The Structure of Evolutionary Theory

by Stephen Jay Gould

  The first phase, reasonably called a synthesis of Mendel and Darwin, re­quired the extinction of Mendelism's first major episode of evolutionary em­ployment in the non-Darwinian saltational theory of de Vries (see Chapter 5). Huxley cites three major steps (1942, p. 25): the recognition that Mendelian principles operate in all organisms, unicellular and multicellular, plant and animal; the key insight that small scale, continuous Darwinian variability also maintains a Mendelian basis; and the mathematical demonstration that small selection pressures acting on minor genetic differences can render evo­lutionary change. This work culminated in the origin of theoretical popula­tion genetics (see Provine, 1971), and led to our almost catechistic invocation of a trinity — Fisher, Haldane, and Wright — as heroes of this first phase. (I do not challenge the attribution, but have long been amused by the almost for­mulaic citation; even the order of names has become invariant — as in another more famous Trinity!)

  The second phase, also a “synthesis” in the vernacular sense of the word, began with Dobzhansky's masterful book (1937) and proceeded as a linking of traditional sub disciplines in biology to the core theory forged during the first phase. This activity, already in full swing when Huxley wrote his book, eventually included such classics as Mayr (1942) for systematics, Simpson (1944) for paleontology, White (1945) for cytology, Rensch (1947) for mor­phology, and Stebbins (1950) for botany. Huxley made a good prediction, and hoped that his own effort would fuel the grand integration: “The time is ripe for a rapid advance in our understanding of evolution. Genetics, devel­opmental physiology, ecology, systematics, paleontology, cytology, mathe­matical analysis, and have all provided new facts or new tools of research: the need today is for concerted attack and synthesis. If this book contributes to such a synthetic point of view, I shall be well content” (1942, p. 8).

  I accept this traditional account of what the Synthesis synthesized, but [Page 505] I prefer to view the history of the Synthesis under a different rubric and terminology developed by historian of science Will Provine and by myself — namely, (1) restriction followed by (2) hardening, with the first process viewed as largely admirable, the second as mostly dubious.

  Provine has argued (1986) that the first phase of synthesizing, the integra­tion of Mendel and Darwin to a core discipline of population genetics, should be viewed as a welcome restriction (a “constriction” in his favored term) — for biologists could now shed the several competing and truly contradictory theories (primarily orthogenetic and saltational) that had made the domain of evolutionary studies seem so anarchic at the Darwinian celebrations of 1909.

  But the first phase also included a vigorously pluralistic range of permis­sible mechanisms within the primary restriction. The original synthesists wanted to render all of evolution by known genetic mechanisms; but they tended to agnosticism about relative frequencies among the legitimate phe­nomena, notably on the issue of drift (and other random phenomena) vs. selection.

  The second phase began with this pluralism intact, as the first wave of books from the late 1930's and 1940's clearly illustrates. But this broad net then tightened, as the leading synthesists promoted natural selection, first to a commanding frequency and then to virtual exclusivity as an agent of evolu­tionary change. This consensus hardened to orthodoxy, often accompa­nied by strong and largely rhetorical dismissal of dissenting views — a position that reached its acme in the Darwinian centennial celebrations of 1959. The pluralism of “consistent with genetics” eventually narrowed to a restrictive faith in what Weismann had called the “all-sufficiency” of natural selection, with the accompanying requirement that phenotypes be analyzed as prob­lems in adaptation. I almost feel as if the arbiters of fashion and theory had distributed sandpaper to all evolutionists — and that ordinary professional ac­tivity shifted to a vigorous smoothing out of all remaining bumps, facets, and channels of Galton's polyhedron (see pp. 342–351), as organisms became un­fettered billiard balls, rolling wherever the pool cue of natural selection dic­tated upon the crowded table of ecology.

