The Structure of Evolutionary Theory

by Stephen Jay Gould

  Interestingly, Darwin's own favorite example for coupling these two princi­ples of redundancy, repeated many times in the Origin of Species and still res­ident in most biology textbooks today, is not only wrong, but backwards. He invoked this coupling to explain the supposed conversion of piscine air blad­ders into lungs. But the same argument works just as well, in reverse order, for the actual transformation of lungs in plesiomorphic fishes to swim blad­ders in highly derived teleosts that did not originate until the Triassic (Liem, 1988). Darwin wrote (1859, pp. 204-205): “A swim-bladder has apparently been converted into an air-breathing lung. The same organ having performed simultaneously very different functions, and then having been specialized for one function; and two very distinct organs having performed at the same time the same function, the one having been perfected whilst aided by the other, must often have largely facilitated transitions.” (Ancestral fish lungs can in­deed also function for buoyancy, whereas, and more obviously, gills work as well as lungs for breathing. Modern lungfishes retain both systems, as their formal name, Dipnoi (or “two breathing”) testifies.)

  The two great historical and structural implications of quirky

  functional shift

  This principle of functional shift deserves far more prominence, and explicit recognition, than it has ever received among evolutionary theorists. I have tried to emphasize its vital role in establishing the contingency and unpre­dictability of evolutionary change by an adjectival strategy of designation as “quirky functional shift.” In operational terms, we should acknowledge, most of all, the property of major functional alteration based upon far more limited (in extreme cases, virtually absent) structural change — another way of expressing the structuralist concept of inherent flexibility in natural forms and designs (to different degrees that should be subject to specification on a case by case basis). In any event, however textually underemphasized, this principle has always played two important roles in standard Darwinian theory:

  AS THE GROUND OF CONTINGENCY FOR LIFE'S HISTORY. Thoughtful Darwinians, no matter how confidently they have identified natural selec­tion as the exclusive cause of evolutionary change, have always recognized that their theory necessarily underpredicts the actual pathways of life's his­tory — and that explanations for the byways of individual lineages (and major aspects of the highways as well) can only be located in the factual record of particulars. This concept potentiated the tacit truce that, until recent years, held between paleontologists and Darwinian theorists under the Modern Synthesis. Under accepted terms, the theorists said to the paleontologists: “give up your old claims about special macroevolutionary mechanisms, and admit our contention that microevolutionary population genetics and natural selection hold full theoretical sufficiency. We will then grant you control over [Page 1225] the actual pageant of life's history by allowing that no nomothetic theory (and ours is 'as good as it gets') can specify actual pathways without factoring in the historical particulars that only your record preserves.” (I have scarcely hid my conviction, either in this book or elsewhere, that this truce always op­erated as a “lousy deal” for the science of paleontology.)

  In a minimal sense, Darwinian theory must grant this space to contin­gency — if only because even the most basic and least sophisticated form of the theory holds that organisms adapt to changing local environments (and do not follow preestablished routes towards “progress” or any other goal). Since we all admit that local environments change on an erratic and contin­gent vector through time, life's overall pathway must be dominated by contin­gent factors, even if every immediate event of natural selection could, in prin­ciple, achieve a deterministic explanation in local environmental terms. (After all, this feature of Darwinism as emphasized in Chapter 2, and as long appre­ciated by intellectual historians, established the most radical aspect of natural selection from the start — as contrary to all earlier evolutionary speculations, with their assumptions about law like directionality, usually regulated by di­vine intent.)

  But if contingency resided only in this basic aspect of environmental scal­ing, then the principle, though sound enough, would not run so deep in Dar­winian traditions. Rather, contingency gains its greatest force through the principle of quirky functional shift: the discordance between historical origin and current utility, and the consequent fallacy of direct inference from mod­ern status to initial meaning. Nietzsche emphasized the primary role of this discordance in the study of history by writing (as quoted more fully on p. 1217) that “the development of a thing, a tradition, an organ is therefore certainly not its progressus towards a goal,” and that the inevitability of func­tional shift makes any important historical sequence “instead ... a succession of more or less profound, more or less mutually independent processes of subjugation exacted on the thing.”

  Even a unidirectional sequence of changing form with basically unaltered function would require explicit knowledge of contingent environmental his­tories for anything close to full or satisfactory explanation. The addition of quirky functional shift, usually in several episodes for each organ in any complex phylogeny, guarantees a cardinal role for historical explanation in any major lineage (again, as Nietzsche recognized). In a personally favorite example, for combining the canonical general case with a particular ending twist, the African black heron, Egretta ardesiaca, uses its wings largely to shade the shallow water of its habitat, thus providing a clear view of available prey (my thanks to E. Vrba for this example, as discussed in Gould and Vrba, 1982).

