AGCI Session II: Characterizing and Communicating Scientific Uncertainty
Session Chairs: Dr. Richard H. Moss and Dr. Stephen H. Schneider
July 31 to August 8, 1996
Terrestrial Ecosystems: Controversies in Assessing Climate Impacts
Allen M. Solomon
U. S. Environmental Protection Agency
Corvallis, Oregon
Solomon discussed the primary controversies in predictions concerning the future dynamics of terrestrial ecosystems in response to climate change. Included in this discussion were controversial elements of the IPCC Second Assessment Report from Working Group II on forestry (Chapter 15) and the differences in the global carbon cycle to be expected when one uses vegetation simulation models that deal with responses to climate change by individual species dynamics versus by whole ecosystems grouped into 15 or 20 primary global biomes.
The fundamental issues in terrestrial biospheric responses to changing climate involve (1) the decline of terrestrial biodiversity and (2) the role of the terrestrial biosphere in the global carbon cycle. Biodiversity was discussed in this AGCI session by Terry Root. The carbon cycle is a problem because too little is known at present to define either the current or future role of the terrestrial biosphere as that of a carbon source or a carbon sink. Based on ice core data and the greater activity of respiration than of photosynthesis with warmer temperatures, there is good reason to suspect the terrestrial biosphere has been a source of CO2 during warmings of the past 250,000 years, producing a positive climate feedback leading to further warming. Yet, the physiological basis for this logic is weak at best. Equally compelling (and uncertain), Solomon says, is the view that CO 2 limits growth of plants under current conditions so that increased CO 2 concentrations will "fertilize" plants, inducing an increased carbon sink with coincident warming, providing a negative feedback on the global carbon cycle.
The carbon cycle
is a problem because too little is known at present to define either
the current or future role of the terrestrial biosphere as that of a
carbon source or a carbon sink.
Since the middle 1970s, the "missing fraction" of CO 2 that calculations of fossil fuel burning and ocean uptake predict in the atmosphere, has been attributed to the terrestrial biosphere, with CO 2 fertilization being the hypothesized mechanism for carbon uptake. Although recent calculations by Dai and Fung (1993, Snowmass Carbon Cycle Meeting) and by Keeling et al. (1996, Nature) indicate that seasonal warming at high latitudes could account for much of the missing fraction, CO2 fertilization is still the preferred mechanism of most oceanographers and others who study the global carbon cycle ( e. g., Tans, Peng, Schimel et al. as IPCC SAR from WG I, etc.).
This dependence on the CO2 fertilization mechanism is not unlike a religious belief; in the absence of either indicative or definitive evidence, one must posit faith that the mechanism seen in herbaceous annuals and tree seedlings in greenhouse pot experiments will also be functioning (and doing so in the same way) across the process scales that link:
1. population dynamics of seedlings to survival, growth and reproduction of mature trees,
2. the interactions among mature trees with one another through competition for light, water and nutrients in forest stands,
3. the interaction of stands via exchange of species and materials to form ecosystems, and
4. the interactions among forest ecosystems through groundwater, and
5. through propagules, to form landscapes and biomes.
Forests are the focus of the carbon fertilization controversy because most of the terrestrial biospheric carbon is tied up in forest trees and soils.
Though seedlings
in greenhouses fumigated with CO 2 absorb extra carbon, this does not
cross the scale to affect processes which are important in mature
trees. Instead, the measured advantage declines continuously such
that no "fertilization" effect is visible after one or two decades.
The necessary long-term (multiple decade) fumigation experiments have not been done to demonstrate such a robust process transfer. Indeed, data from the very few field situations in which mature trees have been fumigated for their lifetimes by adjacent CO 2 fumaroles (experiments in Italy and Iceland) suggest that the excess carbon uptake measured in seedlings in greenhouses also occurs in nature, but that it does not cross the scale to affect processes which are important in mature trees. Instead, the measured advantage declines continuously such that no "fertilization" effect is visible after one or two decades.
Whether or not rising atmospheric CO2 is being increasingly sequestered in terrestrial vegetation, a second group of processes of change may release a significant amount of carbon from terrestrial vegetation during the coming decades to a century or two. These processes are the "transient" lags in response of forests to rapid climate change. First, forest succession (the developmental sequence of forest species from those which grow most quickly and are least tolerant of shade, through those which grow most slowly and are most tolerant of shade) is a slow process. It requires 50-500 years in contrast to the 1 - 10 years required for dieback of trees, such as may be expected from a chronically changing climate.
Second, continuous climate change can be expected to eventually render local tree species "climatically obsolete" requiring reforestation with species not initially present. The immigration of trees (including the transport of seeds, establishment of seedlings, growth of trees to maturity, and transport of seeds from the new trees) has been measured at 15-40 km/century during the last 10,000 years in temperate areas, although occasional "bursts" of speed beyond 100 km/century have been recorded. In contrast, the mean July isotherm in temperate areas is simulated to migrate 200-400 km/century. The lags induced in carbon uptake generated by the lack of suitable trees, and by slow forest development processes, suggests a "pulse" of carbon released from dying forests, but not retrieved from the atmosphere in new forests for many decades or centuries.
The lags induced
in carbon uptake generated by the lack of suitable trees, and by slow
forest development processes, suggests a "pulse" of carbon released
from dying forests, but not retrieved from the atmosphere in new
forests for many decades or centuries.
The means by which these processes (carbon fertilization, transient responses) may control the role of the terrestrial biosphere in the global carbon cycle has been characterized in simulation models designed to determine the implications of those multiple competing processes. One set of models which treat the terrestrial biosphere as a "big leaf" assume the religious tenant of a constant relationship between carbon concentration and carbon uptake, while taking no cognizance of forest "infrastructure" changes (e. g., TEM, GEM, BIOME-BGC). Other more recent models also simulate competing processes of photosynthesis and respiration rates and differences among different physiognomic biomes (e. g., temperate deciduous forests, tropical savannas, etc.), which are in turn defined by climate (BIOME2, DOLY, MAPSS). All of these models either do not redistribute biomes' geography with climate change, or else they do so instantaneously, as though the transient processes do not exist.
Models which simulate life histories of individual tree species or species groups may be more accurate in defining the impacts of transient processes on the temporal dimension of terrestrial carbon. Models which include forest succession (gap models such as JABOWA, FORET, etc.) thus far have not been applied to the globe as a whole because the natural history information needed to parameterize them is lacking in poorly studied regions, such as many tropical areas. Tree migration models are currently under development in several research groups but none are available to apply at the global scale required to calculate implications of these processes on global carbon cycle dynamics. It seems probable that credible, globally-comprehensive vegetation models which simulate transient process mechanisms will be available within the next 5 to 10 years.
Hence, there is considerable optimism that enough is known to construct accurate carbon cycle models, and that the models will soon reveal the critical answers concerning the role of the terrestrial biosphere in the global carbon cycle. However, the answers may come from new ecological data rather than from how we manipulate the data we have. For example, those data may reveal that the processes modeled are not the processes driving the system; perhaps we will learn that the rhizosphere contains the missing fraction of atmospheric carbon, and that carbon allocation to leaves, stems, and roots within individual plants is a central process defining biospheric carbon content. Whatever the result, we must confront the fact that our understanding of the terrestrial carbon cycle is data-limited, not model-limited, Solomon says.
We must confront
the fact that our understanding of the terrestrial carbon cycle is
data-limited, not model-limited, Solomon says.