American Scientist

Stephen Jay Gould 1941-2002

Stephen Jay Gould died on May 20, 2002, after an intense final battle with cancer. He was an extraordinary figure in paleontology and evolutionary biology. He initiated and shaped some of the most heated and productive debates of the second half of the 20th century, helping to return paleontology—and the investigation of evolutionary pattern and process over large temporal scales in general—to the main arena of evolutionary thinking. With his delight in synthesizing across disciplines and his determination to explore linkages and discontinuities across levels in the biological hierarchy, not to mention his muscular prose and astonishing productivity, he made an enormous contribution to the vitality and conceptual richness of the field. At the same time, of course, he was the field's most eloquent and prolific public spokesman. Through his Natural History columns and other nontechnical writings, he helped a vast readership to see the fascination, breadth and power of evolutionary theory, and to appreciate the human enterprise behind that theory—figures great and small, ideas brilliant and lunatic, across the entire history of the discipline. We will miss him.

A small but very real consolation is that Gould published two new books in the last few months. One is I Have Landed, the final collection of his Natural History columns, and the other, his long-awaited magnum opus and a truly spectacular legacy in its own right, is The Structure of Evolutionary Theory. The following review of the latter book was written before his death. As I write this note the day after his death, it is still too soon, personally or professionally, to think of Steve in the past tense, and so I have left the review unchanged. The book was, of course, intended as a springboard to future work, and I can't help but feel that the best way to honor that intention is to retain the debate as well as the praise.—-D.J.

The Structure of Evolutionary Theory. Stephen Jay Gould. xxiv + 1433 pp. The Belknap Press of Harvard University Press, 2002. $39.95.

This is a very big book about a very big subject: the conceptual expansion of evolutionary theory. Here Gould provides a magnificent synthesis and overview of his broad vision of a more complete evolutionary theory. Not since his first book, Ontogeny and Phylogeny, published 25 years ago, has Gould produced a volume aimed primarily at a professional audience. The Structure of Evolutionary Theory is more than twice as long as his first book, but all of the major themes of Gould's writings are here, woven together into a coherent and remarkably readable whole.

Gould sees conventional Darwinism as consisting of three fundamental ideas: that natural selection works almost entirely at the level of organisms (rather than at multiple levels extending from genes to species), that selection is the exclusive shaper of evolutionary change, and that the extrapolation of minute, incremental changes can explain the entire history of life. He argues that all three are in need of, or more properly are in the midst of, conceptual expansion. Each of the central Darwinian claims have faced challenges since the very inception of evolutionary thought, but the more heterodox thinkers of the 19th and 20th centuries were lambasted or ignored—often, Gould notes, for good reason. The major figures in the field also raised such issues, but these were gestures amounting to little more than footnotes, unelaborated caveats and asides—particularly as the Modern Synthesis hardened in the 1950s and 1960s. Gould documents this conceptual shift to a narrower set of evolutionary mechanisms by a close reading of successive editions of the major works of the field. (I can attest from firsthand experience in undergraduate courses and graduate training through the 1970s that Gould's portrait of conventional evolutionary thinking of that not-too-distant era is accurate.) This book, then, is one long argument (to borrow a phrase from Darwin, who unmistakably retains heroic, although not deified, stature here) for the expansion or revision of the three central pillars of Darwinian thought.

First and foremost in Gould's portrayal is the near-exclusive position of the individual organism as the locus of evolution by natural selection. Darwin and many others saw that what Gould calls agency, the level at which selection operates, is central to evolutionary causation. Gould builds a powerful case for the operation of selection at many levels within a genealogical hierarchy, from DNA to species and even clades. This has been a central theme in his writing for decades, and so the major points will come as no surprise to anyone familiar with his work. But his views have themselves evolved (and not always in a punctuational fashion), and here he makes an integrated statement that will be essential reading for anyone grappling with this difficult subject.

