The Regional Impacts of Climate Change

Other reports in this collection

C.1. Introduction

Global vegetation models (GVM) have in the past decade evolved from largely statistical-correlational to more process-based, rendering greater confidence in their abilities to address questions of global change. There are generally two classes of GVMs, biogeography models and biogeochemistry models. The biogeography models place emphasis on determination of what can live where, but either do not calculate or only partially calculate the cycling of carbon and nutrients within ecosystems. The biogeochemistry models simulate the carbon and nutrient cycles within ecosystems, but lack the ability to determine what kind of vegetation could live at a given location. BIOME3 has significantly blurred this model distinction (Haxeltine and Prentice, 1996). There are over 20 biogeochemistry models and about 5 biogeography models. Two of the biogeography models, MAPSS (Neilson, 1995) and BIOME3 were used to provide estimates of changes in vegetation distribution, density and hydrology for this IPCC special report. These are equilibrium models, which simulate the potential 'climax' vegetation that could live at any well-drained, upland site in the world under an 'average' seasonal climate. Equilibrium models provide useful 'snapshots' of what a terrestrial biosphere in equilibrium with its climate might look like, but can provide only inferential information about how the biosphere will make transitions from one condition to another. This is in contrast to other models, which simulate the timeseries of vegetation change at a point (Shugart and Smith, 1996), but which do not produce maps of vegetation distribution and function. Fully dynamic versions of the spatially-explicit GVMs are being developed and incorporate both biogeography and biogeochemistry processes, but the dynamic global vegetation models (DGVM) are not yet ready for assessment purposes (Neilson and Running, 1996). Several model intercomparison projects are underway and can serve to provide some context for the two models used here. One such intercomparison is the VEMAP process.

C.2. VEMAP Model Intercomparison

The Vegetation/Ecosystem Modeling and Analysis Project (VEMAP) compared three biogeography models, MAPSS (Neilson, 1995), BIOME2 (Haxeltine et al., 1996), and DOLY (Woodward and Smith, 1994; Woodward et al., 1995) and three biogeochemistry models TEM (Raich et al., 1991; McGuire et al., 1992; Melillo et al., 1993), CENTURY (Parton et al., 1987; Parton et al., 1988; Parton et al., 1993), and BIOME-BGC (Hunt and Running, 1992; Running and Hunt, 1993). The two classes of global models were intercompared and loosely coupled for an assessment of both model capabilities and the potential impacts of global warming on U.S. ecosystems (VEMAP Members, 1995). The VEMAP process determined that all the models have roughly equal skill in simulating the current environment, but exhibit some divergences under alternative climates, in some cases producing vegetation responses of opposite sign.

Given the timeframe of this IPCC special report, only MAPSS and BIOME3 were able to provide global simulations. MAPSS and BIOME2 (a precursor to BIOME3) were found to produce generally similar results under the future climate scenarios of the VEMAP process. However, MAPSS is consistently more sensitive to water stress, producing a more xeric outcome under future climates and it also has a more sensitive response to elevated CO2. That is, when incorporating a direct, physiological CO2 effect, MAPSS produces a larger benefit to vegetation from increased water-use-efficiency (VEMAP Members, 1995).

C.3. Biogeography Model Description

MAPSS (Mapped Atmosphere-Plant-Soil System; Neilson, 1995) and BIOME3 (Haxeltine and Prentice, 1996) are among a new generation of process-based, equilibrium biogeographic models (IPCC 1996, WG II, Section 1.3.4; VEMAP Members, 1995). The models simulate the distribution of potential global vegetation based on local vegetation and hydrologic processes and the physiological properties of plants. Both models simulate the mixture of vegetation lifeforms, such as trees, shrubs, and grasses, that can coexist at a site while in competition with each other for light and water. A set of physiologically-grounded 'rules' determines whether the woody vegetation will be broadleaved or needleleaved, or evergreen or deciduous, as well as other properties. The models also simulate the maximum carrying capacity, or vegetation density, in the form of leaf area that can be supported at the site, under the constraints of energy and water. Energy constraints, largely applicable to cold ecosystems, are prescribed in MAPSS, but are simulated by an explicit carbon flux model in BIOME3. The two models simulate a similar set of water balance processes, incorporating soil texture effects.

Thus, MAPSS and BIOME3 simulate the distribution of vegetation, such as forests, savannas, shrubland, grasslands and deserts over all non-wetland sites of the Earth, based on the relative densities or productivity of overstory and understory, vegetation leaf characteristics and thermal tolerances. The models simulate the distribution of generalized vegetation lifeforms (e.g. tree, shrub, grass; evergreen-deciduous; broadleaf-needleleaf), rather than species and assemble these into a vegetation type classification. There are currently 45 different vegetation types simulated by MAPSS and 18 by BIOME3. The vegetation types are hierarchical, representing biomes (e.g., boreal forest, temperate savanna, grassland, etc.) at the top and more detailed community-level descriptions at the lower end (e.g., subtropical, xeromorphic woodland). Only the top of the hierarchy is utilized in this analysis. Since both models simulate a full site water balance, they are also calibrated and tested hydrologic models, thereby, allowing estimates of impacts on water resources fully integrated with the simulated impacts on vegetation (Neilson and Marks, 1994; VEMAP Members, 1995). As equilibrium models, MAPSS and BIOME3 simulate vegetation distribution and hydrology under an average seasonal cycle of climate. They simulate an equilibrium land-surface biosphere under current or future climate, but not the transitional vegetation changes from one climate to another. Thus, the models show the long term potential consequences of climate change, but one can only infer immediate (1-10 year) effects.

