Cover Image
close this bookAmazonia: Resiliency and Dynamism of the Land and Its People (UNU, 1995, 253 pages)
close this folder2. Environmental threats
View the document(introduction...)
View the documentThe myth of virginity
View the documentClimatic change
View the documentThe environmental impact of smoke
View the documentSoil erosion and floods
View the documentHydroelectric dams
View the documentThe environmental impacts of mining
View the documentPetroleum extraction
View the documentA blizzard of cocaine
View the documentHabitat destruction and the loss of biodiversity


The myth of virginity
Climatic change
The environmental impact of smoke
Soil erosion and floods
Hydroelectric dams
The environmental impacts of mining
Petroleum extraction
A blizzard of cocaine
Habitat destruction and the loss of biodiversity

Amazonia has a long history of ecological change under human agency. Hunters and gatherers probably penetrated the region tens of thousands of years ago and artificially enriched their campsites with fruit and nut trees (Smith in press). Hunters undoubtedly fired woody savannas in Amazonia, such as in Roraima, to flush game long before farmers started to clear the forest.

Slash-and-burn farming in Amazonia probably began at least 10,000 years ago, based mostly on root crops, thereby creating a rich texture of forest interlaced with second-growth communities of various heights and ages. At first, such interventions were on a minor scale, but, as the population grew denser, more and more of the forest fell to the axe.

The myth of virginity

Contrary to the prevailing idea of a pristine Amazonian forest little disturbed by human activities until recently (Moran 1993a; Revkin 1990: 39; Richards 1977), much of the region has felt the influence of hunters and gatherers and farmers for a considerable time. Many of the forests of tropical America, including the Amazon, are anthropogenic (Denevan 1992; Turner and Butzer 1992). Assertions that the Amazon is being ravaged by development and floods of land-hungry colonists need to be put within the context of a region that has a long history of settlement and human modifications of the environment.

Archaeological research and unusual concentrations of certain economic plants in the forest suggest much higher human population densities in the past than has hitherto been accepted. Even with relatively inefficient stone axes, aborigines have cleared substantial tracts of Amazonian forests in the past (Huber 1910). Charcoal layers in soils of the upper Rio Negro have been dated to 6,000 years B.P. and some ceramic shards mixed with anthrosols are approximately 3,750 years old (Sponsel 1986). Along the Bragantina coast of Pará, charcoal and potsherds have been dated at 5,000 years B.P. (Simões 1981). The oldest recorded pottery in the New World is from the middle Amazon: ceramics from the Taperinha site near Santarém have been dated at 8,000 years B.P. (Roosevelt et al. 1991). Maize reached the Ecuadorian Amazon at least 6,000 years ago, providing farmers with another food option. Upland and flood-plain forests have thus been altered by farming activities for millennia (Bush, Piperno, and Colinvaux 1989).

Population densities reached high levels, particularly along siltladen rivers, well before contact with Europeans (Moran 1990: 148, 1993b: 114). Estimates of human populations in Amazonia around 1500 range from 1 million to 6 million, or even higher (Smith 1980). Only recently has the region's population regained its former numbers, but with a distinct difference: a sizeable proportion of today's population lives in towns and cities. In pre-contact times, the population was much more rural than at present, and therefore more engaged in farming.

The indigenous population did not raise cattle, however, so cleared areas were devoted to crops and managed fallows. The higher population density in rural areas was thus possible with perhaps the same amount of cleared area as today. The overall cleared area in 1500 was probably close to that prevailing in 1990.

If pre-contact aboriginal populations of Amazonia reached in excess of 6 million people, forest fires in the region were probably as common as at present, but on a smaller scale. Indians would not have cleared fields the size of some ranches and plantations being opened up in the region, particularly in southern Pará, northern Mato Grosso, Rondônia, and Acre, but they likely lit many more smaller fires to prepare fields.

Instead of massive fires concentrated in active colonization zones, particularly in the south-eastern and south-western fringes of Amazonia (fig. 2.1), hundreds of thousands of smaller fires would have been scattered over the basin in pre-contact times. Agricultural activities in Amazonia before the arrival of Europeans were akin to a buckshot event (Oldeman 1989); localized, small-scale clearings spread out over a large area. A Dante's inferno characterizes colonization zones today in the dry season, whereas the twinkling of innumerable small fires would have dotted the nocturnal landscape after the main rains in prehistoric Amazonia.

The shift to clearing larger sections of forest is ecologically more damaging than the scattered, insular fires of the past. When thousands of hectares are cleared for a single ranch or plantation, the ecological fabric of the area becomes simpler (Uhf, Buschbacher, and Serrão 1988). Instead of a patchwork quilt of various plant communities, more homogeneous landscapes emerge. Also, seed sources for forest regeneration became scarcer and soil nutrient recycling systems can be disrupted (Buschbacher, Uhl, and Serrão 1987; Nepstad, Uhl, and Serrão 1990,1991; Serrão et al. 1979).

The implications of a dense, pre-contact human population in Amazonia are far-reaching. A salient lesson here is that Amazonia's diverse environments can support relatively large populations, even on nutrient-poor oxisols and ultisols, if resources are managed wisely. Although some have argued that nutrient deficiencies prevent continuous crop production on highly weathered ferrallitic soils of many parts of the lowland, humid tropics (Weischet and Caviedes 1993: 278), pre-contact indigenous populations probably deployed a wide variety of swidden systems that permitted relatively dense populations, even in inland areas.

People in Amazonia have greatly altered plant and animal communities and the distribution and population densities of certain plants and animals, and have probably triggered increased soil erosion and aggrading of smaller rivers and streams for millennia. The notion of a vast, undisturbed wilderness in Amazonia is an artefact of the indigenous population crash after contact with Europeans and the unleashing of introduced diseases such as smallpox and influenza.

Climatic change

The impact of deforestation on regional and global climate has received the most attention when environmental change is discussed in Amazonia (Bunyard 1987; Collins 1990; Dickinson 1987, 1989; Leopoldo, Franken, and Matsui 1985; Myers 1988; Prance 1986; Reis 1972; Wood 1990). Tropical deforestation is often pinpointed as a major culprit in the purported global warming trend and, since Amazonia is the largest stretch of tropical forest, its fate is thought to have an important bearing on the future of the world's climate. Amazonia was probably also heavily cleared in pre-contact times, without triggering the greenhouse effect.

Global warming

Increased atmospheric levels of carbon dioxide and other gases, such as methane, nitrous oxide, and chlorofluorocarbons (CFCs), can potentially trigger a greenhouse effect (Ravel and Ramanathan 1989). Many in the scientific community, and much of the reporting in the media, suggest that we are on a global warming course propelled by human activities such as burning forests and fossil fuels. But claims that a global warming has already begun may be premature (Byrne 1988; Flavin 1989; Schneider 1989). No firm evidence has yet emerged that the world is becoming significantly warmer (Abelson 1990a; Blinder 1992; Hansen and Lacis 1990; Ray 1993: 12; Solow and Broadus 1989; Spencer and Christy 1990). Indeed, surface temperatures over the western Arctic Ocean have become significantly cooler during the 1950-1990 period (Kahl et al. 1993). Such inconsistent reporting on "global warming" points to major defects in current models of global climate and suggests a tenuous basis for drawing policy conclusions (Bryson 1989).

Although some data suggest a recent global warming trend, no valid correlation with greenhouse gases can be made, nor can we be sure how long this trend will last (Barrett 1990). Even if such changes will soon be confirmed, it will be difficult to separate natural climatic cycles from any greenhouse effect (Mitchell, Senior, and Ingram 1989). For example, the subsurface thickness of ice around the North Pole varies markedly from year to year, and no significant trend emerged during the 1977-1990 period (Langereis, Van Hoof, and Rochette 1992). Also, the effects of clouds, volcanic dust, and oceans on any possible greenhouse effect are imperfectly understood (Abelson 1990b; Jarvis 1989; Kerr 1989; Slingo 1989).

