The Structure of Evolutionary Theory

by Stephen Jay Gould

  I do not know if we should make a formal attempt to specify a definite number of tiers in time — as the principle seems sound, whereas definite crite­ria for precise designation seem elusive. (I remain far more comfortable and confident with the concept of hierarchy in levels of selection than with the in­creasing scale of tiers in time — for the genealogical hierarchy can be eluci­dated with fair precision by criteria of Darwinian individuality (see Chapter 8), whereas time's tiers lack such a unitary concept for coordination. For the same reason, I do not follow my closest colleague Niles Eldredge's attempt to identify dual or parallel hierarchies of genealogy and ecology, for the genea­logical units enjoy clear definition, whereas ecological levels, like time's tiers, lack a coherent fundamentum divisionis for unambiguous specification.)

  But if I may make an analogy to the geological time scale for the Phanerozoic eon, we may at least be confident about the few broadest tiers (the three geological eras in my analogy), even while we argue about some boundaries and specifications for the smaller units within these broad domains (geologi­cal periods in my analogy). Darwinian organismal selection, with an overall statistical edge granted to biotic competition in crowded ecosystems, domi­nates the first tier of anagenesis within populations during ecological time. If the fractal principle of Darwin's central belief in extrapolation held, then the norms of this process at the first tier would accumulate in a linear fashion through time to yield a history of life with the same basic form and causal structure, but scaled up in smooth continuity to generate all patterns of phylogeny. I have designated life's failure to display this pattern, particularly its disinclination to feature a clear signal of overall progress, as the “paradox of the first tier.”

  As stated above, I would resolve this paradox by accepting Darwin's ba­sic view of pattern and causality at the first tier, but then asserting that dis­tinct modes of change and balances of causes, operating at higher tiers, intro­duce enough systematic difference to cancel out the first tier's vector. If the first tier governs anagenesis within populations, then the second tier features trends within monophyletic clades. Darwinian tradition holds (as discussed [Page 1329] throughout this book, but primarily in Chapters 2 and 9) that such phyletic trends be explained, by simple extension and extrapolation, as adaptive ana­genesis carried further through time's geological fullness. I shall not here re­hearse the lengthy arguments and documentation of Chapter 9, but I believe that punctuated equilibrium, as the dominant pattern and process of the sec­ond tier during millions of “background” years between pulses of mass ex­tinction, undoes the first tier's vector of progress by supplying a radically different general explanation for phyletic trends: the differential success of certain species within clades, with each species treated as a stable individual during the several million years of its average geological existence. Since the reasons for differential success of species extend so far beyond (while also in­cluding) the traditional citation of adaptive biomechanical advantages for constituent organisms; and since several of these reasons tend to run orthogo­nal, or even counter, to general expectations of organismal progress; the crafting of trends by punctuated equilibrium derails extrapolation from the first tier, thus providing the second tier with a different set of explanations for its central phenomenon of cladal trending.

  This barrier at the second tier would be sufficient, by itself, to resolve the paradox of the first tier, but even the second tier's rules for trending do not ac­cumulate through broader realms of time to explain, by smooth extrapola­tion, the patterns of waxing and waning for major taxonomic groups. That is, the causes of cladal trends through long geological intervals of “normal” time do not accumulate to patterns of success and failure through the full Phanerozoic range. For, as the preceding section of this chapter documented, the “random” and “different rules” models of differential success in epi­sodes of catastrophic extinction then derail the phyletic trends of the second tier for the same basic reason that these cladal trends previously undid the anagenetic accumulations of the first tier: that is, by introducing new patterns and rules at these rare moments of maximal impact. Thus, in summary, adap­tive anagenesis of a single lineage at the first tier cannot be extrapolated to cladal trends within a monophyletic group of species at the second tier; and these cladal trends of the second tier then cannot be extrapolated through episodes of mass extinction to explain patterns of differential success for life's higher taxa throughout Phanerozoic time. Punctuated equilibrium undoes anagenesis, and catastrophic mass extinction derails punctuated equi­librium.

  As with the workings of hierarchical levels in selection, the effects of adja­cent tiers of time may interact in all possible ways. Higher tiers do not auto­matically counteract lower tiers. Adjacent tiers may also reinforce each other to intensify a signal in life's history — as I illustrated, for example, in reporting Jablonski's claim (1987) that species selection on species-rich subclades at the second tier often reinforces the adaptive advantages gained by the organisms of these species in Darwinian selection at the first tier. Similarly, if only for exaptive merits at the higher tier, diatoms flourished by virtue of the same fea­tures at both the first and third tiers (if my argument on page 1319 holds) — as [Page 1330] the evolution of mechanisms for dormancy proved adaptive at the first tier in aiding survival during high-latitude months of darkness and in times of low silica between episodes of upwelling; whereas the same feature may have fa­vored the persistence of diatom species during a prolonged period of darkness imposed by a global dust cloud excavated in bolide impact, the key ingredient advocated by many researchers for the killing scenario of the K-T event.

