The Structure of Evolutionary Theory

by Stephen Jay Gould

  By linking constraints of preferred developmental channels with a punctuational tempo that precludes accumulative incremental selection as the sole cause of extensive evolutionary change, this familiar argument unites the two central themes of this book — hierarchical models of evolutionary mechanics with structuralist accounts of evolutionary stasis and directionality. Several [Page 941] recent volumes have explored the growing power and prestige of this ar­gument, as provided by breakthroughs in unraveling the genetics of devel­opment (see Chapter 10), combined with classical data of allometry and heterochrony (Raff, 1996; Schwartz, 1999; McNamara, 1997; McKinney and McNamara, 1991; McKinney, 1988).

  Wray (1995) has recently summarized an emerging generality that inte­grates all components of the argument across a wide variety of organisms. His chosen title — “Punctuated evolution of embryos” — underscores the puta­tive generality of change in this mode, with punctuated equilibrium as its ma­jor expression at the level of ordinary speciation, and his proposed linkage of development and ecology as its hypothesized primary source for the rarer, but highly consequential, phenomenon of the origin of morphotypic novelty.

  Nearly two centuries of tradition proclaim the conservatism of early larval and embryonic phases of the life cycles — from von Baer's enunciation of his celebrated laws (1828, see discussion in Gould, 1977b) to the standard evolutionary rationale that formative stages of early ontogeny become virtually impervious to change because cascading consequences, even of apparently minor alterations, would discombobulate the subtle complexities of develop­ment. Recent discoveries of “deep” genetic homologies and developmental pathways among animal phyla separated for more than 500 million years (see Chapter 10) have tended to highlight this conventional view.

  But Wray (1995) summarizes several recent studies of broad taxonomic scope — with best examples from the sea urchin Heliocidaris, the frogs Eleutherodactylus and Gastrotheca, and the tunicate Mogula — all showing that “similar species have . . . modifications in a variety of crucial develop­mental processes . . . that have traditionally been viewed as invariant within particular classes or phyla” (Wray, 1995, p. 1115). These substantial changes in the development of closely related forms exhibit three common properties: (1) They usually affect traits of timing and regulation in early development, including specification of cell fates and movement of cells during gastrulation. (2) They yield substantial changes in larval forms and modes of life, but often leave the adult phenotype largely unaltered. (3) They are associated with major changes in the ecology and life history strategies of larval or early developmental forms, and involve such major changes as loss of larval feed­ing ability (the echinoderm and frog examples) or capacity to disperse (tunicates).

  Wray presents evidence that alterations in larval ecology “drive changes in development,” not vice versa. Moreover, and most importantly, compari­son of molecular and phenotypic modification shows that these “function­ally profound changes in developmental mechanics can evolve quite rapidly” (ibid., p. 1116). For species of Heliocidaris with strikingly different develop­mental mechanisms, fewer than 10 million years have elapsed since diver­gence from common ancestry. Wray draws a general and punctuational con­clusion from this evidence for ecologically driven change in mechanisms of early development — events that can occur very rapidly and do not compro­mise the conserved life styles of later development due to greater dissociation [Page 942] than traditional views would allow between successive phases of ontogeny: “Long periods of little net change, with functionally minor modifications in developmental mechanisms and larvae, seem to be the normal mode of evolu­tion. This near stasis is interrupted on occasion by rapid, extensive, and me­chanically significant changes that coincide with switches in life history strat­egy ... Rapid modifications can arise in developmental mechanisms that have been conserved for hundreds of millions of years.”

  Note the striking similarity of language (with analogs of stasis and punctuation) — and of evolutionary style in tempo and mode — between punctuated equilibrium and Wray's description of phenotypic and ecological shift at these much larger scales of morphological change and temporal extent (with the analog of stasis persisting for hundreds of millions of years). These similari­ties in style and import seem to mark a genuine conceptual “homology” based on similar structural principles regulating the nature of change in com­plex systems.

