The Structure of Evolutionary Theory

Home > Other > The Structure of Evolutionary Theory > Page 16
The Structure of Evolutionary Theory Page 16

by Stephen Jay Gould


  5. I summarize the revolutionary empirics and conceptualizations of evo-devo in four themes, united by a common goal: to rebalance constraint and adaptation as causes and forces of evolution, and to acknowledge the perva­siveness and importance — also the synergy with natural selection, rather than opposition to Darwinian themes — of developmental constraint as a positive, structuralist, and internal force. The first theme explores the implications — for internally directed evolutionary pathways and consequent clumping of taxa in morphospace — of the remarkable and utterly unanticipated discovery of extensive “deep homology” among phyla separated at least since the Cam­brian explosion, as expressed by shared and highly conserved genes regulat­ing fundamental processes of development. I first discuss the role and ac­tion of some of these developmental systems — the ABC genes of Arabidopsis in regulating circlets of structures in floral morphology, the Hox genes of Drosophila in regulating differentiation of organs along the AP axis, and the role of the Pax-6 system in the development of eyes — in validating (only par­tially, of course) the archetypal theories of 19th century transcendental mor­phology, long regarded as contrary to strictly selectionist views of life's his­tory — particularly Goethe's theory of the leaf archetype, and Geoffroy's idea of the vertebral groundplan of AP differentiation. I then discuss the even more exciting subject of homologically conserved systems across distant phyla, as expressed in high sequence similarity of important regulators, com­mon rules of development (particularly the “Hoxology” followed in both ar­thropod and vertebrate ontogeny), and similar action of homeotic mutations that impact Hoxological rules by loss or gain of function. Geoffroy was partially right in asserting segmental homology between arthropods and ver­tebrates, particularly for the comparison of insect metameres with [Page 83] rhombomeric segments in the developing vertebrate brain (a small part, perhaps, of the AP axis of most modern vertebrates, but the major component of the earliest fossil vertebrates), where the segments themselves may form differ­ently, but where rules of Hoxology then work in the same manner during later differentiation. I also defend the substantial validity of Geoffroy's other “crazy” comparison — the dorsoventral inversion of the same basic body plan between arthropods and vertebrates.

  6. The second theme stresses the even more positive role of parallelism, based on common action of regulators shared by deep homology, in directing the evolutionary pathways of distantly related phyla into similar channels of adaptations thus more easily generated (thereby defining this phenomenon as synergistic and consistent with an expanded Darwinian theory, and not confrontational or dismissive of selection). I discuss such broad scale examples as the stunning discovery of substantial parallelism in the supposedly classical, “poster boy” expression of the opposite phenomenon of convergence — the development of eyes in arthropods, vertebrates, and cephalopods. The overt adult phenotype, of course, remains largely convergent, but homology of the underlying regulators demonstrates the strong internal channeling of parallel­ism. The vertebrate and squid version of Pax-6 can, in fact, both rescue the development of eyes in Drosophila and produce ectopic expression of eyes in such odd places as limbs. I also discuss smaller-scale examples of “conver­gence,” reinterpreted as parallelism, for even more precise similarities among separate lineages within coherent clades — particularly the independent con­version of thoracic limbs to maxillipeds, by identical homeotic changes in the same Hox genes, in several groups of crustaceans. Finally, I caution against Overextension and overenthusiasm by pointing out that genuine developmen­tal homologies may be far too broad in design, and far too unspecific in mor­phology, to merit a designation as parallelism, as in the role of distal-less in regulating “outpouchings” so generalized in basic structure, yet so different in form, as annelid parapodia, tunicate ampullae and echinoderm tube feet. I designate these overly broad similarities (that should not be designated as parallelism, or used as evidence for constraint by internal channeling) as “Pharaonic bricks” — that is, building blocks of such generality and multi­purpose utility that they cannot be labeled as constraints (with the obvious reductio ad absurdum of DNA as the homological basis of all life). By con­trast, the “Corinthian columns” of more specific conservations define the proper category of important positive constraint by internal channelings of parallelism based on homology of underlying regulators (just as the specific form of a Corinthian column, with its acanthus-leafed capital, represents a tightly constrained historical lineage that strongly influences the particular shape and utility of the entire resulting building).

  7. My third and shorter theme — for this subject, though “classical” throughout the history of evolutionary thought, holds, I believe, less validity and scope than the others — treats the role of homologous regulators in pro­ducing rapid, even truly saltational, changes channeled into limited possibili­ties of developmental pathways (as in Goldschmidt's defense of discontinuous [Page 84] evolution based upon mutations in rate genes that control ontogenetic trajec­tories). I discuss the false arguments often invoked to infer such saltational changes, but then document some limited, but occasionally important, cases of such discontinuous, but strongly channeled, change in macroevolution.

