Note: This website is no longer being updated and is being maintained for archive purposes by the Department of Botany, National Museum of Natural History, Smithsonian Institution. Please see About the Project for further details.

Link to North America map of regional study sites
North America map

Link to Middle America map
Middle America map

Link to South America
South America map

Link to Centres of Plant Diversity home page



Link to South America Regional Overview



III. Amazonia


The geological Amazon Basin is a large horseshoeshaped lowland 3000 km long and 300-800 km wide (Leopoldo et al. 1987). It is bordered to the west and south-west by the Andes, and to the north and south respectively by the Precambrian crystalline shields of the Guianas (elevations to 1000-3000 m) and the Brazilian tableland (elevations mostly 600-1300 m) (Haffer 1987; Takeuchi 1961). over the last 3000 km the river falls only 60 m (Leopoldo et al. 1987). The overall region's geology, topography, dynamics and climate are now known to be neither uniform nor simple, but the myth of a vast and homogenous stable forest is difficult to eradicate (Daly and Prance 1989; Prance 1987; Sioli 1984).

The basin is divided into three approximate regions based on differences in the rivers and the geological history of each region (Junk and Furch 1985; Putzer 1984; Sioli 1984):

1. Upper Amazonia (which drained westward until uplift of the Andes in the later Tertiary) extends as a broad funnel along the Andean foothills, narrowing eastward almost as far as Manaus and the confluence of the Negro and Amazonas/Solimões rivers. Much of its soil is sediment from the Andes (Anderson 1981; Daly and Prance 1989).

2. Much of Lower Amazonia is known as Central Amazonia (and the middle basin). Bound between the ancient crystalline shields, it is narrower (the funnel's spout) and extends from about Leticia in Colombia eastward beyond Manaus to about the mouth of the Xingú River. Central Amazonia is bordered to the north-west and north by white-sand areas, and to the east and south-east by a dry zone (Transverse Dry Belt) following the Trombetas River Basin and the Madeira/Tapajós interfluve, but has an undefined southern and south-western boundary.

3. The small easternmost region of Amazonia, from a bit east of the mouth of the Xingú River to the Atlantic Ocean, includes the river divided into several arms, with delta, estuaries and many islands including Marajó Island (50,000 km², the size of Switzerland). Hydrologically, Amazonia does not include the south-eastern basin (1 million km²) formed by the Araguaia-Tocantins rivers, which is often included in Amazonian estimates, but essentially drains independently into the Atlantic just south-east of the Amazon River (near Belém) (Leopoldo et al. 1987).

There has been considerable discussion as to the boundaries of Amazonia (UNDP 1992). While much of the Andean Cordillera drains into the Amazon River and is sometimes included in a broad concept of Amazonia, the upland Andes are phytogeographically distinct from lowland Amazonia. In contrast, the Guianas are not hydrologically a part of the Amazon Basin, since their rivers drain directly into the Atlantic Ocean, but the lowland Guianas phytogeographically can be considered part of the Amazon.

The broad limits of Amazonian vegetation with its tropical forests (the hylaea) extend well beyond the geological basin. The southern as well as north-western limits are imprecise - gradual, mosaic or abrupt transitions as species are replaced because of changes in climate and soils. To the north-east, the Guianas are included up to 900 m, where the Guayana Highlands are considered to begin (Daly and Prance 1989).

As defined here, Amazonia covers c. 7.6 million km², including lowland Guayana. This northern Amazonian limit is roughly contiguous with the south edge of the Llanos, approximately along the Vichada and Orinoco rivers. To the west Amazonia is delimited by the Andean escarpment. For phytogeographical analysis, the 500 m contour, above which many lowland tropical species do not ascend, may be used to divide the Amazonian and Andean regions. There are also narrow extensions of Amazonian forest along the base of the Andes, most strikingly in Bolivia where the sub-Andean tail of Amazonia reaches 17°S latitude. Amazonia is generally limited southward by the escarpment of the Brazilian Shield and the associated cerrado vegetation.

