The Structure of Evolutionary Theory

by Stephen Jay Gould

  Dear Steve,

  I gave considerable thought to your question how my 1963 book differed from the 1942 one, and why adaptation was so much more fea­tured in the later volume. I think I now have the answer.

  Remember that I consider evolution by and large to consist of two [Page 541] processes: 1) the maintenance and improvement of adaptedness, and 2) the origin and development of diversity.

  Since (2) was so almost totally ignored by the pre-Synthesis geneticists, I focussed in 1942 on (2). By the 1950s the study of diversity had been fully admitted to evolutionary biology, owing to the efforts of Dobzhansky, myself, Rensch and Stebbins, and in my 1963 book I could devote a good deal of attention to (1). This was rather easy because, as you know, I used to be a Lamarckian. And Lamarckians are adaptationists. Hence, it is not that from 1942 to 1963 I had become an adaptationist, rather I reconciled in 1963 my adaptationist inclination with the Darwinian mechanism (Letter of December 20, 1991).

  (Mayr then added a handwritten footnote, demoting to insignificance the one subject for which he did acknowledge a reversal of opinion between the two books: “Neutral polymorphism is an infinitesimal percentage of all evolu­tionary phenomena. Don't make a mountain out of this little mole-hill.”)

  I do not deny Mayr's stable adaptationist preferences (through his ontogenetic change in explanatory preferences from Lamarck to Darwin). This personal stability provides an even better reason for regarding as important, and therefore generally indicative, the textual evidence of transition from plu­ralism in 1942 to adaptationist hardening in 1963 (for Mayr's 1942 text may therefore, by implications of his own testimony, be reporting the conven­tional pluralistic wisdom of the time despite Mayr's own personal preference for adaptationism). On the subject of adaptation — not the major concern of either book (for both treat speciation and the production of diversity as their primary topics) — Mayr recorded a professional consensus both times, largely passively I suspect (hence his personal inattention to the alteration). Scientists must struggle to identify and understand these influences of “shared culture,” for such a “background” consensus fuels the sources of unconscious bias for each of us at every moment of our careers.

  WHY HARDENING?

  I have documented the adaptationist hardening of the Modern Synthesis in some detail, but I have not addressed an obvious and pressing question: why did this conceptual trend occur? Several aspects of an answer seem clear, but I can offer no full or satisfying resolution.

  The culture of science trains us to believe that such major shifts of empha­sis record improvements in knowledge won by empirical research and discov­ery. I do not deny that observation did play a significant role, at least in il­lustrating, with some elegant examples, the power of adaptation. Consider, for example, the “ecological genetics” of E. B. Ford and his panselectionist school in England. Their commitment to adaptationist explanations of effec­tively all variation among populations, and their documentation of strong selection coefficients in nature, buoyed the strict Darwinian faith. Dobzhansky's [Page 542] own empirical work increased his belief in the power of selection. In 1937, he tended to attribute chromosomal inversion frequencies in natural populations of Drosophila to genetic drift, but he then discovered that these frequencies fluctuate in a regular and repeatable way from season to season, and he therefore decided (with evident justice) that such systematic and iter­ated change must be adaptive.

  But empirical discovery cannot supply the entire (or even, I think, the major) reason for adaptationist hardening, for each favorable case can be matched by a failure (often hedged or unacknowledged), and no adequate as­sessment of an overall relative frequency has ever been achieved — to this day. Thus, any judgment, in either direction, must represent the fashionable impo­sition of a few well-documented cases upon an unstudied plethora. For exam­ple, A. J. Cain and colleagues did win a major victory for adaptation by showing that banding-morph frequencies in the land snail Cepaea, a former mainstay for claims about genetic drift, reflected selection based upon visual predation by birds, and upon climatic factors (Cain and Sheppard, 1950, 1952, 1954).

