The ecosystems of shallow marine waters – coral reefs, for example – are the most diverse in the modern oceans, and they have probably been so throughout the history of life. And yet they are under-represented in the fossil record. For environments that are rich in life are also rich in the means of destroying it. When a shrimp or fish dies, it is rapidly devoured by scavenging crustaceans or decomposed by bacteria: every trace of the organism is destroyed in the process. In order for a fossil to be left, the dead organism must somehow be sheltered from the grave-robbing crabs, starfish and bacteria that thrive in shallow-water environments. An animal stands a much better chance of being preserved as a fossil in an anoxic deep-water environment, where there are fewer bacteria and almost no scavengers. The fossil record for oxygen-rich shallow waters is much less complete.
The Burgess Shale is the exception. A lucky accident, or series of accidents, about 530 million years ago transported shallow-water animals by the hundreds of thousand from their normal environment into much deeper water. They were killed in the process but came to lie in a place where they could be preserved. The means was probably a mudslide: the animals, it is thought, were living beneath a cliff made of mud, and when the cliff collapsed it carried down with it an avalanche of shallow-water animals into the anoxic depths.
The Burgess Shale fossil locality is now eight thousand feet up a mountain quarry in British Columbia. Fossils had first been discovered there by a US geologist, Charles Walcott, in 1909 (as Gould convincingly shows, from Walcott’s diaries). In the next eight years, Walcott collected over eighty thousand specimens and deposited them in the Smithsonian Institution in Washington. He never found the time to work on them as thoroughly as he wished, but he did make some preliminary observations. The observations proved highly influential.
Walcott described some of the species, and suggested where they fitted into the Linnaean classification of animal life. He placed, (or, as Gould likes to say, ‘shoehorned’) all the Burgess Shale animals into already-known taxonomic groups. Of 22 arthropods, for example, he thought that eight were fairy shrimps, six were malacostracans (the group containing crabs and prawns), four were trilobites, and four others belonged in another mainly fossil group, the merostomoids (the group containing the horseshoe crab). Not only did Walcott place the Burgess Shale animals in known groups: he also suggested that they were ancestral versions of their modern relatives. After all, you might expect that if you traced life back over 500 million years, you would come to some ancestors. For over half a century, Walcott’s descriptions provided the basis for almost everything that was said about the Burgess fauna.
In 1966, the Cambridge geologist Harry Whittington decided to re-examine the Burgess fossils. His first crucial discovery was that the Burgess fossils have a three-dimensional structure, preserved within the rocks. Walcott had thought that the fossils were squashed flat on the rock surface: Whittington drilled down with a dentist’s drill and found there was more structure below. He was thus able to reconstruct the animals in much more detail. And when Whittington produced his reconstructions, the animals turned out very differently from Walcott’s. They no longer appeared to be primitive versions of modern animals. They were related to modern forms: but as separate early radiations within a broad category, and not as ancestors. For example, many of the fossils are arthropods or annelids, both of which are large modern groups. Walcott had described the Burgess Shale species as ancestral arthropods or worms: but, on Whittington’s reconstructions, the same fossils in many cases turned out not to fit into any modern sub-divisions of annelids or arthropods: they belong in their own sub-divisions. Some of the more astonishing Burgess animals do not even fit into large modern categories. They are recognisable as animals – but not as arthropods, or annelids, or chordates, or any other such group. They belong in major groups of their own, groups no longer represented in our oceans.
