Genesis of Eden

Genesis Home

Leakey, Richard; Lewin, Roger, 1996
The Sixth Extinction,
Weidenfield and Nicolson, London

NOTE: This extract is included as an essential reading for transforming the world. You are requested to purchase the book yourself as it is, without question, essential reading material.

Richard Leakey is the worlds most famous living paleo-anthropologist. He was for many vears Director of the National Museums of Kenya and until recently, Director of' the Kenya Wildlife Service. He has written (with Roger Lewin) Origins, Origins Reconsidered and People of the Lake. His most recent book The Origin of Humankind (in the Science Masters series) was called by the Sunday Times an outstanding account of our current understanding of human evolution'. Richard Leakey lives in Kenya. Roger Lewin is Associate of the Peabody Museum, Harvard University. A successful author in his own right, his books include Bones of Contention, for which he won the British Science Book Award in 1988, and Complexity: Life at the Edge of Chaos. He lives in Cambridge, Massachusetts.

Stability and Chaos in Ecology

IF YOU WERE TO TRAVERSE the globe from pole to equator, you would see what has been called "nature's infinite variety" writ large. From frigid tundra and Alpine meadows, through temperate woodland and pastures, to tropical forests and open savannahs, you would experience a huge range of different ecological communities. This difference is an important motivation for North Americans and Europeans to visit my own country, Kenya, each year. The contrast between the plant, animal, insect, and bird communities in the home and visited environments is enormous. Not only does the number of species in ecological communities increase as one moves from high to low latitudes, but the types of species in those communities change too. (Polar bears are not to be found as members of tropical ecosystems, nor are large primates-humans aside-seen as natural components of temperate or Arctic fauna.) This large pattern is, of course, partly a consequence of the adaptation of species to local conditions, the most noticeable of which are temperature and humidity. A local ecological community, therefore, is a collection of species that in the midst of their individual differences have a common adaptation; that is, to local environmental conditions. You don't have to be a globe trotter to see variety-great, if not infinite-in nature, however. Anywhere you may find yourself in the world you would be among ecological communities that differ from one another, sometimes profoundly so. I've already described the dramatically contrasting ecosystems that cloak the Great Rift Valley, in Kenya. Again, it is easy to recognize an important source of this diversity here: from the floor of the valley to the peaks of the rift's high walls, a myriad of microclimates prevail, offering contrasting conditions for different kinds of species, Evolution and adaptation work at all scales, producing patterns at all scales. One of the goals of ecology, as the University of New Mexico ecologist John Wiens has put it, "is to detect the patterns of natural ecosystems and to explain the causal processes that underlie them." In what I've described so far, the major process is, I've suggested, species' adaption to local conditions. But anyone who has strolled through a wood or across a meadow, and is observant, knows that he or she is passing through a patchwork of nature, not through complete uniformity. One kind of tree will be common for a while on the walk, only to be absent later; one type of flower will be rare in the south end of the meadow and abundant in the north. This patchwork of nature is a patchwork of similar though distinct ecological communities. What are the processes that have shaped these patterns? There's nothing obviously different about the environmental conditions at the two ends of the meadow, so why are the ecosystems there not identical? Perhaps even the keenest observer might miss crucial environmental differences, however, such as soil chemistry or variations in the level of the water table. This kind of explanation is popular among some ecologists, as explicated recently by Seth Rice, a biologist at the University of North Carolina. "All environments in all ecosystems are variable in space and time," he wrote in a major review article. "The environmental patchiness is based on physical and chemical gradients that are ubiquitous."' In other words, the patterns in the biological realm are determined by, and reflect, the underlying patterns in the physical world, through the process of local adaptation.

