The Structure of Evolutionary Theory

by Stephen Jay Gould

  Therefore, under these methodological constraints (which prevailed during most of the 20th century history of genetics), a dominant measured frequency for variable genes taught us nothing about the actual frequency of variable genes across an entire genome — for we knew no way to generate a random or unbiased sample by selecting genes for study prior to any knowledge about whether or not they varied. The fact of variation in all known genes only recorded a methodological limitation that precluded the identification of nonvariable genes.

  I don't, of course, claim that methodological strictures on paleontological lineages have ever been so strong — that is, we could always have selected sta­ble lineages for study, had we chosen to do so. But, in practice, I'm not sure that the actual procedural bias has operated with much less force in paleon­tology than in genetics, so long as researchers confined their attention to lin­eages that appeared (by initial qualitative impression) to evolve by gradual anagenesis. Just as all known genes might be variable (while variable genes actually represent only a few percent of the total complement, because the re­maining 95 percent of invariant genes could not be recognized at all), most studied species might illustrate gradual trends (while gradualistic species rep­resent a small minority of all lineages because no one chooses to study stable species).

  Genetics resolved this problem by inventing techniques — with electrophoresis as the first and historically most important — for identifying genes prior to any knowledge about whether or not they varied. This methodological ad­vance permitted the resolution of several old and troubling questions, most notably the calculation of average genetic differences among human races. [Page 805] This central problem of early Mendelian genetics could never be addressed — even to counter the worst abuses of biological determinism and social Dar­winism — because biologists could not generate random samples of genes, and could therefore only overestimate average distances by ignoring the unknow­able invariant genes among races, while studying the (potentially small) frac­tion capable of recording differences among groups. With electrophoretic techniques, and the attendant generation of a random sample with respect to potential variability, geneticists soon calculated the average genetic differ­ences among races as remarkably small and insignificant — a conclusion of no mean practical importance in a xenophobic world. Similarly, a truly random sample (with respect to the distribution of anagenetic rates) might show predominance for stasis, even if previous studies (with their strong bias for preselection of variable species) had generally affirmed gradualism.

  I am encouraged to accept the probable validity of this argument by the important study of Wagner and Erwin (1995), who used the different and comprehensive technique of compiling full cladograms for two prominent Neogene families of planktonic forams: Globigerinidae and Globorotaliidae. In applying a set of methods for inferring probable evolutionary mode from cladistic topology (see full discussion and details on pp. 820–822), they found that, in both families, branching speciation in the mode favored by punc­tuated equilibrium (divergence of descendants with survival of ancestors in stasis) vastly predominated over the origin of new species by anagenetic transformation. Thus, the literature's apparent preference for anagenesis in tabulated studies of individual lineages may only record an artifact of biased selection in material for research.

  2. Even if gradualism truly does prevail in planktonic forams, we could not infer that the observed predominance of punctuated equilibrium in marine Metazoa must therefore reflect the artifact of an imperfect geological record. The difference might record a characteristic disparity between the taxa, not a general distinction in quality of geological evidence between deep oceanic cores and conventional continental sequences — a proposition defended in the third argument, just below. The deep oceanic record may usually be more complete, but the subset of best cases from conventional sequences surely matches the foram data in quality — and convincing studies of punctuated equilibrium and gradualism generally use these best records. Thus, the subset of most adequate metazoan examples should match, in quality of evidence, the usual records of forams from oceanic cores.

