The Lying Stones of Marrakech

by Stephen Jay Gould

  Nonetheless, new cases of such short-term evolution still maintain enormous and surprising power to attract public attention—for interesting and instructive but utterly invalid reasons, as I shall show. I devote this essay to the three most prominent examples of recent publications that received widespread attention in the popular press as well. (One derives from my own research, so at least I can’t be accused of sour grapes in the debunking that will follow—though I trust that readers will also grasp the highly positive twist that I will ultimately impose upon my criticisms.) I shall briefly describe each case, then present my two general critiques of their prominent reporting by the popular press, and finally explain why such cases teach us so little about evolution in the large, yet remain so important for themselves, and at their own equally legitimate scale.

  1. Guppies from Trinidad. In many drainage systems on the island of Trinidad, populations of guppies live in downstream pools, where several species of fish can feed upon them. “Some of these species prey preferentially on large, mature-size classes of guppies.” (I take all quotes from the primary technical article that inspired later press accounts—“Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata),” by D. N. Reznick, F. H. Shaw, F. H. Rodd, and R. G. Shaw, published in Science, 1977, volume 275, pages 1934–37). Other populations of the same species live in “upstream portions of each drainage” where most “predators are excluded … by rapids or waterfalls, yielding low-predation communities.”

  In studying both kinds of populations, Reznick and colleagues found that “guppies from high-predation sites experience significantly higher mortality rates than those from low-predation sites.” They then reared both kinds of guppies under uniform conditions in the laboratory, and found that fishes from high-predation sites in lower drainages matured earlier and at a smaller size. “They also devote more resources to each litter, produce more, smaller offspring per litter, and produce litters more frequently than guppies from low-predation localities.”

  This combination of observations from nature and the laboratory yields two important inferences. First, the differences make adaptive sense, for guppies subjected to greater predation would fare better if they could grow up fast and reproduce both copiously and quickly before the potential boom falls—a piscine equivalent of the old motto for electoral politics in Boston: vote early and vote often. On the other hand, guppies in little danger of being eaten might do better to bide their time and grow big and strong before engaging their fellows in any reproductive competition. Second, since these differences persist when both kinds of guppies are reared in identical laboratory environments, the distinction must record genetically based and inherited results of divergent evolution between the populations.

  In 1981, Reznick had transferred some guppies from high-predation downstream pools into low-predation upstream waters then devoid of guppies. These transplanted populations evolved rapidly to adopt the reproductive strategy favored by indigenous populations in neighboring upstream environments: delayed sexual maturity at larger size, and longer life. Moreover, Reznick and colleagues made the interesting observation that males evolved considerably more rapidly in this favored direction. In one experiment, males reached their full extent of change within four years, while females continued to alter after eleven years. Since the laboratory populations had shown higher heritability for these traits in males than in females, these results make good sense. (Heritability may be roughly defined as the correlation between traits in parents and offspring due to genetic differences. The greater the heritable basis of a trait, the faster the feature can evolve by natural selection.)

  This favorable set of circumstances—rapid evolution in a predictable and presumably adaptive direction based on traits known to be highly heritable—provides a “tight” case for well-documented (and sensible) evolution at scales well within the purview of human observation, a mere decade in this case. The headline for the news report on this paper in Science magazine (March 28, 1997) read: “Predator-free guppies take an evolutionary leap forward.”

  2. Lizards from the Exuma Cays, Bahama Islands. During most of my career, my fieldwork has centered on biology and paleontology of the land snail Cerion in the Bahama Islands. During these trips, I have often encountered fellow biologists devoted to other creatures. In one major program of research, Tom Schoener (a biology professor at the University of California, Davis) has, with numerous students and colleagues, been studying the biogeography and evolution of the ubiquitous little lizard Anolis—for me just a fleeting shadow running across a snail-studded ground, but for them a focus of utmost fascination (while my beloved snails, I assume, just blend into their immobile background).

  In 1977 and 1981, Schoener and colleagues transplanted groups of five or ten lizards from Staniel Cay in the Exuma chain to fourteen small and neighboring islands that housed no lizards. In 1991, they found that the lizards had thrived (or at least survived and bred) on most of these islands, and they collected samples of adult males from each experimental island with an adequate population. In addition, they gathered a larger sample of males from areas on Staniel Cay that had served as the source for original transplantation in 1977 and 1981.

  This study then benefits from general principles learned by extensive research on numerous Anolis species throughout the Bahama Islands. In particular, relatively longer limbs permit greater speed, a substantial advantage provided that preferred perching places can accommodate long-legged lizards. Trees and other “thick” perching places therefore favor the evolution of long legs. Staniel Cay itself includes a predominant forest, and the local Anolis tend to be long-legged. But when lizards must live on thin twigs in bushy vegetation, the agility provided by shorter legs (on such precarious perches) may outweigh the advantages in speed that longer legs would provide. Thus, lizards living on narrow twigs tend to be shorter-legged. The small cays that received the fourteen transported populations have little or no forest growth and tend instead to be covered with bushy vegetation (and narrow twigs).