Or are they? Although it seems intuitively reasonable, even obvious, that ecosystems should be molded in this way, in recent years it has become apparent that other forces are at work, too, ones that are much less obvious or even intuitively unreasonable. The territory into which we are straying here-that of community ecology-poses what are probably the science's most important and least tractable problems. At its core there is a single, simple question: How does an ecological community come to be the way it is? One response, long popular, is that the community is the way it is because it must be so, because local conditions dictate it to be so. More important, the members of the community are considered to be tightly adapted to these conditions in concert and interact so intimately with one another-are in reality dependent upon each other-that a community of different species' composition could not survive. This is stating the case perhaps a little strongly, but it catches the essence of much of recent thinking in ecological theory. Associated with this view of the fundamental structure of ecological communities is the notion of the "balance of nature." Again, there is something intuitively reasonable about the phrase and its implications, something even reassuring. If ecosystems are the way they should be, then it follows that, when disturbed by some means, nature will work quickly to restore them. Half a decade ago, Fairfield Osborn, son of Henry Fairfield Osborn, encapsulated this sentiment in his book, Our Plundered Planet: "Nature may be a thing of beauty and is indeed a symphony, but above and below and within its own immutable essences, its distances, its apparent quietness and changelessness it is an active, purposeful, coordinated machine." The machine functions to maintain communities in their balanced state. The phrase "balance of nature" became a powerful metaphor in ecological and lay circles for a perceived fundamental natural harmony, evoking the rightness of the world as we experience it. As a result, ecology sported a distinct air of mysticism for a while. Even when that mysticism disappeared, as it did a couple of decades ago, the phrase remained. Stripped of any notion of purposefulness, the balance of nature came to refer to an ecological community's ability to resist or recover from perturbation, which was given the more objective term stability. Whether one uses "balance of nature" or "stability," there is, Stuart Pimm has noted, "something unmistakably fuzzy about the terms.' 12 How an ecological community assembles itself remains a major question, as do the behavior and characteristics of the community once it is formed. These issues are immensely complex, because they involve many variables (that is, individual species) that may interact in many different ways-and all of this is set in an often tumultuous physical environment. As a result, the range of potential patterns is immense, even infinite, and trying to understand why certain patterns emerge rather than others is a daunting challenge. This may sound all very academic, and in a sense there is a strong urge to understand how nature works and what was the source of the biodiversity of which we are a part. But that understanding is also vital to our desire to protect biodiversity, to conserve nature's infinite variety. For instance, in addition to being able to account for the different species' compositions of communities, we need to understand what makes species' populations fluctuate within communities. We need to understand what makes one community vulnerable to perturbation, particularly human-imposed perturbation, while another is resistant. We need to know why some communities recover quickly from disruption while others do so slowly. We need to know why alien species can invade some communities easily, but not others, and the consequences of such invasions. We need to understand which species are vulnerable to extinction and which play such vital roles in their community that their extinction provokes a cascade of further extinctions. Each of these questions finds a place of equal importance in an academic ecologist's text and a conservationist's handbook. This chapter begins with an exploration of the recently discovered, and counter-intuitive, insight into the reasons that species' populations fluctuate the way they do; that is, sometimes regularly and sometimes wildly and erratically. This brings us face to face with the realization that much of the time nature is not in balance at all, but instead is chaotic. For some, this may be too disconcerting an image of nature to accept, as it seems stripped of any fundamental harmony. Then I will discuss the ways ecologists have come to understand something of the processes by which communities form; for the most part, researchers are forced to use powerful computer-generated models of ecosystems to do this. Here we will see that ecological communities seem to have a mind of their own, because they improve themselves over time, becoming ever more resistant to invasion by alien species. I will talk about the dynamics of species invasions-about what makes a successful invader and what determines the impact of such events. This is an important issue for conservation. Lastly, I'll give a cautionary tale about how conservationists should maintain the stability of ecosystems-that is, by allowing them to change. Again, this is counter-intuitive, as is much that is being learned about community ecology. Nature is not all she seems.

In the Smithsonian Institution's National Museum of Natural History, in Washington, there's an exhibit that never fails to make onlookers' flesh creep: every surface of a mock-up kitchen swarms with cockroaches, hundreds of thousands, perhaps millions, of them. Those spectators who manage to go beyond the immediate visceral revulsion and read the exhibit's caption learn that the entomological horde before them is the descendent family of a single female throughout her lifetime-in theory, at least. Luckily, such potential fecundity is rarely realized. As Darwin observed in The Origin of Species, most organisms have the potential to leave more offspring than actually survive. Something keeps this potential in check. (Those not fond of cockroaches should be grateful.) That "something" includes limited food resources, competitors, predators, climatic insults, disease, and other agencies. Although over a long period of time the average number of individuals in a population may be relatively steady, in the short term these numbers bounce around the average. Some of these fluctuations are gentle, some are dramatic, even wild, with population explosions and crashes, or booms and busts, as ecologists call them.

