Modeling interplay between regional net ecosystem carbon balance

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, G04005, doi:10.1029/2007JG000455, 2007

for

Full Article

Modeling interplay between regional net ecosystem carbon balance and soil erosion for a crop-pasture region Qiong Gao,1 Mei Yu,2 Yinghui Liu,1 Hongmei Xu,3 and Xia Xu1 Received 25 March 2007; revised 2 July 2007; accepted 27 July 2007; published 18 October 2007.

[1] The balance between erosion-induced soil carbon loss and the reduction in

heterotrophic respiration caused by carbon removal in semiarid ecosystems that suffer from severe soil erosion is still largely uncertain. In this paper, we revised and applied a simulation model to analyze responses of ecosystem processes in the crop-pasture belt region of northern China to impacts of soil erosion and climate shift. The revised model includes a new module that calculates runoff-induced soil erosion and soil carbon and nutrient losses. The model was validated against long-term field observations on plant productivity at several sites, and sediment yields of experiments with various vegetation covers and slopes. Simulation with historical climate data without considering erosion showed that the average net primary productivity (NPP), heterotrophic respiration (RHE), agricultural harvest (HAV), and net ecosystem carbon balance (NECB) were 210.1 ± 26.9, 169.8 ± 7.7, 35.9 ± 4.1, and 4.4 ± 22.5 gC m
2 a
1, respectively. In contrast, simulation with soil erosion gave an average erosion-induced loss of soil organic carbon (ECL) of 11.0 ± 2.8 gC m
2 a
1, and decreased average NPP, RHE, and HAV by 3.3 ± 0.7, 14.5 ± 0.3, and 0.2 ± 0.0 gC m
2 a
1, respectively. Given NPP maintained by crop fertilization and irrigation for crop fields, the erosion-induced soil carbon loss is thus counterbalanced by the decrease in heterotrophic respiration, resulting in an invariant NECB with respect to soil erosion. Citation: Gao, Q., M. Yu, Y. Liu, H. Xu, and X. Xu (2007), Modeling interplay between regional net ecosystem carbon balance and soil erosion for a crop-pasture region, J. Geophys. Res., 112, G04005, doi:10.1029/2007JG000455.

1. Introduction [2] Long-term regional carbon balance subjected to heavy anthropogenic disturbances has been a concern and challenge for both global change research and regional ecosystem management because of the difficulties in quantification of interplay between climatic changes and managerial disturbance related to agricultural production [Chapin et al., 2002; Murty et al., 2002; Brody, 2003; Ludwig et al., 2005]. In particular, great uncertainties exist about quantitative impacts of large-scale erosion-induced loss in soil organic carbon (SOC) on net ecosystem carbon balance, as reported in a large volume of literature [e.g., Bajracharya et al., 2000; Lal, 2000, 2002b; Houghton, 2003b; Jacinthe et al., 2004; Lal, 2004]. Soil erosion has been known to directly reduce the soil carbon pool by transferring carbon laterally to aquatic systems, by deposing carbon in small lowland areas, and by increasing carbon oxidation and soil efflux [Smith et al., 2005]. Erosion has also been shown to decrease soil nutrient content, degrade soil structure, and 1 Integrated Landscape Analysis and Modeling Group, State Key Lab of Earth Surface Processes and Resources Ecology, College of Resources Science and Technology, Beijing Normal University, Beijing, China. 2 Institute of Botany, Chinese Academy of Sciences, China. 3 National Climate Research Center of China, China.

Copyright 2007 by the American Geophysical Union. 0148-0227/07/2007JG000455$09.00

cause a reduction in net primary production [Tan et al., 2005]. Heterotrophic respiration of regional ecosystems can also be significantly altered by severe soil erosion, because erosion tends to reduce the substrates of heterotrophic respiration, to change soil texture and structure, and thus to modify soil microbe activities [Parton et al., 1987; Murty et al., 2002]. The net effect of soil erosion on the regional carbon budget depends on many factors including climate and managerial activities [Lal, 2000, 2002a; Houghton, 2003a; Liu et al., 2003]. [3] Optimal land use planning and management necessitate comprehensive understanding of the coupling between dynamic soil carbon budget and other terrestrial ecosystem processes. Rational land use planning is especially important for China, due to its dense population and heavy pressure brought about by the demands for rapid economic development [Cao et al., 2003; Li et al., 2003; Gao et al., 2004b]. The issue is even more pressing for the crop-pasture transition regions in northern China, because such areas are in general more sensitive to natural and anthropogenic driving forces, thus are often frontiers for combating desertification. [4] The crop-pasture transition belt in northern China (CCPB) lies within the geographical ranges between 34.7 and 48.6°N and between 100.8 and 124.8°E, with a total area of 725,527.9 km
2 (Figure 1). The long axis of CCPB runs from the northeast verge of the Tibet Plateau, crossing the Loess Plateau, the Yellow River basin, the Mongolia

G04005

1 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

Figure 1. Position of China crop pasture belt (CCPB). Biomass sites are the sites with long term biomass or timber observations (4 – 27 years), Letter site codes are explained and used in Figure 3 for patch-scale model validation. Scaling sites denote the three landscapes (long-term research stations) in CCPB, for which we ran TESim-L (landscape model). The inset figures show the relief map of each sites: Zhifanggou from 1000 to 1425 m, Shihuimiao from 1342 to 1390 m, and Changling from 137 to 145 m. The elevation ranges indicate relative importance of runoff and severity of soil erosion at these sites. Plateau, and ends in the Northeast Plain. The region includes 205 counties/cities in 10 provinces, and has a population of approximately 60 million. Ecosystems in the region are under a typical continental monsoon climate with the precipitation gradient approximately perpendicular to the long axis. Annual precipitation decreases from approximately 580 mm in the Southeast to less than 200 mm in the Northwest. Annual mean temperature decreases from 14.0°C in the South to less than
1.0°C in the North (1959 – 2001). Vegetation patterns largely correlate with precipitation and temperature, with a small portion of coniferous forests in the cold North and mountains, deciduous broadleaf forests in the Southeast and North, large portions of shrubs and grasses on the northwest side of the dry plateaus, and croplands distributed mostly in the middle and the southeast regions. The area has been undergoing severe soil erosion, degradation, and desertification because of the changing climate, inappropriate land use, and overgrazing by livestock. Planning and implementation of ecosystem restoration programs have to be guided by the correct understanding of regional ecosystem processes including soil erosion and quantitative responses of these processes to changes in climate in the past and future. [5] Process-based ecosystem models have been developed and used to analyze ecosystem responses to climate and disturbances on regional and global scales. Biogeochemical models simulate carbon, water, and nutrient cycling within ecosystems [Raich et al., 1991; McGuire et al., 1992, 1993; Running, 1994; Schimel et al., 1997; Xiao et al., 1997; Peng and Apps, 1999; Cao and Li, 2000]. On the other hand, biogeography models determine ecosystem structure represented by vegetation distribution [Prentice et al., 1992; Neilson, 1995]. Development of dynamic

global vegetation models (DGVM) starting in the late 1990s couples ecosystem processes with ecosystem structure [Steffen et al., 1996; Beerling et al., 1997; Peng, 2000]. Models that explicitly take into consideration erosioninduced lateral soil carbon loss in analyses of regional ecosystem carbon balance are still rare, and soil erosion has been widely studied in agricultural rather than ecological sciences. Recent efforts to bridge ecosystem carbon cycle and soil erosion processes provide examples to analyze the impacts of lateral processes on spatially heterogeneous ecosystems on landscape and regional scales [Reid et al., 1999; West and Wali, 2002; Liu et al., 2003; Van Oost et al., 2005; Izaurralde et al., 2006]. [6] One difficulty in incorporating erosion into regional ecosystem models is that water erosion is one of the spatial processes that requires quantification of interactions among neighborhood ecosystems due to spatial heterogeneity in resource demands and ecosystem processes, which cause transfer of mass and energy across neighborhood ecosystems [Rupp et al., 2000; Weir et al., 2000; Loreau et al., 2003; Rastetter et al., 2003; Ludwig et al., 2005]. States of a local ecosystem are controlled both by processes within the system (e.g., vertical movement of soil water and nutrients, and plant growth) and by mass and energy flows across neighborhood ecosystems. Not only does a change in any local ecosystem in a heterogeneous regional mosaic affect the processes of the system itself, but also it may bring about a series of consequences in the neighborhood ecosystems including changes in hydrologic cycles, soil erosion, nutrient loss, and net primary productivity, by decreasing or increasing runoff/run-on flow, and by cutting off or connecting the pathway of spatial plant propagation [Band et al., 2001; Tague and Band, 2001; Tchir et al., 2004].

2 of 16

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

G04005

Figure 2. Diagram of TESim 2.0 model. Rectangular boxes are pools of carbon, nitrogen and water. Lines connecting the pools denote material flow passages. [7] Simulation models have been shown to be ubiquitously scale-dependent because of the nonlinearity of various ecosystem processes [Levin, 1992; Gao et al., 2001; Rastetter et al., 2003]. Models are usually parameterized with local scales of 101 –102 m at research sites, but might have been used on regional or even global scales without considering errors associated with scaling up. Research indicates that cross-scale ecosystem modeling has a high risk of scaling errors with magnitudes comparable to, or even larger than, the values of key ecosystem variables simulated on local scales [Rastetter et al., 2003]. [8] In this paper, we revised, scaled, and applied a regional ecosystem model to simulate net primary production and soil processes in the CCPB region with existing climate, soil, and vegetation data. The model connects local ecosystem processes of carbon assimilation, plant growth, and nutrient cycling, to lateral processes of erosion-induced soil organic carbon loss. The model was run to determine the responses of the CCPB to climatic changes from 1959 to 2001. Effects of soil erosion and climate shifts in the early 1980s on net ecosystem carbon balance (NECB) were analyzed, and the results discussed in the contexts of future global change and regional ecosystem management. We found that soil erosion did not cause significant decreases in regional net ecosystem carbon balance because of decreases in heterotrophic respiration associated with soil organic carbon removal by erosion and relatively unaltered NPP maintained by fertilization and irrigation of agricultural crop fields.

