The Structure of Evolutionary Theory

by Stephen Jay Gould

  The key phenomenon for this entire discussion — whether explanation be offered by Darwinian extrapolation, or by the random or different rules alter­natives — has always resided in the observed selectivity of mass extinctions: why do some taxa flourish and others die, especially since the observed pat­terns of mass extinctions do not simply intensify the tendencies of normal times (that is, mass dyings do not preferentially remove those groups already on the wane by competition with superior forms during “background” times. Mammals were not expanding, with dinosaur retreating, during the long span of Cretacous life).

  Under the “different rules” model, extirpated groups die for definable rea­sons of conventional anatomy, physiology, behavior, or population structure. But their death follows from the unpredictable, and suddenly instituted, “dif­ferent rules” of catastrophically altered environments in episodes of mass ex­tinction, and does not occur because these taxa had evolved properties that would have doomed them in the same manner (albeit more slowly and se­quentially) for ordinary reasons of Darwinian competition during normal times. In fact, and even worse for conventional arguments about progress, the traits that spell doom in catastrophically altered circumstances may just as well have originated as the adaptive features that secured success, and com­petitive superiority in the normal Darwinian times just preceding. In this important sense, if the previous Darwinian “best” often die for unpredictable but deterministic reasons in the suddenly altered worlds of catastrophic mass extinction, then Darwin's crucial argument for progress (always weak and suspect because it could not flow from the abstract logic of natural selection itself, and required an additional ecological belief in plenitude and biotic struggle) collapses through the disruption, or even the reversal of its vector, as imposed by these dramatic episodes with their different rules for who flour­ishes and who goes to the wall. And if these episodes are sufficiently numer­ous, profound, rapid and different (my four criteria of p. 1313), then their ac­cumulated impact may balance, or even reverse, the Darwinian accumulation during much longer stretches of normal times, thereby imbuing the full pattern [Page 1317] of life's history with a very different form (perhaps entirely devoid of any meaningful vector of progress) than the extrapolationist premise of Darwin­ian central logic had assumed.

  Thus, the influence of the “different rules” model in helping to explain the waxing and waning of taxa in macroevolution represents the most interest­ing and far-reaching modification of Darwinian expectations unleashed by catastrophism's renewed respectability, and by the resulting inadequacy of uniformitarian extrapolation from Darwinian microevolution to supply a full explanation for the causes of pattern in life's history. After all, and in admit­ted caricature, if the size of dinosaurs had marked their success over mam­mals in habitats of large terrestrial vertebrates for more than 130 million years; and if this established basis for Darwinian success then became an im­portant causal factor in their differential death, with small size as an equally vital reason for mammalian survival; then the tables truly turned with the in­stitution of these particular different rules of a K-T moment — as marks of failure (or at least of limitation) in the long background of Darwinian compe­tition became fortuitous substrates for survival, while the primary features of former and such prolonged domination became unalterable portents of doom (whether directly or, more probably, via such correlated consequences as gen­erally smaller population sizes and generally greater ecological specializa­tion). “How are the mighty fallen” exclaimed David in his famous lament (2 Samuel: l). But Goliath had died by David's superior wit (and good aim), whereas Saul expired as a consequence of his own madness. Dinosaurs, on the other hand, may have fallen by their might, but surely not by their fault.

  Thus, for these good reasons, the paleontological literature on the biologi­cal implications of mass extinction has, for the past 20 years, rightly focused upon the documentation of “different rules” in such potentially catastrophic episodes, and on their impact upon pattern in the history of life. I cannot, in the context of this chapter, present a compendium of these ever increasing, and ever more sophisticated, studies. I shall therefore cite just a few of the early results that have already become classic in a burgeoning field. I should also at least acknowledge — to open up a real can of worms that I shall not even attempt to close — that, at least partly as a “back formation” from doubts inspired by the debate on mass extinction, the entire subject of the efficacy of competition, in normal times as well as in geological perspective, has fallen under increasing question (see Simberloff, 1983,1984 for some ex­plicitly microevolutionary doubts; Gould and Calloway, 1980, for a paleon­tological example; Benton, 1996, for a general argument from the fossil re­cord; but also Sepkoski, 1996, for a strong and more nuanced defense of competition in paleontologically normal times, and Sepkoski et al., 2000, for a beautifully documented and fascinating example, and Gould, 2000b, for commentary thereupon).

  Jablonski published the most important early work on this subject, estab­lishing a methodology and terminology in his classic paper (Jablonski, 1986b) on “Background and mass extinctions: the alternation of macroevolutionary regimes.” [Page 1318] Some of the specifically K-T patterns will strike no one as surprising, given the context; but the readjustment of diversity under such a pronounced moment of different rules still imposes a major signal upon the overall pattern of life's history. For example, land plants reproduc­ing by seeds, rhizomes, or any other mode of propagation by bodies that can lay dormant under soil, tended to survive at higher frequency (Wolfe, 1990). And Sheehan and Hansen (1986) found higher death rates among animals di­rectly linked in their feeding to a supply of living plants, whereas feeders on dead plant material, scavengers and detritivores tended to fare better.

