The Structure of Evolutionary Theory

by Stephen Jay Gould

  In a striking example, summarizing both the biological argument for grad­ual replacement by competition and the geological claim for false appearance of simultaneity by imperfection of preserved records, Darwin makes his plau­sible case for extrapolation and uniformitarian explanation, even for the two most famous cases of mass extinction for formerly prominent groups (see pp. 1314–1316 for a modern perspective on the demise of these taxa): trilobites at the Permo-Triassic event, and ammonites at the Cretaceous-Tertiary mass dying (pp. 321-322):

  With respect to the apparently sudden extermination of whole families or orders, as of Trilobites at the close of the Paleozoic period and of Ammonites at the close of the secondary period, we must remember [Page 1303] what has been already said on the probable wide intervals of time be­tween our consecutive formations; and in these intervals there may have been much slow extermination. Moreover, when by sudden immigration or by unusually rapid development, many species of a new group have taken possession of a new area, they will have exterminated in a corre­spondingly rapid manner many of the old inhabitants; and the forms which thus yield their places will commonly be allied, for they will par­take of some inferiority in common. Thus, as it seems to me, the manner in which single species and whole groups of species become extinct, ac­cords well with the theory of natural selection.

  I have discussed Darwin's defense of uniformitarian extrapolation in detail because his argument, in this case, proved so successful in directing more than a century of research away from any consideration of truly catastrophic mass extinction, and towards a virtually unchallenged effort to spread the deaths over sufficient time to warrant an ordinary gradualistic explanation in conventional Darwinian terms, with any environmentally triggered accelera­tion of rate only serving to intensify the effects of ordinary competition, spe­cies by species. I can't think of any other prominent subject in paleontology where uniformitarian presuppositions clamped such a tight and efficient lid upon any consideration of empirically legitimate and conceptually plausible catastrophic scenarios. Merely to suggest such a thing (as even so prominent a scientist as Schindewolf, 1963, discovered) was to commit an almost risible apostasy.

  In particular, these uniformitarian assumptions about the extended dura­tion of apparent mass extinctions led geologists and paleontologists to favor earth-based rather than cosmic physical inputs (for most plausible extrater­restrial causes work with greater speed and intensity), and to focus upon tel­luric influences (like changing climates and sea levels) that could most easily be rendered as gradualistic in style. So strongly entrenched did this prejudice remain, even spilling over into popular culture as well, that a few years after Alvarez et al. (1980) published their plausible, and by then increasingly well affirmed, scenario of extraterrestrial impact as a catastrophic trigger for the Cretaceous-Tertiary event, the New York Times even ridiculed the idea in their editorial pages, proclaiming (April 2, 1985) that “terrestrial events, like volcanic activity or changes in climate or sea level, are the most immediate possible causes of mass extinctions. Astronomers should leave to astrologers the task of seeking the cause of earthly events in the stars.”* [Page 1304]

  Thus, for example, when Gilluly (1949) published one of the most famous and influential geological papers of the mid-twentieth century, arguing that one popular physical theory for mass extinction — episodes of orogeny, or mountain building — could not be construed as either global in effect or si­multaneous in occurrence on all continents, but should rather be interpreted as sequences of more limited local events liberally spread out in time, he ended his paper with a stirring manifesto: “Long live Charles Lyell and his doctrine of uniformitarianism!” And, if I may cite an embarrassing incident from my own graduate career, when my mentor Norman Newell decided to invest considerable effort in compiling data from faunal lists in the world's paleontological literature to see if the maligned and effectively abandoned theme of mass extinction held any validity, I thought that the old man had taken leave of his good scientific sense in wasting so much time on a truly set­tled issue. For didn't we all know that the extinctions really spanned consid­erable intervals of time, and that any blip detected from faunal lists could only be recording an artifact of longer periods artificially compressed into si­multaneity by imperfections of the fossil record?

  Only with this understanding of the historical impact and persistence of Darwin's uniformitarian and extrapolationist view of extinction in the fossil record can we grasp the conceptual reforming power (and not merely the phenomenological fascination) of the improving case — from a wild idea re­jected out of hand by nearly all paleontologists in 1980, to a firmly docu­mented virtual fact of nature by 2000 — for the triggering of at least one mass extinction, the Cretaceous-Tertiary event, by impact of a large extraterrestrial object (see Alvarez et al., 1980, for the original proposal, and Glen, 1994, for history of science in progress in a book entitled: The Mass Extinction De­bates: How Science Works in a Crisis).