2. Model Description 2.1. Production, Decomposition, and Nutrient Use [9] The revised model (Figure 2) we used is the outgrowth of our terrestrial ecosystem simulator (TESim),

which has been evolving over the past 10 years [Gao and Zhang, 1997; Gao and Yu, 1998; Gao et al., 2000; Yu et al., 2002a]. The current version of the model includes vertical hydraulic redistribution by plant roots [Caldwell et al., 1998; Ryel et al., 2002], runoff and run-on absorption [Gao and Reynolds, 2003], and a new module of soil erosion [Gao et al., 2002a], in addition to simulation of carbon, water, and nitrogen flow involved in primary production. The well-known CENTURY model [Parton et al., 1987, 1988; Gobron et al., 1999] was adopted to simulate the decomposition processes. The model was implemented for patch, landscape, and regional scales with three standalone programs, TESim-P, TESim-L, and TESim-R, respectively. The main assumptions and treatments are described as follows: [10] Vegetation in a region is grouped into a number of functional types, and each functional type is partitioned into four pools of seeds, leaves, stems, and roots. Soil horizons are divided into a maximum of 8 layers with variable thickness, and total soil thickness is regarded as a spatial variable. The state variables of the model (spatially and temporally variable) include biomass and nitrogen concentration of the plant and litter pools, volumetric water contents of soil layers, mass and nitrogen contents of soil organic matter, and available soil nitrogen concentration. The model has a time step of one day. The spatial resolution, however, is allowed to vary with the extent of applications. TESim-P is a patch-scale model without spatial variation and is mainly used in model parameterization and validation at site scales; TESim-L is designed for application at landscape scales with areas less than or approximately 10 km
2, and is ideal for small watershed analyses. TESim-R, however, is intended for application at regional and continental scales with spatially variable at-

3 of 16

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

mospheric climates. Scaling up from site studies to regional applications may be achieved by controlling points in the region that were analyzed with TESim-L or TESim-P in fine scales [Rastetter et al., 2003]. [11] Carbon assimilation was modeled by using leaf models by Thornley and Johnson [1990] for C3 and C4 species with modification by Gao et al. [2004a], because the leaf models showed stronger coupling with stomatal conductance than the well-known biochemical models [Berry and Farquhar, 1978; Farquhar et al., 1980]. The sensitivity of carbon assimilation to stomatal conductance is especially important for arid and semiarid ecosystems [Reynolds et al., 1993, 1999a]. The net carbon assimilation (mmol m
2 s
1) for plants with C3 carbon pathway (all natural functional types), An3, is defined by the following equation An3 ¼

b3


pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b23 þ 4b3 c3 2a3

ð1Þ

   c4 ¼ gp Oa þ gt Ca 2Rd
ap Ip     gt Rd Pa Rd Pa þ ap Ip Rd þ gp Oa þ þ gt Rd Ca
gsc gsc   d4 ¼ Rd
ap Ip gt Rd Ca

ð8Þ

ð9Þ

where gt is the coefficient of transporting CO2 from mesophyll cells to bundle sheath, which is also assumed to have the same temperature and nitrogen dependence as gx for C3 species. The limitation of transporting capacity of assimilated products from leaf to other organs [Collatz et al., 1992] is not considered in the models from (1) through (9). [13] The stomatal conductance is modeled by Gao et al. [2002a, 2005]. The model calculates stomatal conductance (gs) as an increasing function of incidental light intensity, and a decreasing function of vapor pressure deficit (Vpd), leaf CO2 concentration, and soil water potential (y s). In particular,

in which   a3 ¼ Pa gso gx
gsc gp

G04005

bg þ gs ¼

ð2Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b2g þ 4ag cg 2ag

ð10Þ

in which,

  b3 ¼ gsc gso ap Ip þ gx Ca þ gp Oa
Rd a3   þ ap Ip Pa gsc gp þ gso gx

ag ¼ rz Dv þ b

ð3Þ

ð11Þ 

     c3 ¼ gsc gso Rd ap Ip þ gx Ca þ gp Oa þ ap Ip gp Oa
gx Ca

bg ¼ y s
ðb þ rz Dv Þkg þ p0 þ pp

ð4Þ

 Ip C0 Ca Ip þ kI

ð12Þ

where Pa is air pressure (kPa), Ip is light intensity (mmol m
2 s
1), Ca and Oa are CO2 and O2 pressures on plant leaf surface (kPa), respectively, gsc and gso are stomatal conductance for CO2 and O2 (mmol m
2 s
1), respectively, gx is carboxylation conductance (mmol m
2 s
1 kPa
1), gp is photorespiration conductance (mmol m
2 s
1 kPa
1), Rd is dark respiration coefficient (mmol m
2 s
1), ap is photon efficiency (mol mol
1 ). Parameters ap and g x are considered to depend on leaf nitrogen content and temperature in the way formulated by Medlyn et al. [1999]. Dark respiration Rd is an exponential function of temperature. [12] For crops (mainly maize) with a C4 carbon pathway, the net leaf assimilation, An4, is calculated by the following equation

where Dv is the relative vapor pressure deficit, i.e., leaf vapor pressure deficit (Vpd) divided by air pressure (Pa). Parameters of the model include half saturation stomatal conductance kg (mol m
2 s
1), half-saturation light intensity kI (mmol m
2 s
1), reference air CO2 concentration C0, dark osmotic pressure p0, maximum light-inducible osmotic pressure pp, elastic modulus of guard cell structure b(MPa m2 s mol
1), and soil-to-leaf resistance rz (MPa m2 s mol
1). The model gives a theoretically allowable minimum soil water potential (Ymin) that signifies the tolerance of plants to soil water stress

a4 A3n4 þ b4 A2n4 þ c4 An4 þ d4 ¼ 0

sensitivity of gs to changes in vapor pressure deficit at Ca = C0

ð5Þ

cg ¼ kg ðy s þ p0 Þ

  Ymin ¼
p0 þ pp

in which,  a4 ¼ 2Pa

gp gt
gso gsc

SV ¼



  gs gs þ kg rz @gs  ¼
 @Vpd 2ag gs þ bg Pa

ð13Þ

ð14Þ

ð15Þ

ð6Þ

and sensitivity of gs to changes in soil water potential   a4   b4 ¼ ap Ip þ 2 gp Oa þ gt Ca þ 2Rd
ap Ip 2   ap Ip gp gt Rd
þ Pa gso gsc

Sy ¼ ð7Þ

4 of 16

gs þ kg @gs ¼ @y s 2ag gs þ bg

ð16Þ

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

[14] The leaf assimilation is scaled to canopy by considering light attenuation by leaf shading at different leaf depths from the top to the bottom of the canopy. Thus, daily canopy assimilation is Z

t
day
end

ZLAI

ANC ¼ yc

An ð L; tÞdLdt t
day
start

ð17Þ

0

where ANC is the daily assimilation (gC m
2 d
1), An is An3 or An4 for C3 or C4 plants, respectively, L and LAI are the dummy and true leaf area indices, respectively, t is time, c is a factor for unit conversion, and y is a calibration factor, and t-day-start and t-day-end specify the starting and ending time of the day. [15] Daily net primary productivity, NPP, is calculated as NPP ¼ ANC
RRS

ð18Þ

where RRS is respiration of root and stems. RRS depends on temperature exponentially in the same way as leaf respiration Rd. [16] TESim uses a simplified version of the CENTURY model [Parton et al., 1987, 1988; Gobron et al., 1999] to simulate litter and SOM dynamics, with modification to include the recent research on soil biogeochemistry [Lomander et al., 1998a, 1998b], so that heterotrophic respiration (RHE, gC m
2 a
1) is calculated as carbon release from decomposition and mineralization of litter and SOM,  RHE ¼ Ddec

KS CLitS ½aL LLitS þ bL ð1
LLitS Þ þ ck KM CLitM þ1000ZN Dbulk ½ EðfCS ÞKSOA CSOA þ ck KSOS CSOS 

G04005

[18] Nitrogen uptake by plants was considered an increasing function of whole-body plant nitrogen deficit, fine root biomass, daily carbon assimilation, and available soil nitrogen content. Carbon and nitrogen allocated to these pools were considered controlled by plant phenology, leaf water potential, and organ/component nitrogen deficit. Littering of root and stems was assumed to be a result of natural senescence. Leaf loss, however, was controlled by natural senescence, minimum temperature, and maximum accumulated growth temperature. Symbolic nitrogen fixation was handled in a similar way as that of Kemp et al. [1997] and Yu et al. [2002b]. The reduction of soil nitrogen (both organic and available) by soil erosion was modeled as the amount of sediment yield multiplied by the concentrations of respective nitrogen forms and an enrichment factor. 2.2. Soil Water, Runoff-Run-on Flow, and Soil Erosion [19] Vertical soil water movement is driven by precipitation events, surface evaporation, plant transpiration, gradient of soil water potential, and hydraulic redistribution by roots [Caldwell and Richards, 1989; Caldwell et al., 1998; Ryel et al., 2002]. Runoff, including both surface and subsurface draining flows, is generated when daily rainfall is greater than the field capacity of a calculated average penetrable soil depth, redistributed to and possibly reabsorbed in down slope. (Reabsorption is only calculated in TESim-L.) Discrete pixel-based calculation of run-on flow is formulated as the following [Weltz and Blackburn, 1995; Naeth and Chanasyk, 1996; Kemp et al., 1997; Gao et al., 2002a; Gao and Reynolds, 2003]:   max zi
zj ; 0 =Dij Fon ði ! jÞ ¼ Roff ðiÞ 4 P maxðzi
zk ; 0Þ=Dik



ð21Þ

k¼1

ð19Þ

where Ddec is a decomposition function of soil temperature and soil water content, LLitS is the lignin content in structural litter; CLitS and CLitM are structural and metabolic litter carbons, respectively (gC m
2); CSOS and CSOA are carbon contents in slow and active soil organic matter, respectively (gC kg (soil)
1); fCS is the fraction of silt + clay in soil. E(fCS) is a function of fCS; aL, bL, ck, KS, KM, KSOA, and KSOS are constant decomposition parameters; ZN is the mean depth of soil containing litter and organic matter (0.4 m assumed in this study); Dbulk is the bulk density of soil, and the factor of 1000 ZNDbulk is used for unit conversion purposes. [17] In contrast to the conventional definition of net ecosystem production (NEP) as NPP minus RHE, net ecosystem carbon balance (NECB) is calculated as [Chapin et al., 2006] NECB ¼ NPP
RHE
ECL
HAV

ð20Þ

where HAV is the carbon loss due to crop harvests at the end of the growth season, and ECL is erosion-induced lateral soil carbon loss, which is computed as the water erosion rate multiplied by SOM concentration and an enrichment factor [Starr et al., 2000].

where Roff (i) is the runoff flow generated at pixel i, Fon(i ! j) is the flow of runoff from pixel i to j. z of i, j, and k is the elevation of pixels i, j, and k, respectively. Dij is the spatial distance between the centers of pixels i and j. Equation (21) partitions runoff flow from a pixel to 4 lower neighborhood pixels with steepness of slope as the distribution weights. By definition, the calculation of Fon(i ! j) requires recursive computation. Soil water erosion (Ewater) is computed based on a published model [Gao et al., 2002a], modified to use fine root biomass instead of a remote sensing vegetation index to represent the effects of vegetation on water erosion: Ewater ¼ es Cs Cv

     RON vMR exp
f ðwg Þ# 10RONC 1
exp
RONC MRC ð22Þ

where Ewater is daily erosion by runoff (gSoil m
2 d
1); es is a dimensionless coefficient depending on scale, Cv is a coefficient (g L
1) depending on surface vegetation cover type; Cs is a parameter indicating the difference in erodibility contributed by factors other than vegetation. g is slope angle in radiance. RON is amount of lateral flow (total runoff passing though a grid cell) (cm d
1), and constant RONC was set to 1.33 cm d
1. # is a constant

5 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

Table 1. Distribution of Plant Functional Types in China CropPasture Belt Plant Functional Types (FT)

CODE

Area Occupied, km
2

Percent

Forest Deciduous shrubs Sandy shrub and subshrubs Meadow steppes Typical steppes Desert steppes Crops Other

FOR SHB SSB MDS STP DST CRP OTH

60579.0 91792.7 39537.3 85922.1 104802.8 50718.5 289234.1 1936.2

8.4 12.7 5.5 11.9 14.5 7.0 39.9 0.3

G04005

and temperature-dependent senescence of plant pools, harvesting time, amount and frequency, and irrigation time and rate, are the basic parameters used in the model. The model outputs soil water contents in different layers, soil organic matter and available nitrogen content, total soil carbon storage, plant carbon storage and nitrogen concentrations, net primary productivity, heterotrophic respiration, soil erosion and erosion-induced carbon loss, and net ecosystem carbon balance.

