The Structure of Evolutionary Theory

by Stephen Jay Gould

  HIERARCHY AND THE SIXFOLD WAY

  A literary prologue for the two major properties of hierarchies

  Our vernacular language recognizes a triad of terms for the structural de­scription of any phenomenon that we wish to designate as a unitary item or [Page 674] “thing.” The thing itself becomes our focus, and we call it an object, an entity, an individual, an organism, or any one of a hundred similar terms, depending on the substance and circumstance. The subunits that build the individual are then called “parts” (or units, or components, or organs, etc., depending upon the nature of the focal item); while any recognized grouping of similar indi­viduals becomes a “collectivity” (or aggregation, society, organization, etc.). In other words, and in epitome, individuals are made of parts and aggregate into collectivities.

  The hierarchical theory of selection recognizes many kinds of evolutionary individuals, banded together in a rising series of increasingly greater inclu­sion, one within the next — genes in cells, cells in organisms, organisms in demes, demes in species, species in clades. The focal unit of each level is an in­dividual, and we may choose to direct our evolutionary attention to any of the levels. Once we designate a focal level as primary for a particular study, then the unit of that level — the gene, or the organism, or the species, etc. — be­comes our relevant or focal individual, and its constituent units become parts, while the next higher unit becomes its collectivity. Thus, if I place my focus at the conventional organismic level, genes and cells become parts, while demes and species become collectivities. But if my study enjoins a focus on species as individuals, then organisms become parts, and clades become collectivities. In other words, the triad of part — individual — collectivity will shift, as a three­fold entirety, up and down the hierarchy, depending upon the chosen subjects and objects of any particular study.

  This dry linguistic point becomes important for a fundamental reason of psychological habit. We humans are hidebound creatures of convention, particularly tied to the spatial and temporal scales most palpably familiar in our personal lives. Among nature's vastly different realms of time, from the femtoseconds of some atomic phenomena to the aeons of stellar and geologi­cal time, we really grasp, in a visceral sense, only a small span from the sec­onds of our incidents to the decades of our lives. We can formulate other scales in mathematical terms; we can document their existence and the pro­cesses that unfold in their domains. But we experience enormous difficulty in trying to bring these alien scales into the guts of or our understanding — largely for the parochial reason of personal inexperience.

  We make frequent and legendary errors because we tend to extrapolate the styles and modes of our own scale into the different realms of the incomprehensibly immense or tiny in size, vast or fleeting in time. Geologists, for ex­ample, well appreciate the enormous difficulties that most people encounter (including our professional selves, despite so many years of training and experience) in trying to visualize or understand the meaning of any ordinary statement in “deep” or “earth time” — that a landscape took millions of years to develop, or that a lineage exhibits a trend to increasing size throughout the Cretaceous period. All of us — professionals and laypeople alike — continue to make the damnedest mistakes. I have, for example, struggled for thirty years against the conventional misreading of punctuated equilibrium as a saltational theory in the generational terms usually applied to such a concept in [Page 675] evolutionary studies. The theory's punctuations are only saltational on geo­logical scales — in the sense that most species arise during an unmeasurable geological moment (meaning, in operational terms, that all the evidence ap­pears on a single bedding plane). But geological moments usually include thousands of human years — more than enough time for a continuous process that we would regard as glacially slow by the measure of our lives (see Goodfriend and Gould, 1996, for an example). Thus, punctuated equilibrium represents the proper geological scaling of speciation events that may span several thousand years, not a slavish promotion of “instantaneity,” as con­ventionally measured in a human time frame, to the origin of species.

  As we misunderstand scales of time, we fail just as badly with viscerally unfamiliar realms of size. Our bodies lie in the middle of a continuum ranging from the angstroms of atoms to the light years of galaxies. Individuality exists in all these domains, but when we try to understand the phenomenon of “thingness” at any distant scale, we easily fall under the thrall of the greatest of all parochialisms. We know one kind of individual so intimately and with such familiarity — our own bodies — that we tend to impose the characteristic properties of this level upon the very different styles of being that other scales generate. This inevitable human foible provokes endless trouble, if only be­cause organismal bodies represent a very peculiar kind of individual, serving as a very poor model for the comparable phenomenon at most other scales.

  The “feel” of individuality at other scales becomes so elusive that most of the best exploration has been accomplished by literary figures, not by scien­tists. The tradition extends at least as far back as Lemuel Gulliver, whose “alien” contacts did not depart greatly from our kind of body and our norm of size. This theme has best been promoted, in our generation, within the genre of science fiction. I particularly recommend two “cult” films, Fantastic Voyage and Inner Space, both about humans reduced to cellular size and in­jected into the body of another unaltered conspecific. This ordinary body be­comes the environment of the shrunken protagonists, a “collectivity” rather than a discrete entity — while the “parts” of this body become individuals to the shrunken guests. When Raquel Welch fights a bevy of antibodies to the death in Fantastic Voyage, we understand how location along the triadic continuum of part — individual — collectivity depends upon circumstance and concern. A tiny, if crucial, part to me at about two meters tall becomes an en­tire and ultimately dangerous individual to Ms. Welch at a tiny fraction of a millimeter.