This hierarchical structure of the evolutionary process is the central theme of the book, cast mainly in terms of the pervasiveness of punctuated equilibrium and the consequent role of species selection in shaping large-scale evolutionary patterns. Gould provides a massive overview of the evidence for stasis, punctuation and gradualism in the fossil record, and shows that when large clades or assemblages are censused, morphological stasis and punctuation tend to heavily outweigh substantial directional gradual change. (Recall that stasis involves the absence of net change and not absolute phenotypic immobility.) In addressing the most striking exceptions, Gould makes the intriguing suggestion, inspired by recent work in experimental microcosms, that the more widespread occurrence of gradual species change in the unicellular, planktonic foraminifera, which are predominantly asexual lineages, might be underlain by punctuational evolution of their constituent clones. The frequency of sexual reproduction and the match of stable morphologies to genetic lines are still poorly known in these groups, however, and a further complication is the apparent rarity of gradualistic change in metazoan groups (such as corals and bryozoans) that have clonal reproductive phases, sometimes to the near-exclusion of sexual interludes.

Gould concedes that not all morphological change is concentrated in speciation events. This is not as large a retreat as it first appears, because he embraces Douglas J. Futuyma's argument that speciation promulgates evolutionary change by sequestering, and therefore stabilizing, novel morphologies (or other aspects of the phenotype) that would otherwise be transient, localized features soon swamped by gene flow. Speciation thus retains primacy as the promoter of net evolutionary change. Perhaps a larger concession is the narrowing of potential mechanisms for the crucial phenomenon of evolutionary stasis. Here Gould downplays biologically intrinsic constraints (which are reemphasized for larger-scale patterns) in favor, essentially, of the flip side of Futuyma's model: Local populations may follow their own evolutionary trajectories for a time, but these local trajectories will have little cumulative effect if each population is adjusting to its unique conditions—such locally derived traits will be disfavored in other habitats that are occupied by species members in other persistent populations.

Whatever the mechanism for stasis, the empirical fact that individual species generally fail to exhibit directional trends over geologic timescales means that the differential production or survival of species (that is, species selection in the broad sense) may be responsible for evolutionary trends—Gould's species-level "grand analogy" to Darwinian organism-level selection. The empirical literature on species selection, or the broader category of species sorting, grows slowly, but Gould cites potential new examples and quotes recent evidence for the 25-year-old Wright's Rule (that new species arise randomly with respect to the net directional trend of their lineage). He thereby furthers the analogy with organismal selection, in which mutation is expected to yield random variation relative to selection pressures.

Gould reaffirms his preference for emergent fitness rather than emergent traits as the criterion for species selection, the higher-level analogue of natural selection on organisms. Thus he includes in this category any process driven by differential proliferation of species, rather than just the subset of those trends in which speciation or extinction differentials are shaped by species-level characters such as geographic range or genetic population structure. He concedes that the second, stricter definition represents the "best cases" but widens his concept to include any situation in which the production or survival of species is not simply reducible to the operation of selection on organismal traits. For example, intraspecific variability is not a property of an organism; but in Gould's view it is not an emergent species-level trait, because it entails a sum-of-the-parts aggregate of individual organisms. To some extent this is a semantic argument: Any sorting process irreducible to selection on bodies validates a hierarchical evolutionary theory. But Gould's stance is surprising, given his insistence, when exposing "the fallacy of the selfish gene," that "the causality of selection resides in interaction, not replication." If we are to move from pattern to process, then the hierarchical level of the target of selection—that is, the nature of the interaction between the focal unit and its environment (not just the differential replication of species)—is a key consideration. Broad definitions risk obscuring crucial distinctions.

Gould may assume an unnecessary difficulty (to turn Huxley's warning to Darwin on its head) by pinning his hierarchical theory so heavily on punctuated equilibrium. Punctuation and stasis make the operational recognition of species selection more straightforward. However, Gould fails to cite the classic 1983 paper in Paleobiology in which Montgomery Slatkin showed that the operation of species selection does not actually require punctuation and stasis, thereby even further enlarging its potential scope. Nor does Gould address Sean H. Rice's interesting genetical model for species selection (Journal of Theoretical Biology, 1995), which explores the potential role of species selection in character evolution, a domain that Gould surrenders to organismal selection. Most importantly, he misses a good bet by overlooking so much of ecology. If geographic range or rarity are species-level traits or are at least likely to confer emergent fitnesses, then the burgeoning field of macroecology is rich in potential cases of species selection.