3. Applications of TESim to CCPB exponent; w is a coefficient depending on spatial resolution (scale); MR is plant root biomass; f is a shape parameter describing the protective role of vegetation against sheet and rill erosion; n is the fraction of fine roots; MRC is a critical root biomass. Factor 10 has units of L (water) m
2 cm
1. [20] Hence, equation (22) calculates soil erosion as an increasing function of slope, lateral flow, but a decreasing function of fine root biomass. The effect of slope length is included in RON because longer slopes result in more accumulated flow. The model is not able to describe severe gully erosion that happens episodically in the south portion of the Loess Plateau. However, the large amount of sediment yield with these gully erosion events usually contains a great portion of deep soils with little SOC, and thus contributes little to ecosystem carbon budget. 2.3. Irrigation, Fertilization, and Crop Harvest [21] To simplify the analysis, crops in irrigated land (denoted by the land use map in addition to the vegetation map) are irrigated whenever soil water content is lower than 80% of the field capacity. Irrigation was applied in the model simply by elevating moisture to the field capacity of the soil. Other effects of irrigation were not considered in this version of the model. All crop sites are chemically fertilized within the growth season to the typical available nitrogen level based on field observations. Because this study represents the case with idealized irrigation and fertilization, it may lead to overestimation of cropland production. Harvest of crops is done by removal of all living aboveground biomass from the field at the end of the growing season. With the senescence before harvest, the current harvest design results in approximately half of production returned to crop fields as litter. 2.4. Driving, Input, and Output of the Model [22] The model is driven by daily climate variables including precipitation, minimum and maximum temperatures, daily mean wind speed, and daily mean relative humidity. Although TESim-P and TESim-L use data from one single meteorological station, TESim-R does require daily weather data from multiple stations to interpolate the meteorological variables for each grid cell. Spatial data required by TESim-L and TESim-R include category maps of vegetation, soil, land use, and the geographic relief map of the interested landscape or region. Soil clay and silt fractions, parameters that characterize stomatal dynamics and photosynthesis, specific leaf area (SLA) and leaf phenology, carbon, lignin, and nitrogen contents, natural

3.1. Data Collection and Model Testing at Site Scales [23] To apply TESim to CCPB, we set a spatial resolution of 1/12 degree latitude and 1/12 degree longitude, resulting in an approximate linear resolution of 8 km. The range of CCPB was divided into 164 rows, 286 columns, and 11140 land grid cells (pixels). The spatial data we used for this application include 1:1M vegetation map, 1-km DEM of China with derived slope and aspect maps at the same resolution, and a 1:1M soil map, all converted to grids of the 8 km resolution by the resampling technique in ArcGIS9.0 (ESRI, USA). Daily weather records of 149 meteorological stations from January 1, 1959 to December 31, 2001 were interpolated into the same spatial resolution to drive the model. Air CO2 pressure was assumed to follow Keeling and Whorf (2002) with spatial variations to take into account the decreased pressure with topographical elevation. [24] Vegetation in CCPB is grouped into 7 functional types (FT) including forests (FOR), deciduous shrubs (SHB), deciduous sandy shrubs and subshrubs (SSB), meadow steppes (MDS), typical steppes (STP), desert steppes (DST), and agricultural crops (CRP), based on the 1:1 million vegetation maps of China (Table 1, and Figure 6 later). The FOR FT is a combination of mostly small leaved deciduous trees (birches and poplars), typical deciduous broadleaf trees (oaks), and a small portion of coniferous trees (mostly pines). We grouped all these forest types into one FT to minimize the distraction from our main interests in relationship to grasses, shrubs, and crops in the CCPB region. [25] Leaf physiological (stomata and photosynthesis) parameters were obtained using nonlinear least square fitting of the stomatal and photosynthesis models to the data of diurnal leaf gas exchange data of dominant species in each FT measured in the field with a portable photosynthesis system (model LI-6400, LiCor, Inc., Lincoln, NE, USA). Physiological models (photosynthesis and stomatal conductance) were published in Gao et al. [2002b, 2004a, 2005]. Plant carbon and nitrogen contents are measured in the laboratory with an element analyzer (Model EA 2400 II, PerkinElemer Inc., Shanghai), and are used to parameterize the relationship between photosynthesis rates and nitrogen concentrations of leaves. These physiological parameters of dominant species were used to represent the ecophysiological characteristics of the corresponding functional types. Parameters of net primary productivity, carbon partition, and plant morphology parameters were mostly determined in Gao and Zhang [1997], Gao and Yu [1998], and Gao et al. [2000] and are adjusted for this application based on field-observed long-term NPP data (Table 2, and Figure 3 later). Crop FT is treated as an annual plant with small root

6 of 16

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

G04005

Table 2. Important Parameters of the Model for This Applicationa Symbol

Name

Units

FOR

SHB

SSB

MDS

STP

DST

CRP

Description


1

0.15 0.91 0.01 0.72

0.10 1.79 1.42 0.80

0.17 2.96 2.80 0.82

0.23 16.3 17.1 0.74

0.22 16.6 16.7 0.74

0.13 16.3 17.2 0.80

0.06 0.40 0.00 0.49

Production Model equations (1) – (9), (16) Values are at 25.0 °C for the first four parameters. except y Stomatal model equations (10) – (16)

ap gx/gt gp y

Photon efficiency Carboxylation coeff. Photorespiration coeff. Assimilation yield coeff.

mol mol mmol m
2 s
1 kPa
1 mmol m
2 s
1 kPa
1

Ymin SV Sy V Cv es W # f MRC KS KM KSOS KSOA

Min. soil water potential Sensitivity to air vapor Sensitivity to soil water Fine root proportion Erodibility coefficient Scaling coefficient Slope coefficient Slope exponent Root effect Scaler to fine roots Rate of structural litter Rate of metabolic litter Rate of slow SOM Rate of active SOM

MPa mol m
2 s
1kPa
1 mol m
2 s
1MPa
1 g L
1 L cm
1


3.58
7.14
6.52
0.08
0.32
0.20 0.24 0.15 0.81 0.3 0.5 0.5 108.8 72.9 72.9 0.75 (Landscape scale) 1.0 (Landscape scale)

gDM m
2 d
1 d
1 a
1 a
1


5.35
5.34
5.94
3.14
0.12
0.05
0.03
0.34 1.30 0.86 0.42 0.11 0.8 0.8 0.8 0.8 67.9 67.9 67.9 176.1 0.83 (Regional scale) 2.30 (Regional scale) 0.41 1.82 100.0 0.013 0.036 0.14 2.91

Erosion model equation (22)

Decomposition model equation (18)

a

Codes are the same as Table 1. Sv and Sy (equations (15) and (16)) are evaluated at Ip = 0, y s = 0, and Vpd = 1 kPa, with Pa = 101.3 kPa.

Figure 3. Comparison between simulated (line) and observed (dot) biomasses: (a) Aboveground biomass of a deciduous broadleaved forest (Robinia pseudoacacia community); (b, c) Root and stem biomass of an evergreen coniferous forest (Pinus tabulaeformis community); (d, e) Aboveground biomass of a deciduous shrub (1. Hyppophae rhamnoides community); (f) Aboveground biomass of a deciduous shrubs (2. Caragana korshinskii community); (g) Agoveground biomass of a meadow steppe community (Leymus chinensis community); (h) Aboveground biomass of a typical steppe community (1. Thymus serpyllum L. community); (i) Aboveground biomass of a typical steppe community (2. Stipa gradis community); (j) Aboveground biomass of a desert steppe community (Stipa breviflora Criseb. community); (k) Leaf biomass of a sandy subshrub community (Artemisia ordosica); (l) Aboveground biomass of a desert shrub community (Caragana pygmaea (L.)). Locations of these sites are illustrated in Figure 1. 7 of 16

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

G04005

Table 3. Connections Between Landscape and Regional Simulationsa Landscape

Zhifanggou

Shihuimiao

Changling

fmin (°) fmax (°) g min (°) g max (°) Relief range (m above sea level) Regional grid cells Landscape Area (km2) Landscape resolution (m) Landscape grid cells Time span NPP -Landscape (gC m
2 a
1) NPP-Regional (gC m
2 a
1) Erosion-Landscape (Gg km
2 a
1) Erosion-Regional (Gg km
2 a
1)

36.72 36.77 109.23 109.26 1025 – 1425 CRP(2) 8.09 30 FOR(426), SHB(122), STP(1553), CRP(5309) OTH(1582) 1961 – 2000 234.9 ± 57.6 237.9 ± 19.0 7.3 ± 2.5 7.6 ± 4.2

39.33 39.35 109.73 109.85 1343 – 1390 SHB(1), SSB(1) 13.93 30 MDS(846) SHB(4629) SSB(10004) 1959 – 1993 59.3 ± 25.8 49.7 ± 17.0 1.7 ± 0.7 1.7 ± 0.6

44.62 44.65 123.57 123.61 137 – 145 MDS(1) 11.36 46.5 MDS(5256) 1959 – 2001 361.6 ± 116.6 379.4 ± 101.3 0.0 ± 0.0 0.0 ± 0.0

a Here fmin and fmax are the minimum and maximum latitudes of, g min and g max are the minimum and maximum longitudes of, the landscape, respectively. Depending on the spatial mapping of location, a landscape can traverse 1 to 4 cells in the regional grid. The numbers of grid cells are in the parentheses following functional types. See functional type codes in Table 1.

