, covering approximately 25%of the global land surface. The soil-atmosphereexchange of GHGs in these ecosystems, therefore,may be of significant importance for the compositionof the atmosphere and, thus, for the global climate.Studies dealing with GHG fluxes from grassland soilswere mostly carried out in Australia, Europe andNorth America (e.g., Denmead et al. 1979; Mosier etal. 1991, 1996, 1997; Skiba et al. 1998; Glatzel andStahr 2001; Turner et al. 2008), whereas reports forGHG fluxes for the Inner Asian steppe belt are stillscarce (Du et al. 2006; Galbally et al. 2008).Published reports show that soil moisture, temperatureand grazing as well as N-fertilization have a greatimpact on CH4 and N2O fluxes from the grasslands(e.g., Velthof and Oenema 1995; Holst et al. 2007a, b;Liu et al. 2007; Zhang and Han 2008).

China is covered by approximately 400 millionhectares of grassland, accounting for nearly 12.5% ofthe global grassland area, of which about 87 millionhectares belong to Inner Mongolia. Based on verylimited flux data observed in the semi-arid grasslandarea in China, Chen et al. (2000) roughly estimatedthe annual total N2O emissions from Chinese grass-lands to be approximately 112 Gg N. Estimates forCH4 and CO2 exchange at this scale are missing.However, the regional estimate for N2O is highlyuncertain, since spatial and temporal variability of theN2O fluxes in grassland ecosystems are supposed tobe high and the measurement database is still limited.Understanding GHG fluxes from the steppe soils isnecessary if we are to improve our knowledge of C-and N-trace gas fluxes from global semiarid and aridzones, where there is a poorly quantified understand-ing of their contribution to the global trace gas cycles(Galbally et al. 2008).

It is recognized that these gases are all produced orconsumed as a result of microbial processes in the

soil, but the magnitude of the fluxes between the soiland the atmosphere depends heavily on soil physi-ochemical parameters, such as mineral N content, C:Nratio, temperature and water content (e.g., Smith et al.2003; Lilly et al. 2009). For example, Sozanska et al.(2002) made a prediction of the N2O in Great Britainby establishing a statistical relationship between thefield measurements of N2O emissions and nitrogeninput, soil temperature, and water and carbon contentfrom over 50 field studies from Europe and NorthAmerica. Similarly, Lilly et al. (2009) estimated N2Oemissions in Scotland using soil type, climate, andland-use and management information. Therefore, tounderstand better the spatial variability of soil N2O,CH4 and CO2 fluxes of steppe in Inner Mongolia, wedesigned a soil core study with sampling points beingdistributed across the landscape in such a way thatmajor soil types and different land-use/cover typeswere covered in a representative pattern. Thesedifferent soil and land-use/cover types may havedifferent dynamics of C- and N-cycling which in turnis proximally controlled by environmental and edaphiccharacteristics (e.g., Matson et al. 1991; Corre et al.1996), and, thus, GHG fluxes could also be expectedto vary. Our study focused on the upper Xilin Rivercatchment, i.e., a region where a series of GHGstudies were carried out recently. These studies (e.g.,Wang et al. 2001; Xu-Ri et al. 2003; Holst et al.2007a, b; Liu et al. 2007, 2009a) focused on grazingand topographic effects, but did not investigate apossible regional variability in GHG fluxes. Theabove cited studies showed that, during the growingseason, the upland steppe was a weak, but asignificant net source of N2O (0.8±0.4 μg N m-2 h-1;Holst et al. 2007b) and a significant sink foratmospheric CH4 in the range of 28–77 μg C m-2 h-1

was found (Wang et al. 2005a, 2009; Liu et al. 2007,2009a). Since the studies focused on a few sites inclose vicinity (approximately 10-km radius), wewondered if on a larger scale and for different landuses (steppe, riparian areas, arable land as convertedfrom steppe) and land cover (steppe types: mountainsteppe, Leymus chinensis steppe, and Stipa grandissteppe) types, the reported magnitude of the GHGexchange would be comparable or not.

Our objectives in this study were to investigate theregional spatial pattern of the soil-atmosphere ex-change of N2O, CO2 and CH4 and its regulatingfactors. In particular, we tested the following hypoth-

Plant Soil

eses: (1) Physicochemical soil parameters, i.e., soiltypes are controls of the regional spatial variability ofmidsummer GHG fluxes; (2) Land-use/cover influencesmidsummer fluxes; and (3) land-use types with hot-spot-like high GHG fluxes (e.g., marshland) signif-icantly affect the regional midsummer GHG budget,despite their low coverage of the total area.

Materials and methods

Site description

During the period from July 20 to August 1, inmidsummer 2007, we performed an intensive sam-pling campaign within the upper Xilin River catch-ment (43°26′–44°39′N, 115°32′–117°12′ E). Thecatchment comprises an area of 3,900 km2 (Holst etal. 2007b). Sampling sites were selected in thevicinity of the Inner Mongolia Grassland EcosystemResearch Station (IMGERS), Chinese EcosystemResearch Network (CERN). The station is locatedapproximately 65 km southeast of Xilinhot city,which itself is approximately 450 km north ofBeijing. This area has a cool-temperate semi-aridcontinental monsoon climate, with a frost-free grow-ing season of 90–110 days (May–September). Thelong-term annual mean air temperature is 0.8°C(1982–2006), with the maximum monthly mean of19°C in July and the minimum of −21°C in January.The average annual precipitation for the period 1982–2006 of approximately 330 mm (166–507 mm) isdistributed unevenly among seasons. Sixty to eightypercent of the rainfall falls in June–August (meteoro-logical data provided by IMGERS). The soil types inour study area are predominantly Chernozems asdeveloped on loess parent material (86.4%). Thesesoils mostly have a sandy loam texture. Generally, thedominant land-use management in the Xilin Rivercatchment is non-nomadic sheep pastors, and thus,perennial grazing occurs in all of the land-use/covertypes except for the cultivated arable land sites.

Sampling scheme/ecological units

For the sampling design, we used a stratifiedsampling plan (Brus and de Gruijter 1997) to defineecological units representing the heterogeneity of thecatchment. Land-use and topography were taken as

the stratifying variables. Based on a Landsat TM7image from August 17th, 2005, seven land-use classeswere defined using Idrisi Kilimanjaro (version 14.02,Clark Labs, Worcester MA, USA) and taken as thefirst stratifying variable. The second stratifyingvariable was the topographic wetness index (TWI),defined as lnða=tan bÞ, where a is the specificcatchment area and β is the surface slope (Wilsonand Gallant 2000; Minasny and McBratney 2007). Asa result, the catchment was divided into four mainland-use/cover types and two extra types. The land-use class of water did not play a role in terms of theselection of the sampling sites (Fig. 1):

a) Typical steppe (TS). This is the most commonland cover type and can be found on 85% of thearea of the Xilin River catchment. Typical steppecan be found at all topographic positions (hilltop,slope, foothills) and its spatial distribution in thestudy area coincides with the spatial distributionof Chernozems. The most dominant grasslandspecies is either Leymus chinensis or Stipagrandis.

b) Sand dune (SD). Sand dunes can be found on6.5% of the study area. They are characterized byundulating topography where sand forms asequence of depressions and hills. Dense vegeta-tion, particularly Ulmus pumila and other treegenera, i.e., Betulus spp., Malus spp., Prunus spp.and Populus spp. dominate north- to northwest-facing slopes or in depressions, while south- tosouthwest-facing slopes have generally sparsershrubs and grassland vegetation.

c) Mountain meadow (MM). This vegetation typecovers approximately 4.5% of the study area. It isfound in the east and northeast of the Xilin Rivercatchment. The landscape is characterised bysteep hills with relatively deep valleys. Thevalleys accumulate water and create favourableconditions for meadow grassland (Bromus inermis,Agrostis gigantea, Carex pediformis, Stipa baica-lensis and Calamagrostis epigeios).

d) Marshland (ML). This type covers only a tinyportion of the study area (approximately 0.4%). Itborders most of the Xilin River, its tributaries andsome flat areas of episodic streams. These areasare divided into two parts: dry and wet marsh-land. Dry marshland is frequently wet, but itbecomes dry during one to three months of the

Plant Soil

summer. In contrast, wet marshland is wet allyear round. However, wet marshland has alsobeen noted to dry out during years with lowsummer rainfalls (approximately 100 mm). Thevegetation found in these areas consists mainly ofmarshland species such as Phragmites australis,Carex appendiculata, Iris lactea var. chinensisand Hippuris vulgaris.

e) Arable land (AL). Some areas (approximately1.4% of catchment area), mostly being flat and inthe vicinity of villages or hamlets, have beentransformed from typical steppe into arable landover in the last few decades. The main crop typesare spring wheat, rapeseed, potato and maize.Cropping is done in such a way that one year witha field crop is followed by a fallow period of oneyear. These areas had been in continuous cultiva-tion (mostly) without irrigation or the applicationof nitrogen fertilizer since the typical steppe wasploughed in the 1980s. The weeds of the fallowyear are tilled into soil as green manure.

f) Bare soil (BS). It can be referred to as one land-use/cover type with hardly any vegetation cover. Thistype can be found on 2.1% of the catchment area,which represents an area that is either heavilydegraded or free of vegetation due to topographicor hydrological conditions.

Sample collection and flux measurements

Soil core sampling was done in the midsummer of2007 over a short-term period. Meteorological con-ditions during the entire sampling period were stableand without rainfall. In total, we sampled 30 sitesacross the study area, with more sampling sites forthose land-use/cover types with greater spatial impor-tance. That is, we sampled nine TS sites, six SD sites,six MM sites, three ML sites, three AL sites and threeBS sites. All of the sampling sites except for thearable sites, showed strong indications of grazing, i.e.,aboveground biomass coverage was rather sparse, butthe roots were still present in the soil. With respect tothe arable land, we took the soil cores from thecropped sites but excluded the aboveground biomass.The location of, and the general information about,each site is shown in Fig. 1 and Table 1. At eachsampling site, on average, six intact soil cores (40 cmin depth, and 16 cm in diameter) were randomly takenin an area of approximately 200 m2 by driving 50-cmlong PVC tubes (outer diameter: 16.5 cm) into the soiland carefully digging the tubes up. That means PVCtubes still have 10 cm length headspace left on the topof the intact soil cores for creating a chamber, whichwas used for gas flux measurements. To avoid soilcompression, a cylinder steel frame, which had a

Fig. 1 Distribution ofland-use/cover types andlocation of the 30 samplingsites in the Xilin Rivercatchment

Plant Soil

slightly smaller inner diameter and slightly higherouter diameter as compared to the PVC tube andvertically split into two equal parts, was driven intothe soil part by part and then pulled out. In this way,the roots were cut and the core was separated from thesurrounding soil before the PVC tube was insertedinto the soil. Meanwhile, the PVC tubes were beveledat one end to facilitate insertion into the soil. Thus noobvious compression of the soil was observed. Coreswere closed gas- and water-tight at the bottom andcovered at the top site with a perforated plastic foil toallow gas exchange but prevent quick drying out.

