Studies developed after the Special Report on Regional Impacts of Climate Change (RICC) add new findings on the potential impact of climate change on different sectors in Latin America. Many of these new studies evolved from those mentioned in RICC, indicating that a process is in place in Latin America to reduce uncertainties associated with climate change.
Climate variations produce a variety of impacts on natural ecosystems (see Sections 5.4 to 5.9). Latin America possesses a large quantity of ecosystems, ranging from Amazonian tropical rainforest to cold Andean systems (Paramos). It also hosts remarkable rangelands, shrublands, deserts, savannas, grasslands, coastal wetlands (mainly along the Caribbean and Atlantic coastlines), and inland freshwater wetlands such as Pantanal and Iberá. Natural ecosystems in Latin America can be expected to suffer a variety of impacts from climate change. Latin American humid tropical forests represent an important group of ecosystems for which a great deal of information is offered (e.g., Canziani et al., 1998).
Conversion of large areas of tropical forest to pasture could reduce water cycling and precipitation in the region, in addition to its global role as a contribution to global warming. Pasture has much less leaf area than forest (McWilliam et al., 1993). Because evapotranspiration is proportional to leaf area, water recycled through forest is much greater than that recycled through pastureespecially in the dry season, when pasture is dry but forest remains evergreen (Roberts et al., 1996). This is aggravated by the much higher runoff under pasture, with measured increases of more than 1,000% in small (10m2) plots (Fearnside, 1989). Pasture grasses can partially compensate for reduced evapotranspiration by increasing their efficiency of water use when soil moisture is low, whereas forest trees maintain constant efficiency (McWilliam et al., 1993). Soil under pasture quickly becomes highly compacted, inhibiting infiltration of rainwater into the soil (Schubart et al., 1976). Rain falling on compacted soil runs off quickly, becoming unavailable for later release to the atmosphere through transpiration. The shallower root system of pasture, compared to that of forest, prevents pasture from transpiring during periods of drought (Nepstad et al., 1994, 1999). Precipitation decreases therefore are greatest at the time of year when rain is most needed.
If the extent of deforestation were to expand to substantially larger areas, we have high confidence that reduced evapotranspiration will lead to less rainfall during dry periods in Amazonia and medium confidence that rainfall will be reduced in the center-west, center-south, and south regions of Brazil (Lean et al., 1996). Although the annual rainfall total in Amazonia would decrease by only 7% from conversion to pasture, based on simulations with the Hadley Centre model, in August (dry season) the average rainfall would decrease from 2.2 mm day-1 with forest to 1.5 mm day-1 with pasturea 32% decrease (Lean et al., 1996). Simulations of conversions of Amazonian forest to pasture, using the Météo-France EMERAUDE GCM, indicate reduced volumetric soil moisture in the "arc of deforestation" where clearing activity is concentrated along the southern boundary of the Amazon forest. Rainfall reduction in southern Brazil is greatest for the January-March period (Manzi and Planton, 1996).
Greater dependence of Amazonian rainfall on water derived from evapotranspiration in the dry season means that conversion to pasture would cause this period to become longer and more severea change that could have harsh repercussions on the forest even if total annual precipitation were to remain unchanged (Fearnside, 1995). In patches of forest isolated by cattle pasture, trees on the edges of forest patches die at a much greater rate than do those in continuous forest (Laurance et al., 1997, 1998; Laurance, 1998). Because many trees die while they are still standing, rather than being toppled by wind, dry conditions (particularly in the air) near reserve edges are a likely explanation for mortality. Soil water may partially counterbalance the effect of drier air; as trees die, soil water in the gaps they leave normally increases because the roots that would have removed water from the soil are gone. Increasing vines and decreasing forest biomass are strongly associated near forest edges, probably as a consequence of a positive feedback relationship between these factors (Laurance et al., 2000). Drier microclimatic conditions have been found at forest edges (Kapos, 1989). Increased water stressas indicated by altered 13C in plant leavesextends 60 m into the forest from an edge (Kapos et al., 1993). Tree mortality increases significantly up to 100 m from the forest edge (Laurance et al., 1998). Considering the length of forest edges measured by Skole and Tucker (1993) for the Brazilian Amazon in 1988, a 100-m disturbance buffer to these edges would represent a disturbed area in 1988 of 3.4 x 106 ha, or 15% of the area cleared by that year. Forest edges, which affect an increasingly large portion of the forest with the advance of deforestation, would be especially susceptible to the effects of reduced rainfall.
