Climbing Mount Improbable

by Richard Dawkins

  Figure 3.1 African elephant, Loxodonta africana, and unrelated long-nosed mammals who probably developed their long noses for independent reasons: (anticlockwise from top left) proboscis monkey, Nasalis larvatus; elephant shrew, Rhynchocyon petersi; Malayan tapir, Tapirus indicus; and southern elephant seal, Mirounga leonina. {93}

  'That medium-sized trunk is no good because it is neither one thing nor the other — falls between two stools — but don't worry, give it another few million years and it'll be fine.’ No animal ever made a living purely by being on the evolutionary path to something better. Animals make a living by eating, avoiding being eaten, and reproducing. If a medium-sized trunk were always less efficient for these purposes than either a small nose or a big trunk, the big trunk would never have evolved.

  Just because the trunk had to be useful in all its intermediate stages, this does not mean that it had to be useful for the same purpose throughout all the intermediate stages. Early elongation could have provided an advantage that had nothing to do with picking up objects. Perhaps the nose got longer in the first place to jack up the sense of smell, as in elephant shrews; or as a resonator for calls, as in elephant seals; or an adornment to attract a mate, as in — however surprising it may seem to our aesthetic sense — proboscis monkeys. On the other hand it is also possible that its usefulness as a ‘hand’ came quite early in elephant evolution, when it was still quite short. This guess is made plausible by comparison with tapirs who use their nose to grasp leaves and pull them towards the mouth. Independent evolutions of similar devices in different animals can illuminate our understanding of each other.

  In the specific case of the elephant trunk, there is suggestive evidence from the hard parts of the skull that fossilize, especially the tusks and associated bones. Today's two species of elephants are the sole survivors of a once rich radiation of tusked animals thriving in every continent. Modern elephant tusks are the enormously enlarged upper incisor teeth but many fossil forms, such as some of the mast-odonts, had more prominent lower incisors also pointing forwards. Sometimes they were large and spiky like the tusks that we now see only in the upper jaw. In other kinds they were flat so that the two great teeth together constituted a broad shovel or spade of ivory prolonging the line of the lower jaw, which was probably used for grubbing up roots and tubers. The shovel extended so far in front of the lower jaw that the upper lip could not reach the food that it dug up. It seems probable that the incipient trunk extended originally to work against the shovel and grasp the food that the shovel dug. Later, we {94} may guess, the nascent trunk became so good at this that it started to be used on its own, without the shovel. Later, at least in the lineages that have survived, the shovel itself became reduced while the trunk remained, as if beached by the receding tide. The lower jaw retreated to something like its original proportions, leaving the now fully independent trunk as its legacy. Consult John Maynard Smiths excellent The Theory of Evolution, pages 291 — 4, for a fuller account of elephant-ttunk evolution.

  The word ‘pre-adaptation is used for cases where an organ is originally used for some purpose and then later in evolution is taken over for another purpose. It is an illuminating idea, for it frequently rescues us from puzzlement over evolutionary origins. Porcupine quills ire now formidable weapons. They didn't spring from nothing, but are modified hairs, ‘pre-adapted’ for the completely different purpose of keeping warm. Many mammals have highly developed, specialist scent glands. Where they sprang from might seem a mystery until you look at them closely under the microscope and see that they are modified from a smaller gland with the very different purpose of secreting sweat to cool the body down. Unreconstructed sweat glands are still common in other parts of the very same animal, so the comparison is easy to make. Other scent glands seem to have evolved from sebaceous glands, whose original duty was to protect hairs with I waxy secretion. Often the pre-adaptation and its modern successor are not unrelated. Sweat happens to smell, and it happens to be secreted when the animal is emotionally aroused (people are popularly lupposed to sweat with fear, and I know I do when an important lecture isn't going according to plan). It therefore was natural for the old pre-adaptation to shift into its specialized counterpart.