  3. A third argument completes the trio of logical possibilities (all partially valid, I suspect, though I would grant most weight to this third point) for de­nying that a currently recorded maximal frequency of gradualism for plank­tonic foraminiferal lineages casts doubt on the general importance of punc­tuated equilibrium in evolution. The first argument attributes an apparent high frequency to biased sampling in the preselection, for rigorous study, of lineages already highlighted by taxonomic experts for suspected grad­ual change. The second and third arguments, on the other hand, hold that if high frequency truly characterizes this group, no general rebuttal of punctuated [Page 806] equilibrium follows thereby. The second argument denies the common assumption that high-frequency records uniquely complete geological evi­dence — and that gradualism will therefore prevail whenever the fossil record becomes good enough to preserve its true domination (with a high frequency for punctuated equilibrium then construed, by Darwin's original argument, as the artifact of a gappy record). This second argument maintains that, while foram data may be more complete on average, the best metazoan examples of punctuated equilibrium have been validated with excellent samples from ad­mittedly rarer but equally complete geological sequences, thus precluding the explanation of punctuated equilibrium as artifactual.

  The third argument also grants the reality of higher-relative frequency for gradualism in forams, but argues against extrapolation to larger multicellular organisms on grounds of genuine difference in evolutionary mode, based on important biological distinctions between these single-celled creatures of the oceanic plankton and sexually reproducing metazoan species that, however parochially, have served as the basis for most of our evolutionary theory and, in any case, form the bulk of the known fossil record.

  This third argument should not be viewed as special pleading by partisans, but as a positive opportunity for developing hypotheses about the importance (or insignificance) of punctuated equilibrium based on the correlation be­tween differences in frequency and distinctive biological properties of various taxonomic groups — particularly in features related to speciation, the pre­sumed evolutionary basis of punctuated equilibrium. Planktonic forams, with their asexuality, their small size and rapid turnover of generations, their unicellularity, their vast populations, and their geographic links to water masses, display maximal difference from most metazoans, and may therefore be especially suited for helping us to understand, by contrast, the prevailing mechanisms of evolution in multicellular and sexually reproducing organ­isms. The general nature of these differences does indeed point to a set of factors tied to the definition and division of populations, therefore grant­ing plausibility to the claim that so-called “species” of planktonic forams should show more gradualism than metazoan taxa, while punctuated equi­librium may prevail in sexually reproducing multicellular species. The sub­ject deserves much more attention and rigor, but to sketch a few suggested factors:

  (1) Population characteristics. We conventionally name Linnaean species of asexual protistans, but even if adequately stable “packages” of form or ge­netic distinctness exist in sufficiently extended domains of space and time to merit a vernacular designation as “populations,” what comparison do such entities bear with species of sexually reproducing multicellular organisms? (Needless to say, I raise no new issue here, but only recycle the perennial ques­tion of “the species problem” in asexual organisms.) Punctuated equilibrium posits a link of observed evolutionary rates to properties of branching specia­tion in populations. I don't even know how to think about such issues in planktonic forams, where vast populations may be coextensive with entire oceanic water masses, and where numbers must run into untold billions [Page 807] of organisms for every tiny subsection of a geographic range. How do new populations become isolated? How do favorable (or, for that matter, neu­tral) traits ever spread through popu
lations so extensive in both space and number?

  (2) Morphology and definition. If metazoan stasis can be attributed, at least in part, to developmental buffering, what (if any) corresponding phe­nomenon can keep the phenotypes of simple unicells stable? Perhaps foraminiferal phenotypes manifest substantial plasticity for shaping by forces of temperature, salinity, etc., in surrounding water masses (see Greiner, 1974) — as D'Arcy Thompson (1917, 1942) proposed for most of nature in his won­derfully iconoclastic classic, On Growth and Form — see pp. 1179–1208. (Thompson's claim that physical forces shape organisms directly holds lim­ited validity for complex and internally buffered multicellular forms, but his views may not be so implausible for several features of simpler unicells.) Could many examples of foraminiferal gradualism (compared with metazoan stasis in similar circumstances) reflect the plasticity of these protists in the face of gradual changes in the physical properties of enveloping oceanic water masses through time? If so, such gradual trends would not be recording evolutionary change in the usual genetic sense.