  As a final example of the fruitfulness and detailed testability of punctuational models for the origin of morphogenetic novelty, Blackburn's (1995) re­markable study of “saltationist and punctuated equilibrium models for the evolution of viviparity and placentation” deserves special notice for the au­thor's clarity in specification of hypotheses, and for his richness and rigor in attendant documentation. Blackburn treats the multiple evolution of vivipar­ity in squamate reptiles (lizards and snakes) — a much better case for studying this phenomenon than the group that most of us emphasize for parochial rea­sons (the Mammalia), for extant squamate species include many examples in all stages of the process. This taxonomic richness permits clear distinction of gradualistic vs. punctuational alternatives, and also provides sufficient data to distinguish between punctuated equilibrium and saltation as the dominant punctuational style. Moreover, and largely because the subject has been em­braced as a workable surrogate for unresolvable questions about mammalian evolution, the origin of viviparity in squamate reptiles has become a classical case, and has therefore generated an extensive literature to illustrate (how­ever unintentionally) some major biases of evolutionary argumentation. The power of Blackburn's study resides in three interrelated themes:

  1. Gradualistic scenarios have dominated the classical literature in ways that authors rarely defend, or even recognize. In particular, previous workers have assumed that three apparent stages in the “perfection” of live bear­ing must represent an actual historical sequence gradually and incrementally evolved by all lineages that reach the “last stage” — viviparity or live birth, placentation for gas exchange and water intake, and placentotrophy for em­bryonic nutrition. Blackburn writes (1995, pp. 199-201): “Viviparity and placentation in squamates have stood for over half of a century as examples of gradual evolution ... Even recent reviewers have not considered the appli­cability of alternative evolutionary models.”

  The supposed evidence for such gradualism consists largely of inferences drawn from structural series of extant forms, without affirmation, or even consideration, of an explicit phylogenetic hypothesis that the successive [Page 943] stages represent a cladistic sequence. (Even worse, a phylogenetic inference has often been based only upon the series itself — a flagrantly circular argu­ment that validates the conclusion by the hypothesis supposedly under test.) For example, extant oviparous species do vary substantially in the stage of development at which the eggs are laid — and researchers have generally as­sumed that a linear ordering of such a series must represent an evolutionary continuum “on the way” to viviparity: “The inferred continuum of develop­mental stages at oviposition among squamates commonly is interpreted as ev­idence for a gradual increase in the proportion of development occurring in the female reproductive tract” (ibid., p. 201).

  2. Blackburn marshalls an impressive array of data from a broad range of fields — taxonomy, development and geology, in particular — to affirm an alternative punctuational scenario for the evolution of live birth, with simple viviparity, placentation and placentotrophy as three distinct modes, not three way stations in a progressive sequence. (I suspect that our gradualistic biases have been particularly intrusive in this case because we unconsciously read the squamate story in a mammalian perspective that makes placentotrophy the “obvious” goal of any trend to live bearing.)

  In taxonomy, viviparity has originated more than 100 times among squamate reptiles (ibid., p. 202). But cladistic data have provided not a sin­gle case of correspondence between branching order and the four structural stages of the hypothetical tre
nd: ovipary, vivipary (live birth without pla­centation of embryos), placentation, and placentotrophy. Blackburn writes (1995, pp. 201-202): “Clines of phenotypic variation that can be invoked to support gradualistic evolution of viviparity and placentotrophy tend to be composites of unrelated species representing multiple lineages ... Despite the documentation of over 100 evolutionary origins of viviparity in squamates ... available evidence has not yet permitted construction of a single, complete phenocline of parity modes and embryonic nutritional patterns out of repre­sentatives of a single clade.”

  In structure and development, Blackburn coordinates several lines of evi­dence to argue that intermediary forms between any two stages in the hypo­thetical trend either cannot be found, or exist only rarely and in a tenuous state (because such transitional phenotypes would experience either architec­tural problems in construction or adaptive insufficiencies in function). For ex­ample, if viviparity evolved by progressive delay of oviposition, then we might expect, among extant species, “a full continuum of developmental stages... representing steps in the parallel evolutionary transformations that have occurred independently (and perhaps to different degrees) in various lin­eages” (p. 202). Instead (see Fig. 9-35), the distribution of developmental stages at oviposition shows marked bimodality, with species either depositing eggs containing embryos in the pharygula/limb bud stages (with near normal­ity or minor left skewing for this lower mode, and no right skew in the direc­tion of the putative trend) or else retaining the eggs to term and then giving birth to live young.