  Sometimes it isn't obvious which came first — which is the early pre-idaptation and which the later specialization. Darwin, wondering about tihe evolutionary origin of the lung, sought an answer in the swim bladder of fish. This is a gas-filled bladder, which bony fish use to control their buoyancy on the principle of the Cartesian Diver (those little men in bottles who can be made to sink or rise by means of gentle pressure on the cork). By using muscles to adjust the volume of its swim bladder, a fish is able to change the depth at which it can rest at equilibrium. {95} This applies only to ordinary bony fish. Sharks (which, despite their fishy shape, are actually more distantly related to bony fish than we are) lack swim bladders and they consequently have to do more swimming work to keep themselves at their desired level in the water. The swim bladder looks like a lung, and Darwin thought that it might be the pre-adaptation from which our lungs evolved. Modern zoologists mostly reverse this particular horse and cart, suspecting that the swim bladder is a recent modification of a primitive lung (air-breathing fish are quite common, to this day). Whichever is the more primitive, we need to think about what, even earlier, preceded it. Perhaps the lung/swim bladder arose from a pouch of the gut and had a primitive digestive function. In every stage of its evolution, at every step up the slopes of Mount Improbable, the pouch/cavity/lung had to be useful to the animal that bore it.

  Couldn't the elephant's trunk have shot out in a single, giant step? Why shouldn't an offspring have had a trunk like an elephant, while its parents had trunks like tapirs? There are really three questions here. First, the question whether mutations of very large magnitude — macro-mutations — can happen. Second, the question whether, given that they happen, natural selection would ever favour them. Third is a more subtle question about what we mean by large, when we speak of large mutational change. I shall come back to a distinction I have made in an earlier book, between ‘Boeing 747 macro-mutations’ and ‘Stretched DC8 macro-mutations’.

  The answer to the first of the three question is yes. Macro-mutations do happen. Offspring are sometimes born radically, monstrously different from either parent, and from other members of the species. The toad in Figure 3.2 is said by the photographer, Scott Gardner of the Hamilton Spectator, to have been found by two girls in their garden in Hamilton, Ontario. He says that they put it on the kitchen table for him to photograph. It had no eyes at all on the outside of its head. When it opened its mouth, Mr Gardner said, it seemed to become more aware of its surroundings. He said that it was taken for examination to the Veterinary Department of Guelph University, but I have not so far discovered any full report on it. Such unfortunate freaks are interesting because they often give us clues {96}

  Figure 3.2 Macro-mutations do happen. This freak toad with eyes in the roof of Its mouth is said to have been found surviving wild in a Canadian garden. This photograph was originally published in a local newspaper, The Hamilton Spectator.

  about how embryonic development happens normally. Not all human birth defects are genetic, for instance those caused by thalidomide, but many are. A simple dominant gene causes achondroplasia, a seyere ihortening of the limb bones resulting in low stature and unusual proportions. Mutations of large effect like this — ‘macro-mutations’ — are sometimes called saltations. The achondroplasia gene is Usually inherited from one parent, but it very occasionally springs up ipontaneously by mutation, and this is how it must arise originally. A limilarly dramatic mutation could, in theory though I very much doubt it in practice, have given rise to an abrupt and sudden nose extension from tapir length to elephant length in a single generation.

  Coming to the second question, of whether, once a large ‘freak’ piacro-mutation had arisen, natural selection would ever favour it, you pight think that it is not the sort
of question that has a general answer. Doesn't it vary among particular cases, say, yes for achondroplasia, {97} no for two-headed calves? The dog equivalent of the achondro-plasia gene has in fact been positively favoured in artificial selection by human breeders, not just to service idle whims but to produce useful, working dogs. Dachshunds were bred to go down badger sets, and a significant part of the genetic sculpting that led to the breed was the incorporation of the achondroplasia gene. Perhaps it sometimes happens in nature that major mutations, like achondroplasia, suddenly open up a new way of life or a new diet: a dwarf animal, although heavily penalized when pursuing prey over open country, suddenly discovers that, unlike most of its colleagues, it can follow the prey down a hole.