Cores were transferred at the same day to theIMGERS research station for further processing.

We recognized the possibility of influence on CO2

and N2O fluxes from the soil cores due to decay of rootmaterial cut in the soil and soil disturbance with PVCtube insertion. For example, Matson et al. (1990)suggest that temporary installation of chamber basesmay cause an increase of trace gas fluxes in Amazonianecosystems. However, experiments in a forest in CostaRica did not show a similar chamber installation effect(Keller and Reiners 1994). Prior to gas flux measure-ments at the ambient air in IMGERS, therefore, soil

Table 1 Location and characteristics of sampling sites

Site Land-use/cover Latitude Longitude Elevation (m) Slope Soil typea

BS1 Bare soil (BS) 43°37.8′N 116°43.4′E 1,193 0 o (plain) Arenosol

BS2 Bare soil (BS) 43°43.7’N 116°19.5′E 1,089 3-4 o (slope) Arenosol

BS3 Bare soil (BS) 43°42.9′N 116°18.1′E 1,051 0 o (plain) Arenosol

SD1 Sand dune (SD) 43°39.2′N 116°38.3′E 1,218 15-20 o (slope) Arenosol

SD2 Sand dune (SD) 43°38.4′N 116°49.4′E 1,308 10 o (slope) Arenosol

SD3 Sand dune (SD) 43°48.3′N 116°56.8′E 1,304 3-4 o (slope) Arenosol

SD4 Sand dune (SD) 43°41.5 N 116°26.5′E 1,169 2-4o (slope) Arenosol

SD5 Sand dune (SD) 43°40.7′N 116°33.4′E 1,176 2-3o (slope) Arenosol

SD6 Sand dune (SD) 43°37.1′N 116°53.2′E 1,280 1o (slope) Arenosol

ML1 Marshland (ML) 43°37.0′N 116°42.9′E 1,181 0 o (plain) Gleysol

ML2 Marshland (ML) 43°39.3′N 117°06.4′E 1,326 0 o (valley floor) Phaeozem

ML3 Marshland (ML) 43°46.9′N 116°10.6′E 1,018 0 o (plain) Arenosol

TS1 Typical steppe (TS) 43°39.9′N 116°16.4′E 1,140 6-8 o (slope) Chernozem

TS2 Typical steppe (TS) 43°43.6′N 116°28.8′E 1,237 2-3 o (slope) Phaeozem

TS3 Typical steppe (TS) 43°46.6′N 116°39.7′E 1,286 2-3 o (slope) Phaeozem

TS4 Typical steppe (TS) 43°39.0′N 117°04.9′E 1,338 3-4 o (slope) Phaeozem

TS5 Typical steppe (TS) 43°48.8′N 116°33.6′E 1,136 0 o (plain) Chernozem

TS6 Typical steppe (TS) 43°32.8′N 116°20.6′E 1,299 2-4 o (slope) Arenosol

TS7 Typical steppe (TS) 43°33.9′N 116°26.2′E 1,235 0 o (plateau) Chernozem

TS8 Typical steppe (TS) 43°53.5′N 116°32.1′E 1,152 2-3 o (slope) Chernozem

TS9 Typical steppe (TS) 43°58.4′N 116°42.7′E 1,195 0 o (plain) Chernozem

MM1 Mountain meadow (MM) 43°40.2′N 116°49.8′E 1,401 3-5 o (slope) Phaeozem

MM2 Mountain meadow (MM) 43°42.5′N 116°58.7′E 1,486 2-3 o (slope) Phaeozem

MM3 Mountain meadow (MM) 43°55.1′N 116°51.2′E 1,425 0 o (crest) Chernozem

MM4 Mountain meadow (MM) 43°55.3′N 116°48.7′E 1,423 3-4 o (slope) Chernozem

MM5 Mountain meadow (MM) 43°43.1′N 117°00.9′E 1,372 0 o (plain) Phaeozem

MM6 Mountain meadow (MM) 43°45.3′N 116°57.4′E 1,392 0 o (valley floor) Phaeozem

AL1 Arable land (AL) 43°46.8′N 116°22.5′E 1,078 1-2o (slope) Chernozem

AL2 Arable land (AL) 43°32.7′N 116°39.2′E 1,266 4-5 o (slope) Chernozem

AL3 Arable land (AL) 43°45.2′N 116°36.5′E 1,198 0 o (depression) Chernozem

a Soil type classification of world reference base for soil resources 2006 (IUSS Working Group WRB 2006, http://www.fao.org/ag/Agl/agll/wrb/doc/wrb2006final.pdf). The description in the parentheses for Slope indicates the topography of each sampling site

Plant Soil

cores were equilibrated 24 hours in the outdoor, in orderto reduce soil disturbance effects and keep conditions asclose to the actual ambient status as possible. Duringthis time, we did not make any treatment like irrigationto the soil cores. The gas fluxes (CO2, CH4 and N2O)were measured using a static chamber method. It isnoteworthy that the measured CO2 emissions (ecosys-tem respiration) were mainly dominated by the respira-tion of the soil microbes and living roots remaining inthe soil cores since only few aboveground biomasseswere left in the soil cores. For this, the PVC tubes wereclosed gas-tight with stainless steel lids, enclosing aheadspace volume of ca. 2 L. The lids were equippedwith a circulating fan (A9-V fan, Sinwan Industrial Co.,Ltd, Taiwan) to gently mix the gas headspace during thesampling periods. For gas sampling each lid had twoports, one with a septum for gas sampling, the otherheld a Teflon tube (with outer diameter of 1/8 inchesand length of 20 cm) to ensure the equilibration of airpressure between inside and outside. Five gas sampleswere taken from the chamber headspace during a1-h closure period, at an interval of 15 min. Gassamples were taken with 20 mL plastic syringes whichcould be closed with three-way stopcocks. The airsamples were immediately analysed using a gaschromatograph (Agilent 6890D, Agilent Technologies,Palo Alto, California, USA) equipped with an electroncapture detector (ECD) for N2O detection at 330°C anda flame ionization detector (FID) for CH4 and CO2

detection at 200°C (Wang and Wang 2003). High puritynitrogen (N2) was used as the carrier gas with a flowrate of 25 mL min-1. The CH4 was separated with a 2-mstainless steel column (2 mm in the inner diameter),filled with 13 XMS (60/80 mesh, Sigma-Aldrich Co.,St. Louis MS, USA). The CO2 was separated with a2-m stainless steel column (2 mm in the inner diameter)filled with Porapak Q (60/80 mesh, St. Louis MS,USA). The CH4 was directly measured by FID, whileCO2 was converted to CH4 prior to analysis with theFID, using a Ni catalyst and H2 (“methanizer”) under375°C. The N2O was separated using two stainless steelcolumns (both with an inner diameter of 2 mm, onewith a length of 1 m and another one with a length of2 m). Both were filled with Porapak Q (80/100 mesh).For the N2O analysis, a buffer gas (CO2:N2=10:90)flowed through the ECD cell at a rate of 1–3 mL min-1

to remove the CO2 influence upon the N2O signals.More details about the gas chromatograph configura-tions and setting up for the gas analysis are found in

Zheng et al. (2008). Gas concentration changes in CH4,CO2 and N2O against enclosure time were significantlylinear over the entire incubation period. Thus, thefluxes were calculated from linear regression coeffi-cients of gas concentrations against sampling times.The minimum detection limits of gas fluxes were0.6 μg N m-2 h-1 for N2O, 4.4 μg C m-2 h-1 for CH4

and 0.3 mg C m-2 h-1 for CO2, respectively. The soiltemperature at a depth of 5 cm and the air temperatureduring each chamber enclosure were recorded withthermocouples (JM624, Tianjin Jinming InstrumentCo. and Ltd., China). Air pressure data were deter-mined at the nearby automatic meteorological stationof IMGERS. Measured air temperature of chamberenclosures and air pressure were used to correct the gasdensity in the flux calculation.

The measurements of the GHG fluxes in our studywere based on replicated soil core measurements fromreplicated sites (three to nine) representing differentland-use/cover types (Table 1), totally yielding asample number of approximately 180 for the entirecatchment. Owing to high labour demand for thiswork, gas fluxes were recorded once from each soilcore during the study period. Fluxes with an absolutevalue higher than the detection limit were accepted asvalid data. Over the entire observation period, 165 ofthe N2O data, 164 of the CH4 data and 177 of the CO2

data were accepted as valid data across all of theinvestigated sites, and the remainders were rejected asnull data (<8.0% of total data).

Auxiliary measurements

After the gas flux measurements, we processed half ofthe soil cores (at least three) from each sampling site tomeasure soil moisture at 0–10 cm, 10–20 cm and 20–30 cm depths, using a hand-held TDR (Time DomainReflectometry) probe (Theta Probe, ThetaKit, Delta-TDevices, Cambridge, UK). Meanwhile, soil sampleswere taken from a 0–10 cm depth of each processed soilcore. These samples were sieved carefully with a meshwidth of 3.15 mm, and then extracted with a 1 mol L-1

KCl solution for determination of inorganic NH4+ and

NO3- concentrations. These analyses were done at

IMGERS using a photometric flow-injection analyser(FIA Star 5000, Foss Inc., Hillerød, Denmark).