  J. B. Losos, the principal author of the new study, therefore based an obvious prediction on these generalities. The populations had been transferred from forests with wide perches to bushy islands covered with narrow twigs. “From the kind of vegetation on the new islands,” Losos stated, “we predicted that the lizards would develop shorter hindlimbs.” Their published study validates this expected result: a clearly measurable change, in the predicted and adaptive direction, in less than twenty years. (See details in J. B. Losos, K. I. Warheit, and T. W. Schoener, “Adaptive differentiation following experimental island colonization in Anolis lizards,” Nature, 1997, volume 387, pages 70–73). A news report appeared in Science magazine (May 2, 1997) under the title “Catching lizards in the act of adapting.”

  This study lacks a crucial piece of documentation that the Trinidadian guppies provided—an absence immediately noted by friendly critics and fully acknowledged by the authors. Losos and colleagues have not studied the heritability of leg length in Anolis sagrei and therefore cannot be certain that their results record a genetic process of evolutionary change. The growth of these lizards may feature extensive flexibility in leg length, so that the same genes yield longer legs if lizards grow up on trees, and shorter legs if they always cavort in the bushes (just as the same genes can lead to a thin or fat human being depending upon a personal history of nutrition and exercise). In any case, however, a sensible and apparently adaptive change in average leg length has occurred within twenty years on several islands, whatever the cause of modification.

  3. Snails from Great Inagua, Bahama Islands. Most of Great Inagua, the second-largest Bahamian Island (Andros wins first prize), houses a large and ribby Cerion species named C. rubicundum. But fossil deposits of no great age lack this species entirely and feature instead an extinct form named Cerion excelsior, the largest of all Cerion species. Several years ago, on a mudflat in the southeastern corner of Great Inagua, David Woodruff
(of the University of California, San Diego) and I collected a remarkable series of shells that seemed to span (and quite smoothly) the entire range of form from extinct C. excelsior to modern C. rubicundum. Moreover, and in general, the more eroded and “older looking” the shell, the closer it seemed to lie to the anatomy of extinct C. excelsior.

  This situation suggested a local evolutionary transition by hybridization, as C. rubicundum, arriving on the island from an outside source, interbred with indigenous C. excelsior. Then, as C. excelsior declined toward extinction, while C. rubicundum thrived and increased, the average anatomy of the population transformed slowly and steadily in the direction of the modern form. This hypothesis sounded good and sensible, but we could devise no way to test our idea—for all the shells had been collected from a single mudflat (analogous to a single bedding plane of a geological stratum), and we could not determine their relative ages. The pure C. excelsior shells “looked” older, but such personal impressions count for less than nothing (subject as they are to a researcher’s bias) in science. So we got stymied and put the specimens in a drawer.

  Several years later, I teamed up with paleontologist and geochemist Glenn A. Goodfriend from the Carnegie Institution of Washington. He had refined a dating technique based on changes in the composition of amino acids in the shell over time. By keying these amino acid changes to radiocarbon dates for some of the shells, we could estimate the age of each shell. A plot of shell age versus position on an anatomical spectrum from extinct C. excelsior to modern C. rubicundum produced a beautiful correlation between age and anatomy: the younger the specimen, the closer to the modern anatomy.

  This ten- to twenty-thousand-year transition by hybridization exceeds the time period of the Trinidad and Exuma studies by three orders of magnitude (that is, by a factor of 1,000), but even ten thousand years represents a geological eye-blink in the fullness of evolutionary time—while this transformation in our snails marks a full change from one species to another, not just a small decrement of leg length, or a change in the timing of breeding, within a single species. (For details, see G. A. Goodfriend and S. J. Gould, “Paleontology and chronology of two evolutionary transitions by hybridization in the Bahamian land snail Cerion,” Science, 1996, volume 274, pages 1894–97). Harvard University’s press release (with no input from me) carried the headline “Snails caught in act of evolving.”

  A scanning of any year’s technical literature in evolutionary biology would yield numerous and well-documented cases of such measurable, small-scale evolutionary change—thus disproving the urban legend that evolution must always be too slow to observe in the geological microsecond of a human lifetime. These three studies, all unusually complete in their documentation and in their resolution of details, do not really rank as “news” in the journalist’s prime sense of novelty or deep surprise. Nonetheless, each of these three studies became subjects for front-page stories in either The New York Times or The Boston Globe.

  Now please don’t get me wrong. I do not belong to the cadre of rarefied academics who cringe at every journalistic story about science for fear that the work reported might become tainted with popularity thereby. And in a purely “political” sense, I certainly won’t object if major newspapers choose to feature any result of my profession as a lead story—especially, if I may be self-serving for a moment, when one of the tales reports my own work! Nonetheless, this degree of public attention for workaday results in my field (however elegantly done) does fill me with wry amusement—if only for the general reason that most of us feel a tickle in the funny bone when we note a gross imbalance between public notoriety and the true novelty or importance of an event, as when Hollywood spinmeisters manage to depict their client’s ninth marriage as the earth’s first example of true love triumphant and permanent.

