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

12.1 INTRODUCTION GRASSLANDS AND THE GLOBAL CARBON CYCLE

Grasslands are one of the major vegetation types, covering nearly one-fifth of the world's land surface or approximately 24 X 106 km². Ranging from the savannas of Africa to the North American prairies and the converted grasslands of Latin America and Southeast Asia, these ecosystems are likely to remain roughly constant in area or even see expansion in their areal extent under modified climate conditions. These ecosystems are large reservoirs of carbon (C) globally, containing approximately 30% of global soil C stocks (Anderson 1991; Eswaran et al. 1993) and the soil component is the main C reservoir in these ecosystems. Grasslands have developed under climatic regimes characterized by high climate variability that favor herbaceous or open wooded systems.

Potential effects of climate change on grassland biogeochemistry and soil C stocks have received much less attention compared with forests (Hall and Scurlock 1991; Long 1991a). The effects of changes in factors such as temperature, water and nutrients are relatively well understood, and the interactions with the effects of CO₂ fertilization are now beginning to be sufficiently well-developed to evaluate effects on C fluxes in various ecosystems (Schimel et al. 1990,1991, Long 1991a, b; Owensby et al.1993b). However, most of our current understanding has been necessarily derived from short-term experiments of only one or two of the factors modified simultaneously. To fully understand the impact of global change effects on these grassland ecosystems, it is necessary to study the interactive effects of the biogeochemical, hydrological, and physiological processes to changes in climate and CO₂ in a systematic and integrated fashion. One of the more powerful research tools which has been developed during the past 10 years are ecosystem models robust enough to be used to simulate ecosystem dynamics across a broad range of environmental factors. These models are capable of studying the complex interactions and feedbacks in an integrated plant--soil ecosystem framework, but are also able to provide an evaluation of the long-term implications of change to climate and CO₂ levels on key C storage components, such as soil organic levels. Grassland ecosystems store most of their C in soils, where turnover times of the bulk soil are relatively long (thousands of years), and so changes, though they may occur slowly, will be of significant duration (Schimel et al. 1994). Changes in grassland C storage can thus have a significant and long-lived effect on global C cycles. Several estimates of soil organic matter response to environmental change have recently been developed. Jenkinson et al. (1991) estimated substantial and rapid release of C from soils as a result of global warming (61 Pg over 60 years), while Schlesinger (1990) argued that soils have a low potential for added C storage and can only change slowly.

Most previous studies have had recognized limitations: Schlesinger's (1990) study used inferential and potentially confounded soil chronosequence studies, while the model of Jenkinson et al. (1991) was unable to take account of feedback between soil processes and climate change effects on primary production, a key feature of CENTURY. Jenkinson et al. (1991, p. 306) noted that in order to determine how climate change affects the flux of CO₂ between soil and atmosphere, their model 'would have to be coupled to another which specifies how global plant production is altered by increasing concentrations of green-house gases'.

Under the aegis of a collaborative SCOPE project (Breymeyer and Melillo 1991), the CENTURY model of plant--soil interactions has been further developed in order to apply it to a wide range of grasslands worldwide (Parton et at. 1993; Chapter 11 this volume). The SCOPEGRAM group has assembled a unique synthesis of detailed and long-term data from both temperate and tropical grass-lands, to evaluate the robustness of the CENTURY model and simulated response to climate change scenarios. A comparative analysis of ecosystem dynamics as simulated by a more detailed ecophysiological model, GRASS (Coughenour 1984) and by CENTURY was also conducted for two of these grassland systems.

The analysis of climate and CO₂ changes on biogeochemical processes needs to incorporate the feedbacks linking plant production system (C sequestering processes) and the decomposition processes (C releasing processes). The net change in these two sets of processes will determine the source-sink response of the ecosystem to climate and CO₂ changes. Climate and CO₂ changes may enhance plant production by increasing water or N availability, but these changes may also increase decomposition rates which would offset the gains in storage of C due to elevated CO₂ levels. In addition, physiological changes due to altered water or N use efficiencies under altered CO₂ levels may affect plant productivity and C turnover rates. These changes may often counterbalance changes in system C storage resulting from climate warming.

Over the last few years, it has been suggested that the majority of the 'missing sink' for C in the global CO₂ balance may be accounted for by terrestrial vegetation rather than by oceanic uptake, and, furthermore, that such a sink may be located in both tropical and temperate regions (Tans et al. 1990; Enting and Mansbridge 1991). It is, therefore, imperative to include grasslands with other major terrestrial ecosystems in attempts to model the global C cycle, in order to arrive at a well-founded policy toward climate change. Many approaches to the role of natural ecosystems in climate change involve computer modelling of the C flows between vegetation, soil and atmosphere, and can predict the effects of changes over many years. Here, we demonstrate that a grassland model which has been tested or 'validated' for a wide range of specific study sites can be scaled up to regional levels, and then interfaced with atmospheric general circulation models (GCMs). These GCMs are currently being used to develop national and international policy responses to the threat of global climate change, and it is imperative that these include the latest information on plant/soil response and feedback to climate change.

The primary objective of this chapter is to determine how sensitive grassland ecosystems are to the factors controlling the ecosystem responses (e.g. plant productivity, soil C changes, decomposition, N mineralization) to climate change and increasing atmospheric CO₂ in grasslands found in different ecoregions of the world. This analysis was made using CENTURY and GRASS ecosystem models. A secondary objective is to assess the difficulty of observing change in ecosystem dynamics, especially in arid to semiarid regions where precipitation typically shows great variance. We also analyze the differences between the transient response over 50 years as well as the long-term 'equilibrium' response to CO₂ and climate change for comparison with previous global analyses such as presented by Melillo et al. (1993). In our analysis, we did not attempt to model climate-driven redistribution of grassland regions, although it is recognized that changes in land cover due to climate or land use changes may overshadow climate-induced changes in existing grassland regions.

12.2 MODELLING STRATEGY

12.2.1 Grassland ecoregions

Sites were selected to represent a broad range of grassland regions based on climate factors and ecosystem properties (Figure 12.1). These include 16 sites for which detailed monthly data were available on vegetation biomass and dead matter dynamics, in order to test the model (Parton et al. 1993; Chapter 11 this volume). We assembled at least 25 years of monthly weather data for these 16 sites, plus an additional 15 grassland sites which have been described in the literature (Table 12.1). The regions ranged from temperate ecosystems with highly seasonal temperature patterns to tropical regions where temperature varied little. Precipitation at all of the sites displayed a marked seasonality, and in most cases a high degree of variability between years.

On the basis of temperature, rainfall and seasonality of climate, we grouped 31 sites into 7 bioclimatic regions based on Bailey ecoregions as follows (Table 12.1, Bailey, 1989):

1. Cold desert steppe- dry domain

2. Temperate steppe-dry domain

3. Humid temperate-humid temperate domain

4. Mediterranean -humid temperate domain

5. Dry savanna-dry domain

6. Savanna-humid tropical domain

7. Humid savanna-humid tropical domain

Climatic characteristics of the steppe regions are relatively low annual precipitation (PPT), potential evapotranspiration (PET) rates of around 1000 mm, and a drought index defined by PPT/PET ratios of around 0.3 (Table 12.2). The difference between the cold desert steppes and the temperate steppes are primarily temperature (0 °C vs 10 °C) and the inclusion of some C₄ grasses in the temperate steppes. Steppe ecoregions are characterized by relatively low plant productivity, N mineralization, and abiotic decomposition rates. The cold desert steppes have slightly higher production and N mineralization rates, resulting from higher N deposition and soil N fixation rates. Higher soil C levels in the cold desert steppe are primarily a result of substantially lower decomposition rates.  

