This is apparent from his
statement
that 'the Cambrian explosion exhibited another novelty', and from his phrase long-jump mutations'.
Richard-Dawkins-Unweaving-the-Rainbow
Herbert Spencer, who, to be fair, was the first to use the word evolution in a technical sense, wanted to regard biological evolution as only a special case.
Evolution, for him, was a much more general process, with shared laws at all its levels.
Other manifestations of the same general law of evolution were the development of the individual (the progression from fertilized egg through foetus to adult)- the development of the cosmos, the stars and the planets from simpler beginnings; and progressive changes, over historical time, in social phenomena such as the arts, technology and language.
There are good things and bad about the poetry of general evolutionism. On balance I think it fosters confusion more than illumination, but there is certainly some of both. The analogy between embryonic development and species evolution was artfully exploited by that irascible genius J. B. S. Haldane to make a debating point. When a sceptic of evolution doubted that anything so complicated as a human could have come from single-celled beginnings, Haldane promptly observed that the sceptic himself had done that very thing and the whole process took only nine months. Haldane's rhetorical point is undiminished by the fact, which of course he knew perfectly well, that development is not the same thing as evolution. Development is change in the form of a single object, as clay deforms under a potter's hands. Evolution, as seen in fossils taken from successive strata, is more like a sequence of frames in a cinema film. One frame doesn't literally change into the next, but we experience an illusion of change if we project the frames in succession. With this distinction in place, we can quickly see that the cosmos does not evolve (it develops) but technology does evolve (early aeroplanes are not moulded into later ones but the history of aeroplanes, and of many other
pieces of technology, falls well into the cinema frame analogy). Clothes fashions, too, evolve rather than develop. It is controversial whether the analogy between genetic evolution, on the one hand, and cultural or technical evolution, on the other, leads to illumination or the reverse, and I am not going to get into that argument now.
My remaining examples of bad poetry in evolutionary science come largely from a single author, the American palaeontologist and essayist Stephen Jay Gould. I am anxious that such critical concentration upon one individual shall not be taken as personally rancorous. On the contrary, it is Gould's excellence as a writer that makes his errors, when they occur, so eminently worth rebutting.
In 1977 Gould wrote a chapter on 'eternal metaphors of palaeontology' to introduce a multi-authored book on the evolutionary study of fossils. Beginning with Whitehead's preposterous, though much quoted, statement that all of philosophy is a footnote to Plato, Gould's thesis, in the words of the preacher of Ecclesiastes (whom he also quotes), is that there is nothing new under the sun; 'The thing that hath been, it is that which shall be; and that which is done is that which shall be done. ' Current controversies in palaeontology are just old controversies being recycled. They preceded evolutionary thought and found no resolution within the Darwinian paradigm . . . Basic ideas, like idealized geometric figures, are few in number. They are eternally available for
consumption . . .
Gould's eternally unresolved questions in palaeontology are three in number: Does time have a directional arrow? Is the driving motor of evolution internal or external? Does evolution proceed gradually or in sudden jumps? Historically, he finds examples of palaeontologists who have espoused all eight possible combinations of answers to these three questions, and he satisfies himself that they straddle the Darwinian revolution as though it never happened. But he manages this feat only by forcing analogies between schools of thought which, carefully examined, have no more in common than blood and wine, or helical orbits and helical DNA. All three of Gould's eternal metaphors are bad poetry, forced analogies that obscure rather than illuminate. And bad poetry in his hands is all the more damaging because Gould is a graceful writer.
The question whether evolution has a directional arrow is certainly one that can sensibly be asked, in various guises. But the bedfellows that the different guises bring together are so ill matched that they are not usefully united. Does bodily structure get progressively more complex as evolution goes on? This is a reasonable question. So is the question of whether the total diversity of species on the planet increases progressively as the ages go by. But they are utterly different questions
and it is conspicuously unhelpful to invent a century-spanning school of 'progressivist' thought to unite them. Still less do either of them, in their modern form, have anything in common with the pre-Darwinian schools of 'vitalism' and 'finalism', which held that living things were progressively 'driven' from within, by some mystical life force, towards an equally mystical final goal. Gould forces unnatural connections among all these forms of progressivism, as a device to support his poetic historical thesis.
Much the same is true of the second eternal metaphor, and the question of whether the motor of change is in the external environment, or whether change arises from 'some independent and internal dynamic within organisms themselves'. A prominent modern disagreement is between those who believe that the main driving force of evolution is Darwinian natural selection and those who emphasize other forces such as random genetic drift. This important distinction is not conveyed, not even to the smallest extent, by the internalist/externalist dichotomy that Gould would force upon us in order to maintain his thesis that post- Darwinian argumentation is just a recycling of pre-Darwinian equivalents. Is natural selection externalist or internalist? It depends whether you are talking about adaptation to the external environment or co-adaptation of the parts to each other. I shall return to this distinction later in another context.
Bad poetry is even more evident in Gould's exposition of the third of his eternal metaphors, the one concerning gradual versus episodic evolution. Gould uses the word episodic to unite three kinds of sharp discontinuity in evolution. These are: first, catastrophes such as the mass extinction of the dinosaurs; second, macromutations or saltations; and third, punctuation in the sense of the theory of punctuated equilibrium proposed by Gould and his colleague Niles Eldredge in 1972. This last theory needs more explanation, and I'll come to it in a moment.
Catastrophic extinctions are straightforward to define. Exactly what causes them is controversial and probably different in different cases. For the moment, just notice that a worldwide catastrophe in which most species die is, to put it mildly, not the same thing as a macromutation. Mutations are random errors in gene copying and macromutations are mutations of large effect. A mutation of small effect, or micromutation, is a small error in gene copying, whose effect on its possessors might be too slight to notice easily, say a subtle lengthening of a leg bone, or a hint of reddening in a feather. A macromutation is a dramatic error, a change so large that, in extreme cases, its possessor would be classified in a different species from its parents. In my previous book, Climbing Mount Improbable, I reproduced a photograph from a newspaper of a toad with eyes in the roof of its mouth. If this photograph is genuine (a big if, in
these days of Photoshop and other handy image-manipulation software), and if the error is genetic, the toad is a macromutant. If such a macromutant spawned a new species of toads with eyes in the roofs of their mouths, we should describe the abrupt evolutionary origin of the new species as a saltation or evolutionary jump. There have been biologists, such as the German/American geneticist Richard Goldschmidt, who believed that such saltatory steps were important in natural evolution. I am one of many who have cast doubt on the general idea, but that is not my purpose here. Here I make the much more basic point that such genetic leaps, even if they occur, have nothing in common with earth-shattering catastrophes such as the sudden extinction of the dinosaurs, except that both are sudden. The analogy is purely poetic, and it is bad poetry which doesn't lead to any further illumination. Recalling Medawar's words, the analogy marks the end, not the inauguration, of a train of thought. The ways of being a non- gradualist are so varied as to strip the category of all usefulness.
The same applies to the third category of non-gradualists:
punctuationists in the sense of Eldredge and Gould's theory. The idea here is that a species comes into existence in a time which is short compared with the much longer period of 'stasis' during which it survives unchanged after its initial formation. In the extreme version of the theory, the species, once it has burst into existence, continues unchanged until either it goes extinct or it splits to form a new daughter species. It is
when we ask what happens during the sudden bursts of species formation that the confusion, born of bad poetry, arises. There are two things that might happen. They are utterly different from each other, but Gould makes light of the difference because he is seduced by bad poetry. One is macromutation. The new species is founded by a freak individual, like the alleged toad with eyes in the roof of its mouth. The other thing that might happen - more plausibly, in my view, but I'm not talking
about that now - is what we can call rapid gradualism. The new species comes into existence in a brief episode of rapid evolutionary change which, although gradual in the sense that parents don't spawn an
instant new species in a single generation, is fast enough to look like an instant in the fossil record. The change is spread over many generations of small, step-by-step increments, but it looks like a sudden jump. This
is either because the intermediates lived in a different place (say, on an outlying island) and/or because the intermediate stages passed too rapidly to fossilize - 10,000 years is too short to measure in many geological strata, yet it constitutes ample time for quite major evolutionary change to accumulate gradually in small steps.
There is all the difference in the world between rapid gradualism and macromutational saltation. They depend upon totally different mechanisms and they have radically different implications for Darwinian
controversies. To lump them together simply because, like catastrophic extinctions, they all lead to discontinuities in the fossil record, is bad poetic science. Gould is aware of the difference between rapid gradualism and macromutation, but he treats the matter as though it were a minor detail, to be cleared up after we have taken on board the overarching question of whether evolution is episodic rather than gradual. One can see it as overarching only if one is intoxicated by bad poetry. It makes as little sense as my correspondent's question about the DNA double helix and whether it 'comes from' the earth's orbit. Once again, rapid gradualism no more resembles macromutation than a bleeding wizard resembles a shower of rain.
Even worse is to claim catastrophism under the same punctuationist umbrella. In pre-Darwinian times the existence of fossils became increasingly embarrassing for upholders of biblical creation. Some hoped to drown the problem in Noah's flood, but why did the strata seem to show dramatic replacements of whole faunas, each one different from its predecessor, and all of them largely free of our own, familiar creatures? The answer given by, among others, the nineteenth-century French anatomist Baron Cuvier, was catastrophism. Noah's flood was only the last in a series of cleansing disasters visited upon the earth by a supernatural power. Each catastrophe was followed by a new creation.
Apart from the supernatural intervention, this has something - a little - in common with our modern belief that mass extinctions such as those that ended the Permian and Cretaceous eras were followed by new flowerings of evolutionary diversity to match previous radiations. But to lump the catastrophists in with macromutationists and with modern punctuationists, just because all three can be represented as non- gradualist, is very' bad poetry indeed.
After giving lectures in the United States, I have often been puzzled by a certain pattern of questioning from the audience. The questioner calls my attention to the phenomenon of mass extinction, say, the catastrophic end of the dinosaurs and their succession by the mammals. This interests me greatly and I warm to what promises to be a stimulating question. Then I realize that the tone of the question is unmistakably challenging. It is almost as though the questioner expects me to be surprised, or discomfited, by the fact that evolution is periodically interrupted by catastrophic mass extinctions. I was baffled by this until the truth suddenly hit me. Of course! The questioner, like many people in North America, has learned his evolution from Gould, and I have been billed as one of those 'ultra-Darwinian' gradualists. Doesn't the comet that killed the dinosaurs also blow my gradualistic view of evolution out of the water? No, of course it doesn't. There is not the smallest connection. I am a gradualist in the sense that I don't think
macromutations have played an important role in evolution. More determinedly, I am a gradualist when it comes to explaining the evolution of complex adaptations like eyes (so is any sane person, including Gould). But what on earth have such matters got to do with mass extinctions? Nothing at all. Unless, that is, your mind has been filled up with bad poetry. For the record, I believe, and have believed for the whole of my career, that mass extinctions exert a profound and dramatic influence on the subsequent course of evolutionary history. How could they not? But mass extinctions are not a part of the Darwinian process, except in so far as they clear the decks for new Darwinian beginnings.
There is irony lurking here. Among the facts about extinction that Gould is fond of emphasizing is its capriciousness. He calls it contingency.
When mass extinction strikes, major groups of animals die wholesale. In the Cretaceous extinction, the once mighty group of dinosaurs (with the notable exception of birds) was completely wiped out. The choice of major group for victim is either random or, if non-random, it is not the same non-randomness as we see in conventional natural selection. The normal adaptations to survival do not avail against comets. Grotesquely, this fact is sometimes trotted out as if it were a debating point against neo- Darwinism. But neo-Darwinian natural selection is selection within species, not between species. To be sure, natural selection involves death, and mass extinction involves death, but any further resemblance
between the two is purely poetic. Ironically, Gould is one of the few Darwinians who still think of natural selection as working at levels higher than the individual organism. It would never occur to the rest of us even to ask whether mass extinctions are selective events. We might see extinction as opening up new opportunities for adaptation, by lower- level natural selection choosing between individuals separately within each species that has survived the catastrophe. As a further irony, it is the poet Auden who came nearer to getting it right:
But catastrophes only encouraged experiment.
As a rule, it was the fittest who perished, the mis-fits, forced by failure to emigrate to unsettled niches, who altered their structure and prospered.
'Unpredictable but Providential (for Loren Eiseley)'
I take one further extended example of bad poetic science from paleontology, and once again Stephen Jay Gould is responsible for its popularity even if he has not clearly expressed it himself in its extreme form. Many readers of his elegantly written book Wonderful Life (1989) have been captivated by the idea that there was something special and unique about the whole business of evolution in the Cambrian era, when fossils of most of the great animal groups first appeared, rather over 500 million years ago. It is not just that the animals of the Cambrian were
peculiar. Of course they were. The animals of every era have their peculiarities and the Cambrian ones were arguably more peculiar than most. No, the suggestion is that the whole process of evolution in the Cambrian was odd.
The standard neo-Darwinian view of the evolution of diversity is that a species splits into two when two populations become sufficiently unalike that they can no longer interbreed. Often the populations begin diverging when they chance to be geographically separated. The separation means that they no longer mix their genes sexually and this permits them to evolve in different directions. The divergent evolution might be driven by natural selection (which is likely to push in different directions because of different conditions in the two geographical areas). Or it might consist of random evolutionary drift (since the two populations are not genetically held together by sexual mixing, there is nothing to stop them drifting apart). In either case, when they have evolved sufficiently far apart that they could no longer interbreed even if they were geographically united again, they are defined as belonging to separate species.
Subsequently, the lack of interbreeding permits further evolutionary divergence. What had been distinct species within one genus become, in the fullness of time, distinct genera within one family. Later, families will be found to have diverged to the point where taxonomists (specialists in classification) prefer to call them orders, then classes, then phyla. Phylum (plural phyla) is the classificatory name by which we distinguish really fundamentally different animals like molluscs, nematode worms, echinoderms and chordates (chordates are mostly vertebrates plus a few odds and ends). Ancestors of two different phyla, say vertebrates and molluscs, which we see as built upon utterly different 'fundamental body plans' were once just two species within a genus. Before that, they were two geographically separated populations within one ancestral species. The implication of this widely accepted view is that, as you go back and back in geological time, the gap between any pair of animal groups becomes smaller and smaller. The further back in time you go, the closer you approach the uniting of these different kinds of animals in their single common ancestor species. Our ancestors and mollusc ancestors were once very alike. Later they were not quite so alike. Later again they had diverged further, and so on until eventually they became so different that we should call them two phyla. This general story can scarcely be doubted by any reasonable person who thinks it through, though we do not have to be committed to the view that it occurs at a uniform rate with time. It could have happened in rapid bursts.
The dramatic phrase 'Cambrian explosion' is used in two senses. It can refer to the factual observation that before the Cambrian era, just over half a billion years ago, there are few fossils. Most of the great animal
phyla appear as fossils for the first time in Cambrian rocks, and this looks like a great explosion of new animals. The second meaning is the theory that the phyla actually branched off from each other during the Cambrian, even during as little as 10 million years within the Cambrian. This second idea, which I shall call the branch point explosion hypothesis, is controversial. It is compatible - just - with what I am calling the standard neo-Darwinian model of species divergence. We've already agreed that, as we trace any pair of modem phyla back in time, we eventually converge upon a common ancestor. My hunch is that, for different pairs of phyla, we'll hit the common ancestor in different geological eras: say, the common ancestor of vertebrates and molluscs at 8oo million years ago, the common ancestor of vertebrates and echinoderms at 600 million years, and so on. But I could be wrong, and we can easily accommodate the branch point explosion hypothesis by saying that, for some reason (which is interesting enough to need investigating), most of our backward tracings happen to hit their respective common ancestors during the same relatively short geological period, say, between 540 million and 530 million years ago. This would have to mean that, at least near the beginning of that 10-million-year period, the ancestors of the modem phyla were nowhere near as different from each other as they are today. They were, after all, diverging from common ancestors at the time and were originally members of the same species.
The extreme Gouldian view - certainly the view inspired by his rhetoric, though it is hard to tell from his own words whether he literally holds it himself - is radically different from and utterly incompatible with the standard neo-Darwinian model. It also, as I shall show, has implications which, once they are spelled out, anybody can see are absurd. It is very clearly expressed - betrayed might be a better word - in asides in Stuart Kauffman's At Home in the Universe (1995):
One might imagine that the first multicellular creatures would all he very similar, only later diversifying, from the bottom up, into different genera, families, orders, classes, and so on. That, indeed, would be the expectation of the strictest conventional Darwinist. Darwin, profoundly influenced by the emerging view of geologic gradualism, proposed that all evolution occurred by the very gradual accumulation of useful variations. Thus the earliest multicellular creatures themselves ought to have diverged gradually from one another.
So far, this is a fine summary of the orthodox neo-Darwinian view. Now, in a bizarre passage, Kauffman goes on:
But this appears to be false. One of the wonderful and puzzling features of the Cambrian explosion is that the chart was filled in from the top
down. Nature suddenly sprang forth with many wildly different body plans - the phyla - elaborating on these basic designs to form the classes, orders, families, and genera . . . In his book about the Cambrian explosion, Wonderful Life: The Burgess Shale and the Nature of History, Stephen Jay Gould remarks on this top-down quality of the Cambrian with wonder.
As well he might! You only have to think for one moment about what 'top down' filling in would have to mean for the animals on the ground and you immediately see how preposterous it is, 'Body plans' like the mollusc body plan, or the echinoderm body plan, are not ideal essences hanging in the sky, waiting, like designer dresses, to be adopted by real animals. Real animals is all there ever was: living, breathing, walking, eating, excreting, fighting, copulating real animals, who had to survive and who can't have been dramatically different from their real parents and grandparents. For a new body plan - a new phylum - to spring into existence, what actually has to happen on the ground is that a child is born which suddenly, out of the blue, is as different from its parents as a snail is from an earthworm. No zoologist who thinks through the implications, not even the most ardent saltationist, has ever supported any such notion. Ardent saltationists have been content to postulate the sudden bursting into existence of new species, and even that relatively modest idea has been highly controversial. When you spell out the Gouldian rhetoric into real-life practicalities, it stands revealed as the purest of bad poetic science.
Kauffman is even more explicit in a later chapter. In discussing some of his ingenious mathematical models of evolution on 'rugged fitness landscapes', Kauffman notes a pattern that he thinks sounds a lot like the Cambrian explosion. Early on in the branching process, we find a variety of long-jump mutations that differ from the stem and from one another quite dramatically. These species have sufficient morphological differences to be categorized as founders of distinct phyla. These founders also branch, but do so via slightly closer long-jump variants, yielding branches from, each founder of a phylum to dissimilar daughter species, the founders of classes. As the process continues, fitter variants are found in progressively more nearby neighborhoods, so founders of orders, families, and genera emerge in succession.
Kauffman's earlier, more technical book, The Origins of Order (1993), says something similar about life in the Cambrian:
Not only did a very large number of novel body forms originate rapidly, but the Cambrian explosion exhibited another novelty: Species which founded taxa appear to have built up the higher taxa from the top down.
That is, exemplars of major phyla were present first, followed by progressive filling in at class, order, and lower taxonomic levels. . .
Now, one way of reading this is inoffensive to the point of obviousness. On our 'converging backwards' model it would have to be true that species splittings that are eventually going to become phylum divides would normally precede those that are destined to become divides between orders and lower taxonomic levels. But Kauffman clearly doesn't think he is saying something ordinary and obvious.
This is apparent from his statement that 'the Cambrian explosion exhibited another novelty', and from his phrase long-jump mutations'. He thinks he is attributing to the Cambrian something revolutionary. He really does seem genuinely to intend the alternative reading, in which 'long-jump mutations' give rise, on the instant, to brand new phyla.
I hasten to emphasize that these passages of Kauffman's are embedded in a pair of books that are for the most part interesting, creative and uninfluenced by Gould. The same is true of Richard Leakey and Roger Lewin's The Sixth Extinction (1996), another recent book, admirable in most of its chapters, but sadly marred by one, 'The Mainspring of Evolution', which is explicitly and avowedly influenced by Gould. Here are a couple of relevant passages:
It was as if the facility for making evolutionary leaps that produced major functional novelties - the basis of new phyla - had somehow been lost when the Cambrian period came to an end. It was as if the mainspring of evolution had lost some of its power.
Hence, evolution in Cambrian organisms could take bigger leaps, including phylum-level leaps, while later on it would be more constrained, making only modest jumps, up to the class level.
As I have written before, it is as though a gardener looked at an old oak tree and remarked, wonderingly: 'Isn't it strange that no major new boughs have appeared on this tree for many years. These days, all the new growth appears to be at the twig level! ' Just think once again what a 'phylum-level leap' or even a 'modest' {modest? ) class level leap would have to mean. Animals of different phyla, remember, are animals with different fundamental body plans, like molluscs and vertebrates. Or like starfish and insects. A long-jump, phylum level mutation would have to mean that a couple of parents belonging to one phylum mated and gave birth to a child belonging to a different phylum. The difference between parent and offspring would have to be on the same scale as the difference between a snail and a lobster, or a starfish and a codfish. A class level leap would be equivalent to a pair of birds giving birth to a mammal.
Picture the parents gazing wonderingly into the nest at what they have produced, and the full comedy of the notion becomes apparent.
My assurance in ridiculing these ideas is not based simply upon knowledge of the facts of modern animals. Obviously if it were just that, one could retort that things were different in the Cambrian. No, the argument against Kaufman's long jumps, or Leakey and Lewin's phylum level leaps, is a theoretical one, and an extremely strong one. It is this. Even if mutations on this gigantic scale occurred, the products would not have survived. This is fundamentally because, as I have said before, however many ways there may be of being alive, there are almost infinitely more ways of being dead. A small mutation, representing a minor step away from a parent which has proved its ability to survive by virtue of being a parent, has a good chance of surviving for the same reason, and it may even be an improvement. A gigantic, phylum level mutation is a leap into the wild blue yonder. I said that the long-jump mutation we are talking about would be of the same magnitude as a mutation from a mollusc to an insect. But it would never, of course, have been a jump from a mollusc to an insect. An insect is a highly tuned piece of survival machinery. If a mollusc parent gave birth to a new phylum, the leap would have been a random leap, like any other mutation. And the chance that a random leap of that magnitude would produce an insect, or anything with the faintest chance of surviving, is small enough to be discounted totally. The chance of its being viable is impossibly small, no matter how empty the ecosystem, how wide open the niches. A phylum level leap would be a mess.
I do not believe the authors I am quoting really believe what their printed words undoubtedly appear to be saying. I think they were simply intoxicated by Gould's rhetoric and didn't think it through. The whole point of quoting them in this chapter is to illustrate the power to mislead that a skilled poet can unwittingly exert, especially if he has first misled himself And the poetry of the Cambrian as a blissful dawn of innovation is undoubtedly beguiling. Kauffman gets completely carried away by it:
Soon after multicelled forms were invented, a grand burst of evolutionary novelty thrust itself outward. One almost gets the sense of multicellular life gleefully trying out all its possible ramifications, in a kind of wild dance of heedless exploration
At Home in the Universe (1995)
Yes. One does get exactly that sense. But one gets it from Gould's rhetoric, not from the facts of Cambrian fossils nor from sober reasoning about evolutionary principles.
If scientists of the calibre of Kauffman, Leakey and Lewin can be seduced by bad poetic science, what chance has the non-specialist?
Daniel Dennett has told me of a conversation with a philosopher colleague who had read Wonderful Lifers arguing that the Cambrian phyla did not have a common ancestor - that they had sprung up as independent origins of life! When Dennett assured him that this was not Gould's intention, his colleague's response was, 'Well then, what is all the fuss about? '
Excellence in writing is a double-edged sword, as the distinguished evolutionary scientist John Maynard Smith has noted, in the New York Review of Books, November 1995:
Gould occupies a rather curious position, particularly on his side of the Atlantic. Because of the excellence of his essays, he has come to be seen by non-biologists as the pre-eminent evolutionary theorist. In contrast, the evolutionary biologists with whom I have discussed his work tend to see him as a man whose ideas are so confused as to be hardly worth bothering with, but as one who should not be publicly criticized because he is at least on our side against the creationists. All this would not matter, were it not that he is giving non-biologists a largely false picture of the state of evolutionary theory.
Maynard Smith was reviewing Dennett's book Darwin's Dangerous Idea (1995), which contains a devastating and, one might hope, terminal critique of Gould's influence on evolutionary thinking.
What really happened in the Cambrian? Simon Conway Morris of Cambridge University is, as Gould fulsomely acknowledges, one of the three leading modern investigators of the Burgess Shale, the Cambrian fossil bed which is the subject of Wonderful Life. Conway Morris has recently published his own fascinating book on the subject. The Crucible of Creation (1998), which is critical of almost every aspect of Gould's view. Like Conway Morris, I don't think there's any good reason to think that the process of evolution was different in the Cambrian from the way it is today. But there is no doubt that a large number of major animal groups are seen in the fossil record for the first time in the Cambrian. The obvious hypothesis has occurred to many people. Perhaps several groups of animals evolved hard, fossilizable skeletons around the same time and perhaps for the same reason. One possibility is an evolutionary arms
race between predators and prey, but there are other ideas like a dramatic change in the chemistry of the atmosphere. Conway Morris finds no support at all for the poetic idea of an exuberant and
extravagant flowering of life in a wild dance of Cambrian diversity and disparity, subsequently pruned to today's more limited repertoire of
animal types. If anything, the reverse seems to be true, as most evolutionists would expect.
Where does that leave the question of the timing of the branch points of the major phyla? Recall that this is a separate question from the undoubted Cambrian explosion of fossil availability. The controversial matter is whether the branch points in the divergence of all the major phyla are concentrated in the Cambrian - the branch point explosion hypothesis. I said standard neo-Darwinism was compatible with this hypothesis. But I still don't think it is at all likely.
One possible way to tackle the question is by looking at molecular clocks. 'Molecular clock' refers to the observation that certain biological molecules change at a rather fixed rate over the millions of years. If you accept this, you can take blood from any two modern animals and calculate how long ago their common ancestor lived. Some recent molecular clock studies have pushed the branch points of various pairs of phyla deep into the Precambrian era. If these studies are right, the whole rhetoric of an evolutionary explosion becomes superfluous. But there is controversy over the interpretation of molecular clock results so far back in deep time, and we should wait for more evidence.
Meanwhile, there is a logical argument which I can assert with more confidence. The only evidence in favour of the branch point explosion hypothesis is negative: there aren't any fossils of many of the phyla before the Cambrian. But those fossil animals that have no fossil ancestors must have had ancestors of some kind. They can't have sprung from nothing. Therefore there must have been ancestors that didn't fossilize, absence of fossils does not mean absence of animals. The only question that remains is whether the missing ancestors going back to the branch points, who must have existed, were all compressed into the Cambrian, or whether they were strung out through the previous hundreds of millions of years. Since the only reason to suppose that they were compressed into the Cambrian is the absence of fossils, and since we have just proved logically the irrelevance of that absence, I conclude that there is no good reason at all to favour the branch point explosion hypothesis. Doubtless it has great poetic appeal.
9
THE SELFISH COOPERATOR
Wonder. . . and not any expectation of advantage from its discoveries, is the first principle which prompts mankind to the study of Philosophy, of that science which pretends to lay open the concealed connections that unite the various appearances of nature.
ADAM SMITH, The History of Astronomy' (1795)
The medieval bestiaries continued an earlier tradition of hijacking nature as a source of moral tales. In its modern form, in the development of evolutionary ideas, the same tradition underlies one of the most egregious forms of bad poetic science. I refer to the illusion that there is a simple opposition between nasty and nice, social and antisocial, selfish and altruistic, tough and gentle; that these pairs of binary opposites all correspond to the other pairs, and that the history of evolutionary controversy about society is described by a pendulum swinging back and forth along a continuum between these opposites. I am not denying that there are interesting issues to be discussed hereabouts. What I am criticizing is the 'poetic' idea that there is a single continuum and that worthwhile arguments are to be had between vantage points along its length. To invoke the rainmakers yet again, there is no more connection between a selfish gene and a selfish human than there is between a rock and a rain cloud.
To explain the poetic continuum I am criticizing, I might as well borrow a line from a real poet, Tennyson's 'Nature, red in tooth and claw', from In Memoriam (1850), widely assumed to be inspired by On the Origin of Species but actually published nine years earlier. At one end of the poetic continuum are supposed to stand Thomas Hobbes, Adam Smith, Charles Darwin, T. H. Huxley and all those, such as the distinguished American evolutionist George C. Williams and today's advocates of 'the selfish gene', who emphasize that nature really is red in tooth and claw. At the other end of the continuum are Prince Peter Kropotkin the Russian anarchist and author of Mutual Aid (1902), the gullible but immensely influential American anthropologist Margaret Mead, and today a spate of authors reacting indignantly to the idea that nature is genetically selfish, of
whom Frans de Waal, author of Good Natured (1996), is representative.
De Waal, a chimpanzee expert who understandably loves his animals, is distressed at what he mistakenly sees as a neo-Darwinian tendency to emphasize the 'nastiness of our apish past'. Some of those who share his romantic fancy have recently become fond of the pygmy chimpanzee or bonobo as a yet more benign role model. Where common chimpanzees often resort to violence, and even cannibalism, bonobos say it with sex. They seem to copulate in all possible combinations at every conceivable opportunity. Where we might shake hands, they copulate. Make love not war is their watchword. Margaret Mead would have warmed to them. But the very idea of taking animals to be role models, as in the bestiaries, is a piece of bad poetic science. Animals are not there to be role models, they are there to survive and reproduce.
Moralistic devotees of the bonobo are apt to compound this error with an outright evolutionary falsehood. Probably because of their powerful 'feelgood factor', bonobos are often claimed as more closely related to us than common chimpanzees are. But this cannot be, so long as we accept, as everybody does, that bonobos and common chimpanzees are more closely related to each other than either is to humans. You need no more than that simple and uncontroversial premise to conclude that bonobos and common chimpanzees are exactly equally closely related to us. They are connected to us via the common ancestor that they share and we don't. Certainly we may resemble one of the two species more than the other in some respects (and very probably resemble the other in other respects), but such comparative judgements absolutely cannot be reflections of differential evolutionary closeness.
* I should explain that Margaret Mead is 'gullible but influential' because a large section of American academic culture enthusiastically adopted her rose-tinted environmentalist theory of human nature which, it later transpired, she had built on a somewhat insecure foundation: systematic misinformation fed her as a joke by two mischievous Samoan girls, during her brief period of fieldwork in their island. She didn't stay in Samoa long enough to learn the language well, unlike her professional nemesis, the Australian anthropologist Derek Freeman, who uncovered the whole story years later in the course of a more detailed study of Samoan life.
De Waal's book is full of anecdotal demonstrations (which should surprise nobody) that animals are sometimes kind to each other, cooperate for mutual good, care for each other's welfare, console each other in distress, share food and do other heartwarmingly good things. The position I have always adopted is that much of animal nature is indeed altruistic, cooperative and even attended by benevolent subjective emotions, but that this follows from, rather than contradicts, selfishness at the genetic level. Animals are sometimes nice and sometimes nasty, since either can suit the self-interest of genes at different times. That is precisely the reason for speaking of 'the selfish gene' rather than, say, 'the selfish chimpanzee'. The opposition that de Waal and others have erected, between biologists who believe human and animal nature is fundamentally selfish, and those who believe it is fundamentally 'good- natured', is a false opposition - bad poetry.
It is now widely understood that altruism at the level of the individual organism can be a means by which the underlying genes maximize their self-interest. However, I don't want to dwell on what I have expounded in earlier books such as The Selfish Gene. What I would now re-emphasize from that book - it has been overlooked by critics who appear to have read it by title only - is the important sense in which genes, though in
one way purely selfish, at the same time enter into cooperative cartels with each other. This is poetic science, if you like, but I hope to show that it is good poetic science which aids understanding rather than impedes it. I shall do the same with other examples in the remaining chapters. The key insight of Darwinism can be expressed in genetic terms. The genes that exist in many copies in the population are the ones that are good at making copies, which also means good at surviving. Surviving where? Surviving in individual bodies in ancestral environments. That means surviving in the environment typical of the species: in a desert for camels, up trees for monkeys, in the deep sea for giant squids, and so on. The reason individual bodies are so good at surviving in their environments is mainly that they have been built by genes that have survived in the same environment for many generations, in the form of copies.
But never mind deserts and ice floes, seas and forests; they are only part of the story. A far more prominent aspect of the ancestral environment in which genes have survived is the other genes with which they have had to share a succession of individual bodies. The genes that survive in camels will, to be sure, include some that are particularly good at surviving in deserts, and they may even be shared with desert rats and desert foxes. But more importantly, successful genes will be those that are good at surviving in an environment consisting of the other genes that are typically found in the species. So, the genes of a species become selected to be good at cooperating with each other. Genetic cooperation, which is good scientific poetry whereas universal cooperation is not, will be the subject of this chapter.
The following fact is often misunderstood. It is not the genes of any given individual that cooperate particularly well together. They have never been together before in that combination, for every genome in a sexually reproducing species is unique (with the usual exception of identical twins). It is the genes of a species at large that cooperate, because they have met before, often, and in the intimately shared environment of the cell, though always in different combinations. What they cooperate at is the business of making individuals of the same general type as the present one. There is no particular reason to expect the genes of any particular individual to be especially good at cooperating with one another when compared with any other genes of the same species. It is largely a matter of accident which particular companions the lottery of sexual reproduction has drawn for them from the gene pool of the species. Individuals with unfavourable combinations of genes tend to die. Individuals with favourable combinations tend to pass those genes on to the future. But it is not the favourable combinations themselves that are passed on in the long term. Sexual reshuffling sees to that. Instead, what are passed on are the genes that tend to be good at forming favourable
combinations with the other genes that the species gene pool has to offer. Over the generations, whatever else the surviving genes may be good at, they will be good at working together with other genes of the species.
For all we know, particular camel genes might be good at cooperating with particular cheetah genes. But they are never called upon to do so. Presumably mammal genes are better at cooperating with other mammal genes than with bird genes. But the speculation must remain hypothetical, because one of the characteristics of life on our planet is that, genetic engineering aside, genes are mixed only within species. We can test watered-down versions of such speculations by looking at hybrids. Hybrids between different species, when they exist at all, often survive less well, or are less fertile, than pure-bred individuals. At least part of the reason for this is incompatibilities between their genes. Species A genes that work well against a genetic background or 'climate' of other Species A genes do not work when transplanted into Species B, and vice versa. Similar effects are sometimes seen when varieties or races within one species hybridize.
I first understood this while listening to lectures by the late E. B. Ford, legendary Oxford aesthete and eccentric founder of the now neglected school of Ecological Geneticists. Most of Ford's research was on wild populations of butterflies and moths. Among these was the Lesser Yellow Underwing moth, Triphaena comes. This moth is normally yellowish brown, but there is a variant called curtisii which is blackish. Curtisii is not found in England at all; however, in Scotland and the Isles curtisii coexists with the normal comes. The curtisii dark colour pattern is nearly completely dominant to the normal comes pattern. 'Dominant to' is a technical term, which is why I can't just say 'dominates'. It means that hybrids between the two look like curtisii even though they bear the genes of both. Ford caught specimens from Barra in the Outer Hebrides, west of Scotland, and from one of the Orkney islands, north of Scotland, as well as from the Scottish mainland itself. Each of the two island forms looks exactly like its opposite number at the other island site, and the dark curtisii gene is dominant on the two islands, as well as on the mainland. Other evidence shows that the curtisii gene is the very same gene in all localities. In view of this you'd expect that, when you cross- bred specimens from different islands, the normal dominance pattern would hold up. But it doesn't, and this is the point of the story. Ford caught individuals from Barra and mated them with individuals from Orkney. And the dominance of curtisii completely disappeared. A full range of intermediates turned up in the hybrid families, just as if there was no dominance. What seems to be going on is this. The curtisii gene does not, in itself, encode the formula for the coloured pigment by which we distinguish the moths, nor is dominance ever a property of a gene on its own. Instead, like any other gene, the curtisii gene should be thought
of as having its effects only in the context of a suite of other genes, some of which it 'switches on'. This suite of other genes is part of what I mean by 'genetic background' or 'genetic climate'. In theory, any gene could therefore exert radically different effects on different islands, in the presence of different suites of other genes. In the case of Ford's Yellow Underwings, things are a little more complicated, and very illuminating. The curtisii gene is a 'switch gene' which has what looks like the same effect on both Barra and Orkney, but it achieves it by switching on different suites of genes on the different islands. We notice this only when the two populations are cross-bred. The curtisii switch gene finds itself in a genetic climate which is neither one thing nor the other. It is a mixture of Barra genes and Orkney genes, and the colour pattern that either suite could produce, on its own, breaks down.
What is interesting about this is that either the Barra mixture or the Orkney mixture can put together the colour pattern. There is more than one way of achieving the same result. Both of them involve cooperating suites of genes, but they are two different suites and the members of each suite don't cooperate well with the other. I take this to be a model for what often goes on among working genes within any gene pool. In The Selfish Gene, I used a rowing analogy. A crew of eight oarsmen needs to be well coordinated. Eight men who have trained together can expect to work well together. But if you mix four men from one crew with four from another equally good crew, they don't gel: their rowing fails apart. This is analogous to the mixing of two suites of genes which worked well when each was with its previous companions, but whose coordination breaks down when each is pushed into the foreign genetic climate provided by the other.
Now at this point many biologists get carried away and say that natural selection must work at the level of the whole crew as a unit, the whole suite of genes, or the whole individual organism. They are right that the individual organism is a very important unit in the hierarchy of life. And it really does display unitary qualities. (This is less true of plants than of animals, who tend to have a fixed set of parts, all neatly parcelled inside a skin with a discrete, unitary shape. Individual plants are often harder to delimit as they straggle and vegetatively propagate themselves through meadows and undergrowth. ) But, however unitary and discrete an individual wolf or buffalo, say, may be, the package is temporary and it is unique. Successful buffaloes don't duplicate themselves around the world in the form of multiple copies, they duplicate their genes. The true unit of natural selection has to be a unit of which you can say it has a frequency. It has a frequency which goes up when its type is successful, down when it fails.
