The greater salience of parallelism for non-Darwinian formalists, and for anti-darwinian theorists of various stripes, in late 19th and early 20th century evolutionary debates. We understand why parallelism faded from general consideration when the strict adaptationism of later and hardened versions of the Modern Synthesis pushed the general subject of internal constraint to a periphery of intellectual concern and presumed relevance. Similarly, we should easily comprehend why the same phenomenon — and the importance of distinguishing its component of constraint from the purely adaptational basis of convergence — would have generated more interest and greater clarity of definition during the period of its initial formulation (1890's to 1920's), when non-Darwinian formalist, and more overtly anti-Darwinian orthogenetic and saltational, theories enjoyed considerable vogue as adjuncts or alternatives to natural selection.
Two linguistically and geographically defined traditions of argument reinforce my contention that parallelism has always been understood and debated as a theory of constraint based on homologous generators for the origin of homoplastic similarities. First, the American paleontologists who initially codified the concept of parallelism did so in the context of pluralistic support for non-Darwinian internal mechanisms of evolutionary change (working in conjunction, or potentially in opposition, to natural selection, which they also accepted as a valid mechanism). We find parallelism sufficiently interesting today as an indicator of preferred internal channels that selection can exploit in coordinated evolutionary change. Imagine the even greater theoretical interest of parallelism for evolutionists who hoped to discover, in its workings, new principles and mechanisms of change that might fundamentally enrich or alter the basis of evolutionary theory.
In the article that first defined parallelism, for example, Scott (1891, pp. 370-371) argued that the orthogenetic linearity of parallel series implied a primary nonselectionist cause for phylogenetic transformation, since lineages [Page 1085] under the control of natural selection should exhibit more temporal fluctuation: “So far as the series of fossil mammals which we have been considering are concerned, the developmental history appears to be very direct, and subject to comparatively little fluctuation, advancing steadily in a definite direction, though with slight deviations.”
In his 1902 article, Osborn invoked parallelism more explicitly as a central argument for internal control of phylogenetic directionality, and against natural selection as a primary cause of change. In fact, following a standard tradition of continental non-Darwinian argument, Osborn demoted natural selection to a mere “exciting cause” (“exciting,” that is, in the literal sense of “initiating,” not in the modern meaning of “thrilling”) that can arouse the inherent channels of necessary change, and provoke homoplastic evolution along parallel paths. In his typically regal way, Osborn begins his paper by quoting his own prophetic words of 1897: “My study of teeth in a great many phyla of Mammalia in past times has convinced me that there are fundamental predispositions to vary in certain directions; that the evolution of teeth is marked out beforehand by hereditary influences which extend back hundreds of thousands of years. These predispositions are aroused under certain exciting causes [note his verbal demotion of natural selection] and the progress of tooth development takes a certain form converting into actuality what has hitherto been potentiality.”
Osborn then ends his paper (1902, p. 270) by explicitly citing the “latent or potential” homology of parallelism as an alternative to natural selection among causes of evolutionary change:
These homoplastic cusps [of teeth in independent lineages of mammalian evolution] do not arise from selection out of fortuitous variations, because they develop directly and are not picked from a number of alternates . . . We are forced to the conclusion that in the original tritubercular constitution of the teeth there is some principle which unifies the subsequent variation and evolution up to a certain point. Herein lies the appropriateness of Lankester's phrases, “a likeness of material to begin with.” Philosophically, predeterminate variation and evolution brings us upon dangerous ground. If all that is involved in the Tertiary molar tooth is included in a latent or potential form in the Cretaceous molar tooth we are nearing the emboitement hypothesis of Bonnet or the archetype of Oken and Owen.
Second, continental European theorists in the formalist tradition (see Chapters 4 and 5) had always emphasized constraint channeled by laws of form as a primary alternative to functionalist theories like natural selection. These scientists should therefore have taken a particular interest in parallelism, especially in its distinction from convergence for the origin of homoplastic similarity — for convergence exalts natural selection, while parallelism stresses internal channeling and supports the standard continental view of selection as a mere potentiator, or at most a minor diverter, of predictable and [Page 1086] law like changes that must follow internally specified rules of morphogenetic transformation.