  But Cain and his colleagues then recognized and named the outstand­ing pattern of “area effects” (Cain and Currey, 1963a and b) — abrupt geo­graphic changes in banding-morph frequencies occurring with no perceptible alteration in any environmental factor that might impose a selection pressure. In what can only be labeled an article of faith, Cain attributed area effects to selection based upon “cryptic [meaning truly unmeasured and unperceived by any investigator, not merely subtle] environmental differences” — a re­markable affirmation of an a priori preference based upon not finding the necessary empirical confirmation. Good evidence has since been presented for a non-adaptive explanation of area effects as historical remnants of previous patterns in land use, and not as an outcome of current regimes in selection (Cameron et ah, 1980; see review of the entire case in Gould and Woodruff, 1990). (Area effects rank as anomalies under selectionist presuppositions — hence Cain's need to supply an orthodox adaptationist explanation, even in the absence of required evidence. Under a “legacy of history” explanation, such discordance of morphology with present geography presents no anom­aly and need not even receive a special name.)

  If adaptationist hardening cannot be explained as simply and empiri­cally driven, we might turn to historical and sociological themes. Smocovitis (1996), as previously mentioned (see p. 503), presents the intriguing thesis that renewed optimism following the wreckage of World War II (including the hope inspired by the newly constituted United Nations) launched a strong push for scientific defenses of potential human improvement and evolution­ary progress — an impetus that became a semi-official movement spurred by positivistic theories of knowledge proffered as antidotes for older irrationalisms. Smocovitis writes:

  If selection had enough agency (and at the same time were a mechanical principle) then all the more rapid and possible the “improvement” of humans... [Page 543] More strongly selectionist models would be favored by biolo­gists who modelled themselves after physicists at the same time they pointed the way to the “improvement” of humanity and painted a pro­gressive and optimistic picture of the world . . . Evolutionary models fa­voring random genetic drift, which enforced a stochastic view of evolu­tion — and culture — would not be favored in a post-war frame of mind seeking to “improve” the world. So powerful would be the need for a progressive and selectionist framework in the 1940's that even Dobzhansky and Wright were to adopt more strongly selectionist models.*

  Some complex mixture of empirical and sociological themes may explain the adaptationist hardening of the synthesis, but we must not neglect the ad­ditional impetus of a cultural analog to drift and founder effects in small pop­ulations. The community of evolutionary biologists is sufficiently small, and sufficiently stratified — a few lead and many follow, as in most human activi­ties — that we need not necessarily invoke some deep and general scientific or societal trend to explain a change in opinion by a substantial community of evolutionists in different nations. A reassessment by a few key people, bound in close contact and mutual influence, might trigger a general response. The three leading exponents of hardening in America — Dobzhansky, Simpson, and Mayr — worked together as colleagues in a “New York Mafia” centered at Columbia University and the American Museum of Natural History. Add another seemingly eternal principle of human affairs — that founders tend to be brilliant and subtle, and to keep all major difficulties constantly in mind, while epigones generally promulgate the faith and disregard, or never learn, the problems, exceptions, and nuances — and we may then wish to view the adaptationist hardening as ultimately inadaptive for the broadest goal of understanding evolution aright. Bandwagons might well be construed as cul­tural analogs of internalist drives in nonfunctional orthogenesis. Theories can grow tired. Theories can also harden and lose their bearings when compla­cency occupies the driver's seat.

 
Hardening on the Other Two Legs of the Darwinian Tripod

  To illustrate the hardening of the Modern Synthesis, I have documented its most significant ontogenetic trend in extenso — increasingly exclusive empha­sis on adaptation as the sign of natural selection's pervasive power. But if we epitomize the Synthesis as Darwinism reclothed in Mendelian understanding, [Page 544] then, following this book's focal argument that the minimal commitments of Darwinian logic encompass three central themes, the other two legs of this essential tripod should experience corresponding changes as the Synthesis hardened. I will not provide so extensive a discussion of these other legs — levels of selection and extrapolation into geological time — but I do wish to record that the literature of these subjects also experienced the same ontogeny of solidi­fication (and unjustified neglect of reasonable alternatives).

