SCOPE 56 - Global Change: Effects on Coniferous Forests and Grasslands

15.1 INTRODUCTION

Most climate models predict that the buildup of greenhouse gases such as carbon dioxide (CO₂) and methane are likely to lead to surface air temperature rises of 1.5-4.5 °C and changes in precipitation and cloud patterns over the next century (Houghton et al. 1992). The simultaneous changes in the chemistry of the atmosphere and the climate are expected to affect both the function and the structure of terrestrial ecosystems (Melillo et al. 1990). Functional changes may include changes in processes such as photosynthesis, plant respiration and decomposition. Structural changes may be of various types, including changes in the distribution of carbon and nitrogen between the plant and soil pools, changes in the species composition within an ecosystem, and changes in the distribution of major vegetation groups or biomes.

Process-based models have been used in regional studies to evaluate the functional responses of terrestrial ecosystems to changes in climate (Schimel et al. 1990; Burke et al. 1991; Running and Nemani 1991; McGuire et al. 1992,1993). One of these models, the terrestrial ecosystem model (TEM), has been used at the global scale (Melillo et al. 1993) to explore how carbon and nitrogen cycling in terrestrial ecosystems might change according to predictions of climate change made by several general circulation models (GCMs). In this chapter, we use version 3 of TEM (McGuire et al. 1993; Melillo et al. 1993) to investigate how carbon cycling of grasslands and conifer forests might respond to GCM- predicted climate change. We focus our analysis on the equilibrium response of carbon pools in vegetation and soils and on the response of net primary production (NPP) for the potential vegetation distribution that is appropriate to contemporary climate. Our analysis does not consider how the response of vegetation distribution to climate change (Emanuel et al. 1985; Prentice 1990;

Woodward and McKee 1991; and Prentice et al. 1992) affects carbon pools or NPP. We also investigate how assumptions about soil carbon at the sites used to calibrate TEM influence the response of soil carbon to changes in CO₂ and climate. Finally, we compare the responses predicted by TEM to those of other models and evaluate the relevance of spatial scale for the comparison of model responses.

A dominant feature of GCM-predicted climate for a doubled CO₂ atmosphere is an increase in mean surface temperature of the globe (Mitchell et al. 1990); precipitation and cloudiness are expected to increase in some areas and decrease in others, and there is disagreement among the output of GCMs about the spatial distribution of these changes (Mitchell et al. 1990). Elevated temperature may affect carbon cycling in ecosystems in a variety of ways. It may enhance decomposition of soil organic matter to increase the loss of carbon from the soil. Enhanced decomposition may also increase nitrogen availability through higher rates of nitrogen mineralization. Uptake of this nitrogen by vegetation may enhance NPP. Decreases in NPP may result from elevated temperature by reducing soil moisture or by increasing respiration. Because the ability of vegetation to incorporate elevated CO₂ into production depends on the nutrient and water status of the vegetation, climate change may influence carbon cycling in ecosystems by altering nutrient availability and soil moisture. The TEM has been developed to evaluate how climate change influences simultaneous interactions among the carbon, nitrogen, and water cycles in ecosystems.

15.2 THE TERRESTRIAL ECOSYSTEM MODEL

The terrestrial ecosystem model (TEM; Figure 15.1) is a process-based ecosystem simulation model that uses spatially referenced information on climate, elevation, soils, vegetation, and water availability to make monthly estimates of important carbon and nitrogen fluxes and pool sizes (Raich et al. 1991; McGuire et al. 1992, 1993; Melillo et al. 1993). Because we use TEM to make equilibrium predictions, its estimates of carbon and nitrogen dynamics apply only to mature, undisturbed vegetation; they do not include the effects of land use.

For each monthly time step in a model run, NPP is calculated as the difference between gross primary production (GPP) and plant respiration (RA). The calculation of GPP considers simultaneous interactions among temperature, light, CO₂, water, and nitrogen availiability (McGuire et al. 1992,1993; Melillo et al. 1993). Therefore, the response of GPP to elevated CO₂ is potentially constrained by the availability of light, water, and nitrogen. The calculation of RA considers both maintenance respiration (McGuire et al. 1992, 1993) and construction respiration (Raich et al. 1991).

The data sets used to drive TEM are gridded at a resolution of 0.5° latitude by 0.5° longitude. The sources for the climate data (air temperature, precipitation, and cloudiness), elevation, vegetation, and soil texture are described elsewhere (Raich et al. 1991; Melillo et al. 1993); the climate data represent long-term averages. Hydrological inputs for TEM are determined with a water balance model (Vorosmarty et al. 1989) that uses the climate, elevation, soils, and vegetation data.