  This hardening extended beyond overconfidence in adaptation to a more general, and sometimes rather smug, feeling that truth had now been discov­ered, and that a full account of evolution only required some mopping up and adumbration of details. The paeans of self-satisfaction penned for the 1959 centennial of the Origin (see pp. 569–576) almost invite a parody of Hamlet's insightful comment about his mother's rhetorical overstatements: the writers praised too much, one thinks.

  Synthesis as Restriction

  THE INITIAL GOAL OF REJECTING OLD ALTERNATIVES

  William Bateson, to be sure, had his own particular axe to grind when he so dramatically expressed his pessimism about evolutionary theory in his fa­mous address to the American Association for the Advancement of Science in [Page 506] 1922: “faith has given place to agnosticism” (see p. 412 for the full quote). The first rumblings of synthesis had already occurred. As the most obvious and important example, Fisher's pivotal 1918 paper on “The Correlation Be­tween Relatives on the Supposition of Mendelian Inheritance” had demon­strated the potential Mendelian (polygenic) basis for the small-scale, continu­ous variation that Darwinians had always identified as the primary source of evolutionary change — and that the early Mendelians had either ignored as supposedly non-particulate, or denigrated as evolutionarily insignificant (see pp. 429–431 on de Vries's attitude). Fisher's demonstration established a ba­sis for resolving the increasingly fruitless debate between “biometricians” (Darwinian supporters of continuous variation) and “Mendelians” (uphold­ers of saltational change, in the first evolutionary use of principles that would later blend with Darwinism) — a seminal, if ultimately barren, dichotomy in early 20th century biology. By using the principles of one side (Mendelian particulate inheritance) to explain the supposedly contradictory phenomena of the other (continuous and isotropic variation in phenotypes) — and then by banning the opposite (saltational) phenomena originally attached to the prin­ciples — Fisher staged a brilliant coup that surely deserves the label of “synthesis.”

  If Bateson had fallen a bit behind the times by 1922, he did accurately re­cord the near anarchy that had prevailed in the study of evolutionary mecha­nisms just a few years earlier. Using Kellogg's (1907) framework, as I have done throughout this book (see pp. 163–169 for an account and rationale), we may classify attitudes outside the central core of Darwinism either as help­ful expansions or as contradictory substitutions — auxiliaries and alternatives in Kellogg's terminology. Kellogg identified three major alternatives to Dar­winism — Lamarckism, saltationism and orthogenesis. At the 1909 celebra­tions for Darwin's centennial, all three alternatives enjoyed substantial sup­port, probably equal in extent, and in the reputation of leading supporters, to the popularity of Darwinism itself. As a supreme and well known irony, the original impact of the Mendelian rediscovery — the ultimate source of restric­tion and synthesis — had only increased the range and intensity of evolution­ary debate by revivifying the saltational alternative and thus augmenting the roster of major challenges to Kellogg's triple threat.

  Clearly, the state of evolutionary theory required restriction for further advance — either a settling upon one of the four contenders (Darwinism plus the three challenges), or a new formulation. The first phase of the synthesis accomplished this goal in three major moves: (1) by choosing the Darwinian central core as a proper and fundamental theory; (2) by reading Mendelism in a different way to validate, rather than to confute, this central core; and (3) by utilizing this fusion to ban the three alternatives of Lamarckism, saltation, and orthogenesis.

  Darwinism is a functionalist theory with an operational core that must place primary weight upon building adaptation by the mechanism of natural selection (see Chapters 2 and 4). In logic and principle, therefore, the theory could be confuted in either of two ways: by affirming an alternative mechanism [Page 507] for adaptation, or by denying the primacy of functionalism, and propos�
�ing a dominant role for internal, structuralist shaping of evolutionary change. Provine (1983) has designated the first phase of synthesis as a restriction be­cause the fusion of Darwin with Mendel validated the rejection of all three Kelloggian alternatives, and in both of their more general categories.