The landscape within the basin includes dissected sedimentary plains or plateaux and crystalline terrain as well as hilly landforms and wide alluvial terrains (Bigarella and Ferreira 1985). Amazonian soils are sedimentary deposits, to 300 m deep; in Lower Amazonia they have developed over 20 million years. Most of the soils are mixtures of kaolinite (high in aluminium), iron oxides and quartz; they are nutrient-poor owing to their origins and millennia of intense weathering. Calcium, magnesium and phosphorus are very scarce, with phosphorus a limiting factor for the entire ecosystem (Bigarella and Ferreira 1985; Chauvel, Lucas and Boulet 1987; Irmler 1977; Jordan 1985; Leopoldo et al. 1987). Nevertheless there is an extensive diversity of c. 22 soil units recognized in six classes. Latosols and podzols are frequent, with gleysols (clays and silts) nearer the rivers, and more sandy soils to the north-west, south and east (Brown 1987). Soils appear to correlate poorly with large-scale phytogeography in Amazonia, but can be critical in explaining local distributions of species (Daly and Prance 1989; Jordan 1985).

The tributary rivers of the Amazon differ in the physical and chemical properties of their waters; these variations strongly affect the vegetation. The three basic types are:

1. Whitewater rivers especially of Upper Amazonia, which erode and drain the Andes, carrying a heavy sediment load; they are neutral or slightly acidic. The Solimões River and Amazon River itself are examples. Their floodplains have relatively fertile clay soils (Leopoldo et al. 1987; Sioli 1984).

2. Clear-water rivers mainly of Lower Amazonia, draining the ancient crystalline shields. Already long since stripped of nutrients, they have a very low sediment load, and vary greatly from acidic to alkaline (e.g. Tapajós and Xingú rivers).

3. Black-water rivers (e.g. the Rio Negro), which drain the geologically old, nutrient-poor, low (over 200 m) sandy area especially to the north-west between the Andes and the Guayana Shield; their waters are very dark and acidic (especially with humic and fulvic acids) from incomplete decomposition of the organic forest litter, and also have very low sediment loads. Their floodplains tend to be sandy and poor in nutrients (Leopoldo et al. 1987; Sioli 1984).

Most tributaries of the Amazon are of clear or black water. Most white waters are from the south-western quarter of the basin (the Andean drainage that becomes the Solimões and Madeira rivers). The annual inundation, which affects at least 100,000 km² of forest, may last 2-10 months (rarely all year) and be up to 20 m in depth. In Lower Amazonia "terra firme" (i.e. uplands) and the boundaries of floodplains are generally distinct (Daly and Prance 1989; Goulding 1988).

Annual rainfall over most of Amazonia exceeds 2000 mm, with a single pronounced dry season. The annual precipitation in Central Amazonia is 2000-2500 mm. Rainfall generally increases and seasonality decreases to the west, with the area from Leticia to Iquitos and into Ecuador having little seasonality and rainfall in excess of 3000 mm. There is also a high-rainfall area centred in French Guiana, and there are local areas of relatively high rainfall scattered around the periphery of Amazonia.

Return to Top


Most of Amazonia is covered by various manifestations of tropical rain forest, but there are important phytogeographical differences between forests on different substrates. Seasonally inundated forests occur in an estimated 2% of Amazonia immediately adjacent to the major rivers. Forests inundated by black-water rivers ("igapó") have very different sets of species than those inundated by white-water rivers ("várzea, tahuampa"). Similarly forests on white-sand soils ("campinarana, wallaba, varillal") are floristically very different from those on clay soils, and both are distinct from alluvial-soil terra-firme forests. There are also a rich array of different successional stages of these various forest types. Finally, small patches of savanna or scrubland can occur on white sand or in small areas with less rainfall and stronger dry seasons. The vegetation types of Amazonia have been discussed in detail by a number of authors, including Prance (1979, 1987), Pires and Prance (1985), Daly and Prance (1989), Encarnación (1985) and Duivenwoorden et al. (1988).