A key challenge to understanding the dynamics of ecological communities in the short term-that is, one or a few decades-is to gain an insight into precisely what drives the fluctuation of individual species' populations within them. As Robert May has pointed out, "Such an understanding is not only fundamentally important, but it also has practical applications in trying to predict the likely effects of natural or man-made changes such as occur when a population is harvested or when climate patterns alter."3

Under the rubric of the balance of nature, population fluctuations are readily explained, in principle at least, if not in detail. Species' populations and the communities of which they are a part are viewed as being at or close to equilibrium. Left unperturbed by external interference, the interactions between the plants, herbivores, and carnivores that constitute a community reach a steady state-equilibrium-with species' populations in careful balance. It is Fairfield Osborn's "coordinated machine" humming smoothly. The limited food resources, the competitive interactions, the toll of predators, and even the effects of disease are all part of the workings of this coordinated machine. Once a community of species reaches equilibrium, the major force for disturbing the balance is climate, either in long-term shifts or sudden, forceful episodes, such as storms and temperature changes. Climatic shifts will favor some species and be detrimental to others. If, for instance, a storm decimates the population of a certain plant species, the herbivores that depend on it as a source of food will also suffer; and in their turn, predators that prey on those species will find their food resources reduced, causing a population collapse. At the same time, other prey species of these predators may have a better chance of survival, leading to a rise in their population numbers. A single storm can therefore cause the populations of some species to boom while those of others bust.

Equilibrium is temporarily lost as different species' populations fluctuate. Eventually, however, after populations swing back and forth across the point of balance, equilibrium will be restored-until another external perturbation intrudes. Because ecological communities are rarely blessed with long periods of freedom from external buffeting, populations fluctuate for much of the time. The classic example in the ecological literature is the recent history of the Canadian lynx. For more than two centuries, between 1735 and 1940, the lynx was trapped for its fur, and the record of pelts recovered by trading companies provided ecologists with an unprecedented set of data for studying the species' population history. A strong pattern is easily discernible in the data, showing that the lynx population went through episodes of dramatic booms and busts. For instance, between 1830 and 1910, the population peaked every nine or ten years, after which it collapsed rapidly. The pattern repeated itself with some regularity, but the population peaks varied considerably, between about ten thousand and sixty thousand individuals. When ecologists first analyzed this history, they assumed that the pattern emerged from a predator@prey interaction, between the lynx and the snowshoe hare, its principal prey. When a predator has culled the prey population to low levels, its food resource is dramatically diminished, and the predator population falls, too. With predation pressure now reduced, the prey population can recover; the predator population follows in its trail. This regular oscillation is how it looked with the Canadian lynx and the snowshoe hare. It turned out not to be so simple. The hare population gyrated in size because of fluctuations in its food supply, and the lynx population appeared to be tracking it. This scenario has a logic to it, and it made the chain of interaction longer.

The lynx is the classic example of a species with regularly oscillating numbers, as shown here. It was once believed that lynxes were partners in a dynamically unstable association with their main prey, the snowshoe hare. Recently it has been recognized that the cycle is driven instead by the interaction between hares and their food plants, with the lynxes being carried along more or less passively by changes in the abundance of hares.

But even so, the pattern was not perfectly regular, and even looked erratic in parts. In fact, this combination of some regularity and some apparent randomness is typical of population fluctuations of many species. Insect plagues follow this pattern; so, too, do the population eruptions of sea urchins in the northwestern Atlantic, and of Dungeness crabs in the Pacific Northwest. Wherever you look at ecological communities, not only do you see populations fluctuate, but you also see apparent randomness in these fluctuations. From marine plankton to elephants, from moths to mice, the picture is the same. How is this explained from the point of view of the balance of nature and population equilibrium? The answer is simple: everything that is seen in the history of populations-whether it is regularity or apparent randomness or any combination of these-is said to be a direct result of external forces, such as changes in climate. The fact that population history may at times be unpredictable reflects the complexity of those perturbations. Or so it was argued. Beginning about two decades ago, this interpretation began to be questioned. Perhaps something about the internal dynamics of ecological communities themselves generates these patterns, some ecologists speculated. Perhaps the apparent randomness isn't random after all, but instead is an aspect of the phenomenon known as chaos. When most people hear a system described as chaotic they immediately infer that it is random, that there is no simple, analyzable underpinning to its erratic behavior. But anyone who has read James Gleick's book Chaos knows that mathematicians have recently identified systems that, while erratic and unpredictable, are not random. The behavior of such systems can often be described simply, in terms of mathematical equations. The paradox is that, although it is governed by mathematical rules, the system's behavior can be immensely complex and virtually impossible to predict. This, in fact, is a rough and ready definition of what mathematicians call deterministic chaos. Such systems have now been identified and analyzed in many physical systems, such as weather patterns and turbulence in fluid flow. Few people realize, however, that population fluctuation in ecological communities was among the first phenomena to be studied as potential sources of chaotic behavior. That was done twenty years ago, by Robert May, and described in a classic paper in Nature.