Figure 12.1 Map showing location of grassland sites and boundaries of Bailey ecoregions representing grasslands worldwide. N .indicates sites used in the intial parmeterization of different grassland ecosystems in various ecoregions (see Parton et al. 1996, Chapter 11 this volume)

Table 12.1 Grassland sites modeled in the present study. Each site is assumed to be representative of its surrounding area on the Bailey (1989) ecoregion map

1.T. G. Gilmanov et al. (in press); Parton et al. (1993).2. Mongolia, Chuluun et al. (1996).3. Wang and Jiang (1982); Chen (1988).4. Weather data only; .World Weather Disc' (Weather Disk Associates, Inc., National Climate Data Center, USA). 5. Parton et al. (1993). 6. Haas et al. (1957).7. Fernandez et al. (1991).8. Ojima et al. (1990), Towne and Owensby (1984), Ojima et al. (1994).9. W. Bethgen (pers. comm., Uruguay). 10. P. Hiernaux (pers. comm.). 11. Tropical Soil Biology and Fertility (TSBF) site. 12. Singh and Yadava (1974). 13. TSBF. 14. Long et al. (1989,1992); Parton et al. (1993).15. R.J. Scholes (pers. comm.). 16. San Jose, and Medina (1976).17. CIAT Center. 18. Menaut and Cesar (1979); Parton et al. (1993).

Table 12.2 Long-term mean annual precipitation, temperature, potential evapotranspiration rate and ratios of annual precipitation to potential evapotranspiration rate for the seven grassland regions

The humid temperate ecoregions have similar air temperature and PET to the temperate steppes, but the precipitation is substantially higher. Vegetation type ranges from C₃ in the cooler Russian sites to C₄ in the warmer American sites. Plant productivity is three to four times that of the temperate steppes. Soil C is higher as a result of increased C inputs and only slightly increased abiotic decomposition rate.

The Mediterranean ecoregion is characterized by a warm mild climate with winter rainfall. Annual PET rates are higher than the continental grasslands but the PPT/PET ratio is high during the winter and spring and low during the summer. The vegetation in these regions is dominated by C₃ plants. Plant production is about 50% the level of the humid temperate regions, with reduced N mineralization rates. Decomposition rates are reduced by water stress during the summer period but enhanced during the wet winter period by warmer temperatures. This lower production results in soil C levels about half those of the humid temperate ecoregions.

The savanna ecoregions all have similar temperature (24 °C) but the mean annual precipitation ranges from 400 to 1500 mm, and PET ranges from 2000 mm in the dry savannas to 1400 mm in the savannas and humid savannas. The PPT/PET ratio, therefore, increases markedly from the dry savannas to the humid savannas, mainly as a result of the length of the rainy season. All the savannas are dominated by C₄ grasses. Plant production and N mineralization rates increase rapidly from the dry to the humid savannas, but abiotic decomposition also increases, so the net effect on soil C is only a modest increase from drier to more humid ecoregions. The reduced soil C for the humid tropical ecoregion is a result of the sandy textures for many of the sites considered here.

Overall, in accordance with actual field data, plant production, N mineralization, and the ratio of above- to below ground productivity, all increase with annual precipitation (Long et al. 1992). Differences in soil C levels reflect changes in C inputs to the system and the abiotic decomposition rate, with the highest levels in the humid temperate ecoregion and the lowest in the temperate steppes and dry savannas.

12.2.2 Modelling the effect of climate change and increased CO₂

12.2.2.1 CENTURY model

CENTURY is a simulation model of plant-soil interactions in grasslands, forests, crops and savannas (Parton et al. 1987,1988,1992,1993; 1995, Chapter 11 this volume). Version 3.0 of CENTURY was used for these model runs (Parton et al. 1992, 1993).

The effect of increased atmospheric concentrations of CO₂ on the photosynthetic pathway has been well documented in C₃ plants. In addition to the direct effects of CO₂, it has been observed to increase water and nitrogen use efficiency (NUE) in both C₃ and C₄ plants (Owensby et al. 1993a, b). In CENTURY, we modified the plant production parameters under a 'double-CO₂' climate by changing potential evapotranspiration (PET) and NUE, for both C₃ and (grasslands uniformly. We allowed a 20% decrease in total PET and a 20% (decrease in N content with the change in atmospheric CO₂ concentration from 350 to 700 ppm, via a simple linear function based on results from a tallgrass prairie (Owensby et al. 1993a, b).

For each grassland site, the model was run to equilibrium for 5000 years, using a repeated 25-year pattern of observed current weather data. Site-specific patterns of grassland management (i.e. burning and grazing) were incorporated i these long-term runs as described in Parton et al. (1993).

12.2.2.2 GRASS model

The GRASS model simulates physiological and morphological details of plant growth. Plant growth is simulated with a two-day time step while photosynthesis: and transpiration are simulated with two-hour time steps on alternate days (Coughenour et al. 1984; Coughenour 1984). Its photosynthesis sub model is derived from the C₃ model of Farquhar et al. (1980), with extensions to simulate C₄ photosynthesis (Chen et al. 1993), and linkages with the stomatal conductance sub model of Ball et al. (1987). The GRASS plant growth sub models are linked with a daily time step implementation of the decomposition and nutrient cycling sub model of CENTURY (daily CSOM).

Simulated photosynthetic responses to atmospheric CO₂ involve biochemical and stomatal processes. Rubisco, the main carboxylating enzyme, responds to CO₂ concentration inside the leaf. The optimum temperature for photosynthesis is shifted upwards by elevated CO₂ in C₃ species due to reduced photorespiration (Long 1991a). In C₄ species, CO₂-concentrating processes prevent Rubisco from being CO₂-limited at low CO₂, thus weakening the response of photosynthesis to CO₂. However, stomata close in response to elevated CO₂ (Ball et al. 1987; Collatz et al. 1992). Stomatal closure decreases water loss through transpiration while elevated CO₂ increases gaseous CO₂ flux through stomates. The overall result of stomatal closure responses to elevated CO₂ in both C₃ and C₄ species is an increase in water-use efficiency (WUE).

Increased WUE of C₃ and C₄ plants enhances photosynthesis. Since less water is lost through transpiration, more photosynthate is produced from each millimetre of precipitation. Therefore, increased WUE will increase net primary production in water-limited environments. However, increased WUE will have little or no effect when water is not limiting. The increase in WUE leads to increased soil moisture levels, which ameliorates negative effects of water stress on other processes. Increased soil moisture content increases C allocation to shoots, reduces photosynthesis rate, reduces leaf senescence rate, and increases decomposition and N mineralization rates.

12.2.3 Climate change scenarios

The doubled CO₂ climatologies used to drive the CENTURY model in the present study were derived from the Canadian Climate Centre (CCC) and Geophysical Fluid Dynamics Laboratory, High Scenario (GFHI) models of global climate change, which provide monthly projections of temperature and precipitation for 2.5° grid cells worldwide. We took the values projected for a doubling of atmospheric CO₂, and assumed that these would be reached within 50 years from the present, corresponding roughly to IPCC Scenario A (Houghton et al. 1990). Projected changes in mean annual temperature and precipitation are shown in Figure 12.2.

The two GCMs simulate similar changes to mean annual air temperature (2-5 °C); changes in precipitation were different between the models. The GFHI model showed increases in precipitation for all of the regions whilst the CCC model showed decreases in precipitation for the temperate steppe and humid savanna regions and increases for the other regions (Figure 12.2).