For the analyses of the soil texture of the soil pits, fineearth (< 2 mm) was oxidized with H2O2 to removeorganic material. The remaining material was dispersed

Plant Soil

with Na4P2O7 and shaken for at least 16 hours, followedby wet sieving to isolate sand fractions of 2000 to630 µm, 630 to 200 µm and 200 to 63 µm. Todetermine silt and clay fractions, approximately 3 g ofthe <63 µm fraction was suspended in deionized waterwith Na4P2O7 and an ultra-sonication was conducted forthree minutes with 75 J ml-1. Afterwards the distributionof 63 to 20 µm, 20 to 6.3 µm, 6.3 to 2 µm and <2 µmfractions were obtained by measuring the X-rayabsorption of the soil-water suspension during thesedimentation of the soil particles with a MicromeriticsSedigraph 5100 (Micromeritics, Norcross, USA).

The bulk density was quantified with the mass ofthe oven-dry soil (105°C) divided by the core volume(Steffens et al. 2009). Soil pH values were measuredin 0.01 M CaCl2 at a soil/solution ratio of one to 2.5at room temperature.

The soil organic carbon (SOC) and total nitrogen(TN) were determined in duplicate by dry combustionon a Vario Max CNS elemental analyser (Elementar,Hanau, Germany). The measured C concentrations ofthe samples that were free of carbonate, representingthe SOC concentration. Samples that containedCaCO3 were heated to 500°C for four hours toremove organic carbon and the concentration ofinorganic C of the residual material was determinedby dry combustion. The content of inorganic C wassubtracted from the C concentration of the untreatedmaterial and represented the SOC content.

Statistical analysis

The statistical software packages SPSS 12.0 (SPSSInc., Chicago, USA) and Origin 7.0 (Origin LabCorporation, USA) were used for linear correlationanalysis, multiple linear regression analysis and One-Way ANOVA testing of N2O, CH4 and CO2 fluxes,and the measured physicochemical soil parameters.

Results

Soil parameters according to soil and land use/covertypes

The general soil properties of each site are summarisedin Table 2. Chernozems represent the dominating soiltype covering a large area (86.4%) in the Xilin Rivercatchment. Compared to Chernozems, the investigated

soil properties of Arenosols were not significantlydifferent. In contrast, the average values of SOC, TN,pH, ammonium concentrations and WFPS in theGleysol site were significantly higher than those ofChernozems (P<0.05), whereas the bulk density(1.09 g cm-3) and clay content (3.2%) of the Gleysolsite were significantly lower (P<0.05). The averagevalues of SOC, TN and ammonium concentrations inPhaeozems were greatly larger compared to Cherno-zems (P<0.05), while the mean pH values in theinvestigated Phaeozems (5.9) were significantly lowercompared to the other soil types (P<0.05). Soil C:Nratios ranged from 8.4 to 13.8, with no significantdifferences between the soil types (Table 2).

Typical steppe was the most important land-use/cover type in the Xilin River catchment. Compared tothe typical steppe, the arable land showed significantlyhigher (P<0.05) soil moisture values at the time ofsampling (33.8% vs. 12.3% WFPS on average), butsignificantly lower (P<0.05) soil ammonium concen-trations. The majority of mountain meadow sites werelocated in the Phaeozems areas. As a consequence ofthe differences between Chernozems and Phaeozems,SOC (41.4 g kg-1 on average), TN (3.73 g kg-1 onaverage) and clay (21.2% on average) contents weresignificantly higher (P<0.05) for the mountain mead-ow sites on Phaeozems, whereas soil pH values (5.8on average) were significantly lower (P<0.05) com-pared to the typical steppe with the predominating soiltype Chernozems. For the sites of bare soil and sanddunes, the mean values for WFPS in each depth weresignificantly lower (P<0.05) than those of the typicalsteppe. In contrast, WFPS in marshland sites (49.2%on average) was significantly higher (P<0.01) thanthat of the typical steppe. Soil nitrate contents werenot significantly different among the land-use/covertypes (Table 2).

CH4 fluxes

The overall average CH4 flux across all land-use/cover types (ML3 excluded) was calculated as -33.3±19.2 μg C m-2 h-1. The CH4 fluxes of the 29 sites(ML3 excluded) ranged from −88.6 to −7.8μg Cm-2 h-1,with a coefficient of variation (CV) of 58% (Table 3).We observed no significant difference in the CH4

fluxes (ML3 excluded) among all of the soil types(Table 4). The CH4 fluxes from each land-use/covertype demonstrated high spatial variations, with CVs

Plant Soil

ranging from 31% to 178%. The mean CH4 flux forarable land was −35.9±11.0 μg C m-2 h-1, and, thus,not statistically different from the typical steppe(−33.8±24.1 μg C m-2 h-1). For cores taken frommarshland sites integrated over all three samplingsites, the average CH4 flux of the marshland was

912.0±1,620.2 μg C m-2 h-1. Statistical analysisshowed no significant differences in the CH4 fluxamong all of the land-use/cover types (Table 5,Fig. 2). A significant relationship between CH4 fluxesand soil moisture, temperature, or other soil parame-ters could not be demonstrated.

Table 2 General soil properties of the sampling sites

Site pH SOCa TNb C/N BDc NH4+d NO3

-d %WFPS e Soil texture (%)

0-10 10-20 20-30 Sand Silt Clay

BS1 6.6 2.37(0.01) 0.25(0.02) 9.5 1.62 1.22(0.20) 0.65(0.16) 6.2 3.6 2.8 96.1 2.2 1.7

BS2 5.6 15.48(0.02) 1.55(0.01) 10.0 1.28 1.02(0.26) 2.01(1.80) 7.0 9.5 11.6 67.9 20.9 11.2

BS3 5.6 5.79(0.10) 0.67(0.02) 8.6 1.60 0.46(0.12) 0.23(0.21) 7.7 14.4 20.1 83.5 10.1 6.4

SD1 6.3 62.44(0.01) 5.17(0.02) 12.1 0.79 1.78(1.09) 0.26(0.13) 2.3 7.0 6.6 98.0 0.6 1.4

SD2 6.2 26.33(0.07) 1.91(0.01) 13.8 1.03 0.84(0.35) 1.64(0.78) 8.4 6.6 4.0 90.0 6.7 3.3

SD3 6.6 4.30(0.13) 0.40(0.01) 10.8 1.54 0.90(0.46) 3.02(1.51) 12.4 7.6 0.7 96.1 2.0 1.9

SD4 6.4 8.69(0.04) 0.91(0.02) 9.5 1.44 1.10(0.58) 0.63(0.43) 6.4 7.9 10.6 80.3 13.4 6.3

SD5 6.7 11.32(0.28) 1.17(0.04) 9.7 1.49 0.64(0.29) 0.18(0.17) 6.5 6.2 7.0 73.5 18.2 8.3

SD6 6.2 8.02(0.16) 0.82(0.01) 9.8 1.59 0.37(0.14) 1.69(0.56) 9.3 2.2 3.3 85.1 10.0 4.9

ML1 7.4 44.66(0.06) 4.82(0.02) 10.1 1.09 5.36(3.16) 0.54(0.15) 57.1 44.1 41.0 91.9 4.9 3.2

ML2 6.3 33.63(0.06) 2.85(0.01) 13.2 1.06 3.93(2.96) 0.88(0.45) 64.8 34.1 36.5 87.2 7.5 5.3

ML3 6.7 6.84(0.13) 0.78(0.04) 8.8 1.57 0.79(0.16) 0.21(0.11) 64.9 54.7 45.9 78.7 14.0 7.3

TS1 6.8 8.05(0.01) 0.98(0.02) 8.4 1.40 0.71(0.30) 0.36(0.32) 6.9 9.1 9.0 79.3 12.3 8.4

TS2 6.1 25.77(0.02) 2.46(0.01) 10.6 1.26 1.97(0.30) 0.41(0.30) 10.3 13.4 14.7 58.0 25.7 16.3

TS3 6.3 21.72(0.06) 2.16(0.02) 10.1 1.21 2.42(0.54) 3.64(1.74) 8.8 9.3 8.7 54.7 29.7 15.6

TS4 6.2 19.83(0.22) 2.02(0.05) 9.8 1.32 1.85(0.52) 3.66(2.87) 37.6 8.9 3.5 60.6 25.1 14.3

TS5 6.2 14.12(0.13) 1.53(0.02) 9.2 1.47 1.84(0.27) 0.57(0.29) 17.9 11.4 12.4 60.7 25.5 13.8

TS6 7.4 10.17(0.35) 1.07(0.05) 9.5 1.41 1.30(0.12) 0.54(0.09) 11.2 10.2 7.7 80.8 11.9 7.3

TS7 6.3 10.30(0.12) 1.15(0.02) 9.0 1.34 0.87(0.38) 0.52(0.12) 9.2 12.5 18.6 75.9 14.7 9.4

TS8 6.4 18.37(0.13) 1.95(0.02) 9.4 1.34 2.35(0.31) 3.77(3.13) 10.9 10.0 4.0 58.8 24.9 16.3

TS9 7.1 34.15(0.08) 2.99(0.02) 11.4 1.17 3.22(1.05) 3.73(1.71) 12.3 20.1 22.9 49.9 26.9 23.2

MM1 6.1 21.97(0.45) 2.07(0.06) 10.6 1.15 4.87(2.03) 2.87(0.60) 12.0 14.3 15.6 70.2 18.4 11.4

MM2 5.5 45.22(0.06) 4.03(0.01) 11.2 1.09 8.47(6.10) 5.14(1.89) 19.9 7.6 11.1 34.0 40.3 25.7

MM3 6.9 54.93(0.29) 4.93(0.04) 11.1 0.85 1.27(0.34) 0.61(0.38) 11.8 17.2 16.4 21.4 53.1 25.5

MM4 5.6 27.44(0.13) 2.54(0.01) 10.8 1.09 1.49(0.54) 0.49(0.42) 14.1 10.3 13.4 28.3 47.7 24.0

MM5 5.9 36.14(0.05) 3.47(0.01) 10.4 1.23 2.34(0.34) 1.04(0.54) 14.6 11.3 13.4 56.5 28.2 15.3

MM6 5.0 62.74(0.08) 5.33(0.01) 11.8 0.77 1.56(0.78) 0.58(0.56) 18.2 8.6 9.6 26.8 48.2 25.0

AL1 6.8 6.00(0.04) 0.69(0.01) 8.7 1.54 0.43(0.07) 0.59(0.49) 24.5 42.3 51.9 84.3 8.7 7.0

AL2 7.2 11.69(0.33) 1.21(0.04) 9.7 1.17 0.43(0.20) 2.80(1.01) 26.3 29.3 35.9 59.5 25.5 15.0

AL3 7.5 18.36(0.11) 1.91(0.06) 9.6 1.37 0.44(0.82) 11.90(7.10) 25.1 31.2 38.1 66.9 22.2 10.9

a SOC: soil organic carbon content (g kg-1 ) at 0-20 cm soil depthb TN: total nitrogen content (g kg-1 ) at 0-20 cm soil depthc BD: bulk density (g cm-3 )d NH4