If evapotranspiration rates are substantially reduced in Amazonia as a result of landscape changes, less latent heat may be exported from the region in water vapour (Motion 1987). Whether reductions of the moisture level in warm air circulating from tropical regions to temperate areas in the Hadley cell would affect global climate is unclear. Reduced moisture levels could result in less latent heat being released during condensation, thereby cooling climates. Such a mechanism would help to counteract any greenhouse effect. Deforestation also tends to increase the albedo affect, further reducing energy for atmospheric heating.

In the event that the greenhouse effect takes hold, tropical deforestation will be only partly at fault (Radulovich 1990). Deforestation accounts for less than 20 per cent of greenhouse gas emissions (Flavin 1989: 13). The amount of carbon in initial, undisturbed ecosystems may have been overestimated, thus exaggerating the impact of deforestation on the release of carbon dioxide to the atmosphere (Post et al. 1990).

Carbon dioxide from the burning of fossil fuels, which occurs mostly in temperate countries, is the largest component of greenhouse gases. Just seven industrialized countries produce 40 per cent of carbon dioxide emissions worldwide (Turner et al. 1990a). The industrial countries are responsible for approximately 85 per cent of carbon dioxide build-up in the atmosphere (Parikh 1992).

The release of CFCs, used to make aerosols, refrigerants, and solvents, is responsible for a larger proportion of greenhouse gases entering the atmosphere than CO2 emissions from burning forests. Industrial countries are responsible for most of the CFC emissions. The notion that developing countries must take a large share of the blame for any global warming has been ascribed to environmental colonialism (Agarwal and Narain 1991).

Deforestation in Amazonia may not be implicated in excessive emissions of nitrous oxide, a potent greenhouse gas, as previously thought. Although higher levels of nitrous oxide are released in pasture soils in the first 10 years after forest clearing, emissions of the gas subsequently decline to lower levels than those coming from tropical forest (Keller et al. 1993). This does not mean that cattle-raising is necessarily an appropriate land use for much of Amazonia from the policy standpoint; rather the ecological impacts of different land uses must be weighed according to a "basket" of criteria, and more longterm research is often needed to decipher the environmental effects of habitat change.

The idea that Amazonian countries should arrest forest clearing to save the world's climate while North Americans and Europeans continue to drive their increasingly more powerful cars and burn natural gas and coal does not rest well in Brasilia, Bogota, or Lima (Nisbet 1988). Developed countries will need to do more to reduce their own carbon dioxide emissions if they expect third world countries to tackle the issue (Caccia 1991). Some scientists have urged policy makers to separate "survival emissions," such as resource-poor farmers practising slash-and-burn agriculture, from "luxury emissions," particularly gas-guzzling cars plying the streets of major cities, particularly in the industrial countries (Agarwal and Narain 1991). Some industrial countries, such as Germany, the Netherlands, and Japan, have adopted carbon dioxide stabilization or reduction targets (Miller 1991), but many others are apparently waiting for a global consensus to emerge on appropriate action.

A cautious approach to formulating environmental and economic policy to address global warming has been adopted by several governments. Three main factors account for this wait-and-see attitude (Riebsame 1990). First, climate change predictions are too uncertain, particularly at the regional level. Second, current systems are thought to be capable of absorbing climate change without major disruption, at least for the next few decades. Third, technologies can be deployed to mitigate or compensate for some of the changes wrought by global climate change.

Sceptics about global warming can also point to the fact that C crops, such as rice and wheat, would likely produce higher yields with a doubling of atmospheric carbon dioxide levels if sufficient water and nutrients were available. The C crops, such as sugar cane and maize, are unlikely to be affected by increased carbon dioxide levels, at least for the foreseeable future. Also, some regions, such as the drier tropics, might benefit from a global warming since they could receive more rain.

The ability of countries to respond to global warming will depend in part on the strength of their agricultural research and extension systems to deliver new technologies. Areas with a shift to wetter climates will probably need crop varieties more resistant to fungal and bacterial diseases. Unfortunately, as the need for research institutions to be primed and ready to confront new challenges increases, their scientific capacity is at a low ebb.

In the 1970s and early 1980s, Brazil had one of the strongest agricultural research programmes among the developing countries, with an annual budget of close to US$200 million. Since the mid-1980s, however, high inflation and severe financial constraints have hampered EMBRAPA's (Empresa Brasileira de Pesquisa Agropecuária) ability to raise and sustain agricultural productivity (Ruttan 1991). The agricultural research systems of other countries with territories in Amazonia have also weakened over the past decade or so.

Another constraint on the deployment of technologies to meet the challenge of a possibly warmer world is the loss of biodiversity. As discussed in more detail later, the loss of habitats, particularly in the humid tropics, could have grave consequences for agriculture. The loss of genetic resources for crop improvement, and the disappearance of new crop candidates, could "tie the hands" of plant breeders trying to develop crops adapted to changing environments.

In spite of uncertainties about global warming trends and hazards, pressure is mounting for governments in both industrial and developing countries to take concrete steps to halt the build-up of greenhouse gases in the atmosphere. Tree planting is seen as one way to counteract the greenhouse effect, by providing a carbon sink (Myers and Goreau 1991). Although this popular notion empowers people to do something about a widely perceived problem, the impact of tree planting on the build-up of carbon dioxide in the atmosphere pales compared with what could be accomplished by the more efficient use of fossil fuels. At least 100 million ha would have to be planted to fast-growing trees to sequester a little over 10 per cent of the current annual build-up of carbon in the atmosphere (Myers and Goreau 1991). Once the trees reached maturity, they would no longer act as carbon sinks. That is an area equivalent to Britain, France, and Germany that would then cease to serve as a trap for carbon. The conversion of old-growth forests to fastgrowing plantations would release carbon to the atmosphere, in spite of the greater net photosynthesis of younger trees (Harmon, Ferrell, and Franklin 1990).

Although rehabilitating degraded areas with trees would be desirable, simply filling the landscape with "greenhouse" trees without regard to cultural and economic needs could be counter-productive. Cultural landscapes could not easily accommodate hundreds of millions of hectares of tree planting without disrupting food production and other economic activities, even if the "greenhouse gas" trees formed integral parts of agro-forestry systems. Cleared lands are often fully occupied, and the logistical and managerial implications of massive tree planting would need careful study (Churchill and Saunders 1991).

Reduction of methane emissions, an often overlooked contributor to the greenhouse effect, could help mitigate any global warming. Methane is a far more potent greenhouse gas than carbon dioxide and landfills in the developed world are a major source of methane emissions (Hogan, Hoffman, and Thompson 1991). Recovery of methane from landfills not only would reduce the greenhouse effect, but could supply gas to generate electricity. Improved coal-mining and oil-production techniques would also reduce methane emissions. Ruminant livestock and rice cultivation are also significant sources of methane. Deforestation and burning also increase methane concentrations in the atmosphere, but, again, a large share of the onus for reducing this greenhouse gas rests with the industrial countries.

The spectre of parched deserts

In addition to temperature changes, deforestation has the potential of adversely affecting rainfall regimes. Half of the rain that falls in Amazonia is thought to come from evapotranspiration (Molion 1975: 101; Salati 1987; Salati and Vose 1984; Salati, Marques, and Molion 1978). Accordingly, it has been assumed that continued deforestation might lead to a drier regional climate (Hecht and Cockburn 1989: 43). A linkage between the loss of forests and reduced rainfall has been surmised for centuries, and was much discussed in India and parts of the Caribbean in the nineteenth century (Glacken 1967; Grove 1990). Cattle-ranching and the destruction of Latin America's tropical forests have been blamed for reduced rainfall and increased droughts (Salati1992; Shane 1986: 23). Some computer models predict a sharp drop in rainfall with continued, large-scale deforestation in Amazonia, thus raising the spectre of dust-bowls and desertification (Anderson 1972; Barros 1990: 20; FAO 1991b: 3; Goodland and Irwin 1975; Modenar 1972; Paula 1972; Roddick 1991: 206; Sioli 1987). Cattleranching is sometimes singled out as most likely to provoke desertification (Wesche 1974). But such rainfall models assume that Amazonia will be turned into a barren landscape (EMBRAPA 1989: 6).