  (3) Most interestingly (as a potential illustration of the main theoretical concern of this book), we must consider the potential for strongly allometric scaling of effects from a defined locus of change to other levels of an evo­lutionary hierarchy. To reiterate a claim that runs, almost like a mantra, throughout this text: punctuated equilibrium is a particular theory about a definite level of organization at a specified scale of time: the origin and de­ployment of species in geological perspective. The punctuational character of such change does not imply — and may even, in certain extrapolations to other scales, explicitly deny — a pervasive punctuational style for all change at any level or scale. In particular, punctuated equilibrium posits that tolerably gradual trends in the overall history of phenotypes within major lineages and clades (including such traditional tales as augmenting body size in hominids, increasing sutural complexity in ammonoids, or symmetry of the cup in crinoids) should reveal a punctuational fine texture when placed “under the mi­croscope” of dissection to visualize the individual (speciational) “building blocks” of the totality — what we have long called the “climbing up a stair­case” rather than the “rolling a ball up an inclined plane” model of fine structure for trends.

  Similarly, in asking about evolutionary causality under selective models (see Chapter 8), we need to identify the primary locus of Darwinian individu­ality for the causal agents of any particular process — for only properly de­fined Darwinian individuals can operate as “interactors” in a selective pro­cess: that is, can interact with environments in such a way that their own genetic material becomes plurified in future generations because certain dis­tinctive properties confer emergent fitness upon the individual in its “struggle for existence” (see pp. 656–667). Punctuated equilibrium maintains that spe­cies, as well-defined Darwinian individuals, hold this causal status as irreducible [Page 808] components, or “atoms,” of evolutionary trends in clades. The apparatus of punctuated equilibrium then explains why trends, when necessarily de­scribed as speciational, display a punctuated pattern at geological scales (as expressed in the theory's basic components of stasis and geologically abrupt appearance). In a larger sense, punctuational accounts of trends propose a similar allometric model for any relevant scale — that is, any microscope placed over higher-level smoothness may reveal an underlying “stair-step” pattern among constituent causal individuals acting as Darwinian agents of the trend.

  In sexually reproducing metazoa, species clearly play this role as causal individuals (see Chapter 8). The theoretical validity of punctuated equilibrium depends upon such a claim and model. But when we turn to such asexually reproducing unicells as planktonic forams, designated “species” cannot be construed as proper Darwinian individuals, and therefore cannot be primary causal agents (or interactors) in evolutionary trends. To locate the proper agent, the legitimate analog of the metazoan species, we must move “down” a level to the clone — to what Janzen (1977), in a seminal paper, called the El, or “evolutionary individual.”

  When we execute this conceptual downshift in levels to locate the focal evolutionary individual in asexual and unicellular lineages, we recognize that the foram “species” acts as an analog to the metazoan lineage or clade, not to the metazoan species. The foram “species” represents a temporal collectivity whose evolutionary pattern arises as a summed history of the Darwinian indi­viduals — clones in this case — acting as primary causal agents.

  We can now shift the entire causal apparatus one level down to posit a different locus of punctuational change in planktonic forams. Just as the punctuational history of species generates smooth trends in the collectivity of a lineage or clade in Metazoa, so too might the punctuational history of clones yield gradualism in the collectivities so dubiously designated as “species” in asexual unicells. In other words, foram “species” may exhibit gradualism be­cause these supposed entities are really results or collectivities, not proper Darwinian individuals or causal agents. Eldredge and I first presented this ar­gument in our initial commentary on the debate about punctuated equilib­rium (Gould and Eldredge, 1977, p. 142, and Fig. 9-10):

  We predict more gradualism in asexual forms on biological grounds. Their history should be, in terms of their own unit, as punctuational as the history of sexual Metazoa. But their unit is a clone, not a species. Their evolutionary mode is probably intermediate between natural selec­tion in populations, and species selection in clades: variability arises via new clones produced rapidly (in this case, truly suddenly) by mutation. The phenotypic distribution of these new clones may be random with re­spect to selection within an asexual lineage (usually termed a “species,” but not truly analogous with sexual species composed of interacting individuals). Evolution proceeds by selecting subsets within the group of competing clones. If we could enter the protists' world, we would view this process of “clone selection” as punctuational. But we study [Page 809] their evolution from our own biased perspective of species, and see their gradualism as truly phyletic — while it is really the clonal analog of a gradual evolutionary trend produced by punctuated equilibria and spe­cies selection.

