The Structure of Evolutionary Theory

by Stephen Jay Gould

  Cases of multiple utility for a single feature offer special promise for resolu­tion by these criteria of temporal or cladogenetic sequencing. Arnold cites the case of “the aberrant arboreal tropical African lacertid lizard, Holaspis guentheri,” whose extremely flattened head “allows it to hunt and hide in [Page 1236] narrow crevices beneath bark, and also constitutes an aerofoil which enables it to glide from tree to tree. Phylogenetic analysis shows that the flattening first developed in the context of crevice use and was only later coopted to gliding” (p. 139).

  Moreover, such primary adaptations as head flattening for penetration of crevices usually work in synergy with other coopted features that operate as exaptations in the complex and multifaceted “fit” of the organism to its new environment. For example, when lizard heads become flattened, “the eyes do not usually become correspondingly smaller and, in normal activity, bulge upwards above the skull surface” (p. 139). Arnold then continues to describe the remarkable exaptation of a mouthful of eye: “However, when a lizard flees into a narrow crevice the eyes must be accommodated within the depth of the flattened head. They are most usually pushed downwards by the ceiling of the crevice as the lizard moves deeper into it, so their upper margins are flush with the skull roof and their lower sections bulge through the palate into the buccal cavity (p. 139).” In scincids and lacertids, the eye bulges verti­cally downward into the suborbital foramen. This aptation depends upon the preexistence of this opening (obviously evolved for other reasons), as indicated by its general distribution on the cladogram of lizards. Therefore, “as the occurrence of the foramen on the phylogeny of the forms concerned pre­cedes occupation of crevices, its use for accommodating the eye within the re­duced depth of the skull is an exaptation” (p. 139).

  A common, but unfounded, objection to exaptation enters the logical structure of argument at this point. Several colleagues (Coddington, 1988, for example) have claimed that since almost any exapted structure will un­dergo secondary modification for its new role, and since these subsequent changes must count as adaptations, the concept of exaptation becomes either useless or confusing because any primarily exapted structure must then ac­crete secondary adaptations to be fully “fit” for its new role. I raise this issue here because, as Arnold points out, the suborbital “foramen is initially small and triangular, allowing only limited projection of the eye into the buccal cav­ity” (p. 139). This hole then undergoes a secondary adaptive enlargement to accommodate the eye more completely.

  I am confident that this common objection cannot be sustained, because hierarchical sequences of processes, each with a different name and status, practically define the nature of complex historical change, and pose no con­ceptual problems (but rather help us to understand and sort out these se­quences), provided that we can specify the order of temporal precedence and hierarchical nesting. Exactly the same issue arises for homology and conver­gence, and for plesiomorphy and apomorphy. The front appendages of bats and birds are homologous as forearms and convergent as wings; live birth is plesiomorphic for the clade of marsupial and placental mammals, and apomorphic for the same clade within the Tetrapoda. Similarly, the suborbital fo­ramen of lizards is exaptive as a preexisting receptacle for the pushed-down eyes of lizards with flat heads, whereas the subsequent enlargement of the hole may be adaptive for better accommodation of the eyes. The two aspects [Page 1237] can easily be separated, and their named distinction helps us to understand the probable sequence of evolutionary events. As Arnold states (p. 139): “there is subsequent, presumably adaptive, modification improving the initial exaptation, the foramen becoming larger and more rounded.”

  Interestingly, and as further confirmation of the primarily exaptive nature of these vacuities, crevice-dwelling cordylid lizards, representing another sep­arate evolutionary entrance into this habitat, shift their eyes medially into the interpterygoid vacuity, rather than downward into the suborbital foramen. Arnold argues (see Figure 11-7 on these dual routes to exaptation in crevice-dwelling lizards) that cordylids may have utilized this alternative strategy be­cause this group happens to possess a large interpterygoid space (as lacertids and scincids do not) that can accommodate the eye without the secondary modification required to “house” eyes in the suborbital space. Of this side­wards exaptation of cordylids, Arnold writes (p. 140): “This again turns out to be an exaptation, for examination of the phylogeny of cordylids shows that expansion of the interpterygoid vacuity evolves before crevice use, al­though after the origin of the suborbital foramen. The vacuity may have been utilized instead of the foramen because, being large, it provided immedi­ate housing for a large portion of the eye, whereas the foramen would only have been able to provide this after some modification, as in lacertids and scincids.”