It is highly unlikely that substantial areas of Amazonia will be converted to asphalt or a desert. Second growth soon begins the regeneration path to forest in all but the sandiest soils (Moran 1993a). The widespread struggle to keep pastures and crops free of weeds is a testament to the striking speed of secondary succession. Weeds, not deserts, are a major headache for farmers, ranchers, and plantation owners in Amazonia. How soon mature forest returns depends primarily on the texture and fertility of the soil and the proximity of seed sources. Also, more rainfall may be derived from the flux of water vapour from the Atlantic than has previously been supposed (Paegle 1987).

How much forest can be removed without affecting rainfall is not known. Realistic predictions of climatic change as a result of landscape changes in the humid tropics are fraught with difficulties (Henderson-Sellers 1987). Current models of forest and climate interactions in Amazonia are too imprecise to predict with any degree of certainty the impacts of deforestation on rainfall (Salati 1992). Evapotranspiration from groves of perennial crops and silvicultural plantations may be close to that of forest. Even pastures release substantial quantities of water to the atmosphere during the rainy season. During the dry season, though, pastures transpire less water than forest and experience significantly warmer mean surface temperatures (Nepstad, Uhl, and Serrao 1991; Nobre, Sellers, and Shakla 1991).

No evidence is available to prove that deforestation in Amazonia has led to reduced rainfall. The dry season around Manaus was accentuated in 1976, and again in 1979, when no rain fell for 73 days (Fearnside 1986: 50). The "summer" (verão) of 1992 was also severe, with many farmers experiencing reduced yields or the loss of seedlings of perennial crops. But it is hard to separate little-understood climatic cycles from long-term trends. To keep matters in perspective: after two particularly heavy burning seasons in Amazonia in 1987 and 1988, 1989 was a very wet year. So much rain fell in eastern Amazonia in 1989 that the dry season virtually disappeared.

Rainfall patterns are highly variable in Amazonia. In 1774, a drought assailed the Rio Negro watershed when deforestation rates were much lower than at present (Hemming 1987). River levels in Amazonia were unusually low in 1860 owing to poor rainfall (Chandless 1866). In 1958, 64 days passed without any rain in the Bragantina zone east of Belém (Penteado 1968: 138). The Amazon River was particularly low again in 1963. The big "push" to develop and open up the Amazon started only in the late 1960s.

Unusual weather patterns also prevailed in the southern United States in the 1980s. Record heat and drought seared the southern and eastern parts of the United States during 1987 and 1988, thereby provoking widespread concern about global warming. Summer 1989 and spring 1992 in the eastern United States, however, were wetter and cooler than usual, and talk of global warming in the media subsided. In Western Europe, several hurricane-force gales during the winters of 1988 and 1989 led to speculation among some politicians that global warming was under way (Maddox 1990). Some in the environmental movement may be concerned that a return to more "normal" weather will cool the ardour of politicians to tackle the issue of global warming.

Predictions about how dry the Amazon might become with continued deforestation are fraught with shaky assumptions. With satellite imagery, it should be possible to document the area of forest cleared each year. Through its remote sensing agency (INPE - Instituto Nacional de Pesquisas Espaciais), the Brazilian government monitors forest burning in Amazonia. But what happens to the land after it is cleared is crucial to the question of climatic drying. Landsat and Spot imagery can separate second growth, cropland, grassland, and forest, but it cannot readily differentiate between secondgrowth stages or types of crops. Aerial photography from aircraft would be too expensive on the scale needed to document annual vegetation changes in the region as a whole. More sophisticated satellites in the future may be able to help. Also, more information is needed about evapotranspiration rates in various vegetation communities, including croplands.

The environmental impact of smoke

Smoke from forest cut for agriculture and ranching may have global in addition to regional and local impacts. One component of biomass burning, methyl chloride, attacks ozone, and deforestation in Amazonia has been implicated as partly responsible for holes in the ozone layer over Antarctica (Cutrim 1990). Human-induced fires account for 5 per cent of the ozone-destroying chemicals in the atmosphere (C. Anderson 1990). Nevertheless, industrial countries release far more ozonedepleting aerosols into the atmosphere than do farmers and ranchers in the tropics.

Excessive smoke could lead to temporary climatic disruptions. Smoke reflects some incoming radiation back into the atmosphere, thereby helping to mitigate any greenhouse warming. A reduction in the solar energy reaching the earth's surface may reduce convectional activity and rainfall in some areas. The burning season in most of Amazonia extends from July to October; little smoke is generated during the rainy months.

Smoke from burning fires is thick enough temporarily to close some regional airports during the dry season, particularly in Rondônia and Acre. Temperature inversions can exacerbate the haze. In the dry season, cloudless skies facilitate radiative cooling of the land, thereby helping to create a layer of warmer air above that traps pollutants. Prior to 1970, the Brazilian Air Force had to close airports in parts of the Amazon because of the excessive haziness due to smoke. Although the shutting down of airports because of smoke is not new in Amazonia, the problem has become more acute for airfields in parts of Rondônia and Acre in the past two decades.

Smoke from cleared fields has been implicated in poor harvests of Brazil nut (Bertholletia excelsa) in Pará (Miller 1990). Smoke is believed to interfere with bee pollinators of Brazil nut trees, but such a linkage has not been demonstrated conclusively. Variable yields of Brazil nuts are more likely related to the severity of the dry season when flowers are formed. If the dry season is not pronounced, flowering is reduced and fruit production will be poor the following year. The dry season in 1989 was exceptionally wet in eastern Amazonia, and the 1991 harvest of Brazil nuts was correspondingly poor. At Tomé-Açu, for example, 220 mm of rain fell in October 1989, normally a dry month.

Brazil nut harvests in the municipality of Marabá, the major centre for the Brazil nut trade, have varied from a low of 2,000 tons in 1985 to a high of 17,732 tons in 1970. Marabá was connected to the BelémBrasilia Highway along PA 332 (then PA 70) in 1970, so the peak of production in that year might be attributed in part to improved transportation. Traditionally, the Brazil nut harvest had been taken out by river and along a now defunct railroad from Jatobal to Tucurui on the Tocantins.

Brazil nut harvests have often been highly variable, ever since collecting in earnest began in the Tocantins watershed in the latter half of the nineteenth century. In 1895, for example, some 450 tons of Brazil nuts were collected along the Tocantins, whereas in 1896 the harvested amounted to only 250 tons. In 1987, however, 1,000 tons of nuts were gathered, and the harvest would have been closer to 2,500 tons if the Tocantins had not experienced a massive flood (Moura 1989: 153). Such oscillations in Brazil nut harvests could not be attributed to smoke since few people lived along the river at that time.

If smoke from fires was the principal cause of variation in Brazil nut harvests, one would expect a sharp decline during the 1970s and 1980s. Although harvests appear to have dropped during much of the 1980s, a shift of Brazil nut gatherers to more lucrative occupations, particularly gold mining, probably accounts for lower harvests (Smith et al. 1991a). Fluctuations in Brazil nut prices on world markets have also probably influenced collecting efforts.

Soil erosion and floods

Soil erosion is one of the most serious threats to the sustainability of agriculture, silviculture, and forestry in Amazonia. Although many of the region's soils are deep, extending down several thousand metres in some cases, fertility is usually concentrated in the first few centimetres of topsoil. The need to protect the soil is a major reason that perennial crops, silviculture, and properly managed pastures are among the more viable options for rural development. Soil erosion is a contributing factor in the decision of many farmers to abandon their fields and clear a fresh plot from the forest.

Soil erosion can lead to larger-scale environmental problems by aggrading river beds. Increased run-off from cleared land deposits silt and sand on river beds, thereby provoking more severe floods. A rising and lowering of water levels along streams and rivers in Amazonia is part of the normal seasonal pulse of wet and dry seasons (Sternberg 1975). Some unusually heavy floods along the Amazon in the mid-1970s raised the spectre that deforestation in the foothills of the Andes was having a tangible impact downstream (Fowler and Mooney 1990: 106; Gentry and Lopez-Parodi 1980; Smith 1981a: 122). But statistical analyses of flood peak levels do not reveal any trend to more intense flooding along the Amazon (Richey, Nobre, and Deser 1989; Sternberg 1987b). The greatest flood ever recorded along the Amazon was in 1953, well before major development projects were unleashed in Amazonia. The Amazon River reached unusually high levels again in 1993, although deforestation rates have abated since the 1980s.