  But adjacent tiers may also act in an orthogonal manner, with invocation of the “random” model at the third tier as an obvious general case — for truly random differential survival must run orthogonal to deterministic Darwinian reasons for evolving the clade's distinctive traits at the first tier, or equally de­terministic (but different) reasons for establishing the defining features of cladal trends by punctuated equilibrium at the second tier. Finally, the evident possibility of opposing reasons at adjacent tiers has sparked our interest in catastrophic mass extinction from the start — with dinosaurian superiority over mammals maintained at the first and second tiers throughout Mesozoic times, and mammalian success then achieved by the “different rules” model at the third tier of differential passage through mass extinction, putatively based upon the very features that marked the first and second tier failures of mammals for 130 million previous years.

  Although limited space and personal competence prevent my proceeding beyond this sketchy and cartoonish model of three tiers, I suspect that future work will identify several inhomogeneities and subtiers, particularly between the history of independent and individual clades at the second tier and the co­ordinated impact of environments upon entire biotas at the third tier. Several intermediate modes and processes, affecting groups of species during times of unusual externalities in particular geographic regions, but not global biotas at catastrophic moments, must “intervene” between the pure influence of punctuated equilibrium upon a single clade and the full impact of world­wide catastrophe upon a global biota. For example, the model of coordinated stasis (discussed on pp. 916–922) argues that cladal trends do not always maintain the implied freedom of punctuated equilibrium to proceed indepen­dently, and in an unconstrained manner by differential success in the generation of new species at the second tier, but will often be subject to a form of community stasis that must first be broken by disruptions smaller than mass extinction, and resident within its own tier. Similarly, between the first and second tier, ordinary anagenesis often cannot “push through,” even to the point of disruption by punctuated equilibrium at the second tier, because the ecological communities that set the anagenetic regime for a single species be­come disrupted on a scale of hundreds of thousands of years by the climatic fluctuations of Milankovitch cycles that break up communities and quickly
disperse their elements into new arrangements in different places.

  To conclude this section with a historical example of the profound distinc­tion between traditional untiered views of selective domination smoothly scaled up to all times and magnitudes, and the alternative nonfractal con­cept of a much broader range of potential outcomes engendered by interac­tion among the characteristic modes and processes at different tiers, consider [Page 1331] the famous conclusion from Hatcher, Marsh and Lull's 1907 monograph on the anatomy, taxonomy, evolution and extinction of ceratopsian dino­saurs. In ending their text with a section on “probable causes of extinction,” these three great paleontologists only allowed themselves to consider the two standard hypotheses of the conventional uniformitarian school of pure ex­trapolation, complete with its Darwinian assumption that vectors of general progress must pervade such systems (in this instance by the replacement of di­nosaurs with mammals, for reasons of warm-blooded superiority in anatomi­cal design and mentality). In either case, they argue, the extinction must spread through an extended time in several episodes, step by step. The two reasons themselves — biotic competition from a superior group, and adaptive failure in the face of changes in the physical environment imposed by ordi­nary terrestrial forces (acting at greater than their usual intensity, but still in their characteristic mode) — cause inferior groups to peter out as life gets better and better through each geological day:

  Several theories have been advanced as to the probable causes of extinc­tion of the Ceratopsia ... It seems that animals of another race, or hordes of creatures which emigrated from another region, would be more likely to exterminate their predecessors. The mammals fulfill the requirements of a new foe, and the development of the frill in the Cera­topsia has been considered as meeting the necessity for a better protec­tion of the neck blood vessels from the weasel-like attack of small but bloodthirsty quadrupeds. Another notion . . . was that the Cretaceous mammals sought out the eggs of dinosaurs and destroyed them — Cope even going so far as to suggest the Multituberculata, with their long, sharp anterior teeth, as the probable offenders...

  By far the most reasonable cause . . . seems to be changing climatic conditions and a contracting and draining of the swamp and delta re­gions caused by the orographic upheavals which occurred toward the close of the Cretaceous. The Ceratopsidae and their nearest allies, the Trachodontidae, both highly specialized plant feeders, were unable to adapt themselves to a profoundly changed environment because of this very specialization, and, as a consequence, perished.

  That the Ceratopsia made a gallant struggle for survival seems evident, for they lived through the first series of upheavals at the close of the Laramie and also the second series at the close of the Arapahoe, which were accompanied by great volcanic outbursts in the Colorado region; but the changes accompanying the final upheaval which formed most of the great western mountain chains and closed the Mesozoic era gave the death blow to this remarkable race (Hatcher, Marsh and Lull, 1907, p. 195).

  But new perspectives from two higher tiers have reversed this convention­ality, particularly for a vigorous group like the ceratopsians. At the second tier, these particular dinosaurs, among all other subclades, remained in maxi­mal flourish of expansion and speciation right to the close of Cretaceous [Page 1332] times, and surely cannot be marked as doomed, or even in decline with re­spect to mammals, during such a period of maintenance and expansion by co­pious speciation, or introduction of new Darwinian species-individuals at this macroevolutionary level. And then, with catastrophism reintroduced at the third tier as a hypothesis of renewed respectability, the ceratopsians died, in concert with all other dinosaurs (leaving the anatomically divergent birds as sole survivors of their monophyletic clade), when an unpredictable paroxys­mal change radically altered earthly environments and drove several groups to extinction through no adaptive failure of their own, while imparting fortu­itous exaptive success to creatures that had lived throughout the long reign of dinosaurs, and never made any headway towards displacement, or even to­wards shared domination with one of the most successful vertebrate groups in the history of life.