We assumed that changes in precipitation and temperature at each site were linear over a 50-year period, at which point they stabilized at the modified 'double CO₂' climate, as did atmospheric CO₂ (Figure 12.3). The results from the 25-year period immediately following this represent the transient response to climate change. To determine a 'near-equilibrium' long-term response to change, we continued with 150 years of stable modified weather data. Model results were analyzed for changes in the level of soil organic matter, plant productivity, N mineralization and the model-calculated effect of climate on decomposition (abiotic decomposition factor).

The GRASS model was subjected to 2 x CO₂ climate change scenarios generated from the CCC and GFHI models (Table 12.2). The GRASS model was used to simulate C₄ (Bouteloua gracilis) and C₃ (Pascopyrum smithii) ecosystems in Colorado and a C₄ grass (Themeda triandra) ecosystem in Kenya, with the same repeating patterns of base weather data that were used in the CENTURY experiments, with and without CO₂ doubling. Day and night temperatures were increased equally. Relative humidity and cloud cover were not changed. The model was run for an initial 50 years with no climatic changes. Then, the climate was gradually ramped up to the CCC and GFHI 2 x CO₂ scenarios over a 50-year period, followed by a 25-year period over which the model was allowed to adjust to the scenarios.

12.3 RESULTS

12.3.1 Regional response to climate and CO₂ Changes

Figures 12.4-12.10 show CENTURY simulations of the impact of the two GCM climate change scenarios and the effect of increasing atmospheric CO₂ independently, for each of the ecoregions considered. These values represent the means for the 25 years of 'double-CO₂' climate immediately following the 50-year climate change period; the results for each site are shown, together with the numerical average for each ecoregion, not taking into account any area weighting.

Figure 12.2 Illustration of contrast between changes in temperature and precipitation for each ecoregion  under CCC and GFHI scenarios

Figure 12.3 Time course of 200-year climate transient and elevated CO₂ effects on above ground plant production and soil organic C predicted by the CENTURY model

Statistical analysis of these results shows that changes in total plant production (TPP) are positively correlated to changes in precipitation (PPT) and N mineralization (N_min), with N mineralization being the most important term [TPP = -6.62 + 1.39 (PPT) + 19.62 (N_min); r² = 0.64; p = 0.0001]. The response to N mineralization is consistent with the general observations that grasslands respond positively to addition of N fertilizer (Rains et al. 1975; Lauenroth and Dodd 1978). The response to increased precipitation is due to reduced drought stress and increase in N deposition.

The changes in N mineralization are positively correlated with changes in decomposition and precipitation. Increased decomposition rates in the model are associated with increased mineralization of soil organic N, and increased precipitation increases N availability over the long term. Changes in C are negatively related to changes in decomposition and positively related to changes in productivity. Changes in abiotic decomposition rates are positively correlated with changes in soil temperature and precipitation.

12.3.1.1 Cold desert steppes

For the majority of sites, both GCMs show an increase in precipitation and temperature, with the CCC scenario showing a greater effect (Figure 12.4). This may be attributed to the common geographical area shared by all sites. The increase in temperature results in higher decomposition rates, with the CCC scenario showing a greater effect on decomposition. Plant production is reduced at three of the four sites, and is only weakly correlated with N mineralization, possibly due to the relatively high N input rates for this region. The mineralization of N shows no clear pattern of change. Soil C tends to decrease as a result of reduced production and increased decomposition. The Tumentsogt site (Mongolia) is an exception, showing little climate-induced change in soil C due to smaller increases in decomposition and a small increase in production.

Increased atmospheric CO₂ independently causes a substantial increase in production (30%). The cold desert steppes are characterized by low annual rainfall levels and high PET values; and changes in soil moisture status due to increased WUE appears to accelerate decomposition rates by 25%. This is the only ecoregion where CO₂ also causes a net increase in N mineralization. Together, these effects produce a modest stimulation of soil C storage (1-3%).

Figure 12.4 Cold desert steppe. Figures 12.4-12.10. are ecoregion summary figures showing % change at each site + ecoregion mean, for temperature precipitation, decomposition, NPP, SO M, and N_min Each histogram cluster of three bars shows CCC, GFDL, CO₂ effect

12.3.1.2 Dry steppes and temperate desert

Annual precipitation increases in the GFHI scenario for these regions, but shows both increases and decreases for the CCC scenario (Figure 12.5). Mean annual temperature increases are slightly lower (3.5-4 °C increases relative to 5 to > 6 °C in the cold desert steppe) than those observed for the cold desert steppes, with the GFHI scenario showing slightly higher increases. The combined effect of temperature and precipitation changes is to increase decomposition rates, with larger increases for the GFHI scenario. Increases in plant production are correlated with N mineralization, which itself is correlated with precipitation for most of the sites. Soil C tends to decrease or show little change, correlated with increases in decomposition rate. The Santa Rosa site (Argentina), which has the highest present-day precipitation, shows exceptional climate-induced decreases in N mineralization, plant production and soil C.

The CO₂ effect is a modest increase in production, but also a substantial (30%) increase in decomposition. Carbon dioxide causes a small decrease in N mineralization, and overall, the effect is a slight stimulation of soil C storage (1-2%).

12.3.1.3 Humid temperate

Precipitation is generally increased, with the GFHI scenario showing a greater effect (Figure 12.6). Temperature increases are similar to those for the temperate steppes with no clear difference between the GCMs. Together, these result in about 20% increase in decomposition rates. Changes in plant production are correlated with N mineralization, which appears to be controlled mainly by changes in decomposition rate. Soil C tends to decrease consistently for all sites.

Elevated CO₂ independently causes a 10-15% increase in production, and a 10-15% increase in decomposition. Nitrogen mineralization shows a notable decrease (5-10%) due to elevated CO₂, and overall the effect is a larger stimulation of soil C storage (2-4%) compared with the steppe ecoregions.

12.3.1.4 Mediterranean

Changes in precipitation and temperature were similar to the humid temperature ecoregion, producing a similar increase in decomposition rates (Figure 12.7). Changes in plant production are correlated with N mineralization, which is associated with changes in decomposition rate. Soil C decreases as a result.

As for the humid temperate ecoregion, elevated CO₂ causes a 10-15% increase in production, and about a 10% increase in decomposition. The mineralization of N shows a notable decrease (5-10%) due to elevated CO₂, and overall the effect is a moderate stimulation of soil C storage (3-5%).

Figure 12.5 Temperate Steppe region

Figure 12.6 Humid temperate ecoregion

Figure 12.7 Mediterranean ecoregion

12.3.1.5 Dry savannas

 Substantial increases in precipitation are predicted by both GCMs, with temperature increases slightly lower than for the temperate ecoregions, but the resulting changes in decomposition are low to modest (-10% to + 10%), tending to increase for the CCC and decrease for the GFHI scenario (Figure 12.8). Plant production generally increases and is highly correlated with N mineralization. However, changes in N mineralization were only weakly associated with decomposition rate, and resulting changes in soil C were low ( -3% to +5% ), showing no clear pattern.

The independent effects of elevated atmospheric CO₂ are similar changes in production, decomposition and N mineralization as for the humid temperate ecoregion. However, the net effect of elevated CO₂ on soil C storage balances out as zero.

12.3.1.6 Savannas

Precipitation generally increased, with temperature increases slightly lower than for the dry savannas (Figure 12.9). Projected changes in decomposition are both positive and negative, with a modest increase on average. Plant production shows an increase on balance, and is well correlated with N mineralization. Changes in N mineralization were associated with decomposition rate, and changes in soil C were negative overall.

Elevated CO₂ independently causes a 10-15% increase in production, and a reduced (10%) increase in decomposition relative to the dry savannas. Nitrogen mineralization shows a substantial decrease (10%) due to elevated CO₂ induced reduction in litter quality, and overall the effect is a moderate stimulation of soil C storage (4-6%).

