Massachusetts Institute of Technology
Joint Program on the Science and Policy of Global Change

--> Home Page | Report Series |

Report #12: Net primary production of terrestrial ecosystems in China and its equilibrium responses to changes in climate and atmospheric CO₂ concentration

X. Xiao ^1,2, J.M. Melillo ¹, D.W. Kicklighter ¹, Y. Pan ¹, A.D. McGuire ³ and J. Helfrich ¹

October 1996

¹ The Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA 02543

² Joint Program on the Science and Policy of Global Change, MIT, Cambridge, MA 02139

³ Institute of Arctic Biology, University of Alaska, Fairbanks, AK 99775

Corresponding author: Dr. Xiangming Xiao xiangming.xiao@unh.edu

View Report as PDF file (65 kB)

Abstract We used the Terrestrial Ecosystem Model (TEM, version 4.0) to estimate net primary production (NPP) in China for contemporary climate and NPP responses to elevated CO₂ and climate changes projected by three atmospheric general circulation models (GCMs): Goddard Institute for Space Studies (GISS), Geophysical Fluid Dynamic Laboratory (GFDL) and Oregon State University (OSU). For contemporary climate at 312.5 ppmv CO₂, TEM estimates that China has an annual NPP of 3,653 TgC yr^-1 (1012 gC yr^-1). Temperate broadleaf evergreen forest is the most productive biome and accounts for the largest portion of annual NPP in China. The spatial pattern of NPP is closely correlated to the spatial distributions of precipitation and temperature.

Annual NPP of China is sensitive to changes in CO₂ and climate. At the continental scale, annual NPP of China increases by 6.0% (219 TgC yr^-1) for elevated CO₂ only (519 ppmv CO₂). For climate change with no change in CO₂, the response of annual NPP ranges from a decrease of 1.5% (54.8 TgC yr^-1) for the GISS climate to an increase of 8.4% (306.9 TgC yr^-1) for the GFDL-q climate. For climate change at 519 ppmv CO₂, annual NPP of China increases substantially, ranging from 18.7% (683 TgC yr^-1) for the GISS climate to 23.3% (851 TgC yr^-1) for the GFDL-q climate. Spatially, the responses of annual NPP to changes in climate and CO₂ vary considerably within a GCM climate. Differences among the three GCM climates used in the study cause large differences in the geographical distribution of NPP responses to projected climate changes. The interaction between elevated CO₂ and climate change plays an important role in the overall response of NPP to climate change at 519 ppmv CO₂.

Contents

1. Introduction

2. Model and Data

2.1 The Terrestrial Ecosystem Model (TEM)

2.2 Spatial datafor extrapolation of TEM

2.3 Scenarios of climate change and elevated CO₂

3. Net Primary Production Under Contemporary Climate at312.5 ppmv CO₂

4. Equilibrium Responses of Net Primary Production to Changes in Climate and CO₂

4.1 NPP response to climate change only

4.2 NPP response to elevated CO₂ only

4.3 NPP response to changes in climate and CO₂

5. Discussion

6. Acknowledgments

7. References

1. Introduction | Contents |

Atmospheric concentrations of CO₂ and other long-lived greenhouse gases (e.g., N2O) will continue to increase in the next century as the result of increasing anthropogenic emissions of these trace gases. Increasing greenhouse gases will further increase the radiative forcing of climate. For doubled CO₂, a number of atmospheric general circulation models (GCMs) estimate that globally averaged surface air temperature will increase in the range of 1.5 degrees C to 4.5 degrees C (Mitchell, et al., 1990). Globally, there are also large geographical variations in the changes of precipitation and clouds projected by GCMs. The increasing atmospheric CO₂ concentration and resultant climate change are likely to have significant impacts on net primary production of terrestrial ecosystems (Melillo, et al., 1990). Net primary production of terrestrial ecosystems is important in estimating land carrying capacity, which is critically relevant to the social and economic development of a country like China. China has a very large human population (about 1/5 of the world population) and limited natural resources, especially fertile lands. The total land area in China is only about 1/15 of the world land area and about 65% of this area is hills, mountains and plateau (Xiong and Li, 1988).

In China, the magnitude and spatial patterns of temperature and precipitation will change significantly according to simulations of a GCM for a doubling of atmospheric CO₂ concentration (Zhang and Wang, 1993; Wang and Zhang, 1993). Few studies have examined impacts of climate change on net primary production (NPP) of natural ecosystems in China (Zhang, 1993; Xiao, et al., 1995). Zhang (1993) applied a regression model (Chikugo model, Uchijima and Seino, 1985), which uses annual values of climate variables (e.g., temperature, precipitation, relative humidity) to estimate NPP of various ecosystems in China for contemporary climate. Zhang (1993) also calculated the responses of NPP to: (1) +2 degrees C and 20% increase of annual precipitation, and (2) +4 degrees C and 20% increase of annual precipitation in China. Net primary production increases for all the vegetation types under these climate change conditions (Zhang, 1993). However, the Chikugo model is basically a regression model based on correlation between NPP and climate, and the model has not considered the possible limitations of nutrient availability (e.g., nitrogen) on NPP. Generally, the potential of application of regression models for future projection is limited because the regressions may not be appropriate for climatic conditions that are novel to terrestrial ecosystems (Melillo, et al., 1993).