  Jablonski's studies break more general conceptual ground in their broader scope across several extinction events and in their search for commonalities across mass dyings rather than reactions to a specific impact scenario of the K-T event. He argues (in Jablonski and Bottjer, 1983, for example) that spe­cies-rich clades tend to increase in numbers of taxa during background times (largely by species selection within subclades whose species-individuals gen­erate relatively more daughters, and therefore already a macroevolutionary claim that cannot be extrapolated from Darwinian organismal selection). However, these species-rich clades then tend to fail differentially in mass ex­tinction because the same properties that enhance the capacity for speciation in background times — stenotopy and limited capacity for dispersal, for exam­ple — make taxa more susceptible to removal in catastrophic episodes.

  In later extensions of the same theme, Jablonski (1986a, 1987) established the best-documented case of genuine clade selection in evolution (working contrary, or at least orthogonally, to species selection, and therefore atop the hierarchy of potential levels at an apex that, at least in my judgment, proba­bly operates only rarely in nature — see pp. 712–714 for further discussion). He found no consistent relationship between the properties of species within molluscan clades and the full clade's propensity for survival through mass ex­tinctions. However, the clade's entire geographic range (but not the individual ranges of its component species) correlated strongly and positively with survivorship through mass extinctions.

  Jablonski (1996) then documented a further disconnect between microevolutionary expectations in extrapolation, and macroevolutionary realities. In his rich database of late Cretaceous mollusks, Jablonski could find no evi­dence for the most venerable of all supposed generalities in trending — Cope's rule, or the tendency of lineages to increase in body size (see pp. 902–905 for further discussion). Nonetheless, good microevolutionary reasons — and data — can be cited for claiming a general selective benefit for increased body size that should (ceteris paribus), in a Darwinian world of extrapolation, yield Cope's rule in macroevolutionary extension. The reasons for this extrapolationist failure may be formulated at several levels, including forces op­erating in back
ground times as well. But one important factor also intervenes in the different rules of mass extinction, where susceptibility of clades “does not appear to have been size selective” (Jablonski, 1996, p. 279). Jablonski concludes, also citing his earlier work on geographic ranges (1996, p. 279): “Survivorship [in mass extinction] appears to have hinged on other factors [Page 1319] such as broad geographic range, which shows little or no correlation with body size in marine invertebrates. The analyses presented here reinforce the view that macroevolutionary patterns need not be simple extensions of those seen at the level of individual organisms over microevolutionary time.”

  I would also make the further observation that these discordances between background and mass rules (the “alternation” of regimes in Jablonski's termi­nology) must play a major role in the most general patterning of life's history, absent which the tree of earthly life would have grown in a markedly differ­ent shape. (In this most crucial sense, of course, I reassert the primary theme of this book that macroevolutionary theory matters profoundly.) For if differ­ent rules did not impose their signals at levels above microevolutionary ex­trapolation, the powerful themes of Darwin's world would push through to completion in life's phylogeny. For example, the species-rich clades of back­ground times, with their dual advantages in organismic and species-level se­lection, would eventually eliminate the species-poor clades entirely if a still higher-level component of advantage for at least some species-poor clades did not “kick in” during episodes of different rules in mass extinction. And if the general, albeit slight, statistical edge of larger body size scaled straight up from local populations to the global biota at geological scales, what would guarantee a world enriched with all the little shrews and hummingbirds of our delight, not to mention the continued existence of short people, including the author of this book.

  These generalities enter the corpus of macroevolutionary theory, but the different rules of mass extinction must still work through the specificities of various causes that provoke the rare, but potent, catastrophes of planetary history (with impact at the K-T boundary as the only firmly established case so far). Thus, some of the most interesting, if hypothetical, invocations have been proposed as explanations for specific and otherwise puzzling results of the K-T event. I have long been intrigued, for example, by the striking pattern of differential extinction in the oceanic plankton — with 73 percent of coccolithophorid genera, 85 percent of radiolarians, and 92 percent of forams failing to survive, while diatoms suffered only a 23 percent loss of genera.

  Kitchell, Clark and Gombos (1986) made the interesting argument, later supported by direct data of Griffis and Chapman (1988) on survivorship of phytoplankton in conditions of prolonged darkness, that the differential suc­cess of diatoms probably bears no relationship to any notion of cosmic “betterness” or general “progress,” but may only record the fortuity in exaptive use (under the different rules of K-T darkness) of adaptations evolved for or­dinary, short-term microevolutionary advantages in background times (and not, obviously, in anticipation of any additional edge in forthcoming catas­trophes!) Kitchell et al. (1986) argue that most diatom species have evolved mechanisms of dormancy (formation of resting spores, for example), permit­ting these photosynthetic organisms to survive extended periods of darkness, including several polar months per year for species living at high latitudes. Moreover, since diatoms build their skeletons of silica, which they can extract most readily in oceanic zones of, upwelling that can be uncertain in placement [Page 1320] and impersistent in timing, the capacity to “shut down” for periods of dor­mancy between such favorable moments gains a second important adaptive value.