Haas and Simpson (1946) cite all the major evolutionary theorists among continental paleontologists of the early to mid 20th century — particularly Abel, Dacque and Schindewolf — in support of these weights and definitions. In 1921, for example, Dacque compared parallelism with Eimer's anti-Darwinian concept of orthogenesis (see pp. 355–365 for full discussion on Eimer's views), while stressing the distinction of parallelism and convergence by the predominant causality of constraint vs. adaptation (Haas and Simpson, 1946, p. 335).
G. G. Simpson and the causal vs. geometric definition of parallelism. With parallelism thus falsely depicted as somehow contrary to selection, one can hardly blame the resurgent Darwinians of the Modern Synthesis for their diminished attention to a phenomenon that had been unfairly cited against the cause of change that they now wished to reassert as primary, if not virtually exclusive. (This history provides another concrete illustration of a general argument about older vs. modern versions of constraint that I advance throughout this book. The older versions interpreted constraint as contrary to selection, thus earning the indifference or enmity of Darwinian theorists when they regained ascendancy during the 1930's and afterwards. This unfortunate historical situation clouded the utility of constraint within Darwinian theory as an adjunct, a potentiator, or (at most distinction) an orthogonal source of evolutionary change. Modern versions of constraint can overcome this unfortunate division and reunite these two vital sources, formalist and functionalist, into an expanded and more general theory of Darwinian evolution.)
But the most perceptive of Darwinian theorists would not let such a contingent historical happenstance extinguish an important concept and distinction within the scope of evolutionary causality. In particular, G. G. Simpson — indisputably the most brilliant and biologically sophisticated of 20th century evolutionary paleontologists — continually emphasized the significance of a causal concept of parallelism based upon constraint, and the importance of distinguishing this mode of homoplasy from the opposite style of convergence based entirely upon shared adaptive contexts rather than shared homologous generators.
In his epochal 1945 treatise on principles of taxonomy and classification of mammals, Simpson drew a sharply dichotomous distinction between homology and convergence (1945, p. 9): “Animals may resemble one another because they have inherited like characters, homology, or because they have independently acquired like characters, convergence.” Simpson then spoke of parallelism as “a third sort of process [that] also produces similarities” (p. 9) — for he recognized the “hybrid” nature of a concept that required independent episodes of similar selection, but nonetheless constructed homo-plastic likenesses from homologous generators in two separate lines. With his usual insight, Simpson made the proper theoretical separation, but then ran [Page 1087] right into the old wall of stymied practice — for the biology of his day knew no methods for identifying the homologous generators that could mark a homoplastic similarity as parallel rather than convergent. Unable to cash out his t
heoretical clarity in actual practice, Simpson threw in the towel and admitted operational defeat (1945, p. 9):
It is a complication that a third sort of process also produces similarities: parallelism. The term is descriptive rather than explanatory and refers to the fact that distinct groups of common origin frequently evolve in much the same direction after the discontinuity between them has arisen, so that at a later stage the phyla may have characters in common that were not visible in the common ancestry but that tend, nevertheless, to be more or less in proportion to the nearness of that ancestry. This proportional tendency distinguishes parallelism from convergence, but the distinction is far from absolute. The two phenomena intergrade continuously and are often indistinguishable in practice.
Simpson (1945, p. 10) also stressed the intermediate nature of parallelism in phylogenetic inference, recognizing that even homoplastic characters usually record reasonably close genealogical affinity (in their common origin from homologous generators) in cases of parallelism, but must be regarded as confounders of affinity in cases of convergence: “Homology is always valid evidence of affinity. Parallelism is less direct and reliable, but it is also valid evidence within somewhat broader limits. It may lead to overestimates of degree of affinity, but it is not likely to induce belief in wholly false affinity. Convergence, however, may be wholly misleading, and a principal problem of morphological classification on a phylogenetic basis is the selection of characters that are homologous or parallel and not convergent.”
