Mesoamérica Foundation

The Past and Future Amazon

The climatic history of the Amazon rain forest indicates that the ecosystem is well adapted to certain natural disturbances. Does it have the resilience to tolerate human exploitation?

By Paul A. Colinvaux

The rain forest covering the vast Amazon River basin of South America looks from the air like a uniform green carpet cut here and there by water. Actually the forest is anything but uniform. The "carpet" is the forest's canopy, formed by the broad leaves of the many different kinds of giant trees, and this canopy is but the topmost layer of an ecosystem supporting more species than any other region on earth. The rain forest is home to perhaps 80,000 plant species (including 600 kinds of palm alone) and possibly 30 million animal species, most of them insects.

This remarkable diversity was once presumed to be a product of an ever-lastingly stable climate of abundant rain and warmth. Spared the short-term catastrophe of winter every year and the long-term disaster of glaciation, theory said, the tropical domain known simply as "the Amazon" would lose few species and accumulate many. Yet evidence from the field suggests that the Amazon is subject to changes on all time scales, including cooling when ice-age climates loose glaciers over more northern territories. Moreover, far from being disastrous to life in the Amazon, the moderate climate disturbances in the region may actually help account for the splendid diversity of the Amazon rain forest today.

The influence of past climates on species diversity in the Amazon ecosystem is of more than academic interest; it is an invaluable predictor of the forest's ability to endure change. As human beings lay waste to massive tracts of vegetation, an incalculable and unprecedented number of species are rapidly becoming extinct. Nations with jurisdiction over the rain forest -- notably Brazil, Venezuela, Colombia, Ecuador and Peru -- need to identify management strategies that can save as many species as possible and yet be compatible with the need for economic development. Those who devise such strategies must take into account the kinds of stresses that affect species diversity, and so it is important to learn what insults the ecosystem has tolerated throughout history.

The old belief that climatic stability accounted for species diversity in the Amazon emerged, strangely enough, from observations of the deep sea. Howard L. Sanders of Woods Hole discovered high diversity among the mud-dwelling animals of the deep-ocean floor in spite of the cold darkness and low biological productivity there. He argued that such hostilities to life are offset by the perpetual sameness of the place. Without significant fluctuations in physical conditions, the extinction of species that are adapted to prevailing conditions should be rare. In the course of time new species would continue to evolve, and so the rate of speciation would be greater than the rate of extinction, resulting in the accumulation of great diversity.

Sanders suggested that the Amazon forest and other tropical forests could be thought of as analogous to (albeit more productive than) the deep sea: places with a stable supply of annual moisture and warmth in which extinction should be rare. The absence of winter and glaciation and evidence that rain-forest trees have persisted for some 30 million years or longer somewhere in the 3,000-kilometer-wide, 1,000-kilometer long Amazon supported this view.

Then, in 1969, several observations cast doubt on the validity of the stability theory for the Amazon, implying that the climate there has fluctuated significantly in the past. Jürgen Haffer, then of the Mobil Research and Development Corporation, noted that different corners of the basin held their own, separate bird faunas in spite of the fact that essentially unbroken green forest spread from the western edge of the Amazon to the Atlantic Ocean in the east. Haffer and other biologists also noted species disjunctions among butterflies and a few other groups. This pattern presented a glorious puzzle to workers studying the distributions of plants and animals: Why would the populations have become isolated if the habitat in which they lived was continuous?

Haffer proposed an explanation so bold and logical that many of us wished we had thought of it first. He suggested that the modern isolation has its roots in times past, in the latest ice age. Observing that the regions of species isolation generally centered on outcrops of high ground and that lowlands are drier than uplands, he proposed that in glacial times the Amazon lowlands become a sea of near-desert arid plain; meanwhile the more elevated regions become islands of moisture and hence serve as refuges for the fauna and flora of the rain forest. Populations that were once continuous diverge along this ecological gradient and become permanently separated as they adapt to their upland habitats.

This hypothesis has great appeal because it appears to explain not only species disjunction in the Amazon basin but also the unusual species diversity there. The ice-age refuges would, as the earlier stability hypothesis predicted, protect existing species from extinction. But the periodic geographic isolation of related populations (there have been an estimated 13 ice ages to date) would facilitate ever-increasing divergence.

For the refuge hypothesis to stand there has to be proof that the lowlands of the Amazon, which now support wet forest, do indeed dry out in glacial times. No one has yet found compelling evidence that the Amazon was actually arid, although circumstantial evidence has been provided by computer models and also by sediment samples drawn from several sites in the tropics.