Gould’s new book is a bestiary of these weird animals. They are most singular and delightful. Take Opabinia as an example. It was an inch or two in length. Walcott thought it was a primitive arthropod, with repeated segments and a limb at the front. He guessed the limb might have been analogous to the cephalic horns of modern male brine shrimps, which are used to grasp the female. Indeed, in the 1930s the eminent ecologist Evelyn Hutchinson reconstructed Opabinia as a species of brine shrimp. But Whittington’s Opabinia is not a brine shrimp: it is not even an arthropod. It lacks the characteristic structure of an arthropod – jointed limbs. Whittington reconstructed the anterior ‘limb’ as a flexible nozzle with claws on the end – rather like an elephant’s trunk with pincers, and possibly used more like a vacuum-cleaner than a harpoon. The rest of the animal was a sort of cross between a worm and a shrimp; it had no legs, and its gills were attached laterally on each body segment. It also had a gut with a U-turn in it, and five eyes (four of them on stalks). Opabinia is no longer a primitive arthropod ancestor, but as Gould puts it, an animal ‘to grace the set of a Science Fiction film’.
Opabinia was only the beginning in Whittington’s quest. Consider Wiwaxia next. In the illustration it is like – well, it is so surreal that everyone will come up with a different image – a discus covered with hundreds of nut shells, and with two rows of bent knife blades bristling along its back. ‘About 5 mm from the front end, Conway Morris found two arc-shaped bars, each carrying a row of simple, conical teeth directed toward the rear. The front bar bears a notch at the centre, marking a toothless area between the two side regions, each with seven or eight teeth.’ Wiwaxia would not have been a nice pet.
The strangest animal in the fauna is the aptly named Hallucigenia. Conway Morris reconstructed it as a beast with seven pairs of stiff legs beneath a cylindrical body and an up-turned tail end. The head, or supposed head, is a mere blob. The strangest parts are the seven tentacles arranged in a row down the back, each of which may have had its own mouth. Hallucigenia is so strange that it may not have been an animal at all. Gould suggests that it may have been a broken-off limb from a larger animal: this would explain, among other things, why Hallucigenia’s ‘head’ is so unstructured; it may just be the break-point. But the Burgess Shale’s rich deposits have as yet revealed no hint of the rest of the body.
Even the animals that do fit into modern groups have their own peculiarities. Odaraia, for instance, was a perfectly good arthropod: it had proper arthropod legs and a bivalved carapace like a prawn. However, Odaraia also ‘bears a three-pronged tail, with two lateral flukes and one dorsal projection – a bizarre structure that evokes images of sharks or whales, rather than lobsters’. Then there was Leanchoilia – another arthropod of some kind, looking like a large woodlouse: but it also had two extra large limbs at the front, bearing three-tailed lashes at the tip. And Sanctacaris, in Gould’s words, ‘has a bulbous head-shield, wider than long and extending laterally as a flat, triangular projection’. The head shield alone would serve as a helmet design for one of those films mixing medieval barbarity and futuristic gadgets, but the animal itself gets its name from an even more arresting feature: two mouthparts, designed as large crushing scoops which were presumably powered by muscles anchored in that ‘bulbous head shield’.
Sanctacaris was not the Burgess Shale’s most frightening creature. Anomalocaris carried armaments on an altogether more dangerous scale. It is also the most remarkable of Whittington’s reconstructions. In its present form it unites parts which had previously been classified as four separate animals. Like Sanctacaris, it has two dreadful mouthparts, resembling some engine out of a Dark Age torture-chamber: but these mouthparts had originally been dislocated and reconstructed as ‘appendage-F’. Walcott had described what is now the body of Anomalocaris as a sea cucumber and (most amazingly) its mouth was until recently thought to be a species of jellyfish; it was illustrated as such, swimming in the water column, in a picture in Scientific American in 1979. As Gould says, ‘whoever dreamed about a connection between the rear end of a shrimp, the feeding appendage of Sidneyia [i.e. appendage-F] a squashed sea cucumber, and a jellyfish with a hole in the centre? But such is Anomalocaris, a menacing, seven-inch-long predator. It had crushing teeth in the centre of its mouth (the former supposed jellyfish) as well as grasping limbs; it also had two large bulbous eyes and tail flaps enabling it to hover like a skate or ray. Anomalocaris is another Burgess Shale animal which, like Wiwaxia, Hallucigenia and Opabinia, has no relationship with any modern animal groups.