12.3.1.7 Humid savannas

Precipitation changes showed no clear pattern, and temperate increases were the lowest of any ecoregion (Figure 12.10). Changes in decomposition showed no clear pattern, and plant production was weakly correlated with N mineralization and precipitation changes. Soil C tended to decrease overall (0 to -6%).

Elevated CO₂ produces a modest (10%) increase in production, and a much reduced (5%) increase in decomposition compared with the other savanna ecoregions. The CO₂ effect on N mineralization is the largest decrease for any ecoregion (15%), and the resulting stimulation of soil C storage is the highest for any ecoregion (8-12%).

Figure 12.8 Dry savanna ecoregion

Figure 12.9  savanna ecoregion

Figure 12.10 Humid savanna ecoregion

12.3.2 Site-Level analysis of climate and CO₂ change scenarios

The GRASS model responded quite differently to the CCC and GFHI climate-change-only scenarios at the CPER site (Table 12.3). The GRASS CCC scenario without a CO₂ doubling decreased NPP 52% in the C₃ grass and 48% in the C₄ grass. In contrast, the GFHI scenario increased net primary productivity (NPP) ofC₃ plants by 13% and decreased NPP of C₄ plants by 18%. The CENTURY simulations at the CPER site displayed smaller changes to the two GCM projections for NPP and net N mineralization, and no difference in soil organic matter (SOM) values (Table 12.3). At the Kenya site, the differences between scenarios were much less pronounced. The NPP decreased 3% under the CCC scenario and 6% under the GFHI scenario, without CO₂ doubling. At the Kenya site, CENTURY current values of NPP, net N mineralization, and SOM tended to be 20, 25 and 1%, respectively, of the GRASS simulations. The CENTURY ecosystem response for NPP and net N mineralization to climate changes at Kenya tended to be greater than the GRASS simulations, although the difference between GCMs tended to be less than 5% of each other.

The large discrepancies in model responses to the two climate change scenarios in the GRASSCPER simulations can be attributed mainly to precipitation. While the CCC model predicted a 5.2 mm increase in annual precipitation, the GFHI model predicted a 38.0 mm increase. During the March-August growing season, CCC predicted a 9.2 mm decrease while GFHI predicted a 27.5 mm increase. At the Kenya site, CCC predicted a 2.9 mm annual decrease while GFHI predicted a 72.0 mm annual increase. However, during the November-May wet season, both models predicted increased rainfall: CCC added + 18.6 mm while GFHI added +36.0 mm.

Doubling CO₂ without climatic change increased NPP 6, 5 and 40% in the CPER C₄ and C₃, and the Kenya C₄ grasslands, respectively for the GRASS simulations. Doubling CO₂ ameliorated negative NPP responses to climate change scenarios in CPER and accentuated positive NPP responses in Kenya for both models. The CCC scenario with CO₂ doubling decreased NPP by 20% in the C₃ grass and 37% in the C₄ grass, relative to the control simulation for the GRASS model. The NPP increased 46% in Kenya under both scenarios of CO₂ doubling and climatic change. The CENTURY simulations at CPER and Kenya under current climate with doubled CO₂ increased NPP by approximately 10%. The impact on SOM at CPER was neutral, whereas at Kenya, doubling CO₂ slightly enhanced SOM levels.

Elevated CO₂ concentrations might favor C₃ over C₄ species (e.g. Carter and Peterson 1983; Bazazz et al. 1989). This hypothesis was only supported by the CPER short-term simulations. It was not supported by the long-term simulations. In the long-term simulations, total NPP of C₃ and C₄ grasses responded similarly to CO₂ alone: 2 x CO₂ increased NPP 5% in the C₃ grass and 7% in the C₄ grass. The CCC scenario with 2 x CO₂ decreased NPP by 25 and 37% in C₃ and C₄ grasses, respectively, which would suggest a competitive advantage for C₄ species under this warm, dry scenario. The GFHI scenario with 2 x CO₂ increased NPP 13% in the C₃ and 21% in the C₄ grass, which also indicated that C₄ grasses might be favored. Indeed, the large stimulation of NPP in the tropical C₄ grassland suggested that C₄ species would fare quite well in warm, water- limited ecosystems.

Transpiration rate responses of the GRASS model simulations to climate and CO₂ varied among sites (Figure 12.11). Minor reductions in transpiration rate in response to 2 x CO₂ were simulated at the CPER C₄ site. In contrast, transpiration rates were much reduced by elevated CO₂ at the C₃ site, but less so with the .increased moisture of the GFHI scenario. Differences in transpiration responses likely contributed to different C₃ and C₄ NPP responses to CO₂. The greater reduction in transpiration rate of the C₃ species was due to greater stomatal and carbon assimilation (A_n) response of C₃ species to elevated CO₂. The magnitude of stomatal opening in response to elevated A_n and closing in response to elevated CO₂ is greater in C₃ species (see Ball et al. 1987) than C₄ species (see Collatz et al. 1992).

Transpiration rates at the Kenya site were lower than at CPER sites, due to much higher (e.g. 5 x) leaf area indices. In contrast to the CPER C₄ site, elevated CO₂ reduced transpiration rate significantly in Kenya. The reduction in transpiration rate was much larger than the CPER C₄ response. Stomatal closure due to elevated CO₂ decreased moisture stress on plants and decomposers in Kenya. Responses were not reduced by low temperatures as they were in CPER. This generated a positive feedback as plant growth was accelerated by lower water stress and by increased N availability.

Climate change and CO₂ doubling affected net N mineralization (N_min) rates (Table 12.3b). For both CENTURY and GRASS simulations, N_min responses contributed to differences in NPP among sites. Doubling CO₂ alone decreased N_min for both CPER grass types, but more so for the C₃ type. This response was caused by the negative effect of reduced litter N concentration, which results in greater immobilization of N. Significantly, despite this reduced N availability, NPP was increased by elevated CO₂ due to increased plant N use efficiency (NUE). It is likely that N mineralization would have been stimulated, or that it would not have declined if litter N concentrations were not reduced, since NPP and soil moisture were enhanced by CO₂. Thus, while nutrient limitation did reverse positive responses to CO₂, the reversal could be weakened by increased NUE.

The CCC climate scenario in GRASS simulations alone reduced N mineralization (N_min) on both the C₃ and C₄ sites, due to drier soil conditions and smaller litter inputs. The GFHI climate scenario alone in the GRASS simulations increased N_min at the C₃ and C₄ sites, due to reduced moisture limitations on decomposition rate. The CENTURY simulations showed greater N_min for both GCM projections. Adding 2 x CO₂ to the CCC scenario decreased N_min very slightly at the C₄ site, and increased N_min at the C₃ site. This was probably due to the greater NPP response of the C₃ grassland, with consequently higher litter inputs. Adding 2 x CO₂ to the GFHI scenario decreased N_min slightly at the C₄ site, but greatly decreased N_min at the C₃ site. The reduction in N_min with GFHI suggested increased importance of litter quality when litter inputs are large. The CENTURY simulation at the CPER site with doubled CO₂ plus climatic changes had slight increases in N_min due to slightly better soil moisture conditions.