+ and NO3- , in mg N kg-1 soil dry weight

eWFPS: water-filled pore space at 0-10, 10-20 and 20-30 cm soil depth, respectively. Numbers in parentheses for SOC, TN, NH4+

and NO3- indicates the standard deviation

Plant Soil

Site Land-use/covertypea

AT (°C) ST (°C) CH4 flux CO2 flux N2O flux

BS1 BS 23.4 21.1 -14.5±21.5 36.8±13.4 1.5±1.1

BS2 BS 25.5 22.9 -74.9±24.5 35.7±12.7 3.1±1.0

BS3 BS 25.5 22.9 -26.2±8.1 18.7±7.9 1.3±0.6

SD1 SD 23.4 21.1 -39.6±13.9 50.2±13.8 2.4±1.0

SD2 SD 21.4 18.4 -42.5±8.7 29.3±7.6 5.3±1.9

SD3 SD 19.1 16.7 -20.1±7.3 19.7±16.1 4.4±2.4

SD4 SD 24.5 21.6 -44.4±20.4 28.7±13.1 0.18±2.3

SD5 SD 24.5 21.6 -33.4±35.0 8.7±5.8 1.0±1.3

SD6 SD 19.1 16.7 -19.5±13.3 15.7±5.1 1.4±0.7

ML1 ML 22.6 20.2 -28.0±25.8 95.5±37.3 21.8±30.3

ML2 ML 19.6 17.3 -18.9±9.7 37.7±11.2 2.0±1.5

ML3 ML 20.2 17.1 2,782.8±2,511.5 56.1±18.7 1.6±7.6

TS1 TS 25.5 22.9 -39.8±24.0 25.8±17.8 1.3±1.4

TS2 TS 24.5 21.6 -26.4±13.9 50.7±34.7 6.4±11.2

TS3 TS 18.8 15.8 -27.9±9.9 5.6±7.1 1.4±1.6

TS4 TS 19.6 17.3 -28.4±11.7 109.3±43.3 3.5±3.9

TS5 TS 18.8 15.8 -8.2±45.6 3.6±2.3 0.44±1.2

TS6 TS 26.9 23.7 -45.5±20.1 53.6±40.1 1.5±1.7

TS7 TS 26.9 23.7 -88.6±19.4 38.0±29.3 3.1±1.2

TS8 TS 18.8 15.8 -31.4±10.3 19.2±5.5 1.3±0.71

TS9 TS 21.8 19.8 -7.8±46.2 6.3±2.7 1.9±1.9

MM1 MM 21.4 18.4 -15.6±51.6 49.4±14.3 4.7±3.5

MM2 MM 21.2 19.0 -50.6±16.8 66.1±35.7 3.1±3.2

MM3 MM 21,8 19.8 -58.4±20.9 26.3±9.7 1.4±1.2

MM4 MM 21.8 19.8 -11.1±23.8 18.0±7.5 0.25±1.7

MM5 MM 19.6 17.3 -9.2±10.9 19.6±13.3 1.9±2.1

MM6 MM 19.1 16.7 -45.2±6.8 46.4±11.9 2.8±1.1

AL1 AL 20.2 17.1 -24.1±11.1 55.1±25.1 1.4±1.3

AL2 AL 22.6 20.2 -46.0±5.3 52.4±3.8 4.3±4.3

AL3 AL 20.2 17.1 -37.7±20.1 44.0±34.5 13.8±28.2

Table 3 The measured airtemperature (AT) and soil(5 cm) temperature (ST), aswell as mean fluxes of meth-ane (CH4, in μg C m-2 h-1),carbon dioxide (CO2, in mgC m-2 h-1) and nitrous oxide(N2O, in μg N m-2 h-1) forthe sampling sites

a Definitions of the land-use/cover codes are referred tothe Table 1and in the text.Flux data are given as mean ±standard deviation, calculatedfrom the measurements ofindividual chambers

Table 4 Mean fluxes of methane (CH4, in μg C m-2h-1), carbon dioxide (CO2, in mg C m-2h-1) and nitrous oxide (N2O, in μg N m-2h-1),as well as mean soil water-filled pore space (0-10 cm, %WFPS), soil organic carbon (SOC, in g kg-1) and total nitrogen (TN, in g kg-1) fordifferent soil types

Soil type Number of sites CH4 flux CO2 flux N2O flux WFPS SOC TN

Arenosols 11 -36.1±17.7a 32.1±16.0 2.2±1.5 12.9±17.4 14.7±17.1 1.3±1.4

Gleyosol 1 -28.0±25.8 95.5±37.3 21.8±30.3 57.1±2.8 44.7±19.4 4.8±1.9

Phaeozems 8 -27.8±14.1 48.1±31.1 3.2±1.6 23.3±19.1 33.4±14.7 3.1±1.2

Chernozems 10 -35.3±25.2 28.9±18.1 2.9±4.0 15.9±7.1 20.3±14.9 2.0±1.3

Total 30 -33.3±19.2 37.4±24.7 3.4±4.3 18.2±16.7 22.6±17.2 2.1±1.5

a Not including the data at the ML3 site due to its special high soil moisture under waterlogged conditions resulting in net CH4

emission. The data are given as mean ± standard deviation

Table 4 Mean fluxes of methane (CH4, in μg C m-2h-1),carbon dioxide (CO2, in mg C m-2h-1) and nitrous oxide (N2O,in μg N m-2h-1), as well as mean soil water-filled pore space

(0-10 cm, %WFPS), soil organic carbon (SOC, in g kg-1) andtotal nitrogen (TN, in g kg-1) for different soil types

Plant Soil

CO2 emissions

The CO2 emissions across all of the investigated sitesranged from 3.6 to 109.3 mg C m-2h-1, with a meanvalue of 37.4±24.7 mg C m-2h-1. The coefficient ofvariation (CV) was 66% (Table 3). The CO2 emissionof cores with the Gleysol soil type (95.5 mg C m-2h-1)was significantly higher (P<0.01) compared to thesoil cores with the soil type of Chernozems. Nosignificant difference between soil types was detectedfor all other cores (Table 4). Within each land-use/cover type class, CO2 emissions were considerablyvariable, with a CV ranging from 11% to 97%. Forthe typical steppe, the average CO2 emission was34.7±33.8 mg C m-2h-1. CO2 emissions from thearable land tended to be higher than the typical steppe(50.5 vs. 34.7 mg C m-2h-1). The highest CO2 emissionvalues were measured for cores from the marshlandsites (63.1±29.5 mg C m-2h-1), whereas lowest CO2

emission rates were observed for soil cores taken fromthe sand dune areas (25.4±14.5 mg C m-2h-1, Table 5,Fig. 2).

For all of the investigated sites as well as forindividual land-use/cover types, CO2 emissionswere positively correlated with soil moisture(Fig. 3a-c, Table 6). The variation in soil moistureat 0–10 cm depths could explain 28% of the spatialvariation in CO2 emissions across all land-use/covertypes (P<0.01). Other factors as measured in thisstudy were not significantly correlated with soil CO2

emissions.

N2O emissions

The N2O emissions across all of the investigated sitesranged from 0.18 to 21.8 μg N m-2 h-1, and averagedto 3.4±4.3 μg N m-2 h-1, with a CV of 130%(Table 3). N2O fluxes showed relatively greaterspatial heterogeneity than the CO2 fluxes. Some coresseemed to be hot spots of N2O emissions, withemissions up to 60 μg N m-2 h-1 (individual soil coredata not shown). A comparison of the mean N2Oemissions among soil types showed that significantlyhigher N2O emissions (P<0.01) were observed fromthe Gleysol site (21.8 μg N m-2 h-1) than from coresof other soil types; the latter did not significantlydiffer from each other (Table 4). Spatial variability ofN2O emissions within individual land-use/cover typeswas also high, with CVs ranging from 49% to 137%.The average N2O emission from a typical steppe (2.3±1.8 μg N m-2 h-1) tended to be lower than that of thecultivated land (6.5±6.5 μg N m-2 h-1). The N2Oemission (8.4 μg N m-2 h-1 on average) frommarshland sites tended to be larger than from a typicalsteppe. The N2O emissions from the sand dunes, baresoil and mountain meadow sites were comparable tothe typical steppe (Table 5, Fig. 2).

