The Structure of Evolutionary Theory

by Stephen Jay Gould

  Finally, once we recognize gradualism as an interesting puzzle rather than a dull expectation, we may be led to “dissect” the phylogenetic “anatomy” of such trends more carefully, thus adding an operational benefit to the renewed theoretical interest. In a striking example, Kucera and Malmgren (1998) pub­lished an elegant study of morphological change in the late Cretaceous Contusotruncana lineage of planktonic forams. After several million years of sta­sis, the defining feature of “mean shell conicity” increased in a gradualistic manner (see Fig. 9-16) for 3.5 million years, beginning 68.5 million years ago, in this anagenetic lineage. [Page 837]

  The mean values of Figure 9-16 record a conventional gradualistic se­quence, but the greater detail of Figure 9-17, illustrating the morphology of all specimens, not only the means for each level, reveals fascinating details that suggest novel interpretations. In short, the range of variation, after re­maining stable during the preceding period of morphological stasis, increased rapidly during the half million year interval from 68.5 to 68.0 million years ago. The subsequent gradual trend then developed within the envelope of this expanded range — a spread in variation that had already reached its full extent at the onset of the gradualistic interval. In other words, variation increased rapidly, and the gradual trend then unfurled into the enlarged morphospace of this new range. In fact, as Kucera and Malmgren point out, the upper endpoint of variation never expands after the initial surge, and the trend in mean values records a loss of variation by removal of flattened shells at the lower end.

  I do not mention these details as a punctuational partisan trying to down­grade this example of gradualism, or to reinterpret the trend as a “mere” consequence of a punctuationally expanded range of variation. The gradual trend is both genuine and well documented — but the mapping of variation into its space gives us new insight into potential mechanisms of gradualism (while also imparting an important lesson about the significance of variation, the perils of not recording such data, and the potential for misreading pat­terns in expansion and contraction of variation as conventionally directed trends in mean values — see Gould, 1996a). The gradual trend to greater conicity in Contusotruncana probably warrants a conventional selectionist explanation,

  9-16. From Kucera and Malmgren (1998). A good example of gradualism for mean shell conicity in the planktonic foram Contusotruncana. After several million years of stasis, this trait increases in a grad­ualistic manner for 3.5 million years beginning 68.5 million years ago.

  [Page 838]

  9-17. When we plot the variation for al! specimens abstracted as mean values in Figure 9-16, a fascinating pattern emerges (from Kucera and Malmgren, 1998), the variation remained stable during stasis in the ancestral lineage. Variation then increased rapidly during the half million year interval from 68.5 to 68.0 million years ago. The subsequent gradual trend then developed within the envelope of this expanded range — a spread in variation that had already reached its full extent at the outset of the gradualisric interval. The change in mean values remains gradualistic, but this addition of data on variation provides a new perspective upon the mechanism.

  [Page 839]

  in part — shifting means resulting from selective removal of disadvantaged flatter specimens. But we also need to understand the potentiating condition established by an initial (and geologically rapid) expansion in the range of variation. What mechanisms underlie such change in a variational spectrum? Evolution can't anticipate future needs for altered means, so the enlarged range can only be exaptive for the subsequent trend. What, then, lies behind such rarely documented (but eminently testable) expansions and con­tractions of variational ranges?

  THE PUNCTUATIONS OF PUNCTUATED EQUILIBRIUM: TEMPO AND MODE IN THE ORIGIN OF PALEOSPECIES

  Stasis is data, and potentially documentable in any well-sampled series of persistently abundant fossils spanning a requisite range of time — i.e., most of the duration of an average species in a given taxon, ranging roughly from a mil­lion years or so for such rapidly evolving forms (or perhaps just more closely scrutinized, or richer in visibly complex characters) as ammonites and mam­mals to an average of 5 to 10 million years for “conventional” marine inver­tebrates. But punctuation may only record an absence of intermediary data. Thus, as noted several times before, the second word of our theory stands more open to general test than the first, and this operational constraint inevi­tably skews the relative abundances of published information.