Table 12.3 Climate change and CO₂ effects on simulated (a) total net primary production, (b) annual net N mineralization, and (c) total soil organic matter for CPER and Kenya using CENTURY and G RASS models. G RASS simulation for CPER are parameterized for C₃ and C₄ grasses, the CENTURY simulations for CPER and Kenya are for C₄ plants. The values are the 25-year means and ranges for 'current' climate, doubled 'CO₂' effects under current climate, climate change effects (CC) for the Canadian Climate Centre GCM (c) and the GFDL GCM (g), and the combined effect of 'CC+CO₂' for both GCMs (c and g)

CPER		Current	CO2	CC		CC+CO2
				c	g	c	g
( a) Total net primary production
CENTURY		116	139	122	135	150	164
	Range	75-191	88-217	72-213	81-229	92-237	96-263
GRASS-C3		141	151	75	159	88	171
	Range	49-334	68-286	13-213	92-261	22-218	108-257
GRASS-C4		136	150	68	161	114	170
	Range	72-288	78-246	13-167	113-217	10-256	108-240
Kenya CENTURY		338	368	367	420	398	462
	Range	172-670	216-735	191-694	219-820	223-768	272-912
GRASS		240	316	223	243	305	338
	Range	76-459	97-554	19-397	56-474	21-536	97-643
( b ) Annual net N mineralization
CENTURY		2.3	2.3	2.46	2.70	2.51	2.71
	Range	1.39-3.67	1.52-3.43	1.45-3.70	1.47-4.00	1.55-3.67	1.53-4.00
GRASS-C3		1.62	1.59	1.08	2.09	1.05	1.98
	Range	0.65-7.85	0.69-2.66	0.37-2.17	1.06-3.15	0.12-2.15	1.19-2.66
GRASS-C4		1.72	1.47	1.08	2.25	1.25	1.62
	Range	0.57-3.19	0.80-2.11	0-2.43	1.52-2.94	0.05-2.32	1.14-2.34
Kenya CENTURY		4.14	3.53	4.52	5.03	3.85	4.38
	Range	2.30-5.85	1.45-5.03	2.68-6.42	2.76-7.18	2.07-5.47	1.95-6.11
GRASS		2.13	2.80	2.07	2.38	2.73	3.14
	Range	-0.11-5.36	0.70-5.72	-0.55-4.62	-0.24-6.08	-0.28-5.22	0.83-6.40
(c) Total soil organic matter
CENTURY		2.30	2.28	2.27	2.26	2.26	2.25
	Range	2.26-2.36	2.22-2.34	2.23-2.34	2.21-2.29	2.20-2.31	2.20-2.28
G RASS.C3		2.63	2.63	1.99	2.60	2.06	2.59
	Range	2.56-2.74	2.57-2.73	1.82-2.15	2.53-2.69	1.89-2.23	2.54-2.68
GRASS.C4		2.56	2.58	1.83	2.49	2.39	2.57
	Range	2.47-2.69	2.53-2.64	1.64-2.05	2.42-2.56	2.25-2.55	2.51-2.60
Kenya CENTURY		5.24	5.45	5.24	5.29	5.38	5.41
	Range	5.20-5.29	5.38-5.49	5.19-5.29	5.24-5.34	5.33-5.43	5.34-5.46
GRASS		6.12	6.48	5.75	5.87	6.38	6.37
	Range	5.94-6.28	6.36-6.63	5.65-5.88	5.70-5.98	6.24-6.50	6.21-6.51

Figure 12.11 GRASS-daily CSOM responses of transpiration rate per unit leaf area index (LAI) in the final 25 years of the climate change scenarios in Figure 12.2. Transpiration rate was calculated by dividing season-Iong transpiration by peak LAI

Both climate change scenarios alone elevated N_min greatly at the Kenya site, due to increased soil moisture. The CENTURY simulations under elevated CO₂ levels resulted in a 10% decrease in N_min relative to climate change N_min values alone. Surprisingly, doubling CO₂ in the GRASS simulations increased N_min by 24% despite decreases in litter quality. N_min at the Kenyan site was primarily limited by soil moisture, in contrast to the CPER site where temperature played a larger role. The increased WUE of the C₄ grass under 2 x CO₂ increased soil moisture, which then increased decomposition rate. Similarly, Rice et al. (1992) observed elevated microbial biomass and respiration under elevated CO₂ in a dry year, but not in a wet year. Increased N_min likely contributed to increased plant growth, which was then recycled through decomposition, thus further elevating N_min. This positive feedback loop was undoubtedly enhanced by the high rate of N recycling in this tropical system relative to more temperature-limited systems.

The effect of 2 x CO₂ alone at the Kenya site was to increase NPP by 40%, despite only a 24% increase in N_min and a 19% increase in plant N uptake. This can be explained by the 15% lower N concentrations in dying roots and shoots, and the associated increase in NUE. The total amount of N in the system was actually slightly lower (5%) in the elevated CO₂ simulation due to greater fire losses. Thus, 2 x CO₂ enhanced N recycling rate. Greater total quantities of N uptake (kg N ha^-1) have also been observed in a tallgrass prairie, despite reduced %N concentrations in plant biomass (Owensby et al. 1993a).

Soil organic matter (SOM) in the GRASS simulation was reduced by 25-27% by the CCC scenario without 2 x CO₂ in CPER (Table 12.3c), due to lower C inputs from NPP. With 2 x CO₂, SOM was reduced only 5% under C₃ grass, but still declined by 22% under C₄ grass. In Kenya, the CCC scenario without 2 x CO₂ decreased SOM 4%, and with 2 x CO₂, it increased SOM by 4%. SOM declined by 1-2% under the GFHI scenario in CPER. With 2 x CO₂, however, GFHI decreased SOM 1% under C₄ and increased SOM 3% under C₃ grasses. In Kenya, the GFHI scenario decreased SOM by 4% without 2 x CO₂, and by 4% with 2 x CO₂. Decreased soil C under elevated CO₂ was accompanied by a 2% decrease in total soil N. To summarize, elevated CO₂ alone had little effect on ecosystem C storage in Colorado while it increased C storage 4% in Kenya.

Modeled responses for GRASS and CENTURY to changed climate and CO₂ demonstrated considerable interannual variability (Figure 12.12). For example, under the GFHI climate scenario, Kenya NPP increased by 30-50% in some years, while in other years there were little, no, or even negative responses. Simulations indicated that CO₂ increases may ameliorate the negative effects of certain dry years. For example, periodic dry years in the Kenyan climate data reduced NPP, even under the climate change scenarios (Figure 12.12). However, CO₂ doubling often ameliorated these decreases. Indeed, most of the positive NPP response to CO₂ could be attributed to greater increases in WUE in drier years. Similarly, Owensby et al. (1993b) found that NPP of a C₄ grass was stimulated by CO₂ in a dry rainfall year, while it was not stimulated in a wet year.

12.3.2.1 Climate change sensitivity

For all of the temperate ecoregions, the increase in temperature results in an average increase in PET of about 30-40%, and about 15-20% in the tropical ecoreglons.

Climate change impacts of the two GCM scenarios on total plant production (above- and below ground), soil C, N mineralization and decomposition are summarized for each ecoregion in Figure 12.13. Results show that the cold desert steppe region had a substantial reduction in plant production (about 25%) while the humid temperate sites had increases greater than 12%. All the savanna ecoregions showed relatively small changes ( < 5%). Total plant production is positively correlated to changes in precipitation and N mineralization, the latter being more important. The main exception to this is in the cold desert steppe region. Substantial differences in plant productivity between the GCMs exist for the temperate steppes and dry savannas, and are primarily related to differences in predicted precipitation. These results are in broad agreement with those of Melillo et al. (1993), whose terrestrial ecosystem model (TEM) predicted 10-15% increase in grassland NPP for double-CO₂ climate scenarios; however, TEM does not show up differences between contrasting grassland regions.