In his 1961 book on Principles of Animal Taxonomy, Simpson continued to express his frustration at the conceptual need, but operational impossibility, of distinguishing parallelism from convergence. “The distinction of parallelism from convergence is vital,” he writes (1961b, p. 106). Fifteen years after his joint paper with Haas, and their disagreement over geometrical versus causal definitions of the terms, Simpson stated in frustration (1961b, p. 103): “Parallelism is the independent occurrence of similar changes in groups from a common ancestry and because they had a common ancestry. Some students (for example, Haas in Haas and Simpson, 1946) have preferred a more purely descriptive definition, especially by the geometrical model of parallel lines, symbolizing two lineages both changing but not becoming significantly either more or less similar . . . Most taxonomists do, however, consider that the term parallelism should be used only when community of ancestry is pertinent to the phenomenon.”
Simpson concludes his discussion (1961b, p. 106) with the clearest statement I have ever read for citing homology of underlying generators as the basis of parallelism, and on the joint operation of both overt selection and underlying homology in the evolution of homoplastic structures by parallelism: [Page 1088]
Parallelism has several theoretical bases that help one to understand and also to recognize it. The structure of an ancestral group inevitably restricts the lines of possible evolutionary change. That simple fact greatly increases the probability that among the number of descendant lineages several or all will follow one line. That probability will be further reinforced by natural selection in a geographically expanding and actively speciating group if the ecologies of diverse lineages remain similar in respect to the adaptations involved in the parallelism. The degree of dependence on similar ecology resembles that of convergence, but the retention of homologous characters from the relatively near common ancestry usually distinguishes parallelism. The parallel lineages (unlike those only convergent) furthermore start out with closely similar coadapted genetic systems, and similar changes are more likely to keep the system adequately coadapted.
Parallelism as a “gray zone” between homology and convergence. Despite Simpson's careful separations, and his stress on their theoretical importance, many biologists ignored the important theoretical differences between these two subcategories of homoplasy. If they recognized parallelism and convergence as distinctive terms at all, they often could not state any rationale for the terminology beyond the triviality of an abstract and formal geometric difference between parallel and converging lines.
But thoughtful evolutionists continued to struggle with the “hybrid” character of parallelism. Michener (1949), for example, in the finest technical application of the concept, honored the causal (rather than geometric) distinction: “The potentiality for similar changes, resulting in parallel characters, no doubt results from the fact that related animals have homologous chromosomes and genes” (1949, p. 140).
The cladistic revolution in taxonomic practice also forced renewed attention to the distinction, and to the “intermediate” status of parallelism in producing homoplastic structures based on homologous generators — leading, for example, to Saether's (1983) concept of “underlying synapomorphies,” defined as “the capacity to develop synapomorphy” or “close parallelism as a result of inherited factors within a monophyletic group” (Saether, 1983, p. 343).
The acknowledgment of homologous generators actually led some taxonomists, including such leaders as Mayr (1974), to include parallelism within a broader definition of homology, while most researchers continued to rank parallelism as an uncomfortable subcategory of homoplasy (Patterson, 1988), or as a “hybrid” notion based on homoplastic origin from homologous generators (as in Saether, 1983). Perhaps Patterson (1988, p. 619) put the matter best by writing: “In morphology, the 'gray zone' between homology and nonhomology concerns congruence — or inferred common ancestry — and whether parallelism (which does invoke common ancestry) should be included or excluded from homology.”
