The Structure of Evolutionary Theory

by Stephen Jay Gould

  8-7. Validation of Wright's Rule in a study by Arnold et al. (1995) of 342 ancestral-descendant pairs of Cenozoic planktonic foraminifera. Descendant species originate, with equal frequency, at larger and smaller sizes than their ancestors.

  [Page 734]

  8-8. Top: Clear speciational trends in Lower Paleozoic gastropods towards higher spires (A — measured as shell torque), more inclined apertures (B) and nar­rower sinuses (C). The bottom diagram demonstrates that the first two trends obey Wright's Rule in showing no bias in species origins in the direction of the trend. However, the trend for sinus width does show a bias for new species to originate in the direction of narrower sinuses — thus yielding a complex trend, partly produced by directional speciation, and not entirely by species selection. From Wagner, 1996.

  [Page 735]

  narrower sinuses, falsified Wright's Rule (see Fig. 8-8) and documented a drive of directional speciation. Wagner further demonstrated (1996, p. 1000) that “this bias is distributed throughout the entire clade,” for three major subclades all display the drive. Nonetheless — and showing the power of such data to identify and tease apart the different components of a trend into their relative quantitative strengths — Wagner also documented a component of species sorting in the overall trend to narrower sinuses, for “species with wide sinuses were significantly less likely to survive the end-Ordovician mass ex­tinction” (1996, p. 990).

  I would go further and suggest that synergisms of drive and sorting (as Wagner has documented for the trend to narrower sinuses in Paleozoic gas­tropods) should be common in the history of many clades, and probably mark a powerful mode of macroevolution distinct from conventional microevolution, where such synergism must be rare. Good a priori reasons exist for supposing that features biasing the directionality of speciation might also fa­vor sorting towards the same end. Such synergism should be most evident when the causes of both bias and sorting work at the same (usually organismic) level — as when, for example, a trait under strongly positive organismic selection (like large body size) arises preferentially in speciation events, and then promotes the greater longevity of species so originating. But such syner­gisms may also be common when causes differ in level — as when, for exam­ple, a drive occurs by organismal selection, and species-selection then causes sorting in the same direction. For, unlike the situation at the next lower pair­ing of levels (where genic and cell lineage selection so often run counter to the interest of organismal selection, and consequently become suppressed), selec­tion at the organismal level does not conflict in principle with selection at the species level. Selection at these two levels should, therefore, be synergistic as often as opposed. Such synergisms should therefore be frequent and powerful in macroevolution.

  Species-level drifts as more powerful than the analogous

  phenomena in microevolution

  At the organismal level, the second major mode of sorting — drift by random processes — operates in two ways that should be distinguished both for poten­tially different roles and frequencies at this level, and because the species-level analogs diverge even more clearly. We may distinguish random shift within the collectivity — called genetic drift at the conventional organismal level — from random effects introduced at the founding of new demes or species by small numbers of organisms. Mayr (1942) introduced the term “founder ef­fect” to distinguish this second category (though the basic mechanism does not differ from ordinary genetic drift), and to emphasize that the differences initiating a new species need not arise entirely by natural selection, but may be significantly enhanced by random effects at the outset, because a small number of founders will, for stochastic reasons, surely not begin a new popu­lation with the same gene frequencies as the ancestral population, while some [Page 736] alleles (even if favorable) will be lost by random non-inclusion in all founding organisms.

  Although both genetic drift and founder effects obviously occur at the organismal level, our traditions have tended to downplay the role of random processes vs. selection as sources of sorting — so the phenomena generally re­ceive short shrift. Some conventional arguments for genuine rarity at the organismal level may be valid, particularly given the requirement for either small populations or effective neutrality of drifting sites. (The initiating crite­rion of low N may, however, be quite generally met if Mayr's theory of peripatric speciation holds, hence his emphasis on the “founder principle.” Similarly, if bottlenecking to very small numbers typically occurs during the history of many species, then genetic drift also becomes important in ana­genesis. The argument for effective neutrality, as discussed previously (pp. 684–689), works best at the genic level, where drift may predominate by Kimura's neutral theory of molecular evolution.)

  However, at the species level, these traditional objections to high frequency for drift become invalid, and we should anticipate a major role for this sec­ond cause of sorting. Low population size (number of species in a clade) pro­vides the enabling criterion for important drift in both categories at the spe­cies level. The analog of genetic drift — which I shall call “species drift” — must act both frequently and powerfully in macroevolution. Most clades do not contain large numbers of species. Therefore, trends may often originate for effectively random reasons. Consider a trend produced by random deaths (a comparable argument can easily be made for random birth differentials), based on Raup's “field of bullets” model (1991 and Chapter 12). Suppose, for example, that each of the ten species of a clade lives in a small area, with each species allopatric to all others. Over a certain period of time, a bolide (or some gentler environmental change with power to drive a local species to ex­tinction) strikes half the areas at random and eliminates the resident species of the clade while each of the species in the five safe areas branches off a daughter, thus restoring the cladal population of 10 species. At an N this low, some trends (and perhaps a substantial number) will inevitably arise by this mode of random removal. Perhaps, for example, four of the five species with mean body size below the cladal average will happen to die. A substantial random trend to increased body size then occurs within the clade.

