Box 5-1. Plant Productivity: Terms and Definitions
Plants are responsible for the vast majority of carbon uptake by terrestrial ecosystems. Most of this carbon is returned to the atmosphere via a series of processes, including respiration, consumption (followed by animal and microbial respiration), combustion (i.e., fires), and chemical oxidation. Gross primary productivity (GPP) is the total uptake through photosynthesis, whereas net primary productivity (NPP) is the rate of accumulation of carbon after losses from plant respiration and other metabolic processes in maintaining the plant's living systems are taken into account. Consumption of plant material by animals, fungi, and bacteria (heterotrophic respiration) returns carbon to the atmosphere, and the rate of accumulation of carbon over a whole ecosystem and over a whole season (or other period of time) is called net ecosystem production (NEP). In a given ecosystem, NEP is positive in most years and carbon accumulates, even if only slowly. However, major disturbances such as fires or extreme events that cause the death of many components of the biota release greater than usual amounts of carbon. The average accumulation of carbon over large areas and/or long time periods is called net biome productivity (NBP) (see also Box 3-1 in TAR WGI).
A large literature is developing on modeling the response of ecosystems to climate and global changes. Most of these models simulate changes in a small patch of land. These models are reviewed as appropriate in other sections of this chapter and elsewhere in this report. The focus here is on modeling changes in ecosystem composition, structure, and function at global or regional scales. There are several reasons for developing such models. One is to estimate carbon fluxes and their contribution to the global carbon cycle. This involves making estimates of NPP, net ecosystem productivity (NEP), and net biome productivity (NBP) (see Box 5-1). Another is to develop models of feedbacks between the atmosphere and the land surface (Van Minnen et al., 1995; Foley et al., 1998). Neither of these applications are covered in detail in this chapter (see TAR WGI Chapter 3). Instead, we concentrate on their application in forecasting the impacts of climate change on biodiversity and the provision of other ecosystem goods and services.
There are two paradigms about the way ecosystems (thus biomes) will respond
to global change. The ecosystem movement paradigm assumes that ecosystems will
migrate relatively intact to new locations that are closer analogs to their
current climate and environment. This paradigm clearly is a gross simplification
of what will actually happen, but it has the advantage that the well-demonstrated
relationship between ecosystem range and existing climate can be used to project
new ecosystem distributions under changed climate scenarios.
Basic ecological knowledge suggests that the ecosystem movement paradigm is most unlikely to occur in reality because of different climatic tolerance of species involved, including intra-species genetic variability (Crawford, 1993); different longevities, including clonal regeneration (e.g., survival over 2,000 years by Carex curvula; Steinger et al., 1996); different migration abilities (Pitelka and Plant Migration Workshop Group, 1997); and the effects of invading species (Dukes and Mooney, 1999). It is an idealized working paradigm that is useful for screening scenarios of climate change for potential significant effects.
The alternative paradigm, ecosystem modification, assumes that as climate and
other environmental factors change there will be in situ changes in species
composition and dominance. These changes will occur as some species decline
in abundance or become locally extinct (Jackson and Weng, 1999) and others increase
in abundance. The longevity of individuals, the age structure of existing populations,
and the arrival of invading species will moderate these changes. The outcome
will be ecosystem types that may be quite different from those we see today.
Paleoecological data indicate that ecosystem types broadly similar to those
seen today did exist in the past (Pregitzer et al., 2000), but that there
also occurred combinations of dominant species that are not observed today (Davis,
1981; Jablonski and Sepkoski, 1996; Ammann et al., 2000; Prieto, 2000).
Numerous paleoecological studies provide evidence of important species within
an ecosystem responding differently to climate change. For example, the postglacial
migration pattern for Sierra lodgepole pine (Pinus contorta ssp. murrayana)
was largely elevational, with little migrational lag, whereas the more widely
distributed Rocky Mountain subspecies (P. contorta var. latifolia) migrated
both latitudinally and elevationally (Anderson, 1996). Colinvaux et al.