  LEVELS OF SELECTION

  Darwin, as we have seen (pp. 125–137), viewed organisms as nearly exclusive agents of selection — for deep reasons situated at the core of both the logic (the invisible hand of Adam Smith transferred to nature) and the psychology (the inversion of Paley's world) of his theory. But few Darwinians grasped the rationale or centrality of this principle, and a tradition of vagueness and loose thinking about levels of selection developed. Some, like R. A. Fisher, rode Darwin's wave and wrote explicitly and cogently about reasons for choosing individual organisms as the proper locus, and for disregarding, as effectively impotent, other levels that must be deemed conceivable in theory (1958, on species selection — see my critique of Fisher on pp. 644–652). But others, dat­ing back to A. R. Wallace himself (see pp. 131–132), never understood the full logic and implications of this issue, and ranged indiscriminately up and down potential levels, without grasping the theoretical problems entailed by such excursions.

  Thus, a fluid situation prevailed on this issue at the time of the Darwinian centennial celebrations of 1959 — my point of reference for the triumphal height of the Modern Synthesis in its strongly adaptationist version. Adapta­tion had become all the rage, but vagueness shrouded the key issue of selec­tion's focus and level — and for two reasons.

  First, and less important because the position attracted few supporters, a few evolutionists explicitly advocated a multi-level view of both selection and adaptation. A group of Chicago ecologists, authors of an important text­book known by its acronym of AEPPS (Allee, Emerson, Park, Park and Schmidt, Principles of Animal Ecology, 1949), generated and led this small movement. Emerson spoke at the Chicago centennial symposium, and pre­sented his multilevel view in both content and title: “The evolution of adap­tation in population systems.”

  Emerson begins by acknowledging the conventional Darwinian preference for individual organisms (and reminding us that he will not neglect this usual argument). But he then stakes his higher claim: “It is my intention in this es­say to emphasize the evolution of adaptation in population systems without, however, negating the data or the major interpretations of the roles of indi­viduals in evolutionary history or processes” (1960, p. 307).

  I find Emerson's article frustrating, for his arguments are so reasonable in some places, and so very wrong, to the point of illogic, in others. On the one hand, he presents a defendable and properly philosophical criterion for [Page 545] higher-level selection based on features of populations that cannot be expli­cated as additive results of organismal properties — in other words, “emer­gent” characters. He correctly defines a population ripe for selection at its own level as “an inclusive entity with emergent characteristics that tran­scend the summation of the attributes of the component individuals” (1960, p. 307).

  But, having legitimately defined the problem, he then launches into an almost rhapsodic, and simply illogical, claim that almost anything with defin­able boundaries can be recognized as a unit of natural selection: “Natural selection operates at each level of integration from the gene and complex polygenic characters within the individual, to the whole individual, and to various levels of intraspecific population systems and interspecific inter-adapted community systems and ecosystems” (1960, p. 340).

  This argument can be defended in theory so long as the higher unit oper­ates as an interactor with surrounding environments and remains in a genea­logical nexus engaged in differential reproduction (see Chapter 8). But how can Darwinian selection possibly operate directly on an ecosystem? However we may choose to define such an entity, ecosystems do not mate and do not produce children (see Chapter 8, pp. 597–613 on criteria of Darwinian indi­viduality). No argument can be made about their differential reproductive success, and no Darwinian calculus can therefore be applied to their history through time.

  Emerson doesn't seem to grasp that selection works by differential reproductive success, not by design for immediate, self-serving utility: “It would be extremely difficult,” he writes (1960, p. 319), “to explain the evolution of the uterus and mammary glands in mammals or the nest-building instincts of birds as the result of natural selection of the fittest individual.” But if milk-rich mammary glands promote the survival of offspring, then the mother acts in her own Darwinian interest. In short, Emerson's paper gives us an unin­tended insight into the confusing lack of definition that natural selection has always suffered, even at the moment of its greatest explicit influence.