  But these two reasons for dormancy — to survive seasonal periods of dark­ness at high latitudes and to “wait out” silica-poor times between episodes of upwelling — operate entirely in the microevolutionary world of ordinary Dar­winian selection. In this lucky case, these particular advantages did scale up to good fortune, given the factor favored by many researchers for the pri­mary agent of the K-T killing scenario — extended darkness from a global dust cloud generated by the impacting bolide and its excavated and elevated earthly products. Thus, in this hypothesis, the relative prosperity of diatoms vs. the relative destruction of forams and other planktonic groups arose by the fortuity of how key adaptations for background conditions happened to “play” in an unpredictable and utterly different world of catastrophic im­pact. These “different rules” happened to suit diatoms and spell disaster for forams.

  And if diatoms prevailed by such good fortune, let us not forget a key rea­son behind the possibility of this most immediate interaction between me as writer and you as reader. Dinosaurs and mammals had shared the earth for more than 130 million years, fully double the subsequent period of mamma­lian success that led to the possibility of Homo sapiens among some 4000 other living species in our mammalian clade. If the data of Sheehan et al. (1991) hold, and dinosaurs did persist in respectable abundance right to the moment of impact, then we may reasonably conjecture that, absent this ulti­mate random bolt from the blue, dinosaurs would still dominate the habitats of large terrestrial vertebrates, and mammals would still be rat-size crea­tures living in the ecological interstices of their world. In this most vitally per­sonal of all cases, we really should thank our lucky stars that, at least in one cogent interpretation, certain marks of our ancestral incompetence — persis­tently small size in a dinosaurian world, for example — suddenly turned into a crucial and fortuitous advantage under the different rules of K-T impact, while the former source of triumph for dinosaurs may have spelled their doom under these same newly imposed rules. To be sure, this speculative sce­nario only references a particular event, and its much later impact upon the possibility of origin for one odd species. Yes, of course, we seek general the­ory as the goal of science, not the explanation of such odd particulars. But this tale, above all, happens to be our particular, and the most precious source of our possibility. Enough said.

  THE PARADOX OF THE FIRST TIER: TOWARDS A GENERAL

  THEORY OF TIERS OF TIME

  Although I have tried to present a critical exegesis of both the sources and the logic of Darwin's argument for general progress as a broad statistical and ac­cumulating consequence of biotic competition in a crowded world, I must also confess that Darwin's rationale and development seem basically sound to [Page 1321] me. And yet, I do not think that the actual history of life maps the expecta­tions of his argument, thus leaving us with a central evolutionary paradox. Nearly all-vernacular understanding of evolution, and much professional in­terpretation as well, would deny my last statement and affirm a broad signal of such progress. Although I do appreciate the appeal of an argument based on “ratcheting” for an accretion of levels in complexity through time — for certain kinds of more elaborate aggregation and integration cannot viably disassemble once conjoined, although any level can be lost by extinction (my interpretation of the claims presented by Maynard Smith and Szathmary, 1995, for “the major transitions in evolution”) — I fail to find any rationale beyond anthropocentric hope and social tradition for viewing such a se­quence as a fundamental signal, or an expression of the main weights and tendencies in life's history. After all, two of the three great boughs on life's phyletic tree remain prokaryotic, while all three multicellular kingdoms ex­tend as twigs from the terminus of the third bough. If we regard intellectual skepticism against anthropocentrism as a worthy cause, I don't see how we can deny that the persisting domination, and continued rosy prospects, of prokaryotes epitomize the primary aspect of life's history (see Gould, 1996a, and pp. 897–901 of this book for an elaboration of this argument). And if we must honor animals in our parochialism, arthropods surely hold an enormous edge over vertebrates.

  I have referred (Gould, 1985a) to this failure of Darwin's sensible argument to impress itself upon the actual history of life as “the paradox of the first tier” — thus also giving away my preference between the two majo
r possibili­ties for resolution (see forthcoming discussion, and my defense of nonfractal “tiers” of time with different predominating causes and patterns, with Dar­win's good argument operating only at the first tier, and unable to “push through” to impose a pervasive vector upon the history of life). If we accept my characterization of this situation as a paradox, then we must ask why a valid argument for progress, based upon the uniformitarian extrapolation through geological time of the microevolutionary mechanics of natural selec­tion, fails to make its anticipated mark upon earthly phylogeny. As an ab­stract issue in logic, two “pure” end-member solutions can be specified be­cause the basic proposition includes two assertions, with the falsification of either being sufficient to destroy the full argument even if the other assertion remains entirely valid. (Needless to say, the actual resolution of the paradox in our messy “real world” will, no doubt, combine aspects of both with nu­merous other factors as well.)