Return to Top


The forests of Upper Amazonia are the most diverse in the world, with up to 300 species over 10 cm dbh per ha. Leguminosae is usually the dominant woody family; on the poor sandy soils of the Guayana Shield its subfamily Caesalpinioideae is especially prevalent. On the richest alluvial soils Moraceae may equal the Leguminosae in diversity. Other important families of trees include Lauraceae, Sapotaceae and Annonaceae. The dominant liana family is Bignoniaceae, followed in importance by Leguminosae. Epiphytes and forest-floor herbs are relatively poorly represented in most of Amazonia.

Endemism is often associated with substrate specificity, and concentrated in the habitat mosaic of north-central and north-west Amazonia. While there is also some endemism among lowland terra-firme taxa, a typical distributional pattern is for a species that occurs on clay soil to be widespread over a large portion of the region. There are also concentrations of endemic taxa in the relatively rich-soil area near the base of the Andes, the Guayana area and perhaps other locales. The reputed centres of endemism were instrumental in developing the hypothesis of Pleistocene refugia. Whereas it now appears that some putative centres of endemism are artefacts of collection intensity (Nelson et al. 1990), the areas of endemism associated with habitat differentiation, especially around the periphery of the Guayana Highlands, clearly are real.

Return to Top

Species of economic importance

More than 2000 species of Amazonian plants have been reported to be useful to people (UNDP 1992; cf. Duke and Vásquez 1994; Bennett 1992). For example, at least 1250 species have been recorded to be used in Amazonian Peru (Vásquez 1989). Upper Amazonia is the centre of diversity of a number of important crop species (Clement 1989). In the Iquitos region of Peru c. 220 species of fruit are consumed (Vásquez and Gentry 1989 and unpublished). Floodplain and terra-firme forests near Iquitos can provide profits of US$400 per ha per year, mostly from sale of the native fruits (Peters, Gentry and Mendelsohn 1989), but near Manaus, Brazil one finds neither the high natural productivity of native fruits nor such market tastes. Extensive areas of intact terra-firme forest in Brazil's western State of Acre are utilized and occupied by rubber tappers. In eastern Acre they gather Brazil nuts (from the tree Bertholletia excelsa) in the rainy season when latex cannot be collected. For the country, Brazil nuts amounted to 22,000 tons and US$36 million of export value in 1983 (Müller and Calzavara 1986) and US$20 million in 1989 (Ryan 1991).

According to the Association of Exporters of the Manaus Free Zone, Brazil's State of Amazonas in 1984 exported overseas an FOB (free on board) value of US$26.6 million of regional products, mostly plant products, representing 50% of the total overseas exports from the state. These products are from Central Amazonia (see Lescure et al. 1992), with little from near Manaus itself. The most important extractive cash crop exported through Manaus fluctuates between "sôrva" (Couma spp.) and the Brazil nut ("castanha"), which respectively yielded US$6.2 million and US$3.8 million of annual exports. Aniba rosaeodora and A. duckei provide rosewood oil, an essential oil with US$1.4 million FOB value exported. The Madeira River region provides most of the Copaiba balsam, with US$263,000 FOB value exported.

For the Brazilian Amazon, extractive non-timber products effectively traded and recorded by government census amount to US$110 million per year (Vantomme 1990). This underestimates the real full value of such products and services - for example, a hut constructed from forest materials, fruits and fish consumed at home and probably many products sold in informal markets, all of which do not enter into the calculation.

In the Brazilian Amazon 260 species have some economic significance for timber and 50 are marketed in significant quantity, with c. 4000 tree species estimated to have some forestry potential (Sternadt, Ternadt and Camargos 1988; UNDP 1992). Extractive products provide far less than the US$1.2 billion annual proceeds from timber (Vantomme 1990). Though improperly managed, some of the timber is obtained in a non-destructive way, such as the wood for board and molding lumber floated from tidal forests near Breves in the State of Pará. In the Upper Amazon as well, logging of c. 6 species of fast-growing softwoods for peeler logs by floating them out of várzea floodplains is common and non-destructive. Fishes used for food that are dependent on forest fruits or forest insects are conserved in these areas.

Sustainable economic values

Prance (1990) suggested several sustainable economic activities for the Amazon. These include: "adaptation of indigenous agroforestry techniques to use mixed cultures where crops are grown; restoration with sustainable timber plantations of areas which have already been destroyed; the increase of extractive reserves and development of markets for ["green"] products that are extracted from the forest; the greater use of oligarchic forests which are abundant in some parts of Amazonia; and the development of agriculture only in the few places where the soils are suitable, such as in the flood plains of white-water rivers."

