The Structure of Evolutionary Theory

by Stephen Jay Gould

  2. Most geographic variation in clines:

  It is difficult to see why the gradual decrease from the north to the south in the number of the bridled individuals (ringvia) in populations of the Atlantic murre (Uria aalge) should have an adaptational significance ... The convergent development in several species of Draco also seems to belong to the category of non-adaptive clines (p. 96).

  3. Much geographic variation in general:

  It should not be assumed that all the differences between populations and species are purely adaptational and that they owe their existence to their superior selective qualities... Many combinations of color pat­terns, spots, and bands, as well as extra bristles and wing veins, are prob­ably largely accidental. This is particularly true in regions with many sta­tionary, small, and well-isolated populations, such as we find commonly in tropical and insular species... We must stress the point that not all-geographic variation is adaptive (p. 86).

  Mayr's later book (1963) expanded to more than twice the number of pages, and became even more weighty in its assurances. This work shaped my own evolutionary thinking more than any other book — and I am confident that most naturalists of my generation would offer the same testimony. As I reread Animal Species and Evolution in preparing to write this chapter — and examined my old marginalia, pencilled in preparation for the deciding oral exam of my Ph.D. program — I came to appreciate even more (now that I know the genre's difficulty through personal experience) the enormous labor and creative thought involved in bringing so much material together. And I finally understood the defining word that once puzzled me in Julian Huxley's review of Mayr's book — “magisterial.” (The etymological source does not re­side in “magnificent” or “majestic,” though Mayr's book surely merits either of these accolades, but in magister, the Latin word for teacher. A great magister is not a schoolroom pedant, but a wise preceptor who holds mastery within his teaching authority, or magisterium. Magisterial, above all else, means authoritative. And to what greater virtue, after all, may an author aspire?)

  Although Mayr's 1963 book covers the same general material, and in simi­lar order, as the 1942 version, the works differ profoundly, and Mayr chose a new title (just as Simpson had done in noting the changes between his 1944 and 1953 volumes). I would specify two thematic changes as most important.

  1. The primary role of geographic isolation as a sine qua non, and the consequent near universality of allopatric speciation, has consistently formed the centerpiece of Mayr's worldview. But, in 1942, pure continuationism reigned. Populations split into roughly equal divisions and each subgroup then func­tioned as a microcosm of the ancestral mass — as in the model now called “dumb-bell allopatry” and considered (by Mayr at least) both rare and rela­tively ineffective in producing new species. In other words, Mayr (1942) orig­inally identified no distinctive properties promoting speciation in certain kinds of isolated populations vs. others. Isolation itself, and the severing of [Page 536] gene flow, rendered any population ripe for speciation: “The big gaps which we find between species are preceded by the little gaps which we find between subspecies and by the still lesser gaps which we find between populations. Of course, if these populations are distributed as a complete continuum, there are no gaps. But with the least isolation, the first minor gaps will appear” (1942, p. 159).

  But by 1963, Mayr had developed the full apparatus of the distinctive the­ory that he later called “peripatric speciation” to emphasize a sharp separa­tion from his original, continuationist version of allopatry. For the peripatric model promotes the role of small populations, isolated at the periphery of pa­rental ranges, and subject to a special maelstrom of influences including greatly enhanced selection and random effects of the founder principle — all leading to potential achievement of specific status with relative speed by a “genetic revolution.” Mayr says (personal communication) that he intro­duced this new apparatus in a paper (1954) that achieved no impact, but nonetheless represents his most important idea and best work. (Nihil sub sole novum. He published this paper in a symposium volume — the greatest reposi­tory of unread literature, both then and now.)

  2. Mayr's 1942 book included little explicit material about adaptation, since this volume emphasized the origin and development of discontinuity be­tween species, and said little about anagenetic change within populations. This context of minimal consideration reflects Mayr's pluralism and lack of commitment to strict adaptationism at this time. (This claim may sound para­doxical, but should not be so read. Views expressed in passing — by their sim­ple acknowledgment of an unchallenged belief — tend to record a professional consensus more clearly than material explicitly touted as central and distinc­tive.) But, in 1963, Mayr added a full consideration of variation and change within populations — the main reason for a much longer book. Here the hard­ened, panadaptationist position of the later Synthesis reigns supreme, per­haps more strongly than in any other book of comparable influence.

  In the mid 1990's, Mayr himself (in lift, and personal communications — see end of this section), while continuing to explicate and defend his favored themes of 1963, denies any substantial change between the volumes of 1942 and 1963 on questions of adaptation. This difference between current mem­ory and textual record, previously discussed as a general principle (see p. 521), provides a fascinating illustration of how scholars can slowly and un­consciously imbibe a shifting professional consensus, thus imposing a subjec­tive and personal impression of stability upon a virtual transmogrification. I find this unconscious alteration all the more ironic in Mayr's case because his first category of major change in ideas about speciation — his intellectual move from the dumbbell to the peripatric model — so strongly encourages a widened space for nonadaptationist themes (for many evolutionists have interpreted his notions of genetic revolutions and founder effects in small peripheral isolates as a powerful antidote to the classical panadaptationist model of Fisherian panmixia in large populations). Yet Mayr never translated the implications of these changes in his own ideas about speciation into [Page 537] doubts about adaptation in his chapters on variation and change within pop­ulations.

