Wonderful Life: The Burgess Shale and the Nature of History

by Stephen Jay Gould

  Harry Whittington was clearly the best possible person for the Burgess project. He was not only the world’s leading expert on fossil trilobites (the most conspicuous arthropods of the fossil record), but he had done his most elegant work (Whittington and Evitt, 1953, for example) on rare three-dimensional specimens preserved in silica. The original calcium carbonate of these fossils had been replaced by silica, while the surrounding limestone retained its carbonate base. Since carbonates are dissolved by hydrochloric acid, while silicates are unaffected, the matrix could be dissolved away, providing the rare advantage of three-dimensional preservation completely separable from the surrounding rock. Whittington had therefore been blessed with an ideal, if unwitting, preparation for the Burgess Shale many years later. He had studied three-dimensional structure within rock and then been able to judge his hunches and hypotheses by dissolving the matrix and recovering the fossils intact. These studies “preadapted” Whittington, to use a favorite word in the jargon of evolutionary biology, for his discovery and exploitation of three-dimensional structure in the Burgess Shale fossils.

  * * *

  THE CLASSIFICATION AND ANATOMY OF ARTHROPODS

  Don’t accept the chauvinistic tradition that labels our era the age of mammals. This is the age of arthropods. They outnumber us by any criterion—by species, by individuals, by prospects for evolutionary continuation. Some 80 percent of all named animal species are arthropods, the vast majority insects.

  The higher-level taxonomy of arthropods therefore becomes a subject of much concern and importance. Many schemes have been proposed, and their differences continue to inspire debate. But general agreement can be wrested from most quarters concerning the number and composition of basic subgroups within the phylum. (The evolutionary relationships among subgroups are more problematical, but this subject will not be a major concern of this book).

  1. Representative fossil specimens of the four great groups of arthropods, taken from the most widely used textbook in the history of paleontology, the late-nineteenth-century work of Zittel. (A) A giant dragonfly from the Carboniferous, representing the Uniramia. (B) A fossil eurypterid, representing the Chelicerata. The first pair of head appendages is small and hidden under the carapace; the other five pairs are visible in this figure. (C) A fossil crab, representing the Crustacea. (D) A trilobite.

  The scheme that I follow here is conservative and traditional, the closest to consensus that can be achieved. I recognize four major groups, three still living, one exclusively fossil (figure 1), and I make no proposal about evolutionary connections among them.

  1. Uniramia, including insects, millipedes, centipedes, and perhaps also the onychophores (a small and unusual, but particularly fascinating group, of which a good deal more later on, for the Burgess Shale contains a probable member).

  2. Chelicerata, including spiders, mites, scorpions, horseshoe crabs, and the extinct eurypterids.

  3. Crustacea, primarily marine (the terrestrial pillbug, an isopod, ranks as an exception), and including several groups of small bivalved forms, little known to nonprofessionals, but fantastically diverse and common in the oceans (copepods and ostracodes, for example), the barnacles, and the decapods (crabs, lobsters, and shrimp), whom we eat with relish while regarding their insect cousins as disgusting and unpalatable.

  4. Trilobita, everybody’s favorite invertebrate fossil, extinct for 225 million years, but common in Paleozoic rocks.

  Since the resolution of the Burgess Shale fauna depends so centrally upon an understanding of the amazingly diverse and disparate arthropods, we must enter into some details of arthropod anatomy. Lest this prospect sound daunting, let me assure you that I shall keep the jargon to an absolute and fully comprehensible minimum—only about twenty terms from among more than a thousand available. (I shall not list these terms, but rather define them in the course of discussion. All key terms are underlined at their first use.)

  The basic principle of arthropod design is metamerism, the construction of the body from an extended series of repeated segments. The key to arthropod diversification lies in recognizing that an initial form composed of numerous nearly identical segments can evolve by reduction and fusion of segments, and by specialization of initially similar parts on different segments, into the vast array of divergent anatomies seen in advanced arthropods. Fortunately, we can grasp the complexities of this central theme in arthropod evolution by considering just two matters: the fusion and differentiation of segments themselves, and the specialization of appendages.