Recently, a number of processed-based ecosystem models, which are integrated with geographically referenced spatial data, have been applied to estimate responses of NPP and carbon storage to changes in climate and atmospheric CO₂ concentration at the global scale (Melillo, et al., 1993; McGuire, et al., 1995, 1996; Xiao, et al., 1996a) and the continental scale, e.g., South America (Raich, et al., 1991), North America (McGuire, et al., 1992, 1993) and conterminous United States (VEMAP Members, 1995). Estimates of NPP responses vary among the GCMs used in the studies across the scales of grid cell, biome, continent and the globe (Melillo, et al., 1993; VEMAP, 1995; Xiao, et al., 1996a). The process-based ecosystem models integrate key ecosystem processes such as plant photosynthesis, plant respiration, decomposition of soil organic matter and nutrient cycling, which interactively affect NPP. These ecological processes are controlled by a number of abiotic factors, e.g., water, light, temperature, soil texture and nutrients. The simultaneous interactions among the dynamics of carbon, nitrogen and water are complex and spatially variable.

In this study, we used a process-based global terrestrial biogeochemistry model, the Terrestrial Ecosystem Model (Raich, et al., 1991; McGuire, et al., 1995, 1996). First, we applied TEM to estimate the magnitude and spatial distribution of annual NPP of terrestrial ecosystems in China for contemporary climate. Second, we applied TEM to quantify the equilibrium responses of annual NPP in China to changes in atmospheric CO₂ concentration and climate projected by three GCMs. We examined the TEM simulation results across the spatial scales of grid cell, biome and continent. The analysis will help us to understand how the abiotic factors (e.g., climate, soil texture) control the magnitude and spatial distribution of NPP in China.

2. Model and Data | Contents |

2.1 The Terrestrial Ecosystem Model (TEM) | Contents |

The TEM (Figure 1) is a process-based ecosystem model that simulates important carbon and nitrogen fluxes and pools for various terrestrial biomes (Raich, et al., 1991; McGuire, et al., 1992, 1993, 1995). The TEM has been used to examine patterns of net primary production (NPP) in South America (Raich, et al., 1991) and North America (McGuire, et al., 1992) for contemporary climate. The model has also been used to estimate the responses of NPP of the terrestrial biosphere to climate change projected by four GCMs (Melillo, et al., 1993). In this study, we use version 4.0 of TEM, which was modified from TEM version 3 to improve patterns of soil organic carbon storage along gradients of temperature, moisture and soil texture (McGuire, et al., 1995, 1996). Version 4.0 of TEM has also been used to estimate the responses of NPP and carbon storage of terrestrial ecosystems to climate change in the conterminous United State (VEMAP Members, 1995; Pan, et al., 1996a) and at the global scale (McGuire, et al., 1996; Xiao, et al., 1996a, 1996b). Detailed descriptions of TEM are documented in Raich, et al. (1991) and McGuire, et al. (1992). The recent TEM modifications are presented elsewhere (McGuire, et al., 1995, 1996). To help understand how climate change and elevated CO₂ influence NPP in TEM, we review the key processes related to NPP.

Net primary production (NPP) is calculated as the difference between gross primary production (GPP) and plant respiration (R). The flux GPP is calculated at each monthly time step as follows: GPP = C f (PAR) f (LEAF) f (T) f (CO₂, H2O) f (NA) where C is the maximum rate of C assimilation, PAR is photosynthetically active radiation, LEAF is leaf area relative to maximum annual leaf area, T is temperature, CO₂ is atmospheric CO₂ concentration, H2O is water availability, and NA is nitrogen availability.

The effect of CO₂ and water availability on GPP are interrelated. The function f (CO₂, H2O) is described by the hyperbolic relationship:

f (CO₂, H2O) = Ci / (kc + Ci)

where Ci is the concentration of CO₂ within leaves of the canopy and kc is the half-saturation constant for CO₂ uptake by plants. The relationship between CO₂ concentration inside stomatal cavities (Ci) and in the atmosphere (Ca) is directly proportional to relative moisture availability (Raich, et al., 1991):

Ci = GvCa
Gv = 0.1 + (0.9 EET/PET)

where Gv is unitless multiplier that accounts for changes in leaf conductivity to CO₂ resulting from changes in moisture availability, PET is potential evapotranspiration and EET is the estimated evapotranspiration. The flux PET is calculated as a function of mean air temperature and solar radiation (Jensen and Haise, 1963). The flux EET is equal to PET in wet months but is modeled as a function of rainfall, snowmelt recharge and a change of soil moisture in dry months (Vorosmarty, et al., 1989).

The results from CO₂-enrichment studies indicate that the response of plant productivity to doubled CO₂ ranges from 20% to 50%, given adequate nutrients and water (Kimball, 1975; Gates, 1985). In TEM, the parameter kc (400 ppmv) has been chosen to increase f (CO₂, H2O) by 37% for a doubling of atmospheric CO₂ concentration from 340 ppmv to 680 ppmv, with canopy conductance equal to 1 (McGuire, et al., 1992, 1993). It is important to note that the response of GPP to doubled CO₂ is not a constant 37% for kc of 400 ppmv, because GPP calculation is also affected by f (NA), which represents the limiting effect of vegetation nitrogen status on GPP (McGuire, et al., 1992, 1993; Melillo, et al., 1993). Vegetation nitrogen status is determined by vegetation nitrogen uptake (NUPTAKE) and nitrogen from the labile nitrogen pool of vegetation (NMOBIL). The nitrogen down-regulation of GPP response to elevated CO₂ in TEM is discussed elsewhere (McGuire, et al., 1996).