The diagram shows the computer-simulated history of a population of Dungeness crabs. Notice that from time to time the population size fluctuates wildly, even though no external trigger is involved. This is one illustration of the unexpected and unpredictable property of chaos in living systems. (Courtesy of Alan Hastings and Kevin Higins)

Biologists have been slow to venture down the path that May identified, partly because of the strong adherence to the notion of the balance of nature and populations at equilibrium, and partly because biological systems of this kind are far more complex and difficult to analyze than any physical system. As May once wrote, "To some ecologists [chaos] has an air of black magic."' Obsessed as they were with the notion of equilibrium, ecologists continued to look for evidence in its support, while routinely ignoring erratic behavior that implied something else was going on. In the past year or two, however, long-sought evidence of true chaotic behavior in ecological communities has been discovered, in field experiments and in theoretical models. We are now forced to take a very different view of the world of nature and what shapes the patterns we see and experience. It is deeply counter-intuitive, and therefore difficult to accept.

In the mid-1980s, David Tilman, an ecologist at the University of Minnesota, asked the question: How would different levels of nitrogen in the soil affect the growth of pant-creeper, an American wild grass? He wasn't looking for chaotic phenomena when he designed the experiment, but was open-minded enough to recognize them when he saw them. At low levels of soil nitrogen, the amount of growth was steady over a period of five years, no matter whether seeds were scattered thinly or thickly. At high nitrogen levels, however, a very different picture emerged, which incorporated the classic signal of chaotic behavior: erratic and unpredictable boom and bust. At one point, the grass population crashed to 1/6000 of its previous level, sinking almost into oblivion. A description of what actually happened in the field appears mundane. The high level of soil nitrogen spurred rapid and luxurious growth; this died over winter, producing thick leaf litter that impaired growth the following spring; hence the boom and bust pattern, with varying intensities. More moderate growth, fueled by moderate levels of nitrogen, would produce a steadier population history. When Tilman published these results at the end of 1991, he provoked a mixture of consternation and excitement. This is an area of ecology where theoretical work is strong, not least because experiments of the nature that Tilman conducted are not easy to design and implement. As I said earlier, experimental manipulation of ecosystems is notoriously difficult. The year Tilman reported his findings, Robert May and two colleagues announced results of a mathematical model (of a parasitoid and its host) that behaved very much like Tilman's system. It showed erratic population swings over a period of many "generations" as a result of interaction between the "species," with no external perturbation. The model described the species and their interaction in mathematical equations. The complex dynamics of the system flowed from within it; they were not imposed from without. Equally important here is that apparently erratic, unpredictable behavior resulted from mathematically simple relationships; that's the signature of true chaos. More recently, Alan Hastings and Kevin Higgins, of the University of California at Davis, saw something similar in their model of Dungeness crab populations. Again employing simple mathematical equations, the researchers described the species and its behavior along a stretch of theoretical coastline. And again the behavior was erratic and unpredictable, with periods of stability and periods of wild population fluctuations. "Population eruptions may be an underlying feature of the dynamics without any change in physical or biological condition," they noted in their 5 report in Science early in 1994. The demonstration that the size of a population may vary dramatically and unpredictably as a result of interactions within the system, and in the absence of external change, was a major step in understanding the patterns we see in the world of nature. "The concept of chaos is both exhilarating and a bit threatening," noted William Schaffer and Mark Kot, University of Arizona ecologists who have done much to promote the understanding of chaos in ecosystems. "On the one hand, it offers a deterministic alternative to the idea that population fluctuations are solely the consequence of external perturbations. At the same time, chaos could undermine the conceptual framework of contemporary ecology."' The concepts of ecology are surely going to be battered by these insights, but from the point of view of biodiversity, chaos is a positive force. As I explained, stable communities can become dominated by one or a few species. In contrast, population fluctuations can drive communities to higher levels of species' diversity. We can see, therefore, that the erratic behavior that flows from the internal dynamics of ecological communities is a force for promoting diversity. As surprising as this discovery was, chaos apparently holds more surprises for ecologists. I started the chapter by describing the patchwork nature of many environments, with the patches being similar but distinct ecological communities. The conventional explanation for the differences is that they reflect small but important variations in the physical conditions of the environment. Chaos offers another explanation. In their modelling of parasitoid and host pairs, Robert May and his colleagues discovered that not only do populations fluctuate in size through time, but their distribution through space may be unequal-or patchy -too. Working with models of three species or more distributed over a theoretical, uniform landscape, they found that the dynamics of interaction often kept species separate from each other. "At its most extreme, this . . . [generated] small, relatively static 'islands' within the habitat, giving the appearance of isolated pockets of favorable habitat," they observed in their paper in Nature, published in the summer of 1994. Variation in the distribution of species across habitats (like the trees in the woods and the flowers in the meadows) is common and has been explained by differences in competitive and dispersal abilities, as part of the response to patterns of physical conditions in the habitat. In the counter-intuitive perspective of chaos theory, this no longer holds as a complete answer. The patchiness we see in the world can flow from the internal dynamics of the ecosystem, even when the terrain on which the different communities live is exactly the same. We can now see, therefore, that the world of nature is not in equilibrium; it is not a "coordinated machine" striving for balance. It is a more interesting place than that. There is no denying that adaptation to local physical conditions and such external forces as climatic events helps shape the world we see. But it is also apparent now that much of the pattern we recognize-both in time and space-emerges from nature herself. This is a thrilling insight, even if it means that the work of conservation managers is made more difficult. It was once believed that population numbers could be controlled by managing external conditions (as far as is possible). This must now be recognized as no longer the feasible option it was imagined to be. It is far better to understand and accept the world of nature in its infinite variety and its infinitely complex processes, acknowledging the near futility of attempts to control them, than to imagine through ignorance that it is possible to do so.

