The Structure of Evolutionary Theory

by Stephen Jay Gould

  If the Synthesis viewed the entire history of life, the full tree itself, as grow­ing by an adaptive and stately unfolding, then the history of single branches — trends in lineages, the primary topic of macroevolutionary study — also re­ceived a thoroughly adaptationist reading in the extrapolationist mode. At the same Chicago conference, Simpson defended the adaptationist postulate for all geometries, parallel as well as diverging lineages, and even (in principle and without direct evidence) for “erratic” features where selective control “is not apparent.”

  The selectionist theory is that a trend is adaptive for the lineage involved, that it continues only as long as it is adaptive, that it stops when adapta­tion is as complete as selection can make it in given circumstances, and that it changes or the group becomes extinct if a different direction of evolution becomes adaptive. Often the adaptive nature of a trend seems apparent. Often it is not apparent, but the postulate seems required to account for otherwise erratic features of trends. In instances of parallel evolution the selectionist theory is that changes actually occurring in parallel are adaptive over the whole ecological range occupied by the group, while those divergent (radiating) within the group are adapta­tions to special niches within that range (Simpson, 1960, pp. 170-171).

  Under these precepts, a procedure of building scenarios in the strictly adaptationist mode, based on assumption and conjecture, often passed for ade­quate explanation. The second half of the Chicago panel on the “evolution of life,” supposedly dedicated to the actual record of evolutionary change, in­cluded almost no discussion about paleontology, and relied on theoretical inferences about the past based on knowledge of modern organisms. Ernst [Page 560] Mayr, however, did offer the following conjecture — wrong in many details (as we now know), yet firm in its confident adaptationist scenario — for the evo­lution of lungs. (Devonian fishes already possessed lungs, for the trait is symplesiomorphic in tetrapods and their aquatic ancestors, with the swim bladder of later fishes as its derived homolog. But note Mayr's confidence in his erroneous conjecture for the easy construction of such a novelty — from scratch, gradually, and in pure adaptive continuity with unchanging func­tion):

  I think the development of lungs is now pretty well understood. Certain fishes during the Devonian period lived in stagnant, fresh water swamps, where oxygen was so scant that respiration through the skin and the gills no longer provided the necessary oxygen. Apparently they came to the surface and gulped air, from which the membranes of the digestive tract took up oxygen. When that stage was reached, there was a tremendous selection pressure for developing diverticles and enlarging this respira­tory surface of the digestive tract. As soon as the necessary gene combi­nation providing such diverticles appeared, selection pressure could push this tendency further and further, and this led quite naturally to the de­velopment of lungs (in Tax, 1960, volume 3, p. 136).

  As documented in Chapter 6, the putative domination of biotic over abiotic competition provided Darwin with a rationale for defending general progress in the history of life. The synthesists upheld this orthodoxy as well, thereby imparting broad predictability to the stately unfolding of life. Huxley offered a clear assessment of relative frequencies in the founding document (1942, p. 495): “Sometimes the inorganic environment changes markedly, as when there is a climatic revolution, such as occurred at the end of the Creta­ceous; but in general it is the organic environment which shows the more rapid and important alterations.”

  In his concluding address to the entire Chicago symposium, Huxley then remarked, with the expanded scope and surer resolve of nearly two decades in hardening: “Improved organization gives biological advantage. Accord­ingly, the new type becomes a successful or dominant group. It spreads and multiplies and differentiates into a multiplicity of branches. This new biologi­cal success is usually achieved at the biological expense of the older dominant group from which it sprang or whose place it had usurped. Thus, the rise of the placental mammals was correlated with the decline of the terrestrial rep­tiles, and the birds replaced the pterosaurs as dominant in the air” (1960, p. 250).

