Wonderful Life: The Burgess Shale and the Nature of History (34 page)

If the soft-bodied components had never been found, the Burgess Shale would be an entirely unremarkable Middle Cambrian fauna of about thirty-three genera. It contains a rich assemblage of sponges (Rigby, 1986) and algae, seven species of brachiopods, nineteen species of ordinary trilobites with hard parts, four of echinoderms, and a mollusk and coelenterate or two (Whittington, 1985b, pp. 133–39, presents a complete list). Among the soft-bodied organisms, bringing the total biota to about 120 genera, some are legitimate members of major groups. Whittington lists five certain and two probable species of priapulid worms, six species of polychaetes, and three soft-bodied trilobites (
Tegopelte
and two species of
Naraoia
).

My five-act drama, just concluded, emphasizes a different theme, taught to me by the soft-bodied components alone. The Burgess Shale includes a range of disparity in anatomical design never again equaled, and not matched today by all the creatures in all the world’s oceans. The history of multicellular life has been dominated by decimation of a large initial stock, quickly generated in the Cambrian explosion. The story of the last 500 million years has featured restriction followed by proliferation within a few stereotyped designs, not general expansion of range and increase in complexity as our favored iconography, the cone of increasing diversity, implies. Moreover, the new iconography of rapid establishment and later decimation dominates all scales, and seems to have the generality of a fractal pattern. The Burgess revisions of Whittington and colleagues have specified three ascending levels.

1.
Major groups of a phylum
. No group of invertebrate fossils has received more study, or stands higher in general popularity, than trilobites. The mineralized skeletons of conventional fossils show extraordinary diversity, but all conform to a basic design. One would hardly have anticipated, after all this study, that the total anatomical range of the group could have been far broader in its early days. Yet soft-bodied
Naraoia
is undoubtedly a trilobite in its distinctive series of head appendages (one pair of antennae and three post-oral biramous pairs), and its conventional body appendages of the “right” form and number of segments. Yet the exoskeleton of
Naraoia
, with its two valves, stands far outside the anatomical range of the group as seen in conventional fossils.

2.
Phyla
. We can completely grasp the extent of a surprise only when we also know the full range of conventional possibilities—for we need a baseline of calibration. I find the story of Burgess arthropods particularly satisfying because the baseline has “no vacancy,” and all additional disparity truly supplements a full range of membership in major groups. The orphaned arthropods of the Burgess are spectacular, but the representatives of conventional groups are just as important for documenting the first phrase of the primary theme—“all we could expect and then a great deal more.” The recent discovery of
Sanctacaris
brings the conventional roster to completion. All four great groups of arthropods have representatives in the Burgess Shale:

But the Burgess Shale contains an even greater range of anatomical experiments, equally distinct in design and functionally able, but not leading to subsequent diversity. A few of these orphans may show relationships among themselves—
Actaeus
and
Leanchoilia
, perhaps, on the basis of their distinctive frontal appendages—but most are unique, with defining features shared by no other species.

The monographic work of Whittington and colleagues has identified thirteen unique designs (table 3.3), all discussed in the preceding chronology. But how many more have yet to be described? Whittington lists twenty-two species (and inadvertently omits
Marrella
) in his category “not placed in any phylum or class of Arthropoda” (1985b, p. 138). Therefore, by best estimate, the Burgess Shale contains at least twenty unique designs of arthropods, in addition to the documented representatives of all four great groups within the phylum.
*

3.
Multicellular animal life as a whole
. The weird wonders of the Burgess Shale excite our greatest fascination, though the arthropod story is every bit as satisfying intellectually, especially for its completion of the baseline and consequently firm estimate for the relative frequency of oddballs. Still, whereas
Marrella
and
Leanchoilia
may be beautiful and surprising,
Opabinia, Wiwaxia
, and
Anomalocaris
are awesome—deeply disturbing and thrilling at the same time.

The Burgess revision has identified eight anatomical designs that do not fit into any known animal phylum: in order of publication,
Opabinia, Nectocaris, Odontogriphus, Dinomischus, Amiskwia, Hallucigenia, Wiwaxia
, and
Anomalocaris
. But this list is nowhere near complete—surely less exhaustive than the account of documented oddballs among arthropods. The best estimates indicate that only about half the weird wonders of the Burgess Shale have been described. Two recent sources have provided lists of all potential creatures in this category of ultimate strangeness. Whittington counts seventeen species of “miscellaneous animals” (1985b, p. 139), and I would add
Eldonia
to his total. Briggs and Conway Morris count nineteen “Problematica from the Middle Cambrian Burgess Shale of British Columbia” (1986). Finding no basis for genealogical or anatomical arrangement among the weird wonders, they simply list their nineteen creatures in alphabetical order.

What may the future bring us in further surprises from the Burgess Shale? Consider
Banffia
, namesake of the more famous national park adjoining Yoho and the Burgess Shale. Walcott’s “worm”—with an annulated front portion separated from a saclike posterior—is almost surely a weird wonder. Or
Portalia
, an elongate animal with bifurcating tentacles arrayed along the body axis. Or
Pollingeria
, a scalelike object with a meandering tubelike structure on top. Walcott interpreted
Pollingeria
as a covering plate from a larger organism, akin to the sclerites of
Wiwaxia
, and explained the meandering tube as a commensal worm, but Briggs and Conway Morris think that the object could be an entire organism. The general form of the Burgess story may now be well in hand, but Walcott’s quarry has not yet yielded all its particular treasures.