Increasing photosynthetically active radiation (PAR) increases GPP hyperbolically in the following form: f (PAR) = PAR / (ki + PAR). A mean value of 314 J cm^-2d^-1 for ki from published leaf studies is used and applied to entire leaf canopies independent of vegetation types (Raich, et al., 1991). Cloudiness affects the amount of solar irradiance, including PAR, that reaches the canopy of vegetation.

The effect of air temperature on GPP is described by allowing f (T) to increase in a parabolic fashion to a grid-cell specific optimum temperature. Between the optimum temperature and a maximum vegetation-specific temperature constraint, f (T) is equal to 1.0. Above the maximum temperature constraint, f (T) declines rapidly to 0.0 (McGuire, et al., 1996). Air temperature also affects plant respiration (R). The flux R includes both maintenance respiration (R) and construction respiration (R). The flux R increases logarithmically with temperature using a Q<10> value that varies from 1.5 to 2.5 (McGuire, et al., 1992). The flux R is determined to be 20% of the difference between GPP and R (Raich, et al., 1991). Thus, changes in NPP are directly related to changes in CO₂, temperature, precipitation and cloudiness.

The simulation of TEM requires the use of soil- and vegetation-specific parameters appropriate to an ecosystem type. Although many of the vegetation-specific parameters in the model are defined from published references, some are determined by calibrating TEM to the fluxes and pool sizes of an intensively studied field site. Most of the data used to calibrate TEM for the 18 vegetation types of our global vegetation classification are documented elsewhere (Raich, et al., 1991; McGuire, et al., 1992, 1996).

2.2 Spatial data for extrapolation of TEM | Contents |

The driving variables for the spatial extrapolation of TEM are vegetation types, elevation, soil texture, monthly mean temperature, monthly precipitation, monthly mean cloudiness, and several hydrological variables (potential evapotranspiration, estimated evapotranspiration and soil moisture). Hydrological inputs for TEM are generated by a water balance model (Vorosmarty, et al., 1989) that uses the same climate, elevation, soils and vegetation data.

In this study, the spatial data used to drive TEM are derived from our global data sets (Melillo, et al., 1993; McGuire, et al., 1995, 1996; Xiao, et al., 1996a), which are organized in a grid at a 0.5 degrees (latitude) x 0.5 degrees (longitude) resolution. China has 3,852 grid cells, including six grid cells of inland water body and 26 grid cells of wetlands. The TEM does not estimate NPP and carbon storage for grid cells defined as water or wetland ecosystems, so the extrapolation of TEM to China requires the application of TEM to 3820 grid cells.

We use potential vegetation in this study. Our global classification of potential vegetation includes 18 upland biomes (Melillo, et al., 1993). In developing the global vegetation data, the Vegetation Map of China (Hou, et al., 1982) was used in digitization and aggregation. There are 16 upland biomes in China (Plate 1). Forests dominate in southern China and eastern coastal areas, while arid and semi-arid biomes prevail in northwestern China. The Qinhai-Tibet plateau is dominated by polar desert/alpine tundra.

Plate 1. Geographical distributions of potential natural vegetation in China.

For elevation, we use the NCAR/NAVY global 10-minute dataset (NCAR/NAVY, 1984) aggregated to a 0.5 degree resolution. Elevation increases from east to west in China (Plate 2a). The Qinhai-Tibet plateau has the highest elevation, mostly over 3000 m. The Mongolia Plateau and Loess plateau have elevation of 1000-1500 m.

Plate 2. Geographical distributions of elevation (m), soil texture (%, sand proportion and clay proportion) and contemporary climate, i.e., annual mean temperature (degrees C), annual precipitation (mm) and annual mean cloudiness (%).

For soil texture data, we use the FAO/CSRC/MBL data set (FAO/CSRC/MBL, undated), which is a digitization version of the FAO-UNESCO World Soil Map (FAO/Unesco, 1971) at 0.5 degree grid resolution. There are seven texture classes in the FAO/CSRC/MBL data set that represents "average" soil profiles of the three FAO texture classes. Each FAO/CSRC/MBL soil texture class defines a combination of sand, silt and clay proportions (Table 1). As shown in Plate 2b, soil texture is generally more coarse in northern China than in southern China.

Table 1. Proportions of sand, silt and clay assigned to the FAO/CSRC/MBL soil texture class.

*FAO codes: 1, coarse textured; 2, medium textured; 3, fine textured. No texture was assigned to lithosols.
(From Pan, et al., 1996.)

For contemporary climate, we use the long-term average climate data from the global climate dataset of Cramer and Leemans (Wolfgang Cramer, personal communication), which is an update of an earlier version of global climate database (Leemans and Cramer,1990) that has monthly precipitation, temperature and sunshine duration. Monthly percent cloudiness is calculated as 100 percent minus monthly percent sunshine duration. On the continental scale, annual mean temperature is about 5.8 degrees C, annual precipitation is about 661 mm and annual mean cloudiness is 48%. Most precipitation occurs in the plant growing season (April to September). Geographically, annual mean temperature decreases gradually from south to north (Plate 2c). The Qinhai-Tibet Plateau has very low annual mean temperature, because of high elevation. Annual precipitation decreases from southeast to northwest, as affected by summer monsoon (Plate 2c). Vegetation distribution in China (Plate 1) is closely correlated to spatial patterns of temperature and precipitation (Plate 2). Characteristics of the climate variables for each of the 16 biomes are listed in Table 2. Tropical savanna has the highest annual mean temperature (24.2 degrees C) and annual precipitation (1793 mm), while alpine tundra has the lowest annual mean temperature (-4.1 degrees C), and desert has the smallest amount of annual precipitation (74 mm).

