Leonardo’s Mountain of Clams and the Diet of Worms

by Stephen Jay Gould

  Arthropods concentrate their nervous system on their ventral (belly) side as two major cords running along the bottom surface of the animal. The mouth also opens on the ventral side, with the esophagus passing between the two nerve cords, and the stomach and remainder of the digestive tube running along the body above the nerve cords. In vertebrates, and with maximal contrast, the central nervous system runs along the dorsal (top) surface as a single tube culminating in a bulbous brain at the front end. The entire digestive system then runs along the body axis below the nervecord. (The accompanying figure from Gaskell’s book illustrates this cardinal difference in an unconsciously amusing way.) But could evolution (or a sensible divine creator, for that matter) turn an arthropod with belly above nerve cords into a superior vertebrate with brains on top and gut below?

  Gaskell proposed a pretty wild scheme for such a transformation, and his loss of respect (and students) followed his inability to construct a cogent defense. Gaskell argued that the dorsal gut of arthropods evolved into the vertebrate brain and spinal cord as a proliferation of nervous tissue fueled the upward march of progress. This new nervous tissue began to surround the old gut, eventually choking off all digestive function like a strangler fig around a host tree, or an anaconda squeezing the lifeblood from a pig. The modern vertebrate brain surrounds the old arthropod stomach, thus explaining the ventricles—the interior spaces between the folds of the brain—as remnants of an ancestral digestive space. Similarly, the central canal of the spinal cord represents the old arthropod intestine, now surrounded by nervous tissue.

  But this putative solution only engendered an even more troubling problem: If the arthropod gut became the vertebrate nervous system, then what organ can serve as a precursor for the vertebrate gut? This problem stymied Gaskell, and he opted for a deus ex machina that eventually satisfied no one but himself (and perhaps his one last auditor): the vertebrate digestive tube simply arose de novo, to suit an obvious need. Gaskell concluded:

  Vertebrates arose from ancient forms of arthropods by the formation of a new alimentary canal, and the enclosure of the old canal by the growing central nervous system.

  Can we extract any message from Gaskell’s failed theory beyond a stodgy, if appropriate, warning about the virtues of caution and sobriety? I certainly think so, for I have long held, and expressed as a mainstay of these essays, that when fine scientists devote their careers to theories later judged nutty or crazy, interesting and instructing reasons always underlie the paradoxical advocacy. This principle certainly applies in Gaskell’s case because we can identify both a generally constraining bias and a personally compelling reason that drove Gaskell to the odd idea of stomachs turning into brains and new guts arising from nothing but inchoate potentiality.

  Gaskell’s dubious but unquestioned conviction about linear progress served as the general bias that led him to propose an almost alchemical scheme of transmutation from arthropod to vertebrate. But an understanding of the history of this subject also reveals a particular reason that interacted with his general conviction about progress to lead him down a path of increasing irrelevance and loneliness. In short, Gaskell proposed his own nutty theory because he couldn’t abide the older and standard account, also judged by history as a prime case of nuttiness, for linking arthropods and vertebrates.

  Think about the basic contrast, and the most obvious way to produce alignment. Arthropods grow ventral nerve cords with the gut above; vertebrates develop with a dorsal nerve tube and a gut below. Presto turno—and one becomes the other. Why not just invert a segmented worm or an insect to produce the vertebrate pattern? Turn a bug on its back (as Kafka did, come to think of it, when he changed his protagonist into a roach in “The Metamorphosis”), and the internal geometry of a vertebrate emerges—nerves above guts.

  I don’t mean to be frivolous or cavalier about complex and serious matters. All participants in the history of this debate know perfectly well that an inverted worm or insect doesn’t become a vertebrate, tout simple and all nice and clean. More than a few knotty problems and inconsistencies remain. To cite the dilemma most widely discussed in the literature, the esophagus of an inverted bug runs upward through the nervous system (right in the area that will become the vertebrate brain), to emerge at a mouth on top of the head. Clearly this will not do (and has not done in any real vertebrate)! So the inversion theory for deriving vertebrates from arthropods must argue that the old brain-piercing mouth atrophied and closed up, while a new ventral mouth developed at the front end of the vertebrate gut. Forming a new mouth at the end of an old tube may not be quite so bold or improbable as building an entirely new gut from nothing (as Gaskell’s theory required), but no evidence for such a scenario exists either, and the whole tale smacks of fatuous special pleading to save an otherwise intriguing idea.

  In any case, I am not spinning an abstract fairy tale as a hypothetical alternative to Gaskell’s solution. The inversion theory has a long and fascinating history in the discussion of vertebrate origins. The founding version dates to the early nineteenth century and became the centerpiece of a movement often called “transcendental biology,” and centered on the attempt to reduce organic diversity to one or a very few archetypal building blocks that could then generate all actual anatomies as products of rational laws of transformation. Some of Europe’s greatest thinkers participated in this grand, if flawed, enterprise. Goethe, Germany’s preeminent poet-scientist, tried to explain the varied parts of plants as different manifestations of an archetypal leaf. In France, Etienne Geoffroy Saint-Hilaire attempted to portray the skeleton of vertebrates as a set of modifications upon an archetypal vertebra.