  When we move from the homogenous “field of bullets” model to a scaling of effects in the real world, and consider the consequences of infrequent, but severe, mass extinction on a global scale, the potential role of random trends by elimination only increases — for random effects based on small numbers will be greatly intensified. (The reduction of species number in mass extinc­tion may be conventionally causal, but the final death of the clade, after re­duction to less than a handful of species, may then be effectively random. For example, so few trilobite species still lived when the great Permian extinction occurred that I'm not sure we need to seek a “trilobite specific” cause for the final elimination of this previously dominant group.) [Page 737]

  When we move to the second category of random results achieved by sort­ing in the colonization of new places — the analog of the founder effect — then comparison with the organismal version becomes less straightforward, al­though we may be confident that the species-level version holds potential for great importance in evolution. The species-level analog, which I will call “founder drift” (see lines IIIC2 and IIIC2a), does not work through a sim­ple phenotypic difference between a colonizing species and the parental stay-at-home — for all species differ by definition, and disparities arise by the usual combination of selective and random effects, usually expressed at the organismic level. The stochastic analog to Mayr's “founder effect” at the organismal level lies in random aspects of the differential capacity for prolif­eration of new species in allopatric regions of a clade's full range.

  A hypothetical example will illustrate this unfamiliar concept. Suppose that a clade contains only two species, living in adjacent islands with simi­lar environments. The islands, however, lie on different oceanic plates, and movements of plate tectonics cause the coalescence of one island with a large neighboring continent, while leavi
ng the other island in the midst of the ocean. The species on the continent proliferates into a large subclade of new species, while the species on the island, lacking any room for expansion, re­mains as the only species of its subclade. Because the process of speciation yields phenotypic disparity intrinsically, the founding continental species will differ from its insular sister species. Therefore, the clade will show a strong trend in the direction of autapomorphic traits possessed by the continental founder. But such a trend will often be entirely random with respect to the plurified traits of the continental founder. That is, these spreading traits may be completely neutral in the crucial sense that if the other (insular) species had colonized the island that coalesced with the continent instead, its autapomorphies would have proliferated, and the cladal trend would have pro­ceeded with the same force, but in the opposite direction. Only the luck of residence on one island rather than the other (and not any preferential inter­action of some traits vs. others with the environment) leads to the differential proliferation of one species's traits over those of the sister species.

  Situations of this sort must be common, if not virtually canonical, in evolution. Almost any two geographic regions must maintain differential capacity to house species of a given clade. If both regions are colonized by founding species, and, many million years later, one region holds substantially more species than the other, the random component of spatial and ecological op­portunity must often play a greater role in differential speciation than the se­lective force of greater capacity for differential proliferation in one subclade vs. the other based on interactions of traits with environments. I use the term random in a special, but surely legitimate, sense. Suppose that a large and ecologically diverse Region 1 can accommodate 50 species of a subclade, while smaller and more homogeneous Region 2 can only maintain 10 (I real­ize that species create their own environments, and that regions don't main­tain fixed numbers of available addresses, but I invoke this simplification for [Page 738] the sake of argument). Subclade A invades Region 1, while Subclade B be­gins in Region 2. The resulting strong cladal trend toward the autapomorphic characters of Subclade A cannot be called accidental in the global sense — for Region 1 does predictably accommodate more species. But the trend may be accidental in the sense that Subclade A, rather than Subclade B, happened to invade the more prosperous region — and that if Subclade B had been the colonizer, its progeny would have done equally well, and would have dominated the cladal trend with the same force actually shown by Subclade A. In this case, we call the trend random because A's success does not arise from any superiority of an interacting trait (vs. B's phenotype), but only from the accident of colonizing a more propitious place (see Eble, 1999, and Chap­ter 11 of this book for a discussion of this evolutionary meaning of “ran­dom”).

  As with the relationship between directional speciation and species selec­tion, these two forms of species-level drift must often interact with the other main cause of sorting — i.e., selection — to produce a trend (as when Subclade A, in the example just above, increases both by the good fortune of greater opportunity, and by selective benefits conferred by its traits). The organismic level may experience a higher relative frequency of domination by selective forces, but the world of species evolves by complex interactions among the processes of drive, selection, and drift.

