The Structure of Evolutionary Theory

by Stephen Jay Gould

  * The old canard about advocating a short, even a Mosaic, time scale arises from an il­logical extension of this claim. A short time scale does require paroxysm to encompass events of the geological record within such a limited span. But the converse of this argu­ment — the claim falsely attributed to catastrophists — does not follow: for a dynamics of paroxysm does not require or even imply a doctrine of limited time. The earth may be mil­lions or billions of years old, as the catastrophists believed, and still concentrate its major changes in brief bursts.

  * Cuvier's original text reads: “Y a-t-il des alternatives dans leur retour,” so “alterations,” in Jameson's standard translation, should probably read “alternations,” thus ren­dering Cuvier's inquiry as a question about directionality. He wants to know whether fossil species mark unique episodes of time, a proposition that would be disproved if faunas ap­peared and reappeared in alternation.

  * Huxley could never leave a conflict with such calm and no vitriol. Thus, as a sample of his prose and personality, consider (from the same address) this incisive passage on what a later generation would call GIGO. I would also read this statement as an implied criticism of arrogance among physical scientists, an exercise in wound licking after Kelvin's effective salvos: “Mathematics may be compared to a mill of exquisite workmanship, which grinds you stuff of any degree of fineness; but, nevertheless, what you get out depends upon what you put in it; and as the grandest mill in the world will not extract wheatflour from peascods, so pages of formulae will not get a definite result out of loose data” (1869, in 1894, p. 333).

  * In writing only these two paragraphs on this dismissal of Lamarckism, I imply no judg­ment about relative historical importance. Scarcely any event in the history of evolutionary theory could be more vital or central than the formulation of a rationale for expunging from orthodoxy (and rendering virtually inconceivable in theory) the most venerable of all evolutionary mechanisms. I downplay this subject here only because this book treats the history of valid auxiliaries and alternatives to strict Darwinian functionalism — and Lamarckism, as an invalid functionalism, therefore becomes tangential to my concerns on both grounds (while remaining central to the larger, general history of evolutionary the­ory). Similarly, this chapter gives relatively little space to the formulation of the Modern Synthesis for the same reason: this history, so well covered in other sources (Provine, 1971; Mayr and Provine, 1980), specifies the formulation of modern orthodoxy, while this book treats the persistent sources of heterodoxy.

  * If the Synthesis had retained the pluralism of its early years, Kimura's neutral theory would have been welcomed from the first, under the criterion that any result legitimated by the mechanics and mathematics of known genetic processes thereby secured a rightful place (Wrightful in this particular case) — though Kimura's claim would have been viewed as sur­prising in the light of adaptationist preferences. But when the Synthesis hardened, and adaptationism itself became the primary criterion for acceptability, Kimura's theory seemed beyond the pale to many evolutionists. I shall never forget a decisive moment in my own early career, when I began to understand the difference between theoretical power and po­tentially dangerous overconfidence: Ernst Mayr rising (at the annual meeting of the Evolu­tion Society in New York) to confute the claim for neutralism in synonymous third position substitutions. Such changes could not, a priori and in principle he stated, be neutral. Alter­ations in the third position must impart some difference, perhaps energetic, that selection can “see” even if the coded amino acid does not alter. This must be so, he stated, because we now know that all substantial change is adaptive.

  * I am not generally drawn to sociological proposals in this mode, and I reacted nega­tively at first to Smocovitis's suggestion. But I have since read widely in the just post-World War II literature, and I only now understand the fervor and hope of “never again,” follow­ing all the devastation, and the heartrending impact (and inspired shame) as knowledge of the Holocaust surfaced. I was too young, when the war ended, to experience viscerally both this horror and hope, but I do grasp the character of this unusual time with a pervasive theme and agenda — and I accept the idea that humanistically inclined scientists must have hoped fervently that their own field might contribute to the reconstruction.

  * Although I strongly advocate a hierarchical model of multilevel selection (see Chapters 8 and 9), I regard this restriction to organismic selection as an important and positive re­form. Earlier claims for group and higher-level selection had been formulated so vaguely and falsely that they impeded our understanding of both this important concept and of the theory of selection in general. This salutary reform tore down erroneous standards and in­sisted that no further claims be made until the logical edifice could be properly rebuilt — no examples without a proper substructure; no paintings without a strong frame.

  * Footnote added in proof stage: Just as I submitted this completed book to the publisher, the press conference on Darwin's birthday (Feb. 12, 2001), announcing the very low num­ber of genes in the human genome, struck the deepest blow of our lifetimes against the conventions of reductionism, and for the irreducibility of proteomic (and full phenotypic) explanation to simple properties of codes at lower levels. Combinations, replete with emer­gent properties, and the specifics of contingent phyletic histories, must become a key part­ner, if not a primary locus, for biological explanation (see Gould, 2001).