  In citing the canonical example of dinosaurs and mammals, Huxley ex­poses the heart of the extrapolationist error — the assumption that large-scale pattern can be inferred by extending, through immense time, the small effects of observable processes (in this case the supposed general and overall “superi­ority” of mammals over reptiles in most cases of immediate competition). In explaining trends, the greatest threat to this orthodoxy lies in occasional but profound environmental catastrophe that disrupts and resets the pattern accumulating [Page 561] during “normal” times. The theory and factuality of catastrophic mass extinction has now broken this orthodoxy (see Chapter 12), but sim­ple knowledge of mass extinction had always posed a threat, especially in Cuvier's original paroxysmal interpretation (see pp. 484–492). The synthesists therefore treated this apparent phenomenon in the conventional and congenial way, either by dismissing a catastrophic cause, or by “spreading out” the time of extinction so that all deaths could be encompassed by tradi­tional competitive mechanisms, perhaps enhanced in intensity by rapid envi­ronmental changes, and therefore propelling adaptive evolution even more rigorously. Huxley (1942, p. 446), for example, held that tough physical con­ditions only accelerated the competitive takeover by superior groups: “The worsening of the climate at the end of the Mesozoic reduced the general adaptiveness of the dinosaurs, pterosaurs, and other reptilian groups, while increasing that of the early mammals and birds.”

  Ernst Mayr, in his characteristically forthright way, then linked the denial of catastrophic extinction, via uniformitarianism, to the crucial second state­ment in his definition of the Synthesis (see p. 557), the requirement for ex­trapolation into geological vastness: “Yet it has become clear that there is nothing in the past history of the earth that cannot be interpreted in terms of the processes that are known to occur in the Recent fauna. There is no need to invoke unknown vital forces, mutational avalanches, or cosmic catastro­phes. Geographic speciation, adaptation to the available niches (guided by se­lection), and competition are largely responsible for the observed phenomena” (1963, p. 617).

  In short, by viewing trends as adaptive and anagenetic phenomena, pro­pelled by competition and building, by a lengthy process of stepwise summa­tion, the principal pattern of life's history, the Synthesis encompassed the most salient phenomenon of paleontology within its favored framework of extrapolation. All causality could reside in the accessible here and now. How then, we must ask, did the Synthesis treat the two phenomena — speciation and extinction — now viewed as crucial in breaking the extrapolationist or­thodoxy (for if trends must be expressed as the differential success of some kinds of species vs. others, with most species formed in geological moments, then the adaptive struggles of populations don't extrapolate smoothly to changes of mean and modal phenotypes within clades)?

  The developing orthodoxy generally acknowledged speciation and then demoted its importance and distinctiveness. According to Huxley, for example, all radiations should be treated as adaptive and each event of speciation therefore represents an independent, gradualistic expression of an anagenetic trend (1942, p. 487): “The adaptive radiation is seen to be the result of a number of gradual evolutionary trends, each tending to greater specializa­tion — in other words to greater adaptive efficiency in various mechanisms subservient to some particular mode of life... Each single adaptive trend also shows the phenomenon of successional speciation.”

  In a statement that I find charming, however wrongheaded, Nicholson (1960, p. 518, at the Chicago centennial symposium) extolled speciation [Page 562] as a device to provide more opportunities for adaptation to work its “untrammeled” magic: “The splitting of organisms into the genetically isolated groups we call 'species' has played a very important part in evolution, for it has permitted selection to proceed untrammeled within each group, so per­mitting adaptations of innumerable kinds in th
e different groups. Had organ­isms not divided into genetically isolated groups, the numerous and beautiful adaptations so characteristic of living things could not have evolved, nor could organisms have used the resources of the world in the efficient way they do” (Nicholson, 1960, p. 518).

  More commonly, however, speciation received short shrift rather than glory. Evolution required such a process of multiplication, of course, lest fa­vorable trends disappear through the extinction of single species bearing their fruits. Speciation therefore became a hedge against death by parcelling out, into several iterated lines, a set of adaptations built anagenetically — so that the extinction of one species could not abort the general benefit. The trend itself remains anagenetic (see Fig. 7-2), for speciation does not contribute to the directionality of evolutionary change. (Under later views, including punctuated equilibrium, differential speciation constructs the trend, and anagenetic main trunks do not even exist.)