2.3 Scenarios of climate change and elevated CO₂ | Contents |

We first ran TEM for contemporary climate at 312.5 ppmv CO₂ as the baseline for comparison. We used 312.5 ppmv CO₂ because it is the average value of atmospheric CO₂ concentration for the 1 x CO₂ simulations used by the GCMs described in Melillo, et al., (1993). To help separate the relative importance of elevated CO₂ and climate changes, we then ran TEM under (1) elevated CO₂ only, (2) climate change only, and (3) climate change plus elevated CO₂. For each of the four simulations, we ran TEM to steady state, therefore, its estimates of annual NPP apply only to mature, undisturbed vegetation and ecosystems. We have not considered the impacts of changes in land use and land cover on NPP in this study.

For climate change predictions, we used climate outputs for "current" (1 x CO₂) and "doubled" (2 x CO₂) simulations from three GCMs, i.e., Geophysical Fluid Dynamic Laboratory (GFDL-q, Weatherald and Manabe, 1988), Goddard Institute for Space Studies (GISS, Hansen, et al., 1983, 1984) and Oregon State University (OSU, Schlesinger and Zhao, 1989). These GCMs have very coarse spatial resolutions, i.e., 10.0 degrees (longitude) x 7.83 degrees (latitude) for GISS, 7.5 degrees x 4.44 degrees for GFDL and 5.0 degrees x 4.00 degrees for OSU. In order to be consistent with the spatial resolution of the contemporary climate data, the outputs of the GCMs were interpolated to a 0.5 degree x 0.5 degree grid resolution by applying a spherical interpolation routine (Willmott, et al., 1985) to the data.

We calculated the differences in monthly mean temperature and the ratios in monthly precipitation and mean monthly cloudiness between the 2 x CO₂ and 1 x CO₂ simulations of the GCMs. The OSU climate has the smallest changes in annual mean temperature and annual precipitation, while the GISS climate has the largest changes of annual mean temperature, annual precipitation and annual mean cloudiness. The projected climate changes are the most uniform across China in the OSU climate, whereas the GISS climate has large spatial variations in the magnitude of projected climate changes (Plate 3). Spatial distributions of projected changes in annual mean temperature, annual precipitation and annual mean cloudiness vary among the GCMs (Plate 3).

Plate 3. Geographical distributions of climate changes estimated by three GCMs for a doubled CO₂. (a) Changes (degrees C) in annual mean temperature. (b) Changes (%) in annual precipitation. (c) Changes (%) in annual mean cloudiness.

The "current" (1 x CO₂) climates simulated by GCMs differ considerably from the observed current climate. Therefore, in generating a "future climate," the differences in monthly mean temperature between the 2 x CO₂ and 1 x CO₂ simulations of a GCM are added to the contemporary monthly temperature data, and the ratios in monthly precipitation and cloudiness between the 2 x CO₂ and 1 x CO₂ simulations of a GCM are multiplied by the contemporary monthly precipitation and cloudiness data, respectively.

For elevated CO₂ level, we used an atmospheric CO₂ concentration of 519 ppmv, which corresponds to an effective CO₂ doubling for the radiative forcing, instead of using 625 ppmv CO₂ (Melillo, et al., 1993). An "effective CO₂ doubling" has been defined as the combined radiative forcing of all greenhouse gases having the same radiative forcing as doubled CO₂ (Rosenzweig and Parry, 1994). According to recent studies, CO₂ will be the dominant long-lived greenhouse gas in the next century and its added radiative forcing contributes between 76% and 84% of the total additional radiative forcing (IPCC, 1995). Similar to another study (Xiao, et al., 1996a), we assume that CO₂ contributes 76% of the total radiative forcing in this study, as projected by an economic-emission model, i.e., Emission Prediction and Policy Analysis model (Prinn, et al., 1996).

3. Net Primary Production Under Contemporary Climate at 312.5 ppmv CO₂

At the continental scale, TEM estimated that total annual NPP of terrestrial ecosystems in China for contemporary climate at 312.5 ppmv CO₂ is 3,653 TgC yr^-1 (10^12 gC yr^-1). Among the 16 biomes, annual NPP for a biome ranges from 4 TgC yr^-1 in tropical savanna to 1,274 TgC yr^-1 in temperate broadleaf evergreen forest (Table 3). Temperate forests (temperate coniferous forest, mixed forest, deciduous forest and broadleaf evergreen forest) account for 71.1% of the total annual NPP of China, but only cover about 35.6% of the total land area of China. Tropical forests (tropical deciduous forest and evergreen forest) occur over 4.7% of the total land area, and their annual NPP accounts for 9.1% of the total annual NPP of China. Arid and semi-arid ecosystems (desert, shrubland, shortgrassland and tallgrassland) occur over 28.5% of the total land area, but only account for 6.5% of the total annual NPP of China.