(1997) have interpreted pollen data to show that during glacial cooling, Andean
vegetation did not move upslope and downslope as belts but that plant associations
were reorganized as temperature-sensitive species found different centers of
distribution with changing temperature. Similarly, heat-intolerant plants have
moved in and out of the Amazonian rainforests during periods of cooling and
warming (Colinvaux et al., 2000). Colinvaux et al. (2000) also
argue that expulsion of heat-intolerant species from the lowland forests in
this postglacial warming already is complete and that the forest property of
maintaining its own microhabitat will allow high species richness to survive
more global warming, provided large enough tracts of forest are conserved. This
conclusion is at odds with some modeling studies (e.g., White et al.,
The problem with the ecological modification paradigm is that it is very difficult
to use in practical forecasting of possible trends. Thus, most global and regional
studies assessing the potential impacts of climate change have had to use the
ecosystem paradigm, as illustrated in Box 5-2. They also
tend to be limited to projecting changes in vegetation distributions, with the
implicit assumption that animal populations will track the vegetation components
of an ecosystem. However, observational and experimental studies show many cases
in which animals are responding to climate and environmental change well before
any significant changes in the vegetation (see Section 5.4).
Box 5-2. Illustration of Use and Limitation of Ecosystem Movement Models
The study by Malcolm and Markham (2000) is a good example
of modeling that uses the ecosystem movement paradigm, but it also demonstrates
the inherent weaknesses of the approach.
The study uses two models of existing ecosystem distributions
(MAPSS and BIOME3; Neilson et al., 1998) and compares predicted
distributions at present CO2 levels with the equilibrium
climate associated with doubled CO2 as projected by several
general circulation models (GCMs). It avoids the naive assertion that
the latter climate constitutes a forecast of the future distribution
of ecosystems; instead, it uses the two predictions to calculate the
necessary rate of migration (m yr-1) for species in the ecosystems
to migrate to the new locations within 100 years (other time frames
also are explored in a sensitivity analysis). It then maps these required
rates to show areas where unusually high rates may be required in the
future if a "climatically appropriate" ecosystem is to be
established (referred to as "migration-stressed" locations).
The study predicted that about 20% of the Earth's
surface will require migration rates greater than or equal to 1 km yr-1,
which is equivalent to the highest rates observed in the geological
past. The effects of natural barriers (e.g., lakes) or barriers resulting
from land-cover modification by humans are globally small but can be
regionally significant. Their approach also gives an indication of which
regions of the globe may be most likely to be migration-stressed by
climate change. It shows that much of the Earth's surface will
be "stressed" in at least one of the 14 combinations of vegetation
and climate models used. For some regions in the northern boreal zones
of Eurasia and North America, most of the models predict such stress.
The study then goes on to deal with locations where the
models predict that under climates applying a doubled CO2
scenario, current ecosystems will fall outside their climatic range
(referred to as "climate-stressed" locations). One must be
careful not to attribute specific impacts or changes to climate-stressed
locations. Biomes (or, more correctly, species constituting ecosystems
of the biome) may be able to tolerate the new climatic conditions (i.e.,
new conditions fall within the potential niche) and thus may be relatively
little changed. This same proviso should be applied to migration-stressed
locations (i.e., existing vegetation may continue to occupy the site;
thus, the migratory restriction does not come into play for decades
The authors move on to equate climate-stressed locations with habitat loss and conclude that 36% of the land area will be affected. For aforementioned reasons, this must be regarded as an upper bound. The authors then attempt to estimate the reduction in habitat patch size by counting pixels affected by climate-stress in contiguous blocks of the same biome type, then applying a simple species area relationship to estimate species loss (see McDonald and Brown, 1992). Little reliance should be placed on these estimates given the foregoing provisos and the caveats listed by the authors themselves.
Other reports in this collection