  Any intelligent person with a sense of history's length and meaning could identify the author of the following statement as a modern Parsifal, or perfect fool: “Aha, now I know why herons evolved wings — in order to eat, for they would starve if they couldn't shade the water and see their food.” Our savvy interlocutor would offer the obvious refutation that most birds use their [Page 1226] wings to fly (as does the egret, albeit as a “demoted” secondary utility in the species' current habitat). Any knowledgeable biologist would then add — hence my fondness for this example of double quirky functional shift — that feathers were initially coopted for flight in a much older functional shift, a de­fining transition in avian phylogeny from a different initial role, perhaps in thermoregulation. This old hypothetical argument has now been fortified by two sources of evidence: experimental data on the thermodynamic, with no accompanying aerodynamic, advantages of tiny protowings (Kingsolver and Koehl, 1985); and historical data on the probable origin of birds from the smallest-bodied lineage of running dinosaurs, a group that might have experi­enced the greatest functional need for supplementary thermoregulation, given their highest levels of activity combined with highest surface/volume ratios (as a consequence of minimal body size), of any lineage within the dinosaurian clade. Indeed, this example has served as the canonical illustration, ever since Mivart's defining book of 1871, for “the problem of incipient stages of useful structures,” or quirky functional shift (and also as the eponym for my designation, in nontechnical writing, as “the 5 percent of a wing problem”).

  Faced with this argument, our Parsifal might continue to reject contin­gency, and embrace predictability, on the false assumption that natural knowledge, as “scientific,” must be so constituted. Even after we mock his previous conviction that he knows why egrets developed wings when he un­derstands their present use, we discover that he has not generalized the mes­sage, for he now argues from the opposite temporal end: “But if I, as a great scientist with full knowledge of evolutionary theory, had visited the Earth in early Jurassic times and observed the avian ancestor as a small running dino­saur using feathers on protowings for supplementary thermoregulation, I surely could have concluded that this animal would evolve larger wings and eventually enter a realm of cooptable utility for flight. I would also know that, for 150 million years, the ancestors of African herons would use those wings for flight, and then, on a continent to be called Africa [for our seer also knows the future history of plate
tectonics on Earth] this one little avian lin­eage would redeploy those old thermoregulatory organs for yet another novel task of shading water.” At this point, and continuing the literary analogy, we could only hope that our Parsifal finds the Holy Grail of quirky functional shift, and abandons his foolish ways for the path of wisdom!

  Needless to say, the actual history of any key organ in any major lineage far surpasses this avian cartoon both in complexity and in number of episodes of functional shifting. Just consider, for starters, the passage from a cartillaginous rod-like element, functioning to support the agnathan gill, to the hyomandibula of gnathostome fishes (used primarily to suspend the upper jaw from the cranium), to the stapes of tetrapods (following fusion of the up­per jaw to the skull, and responding to a functional need for a different mode of sound perception in air vs. water).

  This quirky historical character of major evolutionary change in particular lineages — thoroughly explainable after the fact, however unpredictable in principle beforehand — constitutes the greatest fascination of the subject for [Page 1227] many practitioners, myself included. Yet, this same inherent historicity has saddened scientists of other temperaments and predilections. For people who find greatest satisfaction in those aspects of nature that achieve full meaning and explanation under invariant and timeless laws, but who cannot resist the fascination of evolutionary biology as a career, the irreducibly contingent as­pect of their chosen subject defines its least congenial attribute. Such scientists have therefore tended to underplay (or even, in extreme cases, largely to deny) contingency, or to focus on those broader aspects of the subject, far from the fascination of the toings and froings of real history in concrete lin­eages, that do fall into the more conventional realm of predictability under natural law. Indeed, the previous section of this chapter treated this species of structuralist thought — and though I did not hide my own lack of affinity for this approach, I trust that I did grant the subject my genuine respect and ac­knowledgment of partial validity (while also expressing my abiding admira­tion for the sheer iconoclasm and beautiful prose of D'Arcy Thompson). I may be a historian at heart, but I do understand Kauffman's frustration, and his point, when he recognizes the intellectual linkage of natural selection to contingency, and then writes (1993, p. 26):

  We have come to think of selection as essentially the only source of order in the biological world. ... It follows that, in our current view, organ­isms are largely ad hoc solutions to design problems cobbled together by selection. It follows that most properties, which are widespread in or­ganisms, are widespread by virtue of common descent from a tinkered-together ancestor, with selective maintenance of the useful tinkerings. It follows that we see organisms as overwhelmingly contingent historical accidents, abetted by design . . .

  My own aim is not so much to challenge as to broaden the neo-Darwinian tradition. For, despite its resilience, that tradition has surely grown without seriously attempting to integrate the ways in which sim­ple and complex systems may spontaneously exhibit order.

  As ONE OF THE TWO MAJOR SOURCES OF STRUCTURALIST INPUT INTO THE PRIMARILY FUNCTIONALIST BASIS OF DARWINIAN THE­ORY. I treated Darwin's primary acknowledgment of a subsidiary role for structuralist, and at least partly non-adaptationist, thinking within the the­ory of natural selection — his treatment of “correlations of growth,” or nonadaptative side-consequences of adaptive change — in Chapter 4, pages 330–341. Darwin's discussion of quirky functional shift, and his recognition of this principle's indispensability for including the evolution of major novelties within the compass of natural selection by gradual change, marks his second substantial foray into subsidiary themes of a primarily formalist or structuralist character — in modern terms, his acknowledgment of an important role for internal constraint (as a precondition and helpmeet for natural selec­tion) in directing the history of evolutionary lineages.