  * But Huxley's own later contentions belie this strong claim. For example, he argues against uniform internal drives in parallelism, and for control by external selection, by not­ing that characters do not always correlate in the same manner within parallel trends ob­served in different lineages of fossils: “In all cases where fossils are abundantly preserved over a considerable period, we find the same phenomena. The change of form is very grad­ual. It is often along similar lines in related types. And in general it appears that different characters vary independently” (1942, p. 32). But doesn't this statement qualify as an ex­ample of “using data on the course of evolution to make assertions as to its mechanism”?

  * But what scientific good can derive from a theory that includes no possibility of refuta­tion from within? (A Mormon friend once told me that archaeologists of his church would either one day find direct evidence that the people of Mormon and Moroni had migrated from the Near East and lived in the New World until the 4th century ad, which would sup­port the testimony of the Book of Mormon, or they would not find such evidence, which would also support the doctrines of his church by illustrating God's challenge to his people to keep faith in the absence of empirical support.) In such cases, one can only suggest alter­native theories from without, and try to persuade people of good will that these alternatives provide better explanations for the purely empirical evidence.

  * I have struggled with this issue all my professional life, and have often wondered why the questions raised seem so much more recalcitrant, and so much more cascading in impli­cations, than for any other major problem in Darwinian theory. I don't think that mere per­sonal stupidity underlies my puzzlement — or rather, if so, the mental limitations must be largely collective, because other participants share the same struggle and express the same frustrations. I don't mean to sound either grandiloquent or exculpatory, but I seriously wonder if some of the difficulties might not arise largely from limitations in the common mental machinery of Homo sapiens. Levi-Strauss and the French structuralists may well be correct in holding that human brains work best as dichotomizing machines at single levels. We make our fundamental divisions by two (nature and culture or “the raw and the cooked” in Levi-Strauss's terms, night and day, male and female), and we therefore experi­ence great mental difficulty with continua, and with any system other than a two-valued logic (hence Aristotle's law of the excluded middle, and other similar guides). We are espe­cially ill-equipped to think hierarchically, and to juggle simultaneous influences from sev­eral nested levels upon the foci of our int
erest. The hierarchical theory of natural selection rests upon all these intrinsically difficult modes of reasoning.

  * In the late 1970's and early 1980's, I engaged in long and vociferous arguments with my graduate students Tony Arnold and Kurt Fristrup about the criteria of species selection. (As discussed on pp. 656-670, I now believe that they were right, and I was wrong.) In the course of these discussions, we developed this idea and name of “more-making” or pluri­faction. (If manufacture means, literally, making it by hand, and petrifaction means turning it into stone; then plurifaction simply means making more of it.) I do not now remember who first devised the word, or who contributed most to the concept's codification.

  * This kind of textual exegesis, a standard mode of scholarly work in the humanities, should be pursued more often in scientific discussion as well. Scientists tend to reject such an approach, I suppose, because we believe that forms of argument and rhetorical styles only lend a superficial patina to the “real” substance of logic and evidence, and therefore can teach us nothing of interest. I think that we have thereby missed a major source of in­sight about the operation of science — a source that would not only deepen our understand­ing of history and procedure, but would also help us to judge and analyze such contempo­rary issues as the logic of selectionist theory. If we locate consistent slips, foibles, jagged edges, strains, or near apologies — as presented verbally — then we can often pinpoint weak­nesses in logic or failure of empirical support. I show, in this section, that all major support­ers of gene selectionism fall into such verbal patterns at the theory's main loci of inconsis­tency. In previous books, I have tried to use this mode of analysis to explicate such issues as the nature of geological time (Gould, 1987b), the logic of biological determinism (1981a), and the concept of evolutionary progress (1996a) and predictability (1989c).

  * Working through the logic and problems of this vexatious issue has been pursued as a collective enterprise among many biologists and philosophers for more than 20 years. I have used the terminology of bookkeeping and causality for some time (Gould, 1994), and have developed or sharpened some of the arguments. But I do not think that I devised the labels. I believe that I first picked up the terminology of bookkeeping from arguments pre­sented by the University of Chicago philosopher Bill Wimsatt. Many authors have used this fruitful distinction for some time.

  * The logic of this case recalls the celebrated example (see pp. 492-502) of Rutherford's use of radioactivity both to impugn the theoretical basis for Kelvin's young age for the earth and then to provide the empirical basis for measuring a revised and much older age.

  * But it is not “only incidentally true” that genes generally come packaged into organ­isms on our planet — and that, in full extension, organic matter coagulates into evolution­ary individuals at several levels of an inclusive hierarchy: genes, cell lineages, organisms, demes, species, and clades. This process of coagulation has occurred for active and interest­ing structural reasons only dimly understood (Buss, 1987). But how could we regard this most fundamental feature of the organic world, constituting the basis of evolutionary cau­sality in units of selection, as “only incidentally true”? This structure may well be contin­gently true — in the sense that we can imagine an alternative world composed only of naked genes — but our planet's biological reality surely cannot be designated as incidental in the usual sense of unimportant or not fundamental. Indeed, the origin of such hierarchical structure may not even be contingently true, but broadly predictable (see Kauffman, 1993; Maynard Smith and Szathmary, 1995).