In 1789 the Reverend Gilbert White, a clergyman in the south of England, published his wonderful little book, The Natural History of Selborne. The fourth most reprinted book in the English language, White's modest tome is a collection of keen observations of nature in and around the village where he lived and ministered. Interest in the behavior of individual species goes back at least to the time of Aristotle, but the Reverend White was the first to focus on species as components of communities. He didn't use the language of modern community ecology, of course; he didn't speak of assembly rules, food webs, and trophic levels. But his recognition of the interactions between species, with their different levels of dependence on one another, is central to modern community ecology's most challenging question: What governs the way an ecological community comes to be the way it is? A crude dichotomy offers itself as a response here: Is it by design or by chance? More specifically, we can ask: Is there something special about the species in a particular community, such that it and it alone is the optimum assembly of species for that habitat? In other words, what kind of order underlies Darwin's "entangled bank"? There is no easy way to answer this by looking at the real world, because the scale-temporal and spatial-of ecosystems defies easy analysis. For this reason, much of the ground-breaking work in community ecology goes on inside computers, where researchers manipulate experimental ecosystems. Such systems are simple compared with the natural world, but in recent years powerful messages have flowed from them, and, as with the insights from chaos, they are distinctly counter-intuitive. A decade ago, for instance, Stuart Pimm and Mac Post, at the University of Tennessee, set out to assemble one such ecosystem, adding one species at a time (plants, herbivores, and carnivores). Each species was described mathematically, with a suite of behaviors that included its type, its size, and its requirements in terms of territorial range and food resources. Pimm and Post were essentially doing in a computer what occurs in nature when virgin territory is colonized, such as after a fire or on a volcanic island. An ecological community is slowly assembled, in a process called succession, beginning with the simplest of organisms that can thrive in an impoverished habitat and gradually including more species that may depend on those already present. You can't have a herbivore until you have established plant species, for instance; and you can't have predators until prey species are thriving. In the computer model, species were randomly added to the community. There was no attempt to build a particular community; rather, the idea was to allow a community to develop as it would. And, just as in nature, plant species had to precede herbivores, and herbivores predators. The dynamics of building the community were quite striking. At first, species could be added easily (provided they were ecologically logical). But as the community grew in size (that is, the total number of species), new species found it more difficult to become part of the community.