  Two comments on the K-T (Cretaceous-Tertiary) transition, one new and one old, may be taken as emblematic of the magnitude of both theoretical and practical reformulation. First, M. J. S. Rudwick, a prominent systematist of fossil brachiopods early in his career and the world's leading histo­rian of geology in later years, commented to Glen with the professional skills and “feel” of both segments of his ontogeny: “It never crossed my mind that . . . the brachiopod groups I worked with expired suddenly by modern K/T boundary standards. I thought that the brachs went out suddenly, but that 'suddenly'... in 1967 meant a few million years, which was considered geo­logically sudden” (quoted in Glen, 1994, p. 41).

  Second, an argument prominently advanced by Charles Lyell himself dramatically illustrates the difference between strictly uniformitarian expecta­tions and the implications of truly catastrophic triggers for mass extinction. [Page 1305] Lyell, as well known and recorded (see Gould, 1987b, for example), named the epochs of the Tertiary Era by a statistical method based on the percentage of molluscan species still extant — from Eocene (or “dawn of the recent” for the lowest percentage) to Pliocene (or “more of the recent” for the much higher percentages of later strata). He then noted that the uppermost Creta­ceous (Maastrichtian) and lowermost Tertiary beds held no species in com­mon at all. By his argument of statistical gradualism, this complete non-over­lap could only be explained by a vast gap of missing time — a period long enough to remove all Cretaceous species, one by one, at the same rate as the Tertiary demise. But since the entire Tertiary did not suffice to overturn the molluscan fauna completely (as a few Eocene species still survive), Lyell rea­soned that a globally unrecorded interval of time, longer than the full Tertiary span (so well represented by a voluminous paleontological record throughout the world), probably intervened between the latest Cretaceous and earliest Tertiary strata. As Lyell's hypothetical missing interval of more than 65 mil­lion years actually spans only a geological moment under the impact sce­nario, I would nominate Lyell's following statement (1833, p. 328) as the worst forecast ever made under the uniformitarian method of extrapolating from a range of observed rates!

  There appears, then, to be a greater chasm between the organic remains of the Eocene and Maastricht beds, than between the Eocene and Recent strata; for there are some living shells in the Eocene formations, while there are no Eocene fossils in the newest secondary [that is, Maastricht­ian or uppermost Cretaceous] group. It is not improbable that a greater interval of time may be indicated by this greater dissimilarity in fossil re­mains ... We may, perhaps, hereafter detect an equal, or even greater se­ries, intermediate between the Maastricht beds and the Eocene strata.

  Despite the uniformitarian consensus from Darwin's time until the late 20th century, occasional scholars of high reputation continued to float cata­strophic proposals for the unresolved puzzle of mass extinctions that, despite the orthodox conviction about “spreading out” into missing geological time, never meshed well with gradualist presu
ppositions and continued, like the proverbial sore thumb, to stick out above a comfortable background. But, in fairness, we cannot blame geologists and paleontologists for rejecting these proposals because, to cite a familiar motto, extraordinary claims require ex­traordinary evidence — and these attempts to resuscitate catastrophism re­mained entirely speculative (or at least undocumented by anything beyond the basic data for mass extinction itself, an evidentiary source that had al­ready, to the satisfaction of an entire profession, been rendered consistent with uniformitarian presuppositions).

  To cite the two most notable examples from the generation before Alvarez, Schindewolf (1963), in an article entitled “Neokatastrophismus,” proposed bursts of cosmic radiation as the paroxysmal mechanism of mass extinc­tion — with direct nuclear death (for the exterminations) and vast increases in mutation rates among survivors (for subsequent replacements by highly altered forms). [Page 1306] But, to show the frustration (and scientific nonoperationality) of such proposals, Schindewolf actually stated — thus providing a favorite case that I have used for decades to illustrate the difference between science and speculation — that he had postulated cosmic radiation explicitly because such a cause would leave no empirical sign (then known to geologists) in the record of strata and fossils. (For Schindewolf had to admit that the empirical record revealed no direct evidence at all for a catastrophic mechanism of mass extinction, and he therefore had to seek a potential cause that would leave no testable sign of its operation! Can one possibly imagine an unhappier situa­tion for science? — to face the prospect of a plausible explanation that does not, in principle, leave evidence for its validation.)