  * This interesting idea of parallel hierarchies to separate the replicative and interactive criteria of evolutionary individuality originated with Eldredge (1989; see also Vrba and Eldredge, 1984), who spoke of genealogical and economic hierarchies. The scheme contin­ues here with Williams's similar distinction of material and codical systems. I find the idea of dual hierarchies both interesting and challenging, but ultimately flawed and counterpro­ductive in the introduction of unnecessary complexity. (My rejection of this scheme defines my only major difference with my closest colleague Niles Eldredge, who has worked with me for 25 years on problems of macroevolutionary theory.)

  Eldredge's “economic” and Williams's “material” hierarchies include the interactors defined as proper units of selection in this book — and also in Wilson and Sober (1994), and (by unintended verbal admission, though not explicitly) by such gene selectionists as Dawkins and Williams, as I have shown throughout this section. (Eldredge calls this hierar­chy “economic” to stress the doing and dying of such entities in nature's ecosystems.) Eldredge's “genealogical” and Williams's “codical” hierarchies express the concept of rep­lication (as nonmaterial units of information for Williams, but as an alternative hierarchy of replicating material entities for Eldredge).

  I find the framework of dual sequences unnecessarily complex and divisive because a sin­gle theme unites our search to define units of selection, and a single hierarchy expresses this theme in the best and clearest way. Units of selection must be evolutionary individuals by the criteria outlined on pages 608-613. Above all, such individuals must be interactors in order to function as units of selection in a causal process. They must also possess a mecha­nism of plurifaction — that is, interactors must be able to bias the heredity of subsequent generations towards more of their own contribution, however these contributions be pack­aged. This need for plurifaction underlies our sense that replication plays a vital role in evolutionary individuality — a role sufficiently important to be mistaken as causal and primary by gene selectionists, or at least to warrant a separate hierarchy (by Eldredge). But I raise two points to obviate the need for a separate hierarchy of replicators: (1) replication (or some other form of hereditary passage) constitutes only one of several necessary criteria for de­fining evolutionary individuality; and (2) this criterion of hereditary passage only demands that interactors possess a means of plurifaction; faithful replication represents one style of hereditary passage, but not a necessary mode for attribution of evolutionary individuality or designation as a unit of selection. Sexual organisms plurify by disaggregation and differ­ential passage of genes; other kinds of evolutionary individuals plurify by faithful passage.

  We should formulate a single hierarchy — call it material, genealogical, or perhaps simply evolutionary — composed of interactors with adequate modes of plurifaction. These evolu­tionary individuals build a hierarchy of inclusion, with each higher level encompassing the individuals beneath as parts. Most units in Eldredge's parallel hierarchies appear in both his economic and genealogical arrays — and therefore represent the evolutionary individuals we seek for a single hierarchy — for these are the entities that possess both the interactive (economic) and hereditary (genealogical) properties required of any evolutionary indi­vidual.

  * Anyone, like me, who grew up in America with fiercely traditional immigrant grand­parents from “the old country” will appreciate the humor of such limited and inappropri­ate reference points. My grandmother's only concern for any cultural or historical event (all of which she followed with great interest and intensity) stood out in her single, invariant question: “Is it good for the Jews?”

  * In my favorite more specific example, Williams (1992, p. 129) — who is, to say the least, no general champion of punctuated equilibrium or detractor of anagenesis — points out that mean morphological changes in some North American populations of English spar­rows during their century of residence on our side of the Atlantic reach a maximum of about 5% increase for the lengths of long bones of the wings and legs. This anagenetic in­crement, so small that “no birdwatchers will notice in their old age that the bird looks any different from what they remember from childhood” (Williams, 1992, p. 129), would, nonetheless, if maintained only for the geologically trivial interval of one million years, be “capable of turning sparrow-size into ostrich-size bones, and back again, about 54 times.” Clearly, anagenesis at virtua
lly any rate high enough to stand out above measurement error over a human lifetime cannot be sustained in a unidirectional manner for meaningful inter­vals in geological time.

  * At the risk of an unwarranted metaphorical excursion into anthropomorphic imagery, one might contrast limited change at organismal birth with necessary change at species birth in the following manner: New metazoan organisms arise by a process of complex de­velopment, which must discourage change for reasons recognized ever since von Baer formulated his laws of embryology (1828). At the organismal level, the new individual separates intrinsically from the parent; how then, may this offspring be kept sufficiently like the parent to preserve the collectivity of the population? An opposite problem attends the birth of species. At the species level, new individuals are born by speciation, which enhances change. But species do not separate intrinsically from their parents. They are born in fuzzy continuity. Their separation may be difficult. They must be cast out, or they will reinte­grate. Necessary change at speciation enhances this defining process of casting out from the parent. The newly born species faces a structural problem opposite from the neonatal or­ganism's dilemma: how may the new species-individual become sufficiently unlike the par­ent to be cast out, thus enhancing the collectivity of the clade by adding another part? In short, the new metazoan organism forms outside the parent: how can it be kept close? The new species separates with difficulty from the parent: how can it be cast out?