The Biosphere as Complex Adaptive System

March 9, 2001

Book Review
James Case

Fragile Dominion: Complexity and the Commons. By Simon Levin, Helix Books (a division of Perseus Books), Reading, Massachusetts, 1999, 250 pages, $27 hardcover, $17 paperback.

Simon Levin---whose PhD and earliest positions were in mathematics---is the Moffett Professor of Biology at Princeton University, as well as the founding director of the Princeton Environmental Institute and a member of the Science Board of the Santa Fe Institute. As the author or editor of more than 25 books on biology, ecology, and biodiversity,* he is a leading contributor to some of the most politically sensitive branches of science. In the book under review, Levin explains why he believes the biosphere to constitute a complex adaptive system (CAS)---in the sense of the systems studied by John H. Holland---and why he deems biodiversity to be an essential ingredient thereof.

Since bursting on the scene little more than a decade ago, Holland's brain-children (which include genetic algorithms as well as complex adaptive systems) have surged to the forefront of research in an impressive variety of sciences. Their impact on descriptive sciences like biology, ecology, and evolution---in which there were few pre-existing theories to serve as barriers to acceptance---has been particularly remarkable.

Without pausing to catalogue the indispensable services with which nature supplies humankind---a task admirably performed by older books, including Nature's Services, by Sandra Postel and Stephen Carpenter---Levin proceeds immediately to the manner in which nature literally chances to deliver those services. To understand the workings of the delivery system is to understand the ease and abruptness with which any number of crucial services could be interrupted. For instance, the "nitrogen cycle," on which all higher life forms depend, would cease forthwith if any of several participating microbe populations should crash.

Nitrogen, which constitutes some 80% of the atmosphere, is essential to the survival of all higher organisms. Yet it is useless to those organisms in its gaseous form. Bacteria must first combine it with other elements, or "fix" it, to form such compounds as ammonium and ammonia. These in turn must be oxidized by other bacteria to form the nitrates that fuel the growth of grasses, trees, and legumes. Finally, still other bacteria must break down the nitrates in decaying plants and animals, releasing gaseous nitrogen back into the atmosphere and restarting the cycle. The elimination of the microbes responsible for any one of these transformations would doom all higher forms of life. Fortunately, many different microbial species are known to participate in each transformation, so that a very major catastrophe indeed would be required eliminate them all.

In addition to the diversity of species in the "functional group" of nitrogen-fixing bacteria, a diversity of genotypes within each individual species produces a second level of redundancy in the functional group. Yet no one claims to know how diverse the group actually is. Is it so homogeneous that a single random event might doom it to extinction? Or is it so heterogeneous that many such events would be required? Do it and other key functional groups survive with comfortable margins of safety, or do some of them stand on the very brink of extinction? All that seems certain is that the level of genetic diversity within a population measures its potential to survive and adapt. So it is that (black and white) Holstein-Frisian milk cattle have survived and prospered in the age of selective breeding, while formerly competitive breeds---such as Ayrshires, Brown Swiss, Jerseys, and Guernseys---are no longer profitable. The large and diverse Holstein-Frisian gene pool has enabled the breed to prosper in circumstances under which the other breeds could not.

To illustrate the role that model building can play in assessing the contributions made by individual populations to the plant or animal communities in which they reside, Levin draws heavily on his personal experience with forest-growth models. Such models have existed for at least thirty years, and some are surprisingly elaborate. One he mentions considers the direct and reflected light incident from each of 256 directions on each tree in an entire stretch of (simulated) forest! Functional groups are particularly identifiable in recently disturbed---by windfall, fire, or clear cutting---parts of a simulated forest. First to thrive in the openings created are varieties like larch and scrub oak, which---because of their rapid respiration rates---are best able to exploit abundant sunlight. Later on, maple and hemlock, which require shade protection in infancy but grow rapidly to sunlight once their roots are established, become dominant. Eventually, they too are replaced by slower growing but longer-lived "climax" varieties, such as beech, ash, and white oak.

Models exploring the effects in question are often run for as many as a thousand "model years," to eliminate the influence of initial conditions. The resulting "mature forests" clearly show the effects of natural disturbances past, in the form of "islands" of maples, hemlocks, or other short-lived fast-growing "colonists" in a sea of climax varieties. These islands are important contributors to forest renewal and diversity. That is why fires in national forests, once extinguished with all available manpower, are often now allowed to burn: The resulting colonies have been shown to foster overall forest health. Similar colonies are plainly visible at places along the Alaskan coast where glaciers are currently receding. Whereas the most recently uncovered zones consist of bare rock, previously vacated zones host lichens---mixtures of fungi and (photo-synthesizing) mosses---which are later replaced by short-lived grasses, shrubs, and larches, before long-lived conifers assert dominance.

Levin's strategy is to persuade readers that seemingly robust populations---perhaps including our own---can crash abruptly; he does so by explaining how such events can happen. The mechanism is particularly simple in marine communities located in the intertidal zones of the Pacific Northwest, which are among the most studied and best understood of all such communities. The local bully is the large mussel Mytilus Californianus, which can out-compete any of the other native species that attach to rock. Its only natural enemy is the multicolored sea star Pisaster ochraceous.

High up on the rock, where the surface is only intermittently wet, and where the hot sun would result in baked starfish, the mussel dominates completely. But a little lower down, in the starfish comfort zone, the mussel density is much reduced, leaving plenty of room for barnacles and other sea creatures to thrive. Levin describes an actual experiment confirming that, in the absence of starfish, the mussel becomes dominant on all parts of the rock surface, save those that are permanently submerged. Because its extinction would bring about the extinction of numerous other species, the multicolored starfish is known as a "keystone," or "trigger," species.