As shown in Fig. 3d-f and Table 6, both site meansof N2O emissions and land-use/cover type-specificmeans of the N2O emissions were significantlypositively correlated with soil moisture. Soil moisture(WFPS) explained 16–18% of the spatial variance inthe N2O emissions (P<0.05). Furthermore, the site

Table 5 Mean fluxes of methane (CH4, in μg C m-2h-1), carbon dioxide (CO2, in mg C m-2h-1) and nitrous oxide (N2O, in μg N m-2

h-1), as well as mean soil water-filled pore space (0-10 cm, %WFPS), soil organic carbon (SOC, in g kg-1) and total nitrogen (TN, ing kg-1) for different land-use/cover types

Land-use typea Number of sites CH4 flux CO2 flux N2O flux WFPS SOC TN

BS 3 -38.6±32.1 30.4±10.1 2.0±1.0 7.0±0.8 7.9±6.8 0.8±0.7

SD 6 -33.3±11.1 25.4±14.5 2.4±2.0 7.6±3.4 20.2±22.1 1.7±1.8

ML 3 912±1,620 63.1±29.5 8.4±11.5 62.3±4.4 28.4±19.4 2.8±2.0

TS 9 -33.8±24.0 34.7±33.8 2.3±1.8 13.9±9.4 18.1±8.5 1.8±0.7

MM 6 -31.7±22.1 37.6±19.3 2.4±1.5 15.1±3.3 41.4±15.8 3.7±1.3

AL 3 -35.9±11.0 50.5±5.8 6.5±6.5 25.3±0.9 12.0±6.2 1.3±0.6

Total 30 -33.3±19.2b 37.4±24.7 3.4±4.3 18.2±16.7 22.6±17.2 2.1±1.5

a Definitions of the land-use/cover codes are referred to the Table 1 and the textb Not including the data at the ML3 site

The data are given as mean±standard deviation

Table 5 Mean fluxes of methane (CH4, in μg C m-2h-1),carbon dioxide (CO2, in mg C m-2h-1) and nitrous oxide (N2O,in μg N m-2h-1), as well as mean soil water-filled pore space

(0-10 cm, %WFPS), soil organic carbon (SOC, in g kg-1) andtotal nitrogen (TN, in g kg-1) for different land-use/cover types

Plant Soil

means of N2O emissions were significantly positivelycorrelated with soil mineral N content (NH4

+ + NO3-)

(R2=0.19, P<0.05, Fig. 4a). A multiple linearregression model that combines soil mineral Ncontent and moisture yielded a higher R2 value(29%). In addition, we observed a significantlypositive relationship (P<0.01) between soil N2O andCO2 emissions over all of the investigated sites(Fig. 4b).

Discussion

Patterns, magnitudes and environmental controlsof GHG fluxes

The measured magnitude of N2O fluxes in our study(mean: 3.4±4.3 μg N m-2 h-1) is within the range ofN2O fluxes for semi-arid and arid zones (ranging from0.05 to 9 μg N m-2 h-1) as reported by Galbally et al.

0

5

10

15

N2O

em

issi

ons

(µg

N m

-2 h

-1)

Land use typeBS SD ML TS MM AL

0

30

60

90

120

CO

2 em

issi

ons

(mg

C m

-2 h

-1)

-50-25

0800

900

1000

CH

4 flu

xes

(µg

C m

-2h-1

)

Fig. 2 Mean fluxesof methane (CH4), carbondioxide (CO2) and nitrousoxide f (N2O) of each land-use/cover type in the XilinRiver catchment during theentire study period. Verticalbars are standard deviation.Definitions of the land-use/cover codes are referred tothe Table 1 and the text

0

40

80

120

160

CO

2 flux

es (

mg

C m

-2 h

-1) a 0-10 cm

Y=0.79X+23.2 (R2=0.28, P<0.01)

b 10-20 cm

Y=0.72X+25.9 (R2=0.15, P<0.05)

0

40

80

120

160

CO

2 flux

es (

mg

C m

-2 h

-1)c 20-30 cm

Y=0.57X+28.1 (R2=0.11, P<0.10)

0 20 40 60 80

0

10

20

30

N2O

flux

es (

µg N

m-2 h

-1) d 0-10 cm

Y=0.10X+1.5 (R2=0.16, P<0.05)

0 20 40 60 80

Soil moisture (%WFPS)

e 10-20 cm

Y=0.14X+1.1 (R2=0.18, P<0.05)

0 20 40 60 80

0

10

20

30

N2O

flux

es (

µg N

m-2 h

-1)f 20-30 cm

Y=0.12X+1.4 (R2=0.16, P<0.05)

Fig. 3 Correlation between soil moisture (water-filled pore space, WFPS) at different depths and (a–c) carbon dioxide (CO2)emissions, (d–f) nitrous oxide (N2O) emissions at 30 investigated sites in the Xilin River catchment

Plant Soil

(2008). Similarly, the average N2O emission of 2.3 μgN m-2 h-1 from a typical steppe was comparable toreported ranges of N2O fluxes for a typical steppe byHolst et al. (2007b) (generally, 1–3 μg N m-2 h-1) orXu-Ri et al. (2003) for five typical steppe sites in theXilin River area (1.6–4.4 μg N m-2 h-1) in InnerMongolia. Across all of the investigated sites, N2Oemissions were significantly positively correlated withsoil moisture and soil inorganic N (NH4

++NO3-)

concentration (Figs. 3d-f and 4a). This indicates thatN2O fluxes from semiarid ecosystems are mostlylimited by soil moisture and inorganic N content,which is in agreement with results from previousstudies (e.g., Mummey et al. 1994). Wang et al.(2005a) suggested that soil moisture is the primary

driving factor determining the spatial variability inN2O emissions from various semi-arid grassland types.Similar results were also reported by Corre et al.(1996) for landscape N2O fluxes in the Black SoilZone in Canada, and by Holst et al. (2007b) for alandscape transect study in the Xilin River catchment.In our study, only less than 20% of the variance in N2Oemissions could be explained by variations in WFPS orsoil inorganic N contents alone. Matson et al. (1991)examined the spatial variation in N2O fluxes from aWyoming shrub-steppe ecosystem and found a positiverelationship between N2O flux and soil nitrate concen-tration. However, also in this study the variation inN2O fluxes could not be explained on the basis ofvariation in soil N pools alone.

According to the thorough review of Wang andFang (2009), soil CO2 efflux ranged from 5.9 to114.6 mg C m-2 h-1 for temperate grasslands. Hence,the CO2 emissions during midsummer as determinedin this study (mean: 37.4 mg C m-2 h-1) are within therange of the published values. Likewise, the mean soilCO2 emission (34.7±33.8 mg C m-2 h-1) for corestaken from typical steppes was in agreement withfindings from previous studies conducted in thetypical steppe of Inner Mongolia (ranging from 5.9–44.5 mg C m-2 h-1; Li et al. 2000; Cui et al. 2000;Zhang et al. 2003; Wang et al. 2004; Qi et al. 2007;Zou et al. 2007). Temperature and precipitation areconsidered to be the most important environmentalfactors in determining spatial variations of soil CO2

emission (Xu and Qi 2001; Wang and Fang 2009). In

Table 6 Relationships across the six investigated land-use/cover types between soil moisture contents (M, in %WFPS) atdifferent depths and the fluxes of CO2 or N2O, as well asbetween N2O and CO2 fluxes

Regression function R2 P

CO2=26.5+0.63 M0–10 0.89 < 0.01

CO2=23.3+0.87 M10–20 0.96 < 0.01

CO2=23.9+0.80 M20–30 0.88 < 0.01

N2O=1.4+0.12 M0–10 0.85 < 0.01

N2O=0.62+0.17 M10–20 0.98 < 0.01

N2O=0.71+0.16 M20–30 0.91 < 0.01

N2O=-3.5+0.19 CO2 0.91 < 0.01

The subscripts for M indicate soil depth. P values indicatesignificant level

0 6 12 18

0

10

20

30

N2O

flux

es (

µg N

m-2 h

-1)

Soil mineral N (NH4

++NO3

-) (mg N kg-1SDW)

aY=0.58X+1.2 (R2=0.19, P<0.05)

0 40 80 120 160

0

10

20

30

N2O

flux

es (

µg N

m-2 h

-1)

CO2 fluxes (mg C m-2 h-1)

bY=0.092X-0.081 (R2=0.27, P<0.01)

Fig. 4 Correlation betweennitrous oxide (N2O) emis-sions and a soil mineral N(NH4

++NO3-), b carbon

dioxide (CO2) emissions at30 investigated sites in theXilin River catchment.SDW is soil dry weight

Plant Soil

our study, the significantly positive correlation be-tween WFPS and CO2 emissions (Fig. 3a-c), stronglysuggested that soil moisture is the major driver forplant and microbial activities and, thus, for the CO2

emission during the summer, which is in agreementwith a series of studies in other steppe and prairiesystems (Zak et al. 1993; Williams et al. 2000; Liu etal. 2009b). Surprisingly, no clear influence of soiltemperature on spatial variation of CO2 emissionscould be found in the present study. This may be dueto the fact that the range in temperature wascomparably low at high temperature levels (airtemperature: 18.8–26.9°C, soil temperature: 15.8–23.7°C) and that microbial activity under the envi-ronmental conditions of the sampling period of thisstudy is moisture rather than temperature-limited.

The magnitude and the direction of fluxes ofCH4 varied across the different sampling sites of theXilin River catchment. Only soils with a very highsoil water content were sources of CH4, i.e., the ML3site soils characterised by 55% WFPS on average; allothers were CH4 sinks. The mean CH4 flux for all ofthe sites in the Xilin River catchment (–33.3±19.2 μg C m-2 h-1, n=29) was in agreement withthe results of extensive studies in semi-arid and aridzones (ranging from –10.8 to –77 μg C m-2 h-1;Mosier et al. 1991, 1996, 1997; Striegl et al. 1992;Wang et al. 2005a, b; Liu et al. 2007, 2009a). Ourobserved CH4 flux of –33.8 μg C m-2 h-1 in thetypical steppe was also similar to that reported forgrazed sites (approximately –28 μg C m-2 h-1; Liu etal. 2007), and to those reported for a hill slopetransect in our study region (approximately –30 μgC m-2 h-1; Liu et al. 2009a). In our study, CH4 fluxesdid not display a pronounced dependency on soiltemperature and moisture. This was probably due tothe relatively low average soil moisture values(6.4%), which may have caused drought stress formethanotrophs (Mosier et al. 1996; Liu et al. 2007)at some sampling sites, e.g., the sand dune sites. Atother sites, such as the marshland sites, high valuesof soil moisture even led to net CH4 emissions, sothat moisture effects on CH4 fluxes cannot begeneralised across all sites. Similarly, Fang et al.(2009) reported that the spatial pattern of CH4

uptake in an evergreen broadleaved forest was notsignificantly influenced by soil moisture, thoughthese authors found that the spatial pattern of N2Oemissions followed that of soil moisture.