MAPSS and BIOME3 contain algorithms that allow the incorporation of a direct physiological CO2 effect. Elevated CO2 concentrations can, among other effects, enhance productivity and increase the water-use-efficiency (WUE, carbon fixed per unit water transpired) of the vegetation thereby reducing the sensitivity of the vegetation to drought stress (IPCC 1996, WG II, Section A.2.3; Bazzaz et al., 1996; Eamus, 1991). BIOME3 allows a direct CO2 effect on both productivity and water-use-efficiency. MAPSS accomplishes the same effect by reducing stomatal conductance, which then results in increased leaf area, thus indirectly incorporating a productivity effect. Individual species exhibit variations in their expressions of direct CO2 effects. For example, the productivity and WUE effects are not necessarily tightly coupled (Eamus, 1996a). However, since the models simulate functional types, rather than species, both models have generalized the direct CO2 effects to all vegetation, with the exception in BIOME3 of differentiating C3 and C4 physiological types. In BIOME3 C4 plants do not experience the elevated growth of C3 plants, but do experience increased water-use-efficiency. The realized importance of the direct CO2 processes in complex, mature ecosystems remains a matter of debate (Bazzaz et al., 1996).

A review of 58 studies indicated an average 32% increase in plant dry mass under a doubling of CO2 concentration (Wullschleger, Post, and King, 1995). Norby (1996) documented an average 29% increase in annual growth per unit leaf area in seven broadleaf tree species under 2 x CO2 over a wide range of conditions. Increased WUE, examined in another review, averaged about 30-40% as indexed by reductions in leaf conductance to water vapor (Eamus, 1991). If such responses were maintained in forests over many decades, they would imply a substantial potential for increased storage of atmospheric carbon, as well as conferring some increased tolerance to drought due to increased WUE. However, some species or ecosystems exhibit acclimation to elevated CO2 by downregulating photosynthesis (Bazzaz, 1990; Grulke et al., 1993; Grulke et al., 1990); while others do not exhibit acclimation (Bazzaz, 1990; Teskey, 1997). Most of the early CO2 research was done on juvenile trees in pots and growth chambers. New research is beginning to emerge which focuses on larger trees or intact forested ecosystems. Recent reviews of this newer literature (Eamus, 1996a; Curtis, 1996) indicate that acclimation may not be as prevalent when roots are unconstrained and also that leaf conductance may not be reduced and that both responses are dependent on the experimental conditions, the length of exposure and the degree of nutrient or water stress. These results imply that forests could produce more leaf area under elevated CO2, but may not gain a benefit from increased WUE. In fact, with increased leaf area, transpiration should increase on a per tree basis and the stand would use more water. Elevated temperatures would increase transpiration even further, perhaps drying the soils and inducing a drought effect on the ecosystem (ibid). Prominent among the environmental influences that are thought to moderate long-term responses to elevated CO2 is nitrogen supply (Kirschbaum et al., 1994; McGuire et al., 1995; Eamus, 1996b). Unless CO2 stimulates an increase in N mineralization (Curtis et al., 1995), productivity gains in high CO2 are likely to be constrained by the system's N budget (K�rner, 1995). Nitrogen limitations may constrain carbon gains to structural tissue, rather than leaves (Curtis et al., 1995). Thus, in areas receiving large amounts of N deposition, a direct CO2 response could result in large increases in leaf area, increasing transpiration and possibly increasing sensitivity to drought via rapid soil water depletion. Early growth increases may disappear as the system approaches carrying capacity as limited by water or nutrients (K�rner, 1995). Shifts in species composition will likely result from different sensitivities to elevated CO2 (Bazzaz et al., 1996; K�rner, 1995). Both MAPSS and BIOME3 have been operated with and without the direct CO2 effects in this study in order to gauge the importance and sensitivity of the processes within the modeling framework. The direct effects of elevated CO2 are imparted only in the ecological model processes and not in the GCMs. There are no feedbacks between the ecological and atmospheric models

The MAPSS model also contains a fire model that shifts some vegetation to a 'fire climax' state, such as in many grasslands or savannas. BIOME3 embeds these processes in the calibration. Neither model considers current or past land-use practices. Thus, some areas that the models indicate as grassland, for example, might actually be shrublands, due to either grazing or fire suppression. Although the two models do not simulate actual land-use, the 'potential' land-cover simulated by the models should provide an accurate estimate of the land-surface potential. That is, forests cannot be grown in deserts or shrublands without irrigation; and, agricultural productivity should be higher in a potential forest landscape than in a potential shrubland landscape (given similar soils). Changes in LAI can be interpreted as a change in the overall carrying capacity or standing crop of the site, regardless of whether it is in potential natural vegetation or under cultivation. Additions from irrigation or nitrogen could alter this conclusion, but it should hold for non-irrigated, upland systems. Thus, simulated changes in potential natural vegetation should be valuable indicators of general shifts in agricultural potential.

Other reports in this collection