The Structure of Evolutionary Theory

by Stephen Jay Gould

  Moreover, one might even stress the bright side and recognize that such biases may exist for interesting reasons in themselves — reasons that might even enhance the importance of punctuated equilibrium and its implications. I doubt that Levinton (1988, p. 379) intended the following passage in such a positive light, but I would suggest such a reading: “One cannot rule out the possibility that speciation is rampant, but morphological evolution only oc­curs occasionally when a population is forced into a marginal environment and subjected to rapid directional selection. What then becomes interesting is why the character complexes evolved in the daughter species remain con­stant. This is, again, the issue of stasis, which I believe to be the legitimate problem spawned by the punctuated equilibrium model.”

  Finally, I am not sure that fossil species do strongly and generally underesti­mate the frequency of true biospeciation — although I do accept that a bias, if present at all, probably operates in this direction. The most rigorous empiri­cal studies on correspondence between well-defined paleospecies and true biospecies — the works of Michaux and of Jackson and Cheetham discussed above — affirm a one-to-one link between paleontological morphospecies and extant, genetically defined biospecies.

  Reasons for a potential systematic overestimation of biospecies

  by paleospecies

  If a bias did exist in this opposite direction, the consequences for punctuated equilibrium would be troubling (as implied in the previous section on accept­able and unacceptable forms of unavoidable natural biasing). For if we sys­tematically name too many species by paleontological criteria, then we might be affirming punctuated equilibrium by skewing data in the direction of our favored theory, rather than by genuine evidence from the fossil record. How­ever, I doubt that such a problem exists for punctuated equilibrium, especially since all experts — both strong advocates and fierce critics alike (as the preced­ing discussion documented) — seem to agree that if any systematic bias exists, the probable direction lies in the acceptable opposite claim for underestima­tion of biospecies by paleospecies.

  I don't doubt, of course, that past taxonomic practice, often favoring the [Page 793] erection of a species name for every recognizable morphological variant (even for odd individuals rather than populations), has greatly inflated the roster of legitimate names in many cases, particularly for fossil groups last mono­graphed several generations ago. (Our literature even recognizes the half-facetious term “monographic burst” for peaks of diversity thus artificially created. But this problem of past over splitting cannot be construed as either uniquely or even especially paleontological, for neontological systematics then followed the same practices as well.) The grossly uneven, and often greatly over split, construction of species-level taxonomy in paleontology has acted as a strong impediment for the entire research program of the promi­nent school of “taxon-counting” (Raup, 1975,1985). For this reason, the ge­nus has traditionally been regarded as the lowest unit of rough comparability in paleontological data (see Newell, 1949). Sepkoski (1982) therefore com­piled his two great compendia — the basis for so much research in the his­tory of life's fluctuating diversity — at the family, and then at the genus, level (but explicitly not at the species level in recognition of frequent over splitting and extreme imbalance in practice of research among specialists on various groups).

  Although this problem has proved far more serious for taxon-counters than for proponents of punctuated equilibrium, a potential bias towards overrepresentation also poses a threat for our theory, as Levinton (1988, p. 364) rightly recognizes: “The problem is not very new. Meyer (1878) claimed that the ability to recognize gradual evolutionary change in Micraster [a famous sequence of Cretaceous echinoids] was obscured by the rampant naming of separate species by previous taxonomists.”

  This issue would cause me serious concern — for the claim of overestimation does, after all, fall into the worrisome category of biases favoring a pre­ferred hypothesis under test — if two arguments and realities did not obvi­ate the danger. First, if supporters of punctuated equilibrium did try to affirm their hypothesis by using names recorded in the literature as primary data for judging the strength and effect of speciation upon evolutionary trends, then we would face a serious difficulty. But I cannot think of any study that utilized this invalid approach — for paleontologists recognize and generally avoid the dangers of this well-known directional bias. Punctuated equilib­rium, to my knowledge, has never been defended by taxon counting at the species level. All confirmatory studies employ measured morphometric pat­terns, not the geological ranges of names recorded in literature.