Effect of soil type on GHG fluxes

Even though our study is somewhat limited due to thenumber of sites and temporal coverage, it providessome clear indications that the magnitude of N2Oemissions was affected by the different soil types insteppe environments. Our study shows that N2Oemissions tend to increase with increasing SOC, TNand WFPS at 0–10 cm depths (Table 4). However,multiple regression analysis revealed that only WFPSwas significantly positively correlated with N2Oemission (R2=0.97, P<0.01). This suggested that soilmoisture content was the most important factorcontrolling N2O emissions. When soil moisture wasfavourable, N and C availability became important(Corre et al. 1996). The highest N2O emission wasobserved in the Gleysol site, which showed higherWFPS, SOC and TN. Our interpretation that soiltypes and soil properties are good indicators for themagnitude of N2O emissions is in agreement withearlier studies on N2O dynamics in grassland soils byMosier et al. (1996, 1997) and Velthof and Oenema(1995). These authors reported that the soils withhigher total C and N content typically emitted moreN2O than those grassland soils with lower total C andN contents. Similarly, Xu-Ri et al. (2003) noticed thatN2O fluxes from the typical steppe ecosystems inInner Mongolia generally decreased with decreasingSOC and TN contents.

In our study, the highest CO2 emissions wereobserved in the Gleysol site. This soil type exhibitedon average higher WFPS and SOC contents comparedto all other soil types (Table 4). The higher soilmoisture content could have stimulated activities ofplant roots and soil microorganisms (Schimel et al.1999; Qi et al. 2007), which further affected themetabolism of plant roots and the decomposition ofSOC. Furthermore, the higher soil moisture content caninfluence CO2 emission through stimulating plantgrowth and consequent belowground C allocation andC substrate (Zak et al. 1993; Liu et al. 2009b). Oneneeds to consider that, in steppe environments, wateravailability in general limits plant and microbialactivities. The Gleysol is formed under waterloggedconditions produced by rising groundwater and, thus,indicates that these sites have a higher water availabilityas compared to other sites in our study area. Accord-ingly, it is likely that the combined influences of WFPSand SOC alter soil CO2 emissions in the present study.

Plant Soil

The average CH4 fluxes stratified by the soil typeranged from –27.8 to –36.1 μg C m-2 h-1 μg C m-2 h-1.The rates of CH4 uptake for the Gleysol and Phaeozemswere slightly lower than those for Arenosols andChernozems (Table 4). It is likely that lower CH4

uptake rates in the Gleysol and Phaeozems sites resultfrom higher amounts of total nitrogen and higheraverage soil water contents. These two factors probablyresulted in higher N turnover rates, indicated by thegenerally higher N2O emissions from the Gleysol andPhaeozem sites (Table 4). Our present results mightindirectly confirm the observations of earlier studiesthat CH4 uptake rates are negatively correlated with Nturnover for grassland soils (Mosier et al. 1991, 1996).

Effect of land-use/cover on GHG fluxes

Similar to the procedure used to identify the effects ofsoil types on soil GHG exchange, we regrouped ourdata to identify possible land-use/ cover effects. Also,for different land-use/cover types, we observed asignificant correlation between N2O emissions andWFPS (Table 6). As shown in other studies performedin semi-arid ecosystems, soil water content is theprimary factor controlling the dynamics of N cycling(Matson et al. 1991; Mummey et al. 1994), and, thus,soil N2O fluxes. The lowest N2O emissions wereobserved for bare soil sites, which also have thelowest values of WFPS, SOC and TN. This findingwas in agreement with our measurements of soil CO2

fluxes, which were low for soil cores taken from baresoil sites as well, indicating low microbial activity(Table 5). In contrast, the highest N2O emissions werefound for the marshland sites with higher WFPS,SOC and TN. Compared to the typical steppe,cultivated arable land emitted more N2O even thoughthose sites were depleted in SOC and TN (6.5 vs.2.3 μg N m-2 h-1). This was most likely due to higherWFPS (Table 5), with soil moisture values beingpositively affected by agricultural practices (Mosier etal. 1997; Wang et al. 2001). Here, one needs toconsider that cropping is done in such a way that ayear with a field crop is followed by a fallow periodof one year in order to increase soil moisture. Thestimulating effect of agriculture on N2O emissionsseems to be transient, since the study by Zhang andHan (2008) for a comparable region in InnerMongolia shows that N2O emissions in abandonedcropland returned to a level similar to a typical steppe

within a five-year period, after agricultural practiceshad ceased. It is noteworthy that N2O emissions frommountain meadow and sand dune sites were compa-rable to the typical steppe. In the sand dune sites, thiscan probably be explained by comparable SOC andTN contents. In the mountain meadow sites, SOC andTN were higher, but the soil pH was significantlylower compared to the typical steppe. Obviously, theobserved similar N2O emissions from the mountainmeadow sites can hardly be explained by either the Cor the N pool size or by WFPS. Studies have shownthat nitrification is a significant, if not, even thedominating source of N2O production in semi-aridecosystems (Parton et al. 1988; Mummey et al. 1994;Xu-Ri et al. 2003, Du et al. 2006; Zhang and Han2008). If denitrification is the main source, N2Oemissions tend to decrease with increasing pH, at leastin acid soils (pH below 5-6), but where nitrification isthe main source, emissions of N2O tend to increasewith increasing pH at least in the range of pH 6-8(Granli and Bockman 1994). Yamulki et al. (1997)also reported for grassland soils in the UK that N2Ofluxes decreased appreciably with increasing acidity(pH ranging from 3.9 to 7.6). We need to speculatethat it is the offset of these positive (SOC and TN)and negative (pH) effects that led to similar N2Oemissions between mountain meadows and typicalsteppes. On the other hand, one needs to note that wewere only able to capture snapshots of the emissionsituation and that repeated sampling processes wouldbe necessary (even though not visible due to theenormous labour demand) to underline further ourfindings.

In comparison to typical steppe, CO2 emissions inthe arable land were slightly higher, indicating thatsoil C pools are mobilised due to agricultural activity.This observation is in agreement with the study ofBuyanovsky et al. (1987) who reported that soil CO2

efflux from a wheat field was greater than that fromthe native grassland in Missouri, USA. Typicalagricultural practices such as ploughing destroy soilaggregates, improve soil aeration and expose nativesoil organic matter to microbial decomposition (Six etal. 2002), thereby enhancing CO2 emission. However,some other studies showed that the conversion ofgrassland to cropland resulted in decreased CO2

emission rates after some years, since SOC poolswere depleted (Frank et al. 2006; Qi et al. 2007).Also, in our study, the soil cores taken from arable

Plant Soil

land had lower SOC contents as compared to corestaken from the typical steppe (Table 5). However,since WFPS was higher for cores from the arable sitesas compared to cores from the steppe sites, thestimulating effect of increased soil moisture may haveoverridden the effect of lower SOC contents on CO2

emission. This interpretation is supported by thesignificantly positive relationships between CO2

emissions and WFPS (Table 6). Maestre and Cortina(2003) suggested that the differences in soil moisturemay be more relevant to explain land cover differ-ences in soil CO2 efflux than soil organic matter androot density. Conant et al. (2000) also reported thatsoil moisture was the main factor influencing soilCO2 efflux in three semi-arid ecosystems in Arizona.Compared to the typical steppe, the higher CO2

emissions in marshland sites may be due to thehighest WFPS and relatively higher SOC contents(Table 5). In contrast, lower CO2 emissions in baresoil and sand dune sites can be interpreted as aconsequence of lower soil moisture, lower SOCcontents and lower plant cover, which negativelyaffect plant litter production and root exudation, andthereby decrease carbon turnover rates and CO2

emission (Ishizuka et al. 2002).Net CH4 emission was only observed from the ML

sites with higher soil moisture. All of the other land-use/cover types showed a net consumption of atmo-spheric CH4 (see Fig. 2). CH4 fluxes for arable soilswere not different from those of the typical steppe,although WFPS in arable land sites was relativelyhigher. A comparable result, i.e., that CH4 uptake insteppe regions is not significantly affected by agri-cultural practice, has also been reported by Wang etal. (2001). This is somewhat surprising, since e.g.,Mosier et al. (1997) found that the conversion ofNorth American shortgrass steppe to croplandstypically decreased soil CH4 uptake due to thenegative effects of ploughing and fertilization onmethanotrophic activity. Since in our study regionagricultural practice is rather extensive in terms offertilization (low doses if at all) or ploughing (onceevery two years), we assume that such low activitiesprevent agricultural soils in our study region fromlosing their CH4 uptake activity (Schimel andGulledge 1998). The bare soil, sand dune andmountain meadow sites showed comparable rates ofCH4 uptake (ranging from 31.7 to 38.6 μg C m-2 h-1)compared to typical steppes, although soil moisture

values in these sites have a wide range from 7% to25% WFPS. It is possible that differences in soilmoisture among sites were not large enough to resultin a significant difference in CH4 uptake intensity.Wang et al. (2005a) also suggested that the usualdifference in soil moisture among various semi-aridgrasslands does not lead to a significant difference inCH4 uptake rates. Alternatively, it is also likely thatthe substantial spatial variations in CH4 fluxes mayoverride any land-use/cover influences.

Snapshot-estimation of site and regional GHG fluxes

On the basis of our soil cores study, we calculatedregional soil GHG fluxes according to land-use/covertypes (Tables 5 and 7). In view of the high temporalvariability of GHG fluxes (e.g., Du et al. 2006; Liu etal. 2007; Qi et al. 2007), this estimate does notrepresent annual GHG fluxes from the Xilin Rivercatchment but is a snapshot-estimation of the regionalGHG budget and the relative contribution of theindividual GHG at the time of sampling. Furthermore,our data allow very roughly evaluating the relativecontribution of individual land-use/cover type to theGHG budget of the entire Xilin River catchment. Theestimated N2O emission in the investigated upperXilin River catchment (3,900 km2) averaged 223±178 kg N d-1, with the contribution of individual land-use/cover types ranging from 1.1% to 81.6%. Simi-larly, the amount of calculated CO2 emission for theupper Xilin River catchment was 3,223±2,897 t C d-1,with contribution of individual land use/cover typesranging from 1.0% to 85.6%. The dominant compo-nent of N2O and CO2 emissions in the Xilin Rivercatchment was the typical steppe due to its largeproportion in the catchment area (approximately85%). With regard to regional CH4 fluxes, weconsidered the specific role of marshland, whichcovers only about 0.4% of the Xilin River catchment,as a net source for CH4. The estimated regional CH4

uptake over marshland and other land-use/cover typesin midsummer was 2,808±2,758 kg C d-1. The CH4

emission from the marshland was about 11% of theCH4 uptake by the soils of other land-use/cover types,which covered 99.5% of the catchment area (Table 7).Similarly, Wang et al. (2005b) reported that the smallriparian wetland was an important CH4 emissionsource in the Xilin River catchment. However, Wanget al. (2005b) estimated that CH4 emission from

Plant Soil

wetlands (10.1 Mg CH4 d-1) may offset more than50% of CH4 uptake by upland areas. The higher CH4

emissions by Wang et al. (2005b) obviously resultedfrom only wet marshland (waterlogged or swampyparts), whereas the average CH4 flux of this studyintegrated wet and dry marshlands.