For example, John E. Kutzbach and Peter J. Guetter of the University of Wisconsin at Madison, have experimented with computer models of the global ice-age climate, predicting reductions of up to 20 percent in the monsoon rains in the tropics. Such a reduction would certainly spread aridity into strongly seasonal regions ­­ those with true dry periods each year ­­ including at least some parts of the Amazon. (Many so-called seasonal regions in the tropics do not have true annual dry periods, merely ones that are less rainy than other seasons).

Daniel A. Livingstone of Duke University published some of the earliest field data addressing the issue of aridity. He found by examining fossilized tree pollen, which is a good indicator of the local flora, that several tropical rain forests in modern Africa grow where dry woodland or savanna grew between 12,000 and 20,000 years ago; that is, late in the last ice age (from about 10,000 to 70,000 years ago). Hints of similar ice-age aridity in what are now wet tropical regions also came from outside the periphery of the Amazon basin, suggesting that what was true for ice-age Africa might also have been the case of the Amazon basin.

I myself gathered data suggestive of Amazonian aridity from the Galápagos Islands, which are about 2,000 kilometers directly west of the western-most edge of the Amazon basin on the Equator. A well-dated sediment core, or pipeful of undisturbed sediment, from under the floor of the Galápagos' one freshwater lake shows that the lake was without water in the last ice age. It did not fill until the onset of the Holocene (the current, postglacial period).

On the continent itself others have found similar evidence at sites north of the Amazon. In northern Venezuela what is now deep Lake Valencia was without water at the end of the last ice age, and what is now a swamp on the coast of Guyana held dry vegetation. Different work suggests that a strongly seasonal forest to the south of the Amazon basin, near Säo Paulo, Brazil, was arid as well.

More evidence that the glacial-age Amazon lowlands might have been arid has been obtained from deep-sea cores from the Atlantic Ocean, right near the mouth of the Amazon River. John E. Damuth and Rhodes W. Fairbridge, then of Columbia University, have found that sediment delivered from the east-flowing river during the last ice age included a small fraction of feldspars, minerals from crystalline rocks, that could conceivably have been the erosion products of an arid landscape.

In spite of the abundance of the circumstantial evidence for ice-age aridity, I have become convinced that the evidence is not strong. Georg Irion of the Research Institute and Natural History Museum in Frankfurt has shown that he feldspars in the Atlantic deposits probably came from rocks in the Amazon basin, but not from rocks on the surface. The sea level fell in the ice age, and the Amazon's many tributaries presumably cut their channels deeper at the same time, carving the feldspars in the process.

The sediment-core data from the Galápagos Islands and around the Amazon on the mainland are also weak because the regions from which the sediments were taken, although closer than Africa, are still quite a distance from the Amazon rain forest. Moreover, the regions frequently come under the influence of different climatic forces. Between the study sites and the Amazon basin are the tall Andes, the Venezuelan and Guiana Highlands and another immense line of hills known as the Mato Grosso (the great scrub), which constitute respectively the western, northern and southern boundaries of the basin. Each of those boundaries is a site of climatic tension, so that the climate on one side of the heights is often quite different from the climate on the other side.

It may be that in the strongly seasonal regions of the modern Amazon reductions in the monsoon rains do result in semiaridity during glacial times -- perhaps just below the Guiana Highlands in the northeastern corner, immediately to the west of the Mato Grosso in the southwest and in some parts of the central basin in Brazil. For most other, wetter regions of the Amazon, where annual rainfall is from two to five meters, it is quite unlikely that even a 20 percent reduction in the rains would make any practical difference to life.

In 1985 the fact that there were no radiocarbon-dated, glacial-age fossil data of any kind from Haffer's postulated refuges or the lowlands of the Amazon prompted me and my colleagues at Ohio State University to seek data from the Amazon proper. We found the only fossils that have yet been dated to an ice age. They suggest that the Amazon was not dry. The region did undergo a significant change, however; it became cooler by several degrees Celsius.

We found our specimens by serendipity, while searching the western-most edge of the rain forest, in eastern Ecuador, for old lakes whose sediments we hoped would hold ancient records. At one point we found ourselves on a dirt road in Mera, whose 1,100-meter elevation in the eastern foothills of the Andes places it close to the upper limits of today's rain forest and within one of the major putative refuges. There we noticed old tree stumps and logs that we embedded in an exposed patch of sediment. Because we had a bit of time on our hands as we waited for a flight to a distant lake, we took a collection of sediment and wood samples. Radiocarbon dating of the wood indicated that some samples were 35,000 years old and others 26,000 years old. Although earlier than the time of maximum glaciation than the time of maximum glaciation in the most recent ice age (18,000 years ago), these are dates of glacial times.