Figure 15.1 The Terrestrial Ecosystem Model (TEM). The state variables are: carbon in the vegetation (C_V); structural nitrogen in the vegetation (N_VC); labile nitrogen in the vegetation (N_VL); organic carbon in soils and detritus (C_S); organic nitrogen in soils and detritus (N_S); and available soil inorganic nitrogen (N_AV). Arrows show carbon and nitrogen fluxes; GPP, gross primary production; R_A, autotrophic respiration; RH, heterotrophic respiration; L_C, litterfall carbon; L_N, litterfall nitrogen; NUPTAKE_S, nitrogen uptake into the strutural nitrogen pool of the vegetation; NUPTAKE_L, nitrogen uptake into the labile nitrogen pool of the vegetation; NRESORB, nitrogen resorption from dying tissue into the labile nitrogen pool of the vegetation; NMOBIL, nitrogen mobilized between the structural and labile nitrogen pools of the vegetation; NETNMIN, net nitrogen mineralization of soil organic nitrogen; NINPUT, nitrogen input from outside the ecosystem; and NLOST, nitrogen losses from the ecosystem

The application of TEM to a grid cell requires the use of monthly climatic and hydrological data and the soil-and vegetation-specific parameters appropriate to the grid cell. Although many of the vegetation-specific parameters in the model are defined from published information (Raich et al. 1991; McGuire et al. 1992), some are determined by calibrating the model to the steady-state fluxes and pool sizes of an intensively studied field site. Most of the data used to calibrate the model for the vegetation types considered in this study are documented elsewhere (McGuire et al. 1992). In most of the analyses we present, TEM has been calibrated to the soil organic carbon and nitrogen found to approximately 1 m depth at the calibration site, the '1 m' calibration. We also calibrated the model to 20 cm soil carbon and nitrogen (the '20 cm' calibration) to examine how different assumptions about the actively decomposing soil organic carbon affect responses of NPP and carbon pools to changes in climate and CO₂.

Table 15.1 Areal extent of grasslands and coniferous forests

Figure 15.2 The global distribution of grasslands (tall and short) and conifer forests (boreal, temperature, and temperature mixed)

15.3 POTENTIAL DISTRIBUTION OF GRASSLANDS AND CONIFER FORESTS

The global distribution of grasslands and conifer forests (Figure 15.2), which represents 22% of the area occupied by the terrestrial biosphere (Table 15.1), is determined from a global georeferenced data base (0.5° spatial resolution) of potential vegetation developed from extant maps (Melillo et al. 1993). To develop this data base we modified the potential vegetation data of Matthews (1983), which has 1° spatial resolution, by incorporating information from several regional sources (White 1981; Institute of Geography of the Siberian Department of USSR Academy of Science et al. 1990; Hou et al. 1979; Australian Surveying and Land Information Group 1990; Rowe 1972; Joint Federal-State Land Use Planning Commission for Alaska 1973; Kiichler 1964; UNESCO 1981).

The global grassland communities are aggregated into two major vegetation types, tall and short grasslands, based on the relative height of the dominant vegetation. Tall grasslands contain grasses with heights greater than 1 m and occur in more mesic sites than short grasslands. Although the northern meadow-steppe of Euro-Asia contains relatively 'short' species, the dynamics of this community are thought to be similar to the tallgrass prairie of North America (Walter 1979). Therefore, northern meadow-steppes have been classified as tall grassland. Both grassland types are found throughout the temperate and tropical zones (Figure 15.2). We do not consider grasslands found in savannas of the temperate and tropical regions.

Table 15.2 Estimates by the TEM of annual NPP for grasslands and conifer forests in the terrestrial biosphere at an atmospheric CO₂ concentration of 355 ppmv (parameterized for 1 m soil carbon)

^a Units are Pg C( 10¹⁵ g C ) yr^-1
b Units are g Cm^-2 yr^-1

Conifer forests are aggregated into three major vegetation types based on physiognomy and climate: boreal forests, temperate mixed forests, and temperate conifer forests. Boreal forests (located mainly in Canada and Alaska, northern Europe, and the Commonwealth of Independent States) contain both conifer and deciduous dominant species and represents 62% of all potential conifer forests (Table 15.1). The next abundant forest type, temperate mixed forests (26%), also contains both conifer and deciduous dominant species and is found mainly in the United States, Europe, and China. Temperate conifer forests have a global distribution similar to temperate mixed forests, but are also found in many mountainous regions.

15.4 EXTRAPOLATION OF TEM FOR CONTEMPORARY CLIMATE

To estimate fluxes and pools of grasslands and conifer forests for 'contemporary' conditions we applied TEM at 355 ppmv CO₂ using the long-term climate data with the 1 m calibration. For grasslands, TEM estimates an annual NPP of 2.2 Pg C (10¹⁵ g C; Table 15.2). The vegetation and soil carbon estimates for grassland are 3.4 Pg C (Table 15.3) and 75.7 Pg C (Table 15.4). Although conifer forests in our analysis occupy 2.4 times the area of grasslands, the estimated annual NPP is 3.4 times that of grasslands (7.4 Pg C; Table 15.2), vegetation carbon is 84.3 times greater (286.7 Pg C; Table 15.3), and soil carbon is 3.1 times greater (231.4 Pg C; Table 15.4). The higher vegetation carbon of conifer forests reflects the ability of forests to store carbon in woody tissue. Per unit area, TEM estimates that total carbon storage in global conifer forests is 2.8 times that in global grasslands.