To evaluate the impact of individual land-use/cover types on the GHG budget of the Xilin Rivercatchment in Inner Mongolia in a better way, weused the global warming potential (GWP) methodto compare the contribution of each land-use/covertypes to the GHG budget. We expressed the GWPof CH4 and N2O as CO2-C-eq (Table 7; IPCC 2007).The marshland has the highest contribution to globalwarming if only CH4 and N2O are considered.However, despite its hot-spot character of area-related fluxes, the contribution of marshland to theregional GHG budget remains the lowest among allland-use/cover types and plays of minor importance(Table 7). In addition, the average fluxes for thisregion in midsummer, weighted by the areal extentof the land-use/cover types, are –30.0±29.5 μgC m-2 h-1 for CH4, 2.4±1.9 μg N m-2 h-1 for N2Oand 34.5±30.9 mg C m-2 h-1 for CO2, which also fallwithin the range of GHG estimates for other semi-arid and arid climate zones worldwide (see discussedabove).

Nevertheless, our present snapshot-estimation wasquite preliminary. We extrapolated limited site resultsobtained from soil core measurements of relativelyshort periods to a regional scale primarily based onland-use/cover type. Many important environmentalfactors, such as soil water availability, were notseparately considered. Meanwhile, this estimationdid not reflect the impacts of temporal variation andlandscape (e.g., topography and soil texture) on GHGfluxes from semi-arid grassland, as revealed by long-term field studies (Corre et al. 1996; Mosier et al.1996). Some researchers reported empirical methodsfor estimating N2O emissions from temperaturegrassland with the consideration of backgroundemission and N2O loss from the data of the total Nin N-fertilizer and animal (e.g., Bouwman et al.1995). However, these methods are not suitable forestimating of N2O emissions from normally unfertil-ized grasslands of China, due to their simplifiedestimation of the background emissions as constants.Site simulations of N2O emissions from temperategrassland by general process-based models were alsoreported, but two of the four test process models fitwell with the field-measured original data (Frolking etal. 1998). Accordingly, a more holistic view ofregional fluxes of the Xilin River catchment may bereached by full-factorial parameterization of the GHG

Table 7 Estimated regional daily soil fluxes of CH4, N2O and CO2 for individual land-use/cover types of the upper Xilin Rivercatchment in Inner Mongolia for mid-summer 2007

Land-use typea Percent of areab (%) CH4 fluxes(kg C d-1)

GWPc

(t C-eq d-1)N2O fluxes(kg N d-1)

GWPc

(t C-eq d-1)CO2 fluxes(t C d-1)

BS 2.1 -76.8±62.4 -0.70 4.8±2.4 0.61 60.0±19.2

SD 6.5 -202±67 -1.8 14.4±12.0 1.8 154±89

ML 0.4 341±607 3.1 2.4±4.8 0.31 24.0±12.0

TS 85 -2,688±1,913 -24.4 182±142 23.2 2,760±2,688

MM 4.5 -134±94 -1.2 9.6±7.2 1.2 158±82

AL 1.4 -48.0±14.4 -0.44 9.6±9.6 1.2 67.2±7.2

Total -2,808±2,758 -25.5 223±178 28.5 3,223±2,897

Area weighted CH4 flux -30.0±29.5μg C m-2h-1

Area weighted N2O flux 2.4±1.9 μg N m-2 h-1

Area weighted CO2 flux 34.5±30.9 mg C m-2 h-1

a Definitions of the land-use/cover codes are referred to the Table 1 and the textb The percentage of each land-use/cover type in the Xilin River catchment area was compiled from the sources of Wang et al. (2001)and Liu (2007). The total catchment land area is 3,900 km2 (Holst et al. 2007b)c Global warming potential (GWP) was calculated on the basis of mass factors of 25 for CH4 and 298 for N2O of 100-year timehorizon (IPCC 2007). The data are given as mean ± standard deviation

Plant Soil

fluxes of soil cores of the selected sites based onchanging temperature and soil moisture and subse-quent climate-data-driven biogeochemical modelling.

Conclusions

Against our expectations the spatial variability ofgreenhouse gas (GHG) exchange at the regional scalewas rather low and often even lower than that forcores taken at one sampling site. However, we founda Gleysol site emitted more N2O and CO2 than othersoil types, though not affecting significantly theregional midsummer GHG budget. Surprisingly,cultivated agricultural soils only tended to emit moreN2O and were not characterised by a decrease in CH4

sink strength compared to other land-use/cover types,e.g., typical steppes. Low intensity agricultural prac-tices, with low fertilization rates and ploughing in twoyear intervals, is most likely the main reason for thisobservation, which is in contrast to most studieswhich investigated changes in CH4 uptake followingthe conversion of native grasslands into croplands.The predominant factor affecting the regional vari-ability of GHG fluxes in our study was a difference inthe soil moisture, with the soil moisture beingobviously affected by land-use (e.g., arable versussteppe soils), soil properties (e.g., light texture sanddune soils versus steppe soils) or soil types (Gleysolversus Chernozems). The calculation of a regionalGHG estimate, in which the importance of differentland-use/cover types was considered, revealed that themidsummer net GHG balance was strongly dominat-ed by CO2 exchange, with N2O fluxes or CH4

exchanges playing only a marginal role even ifmarshlands are considered as net sources for CH4.The results from this study, however, may notnecessarily represent the situation of biosphere-atmosphere exchange of GHGs in the Xilin Rivercatchment, as a result of the limited number ofsampling sites and temporal coverage. Therefore,future long-term field monitoring is needed to confirmwhether our findings represent a short-term responseof the catchment, or long-term changes in C- andN-cycling.

Acknowledgements This study was financially supported bythe National Natural Science Foundation of China (40805061,40425010), the German Research Foundation (DFG) (Research

Group MAGIM, FG 536) and the Helmholtz-CAS jointlaboratory project ENTRANCE. We thank Ms. Zhihong Yufor her assistance in gas sampling and analysis. Thanks are alsogiven to the Inner Mongolia Grassland Ecosystem ResearchStation for providing meteorological data.

References

Bouwman AF, Vanderhoek KW, Olivier JGJ (1995) Uncertain-ties in the global source distribution of nitrous oxide. JGeophys Res 100:2785–2800

Brus DJ, de Gruijter JJ (1997) Random sampling or geo-statistical modelling? Choosing between design-based andmodel-based sampling strategies for soil (with discussion).Geoderma 80:1–44

Buyanovsky GA, Kucera CL, Wagner GH (1987) Comparativeanalysis of carbon dynamics in native and cultivatedecosystems. Ecology 68:2023–2031

Chapius-Lardy L, Wrage N, Metay A, Chotte J, Bernoux M(2007) Soils, a sink for N2O? A review. Global ChangeBiol 13:1–17

Chen G, Huang B, Xu H, Zhang Y, Huang G, Yu K, Hou A, DuR, Han S, VanCleemput O (2000) Nitrous oxide emissionsfrom terrestrial ecosystems in China. Chemosphere-GlobalChange Sci 2:373–378

Conant RT, Klopatek JM, Klopatek CC (2000) Environmentalfactors controlling soil respiration in three semiaridecosystems. Soil Sci Soc Am J 64:383–390

Corre MD, van Kessel C, Pennock DJ (1996) Landscape andseasonal patterns of nitrous oxide emissions in a semiaridregion. Soil Sci Soc Am J 60:1806–1815

Cui XY, Chen SQ, Chen ZZ (2000) CO2 release from typicalstipa grandis grassland soil. Chin J Appl Ecol 11:390–394

Denmead OT, Ereney JR, Simpson JR (1979) Studies of nitrousoxide emission from a grass sward. Soil Sci Soc Am J43:726–728

DuR, Lu DR,WangGC (2006) Diurnal, seasonal and inter-annualvariations of N2O fluxes from native semi-arid grasslandsoils of Inner Mongolia. Soil Biol Biochem 38:3474–3482

Fang Y, Gundersen P, Zhang W, Zhou G, Christiansen JR, MoJ, Dong S, Zhang T (2009) Soil-atmosphere exchange ofN2O, CO2 and CH4 along a slope of an evergreen broad-leaved forest in southern China. Plant Soil 319:37–48

Frank AB, Liebig MA, Tanaka DL (2006) Management effectson soil CO2 efflux in northern semiarid grassland andcropland. Soil Till Res 89:78–85

Frolking SE, Mosier AR, Ojima DS, Li C, Parton WJ, PotterCS, Priesack E, Stenger R, Haberbosch C, Dorsch P,Flessa H, Smith KA (1998) Comparison of N2O emissionsfrom soils at three temperate agricultural sites: simulationsof year-round measurements by four models. Nutr CyclAgroecosyst 52:77–105

Galbally IE, Kirstine WV, Meyer CP, Wang YP (2008) Soil-atmosphere trace gas exchange in semiarid and arid zones.J Environ Qual 37:599–607

Glatzel S, Stahr K (2001) Methane and nitrous oxide exchangein differently fertilized grassland in southern Germany.Plant Soil 231:21–35

Plant Soil

Granli T, Bockman OC (1994) Nitrous oxide from agriculture.Norweg J Agr Sci. Norsk Hydro, Oslo, pp 57–58,Supplement No. 12

Holst J, Liu CY, Brüggemann N, Butterbach-Bahl K, ZhengXH, Wang YS, Han SH, Yao ZS, Yue J, Han XG (2007a)Microbial N turnover and N-Oxide (N2O/NO/NO2) fluxesin semi-arid grassland of Inner Mongolia. Ecosystems10:623–634