  An Epilog on Theory and History in Creating the Grandeur of This View of Life

  This comfortable view of ceratopsian (and all dinosaurian) demise engen­dered smug feelings among evolutionists and paleontologists of previous gen­erations for two reasons, both lamentable. First, the implied pattern of a lawlike and predictable vector of progress, culminating in mammalian vic­tory over dinosaurs and crowned by the eventual evolution of a single con­scious scribe within the triumphant clade, validated the oldest social tradi­tions and deepest psychological hopes of Western cultures — the strongest possible reason for turning our brightest beacon of skepticism upon so conge­nial a conclusion defended by so little beyond emotional satisfaction. Second, the supposed underpinning (and virtual guarantee) of this happy result by a putative general law of nature, enhanced the meaning and centrality of the particular outcome as a dictate of universal science, not merely a fortuitous circumstance, or even a special dictate of an arcane controlling power whose comprehensive reasons can never be entirely known (and whose future ac­tions can therefore never be fully anticipated).

  If, however, as the central thesis of this book maintains and the Zeitgeist of our dawning millennium no longer rejects, we cannot validate the actuality of mammalian success by general principles, but only as a happy (albeit entirely sensible) contingency of a historical process with innumerable alternatives that didn't happen to attain expression (despite their equal plausibility before the fact), then we must face the philosophical question of whether we have surrendered too much in developing a more complex and nuanced view of causality in the history of life.

  What is science, after all, if not the attempt to understand the natural world by explaining its phenomenology as causal consequences of spatio-temporally invariant laws? We may need to know the particularities of a given set of initial conditions in order to infer the details of later states reached by the operation of these laws, but we do not regard the resolution of [Page 1333] such details as essential or causal components of the explanation itself. (I con­fess that, after 30 years of teaching at a major university, I remain surprised by the unquestioned acceptance of this view of science — which, by the way, I strongly reject for reasons exemplified just below — both among students headed for a life in this profession, and among intellectually inclined people in general. If, as a teacher, I suggest to students that they might wish to con­strue probability and contingency as ontological properties of nature, they of­ten become confused, or even angry, and almost invariably respond with some version of the old Laplacean claim. In short, they insist that our use of probabilistic inference can only, and in principle, be an epistemological con­sequence of our mental limitations, and simply cannot represent an irreduc­ible property of nature, which must, if science works at all, be truly determin­istic.)

  Natural historians have too often been apologetic — but most emphatically should not be — in supporting a plurality of legitimately scientific modes, in­cluding a narrative or historical style that explicitly links the explanation of outcomes not only to spatiotemporally invariant laws of nature, but also, if not primarily, to the specific contingencies of antecedent states, which, if con­stituted differently, could not have generated the observed result. As these an­tecedent states are, themselves, particulars of history rather than necessary expressions of law, and as subsequent configurations can cascade in innumer­able directions, each crucially dependent upon tiny differences in the anteced­ent states, we regard these subsequent outcomes as unpredictable in principle (as an ontological property of nature's probabilistic constitution, not as a lim­itation of our minds, or as a sign of the inferior status of historical science), however fully explainable they will become, at least in principle, after their occurrence as the single actualized result among innumerable unrealized pos­sibilities.

  In order
to gain entry into the hallowed halls of science (often defined, far too parochially, in terms of quantified predictability as a summum bonum), natural historians have often been too willing to accept an inferior status, based on the principled unpredictability of their largely contingent phenom­enology, in order to gain recognition as practitioners of science at all. (For in this Laplacean construction, the frequency of probabilistic inference corre­lates directly with the weakness of scientific apparatus — for we live, under this fallacy, in a genuinely deterministic world, and the extent of our recourse to probability therefore maps our relative inability to define the true deter­minisms of any particular process.)

  Wise natural historians, with Darwin himself as a most articulate cham­pion, have always rejected this disparagement, and its attendant relegation to inferior status — and have defended historical explanation, with its claims for contingency and the ontological status of probabilistic structure, as a fasci­nating, even inspirational, property of complex nature. Such contingency, moreover, in no way compromises the power of legitimate explanation, for our inability to predict before the fact only records the true character of this complexity, whereas our subsequent capacity to explain after the fact can [Page 1334] reach the same level of confidence as any physical resolution under invari­ant law, provided that we can obtain enough factual detail about anteced­ent states to resolve their causal relation to the observed outcome. In fact, and on this very subject, Darwin made a striking exception to his astonishingly calm and genial temperament, and permitted himself a rare excursion into satire and sharp criticism. Moreover, he expressed these partisan thoughts in the most prominent of all possible places — the very last line of the Origin of Species, where he rejected the traditional claims of quantitative physical sci­ence to represent the apotheosis of sophistication, and awarded higher honor to his own discipline of natural history and evolutionary biology, as embod­ied in the gnarly and meandering icon of the luxuriantly, but contingently, branching tree of life.