Extractivism and forest enrichment work well in certain situations, such as in the tidal forests near Belém where few crops could be grown and the large city provides markets for perishable palm hearts and fruits. As many as 40 people per km can be found living off the tidal forests (A. Anderson 1989, pers. comm. to B. Nelson), where the most important economic species is the "açaí" palm (Euterpe oleracea). US$20 million per year is obtained from export of palm hearts of this species (Ryan 1991), which can be harvested in a sustainable manner, as the trees grow in clumps with juveniles around the base.

Often the major extractive activities have a questionable future. They are destined to be replaced by more efficient plantation or agricultural systems (Bunker 1985; Homma 1992), as is now happening with rubber (Hevea brasiliensis) planted south of Amazonia (Parfit 1989; Ryan 1991). Plantation technology has been developed for the Brazil nut (Müller and Calzavara 1986) but has not yet replaced native harvests. Selected graft stock in Brazil-nut plantations can produce consistently larger seeds, which have a much higher per-kg sale value than the average natural tree of some regions. Just 30,000 ha of such plantations could replace all of Brazil's present extractive production.

So far, very little agricultural activity is being developed on terra firme around Manaus. Most meat and garden vegetables can be transported economically from the south via Belém, because barged trucks return empty after delivering 80% of Brazil's consumer electronic items. This situation permitted Manaus to generate US$4-5 billion in goods and services each year until recently. Every 48 hours the city thus generates receipts equivalent to the total annual Brazil-nut exports. The extremely limited deforestation around such a large city is quite remarkable. An essential part of this system is the use of barge traffic rather than roads, which avoids the inevitable deforestation that penetrates from all trafficable roads in Amazonia.

Before 1991, 6% of all foreign tourism money spent in Brazil was spent in Manaus and the number of visitors grew by 5.8% per year. Two-thirds were attracted by the natural environment, and each spent US$760 during an average 5-day stay (A Critica 17/5/1992). The most spending is in the city itself, even by those tourists taken to jungle lodges a few hours away. Adventure cruises based on foreign vessels come fully outfitted and self-contained, but leave some receipts in small towns by purchasing handicrafts and hiring local entertainment and guides. Though tourism is a large business, actual ecotourism is an extremely small fraction, and employment in backwoods areas for potential deforesters constitutes a similarly small fraction of the ecotourism money.

Tourism receipts are not a significant factor in land-use choices made by those wielding axes in the far reaches of Amazonas State, nor by squatters, ranchers and timber harvesters along the hot dusty roads of southern Pará. As the population of the region increases, any green options for making a livelihood must produce greater receipts. Even if rubber tapping were to revive, it could only employ a small portion of the more than 15 million people in Brazil's designated "Legal Amazon".

Threats to extractive reserves have prompted proponents of extractivism to search for additional non-timber cash crops from intact forest. Over 80 promising species have been identified for research and promotion by the Instituto Nacional de Pesquisas da Amazônia (INPA) in Manaus (M. Allegretti 1991, pers. comm. to B. Nelson).

Exploitation of hydroelectric potential as well as mining of iron ore, bauxite, cassiterite and gold are furnishing the largest economic returns from the Amazon's natural resources. Access controlled by the Companhia Vale do Rio Doce (CVRD) to the large Carajás mining concession has kept that area an island of green within a patchwork of farms and pastures. When properly controlled and conducted by well-capitalized and regulated companies, hydroelectric activity and mining can be relatively benign, although synergistic effects, such as charcoal production for pig-iron smelting outside the Carajás concession, can be devastating. Deforestation for ranching and small farms far exceeds the loss of forest to hydroelectric and mining activities, which also generate much more wealth per hectare of deforested or flooded land. Unfortunately, such wealth is not well distributed, so tens of thousands of urban and rural poor turn to less efficient and environmentally damaging gold prospecting or itinerant agriculture.