  No good naturalist, living in our complex universe of relative frequencies, could ever become an uncompromising dogmatist on the subject of adapta­tion. Mayr therefore mentions occasional inadaptive features (1963, p. 156), or acknowledges the importance of developmental constraint (p. 608). But these statements function more as footnotes or placeholders in the logic of an argument; for Mayr does not treat alternatives to adaptation as operational imperatives in the ordinary analysis of cases. Moreover, Mayr laces his plu­ralistic admissions with hedges and caveats. Note, for example, how Mayr frames his main admission of potential nonadaptation only as an argument against optimality, not as a denial of selection — and how his closing hedge anticipates a movement of even these least promising cases into the adaptationist camp:

  Each local population is the product of a continuing selection process. By definition, then, the genotype of each local population has been se­lected for the production of a well-adapted phenotype. It does not follow from this conclusion, however, that every detail of the phenotype is max­imally adaptive. If a given subspecies of ladybird beetles has more spots on the elytra than another subspecies, it does not necessarily mean that the extra spots are essential for survival in the range of that subspecies. It merely means that the genotype that has evolved in this area as the result of selection develops additional spots on the elytra ... Yet close analysis often reveals unsuspected adaptive qualities even in minute details of the phenotype (1963, p. 311).

  Selection holds primacy of place as the ruling force of evolution: “Every species is the product of a long history of selection and is thus well adapted to the environment in which it lives. There is no doubt that the phenotype as a whole, including its physiological properties, is adaptive and is produced by a genotype that is the result of natural selection. This is not contradicted by t
he fact that an occasional component of the phenotype is adaptively irrelevant” (1963, p. 60).

  Above all else, Mayr regards one conclusion as especially well confirmed by observation: adaptation rules in “every local population” as selection to “exacting requirements” of local environments produces an “optimal pheno­type.” One could hardly state the adaptationist position more boldly: “One conclusion emerges from these observations more strongly than any other: every local population is very precisely adjusted in its phenotype to the exact­ing requirements of the local environment. This adjustment is the result of a selection of genes producing an optimal phenotype” (1963, p. 318).

  Mayr's treatment of potential alternatives illustrates his adherence to the rule of adaptation, both as a methodological preference and an empirical claim. Geographic trends that he formerly attributed to incidental allometries have now become active adaptations: “A particularly impressive result of studies of ecogeographical rules is the discovery of the extreme sensitivity of [Page 538] body proportions to natural selection. The former belief that proportions are determined by 'built-in' allometry factors and change automatically with changes in body size is not supported by these findings” (1963, p. 324).

  Neutral genes become improbable, almost nonsensical in principle, once we recognize the pervasive monitoring of nature by selection:

  Entirely neutral genes are improbable for physiological reasons. Every gene elaborates a “gene product,” a chemical that enters the develop­mental stream. It seems unrealistic to me to assume that the nature of the particular chemical (enzyme or other product) should be without any ef­fect whatsoever on the fitness of the ultimate phenotype. A gene may be selectively neutral when placed on a particular genetic background in a particular temporary physical and biotic environment. However, genetic background as well as environment change continually in natural populations and I consider it therefore exceedingly unlikely that any gene will remain selectively neutral for any length of time (1963, p. 207).

  Consequently, even the most apparently trivial features probably origi­nated by direct selection. “One can never assert with confidence that a given structure does not have selective significance. The peculiar tarsal combs of the males in certain species of Drosophila turned out to have an important func­tion during copulation; the color patterns of Cepaea snails have cryptic sig­nificance, mitigating predator pressure” (1963, p. 190).

  In 1963, Mayr repudiated all three major classes of nonadaptation that he had defended in 1942: polymorphisms, clines, and much geographic varia­tion in general. Explicitly refuting his own former view, Mayr now argues (1963, p. 162) that the ubiquity of selection must imply an adaptive basis for polymorphisms (see also pp. 158 and 167): “Such neutral polymorphism, it was claimed, was maintained by 'accident.' Now that the cryptic physiologi­cal effects of 'neutral' genes have been discovered, it is evident that such genes are anything but selectively neutral. It is altogether unlikely that two genes would have identical selective values under all the conditions in which they may coexist in a population.”

  In a remarkable statement, he then urges that polymorphisms and clines be viewed as evidence for adaptation a priori: “Selective neutrality can be ex­cluded almost automatically wherever polymorphism or character clines are found in natural populations . . . Virtually every case quoted in the past as caused by genetic drift due to errors of sampling has more recently been rein­terpreted in terms of selection pressures” (1963, pp. 207-208).

  As for geographic variation, what else could such a phenomenon represent but adaptation to an altered environment, with selection as an efficient and omnipresent watchdog: “The geographic variation of species is the inevitable consequence of the geographic variation of the environment. A species must adapt itself in different parts of its range to the demands of the local environ­ment. Every local population is under continuous selection pressure for maxi­mal fitness in the particular area where it occurs... Each local environment [Page 539] exerts a continuous selection pressure on the localized demes of every species and models them thereby into adaptedness” (1963, pp. 311-312).