biomass and is assumed to have approximately the same NPP as meadow steppes. Soil texture (clay and silt fractions) and nutrients were obtained from the China soil horizon data book [National Soil Survey, 1996]. Soil hydraulic properties are calculated from the clay fraction [Campbell et al., 1993]. Litter and SOM decomposition also used clay and silt fractions as parameters. [26] To test the processes of primary production and biomass dynamics in TESim, TESim-P was run for eleven sites with long-term observations on biomass and productivity within or nearby the CCPB region (Figure 1), with time spans varying from 4 to 27 years [Ma et al., 1990; Xiao, 1990; Liu, 1993; Liu and Xun, 1993; Liu et al., 1993; Xing and Lu, 1993; Chen et al., 1996; Hou, 1997; Yuan and Li, 2001]. Weather data from meteorological stations closest to these sites were used to drive the model. Soil and land use information was found from the 1:1M soil and land use maps. The water erosion model (equation (22)) was run for the observed sediment yield data in several water erosion experiments with various combinations of vegetation cover, fine root biomass, and slope angles, at the Loess Plateau, an area with the most severe water erosion in China [Liu et al., 1990; Hou et al., 1996]. These erosion experiments were performed at plots with spatial scales of 10– 100 m with various slopes and vegetation coverage. Water and sediments were collected after simulated rainfall events, together with plant fine roots sampled and measured at variable soil depths, and data are averaged over all events for each treatment combination. Parameters of the erosion model were adjusted to minimize the difference between the observed and simulated erosion rates (Table 2). 3.2. Scaling the Soil Erosion Model to the Regional Scale [27] Because the calculation of soil erosion strongly depends on a spatial scale (resolution of the grid), we started our spatial simulation with the landscape version of TESim (TESim-L), with erosion parameters on the site scale, to simulate ecosystem processes and soil erosion for three landscapes within CCPB (Figure 1), and then used TESim-R, with try-on erosion parameters for regional scales, to simulate the regional carbon dynamics and ero-

sion for the entire CCPB. The three landscapes are shown in Figure 1. Zhifanggou, a small catchment in the south of the Loess Plateau, is a research site of the Institute of Soil and Water Conservation in the Chinese Academy of Sciences (CAS); Shihuimiao is where the Erdos Sandy Grassland Ecosystem Station of CAS is located; and Changling is a research site of the Institute of Grassland Research of the Northeast Normal University. We chose these sites because they feature heavy (Zhifanggou), moderate (Shihuimiao), and minimal (Changling) water erosion in CCPB, because long-term pertinent ecosystem research has been done at the sites, and because we have been involved in research at these sites over the past 10 years [Gao et al., 2001, 2004a, 2005]. The areas of these landscapes are about 10 km2, much smaller than the area of 1 pixel on the regional grid. However, depending on the particular position of the landscapes within the regional grid, a landscape can traverse up to four pixels in the regional grid. To scale the ecosystem model to the regional scale, we assumed that each of the three landscapes is representative of a relatively large landmass surrounding the landscape with similar geological, climatic, and edaphic properties. This was an important guide for institutions to select their long-term study sites per se. The assumption allowed us to compare the results of the landscape simulation with those of the regional simulation. If the model at the site scales is correctly scaled to the regional scale, the spatial averages of key variables estimated by the landscape model, such as NPP and soil water erosion, should not significantly differ from those by the regional model at the regional grid cells where the landscapes are located [Rastetter et al., 2003]. In our cases, Zhifanggou spans two regional grid cells with different latitudes (Rows), Shihuimiao is within two regional grid cells with different longitudes (Columns); Changling, however, happens to be located within one grid cell (Table 3). Table 3 also shows the number of grid cells occupied by respective functional types on the regional and landscape scales for each of the three landscapes. The parameters depending on spatial scales in the soil erosion model were adjusted to make the results of the simulations on the regional scale as close as possible to those on the landscape scales (Table 2).

8 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

Comparisons between the two scenarios revealed the effects of soil erosion on the regional net ecosystem carbon balance.

4. Results

Figure 4. Observed versus predicted soil erosion rates by equation (22) for different vegetation cover types. Shrub1 and Shrub2 indicated two shrub sites with different rooting patterns. Regression of the modeled against observed yielded a slope of 1.02 not significantly different from 1, an intercept of
1.14 (g L
1) not significantly different from 0, and R2 = 75.2%.

4.1. Comparison Between Observed and Simulated Biomass and Erosion Rates [29] Statistical analysis indicated that the model at the patch scales explained 32% to 99% of the variation in the biomass data for the 11 sites (Figure 3), and the model is a statistically significant descriptor of the field observations at all these sites. The water-erosion model explained 75% of the variations in the observed sediment yields measured as eroded soil mass per volume of runoff flow [Liu et al., 1990; Hou et al., 1996]. The results (Figure 4) show that the model captured the main characteristics of water erosion in the Loess Plateau so that grass-covered land surfaces (steppes) are the most resistant, crop fields and abandoned fields with forbs are the least resistant, and shrubs and planted trees are moderately resistant to water erosion.

3.3. Simulation for the Period Between 1959 and 2001 [28] To investigate the impacts of erosion-induced soil carbon loss and climate shift on the regional carbon balance of CCPB, the parameterized, scaled model was initialized with zero litter, typical SOM and available soil nitrogen contents, and small plant biomasses, and was run for approximate equilibrium with climate data from 133 meteorological stations within and near CCPB from 1959 to 1968 for two scenarios, one of which did, and one of which did not, calculate soil erosion. The end values of the state variables of the model were used as initial values for formal simulation runs from 1959 to 2001 (with daily weather data from the same 133 stations) to obtain the response of ecosystem processes to the climate change over the 43 years.

4.2. Comparison of the Simulations of Landscape and Regional Scales [30] The simulated spatial averages of NPP and soil erosion rates for the three landscapes on the landscape and regional scales are shown in Figure 5 and Table 3. For most years, both simulations showed similar temporal patterns of variation (increase and decreases). However, the landscape simulations showed greater variation in NPP than did the regional simulation for Zhifanggou and Shihuimiao landscapes. The application of TESim-L at Zhifanggou in the Loess Plateau provided average NPP and soil erosion rates of 234.9 ± 57.6 gC m
2 a
1 and 7.3 ± 2.5 Gt (soil mass) km
2 a
1, respectively, in comparison with the regionally simulated NPP of 237.9 ± 19.0 gC m
2 a
1 and soil water erosion of 7.6 ± 4.2 Gg (soil mass) km
2 a
1 in the two regional grid cells that the landscape traverses.

Figure 5. Average net primary productivity and soil erosion simulated by TESim-L at landscape scale and TESim-R at regional scale, for three landscapes of Zhifanggou, Shihuimiao, and Changling. 9 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

Table 4. Effects of Erosion on Net Ecosystem Carbon Balance of CCPBa Average No erosion calculated

NPP 210.1 ± 26.9 HAV 35.9 ± 4.1 RH 169.8 ± 7.7 NECB 4.4 ± 22.5 With erosion calculation NPP 206.8 ± 26.3 imposed EWT 4.0 ± 1.0 ECL 11.0 ± 2.8 HAV 35.6 ± 4.2 RHE 155.3 ± 7.4 NECB 4.8 ± 20.0 Difference DNPP
3.3 ± 0.7 DECL 11.0 ± 2.8 DHAV
0.2 ± 0.0 DRHE
14.5 ± 0.3 DNECB 0.4 ± 3.0

FOR

SHB

SSB

MDS

STP

DST

CRP

417.3 ± 85.2 279.5 ± 46.2 190.3 ± 31.0 198.8 ± 44.0 143.3 ± 29.0 81.2 ± 17.8 199.7 ± 17.8 90.0 ± 10.3 427.7 ± 13.8 264.8 ± 16.1 186.5 ± 8.8 185.9 ± 17.8 135.7 ± 10.1 76.3 ± 7.0 109.2 ± 4.4
10.5 ± 88.2 14.7 ± 42.2 3.9 ± 29.8 12.9 ± 39.1 7.6 ± 27.4 4.9 ± 15.4 0.5 ± 5.2 416.8 ± 85.1 268.5 ± 43.6 170.3 ± 27.5 198.2 ± 43.9 142.5 ± 28.9 80.6 ± 17.7 198.3 ± 17.7 0.0 ± 0.0 1.3 ± 0.3 1.4 ± 0.4 1.6 ± 0.3 1.3 ± 0.3 2.2 ± 0.6 8.2 ± 2.1 0.7 ± 0.2 1.3 ± 0.7 0.9 ± 0.4 0.1 ± 0.1 0.0 ± 0.1 0.1 ± 0.1 26.9 ± 6.7 89.4 ± 10.3 426.6 ± 13.8 254.0 ± 14.3 166.1 ± 7.7 185.2 ± 17.8 134.9 ± 10.0 75.6 ± 6.9 79.8 ± 4.4
10.5 ± 88.2 13.1 ± 40.4 3.2 ± 26.3 12.9 ± 38.9 7.5 ± 27.2 4.9 ± 15.3 2.2 ± 6.2
0.5 ± 0.2
11.0 ± 3.6
20.1 ± 4.3
0.6 ± 0.2
0.8 ± 0.2
0.6 ± 0.2
1.4 ± 0.1 0.7 ± 0.2 1.3 ± 0.7 0.9 ± 0.4 0.1 ± 0.1 0.0 ± 0.0 0.1 ± 0.1 27.0 ± 6.7
0.5 ± 0.1
1.2 ± 0.1
10.7 ± 1.8
20.4 ± 1.2
0.7 ± 0.1
0.8 ± 0.1
0.6 ± 0.1
29.4 ± 0.4
0.0 ± 0.3
1.6 ± 2.3
0.7 ± 4.3 0.0 ± 0.2
0.1 ± 0.2
0.0 ± 0.2 1.7 ± 6.7

a Units are Gg km
2 a
1 for soil erosion (EWT), and gC m
2 a
1 for other variables. NPP, net primary productivity; HAV, agricultural harvest; ECL, erosion caused lateral carbon export; RHE, heterotrophic respiration. Difference = with erosion – without erosion. Standard errors indicate temporal variations. Bold numbers indicate significant differences.

Statistical comparisons between the two means indicated that the two estimates did not differ significantly from each other. Past research showed that the spatial average erosion modulus of Zhifanggou varied with land use patterns from 10.5 to 7.5 Gg km
2 a
1 [Lu et al., 1997], indicating the model worked properly for the areas of CCPB with severe water erosion. The simulated NPP of Shihuimiao on the landscape scale is larger than, but not significantly different from, that on the regional scale (59.3 vs. 49.7 gC m
2 a
1). The larger landscape-scale NPP was mostly because a small portion of meadow steppes in the lower area of the landscape was not properly represented in the regional vegetation grid. The difference in overall average water-induced soil erosion (1.7 vs. 1.7 Gg km
2 a
1), however, is minimal. A previous study indicates that the county that includes the Shihuimiao landscape had an average water erosion rate of 1. 1 1.7 Gg m
2 a
1 [Gao et al., 2002a], which is lower than the present calculation, because the landscape is located in the eastern part of the county with higher precipitation and water erosion rates than those in the other areas of the county. The meadow steppe landscape of Changling in east CCPB has higher productivity than the two landscapes in the west, with little soil erosion simulated on both scales and reported in the literature. No significant difference in average NPP between the landscape and regional simulations was found (361.6 vs. 379.4 gC m
2 a
1) for the site. In 1991, a year with average precipitation and temperature, we conducted field measurements on seasonal biomass changes of Leymus chinensis communities. Such communities have been recognized as the dominant community type within the landscape. And we found 412.5 gDM m
2 to be the maximum aboveground biomass in August [Gao et al., 1996], in comparison with the simulated 357.8 gDM m
2 (spatially averaged over the landscape) for the year. 4.3. Impacts of Erosion-Induced Carbon Loss on Regional Carbon Balance From 1959 to 2001 [31] The simulated averaged NPP of the 7 FTs in the 43 years, shown in Table 4, varied from 81.2 gC m
2 a
1 for desert steppes (DST) with erosion calculated, to 417.3 gC m
2 a
1 for the forest FT (FOR) without erosion being considered, resulting in 210.1 and 206.8 gC m
2 a
1 as the average regional NPP without and with soil water