There are good things and bad about the poetry of general evolutionism. On balance I think it fosters confusion more than illumination, but there is certainly some of both. The analogy between embryonic development and species evolution was artfully exploited by that irascible genius J. B. S. Haldane to make a debating point. When a sceptic of evolution doubted that anything so complicated as a human could have come from single-celled beginnings, Haldane promptly observed that the sceptic himself had done that very thing and the whole process took only nine months. Haldane's rhetorical point is undiminished by the fact, which of course he knew perfectly well, that development is not the same thing as evolution. Development is change in the form of a single object, as clay deforms under a potter's hands. Evolution, as seen in fossils taken from successive strata, is more like a sequence of frames in a cinema film. One frame doesn't literally change into the next, but we experience an illusion of change if we project the frames in succession. With this distinction in place, we can quickly see that the cosmos does not evolve (it develops) but technology does evolve (early aeroplanes are not moulded into later ones but the history of aeroplanes, and of many other
pieces of technology, falls well into the cinema frame analogy). Clothes fashions, too, evolve rather than develop. It is controversial whether the analogy between genetic evolution, on the one hand, and cultural or technical evolution, on the other, leads to illumination or the reverse, and I am not going to get into that argument now.
My remaining examples of bad poetry in evolutionary science come largely from a single author, the American palaeontologist and essayist Stephen Jay Gould. I am anxious that such critical concentration upon one individual shall not be taken as personally rancorous. On the contrary, it is Gould's excellence as a writer that makes his errors, when they occur, so eminently worth rebutting.
In 1977 Gould wrote a chapter on 'eternal metaphors of palaeontology' to introduce a multi-authored book on the evolutionary study of fossils. Beginning with Whitehead's preposterous, though much quoted, statement that all of philosophy is a footnote to Plato, Gould's thesis, in the words of the preacher of Ecclesiastes (whom he also quotes), is that there is nothing new under the sun; 'The thing that hath been, it is that which shall be; and that which is done is that which shall be done. ' Current controversies in palaeontology are just old controversies being recycled. They preceded evolutionary thought and found no resolution within the Darwinian paradigm . . . Basic ideas, like idealized geometric figures, are few in number. They are eternally available for
consumption . . .
Gould's eternally unresolved questions in palaeontology are three in number: Does time have a directional arrow? Is the driving motor of evolution internal or external? Does evolution proceed gradually or in sudden jumps? Historically, he finds examples of palaeontologists who have espoused all eight possible combinations of answers to these three questions, and he satisfies himself that they straddle the Darwinian revolution as though it never happened. But he manages this feat only by forcing analogies between schools of thought which, carefully examined, have no more in common than blood and wine, or helical orbits and helical DNA. All three of Gould's eternal metaphors are bad poetry, forced analogies that obscure rather than illuminate. And bad poetry in his hands is all the more damaging because Gould is a graceful writer.
The question whether evolution has a directional arrow is certainly one that can sensibly be asked, in various guises. But the bedfellows that the different guises bring together are so ill matched that they are not usefully united. Does bodily structure get progressively more complex as evolution goes on? This is a reasonable question. So is the question of whether the total diversity of species on the planet increases progressively as the ages go by. But they are utterly different questions
and it is conspicuously unhelpful to invent a century-spanning school of 'progressivist' thought to unite them. Still less do either of them, in their modern form, have anything in common with the pre-Darwinian schools of 'vitalism' and 'finalism', which held that living things were progressively 'driven' from within, by some mystical life force, towards an equally mystical final goal. Gould forces unnatural connections among all these forms of progressivism, as a device to support his poetic historical thesis.
Much the same is true of the second eternal metaphor, and the question of whether the motor of change is in the external environment, or whether change arises from 'some independent and internal dynamic within organisms themselves'. A prominent modern disagreement is between those who believe that the main driving force of evolution is Darwinian natural selection and those who emphasize other forces such as random genetic drift. This important distinction is not conveyed, not even to the smallest extent, by the internalist/externalist dichotomy that Gould would force upon us in order to maintain his thesis that post- Darwinian argumentation is just a recycling of pre-Darwinian equivalents. Is natural selection externalist or internalist? It depends whether you are talking about adaptation to the external environment or co-adaptation of the parts to each other. I shall return to this distinction later in another context.
Bad poetry is even more evident in Gould's exposition of the third of his eternal metaphors, the one concerning gradual versus episodic evolution. Gould uses the word episodic to unite three kinds of sharp discontinuity in evolution. These are: first, catastrophes such as the mass extinction of the dinosaurs; second, macromutations or saltations; and third, punctuation in the sense of the theory of punctuated equilibrium proposed by Gould and his colleague Niles Eldredge in 1972. This last theory needs more explanation, and I'll come to it in a moment.
Catastrophic extinctions are straightforward to define. Exactly what causes them is controversial and probably different in different cases. For the moment, just notice that a worldwide catastrophe in which most species die is, to put it mildly, not the same thing as a macromutation. Mutations are random errors in gene copying and macromutations are mutations of large effect. A mutation of small effect, or micromutation, is a small error in gene copying, whose effect on its possessors might be too slight to notice easily, say a subtle lengthening of a leg bone, or a hint of reddening in a feather. A macromutation is a dramatic error, a change so large that, in extreme cases, its possessor would be classified in a different species from its parents. In my previous book, Climbing Mount Improbable, I reproduced a photograph from a newspaper of a toad with eyes in the roof of its mouth. If this photograph is genuine (a big if, in
these days of Photoshop and other handy image-manipulation software), and if the error is genetic, the toad is a macromutant. If such a macromutant spawned a new species of toads with eyes in the roofs of their mouths, we should describe the abrupt evolutionary origin of the new species as a saltation or evolutionary jump. There have been biologists, such as the German/American geneticist Richard Goldschmidt, who believed that such saltatory steps were important in natural evolution. I am one of many who have cast doubt on the general idea, but that is not my purpose here. Here I make the much more basic point that such genetic leaps, even if they occur, have nothing in common with earth-shattering catastrophes such as the sudden extinction of the dinosaurs, except that both are sudden. The analogy is purely poetic, and it is bad poetry which doesn't lead to any further illumination. Recalling Medawar's words, the analogy marks the end, not the inauguration, of a train of thought. The ways of being a non- gradualist are so varied as to strip the category of all usefulness.
The same applies to the third category of non-gradualists:
punctuationists in the sense of Eldredge and Gould's theory. The idea here is that a species comes into existence in a time which is short compared with the much longer period of 'stasis' during which it survives unchanged after its initial formation. In the extreme version of the theory, the species, once it has burst into existence, continues unchanged until either it goes extinct or it splits to form a new daughter species. It is
when we ask what happens during the sudden bursts of species formation that the confusion, born of bad poetry, arises. There are two things that might happen. They are utterly different from each other, but Gould makes light of the difference because he is seduced by bad poetry. One is macromutation. The new species is founded by a freak individual, like the alleged toad with eyes in the roof of its mouth. The other thing that might happen - more plausibly, in my view, but I'm not talking
about that now - is what we can call rapid gradualism. The new species comes into existence in a brief episode of rapid evolutionary change which, although gradual in the sense that parents don't spawn an
instant new species in a single generation, is fast enough to look like an instant in the fossil record. The change is spread over many generations of small, step-by-step increments, but it looks like a sudden jump. This
is either because the intermediates lived in a different place (say, on an outlying island) and/or because the intermediate stages passed too rapidly to fossilize - 10,000 years is too short to measure in many geological strata, yet it constitutes ample time for quite major evolutionary change to accumulate gradually in small steps.
There is all the difference in the world between rapid gradualism and macromutational saltation. They depend upon totally different mechanisms and they have radically different implications for Darwinian
controversies. To lump them together simply because, like catastrophic extinctions, they all lead to discontinuities in the fossil record, is bad poetic science. Gould is aware of the difference between rapid gradualism and macromutation, but he treats the matter as though it were a minor detail, to be cleared up after we have taken on board the overarching question of whether evolution is episodic rather than gradual. One can see it as overarching only if one is intoxicated by bad poetry. It makes as little sense as my correspondent's question about the DNA double helix and whether it 'comes from' the earth's orbit. Once again, rapid gradualism no more resembles macromutation than a bleeding wizard resembles a shower of rain.
Even worse is to claim catastrophism under the same punctuationist umbrella. In pre-Darwinian times the existence of fossils became increasingly embarrassing for upholders of biblical creation. Some hoped to drown the problem in Noah's flood, but why did the strata seem to show dramatic replacements of whole faunas, each one different from its predecessor, and all of them largely free of our own, familiar creatures? The answer given by, among others, the nineteenth-century French anatomist Baron Cuvier, was catastrophism. Noah's flood was only the last in a series of cleansing disasters visited upon the earth by a supernatural power. Each catastrophe was followed by a new creation.
Apart from the supernatural intervention, this has something - a little - in common with our modern belief that mass extinctions such as those that ended the Permian and Cretaceous eras were followed by new flowerings of evolutionary diversity to match previous radiations. But to lump the catastrophists in with macromutationists and with modern punctuationists, just because all three can be represented as non- gradualist, is very' bad poetry indeed.