Similar findings have been made along the Ganges, where flooding in Bangladesh is not a result of deforestation in the Himalayas, as is commonly thought (Ives and Messerli 1989). An important variable in runoff from deforested lands is the infiltration rate of the soil (Newson and Calder 1989). Some compaction of the soil is likely after the forest is cut, but how adversely infiltration is affected over the long term depends on land management and soil type.

Destruction of forests along streams and some river banks is surely affecting water quality and flow on a local scale. But the vast scale of Amazonia's forests appears to be masking the impact of deforestation on smaller watersheds. Landscape changes are not currently radical enough to affect Amazon hydrology on a large scale (Bayley 1989). Also, the variability of rainfall in different parts of the Amazon basin could lead to premature conclusions that floods are more pronounced along the Amazon.

Hydroelectric dams

Disruption of fisheries as a result of dam-building could be a serious threat to the livelihoods of many rural and urban folk in Amazonia (fig. 2.2). A number of fish important in commerce and subsistence, such as jaraqui (Semaprochilodus spp.) and catfish (Brachyplatystoma flavicans, B. filamentosum), migrate from the Amazon to spawn in tributaries (Barthem, Brito, and Petrere 1991; Goulding 1981, 1989). Changes in water quality and flood cycles are likely to interfere with the reproduction and feeding of many of the 2,000 or more species of fish inhabiting the myriad waters of Amazonia.

Little is known about the impact of dams built thus far on Amazonia's fisheries (Bayley and Petrere 1990). On 2 October 1992, a school of jaraqui (Semaprochilodus sp.) was observed swimming back and forth along the foot of the dam at Curuá-Una in a fruitless at tempt to move upstream. The dam closed in 1976, so it is curious to see fish still trying to move upstream. Several other species have apparently disappeared above the dam, such as pirapitinga (Colossoma bidens) and jatuarana (Brycon sp.), but one interviewed farmer felt that fishing yields overall had not declined. Tucunaré (Cichla ocellaris) and charuto, in particular, are reportedly plentiful in the reservoir. The Curuá-Una reservoir spawned a population explosion of piranhas (Serrasalmus spp.) during the first two decades of operation (Ferreira 19B4; Junk et al. 1981). For all their fame as dangerous fish, piranhas are eaten by locals and sell briskly in urban markets.

Fig. 2.2 Major hydroelectric darns in Amazonia

Some fish were killed by the lack of oxygen and hydrogen sulphide when the Tucurui dam closed in 1984 (Sioli 1986), but the reservoir has become a significant fishery for the highly prized tucunare. Tucunaré from the Tucurui reservoir are marketed at least as far south as Carajás. The 2,100 km Balbina reservoir on the Uatumã River, completed in 1987 to provide electricity for Manaus, has also become a significant fishery for this spirited predator.

Known as peacock bass to English-speaking sport fishermen, tucunaré may accumulate mercury released by gold miners. Also, tucunaré in the Curuá-Una resrvoir near Santarém, Pará, became so heavily infested with parasitic nematodes that some locals declined to eat the highly prized fish (Junk and Nunes de Mello 1987).

Some fisheries downstream from the Tucurui appear to have suffered from the dam (Dwyer 1990: 44; Magee 1989). The productivity of fisheries appears to have declined mostly in the lower regions of the Tocantins in the vicinity of Cametá. One migratory species, Anodus elongatus, has virtually disappeared from the lower Tocantins (Merona, Carvalho, and Bittencourt 1987). Populations of Curimata cyprinoides have also diminished, at least temporarily. The Tucurui dam contributed to the collapse of the mapará fishery on the lower Tocantins by closing off a spawning route and reducing plankton biomass (Goulding, Smith, and Mahar in press).

Although the composition of fish communities has shifted below the Tucurui dam, the overall impact of this formidable barrier on fisheries has not proved especially serious from the perspective of local nutrition. A shrimp fishery based on Macrobrachium amazonicum was waning along the lower Tocantins well before the Tucurui dam closed; furthermore, this freshwater shrimp, which is used in a variety of regional dishes, thrives in the Tucurui reservoir (Odinetz-Collart 1987). The rapid turnover of water in the Tucurui reservoir, some six or seven times a year, helps avoid drastic changes in water chemistry and thermal stratification, thus reducing the danger of intoxicating fish (Barrow 1987).

Turtles, especially Podocnemis expansa, were abundant along the Tocantins in the seventeenth century and served as an important food for locals (Heriarte 1964: 30), but this resource had dwindled considerably long before construction of the Tucurui dam. The resulting reservoir, however, has covered many former nesting beaches and will probably preclude the chances of re-establishing sizeable populations of P. expansa along the Tocantins.

One concern about reservoirs in the Amazon is their potential role in favouring the population build-up of disease vectors. Soon after the Tucurui dam closed, some inhabitants and their livestock near the margin of the reservoir were plagued by swarms of mosquitoes (Mansonia titillans) and, to a lesser extent, horseflies (Lapiselaga grassipes). The former are known to carry two arboviruses, but no outbreaks of disease attributed to M. titillans have occurred near the Tucurui reservoir (Marques 1992). Populations of both flies appear to have dwindled, presumably as the new lacustrine ecosystem and surrounding areas have become more stable.

The Samuel dam on the Jamari River in Rondonia, which filled in 1989 to supply electricity to Porto Velho, has reportedly disrupted the upstream migration of some large catfish (João Paulo Viana, pers. comm.). Fisheries have also allegedly suffered downstream from the Balbina dam, but no quantitative data support such claims (Gribel 1990).

Some farmers have apparently benefited downstream from the Tucurui dam, whereas others have lost fertile planting ground. Regulation of water flow facilitates the irrigation of rice at the mouth of the Tocantins, but the reduced sediment load has resulted in a loss of flood plain for agriculture along the lower Tocantins (Barrow 1988).

No reservoirs created for hydroelectric dams in Amazonia are in imminent danger of losing electrical generating capacity because of siltation. The oldest dam, Curuá-Una, is nearly 20 years old and is still operational (fig. 2.3). All the other hydroelectric dams were built in the late 1970s and in the 1980s (table 2.1). Although the Tocantins appears to be getting cloudier as a result of forest clearing, "storage pockets" abound in the reservoir bed and this dead space will take some time to fill with sediment.

Table 2.1 Major hydroelectric dams operating in Amazonia

Dam(a) River Operational Capacity (MOO) Reservoir
area (km)
Tucurui Tocantins 1984 2,000(b) 2,400
Balbina Uatumã 1989 250 2,300
Samuel Jamari 1989 216 560
Curuá-Una Curuá-Una 1977 30 78
Paredão Araguari 1975 40 23
TOTALS   2,536 5,361

Sources: Barrow (1988); Ledec and Goodland (1988); Serra (1992); Sioli (1986); field notes of NJHS at the Curuá-Una dam on 2 October 1992.

a. See fig. 2.1 for dam locations.

b. The Tucurui dam is expected eventually to generate 4,000 MW.

The Tucurui reservoir has drowned tens of thousands of Brazil nut trees (fig. 2.4). The Tocantins valley has always been the most important centre for the Brazil nut trade, and the 2,000 km reservoir has destroyed some valuable plant resources. It is difficult to measure the impact of such losses on yields, but some unique germplasm has surely been lost. As the Brazil nut's genepool shrinks, genes that could be useful for future improvement efforts also vanish.

Amazonia is relatively flat, so dam-building along major rivers leads to the drowning of substantial tracts of flood plain and upland forest. Iquitos on the Upper Amazon is 3,600 km from the Atlantic, yet only 80 metres above sealevel (Irion 1989). From Iquitos to the confluence with the Negro, the Amazon drops only 57 metros. Some dams are more "cost effective" in terms of kilowatts generated per flooded area (Ledec and Goodland 1988: 60). The Tucurui dam generates 30 kW/ha of reservoir, whereas Balbina delivers only 2 kW/ha of flooded area (table 2.1). Both dams are low in electricity generated per flooded area when compared with rivers with steeper valleys, such as the Itaipu dam on the Paraná (77 kW/ha) or the Grande Coulee on the Colorado (63 kW/ha).