  In the 1820s, Geoffroy extended his ambitious program to include annelids and arthropods under the same rubric. With boldness verging on a mania too sweeping to be entirely right but also too ingenious to be completely wrong, he argued that arthropods also built their bodies on a vertebral plan, but with one central difference. Vertebrates support their soft parts with an internal skeleton, but insects, with their external skeletons, must live within their own vertebrae (a reality, not a metaphor, for Geoffroy). This comparison led to other strange consequences, all explicitly defended by Geoffroy, including the claim that a vertebrate rib must represent the same organ as an arthropod leg—and that insects must therefore walk on their own ribs!

  Geoffroy also recognized that the opposite orientations of gut and nervous system posed a problem for his claim that insects and vertebrates represent different versions of the same archetypal animal—and he proposed the first account of the inversion theory to resolve this threat to unification. Geoffroy’s initial version of 1822 makes much more sense than the later evolutionary scenarios of linear transformation that so enraged Gaskell. Geoffroy was an early evolutionist in these decades before Darwin, but he did not devise the inversion theory as a genealogical proposition—that is, he did not argue that an arthropod ancestor evolved directly into a primitive vertebrate by turning over. Geoffroy pursued the quite different aim of establishing a “unity of type” that could generate both arthropods and vertebrates from the same basic blueprint.

  He then argued, quite cogently within his own framework, that this grand Platonic blueprint paid scant attention to such “insignificant” questions of nitty-gritty daily reality as which side of a universal design happened to point toward the sun. The single grand design includes a gut in the middle and the main nerve cords somewhere on the periphery. Arthropods orient this peripheral region down and away from the sun—so we call their nerve cords ventral. But vertebrates orient their spinal cord up and toward the sun—so we call the same structure dorsal in our own kin. In other words, arthropods and verterbates express one common design in two orientations, insignificantly inverted with respect to an external axis of sunlight and gravity.

  But later evolutionary theorists of linear progress had to advance the overtly physical and historical claim that an ancestral lineage of arthropods actually turned over to become the firs
t vertebrates (for the classic statement of the inversion theory in this genealogical form, see William Patten, The Grand Strategy of Evolution, 1920). Gaskell could not abide this indecorous version of his beloved linear progress theory. He could not bear to imagine that the grand procession from jellyfish to man, orchestrated by an ever-increasing mass of nervous tissue, once paused in its stately and orderly march toward human consciousness in order to execute a fancy little flip, a clever jig of inversion, just at the sublime and definitive moment of entrance into the vertebral home stretch.

  Gaskell therefore had to keep his stately soldiers upright and uniformly oriented throughout their journey toward the human pinnacle—and he fulfilled this need by crafting the vertebrate brain and spinal cord from an arthropod digestive tube, while forming a completely new gut below. By this device, he could keep tops on top and bottoms at the bottom throughout the linear history of animal life, while placing nerves below the gut in arthropods, but above the gut in vertebrates. Gaskell thought that his move would rescue the theory of linear progress, with its necessary transition of arthropod into vertebrate, from the absurdities of the old inversion theory. “How is it then,” he wrote, “that this theory has been discredited and lost ground? Simply, I imagine, because it was thought to necessitate the turning over of the animal.” Gaskell therefore invented his peculiar alternative as a refutation of the venerable inversion theory. He wrote of the first vertebrate: “If the animal is regarded as not having been turned over . . . then the ventricles of the vertebrate brain represent the original stomach, and the central canal of the spinal cord the straight intestine of the arthropod ancestor.”

  How ironic. In order to avoid the “nutty” theory of inversion, Gaskell invented the even odder notion of stomachs turning into brains with new guts forming below. No wonder, then, that later biologists cast a plague on both speculative houses and opted instead for the obvious alternative: arthropods and vertebrates do not share the same anatomical plan at all, but rather represent two separate evolutionary developments of similar complexity from a much simpler common ancestor that grew neither a discrete gut nor a central nerve cord. After all, we now know that arthropods and vertebrates have been separated for more than 500 million years, and that “simpler” arthropods did not turn into “complex” vertebrates at some halfway point on a march to a single evolutionary apex.

  Furthermore, this sensible idea of independent derivation meshed beautifully with the triumph, from the 1930s on, of a strict version of Darwinism based on the near ubiquity of adaptive design built by natural selection with little constraint imposed by strictures of a common anatomic ground plan like Goethe’s leaf or Geoffroy’s vertebra. If adaptation and natural selection wield such unimpeded power over the fate of each evolutionary sequence, why search for deeper commonalities in lineages long separate? Arthropods and vertebrates do share several features of functional design. But those similarities only reflect the power of natural selection to craft optimal structures independently in a world of limited biomechanical solutions to common functional problems—an evolutionary phenomenon called convergence.

  After all, if you want to fly, you have to develop wings of some sort, because nothing else can work. Bats, birds, and pterosaurs (flying reptiles of dinosaur times) all evolved wings independently because natural selection knows no other solution, and holds the capacity to build such intricate convergences as independent illustrations of its predominant power. Therefore, if both arthropods and vertebrates evolved guts and nerves in reversed orientations, why worry about different expressions of a common constraint? The two phyla have been separate for half a billion years and undoubtedly evolved their digestive and neurological organs along separate pathways of adaptation.