Another analysis of the wood samples showed they included softwoods, meaning conifers. Now, the only conifers in modern Ecuador are Podocarpus trees of the Andean forests, which today grow at elevations that are at least 700 meters above the Mera site. Indeed, pollen analyses of the sediment samples showed that the ancient forest at Mera had a strong Andean cast to it and that it has no direct analogue in modern tropical America.

We conclude that the Podocarpus trees, which require both moisture and the relative coolness now found only in the mountains, grew at least 700 meters lower in glacial times. Taken together with the pollen data, this conclusion suggested to us that Mera was moist during the last ice age but was too cool to support a modern kind of rain forest. A standard formula suggests that the temperature in the Ecuadoran foothills was at least 4.5 degrees Celsius cooler than it is today. (The formula assumes that in moist air the temperature cools six degrees Celsius for every 1,000 meters of ascent up a mountain.)

There was, then, no refuge for warmth-loving species in Mera. If the temperature depression was widespread in the Amazon, as seems reasonable, then other high-lying parts of today's rain forest might also have been too cold to support rain-forest refuges. Given the well-established fact that typical rain-forest trees persisted somewhere in the Amazon basin, the data suggest that the rain forest was restricted to somewhere in the lowest region, where refuge theorists had predicted aridity. If the trees did in fact survive only in the lowlands, it follows that the lowlands were not arid, or at least were not uniformly so.

Obviously more than one data set from the ice age and data from the coldest part of the period are needed before a picture of the ice-age Amazon can be reconstructed with confidence. Still, the existing evidence suggests that the refuge scenario is no more supportable than the notion of imperturbability that it replaced. It also suggests that the rain forest shrinks in response to ice-age cooling and expands again during warmer interglacial periods, such as the one we are in now. In essence we have the inverse of the refuge picture: some part of the lowlands remains as a relatively warm, wet preserve for rain-forest trees and the uplands -- formerly thought to be refuges -- become inhospitable.

What then explains the species diversity and the species disjunctions identified by biologists? Changes in the climate between glacial and interglacial periods may still be important, albeit not in the way the refuge theorists predict. The migration I propose of rain-forest species to higher ground in warm interglacial periods such as the present one would provide opportunities for the formation of novel mixtures of species. Populations from the warm lowlands of glacial times could then become adapted to their new upland conditions and diverge from their ancestors. Perhaps something like that happened to the birds and butterflies now living in the uplands.

Another part of the answer certainly has to do with the varied climate and geography of the Amazon. The game of speciation has always been played across a huge playing board: the Amazon basin is almost the size of the continental U.S. All parts of the board are superficially alike and indeed have some species in common, but the territory does not hold one monolithic ecosystem. Rather, there are large regional and local differences in rainfall, seasonality, soil and susceptibility to flooding, all of which can affect the mixture and evolution of species in any given area.

A further explanation for the diversity is known as the intermediate-disturbance hypothesis, put forward by Joseph H. Connell of the University of California at Santa Barbara and independently by Stephen P. Hubbel, now of Princeton University. This hypothesis suggests that the highest species richness will be found not where the climate is stable but where environmental disturbance is frequent but not excessive.

The intermediate-disturbance hypothesis concedes that massive catastrophes lead to large-scale extinction. This is what would happen if asteroids struck the earth, as they may have done at the time of dinosaur extinction. But lesser and local catastrophes, such as storms and flooding, do not generally extinguish entire species. Instead, by killing some fraction of a dominant species, the smaller hazards prevent winner-take-all competitions from going to their conclusion (extinction for one species) and give initially weaker organisms the opportunity to establish themselves.

The effect of intermediate disturbances can be seen quite readily in rain-forest trees. Great gashes and succession communities are everywhere in the Amazon and other tropical rain forests wherever gaps are cut in the forest by the felling of large, sun-blocking trees (such as rubber trees and balsa), various species of short-lived, sun-loving plants, together with their associated fauna, populate the gaps. This process initiates a series of successions that, if they go undisturbed, culminate centuries later in the groupings of giant trees characteristic of mature rain forests. The big rain-forest trees actually fall fairly easily because their roots are shallow: often half of their root mass is in the top 20 centimeters of soil. All it takes is a strong push by wind or the washing away of topsoil at the base of the tree by floods or moving streams.

A growing body of data suggests that at least throughout the Holocene (and probably throughout history) the Amazon has been perturbed in one region or another by storms, erosion and other forces. Indeed, the Amazon's topography apparently changes from century to century and even from decade to decade.