The apparent and hidden ecological costs of building major dams in Amazonia must be weighed against the benefits hydroelectric dams bring to the region. Brazil's desire to tap the hydroelectric potential of waters in Amazonia is understandable in view of the burdensome bill for imported petroleum. Brazil produces less than one-third of its petroleum needs, and most of the electricity generated in Amazonia has historically come from diesel-powered turbines. The Tucurui dam has benefited Belem and environs with reliable electricity and has created jobs, such as at the aluminium smelting plant at Bacarena.

Nevertheless, a series of smaller, more environmentally benign hydroelectric projects might prove more suitable over the long term. To supply power to the pulp mill at Jari, for example, a proposed hydroelectric plant at the Santo Antonio Falls will divert part of the river through a turbine and thus will not involve any flooding.


The environmental impacts of mining

Mining has become a major economic activity in Amazonia. The environmental impacts of mining operations by corporations are largely localized and of minimal significance, particularly since some earlier water pollution concerns have been addressed. At most, 4,500 km of forest is likely to be cleared to gain access to all known exploitable mineral deposits in Amazonia (Hoppe 1992). Settlement and development activities associated with the poles of growth generated by mining concerns are likely to have more widespread impacts.

Corporate operations

Bauxite mining requires the removal of large quantities of overlying soil, and if precautions are not taken sediment can be washed into nearby watercourses. At one point, seven miles of Lake Batata near Minerac,ao Rio Norte on the Trombetas had filled in with reddishbrown soil, thereby killing trees and destroying fish and wildlife habitats (Mee 1988: 279). Corrective measures have been taken by building a siltation pond, and Lake Batata is being restored.

Mineração Rio Norte, which operates the bauxite mine along the Trombetas, eventually replants areas scraped to gain access to the aluminium ore. Topsoil is stockpiled and then spread back once an area has been mined. Several native trees are planted to speed up restoration of the land (Gradwohl and Greenberg 1988:173). Such recuperation efforts are costly, but Mineração Rio Norte is demonstrating leadership in environmental management, and technologies developed by the company are likely to prove useful at many other mining sites in the humid tropics.

The manganese field at Serra do Navio in Amapa has been mined by ICOMI since the early 1960s, and is nearing the end of its economic life. The ore is taken by rail to Porto Santana, a deep-water port on the northern bank of the Amazon. Sizeable oil-palm plantations and smallscale settlement have sprung up along the railroad. The manganese mines still operating in the vicinity of Serra do Navio are well managed and do not provoke any significant ecological damage. Oilpalm plantations along the railroad are well adapted to the climate and soils of the region and provide good ground cover.

At Carajás, forest-clearing around mines is minimized, and road and rail sidings are planted to Brachiaria humidicola, a perennial grass from Africa. Holding ponds to decant mining sediment have also been established by Companhia Vale do Rio Doce (CVRD) at Carajas. Outside the 400,000 ha concession granted to CVRD at Carajás, forest-clearing by farmers and ranchers is rampant, particularly along the 890 km railroad to Itaqui in Maranhão. The lesson learned here is that sound environmental management practices within a concession are no guarantee that natural resources will not be destroyed around the periphery. A broader, more integrated approach to regional development is thus called for that explores the linkages and interactions between land-use systems.

Itinerant gold miners and mercury pollution

In contrast to most corporate mining operations, small-scale gold-mining activities in Amazonia are causing widespread ecological damage. According to official figures, itinerant gold miners unearthed US$13 billion worth of gold in the Brazilian Amazon between 1980 and 1988 (Almeida 1992). The gold rush in Amazonia during the 1980s rivalled the California gold rush of the nineteenth century and is exceeded only by production from the mines of South Africa (Godfrey 1992). Clandestine gold trading probably increased that figure several fold. Little wonder that gold mining has attracted so much attention in the Amazon, even though few miners ever become wealthy.

Itinerant gold mining is causing one of the most serious environmental problems in Amazonia today: mercury pollution. Itinerant miners use the toxic element to precipitate gold when washing gravel. Between 5 and 30 per cent of the mercury is lost during this process, and much of the mercury finds its way back to water (Maim et al. 1990). A further 20 per cent vaporizes when the amalgamate is torched to obtain the gold, both in the field and in goldbuying stores. The high rainfall and humidity in the region facilitate reoxidation of vaporized mercury. Mercury pollution of the air is particularly acute in the districts of certain towns where gold stores tend to concentrate (Biller 1994: 8).

Approximately half a million gold miners (garimpeiros) were operating in Amazonia during the early to mid-1980s (Mallas and Benedicto 1986). In the 1990s, the gold fever has cooled somewhat, especially in Rondônia and southern Pará, but several hundred thousand men and women are probably still engaged in gold mining in the Brazilian Amazon. By the late 1980s, about 100 metric tons of gold were being exported annually from the Brazilian Amazon. For every kilogram of gold produced, an estimated 1.32 kg of mercury is lost to the environment - 45 per cent into rivers and streams and 55 per cent into the atmosphere (Pfeiffer and Lacerda 1988). Some 100 tons of mercury were thus finding their way into the region's ecosystems at the close of the 1980s (Nriagu et al. 1992).

Mercury has been employed in gold and silver mining in Latin America since early colonial times, but the current scale of mercury use for mining is unmatched. A Mexican miner, Bartolome de Medina, devised a process for extracting silver from its encasing ore with mercury in the mid-sixteenth century (McAlister 1984: 228). Mercury was used extensively as an amalgam in silver mining in Mexico and Peru during the sixteenth and seventeenth centuries. The Spanish employed quicksilver to refine a little gold in Hispaniola in the sixteenth century.

Brazil depends on imports for its mercury needs, and most of it is consumed by the informal gold-mining sector (Biller 1994: 6). Itinerant miners account for more than 70 per cent of Brazil's gold production, so the use of mercury is highly diffused in Amazonia and would be hard to control. Technologies exist for reducing mercury contamination, such as the use of retorts when torching amalgam, but thus far they have not been widely adopted.

During the colonial period in Colombia, placer miners separated gold from sediment rich in iron oxide by employing the glutinous sap of several plants, including crushed leaves of cordoncillo (Piper sp.), encinillo (Weinmannia sp.), and chica (Jacquinia aurantiaca). When mixed with water, the foamy sap captured the iron oxide flakes, allowing the gold particles to settle to the bottom of the pan (West 1952). Organic precipitates would be much more environmentally benign than mercury, although Jacquinia aurantiaca is a piscicide. Such ancient practices are worth investigating, particularly to compare gold recovery rates with mercury.

Gold mining in rain forests is not confined to Amazonia, although it is most prevalent there. Large numbers of itinerant miners are operating with mercury in other regions, such as southern Guyana, Venezuela, and parts of Central America (Parsons 1955). Gold miners had to be evicted from a park on the Osa Peninsula of Costa Rica where they were clogging rivers with sediment. The Costa Rican government relocated the miners at considerable expense. In southern Guyana, dredgers are sucking up river beds and discharging large quantities of sediment back into the water, thereby altering its turbidity.

Although gold has been panned from alluvial deposits in the Tapajos valley since the 1950s, the Amazon gold rush started in earnest in 1980 when gold prices soared to US$850 per ounce. International investors had apparently grown anxious about the Iran-Iraq war and were buying substantial quantities of gold on the world market.

The most celebrated gold find in Amazon was at Serra Pelada, approximately 100 km south-west of Marabá in January 1980; within a few months, 20,000 garimpeiros had swarmed to the mountain once isolated in a sea of forest (Santos 1981: 161; Schmink 1985). In June 1980, Serra Pelada was producing about 6 kg of gold per day, and by 1986 the mountain had been reduced to a gaping pit, 110 metres deep (Mallas and Benedicto 1986). By 1990, most of the gold miners had moved on, but much of the forest around Serra Pelada had been cleared by settlers.