Table 15.3 Estimates by the TEM of vegetation carbon for grasslands and conifer forests in the terrestrial biosphere at an atmospheric CO₂ concentration of 355 ppmv (parameterized for 1 m soil carbon)

^a Unites are Pg C( 10¹⁵ g C ) yr^-1
b Unites are g C m^-2 yr^-1

Table 15.4 Estimates by the TEM of soil carbon for grasslands and conifer forests in the terrestrial biosphere at an atmospheric CO₂ concentration of 355 ppmv (parameterized for 1 m soil carbon)

^a Unites are Pg C( 10¹⁵ g C ) yr^-1
b Unites are g Cm^-2 yr^-1

Because of the way we aggregated grasslands and forests, our estimates of total NPP and carbon pools for grasslands are not always directly comparable to other published estimates. However, the area-based estimates by TEM may be appropriately compared. The grassland NPP estimate of 267 g C m^-2 yr^-1 by TEM (Table 15.2) is similar to the estimate of 225 g C m^-2 yr^-1 by Whit taker and Likens (1973) for temperate grasslands. However, the vegetation carbon estimate of 413 g C m^-2 (Table 15.3) is substantially lower than their estimate of 700 g C m^-2. The soil carbon estimate of 9156 g C m^-2 by TEM for grasslands (Table 15.4) is similar to 10692 g C m^-2 (417 Pg / 3900 106 ha) reported by Ojima et al. (1993a) for soil carbon to 1 m depth in potential world grasslands.

The TEM estimate of 238 g C m^-2 yr^-1 for boreal forest NPP (Table 15.2) is substantially lower than the Whittaker and Likens (1973) estimate of 360 g C m^-2 yr^-1. However, the estimate for boreal forest vegetation carbon (9739 g C m^-2; Table 15.3) is similar to their estimate of 9000 g C m^-2. The TEM estimate of 10878 g C m^-2 for boreal forest soil carbon (Table 15.4)is lower than the 14900 g C m^-2 estimate of Schlesinger (1977); the TEM estimate is lower because the model does not represent anaerobic processes that cause higher carbon storage to occur in northern peatlands (see Schlesinger 1977).

The Whittaker and Likens (1973) NPP estimate of 585 g Cm^-2 yr^-1 for temperate evergreen forest is intermediate between the TEM estimates for temperate conifer and temperate mixed forest (Table 15.2). Their estimate of 16000 g Cm^-2 for vegetation carbon of temperate evergreen forest is similar to the TEM estimate of 15224 g Cm^-2 for temperate mixed forest (Table 15.3); the TEM estimate for temperate conifer forest (37 803 g Cm^-2; Table 15.3) is much higher because the calibration site is an old-growth forest in the Pacific Northwest of the United States (McGuire et al. 1992). The Schlesinger (1977) estimate of 11 800 g Cm^-2 for temperate forest soil carbon is similar to the TEM estimate for temperate mixed forest (10612 g Cm^-2; Table 15.4). Again the TEM estimate for temperate conifer forest (18804 g Cm^-2; Table 15.4) may be higher because the calibration site is an old-growth forest.

15.5 DEVELOPMENT' OF FUTURE CLIMATE SCENARIOS FOR USE WITH TEM

We obtained the output of two general circulation models (GCMs) from the National Center for Atmospheric Research (Jenne 1992). The simulations estimate equilibrium climates that correspond to a doubling of the atmospheric CO₂ concentration and include one from the Geophysical Fluid Dynamics Laboratory (GFDL; Manabe and Wetherald 1987) and one from Oregon State University (OSU; Schlesinger and Zhao 1989). The GFDL GCM represents a high temperature impact scenario and predicts increases of 4.9 °C for the grassland biome and 6.2 °C for the conifer forest biome. The OSU GCM represents a low impact scenario and predicts increases of 3.2 and 3.4 °C for the grassland and conifer forest biomes. Precipitation for grasslands is predicted to increase by both GCMs with larger increases predicted by OSU (82.0 mm) than by GFDL (46.5 mm). Precipitation is also predicted to increase for conifer forests, with similar increases predicted by GFDL (67.7 mm) and OSU (75.9 mm). Cloudiness is predicted by GFDL and OSO to decrease 1.4 and 2.1% for grasslands, respectively. For conifer forests, cloudiness is predicted by GFDL to increase 0.6%, but by OSO to decrease 2.5%.

Because we were interested in the implications of climate change for NPP and carbon pools, we generated 'GCM climates' for TEM by using the output variables of surface air temperature, precipitation, and total cloud cover for the current and 2 X CO₂ simulations of each GCM to modify the contemporary climate data for TEM. We organized each of the output variables of each GCM simulation at 0.5° spatial resolution with a spherical interpolation procedure (Willmott et al. 1985). Next, similar to the method used in a study of the potential effects of climate change on US agriculture (Adams et al. 1990), we calculated for each grid cell the ratio of the monthly output of the 2 X CO₂ simulation to that of the 1 X CO₂ simulation for each of the three output variables; temperature was converted to Kelvin before calculating monthly temperature ratios. We then multiplied each ratio by the corresponding variable in our data for contemporary climate to determine the input data for TEM that represent the 2 X CO₂ climate for each GCM.

To help separate the effects of changes in CO₂ concentration from those of the GCM climates on estimates of NPP and carbon pools we performed a factorial experiment with the 1 m calibration of TEM involving two levels of CO₂ (312.5 and 625.0 ppmv) and three climate scenarios (contemporary and the two GCM climates). We chose the CO₂ level of 312.5 ppmv because it was the average baseline concentration of four GCMs that we used in a previous study (Melillo et al. 1993).