erosion considered, respectively. The largely unaltered NPP under soil erosion, in comparison with the scenario without erosion calculation, had something to do with our assumed optimal fertilization scheme which maintained available soil nutrients under severe soil erosion in the crop FT. Heterotrophic respiration is comparable in magnitude to NPP values for all FTs except the crops (CRP). NPP of the crop FT is balanced by the sum of heterotrophic respiration (RHE), agricultural crop harvest (HAV), and erosion-induced soil carbon loss (ECL). The average net ecosystem carbon balance (NECB) ranged from
10.5 gC m
2 a
1 for the forest FT (FOR) to 12.9 gC m
2 a
1 for the meadow steppe FT (MDS) without erosion calculation. The negative NECB for the forests reported in Table 4 is mostly because of the simulated decrease in NPP and rapid increases in heterotrophic respiration with the temperature after the 1980s (later). The average regional NECB calculated to 4.4 and 4.8 gC m
2 a
1, respectively, for the scenarios without and with soil erosion considered. [32] The calculated average erosion-induced soil organic carbon loss (ECL) was 26.9 for the crop FT, which is equivalent to 11.0 gC m
2 a
1 for the CCPB region (Figure 6). The calculated ECL values for the ‘natural’ ecosystems without disturbance imposed are one order of magnitude smaller than that for the crop FT, despite large soil erosion rates being calculated at some patches with little biomass within these undisturbed FTs. The high erosion SOC loss for the crop FT is mostly the result of regular irrigation and fertilization to maintain large NPP under high soil erosion rates. The large NPP in turn maintains relatively constant SOC pool with respect to soil erosion, thus represents a feedback to erosion and carbon balance. In the natural ecosystems, however, great spatial heterogeneity exists among patches within FTs. Heavy soil erosion exists in some pixels with little plant biomass and soil organic matter. Severe soil erosion caused a heavy reduction in soil organic matter and available nitrogen, thus preventing vegetation from growing normally in these pixels. On the other hand, most pixels in undisturbed FTs with plants established and growing normally had negligible soil erosion. The erosion-induced SOC loss, though small compared with that induced by plant carbon fixation (NPP), is a

10 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

Figure 6. Simulated annual net primary production (NPP), net ecosystem production (NECB), erosioninduced SOC carbon loss (ECL), heterotrophic respiration (RHE), all averaged over China Crop-Pasture Belt (CCPB), plotted against time in years. Annual mean temperature and precipitation were also presented. Dashed lines are predictions by linear regressions of simulated/interpolated variables on time. significant component of net ecosystem carbon balance for both natural undisturbed and managed agricultural crop systems. [33] Erosion-induced soil organic carbon losses are balanced by the decreased heterotrophic respiration for all FTs and the whole region by direct removal of substrates for heterotrophic respiration, resulting in a statistically insignificant comparison in NECB for most FTs and the entire region between the two scenarios. In particular, large ECL in the erosion scenario even caused increases in NECB from 0.5 to 2.2 gC m
2 a
1 for the crop FT, which caused the regional NECB to increase from 4.4 to 4.8 gC m
2 a
1. The contrast between the scenarios with and without erosion is expected to increase with agricultural land use. Hence, erosion did not decrease the regional carbon balance, but only ‘‘converted’’ about 9% of RHE to ECL and caused a reduction in NPP. [34] Simulated averages over the 43-year period indicate that both NPP and RHE follow the pattern of rainfall distribution (Figure 7, NPP, RHE), and prescribed ground water condition for meadow steppes FT(MDS) in north CCPB, showing that rainfall is the strongest control on plant carbon fixation in this area. Large negative values of NECB (NECB in Figure 7) were simulated in the northeast at sites of crops and in the southeast at sites of forests. Vegetation, other than crops in the northern portion of CCPB in general, contributes to large, positive NECB because of its lowtemperature limited heterotrophic respiration. Soil erosion prevented establishment of vegetation at some sites in the northwest side of CCPB and caused large differences in NPP between the scenarios with and without soil erosion for these sites. For established vegetation, the effect of soil erosion was in general slightly negative or even slightly positive in some locations (dNPP in Figure 7). The calculated positive effects of soil erosion on NPP might be caused by the nonlinearity of ecosystems. Erosion may cause a decrease in productivity in years with high rainfall in these places;

however, the decreased vegetation activities in wet years because of erosion may leave soil water for better sprouting and growth in the successive dry years. The spatial patterns of simulated SOC loss (ECL) and the effects of erosion on heterotrophic respiration (dRHE) are closely related to the distribution in agricultural crops (VEGT), and the effect of erosion-induced SOC loss on regional carbon balance is offset by the decreases in heterotrophic respiration, which results in relatively small differences in NECB between the two scenarios. 4.4. Shifts in Ecosystem Processes in the Early 1980s [35] Temporal regional carbon balance with the calculation of soil erosion (Figure 6) showed that regional NPP is highly variable with time, varying from 145.5 gC m
2 a
1 in 1972 to 277.4 gC m
2 a
1 in 1998; the two years also had the minimum and maximum rainfall, respectively. Regional average NECB, also highly variable, had its minimum of
35.1 gC m
2 a
1 and maximum of 53.0 gC m
2 a
1 in the same two years. Statistically insignificant slopes of
0.06 gC m
2 a
2 and
0.7 gC m
2 a
2 were detected for the regional NPP and NECB from 1981 to 2001, respectively. The simulated regional SOC loss showed a minimum value of 5.7 gC m
2 a
1 in 1972 with lowest soil erosion, and the maximum value of 17.9 gC m
2 a
1 in 1964 with the second highest rainfall, and there has been a systematic decreasing trend in ECL over the past 43 years of simulation. Compared with NECB, NPP, and ECL, regional heterotrophic respiration seems much less variable with a minimum 145.7 and a maximum 170.7 gC m
2 a
1. [36] The simulated regional averages indicated that a significant shift associated with changes in climate variables occurred in ecosystem processes in the early 1980s. The interpolated average annual mean temperature (Figure 6e) did not show a systematic increase until 1980 when it started to increase rapidly with an average slope of 0.03°C a
1. The annual precipitation, however, also showed

11 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

Figure 7. Simulated NPP, NECB, RHE, without erosion calculation, the differences in NPP and RHE between two scenarios with and without erosion calculation (With – Without), dNPP and dRHE, and ECL simulated in the scenario with erosion calculation, all averaged over the 43 years from 1959 to 2001. Digital elevation map (ELEV), vegetation distribution (VEGT) and interpolated average precipitation (PREC) in CCPB are included. two-phase processes separated at 1980, although the change in the precipitation is not as evident as that of the temperature. There are no significant systematic trends in the simulated regional carbon fluxes detected from 1959 to 1980. In contrast, the simulation indicated significant increases in NPP for shrubs (SHB), meadow steppes (MDS), typical steppes (STP), desert steppes (DST), and agricultural crops (CRP) from 1981 – 2001 compared with the period from 1959 – 1980 (Table 5). The largest increases of 47.4 and 35.6 gC m
2 a
1 in NPP were found for meadow steppes (MDS) and shrubs (SHB), respectively. An almost statistically significant increase in NPP was calculated for sandy shrubs (SSB) with p = 0.06, and an insignificant decrease in NPP for the forest FT (FOR) was detected. The regionally averaged NPP was also found to increase significantly by 22.1 gC m
2 a
1. The climate shift caused significant changes in average heterotrophic respiration, which varies from
5.1 gC m
2 a
1 for the forests (FOR) to 29.9 gC m
2 a
1 for the meadow steppes (MDS), respectively. And the regional heterotrophic respiration was increased by 11.8 gC m
2 a
1. Carbon loss due to agricultural crop harvesting was calculated to increase by 13.1 and 5.2 gC m
2 a
1 for the crop FT and the entire region, respectively. Decreases in erosion-induced SOC loss (ECL), though not statistically significant, were found for all FTs and the regional average. The magnitudes of decreases varied from 0 gC m
2 a
1 for the forest FT to 1.1 gC m
2 a
1 for crop FT. In contrast, the climate shift brought

about statistically significant increases in soil erosion for meadow steppes (MDS), and insignificant increases in soil erosion of the forests (FOR), typical steppes (STP), desert steppes (DST), and statistically insignificant decreases in erosion for the crops (CRP). The regional average soil erosion seems unchanged with the climate shift. The climate shift also caused significant increases of 17.5, 9.0 and 3.9 gC m
2 a
1 in NECB for the meadow steppes (MDS), the desert steppes (DST), and the crops (CRP), respectively, a statistically insignificant decrease of
24.6 in NECB for the forests (FOR) and statistically insignificant increases of 13.4, 5.9, and 9.7 gC m
2 a
1 for the shrubs (SHB), sandy shrubs (SSB), and typical steppes (STP), respectively, resulting in a statistically insignificant increase of 5.7 gC m
2 a
1 in the regionally averaged NECB with the climate shift.

5. Discussion [37] The simulation for ecosystem carbon balance in Ohio mine lands with various initial SOC conditions [West and Wali, 2002] shows that carbon displaced by soil erosion varied from 9.2 to 93.5 gC m
2 a
1. Quinton et al. [2006] found that soil loss by water erosion in England and Wales was between 1.3 and 4.8 gC m
2 a
1. Polyakov and Lal [2004] used the CENTURY model to simulate carbon loss by water erosion in meadow and corn-soybean rotation and found that an average soil organic carbon loss of about

12 of 16

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

G04005

G04005

Table 5. Responses of the Carbon Budgets of CCPB to Climate Shift in the Early 1980s of the Last Centurya Units NPP59-80 NPP81-01 EWT59-80 EWT81-01 ECL59-80 ECL81-01 HAV59-80 HAV81-01 RHE59-80 RHE81-01 NECB59-80 NECB81-01


2


1

gC m a gC m
2 a
1 Gg km
2 a
1 Gg km
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1 gC m
2 a
1

Region

FOR

SHB

SSB

MDS

STP

DST

CRP

196.0 ± 20.7 218.1 ± 27.3 4.0±1.1 4.0 ± 0.9 11.3 ± 2.9 10.8 ± 2.7 33.1 ± 3.3 38.3 ± 3.0 149.5 ± 3.0 161.3 ± 5.6 2.0 ± 16.2 7.7 ± 23.3

431.2 ± 86.1 401.6 ± 83.5 0.0 ± 0.0 0.1 ± 0.0 0.7 ± 0.2 0.7 ± 0.2

251.1 ± 32.0 286.7 ± 47.2 1.3 ± 0.3 1.3 ± 0.3 1.1 ± 0.7 0.9 ± 0.3

162.9 ± 23.1 178.0 ± 30.0 1.4 ± 0.4 1.4 ± 0.3 1.1 ± 0.4 0.8 ± 0.3

175.0 ± 25.5 222.4 ± 46.4 1.4 ± 0.2 1.7 ± 0.3 0.1 ± 0.1 0.0 ± 0.1

130.1 ± 19.7 155.5 ± 31.5 1.2 ± 0.4 1.3 ± 0.3 0.1 ± 0.1 0.0 ± 0.0

71.7 89.9 2.1 2.3 0.1 0.0

± ± ± ± ± ±

187.7 ± 14.3 209.5 ± 13.8 8.3 ± 2.3 8.1 ± 1.9 27.5 ± 7.0 26.4 ± 6.6 83.0 ± 8.3 96.1 ± 7.6 76.9 ± 3.1 82.8 ± 3.6 0.3 ± 7.1 4.2 ± 4.6

429.0 ± 13.3 423.9 ± 14.1 1.5 ± 91.4
23.1 ± 85.2

242.8 ± 5.6 265.8 ± 10.7 6.6 ± 30.8 20.0 ± 48.4

161.4 ± 5.3 171.0 ± 6.8 0.3 ± 22.1 6.2 ± 30.4

170.6 ± 6.1 200.5 ± 12.0 4.4 ± 25.9 21.9 ± 48.1

127.2 ± 4.0 143.0 ± 7.7 2.8 ± 20.1 12.5 ± 32.9

71.2 ± 2.2 80.3 ± 7.1 0.5 ± 11.5 9.5 ± 17.5

11.3 18.7 0.7 0.5 0.1 0.1

a

Variable codes are the same as Table 3. Numbers in the first column mark the starting and ending years, i.e., 59 for 1959, 80 for 1980, 81 for 1981, and 01 for 2001. Bold numbers indicate significant differences.