Holst J, Liu CY, Yao ZS, Brüggemann N, Zheng XH, Han XG,Butterbach-Bahl K (2007b) Importance of point sourceson regional nitrous oxide fluxes in semi-arid steppe ofInner Mongolia, China. Plant Soil 296:209–226

IPCC (Intergovernmental Panel onClimate Change) (2007) ClimateChange 2007: The physical science basis, contribution ofworking group I to the fourth assessment report of theintergovernmental panel on climate change Solomon S,Qin D, Manning M, Chen Z, Marquis M, Averyt K B,Tignor M, and Miller H L (eds.), Cambridge UniversityPress, Cambridge, United Kingdom and New York, NY,USA

Ishizuka S, Tsuruta H, Murdiyarso D (2002) An intensive fieldstudy on CO2, CH4 and N2O emissions and soil properties atfour land-use types in Sumatra, Indonesia. Global Biogeo-chem Cycles 16:1049. doi:10.1029/2001GB001614

Keller M, Reiners WA (1994) Soil-atmosphere exchange ofnitrous oxide, nitric oxide, and methane under secondarysuccession of pasture to forest in the Atlantic lowlands ofCosta Rica. Global Biogeochem Cycles 8:399–409

Li LH, Wang QB, Bai YF, Zhou GS, Xing XR (2000) Soilrespiration of a Leymus chinensis grassland stand in theXilin River Basin as affected by over-grazing and climate.Acta Phytoecological Sinica 24:680–686

Lilly A, Ball BC, McTaggart IP, DeGroote J (2009) Spatialmodeling of nitrous oxide emissions at the national scaleusing soil, climate and land use information. GlobalChange Biol 15:2321–2332

Liu C (2007) The study of biosphere-atmosphere exchange ofCH4, N2O and NOx in a semi-arid steppe in InnerMongolia, China. Ph. D. Thesis, Institute of AtmosphericPhysics, Chinese Academy of Sciences, Beijing, China

Liu C, Holst J, Bruggemann N, Butterbach-Bahl K, Yao Z, Yue J,Han S, Han X, Krummelbeind J, Hornd R, Zheng X (2007)Grazing reduces methane uptake by soils in a semi-arid steppein Inner Mongolia, China. Atmos Environ 41:5948–5958

Liu C, Holst J, Yao Z, Bruggemann N, Butterbach-Bahl K, HanS, Han X, Tas B, Susenbeth A, Zheng X (2009a) Growingseason methane budget of an Inner Mongolian steppe.Atmos Environ 43:3086–3095

Liu WX, Zhang Z, Wan SQ (2009b) Predominant role of waterin regulating soil and microbial respiration and theirresponses to climate change in a semiarid grassland.Global Change Biol 15:184–195

Maestre FT, Cortina J (2003) Small-scale spatial variation insoil CO2 efflux in a Mediterranean semiarid steppe. ApplSoil Ecol 23:199–209

Matson PA, Vitousek PM, Livingston GP, Swanberg NA (1990)Sources of variations in nitrous oxide flux from Amazonianecosystems. J Geophys Res 95:16789–16798

Matson PA, Volkmann C, Coppinger K, Reiners WA (1991)Annual nitrous oxide flux and soil nitrogen characteristicsin sagebrush steppe ecosystems. Biogeochemistry 14:1–12

Minasny B, McBratney AB (2007) Incorporating taxonomicdistance into spatial prediction and digital mapping of soilclasses. Geoderma 142:285–293

Mosier A, Schimel D, Valentine D, Bronson K, Parton W(1991) Methane and nitrous oxide fluxes in native,fertilized and cultivated grasslands. Nature 350:330–332

Mosier AR, Parton WJ, Valentine DW, Ojima DS, Schimel DS,Delgada JA (1996) CH4 and N2O fluxes in the Coloradoshortgrass steppe 1. Impact of landscape and nitrogenaddition. Global Biogeochem Cycles 10:387–399

Mosier AR, Parton WJ, Valentine DW, Ojima DS, Schimel DS,Heinemeyer O (1997) CH4 and N2O fluxes in theColorado shortgrass steppe 2. Long-term impact of landuse change. Global Biogeochem Cycles 11:29–42

Mummey DL, Smith JL, Bolton H Jr (1994) Nitrous oxide fluxfrom a shrub-steppe ecosystems: sources and regulation.Soil Biol Biochem 26:29–286

Parton WJ, Mosier AR, Schimel DS (1988) Rates andPathways of nitrous oxide production in a shortgrasssteppe. Biogeochemistry 6:45–58

Qi YC, Dong YS, Liu JY, Domroes M, Geng YB, Liu LX, LiuXR, Yang XH (2007) Effect of the conversion of grasslandto spring wheat field on the CO2 emission characteristicsin Inner Mongolia, China. Soil Till Res 94:310–320

Schimel JP, Gulledge J (1998) Microbial community structureand global trace gases. Global Change Biol 4:745–758

Schimel JP, Gulledge JM, Clein-Curley JS, Lindstrom JE,Braddock JF (1999) Moisture effects on microbial activityand community structure in decomposing birch litter in theAlaskan taiga. Soil Biol Biochem 31:831–838

Six J, Conant RT, Paul EA, Paustian K (2002) Stabilizationmechanisms of soil organic matter: implications for C-saturation of soils. Plant Soil 241:155–176

Skiba UM, Sheppard LJ, Macdonald J, Fowler D (1998) Somekey environmental variables controlling nitrous oxideemissions from agricultural and semi-natural soils inScotland. Atmos Environ 32:3311–3320

Smith KA, Ball T, Conen F, Dobbie KE, Massheder J, Rey A(2003) Exchange of greenhouse gases between soil andatmosphere: interactions of soil physical factors and biolog-ical processes. Eur J Soil Sci 54:779–791

Sozanska M, Skiba U, Metcalfe S (2002) Developing aninventory of N2O emissions from British soils. AtmosEnvion 36:987–998

Steffens M, Kölbl A, Giese M, Hoffmann C, Totsche KU, BreuerL, Kögel-Knabner I (2009) Spatial variability of topsoils andvegetation in a grazed steppe ecosystem in Inner Mongolia(PR China). J Plant Nutr Soil Sci 172:78–90

Striegl RG, McConnaughey TA, Thorstenson OC, Weeks EP,Woodward JC (1992) Consumption of atmospheric methaneby desert soils. Nature 357:145–147

Turner DA, Chen D, Galbally IE, Leuning R, Edis RB, Li Y,Kelly K, Phillips F (2008) Spatial variability of nitrousoxide emissions from an Australian irrigated dairy pasture.Plant Soil 309:77–88

Velthof GL, Oenema O (1995) Nitrous oxide fluxes fromgrassland in Netherlands: II. Effects of soil type, nitrogenfertilizer application and grazing. Eur J Soil Sci 46:531–540

Wang YS, Wang YH (2003) Quick measurement of CH4, CO2

and N2O emissions from a short-plant ecosystem. AdvAtmos Sci 20:842–844

Plant Soil

Wang W, Fang J (2009) Soil respiration and human effects onglobal grasslands. Global Planet Change 67:20–28

Wang Y, Ji B, Huang Y, Hu Y, Wang Y (2001) Effects of grazingand cultivating on emission of nitrous oxide, carbon dioxideand uptake of methane from Grasslands. Environ Sci 22:7–13

Wang GC, Du R, Kong QX, Lu DR (2004) Experimental studyon soil respiration in temperate grassland in China.Chinese Science Bulletin 49:642–646

Wang Y, Xue M, Zheng X, Ji B, Du R, Wang Y (2005a) Effectsof environmental factors on N2O emission from and CH4

uptake by the typical grasslands in the Inner Mongolia.Chemosphere 58:205–215

Wang ZP, Han XG, Li LH, Chen QS, Duan Y, Cheng WX(2005b) Methane emission from small wetlands andimplications for semiarid region budgets. J Geophys Res110, doi:10.1029/2004JD005548

Wang ZP, Song Y, Gulledge J, Yu Q, Liu HS, Han XG (2009)China’s grazed temperate grasslands are a net source ofatmospheric methane. Atmos Environ 43:2148–2153

Williams MA, Rice CW, Owensby CE (2000) Carbon dynamicsand microbial activity in tallgrass prairie exposed toelevated CO2 for 8 years. Plant Soil 227:127–137

Wilson J, Gallant J (eds) (2000) Terrain Analysis: Principlesand Applications. Wiley, New York

Xu M, Qi Y (2001) Spatial and seasonal variations of Q10

determined by soil respiration measurements at a SierraNevadan forest. Global Biogeochem Cycles 15:687–696

Xu-Ri WYS, Zheng XH, Ji BM, Wang MX (2003) Acomparison between measured and modeled N2O emis-sions from Inner Mongolia semi-arid grassland. Plant Soil255:513–528

Yamulki S, Harrison RM, Goulding KWT, Webster CP (1997)N2O, NO and NO2 fluxes from a grassland: effect of soilpH. Soil Biol Biochem 29:1199–1208

Zak DR, Pregitzer KR, Curtis PS, Teeri JA, Fogel R, RandlettDL (1993) Elevated CO2 and feedback between carbonand nitrogen cycles. Plant Soil 151:105–107

Zhang JF, Han XG (2008) N2O emission from the semi-aridecosystem under mineral fertilizer (urea and superphos-phate) and increased precipitation in northern China.Atmos Environ 42:291–302

Zhang Y, Li LH, Wang YF, Tang F, Chen QS, Yang J, YuanYZ, Dong YS (2003) Comparison of soil respiration intwo grass-dominated communities in the Xilin RiverBasin: correlations and controls. Acta Botanica Sinica45:1024–1029

Zheng X, Mei B, Wang YH, Xie B, Wang YS, Dong H, Xu H,Chen G, Cai Z, Yue J, Gu J, Su F, Zou J, Zhu J (2008)Quantification of N2O fluxes from soil-plant systems maybe biased by the applied gas chromatograph methodology.Plant Soil 311:211–234

Zou CJ, Wang KY, Wang TH, Xu WD (2007) Overgrazing andsoil carbon dynamics in eastern Inner Mongolia of China.Ecol Res 22:135–142

Plant Soil