  Lenski's remarkable studies on controlled evolution of bacteria under laboratory conditions of replication provide striking evidence for this claim (see full discussion of this work on pp. 931–936). Lenski and colleagues (Lenski and Travisano, 1994; Papadopoulos et al., 1999) monitored average cell size for 10,000 generations in 12 lineages of E. coli. Cell size increased asymptoti­cally in each lineage, steadily for the first 3000 generations or so, but remain­ing relatively stable thereafter. The fine structure of increase, however, pro­ceeded in a punctuational manner in each lineage — a step-like pattern of stability in average cell size, followed by rapid ratcheting of the full popula­tion up to larger dimensions. This punctuational pattern presumably oc­curred because clones act as primary Darwinian individuals in this system. The full lineage must “wait” for sudden introduction of favorable variation in the form of occasional mutations, initiating novel clones that can then sweep through the entire lineage to yield a punctuational step in the overall phylogeny (at a scale of 10,000 generations in phenotypic history). Predict­able, replicable size increase occurs by punctuational clone selection in each case (see Fig. 9-11). Lenski's powerful result does not illustrate a case of punc­tuated equilibrium, sensu stricto, but he does provide a challenging and instructive

  9-10. The supposed gradualism noted in many foram species may represent a view, from too high a level, of an overall trend within a phyletic sequence prop­erly analyzed in terms of punctuational events at the level of clone selection — the appropriate mode in such asexual forms. 1 shows a conventional metazoan lin­eage in punctuated equilibrium. 2 shows the apparent gradualism in a foram lin­eage. 3 shows a gradualistic segment between B and C magnified so that the ap­propriate process of punctuational clone selection becomes visible. From Gould and Eldredge, 1977.

&
nbsp; [Page 810]

  argument for considering the validity of punctuational change at all levels.

  Just as the careful watchdog at any scientific meeting will unhesitatingly call out “what's the scale” when a colleague fails to include a measurement bar on a slide of any important object, we must always ask, “what's the level” when we analyze the causal basis of any evolutionary pattern. Punctuational clone selection can yield gradualism within collectivities conventionally (if dubiously) called “species,” just as punctuated equilibrium, acting on species as Darwinian individuals, can produce gradual trends in the overall history of lineages and clades.

  GENOTYPES. Punctuated equilibrium is a theory about the evolution of phenotypes (both in concept and in operational testability for paleontological hypotheses), and correlations with genotypic patterns provide neither a crucial test nor even any necessary prediction. For example, critics of punctuated equilibrium have often argued that the apparently cumulative character of overall genetic distances among members of an evolving clade, expressed as a high correlation between measured disparities and independently derived times since divergence from a common ancestor — the kind of information that, in idealized (but rarely encountered) situations, yields a rough “molecu­lar clock” — should argue strongly against punctuational styles of evolution, while affirming anagenetic gradualism.

  But, leaving aside the highly questionable empirical status of these claims, the hypothesis of punctuated equilibrium would not be affected by positive outcomes, even at much higher relative frequency than the known history of life apparently validates. In supposing that “molecular clocks” tick against the requirements of punctuated equilibrium, we fall into two bad habits of thinking that impede macroevolutionary theory in general, and therefore rank as important conceptual barriers against the theses of this book. First, reductionistic biases often lead us to seek an “underlying” genetic basis for any overt phenomenon at any scale, and then to view data at this level as a fundamental locus for proper evolutionary explanation. (But consider only two among many rebuttals of such a position: (1) a genetic pattern may be