  Second, as stated above, all students of this subject seem to agree that if a systematic bias exists in relative numbers of paleospecies and biospecies, fos­sil data should be skewed in the opposite direction of recognizing fewer paleospecies than biospecies — an acceptable bias operating against the confir­mation of punctuated equilibrium.

  Reasons why an observed punctuational pattern might not

  represent speciation

  Suppose that we have empirical evidence for a punctuational event separating two distinct morphological packages regarded as both different enough to be [Page 794] designated as separate paleospecies by any standard criterion, and also genealogically close enough to support a hypothesis of direct ancestry and descent. What more do we need? Does this situation not affirm punctuated equilib­rium ipso facto?

  But critics charge (and I must agree) that such evidence cannot be persua­sive by itself, because punctuated equilibrium explicitly links punctuational patterns to events of branching speciation. Therefore, recorded punctuations produced for other reasons do not affirm punctuated equilibrium — and may even challenge the theory if their frequency be high and, especially, if they cannot be distinguished in principle (or frequently enough in practice) from events of cladogenetic branching.

  Punctuational patterns often originate (at all scales in evolutionary hierar­chies of levels and times) for reasons other than geologically instantaneous speciation — and I welcome such evidence as an affirmation of pervasive im­portance (see p. 922 et seq.) for a general style of nongradualistic change, with punctuated equilibrium as its usual mode of expression at the speciational scale under consideration in this chapter. But testable, and generally applicable, criteria have been formulated for distinguishing punctuated equi­librium from other reasons for punctuational patterns — and available evi­dence amply confirms the importance and high relative frequency of punctu­ated equilibrium.

  Of the two major reasons for punctuational patterns not due to speciation, Darwin's own classic argument of imperfection — geological gradualism that appears punctuational because most steps of a continuum have not been pre­served in the fossil record — retains pride of place by venerable ancestry. I have already presented my reasons for regarding this argument as inconse­quential (see pp. 765–774). I do not, of course, deny that many (or most) breaks in geological sequences only reflect missing evidence. But proponents of punctuated equilibrium do not base their claims on such inadequate exam­ples that cannot be decided in either direction. The test cases of our best liter­ature — whether their outcomes be punctuational or gradualistic — have been generated from stratigraphic situations where temporal resolution and den­sity of sampling can make appropriate distinctions by recorded evidence, not conjectures about missing data.

  The second reason has been highlighted by some critics, but unfairly I think, because punctuated equilibrium has always recognized the argument and has, moreover, enunciated and explicitly tested proper criteria for mak­ing the necessary distinctions. To state the supposed problem: what can we conclude when we document a truly punctuational sequence that cannot be attributed to imperfections of the fossil record? How do we know that such a pattern records an event of branching speciation, as the theory of punctuated equilibrium requires? When ancestral Species A abruptly yields to descendant Species B in a vertical sequence of strata, we may
only be witnessing an anagenetic transformation through a population bottleneck, or perhaps an event of migration, where Species B, having evolved gradualistically from Species A in another region, invades the geographic range, and abruptly wipes out its ancestor. [Page 795]

  But an appropriate and non-arbitrary criterion exists — and has been fully enunciated, featured as crucial, and subjected to frequent test, from the early days of punctuated equilibrium. We can distinguish the punctuations of rapid anagenesis from those of branching speciation by invoking the eminently test­able criterion of ancestral survival following the origin of a descendant spe­cies. If the ancestor survives, then the new species has arisen by branching. If the ancestor does not survive, then we must count the case either as indeci­sive, or as good evidence for rapid anagenesis — but in any instance, certainly not as evidence for punctuated equilibrium.

  Moreover, by using this criterion, we obey the methodological requirement that existing biases must work against a theory under test. When ancestors do not survive following the first appearance of descendants, the pattern may still be recording an event of branching speciation — hence affirmation for punctuated equilibrium. But we cannot count such cases in our favor, for the plausible alternative of rapid anagenesis cannot be disproven. By restricting affirmations to cases where ancestors demonstrably survive, we accept only a subset of events actually caused by speciation. Thus, we underestimate the frequency of punctuated equilibrium — as we must do in the face of an unresolvable bias affecting a hypothesis under test.