15.6 RESPONSES TO DOUBLED CARBON DIOXIDE

For doubled CO₂ and no climate change, TEM predicts that NPP, vegetation carbon, and soil carbon for individual grid cells of grasslands and conifer forests either do not change or increase (Figures 15.3-15.5). For global grasslands, TEM predicts that NPP increases 0.2 Pg C (9.1 %; Table 15.5), vegetation carbon increases 0.3 Pg C (9.1 %; Table 15.6), and soil carbon increases 3.4 Pg C (4.3%; Table 15.7). For global conifer forests, TEM predicts NPP to increase 0.4 Pg C (5.5%; Table 15.5), vegetation carbon to increase 17.5 Pg C (6.2%; Table 15.6), and soil carbon to increase 9.0 Pg C (3.9%, Table 15.7). Thus, for doubled CO₂ and no climate change conifer forests are potentially a more responsive carbon sink than grasslands.

The response of conifer forest NPP and soil carbon to elevated CO₂ predicted by TEM depends on the degree to which NPP is limited by nitrogen availability (McGuire et al. 1993). In moist regions of temperate conifer forest, where NPP is predicted by TEM to be limited by nitrogen availability more than by soil moisture, there is little response to elevated CO₂ (McGuire et al. 1993). In dry regions elevated CO₂ promotes enhanced water-use efficiency in TEM which translates into increased NPP (McGuire et al. 1993). Because most of the conifer forest region considered in this study is moist conifer forest, i.e. in the boreal and temperate-mixed regions, the enhancement of NPP and soil carbon in response to elevated CO₂ is small.

15.7 RESPONSES TO CHANGES IN CLIMATE

With no change in CO₂ concentration, TEM predicts for both the GFDL and OSU climates that responses of NPP, vegetation carbon, and soil carbon for individual grid cells of grasslands and conifer forests can be either positive or negative (Figures 15.3-15.5). For global grasslands, TEM predicts annual NPP to increase 0.4 Pg C (18.2%) for the GFDL climate and 0.2 Pg C (9.1 %) for the OSU climate (Table 15.5). Vegetation carbon is predicted to increase 0.7 Pg C (21.2%) for the GFDL climate and 0.6 Pg C (18.2%) for the OSU climate (Table 15.6). For the GFDL climate, soil carbon decreases 3.8 Pg C(5.1 %; Table 15.7). In contrast, soil carbon is predicted to increase slightly for the OSU climate (0.7 Pg C, 0.9%; Table 15.7).

For global conifer forests, TEM predicts annual NPP to increase 1.0 Pg C (13.7%) for the GFDL climate and 0.9 Pg C (12.3%) for the OSU climate (Table 15.5). Increases in vegetation carbon are slightly higher for the GFDL climate (38.1 Pg C, 13.5%; Table 15.6) than for the OSU climate (34.4 Pg C, 12.2%; Table 15.6). The decreases in soil carbon predicted for the GFDL climate (31.3 Pg C, 13.7%; Table 15.7) are more than three times those predicted for the OSU climate (9.2 Pg C, 4.0%; Table 15.7). Thus, for the climates considered with no change in CO₂, grasslands are potentially either carbon sources or sinks and conifer forests are potentially sinks. The sink strength is greater for conifer forests than grasslands because of the ability of forests to store carbon in woody tissue.

In grasslands, and in boreal and cool-moist temperate regions of conifer forest, elevated temperature generally increases the NPP predicted by TEM through enhanced nitrogen availability (McGuire et al. 1992, 1993; Melillo et al. 1993). Schimel et al. (1990), based on 50-year simulations of climate change with the CENTURY model for sites in the Great Plains, attributed increased NPP to elevated nitrogen availability because of enhanced decomposition, but indicated that nitrogen losses related to higher decomposition could decrease NPP in the long term. Burke et al. (1991) applied the doubled-CO₂ climate predicted by the Goddard Institute for Space Studies (GISS) GCM to the central Great Plains and reported that aboveground NPP for the region increased less than 10% after 50 years of simulation with CENTURY. This result is similar to the equilibrium response predicted by TEM for the OSU climate applied to global grasslands. Linked models of forest productivity and soil processes have also predicted that elevated temperature enhances conifer growth through increased nitrogen availability for simulations at specific sites (Pastor and Post 1988; Bonan and Van Cleve 1992).

Figure 15.3 The range and mean of responses in annual NPP for grasslands (filled symbols) and conifer forests (open symbols) at two levels of atmospheric carbon dioxide (312.5 and 625.0 ppmv) and three levels of climate (contemporary, circles; GFDL, squares; OSU, triangles). The response is relative to NPP for contemporary climate and atmospheric carbon dioxide of 312.5 ppmv

Figure 15.4 The range and mean of responses in vegetation carbon for grasslands (filled symbols) and conifer forests (open symbols) at two levels of atmospheric carbon dioxide (312.5 and 625.0 ppmv) and three levels of climate (contemporary, circles; GFDL, squares; OSU, triangles). The response is relative to vegetation carbon for contemporary climate and atmospheric carbon dioxide of 312.5 ppmv

Figure 15.5 The range and mean "of responses in soil carbon for grasslands (filled symbols) and conifer forests (open symbols) at two levels of atmospheric carbon dioxide (312.5 and 625.0 ppmv) and three levels of climate (contemporary, circles; GFDL, squares; OSU, triangles). The response is relative to soil carbon for contemporary climate and atmospheric carbon dioxide of 312.5 ppmv