The relative importance of a biome to the total annual NPP of China reflects the relative productivity among biomes. Area-weighted mean biome NPP varies considerably among the 16 biomes (Table 3). Desert has the lowest mean biome NPP of 57 gC m^-2yr^-1. In desert and arid shrublands, NPP is controlled primarily by water availability, where annual precipitation is only 74 mm and 124 mm, respectively. In alpine desert/alpine tundra and wet/moist tundra, NPP is controlled primarily by low temperature, which results in a short plant growing season and low nitrogen availability because of lower rate of net nitrogen (N) mineralization (NMIN). Temperate broadleaf evergreen forest is the most productive biome with a mean biome NPP of 890 gC m^-2yr^-1. In general, NMIN is higher in warm temperate and tropical biomes, but lower in high latitude and plateau biomes (Table 3). This indicates that mean biome NPP is closely related to annual net nitrogen mineralization rate in soils.

Spatial variation of NPP within a biome differs among the 16 biomes (Table 3). Desert has the largest spatial variation in NPP with a coefficient of variation of 63%. In contrast, tropical deciduous forest and evergreen forest have small spatial variations in NPP with a coefficient of variation of 10%. Spatial variation of NPP within a biome is primarily determined by spatial variations of climate. As shown in Table 2, spatial variations in temperature and precipitation within a biome are large and differ among the biomes. The coefficient of variation in annual precipitation is large in desert, xeromorphic forest and arid shrubland, but small in tropical forests (Table 2).

At the scale of grid cell, annual NPP is significantly correlated with annual precipitation (Figure 2), annual mean temperature (Figure 3) and annual mean cloudiness (Figure 4). Annual precipitation accounts for 60% of the total variation in annual NPP among the 3,820 pixels, annual mean temperature accounts for 52%, and annual mean cloudiness accounts for 52%, respectively. Annual precipitation (P) and annual mean temperature (T) together accounts for 70% of the total variation in annual NPP (NPP = 54.0 + 0.35 P + 1.76 T, r^2 = 0.70, p <0.001). Annual precipitation, annual mean temperature and annual mean cloudiness (C) together account for 72% of the total variation in annual NPP (NPP = -127.1 + 0.26 P+ 16.2 T + 5.2 C, r^2 = 0.72, p <0.001).

Geographically, annual NPP decreases from southeast to northwest (Plate 4a). Southern China has high annual NPP of more than 1,200 gC m^-2yr^-1. The spatial patterns of NPP are correlated with the spatial patterns of precipitation and temperature (Plate 3). Annual NPP in Qinhai-Tibet Plateau is very low because of low temperature and the short plant growing season. Annual NPP in Xinjian, Ninxia and Inner Mongolia is also very low, because of the small amount of annual precipitation.

Plate 4. Geographical distributions of annual net primary production (NPP) for contemporary climate at 312.5 ppmv CO₂ and its responses to changes in climate and CO₂. (a) Annual NPP (gC m^-2 yr^-1) for contemporary climate at 312.5 ppmv CO₂. (b) Changes (gC m^-2 yr^-1) in annual NPP for climate change at 312.5 ppmv CO₂. (c) Changes in annual NPP for contemporary climate at 519 ppmv CO₂. (d) Changes in annual NPP for climate change at 519 ppmv CO₂. The change in annual NPP is defined as the difference in NPP estimates between the scenarios with either elevated CO₂ or climate change or elevated CO₂ plus climate change and the scenario of contemporary climate at 312.5 ppmv CO₂. The positive values indicate an increase of annual NPP; negative values indicate a decrease of annual NPP. The color bar in the center of the plate is for panels (b), (c) and (d).

4. Equilibrium Responses of Net Primary Production to Changes in Climate and CO₂

4.1 NPP response to climate change only | Contents |

For climate change with 312.5 ppmv CO₂, TEM estimates that the continental response of annual NPP in China ranges from a decrease of 54.8 TgC yr^-1 (-1.5%) for the GISS climate to an increase of 306.9 TgC yr^-1 (8.4%) for the GFDL-q climate (Table 4). These differences in NPP response are determined by the differences among the three GCM climates. According to the GCMs projections, changes in annual mean temperature in China ranges from an increase of 2.9 degrees C for the OSU climate to an increase of 4.2 degrees C for the GISS climate (Table 5). Annual precipitation increases considerably but varies little among the GCMs, i.e. 15.5% for the GFDL-q climate to 17.6% for the GISS climate (Table 5). However, annual mean cloudiness declines slightly, ranging from -0.2% for the OSU climate to -1.2% for the GFDL-q climate (Table 5).

In TEM, climate change affects NPP in a number of ways. Temperature change will affect plant photosynthesis. Elevated temperature will increase potential evapotranspiration (PET) and plant respiration. Higher PET will generally result in lower soil moisture, which enhances water stress of plants. Higher plant respiration and stronger water stress of plants may reduce NPP. Elevated temperature will enhance decomposition of soil organic matter, which releases more mineralized N available for plant uptake. Enhanced uptake of N by plants may increase NPP. An increase of precipitation may increase soil moisture and thus alleviate water stress of plant, and it also enhances decomposition of soil organic matter. A decrease of cloudiness will increase amount of total solar radiation and photosynthetic active radiation (PAR) that reaches the canopy of vegetation. Higher PAR may increase GPP and NPP. Higher total solar radiation will increase PET. As described earlier, higher PET may enhance water stress of plants, which leads to a decrease of GPP and NPP.