  In our first papers, we did not recognize or articulate the importance of tabulating cases of ancestral survival following punctuational origin of a de­scendant as a criterion for distinguishing punctuated equilibrium from other forms of punctuational change. (Both of our original examples in Eldredge and Gould, 1972, did feature — and prominently discuss — ancestral survival as an important aspect of the total pattern. We had a proper “gut feeling” about best cases, but we did not formalize the criterion.) But, beginning in 1982, and continuing thereafter, we have stressed the centrality of this crite­rion in claims for speciation as the mechanism of punctuated equilibrium. Contrasting the difference in paleontological expression between Wright's shifting balance and punctuated equilibrium by speciation, for example, I wrote (Gould, 1982c, p. 100): “Since punctuational events can occur in the phyletic mode under shifting balance, but by branching speciation under punctuated equilibrium, the persistence of ancestors following the abrupt ap­pearance of a descendant is the surest sign of punctuated equilibrium.”

  This criterion has been actively applied, in an increasingly routine man­ner (as researchers recognize its importance), in the expanding literature on empirical study of evolutionary tempos and modes in well-documented fos­sil sequences. Cases of probable anagenetic transformation have been doc­umented (no ancestral survival when good stratigraphic resolution should have recorded such persistence, had it occurred), especially in planktic marine Foraminifera, where long oceanic cores often provide unusually complete evi­dence (Banner and Lowry, 1985; Malmgren and Kennett, 1981, who coined the appropriate term “punctuated anagenesis” for this phenomenon).

  However, abundant cases of ancestral survival, and consequent punctua­tional origin of descendant taxa by branching speciation, have also been affirmed as [Page 796] illustrations of punctuated equilibrium. These examples span the gamut of taxonomies and ecologies, ranging from marine microfossils (Cronin, 1985, on ostracodes); to “standard” macroscopic marine inverte­brates (with Cheetham's famous studies of bryozoans, 1986 and 1987, as classic and multiply documented examples), to freshwater invertebrates (Wil­liamson's 1981 work on multiple events of speciation in African lake mollusks, where ancestral species reinvade upon coalescence of lakes following periods of isolation that provided conditions for speciation); to terrestrial vertebrates (Flynn, 1986, on rodents; Prothero and Shubin, 1989, on horses). I shall discuss this important issue in more detail within the forthcoming sec­tion on evidence for punctuated equilibrium (see pp. 822–874), but I have been particularly (if parochially) gratified by the increasing application of punctuated equilibrium to the resolution of hominid phylogeny. The criterion of ancestral survival has been prominently featured in this literature, as by McHenry (1994), who notes “ancestral species overlap in time with de­scendants in most cases in hominid evolution, which is not what would be ex­pected from gradual transformations by anagenesis.”

  In any case, punctuated equilibrium can be adequately and generally recognized by firm evidence linking observed punctuational patterns to branching speciation as a cause. The theory of punctuated equilibrium is eminently test­able and has, indeed, passed such trials in cases now so numerous that a high relative frequency for this important evolutionary phenomenon can no longer be denied (see Gould and Eldredge, 1993).

  CRITIQUES BASED ON DENYING EVENTS OF SPECIATION

  AS THE PRIMARY LOCUS OF CHANGE

  Once we overcome the problem of definability for species in the fossil record, punctuated equilibrium still faces a major issue rooted in the crucial subject of speciation. Punctuated equilibrium affirms, as a primary statement, that ordinary biological speciation, when properly scaled into geological time, produces the characteristic punctuational pattern of our fossil record. We must therefore be able to defend the central implication that morphological change should be preferentially associated with events of branching specia­tion. Our critics have strongly argued that such a proposition cannot be justi­fied by our best understanding of evolutionary processes and mechanisms.