  Moreover, the supposed development of placentation, and then of placentotrophy, [Page 944] only after the origin of live birth also derives no support from documented intermediary stages. In the traditional view, the shell membrane between fetal and maternal tissues must thin gradually, permitting an initial function of placental organs in uptake of water and exchange of gases. Placentotrophy then evolves later “as the placental supply of nutrients first sup­plements and then supplants provision by the yolk” (p. 208). But evidence from at least 19 independent clades of viviparous squamates indicates that all “have anatomically recognizable placentae derived from both the chorioallantois and the yolk sac” (p. 208). Thus Blackburn concludes, “the exis­tence of a truly non-placental viviparous squamate has not been documented in over a century of investigation ... The universal occurrence of placentae in viviparous squamates is most consistent with the view that placental organs that accomplish gas exchange and water uptake evolve simultaneously with viviparity” (p. 209).

  Similarly, no purely lecithotrophic (yolk feeding) placental squamates have been discovered, and all viviparous forms derive at least some nutrition through the placental organs. Thus, “available data are most consistent with the hypothesis that incipient placentotrophy is a necessary correlate of vivi­parity.” The three “logical” steps of the hypothetically gradual trend become telescoped into a single structural transition, with an evident implication of punctuational origin.

  9-35. Punctuational change in the morphological evolution of lineages in squamate reptiles, Ovoviviparity does not evolve by progressive and gradualistic delay of oviposition, but rather shows marked bimodality with females either de­positing eggs with embryos in their early limb bud stages or else retaining the eggs within their body to term, and then giving birth to live young (the right mode). The existence of a few intermediary species shows that the full sequence proceeds by punctuational steps and not by full saltation.

  [Page 945]

  Blackburn supplements this empirical evidence for a punctuational divide with both structural and functional rationales for the inviability of putative intermediary stages. Viviparity without placentation may be structurally unattainable because live birth requires that the eggshell become sufficiently thin “to permit gas exchange in the hypoxic uterine environment” (p. 211) — while such reduction may entail gas and water exchange (that is, incipient placentation) as a virtually automatic consequence. Thus, Blackburn argues (p. 210), “placentation is best viewed as a necessary correlate of viviparity, not as a 'reproductive strategy' per se.”

  Intermediacy may be equally unlikely in functional and adaptationist terms as well. Both endpoints entail costs as well as putative benefits — oviparity in dangers and energetic requirements of nesting behavior, and in maternal loss of calcium in making eggshells (p. 211); viviparity in decreased maternal mo­bility, fecundity, or clutch size. A hypothetical intermediate that incurs both sets of costs — for example, the calcium drain from internal shells of hypothet­ical stage one (still too thick for placentation), combined with a heightened susceptibility to predators caused by compromised mobility — without win­ning greater compensation in combined benefits, could not compete against either end member of the supposed trend, and probably would not survive even if patterns of development permitted evolutionary access to this putatively transitional design.

  Finally, in geology, the recent origin of most viviparous lines strongly supports a punctuational inference. A few origins of squamate viviparity may date to late Mesozoic or early Cenozoic times (p. 207), but most represent Pliocene or Pleistocene events. Moreover, taxonomic distribution fully sup­ports the rapidity of full transition to placentotrophy. More than 60 percent of origins for viviparity “have occurred at subgeneric levels, and virtually all have arisen at subfamilial levels.” Several origins can be traced to populations of a single species (with other populations remaining oviparous) — for exam­ple, a Pleistocene event within Lacerta vivipara, and an origin within the past 11 to 25 thousand years within the Sceloporus aeneus complex (p. 207). The extent of structural differences between oviparous and viviparous popula­tions of these minimally distant forms (both temporally and phylogenetically) fully matches the phenotypic separation noted for the same features in cladistically distant lineages.