Figure 12.12 Annual dynamics of (a) precipitation, (b) temperature, (c) NPP-GRASS, (d) NPP-CENTURY (e) SOM for the climate change and CO₂ perturbation for Kenya. The climate time series for 100 years are depicted in (a) and (b) for the CCC and GFHI GCMs, the first 25 years is the current observed climate, followed by the 50 year transient ramp, and ending with a 25 year stabilized 2 X CO₂ climate as projected by CCC and GFHI GCMs. Simulated NPP from the (c) CENTURY model and (d) from GRASS are depicted for the 100-year times series. The simulated SOM are depicted in (e)

Figure 12.13 Effect of climate change alone on (a) production, (b) SOM, (c) N mineralization, (d) decomposition, for each ecoregion, for the CCC and GFHI scenarios

Soil C results show substantial losses of soil C for all the temperate sites, including the Mediterranean region, with the largest losses occurring in the cold desert steppe and temperate steppe regions. Soil C losses are low in all three savanna regions, with the highest losses being in the humid savannas (7%). The changes in soil C are negatively related to changes in decomposition and positively related to changes in productivity. Soil C changes do not vary much between the two GCMs.

The N mineralization rates tended to increase for all sites, with the largest increases observed for the humid temperate, Mediterranean and temperate steppe ecoregions. The N mineralization changes were positively correlated with changes in abiotic decomposition rates and precipitation, with decomposition rates having the greater impact.

Decomposition rates at the temperate sites are greatly modified under climate change simulations compared with the relatively small changes simulated in the tropical regions of the study. The apparent differences between the temperate and tropical regions in this analysis are the result of the relative changes in the temperature projections; more than 25% increase in mean annual temperature for the temperate regions, compared with approximately 10% increase for the tropics. There is also a trend for abiotic decomposition rates to increase with increasing precipitation.

12.3.2.2 Effect of increased CO₂ and CO₂/climate interaction

We simulated the impact of CO₂ on grassland ecosystems with and without the inclusion of climate change (Figure 12.14). Overall, the effect of increasing CO₂ under current climatic patterns was to increase total plant production and soil C storage. Production increased the most in the cold desert steppe region, and C storage increased most for the humid savannas. There is a trend for soil C changes to be inversely related to changes in abiotic decomposition rate (Figure 12.14). Nitrogen mineralization rates decreased in most regions, resulting from increased input of lower quality litter with higher C/N ratios (Melillo et al. 1982), but increased for the cold desert steppe region. The relatively high N inputs for the cold desert steppe regions may have influenced the response to elevated atmospheric CO₂. Changes in relative N mineralization rates appear to be positively related to changes in relative decomposition rates due to the release of N from decomposing SOM. Simulated increases in decomposition under elevated CO₂ resulted from improved soil water relationships at the cold dry sites cold desert steppe, temperate steppe), with smaller increases for the warmer and more humid sites.

Figure 12.14 Effect of CO₂ alone, climate alone (GFHI scenario), and combined CO₂ and climate for each ecoregion, for (a) production, (b) SOM, (c) N mineralization, (d) decomposition

The combined effects of CO₂ and climate change were additive for each ecosystem property studied. Relative changes in total plant production, soil C, N mineralization, and decomposition were compared statistically between (i) the combined impact of CO₂ and climate and (ii) the independent impacts of CO₂ and climate change; a linear additive effect was demonstrated (r₂ = 0.97,0.98, 0.97, and 0.92, respectively).

Increased CO₂ enhanced total plant production regardless of the climate change impact, consistent with the results of Melillo et al. (1993). This enhancement in plant production reduces soil C losses resulting from climatically driven changes in SOM decomposition. The direct CO₂ effect reduces C losses throughout the grassland systems and, in fact, results in a soil C sink region in the tropical savanna and humid savanna regions.

The impact of these climate and CO₂ changes on current grassland soil C stocks indicates that purely climatic effects resulted in a net loss of over 4 Pg C (Table 12.3), with all regions becoming C sources as a result of increased soil C decomposition. However, the effect of photosynthetic CO₂ increases is to offset the soil C losses due to climatic changes. The CO₂ effect reduced the net global losses from grasslands by approximately 50% in the CCC case and 66% in the GFHI case, suggesting a net C source of 1-2 Pg C over 50 years, or 0.02-0.04 Pg C yr^-l. In the tropical regions, savannas and humid savannas actually become soil C sinks with the inclusion of CO₂ and climate changes together, regardless of GCM scenario (Table 12.3; Figure 12.15).

These results are in marked contrast to those of Jenkinson et al. (1991), who estimated a net source of 61 Pg C over the next 60 years from SOM worldwide. Grassland soils (temperate steppes and tropical savannas) accounted for about 20% of present soil C stocks in this analysis, suggesting a significant grassland source of about 12 Pg C over 60 years, or 0.2 Pg C yr^-1.

12.3.2.3 Transient responses

The discussion to this point has been focused on net changes to system properties after a 'doubled-CO₂' climate impact or atmospheric CO₂ increase has been established. The short-term transient nature of ecosystem responses and the long-term stabilization of ecosystem properties due to the climate and CO₂ effects are also critical to understanding grassland responses to climate and atmospheric perturbations.

Analysis of the dynamics of the NPP and net N mineralization rates during the ramp and the near-term response (i.e. initial 50 years of climate change) indicate a strong correspondence to climatic changes. The NPP is positively correlated to rainfall dynamics. The SOM C and N dynamics appear to be more related to temperature changes; however, in the drier sites, the SOM dynamics appear to also be controlled by moisture conditions. The long-term average increase in NPP is controlled by N availability from SOM, so that changes in litter quality over time has a greater impact on N turnover. This long-term impact of litter quality changes is observed after several l00 years of enriched CO₂ simulations.

Plant production has stabilized 150 years after the start of climate change, SOM has not reached a new steady state, although the rate of change in SOM has decreased. Comparison of transient and long-term (5000-year) equilibrium levels of SOM shows that only 30% of the long-term change in SOM has occurred after 150 years. These comparisons made across the individual ecoregions show that the 50-75-year changes in SOM levels are greatest in the warm humid zones (50, 35,30,25,20, 15, and 10% of the long-term equilibrium SOM changes, for the humid savanna, savanna, humid temperate, dry savanna, Mediterranean temperate steppe, and cold desert steppe ecoregions, respectively). These changes correspond to the regional decomposition rates and agree with analyses indicating that due to faster turnover rates in these sites will approach equilibrium SOM levels more rapidly (Schimel et al. 1994).

12.3.2.4 Detecting changes in grasslands

The ability to detect the impact of climate changes on plant production and soil C levels was evaluated using the CENTURY model results. The 25-year average annual plant production and soil C levels for all of the 31 sites prior to climate change were compared with the 25-year average values following the start of climate change for both of the GCM scenarios, and a statistical t-test was used to determine if the mean values were significantly different at the 5% level.

Only 8% of the 62 model runs showed significantly different plant production (predominantly those with low productivity), with plant production changes averaging 22% over 25 years (range 17-31 %). The nonsignificant comparisons have an average change in plant production of 7% (range 0-16%); 81% of the comparisons showed significant changes in mean soil C, with soil C changes averaging 7% in 25 years (range 1-33%). Nonsignificant comparisons had soil C changes of less than 1 %.

These results suggest that soil C and plant production would have to change by at least 1 and 16%, respectively, over 25 years in order to be statistically detectable. Note that it would be impossible to detect a 1% change in soil C in the field because of the large spatial variability in soil C. We repeated the analysis for years 25-50, 50-75, and 75-100 following the start of climate change, but the number of comparisons showing significant differences in mean plant production and soil C did not change with time.

Changes in plant production or soil C induced by climate change and increasing atmospheric CO₂ will, therefore, have to be substantial, overcoming interannual variability and sampling variance, respectively, in order to be detectable. Such changes are likely to occur only at selected grassland locations.