In his final chapter, Levin proposes eight "commandments of environmental management," of which the first is to "reduce uncertainty" by exploring (and funding) the earth sciences. It irks and mystifies him that national leaders seem willing to spend more to search for life on other planets than to explore the endless variety of life forms found on this one. The majority of earthly species have yet to be counted, much less studied in detail, either individually or in relation to one another. Knowledgeable estimates range from a few million to as many as 100 million earthly species in all, with a general consensus of 10-20 million.

What sense does it make to eradicate species in an age in which, according to Levin, the active ingredients of most prescription medicines---many of them still unsynthesized---were first obtained from plants and animals native to the tropical rainforests? Why lose them now, just as techniques are becoming available for splicing (say) a tarantula gene into the DNA of a wheat or cotton plant---thus making those valuable crops less appetizing to pests? Even though genetically engineered crops are not yet particularly successful, it seems foolish to shrink the existing pool of potentially useful genes at a time when the means for exploiting them are only beginning to be developed. Why not preserve and study this potential treasure trove (which took about four billion years to create) until science can provide at least a preliminary estimate of its potential value to mankind?

Levin's second commandment is to "expect surprise" by stockpiling contingency plans, along with the funds to carry them out. Even when science has far exceeded its current limits, he expects that serious threats will elude prediction!

The third, fourth, and fifth commandments form the core of Levin's management program. They concern heterogeneity, modularity, and redundancy, which have been consistently shown by CAS models to encourage robustness and system sustainability. As heterogeneity in the species of nitrogen-fixing bacteria decreases the probability that a single cataclysmic event---perhaps a meteor impact, an atomic war, or an abrupt change in global temperatures---will interrupt a process as essential as the nitrogen cycle, so heterogeneity in the local crop mix decreases the probability that a single blight or pest will wipe out the harvest in any one region, much less spread beyond it. Heterogeneity in tree varieties, breeds of pets, and farm animal populations is equally prudent, according to Levin.

Modularity and compartmentalization retard or even prevent the spread of blight and disease. Moreover, the evolutionary process, like the more successful models thereof, tends to respect the barriers between nature's modules and compartments, often permitting distinct populations (be they animal species or human cultures) to evolve independently on opposite sides of such seemingly inconsequential barriers as fordable rivers, penetrable forests, and low-lying ranges of hills. Yet the current rush toward globalization is tearing down bar-riers against such scourges as HIV and the ebola and mad cow viruses. Levin warns that the eventual collateral damage could far exceed the short-term pecuniary gain.

Redundancy is another potent weapon against the sudden collapse of an essential life-support system like the nitrogen cycle. As mentioned earlier, the nitrogen-fixing bacteria contain two levels of redundancy, given the participation of many species in the activity and the genetic diversity that appears to exist within those species. Yet the existing catalogue of essential services is far from complete, and may well contain items provided by only a few homogeneous species. Does it not seem unwise, Levin asks, without even exploring the number and function of such species, to risk losing them in the current stampede to privatize and globalize every aspect of the human condition?
Whereas Levin's seventh and eighth commandments concern the building of trust and his faith in the "golden rule," both of which are self-explanatory, the sixth has to do with feedback loops, and requires some explanation.

Perhaps the clearest lesson yet to emerge from CAS analysis of evolving interactions between organisms is that "tight" reward-and-punishment loops---those that cause agents to feel, without delay, both the fortunate and the unfortunate consequences of their actions---are essential for any adaptive change. That is so because, although the agents that inhabit a CAS are only computer programs, they share the human tendency to ignore the small, remote, and uncertain consequences of their actions while overreacting to the more immediate ones.

Unfortunately, the loops that govern the utilization of scarce common resources are naturally loose. Though real, the rewards for individual restraint in the use of such resources are---along with the punishments for profligate use---too small, remote, and uncertain to weigh heavily at the moment of decision. To Levin, the conclusion is obvious: In order to discharge their constitutional duty to further the common good, governments and legislatures must find the will to tighten artificially the feedback loops through which the already realized costs and benefits to society of individual action shape future individual actions.

Levin does not address the fact that national governments and legislatures have recently ceded, to the World Trade Organization, their constitutional right to tighten what they consider to be virtuous feedback loops. In fact, WTO agreements have rolled back years of international agreements and domestic legislation meant to protect diversity and the environment. Even measures agreed upon as recently as the 1992 Rio summit have been weakened or repealed. No law or regulation aimed at preserving (currently) noncommercial species or their habitats is safe, because any such instrument can be voided overnight by a WTO tribunal persuaded that it will reduce corporate profit.

Until recently, agendas like the one enunciated by Levin were going nowhere against the forces of globalization. The 1999 WTO meeting in Seattle changed that to some extent, but time alone will tell how much and for how long. Even Levin's seemingly innocuous plea to reduce uncertainty via research in the earth sciences faces powerful opposition in Congress. Some there still argue---as Newt Gingrich occasionally did---that national needs would be amply served by corporate research alone.

All in all, Fragile Dominion is well worth reading. It describes the empirical foundations of ecological science in somewhat greater detail than do other invitations to CAS modeling. It does so, moreover, without getting bogged down in unwelcome detail. For Levin, computer models are just one more way to study the plant and animal communities he yearns to comprehend. They permit him to watch the unfolding of dramas that, in nature, would play out over centuries. And by so doing, the models permit him to fashion informed opinions on questions of policy unclarified by conventional methods.

James Case writes from Baltimore, Maryland.

*The list includes Ecosystem Analysis and Prediction, the 1975 book edited by Levin for SIAM/SIMS (the SIAM Institute for Mathematics and Society).

Renew SIAM · Contact Us · Site Map · Join SIAM · My Account
Facebook Twitter Youtube