  In his richest and most extensive analysis, Arnold then discusses six sepa­rate evolutionary innovations of “sand-diving” among lizards (quick entry into aeolian dunes to escape predators). He finds all six to be equivalent, both in efficacy and as solutions to the same functional requirement. “In no in­stance,” he writes (p. 156), “is there evidence that the different methods employed reflect different mechanical problems.” In his combination of func­tional and cladistic analysis, he interprets the mechanisms used for sand-

  11-7. Two pathways to the exap­tive use of preexisting vacuities in the skull to house the eyes tem­porarily, so that lizards can squeeze into crevices in rocks. From Arnold, 1994.

  [Page 1238]

  diving as directly adaptive (that is, first evolved in conjunction with, and pre­sumably for, the behavior) in two cases, but as exaptive in the other four se­quences. Interestingly, the exaptive lineages coopted their sand-diving move­ments from two different functional sources in ancestral lineages: from “part of the drilling mechanism used in firm substrates” (p. 161) by ancestral lin­eages that hide themselves in harder grounds, and “from the burial pattern adopted before periods of inactivity” (p. 161) in other ancestral lineages (that is, from movements carried out much more slowly, in a clandestine fashion, and for purposes of dormancy rather than escape from predators).

  Two additional criteria might be cited as evidence for both the use and usefulness of exaptation as a concept in evolutionary biology, and in other forms of historical study as well.

  1. Utility in fields distant from evolutionary biology. Markey (1997) in­voked our concept of exaptation to explain the peculiar history of the letter perth in the runic alphabet of futhark. (The word “futhark” is an acronym for the first six letters of the runic alphabet, just as “alphabet” itself combines the first two letters of the Greek sequence, alpha and beta.) Perth must have had ordinary phonemic value in a still earlier system, but the letter is never glossed in futhark texts and has left no descendants in any Germanic lan­guage. (Actually, the letter occurs only one time in all runic literature — in the English Rune Poem.)

  Markey (1997) argues that, having lost its original phonemic use, “the p-rune appears to have been a redundant luxury” (p. 10). Some versions of later futhark alphabets simply eliminated the symbol, but others retained the p-rune, apparently for an interesting structural reason with excellent literary analogy to quirky functional shift in biological features. The p-rune hap­pened to stand right at the middle of the futhark alphabet, where its phone­mic suppression encouraged a different use as a place marker or mnemonic guide, at the halfway point of a long sequence more easily recalled in two di­vided halves. (Markey shows that several versions of futhark added letters, but always kept the p-rune right in the middle of the sequence.)

  Markey argues that the p-rune, phonemically extinct in Germanic tongues, then lost its exapted function as a place marker when these languages re­placed the runic alphabet with our current Latin system. At this point, the p-rune again resisted extinction by another exaptation, this time for spelling Latin loan words with an initial p sound before a vowel (as in papa for “pope,” or pater for “priest”), a phonemic combination not found
in Ger­manic words. (Runic-p served the same function for some English loan words of non Indo-European origin, as in “pebble.”) In any case, Markey (p. 11) found our biological concept of exaptation useful in describing this complex double quirky functional shift from ordinary phonemic value in a hypotheti­cal ancestor, to place marking in futhark, to renewed phonemic value for loan words when the Latin alphabet replaced futhark: “Exaptation is manifested by functional bifurcation. The primary function of feathers was warmth, the secondary function flight. The primary function of runic-p appears to have [Page 1239] been that of a boundary marker, while its secondary function was loan­word spelling. Runic-p has every appearance of having been coopted. Non-essential in the runic system, it must have been essential in some system, pre­sumably the parent of Older Futhark.”