  Alternative theories of functionalism. Most importantly, this first-stage fusion finally gained enough knowledge and proof to depose the longest and most severe of all challenges to Darwinism — the doctrine that preceded natural selection, and that Darwin himself had both accepted as a subsidiary player and granted an ever-increasing role through subsequent re­visions of The Origin of Species. This so-called, if misnamed, “Lamarckian” theory of soft inheritance and direct production of adaptation by inheritance of acquired characters, had stubbornly persisted through all vicissitudes of evolutionary debate to remain a favorite in certain circles (at least among field naturalists, if not among experimentalists) well into the 1920's.

  Mendelians had always rejected Lamarckism as inconsistent with the mechanisms of inheritance, but Darwinians before the synthesis had gener­ally downplayed, ignored or actively rejected Mendelism. The first-stage fu­sion finally gave Darwinian functionalists a tight rationale for outright rejec­tion of this alternative Lamarckian route to adaptation. This clear solution struck most Darwinians as far more satisfactory than the uneasy pluralism that functionalists (including Darwin himself) had been forced to espouse be­fore — for Darwinism and Lamarckism, as pathways to the same result of ad­aptation, had coexisted no more peacefully than most competitors for the same prize. An old motto proclaims that the enemy without should always be preferred to the enemy (or apparent helpmate) who saps your strength from within.*

  Demoting internalism. The dismissal of Lamarckism left two strong alternatives in Kellogg's triad: saltationism and orthogenesis (see Chapters 4 and 5). These two challenges to Darwinian functionalism have al­ways been linked in their common focus on internal sources of change and di­rection — as jointly embodied in the metaphor of Galton's polyhedron (see Chapter 5. The reformulation of Mendelism as a source for pervasive, small-scale, continuous variability (and not only for mutations of substantial ex­tent) provoked a strong distaste for saltationism, as early population geneticists [Page 508] denied evolutionary importance to mutations of large effect. At the same time, the Mendelian explanation of this copious and small-scale variability left little scope for orthogenesis. Darwinism surely welcomed this further Mendelian aid, for all functionalist theories must try to smooth Galton's polyhedron.

  Yet, while internalist theories endured a pronounced demotion during this first phase of restriction, these alternatives to Darwinism did not suffer the Lamarckian fate of complete dismissal in theory. Saltations could not be ban­ished or denied, but only declared unimportant in evolution. Orthogenesis could not be overthrown in principle, for mutation pressure could conceiv­ably boost the frequency of alleles from within. Moreover, and more impor­tantly as an enjoined consequence of its own premises, population genetics had to acknowledge another potentially substantial source for non-adaptive evolutionary change: the effects of sampling errors, primarily in small popu­lations.

  Thus, the first phase of restriction invoked a fusion of Mendel and Darwin to dismiss or downplay the traditional roster of alternatives to Darwinian functionalism. But the resulting theory remained open and pluralistic in wel­coming all notions consistent with the new formulation of Mendelism. I shall devote the rest of this section to three illustrations of the two key properties in this first phase of the synthesis: (i) the revivified Darwinian core, and (ii) toleration of a broad phenomenology, including substantial nonadaptation, so long as results could be rendered by known genetic mechanisms. I will discuss the two most important books of early population genetics: Fisher (1930) and Haldane (1932). Huxley's compendium (1942) should be read as a transitional document, and may belong more properly to the begin­ning of the second phase of hardening, but I include Huxley's book here as a summation of early ideas in the Synthesis, if only for its symbolic role in sup­plying the developing theory with a name.

  R. A. FISHER AND THE DARWINIAN CORE

  The history of the Modern Synthesis holds special, one might almost say inspirational, interest in the good fortune of its construction by such a fascinat­ing group of scientists — so different in personality, so diverse in philosophical attitude amidst their defining agreement, and so brilliant. I experienced the great intellectual privilege of knowing most of the second-phase founders — and I have tried to understand the personal components of scientific greatness by, for example, contrasting Dobzhansky's infectious enthusiasm with Mayr's fierce commitment and encyclopedism. I have also pondered the intra-individual variety in trying to square the warm and expansive humanism of Simpson's writing with the irascibility of his personality.