Dry-season Landsat images indicate that in Maranhão, Tocantins, Pará east of Belém, Amazonas along várzea floodplain lakes, and parts of Roraima, most of the terra firme deforested for both pasture and small farms has been abandoned to regrowth (or is in a long fallow cycle). However in north-eastern Mato Grosso, in Rondônia and much of Acre, most deforested land is under intensive use, with satellite spectral signatures suggestive of dry pasture or bare soil.

In the Lower Amazon below Manaus to the mouth of the Xingú River, grazing of cattle takes place during low water on fertile floodplain savannas and deforested várzea. Terra-firme pastures are maintained as holding areas for grazing during the high-water months while the lowlands are flooded. This seasonal alternation means that the upland deforestation will penetrate only a limited distance from rivers, until (or unless) good roads are established. Large várzea/terra-firme landholdings often belong to owners located in Manaus, Parintins or Santarém where the beef is consumed. This extensive type of cattle-ranching is successful and has been established and maintained for generations, without infamous subsidies as are associated with non-sustainable terra-firme ranching in south-eastern Pará. Ranching also appears to be sustainable on the more fertile terra-firme soils of the Amazon Basin, such as in large parts of Acre and Rondônia to the south-west.

Return to Top


Extractivism and ecotourism are often suggested as ideally benign and sustainable activities for the Amazon, but their economic potential appears grossly insufficient. This suggests that Amazonia, far from being a demographic void, may already have exceeded an ideal population size compatible with maintaining most of its forest intact. Furthermore, masses of poverty striken farmers from the populous north-east of Brazil and rural poor in the extreme south stand ready to swell the Amazonian population at the slightest evidence that one can make a marginal living.

After depleting timber in the Atlantic coastal forest, hundreds of lumber mills and mechanized timber harvesting operations are now using the closest available Amazonian hardwood stands, attracting labourers from the north-east for a US$112 monthly wage (Veríssimo et al. 1992). Eighty thousand jobs offered by the factories of the Manaus Free Zone at similarly low wage levels led in just 15 years to a large sprawl of shanty towns. And 350,000 gold and tin prospectors readily shifted into remote areas, ignoring laws, Amerindian rights and international boundaries (Veja 12/2/92).

Deforestation of the Brazilian portion of Amazonia in 1978, 1988, 1989, 1990 and 1991 has been measured precisely by Brazil's Instituto Nacional de Pesquisas Espaciais (National Space Research Institute) (INPE) (Fearnside, Tardin and Meira Filho 1990; INPE 1992; ESP 1992). The region includes 4.9 million km², of which 3.9 million km² were originally covered in natural forest according to INPE (or 3.6 million km² according to Nascimento and Homma 1984). The total area deforested was 426,000 km² as of August 1991 - 10.9% of the forests and 8.7% of the region. The rate was highest in 1987, although there are no reliable measurements for that year. Annual deforestation estimates were 21,500 km²/yr for the 11.6-year period 1978-1989, 18,800 km² for 1988-89, 13,800 km² for 1989-90 and 11,100 km² for 1990-91. The rates dropped after 1987 due to a number of possible factors, including a reduced threat of agrarian reform by 1988, heavy rainfall in 1989, drastic reduction in the money supply for investments in 1990 and widespread popular sentiment in Brazil and abroad opposed to deforestation, spurring more effective law enforcement. Yet over 10% of the c. 4 million km² of Amazonian forest has been cleared, and deforestation has continued at a rate of more than 10,000 km² per year (Fearnside 1991; Brown and Brown 1992).

Habitat degradation covers a larger area than deforestation sensu stricto, since the forests between closely spaced "fish-spine" colonization roads are intensively logged and hunted. Deforestation and habitat degradation are concentrated along southern and eastern flanks of the naturally forested areas (Skole, Chomentowski and Nobre 1992), where extensive road building, better soils, higher populations of nearby rural poor and proximity to a considerable domestic market for sawn lumber are the main causes. Large ranches, timber harvesting and small holdings of poor farmers are about equally important agents of deforestation/degradation in the Brazilian Amazon, which vary widely in relative importance from state to state.