  The scaling of external and internal environments

  I have not tried to develop an exhaustive comparison between levels for influ­ences of external and internal environments upon the modes of change dis­cussed in previous sections. But I offer a few sketchy comments to encourage further work in this area.

  For environmental factors that induce competition among individuals and therefore establish selection pressures (line IVA of the chart), I contrast modes that involve direct contact among individuals with those that can proceed in allopatry. At the organismic level, this contrast exposes a strong correlation between prevalence of biotic factors in direct contact and abiotic factors in allopatry. At the species level, a different correlation may dominate: the asso­ciation of selection by differential elimination with direct contact, and selec­tion by differential birth with allopatry (lines IVA1 and IVA2).

  This contrast also leads to different implications at the two levels. At the organismic level, as Darwin himself argued in his primary justification for progress in the history of life (see Chapter 6), the biotic mode correlates more often with adaptation by general biomechanical improvement, and the abiotic mode with adaptation to local circumstances of the physical envi­ronment, with no vectorial component as environments fluctuate randomly through time. At the species level, we may expect to find a strong correlation of selection by differential elimination with potential reduction to the organismal level, while selection by differential birth represents the most promising domain for true and irreducible species selection. [Page 739]

  For constraints of internal environments (line IVB), I make a distinction between negative factors that limit amounts and directions of change, and positive properties that channel change in certain directions, or provide par­ticular opportunities for evolutionary novelties and breakthroughs. (I also base Chapter 10, this book's major discussion of constraint, on the same dis­tinction.) The operation of these constraints often differs in interesting ways at the two levels.

  For some of the important limits, line IVB1 specifies a major shaping force of life's structure, a factor not often explicitly acknowledged. Why does the world contain stable individuals at all, and at any level? Why doesn't evo­lution work as continuous flux at all scales, rather than primarily by selec­tion upon individuals stable enough to persist, at least through one round of differential sorting? Comparable reasons can be stated at both the organismic and species levels, thus giving evolution its primary shape or structure: Lamarckian inheritance does not occur at the organismal level, thus stabiliz­ing the ontogeny of heritable variability. At the species level, punctuated equi­librium suppresses anagenesis by maintaining species-individuals in stasis.

  When we explore the structural brakes that limit amounts of change in most trends (line IVB2), several factors could be mentioned, but I just list, as an example, the single property that I consider most important. For organ­isms, those paragons of individuality by the criterion of structural and func­tional integrity, design limits of the Bauplan (both internally by structural constraint, and externally by adaptive possibilities) place strong brakes upon almost any evolutionary trend. Contrary to the themes of several popular films, elephants will never fly, and insects will not reach elephantine propor­tions and engulf our cities as a plague of megalocusts.

  At the species level, Stanley (1979) made an important observation that has not been sufficiently appreciated for its defining force in limiting the possibili­ties of species selection. If we consider the two major modes of positive spe­cies selection — enhancing the rate of production for new species, and extend­ing the geological longevity of existing species — why shouldn't some lineages be able to maximize both properties simultaneously, thus becoming gigantic megaclades, dominating the earth's biota? (Perhaps, of course, a few clades have been able to approach this ideal — thus explaining the great success of beetles and nematodes.) In other words, why don't clades ratchet themselves towards this pinnacle by species selection — by working both ends of the game, and evolving species of extraordinary durability and fantastic rates of branching, superspecies that live for several geological periods and spawn large numbers of daughters all along the way?

  Stanley (1979) argues, with extensive data in support, that the nature of speciation as a process, and the general rules of ecology, engenders a strong, and effectively unbreakable, negative correlation between speciation and ex­tinction rates. Unfortunately for ambitious species with dreams of mega-cladal domination (but happily for any ideal of a richly varied biota), the ma­jor factors that boost speciation rates also raise the probability of extinction; [Page 740] while features that
enhance longevity also suppress the rate of speciation. For example, small populations in stressful environments are especially prone to both speciating and dying; while large, global populations of marked stability and great mobility (like Homo sapiens and Rattus rattus) are remarkably re­sistant to extinction (unless, like one of the above, they evolve an odd capac­ity for potential self-destruction), but ill-equipped to form the isolated popu­lations that can generate new species.

  For a third limiting constraint of brakes on the amount of available varia­tion for selectional processes (line IVB3), infrequency of new mutation may play an important role at the organismal level (not so often in sexual forms, where recombination greatly boosts the amount of variability among individ­uals, but usually a defining limit in asexuals, and perhaps the major reason for the rarity and marginal status of asexuals among complex Metazoa, but not in unicells with short generations). At the species level, variation per indi­vidual may be more than adequate (given the forced correlation of birth with change), but many clades contain too few individuals, giving birth too rarely, for very efficient selection (Fisher's argument — see page 645).