By the time the ecosystem had about twelve species, invasion was quite difficult and, if successful, often caused the loss of one or more previously established species. This was reminiscent of Darwin's wedge analogy, in which species were viewed as being tightly packed, so that the driving in of a new wedge pushed out an existing wedge. In ecological terms, what Pimm and Post were seeing was the success of alien species invading an existing community, and the effects on the community. Success was easy in species-poor communities and difficult in species-rich communities. The British ecologist Charles Elton had suggested more than three decades ago that this pattern would be seen in nature. Why should this be so? A traditional explanation is that as the number of species in the community increases, the ecological niches become filled. A potential invader seeking a niche that is already filled will have more difficulty in establishing itself than one whose niche is empty. In the first case, the invader would seem to have to outcompete the established species in order to become part of the community; in the second case, the invader faces no such competition. This seems ecologically logical, but apparently it is wrong. The challenge for the potential invader is not with the established species occupying its preferred niche, but with the community as a whole. This was very clear in a computer model of communities constructed by Ted Case, of the University of California at San Diego. He constructed several different computer communities, and manipulated the degree of interaction among the species in each of them; in some, interaction was strong, while in others it was weak. "Communities of many strongly interacting species limit the invasion possibilities of most species," he wrote of the work. "These communities, even for a superior invading competitor, set up a sort of 'activation barrier' that repels competitors when they invade at low numbers."' If the niche hypothesis was correct, then a potential invader that is competitively superior to an established species would be expected to succeed. And yet it doesn't. Communities with strongly interacting species are less vulnerable to invasion by alien species than those with weakly interacting species, even when the would-be invader is a superior competitor. "These models point to community-level rather than invader properties as the strongest determinant of differences in invader success rates," Case concluded.' If correct, this result is extremely important not only for a deeper understanding of the dynamics of ecosystems, but also in the realm of conservation. Frequently, conservation managers are faced with trying to preserve a species in a community that is competitively inferior to an invading exotic species. From Case's models, it is evident that the inferior competitor has the best chance of surviving if the community of which it is a part is species-rich; that is, intact and undisturbed. Preventing disturbance of the community as a whole therefore offers security for its weakest members, by creating a protective network within the community. The phrase "protective network" has an unmistakably mystical ring to it, so we have to understand from where it derives. The answer is food webs, which have been described as "the roadmaps through Darwin's famous entangled bank . . . [which show] how a community is put together and how it works."10 The maps reveal the interactions between species in the community, such as who eats whom. Biologists have long been fascinated by these maps. Often, the food-web patterns look bewilderingly complex, and initially biologists believed that each community had a unique food web. As they cut through the superficial complexities, however, biologists came to realize that all food webs are very similar, no matter what kind of community they represent, and display the same few common properties. These include the length of food chains-that is, a description of who consumes whom in the community-and the ratio of predator species to prey species. Wherever you look in nature, similar patterns exist. The fact that such commonalities among disparate communities are to be seen even where there is the potential for limitless variety suggests something fundamental about the underlying order in nature. That order seems to emerge from the internal dynamics of the system itself, and is not imposed by external circumstances. The interactions between species that Ted Case manipulated in his computer model represent food webs of the real world. The protective network that he saw in the communities with strong interactions can therefore be explained as a property of the underlying food-web pattern. No mystical force need be adduced to explain the counter-intuitive observation he made.