  This new consensus seemed so compelling that Ernst Mayr, the dean of modern Darwinians, opened the ashcan of history for a deposit of Geoffroy’s ideas about anatomical unity. We now appreciate the immense power of natural selection to build and rebuild every feature; to change, and then to alter again, nearly every nucleotide of every gene in the interest of better adaptation. Lineages that have been separate for 500 million years cannot possibly retain enough genetic identity to encode any important common constraint of design. In his epochal book of 1963, Animal Species and Evolution, Mayr wrote:

  In the early days of Mendelism there was much search for homologous genes that would account for such similarities. Much that has been learned about gene physiology makes it evident that the search for homologous genes is quite futile except in very close relatives.

  The verdict of history had descended. Gaskell had proposed a bizarre theory to reject Geoffroy’s union of arthropods and vertebrates by inversion. But Geoffroy’s theory turned out to be quite weird enough all by itself. Evolutionary studies would finally abandon such romantic nonsense and move into the light of unimpeded natural selection.

  Except for one small matter. Darwin himself told us in his last book (The Formation of Vegetable Mould Through the Action of Worms) that we should never underestimate the collective power of worms on the move. Our general culture also recognizes two primary metaphors, one inorganic and one organic, for the reversal of received opinion. Well may traditionalists fear the turning of these two objects: tables and worms. The inversion of a humble worm, especially when disturbed, may bring down empires. Shakespeare told us that “the smallest worm will turn being trodden on.” And Cervantes wrote in his author’s preface to Don Quixote that “even a worm when trod upon, will turn again.”

  How wonderfully symbolic and real in the double meaning. Geoffroy proposed a theory to unite the architecture of complex animals by comparing vertebrates with segmented worms and arthropods turned over. This theory for the archetype of complex animals became, instead, the archetype of nutty ideas in biology. But turning worms also serve as a leading cultural metaphor for upheaval of accepted ways and thoughts. I have always loved the boldness of Geoffroy’s theory, but I never dreamed that he might have been right—even though I have long embraced, as a centerpiece of my own career, his larger view about the importance of inherited architectural pathways as constraints upon the optimizing power of natural selection. Well, the worm turned twice during the past year or two—in both actual and symbolic styles. Geoffroy, it seems, was correct after all—not in every detail, of course, but at least in basic vision and theoretical meaning. And the triumph of this surprise, the inversion of nuttiness to apparent truth, stands as a premier example of the most exciting general development in evolutionary theory during our times.

  I published my first technical book, Ontogeny and Phylogeny, in 1977. I took pride in this long work on the relationship between embryology and evolution, but also became quite frustrated because we then knew so preciously little about the potential key to a resolution: the genetic basis of development. How does the genetic code help to orchestrate this greatest miracle of everyday biology—the regular and usually unerring production of adult complexity from the apparent formlessness of a tiny fertilized egg? We knew practically nothing, but we assumed (as documented above) that the major animal phyla, all evolutionarily separate for at least 500 million years, could share no constraining common plan or genetic architecture. Pure Darwinism reigned triumphant, and natural selection had built each basic anatomy for its own adaptive utility.

  But we can now determine, easily and relatively cheaply, the detailed chemical architecture of genes; and we can trace the products of these genes (enzymes and proteins) as they influence the course of embryology. In so doing, we have made the astounding discovery that all complex animal phyla—arthropods and vertebrates in particular—have retained, despite their half-billion years of evolutionary independence, an extensive set of common genetic blueprints for building bodies. Many similarities of basic design among animal phyla, once so confidently attributed to convergence, and viewed as testimony to the power of natural selection to craft exquisite adaptation, demand the opposite interpretation that Mayr labeled as inconceivable: the
similar features are homologies, or products of the same genes, inherited from a common ancestor and never altered enough by subsequent evolution to erase their comparable structure and function. The similarities record the constraining power of conserved history, not the architectural skills of natural selection independently pursuing an optimal design in separate lineages. Vertebrates are, in a certain sense, true brothers (or homologs)—and not mere analogs—of worms and insects.

  Examples of this primary reversal of standard theory have been accumulating for the past fifteen years. In the first pathbreaking case, the homeotic genes of insects, responsible for specifying the separate identities of segments along the main body axis (by orchestrating the growth of antennae, mouthparts, legs, and so on in their proper places), were also discovered, in minimally altered form, in vertebrates. (The homeotic genes were first recognized by oddball mutants with body parts in the wrong places—legs growing out of the head where antennae should be, for example. In Drosophila, the homeotic genes occur in two arrays on a single chromosome. Interestingly, in vertebrates, these same arrays exist in multiple copies, as four sequences on four separate chromosomes.) These vertebrate homologs do not control the basic segmentation of the vertebral column (so insect segments are not simple homologs of vertebrae, as Geoffroy had originally proposed). But the homeotic genes of vertebrae do regulate the embryonic segmentation of the mid- and hind brain, and they do strongly influence other important repetitive structures, including the positioning of cranial nerves along the body axis.