The Flamingo’s Smile

by Stephen Jay Gould

  Contrary to the romantic image of science and exploration, many important discoveries are made in museum drawers, not under adverse conditions in the parched Gobi or the freezing Antarctic. And so it must be, for the nineteenth century was the great age of collecting—and leading practitioners shoveled up material by the ton, dumped it in museum drawers, and never looked at it again. One of the great zoological discoveries of our century, the primitive segmented mollusk Neopilina, had been dredged from the deep sea, placed in a vial and labeled with the name of a limpet-shaped snail (for its external shell maintains such a shape)—where it remained for several years until H. Lemche turned the vial over to look at the soft parts and saw the segmented gills.

  I am delighted to report that the conodont animal has now, and apparently truly this time, been found—in a museum drawer in Scotland. My friend Euan Clarkson was rummaging through some material of Carboniferous age (about 340 million years old) collected by D. Tait during the 1920s, when he noticed the impression of a worm-shaped creature with conodonts at the front end, right where the mouth should be. Since Clarkson is not an expert on conodonts, he called in some colleagues to verify and extend his discovery. Their results have just been published (Derek E.G. Briggs, Euan N.K. Clarkson, and Richard J. Aldridge, in bibliography).

  Our fossil record is almost entirely the history of hard parts—bones, teeth, shells, and plates—because soft structures decay quickly and do not fossilize. Under very special circumstances, soft parts can be preserved, and these rare windows on the true diversity of past life are among the most precious of our fossil localities. For the 600 million years that multicellular animals have dominated our earth’s fauna, we have no more than a dozen or so extensive deposits of soft-bodied creatures. Most famous are the carbonized films of weird and wonderful creatures from the Burgess Shale, Cambrian of Alberta (some 550 million years old, and most ancient of our extensive windows); animals preserved within ironstone concretions from the Mazon Creek Formation of Illinois, Carboniferous period (350–270 million years old); and the Jurassic (180–130 million years) lithographic limestones of Solnhofen, Germany, where remains of Archaeopteryx, the first bird, feathers and all, were discovered.

  The conodont animal comes from one of our smaller windows, the so-called shrimp band within the Granton Sandstones found east of Edinburgh. The Granton Sandstones are a sequence of lake and lagoonal sediments deposited in fresh or slightly saline water. This basin was occasionally flooded by the sea, and the shrimp band represents one such marine incursion. Its soft-bodied fauna was preserved because two unusual conditions prevailed during this brief flood. First, the bottom waters apparently lacked oxygen. No animal scavengers or bacteria could live on the lake floor, and dead animals floating down from above were not dismembered or decomposed. (We make these inferences because the shrimp band displays continuous, fine-layered sedimentation, an indication that no creatures burrowed or plowed through the bottom muck.) Second, the basin was stagnant and virtually devoid of currents. Thus, fragile, soft-bodied creatures were not pulled apart but floated gently down to be buried intact.

  Fossil of a conodont animal (here outlined with dashes) recently found in a museum drawer in Scotland. Its taxonomic position is controversial; it may best be placed in a new phylum of its own. REPRINTED FROM NATURAL HISTORY.

  The conodont animal is wormlike in appearance, some 40.5 mm long, but no more than 2 mm wide (see photo on chapter 16). Its head end seems to be cleft, with two broad lobes surrounding a central depression (entrance to the mouth, perhaps). Just behind the head, conodonts are affixed along one edge in a sensible position for the mouth. They occur in three groups and contain elements of a well-known assemblage. Thus, Clarkson and his colleagues did not need to invent a name for their creature; they included it within the genus Clydagnathus, first established in 1969 for the skeletonized conodonts alone. A few faint lines run along the interior of the animal, parallel to its sides. Whether these represent a gut, a nerve tube, perhaps even the chordate notochord, we do not know. About two-thirds of the way back, and extending nearly to the posterior end, we find an intriguing sequence of repeated segments, some thirty-three in all, sloping at an angle to the midline of the body. Finally, one edge of the posterior end seems to sport a sequence of projections, interpreted as fin rays. Not much else worthy of mention has been preserved. At least the structures of Clydagnathus confirm one old assumption about conodont elements—they represent the only hard parts of an otherwise entirely soft-bodied creature. No wonder we had so little previous success in determining their affinity.