40 gC m
2 a
1 for crop rotation over 35 years, and a very small rate of carbon loss was found for meadows over 15 years. It was interesting to see that in their experiments, for a given initial SOC value, the ultimate long-term carbon removed by water erosion is far from directly proportional to the soil erosion rate. When soil erosion increased 10-fold, carbon removal over 90 years increased less than 40%. These findings using experimental and modeling approaches are comparable to our simulated SOC loss by water erosion in CCPB. This study also found that soil erosion has much smaller effects on the carbon balance of undisturbed natural ecosystems than on those managed agricultural ecosystems. The differences in the erosion-induced soil organic carbon loss between undisturbed natural ecosystems and managed crop systems have also been demonstrated in several studies [West and Wali, 2002; Polyakov and Lal, 2004; Quinton et al., 2006]. [38] Cao et al. [2003] simulated carbon balance in China for the period from 1981 to 1998 and concluded that the average NPP in northern China, with a total area around 1,560,000 km2, over the 18-year period is 390 TgC a
1, which is equivalent to approximately 250 gC m
2 a
1. Our estimate of NPP in CCPB with soil erosion calculation is 196.0 and 218.1 gC m
2 a
1 for the periods between 1959 and 1980 and between 1981 and 2001, respectively. Our lower estimation of average NPP in CCPB than reported by Cao et al. [2003] might have something to do with inclusion of a low productivity functional type of sandy shrubs (SSB) in grassland ecosystems, and the three years of severe drought in northern China from 1999 to 2001. The simulated NPP also agreed with our earlier calculation using climate data in the same region from 1986 to 1990 [Gao et al., 2000]. The simulated increasing trends in vegetation activities from the early 1980s in this study is also supported by several studies based on analyses of long-term satellite vegetation indices [Piao et al., 2003, 2004], mechanistic simulations [Cao et al., 2003], and inventory calculation [Fang et al., 2003, 2004]. Our analyses revealed that NPP of all functional types had an increasing slope from 1981 to 1998 except for the forest FT (not shown in the tables), the NPP of which showed a decreasing trend with time. The changes in NPP of the functional types are largely a result of interplay between increased temperature and changes in precipitation. High temperature increases vegetation productivity only when water is sufficient with high precipitation. High temperature with low water supply, however, is disastrous for plants. Therefore, increasing temperature may bring about

instability for arid and semiarid ecosystems, as concluded by Cao et al. [2003]. [39] Soil erosion is shown in this study to increase regional NEP, because ECL directly removed substrates of heterotrophic respiration. The calculated average NEP for entire CCPB is 40.2 and 51.5 gC m
2 a
1 for scenarios without and with erosion calculation, respectively. However, the erosion-induced SOC loss of and reduction in heterotrophic respiration offset each other to render a much smaller difference in net ecosystem carbon balance between the two scenarios with and without erosion calculation. The simulated NECB of 7.7 gC m
2 a
1 for the period from 1981 to 2001 for the scenario with soil erosion is comparable to the calculated carbon sequestration rates of the northern United States of America (USA) from 1980 to 1993 with three renowned global ecosystem models [Schimel et al., 2000], which gives average rates of around 10 gC m
2 a
1 for Midwest and Central grasslands in the USA (read from the charts in the paper) with similar climate and vegetation distribution to those of CCBP. The value of 7.7 gC m
2 a
1 in this study is also slightly higher than the estimation of net ecosystem carbon balance of 6.7 gC m
2 a
1 by Cao et al. [2003] for northern China. This research, however, gave a lateral carbon loss up to 11.0 gC m
2 a
1 by means of soil erosion, which can be a large source of carbon for aquatic ecosystems. The large lateral carbon loss also means significantly smaller local soil carbon efflux than the previous research because of the erosion-induced removal of SOC from the regional ecosystem. These calculations need to be validated by various experimental and observational techniques in the future. The climate shift in the late 1970s to early 1980s in China has been recognized in studies [Gong and Shi, 2001; Gong, 2002], and this is also likely to be both a consequence of global [Cramer et al., 2001] events and rapid regional land use change during that period which feedback to climate [Chapin et al., 2002]. The regional ecosystem responded to the increases in temperature and precipitation by increasing its net primary production and net ecosystem carbon balance. [40] In terms of offsetting global warming, the regional ecosystem of CCPB increased its carbon pools by 5.6 TgC a
1 during the period between 1981 and 2001, about 8.6% of the carbon sink capacity of vegetation in China [Cao et al., 2003], 2.7% of that in middle northern latitudes [Houghton, 2003b], and 2.7% of that worldwide. However, NECB is simulated to decrease with time after

13 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

the 1980s at a rate of 0.7 gC m
2 a
2, although not statistically significant because of the high inter-annual variability, largely due to the decreased precipitation and increased temperature during the period from 1981 to 2001. [41] Because we used field measured leaf-photosynthesis and stomatal parameters in the regional ecosystem simulation model, we were able to connect leaf-scale ecophysiological characteristics of vegetation functional types to macro-scale regional ecosystem behavior. Our simulation results indicated that shrub lands are potential sinks of carbon in arid and semiarid ecosystems in the future with rising atmospheric temperature, because of their better drought resistance and capacity for water conservation [Polley et al., 1996; Grammatikopoulos, 1999; Reynolds et al., 1999b]. This is illustrated by comparison of minimum soil water potential (Ymin), which represents drought tolerance of the plants, among functional types in Table 2. The two shrub FTs have the lowest Ymin (
7.14 and
6.52 Mpa), thus the highest drought tolerance, while the crops and forest have the highest Ymin (
3.14 and
3.58 Mpa, respectively) and hence the lowest drought tolerance. The grass FTs have Ymin ranging from
5.34 to
5.94, hence showed moderate tolerance to drought. Shrubs have been reported to have gained popularity in the past century with many hypothetical explanations including global warming climatic changes [Neilson, 1986; Grover and Musick, 1990; Schlesinger et al., 1990; Reynolds et al., 1999a]. Grasslands, especially the meadow steppes in high latitudes, may also be large carbon sinks because of their increased net primary productivity with temperature [Parton et al., 1995]. Our analysis also indicates that agricultural crop lands can be potential carbon sinks with appropriate management of residual and soil [Lal, 2002b]. Sandy shrubs, a functional type within CCPB, showed less capacity as carbon sinks because of the limited nutrient supply in sandy soils [Oren et al., 2001]. By contrast, undisturbed forests in arid and semiarid ecosystems can be potential carbon sources because of their decreased primary productivity and increased heterotrophic respiration with rising temperature. Trees have been regarded as major carbon sinks because of their high productivity but low decomposition rates of woody tissues [Schimel et al., 2001; Houghton, 2003b]. But their large leaf area index often represents low water use efficiency and poor water conservation capacity, and hence less adaptation to arid and semiarid environments. [42] The results reported in this article do not include the effects of episodic natural disturbances, such as fires, and anthropogenic managerial activities, such as livestock grazing and land use changes. These disturbances can affect the regional ecosystem by reducing/increasing net primary production and rates of nutrient and carbon cycling, and bring about changes in processes and carbon storage in soil. The large sink capacity of shrubs and grasses should be attenuated by the grazing and utilization that cause carbon loss. In CCPB, the dominant land use change over the past 100 years is the change from wild forests and grasslands into crop fields with the increasing residential population and economic demands. The trend is reversed by the recent efforts to combat desertification and land degradation. Problems with land use change and livestock grazing should

G04005

be addressed in our next analysis for purposes of regional ecosystem management.

6. Summary [43] This research attempted to quantify net carbon balance of various vegetation functional types within the crop-pasture transition region in northern China (CCPB) during the period from 1959 to 2001, by using a spatially explicit regional ecosystem model that simulates carbon assimilation by vegetation, litter, and soil carbon decomposition, and soil erosion-induced carbon loss. We used field measured ecophysiological data of major plants and longterm observations of community biomass within the region or close to it to test the model. The results of the simulation showed that soil erosion induced SOC loss in the undisturbed natural ecosystems is one order of magnitude smaller than that of the managed crops systems, lateral carbon loss by means of soil erosion may not significantly alter the regional net ecosystem carbon balance in arid and semiarid regions, but may decrease soil carbon efflux by transporting the SOC with sediments to aquatic ecosystems. The simulated average net ecosystem carbon balance of CCPB from 1981 to 2001 is 5.6 TgC a
1, which is 8.6% of the carbon sink capacity of vegetation in China. The simulation detected a significant shift in ecosystem processes in the early 1980s, indicated by increased net carbon fixation and the enhanced carbon sink capacity of vegetation and soils driven by rapid increases in atmospheric temperature. Shrub lands and meadow steppes in the north latitude of CCPB have been shown to have the greatest potential sink capacity because of their drought resistance or their location. Undisturbed natural forests in CCPB may become carbon sources because of their lower resistance to drought with increasing temperature and inadequate water supply. Regional ecosystem restoration within and in the neighborhood of CCPB may choose shrubs with better adaptation to droughts with increasing temperature to enhance carbon sink capacity, instead of the arbor species that have been widely used in ecosystem restoration in northern China over the past years. [44] Acknowledgments. This research was jointly supported by the National Science Foundation of China grants 30590384, 40671071, and BNU Innovation Team Funds for Synthetic Landscape Dynamics.

References Bajracharya, R. M., R. Lal, and J. M. Kimble (2000), Erosion effects on carbon dioxide concentration and carbon flux from an Ohio alfisol, Soil Sci. Soc. Am. J., 64, 694 – 700. Band, L. E., C. L. Tague, P. Groffman, and K. Belt (2001), Forest ecosystem processes at the watershed scale: hydrological and ecological controls of nitrogen export, Hydrol. Processes, 15, 2013 – 2028. Beerling, D. J., F. I. Woodward, M. Lomas, and A. J. Jenkins (1997), Testing the responses of a dynamic global vegetation model to environmental change: A comparison of observations and predictions, Global Ecol. Biogeogr. Lett., 6, 439 – 450. Berry, J. A., and G. D. Farquhar (1978), The CO2 concentrating function of C4 photosynthesis: a biochemical model, in The 4th International Congress on Photosynthesis, edited by T. Goodwin, pp. 119 – 131, Biochem. Soc., London. Brody, S. D. (2003), Implementing the principles of ecosystem management through local land use planning, Population Environ., 24, 511 – 540. Caldwell, M. M., and J. H. Richards (1989), Hydraulic lift: water efflux from upper roots improves effectiveness of water uptake by deep roots, Oecologia, 79, 1 – 5.