  The role played by historical constraint in quirky functional shift lies im­plicit within the previous discussion of contingency, and therefore needs little [Page 1228] additional elaboration. Suffice to say that if a capacity for utilization in mark­edly different ways did not lie within the inherent or formal structure of most primary adaptations, then evolution would never be able to reach a novel “there” from its present “here” — and life's history would stagnate in tran­sient perfection (and then expire when surrounding environments underwent their occasional substantial alterations).

  After all, natural selection cannot act as a magic wand for the immediate construction of any urgent need. The adaptability — or, in the more general term now finally receiving substantial and deserved attention from organismal biologists (see p. 1270), the “evolvability” — of any phenotype must de­pend, in large part, on a flexibility for future change that simply cannot arise, if we understand the nature of causality itself aright, by direct natural selec­tion at the usual Darwinian level of organismal phenotypes. Therefore, a large component of evolvability must be attributed to inherent structural properties of features that originated by natural selection for one reason, but also manifest a capacity for subsequent recruitment (with minimal change) to substantially different and novel functions. The study and systematization of these formal and structural reasons for evolvability sets an important agenda, now largely unfulfilled but attracting considerable interest, for evolutionary biology.

  To return to my previous example, the agnathan ancestor that built a series of v-shaped, backward-pointing gill arches, each made of several rod-like ele­ments, for pumping water to breathe and feed, evolved these features for its own immediate needs, and not (obviously) with any forethought about modifiability into jaws that might one day surround its unsupported mouth. But if the elements of the foremost arch had not inherently possessed the form, the positioning, the coordination, and the developmental potential to move to a more anterior position surrounding the mouth, the gnathostome lineage would never have emerged, the agnathans might have remained a relatively minor component of marine faunas (or become extinct entirely), and terres­trial environments, to this day, might have remained the domain of plants and insects — perfectly competent and “happy” ecosystems, building a lovely earth teeming with life, but evolving nothing conscious to proclaim its aes­thetic, extol its virtues, or to record, perhaps even to seal, its doom. We must thank both this contingent good fortune, and the latent structural possibili­ties of gill arches, for this shot at our own particular brand of record keeping (even of “immortality” in some operationally meaningful sense of the term).

  The story, of course, continues from there (and for each lineage), with a constant twinning of contingency and structural potentiality. If one marginal group of fishes had not evolved a peculiar fin, with a branching central ele­ment orthogonal to the body's antero-posterior axis (rather than parallel to the axis, as in most self-respecting members of the clade and guild), no sup­port firm enough to build the centerpiece of a limb for terrestrial life might ever have emerged within the lineage of vertebrates. And if these resulting tetrapods had never evolved their forelimb for terrestrial locomotion, the cel­ebrated convergence of aerodynamic form in the wings of bats, birds and [Page 1229] pterosaurs — the supposed disproof of contingency's dominant role in evolu­tion! — would have died aborning for want of a common and contingent sub­strate on which to hang these adaptive marvels of similarly excellent design.

  In wondering why this principle of quirky functional shift, or the discor­dance between reasons for origin and current utility, has received such short shrift in Darwinian traditions, despite its pivotal importance for these two central aspects of natural selection, I can only conclude that its status has, heretofore, rested only upon its acknowledged capacity for auxiliary aid, and not upon a claim for conceptual novelty thus supplied to evolutionary theory. Such an evaluation flows easily from an understanding that quirky functional shift, when confined to this Darwinian formulation, remains entirely within the ordinary functionalist and adapta
tionist framework of the general theory. That is, Darwin's version of the principle, overly restrictive as I shall show in the next subsection, remains fully adaptational in confining its compass to functional shift from one utility to another. The feature in question initially evolves as a conventional adaptation for one function, and then becomes co-opted for a different role. This shift may validate the centrality of contingency for a Darwinian explanation of history, but the process remains under adap­tational direction at all times — and the fundamental mechanism of Darwin­ian evolutionary change never cedes any control. In other words, features that undergo this Darwinian style of quirky shift retain full functionality throughout, and their changes remain under the government of natural selec­tion at all times. Thus, the analysis of history may be enriched, but the mech­anisms of evolution do not alter or augment.

  Nonetheless, although I may recognize why Darwinian tradition has underemphasized quirky functional shift, I still believe that this inattention has created substantial problems in our understanding of the logic of evolution­ary change. (As a psychological inference, I also suspect that this neglect flows from the status of quirky functional shift as a slightly uncomfortable “odd man out,” exuding a structuralist odor within an apparatus deemed powerful and intellectually intriguing for its functionalist basis and mechan­ics — just as Darwin's other structuralist principle of “correlations of growth” has received similarly little regard in the history of our field, at least until evo-devo made constraint an operational concept, thus inspiring both our interest and attention.)