Decreases in NPP may result from elevated temperature by reducing soil moisture. This mechanism primarily affects the NPP response of TEM in dry regions of conifer forest (McGuire et al. 1993). Other models have predicted that elevated temperature may increase evapotranspiration to decrease forest growth in both dry and wet regions of present-day conifer forest (Pastor and Post 1988; Bonan et al. 1990; Running and Nemani 1991). Elevated temperature may also decrease NPP by enhancing respiration costs relative to carbon uptake. This mechanism has been observed to primarily affect the NPP response of TEM in warm moist regions of temperate conifer forest (McGuire et al. 1993); it has also been observed to be an important factor influencing the NPP response predicted by the Forest-BGC model to elevated temperature at a site in warm-moist temperate forest (Running and Nemani 1991). Because elevated temperature influences processes that can both enhance or decrease the NPP predicted by TEM, the NPP response to the high-temperature GFDL climate did not substantially differ from the NPP response to the low-temperature OSU climate.

Table 15.5 Response of NPP (10¹⁵ 9 C yr^-1) by region for experiment involving two levels of atmospheric CO₂ and three levels of climate (parameterized for 1 m soil carbon)

The NPP response of TEM to elevated temperature influences the response of vegetation carbon; increased NPP translates into increased vegetation carbon for both the GFDL and OSU climate scenarios. Linked models of forest productivity and soil processes have also predicted increased aboveground vegetation carbon in response to climate warming for conifers of both the temperate and boreal region (Pastor and Post 1988; Bonan and Van Cleve 1992), although the response depends on whether or not soil moisture is affected (Pastor and Post 1988; Bonan et al. 1990).

Table 15.6: Response of vegetation carbon(10¹⁵ g C) by region for experiment involving two levels of atmospheric CO₂ and three levels of climate (parameterized for 1 m soil carbon)

Table 15.7 Response of soil carbon (10¹⁵ g C) by region for experiment involving two levels of atmospheric CO₂ and three levels of climate (parameterized for 1 m soil carbon)

The response of soil carbon to elevated temperature will depend on the NPP response, which influences inputs into the soil, and on the decomposition response per unit soil carbon, which influences CO₂ losses from the soil organic pool. In TEM, if elevated temperature does not substantially decrease available soil moisture, then it will increase decomposition per unit soil carbon. Elevated temperature is predicted by TEM to decrease soil carbon for the high- temperature GFDL climate, but not for the low-temperature OSU climate. Schimel et al. (1990) indicated that soil carbon levels decreased in response to elevated temperature at sites in both the northern and southern Great Plains of the United States. Similarly, Burke et al. (1991) reported that soil carbon levels of the central Great Plains decreased approximately 3% after running CENTURY for 50 years with the GISS climate. This decrease is intermediate to the responses predicted by TEM for the GFDL and OSU climates; the temperature increase predicted by the GISS climate is intermediate between the GFDL and OSU climates. Elevated temperature is predicted by TEM to decrease the soil organic pool of conifer forests for both the GFDL and OSU climates, with greater decreases predicted for the high-temperature GFDL scenario. A linked model of boreal forest productivity and soil processes also predicts that soil organic carbon of boreal conifers decreases in response to climatic warming (Bonan and Van Cleve 1992).

15.8 RESPONSES TO CHANGES IN CLIMATE AND CARBON DIOXIDE

With changes in both climate and CO₂ concentration, TEM predicts that responses of NPP, vegetation carbon, and soil carbon for individual grid cells of grasslands and conifer forests may be positive or negative (Figures 15.3-15.5). For global grasslands, TEM estimates annual NPP to increase 0.6 Pg C (27.3%) for the GFDL climate (Table 15.5). The predicted increases for the OSU climate are slightly less (0.4 Pg C, 18.2%; Table 15.5). Vegetation carbon increases 1.2 Pg C (36.4%) for the GFDL climate and 0.9 Pg C (27.3%) for the OSU climate (Table 15.6). The predicted increases in soil carbon are less for the GFDL climate (3.0 Pg C, 4.0%; Table 15.7) than for the OSU climate (5.1 Pg C, 6.8%; Table 15.7).

For global conifer forests, TEM predicts annual NPP to increase 2.4 Pg C (32.9%) for the GFDL climate, which is about 40% more than the 1.7 Pg C (23.3%) increase for the OSU climate (Table 15.5). Increases in vegetation carbon predicted for the two climates show a similar pattern; for the GFDL climate, the 97.1 Pg C increase (34.4%) is more than 40% higher than the 66.0 Pg C increase (23.4%) predicted for the OSU climate (Table 15.6). Soil carbon decreased slightly for the GFDL climate (0.1 Pg C; < 0.1 %; Table 15.7), but increased for the OSU climate (6.6 Pg C; 2.9%; Table 15.7). Thus, for the climates considered with elevated CO₂ concentration, conifer forests are potentially much stronger carbon sinks than grasslands because of the ability to store carbon in woody biomass.