The ecosystems that assembled themselves in Stuart Pimm and Mac Post's computer models displayed networks of interactions between species that closely resemble food-web patterns in the real world. This promoted confidence that, although they were simple, the models were realistic. But it also led to further insights. The first result, remember, was that species-poor communities are easily invaded, whereas with species-rich communities successful invasion is more difficult. Difficult, but not impossible. If a species-rich community is allowed to mature, it does not remain static but instead experiences a slow turnover of species. In other words, some invasions are successful, usually propelling the loss of existing species; community composition is dynamic, not static. A successful species may, however, become a victim of a later invasion and get pushed out of the community. But its temporary presence leaves a mark on the community, like a footprint in the sand. The second result from Pimm and Post's work, therefore, was that mature, species-rich communities are much more difficult to invade than newly established ones. There seems to be something in the maturation process that strengthens the emerging protective network within the community. The community seems to be improving itself, almost purposefully getting better in a way that is difficult to define. This result is by no means the whimsy of an esoteric computer model, because precisely the same thing happens in nature. In Hawaii, for instance, two types of ecosystem exist. The first is the highland forests, which have been untouched by human interference. They therefore represent a mature, species-rich ecosystem. The second is the lowland forests, which have been disturbed through human activity in and around them. In the process of recovering from such disturbance, they are at an immature stage of assembly, even though they are rich in species. Because the islands have been colonized many times since the first Polynesian settlers arrived fifteen hundred years ago, many alien species came along, too, either as deliberate or accidental passengers. More species of birds and plants, for instance, have been introduced to Hawaii than anywhere else in the world. Twenty-eight percent of the archipelago's insects and 65 percent of its plants are non-native. All its mammals are recent arrivals. Three decades ago, in The Ecology of Invasions by Animals and Plants, Charles Elton described the situation there as "one of the great historical convolutions of the world's fauna and flora." Each time an alien species succeeded in establishing itself, it triggered reductions in the population size of native species or it pushed them, cascades of them, to extinction. The issue here is: Where did the alien species become established? Was it in the immature ecosystems of the lowlands or the mature ecosystems of the highlands? Overwhelmingly, the answer is the former. The mature ecosystems were evidently better able to resist invasion than the immature ones. In the terms of ecological theory, the mature ecosystems had reached a persistent state. Mature communities-whether in the real world or inside computers-clearly have important ecological properties that are denied to immature ones. And the obvious inference is that during the assembly process there is a selection for species that are superior in some ways. Superior, for instance, in productivity, in the case of plant species; in speed or stealth, in the case of predator species; and so on. Self-evidently, a community of superior species will be superior ecologically to one made up of inferior species. However, when Pimm and Post examined the behavioral characteristics of the species in the persistent communities in their computer model, they could find no indication of superiority. In ecological terms, these species were no different from ones that had failed to become part of the community. Perhaps they weren't looking at the right parameters, they speculated. As it turned out, they had not made a mistake, which became clear when another ecologist, Jim Drake, then of Purdue University, performed similar computer modelling. Like Pimm and Post, Drake promoted assembly of an ecological community by randomly adding species, one by one. But he did it by drawing from a finite pool of species, 125 in all. If a species failed to penetrate on one occasion, it was available for a later attempt. Again, a persistent community matured with about a dozen species. Then Drake started again, using the same pool of species, and again a persistent community matured with about a dozen species. But it was a different community. Less than half the species in the second community were in common with the first. He repeated the process dozens of times, each time getting a mature persistent community, each one different in composition from the others. Again, none of the species in the communities was identifiably "better" in any respect than the others in the pool. Any species could become a member of a persistent community, if it was added at the right time. These results are as fascinating as they are important. First of all, we can see that persistent communities can form through a process of random addition of species. Second, the ecologically crucial property of persistence, or stability, emerges from the interactions of the species in the community, not through superior qualities of those species. As significant, and perhaps more so, are the implications of these observations on the patchiness of nature. We saw earlier that, traditionally, differences among neighboring ecosystems are explained as a response to differing physical conditions. We saw, too, that chaos theory leads us to expect patchiness, even in the absence of physical differences in the environment. And now, with the work of Pimm, Post, and Drake, we have a second source of patchiness that also does not include-adaptation to local conditions: history. The final composition of a persistent ecosystem clearly depends on the order in which species attempt to become part of that system as it matures. Sometimes arriving on the scene early will confer an advantage; sometimes it is better to be late in the day. It all depends on which species are already part of the community. We saw in an earlier chapter that history, or contingency, is becoming recognized as a powerful agency in shaping the path of evolution, while adaptation is seen as having a lesser role than was once assumed. Here we have an analogous situation, with history a powerful agent in shaping the evolution of ecosystems, while adaptation plays a lesser role. That's a very different way of looking at nature from the traditional view. If it's true of nature, that is. Jim Drake has put it to the test experimentally, by doing with microorganisms (mostly algae of various types) what he does with computer species. He adds the species randomly, and obtains many different persistent communities. History matters. just recently, two paleontologists produced a fossil-record perspective on this phenomenon. Martin Buzas and Stephen Culver, of the Smithsonian Institution and the Natural History Museum of London, looked at the composition of near-shore marine communities on the North American Atlantic coastal plain over a period of fifty-five million years, during which time the sea level dropped and rose six times. Six times new communities formed in the near-shore habitat, drawn from the pool of species in the region. And six times the composition of the communities was different. Commenting on the observations, the Smithsonian Institution ecologist Jeremy Jackson stated, "This is surely the death knell for the concept of tightly integrated marine ecological communities." 1 1 It is, and it is also a vote for the importance of historical contingency. If these results seem counter-intuitive, one further observation made by Jim Drake is doubly so. Effectively he said to himself the following: "The persistent communities I make in my computer model clearly work very well. I'll therefore take one of these communities and try to rebuild it from scratch, using only the dozen or so species that make up the community." He could not do it. Once he took the community apart, he couldn't put it back together again, no matter in what order he added the species. Stuart Pimm calls this the Humpty Dumpty effect, for good reason. The explanation is somewhat esoteric mathematically, but it boils down to saying that in order to reach the persistent state, Z, the ecosystem has to pass through states A through Y. You can't jump to Z in one step. If this sounds like the stuff of late-night bar talk at ecology conferences, it's not. These days there is growing interest in restoring ecosystems that have been degraded or destroyed. The prairies of the Midwest and the Florida Everglades are two such examples. In these and other cases, ecologists often know the species composition of the pristine communities, from historical documentation. Until the work I've just described had been done (and it is still being refined), ecologists' inclination was simply to gather the requisite species for the ecosystem they were planning to restore, and then let them loose in the chosen habitat. They were puzzled when they repeatedly discovered it didn't work. Now we know why.