Every state in the Brazilian Amazon is currently experiencing a rapid influx of fortune-seekers. Regional airports, such as at Monte Dourado in Para and Boa Vista in Roraima, are hives of activity, with small planes constantly ferrying miners and their supplies to and from remote airstrips. During the height of the gold rush in the 1980s, other regional airports, such as Santarém, Itaituba, and Maraba in Para, Alta Floresta in Mato Grosso, and Porto Velho in Rondônia, also witnessed swarms of single- and twin-engined aircraft serving far-flung gold-mining camps.

The gravity of the mercury pollution problem has attracted the attention of the national press in Brazil. Few data are available, though, to assess the dimensions of mercury accumulation in fish. Mercury concentration in rivers and streams varies markedly by location, and mercury contamination in fish also varies according to species (Biller 1994: 8). Some Kayapó Indians have acquired dangerously high levels of mercury, presumably from eating fish and drinking water from polluted rivers and streams (Hecht and Cockburn 1989: 143). Of 106 individuals sampled in four communities in the Tapajós watershed, over 60 per cent had mercury levels in their urine high enough to warrant regular testing as recommended by the World Health Organization (Thornton et al. 1991). Around half of the 97 river sediment samples taken from the Tapajós and its effluents exceeded the 1 ppm mercury considered the limit for safety by the Brazilian environmental secretariat (Thornton et al. 1991).

Predatory fish, such as tucunare, pirarucu (Arapaima gigas), aruanã (Osteoglossum bicirrhosum), most species of piranha, and many species of catfish, are likely to accumulate mercury more rapidly than herbivorous species. In the Madeira river system, several species of predatory fish, such as dourado (Brachyplalystoma flavicans), filhote (B. filamentosum), and other catfish (Pseudoplatystoma spp.), have accumulated high levels of mercury, up to 3.81 fig mercury/g (Maim et al. 1990; Pfeiffer et al. 1989). Levels of mercury in excess of 0.5 Gig are generally considered a potential threat to human health. These large catfish are frozen and sent to markets in the United States and central and southern Brazil (Goulding 1981). Omnivorous fish, such as highly prized tambaqui (Colossoma macropomum), and eggs of detritus feeders, such as Scarf catfish (Loricariidae), have also been found with high levels of mercury along the Madeira River (Martinelli et al. 19B8). Preliminary analyses of fish and human hair along the margin of the Tucurui reservoir also reveal that mercury is entering the food chain along the Tocantins River (Braunschweiler 1991).

If mercury levels rise in fish, an important source of protein in the region, then the poor are especially likely to suffer. If the region's fish become unsafe to eat, wealthier inhabitants can more easily turn to beef, chicken, or imported fish. The long residence time of mercury in river sediments can contribute to health hazards long after the goldmining frontier has moved on (Fuge et al. 1992), so damage from gold mining in Amazonia may be felt for decades to come. A trend towards gold mining from barges in rivers, such as the Madeira and the Negro, is particularly worrisome in this regard. In 1985, some 1,400 boats were pumping up sediment from the bottom of the Madeira in search of gold (Pfeiffer and Lacerda 1988).

The spectacular surge in gold mining is having other adverse environmental effects. Gold-mining camps are among the worst areas for malaria transmission in Amazonia. Gold miners may be introducing new strains of malaria (Plasmodium falciparum and P. vivax) to which local populations have no resistance. Thus Indians are likely to suffer more severe malaria symptoms if gold miners are operating nearby. In Roraima, for example, hundreds of Yanomami Indians have succumbed to malaria, owing in part to infections brought in by the nearly 20,000 gold miners who started arriving in 1987 (Robinson 1991). In a two-week period in 1992, 44 Yanomamo died from malaria in the Parafuri village alone.

Gold miners are likely accelerating natural selection among malaria parasites for resistance to drugs used for prophylaxis and treatment. In some areas, miners are apparently employing drugs that are usually reserved for treating chloroquine-resistant strains of falciparum malaria. In the early 1970s, Fansidar was the drug of last resort for treating infections of falciparum malaria that would not respond to chloroquine treatment. Now Fansidar is largely ineffective in most areas of Amazonia, in part because miners have used the drug as a prophylaxis, thus increasing selection pressures for resistance.

The treatment and prophylactic strategies of gold miners warrant further study. A first impression was that gold miners often suffered so much from malaria because they were not taking adequate precautions, ranging from the use of mosquito nets to standard prophylaxis with chloroquine. If, on the other hand, significant numbers of miners are using last resort treatments for chloroquine-resistant falciparum malaria, such as mefloquine and doxycycline, as a prophylaxis then malaria is likely to become an even more widespread public health problem in the Amazon basin.

Although mercury contamination and invasion of indigenous lands remain serious issues in Amazonia, these threats have abated to some degree in the 1990s. Gold prices plunged to about US$340 an ounce in early 1993, and many garimpeiros have returned from mining camps to seek their fortunes elsewhere. The end of the Cold War has probably calmed investors' fears and the lustre of gold has diminished somewhat. The closing of many gold shops in Santarem and Porto Velho, as well as diminished small plane activity at such airports as Santarém and Itaituba, are symptoms of the decline in goldmining activity. The extent to which gold mining has declined overall is hard to tell, and in some parts it is actually on the increase. If gold prices climb steeply once again, gold miners are likely to return in force to the backlands of Amazonia.

Petroleum extraction

Western Amazonia has witnessed some sizeable oil strikes within the past 30 years, and prospecting is under way in several parts of the basin. Most of Ecuador's oil production, some 300,000 barrels/day, comes from the Amazon region. Amazonia may yet become a major oil-producing area.

With current technologies and safeguards, oil drilling does not usually pose undue hazards to the environment. Some water pollution occurs around soil extraction sites, but overall such damage is usually localized. Major pipeline leaks cause much greater damage. Two significant leaks have occurred along the Trans-Ecuadorian Petroleum Pipeline: an earthquake ruptured a 30 km section of the pipeline in March 1987, while a massive landslide caused by heavy rains and deforestation sheared the pipeline in May 1989 (Hicks et al. 1990: 9). Civil unrest has occasionally led to sabotage of oil pipelines in Colombia and Peru; should more pipelines be built in Amazonia, such dangers could increase. On the whole, though, the oil industry is not causing any large-scale pollution of the air or waters in Amazonia. An even greater risk would ensue if large numbers of oil tankers began plying the Amazon. The dangers of running aground or colliding with other ships would increase and oil spills would be disastrous for fish production and many agricultural activities.

Roads created to facilitate petroleum extraction help settlers penetrate the rain forest. Roads built to oil fields operated by TexacoGulf in the Ecuadorian Amazon, for example, have opened up some 2 million ha of formerly undeveloped forest lands (Hiraoka and Yamamoto 1980). By the late 1970s, over 30,000 people had taken advantage of the oil company roads to eke out homesteads in the forest. Currently proven reserves of sweet crude in the Ecuadorian Amazon are expected to be depleted by the late 1990s, but substantial reserves of heavier grades of petroleum have been found in eastern Ecuador, which may be tapped in the twenty-first century.

A blizzard of cocaine

All the world's cocaine is produced from coca plantations on the eastern slopes of the Andes. The seemingly endless appetite for recreational drugs in industrial countries, and increasingly in developing nations, has spurred coca plantings in areas formerly in forest or planted to food crops. A crop once grown for local consumption has become so profitable that it is now much more widely grown than in the past. Traditionally, coca has been used in two forms: a mild stimulant by chewing the leaves or making tea; and as a snuff in rituals by certain indigenous groups of western Amazonia. Now large quantities of coca leaves are processed to concentrate cocaine. Coca cultivation is consequently spreading further into the Amazon lowlands, including Brazil.

Three latent environmental problems may surface because of the cocaine business: deforestation, coca eradication, and the dumping of chemicals used to process coca leaves into paste. Given the secrecy and peril involved in investigating the coca industry, information on such negative environmental impacts will be difficult to obtain. Presumably, they will have mainly local or regional impacts.