The response of NPP to elevated CO₂ and temperature predicted by TEM is influenced by moisture availability. In moist regions of temperate forest, elevated temperature enhances decomposition to increase nitrogen availability. The increased nitrogen availability allows the vegetation to generally incorporate elevated CO₂ into production, but the overall effect on NPP is sensitive to the plant respiration response (McGuire et al. 1993). In contrast, the FOREST -BGC model predicts a slight decrease in NPP for a site in warm-moist conifer forest because the enhanced respiration costs of elevated temperature more than offset the photosynthetic gains from elevated CO₂ (Running and Nemani 1991). In dry regions of temperate forest, NPP is predicted by TEM to increase because enhanced carbon uptake in response to elevated CO₂ generally more than compensates for decreased soil moisture or increased plant respiration caused by elevated temperature (McGuire et al. 1993). Similarly, for a site in a dry conifer forest the Forest-BGC model predicts increased NPP in response to elevated temperature and CO₂ because photosynthetic gains more than offset respiration costs (Running and Nemani 1991). The increases in NPP predicted by TEM for conifer forests in response to elevated temperature and CO₂ translate into increased vegetation carbon for both the GFDL and OSU climates. However, the enhanced NPP is able to compensate for the increased decomposition per unit soil carbon for the low-temperature OSU climate, but not for the GFDL climate. Thus, soil carbon increases are predicted for the OSU climate and decreases for the GFDL climate.

Table 15.8 Response of soil carbon (10¹⁵ g C) in tall grasslands and temperate conifer forest for the 1 ill and 20 cill calibrations in the experiment involving two levels of atmospheric CO₂ and three levels of climate

15.9 SENSITIVITY OF SOIL CARBON RESPONSE TO CALIBRATION DEPTH

The responses of NPP and vegetation carbon to changed climate or CO₂ do not demonstrate any sensitivity to the calibration depth of soil carbon for either tall grasslands or temperate conifer forests. However, the absolute responses of soil carbon for the 1 m calibrations of tall grasslands and temperate conifer forest are approximately three to four times larger than for the 20 cm calibrations (Table 15.8). Although the absolute response of soil carbon is always greater for the 1 m calibration, the proportional responses are essentially identical for both the 1 m and 20 cm calibrations. For models that make equilibrium estimates, these results indicate the importance of identifying at the calibration site the soil carbon that is likely to be actively decomposing over the time frame of interest. For climate change studies involving a doubled CO₂ atmosphere, the appropriate time scale is decades to centuries. The inclusion of soil carbon that turns over on the time scale of millennia ('old carbon') will overestimate the response of soil carbon. The CENTURY model (Schimel et al. 1990; Burke et al. 1991; Ojima et al. 1993b) estimates soil carbon to a depth of 20 cm. This depth may be approximately appropriate for identifying the relevant soil carbon in grasslands where most inputs are near the surface, but in forests the rooting zone may be much deeper than 20 cm. Because both recent and old carbon may occur at all depths in a forest soil, depth may not be the best metric to identify the actively decomposing soil carbon that is appropriate to doubled CO₂ climate studies.

Table 15.9 Comparison of the responses of grasslands and conifer forest to climate change and elevated CO₂ among various models

^aNR indicates no response

15.10 RELEVANCE OF SPATIAL SCALE FOR THE COMPARISON OF MODEL RESPONSES

During the next century, substantial simultaneous changes are predicted to occur in several climatic variables including CO₂ temperature, precipitation, and cloudiness (Mitchell et al. 1990; Watson et al. 1992). To assess the influence of these changes on regional carbon cycling, it is desirable to represent how the interactive effects of climate change and elevated CO₂ influence ecosystem processes in a spatially continuous manner. Although several models have been used to study the potential effects of climate changes on carbon cycling in grasslands and conifer forests, few have been used to study the interactive effects of climate with elevated CO_`. Also, most investigations have focused on potential responses at specific sites rather than the responses at larger spatial scales; the results of site-specific investigations can appear contradictory so that responses at the regional scale are difficult to assess. In this discussion we compare the directions of response of our study to those of other modeling studies with respect to the changes in climate and CO₂ and with respect to the spatial scale of response (Table 15.9). The spatial scale is considered at two resolutions. The fine resolution is 'site', which may include a grid cell or polygon for which the climate variables were treated in the study as having no spatial variability. The coarse spatial scale is 'regional', which we define as an aggregation of grid cells or polygons.

The CENTURY model has been used to study potential responses of NPP and soil carbon in the grassland biome at both the site (Schimel et al.1990; Ojima et al. 1993b) and regional (Burke et al. 1991) scales. The region considered by Burke et al. (1991) is the central Great Plains of the United States, which is contained within the total grassland region considered by TEM. The climatic responses predicted by both models are consistent at both the site and the regional spatial scales, although the regions considered by CENTURY and TEM are not identical. Of the CENTURY studies, only Ojima et al. (1993b) consider responses to changes in both climate and CO₂; these are consistent with the TEM results at the site scale.

The application of models to study the potential responses of conifer forests has primarily focused on the effects of climate change at the site scale (Pastor and Post 1988; Running and Nemani 1991; Bonanand Van Cleve 1992). The results of these studies are consistent with responses predicted by TEM at the site scale, but only one of the studies has examined the response of soil carbon (Bonan and Van Cleve 1992). Also, only one of these studies has examined the potential response of conifer forest NPP to changes in both CO₂ and climate (Running and Nemani 1991); the responses are consistent with those of TEM at the site scale.