  Throughout Mayr's 1963 book — with a cadence that sounds, at times, almost like a morality play — phenomenon after phenomenon falls to the explanatory unity of adaptation, as the light of nature's truth expands into previous darkness: non-genetic variation (p. 139), homeostasis (pp. 57, 61), prevention of hybridization (p. 109). Former standard bearers of the opposi­tion fall into disarray, finally succumbing to defeat almost by definition: “It is now evident that the term 'drift' was ill-chosen and that all or virtually all of the cases listed in the literature as 'evolutionary change due to genetic drift' are to be interpreted in terms of selection” (p. 214). All particular Goliaths have been slain (although later genetic studies would revivify this particular old warrior): “The human blood-group genes have in the past been held up as an exemplary case of 'neutral genes,' that is, genes of no selective significance. This assumption has now been thoroughly disproved” (p. 161).

  However, Mayr's most interesting expression of movement towards a hardened adaptationism occurs not so much in these explicit claims for near ubiquity, but even more forcefully in the subtle redefinition of all evolutionary problems as issues in adaptation. The very meaning of terms, questions, groupings and weights of phenomena, now enter evolutionary discourse un­der adaptationist presumptions. Not only have alternatives to adaptation been routed on an objective playing field, Mayr claims in 1963, but the con­ceptual space of evolutionary inquiry has also become so reconfigured that hardly any room (or even language) remains for considering, or even formu­lating, a potential way to consider answers outside an adaptationist frame­work.

  Major subjects, the origin of evolutionary novelty for example, now reside exclusively within an adaptationist framework by purely functional defini­tion: “We may begin by defining evolutionary novelty as any newly acquired structure or property that permits the performance of a new function, which, in turn, will open a new adaptive zone” (p. 602). In a world of rapid and pre­cise adaptation, morphological similarity between distantly related groups can only arise through convergence imposed by similar adaptive regimes upon fundamentally different genetic material. The older, internalist view (constraint-based and potentially nonadaptationist) — the claim that we might attribute such similarities to parallelism produced by homologous genes — is dismissed as both old-fashioned and wrong-headed. (In modern hindsight, this claim provides a particularly compelling example of how hardened adaptationism can suppress interesting questions — for such homologues have now been found in abundance. Their discovery ranks as one of the most important events in modern evolutionary science — see Chapter 10, p. 1092, where we will revisit this particular Mayrian claim): “In the early days of Mendelism there was much search for homologous genes that would account for such similarities. Much that has been learned about gene physiol­ogy makes it evident that the search for homologous genes is quite futile ex­cept in very close relatives” (1963, p. 609). [Page 540]

  Subjects that might have seemed challenging or exceptional now achieve a place within the adaptationist framework by expanded definition. Nonfunc­tional pleiotropic consequences, for example, become an aspect of orthodoxy because they now enter a hardened discourse in the redefined guise of features subsidiary to a main effect of adaptive significance. (I do not challenge the particular assertion in this case, but I do feel that such an important subject deserves consideration from a structuralist perspective as well): “Pleiotropic gene action is the key to the solution of many other puzzling phenomena ... Color, pattern, or some structural detail may be merely an incidental by-prod­uct of a gene maintained in the gene pool for other physiological properties. The curious evolutionary success of seemingly insignificant characters now appears in a new light” (1963, p. 162).

  All potential anomalies yield to a more complex selectionist scenario, often presented as a “just-so-story.” Why did
the crown height of molars increase so slowly, if hypsodonty became so advantageous once horses shifted to vegetational regimes of newly evolved grasses with high silica content? Mayr devises a story — sensible, though empirically wrong in this case — and regards such a hypothetical claim for plausibility as an adequate reason to affirm a selectionist cause. (The average increase may have been as small as the figure cited by Mayr, but horses did not change in anagenetic continuity at constant rates. Horses probably evolved predominantly by punctuated equilibrium — see Prothero and Shubin, 1989, and Chapter 9. The average of a millimeter per million years represents a meaningless amalgam of geological moments of rapid change during speciation mixed with long periods of stasis): “An in­crease in tooth length (hypsodonty) was of selective advantage to primitive horses shifting from browsing to grazing in an increasingly arid environment. However, such a change in feeding habits required a larger jaw and stronger jaw muscles, hence a bigger and heavier skull supported by heavier neck mus­cles, as well as shifts in the intestinal tract. Too rapid an increase in tooth length was consequently opposed by selection, and indeed the increase aver­aged only about 1 millimeter per million years” (1963, p. 238).

  In 1991, I asked Ernst Mayr about changes between his 1942 and 1963 books. He acknowledged the structural alterations, of course — particularly his addition of several chapters emphasizing adaptational themes. But he strongly denied any personal augmentation of adaptationist preferences through the intervening years, citing the interesting argument that, as a Lamarckian in his evolutionary youth (well before both books), he had al­ways favored adaptationism. He even wrote me a fascinating letter the day af­ter our lunchtime conversation: