The Structure of Evolutionary Theory

by Stephen Jay Gould

  Mayr's version of allopatry fit the paleontological pattern of punctuation and stasis particularly well. If most new species arise from small populations peripherally isolated at the edges of a parental range, then we cannot expect to document a gradual transition by analyzing the stratigraphic sequence of samples for a common species. For we will usually be collecting from the population's central range during its period of stability. Daughter species [Page 780] originate in three circumstances that virtually guarantee a punctuational ex­pression in the fossil record: (1) they arise rapidly (usually instantaneously) in geological time, and they originate both (2) in a small geographic region (the peripheral isolate), and (3) elsewhere (beyond the borders of the parental range that provides the exclusive source for standard paleontological collec­tions). The “sudden” entrance of a daughter species into strata previously oc­cupied by parents usually represents the inward migration of a peripheral iso­late, now “promoted” by reproductive isolation to full separation, not the origin of a new species in situ.

  Eldredge and I have often been asked what we think of sympatric spe­ciation, or of various models, like polyploidy, for rapid origin even in hu­man time. We do not mean to be evasive or obscure in our assertions of ag­nosticism. (I am intensely interested in the literature on speciation, and I would love to know the relative frequencies of these other models vs. classical Mayrian peripatry. But this important issue does not strongly impact punctu­ated equilibrium, and surely cannot be resolved by paleontological data.) Punctuated equilibrium simply requires that any asserted mechanism of spe­ciation, whatever its mode or style, be sufficiently rapid and localized to ap­pear as punctuation when scaled into geological time. If I understand them correctly, most alternative models to peripatry generally operate even more rapidly than the conventional Mayrian mode that we invoked to anchor our theory — as obviously true for polyploidy, and also for most versions of sympatric speciation (if only because the constant threat of dilution by gene flow from surrounding parentals can best be overcome by rapid achievement of reproductive isolation in ecological time). Therefore, punctuated equilibrium can only gain strength if these alternative mechanisms become validated at meaningful relative frequencies. (The faster the better, one might say.) But punctuated equilibrium does not require this boost — and we therefore remain agnostic — because the most conventional form of Mayrian peripatry already yields the full set of phenomena predicted by punctuated equilibrium when properly scaled into the immensity of geological time. (Punctuated equilib­rium, on the other hand, does not maintain a similar agnosticism towards any putative mechanism of speciation that conceives the process of splitting as no more rapid than imagined rates for the gradual anagenesis of large central populations. Some models of so-called “dumbbell allopatry” — or the split­ting of a parental population into two effectively equal moieties, with subse­quent anagenesis in each — do construe speciation as consequently slow in geological expression, and therefore do threaten punctuated equilibrium. But I do not think that such models enjoy much support among biologists, espe­cially for operation at a high relative frequency.)

  Geological time can be both a wonder and a snare because we grasp the idea in our heads (all scientists know how many zeroes follow the one in ex­pressing millions or billions), but we face a primal, and fundamentally psy­chological, difficulty in trying to incorporate this central concept into the guts of our intuition. We can lose information in upward scaling when glacial slowness in human history becomes a passing and unresolvable geological [Page 781] moment. But we can also gain when operational invisibility at our scale (in­ability to distinguish a small effect from measurement error) becomes palpa­ble and prominent in the large, or when the almost inconceivable rarity of an event that averages one expression in ten thousand years achieves guaranteed repetition across millions.

  MACROEVOLUTIONARY IMPLICATIONS

  If punctuated equilibrium has broader utility beyond the reform of paleontological practice, then we must look to potential implications for macroevolu­tionary theory, and for consequent enrichment in our general understanding of mechanisms that regulate the history of life. I have linked my treatments of punctuated equilibrium and the hierarchical theory of natural selection to form the longest section of this book (presented as two chapters, 8 and 9) be­cause I believe that punctuated equilibrium supplies the central argument for viewing species as effective Darwinian individuals at a relative frequency high enough to be regarded as general — thereby validating the level of species as a domain of evolutionary causality, and establishing the effectiveness and inde­pendence of macroevolution by two of the three criteria featured throughout this book as indispensable foundations of Darwinism.

  First, punctuated equilibrium secures the hierarchical expansion of selectionist theory to the level of species, thus moving beyond Darwin's preference for restricting causality effectively to the organismic realm alone (leg one on the essential tripod). Second, by defining species as the basic units or atoms of macroevolution — as stable “things” (Darwinian individuals) rather than as arbitrary segments of continua — punctuated equilibrium precludes the ex­planation of all evolutionary patterns by extrapolation from mechanisms operating on local populations, at human timescales, and at organismic and lower levels (leg three on the tripod of Darwinian essentials). Thus, as em­phasized in the last section, punctuated equilibrium presents no radical pro­posal in the domain of microevolutionary mechanics — in particular (and as so often misunderstood), the theory advances no defenses for saltational models of speciation, and no claims for novel genetic processes. Moreover, punctuated equilibrium does not attempt to specify or criticize the conven­tional mechanisms of microevolution at all (for punctuated equilibrium emerges as the anticipated expression, by proper scaling, of micro-evolutionary theories about speciation into the radically different domain of “deep,” or geological time). But punctuated equilibrium does maintain, as the kernel of its potential novelty for biological theory, that these unrevised microevolutionary mechanisms do not hold exclusive sway in evolutionary explanation, and that their domain of action must be restricted (or at least shared) at the level of macroevolutionary pattern over geological scales — for punctuated equilibrium ratifies an effective realm of macroevolutionary mechanics based on recognizing species as Darwinian individuals. In other words, punctuated equilibrium makes its major contribution to evolutionary theory, not by revising microevolutionary mechanics, but by individuating [Page 782] species (and thereby establishing the basis for an independent theoretical do­main of macroevolution).

  As discussed in Chapter 8 (see pp. 648–652), punctuated equilibrium wins this role by refuting Fisher's otherwise decisive argument for the impotence (despite the undeniable existence) of species selection. So long as most new species arise by branching (speciation) rather than by transformation (ana­genesis), species can be individuated by their uniquely personal duration, bounded by birth in branching and death by extinction. But if anagenesis, fueled by Darwinian organismic selection, operates to substantial effect dur­ing the lifetimes of most species, then, by Fisher's argument, such micro-evolutionary transformation must overwhelm species selection in building the overall pattern of macroevolutionary change — for the number of organ­ism-births must exceed species-births by several orders of magnitude, and if every event of birthing, at each level, supplies effective variation for evo­lutionary transformation, then the level of species can contribute virtually nothing to the totality of change. But if stasis rules and anagenesis rarely oc­curs, then speciation becomes the more effective level of evolutionary varia­tion. And if speciation unfolds in geological moments, then species in geologi­cal time match organisms on our ordinary yearly scales in both distinctness and discreteness. Thus, the pattern of punctuated equilibrium establishes spe­cies as effective individuals and potential Darwinian agents in the mecha­nisms of macroevolution.

  In summary, G. G. Simpson gave a singularly appropriate
title to his ep­ochal 1944 book that defined the potential of paleontology to devise insights about evolutionary mechanisms: Tempo and Mode in Evolution. If we accept Simpson's focus on tempo and mode as primary subjects, then punctuated equilibrium has provoked substantial revisions of macroevolutionary theory and practice in both domains.

  Tempo and the significance of stasis

  For tempo, punctuated equilibrium reverses our basic perspective. We must abandon our concept of constant change operating within a sensible, stately range of rates as the normal condition of an evolving entity. We must then reformulate evolutionary change as a set of rare episodes, short in duration relative to periods of stasis between. Stability becomes the normal state of a lineage, with change recast as an infrequent and concentrated event that, nonetheless, renders phylogeny as a set of summed episodes through time. The implications of this fundamental shift resonate afar by impacting a set of issues ranging from the most immediately practical to the most broadly philo­sophical (including, in the latter category, an interesting consonance with the atomism and quantization invoked to define the general intellectual move­ment known as “modernism” — as expressed in disparate disciplines from Seurat's pointillism in art, to Schonberg's serial style in music; and as opposed to the smooth continuationism favored by earlier mechanistic views of cau­sality). In a theme more immediately relevant to biology, the same shift ineluctably places much greater emphasis upon chance and contingency, rather [Page 783] than predictability by extrapolation — for the ordinary condition of stasis provides little insight into when and how the next punctuation will occur, whereas the fractal character of gradualism suggests that causes of change at any moment will, by extrapolation, predict and explain the larger effects ac­cumulated through longer times.

  On the practical side, punctuated equilibrium's formulation of tempo has validated the study of stasis — paleontology's prevalent pattern within spe­cies — as a source of insight about evolution, rather than a cause of chagrin best bypassed and ignored as a testimony to an embarrassing poverty of evi­dence. Punctuated equilibrium has broken “Cordelia's Dilemma” of silence about the supposed “nothing” of stasis, and has established a burgeoning subfield of research in the documentation of stability at several levels. In pur­suing and valuing this documentation, scientists then feel compelled to postu­late explanations for the puzzling frequency of this previously “invisible” phenomenon — and theoretical inquiry about the “why” of stasis has also flourished following the prod from punctuated equilibrium (see pp. 877–885 for fuller discussion).

  Mode and the speciational foundation of macroevolution

  For mode, as discussed throughout this chapter, punctuated equilibrium has established a speciational basis for macroevolution. By supplying crucial data and arguments for defining species as effective Darwinian individuals — that is, as basic units for describing macroevolution in Darwinian terms as an out­come of patterns in differential birth and death of species treated as stable in­dividuals, just as microevolution works by the same process applied to births and deaths of organisms — punctuated equilibrium validates the hierarchical theory of selection. This hierarchical theory (explicated in Chapter 8) es­tablishes the independence of macroevolution as a theoretical subject (not just as a domain of description for accumulated microevolutionary mechan­ics), thereby precluding the full explanation of evolution by extrapolation of microevolutionary processes to all scales and times.

  In practical terms, the implications of punctuated equilibrium for evolu­tionary mode have strongly impacted two prominent subjects, heretofore al­most always rendered by extrapolation as consequences of adaptation within populations writ large: evolutionary trends within clades, and relative wax­ing and waning of diversity within supposedly competing clades through time. Punctuated equilibrium suggests novel, and irreducibly macroevolu­tionary, explanations for both phenomena (see pp. 885–916).

  Finally, the role of punctuated equilibrium in establishing an independent field of macroevolution includes both a weak and a strong version. The first, undoubtedly valid as a generality, “uncouples” macro from microevolution as a descriptive necessity, while not establishing independent causal principles of macroevolution. The second clearly regulates many cases, but has not yet been validated as commanding a high relative frequency; this second, or strong, version establishes irreducible causal principles of macroevolution.

  The weak version, based on “species sorting” rather than “species selection,” [Page 784] holds that evolution must be described as differential success in birth and death of stable species, but allows that the causality behind reasons for differential success might emerge from the conventional Darwinian level of struggling organisms within successful populations — the effect hypothesis of Vrba (see p. 658). In this version, we need a descriptive, but not a causal, ac­count of macroevolution based on species as individuals.

  However, in the strong version, based on true species selection, the differ­ential success of species arises from irreducible fitness defined by the interac­tion of species-individuals with their environments. Chapter 8 presents an ex­tensive argument for the efficacy of true species selection at high relative frequency. Validation of this argument would establish a genuinely causal and irreducible theory of macroevolution. This difficult issue stands far from resolution, but represents the most exciting potential for punctuated equilib­rium as an impetus in formulating a revised structure for evolutionary theory.