Some of what is known about the Holocene's history comes from the sediment cores my colleagues and I obtained from the Ecuadoran lakes alluded to above. Our oldest data are from lakes in volcanic-explosion craters, the only crater lakes yet known in the Amazon. One of these lakes, known as Kumpak in the Shuar language of the local populace, has muddy, oxygen-depleted water typical of an Amazon river-fed lake. The other lake, Ayauch, is transparent and blue, with oxygen deep in the water. It is almost like an alpine lake -- a wildly exciting and improbable thing to find in the jungle.

Mark B. Bush, a postdoctoral fellow in my laboratory, has completed pollen analyses of the 7,000-year record from Ayauch, finding evidence for rain forest throughout most of that time, although there was a prolonged local drought 4,000 year ago, long after the ice age ended. (He also found evidence that people had grown maize in the vicinity 3,000 years ago -- earliest record of maize cultivation in the Amazon.)

Kam-biu Liu, now of Louisiana State University, found a different kind of history in the Kumpak sediments. These are banded throughout with layers of different textures, suggesting a history of heavy rainstorms that washed sediment in pulses from the banks into the lake. In other words storms have been remodeling the land Kumpak for at least the past 5,000 years.

Again in the Ecuadoran rain forest, but from the north, cores we obtained from four lowland lakes abandoned by their parent rivers centuries ago have also yielded clear evidence of at least one extended period of unusually powerful storms and forest perturbation in the west. The sediment samples from each lake were topped by a layer of gyttja, typical lake mud, first laid down 8,000 years ago, according to radiocarbon dating. Beneath this layer in all four lakes was river sediment that had been deposited continuously since about 1,300 years ago. The lakes, therefore, were left by their parent rivers roughly 800 years ago and have not been reentered since.

We postulated that between 800 and 1,300 years ago excessive rains in the mountains to the west caused massive flooding and forced rivers to spread back to old, once-abandoned channels in the lowlands, which became largely inundated. Our pollen analysis of the old river mud indicates that early succession forests were present on temporary sandbanks that formed when the floods, which presumably killed mature trees, subsided temporarily.

Marcia L. Absy of the Amazon Fisheries Institute in Manaus, Brazil, has also found evidence of flooding at the appropriate time in lakes of the central Amazon. Absy cored five varzea lakes (which are filled by rivers during the wettest and then are abandoned to evaporate slowly in the driest seasons from different water sheds near Manaus, including lakes spawned by south-flowing rivers from the dry Guiana Highlands, by north-flowing rivers from the Mato Grosso and by east-flowing rivers draining the Andes of Ecuador and Peru. Her data show that only lakes in the path of the western drainage hold the flood record -- not those in the paths of the north- and south-flowing streams. In other words the stormy period was a local phenomenon of the west and had long-range effects solely on regions in the path of the swollen Amazon tributaries.

The western Amazon, then, has been disturbed in the Holocene by at least one unusually drastic climate change, whose effects lasted almost a millennium. Doubtless other climatic events of equal long-term effect await discovery in the turbulent basin.

On time scales of a few centuries the perturbing effects of erosion by moving streams are particularly evident. Some rivers are fast-moving and wear away sediment rapidly; others are sluggish but nonetheless erode sediment as they move. Many rivers flow now through one channel, now through another. As they change direction they topple trees in their path; they also leave behind sediment ripe for colonization. All are subject to flooding in rainy periods, at which time they wash away trees on their banks and redecorate the landscape. The turbid western tributaries of the Amazon are particularly active. Indeed, various estimates suggest that 80 percent of suspended solids in the lower Amazon comes from the west.

On the basis of published measures of the sediment discharged at the mouth of the Amazon River each year, my group tried to calculate how fast the huge rivers draining the Andes are eroding the western region. That is an exercise in unsafe extrapolation, but it does suggest that many centimeters of land surface are lost every century. Perhaps half the rooting depth of a typical, mature Amazon tree can be washed away in less than the tree's normal lifetime.

Jukka S. Salo and his team at the University of Turku in Finland, working in the Peruvian Amazon, have obtained better evidence of the disturbance caused by erosion in the west. With the help of satellites they have constructed maps of different forest types, including both early succession communities (which typically grow on sediment left behind as rivers progressively more their channels) and mature communities (which become established only centuries after an old riverbed has been abandoned). From these maps they determined that in the past couple of centuries as much as a quarter of the Peruvian forest has been swept away and rebuilt as new communities.