  Of course I’m delighted that some ordinary, albeit particularly well done studies of small-scale evolution struck journalists as front-page news. But I still feel impelled to ask why these studies, rather than a hundred others of equal care and merit that appear in our literature every month, caught this journalistic fancy and inspired such prime attention. When I muse over this issue, I can only devise two reasons—both based on deep and interesting fallacies well worth identifying and discussing. In this sense, the miselevation of everyday good work to surprising novelty may teach us something important about public attitudes toward evolution, and toward science in general. We may, I think, resolve each of the two fallacies by contrasting the supposed meaning of these studies, as reported in public accounts, with the significance of such work as viewed by professionals in the field.

  1. The fallacy of the crucial experiment

  In high school physics classes, we all learned a heroically simplified version of scientific progress based upon a model that does work sometimes but by no means always—the experimentum crucis, or crucial experiment. Newton or Einstein? Ptolemy or Copernicus? Special Creation or Darwin? To find out, perform a single decisive experiment with a clearly measurable result replete with decisive power to decree yea or nay.

  The decision to treat a limited and particular case as front-page news must be rooted in this fallacy. Reporters must imagine that evolution can be proved by a single crucial case, so that any of these stories may provide decisive confirmation of Darwin’s truth—a matter of some importance given the urban legend that evolution, even if valid, must be invisible on human timescales.

  But two counterarguments vitiate this premise. First, as a scientific or intellectual issue, we hardly need to “prove” evolution by discovering new and elegant cases. We do not, after all, expect to encounter a page-one story with the headline “New experiment proves earth goes around sun, not vice versa. Galileo vindicated.” The fact of evolution has been equally well documented for more than a century.

  Second, and more generally, single “crucial” experiments rarely decide major issues in science—especially in natural history, where nearly all theories require data about “relative frequencies” (or percentage of occurrences), not pristine single cases. Of course, for a person who believes that evolution never occurs at all, one good case can pack enormous punch—but science resolved this basic issue more than one hundred years ago. Nearly every interesting question in evolutionary theory asks “how often?” or “how dominant in setting the pattern of life?”—not “does this phenomenon occur at all?” For example, on the most important issue of all—the role of Darwin’s own favored mechanism of natural selection—single examples of selection’s efficacy advance the argument very little. We already know, by abundant documentation and rigorous theorizing, that natural selection can and does operate in nature. We need to determine the relative strength of Darwin’s mechanism among a set of alternative modes for evolutionary change—and single cases, however elegant, cannot establish a relative frequency.

  Professionals also commit this common error of confusing well-documented single instances with statements about relative strength among plausible alternatives. For example, we would like to know how often small and isolated populations evolve differences as adaptive responses to local environments (presumably by Darwin’s mechanism of natural selection), and how often such changes occur by the random process known as “genetic drift”—a potentially important phenomenon in small populations (just as a small number of coin flips can depart radically from fifty-fifty for heads and tails, while a million flips with an honest coin cannot stray too far from this ideal). Losos’s study on lizard legs provides one vote for selection (if the change turns out to have a genetic basis)—because leg length altered in a predicted direction toward better adaptation to local environments on new islands. But even such an elegant case cannot prove the domination of natural selection in general. Losos has only shown the power of Darwin’s process in a particular example. Yet the reporter for Science magazine made this distressingly common error in concluding: “If it [change in leg length] is rooted in the genes, then the study is strong evidence that isolated popul
ations diverge by natural selection, not genetic drift as some theorists have argued.” Yes, strong evidence for these lizards on that island during those years—but not proof for the general domination of selection over drift. Single cases don’t establish generalities, so long as alternative mechanisms retain their theoretical plausibility.

  2. The paradox of the visibly irrelevant

  As a second reason for overstating the centrality of such cases in our general understanding of evolution, many commentators (and research scientists as well) ally themselves too strongly with one of the oldest (and often fallacious) traditions of Western thought: reductionism, or the assumption that laws and mechanics of the smallest constituents must explain objects and events at all scales and times. Thus, if we can render the behavior of a large body (an organism or a plant, for example) as a consequence of atoms and molecules in motion, we feel that we have developed a “deeper” or “more basic” understanding than if our explanatory principles engage only large objects themselves, and not their constituent parts.

  Reductionists assume that documenting evolution at the smallest scale of a few years and generations should provide a general model of explanation for events at all scales and times—so these cases should become a gold standard for the entire field, hence their status as front-page news. The authors of our two studies on decadal evolution certainly nurture such a hope. Reznick and colleagues end their publication on Trinidadian guppies by writing: “It is part of a growing body of evidence that the rate and patterns of change attainable through natural selection are sufficient to account for the patterns observed in the fossil record.” Losos and colleagues say much the same for their lizards: “Macroevolution may just be microevolution writ large—and, consequently, insight into the former may result from study of the latter.”