Greater severity of droughts reinforced by deforestation effects could lead to erosion of the remainder of the forest once a substantial portion of the region had been converted to pasture. The greatest effects are likely to occur during occasional complex phenomena such as El Niño (Tian et al., 1998; Nepstad et al., 1999). Precipitation in Amazonia is characterized by tremendous variability from 1 year to the next, even in the absence of massive deforestation (e.g., Fearnside, 1984; Walker et al., 1995). If the forest's contribution to dry-season rainfall were to decrease, the result would be to increase the probability of droughts that are more severe than those experienced in the centuries or millennia over which the present forest became established. Occasional severe droughts would kill many trees of susceptible species. The result would be replacement of tropical moist forest with more drought-tolerant forms of scrubby, open vegetation resembling the cerrado (scrub savanna) of central Brazil (Shukla et al., 1990)
Until recently, burning in Amazonia has been almost entirely restricted to areas where trees have been felled and allowed to dry before being set alight. Fire normally stops burning when it reaches the edge of a clearing rather than continuing into unfelled forest. Archaeological evidence suggests that catastrophic fires have occurred in Amazonia during major El Niño events four times over the past 2,000 years: 1,500, 1,000, 700, and 400 BP (Meggers, 1994). Human action could now turn less intensive El Niño events, which are much more frequent than major ones, into catastrophes. Increased fire initiation foci, together with increased forest flammability from logging, already have resulted in substantial incursions of fires into standing forest in eastern and southern Amazonia during dry years (Uhl and Buschbacher, 1985; Uhl and Kauffman, 1990; Cochrane and Schulze, 1999; Cochrane et al., 1999; Nepstad et al., 1999). The 1998-1999 fires in Roraima, in the far northern portion of Brazil, reflect the vulnerability of standing forests in Amazonia during El Niño events now that settlement areas in the forest provide permanent opportunities for fire initiation (Barbosa and Fearnside, 1999).
Increases in the amount of biomass burning could affect nutrient cycling in Amazonian forest ecosystems (medium confidence) (Fearnside, 1995). Droughts lead to increases in the area and completeness of burning in clearings in Amazonia, contributing to smoke and dust that function as sources of wind-borne nutrients to the surrounding forest (Talbot et al., 1990). Climatic change also could increase nutrient supply via long-range transport of dust. African dust transported across the Atlantic Ocean by winds may be supplying significant amounts of phosphorus and calcium to Amazonia (Swap et al., 1992). Amazonian soils are very poor in these elements. Soil nutrients are among the factors that limit growth, recruitment, and mortality of trees (Laurance et al., 1998; Sollins, 1998). Smoke and ash particles from burning in savannas, possibly including those in Africa, also contribute nutrients (Talbot et al., 1990). The extent to which these nutrient sources could increase the growth of Amazonian forests is not known. Increases undoubtedly differ by tree species, thereby altering forest composition. Burning is affected by climate, as well as by the size and behavior of the human population. Factors that influence the growth of intact forests in Amazonia are particularly important because of the large amounts of carbon that could be released to or removed from the atmosphere if the balance between forest growth and decay is altered.