At the biome scale, the response of annual NPP varies substantially within a GCM (Table 4). For the GISS climate, biome NPP ranges from a decrease of 13.9% in temperate coniferous forest to an increase of 28.0% in polar desert/alpine tundra. The TEM estimates that annual NPP in tundra and boreal forest increase substantially for the three GCM climates, because all the GCMs project large increases of temperature and precipitation in these biomes (Table 5). Elevated temperature in tundra and boreal regions will significantly prolong the plant growing season and increase nutrient availability from enhanced decomposition of soil organic matter. Annual NPP for arid biomes (desert and arid shrubland) decreases moderately, except for arid shrubland in the GISS climate, in which there are large increases of annual precipitation and annual mean cloudiness (Table 5). Annual NPP of tropical deciduous forest and evergreen forest increases moderately for the OSU climate but declines for both the GISS and GFDL-q climate. These responses are attributable to the larger increases of annual precipitation and annual mean cloudiness as well as a smaller increase in annual mean temperature in the OSU climate (Table 5). The most significant differences in annual NPP response among the GCM climates occur in temperate broadleaf evergreen forest, i.e., -135 TgC yr^-1 (-10.6%) for the GISS climate, 96 TgC yr^-1 (7.5%) for the OSU climate and 107 TgC yr^-1 (8.4%) for the GFDL-q climate. These differences can be attributed to the substantial increase of annual mean temperature (4.8 degrees C), large decrease of annual mean cloudiness (-10.9%) and slight decrease of annual precipitation (-1.2%) in the GISS climate (Table 5). In contrast, both the GFDL-q and OSU project smaller increases in temperature and larger increases in precipitation (Table 5).

The response of NPP to climate change has large regional variations within a GCM climate (Plate 4b). For the GISS climate, NPP decreases substantially in Southern China but increase in Northeastern China. In Southern China, the GISS GCM predicts the largest increase of annual mean temperature and a decrease of annual precipitation and annual mean cloudiness (Plate 3). For the three GCMs climate, annual NPP decreases less than 20 gC m^-2yr^-1 in the arid and semi-arid regions in the Northwestern China, including the Loess Plateau. The Qinhai-Tibet Plateau has an increase of annual NPP less than 40 gC m^-2yr^-1. However, the geographical distribution of NPP response in Southern China varies considerably among the GCM climates (Plate 4b). In the GISS and OSU climates, annual NPP decreases significantly in the Chendu Plain in the Shichuan Province, where is the most productive agricultural zone and the most dense settlement of human population. In contrast, annual NPP increases in the Chendu Plain for the GFDL climate.

4.2 NPP response to elevated CO₂ only | Contents |

For contemporary climate with elevated CO₂ (519 ppmv), TEM estimated that annual NPP increases by 219 TgC yr^-1 (6.0%) in China. In TEM, a direct effect of elevated atmospheric CO₂ level is to increase the intercellular CO₂ concentration within the canopy, which potentially increases GPP via a Michaelis-Menton (hyperbolic) function. Elevated atmospheric CO₂ level may also indirectly affect GPP by altering the carbon-nitrogen status of the vegetation increase effort toward nitrogen uptake; increased effort is generally realized only when GPP is limited more by carbon availability than by nitrogen availability (VEMAP Members, 1995).

At the biome scale, annual NPP increases for all of the 16 biomes (Table 4). The NPP responses vary substantially among biomes, ranging from an increase of 1.9% in tall grasslands to an increase of 30.2% in desert (Table 4). Biomes in the arid regions (desert, arid shrubland) have the largest percentage increase in annual NPP. This is attributed mostly to a significant increase of water use efficiency of plants in the arid regions, where water is the primary limiting factor on NPP. Temperate forests and boreal forests are generally limited by the availability of inorganic nitrogen in the soils, and as the result, their responses to elevated CO₂ is strongly constrained by nitrogen availability. Tropical savanna, xeromorphic forest, tropical deciduous forest and tropical evergreen forest have large increases of annual NPP in the range of 7.5% to 18.8%. In the tropical regions, annual NPP of ecosystems is generally not limited by nitrogen, because of high rates of net N mineralization resulting from high temperatures.

The geographical distribution of NPP responses to elevated CO₂ has distinct spatial patterns (Plate 4c). Annual NPP increase from about 20 gC m^-2yr^-1 to 300 gC m^-2yr^-1 in Southeastern China. Annual NPP in arid regions increases in the range of 20 to 40 gC m^-2yr^-1. In the Qinhai-Tibet Plateau, the response of annual NPP is within ±20 gC m^-2yr^-1 (Plate 4c). Annual NPP in the Qinhai-Tibet Plateau is primarily controlled by low temperature and constrained by low nutrient availability.

4.3 NPP response to changes in climate and CO₂ | Contents |

At the continental scale, annual NPP of China for climate change at 519 ppmv CO₂ increases substantially but varies little among the GCM climates, ranging from 683 TgC yr^-1 (18.7%) for the GISS climate to 851 TgC yr^-1 (23.3%) for the GFDL-q climate. Annual NPP increases considerably for all of the 16 biomes among the three GCM climates (Table 4). Dryland ecosystems (desert, arid shrubland) have the largest percentage increases in annual NPP (more than 30%). Annual NPP in tundra and boreal forest also increases substantially (more than 20%). Although the percentage increases in annual NPP for temperate broadleaf evergreen forest are moderate, this biome accounts for the largest portion of the increase of annual NPP in China: 238.2 TgC yr^-1 for the OSU climate, 179.6 TgC yr^-1 for the GISS climate and 284.1 TgC yr^-1 for the GFDL-q climate.