After giving lectures in the United States, I have often been puzzled by a certain pattern of questioning from the audience. The questioner calls my attention to the phenomenon of mass extinction, say, the catastrophic end of the dinosaurs and their succession by the mammals. This interests me greatly and I warm to what promises to be a stimulating question. Then I realize that the tone of the question is unmistakably challenging. It is almost as though the questioner expects me to be surprised, or discomfited, by the fact that evolution is periodically interrupted by catastrophic mass extinctions. I was baffled by this until the truth suddenly hit me. Of course! The questioner, like many people in North America, has learned his evolution from Gould, and I have been billed as one of those 'ultra-Darwinian' gradualists. Doesn't the comet that killed the dinosaurs also blow my gradualistic view of evolution out of the water? No, of course it doesn't. There is not the smallest connection. I am a gradualist in the sense that I don't think
macromutations have played an important role in evolution. More determinedly, I am a gradualist when it comes to explaining the evolution of complex adaptations like eyes (so is any sane person, including Gould). But what on earth have such matters got to do with mass extinctions? Nothing at all. Unless, that is, your mind has been filled up with bad poetry. For the record, I believe, and have believed for the whole of my career, that mass extinctions exert a profound and dramatic influence on the subsequent course of evolutionary history. How could they not? But mass extinctions are not a part of the Darwinian process, except in so far as they clear the decks for new Darwinian beginnings.
There is irony lurking here. Among the facts about extinction that Gould is fond of emphasizing is its capriciousness. He calls it contingency.
When mass extinction strikes, major groups of animals die wholesale. In the Cretaceous extinction, the once mighty group of dinosaurs (with the notable exception of birds) was completely wiped out. The choice of major group for victim is either random or, if non-random, it is not the same non-randomness as we see in conventional natural selection. The normal adaptations to survival do not avail against comets. Grotesquely, this fact is sometimes trotted out as if it were a debating point against neo- Darwinism. But neo-Darwinian natural selection is selection within species, not between species. To be sure, natural selection involves death, and mass extinction involves death, but any further resemblance
between the two is purely poetic. Ironically, Gould is one of the few Darwinians who still think of natural selection as working at levels higher than the individual organism. It would never occur to the rest of us even to ask whether mass extinctions are selective events. We might see extinction as opening up new opportunities for adaptation, by lower- level natural selection choosing between individuals separately within each species that has survived the catastrophe. As a further irony, it is the poet Auden who came nearer to getting it right:
But catastrophes only encouraged experiment.
As a rule, it was the fittest who perished, the mis-fits, forced by failure to emigrate to unsettled niches, who altered their structure and prospered.
'Unpredictable but Providential (for Loren Eiseley)'
I take one further extended example of bad poetic science from paleontology, and once again Stephen Jay Gould is responsible for its popularity even if he has not clearly expressed it himself in its extreme form. Many readers of his elegantly written book Wonderful Life (1989) have been captivated by the idea that there was something special and unique about the whole business of evolution in the Cambrian era, when fossils of most of the great animal groups first appeared, rather over 500 million years ago. It is not just that the animals of the Cambrian were
peculiar. Of course they were. The animals of every era have their peculiarities and the Cambrian ones were arguably more peculiar than most. No, the suggestion is that the whole process of evolution in the Cambrian was odd.
The standard neo-Darwinian view of the evolution of diversity is that a species splits into two when two populations become sufficiently unalike that they can no longer interbreed. Often the populations begin diverging when they chance to be geographically separated. The separation means that they no longer mix their genes sexually and this permits them to evolve in different directions. The divergent evolution might be driven by natural selection (which is likely to push in different directions because of different conditions in the two geographical areas). Or it might consist of random evolutionary drift (since the two populations are not genetically held together by sexual mixing, there is nothing to stop them drifting apart). In either case, when they have evolved sufficiently far apart that they could no longer interbreed even if they were geographically united again, they are defined as belonging to separate species.
Subsequently, the lack of interbreeding permits further evolutionary divergence. What had been distinct species within one genus become, in the fullness of time, distinct genera within one family. Later, families will be found to have diverged to the point where taxonomists (specialists in classification) prefer to call them orders, then classes, then phyla. Phylum (plural phyla) is the classificatory name by which we distinguish really fundamentally different animals like molluscs, nematode worms, echinoderms and chordates (chordates are mostly vertebrates plus a few odds and ends). Ancestors of two different phyla, say vertebrates and molluscs, which we see as built upon utterly different 'fundamental body plans' were once just two species within a genus. Before that, they were two geographically separated populations within one ancestral species. The implication of this widely accepted view is that, as you go back and back in geological time, the gap between any pair of animal groups becomes smaller and smaller. The further back in time you go, the closer you approach the uniting of these different kinds of animals in their single common ancestor species. Our ancestors and mollusc ancestors were once very alike. Later they were not quite so alike. Later again they had diverged further, and so on until eventually they became so different that we should call them two phyla. This general story can scarcely be doubted by any reasonable person who thinks it through, though we do not have to be committed to the view that it occurs at a uniform rate with time. It could have happened in rapid bursts.
The dramatic phrase 'Cambrian explosion' is used in two senses. It can refer to the factual observation that before the Cambrian era, just over half a billion years ago, there are few fossils. Most of the great animal
phyla appear as fossils for the first time in Cambrian rocks, and this looks like a great explosion of new animals. The second meaning is the theory that the phyla actually branched off from each other during the Cambrian, even during as little as 10 million years within the Cambrian. This second idea, which I shall call the branch point explosion hypothesis, is controversial. It is compatible - just - with what I am calling the standard neo-Darwinian model of species divergence. We've already agreed that, as we trace any pair of modem phyla back in time, we eventually converge upon a common ancestor. My hunch is that, for different pairs of phyla, we'll hit the common ancestor in different geological eras: say, the common ancestor of vertebrates and molluscs at 8oo million years ago, the common ancestor of vertebrates and echinoderms at 600 million years, and so on. But I could be wrong, and we can easily accommodate the branch point explosion hypothesis by saying that, for some reason (which is interesting enough to need investigating), most of our backward tracings happen to hit their respective common ancestors during the same relatively short geological period, say, between 540 million and 530 million years ago. This would have to mean that, at least near the beginning of that 10-million-year period, the ancestors of the modem phyla were nowhere near as different from each other as they are today. They were, after all, diverging from common ancestors at the time and were originally members of the same species.
The extreme Gouldian view - certainly the view inspired by his rhetoric, though it is hard to tell from his own words whether he literally holds it himself - is radically different from and utterly incompatible with the standard neo-Darwinian model. It also, as I shall show, has implications which, once they are spelled out, anybody can see are absurd. It is very clearly expressed - betrayed might be a better word - in asides in Stuart Kauffman's At Home in the Universe (1995):
One might imagine that the first multicellular creatures would all he very similar, only later diversifying, from the bottom up, into different genera, families, orders, classes, and so on. That, indeed, would be the expectation of the strictest conventional Darwinist. Darwin, profoundly influenced by the emerging view of geologic gradualism, proposed that all evolution occurred by the very gradual accumulation of useful variations. Thus the earliest multicellular creatures themselves ought to have diverged gradually from one another.
So far, this is a fine summary of the orthodox neo-Darwinian view. Now, in a bizarre passage, Kauffman goes on:
But this appears to be false. One of the wonderful and puzzling features of the Cambrian explosion is that the chart was filled in from the top
down. Nature suddenly sprang forth with many wildly different body plans - the phyla - elaborating on these basic designs to form the classes, orders, families, and genera . . . In his book about the Cambrian explosion, Wonderful Life: The Burgess Shale and the Nature of History, Stephen Jay Gould remarks on this top-down quality of the Cambrian with wonder.
As well he might! You only have to think for one moment about what 'top down' filling in would have to mean for the animals on the ground and you immediately see how preposterous it is, 'Body plans' like the mollusc body plan, or the echinoderm body plan, are not ideal essences hanging in the sky, waiting, like designer dresses, to be adopted by real animals. Real animals is all there ever was: living, breathing, walking, eating, excreting, fighting, copulating real animals, who had to survive and who can't have been dramatically different from their real parents and grandparents. For a new body plan - a new phylum - to spring into existence, what actually has to happen on the ground is that a child is born which suddenly, out of the blue, is as different from its parents as a snail is from an earthworm. No zoologist who thinks through the implications, not even the most ardent saltationist, has ever supported any such notion. Ardent saltationists have been content to postulate the sudden bursting into existence of new species, and even that relatively modest idea has been highly controversial. When you spell out the Gouldian rhetoric into real-life practicalities, it stands revealed as the purest of bad poetic science.
Kauffman is even more explicit in a later chapter. In discussing some of his ingenious mathematical models of evolution on 'rugged fitness landscapes', Kauffman notes a pattern that he thinks sounds a lot like the Cambrian explosion. Early on in the branching process, we find a variety of long-jump mutations that differ from the stem and from one another quite dramatically. These species have sufficient morphological differences to be categorized as founders of distinct phyla. These founders also branch, but do so via slightly closer long-jump variants, yielding branches from, each founder of a phylum to dissimilar daughter species, the founders of classes. As the process continues, fitter variants are found in progressively more nearby neighborhoods, so founders of orders, families, and genera emerge in succession.
Kauffman's earlier, more technical book, The Origins of Order (1993), says something similar about life in the Cambrian:
Not only did a very large number of novel body forms originate rapidly, but the Cambrian explosion exhibited another novelty: Species which founded taxa appear to have built up the higher taxa from the top down.
That is, exemplars of major phyla were present first, followed by progressive filling in at class, order, and lower taxonomic levels. . .
Now, one way of reading this is inoffensive to the point of obviousness. On our 'converging backwards' model it would have to be true that species splittings that are eventually going to become phylum divides would normally precede those that are destined to become divides between orders and lower taxonomic levels. But Kauffman clearly doesn't think he is saying something ordinary and obvious.