Reliable data on planting rates and the cultivated area in coca are understandably hard to gather. More than 700,000 ha of montane forest have apparently been cut in Peru to grow and process coca (Goodman 1993). More tropical forest in Amazonia has surely fallen at the hands of small farmers to grow coca than has been cleared by miners, both corporate and itinerant. The Andean portion of the Amazon basin is especially rich in biodiversity; it would be ironic if some individuals who partake of cocaine also buy albums and "rainforest" products that purportedly support "sustainable" use of tropical forests. Discussion about using defoliants, such as Spike, to eradicate coca plantations has elicited some concern, particularly from environmentalists and ecologists concerned about the impact of herbicides on non-target plants. Biocontrol efforts may prove more effective, provided they do not attack other plant species. On the other hand, coca bushes may do a better job of securing soil on steep Andean slopes than most food crops. While the international market for cocaine remains strong, farmers will find a way to continue planting a crop that produces a handsome profit.

Poppy fields are sprouting now in the Andes, in response to a resurgent demand for opium and a desire by drug cartels to diversify their product lines. Poppy fields cover at least 20,000 ha in the Colombian Andes (Goodman 1993), and cultivation may spread south into the Amazon basin. This introduced annual does not secure soils well on steep slopes. In discussions about the need for people in developed countries to reconcile their lifestyles with the planet and biodiversity, little attention has been paid to the hedonistic use of drugs. Efforts should be redoubled to educate people about the folly of taking recreational drugs in industrial countries, rather than try and blame coca growers in western Amazonia.

One of the most serious environmental issues associated with coca in Amazonia is the nature and quantity of chemicals used to process coca leaves. In 1986, Peru exported an estimated 6,400 tons of coca paste. Such big quantities would have involved the use of 32 million litres of sulphuric acid, 16,000 tons of quicklime, and 6.5 million tons of acetone.6 In the Upper Huallaga valley alone, coca processors annually dump an estimated 56 million litres of kerosene, 8 million tons of sulphuric acid, and large quantities of acetone, toluene, and carbide (FAO 1990: 12). Coca is also processed in Colombia, Ecuador, and Bolivia. Most of these chemicals and compounds probably found their way into streams and the groundwater in various parts of western Amazonia. If these chemicals are not eliminating fish populations, they may render them unsafe to eat. Such compounds may also trigger mutations in at least some of the fish species important for subsistence and commerce.

The threat of coca-processing chemicals may be diminishing, at least in some areas. In Colombia, for example, coca refiners have recently begun recycling chemicals, rather than ditching them in streams. Such measures do not reflect a concern for the environment; rather they are an effort to circumvent restrictions on the importation or production of precursor chemicals used in processing coca paste.

Habitat destruction and the loss of biodiversity

Of all the issues surrounding environmental change in Amazonia, threats to biodiversity are arguably the most serious in the long run. Species loss as a result of drastic habitat modification, such as logging, is an issue in many parts of the humid tropics, such as in Malaysia (Brookfield, Potter, and Byron 1995). Particularly worrisome is that such biodiversity losses are often not accompanied by any long-term economic benefits to the local people.

The effects of species loss may not be immediately obvious, and are thus not usually considered in economic development plans. Species loss can be hard to quantify in economic terms, and is considered an externality. The impoverishment of habitats is sure to reduce options for future development. Air and water pollution may be more tangible assaults on our living space, but the haemorrhaging of species will drain resources for future generations, quite apart from the ethical and moral questions posed by human-induced extinctions.

Some have argued that we do not really need nature's storehouse of genes because ingenious scientists can concoct novel genes in laboratories at will (Huber 1992). This notion is fallacious. The idea that we can safely dispense with tropical forests or other ecosystems because modern biotechnology has made them redundant is dangerous thinking. Genes need to be synthesized from models. Laboratories and computer memories cannot replicate dynamic evolutionary processes under way around the world.

Impressive advances in biotechnology only underscore the importance of conserving biodiversity since desirable genes can increasingly be switched from one organism to another. The glamour of genetic engineering should not blind the public and policy makers to the need to safeguard the integrity of natural ecosystems.

Loss of biodiversity as a result of development and environmental degradation has emerged as a global concern (Raven 1990; Wilson and Peter 1988). Concern over erosion of biodiversity was initially confined mostly to the scientific community, but is now spreading to the general public and politicians. Tropical deforestation is often at the forefront in debates on biodiversity loss; at least 27,000 species are thought to be lost from such widespread destruction every year (Myers 1993). Actual species loss may be much higher considering that many plant and animal groups are still imperfectly known (Wilson 1992). Amazonia has attracted particular attention in this regard because of its high degree of endemism and vast numbers of animal and plant species, many of them undocumented or poorly studied (Adds 1990). The region's diverse array of plant and animal communities contains many unique species and genes.

Amazonia contains the largest stretch of tropical forest, spanning 5,000 km from the Andes to the Atlantic, and some 4,000 km from the Guianas and the Upper Orinoco to the scrub cerrado of the Brazilian shield and the seasonally flooded grasslands of the Pantanal. This vast mosaic of forest communities, second growth, natural and man-induced grasslands and swamps contains the richest assortment of plant and animal species in the world, as well as a rich storehouse of genes for crop improvement.

Deforestation can adversely affect biodiversity on two accounts: outright habitat destruction, and ecological changes along the contact zone of remaining forest stands. Whereas deforestation rates in Amazonia appear to have been less dramatic over the past decade than has previously been thought, perhaps in the order of 15,000 km per year during the 1978-1988 period, some 38,000 km of forest may have been degraded (Skole and Tucker 1993). The edge effect of disturbance, ranging from microclimatic changes to the encroachment of swidden and pasture fires, can allegedly alter plant and animal communities as much as 1 km into the forest. Although some may dispute how deeply human influences penetrate the forest in terms of significant ecological disturbance, it is clear that outright deforestation alone is an insufficient measure of biodiversity change. In drier areas of the Amazon, such as around Maraba in southern Pará and in southern fringes of the rain forest in Mato Grosso, the border of isolated patches of forest can easily catch fire when fields or grazing areas are torched in preparation for planting or to destroy weeds and promote new growth of pasture grasses.

Endangered gene pools

Rampant deforestation fuelled by development schemes and pioneer farmers now threatens to destroy the genetic resources of many economic plants and potential crops before they can be tapped for the benefit of people throughout the world (NRC 1991; Smith et al. 1991b, 1992). Also, loss of tribal cultures is resulting in the disappearance of unique varieties of many annual and perennial crops.

Wild populations of crops, and in some cases their near relatives, are increasingly sought by plant breeders for desirable traits, such as pest and disease resistance and tolerance to problem soils. To help make farming more sustainable, researchers, development organizations, and farmers are increasingly seeking genetic solutions to agricultural constraints rather than costly, and sometimes environmentally damaging, chemical applications.

Ironically, forest-clearing to establish farms and plantations can eliminate the very genes that could be used to improve the crops being planted. The shrinking of wild populations of over 47 perennial crop species is currently under way in Amazonia (Smith and Schultes 1990). Perennial crops are important to the livelihoods of most small farmers in the tropics as well as operators of large plantations. Perennial crops that originated in Amazonia provide food, beverages, shelter, medicines, oils, resins, cosmetics, food colourants, and latex for citizens on every continent.

The Amazon contains wild gene pools of such commercially important crops as rubber (Hevea brasiliensis) and cacao (Theobroma cacao), as well as regionally important food and beverage crops such as peach palm (Bactris gasipaes) and guarana (Paullinia cupana), a popular soda in Brazil and now exported to developed countries such as Canada. As the heart-of-palm trade destroys many wild stands, particularly species of Euterpe, entrepreneurs in several Latin American countries are planting peach palm for palmito. Plantations of peach palm for the heart-of-palm trade are especially well developed in Costa Rica and Mexico. The ability of developing countries in the Amazon region as well as in other parts of Latin America, Africa, and South-East Asia to raise and sustain yields of several important cash and food crops will hinge to a large extent on their ability to marshal genetic resources to overcome constraints to production.

