C.7. Conclusions

MAPSS and BIOME3 produce qualitatively similar results under alternative future climate scenarios, with or without including a direct, physiological CO₂ effect. However, MAPSS produces consistently stronger drought effects with increasing temperature than does BIOME3. When under common scenarios, the two models produce similar subregional sensitivities. That is, if two adjacent subregions show opposite sign responses under a future climate in MAPSS, they will tend to exhibit the same relationship in BIOME3, but the overall sensitivity will be lower in BIOME3. The newer scenarios, constructed from transient GCM experiments, are consistently less xeric, as measured by simulated changes in leaf area index (LAI), than the older, equilibrium GCM scenarios. Under the newer scenarios, both ecological models indicate an overall increase in LAI (although nutrient constraints could limit or delay the increase), when a direct physiological CO₂ effect is included. However, if the direct CO₂ effects are not included, both models indicate a general reduction in global vegetation density.

The changes in vegetation leaf area (LAI) simulated by both MAPSS and BIOME3 are analogous to the changes in soil water content reported by earlier GCM experiments (IPCC 1996, WG I, Section 6). Those earlier experiments maintained a fixed vegetated land surface. That is, the vegetation type and density were not allowed to respond to changes in either climate or elevated CO₂ concentration. Therefore, as evaporative demand went up in those simulations, soil water content decreased, or was totally depleted. MAPSS and BIOME3, however, absorb those processes directly in the vegetation response. Both models simulate water-limited LAI by maximizing the LAI that can be supported and just barely transpire available soil water. Thus, in the equilibrium solution to LAI, soil water is fully utilized and can't change much under altered climates. Changes in the site water balance are, therefore, indicated by changes in LAI.

The newer climate scenarios used in this analysis are relatively cool in comparison to other possible new scenarios (IPCC 1996, WG I, Section 6). Therefore, the analyses presented here must be considered as a relatively conservative subset of the possible future ecological responses.

Although many of the simulations from both MAPSS and BIOME3 indicate potentially large expansions of tropical and in some cases temperate forests, actual expansions would be limited by urban and agricultural land-use constraints, unsuitable soils in some areas and slow dispersal rates, among other factors. Even so, if a forest is anticipated to expand into a region formerly indicated as shrubland, any agriculture in the region might expect an increase in potential productivity, and vice versa. Such changes between forest and shrubland are usually underlain by a change in LAI, which reflects the site water balance. An increase or decrease in LAI indicates a change in the water or energy balance and the potential biomass density or carrying capacity that could be supported on the site, regardless of whether the biomass is 'natural' or agricultural (dryland agriculture only).

The results presented here are for steady-state, or equilibrium conditions and do not directly indicate how the systems would behave in their transient responses toward a new equilibrium. For example, in areas where LAI is indicated to decline, it may be that equilibrium runoff is indicated to increase. However, one hypothesis is that during the processes of LAI decline, increased evaporative demand could cause reductions in runoff, before the vegetation becomes sufficiently drought-stressed for the LAI to be reduced. After further time, if the vegetation is sufficiently drought stressed, a rapid dieback could occur and might be facilitated by pests and fire. Were vegetation to undergo such a large dieback, then transpiration demand would be temporarily reduced and runoff could increase substantially. Thus, before a new equilibrium is attained with new vegetation growth, streams could go through a dry to wet oscillation. These possible hydrologic responses to vegetation change are, however, of a different timeframe (years) than possible short term floods and droughts that could occur simply due to increased variation of extreme weather events (IPCC 1996, WG I, Section 6).

At least two contrasting, transient trajectories of vegetation change are possible. If a large, direct CO₂ benefit were to occur, vegetation could increase in growth and biomass under relatively cool, early warming conditions, only to experience drought-stress and decline or dieback under the hotter, later stages of warming. Alternatively, if direct CO₂ benefits are more muted, vegetation could become drought-stressed and experience decline or dieback within the next few decades even under mild warming. Expansion of forests into cooler zones would likely lag behind decline and dieback in warmer zones, producing a transient reduction in forest area, possible increases in pests and fire, and possibly large releases of CO₂ to the atmosphere (King and Neilson, 1992; Smith and Shugart, 1993).

Other reports in this collection