CO2 enrichment is believed to contribute to observed imbalances between CO2 uptake and release by forest biomass in Amazonia; forest recovery from past disturbances also may contribute to these imbalances. Eddy correlation measurements (studies of gas movements in air flows inside and immediately above the forest) at one site in Rondônia indicated an uptake of 1.0 ± 0.2 t C ha-1 yr-1 (Grace et al., 1995). A similar eddy correlation study near Manaus found uptake of 5.9 t C ha-1 yr-1 (Malhi et al., 1999). An estimate based on reviewing existing measurements of tree growth and mortality in permanent plots found mean uptake of 0.62 ± 0.37 t C ha-1 yr-1 (Phillips et al., 1998). On the other hand, at the Biological Dynamics of Forest Fragments site near Manaus (the largest and longest running study included in the forest growth measurement data set), no uptake or loss was found in 36 1-ha control plots located >100 m from a forest edge (Laurance et al., 1997). The Rondônia and Manaus eddy correlation studies and the basin-wide review of tree growth indicate uptakes of 0.63, 3.66, and 0.38 x 109 t C, respectively, when extrapolated on the basis of consistent definitions of forest that indicate a total area of forest in the Amazon Basin of 620.5 x 106 ha (Fearnside, 2000). A process-based model of undisturbed ecosystems in the Amazon Basin, including savannas and forests (not necessarily defined as above), indicates wide interannual variations in net carbon flux from vegetation and soil, ranging from emissions of 0.2 x 109 t C in El Niño years to a sink of as much as 0.7 x 109 t C in other years; mean annual flux simulated over the 1980-1994 period gives an uptake of 0.2 x 109 t C (Tian et al., 1998). If the frequency of El Niño events increases as a consequence of global warming (Timmermann et al., 1999), these forests may release some of their large carbon stocks to the atmosphere. The future course of accumulation of CO2 in the atmosphereand consequently the time when concentrations would reach "dangerous" levelsdepends heavily on continued uptake of carbon by the biosphere, including an important contribution from Amazonian forests. Climate effects contribute to making the sink in Amazonian forests unreliable as a brake on atmospheric carbon accumulation.
Although temperature changes from global warming are expected to be modest in the tropics as compared to temperate regions, it is important to realize that each degree of temperature alteration in a tropical environment may be "perceived" by forest species there as a greater change than would be the case for the same temperature shift in a temperate forest (Janzen, 1967). The direct effects of global warming on ecosystems at relatively low latitudes, if not at the equator itself, therefore may be greater than the small predicted temperature alterations at these sites might lead one to believe. In addition, direct effects of global warming through temperature change are likely to be less pronounced than effects that temperature can have through its influence on other climatic parameters, such as rainfall (Fearnside, 1995).
GCMs indicate a range of results for the effect of global warming on precipitation in Amazonia. Drying generally is expected; some models indicating greater drying than others. The Hadley Centre's HadCM2 model indicates especially dry climate over Amazonia. Process-based ecosystem models that use this simulated climate show large declines in net primary productivity (NPP) and release of carbon as a result of Amazonian forest dieback (Friend et al., 1997). The varied GCM results suggest the need for a range of climate scenarios as inputs to ecosystem simulations (Bolin et al., 2000). It should be noted that available scenarios (e.g., Nakicenovic et al., 2000) represent the change in climate resulting from altered composition of the atmosphere only, not the additional impacts of regional land-use changes such as replacement of Amazonian forest by pasture.
Globally, models show that a doubling of GHGs may lead to a 10-15% expansion of the area that is suitable for tropical forests as equilibrium vegetation types (Solomon et al., 1993). For tropical rainforest, the suitable area would expand 7-40%, depending on the GCM employed in estimating the future distribution of climatic zones. The GCM studies used by Solomon et al. (1993) assess the effects of doubled GHGs (i.e., CO2-equivalence), through the direct effects of temperature and through temperature-driven alteration of precipitation regimes (but not rainfall changes provoked by deforestation). These results are indicators of potential for forest expansion and are not intended to reflect expected landscapes in the future; they do not include the influence of human populations in converting to other uses land that is climatically suitable for tropical forests.
One model that includes climate-induced and human changes to the year 2050 points to decreases in forest areas by about 5% in Latin America (Zuidema et al., 1994). The deforestation estimates used in these calculations are based on the areas needed to satisfy expected demands for agricultural products. In the case of Brazil, deforestation is likely to exceed these forecasts because much of the forest clearing stems from motivations other than consumption of agricultural products (Hecht et al., 1988; Reis and Margulis, 1991; Hecht, 1993; Fearnside, 1997). In any case, the combination of forces driving deforestation makes it unlikely that tropical forests will be permitted to expand to occupy the increased areas that are made climatically suitable for them by global warming. Land-use change interacts with climate through positive feedback processes that accelerate the loss of Brazil's Amazonian forests.
Other reports in this collection