The Amazon forest also contains wild populations of hitherto minor crops such as annatto (Bixaorellana) and cupuaçu (Theabroma grandiflorum). Annatto is used by some Indians for body paint and is commonly used as a food colourant in Latin America. Known as achiote in Spanish and urucú in Brazil, annatto occurs spontaneously in various parts of Amazonia and probably originated in Acre. Traditionally, annatto has been grown as a backyard plant but it is now being cultivated on a commercial scale in several Latin American countries, particularly Brazil. After artificial Red Dye No. 3 was banned in the United States because it is a carcinogen, interest in annatto rebounded. Natural red bixin from annatto is increasingly used to colour foodstuffs and cosmetics in both industrial nations and developing countries. Annatto is used to enhance the colour of some peach-flavoured yogurts in the eastern United States, although few consumers probably make any connection with Amazonia when they relish that healthy treat.

Football-sized cupuaçu, a relative of cacao, grows in the forests of eastern Amazonia and is cultivated in backyards and fields for its refreshing pulp, which is used to make drinks, ice-cream, cakes, and puddings . Cupuaçu sells briskly in Amazonia and is penetrating markets in southern Brazil, and more recently in the United States and Japan. Cupuaçu has made the transition from an extractive product, to an occasional plant in home gardens, to a full-fledged crop, often grown in agro-forestry systems. How many other "cupuaçus" linger in the forest that could one day delight the palate of people in the region and abroad?

Amazonia's lush and diverse forests also contain many plants that could be incorporated into our menu of cultivated species, or are currently on the threshold of domestication. The nuts of patauá (Jessenia bataua) palm, for example, contain an oil similar to the quality of olive oil. The fruits are collected in the wild to make refreshing drinks. Brazil nut is now being grown on a small scale on several private landholdings in the Brazilian Amazon as well as in Malaysia. Domestication of Brazil nut is an important step considering that wild stands of the giant forest tree are falling for ranches, farms, and reservoirs.

Forests also contain pollinators and dispersal agents of wild populations of many crop plants as well as their near relatives . Intricate and often fine-tuned relationships between plants and animals need to be maintained if the integrity of many wild populations of our crop plants is to be ensured. For example, Brazil nut is dispersed by agoutis (Dasyprocta spp.), which bury many of the nuts after gnawing open the hard capsules that encase the protein- and oil-rich nuts (Huber 1910). Some bats pollinate forest trees while others disperse their seeds, and many birds, such as toucans, guans, and curassows (Cracidae), also disseminate many forest fruits. Conservation of forest environments, as well as field gene banks, is thus essential for the long-term viability of many crops important for subsistence and commerce.

Conservation of genetic resources and regional development in Amazonia are intertwined. More studies are needed, however, to document genetic variation in wild populations and domesticated gene pools of Amazonian crops. Also, hard economic data are needed to convince policy makers of the value of preserving forest habitats as resources for development.

Another dimension to the loss of botanical resources in Amazonia is that an invaluable medicine chest is literally being depleted before we can assess even a fraction of the potentially useful drugs. The rapid decline of many indigenous societies represents a loss not only of crop plants and unique varieties, but of ethno-botanical knowledge about which trees, shrubs, and herbs in the forest might offer cures for a wide variety of health complaints (Plotkin 1993). Many drugs are eventually synthesized in laboratories but, as in the case of genes for crop improvement, they have to be discovered first.

The indigenous knowledge base

A salient lesson from global efforts to conserve crop genetic resources is the need to maintain the cultural integrity of indigenous groups. People with a long history of interaction with the forest have much to teach us about sustainable agricultural practices and plant resources. Rural folk are particularly knowledgeable about the location and natural history of wild populations of crops and their near relatives. Tribal peoples are also knowledgeable about the medicinal value of plants in the forest and in their home gardens, a priceless heritage that is increasingly threatened by encroaching civilization (Schultes 1988; Schultes and Raffauf 1990, 1992). Biodiversity and cultural heterogeneity are vital to sustainable development.

Although rural peoples often have affinity with the surrounding plant and animal world, their practices do not always result in conservation of resources. Peasants and ranchers alike frequently clear fields and home gardens up to the margins of streams or rivers. If water courses were left in forest, natural corridors would remain for animal and plant dispersal in cleared areas.

Farmers in the Brazilian Amazon report three main reasons why they fell riparian forests, even if they contain economically important species such as açai palm (Euterpe oleracea). First, several crops important for domestic consumption and commerce grow well in the moist, organic-rich soils flanking streams and small rivers in contrast to the generally highly acidic, nutrient-poor upland soils. The common bean (Phaseolus vulgaris), maize (Zea mays), and vegetables reportedly yield better when grown in humid, valley bottoms. It is also easier to irrigate vegetables when they are close to a perennial water source. Second, farmers want to eliminate habitat for predators on their small livestock, particularly ocelot (Felis pardalis), margay (Felis wiedii), and various hawks. Third, newcomers to the region do not yet appreciate the value of some riparian trees and wish to create open, more "productive" landscapes.

Faunal resources

The forests and rivers of Amazonia also contain an abundant array of wildlife important for hunting (Ayres and Ayres 1979; Ayres et al. 1991; Bodmer, Fang, and Ibanez 1988, 1991; Dourojeanni 1974; Rios, Dourojeanni, and Trovar 1973; Smith 1976). Some game animals, such as various species of turtle (Podocnemis spp.), could be domesticated or reared in captivity and the young released to the wild. Of the many species of mammals, birds, and reptiles of lowland South America, only one species has been domesticated, the muscovy duck (Cairina moschata). Muscovy ducks still grace the extensive lakes and lagoons of the Amazon flood plain, and some of these wild populations could be helpful to further breeding efforts.

Several wild animals are kept around homes for eventual consumption and some of them are in the process of being domesticated. Most are captured while still young, particularly when their mothers are killed during hunts. Capybara (Hydrochaeris hydrochaeris) and black-bellied tree ducks (Dendrocygna autumnalis) are kept around homes by some farmers along the Amazon flood plain, and they sometimes breed in captivity. Options for animal domestication will be foreclosed if wild populations of such species are drastically reduced.

Habitat destruction is the principal threat to wildlife in the region. The issue of dams and fisheries has already been highlighted, but the clearing of floodplain forests by farmers and ranchers also eliminates breeding and feeding grounds for fish. At least three-quarters of the fish species important in commerce and subsistence derive their nutrition directly or indirectly from flooded forests (Goulding 1980, 1993). A better understanding of land-use systems is thus essential for improved development and conservation efforts.

Policy implications

Although Amazonia is rich in plant and animal species, particularly in forests cloaking the eastern slopes of the Andes, policies that might deflect settlement and development to relatively empty areas, such as cerrados and thorn scrub, need to be considered carefully. Some of the drier areas of South America have higher levels of mammal endemism than the rain forests (Mares 1992). The campos, cerrados, and thorn scrub woodlands (caatinga, Chaco) of South America are much richer in animal and plant species than is generally appreciated (Pimm and Gittleman 1992). Wholesale destruction of low rainfall environments would also greatly reduce biodiversity - and future options for sustainable development in those areas.

The expansive flood plain of the sinuous Amazon River is another perceived venue for increased development efforts. The Amazon flood plain has always been regarded as an underutilized environment in Amazonia with enormous potential for raising food crops, livestock, and fish. Yet the idea that the flood plains can help absorb development pressures from the "fragile" upland forests warrants careful scrutiny. Flood-plain forests along the Amazon have already been largely logged out and extensively cleared. Endemism is especially high in the seasonally flooded forests of the Amazon River. Efforts to boost the productivity of flood-plain areas should focus on already cleared areas, rather than promote the wholesale destruction of the remaining forest and aquatic habitats with the* rich assortment of wildlife.

To help preserve the remaining biodiversity in Amazonia and to reduce pressures to develop contiguous regions, the productivity of agriculture, managed forests, plantations, and ranches must be raised within the well-watered basin. Whenever feasible, such land-use systems should be ecologically diverse. Agro-forestry systems, highlighted later as one of the more viable options for agricultural development in the region, could help maintain some level of biodiversity. Compared with simple monocultures, agro-forestry permits the survival of more animals and plants (Holloway 1991). Also, monocultures of perennial crops, such as oil-palm, create microenvironments for such plants as lichens and mosses that cannot survive in fields of rice or maize.