  I believe that our critics have been correct in this argument, and that Eldredge and I made a major error by advocating, in the original formulation of our theory, a direct acceleration of evolutionary rate by the processes of speciation. This claim, I now think, represents one of the two most important errors that we committed in advocating punctuated equilibrium during the past 25 years. (The other error, as discussed and corrected on pages 670–673, lay first in our failure to recognize the phenomenon of species selection as dis­tinct (by hierarchical reasoning) from classical Darwinian organismic selec­tion, and then (see Gould and Eldredge, 1977) in our decision to advocate an overly broad and purely descriptive definition rather than a properly limited meaning based on emergent characters or fitnesses — see pages 656–670.) [Page 797]

  We did not urge this correlation between speciation events and morpholog­ical change in a self-serving and circular manner — i.e., only because the pat­tern of punctuated equilibrium could be best defended thereby. We did, of course, recognize the logical link, as in the following statement from Gould, 1982c, p. 87 (see also Gould and Eldredge, 1977, p. 137): “Reproductive iso­lation and the morphological gaps that define species for paleontologists are not equivalent. Punctuated equilibrium requires either that most morphologi­cal change arise in coincidence with speciation itself, or that the morphologi­cal adaptations made possible by reproductive isolation arise rapidly thereaf­ter.” But we based our defense of this proposition upon a large, and then quite standard, literature advocating a strong negative correlation between capacity for rapid evolutionary change and population size. Small popula­tions, under these models, maintained maximal prospects for rapid transfor­mation based on several factors, including potentially rapid fixation of favor­able variants, and enhancement of differences from ancestral populations by interaction of intense selection with stochastic reasons for change (particu­larly the founder effect) that can only occur with such effective speed in small populations. Large and stable populations, by the converse of these argu­ments, should be sluggish and resistant to change.

  This literature culminated in Mayr's spirited defense for “genetic revolu­tion” as a common component of speciation (first proposed in a famous 1954 article, and then defended in extenso in the 1963 book that served as the clos­est analog to a “bible” for graduate students of my generation). Since Mayr (who coined the name “founder effect” i
n this context) also linked his con­cept of “genetic revolution” to the small, peripherally isolated populations that served as “incipient species” in his influential theory of peripatric spe­ciation — and since we had invoked this theory in our original formulation of punctuated equilibrium (Eldredge and Gould, 1972) — our defense of a link between speciation and concentrated episodes of genetic (and phenotypic) change flowed logically from the evolutionary views we had embraced. Thus, we correlated punctuations with the extensive changes that often occurred during events of speciation in small, peripherally isolated populations; and we linked stasis with the expected stability of large and successful popu­lations following their more volatile and punctuational origins as small iso­lates.

  I can claim no expertise in this aspect of neontological evolutionary theory, but I certainly acknowledge, and must therefore provisionally accept, the re­vised consensus of the past twenty years that has challenged this body of thought, and rejected any general rationale for equating the bulk of evolu­tionary change with events of speciation in small populations, or with small populations in any sense. As I read the current literature, most evolutionists now view large populations as equally prone to evolutionary transformation, and also find no reason to equate times of speciation — the attainment of re­productive isolation — with acceleration in general rates of genetic or pheno­typic change (see, for example, Ridley, 1993; and Williams, 1992). (I do, however, continue to wonder whether the Mayrian viewpoint might still hold [Page 798] some validity, and might now be subject to overly curt and confident dis­missal.)

  This situation creates a paradox for our theory. The pattern of punctuated equilibrium has been well documented and shown to predominate in many situations (see pp. 822–874), but its most obvious theoretical rationale has now fallen under strong skepticism. So either punctuated equilibrium is wrong — a proposition that this partisan views as unlikely (although obvi­ously possible), especially in the face of such strong documentation — or we must identify another reason for the prominence of punctuated equilibrium as a pattern in the history of life. In our article on the “majority” (21st birth­day!) of punctuated equilibrium, Eldredge and I expressed this dilemma in the following manner (Gould and Eldredge, 1993, p. 226): “The pattern of punctuated equilibrium exists (at predominant relative frequency, we would argue) and is robust. Eppur non si muove; but why then? For the association of morphological change with speciation remains as a major pattern in the fossil record.” (Our Italian parody, missed by many readers of the original ar­ticle, alters Galileo's famous, but almost surely legendary, rebuke to the In­quisition, delivered secretly and sotto voce after he had been forced to recant his Copernican views in public: Eppur si muove — nevertheless it does move. Our parody says “nevertheless it does not move” — a reference to the over­whelming evidence for predominant stasis in the history of species, even if our original evolutionary rationale, based on population size, must be reas­sessed.)