  If punctuational claims were truly untestable, or subject to empirical documentation only in the rarest of special circumstances, then the entire theory would be severely compromised. Fortunately, many cases at the upper end of a spectrum in richness of data (both in abundance of specimens and fine-scale temporal resolution) offer adequate materials for distinguishing the causes and modalities of punctuation. The testable cases do not nearly approach a majority of available species, but neither do they stand out as preciously unusual. Thus, we face a situation no different from most experimental test­ing in science, especially when we cannot construct ideal conditions in a labo­ratory, and must use nature's own “experiments” instead. That is, we must pick and choose cases with adequate information for resolution, and with­out inherent biases that falsely presuppose one solution over others. But experimentalists in “hard science” seek the same unusual resolvability when they “improve” upon the ordinary situations of nature by establishing fixed, simplified, and measurable circumstances in a laboratory. Natural histori­ans proceed no differently, and with no more artificiality or rarity of ac­ceptable conditions for testing — except that we must ask nature to set the controls (and must therefore live by her whims rather than our manipula­tions. In this sense, the naturalist's tactic of “choosing spots” selectively cor­responds with the experimentalist's strategy of establishing controls in labo­ratories).

  The testing of punctuations has generated two primary themes of research: the establishment of criteria for distinguishing among the potential causes of [Page 840] literal punctuations in the fossil record; and the specification of fine-scale “anatomy” (data of timing, mode, morphology, geography, etc.) for punctuational events.

  The inference of cladogenesis by the criterion of ancestral survival

  I doubt that any professional paleontologist would dispute the statement that a great majority of paleospecies makes a geologically abrupt first appearance in the fossil record. But this statement about an observed, literal pattern car­ries almost no interpretive weight because the phenomenon so described can be explained by such a wide variety of putative causes, including the follow­ing distinctly different proposals:

  The traditional gradualist view. The species arose by geologi­cally gradual transformation of an ancestral population, but our woefully im­perfect fossil record did not preserve the intermediary stages.

  The cladogenetic proposal of punctuated equilibrium. The literal record represents the expected geological scaling of biological pro­cesses responsible for the origin of species. New species arise by isolation and branching of a segment of the ancestral population. The branch evolves to a new species by continuous transformation, but at a rate that, however “slow” by the inappropriate standard of a human lifetime, runs to comple­tion during the “geological moment” of a bedding plane in most cases.

  Punctuated anagenesis. The new species arises by continuous transformation, in toto and without branching, of an ancestral species, but at a rate too rapid for geological resolution of intermediary stages.

  Sudden appearance by migration into a local section. The new species arose in another region at a rate and mode that cannot be de­termined from local evidence. A punctuational first appearance in any partic­ular geographic region records a process of migration from another area of earlier origin.

  These four proposals have strikingly different implications for the valida­tion of punctuated equilibrium. The first opposes punctuated equilibrium unambiguously and would disprove the theory, o
r at least consign it to irrele­vancy as a cause of pattern in the history of life, if this mode of classical gradualism could be affirmed at dominant relative frequency for the origin of new species.

  Punctuated equilibrium predicts that the second explanation must hold as the primary generator of the dominant empirical signal of punctuational ori­gin for paleospecies. If most species did not arise by rapid cladogenesis at ap­propriate geological scales, then punctuated equilibrium would be disproven as a major cause of evolutionary pattern (and would be relegated to a status of marginality and insignificance in the history of life).

  The third explanation may fall within the “spirit” of punctuated equilib­rium, by identifying a genuine geological punctuation, rather than a false ap­pearance based on missing data in gradualistic sequences, as the source for an empirical observation of abrupt origin. But if punctuated anagenesis could be validated at high relative frequency for the origin of paleospecies, then punctuated [Page 841] equilibrium — a theory about cladogenesis — would be demoted or ne­gated, while the important ancillary concept of explaining trends, within a hierarchical model (see Chapter 8), as differential success of species within clades, would also become marginalized.