This book, long enough already, cannot become an abstract treatise on the rules of evolutionary inference. But I do need to provide a few explicit comments on how paleontologists move from descriptions of anatomy to proposals about genealogical relationships—so that my numerous statements on this subject receive some underpinning and do not stand as undefended pronouncements
ex cathedra
.

Louis Agassiz, the great zoologist who founded the institution that now houses both me and the Raymond collection of Burgess Shale fossils, picked a superficially peculiar name that we retain with pride—the Museum of Comparative Zoology. (Anticipating the hagiographical urges of his contemporaries, he even explicitly requested that his chosen title be retained in perpetuity, and that the museum not be renamed for him upon his demise.) Experiment and manipulation may form the stereotype of science, Agassiz argued, but disciplines that treat the inordinately complex, unrepeatable products of history must proceed differently. Natural history must operate by analyzing similarities and differences within its forest of unique and distinctive products—in other words, by comparison.

Evolutionary and genealogical inferences rest upon the study and meaning of similarities and differences, and the basic task is neither simple nor obvious. If we could just compile a long list of features, count the likenesses and unlikenesses, gin up a number to express an overall level of resemblances, and then equate evolutionary relationship with measured similarity, we could almost switch to automatic pilot and entrust our basic job to a computer.

The world, as usual, is not so simple—and thank goodness, for the horizon would probably be a disappointing place anyway. Similarities come in many forms: some are guides to genealogical inferences; others are pitfalls and dangers. As a basic distinction, we must rigidly separate similarities due to simple inheritance of features present in common ancestors, from similarities arising by separate evolution for the same function. The first kind of similarity, called homology, is the proper guide to descent. I have the same number of neck vertebrae as a giraffe, a mole, and a bat, not (obviously) because we all use our heads in the same way, but because seven is the ancestral number in mammals, and has been retained by descent in nearly all modern groups (sloths and their relatives excepted). The second kind of similarity, called analogy, is the most treacherous obstacle to the search for genealogy. The wings of birds, bats, and pterosaurs share some basic aerodynamic features, but each evolved independently; for no common ancestor of any pair had wings. Distinguishing homology from analogy is the basic activity of genealogical inference. We use a simple rule: rigidly exclude analogies and base genealogies on homology alone. Bats are mammals, not birds.

Using this cardinal rule, we can go a certain distance with the Burgess Shale. The tail flukes of
Odaraia
bear an uncanny resemblance to functionally similar structures of some fishes and marine mammals. But
Odaraia
is clearly an arthropod, not a vertebrate.
Anomalocaris
may have used its overlapping lateral flaps to swim by undulation, much as certain fishes with continuous lateral fins or flattened body edges do—but this functional similarity, evolved from different anatomical foundations, indicates nothing about genealogical relationship.
Anomalocaris
remains a weird wonder, no closer to a vertebrate than to any other known creature.

But the basic distinction between homology and analogy will not carry us far enough. We must make a second division, among homologous structures themselves. Rats and people share both hair and a vertebral column. Both are homologies, structures inherited from common ancestors. If we are searching for a criterion that will properly unite rats and people into the genealogical group of mammals, we can use hair, but the shared vertebral column will not help us at all. Why the difference? Hair works because it is a
shared-and-derived
character, confined to mammals among the vertebrates. A vertebral column is no help because it is a
shared-but-primitive
character, present in the common ancestor of all terrestrial vertebrates—not just mammals—and most fish.

This distinction between properly restricted (shared and derived) and overly broad homologies (shared but primitive) lies at the core of our greatest contemporary difficulties with Burgess organisms.
*
For example, many Burgess arthropods have a bivalved carapace; many others share the basic “merostomoid” form, a broad head shield followed by numerous short and wide body segments capped by a tail spike. These two features are, presumably, genuine arthropod homologies—each bivalved lineage doesn’t start from scratch and develop the same complex structure, slowly and separately. But neither the presence of a bivalved carapace nor “merostomoid” body form can identify a genealogically coherent group of Burgess arthropods because both are shared-but-primitive characters.

Figure 3.71 should clarify the reason for rejecting shared-but-primitive traits as a guide to genealogy. This evolutionary tree represents a lineage that has diversified into three great groups—I, II, and III—by the time marked by the dashed line. A star indicates the presence of a homologous trait—call it five digits on the front limb—inherited from the distant common ancestor (A). In many branches, this trait has been lost or modified beyond recognition. Every loss is marked by a double-headed arrow. Note that at the selected time, four species (1–4) still retain the shared-but-primitive trait. If we united these four as a genealogical group, we would be making the worst possible error—missing the three true groups entirely, while taking members from each to construct a false assemblage: species 1 might be the ancestor of horses; species 2 and 3, early rodents; and species 4, an ancestor of primates, including humans. The fallacy of basing groups on shared-but-primitive traits should be apparent.
*