  As I said at the outset, Clarkson and his colleagues have solved only half the conodont problem. They have found the elusive animal, but they do not know where it belongs. Of modern animal phyla, only two seem worthy of discussion as a possible taxonomic home for the conodont animal. Perhaps it is a chordate—that is, a prevertebrate member of our own phylum. Yet, each potential similarity with chordates scarcely carries conviction. The slender and flattened eel-shaped body reminds us of some chordates, but we find the same general shape in several other phyla as well. The faint lines parallel to the animal’s side could represent such chordate structures as the notochord, but they may simply be remnants of the gut as well, an organ shared by virtually all “higher” animals. The apparent fin rays of the posterior end suggest chordate affinities, but similar structures occur in several other phyla as well. The V-shaped segments seem to say “chordate,” but these structures are so poorly preserved that we cannot really distinguish between a chordate style of segmentation and the patterns of several other phyla with serially repeated elements. In short, we find a few general and superficial similarities with chordates but nothing specific, and surely nothing that would warrant any firm, or even tentative, placement within our phylum.

  The Chaetognatha, or arrow worms, a small marine group located not far from chordates on our evolutionary tree, include the only other viable candidates for a link between the conodont animal and some modern group. Chaetognaths are armed with grasping spines that flank the mouth in two lateral sets. These spines bear a superficial resemblance to conodonts, but they are made of chitin, not calcium phosphate. Chaetognaths also have tail fins not unlike those of the conodont animal. But they also have lateral fins, and such structures are not present on the conodont animal (in an area of the body—the posterior—where preservation is detailed and excellent). In short, chaetognaths seem an even less worthy prospect than chordates as a home for the conodont animal.

  Briggs, Clarkson, and Aldridge therefore conclude, with ample justice in my opinion, that the conodont animal is unique and previously unknown. It must be placed in a separate phylum—the Conodonta. After all, they argue, if a century of efforts to squeeze them into some modern group have been dashed on the enigma of their peculiar hard parts, why should the discovery of equally ambiguous soft parts comfortably fit them into some well-established pigeonhole of our taxonomy? They write: “The lack of a definitive solution to this problem in 125 years of research emphasizes the uniqueness of conodonts.” And with this conclusion—that conodonts must be placed in a new and separate phylum of their own—we finally come to the general message that inspired me to write this essay.

  Paleontologists are, in general, a conservative lot. Problematica of uncertain taxonomic affinity and few species are an embarrassment and an untidy bother; nothing makes an old-style paleontologist happier than the successful housing of problematical organisms within a well-known group. The admission that Problematica must be treated by erecting new phyla flies in the face of hope and tradition and represents a last resort. In recent years, that resort has been followed more and more often because—well, damn it—many Problematica are weird, wonderful, and unique and simply do not fit into any known group. This most unwilling admission reflects an important and little-known fact about the history of life.

  To appreciate this fact and its implications, we must study the distribution in time of Problematica that
cannot be placed into conventional phyla. The history of life has featured multicellular animals only during the past 600 million years. We divide this time into three great eras—the Paleozoic (or ancient life), the Mesozoic (or middle life), and the Cenozoic (or recent life). Virtually all Problematica now begrudgingly granted their own phylum lived during the oldest era, the Paleozoic (although conodonts, after living throughout the Paleozoic, just sneaked into the Triassic, the first period of the Mesozoic). This fact, the focal point of my essay, may not strike you, at first, as strange. After all, the further back we go, the more different should life become from our modern phyla. But two aspects of this distribution in time are surprising and point to a major pattern. First, although we might expect a general decrease in the number of problematic groups through time, we would not anticipate an abrupt disappearance of oddities after the Paleozoic. We do not find a gradual decline in curious creatures. Instead, they abound in the lower Paleozoic, become rare by the end of the Paleozoic, and cease thereafter. Of the three windows I mentioned, the Burgess Shale (lower Paleozoic) is chock-full of Problematica, the Mazon Creek (upper Paleozoic) sports two, the Solnhofen lithographic limestone (Mesozoic), none. Something about the earliest history of multicellular life encouraged a flowering of Problematica. Something about its later history (and not much later) dried the well completely.