  The leaders of the first phase were equally fascinating and diverse. Their differences have been discussed extensively, perhaps most cogently by one of the actors himself (Wright, 1978). But R. A. Fisher holds a special status as author of the movement's first major book (1930), and as the most thoroughgoing [Page 509] and uncompromising strict Darwinian among early synthesists. As a great logician and statistician, Fisher explored the Darwinian consequences of small changes in large populations with a consistency and completeness never achieved before. We may choose, in retrospect, to regard the views of other evolutionists as more subtle, or more attuned to nature's diversity and ambiguity — but no one can match Fisher as an advocate of the main line in its pure form. We must therefore treat him as inaugural, in both senses of prior­ity and centrality.

  Fisher begins The Genetical Theory of Natural Selection by explaining how the logic of Darwinism requires a particulate theory of inheritance in or­der to achieve self-sustainability as a mechanics of change without special pleading and additional forces. He states in his introduction (1930, pp. vii-viii): “That an independent study of Natural Selection is now possible is prin­cipally due to the great advance which our generation has seen in the science of genetics.”

  As an opening gambit, Fisher develops the intriguing argument that Dar­win's forced allegiance to blending inheritance entailed far more than mere inconvenience and surmountable difficulty, thus confuting the usual claim, advanced by scientists who have traditionally viewed the Origin as a com­plete and fully workable theory of natural selection. Rather, Fisher maintains, blending inheritance represented a central impediment, debarring natural se­lection as the chief agent of evolution. By constantly degrading favorable ge­notypes, blending requires an enormous input of new mutation to fuel any process of change. But mutation rates this high could easily overwhelm any reasonable force of natural selection. Evolution would then be propelled largely from within, and natural selection, as a theory of “trial and error externalism,” could not operate as the major cause of change. But a particu­late theory provides raw material for favorable change without degrada­tion — and a vastly lowered rate of new input now suffices. Evolution need not be driven from within, and may now be pulled by the external force of natural selection.

  Fisher conceived his argument as a proof by elimination. Once we dispense with a need for such lavish internal stirring, no credible force but natural selection remains. Mendelism therefore validates Darwinism: “The whole group of theories which ascribe to hypothetical physiological mechanisms, controlling the occurrence of mutations, a power of directing the course of evolution, must be set aside, once the blending theory of inheritance is aban­doned. The sole surviving theory is that of Natural Selection” (1930, p. 20).

  Fisher illustrates the restrictive character of this first synthetic phase by invoking the fusion of Darwin and Mendel to discard each alternative of Kellogg's triad. Lamarckism (as noted before) goes quickly and gently into that good night — for soft inheritance cannot exist under the newly validated mechanics of Mendelism. Saltationism surrenders when small-scale continu­ous variability gains a Mendelian ba
sis and wins evolutionary effectiveness thereby. This small-scale Mendelian component provides superior evolution­ary raw material for two reasons: (1) overwhelming predominance in sheer [Page 510] amount (saltations occur only rarely, but minor variants are ubiquitous); and (2) greater potential for utility (small changes can often be advantageous, but large excursions will almost always be disruptive — see Fig. 7-1, from Fisher). When De Vries and other biologists had disparaged small-scale variation as ineffective and different in kind from Mendelian effects, saltations had pre­vailed faute de mieux. But once we can assert a continuum in effect and a uni­formity in genetic mechanism for variation at all scales, then we must prefer an omnipresent and potentially advantageous mode to an exceedingly rare and almost always deleterious extreme in the spectrum. Fisher writes: “The chance of improvement, for very small displacements, tends to the limiting value one-half, while it falls off rapidly for increasing displacements (p. 39). In any highly adapted organism the probability of advantage through any considerable evolutionary step (saltation) rapidly becomes infinitesimal as the step is increased in magnitude” (p. 114).