The second of the central tenets of classical Darwinism is the unfettered efficacy of natural selection. Gould devotes two chapters, totaling 270 pages, to the integration of constraint and adaptation. He does not challenge the role of selection, predominantly at the level of the organism, in fashioning complex adaptations. But he does challenge the assumptions that variation, the raw material presented to selection, is effectively random (more precisely, "isotropic") and that it always occurs in infinitesimally small increments.

Gould plunges gleefully into the extraordinary new literature of molecular developmental biology and its macroevolutionary implications. This work is beginning to reveal in exquisite detail the network of regulatory genes behind complex structures, thereby providing a basis for what Gould terms top-down channeling of evolutionary directions, as opposed to bottom-up probing of equiprobable alternatives. In other words, the intrinsic developmental machinery of organisms produces an uneven probability profile of evolutionary transitions around any given starting point. The more we know about the developmental mechanisms, the better we will be able to predict the shape of that profile and the ability of a given lineage to break out of such channels.

Gould elaborates his longstanding thesis that many features start as spandrels (architectural or developmental byproducts of other structures) and are later coopted to functional status. For him, spandrels are ubiquitous. He uses the term to refer not to standard Darwinian "preadaptions" (feathers evolving under selection for insulation and being coopted as flight apparatus), but to features of nonadaptive origin, such as the open space at the base of a snail shell. (This umbilicus is geometrically inevitable for all but the most tightly coiled shells, but it can then provide a home for brooded young.) Gould makes two points, logical but untested, that will provide fodder for research programs for years: The more complex the morphology, the more potential spandrels and thus the greater evolvability; and novelty introduced at one level will propagate effects upward and downward in the hierarchy—that is, will yield nonadaptive (not necessarily maladaptive) consequences of adaptation or other processes at their level of origin.

Finally, Gould challenges the smooth extrapolation of short-term evolutionary processes to the full sweep of geologic time. Not only can the sorting of species and more inclusive groups occur in opposition to—or independent of—selection at the level of individual bodies, but changes in survivorship rules during rare mass extinction events can derail evolutionary trajectories set at organismic and species levels. Recent work suggests that Gould puts too heavy an emphasis on the mass-extinction filter. The sorting that occurs in the aftermath of mass extinctions, with its own set of winners, losers and marginalized players, may be as important in shaping long-term evolutionary patterns as the extinction event itself.

I fear that the sheer mass of this book will discourage readers, and that would be a shame. Yes, it would probably have been more effective in shorter form, and some of the personal asides and metaphoric flourishes are more telling than others. But in a very real sense it is three books, interweaving a rich history of evolutionary thought, a state-of-the-art formulation of hierarchical evolutionary theory and a professional autobiography. The first 500 pages are predominantly historical, providing fresh perspectives not only on the major players (Darwin, Lamarck, Cuvier, Owen and others), but also on less familiar figures such as August Weismann (who promulgated an explicitly hierarchical evolutionary theory); Hugo de Vries (who, as the first to develop the concept, coined the term species selection); Charles Otis Whitman (who used Darwin's pigeons as evidence for the heresy of orthogenesis); and many more. Gould insightfully builds a case for viewing Darwin's work as an effort to develop a comprehensive strategy for historical science, as valid and rigorous as its experimental cousin.

If this book does not have the same galvanizing effect as Ontogeny and Phylogeny, it will be for the happiest of reasons: Many of the views presented here have become part of the working toolkit of evolutionary biology, respectable hypotheses for testing or elaboration—for example, the potential role of intrinsic constraints in channeling evolution; sorting at levels above and below the organism (differential proliferation is routinely tested, for example, in many comparative analytical methods and in the study of transposable genetic elements); critical approaches to adaptations and their origins; and mass extinctions as profound bottlenecks in the history of life. Needless to say, there is still plenty in the book to stimulate debate; Gould does not avoid such hot-button issues as nonadaptation and the status of replicators and interactors in the operation of selection.

I expect this important book will serve a different primary role, as a deeply considered, conceptually rich and well-articulated synthesis of the new approaches and newly invigorated older themes that provide the foundation for a truly hierarchical evolutionary theory.