  Second—although the conodonts are an exception to this generality—most Problematica are rare, restricted in time, and represented by only a few species. Phyla are supposed to be big groups—arthropods with their 750,000 species of insects, or chordates with their 20,000 species of fishes. They are also supposed to endure for a long time. Taxonomists are stingy; they do not like to establish a group just below the highest level of kingdom for just a few species that lived but a few million years. If Problematica were restricted to the Paleozoic but were all as abundant and long-lived as conodonts, the pattern would not be as troubling or curious. But some Problematica, now housed in their own phylum, are known as only one species found in a single place. And some are surpassingly strange. Consider the animal so formidably curious that it goes by the Latin name Hallucigenia, coined by its author, Simon Conway Morris, for “the bizarre and dream-like appearance of the animal.” (Simon once told me that it resembled something he had seen on a trip—and I don’t mean to Boston.) Hallucigenia (from that first and most famous window, the Burgess Shale) has an elongate body, nearly an inch in length, supported by seven pairs of spines that look nothing like the legs of any known creature. It has a bulbous head and, behind it, a row of tentacles, each forked at the tip, running along the back. Behind the tentacles lies a smaller and bunched array of projections recalling the spines on a Stegosaurus’s tail. An anal tube projects upward at the rear end (see figure on chapter 16). Damned strangest thing I’ve ever seen in my life. Or consider the peculiar Problematica from the second window, our own Mazon Creek Formation of Illinois. It also bears a whimsical formal name, a Latinization of its discoverer, a Mr. Tully, and its appearance. It is called Tullimonstrum. The Tully monster is a peculiar, roughly banana-shaped creature, some three to six inches long. Like Hallucigenia, it is so different from anything else we know that it seems to demand a phylum of its own.

  We tend to think of evolution as progressive change within lineages—fish become amphibians, reptiles, mammals, and finally humans—and we therefore miss important themes related to a different and more pervasive aspect of evolution: changing diversity, considered as absolute numbers of species and their relative abundance through time. The predominance of Paleozoic Problematica records an outstanding theme in the history of diversity. This theme imparts a direction to time that is more clear and reliable than any statement we can make about change within lineages. It also probably reflects a more general and basic law about the history of change in natural systems.

  During the past decade, paleontologists have hotly debated the pattern of change through time in the diversity of marine animals. Do more species live now (as the “progressivist” view of evolution might suggest) or has the number of species remained roughly constant following a quick achievement of some equilibrium value after the Cambrian explosion? The problem is not so easy to solve as it may seem. You can’t simply count the number of species described for each interval of time. The fossil record is notoriously imperfect, and it tends to get worse the further back you go. Thus, an empirical increase in abundance of known fossils could actually reflect a decrease of true diversity.

  Simon Conway Morris’s restoration of Hallucigenia sparsa. Note the seven pairs of spines below (labeled S), the bulbous “head” (in front, labeled Hd), the single row of forked tentacles on the back (labeled Tt), the bunched array of projections at the rear (labeled St. Tt.) and the upstanding “anal tube” (labeled An). FROM PALAEONTOLOGY, VOLUME 20, 1977, p. 628.

  Arguments have therefore raged back and forth, but in 1981 the four leading debaters buried the hatchet and published a joint paper of welcome agreement (J.J. Sepkoski, R.K. Bambach, D.M. Raup, and J.W. Valentine, in bibliography). Several sources of data (all corrected as best we can for imperfection of the record) now point to a clear pattern of real increase through time—not steady and progressive but unmistakable as a general direction. Modern oceans contain at least twice the number of species as our average Paleozoic seas.