  Second — and more important in its virtual ubiquity — leading evolutionists, though well aware that orthodoxy identified individual organisms as the fo­cus of selection, did not grasp the logical necessity or centrality of such a claim in Darwinian theory, and therefore often indulged in vague, perhaps unconscious, and often fuzzy, statements about the efficacy of higher levels. (I say “fuzzy” because most of these claims about populations and groups only invoked the non-emergent effects of organismic characters — and therefore do not necessarily qualify as valid statements about higher-level selection. I don't think that many evolutionists had properly formulated this crucial issue at the time.)

  Dobzhansky, for example (1957, p. 392), states that selection operates on organisms, but then proposes that such phenomena as heterozygote advan­tage might record some populational “extra” in exposing the reduced fitness of homozygotes as a kind of organismic sacrifice “for” the group: “Natural selection operates through differential survival and differential fertility of individuals, [Page 546] and yet at some times brings about such forms of integration of the gene pool of the population which lead to the sacrifice of some of the indi­vidual members of the population. The phenomenon of balanced polymor­phism, with highly fit heterozygotes contrasting with less fit homozygotes, is one of such forms of genetic integration of Mendelian populations.”

  Mayr's most authoritative book (1963) provides an excellent illustration of organismic orthodoxy amidst statements, lacking clear definition, about se­lection at higher levels. Mayr surely recognizes the usual form of proper Dar­winian argument — that apparent benefits to populations should be ex­plained, whenever possible, as effects of selection upon organisms: “The solution usually proposed for the difficulty raised by the conflict between a benefit for the individual and one for the population is to make the popula­tion rather than the individual the unit of selection.... It would seem prefera­ble to search for solutions based on the selective advantage of individual ge­notypes, such as Fisher's explanation of an even sex ratio” (1963, pp. 198-199).

  In rereading Mayr's 1963 book with the hindsight of thirty years, however, I was struck by the number of passages and arguments that either speak loosely about explicit advantages to groups and populations (rather than for­tuitous beneficial effects arising as side consequences of selection on organ­isms), or seem to state an explicit claim for selection at the population level. Most of these statements focus on the virtues of genetic variability. Mayr asks (1963, p. 158): “Why are not all individuals of a population identical in ap­pearance? Is it because diversity is of selective advantage to the population?” He then argues (1963, p. 308) that the primary func
tion of chromosomal variation lies in the flexibility thus accorded to populations: “They appear to have, as primary function, either the increase of adaptability and adaptedness of these populations through balanced polymorphism of entire chromosome sections or the regulation of the amount of recombination.” In fact, Mayr's main justification for regarding polymorphism as adaptive — a major shift in his own adaptationist hardening from the examples used to support non-selectionist claims in his 1942 book — focuses on advantages to populations (1963, p. 251).

  Polymorphism is based on and produced by definite genetic mechanisms, such as genes for differential niche selection and the heterosis of heterozygotes. A population that has not responded to selection for such mech­anisms and therefore lacks polymorphic diversity is more narrowly adapted, more specialized, and therefore more vulnerable to extermina­tion. The widespread occurrence of genetic mechanisms that produce and maintain polymorphism is directly due to selection and is in itself a component of adaptiveness. It seems appropriate, therefore, to speak of “adaptive polymorphism.”

  A rally around the flag of organismic selection, and an explicit (and vociferous) denial of higher levels, became a major movement in evolutionary theory during the 1960's. The hardening of adaptationism had occurred largely [Page 547] during the 1940's and 1950's (in time for the Darwinian centennial of 1959); but the refinement of adaptationist arguments to nearly exclusive operation at the organismic level followed later. This reform* emerged largely within the field of animal behavior, where the ethological tradition, particularly in the work of Konrad Lorenz, had long promulgated a loose and largely unconsidered approach to multilevel selection.