  The parochiality of time has served us badly enough, but the parochiality of bodily size has, for two reasons, placed even more imposing barriers in our path to an improved and generalized evolutionary theory — a formulation well within our grasp if we can learn how to expand the Darwinian perspec­tive to all levels of nature's hierarchy. First, we know almost viscerally what our bodies do best as Darwinian agents — and we then grant universal impor­tance to these properties both by denying interest to the different “best” properties of individuals at other levels, and by assuming that our “bests” must, by extension, power Darwinian systems wherever they work. Our bodies [Page 676] are best at developing adaptations in the complex and coordinated form that we call “organic.” Many evolutionists therefore argue, in the worst pa­rochialism of all, that only adaptations matter as an explanatory goal of Dar­winism, and that such adaptations must therefore drive evolution at all levels. I don't even think that such a perspective works well for organisms — surely the locus of most promising application (Gould and Lewontin, 1979) — but this attitude will surely stymie any understanding of individuality at other levels, where complex adaptations do not figure so prominently. Evolutionists will not be able to appreciate the different individuality of species, where exaptive effects hold at least equal sway with adaptations, if they continue to regard spandrels, sequelae, and side consequences as phenomena generated by “the boring by-product theory” (Dawkins, 1982, p. 215).

  Second, we just don't comprehend the scale-bound realities in other do­mains of size, and we err by imposing our own perceptions when we try to think about the world of a gene, or of a species. In one of the most famous statements of 20th century biology, D'Arcy Thompson (1942, p. 77) ended his chapter “On Magnitude” (in his classic work, On Growth and Form — see the first section of Chapter 11 for a general analysis of his work) by noting how badly we misread the world of smaller organisms because our large size places us in gravity's domain (a result of low surface/volume ratios in larger creatures, but not a significant fea
ture in other realms of size). If we encoun­ter so much trouble for extremes within our own level of organismic individ­uality, how will we grasp the even more distant worlds of other kinds of evo­lutionary individuals? D'Arcy Thompson wrote (1942, p. 771):

  Life has a range of magnitude narrow indeed compared to that with which physical science deals; but it is wide enough to include three such discrepant conditions as those in which a man, an insect, and a bacillus have their being and play their several roles. Man is ruled by gravitation, and rests on mother earth. A water beetle finds the surface of a pool a matter of life and death, a perilous entanglement or an indispensable support. In a third world, where the bacillus lives, gravitation is forgot­ten, and the viscosity of the liquid, the resistance defined by Stokes's law, the molecular shocks of the Brownian movement, doubtless also the electric charges of the ionized medium, make up the physical environ­ment and have their potent and immediate influence on the organism. The predominant factors are no longer those of our scale; we have come to the edge of a world of which we have no experience, and where all our preconceptions must be recast.

  As one example, consider the difficulty we experience, despite our preferences for reductionism in science, when we try to visualize the world of our genes, where nucleotides function as active and substitutable evolutionary parts — and where chromosomes build a first encasement, followed by nuclei and cells, with our body now serving as an entire universe, whose death will also destroy any gene still resident within. Think of the initial resistance that most of us felt towards Kimura's neutralist theory — largely because we falsely [Page 677] “downloaded” our adaptationist views about organisms into this different domain, where high frequencies of neutral substitution become so reasonable once we grasp the weirdly (to us) divergent nature of life at such infinitude. And if we fare so badly for the small and immediate, supposedly so valued by our reductionist preferences, how can we comprehend an opposite extension into the longer life, the larger size, and the markedly different character of species-individuals — a world that we have usually viewed exclusively as a col­lectivity, an aggregation of our bodies, and not as a different kind of individ­ual in any sense at all?

  I like to play a game of “science fiction” by imagining myself as an individ­ual of another scale (not just as a human being shrunken or enlarged for a visit to such a terra incognita). But I do not know how far I can succeed. As organisms, we have eyes to see the world of selection and adaptation as ex­pressed in the good design of wings, legs, and brains. But randomness may predominate in the world of genes — and we might interpret the universe very differently if our primary vantage point resided at this lower level. We might then note a world of largely independent items, drifting in and out by the luck of the draw — but with little islands dotted about here and there, where selec­tion slows down the ordinary tempo and embryology ties things together. How, then, shall we comprehend the still different order of a world much larger than ourselves? If we missed the strange world of genic neutrality be­cause we are too big, then what passes above our gaze because we are too small? Perhaps we become stymied, like genes trying to grasp the much larger world of change in bodies, when we, as bodies, try to contemplate the do­main of evolution among species in the vastness of geological time? What are we missing in trying to read this world by the inappropriate scale of our small bodies and minuscule lifetimes?

  Once we have become mentally prepared to seek and appreciate (and not to ignore or devalue) the structural and causal differences among nature's richly various scales, we can formulate more fruitfully the two cardinal prop­erties of hierarchies that make the theory of hierarchical selection both so in­teresting and so different from the conventional single-level Darwinism of organismal selection. The key to both properties lies in “interdependence with difference” — for the hierarchical levels of causality, while bonded in in­teraction, are also (for some attributes) fairly independent in modality. More­over, these levels invariably diverge, one from the other, despite unifying prin­ciples, like selection, applicable to all levels. Allometry, not pure fractality, rules among the scales of nature.