14 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

Caldwell, M. M., T. E. Dawson, and J. H. Richards (1998), Hydraulic lift: consequences of water efflux from the roots of plants, Oecologia, 79, 1 – 5. Campbell, G. S., J. D. Jungbauer, S. Shiozawa, and R. D. Hungerford (1993), A one-parameter equation for water sorption isotherms of soils, Soil Sci., 156, 302 – 305. Cao, M., and K. Li (2000), Perspective on terrestrial ecosystem-climate interaction (in Chinese), Adv. Earth Sci., 15, 446 – 452. Cao, M., B. Tao, K. Li, X. Shao, and S. D. Prience (2003), Interannual variation in terrestrial ecosystem carbon fluxes in China from 1981 to 1998, Acta Bot. Sin., 45, 552 – 560. Chapin, F. S., P. A. Matson, and H. A. Mooney (2002), Principles of Terrestrial Ecosystem Ecology, Springer, New York. Chapin, F. S., et al. (2006), Reconciling carbon-cycling concepts, terminology, and methods, Ecosystems, 9, 1 – 11, doi:10.1007s10021-1000510105-10027. Chen, Y., X. Wu, X. Liu, and H. Zhao (1996), Correlation analysis between Pinus Tabulaeformis growth and climatic factors (in Chinese), Bull. Soil Water Conserv., 16, 38 – 42. Collatz, G. J., C. M. Ribas, and J. A. Berry (1992), Coupled photosynthesisstomatal conductance model for leaves of C-4 plants, Aust. J. Plant Physiol., 19, 519 – 538. Cramer, W., et al. (2001), Global response of terrestrial ecosystem structure and function to CO2 and climate change: results from six dynamic global vegetation models, Global Change Biol., 7, 357 – 373. Fang, J., S. Piao, C. Field, Y. Pan, Q. Guo, L. Zhou, C. Peng, and S. Tao (2003), Increasing net primary production in China from 1982 to 1999, Frontier Ecol. Environ., 1, 293 – 297. Fang, J., S. Piao, J. He, and W. Ma (2004), Increasing terrestrial vegetation activity in China, 1982 – 1999, Sci. China, Ser. C, 47, 229 – 240. Farquhar, G. D., S. V. Caemmerer, and J. A. Berry (1980), A biochemical model of photosynthetic CO2 assimilation in leaves of C3 species, Planta, 149, 78 – 90. Gao, Q., and J. F. Reynolds (2003), Historical shrub-grass transition in the northern Chihuahuan desert: modeling the effects of shifting rainfall reasonality and event size over a landscape gradient, Global Change Biol., 9, 1475 – 1493. Gao, Q., and M. Yu (1998), A model of regional vegetation dynamics and its application to the study of Northeast China Transect (NECT) responses to global change, Global Biogeochem. Cycles, 12, 329 – 344. Gao, Q., and X. Zhang (1997), A simulation study of responses of the northeast China transect to elevated CO2 and climate change, Ecol. Appl., 7, 470 – 483. Gao, Q., X. Yang, R. Yun, and C. Li (1996), MAGE, a dynamic model of alkaline grassland ecosystems with variable soil characteristics, Ecol. Modell., 93, 19 – 32. Gao, Q., M. Yu, and X. Yang (2000), An analysis of sensitivity of terrestrial ecosystems in China to climatic change using spatial simulation, Clim. Change, 47, 373 – 400. Gao, Q., M. Yu, X. Yanng, and J. Wu (2001), Scaling simulation models for spatially heterogeneous ecosystems with diffusive transportation, Landscape Ecol., 16, 289 – 300. Gao, Q., L. Ci, and M. Yu (2002a), Modeling wind and water erosion in northern China under climate and land use changes, J. Soil Water Conserv., 57, 46 – 55. Gao, Q., P. Zhao, X. Zeng, X. Cai, and W. Shen (2002b), A model of stomatal conductance to quantify the relationship between leaf transpiration, microclimate and soil water stress, Plant Cell Environ., 25, 1373 – 1381. Gao, Q., X. Zhang, Y. Huang, and H. Xu (2004a), A comparative analysis of four models of photosynthesis for 11 species in the Loess Plateau, Agric. For. Meteorol., 126, 203 – 222. Gao, Z., J. Liu, M. Cao, K. Li, and B. Tao (2004b), Impacts of land use and climate change on regional net primary productivity, Acta Geogr. Sin., 59, 581 – 591. Gao, Q., X. Zhang, H. Xu, and Y. Huang (2005), Modeling seasonal and diurnal dynamics of stomatal conductance of plants in a semiarid environment, Functional Plant Biol., 32, 583 – 598. Gobron, N., B. Pinty, M. Verstraete, and Y. Govaerts (1999), The MERIS Global Vegetation Index (MGVI): Description and preliminary application, Int. J. Remote Sens., 20, 1917 – 1927. Gong, D. Y. (2002), Shift in the summer rainfall over the Yangtze River valley in the late 1970s, Geophys. Res. Lett., 29(10), 1436, doi:10.1029/ 2001GL014523. Gong, D. Y., and P. Shi (2001), Variability of summer rainfall over northern China and its accociation with thermal conditions of early stage underlying surface, J. Nat. Resour. (China), 16, 211 – 215. Grammatikopoulos, G. (1999), Mechanisms for drought tolerance in two Mediterranean seasonal dimorphic shrubs, Aust. J. Plant Physiol., 26, 587 – 593.

G04005

Grover, H. D., and H. B. Musick (1990), Shrubland encroachment in southern New Mexico, USA: An analysis of desertification processes in the American southwest, Clim. Change, 17, 305 – 330. Hou, X. (1997), On the mechanics of soil and water conservation in forest regions of the Loess Plateau, in Eco-Agriculture in the Loess Plateau, edited by Z. Lu, pp. 45 – 67, Shaanxi Sci. and Technol. Press, Xi’an, China. Hou, X., G. Bai, and Q. Cao (1996), Observations on soil moisture dynamics of arbor and shrubbery in sinking lands in Loess hilly regions (in Chinese), Res. Soil Water Conserv., 3, 57 – 65. Houghton, R. A. (2003a), Revised estimates of the annual net flux of carbon to the atmosphere from changes in land use and land management 1850 – 2000, Tellus, Ser. B, 55, 378 – 390. Houghton, R. A. (2003b), Why are estimates of the terrestrial carbon balance so different?, Global Change Biol., 9, 500 – 509. Izaurralde, R. C., J. R. Williams, W. B. McGill, N. J. Rosenberg, and M. C. Q. Jakas (2006), Simulating soil C dynamics with EPIC: Model description and testing against long-term data, Ecol. Modell., 192, 362 – 384. Jacinthe, P. A., R. Lal, L. B. Owens, and D. L. Hothem (2004), Transport of labile carbon in runoff as affected by land use and rainfall characteristics, Soil Tillage Res., 77, 111 – 123. Kemp, P. R., J. F. Reynolds, Y. Pachepsky, and J.-L. Chen (1997), A comparative modelling study of soil water dynamics in a desert ecosystem, Water Resour. Res., 33, 73 – 90. Lal, R. (2000), Soil management in the developing countries, Soil Sci., 165, 57 – 72. Lal, R. (2002a), Soil carbon dynamics in cropland and rangeland, Environ. Pollut., 116, 353 – 362. Lal, R. (2002b), Soil carbon sequestration in China through agricultural intensification, and restoration of degraded and desertified ecosystems, Land Degrad. Dev., 13, 469 – 478. Lal, R. (2004), Soil carbon sequestration impacts on global climate change and food security, Science, 304, 1623 – 1627. Levin, S.A. (1992), The problem of pattern and scale in ecology, Ecology, 73, 1943 – 1967. Li, K., S. Wang, and M. Cao (2003), Carbon storage in vegetation and soil in China, Sci. China, Ser. D, 33, 72 – 80. Liu, D. (1993), Aboveground biomass of desert steppes in Inner Mongonia, in Dynamic Monitoring Studies of Husbandry in Northern China Grasslands, edited by B. Li, Sci. Press, Beijing. Liu, D., and Q. Xun (1993), Aboveground biomass of Inner Mongolia meadow steppes, in Dynamic Monitoring Studies of Husbandry in Northern China Grasslands, edited by B. Li, Sci. Press, Beijing. Liu, S. G., N. Bliss, E. Sundquist, and T. G. Huntington (2003), Modeling carbon dynamics in vegetation and soil under the impact of soil erosion and deposition, Global Biogeochem. Cycles, 17(2), 1074, doi:10.1029/ 2002GB002010. Liu, Y., K. Tang, X. Cha, and R. Shi (1990), Experiment on the soil and water loss of slope land with different ground cover, Acta Conserv. Soli Aquae Sin., 4, 25 – 29. Liu, Z., L. Jia, Q. Xing, Y. Dai, Y. Liu, H. Li, J. Wu, and B. Sen (1993), Aboveground biomass in Erdos Plateau, Inner Mongolia, in Dynamic Monitoring Studies of Husbandry in Northern China Grasslands, edited by B. Li, Sci. Press, Beijing. Lomander, A., T. Katterer, and O. Andren (1998a), Carbon dioxide evolution from top- and subsoil as affected by moisture and constant and fluctuating temperature, Soil Biol. Biochem., 30, 2017 – 2022. Lomander, A., T. Katterer, and O. Andren (1998b), Modelling the effects of temperature and moisture on CO2 evolution from top- and subsoil using a multi-compartment approach, Soil Biol. Biochem., 30, 2023 – 2030. Loreau, M., N. Mouquet, and R. D. Holt (2003), Meta-ecosystems: a theoretical framework for a spatial ecosystem ecology, Ecol. Lett., 6, 673 – 679. Lu, Z., Y. Liang, and G. Liu (1997), Eco-Agricultural in Loess Plateau, Shaanxi Sci. and Technol. Press, Xi’an, China. Ludwig, J. A., B. P. Wilcox, D. D. Breshears, D. J. Tongway, and A. C. Imeson (2005), Vegetation patches and runoff-erosion as interacting ecohydrological processes in semiarid landscapes, Ecology, 86, 288 – 297. Ma, Y., W. Yang, and X. Yang (1990), A study on the water conditions and productivity of Robinia psedoacacia in hilly-gully northern Shaanxi, China (in Chinese), Bull. Soil Water Conserv., 10, 71 – 77. McGuire, A. D., J. M. Melillo, L. A. Joyce, D. W. Kicklighter, A. L. Grace, B. I. Moore, and C. J. Vorosmarty (1992), Interactions between carbon and nitrogen dynamics in estimating net primary productivity for potential vegetation in North America, Global Biogeochem. Cycles, 6, 101 – 124. McGuire, A. D., L. A. Joyce, D. W. Kicklighter, J. M. Melillo, G. Esser, and C. J. Vorosmarty (1993), Productivity response of climax temperate forests to elevated temperature and carbon dioxide: A North American comparison between two global models, Clim. Change, 24, 287 – 310.