Wonderful Life is a superb piece of scientific popularisation. The original descriptive monographs are dry and professional, and would hardly have been accessible to Gould’s scientific colleagues, let alone the public, without his lively and enthusiastic account. But he aims to do more than just describe the content, and discovery, of a remarkable fauna. He has two further purposes. One is to establish a general point about the nature of history – a point which I think all biologists would agree with. The other is to make a factual claim about the diversity of the Burgess Shale fauna – a claim which I suspect is a mistake.
The point about history is this. When we look at the Burgess Shale, some of the species (like Sanctacaris) have modern relatives; many others (like Wiwaxia) do not. The fauna must have been ‘decimated’ (the term Gould prefers) by extinction and only a few of its members survived to populate the Earth. It was probably mainly a matter of luck which groups survived and which went extinct. The groups (including Chordates and, by descent, ourselves) which happen to be alive today are probably the fortunate survivors of a series of contingent accidents. We are not the climax of a predictable, progressive evolutionary drama.
In the philosophy of human history, this is a familiar distinction. In the works of Marx, Spengler, Toynbee (and many now less well-known writers), history steers an inevitable course, in which individual accidents hardly matter. Liberal historians, by contrast, have seen more influence in accidents. There are marvellous examples in Gibbon. The Moslem invasion of Europe, for instance, was checked, and reversed, by the battle of Poitiers in 732. But for that battle, Gibbon suggested, things could have turned out very differently. At Poitiers, the Arabs had already advanced a thousand miles from Gibraltar, and, in Gibbon’s words, ‘the repetition of an equal space would have carried the Saracens to the confines of Poland and the highlands of Scotland; the Rhine is not more impassable than the Nile or Euphrates, and the Arabian fleet might have sailed without a conflict into the mouth of the Thames. Perhaps the interpretation of the Koran would now be taught in the schools of Oxford, and her pulpits might demonstrate to a circumcised people the sanctity and truth of the revelations of Mohammed.’
Gould describes, in a rather different idiom, several analogous accidents in the history of life. The origin of modern humans, Gould argues, required the Homo erectus of Europe and Asia to go extinct and leave the future to the Homo erectus of Africa; before that, the accidental extinction of the dinosaurs had to leave the Earth to a (at that time) side-lined group called the mammals; another accident had to send a fish out onto the land to produce the terrestrial vertebrates, and so on back to the origin of life. Indeed, it is a notorious result of certain test-tube experiments, designed to simulate the conditions of the earliest life forms, that molecular evolution simply comes to a stop with a nucleic acid only a few units long. There is nothing inevitable about the evolution of complex living things, let alone of us.
Evolutionary biologists would almost all agree with Gould that accidents have had a strong influence in the history of life. Many would agree with his other general claim: namely, that ‘with far fewer species, the Burgess Shale – one quarry in British Columbia, no longer than a city block – contains a disparity in anatomical design far exceeding the modern range throughout the world.’ This (constantly reiterated) theme is the most original idea in the book: the anatomical reconstructions were done by others, and the idea of historical contingency, rather than inevitability, is one of the oldest Darwinian chestnuts. Perhaps because Wonderful Life is a popular book, Gould provides no formal evidence that the Burgess Shale had a relatively high diversity, but his reasoning is clear enough. The passage quoted above talked about disparity in anatomical design. This is a subjective concept, however: a modern coral reef contains a great variety of forms, and it is anyone’s (or rather, no one’s) guess whether the range of form between Wiwaxia and Anomalocaris in the Burgess Shale is higher or lower than that between, say, a shark and a brain coral on the Great Barrier Reef today. The comparison is meaningless.
