The Structure of Evolutionary Theory

by Stephen Jay Gould

  1. We must not view upwardly cascading effects as uniquely detrimental for species, but generally neutral or even positive at other levels. I suspect, rather, that a majority of upwardly cascading effects will be negative at any higher level of expression. Indeed, as argued previously, we have long recog­nized this phenomenon as a fundamental property of Darwinism (though, again, we have lacked the conceptual apparatus for explaining the results in these appropriate terms). The phenotypic expressions of mutations are span­drels at the organismal level, and we have long recognized the vast majority as deleterious for the organism. But we do not regard this inevitable property (of anything injected adventitiously into a different level) as globally detri­mental to organisms. Populations of organisms are large enough, and the generational cycling time of organisms short enough, to tolerate a substantial load of general disadvantage in exchange for the occasional opportunity [Page 1291] thereby achieved in encountering a favorable spandrel that can then be exapted in the canonical organismal process of natural selection. (This phe­nomenon gains its extreme expression in bacterial evolution, where the rarity of favorable mutants hardly curtails rapid evolution because populations are so large, and generations so short, that highly infrequent exaptable effects oc­cur often enough to drive substantial evolutionary change by sheer brevity of the waiting time between favorable injections, even in the absence of any mechanism of recombination to spread the benefits among individuals.)

  Thus, the organismal level can usually well afford this carnage of generally deleterious mutational effects in order to win its fuel of positive variants for natural selection. Indeed, Darwin's world offers the organism no other choice — for natural selection makes nothing directly and can only operate cre­atively in the generation of evolutionary change if some other process sup­plies enough undirected variation to power its odd mode of negative con­struction by elimination. One might even “applaud” this inevitable property of cross-level spandrels as “just the thing” that natural selection needs to be­come nature's potent driver of evolution despite its weakness, its dependen­cies, and its peculiar style of operation. Three interrelated facts establish and undergird natural selection's capacity to power organismal evolution in the face of these limitations: (1) natural selection requires “random” fuel undi­rected towards adaptive states (lest such an internal force for automatic or­ganic good overwhelm the weak power of selection to produce similar results in a characteristically slow, gradual and roundabout way); (2) the fuel sup­plied by phenotypic effects of mutation expresses itself as cross-level span­drels injected from the gene level into organismal phenotypes; and (3) cross-level spandrels manifest the required property of noncorrelation with benefits at their new level of injection. Would nature be Darwinian at all, absent these interesting properties of cross-level spandrels that must supply the fuel of nat­ural selection — thus establishing the category of “chance” in the duality of “chance” (effects of cross-level spandrels manifested as mutational pheno­types) and “necessity” (direct action of natural selection for adaptation to lo­cal environments) that we generally cite as an epitome of the Darwinian mechanism and worldview? If we just remember that the phenotypic expres­sions of mutations are cross-level spandrels, we will hold an important key for unlocking the curiosities of Darwinism.

  2. I don't wish to imply that species-individuals can only be weakened, given their generally small population sizes within their clades, and their slow cycling times, by the nonadaptive (and perhaps usually inadaptive) char­acter of most effects that supply a component of emergent fitness to species selection (Gould and Lloyd, 1999), and that arise as cross-level spandrels injected upwardly from the organismal level below. On the contrary, al­though species may often suffer (in terms of their ability to overcome in-adaptive spandrels) by their small population numbers and slow cycling times — whereas organisms vanquish this impediment by large populations and quick cycling — species also, and in “trade-off,” obtain a tremendous “leg up” over the organismal level by expressing an inherent “allometric” property [Page 1292] (see Gould and Lloyd, 1999 for this extension of allometry) possessed by virtue of their sizes and cyclings, but absent in organisms for equally inherent reasons of their own constitution. As argued at length in Chapter 8, organ­isms maintain their biological individuality as discrete entities by strategies of intricate and precise functional interrelationships among constituent parts, and by maintenance of internal and external boundaries (immune systems and skins) to exclude the subparts of other individuals from their geographic space. Thus, and most characteristically, if not uniquely, the organism main­tains its integrity by rigorous policing, and by active suppression of differen­tial proliferation among its subparts, a result that could otherwise follow from positive selection upon the individuals of lower levels within the organ­ism itself (especially upon cell lineages, with the failure of such organismal policing leading to the result that we call cancer).

  Thus, organisms sacrifice the benefits of including more upwardly cascad­ing effects as components of their exaptive pool because they work so hard to preserve their distinctive style of individuality by suppressing the churning of lower levels of selection within their bodies. But, by important contrast, spe­cies construct their equally powerful integrity and discreteness by different means that do not require such suppression of upwardly cascading effects. Species maintain their boundaries primarily by reproductive isolation of their subparts (the organisms that constitute their populations) from subparts of other species-individuals. By permitting their subparts to reproduce only with each other, and not with subparts of other species-individuals, a species main­tains its integrity, and constructs its boundary, with just as much clarity and efficiency as organisms can muster by different strategies of functional inte­gration among organs and active exclusion of invaders. This different, but equally efficient, modality of species does not include the suppression of lower level selection as a consequence. Therefore, species remain open, as or­ganisms do not, to experiencing a full and rich range of cross-level spandrels, injected as consequences of selection acting on lower-level individuals (pri­marily upon organisms) within their boundaries.