  1. Selection at one level may enhance, counteract, or just be orthogonal to selection at any adjacent level. All modes of interaction prevail among levels and make prominent imprints in nature.

  I emphasize this crucial point because many students of the subject have focussed so strongly on negative interaction between levels — for a sensible and practical reason — that they verge on the serious error of equating an opera­tional advantage with a theoretical restriction, and almost seem to deny the [Page 678] other modes of positive (synergistic) and orthogonal (independent) interac­tion. Negative interaction wins primary heuristic attention because this mode provides our most cogent evidence, not merely for simultaneous action of two levels, but especially for the operation of a controversial or unsuspected level. If two levels work in synergism, then we easily miss the one we do not expect to see, and attribute the full effect to an unsuspected strength for the level we know. But if the controversial level yields an unexpected effect con­trary to the known direction of selection at a familiar level, then we may be able to specify and measure the disputed phenomenon.

  In the example cited previously, individual selection favors a balanced sex ratio, while interdemic selection leads to female bias in many circumstances. Our best evidence for the reality of interdemic selection emerges from the dis­covery of such biases — not so strong as purely interdemic selection would produce (for organismic selection operates simultaneously in the other direc­tion), but firm enough to demonstrate the existence of a controversial phe­nomenon. But if interdemic selection also worked towards a 1:1 ratio, we could attribute such an empirical finding exclusively to the conventional op­eration of organismic selection.

  Negative interaction, however, does yield one distinguishing consequence to highlight this mode as especially important in the revisions to evolutionary theory that the hierarchical model will engender. In conventional one-level Darwinism, stabilities generally receive interpretation as adaptive peaks or optima, thus enhancing the functionalist bias inherent in the theory. The ma­jor structuralist intrusion into this theme ordinarily occurs when we have been willing to allow that natural selection can't surmount a constraint — ele­phants too heavy to fly even if genetic variability for wings existed; insects confined to small sizes by the inherited Bauplan of an exoskeleton that must be molted, and a respiratory system of skeletal invaginations that would be­come too extensive at the surface/volume ratio of large organisms. But the constraints in these cases act as passive walls, not active agents.

  The hierarchical theory of selection suggests a theoretically quite different and dynamic reason for many of nature's stabilities: an achieved balance, at an intermediary point optimal for neither, between two levels of selec­tion working in opposite directions. Several important phenomena may be so explained: weak female bias as the negative interaction of organismal and interdemic selection (see above); restriction of multiple copy number in “selfish DNA” as a balance between positive selection at the gene level, sup­pressed by negative selection (based, perhaps, on energetic costs of produc­ing so many copies irrelevant to the phenotype) at the organismic level. I also suspect that stable and distinctive features of species and clades must repre­sent balances between positive organismic selection that would drive a fea­ture to further elaboration, and negative species selection to limit the geo­logical longevity of such “overspecialized” forms. In any case, a world of conceptual difference exists between stabilities read as optima of a single pro­cess, and stabilities interpreted as compromises between active and opposed forces. [Page 679]

  As an example of overemphasis upon negative interaction, Wilson and Sober (1994, p. 592) ask: “Why aren't examples of within-individual [organ­ism] selection more common?” They mention the most familiar case of meiotic drive, and then discuss the conventional argument for rarity of
such phenomena: the integrity of complex organisms implies strong balance and homeostasis among parts; therefore, any part that begins to proliferate inde­pendently will threaten this stability, and must therefore be disfavored by organismic selection, a force generally strong enough to eliminate such a threat from below.

  If selection within bodies generally opposes the organismic level, as this discussion implies, then we properly expect a low frequency for the phenome­non, since evolution has endowed the organismic level with a plethora of de­vices for resisting such dysfunctional invasion from within. Although I accept this argument for a low frequency of selection contrary to the interests of en­closing organisms, selection within bodies may not be so rare when we in­clude the other modalities of synergistic and orthogonal directions. The most interesting hypothesis for extensive selection at the gene level, the notion originally dubbed “selfish DNA” (Orgel and Crick, 1980; Doolittle and Sapienza, 1980), attributes the observed copy number of much middle-repetitive DNA to orthogonal gene-level selection initially “unnoticed” by the organ­ism, though eventually suppressed by negative selection from above when copies reach sufficient numbers to exact an energetic drain upon construction of the phenotype (see fuller discussion on pp. 693–695). In fact, I suspect that organismic complexity could never have evolved without extensive gene-level selection in this orthogonal (or synergistic) mode. For if we accept the com­mon argument that freedom to evolve new phenotypic complexity requires genetic duplication to “liberate” copies for modification in novel directions, then how could such redundancy arise if organismic selection worked with such watchdog efficiency that even a single “extra” copy, initially unneeded by the organismic phenotype, induced strong negative selection from above, and immediately got flushed out — thus, in an odd sense, making the organism a delayed Kamikaze, killing its “invader” now and, by summation of such consequences, itself later?