The Structure of Evolutionary Theory

by Stephen Jay Gould

  This paradox permits several approaches, including the following two that I would not favor. One might simply argue that the pattern of punctuated equilibrium demonstrably exists, so the task falls to evolutionary theorists to find a proper explanation. The current absence of a satisfactory account does not threaten the empirical record, but rather directs inquiry by posing a prob­lem. Or one might doubt that any single explanation can render the phenom­enon, and suspect that many rationales will yield the observed pattern (in­cluding Mayrian genetic revolutions, even if we now regard their relative frequency as low). Thus, we need to identify a set of enabling criteria from evolutionary theory, and then argue that their combination may render the observed phenomena of the fossil record.

  Most researchers would regard a third approach as preferable in science: an alternate general explanation of different form from the previous, but now rejected, leading candidate. I believe that such a resolution has been provided by Douglas Futuyma (1986,1988a and b, but especially 1987),* although his [Page 799] simple, yet profound, argument has not infused the consciousness of evolu­tionists because the implied and required hierarchical style of thinking re­mains so unfamiliar and elusive to most of us. (In fact, and with some shame, I am chagrined that I never recognized this evident and elegant resolution my­self. After all, I am supposedly steeped in this alternative hierarchical mode of thinking — and I certainly have a strong stake in the problems of punctuated equilibrium.)

  In short, Futuyma argues that we have been running on the wrong track, and thinking at the wrong level, in trying to locate the reason for a correla­tion between paleontological punctuations and events of speciation in a di­rect mechanism of accelerated change promoted by the process of speciation itself. Yet Futuyma does agree that a strong correlation exists (and has been demonstrated, in large part by research and literature generated by debate about punctuated equilibrium). Since we all understand (but do not always put into practice!) the important logical principle that correlation does not imply causality (the post hoc fallacy), an acknowledgement of the genuine link doesn't commit us to any particular causal scheme — especially, in this case, to the apparently false claim that mechanisms of speciation inherently enhance evolutionary rates.

  Futuyma begins by arguing that morphological change may accumulate anywhere along the temporal trajectory of a species, and not exclusively (or even preferentially) during the geological moment of its origin. What then could produce such a strong correlation between events of branching specia­tion and morphological change from an ancestral phenotype to the subse­quent stasis of an altered descendant? Futuyma — and I am somewhat re­phrasing and extending his argument here — draws an insightful and original analogy between macroevolution and the conventional Darwinism of natural selection in populations.

  The operation of natural selection requires that Darwinian individuals in­teract with environments in such a manner that distinct features of these indi­viduals bias their reproductive success relative to others in the population. As a defining criterion of Darwinian individuality, entities that interact with the environment must show “sufficient stability” (see discussion on pp. 611–613) — defined in terms of the theory and mechanism under discussion as enough coherence to perform as an interactor in the process of natural selec­tion.

  Darwin recognized that organisms operate as fundamental interactors for [Page 800] microevolution within populations. (Gene selectionists make a crucial error in arguing that sexual organisms are not stable enough to be regarded as units of selection because they must disaggregate in forming the next generation. But units of selection are interactors, and the “sufficient stability” required by the theory only demands persistence through one episode (generational at this level) of selective interaction to bias reproductive success — as organisms do in the classical Darwinian “struggle for existence,” see full discussion on pages 619–625.) Organisms achieve this stability through ordinary mecha­nisms of bodily coherence (a protective skin, functional integration of parts, a regulated developmental program, etc.).

  What, then, produces a corresponding stability for units of macroevolution? Species-individuals are constructed as complex units, composed of nu­merous local populations, each potentially separate (at any moment) due to limited gene flow, and each capable of adaptation to unique and immediate environments. Thus, in principle, substantial evolution can occur in any local population at any time during the geological trajectory of a species. A large and developing literature, much beloved by popular sources (media and text­books) for illustrating the efficacy of evolution in the flesh of immediacy (that is, within a time frame viscerally understood by human beings), has docu­mented these rapid and adaptive changes in isolated local populations — sub­stantial evolution of body size in guppies (Reznick et al., 1997), or of leg length in anolid lizards (Losos et al., 1997), for example (see Gould, 1997f).

  But these changes in local populations cannot gain any sustained macroevolutionary expression unless they become “locked up” in a Darwinian indi­vidual with sufficient stability to act as a unit of selection in geological time. Local populations — as a primary feature of their definition — do not maintain such coherence. They can in principle — and do, in the fullness of geological time, almost invariably in practice — interbreed with other local populations of their species. The distinctively evolved adaptations of local populations must therefore be ephemeral in geological terms, unless these features can be stabilized by individuation — that is, by protection against amalgamation with other Darwinian individuals. Speciation — as the core of its macroevolutionary meaning — provides such individuation by “locking up” evolved changes in reproductively isolated populations that can, thereafter, no longer amalgamate with others. The Darwinian individuation of organisms occurs by bodily coherence for structural and functional reasons. The Darwinian in­dividuation of species occurs by reproductive coherence among parts (organ­isms), and by prevention of intermingling between these parts and the parts of other macroevolutionary individuals (that is, organisms of other species).