We have seen that nature is not all she seems. There is a dynamic within ecological communities that is counter-intuitive and therefore was unsuspected. Communities are always changing, apparently purposefully improving themselves, but we now know that chance and history play a large part. I will finish this chapter with a story of a real ecosystem that reveals these dynamics, shows the importance of change through time, and is a salutary tale for would-be conservationists.

The Chobe National Park in northern Botswana is typical of several ecosystems in southern and eastern Africa. There are many large herbivores, some of them migratory, including giraffe, buffalo, elephant, zebra, wildebeest, and impala. Lions, hyenas, wild dogs, and jackals form a rich carnivore guild. A mosaic habitat of grassland and acacia woodland harbors a rich array of bird and insect species. Altogether, the park offers an abundant diversity of species of the sort that people think of when they hear the word wildlife. Managers of the park would like to maintain this diversity, because it is attractive to tourists and because it is perceived of as being the way it should be. They are, however, facing a severe challenge: the acacia woodlands are being destroyed, principally by elephants, and no new trees are growing. If the woodlands shrink to mere remnants of their present selves, the managers believe they will have failed, because they want to keep things as they are. To do so would, however, not only be wrong ecologically; it is also probably impossible. A look at the ecological history of the park reveals why.

Most environments undergo cycles of change, driven by internal and external forces, as seen here in the history of Chobe National Park, Botswana. (Courtesy of Brian Walker)

The Savuti channel is the major source of surface water in the area. When full, it flows from Angola via the Linyanti swamps and empties into the Savuti marsh (which currently is grassland). It was full in the late 1800s, dried up around the turn of the century, and remained dry until the mid-1950s. In 1982 it dried up again, and remains so. Soon after the channel dried up in the early years of this century, there was a massive outbreak of rinderpest in the area. These two events played midwife to the current acacia woodlands, as follows. The lack of water encouraged the elephants to seek water elsewhere (hunting also reduced their numbers). And the rinderpest epidemic devastated the ungulate population. As a result, browsing pressure in the area was suddenly very light, which allowed acacia seedlings (a favorite food of many browsers) to mature into trees. By the time the elephants and ungulates returned, extensive acacia woodlands had been established. "What we observe today, the coexistence of lots of elephants and extensive Acacia woodlands, represents a very narrow window in time and is apparently not sustainable," observes Brian Walker, who has made a detailed study of the region.'2 It is not sustainable because as long as there are healthy populations of elephants and ungulates in the area, no acacia seedlings will survive to maturity. The animals would have to be removed if the woodlands are to thrive again. "The question," states Walker, "is whether managers and tourists are prepared to accept a ten to fifteen-year period with virtually no animals to see." Probably not. The current species diversity of the park is natural, of course, but it is generated by substantial environmental change that took place over many decades. And change is what park managers often resist; at least they do when they see something of value apparently disappearing. Ecosystems are in a constant state of turmoil, both in space and time, and at any point some populations will be in decline while others may be booming. And constant change is vital as an engine of species diversity. "Conservationists should spend less time worrying about the persistence of particular plant or animal species," warns Walker, "and begin to think instead about maintaining the nature and diversity of ecosystem processes." 1 3 Armed with the perspective we've gained about the nature of ecosystems, from an understanding of chaos and the dynamics of the assembly of communities, we can see that what Walker exhorts us to do is sound. But, as with all of human affairs, it is very difficult to manage processes that take many decades to occur. And no one likes to stand idle and watch woodlands shrink or animals die of hunger or thirst. Ultimately, however, that may be what we shall have to do.