  The Scientific Debate on Punctuated Equilibrium:

  Critiques and Responses

  CRITIQUES BASED ON THE DEFINABILITY OF

  PALEONTOLOG1CAL SPECIES

  Empirical affirmation

  The issue of whether true biospecies (or entities operationally close enough to biospecies) can be recognized in fossils has prompted long and intense debate in paleontology (see Sylvester-Bradley, 1956, and other references previously cited), and does not represent a new or special difficulty raised by punctuated equilibrium. But given the reliance of punctuated equilibrium on speciation as the mechanism behind the pattern, this old problem does legitimately as­sume a central place in debates about our theory (as emphasized in all nega­tive commentary, particularly clearly by Turner, 1986, and in the book-length critiques of Levinton, 1988, and Hoffman, 1989).

  At least we may begin by exposing the canonical issue of the older litera­ture as a Scheinproblem (literally an “appearance problem” with no real content): the logical impossibility of defining a species boundary within a gradualistic continuum (see my previous discussion on p. 775). I think we may now accept that the punctuational pattern exists at high relative fre­quency, and that few gradualistic and anagenetic continua have been docu­mented between fossil species. Turner's (1986) sharp critique, for example (and I do accept his formulation, though not his resolution), depicts the chief claims of punctuated equilibrium as a three-pronged fork. He accepts the first tine — the existence of the punctuational pattern itself — as sufficiently demon­strated by enough empirical cases in the fossil record. He regards the third tine — macroevolutionary invocation of the theory to explain trends by spe­cies sorting — as “an important extension of evolutionary theory into a hith­erto little explored territory” (1986, p. 206). But he then rejects the second [Page 785] tine as both unlikely and too difficult to test in any case — explanation of the punctuational pattern as a consequence of speciation scaled into geological time.

  If we accept that temporal sequences of fossils generally don't appear in the geological record as unbreakable continua, but usually as morphologi­cal “packages” with reasonably defined boundaries and sufficient stability within an extended duration, how can we assert that these packages represent biospecies, or at least that they approximate these neontologically defined units with sufficient closeness to bear comparison? After all, we cannot apply conventional tests of observed ecological interaction or interbreeding to fos­sils — and, whereas biospecies may
be recognized by morphological differ­entia in everyday practice, they are not supposed to be so defined. Can the temporally extended “morphospecies” of paleontology really be equated with the “nondimensional species concept” (Mayr's words) of neontology?

  I certainly accept the centrality and difficulty of these issues, but I do not regard them as insuperable, and I do not view the species concept as untestable with fossils. After all, the overwhelming majority of modern species in our literature and museum drawers have also been phenotypically, not ecologically, defined. Once we accept that no special paleontological riddles arise from the Scheinproblem of temporal continua, and then most paleospecies have been no worse characterized than the majority of neospecies. Still, I will not advance this excuse as exculpatory for the fossil record, for a neontologist could reply, with impeccable logic, that neospecies so defined should also be regarded as uncertain, if not vacuous, and that no paleontological de­fense can be mounted by arguing that ordinary practice with fossils follows the worst habits (majoritarian though they may be) of neontological tax­onomy.

  But a best defense of phenotypically defined neospecies would follow from demonstrations that taxa so established usually do match true biospecies upon proper behavioral and ecological study — a line of research often pur­sued with success (see references in Jackson and Cheetham, 1994, and in Jablonski, 1999). Similarly, my main source for confidence about paleo­species arises from proven correspondences with true biospecies in favorable cases providing sufficient information for such a test (particularly for extant species with lengthy fossil records). I do not, of course, argue that all named paleospecies are true biospecies, or that I can even estimate the percentage properly so defined (any more than we know the relative frequency of mod­ern taxa that represent true biospecies). But I do not see why the probability that well-defined paleospecies, based on good collections from many times and places, might represent proper biospecies should be any lower than the corresponding figure for equally well documented, but entirely morphologi­cally defined, modern taxa. (In fact, one might argue that well-documented paleospecies probably maintain a higher probability for representing bio­species, because we know their phenotypes, and have measured their stability, across long periods of time and wide ranges of environment — whereas mod­ern “morphospecies” may arise as ecophenotypic expressions of a single time [Page 786] and place, therefore ranking only as local populations, rather than true spe­cies.)