  Rapid evolution in local population of guppies and anoles illustrates a fascinating phenomenon that teaches us many important lessons about the gen­eral process of evolution. But such changes can only be ephemeral unless they then become stabilized in coherent higher-level Darwinian individuals with sufficient stability to participate in macroevolutionary selection. These local populations usually strut and fret their short hour on the geological stage, and then disappear by [Page 801] death or amalgamation. They produce the ubiquitous and geologically momentary fluctuations that characterize and embellish the long-term stasis of species. They are, to use Mandelbrot's famous metaphor for fractals, the squiggles and jiggles on the coastline of Maine depicted at a scale that measures the distance around every boulder on every beach along the shore, and not at the resolution properly enjoined when the entire state appears on a single page in an atlas. Macroevolution represents the page of the atlas. The distance around each boulder (marking substantial but ephem­eral changes in local populations of guppies and lizards) — however important in the immediacy of an ecological moment — becomes invisible and irrelevant (as the transient fluctuations of stasis) in the domain of sustained macroevo­lutionary change (Fig. 9-9).

  In other words, morphological change correlates so strongly with speciation not because cladogenesis accelerates evolutionary rates, but rather be­cause such changes, which can occur at any time in the life of a local popula­tion, cannot be retained (and sufficiently stabilized to participate in selection) without the protection provided by individuation — and speciation, via repro­ductive isolation, represents nature's preeminent mechanism for generating macroevolutionary individuals. Speciation does not necessarily promote evo­lutionary change; rather, speciation “gathers in” and guards’ evolutionary change by locking and stabilization for sufficient geological time within a Darwinian individual of the appropriate scale. If a chang
e in a local popula­tion does not gain such protection, it becomes — to borrow Dawkins's meta­phor at a macroevolutionary scale — a transient duststorm in the desert of time, a passing cloud without borders, integrity, or even the capacity to act as a unit of selection, in the panorama of life's phylogeny.

  To cite Futuyma's summary of his powerful idea (1987, p. 465): “I propose that because the spatial locations of habitats shift in time, extinction of and interbreeding among local populations makes much of the geographic differentiation of populations ephemeral, whereas reproductive isolation confers sufficient permanence on morphological changes for them to be discerned in

  9-9. Stasis does not imply absolute stability, but rather directionless fluctuation that generally does not stray beyond the boundaries of graphic variation within similar species and, particularly, does not trend in any given direction, especially towards the modal morphology of descendant forms. This figure shows that, when a small segment in geological stasis becomes magnified so that change may be visualized on a generational scale, the natural fluctuations within local populations become more visible — but still do not, at the proper geological focus, exceed the bounds of stasis within the species.

  [Page 802]

  the fossil record.” Futuyma directly follows this statement with the key impli­cation of punctuated equilibrium for the explanation of evolutionary trends: “Long-term anagenetic change in some characters is then the consequence of a succession of speciation events.”

  Later in his article, Futuyma (p. 467) explicitly links speciation with suf­ficient stability (individuation) for macroevolutionary expression: “In the ab­sence of reproductive isolation, differentiation is broken down by recombina­tion. Given reproductive isolation, however, a species can retain its distinctive complex of characters as its spatial distribution changes along with that of its habitat or niche . . . Although speciation does not accelerate evolution within populations, it provides morphological changes with enough permanence to be registered in the fossil record. Thus, it is plausible to expect many evolu­tionary changes in the fossil record to be associated with speciation.” And, at the end of his article, Futuyma (p. 470) notes the crucial link between punctu­ated equilibrium and the possibility of sustained evolutionary trends: “Each step has had a more than ephemeral existence only because reproductive iso­lation prevented the slippage consequent on interbreeding with other popula­tions . . . Speciation may facilitate anagenesis by retaining, stepwise, the ad­vances made in any one direction . . . Successive speciation events are the pitons affixed to the slopes of an adaptive peak.”

  I hope that Futuyma's simple yet profound insight may help to heal the remaining rifts, thereby promoting the integration of punctuated equilibrium into an evolutionary theory hierarchically enriched in its light.