Gould does also have a source of more objective evidence. At the lowest level of the Linnaean hierarchy there are species, and above them there are the successively more inclusive categories of genera, families, orders, classes, phyla. Humans, for instance, belong to the species Homo sapiens; it and a number of other species are included in the genus Homo; Homo and a number of other genera are combined in the family Hominidae; the Hominidae and other families of monkeys make up the order Primata; which in turn is part of the Class Mammalia, and the phylum Chordata. Now, Gould suggests that the Burgess Shale fauna shows a relatively high diversity because it contained representatives of more of the higher taxonomic groups than would a sample of the same number of organisms from the modern ocean: the Burgess Shale fauna contains more phyla, and more classes within phyla, than would an equivalent modern sample of species. The Burgess Shale is more diverse in the same sense that a sample of five organisms made up of a lizard, a frog, a bison, a porcupine fish and a parrot would be more diverse than a sample containing five different apes.
In formal terms, a relatively high number of phyla and classes has indeed been found in the Burgess Shale. The fact is not in doubt: only its meaning. I believe it is a taxonomic artifact. An increasing diversity at higher taxonomic levels will arise automatically if you take the modern definitions of higher categories and apply them retrospectively. For example, all modern Crustacea have two pairs of antennae: the character ‘two pairs of antennae’ therefore defines a higher category – a Class, in fact. In modern life, the character is highly significant. It defines a group of over 25,000 species. But now imagine tracing the crustacean evolutionary lineage back through time. Twenty-five, fifty or even 250 million years ago there may have been more (or there may have been fewer) crustacean species than now: but if we go back far enough we must eventually come to a time when there were only a few crustaceans. At some point there will be only one crustacean species. Imagine that, at that time, there were, in some inter-tidal rock-pool, three crustacean-like arthropod species: one with two pairs of antennae, one with four pairs and one with six pairs; in all other respects (let us imagine) they were very similar. The forms with four and six pairs of antennae then went extinct and those with two pairs gave rise to the 25,000 species of modern Crustacea.
What happens if we apply the modern definition of the class Crustacea to this ancestral rock-pool? We find at least two, and probably three whole classes of arthropods in it. If we are Stephen Gould, we should duly conclude that this one rock pool had a diversity of crustacean-like arthropods ‘not matched today by all the creatures in all the world’s oceans’, ‘exceeding the range throughout the modern world’, etc. The diversity, however, is artifactual. The three ‘classes’ could show no more diversity than three modern species of shrimp. If we did not know that two pairs of antennae was destined to become a highly significant character, we should not attach any great importance to the difference between the arthropods with two pairs and those with other numbers. But if we use the modern definitions, we have to say that the arthropods with two pairs of antennae are in one class and the others in another class (or classes): we then apparently find great diversity at high taxonomic levels.
Today, two pairs of antennae just happens to be a reliable indicator of a major branch in the tree of life. This is a mere evolutionary fluke: in a million years time, a crustacean may evolve with one, or three, pairs of antennae, and the successful Crustacea of the future may be descended from a three-antennaed, rather than a two-antennaed shrimp. In 530 million years time, the character ‘two pairs of antennae’ may define only an obscure group of arthropods. Likewise, 530 million years in the past, in the Burgess Shale, this character did not define a big group of species. To recognise them as a major group because they possess a character that defines a modern class is an anachronism. Their status as a higher taxon in the Burgess Shale stands only on their future importance. The diversity of the Burgess Shale fauna at higher levels is therefore quite spurious. If you apply the characters used to define modern higher taxa backwards through the tree of life, you inevitably find more higher taxa earlier in time.
I can make the same point in another way. Imagine a zoologist who has been magically educated in the principles of Linnaean classification, but who knows nothing about the particular array of modern living species. We transport him 530 million years into the past and set him to work. Being a follower of Linnaeus, he will classify the animals into species and genera, classes and phyla. But the major groups of modern classifications will be reduced in taxonomic level. The class Crustacea may be no more than the genus Biantennus within the order Arthropoda. What Gould needs to show is that this time-transported taxonomist would find a disproportionate number of phyla and classes relative to species and genera. The work has not been done, but Gould’s naive finding of a high diversity at high taxonomic levels is exactly what would be predicted if a sample of 80,000 specimens from the Burgess Shale had a similar range of real diversity as 80,000 specimens plucked today from the Great Barrier Reef.