  The fourth, or migrational, alternative may resolve a local issue in a given section, but can only indicate, for questions about tempo and mode of speciation, a need for additional information of wider geographic scope. For we must still learn whether the new species, arriving as a punctuational migrant, arose by anagenesis or cladogenesis, and at either a gradual or punctuational tempo, in its natal area. But if the migrant invades the territory of a surviving ancestor — a common pattern in recorded literature — then, at least, we have documented the cladogenetic origin that punctuated equilibrium requires.

  Against the charge that our theory cannot be adequately tested, partic­ipants in the empirical debate about punctuated equilibrium have long rec­ognized, and generally utilized, an excellent criterion possessing the two car­dinal virtues of a probing agent for scientific hypotheses: ready (and unambiguous) application in most cases, and an inherent bias against punctu­ated equilibrium by underrepresentation of actual cases. I presented this tool — ancestral survival following punctuational origin — earlier in the chap­ter (see pages 793–796), while leaving the primary documentation for this section.

  The criterion of ancestral survival invokes paleontological data of the most conventional and easily acquired kind — specimens in local sections, forming samples of sufficient size for basic taxonomic identifications — and not distant inferences from models or from fossil data to unobservable correlates in be­havior or physiology. Moreover, the criterion is properly biased against punc­tuated equilibrium in recognizing only the subset of legitimate cases with doc­umented ancestral survival (therefore leaving in limbo all genuine cases where ancestors may have survived in other regions, or may not yet have been found). Any tabulation based on the criterion of ancestral survival must therefore underestimate the true relative frequency of punctuated equilib­rium.

  One potential biasing factor, however, might lead to an overestimate for punctuated equilibrium under this criterion, and must therefore be scruti­nized and avoided. Under the fourth explanation presented above for literal observations of punctuation between a descendant and a surviving ancestor in a local section, migration of the descendant from a different region (where it might have originated gradually), rather than punctuational evolution in situ, could produce an artificial boost in frequency if falsely counted as a proven case of punctuated equilibrium. (I suspect that most cases in this mode do represented punctuated equilibrium, based on general arguments that most speciation events in unobserved regions of the geographic range will themselves be punctuational, but we obviously cannot count these exam­ples favorably, because our entire case would then become circular by assum­ing the premise supposedly under test.)

  The proper solution to such unresolvable cases lies in proper scrutiny, and [Page 842] in declining to count them as proven support for punctuated equilibrium. Most paleontologists recognize and follow this recommended practice. For example, to cite three titles from opposite ends of the conventional taxonomic spectrum, Sorhannus (1990) could not determine whether the punc­tuational origin of the diatom Rhizosolenia praebergonii from ancestral R. bergonii 2.9 million years ago in the Indian Ocean occurred in situ or by mi­gration from the central Pacific. He entitled his article: “Punctuational mor­phological change in a Neogene diatom lineage: 'local' evolution or migra­tion?” Schankler (1981), as previously reported, attributed punctuational patterns of Eocene condylarth Phenacodus to probable migration, and called his paper: “Local extinction and ecological re-entry of early Eocene mam­mals.” And Flynn (1986) documented an excellent case of ancestral survival in Miocene rodents from Pakistan (in a group frequently cited for high rela­tive frequencies of gradualism), but couldn't distinguish evolution in situ from migration as the cause of observed cladogenesis. He therefore only cited the literal pattern itself in his title: “Species longevity, stasis, and stairsteps in rhizomyid rodents.”

  Among affirmations of punctuated equilibrium by the criterion of ancestral survival, and ordering my discussion along a conventional taxonomic spec­trum (for no reason beyond antiquated custom), Wei and Kennett's classic study (1988) illustrates how geographic data can be integrated with vertical sequences to resolve evolutionary modes not deducible from data of single sections. These authors showed that the upper Miocene planktonic foram Globorotalia (Globoconella) conomiozea terminalis evolved gradually into G. (G.) sphericomiozea during a 0.2 million-year interval in central parts of its geographic range.