  8. The fourth theme of top-down channeling from full ancestral comple­ments, rather than bottom-up accretion along effectively unconstrained path­ways of local adaptation, explores the role of positive constraint in establish­ing the markedly non-random and inhomogeneous population of potential morphospace by actual organisms throughout the history of life. Ed Lewis, in brilliantly elucidating the action of Hox genes in the development of Drosophila, quite understandably assumed (albeit falsely, as we later discov­ered to our surprise) that evolution from initial homonomy to increasing complexity of AP differentiation had been achieved by addition of Hox genes, particularly to suppress abdominal legs and convert the second pair of wings to halteres. In fact, the opposite process of tinkering with established rules, primarily by increased localization of action and differentiation in tim­ing (and also by duplication of sets, at least for vertebrate Hox genes), has largely established the increasing diversity and complexity of differentiation in bilaterian phyla. The (presumably quite homonomous) common ancestor of arthropods and vertebrates already possessed a full complement of Hox genes, and even the bilaterian common ancestor already possessed at least seven elements of the set. Moreover, the genomes of the most homonomous modern groups of onycophorans and myriapods also include a full set of Hox genes — so differentiation of phenotypic complexity must originate as a de­rived feature of Hox action, exapted from a different initial role. The Cam­brian explosion remains a crucial and genuine phenomenon of phenotypic diversification, a conclusion unthreatened by a putatively earlier common an­cestry of animal phyla in a strictly genealogical (not phenotypic) sense. The further evolution of admittedly luxuriant, even awesome, variety in major phyla of complex animals has followed definite pathways of internal channel­ing, positively abetted (as much as negatively constrained) by homologous developmental rules acting as potentiators for more rapid and effective selec­tion (as in the loss of snake limbs and iteration of prepelvic segments), and not as brakes or limitations upon Darwinian efficacy.

  Chapter 11: The integration of constraint and adaptation: structural constraints, spandrels, and exaptation

  1. D'Arcy Thompson's idiosyncratic, but brilliantly crafted and expressed, theory of form (1917,1942) presents a 20th century prototype for the generalist, or ahistorical, form of structural constraint: adaptation produced not by a functionalist mechanism like natural selection (or Lamarckism), but di­rectly and automatically impressed by physical forces operating under invari­ant laws of nature. This theory enjoyed some success in explaining the corre­lation of form and function in very simple and labile forms (particularly as influenced by scale-bound changes in surface/volume ratios). But similarly nongenetic (and nonphyletic) explanations do not apply to complex [Page 85] creatures, and even D'Arcy Tho
mpson admitted that his mechanism could not encompass, say, “hipponess,” but, at most, only the smooth transformations of these basic designs among closely related forms of similar Bauplan (the true theoretical significance of his much misunderstood theory of transformed coordinates). In summary, D'Arcy Thompson, the great student of Aristotle, erred in mixing the master's modes of causality — by assuming that the adap­tive value (or final cause) of well designed morphology could specify the physical forces (or efficient causes) that actually built the structures.

  2. Stuart Kauffman and Brian Goodwin have presented the most cogent modern arguments in this tradition of direct physical causation. These argu­ments hold substantial power for explaining some features of relatively sim­ple biological systems, say from life's beginnings to the origin of prokaryotic cells, where basic organic chemistry and the physics of self-organizing sys­tems can play out their timeless and general rules. Such models also have sub­stantial utility in describing very broad features of the ecology and energy dynamics of living systems in general terms that transcend any particular taxonomic composition. But this approach founders, as did D'Arcy Thompson's as well, when the contingent and phyletically bound histories of particular complex lineages fall under scrutiny — and such systems do constitute the “bread and butter” of macroevolution. Nonetheless, Kauffman's powerful notion of “order for free,” or adaptive configurations that emerge from the ahistoric (even abiological) nature of systems, and need not be explained by particular invocations of some functional force like natural selection, should give us pause before we speculate about Darwinian causes only from evidence of functionality. This “order for free” aids, and does not confute, such func­tional forces as selection by providing easier (even automatic) pathways to­wards a common desideratum of adaptive biological systems.

  3. I then turn to the second, and (in my judgment) far more important, theme of structural constraint in the fully historicist and phyletic context of aptive evolution by cooptation of structures already present for other reasons (often nonadaptive in their origin), rather than by direct adaptation for cur­rent function via natural selection. The central principle of a fundamental logical difference between reasons for historical origin and current functional utility — a vital component in all historical analysis, as clearly recognized but insufficiently emphasized by Darwin, and then unfortunately underplayed or forgotten by later acolytes — was brilliantly identified and dissected by Friedrich Nietzsche in his Genealogy of Morals, where he contrasted the ori­gin of punishment in a primal will to power, with the (often very different) utility of punishment in our current social and political systems.