  In a fascinating exegesis of Emil Durkheim's seminal, late 19th century sociological studies of the division of labor, Catton (1998) invoked our concept of exaptation to explain a central error that Durkheim might not have com­mitted if he had recognized the principle of functional shift, either from Dar­win (whom he studied intensely), or from the more nearly contemporary Nietzsche. Durkheim recognized (correctly) that division of labor, and the at­tendant specialization of tasks in society, can greatly reduce competition and lead to “organic solidarity” (Catton, 1998, p. 89). But he then erred in as­suming that this current utility also permits the inference that division of la­bor arose, in explicit analogy with speciation, as a direct adaptation for its current function of reducing competition and stabilizing both social and eco­nomic systems. “To Durkheim,” Catton explains (p. 117), “it seemed abate­ment of competition by means of differentiation was the necessary removal of an otherwise insurmountable barrier to mutualistic interdependence. That was why division of labor was supposed to result in organic solidarity.”

  But Catton then exposes the dilemma and logical error entailed by Durk­heim's commitment to an evolutionary analogy with speciation. For biolo­gists argue, and have demonstrated in many cases, that mutualistic interac­tions often evolve from initial antagonisms and exploitations: “Evolutionary ecologists now know that mutualism can evolve from antagonism ... by some modification of structure or behavior that changes the outcome of an interaction from which the parties cannot withdraw.”

  The secondarily evolved cooperation may remain “good” for both parties, while so altering the initial state of the system that origins cannot be inferred from this current utility. Catton (1998, p. 118) found our discussion of exap­tation useful in explaining this important concept to his colleagues. (He also shows his appreciation of the corollary that secondary adaptation for a new role does not impeach the exaptive origin of the coopted utility): “An adapta­tion has a function. An exaptation has an effect. Once that effect becomes im­portant in the life of an organic type (in its new environment), natural selec­tion may 'improve' the exapted trait, eventually making it an adaptation, and converting the effect into a true function.”

  2. Passage of the term from explicitly cited novelty to general and unrefer­enced usage in evolutionary literature. The sequence may be bittersweet for originators, but only the most narcissistic or insecure scientist could fail to take pleasure when a concept of his invention, or an experiment of his doing, loses explicit connection to his authorship by “evolution” into an ordinary term of art within the profession. This form of acquired anonymity crowns the diffusion to general success of a suggestion or innovation with a “point source” of origin, now happily forgotten and relegated to the domain of anti­quarian or historical studies. [Page 1240]

  Exaptation has already passed through the three major stages of this se­quence. In a first stage, exaptation, as a novel term, became an explicit focus for studies to test or illustrate its utility — as in Arnold's title (to his 1994 pa­per, discussed above): “Investigating the origins of performance advantage: adaptation, exaptation and lineage effects”; or in Almada and Santos (1995): “Parental care in the rocky intertidal: a case study of adaptation and exap­tation in Mediterranean and Atlantic blennies.”

  In a second stage, the concept serves, inter alia, as part of an ordinary anal­ysis, and not as an explicit focus of study — but the term still requires a cita­tion to its source of origin, or at least must be defined and presented within quotation marks. For example, Chatterjee (1997), advocating the currently less popular “trees down” arboreal (rather than the “ground up” terrestrial) theory for the origin of birds, argued that many climbing adaptations of tree-dwelling ancestors “were exapted for gliding” in the transitional stages to­wards full flight. The language of Chatterjee's full sentence records an inter­esting, and undoubtedly unconscious, intermediary stage in the acceptance of functional shift as a principle in evolutionary analysis: “Surprisingly, many of these arboreal innovations were exapted for gliding” (p. 311). But such cooptations and functional shifts can only be deemed “surprising” when con­trasted with expectations of continuous improvement within a single “adap­tive zone” (to use Simpson's classical terminology of 1944). Once we recog­nize functional shift and cooptation as important components in almost any extensive evolutionary sequence, we will no longer label exaptations as sur­prising.