Although there is much consistency among models for the results reported in Table 15.9, there are limitations in comparing the results. Most of the studies have focused on the responses of carbon cycling to climate change without considering elevated CO₂. Responses at both site and regional scales are sensitive to the interaction of elevated CO₂ with climate change, and few studies are available for this comparison. Also, the responses at the site scale depend on the sites chosen. Studies that use few sites are more likely to miss some variability in potential responses across a biome. By examining site responses in a spatially continuous manner, e.g. grid cells or polygons, models can document the range of responses to a particular climate scenario for the region of interest; the aggregation of responses can be used to examine the overall response for the region. The aggregation error associated with regional estimates developed from a spatially continuous model application is generally less than for regional estimates developed from sparse site-specific applications (see Kicklighter et al. 1994).

15.11 CONCLUSIONS

The potential distribution of grasslands and conifer forests in our analysis represents 22% of the area occupied by the terrestrial biosphere. Globally, TEM estimates that conifer forests are 1.4 times more productive per unit area than grasslands. Carbon storage per unit area in global conifer forests is estimated by TEM to be approximately 2.8 times that of global grasslands, primarily because of carbon storage in woody tissue. In response to climate change and elevated CO₂, conifer forests are potentially much stronger carbon sinks than grasslands because of the ability to store carbon in woody tissue. The responses of NPP and vegetation carbon in both grasslands and conifer forests appear insensitive to assumptions about the soil carbon at the calibration site. However, the response of soil carbon to changes in climate or CO₂ is three to four times greater for carbon considered to a 1 m depth than for carbon considered to a 20 cm depth. For models that make equilibrium predictions, these results indicate the importance of identifying at the calibration site the soil carbon that is likely to be actively decomposing over the time frame of interest.

The application of process-based models in a spatially continuous manner represents an important tool for making assessments of ecosystem response to climate change because it provides scientists and policy makers with the capability to investigate the potential effects in a quantitative and geographically specific way. However, it is important to recognize that there is uncertainty surrounding the responses predicted by any model. There is a need to compare different models applied in a spatially continuous fashion to help quantify some of the uncertainty. Progress in understanding the potential responses of grasslands and conifer forests to changes in climate and CO₂ requires this activity. Also, there is a need to link models of vegetation distribution with models of ecosystem function to more fully understand the potential effects of climate change on carbon cycling in grasslands and conifer forests.

15.12 REFERENCES

Adams, R. M" Rosenzweig, C., Peart, R. M., Ritchie, J. T., McCarl, B. A., Glyer, J. D., Curry, R. B., Jones, J. W., Boote, K. J. and Allen, L. H. Jr (1990) Global climate change and US agriculture. Nature 345, 219-224.

Australian Surveying and Land Information Group (1990) Vegetation. Atlas of Australian Resources, Third Series, Vol. 6, Commonwealth of Australia, Canberra.

Bonan, G. B., Shugart, H. H. and Urban, D. L. (1990) The sensitivity of some high-latitude boreal forests to climatic parameters. Clim. Change 16, 9-29.

Bonan, G. B. and Van Cleve, K. (1992) Soil temperature, nitrogen mineralization, and carbon source-sink relationships in boreal forests. Can. J. For. Res. 22,629-639.

Burke, I. C.,Kit tel, T. G. F., Lauenroth,W. K., Snook,P., Yonker,C. M. and Parton,W. J. (1991) Regional analysis of the Central Great Plains. BioScience 41, 685-692.

Emanuel, W. R., Shugart, H. H. and Stevenson, M. P. (1985) Climatic change and the broad-scale distribution of terrestrial ecosystem complexes. Clim. Change, 7,29-43.

Hou,H. Y.,Sun,S. C., Chang,J. W.,Ho, M. K.,Wang,Y. F., Kong, D. C. and Wang,S. C. (1979) Vegetation Map of China. The Laboratory of Plant Ecology and Geobotany, Institute of Botany, Chinese Academy of Sciences. Map Publisher of the People's Republic of China, Beijing, China.

Houghton,J. T., Callander, B. A. and Varney,S. K. (Eds)(1992) Climate Change 1992- The Supplementary Report to the I PCC Scientific Assessment. Cambridge University Press, Cambridge.

Institute of Geography of the Siberian Department of USSR Academy of Sciences, Botanical Institute of USSR Academy of Sciences, and Moscow State University Geography Department (1990) Vegetation of the USSR. GVGK, Minsk, USSR.

Jenne, R. L. (1992) Climate model description and impact on terrestrial climate. In: Majumdar et al. (Eds) Global Climate Change: Implications, Challenges and Mitigation Measures, pp. 145-164. Pennsylvania Academy of Science, Easton, Pa.

Joint Federal-State Land Use Planning Commission for Alaska (1973) Major Ecosystems of Alaska. United States Geological Survey, Fairbanks, Ala.

Kicklighter, D. W., Melillo, J. M., Peterjohn, W. T., Rastetter, E. B., McGuire, A. D., Steudler, P. A. and Aber, J. D. (1994) Aspects of spatial and temporal aggregation in estimating regional carbon dioxide fluxes from temperate forest soils. J. Geophys. Res. 99, Dl, 1303-1315.

Kiichler, A. W. (1964) Potential Natural Vegetation of the Conterminous United States (map). American Geographical Society, New York.

McGuire, A. D., Joyce, L. A., Kicklighter, D. W., Melillo, J. M., Esser, G. and Vrsmarty, C. J. (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.