  The numerous separate and similar segments of ancestral arthropods (figure 2) tended to coalesce into fewer specialized groups. The most common arrangement is a three-part division, into head, middle, and rear (called by various names, such as cephalon, thorax, and pygidium in trilobites, or head, thorax, and abdomen in insects and crustaceans). Most chelicerates have a two-part division, with a prosoma followed by an opisthosoma. The fused tailpiece of many crustaceans is called a telson.

  Arthropods have external skeletons, or exoskeletons (stiff, but unmineralized in most groups, thus explaining the rarity of many arthropods as fossils). As segments fused, their exoskeletal parts joined to form discrete skeletal units called tagma. This process of fusion is called tagmosis. Different patterns of skeletal tagmosis provide a primary criterion for identifying fossil arthropods.

  Just as important, and as crucial to the Burgess story, is the specialization and differentiation of appendages. Each segment of the original, unspecialized, many-segmented arthropod bore a pair of appendages—one on each side of the body. Each appendage consisted of two branches, or rami (singular ramus). These rami are named according to their position—the inner ramus and the outer ramus—or according to their usual function. Since the outer branch often bears a gill used in respiration or swimming (or both), it is often called the gill branch. The inner branch is usually used in locomotion, and may be called the leg branch, walking branch, or walking leg. (The common term “walking leg” may strike readers as amusingly redundant, but “leg” is an anatomical, not a functional term, and not all arthropods use their legs for walking; insect mouth parts, for example, are slightly modified legs.)

  2. The numerous similar segments of a primitive arthropod, as seen in the trilobite Triarthrus. With the exception of the frontal antennae, all pairs of appendages are similar and biramous, and each body segment has a single pair. (A) Top view. (B) Bottom view. From Zittel.

  3. Cross section through a body segment of an arthropod, showing a pair of typical biramous limbs. Drawn by Laszlo Meszoly.

  This original structure (figure 3) is called a biramous (literally, “two-branched”) limb. (If you retain no other term from this discussion, please inscribe the definition of a biramous limb in your long-term memory. It is the single most important facet of arthropod anatomy in our Burgess discussion.) Specialized arthropods often lose one of the two branches, retaining the other as a uniramous (“one-branched”) limb. (Please place “uniramous” next to “biramous” in your long-term memory.) The higher-level taxonomy of arthropods records the different mixes of uniramous and biramous limbs on various parts of the body.

  The walking legs of most marine arthropods perform an additional function that seems odd from our vertebrate-centered perspective. Some marine arthropods feed as we do by seizing food items in front of their head and passing them directly to the mouth. But most use their walking legs to grasp food particles and pass them forward to the mouth along a food groove situated in the ventral (bottom) midline, between the legs. (The top side of an animal is called dorsal.) Arthropod means “jointed foot,” and the appendages are composed of several segments. Segments located near the body are proximal; those far a way at the ends of the appendage are distal. The most proximal segment of the walking leg is called a coxa. The edge of the coxa bordering the food groove is often armed with teeth, used to capture and move the food forward (see figure 3) and called a gnathobase (literally, “jawed foundation”).


  We form the higher-level taxonomy of arthropods by joining the two principles discussed above: patterns of tagmosis, or fusion of segments, and specialization of appendages by loss of one ramus and differentiation of the other. Beginning with an ancestral arthropod built of many unfused segments, each bearing a biramous limb, the major groups have evolved along different routes of tagmosis and specialization. Consider the four major kinds of arthropods:

  1. Uniramia. As the name implies, insects and their kin have invariably lost the gill branch of the original biramous limb; they build their appendages (antennae, legs, mouth parts) exclusively from leg branches. (Insects breathe through invaginations of the external body surface, called tracheae.)