Figure 12.15 Worldwide distribution of changes in soil C for GFHI climate change scenario and CO₂ effect combined. Map showing areal distribution of soil C changes (+5 to +10%, 0 to –5%, -5 to –10%, –10 to –15%, –15 to –20%, <–20%)

12.4 CONCLUSIONS

GRASS and the CENTURY modeling experiments indicated that GCM predictions of precipitation have considerable influence on predicted ecosystem responses. The results indicate that climate change alone accounted for an increase in total above- and below ground production for the mesic regions (humid temperate and Mediterranean), mainly attributable to increased N mineralization, and decreased plant production in the cold desert steppe regions. Soil organic matter decreased in all the mesic and colder regions, due to increased decomposition. In line with most GCM predictions, the tropical savanna regions were affected the least. However, there is large uncertainty in GCM precipitation predictions. Since many of the predicted ecosystem responses to changed climate are due to altered precipitation, this uncertainty undermines confidence in ecological model predictions.

Elevated CO₂ may cause relatively small increases in NPP at some sites and large increases at others. Soil organic matter had the greatest proportional increase in tropical savanna regions. Plant responses to CO₂ are modified in complex ways by moisture and nutrient availabilities. In general, the results suggested that CO₂-enrichment has a greater effect with higher soil moisture stress. However, nutrient limitations reduced CO₂ responses. Similarly, response predictions that are based solely on C₃ vs C₄ photosynthetic characteristics are likely to be oversimplifications. Responses of C₄ and C₃ are each modified by moisture stress and temperature.

Elevated CO₂ counteracts negative effects of increased temperatures and decreased precipitation. The reversed negative impact of elevated temperatures on C₃ species is particularly notable and is probably caused by a shift in photosynthetic temperature optima (Long 1991a, b). The net effect of climate change and CO₂ was a significant increase in NPP in mesic regions (attributable to N mineralization) as well as in dry savannas, with little or no net change in cold desert steppe or humid tropical regions. Overall, SOM showed a decrease, especially in temperate steppes and cold desert steppes due to stimulation of decomposition by both climate change and CO₂, but tropical savanna and humid savanna regions were actually soil C sinks, regardless of GCM scenario. The increased importance of CO₂ in dry years is important, in that negative effects of periodic droughts could be ameliorated, thus reducing interannual variability in NPP.

Ecosystem modeling clearly showed that indirect effects and interactions among the many processes involved are important. Responses to single variables were modified by responses to other variables. Positive effects of predicted increases in rainfall are diminished by increased temperature, for example. At the Kenyan site, the GRASS model predicted a 6% decrease in NPP due to the GFHI climate change alone, despite a + 72.0 mm yr^-1 increase in rainfall. Interactions among plant and soil processes elicited system-level responses. Thus, while 2 x CO₂ increased N_min in Kenya it decreased N_min in Colorado due to differences in the indirect effects of plants on soil moisture and decomposition rate at the two sites.

Climate change alone predicts a C loss of 3-4 Pg after 50 years of climate change. However, the CO₂ enhancement effect, amounting to 2 Pg over the same time, results in a smaller net loss of 1-2 Pg over 50 years. These numbers are substantially lower than previous estimates of Jenkinson's.

12.5 FURTHER INTER-SITE COLLABORATION

Members and associates of the SCOPE grassland modelling group (SCOPE- GRAM) have contributed data from established grassland study sites in a wide variety of locations, covering most of the major grassland regions worldwide. The SCOPEGRAM group continues to actively seek collaboration with grassland specialists worldwide who have sufficient data to parameterize the CENTURY model. Peak aboveground biomass is required for several years at the same site, or more complete data on monthly biomass and dead matter (above- and below ground). At least 10 years of temperature and precipitation data are needed, together with determinations of plant and soil C and N. Opportunities for exchange of staff and data between collaborating research groups have been encouraged, in keeping with the aims of the International Geosphere-Biosphere Programme (IGBP) strategy of establishing regional research networks. The SCOPEGRAM group is keen to promote cross-model comparisons and the integration of compatible modules from different models, with a view to developing a generic terrestrial community model framework.

12.6 ACKNOWLEDGEMENTS

Data analyses conducted by Mr Brian Newkirk and Ms Song Bo (visiting scholar sponsored by the China Committee of Scholarly Exchange). Graphics prepared by Becky Techau and Michele Nelson. Manuscript prepared by Kay McElwain.

Measurement of monthly biomass dynamics, NPP and SOM at the tropical grassland sites in Kenya, Thailand and Mexico was carried out UNEP Project FP/6108-88- 01(2855) 'Environment Changes and Productivity of Tropical Grasslands' (1989-92), and more recently at the Kenya and Mexico sites under the UK Overseas Development Administration (ODA) Project R4744 'Productivity of Tropical Grasslands' (1991-94). Analysis of data from the sites in the former USSR was carried out under the Russian National Scientific and Technical Programme No.18 'Changes of Natural Environment and Climate'. Data synthesis and model validation were made possible by the SCOPE (Scientific Committee on Problems of the Environment) Project 'Effects of Climate Change on Production and Decomposition in Coniferous Forests and Grasslands' (1989-92); and the National Institute for Global Environmental Change (NIGEC) Midwestern Region Center, US-DOE project on 'Biological Hysteresis in Climate change Models for Grasslands: Implications of plant community dynamics on biogeochemical feedbacks'. Model development was primarily funded by the US National Science Foundation (NSF) project BSR 9013888 'Coupling Ecosystem Processes and Vegetation Patterns across Environmental Gradients' and the US NASA Earth Observing System project NACW-2662 'Using Multi-Sensor Data to Model Factors Limiting Carbon Balance in Global Grasslands', and US NSF project BSR 9011659 'Long Term Ecological Research Program: Shortgrass Steppe' and the Ecological Research Division, Office of Health and Environmental Research, US Department of Energy (DOE) project, 'Grassland-Atmosphere Response to Changing Climate: Coupling Local and Regional Scales'.

12.7 REFERENCES

Anderson, I. M. (1991) The effects of climate change on decomposition processes in grassland and coniferous forests. Ecol. Appl. 1, 326-347.

Bailey, R. G. (1989) Explanatory supplement to ecoregions map of the continents. Environ. Conserv. 16, 307-309.

Ball, J. T., Woodrow, I. E. and Berry, I. A. (1987) A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions. In: Biggins, I. (Ed.) Progress in Photosynthesis Research, pp. 221-224. Vol. IV. Martinus Nijhoff, Dordrecht.

Bazazz, F. A., Garbutt, K., Reekie, E. G. and Williams, W. E. (1989) Using growth analysis to interpret competition between a C₃ and a C₄ annual under ambient and elevated CO₂. Oecologia 79, 223-235.

Breymeyer, A. and Melillo, J. M. (1991) The effects of climate change on production and decomposition in coniferous forests and grasslands. Ecol. Appl. 1, 111.

Carter, D. R. and Peterson, K. M. (1983) Effects of a CO₂-enriched atmosphere on the growth and competitive interaction of a C₃ and C₄ grass. Oecologia 58, 188-193.

Chen, D., Coughenour, M. B., Knapp, A. K. and Owensby, C. E. (1993) Mathematical simulation of C₄ grass photosynthesis in ambient and elevated CO₂. Ecol. Model. 73, 63-80.

Chen, z. (1988) Topography and climate of the Xilin river basin. In: Inner Mongolia Research Station (Ed.) Research on Grassland Ecosystem, Vol. 3, pp. 13-22. Science Press, Beijing (in Chinese with English abstract).

Chuluun, T., Ojima, D. S., Luvsandorjiin, I., Dodd, I., and Williams, S. (1996) Simulation studies of grazing in the Mongolian steppe. In: West, N. E. (Ed.) Proceedings of the Fifth International Rangeland Congress, pp. 561- 562 (Volume 1 ). Society of Range Management, Denver, CO, USA.