The operational rescue of parallelism by evo-devo. The [Page 1089] culmination of more than a century of conceptual and terminological struggle may now be epitomized in a triumphalist tone usually shunned in science, but clearly justified in this rare case: the development of genetic and developmental techniques that established the field of evo-devo have finally allowed biologists to identify the homologous generators that always specified the concept of parallelism in theoretical terms. Parallelism has now, and finally after a century of terminological recognition, become an operational subject for evolutionary research. Moreover, the first flood of results has revealed a depth and extent of parallelism among distant phyla that strict Darwinians had explicitly deemed inconceivable, and that even the most enthusiastic well-wishers and partisans of constraint did not dare to imagine in their fondest dreams (unless their capacity for imagination greatly exceeded the scope of this particular rooter — see Gould, 1977b).
A SYMPHONY IN FOUR MOVEMENTS ON THE ROLE OF HISTORICAL CONSTRAINT IN EVOLUTION: TOWARDS THE HARMONIOUS REBALANCING OF FORM AND FUNCTION IN EVOLUTIONARY THEORY
As a literary device, metaphor spans a particularly broad band of relative merit — from treacherous comparisons virtually guaranteed to confuse or misstate a causal analysis, to illuminating analogies intended to explicate the unfamiliar or to impose a useful and sensible order upon an otherwise inchoate mass of ideas and information. By invoking the following risky comparison of the major ideas and putative theoretical reforms of evo-devo to the four movements of a classical symphony, I mean to highlight some aspects of the comparison, while abjuring others. I do not, in the most obviously non-adaptive feature of the metaphor, claim any chronological basis, or any numerical ordering of importance, for the four sequential themes.
Rather, I rest my case for the utility of this organizing device upon an admittedly peculiar isomorphism between these disparate realms. I believe that the burgeoning literature on the genetics of development can be explicated most usefully (in terms of a probable enduring influence upon evolutionary theory) as a set of four subjects — and that these subjects, presented in their most sensible and logical order, invite a close comparison with the “standard” sequence and thematic progression of the four movements in a classical symphony: statement, development, scherzo, and generalization — or, for the literature of evo-devo, deep homology, pervasive parallelism (for features once deemed
convergent), saltational musings, and reasons for the markedly inhomogeneous occupation of morphospace among animal phyla.
Movement one, Statement: deep homology across phyla: Mayr's
functional certainty and Geoffroy's structural vindication.
DEEP HOMOLOGY, ARCHETYPAL THEORIES, AND HISTORICAL CONSTRAINT. In the most important general book on evo-devo written in the last decade of a millennium, Raff (1996, p. 428) astutely epitomized the [Page 1090] importance of constraint for an enriched and revised version of Darwinian theory.
A long-standing and important theoretical conception of the relationship between development and evolution is that of developmental constraints. The idea that developmental rules can direct or constrain the course of evolution has two origins. A number of evolutionists, particularly in the generation following Darwin, took antiselectionist positions, and posited that internal forces direct evolution and produce long-term trends independent of the external environment. That is not a tenable position, but neither is extreme selectionism. Internal genetic and developmental constraints of various kinds must exist, but... they are diverse and poorly understood. Yet if internal factors constrain evolution, they are hardly a minor issue. The acceptance of internal constraints does not mean that Darwinian selection is unimportant, but it does mean that the variation presented to selection is not random.
Two aspects of this statement capture both the optimism and the theoretical importance of this emerging field. By defining the subject of constraint as collateral and helpful to selectionism (rather than oppositional, if not substitutional, as in most 19th century versions of internalism, as Raff mentions above and as I document extensively in Chapters 4 and 5), Raff depicts the growth of evo-devo as interactive building in a different architectural style, rather than as demolition. Secondly, by summarizing the main import of constraint for Darwinian theory in the claim “that the variation presented to selection is not random,” Raff correctly identifies the locus of greatest importance for evolutionary theory — for the logic of pure selectionism does presuppose nondirectional variability (see pp. 144–146), and the existence of strongly preferred channels, based on the architecture and history of development, does require an important restructuring (not just a minor nuancing) of Darwinian logic.
The Structure of Evolutionary Theory Page 173