Rivers also perturb the land on smaller time scales. With aerial photographs taken 13 years apart, Salo and his group showed that a single small river had reworked 3.7 percent of its floodplain in these few years; on the average, it had eroded 12 meters of land per year.

Most people are not surprised to learn that the Amazon forest is disturbed by storms, flooding and erosion. Few, however, expect a rain forest to be perturbed by natural fires, and yet rain-forest trees apparently do burn in the Amazon. Recently Robert L. Sanford of the University of California at Berkeley and other investigators found layers of charcoal in soil pits in southern Venezuela, at the northern tip of the Amazon basin. Radiocarbon dating showed that some of the samples were 6,000 years old and hence were deposited before human beings are thought to have entered the region. At least some of the charcoal layers, then, must be products of natural wildfires.

What would cause such fires? Rain-forest trees can be thought of as water-cooled energy traps. They spend their days absorbing intense solar radiation on the huge surface areas of their thin, broad leaves, an undertaking that is possible only if the trees can dissipate heat by evaporating immense quantities of water. Sanford thinks it possible that in a mere month without rain the trees might expend all the water within reach of their shallow roots, after which their leaves would wilt, overheat and be lost. The sun could then scorch litter under the leafless trees, enabling spontaneous combustion or lightning to set the dried forest alight. The random properties of weather can be expected to produce such a dry spell in the rain forest once or twice in a span of several thousand years.

It becomes clear that, indeed, the Amazon basin has always been a place disturbed. In ice ages, if our single data set is representative of the region, the land apparently cools and the forest is displaced. In warmer times the Amazon forest is probably subject to the kinds of disturbance identified from lake-sediment records and other records of the Holocene. Different places at different times are beset by storms, flooding and erosion, and they burn in the rare periods when no rain falls for days on end. Few patches are likely to go undisturbed for more than a century or two. The result today is a mosaic of gaps, successions and mature forest ­­ and a wonderful assortment of plant and animal species.

What does all of this say about the future of the Amazon ecosystem, now severely imperiled by humankind? The fact that species have accumulated in a place of constant change suggests that the fauna and flora can tolerate some human activity, if the activity is on the order of natural intermediate disturbances that always leave survivors. It must be emphasized, however, that nothing in the history of the Amazon seems to have approached the clear-cutting now being inflicted on the system by human beings. Such activities are more akin to the catastrophic forms of natural disturbance that have led to the extinction of vast numbers of species in the past.

Moreover, the larger animals that dwell in the rain forests cannot survive the devastating effect of modern firearms. Indeed, plant-eating primates and sloths that graze in the canopy and their flying predators, such as the harpy eagle (an engine of destruction so powerful that it can tear monkeys out of the canopy), are desperately vulnerable to shotguns. One person with a 16-gauge shotgun can remove all the harpy eagles and less motile primates within 10 kilometers of a camp in a year, and thousands of people have done just that. Protected preserves are the only hope for such animals, and the governments of the Amazon are setting aside land for this purpose. The proper size for these refuges is now an active area of research.

The trees are another story. The contemporary reality is that much of the Amazon basin will be turned into pasture as people clear land for cattle grazing. The only hope for trees and perhaps for other plants and insects probably lies in the development of wise uses: ways to generate cash flow from the remaining forest that inflict no more disturbance than the rain-forest species are accustomed to. Perhaps parts of the forest could be set aside for vacation retreats or retirement communities or for industries that manufacture products that do not require enormous amounts of power and would not pollute the ecosystem. History does suggest that parts of the Amazon can be exploited productively without causing mass extinction, but wise use must be the overriding theme.

Paul A. Colinvaux, author of several books on ecology, has been professor of zoology and anthropology at Ohio State University since 1972. A native of Great Britain, he obtained bachelor's and master's degrees at the University of Cambridge before moving to the U.S. and earning a Ph.D. in zoology from Duke University in 1962; he went to Ohio State in 1964. When he is not engaged in his work, Colinvaux enjoys accompanying his wife, a coral reef biologist at Ohio State, on her diving expeditions.

This article appears here courtesy of Dr. Colinvaux. A translation in Spanish will soon be published by Mesoamerica Foundation; the English text was first published in Scientific American in 1989.

Mesoamerica Foundation will have four representatives present at the COP16/CM6 Conference in Cancun, November 29 to December 10, 2010.

This is the 16th Conference of the Parties of the United Nations Framework Convention on Climate Change and 6th Conference of the Parties serving as the Meeting of the Parties to the Kyoto Protocol (COP16/CMP6).

For more information, click on the image to the right.