  Now — and here's the rub — we have generally viewed this openness of spe­cies as a negative sign of impotence at this higher level of biological organiza­tion. For, reasoning from our parochial perspective as organisms — and falsely supposing that organismal ways must be universally best ways — we have viewed the species's nonsuppression of upwardly cascading spandrels as a mark of its failings as a potential unit of selection. But perhaps we should re­verse this perspective and learn to respect the distinct allometric properties of species individuality as potential strengths for a different mode of selection — with the higher frequency of cross-level spandrels viewed as a source of rich potential denied to organisms, rather than as a mark of inefficiency imposed upon species.

  The species-individual, as a Darwinian interactor in selection at its own level, operates largely with cross-level exaptations arising from unsuppressed evolution of subparts (primarily organisms) at lower levels within itself. Such [Page 1293] nonsuppression acts as a source of evolutionary potential by permitting spe­cies to draw upon a wider pool of features than organisms can access.

  By not suppressing this evolutionary churning from within, the species-individual gains enormous flexibility in remaining open to help from below, expressed as exaptive effects that confer emergent fitness. Rather than view­ing this nonsuppression of aid from other levels, with the accompanying fail­ure to build many active adaptations at its own level, as a sign of wimpy weakness for the species — construed as a “poor organism” in the implication of most traditional thought — we should rather interpret these allometrically driven properties as cardinal strengths, and recognize the species as a “rich-but-different” Darwinian individual. The species, in this view, acts as a shel­ter or arbor that holds itself fast by active utilization of the proper
ties that build its well-defined individuality. By fostering internal change, and thereby gaining a large supply of upwardly cascading exaptive effects, species use the features of all contained lower-level individuals through the manifestation of their effects on the shelter itself. The species, through its own distinctive fea­tures of individuality, and requiring neither indulgence nor apologia from hu­man understanding, will continue to operate as a powerful agent in Darwin's world whether or not we parochial organisms, limited by our visceral feelings and traditions of language, choose to expand our view and recognize the sources of evolutionary potency at distant scales of nature's hierarchy.

  In conclusion, and to reiterate my rationale for placing so much emphasis on cross-level spandrels in evolution, this single theme, more than any other in this book, unites and exemplifies the weaknesses on all three legs of the tri­pod of essential postulates in conventional Darwinian logic, while also point­ing the way towards revisions that will expand and strengthen these three supports to produce an improved, and more comprehensive, evolutionary theory by retaining its Darwinian basis in the expanded form of a fully hierar­chical theory, while adding, to its preserved selectionist core, several aids and flavors from alternative traditions of structuralist thought.

  Thus, the hierarchical theory of selection (for strengthening and expansion of the first leg) greatly augments the role of spandrels by adding the cross-level category as more potent in numbers and more various in potential re­sults than the traditional at-level category. For at-level spandrels, in the usual Darwinian account of selection as a uniquely organismal process, must re­main confined to structural byproducts and space fillers within a context of integral adaptation of the body. But if selection works simultaneously at sev­eral hierarchically ordered levels of biological individuality, then the domain of spandrels expands to include any enjoined expression (upon Darwinian individuals at other levels) of changes causally introduced at a focal level — for these injected and adventitious expressions must originate nonadaptively (and “randomly” in our usual loose parlance) relative to their causal reasons of origin at the focal level of their construction.

  By the same token, spandrels become structural constraints rather than di­rect adaptations (or even alternative potentials of direct adaptations) — that [Page 1294] is, they enter the exaptive pool as miltonic things rather than franklinian po­tentials. Moreover, as constraints of this type, they add a structuralist and nonadaptationist component to the workings of evolution — thus strengthen­ing the tripod on leg two by including aspects of this formerly rejected mode of causation among the totality of devices that generate creative change in evolution. On a more specific note, if cross-level spandrels maintain an im­portant relative frequency among the components of evolutionary change, then these automatic expressions at other levels — introduced separately from, and simultaneously with, the primary changes that generate them at a differ­ent focal level — may largely control the possibilities and directions of evolution from a structural “inside,” rather than only from the functional “out­side” of natural selection.

  Thus, if the positive structural constraints of spandrels — particularly in their cross-level mode as effects propagated to various levels of the evolution­ary hierarchy — can help to explain the phenomenon of evolvability and the parsing of categories in the exaptive pool, then reforms on legs one and two of the essential Darwinian tripod will also illustrate why the extrapolationist premise of the third leg cannot suffice to explain evolution either. For macro-evolution cannot simply be scaled-up from microevolutionary mechanics if the phenomenology of this larger scale depends as much upon the potentials of evolvability as upon the impositions of selection, and if the exaptive pool promotes evolvability largely by the later utility at the same level (or simulta­neous exaptive use at other levels) of spandrels that originate for nonadaptive reasons. The explanation of macroevolution requires structuralist and hierar­chical inputs from various scales, and cannot be fully rendered as an exten­sion of organismal adaptation, smoothly scaled up through the immensity of geological time.