  At the same time, intensification of the Tasman Front (Subtropical Divergence) separated peripheral populations of the warm subtropics from the central stock. The isolated population then branched rapidly into a new species, G. (G.) pliozea, in less than 0.01 million years, or 5 percent of the time taken for anagenetic transformation of the ancestral stock at the center of its range. The anagenetic trend proceeded in a direction (loss of keel and development of a more conical test) opposite to the morphological innovations (flattened test and more pronounced keel) of the allopatrically speciating peripheral form. The new species, following its punctuational origin, persisted in stasis for more than a million years. About halfway through this interval, a descen­dant of the central stock migrated into the warm subtropical region of G. (G.) pliozea. The two species then coexisted for half a million years without apparent intermixing, and with no interruption of stasis.

  The rich data of microfossils from oceanic cores, often providing good resolution for both geographic and temporal variation, have also documented punctuational speciation (usually allopatric) with ancestral survival in sev­eral other cases. Cronin (1985) correlated the punctuational origin of six spe­cies in the ostracode Puriana with changes in oceanographic circulation en­gendered by the Pliocene rise of the Isthmus of Panama. Cronin comments [Page 843] (p. 60) that “since speciation occurred, ancestral species and their descen­dants have coexisted, in some cases sympatrically.” In a study of Miocene deep-sea ostracodes from the southwest Pacific, Whatley (1985, p. 109) docu­mented two cases of allopatric and punctuational origin for new species fol­lowed by migration back to the parental range and subsequent coexistence with the ancestral species.

  Alan Cheetham's work on American Cenozoic clades of the cheilostome bryozoan genera Metrarabdotos and Stylopoma (Cheetham, 1986, 1987; Jackson and Cheetham, 1994; Cheetham and Jackson, 1995) merits citation at several points in this chapter for its unparalleled documentation of all major tenets of punctuated equilibrium — both in clarity of conclusions and richness of empirical evidence. I present a general summary in the section on relative frequency (p. 868), but Cheetham's fruitful use of ancestral per­sistence should be noted here. Jackson and Cheetham (1995, p. 204) cite three primary empirical sources for documenting punctuated equilibrium from paleontological data: “The geologically ab
rupt appearance of species in the record, the static morphologies of species for millions of years, and the extensive temporal overlap between apparent ancestor-descendant species pairs.”

  Their summary of overwhelming support for punctuated equilibrium from the last source (Jackson and Cheetham, 1994, p. 407) states that “most well-sampled Metrarabdotos and Stylopoma species originated fully differentiated morphologically and persisted unchanged for > 1 to > 16 million years, typi­cally alongside their putative ancestors.”

  On Cheetham's celebrated and frequently reprinted diagram of evolution and cladogenesis in the Metrarabdotos clade (Fig. 9-18, and redundant in cit­ing “evolution and cladogenesis” because all phyletic change occurs by clado­genesis in this lineage), ancestors persist after the origin of descendants in 7 of the 9 cases where Cheetham felt confident enough to assert a phylogenetic claim for direct filiation. (Marshall's important challenge (1995) to assess­ments of stratigraphic range in several cases does not counter Cheetham's hy­potheses about filiation, and certainly does not challenge the assertion of overlap, a claim based on direct observation of joint occurrence, not on in­ference.) The two cases where ancestral persistence has not been directly observed (see Fig. 9-18), but may well have occurred (the derivation of M. tenue from sp. 10, and of M. unguiculatum from M. lacrymosum), both fall “outside the interval of dense sampling” (Cheetham, 1986, p. 201), where Cheetham achieved a stratigraphic resolution by Sadler's (1981) criteria of 0.63. For Stylopoma, “eleven of the nineteen species originate fully formed at p >