Geographically, the responses of annual NPP decrease from southeast to northwest within a GCM climate (Plate 4d). The spatial patterns of NPP responses are generally similar among the three GCM climates. Annual NPP has a slight increase (<40 gC m^-2yr^-1) in the Qinhai-Tibet plateau and arid regions, a moderate increase (40 to 60 gC m^-2yr^-1) in grasslands, and a large increase (> 60 gC m^-2yr^-1) in forest regions (Plate 4d). Increases of annual NPP in forest regions are the largest in the GFDL-q climate, up to 200-300 gC m^-2yr^-1 in the southern China. For the GISS climate, NPP decreases in scattered areas of the Southern China (Plate 4). There are slight differences in both the magnitude and spatial distributions of NPP responses in the transitional zones between forests and grasslands among the three GCM climates (Plate 4).

It is important to note that at the continental scale, the response of NPP to climate change at 519 ppmv CO₂ is much larger than the sum of the NPP response to elevated CO₂ only and the NPP response to climate change only (Table 4). For the OSU climate, the sum of the NPP response for climate change at 315 ppmv CO₂ (7.5%) and the NPP response for contemporary climate at 519 ppmv CO₂ (6.0%) is much lower than the 20.0% NPP response for climate change at 519 ppmv CO₂; the difference is attributed to the interaction between climate change and elevated CO₂. Similarly, the interaction between climate change and elevated CO₂ contributes about 9.0% to the overall response of NPP to climate change at 519 ppmv CO₂ for the GFDL-q climate and 14.2% for the GISS climate. The interaction between elevated CO₂ and climate change is primarily caused by enhanced plant nitrogen uptake and increasing availability of mineralized nitrogen and carbon (CO₂) resources. These results show clearly that the interaction between elevated CO₂ and climate change contributes significantly to the overall equilibrium response of NPP to changes in climate and CO₂ (Table 4).

The relative role of the interaction between elevated CO₂ and climate change varies among the 16 biomes, ranging from a large role in arid biomes (e.g., desert, arid shrubland) to a small role in alpine tundra (Table 4). The relative role of the interaction between elevated CO₂ and climate change also varies among the three GCMs climates (Table 4). This indicates that geographical distributions of climate changes projected by the GCMs affect the interaction between elevated CO₂ and climate change at larger spatial scales.

5. Discussion | Contents |

There are numerous field measurements of plant biomass and NPP for various ecosystems in China, however, few studies have estimated the spatial distribution of NPP in China on large spatial scales. Zhang (1993) calculated annual NPP for 671 weather stations in China, using the Chikugo model (Uchijima and Seino, 1985). Fang, et al., (1995) and Xiao, et al., (1996) estimated annual NPP in China in 1990, using Normalized Difference Vegetation Index (NDVI) derived from remote sensing data of NOAA AVHRR. In our study, the TEM-estimated annual NPP of all the terrestrial ecosystems in China under contemporary climate at 312.5 ppmv CO₂ is 3,653 TgC yr^-1, about 8% of global terrestrial NPP (see Melillo, et al., 1993). Application of process-based ecosystem models such as TEM provides an important and useful tool to study spatial patterns and dynamics of NPP in China, although NPP estimates can be improved with the collection of additional field data and remote sensing data to further develop and validate ecosystem models.

The TEM results show that annual NPP in China has a large spatial variation. The geographical distribution of annual NPP in this study is similar to Zhang (1993) and Xiao, et al. (1996), characterized by a gradient of high NPP in southeastern China to low NPP in northwestern China. The spatial patterns of NPP are closely correlated to the spatial distributions of abiotic factors, particularly climate. This indicates that the spatial representation of contemporary climate and other abiotic factors used in extrapolation of TEM for China may affect estimates of NPP. In an earlier study that examines three alternative geo-referenced data sets of climate, solar radiation and soil texture in estimating NPP of the conterminous United States, Pan, et al. (1996a) found that NPP estimates of TEM are sensitive to variations in the input data, and the relative importance of climate, soil texture and solar radiation to NPP estimates differs among the vegetation types.

The spatial input data used in this study are from our global data base and need to be updated with the new data available in China. For instance, the climate data used in this study are the long-term average values of weather records from about 300 weather stations in China (W. Cramer, personal communication). Leemans and Cramer (1990) compiled weather record data mainly for the period of 1931-1960. Zhang (1993) calculated biotemperature and potential evapotranspiration using long-term average data from about 671 weather stations in China. Using more weather stations in the spatial interpolation will improve the spatial and temporal (seasonal) representations of contemporary climate.

Major soils in Xinjian, Qinhai-Tibet plateau and Inner Mongolia have generally about 40 to 70% of sand proportion in soil texture (Xiong and Li, 1987), which is higher than the 35 or 45% of sand proportion we used in this study to drive TEM (Plate 2). In arid and semi-arid regions, coarse texture soils, which allow deep infiltration of water, may have lower evaporation but higher transpiration rate than fine texture soils (Noy-Meir, 1973; Sala, et al., 1988). Primary production is therefore higher in coarse texture soils than in fine texture soils in arid and semi-arid regions (Noy-Meir, 1973). In grasslands of the Central Plains of United States, aboveground production increases with coarser soil texture when annual precipitation is less than 370 mm, but decreases when annual precipitation is greater than 370 mm (Sala, et al., 1988).