  A CLOSING COMMENT TO RESOLVE THE MACROEVOLUTIONARY PARADOX THAT CONSTRAINT ENSURES FLEXIBILITY WHEREAS SELECTION CRAFTS RESTRICTION

  In closing this section by reiterating the opening argument (p. 1270) in an­other context, I should extend my previous statement on the bounded inde­pendence of macroevolution to stress the positive theme of interesting differ­ences, and not only a negative claim for the necessary limitation of any explanation based upon pure extrapolation from microevolutionary mechan­ics. For the failure of microevolutionary extrapolation resides in something far deeper than mere insufficiency. Rather, and thus operating to intensify the explanatory gap, a cardinal feature of microevolution works directly against the potentials for macroevolution defined by the exaptive pool — thereby re­quiring that macroevolution proceed by actively overcoming this microevolu­tionary limitation, and not only by “adding value” to its mere insufficiency.

  Darwinian evolutionists have known this all along in their heart of hearts, and have tended to escape the resulting paradox by a leap of faith into the en­abling power of geological time to accomplish anything by accumulation [Page 1295] of small inputs. Most events of microevolutionary adaptation — that is, of or­dinary Darwinian natural selection in the organismal mode — work against evolvability by locking organisms into transient specializations and reducing the flexibility of the exaptive pool. This fact engenders the central paradox noted above: that immediate organismal processes tend to derail prospects for longterm evolutionary success at the species and clade level. Darwinian traditions have tried to surmount this stumbling block by arguing that, how­ever the process of specialization might restrict future prospects, natural se­lection still makes “better” organisms by rewarding success in direct competi­tion against conspecifics. And at least one component of this “betterness,” albeit a minority component — the occasional achievement of local adapta­tion by general biomechanical improvement, rather than by limiting specialization — must provide the major source of increments for macroevolutionary patterning by extrapolation. But this argument is bankrupt, and I have, throughout this book, chronicled a host of reasons for its failure.

  Therefore, the macroevolutionary success of species and clades must arise, in large part, by active utilization of selective processes at their own higher levels, and in opposition to the generally restricting implications and sequelae of microevolution. Moreover, in fueling these macroevolutionary successes, species must exapt the rich potentials supplied by structural and historical constraints of spandrels and other miltonic “things” emplaced into the exap­tive pool against (or orthogonally to) these restricting tendencies of natural selection — in other words, by exploiting the components of a phenomenon that we have loosely called “evolvability” and vaguely recognized as some­thing apart from natural selection. And thus, an expansion of the first leg by hierarchical selection, and a strengthening of the second leg by structural con­straint, really does build a “higher Darwinism” of greater sophistication and explanatory power — an indispensable basis in our struggles to understand “this view of life,” the evolutionary process that made us, and imbued us with all the spandrels of body and soul that force us to ask such difficult ques­tions about the meaning of our own existence and of nature's ways. These spandrels of historical ancestry and structural inevitability may impede our search for solutions by imposing such quirky modalities upon our mental op­erations, but they also grant us more than sufficient power to overcome and prevail. Sweet, and adaptive, are the uses of adversity. Shakespeare, after all, in the words that follow this famous statement, parodied just above, prom­ised us salvation, or at least succor, in natural history, where we would find “tongues in trees, books in the running brooks, sermons in stones, and good in everything.”

  [Page 1296]

  CHAPTER TWELVE

  Tiers of Time and Trials of

&nb
sp; Extrapolationism, With an Epilog

  on the Interaction of General

  Theory and Contingent History

  Failure of Extrapolationism in the Non-Isotropy

  of Time and Geology

  THE SPECTER OF CATASTROPHIC MASS EXTINCTION:

  DARWIN TO CHICXULUB

  Greatness shines brightest in adversity. The logic of the Origin's 9th Chapter (1859, pp. 279-311) — “On the Imperfection of the Geological Record” — shows Darwin's reasoning at its very best and most systematic, all in the ser­vice of resolving his worst problem. For, in this chapter, he must explain why the subject that should, at first glance, have provided the strongest and most direct confirmation of evolution, seems to mock, in its opposite message (at least if read in an empirically literal manner), the gradual and incremental style of change touted throughout his book both as a validation of natural se­lection, and as the primary empirical basis for confidence in the factuality of evolution itself.

  Darwin's logic proceeds in a systematic way within the norms of scienti­fic discourse, moving linearly and relentlessly through the chapter from the problem with the easiest resolution (why do we not find living intermediary forms between modern species) to the most difficult appearance of all (epi­sodes of mass origination and extinction in the fossil record). The opening is­sue, representing a misconception and not a true threat, achieves an easy so­lution: evolution is a process of branching, not of linear transformation, as the brilliant “tree of life” metaphor, closing the operative chapter four on the mechanics of natural selection, had so well exemplified (1859, pp. 129-130). Few, if any, living species are the unaltered forebears of another modern form; rather, any two sister species have branched and diverged from an ancient common ancestor. Therefore, we do not expect to encounter living intermediates between extant species, for, although transitional forms surely existed, they died long ago and should only be found in the fossil record. [Page 1297]