  Simpson held strongly to this view, and even ventured a quantitative de­fense, in his repeated assertions that speciation represents only a minor mode in evolution because 90 percent of important changes arise anagenetically in the phyletic mode (1944, 1953; Simpson recognized three major modes of change: speciation, phyletic evolution, and quantum evolution). Huxley, in a grand prose flourish, then branded speciation as a pretty little epiphenomenon, a luxurious patina upon the grand pattern of evolution — never realizing

  7-2. Standard view of the role of speciation in evolutionary trends under the Modern Synthesis. Speciation certainly plays an important part in iterating favorable variations produced by anagenesis within species. (If this iteration did not occur, lineages would quickly become extinct because individual species must eventually die.) But the trend in morphology arises almost entirely by anagenetic directionalism within the geological duration of individual species.

  [Page 563]

  that the pattern itself might be built by higher-level sorting, operating through the differential success of certain kinds of species!

  The formation of many geographically isolated and most genetically isolated species is thus without any bearing upon the main processes of evo­lution ... Much of the minor systematic diversity to be observed in nature is irrelevant to the main course of evolution, a mere thrill of variety superimposed upon its broad pattern. We may thus say that, while it is inevitable that life should be divided up into species, and that the broad processes of evolution should operate with species as units of organiza­tion, the number thus necessitated is far less than the number, which ac­tually exist. Species-formation constitutes one aspect of evolution; but a large fraction of it is in a sense an accident, a biological luxury, without bearing upon the major and continuing trends of the evolutionary pro­cess (Huxley, 1942, p. 389).

  Amidst this attempt to relegate the origin of the primary unit of macroevolution to irrelevancy at larger scales, one prominent voice within the Syn­thesis spoke up for the centrality of speciation in constructing large-scale pat­tern. In a cautious, but prophetic statement, Ernst Mayr (1963, p. 587) wrote: “To state the problems of macroevolution in terms of species and pop­ulations as 'units of evolution' reveals previously neglected problems and sometimes leads to an emphasis on different aspects.” (Much of macroevolutionary theory, as developed later, begins with this proposition, and Mayr therefore becomes an inspiration — ironically in a sense, for several key con­cepts in this developing body of thought have challenged other aspects of the Synthesis that Mayr so strongly championed. For example, the theory of punctuated equilibrium rests upon a proper translation into geological time of Mayr's peripatric theory of speciation — see Eldredge and Gould, 1972, and Chapter 9.

  Directly refuting Huxley's charge that speciation only ranks as a frill and luxury in the overall pattern of evolutionary change, Mayr wrote:

  I feel that it is the very process of creating so many species, which leads to evolutionary progress. Species, in the sense of evolution, are quite comparable to mutations. They also are a necessity for evolutionary progress, even though only one out of many mutations leads to a significant improvement of the genotype. Since each coadapted gene complex has different properties and since these properties are, so to speak, not predictable, it requires the creation of a large number of such gene complexes before one is achieved that will lead to real evolutionary advance. Seen in this light, it appears then that a prodigious multiplication of spe­cies is a prerequisite for evolutionary progress... Without speciation, there would be no diversification of the organic world, no adaptive radiation, and very little evolutionary progress. The species, then, is the key­stone of evolution (1963, p. 621). [Page 564]

  A world of difference separates the negative view held by most synthesists — that speciation merely iterates (and therefore buffers) adaptations pro­duced by a different, anagenetic process — from Mayr's recognition that adap­tations may be pieced together through accumulated events of speciation, each chancy in itself and not directed towards the eventual novel phenotype. In this sense, Mayr's view becomes the root for those branches of modern macroevolutionary theory that treat speciation as a higher-order analog of organismic birth — leading to a concept of trends as the product of differential sorting within the multitude of units thereby produced, and not as the extrap­olated result of organismic selection within anagenetic lineages.