McGuire, A. D., Melillo, J. M., Joyce, L. A., Kicklighter, D. W., Grace, A. L., Moore, B. III and Vorosmarty, C. J. (1992) Interactions between carbon and nitrogen dynamics in estimating net primary productivity for potential vegetation in North America. Glob. Biogeochem. Cyc. 6, 101-124.

Manabe, S. and Wetherald, R. T. (1987) Large scale changes in soil wetness induced by an increase in carbon dioxide. J. Atmos. Sci. 44,1211-1235.

Mat thews, E. (1983) Global vegetation and land use: new high resolution data bases for climate studies. J. Clim. Appl. Meteorol. 22,474-487.

Melillo, J. M., Callaghan, T. V., Woodward, F. I., Salati, E. and Sinha, S. K. (1990) Effects on ecosystems. In: Houghton, J. T. et al. (Eds) Climate Change: The IPCC Scientific Assessment, pp. 283-310. Cambridge University Press, Cambridge.

Melillo, J. M., McGuire, A. D., Kicklighter, D. W., Moore, B. III, Vörösmarty, C. J. and Schloss, A. L. (1993) Global climate change and terrestrial net primary production. Nature 363,234-240.

Mitchell, J. F. B., Manabe, S., Meleshko, V. and Tokioka, T. (1990) Equilibrium climate change- and its implications for the future. In: Houghton, J. T. et al. (Eds) Climate Change: The IPCC Scientific Assessment, pp. 131-172. Cambridge University Press, Cambridge.

Ojima, D. S., Dirks, B. 0. M., Glenn, E. P., Owensby, C. E. and Scurlock, J. 0. (1993a) Assessment of C budget for grasslands and drylands of the world. Water Air Soil Pollut. 70,95-109.

Ojima, D. S., Parton, W. J., Schimel, D. S., Scurlock, J. M. 0. and Kittel, T. G. F. (1993b) Modeling the effects of climatic and C02 changes on grassland storage of soil C. Water Air Soil Pollut. 70,643-657.

Pastor, J. and Post, W. M. (1988) Response of northern forests to C02-induced climate change. Nature 334,55-58.

Prentice, I. C., Cramer, W., Harrison, S. P., Leemans, R., Monserud, R. A. and Solomon, A. M. (1992) A global biome model based on plant physiology and dominance, soil properties and climate. J. Biogeogr. 19,117-134.

Prentice, K. C. (1990) Bioclimatic distribution of vegetation for general circulation model studies. J. Geophys. Res. 95, D8, 11811-11830.

Raich, J. W., Rastetter, E. B, Melillo,J. M., Kicklighter, D. W., Steudler, P. A., Peterson, B. J., Grace, A. L., Moore, B. III and Vörösmarty, C. J. (1991) Potential net primary productivity in South America: application of a global model. Ecol. Appl. 1,399-429.

Rowe, J. S. (1972) Forest Regions of Canada. Department of Fisheries and the Environment, Canadian Forestry Service, Publication No.1300, Ottawa.

Running, S. W. and Nemani, R. R. (1991) Regional hydrologic and carbon balance responses of forests resulting from potential climate change. Clim. Change 19, 349- 368.

Schimel,D. S., Parton,W. J.,Kittel, T. G. F., Ojima, D. S. and Cole, C. V. (1990) Grassland biogeochemistry: links to atmospheric processes. Clim. Change 17, 13-25.

Schlesinger, M. E. and Zhao, Z.-C. (1989) Seasonal climatic changes induced by doubled C02 as simulated by the OSU atmospheric GCM/mixed-Iayer ocean model. J. Clim. 2, 459-495.

Schlesinger, W. H. (1977) Carbon balance in terrestrial detritus. Ann. Rev. Ecol. Syst. 8, 51-81.

UNESCO (1981) Vegetation Map of South America. UNESCO, Paris.

Vörösmarty, C. J., Moore, B. III, Grace, A. L., Gildea, M. P., Melillo, J. M., Peterson, B. J., Rastetter, E. B. and Steudler, P. A. (1989) Continental scale models of water balance and fluvial transport: an application to South America. Glob. Biogeochem. Cyc. 3, 241-265.

Walter, H. (1979) Vegetation of the Earth and Ecological Systems of the Geo-biosphere. Springer-Verlag, New York.

Watson, R.H., Filho, L.G.M., Sanhueza, E. and Janetos, A. (1992) Greenhouse gases: sources and sinks. In: Houghton, J.T. et al. (Eds) Climate Change 1992: The Supplementary Report to the I PCC Scientific Assessment, pp. 25-46, Cambridge University Press, New York.

White, F. (1981) Vegetation of Africa. UNESCO, Paris, France.

Whit taker, R. H. and Likens, G.E. (1973) Carbon in the biota. In: Woodwell, G. W. and Pecan, E. V. (Eds) Carbon and the Biosphere, pp. 281-302. United States Atomic Energy Commission, Washington, DC.

Willmott, C. J., Rowe, C. M. and Philpot, W. D. (1985) Small-scale climate maps: a sensitivity analysis to some common assumptions associated with a grid-point interpolation and contouring. J. Am. Cartog. 12,5-16.

Woodward,F. I. and McKee,I. F. (1991) Vegetation and climate. Environ. Int.17, 535-546.