  In a second example, well remembered by paleontologists of my genera­tion, Digby McLaren used his presidential address to the Paleontological So­ciety in 1970 to hypothesize that a bolide impact had triggered the Devonian mass extinction. In the light of Alvarez's later triumph with a similar explana­tion for the K-T event, one might be tempted to view this address as pro­phetic. But, much as I admire both Digby himself and iconoclasts in general, I'm sure McLaren would admit that he simply “lucked out” in this case. For, like Schindewolf, McLaren could present no evidence at all for his bolide, and simply slipped this proposal into the end of his talk in an almost apologetic manner, after documenting the style and extent of the extinction itself (the main focus of his paper). (Moreover, to this day and despite excellent evidence for bolide triggering of the K-T event, we have no satisfactory expla­nation of the Devonian extinction, and no credible data for its causation by impact.)

  By contrast, the genesis of the Alvarez's hypothesis for the K-T mass extinc­tion could not have been more different, or more exemplary for science. For the K-T bolide proposal began with an unanticipated empirical discovery — generated, ironically, during a test for an opposite hypothesis, and therefore surely not gathered under the aegis of any iconoclastic theoretical thoughts. Geologist Walter Alvarez, trying to test an idea about latest Cretaceous sedi­mentation rates under traditional gradualist views of the extinction, asked his father, the Nobel laureate in physics Luis Alvarez, whether any isotopic signa­ture might provide evidence for the following conjecture: Walter wondered whether a false appearance of rapidity in extinction might arise from an un­usual slowdown in sedimentation rates, thereby compressing a “standard” amount of extinction into an unusually short stratigraphic interval.

  Luis proposed a measurement or iridium, an element virtually absent from the earth's indigenous surficial rocks. (Presumably, the earth formed with iridium at standard cosmic abundance, but this indigenous iridium, as such a heavy and unreactive element, quickly sank well below the surface, especially since the earth's crust was effectively molten early in the planet's history.) Luis therefore supposed that throughout Phanerozoic time, iridium has en­tered the earth's surface only through cosmic influx, and at the effectively constant rate of uniformitarian assumptions — the “gentle cosmic rain from heaven” in radiation and tiny particles, as standard views and terminology [Page 1307] then held. Luis reasoned that Walter's hypothesis could therefore be tested by measuring iridium concentrations in latest Cretaceous sediments, arguing that a small positive excursion would validate Walter's supposition, as the constant cosmic influx became diluted by less than the usual amount of iridium-free terrestrial sediment.

  But when the Alvarez team measured iridium in boundary-layer sediments, they found a value so high that they had to invert their initial assumption in the most radical manner. Terrestrial sedimentation would have to cease for longer than the earth's entire history to produce such a high spike from a low and constant cosmic influx. Rather, they now reasoned, a true and sudden in­flux of iridium must have occurred right at the K-T boundary itself — with the obvious “culprit” as a large extraterrestrial body striking the earth and im-placing an enormous and momentary dose of cosmic iridium. Thus, the re­vival of catastrophic theories for mass extinction began with an empirical surprise generated during a test for a conventional gradualistic hypothesis — the exact opposite (in both form and utility) of previous exercises in evidence-free and catastrophically driven speculation.

  This thoroughly different character of the Alvarez hypothesis — as an evi­dence-driven claim bursting with seeds of testability, rather than a sterile speculation — should have caught the attention and intrigue of all scientists from the start. But the anti-catastrophic biases of Lyellian and Darwinian tra­ditions ran so deep, and the knee-jerk fear and disdain of paleontologists therefore stood so high, that even this welcome novelty of operationality did not allay rejection and outright disdain from nearly all established profes­sional students of the fossil record (whereas other relevant subdisciplines with other traditions, planetary scientists and students of the physics and en­gineering of impacts, for example, reacted in markedly more mixed or posi­tive ways — see Glen, 1994). I will never forget a 1979 phone conversation (as the preprints of Alvarez et al., 1980, circulated) with David Raup, perhaps the only other invertebrate paleontologist of my generation who reacted with initial warmth to the impact hypothesis:* It was one of those laconic affairs, [Page 1308] where no more need be said: Raup: “This time it's different you know.” Me, in reply: “Yes, of course, the iridium.”