This is apparent from his statement that 'the Cambrian explosion exhibited another novelty', and from his phrase long-jump mutations'. He thinks he is attributing to the Cambrian something revolutionary. He really does seem genuinely to intend the alternative reading, in which 'long-jump mutations' give rise, on the instant, to brand new phyla.
I hasten to emphasize that these passages of Kauffman's are embedded in a pair of books that are for the most part interesting, creative and uninfluenced by Gould. The same is true of Richard Leakey and Roger Lewin's The Sixth Extinction (1996), another recent book, admirable in most of its chapters, but sadly marred by one, 'The Mainspring of Evolution', which is explicitly and avowedly influenced by Gould. Here are a couple of relevant passages:
It was as if the facility for making evolutionary leaps that produced major functional novelties - the basis of new phyla - had somehow been lost when the Cambrian period came to an end. It was as if the mainspring of evolution had lost some of its power.
Hence, evolution in Cambrian organisms could take bigger leaps, including phylum-level leaps, while later on it would be more constrained, making only modest jumps, up to the class level.
As I have written before, it is as though a gardener looked at an old oak tree and remarked, wonderingly: 'Isn't it strange that no major new boughs have appeared on this tree for many years. These days, all the new growth appears to be at the twig level! ' Just think once again what a 'phylum-level leap' or even a 'modest' {modest? ) class level leap would have to mean. Animals of different phyla, remember, are animals with different fundamental body plans, like molluscs and vertebrates. Or like starfish and insects. A long-jump, phylum level mutation would have to mean that a couple of parents belonging to one phylum mated and gave birth to a child belonging to a different phylum. The difference between parent and offspring would have to be on the same scale as the difference between a snail and a lobster, or a starfish and a codfish. A class level leap would be equivalent to a pair of birds giving birth to a mammal.
Picture the parents gazing wonderingly into the nest at what they have produced, and the full comedy of the notion becomes apparent.
My assurance in ridiculing these ideas is not based simply upon knowledge of the facts of modern animals. Obviously if it were just that, one could retort that things were different in the Cambrian. No, the argument against Kaufman's long jumps, or Leakey and Lewin's phylum level leaps, is a theoretical one, and an extremely strong one. It is this. Even if mutations on this gigantic scale occurred, the products would not have survived. This is fundamentally because, as I have said before, however many ways there may be of being alive, there are almost infinitely more ways of being dead. A small mutation, representing a minor step away from a parent which has proved its ability to survive by virtue of being a parent, has a good chance of surviving for the same reason, and it may even be an improvement. A gigantic, phylum level mutation is a leap into the wild blue yonder. I said that the long-jump mutation we are talking about would be of the same magnitude as a mutation from a mollusc to an insect. But it would never, of course, have been a jump from a mollusc to an insect. An insect is a highly tuned piece of survival machinery. If a mollusc parent gave birth to a new phylum, the leap would have been a random leap, like any other mutation. And the chance that a random leap of that magnitude would produce an insect, or anything with the faintest chance of surviving, is small enough to be discounted totally. The chance of its being viable is impossibly small, no matter how empty the ecosystem, how wide open the niches. A phylum level leap would be a mess.
I do not believe the authors I am quoting really believe what their printed words undoubtedly appear to be saying. I think they were simply intoxicated by Gould's rhetoric and didn't think it through. The whole point of quoting them in this chapter is to illustrate the power to mislead that a skilled poet can unwittingly exert, especially if he has first misled himself And the poetry of the Cambrian as a blissful dawn of innovation is undoubtedly beguiling. Kauffman gets completely carried away by it:
Soon after multicelled forms were invented, a grand burst of evolutionary novelty thrust itself outward. One almost gets the sense of multicellular life gleefully trying out all its possible ramifications, in a kind of wild dance of heedless exploration
At Home in the Universe (1995)
Yes. One does get exactly that sense. But one gets it from Gould's rhetoric, not from the facts of Cambrian fossils nor from sober reasoning about evolutionary principles.
If scientists of the calibre of Kauffman, Leakey and Lewin can be seduced by bad poetic science, what chance has the non-specialist?
Daniel Dennett has told me of a conversation with a philosopher colleague who had read Wonderful Lifers arguing that the Cambrian phyla did not have a common ancestor - that they had sprung up as independent origins of life! When Dennett assured him that this was not Gould's intention, his colleague's response was, 'Well then, what is all the fuss about? '
Excellence in writing is a double-edged sword, as the distinguished evolutionary scientist John Maynard Smith has noted, in the New York Review of Books, November 1995:
Gould occupies a rather curious position, particularly on his side of the Atlantic. Because of the excellence of his essays, he has come to be seen by non-biologists as the pre-eminent evolutionary theorist. In contrast, the evolutionary biologists with whom I have discussed his work tend to see him as a man whose ideas are so confused as to be hardly worth bothering with, but as one who should not be publicly criticized because he is at least on our side against the creationists. All this would not matter, were it not that he is giving non-biologists a largely false picture of the state of evolutionary theory.
Maynard Smith was reviewing Dennett's book Darwin's Dangerous Idea (1995), which contains a devastating and, one might hope, terminal critique of Gould's influence on evolutionary thinking.
What really happened in the Cambrian? Simon Conway Morris of Cambridge University is, as Gould fulsomely acknowledges, one of the three leading modern investigators of the Burgess Shale, the Cambrian fossil bed which is the subject of Wonderful Life. Conway Morris has recently published his own fascinating book on the subject. The Crucible of Creation (1998), which is critical of almost every aspect of Gould's view. Like Conway Morris, I don't think there's any good reason to think that the process of evolution was different in the Cambrian from the way it is today. But there is no doubt that a large number of major animal groups are seen in the fossil record for the first time in the Cambrian. The obvious hypothesis has occurred to many people. Perhaps several groups of animals evolved hard, fossilizable skeletons around the same time and perhaps for the same reason. One possibility is an evolutionary arms
race between predators and prey, but there are other ideas like a dramatic change in the chemistry of the atmosphere. Conway Morris finds no support at all for the poetic idea of an exuberant and
extravagant flowering of life in a wild dance of Cambrian diversity and disparity, subsequently pruned to today's more limited repertoire of
animal types. If anything, the reverse seems to be true, as most evolutionists would expect.
Where does that leave the question of the timing of the branch points of the major phyla? Recall that this is a separate question from the undoubted Cambrian explosion of fossil availability. The controversial matter is whether the branch points in the divergence of all the major phyla are concentrated in the Cambrian - the branch point explosion hypothesis. I said standard neo-Darwinism was compatible with this hypothesis. But I still don't think it is at all likely.
One possible way to tackle the question is by looking at molecular clocks. 'Molecular clock' refers to the observation that certain biological molecules change at a rather fixed rate over the millions of years. If you accept this, you can take blood from any two modern animals and calculate how long ago their common ancestor lived. Some recent molecular clock studies have pushed the branch points of various pairs of phyla deep into the Precambrian era. If these studies are right, the whole rhetoric of an evolutionary explosion becomes superfluous. But there is controversy over the interpretation of molecular clock results so far back in deep time, and we should wait for more evidence.
Meanwhile, there is a logical argument which I can assert with more confidence. The only evidence in favour of the branch point explosion hypothesis is negative: there aren't any fossils of many of the phyla before the Cambrian. But those fossil animals that have no fossil ancestors must have had ancestors of some kind. They can't have sprung from nothing. Therefore there must have been ancestors that didn't fossilize, absence of fossils does not mean absence of animals. The only question that remains is whether the missing ancestors going back to the branch points, who must have existed, were all compressed into the Cambrian, or whether they were strung out through the previous hundreds of millions of years. Since the only reason to suppose that they were compressed into the Cambrian is the absence of fossils, and since we have just proved logically the irrelevance of that absence, I conclude that there is no good reason at all to favour the branch point explosion hypothesis. Doubtless it has great poetic appeal.
9
THE SELFISH COOPERATOR
Wonder. . . and not any expectation of advantage from its discoveries, is the first principle which prompts mankind to the study of Philosophy, of that science which pretends to lay open the concealed connections that unite the various appearances of nature.
ADAM SMITH, The History of Astronomy' (1795)
The medieval bestiaries continued an earlier tradition of hijacking nature as a source of moral tales. In its modern form, in the development of evolutionary ideas, the same tradition underlies one of the most egregious forms of bad poetic science. I refer to the illusion that there is a simple opposition between nasty and nice, social and antisocial, selfish and altruistic, tough and gentle; that these pairs of binary opposites all correspond to the other pairs, and that the history of evolutionary controversy about society is described by a pendulum swinging back and forth along a continuum between these opposites. I am not denying that there are interesting issues to be discussed hereabouts. What I am criticizing is the 'poetic' idea that there is a single continuum and that worthwhile arguments are to be had between vantage points along its length. To invoke the rainmakers yet again, there is no more connection between a selfish gene and a selfish human than there is between a rock and a rain cloud.
To explain the poetic continuum I am criticizing, I might as well borrow a line from a real poet, Tennyson's 'Nature, red in tooth and claw', from In Memoriam (1850), widely assumed to be inspired by On the Origin of Species but actually published nine years earlier. At one end of the poetic continuum are supposed to stand Thomas Hobbes, Adam Smith, Charles Darwin, T. H. Huxley and all those, such as the distinguished American evolutionist George C. Williams and today's advocates of 'the selfish gene', who emphasize that nature really is red in tooth and claw. At the other end of the continuum are Prince Peter Kropotkin the Russian anarchist and author of Mutual Aid (1902), the gullible but immensely influential American anthropologist Margaret Mead, and today a spate of authors reacting indignantly to the idea that nature is genetically selfish, of
whom Frans de Waal, author of Good Natured (1996), is representative.