There is a possibility as the eastern Amazon Basin continues to be deforested that the naturally dry zone could extend westward towards Manaus - with consequent massive forest mortality. Water vapour input to the Amazon Basin is from the Atlantic Ocean, carried westward by prevailing winds. Through evaporation and transpiration, terra-firme forests recycle 50-60% of the rain received back into the atmosphere, to fall again farther west (Salati and Marques 1984).

Although only four to eight trees are saleable in 1 ha of terra firme in eastern Pará, mechanized logging kills or damages 26% of preharvest standing trees (Uhl and Vieira 1989) and allows greater penetration of sunlight to the forest floor. Abundant dry slash fosters deep penetration of fires from nearby pastures (Uhl and Buschbacher 1985). No significant replanting of native terra-firme species occurs in the harvested heterogeneous forests, nor is it likely to occur until accessible primary forests are depleted. Extrapolation of recent logging rates indicates the depletion of commercial stocks for the entire State of Pará in 80 years (Uhl and Vieira 1989).

Road access along the southern and eastern rim of Amazonia almost always sets off an acceleration of deforestation for pastures and subsistence farming. The phenomenon has metastasized to more centrally located foci of road-based colonization in eastern and south-central Roraima, eastern Acre, along the Trans-Amazon highway at Sucunduri in south-eastern Amazonas, and between Itaituba and Altamira in central-western Pará.

Where there are few roads, poorer soils and great distance from domestic hardwood timber markets, as in Amazonas State, deforestation is largely limited to a 5-km deep rim on the terra-firme shores of dendritic lakes connected to the Amazon. These productive lakes have both abundant food fish and permanently dry shores. The 5-km limit is the distance subsistence farmers walk daily from their lakeside homes in search of primary forest to burn and obtain a better nutrient input for swidden manioc fields. Though much richer, floodplain soils do not permit a reliable manioc harvest (Manihot esculenta) due to occasional higher floods.

Attempts at ecologically benign development throughout Amazonia probably are close to futile if Brazil and the other Amazonian countries do not solve the problems of poverty, rapid population growth and overpopulation outside of this region.

Return to Top


Some kind of protected status has been given to at least 1,812,510 km² or c. 23% of Amazonia (UNDP 1992). However, the total includes such areas as indigenous territories, forest reserves and extractive reserves, which often function more as sources of timber or other forest products than as conservation units for nature. Of that total, 245,560 km² or 3.2% of Amazonia constitutes National Parks and perhaps twice as much (562,230 km² or 7.4% of Amazonia) is included in various Ecological Reserves, Faunal Refuges and Biosphere Reserves with a primarily conservational purpose. Peres and Terborgh (1995) found 117 of all 459 conservation units in Amazonia or 25% to be strict nature reserves, accounting for 41% of the region's land that is under some form of institutional protection; their average size is 4765 km². The largest portion of Amazonia by far is in Brazil. About 143,390 km² or 3.6% of Brazilian Amazonia is totally protected (Brown and Brown 1992), including 97,000 km² in National Parks (UNDP 1992). While the total reserve area in Amazonia is impressive, nearly all the conservation units have been created in the last decade, and many exist only on paper. A map prepared by Conservation International (1991) from Workshop 90 (Manaus) pinpoints specific areas for conservational focus in Amazonia (cf. Rylands 1991).

Return to Top

Centres of Plant Diversity and Endemism


French Guiana

SA3. Saül region


Brazil, Guyana, Surinam

SA4. Transverse Dry Belt of Brazil



SA5. Manaus region

Brazil, Colombia, Venezuela

SA6. Upper Rio Negro region


SA7. Chiribiquete-Araracuara-Cahuinarí region



SA8. Yasuní National Park and Waorani Ethnic Reserve

Peru, Colombia

SA9. Iquitos region



SA11. Lowlands of Manu National Park: Cocha Cashu Bioplogical Station


S.A. Regional Overviews
I. Caribbean (of South America) V. Interior dry and mesic forests
II. Guayana Highlands VI. Andes
III. Return to Top This Region VII. Pacific Coast
IV. Mata Atlântica VIII. Southern Cone

Return to: South America Overview

North | Middle | South

CPD Home  

Botany Home Page | Smithsonian Home Page