  4. Darwin himself invoked this principle of disconnection between histori­cal origin and current utility both in the Origin's first edition, and particularly in later responses to St. George Mivart's critique (the basis for the only chap­ter that Darwin added to later editions of the Origin) on the supposed inabil­ity of natural selection to explain the incipient (and apparently useless) stages of adaptive structures. Darwin asserted the principle of functional shift to ar­gue that, although incipient stages could not have functioned in the manner [Page 86] of their final form, they might still have arisen by natural selection for a dif­ferent initial utility (feathers first evolved for thermoregulation and later co-opted for flight, for example). Darwin used this principle of cooptation, or functional shift, in two important ways that enriched and expanded his the­ory away from a caricatured panselectionist version — as the primary ground of historical contingency in phyletic sequences (for one cannot predict the di­rection of subsequent cooptation from different primary utilities), and as a source of structural constraint upon evolutionary pathways. But these Dar­winian invocations stopped short of a radical claim for frequent and impor­tant nonadaptive origins of structures co-opted to later utility. That is, Darwin rarely proceeded beyond the principle of originally adaptive origin for differ­ent function, with later cooptation to altered utility.

  5. This important principle of cooptation of preexisting structures origi­nally built for different reasons has been so underemphasized in Darwinian traditions that the language of evolutionary theory does not even include a term for this central process — which Elisabeth Vrba and I called “exaptation” (Gould and Vrba, 1982). (The available, but generally disfavored, term “pre-adaptation” only speaks of potential before the fact, and has been widely re­jected in any case for its unfortunate, but inevitable, linguistic implication of foreordination in evolution, the very opposite of the intended meaning!)

  6. I present a list of criteria for recognizing exaptations and separating them from true adaptations. I also discuss some outstanding examples of exaptation from the recent literature, with particular emphasis on the multi­ple exaptation of lens crystallins (in part for their fortuitous transparency, but for many other cooptable characteristics as well) in so many vertebrates and from so many independent and different original functions.

  7. The exaptation of structures that arose for different adaptive reasons remains within selectionist orthodoxy (while granting structural constraint a large influence over historical pathways, in contrast with crude panadaptationism) by confirming a Darwinian basis for the adaptive origin of struc­tures, whatever their later history of exaptive shift. On the other hand, the theoretically radical version of this second, or historicist, style of structural constraint in evolution posits an important role for an additional phenome­non in macroevolution: the truly nonadaptive origin of structures that may later be exapted for subsequent utility. Many sources of such nonadaptive origin may be specified (see point 10 below), but inevitable architectural con­sequences of other features — the spandrels of Gould and Lewontin's termi­nology (1979) — probably rank as most frequent and most important in the history of lineages.

  8. Spandrels (although unnamed and ungeneralized) have been acknowl­edged in Darwinian traditions, but relegated to insignificant relative frequen­cies by invalid arguments for their rarity, their structural inconsequentiality (the mold marks on an old bottle, for example), or their temporally subse­quent status as sequelae — with the first two claims empirically false, and the last claim logically false as a further confusion between historical origin and current utility. [Page 87]

  9. I affirm the importance and high relative frequency of spandrels, and therefore of nonadaptive origin, in evolutionary theory by two major argu­ments for ubiquity. First, for intrinsic structural reasons, the number of po­tential spandrels greatly increases as organisms and their traits become more complex. (The spandrels of the human brain must greatly outnumber the im­mediately adaptive reasons for increase in size; the spandrels of the cylindri­cal umbilical space of a gastropod shell, by contrast, may be far more limited, although exaptive use as a brooding chamber has been important in several lineages.) Second, under hierarchical models of selection, features evolved for any reason at one level generate automatic consequences at other levels — and these consequences can only be classified as cross-level spandrels (since they are “injected into” the new level, rather than actively evolved there).

  10. The full classification of spandrels and modes of exaptation offers a resolving taxonomy and solution — primarily through the key concept of the “exaptive pool” — for the compelling and heretofore confusing (yet much discussed) problem of “evolvability.” Former confusion has centered upon the apparent paradox that ordinary organismal selection, the supposed ca­nonical mechanism of evolutionary change, would seem (at least as its pri­mary overt effect) to restrict and limit future possibilities by specializing forms to complexities of immediate environments, and therefore to act against an “evolvability” that largely defines the future macroevolutionary prospects of any lineage. The solution lies in recognizing that spandrels, al­though architecturally consequential, are not doomed to a secondary or un­important status thereby. Spandrels, and all other forms of exaptive potential, define the ground of evolvability, and play as importa
nt a role in macro-evolutionary potential as conventional adaptation does for the immediacy of microevolutionary success. I emphasize the centrality of the exaptive pool for solving the problem of evolvability by presenting a full taxonomy of catego­ries for the pool's richness, focusing on a primary distinction between “frank­lins” (or inherent potentials of structures evolved for other adaptive roles — that is, the classical Darwinian functional shifts that do not depart from adaptationism), and “miltons” (or true nonadaptations, arising from several sources, with spandrels as a primary category, and then available for later cooptation from the exaptive pool — that is, the class of nonadaptive origins that does challenge the dominant role of panadaptationism in evolutionary theory).

 

‹ Prev