  3. As an indication that data of natural history can provide combined crite­ria to permit fine and testable distinctions, Blackburn has been able to reject saltation and defend punctuated equilibrium as the probable cause and tem­poral basis of this well-documented punctuational pattern. Blackburn notes that “under the punctuated equilibrium model, typical oviparity and vivipar­ity could represent regions of stasis, with prolonged oviparous egg-retention being a transitory, intermediate stage between them” (p. 206). The structur­ally well integrated and functionally well adapted end members “would con­trast with the instability of the evolutionary intermediate, and prolonged oviparous egg-retention would be a relatively short-lived (and hence scarce) pattern” (p. 206). By contrast, of course, saltational models predict the structural [Page 946] unattainability and adaptive inviability of intermediates, and therefore envisage a one-step transition (with substantial opportunity, perhaps, for later adaptive finetuning).

  In favor of punctuational origins (rapid transition between domains, based on structural properties of endpoints as well coordinated states that actively resist change, and with intermediary forms as unstable, and “driven” to­wards one of the endpoints) versus saltational events (truly sudden transition, scaled to the magnitude of the unit of change and enforced by the absolute structural inaccessibility of intermediary states), Blackburn cites several fea­tures of squamate viviparity. The putative intermediary stage of “prolonged oviparous egg retention” (p. 206), while empirically rare, structurally unsta­ble, and adaptively compromised (as discussed above), does exist as the char­acteristic form of a few species in nature, not just as a facultative or transient state of a population in transformation. As the bimodality of Figure 9-35 shows, prolonged oviparous egg retention does represent an attainable inter­mediary state between two endpoints. But few forms occupy this uncertain ground between advantageous configurations.

  In a second indication of punctuational rather than saltational change, Blackburn notes that a viable first step to intermediacy does exist in nature as a strategy that can be activated under certain environmental conditions: “fac­ultative egg-retention with co
ntinued intraoviductal development” (p. 206). Several squamate species exhibit this phenotypic flexibility in development. “Such facultative retention could provide raw material for natural selection, making more likely the evolution of a pattern in which prolonged egg-reten­tion was obligative.”

  As a third confirmed prediction, favoring punctuation over saltation when joined with the two previous arguments (but unable to make the distinction otherwise, while confuting gradualism in any case), the phenotypic similarity of oviparous and viviparous congeners affirms the relative ease and accessi­bility of transition: “As an evolutionary unstable pattern, prolonged egg-re­tention might lead to viviparity, or might revert to typical oviparity; thus the less genotypic change involved, the more probable the origin of viviparity would be” (p. 206).

  Blackburn's final paragraph (p. 212) serves as an apt reminder about the restrictive nature of gradualistic bias, and of the power and inherent proba­bility of punctuational alternatives in a world that may favor this mode of change as a general structural property of material organization at all scales: “For over 60 years, research on amniotes has assumed that gradual­istic change is the sole mechanism by which viviparity, placentation, and placentotrophy could have evolved. Future empirical and theoretical analyses of reproduction in squamates and other vertebrates should not overlook the po­tential of non-gradualistic models as explanations for evolutionary change and the biological patterns it has produced.”

  Punctuational analogs in faunas and ecosystems. If stable locations, reached by rapid movement though “perilous” intermediary ter­rain, sets the structural basis of punctuational change for the “internalities” [Page 947] of evolution in complex organic phenotypes, then a similarly episodic, rather than evenly flowing, mode of change might characterize the “externalities” of environments that regulate any coordinated evolutionary tempo among components of biotas and ecosystems. I have already considered the scale of environmental punctuation most immediately relevant to punctuated equilib­rium — the geological history of regional biotas (see pp. 916–922 on coor­dinated stasis, the turnover-pulse hypothesis, and other notions of faunal transition by coincidence of punctuational extinction and origination in a high percentage of species within a previously stable biota). But the generality of such punctuational tempos in external controls might also extend to levels both below and above the direct mechanics of speciation itself.