  2. Chelicerata. Most modern chelicerates have six uniramous appendages on the prosoma. The first pair—chelicerae—are jawlike at the distal end and are used for grasping. (Antennae are absent in this group.) The second pair—pedipalps—are usually sensory in function. The last four pairs are usually leglike (giving spiders their eight legs). All these anterior appendages have evolved from leg branches. The situation is reversed on the posterior section. The opisthosomal appendages are also uniramous, but have been built from gill branches only. (The “lung-books,” or breathing organs, of spiders are on the abdomen.)

  3. Crustacea. Despite an enormous diversity of form, from barnacles to lobsters, all crustaceans are distinguished by their stereotypical pattern of five pairs of appendages on the head (indicating that the head was formed by a tagmosis of at least five segments). The first two pairs, usually called antennae and antennules, are uniramous; they lie in a pre-oral position, in front of the mouth, and have sensory functions. The last three lie in a post-oral position, behind the mouth, and are usually used in feeding, as mouth parts. Appendages on the trunk often retain the original biramous form.

  4. Trilobita. The trilobite head bears one pre-oral pair of appendages (antennae) and three post-oral pairs. Each body segment usually bears a pair of biramous limbs very little modified from the presumed ancestral form.

  The stereotypy of these patterns is, perhaps, the most striking phenomenon in modern arthropods. Of nearly a million described species of insects, none has a biramous appendage, and nearly all have exactly three pairs of limbs on the thorax. Marine crustacea display incredible diversity of form, but all have the same pattern of tagmosis in the head—two pre-oral and three post-oral pairs of appendages. Apparently, evolution settled upon just a few themes or ground plans for arthropods and then stuck with them through the greatest story of diversification in the entire animal kingdom.

  The story of the Burgess Shale ranks as perhaps the most amazing in the history of life largely in relation to this phenomenon of later restriction in arthropod ground plans—for in addition to early representatives of all four later groups, the Burgess Shale, one quarry in British Columbia, contains fossils of more than twenty additional basic arthropod designs. How could such disparity originate so quickly? Why did only four basic designs survive? These questions form the primary subject of this book.

  * * *

  If Harry Whittington had known at the outset what a restudy of the Burgess Shale would require in time and commitment, he would probably not have started. He was fifty years old during the first field season of 1966, and already had enough commitments to last a lifetime. Moreover, as professor of geology at Cambridge he had oppressive administrative responsibilities that could not be delegated.

  But the Burgess was too beautiful and variegated a plum to resist. Besides, everybody knew that its arthropods—the focus of Whittington’s proposed work—posed no major taxonomic dilemmas. Harry told me that when he first decided to work on the Burgess, he “expected to spend a year or two describing some arthropods—full stop.” In England, a “full stop” is a period—ending the sentence, and ending the project.

  It was not to be. Harry Whittington spent four and a half years just writing his first monograph on the genus Marrella. Surprise cascaded upon surprise, starting slowly with doubts about the identity of certain arthropods, and accelerating until a new interpretation jelled in the mid-1970s. This view blossomed to guide all subsequent work toward a new conception for the history of early life. As I read the taxonomic monographs in chronological order, I came to see this story as a classical drama in five acts. No one was killed; few people even got angry. But just as Darwin let his theory gestate for twenty-one basically quiet years between formulation and publication, the similar time for the reevaluation of the Burgess Shale has produced, behind a placid exterior, an intellectual drama of the highest order.

  The Burgess Drama

  ACT 1. Marrella and Yohoia: The Dawning and Consolidation of Suspicion, 1971–1974

  THE CONCEPTUAL WORLD THAT WHITTINGTON FACED

  Harry Whittington is, by nature, a cautious and conservative man. To this day, though he served as midwife to a major transformation of thought, he views himself as an empiricist, with skill in the meticulous description of arthropod fossils. His favorite motto exhorts his younger colleagues to place fact and description before theory, for “one should not run before one can walk.”

  Whittington began, as would any paleontologist who believes in cranking up slowly and deliberately, with the genus Marrella, the most common organism in the Burgess Shale. Marrella splendens overwhelms anything else in the Burgess by sheer abundance. Walcott collected more than 12,000 specimens. Whittington’s party gathered another 800, and I am custodian to 200 more, collected by Percy Raymond in 1930. Many Burgess species are known from fewer than ten specimens, some from only one. But with nearly 13,000 potential views, one need hardly worry about destroying unique evidence by dissection, or failing to find a crucial orientation.