De Waal, a chimpanzee expert who understandably loves his animals, is distressed at what he mistakenly sees as a neo-Darwinian tendency to emphasize the 'nastiness of our apish past'. Some of those who share his romantic fancy have recently become fond of the pygmy chimpanzee or bonobo as a yet more benign role model. Where common chimpanzees often resort to violence, and even cannibalism, bonobos say it with sex. They seem to copulate in all possible combinations at every conceivable opportunity. Where we might shake hands, they copulate. Make love not war is their watchword. Margaret Mead would have warmed to them. But the very idea of taking animals to be role models, as in the bestiaries, is a piece of bad poetic science. Animals are not there to be role models, they are there to survive and reproduce.
Moralistic devotees of the bonobo are apt to compound this error with an outright evolutionary falsehood. Probably because of their powerful 'feelgood factor', bonobos are often claimed as more closely related to us than common chimpanzees are. But this cannot be, so long as we accept, as everybody does, that bonobos and common chimpanzees are more closely related to each other than either is to humans. You need no more than that simple and uncontroversial premise to conclude that bonobos and common chimpanzees are exactly equally closely related to us. They are connected to us via the common ancestor that they share and we don't. Certainly we may resemble one of the two species more than the other in some respects (and very probably resemble the other in other respects), but such comparative judgements absolutely cannot be reflections of differential evolutionary closeness.
* I should explain that Margaret Mead is 'gullible but influential' because a large section of American academic culture enthusiastically adopted her rose-tinted environmentalist theory of human nature which, it later transpired, she had built on a somewhat insecure foundation: systematic misinformation fed her as a joke by two mischievous Samoan girls, during her brief period of fieldwork in their island. She didn't stay in Samoa long enough to learn the language well, unlike her professional nemesis, the Australian anthropologist Derek Freeman, who uncovered the whole story years later in the course of a more detailed study of Samoan life.
De Waal's book is full of anecdotal demonstrations (which should surprise nobody) that animals are sometimes kind to each other, cooperate for mutual good, care for each other's welfare, console each other in distress, share food and do other heartwarmingly good things. The position I have always adopted is that much of animal nature is indeed altruistic, cooperative and even attended by benevolent subjective emotions, but that this follows from, rather than contradicts, selfishness at the genetic level. Animals are sometimes nice and sometimes nasty, since either can suit the self-interest of genes at different times. That is precisely the reason for speaking of 'the selfish gene' rather than, say, 'the selfish chimpanzee'. The opposition that de Waal and others have erected, between biologists who believe human and animal nature is fundamentally selfish, and those who believe it is fundamentally 'good- natured', is a false opposition - bad poetry.
It is now widely understood that altruism at the level of the individual organism can be a means by which the underlying genes maximize their self-interest. However, I don't want to dwell on what I have expounded in earlier books such as The Selfish Gene. What I would now re-emphasize from that book - it has been overlooked by critics who appear to have read it by title only - is the important sense in which genes, though in
one way purely selfish, at the same time enter into cooperative cartels with each other. This is poetic science, if you like, but I hope to show that it is good poetic science which aids understanding rather than impedes it. I shall do the same with other examples in the remaining chapters. The key insight of Darwinism can be expressed in genetic terms. The genes that exist in many copies in the population are the ones that are good at making copies, which also means good at surviving. Surviving where? Surviving in individual bodies in ancestral environments. That means surviving in the environment typical of the species: in a desert for camels, up trees for monkeys, in the deep sea for giant squids, and so on. The reason individual bodies are so good at surviving in their environments is mainly that they have been built by genes that have survived in the same environment for many generations, in the form of copies.
But never mind deserts and ice floes, seas and forests; they are only part of the story. A far more prominent aspect of the ancestral environment in which genes have survived is the other genes with which they have had to share a succession of individual bodies. The genes that survive in camels will, to be sure, include some that are particularly good at surviving in deserts, and they may even be shared with desert rats and desert foxes. But more importantly, successful genes will be those that are good at surviving in an environment consisting of the other genes that are typically found in the species. So, the genes of a species become selected to be good at cooperating with each other. Genetic cooperation, which is good scientific poetry whereas universal cooperation is not, will be the subject of this chapter.
The following fact is often misunderstood. It is not the genes of any given individual that cooperate particularly well together. They have never been together before in that combination, for every genome in a sexually reproducing species is unique (with the usual exception of identical twins). It is the genes of a species at large that cooperate, because they have met before, often, and in the intimately shared environment of the cell, though always in different combinations. What they cooperate at is the business of making individuals of the same general type as the present one. There is no particular reason to expect the genes of any particular individual to be especially good at cooperating with one another when compared with any other genes of the same species. It is largely a matter of accident which particular companions the lottery of sexual reproduction has drawn for them from the gene pool of the species. Individuals with unfavourable combinations of genes tend to die. Individuals with favourable combinations tend to pass those genes on to the future. But it is not the favourable combinations themselves that are passed on in the long term. Sexual reshuffling sees to that. Instead, what are passed on are the genes that tend to be good at forming favourable
combinations with the other genes that the species gene pool has to offer. Over the generations, whatever else the surviving genes may be good at, they will be good at working together with other genes of the species.
For all we know, particular camel genes might be good at cooperating with particular cheetah genes. But they are never called upon to do so. Presumably mammal genes are better at cooperating with other mammal genes than with bird genes. But the speculation must remain hypothetical, because one of the characteristics of life on our planet is that, genetic engineering aside, genes are mixed only within species. We can test watered-down versions of such speculations by looking at hybrids. Hybrids between different species, when they exist at all, often survive less well, or are less fertile, than pure-bred individuals. At least part of the reason for this is incompatibilities between their genes. Species A genes that work well against a genetic background or 'climate' of other Species A genes do not work when transplanted into Species B, and vice versa. Similar effects are sometimes seen when varieties or races within one species hybridize.
I first understood this while listening to lectures by the late E. B. Ford, legendary Oxford aesthete and eccentric founder of the now neglected school of Ecological Geneticists. Most of Ford's research was on wild populations of butterflies and moths. Among these was the Lesser Yellow Underwing moth, Triphaena comes. This moth is normally yellowish brown, but there is a variant called curtisii which is blackish. Curtisii is not found in England at all; however, in Scotland and the Isles curtisii coexists with the normal comes. The curtisii dark colour pattern is nearly completely dominant to the normal comes pattern. 'Dominant to' is a technical term, which is why I can't just say 'dominates'. It means that hybrids between the two look like curtisii even though they bear the genes of both. Ford caught specimens from Barra in the Outer Hebrides, west of Scotland, and from one of the Orkney islands, north of Scotland, as well as from the Scottish mainland itself. Each of the two island forms looks exactly like its opposite number at the other island site, and the dark curtisii gene is dominant on the two islands, as well as on the mainland. Other evidence shows that the curtisii gene is the very same gene in all localities. In view of this you'd expect that, when you cross- bred specimens from different islands, the normal dominance pattern would hold up. But it doesn't, and this is the point of the story. Ford caught individuals from Barra and mated them with individuals from Orkney. And the dominance of curtisii completely disappeared. A full range of intermediates turned up in the hybrid families, just as if there was no dominance. What seems to be going on is this. The curtisii gene does not, in itself, encode the formula for the coloured pigment by which we distinguish the moths, nor is dominance ever a property of a gene on its own. Instead, like any other gene, the curtisii gene should be thought
of as having its effects only in the context of a suite of other genes, some of which it 'switches on'. This suite of other genes is part of what I mean by 'genetic background' or 'genetic climate'. In theory, any gene could therefore exert radically different effects on different islands, in the presence of different suites of other genes. In the case of Ford's Yellow Underwings, things are a little more complicated, and very illuminating. The curtisii gene is a 'switch gene' which has what looks like the same effect on both Barra and Orkney, but it achieves it by switching on different suites of genes on the different islands. We notice this only when the two populations are cross-bred. The curtisii switch gene finds itself in a genetic climate which is neither one thing nor the other. It is a mixture of Barra genes and Orkney genes, and the colour pattern that either suite could produce, on its own, breaks down.
What is interesting about this is that either the Barra mixture or the Orkney mixture can put together the colour pattern. There is more than one way of achieving the same result. Both of them involve cooperating suites of genes, but they are two different suites and the members of each suite don't cooperate well with the other. I take this to be a model for what often goes on among working genes within any gene pool. In The Selfish Gene, I used a rowing analogy. A crew of eight oarsmen needs to be well coordinated. Eight men who have trained together can expect to work well together. But if you mix four men from one crew with four from another equally good crew, they don't gel: their rowing fails apart. This is analogous to the mixing of two suites of genes which worked well when each was with its previous companions, but whose coordination breaks down when each is pushed into the foreign genetic climate provided by the other.
Now at this point many biologists get carried away and say that natural selection must work at the level of the whole crew as a unit, the whole suite of genes, or the whole individual organism. They are right that the individual organism is a very important unit in the hierarchy of life. And it really does display unitary qualities. (This is less true of plants than of animals, who tend to have a fixed set of parts, all neatly parcelled inside a skin with a discrete, unitary shape. Individual plants are often harder to delimit as they straggle and vegetatively propagate themselves through meadows and undergrowth. ) But, however unitary and discrete an individual wolf or buffalo, say, may be, the package is temporary and it is unique. Successful buffaloes don't duplicate themselves around the world in the form of multiple copies, they duplicate their genes. The true unit of natural selection has to be a unit of which you can say it has a frequency. It has a frequency which goes up when its type is successful, down when it fails.