  In an example from the most fecund realm of exaptation in molecular evolution, Weiner and Maizels (1999) explained to their biochemical col­leagues who may not be au courant with the literature of evolutionary theory: “Those with an evolutionary bent sometimes use the word 'exaptation' to describe the appropriation of a molecule with one job for a completely dif­ferent purpose. Exaptation contrasts with 'adaptation,' a seemingly natu­ral extension of preexisting functions” (1999, p. 64). Their article, entitled “a deadly double life,” documents the fascinating “remarkable discovery” (1999, p. 63) that the carboxyl-terminal domain of human tyrosyl-transfer RNA synthetase (the enzyme that catalyzes the attachment of the amino acid tyrosine to the appropriate tRNA molecule prior to protein synthesis) shows clear homology (49 percent sequence similarity) with a cytokine performing the quite different — one might say conceptually opposite — function of at­tracting phagocytic cells to sites of apoptosis, suggesting in a broader sense that “secretion of tyrosyl-tRNA synthetase may help to shut down residual protein synthesis in the dying cell” (p. 63).

  In this case, the synthetase activity seems primary (for a set of reasons elaborated in Weiner and Maizels, 1999), and the “opposite” role in cell death secondary, following gene duplication, and leading to the molecule's “deadly double life.” Weiner and Maizels (1999) argue that the utility in apoptosis originates as an exaptation recruited from an “accidental” effect of the gene's primary activity: “The recruitment of tryrosyl-tRNA synthetase as an extracellular [Page 1241] death messenger” (p. 64) follows from its primary role and allows these molecules to “serve as harbingers of impending cell death when released from their normal cellular compartments”: “Release of proteins from their normal locations in the cell may have originally been a symptom of cell death, rather than a cause of it. Evolution may then have exploited the accidental [their italics] release of these proteins (and possibly others) to build, amplify, and eventually fine-tune the death circuitry.”

  Finally, in a third phase, the term enters the literature as a standard item of professional lingo, requiring no further citation of original sources (probably unknown to the authors in any case) than any other word of professional jar­gon. Thus, Jablonski and Chaplin (1999, p. 836) view manual dexterity and eventual tool use as human exaptations of a bipedal posture that originally arose as part of a common threat display in ancestral apes, and Roy (1996) analyzes exaptations for defense in the fossil history of stromboid gastropod shells.

  I take greatest pleasure, however, in the spread of exaptation as a term of art in the most pervasively expressed domain of molecular evolution. For example, in a review article on “interspersed repeats and other mementos of transposable elements in mammalian genomes,” Smit (1999) consi
stently uses exaptation to describe coopted utilities of multiply repeated and dis­persed transposable elements (the classic molecular items that inspired the concept of “selfish DNA,” see pp. 693–695 for further discussion).

  In a section of a paper on “domestication of individual transposable ele­ments” (p. 661, and I do appreciate his witty and apposite metaphors from the vernacular), Smit writes: “Throughout time, host genomes have rum­maged through the novel sequences accumulated by transposition and have recruited numerous elements . . . Far from merely expanding genomes with interspersed repeats, their legacy ranges from spliceosomal introns and antigen-specific immunity to many recent recruits in highly specialized func­tions” (1999, pp. 661-662). Although Smit notes an apparently reduced exaptive role for such transposons in humans vs. mice, “leading to speculations on host defense mechanisms” (p. 657) in our species, he also lists an impressive array of potential human examples, “some with household names” (p. 661). For example, Smit cites a fascinating exaptation of all higher pri­mates, probably essential to the existence of this book and any reader's kind and current attention thereto: “BC200, the only known fully recruited SINE in humans, is a brain specific RNA that is part of a ribonucleoprotein com­plex preferentially located in the dendrites of all higher primates. It is presum­ably derived about 50 million years ago from a monomeric Alu and has since been selectively conserved in all studied descendants.”