Therefore, to improve estimates of NPP in China, geographically referenced spatial data with better spatial-temporal representations are critically needed. Improved data set will allow ecosystem models such as TEM to make more realistic estimates of NPP for contemporary climate. As the contemporary climate data sets are also used in generating a "future climate" for impact analyses of climate change, improved contemporary climate data sets are likely to reduce the uncertainty in estimating responses of NPP in China to climate change.

The TEM results show that NPP of terrestrial ecosystems in China is sensitive to changes in atmospheric CO₂ concentration and climate as projected by the GCMs. These results are consistent with an earlier study (Xiao, et al., 1995) that used the CENTURY ecosystem model (Parton, et al., 1993) at individual sites and showed that NPP in grasslands of Inner Mongolia of China is sensitive to changes in climate and atmospheric CO₂ concentration. The TEM results also show that the magnitude and spatial distribution of NPP responses vary considerably among the GCMs projections. The largest uncertainty in the NPP response to climate change among the three GCM climates occurs in Southern China. The uncertainty in NPP responses is related to the uncertainty in the GCMs projections for future climate change.

There is a large uncertainty in the GCMs projections for future climate changes. A number of GCMs are available (Mitchell, et al., 1990). Most GCMs project a similar range of global surface temperature increase, but they vary significantly in the spatial patterns of projected changes in temperature and precipitation. Zhang and Wang (1993) examined the simulation results for 1 x CO₂ and 2 x CO₂ scenarios from the Community Climate Model (CCM) at the National Center for Atmospheric Research (NCAR) of the United States. Annual mean temperature in China increases by 2.5-3.0 degrees C. There is a large spatial variation in changes of annual precipitation in China. In particular, the NCAR CCM predicts a considerable decrease (about 200 mm) of annual precipitation in the semi-arid Loess Plateau, where serious soil-water erosion and land degradation occur extensively.

Because of the very coarse spatial resolutions of GCMs, GCMs are generally poor in representing regional climate change, especially in China, which has very complex topography and the climate is dominated by summer monsoon. Monsoon climate is characterized by seasonal jumps, large interannual variability and abrupt changes on decade and longer time scales. There are strong interactions between climate and ecosystems in the Asian monsoon region on various time scales (Fu, 1995a). For instance, the normalized difference vegetation index (NDVI) for southern China varied significantly in the period of 1985-1990 (Fu, 1995a). The coupling of GCMs and meso-scale climate models with detailed representations of topography and ecosystem dynamics may generate better spatial-temporal representations of climate change in China (Fu, 1995b).

In this study, we have not considered the responses of vegetation distribution to climate change and elevated CO₂. Climate change is likely to affect the spatial distribution of vegetation in China (Zhang and Liu, 1993; Zhang and Song, 1993; Zhang, 1993; Wang and Zhao, 1995; Zhou and Zhang, 1996). Vegetation redistribution can affect regional estimates of NPP of terrestrial ecosystems (VEMAP Members, 1995). How climate change and elevated CO₂ affect vegetation distribution in China and consequently NPP is discussed in another paper (Pan, et al., 1996b).

We used the potential vegetation of China in this study. Thus, we have not considered the effects of land use and management on NPP in China. In the last several thousand years of agricultural practice, a large portion of uplands has been converted to croplands. According to the Vegetation Map of China (Hou, et al., 1982), the area of croplands is approximately 1.5 million km^2, or about 15% of the total land area of China. Based on the Land Use Map of China at the scale of 1:1,000,000 (Wu, 1990), cultivated lands account for about 14.2% of the total land area of China (Wu and Guo, 1994). Conversion of forests to croplands results in significant losses of NPP and carbon storage (Houghton and Skole, 1990). Climate change is likely to have significant impacts on physical geographical zones (Zhao, 1993) and agriculture in China (Wang, 1993; Gong and Hameed, 1993; Ohta, et al., 1995). This study provides a baseline for future studies to investigate the impacts of changes in land use and land cover on NPP and carbon storage of China.

In summary, this study demonstrates the usefulness of integrating a process-based ecosystem model with geo-referenced spatial data in China for examining the spatial patterns and dynamics of net primary production of terrestrial ecosystems. However, the analyses in this study are based on an equilibrium changes in climate and carbon fluxes of terrestrial ecosystems. The time course of changes in atmospheric CO₂ concentration, climate, land use and anthropogenic nitrogen deposition from the last several centuries to next century have not been considered. We are beginning to conduct transient simulations of TEM to track the path and magnitude of the responses of NPP over time (Melillo, et al., 1996), but progress is hindered by a lack of both historical and future data with better spatial-temporal representation on climate, land use and N deposition in China. As these data sets become available in the near future, we should be able to improve the estimates of NPP and to better assess the effects of land use and climate changes on NPP of terrestrial ecosystems in China.

6. Acknowledgments | Contents |

The study was supported by the National Aeronautics and Space Administration's Earth Observing System (NAGW-2669), the Department of Energy's National Institute for Global Environmental Change (No: 901214-HAR) and the Joint Program on Science and Policy of Global Change at the Massachusetts Institute of Technology (CE-S-462041).

7. References

--> Home Page | top of page |