  CRITIQUES BASED UPON SUPPOSED FAILURES OF

  EMPIRICAL RESULTS TO AFFIRM PREDICTIONS

  OF PUNCTUATED EQUILIBRIUM

  I shall treat the specifics of this topic primarily in the next section on “the data of punctuated equilibrium.” But the logic of this chapter's development also requires that I state the major arguments and my responses in this ac­count of principal critiques directed at the theory — for the totality of at­tempted rebuttals has not only posited theoretical objections in an effort to undermine the theory's logic or testability (as discussed in the first two parts of this section), but has also proceeded by accepting the theory's program of research as valid, and then arguing that the bulk of data thus accumulated re­futes punctuated equilibrium empirically. I shall summarize discussion on the two major strategies pursued under this rubric: refutation by accumulation of important cases, and rejection by failure of actual data to fit models for pre­dicted phylogenetic patterns.

  Claims for empirical refutation by cases

  PHENOTYPES. Despite some early misunderstandings, long since resolved by all parties to the discussion, we recognize that no individual case for or against punctuated equilibrium, however elegantly documented, can serve as a “crucial experiment” for questions in natural history that must be decided [Page 803] by relative frequencies. No exquisite case of punctuated equilibrium — and many have been documented — can “prove” our theory; while no beautiful example of gradualism — and such have been discovered as well — can refute us. The key question has never been “whether,” but rather “how often,” “with what range of variation in what circumstances of time, taxon, and en­vironment,” and especially, “to what degree of control over patterns in phylogeny?” A single good case can only validate the reality of the phenome­non — and the simple claim for existence has not, surely, been an issue for more than 20 years. Similarly, an opposite case of gradualism can only prove that punctuated equilibrium lacks universal validity, and neither we nor any­one else ever made such a foolish and vainglorious claim in the first place. The empirical debate about punctuated equilibrium has always, and properly, focussed upon issues of relative frequency.

  I shall present the empirical arguments for asserting dominant relative frequency, rather than mere occurrence, for punctuated equilibrium on pages 854–874. If we ask, by contrast, whether strong evidence for predomi­nant gradualism has been asserted for any major taxon, time or environment, one case stands out as a potentially general refutation of punctuated equilib­rium in one important domain at least: the claim for anagenetic gradualism as a primary phylogenetic pattern in the evolution of Cenozoic planktonic Foraminifera.

  This case gains potential power and generality from the unusually favor­able stratigraphic context, and the consequent nature of sampling, in such studies. The data come from deep oceanic cores, with stratigraphic records presumably unmatched in general completeness, for these environments re­ceive a continuous supply of sediment (including foraminiferal tests) from the water column above. Moreover, these microscopic organisms can usually be extracted in large and closely spaced samples (sieved from disaggregated sedi­ments), even from the restricted volume of a single oceanic core. Thus, forams in oceanic cores should provide our most consistently satisfactory informa­tion — in terms of large samples with good stratigraphic resolution — for the study of phylogenetic pattern. If gradualistic anagenesis prevails in such situ­ations of maximal information — even if punctuated equilibrium predomi­nates in the conventional fossil record of marine invertebrates from shallow water sediments — shouldn't we then conclude that Darwin's old argument must be valid after all; that punctuational patterns represent an artifact of missing data; and that more complete information will affirm genuine gradu­alism as the characteristic signal of phylogeny?

  I acknowledge the highest relative frequency of recorded gradualism for foraminiferal data of this type, and I also admire the procedural rigor and informational richness in several of these studies. But I do not regard this case as a general argument against punctuated equilibrium — and neither, I think, do most of my paleontological colleagues, whatever their overall opin­ion about our theory, for the following reasons based upon well-known features of the fossil record in general, and the biology of forams in par­ticular. [Page 804]

  1. As emphasized in my previous discussion of publication bias (see p. 763), I remain unconvinced that a predominant relative frequency for gradual­ism — as opposed to genuine documentation of several convincing cases — has been established, even for this maximally promising taxon. No one has ever compiled an adequately random, or even an adequately numerous, sample of planktonic species drawn from the entire clade. Gradualistic lineages have been highlighted for study as a consequence of their greater “interest” under conventional views, while putatively stable lineages have tended to remain in unexamined limbo as supposedly uninformative, or even dull. Thus, the fact that gradualism prevails in a high percentage of published studies tells us little about the relative frequency of gradualism in the clade as a whole.

  A telling analogy may be drawn with a crucial episode in the history of genetics. With classical techn
iques based on the Mendelian analysis of pedi­grees, only variable genes could be identified. (If every Drosophila individual had red eyes, earlier researchers could legitimately assume some genetic bases for the invariance, but no genes could be specified because traits could not be traced through pedigrees. But once a white-eyed mutant fly appeared in the population, geneticists gained a necessary tool for identifying relevant genes by crossbreeding the two forms and tracing the alternate phenotypes through successive generations. In other words, genes had to vary before they could be specified at all.)