  If most of the synthesists viewed speciation as trivial, they didn't grant even this modicum of concern to the counterpart process of extinction. Although they acknowledged the death of species (for a process affecting 99 percent of all species that ever lived can't be entirely ignored), they viewed extinction en­tirely in a negative light — as a loss of adaptation, and therefore as a failure in evolution, something to be recognized but not extensively discussed in polite company. Even Ernst Mayr, who understood so clearly how speciation could enter a higher-level process of sorting, didn't grasp the logical corollary — that any selective process must pair survivals with eliminations, and that “de­feats” can therefore teach us as much as “victories.” Instead, Mayr professed puzzlement as to why such a profoundly negative phenomenon should be so common:

  We find so many cases of extreme sensitivity of natural selection, doing the most incredible and impossible things; and yet the whole pathway of evolution is strewn left and right with the bodies of extinct types. The frequency of extinction is a great puzzle to me (in Tax, 1960, volume 3, p. 141).

  Natural selection comes up with the right answer so often that one is sometimes tempted to forget its failures. Yet the history of the earth is a history of extinction, and every extinction is in part a defeat for natural selection... Natural selection does not always produce the needed improvements (1960, pp. 375-376).

  The Synthetic approach to macroevolution can be encapsulated in a few dicta: view life as stately unfolding under adaptive control; depict trends as accumulative and anagenetic within lineages according to the extrapolationist model; downplay or ignore the macroevolutionary calculus of birth and death of species. These propositions leave little role for the actual archives of life's history — the fossil record — beyond the documentation of change. The causes of change must be ascertained elsewhere, and entirely by neontologists (my profession's term for the folks who study modern organisms). Thus the Synthesis held paleontology at arm's length. (I suppose we deserved this deni­gration in retaliation for the plethora of poorly conceived, anti-Darwinian assertions and speculations that so many earlier paleontologists had falsely based upon the fossil record — see Chapter 4. In this sense, our later demo­tion, however unfairly extended, became part of the salutary cleansing accomplished [Page 565] by the early Synthesis in its first phase of restriction — see pp. 505–505.) Huxley (1942, p. 41) spoke of “the illegitimacy of using data on the course of evolution to make assertions as to its mechanism.”* He continued:

  As admitted by various paleontologists ... a study of the course of evo­lution cannot be decisive in regard to the method of
evolution. All that paleontology can do in this latter field is to assert that, as regards the type of organisms, which it studies, the evolutionary methods suggested by the geneticists, and evolutionists shall not contradict its data. For in­stance, in face of the gradualness of transformation revealed by paleon­tology in sea urchins or horses it is no good suggesting that large muta­tions of the sort envisaged by de Vries shall have played a major part in providing the material for evolutionary change (1942, p. 38).

  Even so iconoclastic a morphologist as D. Dwight Davis, who would later tweak strict adaptationism so effectively in discussing formal and historical constraints in his classic monograph on the giant panda (Davis, 1964), wrote for the Princeton meeting on genetics, paleontology, and evolution (1949, p. 77): “Paleontology supplies factual data on the actual rates of change in the skeleton and the patterns of phyletic change in the skeleton. Because of the inherent limitations of paleontological data, however, it cannot perceive the factors producing such changes. Attempts to do so merely represent a superimposition of neobiological concepts on paleontological data.”

  I admit, of course, that paleontologists have no access to mechanisms requiring direct observation of ontogeny and ecological interaction. But to say, as Davis does, that we cannot ever derive concepts of evolutionary mecha­nisms from paleontological data — and must therefore gain all our causal un­derstanding from “neobiology” — seems excessively pessimistic, and consigns paleontology to impotence. If paleontologists cannot gain insights about mechanisms, then historical science of any kind becomes impossible, for all scientific study of the past must make causal inferences from results of pro­cesses that cannot be directly observed.