  We might therefore expect—indeed it seems unavoidable—that modern seas would not only contain more species but also more distinct kinds of creatures, more basically different body plans. Yet it is not so. Today, double the number of species are crammed into far fewer groups of higher taxonomic rank. To be sure, we still find several phyla of distinct body plan and low membership—all the wormlike groups with the funny names that no one but specialists know and love: the kinorhynchs, the gnathostomulids, the priapulids, the chaetognaths, already mentioned as a possible home for conodonts, and several others. But our modern seas are dominated by just a few groups—primarily clams, snails, crabs, fishes, and echinoids—each with far more species than any Paleozoic phylum ever attained (with the possible exception of trilobites in the Ordovician and crinoids in the Carboniferous). Paleozoic seas may have contained only half the species that grace our modern oceans, but these species were distributed over a greatly expanded range of basic body plans. This steady decrease in kinds of organic designs—all in the face of a strong increase in numbers of species—may represent the most outstanding trend of our fossil record.

  This steady decrease is well recorded by the pattern of Problematica already discussed. Most of the really weird and wonderful creatures lived exclusively during the Paleozoic. (Don’t be too impressed by the oddity of some modern minor phyla, for many of them did not arise recently but also have records extending back into the Paleozoic.) It is perhaps even better recorded by changes in the number of classes (next lower taxonomic level) within common phyla. Consider just one example, based on a highly conservative counting of classes made by J.J. Sepkoski of the University of Chicago. Modern echinoderms come in four classes, all of respectable to high diversity: sea urchins (the echinoids already cited as a dominant group), starfishes, sea cucumbers, and crinoids. Yet, sixteen additional classes lived and died within the Paleozoic, and sixteen of the total twenty coexisted during the Ordovician period, some 500 million years ago. None of these sixteen classes (with two possible exceptions) ever reached the diversity displayed today by any of the modern survivors.

  The Paleozoic world was very different from ours, with few of a kind distributed over a greatly increased range of basic body forms. Hallucigenia is gone, the Tully monster lives no longer, even the abundant conodonts are extinct. Why has our world of life undergone this profound shift from few species in many groups to many species in fewer groups?

  Of the two general answers, the first is conventional and causal (the second will be based on random processes). It invokes what may be a common property of nearly all natural systems and may therefore have an importance far transcending this p
articular example. The principle might be called “early experimentation and later standardization.” Some 600 million years ago, the Cambrian explosion filled the oceans with their first suite of multicellular animals. Evolution probed all the limits of possibility. Each basic body plan experimented with a great array of potential variants. The pattern of many groups, each with few members, was established. Some of these experiments worked well, but, inevitably, most didn’t—and a gradual sorting-out ensued.

  Many of the failures were flawed from the start and never reached high diversity. They are our taxonomic embarrassments—highly distinct body plans with few species. We call them Problematica and grant them their own phyla only begrudgingly (although if we understood the principle that they represent, we would propose and accept their special names with more equanimity). Others, like the small and extinct classes of Paleozoic echinoderms, are failed experiments with a basic design that does work well in a few successful classes. Thus, sea urchins and starfishes use the echinoderm ground plan to great advantage, while a host of early experiments, bearing such strange names as ctenocystoids, helicoplacoids, and edrioblastoids, quickly bit the dust. Our modern faunas are the winnowed and well-honed survivors of a grand sorting-out based on principles of good engineering.

  The same principle applies to any system free to experiment but ultimately regulated by good and workable design. Electric and steam cars, and a variety of other experiments, yielded to the internal combustion engine (although someday, if we ever run out of oil, they may reemerge like the phoenix). Cars now come in hundreds of brands, each built on the same principle. In 1900, far fewer brands used a much greater variety of basic designs. And consider the blimps, gliders, and variety of powered planes before we settled upon 747s and their ilk.