Collatz, G. I., Ribas-Carbo, M. and Berry, I. A. (1992) Coupled photosynthesis-stomatal conductance model for leaves of C₄ plants. Austr. J. Plant Physiol. 19, 519-538.

Coughenour, M. B. (1984) A mechanistic simulation analysis of water use, leaf angles, and grazing in east African graminoids. Ecol. Model. 26,203-230.

Coughenour, M. B., McNaughton, S. J. and Wallace, L. L. (1984) Modelling primary production of perennial graminoids: uniting physiological processes and morphometric traits. Ecol. Model. 23, 101-134.

Enting, I. G. and Mansbridge, I. V. (1991) Latitudinal distribution of sources and sinks of CO₂: results of an inversion study. Tellus 43B, 156-170.

Eswaran, H., van den Berg, E. and Reich, P. (1993) Organic carbon in soils of the world. Soil Sci. Soc. Am. J. 57,192-194.

Farquhar, G. D., von Caemmerer, S. and Berry, J. A. (1980) A biochemical model of photosynthetic CO₂ assimilation in leaves of C₃ species. Planta 149, 78-90.

Fernandez, R. J., Sala, O. E. and Golluscio, R. A. (1991) Woody and herbaceous above-ground production of a Patagonian steppe. J. Range Manage. 44, 434--437.

Gilmanou, T. G., Parton, W. J. and Ojima, D. S. (in press). Testing the CENTURY ecosystem level model on data sets from eight grassland sites in the former USSR representing a wide climatic/soil gradient. Ecological Modeling (in press).

Haas, H. J., Evans, C. E. and Miles, E. F. (1957) Nitrogen and carbon changes in Great Plains soils as influenced by cropping and soil treatments. Tech. Bull. 1164. United States Department of Agriculture. US Printing Office, Washington, DC, USA.

Hall, D. O. and Scurlock, J. M. O. (1991) Climate change and productivity of natural grasslands. Ann. Bot. 67(suppl.), 49-55.

Houghton,J. T.,Jenkins, G. J. and Ephraums,J. J. (Eds)(1990) Climate Change: The IPCC Scientific Assessment. Cambridge University Press, Cambridge, UK.

Jenkinson, D. S., Adams, D. E. and Wild, A. (1991) Model estimates of CO₂ emissions from soil in response to global warming. Nature 351, 304-306.

Lauenroth, W. K. and Dodd, J. L. (1978) The effects of water- and nitrogen-induced stresses on plant community structure in a semiarid grassland. Oecologia 36, 211-222.

Long, S. P. (1991a) Modification of the response of photosynthetic productivity to rising temperature by atmospheric CO₂ concentrations: has its importance been underestimated? Plant Cell Environ. 14, 729-739.

Long, S. P. (1991b) Primary production in grasslands and coniferous forests with climate change: an overview. Ecol. Appl. 1, 139-156.

Long, S. P., Garcia Moya, E., Imbamba, S. K., Kamnalrut, A., Piedade, M. T. F., Scurlock, J. M. O., Shen, Y. K. and Hall, D.O. (1989) Primary productivity of natural grass ecosystems of the tropics: a reappraisal. Plant Soi1 115, 155-166.

Long, S. P., Jones, M. B. and Roberts, M. J. (1992) Primary Productivity of Grass Ecosystems of the Tropics and Sub-tropics. Chapman & Hall, London. 267 pp.

Melillo, J. M., Aber, J. D. and Muratore, J.F. (1982) Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecology 63, 621-626.

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

Menaut,J.-C. and Cesar,J. (1979) Structure and primary productivity of Larnta savannas, Ivory Coast. Ecology 60, 1197-1210.

Ojima, D. S., Parton, W. J., Schimel, D. S. and Owensby, C. E. (1990) Simulated impacts of annual burning on prairie ecosystems. In: Collins, S. L. and Wallace, L. (Eds) Fire in the North American Prairies, pp. 118-132. Univ. of Oklahoma Press, Norman.

Ojima, D. S., Schimel, D. S., Parton, W. J. and Owensby, C. E. (1994) Long- and short-term effects of fire on nitrogen cycling in tallgrass prairie. Biogeochemistry 24, 67-84.

Owensby, C. E., Coyne, P. I. and Auen, L. M. (1993a) Nitrogen and phosphorus dynamics of a tall grass prairie ecosystem exposed to elevated carbon dioxide. Plant Cell Environ. 16, 843-850.

Owensby, C. E., Coyne, P. I., Ham, J. M., Auen, L. M. and Knapp, A. (1993b) Biomass production in a tallgrass prairie ecosystem exposed to ambient and elevated levels of CO₂. Ecol. Appl. 3, 644-653.

Parton, W. J., McKeown, B., Kirchner, V. and Ojima, D. S. (1992) CENTURY Users' Manual. NREL Publication. CSU, Fort Collins, Colo.

Parton, W. J., Schimel, D. S., Cole, C. V. and Ojima, D. S. (1987) Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci. Soc. Am. J. 51, 1173-1179.

Parton,W.J., Scurlock,J. M.O., Ojima, D. S., Gilmanov, T. G., Scholes, R. I., Schimel, D. S., Kirchner, T., Menaut, J. -C., Seastedt, T., Garcia Moya, E., Kamnalrut, A. and Kinyamario, J. I. (1993) Observations and modelling of biomass and soil organic matter dynamics for the grassland biome worldwide. Glob. Biogeochem. Cyc. 7(4), 785-809.

Parton, W. J., Stewart,J. W. B. and Cole, C. V. (1988) Dynamics of C, N, P and S in grassland soils: a model. Biogeochemistry 5, 109-131.

Rains, J. R., Owensby, C. E. and Kemp, K. (1975) Effects of nitrogen fertilization, burning, and grazing on reserve constituents of big bluestem. J. Range Manage. 28,358-362.

Rice, C. W., Garcia, F. O., Hampton, C. O. and Owensby, C. E. (1992) Soil microbial biomass and respiration under increased levels of atmospheric CO₂. Agron. Abstr. 84: p.258.

San Jose,J.J. and Medina, E. (1976) Organic matter production in the Trachypogon savanna at Calbozo, Venezuela. Trop. Ecol. 17,13-124.

Schimel, D. S., Braswell, B. H., Holland, E. A., McKeown, R., Ojima, D. S., Painter, T. H., Parton, W. J. and Townsend, A.R. (1994) Climatic, edaphic, and biotic controls over storage and turnover of carbon in soils. Glob. Biogeochem. Cyc. 8, 279-293.

Schimel, D. S., Kittel, T. G. F. and Parton, W. I. (1991) Terrestrial biogeochemical cycles: global interactions with the atmosphere and hydrology. Tellus 43, 188-203.

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

Schlesinger, W. H. (1990) Evidence from chronosequence studies for a low carbon-storage potential of soils. Nature 348, 232-234.

Singh,J. S. and Yadava, P. S. (1974) Seasonal variation in composition, plant biomass and net primary productivity of a tropical grassland at Kurukshetra, India. Ecol. Monogr. 44,351-376.

Tans, P. P., Fung, I. Y. and Takahashi, T. (1990) Observational constraints on the global atmospheric CO₂ budget. Science 247, 1431-1438.

Towne, G. and Owensby, C. E. (1984) Long-term effects of annual burning at different dates in ungrazed Kansas tallgrass prairie. J. Range Manage. 37,392-397.

Wang, Y. and Jiang, S. (1982) Effects of arid climate on community structure and aerial biomass of Stipa grandis steppe. Acta Phytoecol. Geobot. Sin. 13, 297-308.