  Marrella splendens is the first Burgess organism that Walcott found and drew; it virtually identifies the Burgess Shale. When Walcott described Marrella formally in 1912, he recognized that his “lace crab” was not a conventional trilobite, but still placed Marrella in the class Trilobita, order previously unknown. Following his need to view Burgess organisms as primitive members of later successful groups, he wrote: “In Marrella the trilobite is foreshadowed” (1912, p. 163).

  Not all of Walcott’s colleagues were convinced. In the Smithsonian archives, I found some interesting correspondence with Charles Schuchert, celebrated Yale paleontologist and codifier of the canonical legend about Walcott’s discovery of the Burgess Shale. After reading his friend Walcott’s paper on the Burgess arthropods, Schuchert wrote to him on March 26, 1912:

  To you personally I want to say that from the first time that I saw Marrella and now with your many excellent pictures of this animal I still cannot get it into my head that this is a trilobite.… I cannot see how it can be a trilobite. Such gills are unknown, I believe, in any trilobite. However, I am only throwing out these half-digested ideas for your consideration rather than to convince you that Marrella is not a trilobite.

  Yet Schuchert, as committed as Walcott to the larger theme that all Burgess creatures belong in known groups, never suggested uniqueness for Marrella, but only hinted at a different home among well-known arthropods.

  To give some idea of the conceptual barriers that Whittington faced when he began to redescribe the arthropods of the Burgess Shale, I must now exemplify what I shall call, throughout this volume, “Walcott’s shoehorn”—his decision to place all Burgess genera in established major groups. Most readers will need to consider these pages in conjunction with the insets on taxonomy and arthropod anatomy (pages 98 and 102). I am asking some investment here from readers with little knowledge of invertebrate biology. But the story is not difficult to follow, the conceptual rewards are great, and I shall try my best to provide the necessary background and guidance. The material is not at all conceptually difficult, and the details are both beautiful and fascinating. Moreover, you can easily retain the thread of argument without completely following the intricacies of classification—as long as you realize that Walcott and all students of the Burg
ess before Whittington placed these organisms in conventional groups, and that Whittington slowly weaned himself away from this tradition, and toward a radical view about the history of life’s diversification.

  Walcott presented his complete classification of Burgess arthropods on page 154 of his 1912 paper (reproduced here as table 3.1). He scattered his Burgess genera widely among four subclasses, all placed within his version of the class Crustacea. Walcott defined Crustacea far more broadly than we do today. He included virtually all marine and freshwater arthropods, organisms that span the entire arthropod phylum as defined today. Of his four subclasses, the modern branchiopods (1) are a group of predominantly freshwater crustaceans, including the brine shrimp and the cladocerans, or water fleas; malacostracans (2) form the great group of marine crustacea, including crabs, shrimp, and lobsters; trilobites (3) are, of course, the most famous of fossil arthropods; while merostomes (4), including fossil eurypterids and modern horseshoe crabs, are closely related to terrestrial scorpions, mites, and spiders.

  The fate of Walcott’s 1912 chart is a striking epitome of the entire Burgess story. Of his twenty-two genera, only two are legitimate members of their groups. Nathorstia (now called Olenoides serratus) is an uncontroversial trilobite (Whittington, 1975b); Hymenocaris (now called Canadaspis) is a true crustacean of the malacostracan line (see Act 3). Three genera (Hurdia, Tuzoia, and Carnarvonia) are bivalved arthropod carapaces with no soft parts preserved; they cannot be properly allocated to any arthropod subgroup, and remain unclassified today. Three other names do not belong to the story of Burgess arthropods: Tontoia, position still unresolved and possibly inorganic, comes from the Grand Canyon, not the Burgess Shale; Bidentia is an invalid name, and these specimens belong to the genus Leanchoilia; Fieldia, misidentified by Walcott, is a priapulid worm, not an arthropod.