15 of 16

G04005

GAO ET AL.: IMPACT OF SOIL EROSION ON CARBON BUDGET

Medlyn, B. E., et al. (1999), Effects of elevated CO2 on photosynthesis in European forest species: a meta-analysis of model parameters, Plant Cell Environ., 22, 1457 – 1495. Murty, D., M. U. F. Kirschbaum, R. E. McMurtrie, and H. McGilvray (2002), Does conversion of forest to agricultural land change soil carbon and nitrogen, a review of the literature, Global Change Biol., 8, 105 – 123. Naeth, M. A., and D. S. Chanasyk (1996), Runoff and sediment yield under grazing in foothills fescue grasslands of Alberta, Water Resour. Bull., 32, 89 – 95. National Soil Survey (1996), China Soil Species Data Book, China Agric., Beijing. Neilson, R. P. (1986), High-resolution climatic analysis and southwest biogeography, Science, 232, 27 – 34. Neilson, R. P. (1995), A model for predicting continental-scale vegetation distribution and water balance, Ecol. Appl., 5, 362 – 385. Oren, R., et al. (2001), Soil fertility limits carbon sequestration by forest ecosystems in a CO2-enriched atmosphere, Nature, 411, 469 – 472. Parton, W. J., D. S. Schimel, C. V. Cole, and D. S. Ojima (1987), Analysis of factors controlling soil organic matter levels in Great Plains grasslands, Soil Sci. Soc. Am. J., 51, 1173 – 1179. Parton, W. J., A. R. Mosier, and D. S. Schimel (1988), Dynamics of C, N, P, and S in grassland soils: a model, Biogeochemistry, 5, 109 – 131. Parton, W. J., J. M. O. Scurlock, D. S. Ojima, D. S. Schimel, and D. O. Hall (1995), Impact of climate-change on grassland production and soil carbon worldwide, Global Change Biol., 1, 13 – 22. Peng, C., and M. J. Apps (1999), Modelling the response of net primary productivity (NPP) of boreal forest ecosystems to changes in climate and fire disturbance regimes, Ecol. Modell., 122, 175 – 193. Peng, C. H. (2000), From static biogeographical model to dynamic global vegetation model: a global perspective on modelling vegetation dynamics, Ecol. Modell., 135, 33 – 54. Piao, S., J. Fang, L. Zhou, Q. Guo, M. Henderson, W. Ji, Y. Li, and S. Tao (2003), Interannual variations of monthly and seasonal normalized difference vegetation index (NDVI) in China from 1982 to 1999, J. Geophys. Res., 108(D14), 4401, doi:10.1029/2002JD002848. Piao, S., J. Fang, W. Ji, Q. Guo, J. Ke, and S. Tao (2004), Variation in a satellite-based vegetation index in relation to climate in China, J. Veg. Sci., 15, 219 – 226. Polley, H. W., H. B. Johnson, H. S. Mayeux, C. R. Tischler, and D. A. Brown (1996), Carbon dioxide enrichment improves growth, water relations and survival of droughted honey mesquite (Prosopis Grandulosa) seedlings, Tree Physiol., 16, 817 – 823. Polyakov, V., and R. Lal (2004), Modeling soil organic matter dynamics as affected by soil water erosion, Environ. Int., 30, 547 – 556. Prentice, C., W. Cramer, S. P. Harrison, R. Leemans, R. A. Monserud, and A. M. Solomon (1992), A global biome model based on plant physiology and dominance, soil properties and climate, J. Biogeogr., 19, 117 – 134. Quinton, J. N., J. A. Catt, G. A. Wood, and J. Steer (2006), Soil carbon losses by water erosion: Experimentation and modeling at field and national scales in the UK, Agric. Ecosyst. Environ., 112, 87 – 102. Raich, J. W., E. B. Rastetter, J. M. Melillo, D. W. Kicklighter, P. A. Stevdler, B. J. Peterson, A. L. Grace, B. I. Moore, and C. J. Vorosmarty (1991), Potential net primary productivity in South America: Application of a global model, Ecol. Appl., 1, 399 – 429. Rastetter, E. B., J. D. Aber, D. P. C. Peters, D. S. Ojima, and I. C. Burke (2003), Using mechanistic models to scale ecological processes across space and time, Bioscience, 53, 68 – 76. Reid, K. D., B. P. Wilcox, D. D. Breshears, and L. MacDonald (1999), Runoff and erosion in a Pinon-Juniper woodland: Influence of vegetation patches, Soil Sci. Soc. Am. J., 63, 869 – 1879. Reynolds, J. F., D. W. Hilbert, and P. R. Kemp (1993), Scaling ecophysiology from the plant to the ecosystem: A conceptual framework, in Scaling Processes Between Leaf and the Globe, edited by J. Ehleringer and C. Field, pp. 127 – 140, Academic, New York. Reynolds, J. F., R. J. Fernandez, and P. R. Kemp (1999a), Drylands and global change: Rainfall variability and sustainable rangeland production, in Proceedings of the 12th Toyota Conference: Challenge of Plant and Agricultural Sciences to the Crisis of Biosphere on the Earth in the 21st Century, edited by K. N. Watanabe and A. Komanine, pp. 73 – 86, R. G. Landes, Austin, Tex. Reynolds, J. F., R. A. Virginia, P. R. Kemp, A. G. D. Soyza, and D. C. Tremel (1999b), Impact of drought on desert shrubs: effects of seasonality and degree of resource island development, Ecol. Monogr., 69, 69 – 106. Running, S. W. (1994), Testing FOREST-BGC ecosystem process simulations across a climatic gradient in Oregon, Ecol. Appl., 4, 238 – 247.

G04005

Rupp, T. S., A. M. Starfield, and F. S. Chapin (2000), A frame-based spatially explicit model of subarctic vegetation response to climatic change: comparison with a point model, Landscape Ecol., 15, 383 – 400. Ryel, R. J., M. M. Caldwell, C. K. Yoder, D. Or, and A. J. Leffler (2002), Hydraulic redistribution in a stand of Artemisia tridentata: evaluation of benefits to transpiration assessed with a simulation model, Oecologia, 130, 173 – 184. Schimel, D. S., P. Vemap, and B. H. Braswell (1997), Continental scale variability in ecosystem processes: Model, data, and the role of disturbance, Ecol. Monogr., 67, 251 – 271. Schimel, D., et al. (2000), Contribution of increasing CO2 and climate to carbon storage by ecosystems in the United States, Science, 287, 2004 – 2006. Schimel, D. S., et al. (2001), Recent patterns and mechanisms of carbon exchange by terrestrial ecosystems, Nature, 414, 169 – 172. Schlesinger, W. H., J. F. Reynolds, G. L. Gunningham, L. F. Hueneke, W. M. Jarrell, R. A. Virginia, and W. G. Whitford (1990), Biological feedbacks in global desertification, Science, 247, 1043 – 1048. Smith, S. V., R. O. Sleezer, W. H. Renwick, and R. Buddemeier (2005), Fates of eroded soil organic carbon: Mississippi basin case study, Ecol. Appl., 15, 1929 – 1940. Starr, G. C., R. Lal, R. Malone, D. Hothem, L. Owens, and J. Kimble (2000), Modeling soil carbon transported by water erosion processes, Land Degrad. Dev., 11, 83 – 91. Steffen, W. L., W. Cramer, M. Ploechl, and H. Bugmann (1996), Global vegetation models: Incorporating transient changes to structure and composition, J. Veg. Sci., 7, 321 – 328. Tague, C. L., and L. E. Band (2001), Evaluating explicit and implicit routing for watershed hydro-ecological models of forest hydrology at the small catchment scale, Hydrol. Processes, 15, 1415 – 1439. Tan, Z. X., R. Lal, and K. D. Wiebe (2005), Global soil nutrient depletion and yield reduction, J. Sustainable Agric., 26, 123 – 146. Tchir, T. L., E. A. Johnson, and K. Miyanishi (2004), A model of fragmentation in the Canadian boreal forest, Can. J. For. Res., 34, 2248 – 2262. Thornley, J. H. M., and I. R. Johnson (1990), Plant and Crop Modelling, Clarendon, Oxford, U. K. Van Oost, K., G. Govers, T. A. Quine, G. Heckrath, J. E. Olesen, S. De Gryze, and R. Merckx (2005), Landscape-scale modeling of carbon cycling under the impact of soil redistribution: The role of tillage erosion, Global Biogeochem. Cycles, 19, GB4014, doi:10.1029/2005GB002471. Weir, J. M. H., E. A. Johnson, and K. Miyanishi (2000), Fire frequency and the spatial age mosaic of the mixed-wood boreal forest in western Canada, Ecol. Appl., 10, 1162 – 1177. Weltz, M. A., and W. H. Blackburn (1995), Water budget for south Texas rangelands, J. Range Manage., 48, 45 – 52. West, T. O., and M. K. Wali (2002), Modeling regional carbon dynamics and soil erosion in disturbed and rehabilitated ecosystems as affected by land use and climate, Water Air Soil Pollut., 138, 141 – 163. Xiao, X., D. W. Kicklighter, J. M. Melillo, A. D. McGuire, P. H. Stone, and A. P. Sokolov (1997), Linking a global terrestrial biogeochemical model and a 2-dimensional climate model: Implications for the global carbon budget, Tellus, Ser. B, 49, 18 – 37. Xiao, Y. (1990), Comparative studies on biomass and productivity of Pinus Tabulaeformis plantations in different climatic zones in Shaanxi province (in Chinese with Engilsh abstract), Acta Phytoecol. Geobot. Sin, 1990, 237 – 246. Xing, Q., and X. Lu (1993), Aboveground biomass of typical steppes in Inner Mongolia, in Dynamic Monitoring Studies of Husbandry in Northern China Grasslands, edited by B. Li, Sci. Press, Beijing. Yu, M., Q. Gao, Y. Liu, H. Xu, and P. Shi (2002a), Responses of vegetation structure and primary production of a forest transect in eastern China to global change, Global Ecol. Biogeogr., 11, 223 – 236. Yu, M., Q. Gao, and M. J. Shaffer (2002b), Simulating interactive effects of symbiotic nitrogen fixation, carbon dioxide, elevation, and climatic change on legume growth, J. Environ. Qual., 31, 634 – 641. Yuan, C. J., and D. Li (2001), Biomass and net primary productivity of aboveground parts of artificial Hyppophae rhamnoides L. stand in Ansai (in Chinese), J. Plant Resour. Environ., 10, 38 – 41.























Q. Gao, Y. Liu, and X. Xu, Integrated Landscape Analysis and Modeling Group, State Key Lab of Earth Surface Processes and Resources Ecology, College of Resources Science and Technology, Beijing Normal University, 100875 Beijing, China. ([email protected]; [email protected]) H. Xu, National Climate Research Center of China, China. M. Yu, Institute of Botany, Chinese Academy of Sciences, China.

16 of 16