  In the retrospect of a mere twenty years between initial proposal and such substantial success (although by no means total, in either its own hopes or terms), we may identify several reasons to honor the conventional criteria used by scientists to judge the strength and importance of hypotheses — crite­ria based on empirical affirmation, fruitful extension, and widening intellec­tual scope, rather than on such nonoperational notions as progress towards absolute truth. Science is, as P. B. Medawar stated in the title to his finest book, the Art of the Soluble:

  1. At their initial decision to publish, the Alvarezes had detected an iridium spike only at two nearby localities in Denmark and Italy, and couldn't even be confident in their theory's crucial prediction of a worldwide enhancement. (Fortunately for them, evidence for a third and virtually antipodal spike from New Zealand arrived in time for inclusion in the original publication.) But, within a decade, affirmation had accumulated from so many collateral sources — all independent of, and unpredictable from, the iridium spike it­self — that the case for impact had effectively been sealed. These additions ranged from shocked quartz in the K-T boundary layer throughout the world (with silica tetrahedra arranged in an unusual manner only associated, so far as we know, with high pressures of impact, including the initial discovery of such forms in nuclear bomb craters), to the “smoking gun” of a gigantic crater of exactly appropriate age — the Chicxulub structure off the Yucatan peninsula in Mexico. As this evidence accumulated, the alternative volcanist scenario (with iridium recruited from the earth's interior in extrusive events of a magnitude never witnessed in historical times), which had provided such good material for fruitful debate, yielded the floor (although volcanic action initiated by impact may have played an import
ant role in the full scenario of explanation for mass extinctions). Needless to say, other exciting hypotheses generated by the impact debate — particularly Raup and Sepkoski's claim for a 26 million year periodicity in extinction, with a subsequent set of wondrous astronomical hypotheses as potential explanations, including the actions of a previously unknown dwarf companion star to the sun — have not fared so well (although a few jurors, with a few good arguments, are still holding out).

  2. Enhanced and surprising interdisciplinary communication offers no guarantee of scientific rectitude or success, but we can only celebrate the veri­table orgy of exciting, and at least intellectually fruitful, discussion and col­laboration inspired by the impact hypothesis among scientists in subdisciplines that had never read each other's work, hardly even knew the names of the most reputable leaders in the disparate domains, and could barely speak [Page 1309] the same scientific language (most notably of Linnaean Latin vs. Newtonian triple integral signs). Paleontologists met, and even eventually published pa­pers with, colleagues who had formerly received little more than a grunted “hello” in 20 years of hallway passing, or who had only been seen as strang­ers across a crowded room during some unenchanted evening at a faculty party. The several Snowbird conferences in Utah will never be forgotten by anyone who enjoyed the privilege of attendance, and who participated in the warm fellowship, and sometimes-antic debate, with (among many others) nu­clear physicists, taxonomic paleontologists, and historians of science.

  3. Most importantly, and diagnostically for scientific practice, the impact hypothesis proved its mettle (at least for me) in the explicit suggestions and prods that it provided for particular (and ultimately highly fruitful and excit­ing) paleontological research that would never even have been conceptualized without its nudge and encouragement. I have argued throughout this book that the broad world-views of scientists (with gradualism, uniformitarianism, and strict Darwinian adaptationism as the major examples in this context) do not merely act as passive summaries of general beliefs, but serve as active definers of permissible subjects for study, and modes for their examination. At best, a potent context continually provokes more fruitful work. But at worst, and (unfortunately) ever so often in the history of science, such world-views direct and constrain research by actively defining out of existence, or simply placing outside the realm of conceptualization, a large set of interesting sub­jects and approaches, often including the very classes of data best suited to act as potential refutations of the world-view. Such self-referential affirma­tions are not promoted cynically, or (for the most part) even consciously, but they do, nonetheless, operate as strong impediments to scientific change.