  Evolutionary theorists have sometimes suggested that major saltations are incorporated into evolutionary change in nature. The famous German-American geneticist Richard Goldschmidt advocated the theory under the memorable catchphrase of the ‘hopeful monster’ theory. I'll mention one possible example in Chapter 7. But Goldschmidt's theory has never been widely supported, and there are general reasons for doubting whether macro-mutations or freaks really are important in evolution. Organisms are extremely complicated and sensitively adjusted pieces of machinery. If you take a complicated piece of machinery, even one which is not working all that well, and make a very large, random alteration to its insides, the chance that you will improve it is very low indeed. If, on the other hand, you make a very small random alteration to its insides, you may improve it. If your television aerial (antenna) is not quite properly aligned, a very slight random twist to the aerial has about a 50—50 chance of improving matters. This is because, whichever direction the aerial ought to be pointing in, there's a 50 per cent chance that your slight random twist will be in that direction. But a very large random assault on the aerial, wrenching it round violently through a very large angle, is more likely to make matters worse. This is partly because, even if your twist is in the right direction, it will probably overshoot the correct angle. More generally it is because there are so many more ways of being wrongly adjusted than of being rightly adjusted. A complicated mechanism that is working {98} at all cannot be too far from correct adjustment. A small random change may improve it; or, if it makes matters worse, it will still not move it too far from the correct arrangement. But a very large random change has the effect of sampling the gigantic set of all possible arrangements. And the vast majority of all possible arrangements are wrong.

  Even the common experience that a machine that is on the blink can often be improved by a good kick does not contradict my argument. Violent though the kick may be, the television set is a robust piece of hardware and the kick doesn't necessarily have a large effect on the arrangement of the parts. What it can do is slightly change the position of any part that is slightly loose; and this loose part is quite likely to be the very part that is causing the faulty behaviour.*

  Turning to living creatures, I wrote in The Blind Watchmaker that however many ways there may be of being alive, it is certain that there are vastly more ways of being dead. (I'd be inhuman not to confess my delight that this remark has made it into the Oxford Dictionary of Quotations!) If you think of all possible ways of arranging the bits of an animal, almost all of them would turn out to be dead; more accurately they'd mostly never be born. Each species of animal and plant is an island of workability set in a vast sea of conceivable arrangements most of which would, if they ever came into existence, die. The ocean of all possible animals includes animals with their eyes in the soles of their feet, animals with lenses in their ears instead of their eyes, animals with one left wing and one right fin; animals with skulls around their stomachs and nothing around their brains. There is no point in going on inventing. I have said enough to demonstrate that the islands of survivability, however large and however numerous they may be, are {99} minuscule in size and infinitesimal in number when compared with the ocean of dead unworkability.

  When a parent has a mutant child, the parent, being alive, must be safely ensconced on one of the islands. A small mutation — a fractional lengthening of a leg-bone here, a delicate adjustment to a jaw angle there — simply moves the child to a different part of the same island. Or it may reclaim a small offshore sandbank and join it to the dry land. But a large mutation, a drastic, freakish, revolutionary change, is equivalent to a mad leap into the wild blue yonder. The macro-mutant is catapulted in a random direction, leagues away from its home island. It is just possible that it will chance to land on another island. But since islands are so few and small, and the sea so large, the chances are very very low. It may happen very occasionally once every few million years, and when it does happen it may have a dramatic impact on the course of evolution.

  We mustn't push the metaphor of the islands too far. There is much wrong with it. All species are related to each other, which means there must be ways of travelling, through the ocean of possibilities, from any one way of being alive to every other. The metaphor of the islands does not help us here, and the metaphor of Mount Improbable is better. The islands serve the particular purpose of dramatizing the point that the more drastic and freakish the mutation the less likely it is to be favoured.

  We also need to distinguish different kinds of macro-mutation. By invoking imaginary animals with eyes in their soles and lenses in their ears, I concentrated attention on changes in arrangement of parts. Large changes of this kind are certainly very unlikely indeed to strike lucky and survive. But there are also large changes in magnitude of a part, which do not involve rearrangements of parts. A sudden shooting out of a tapir-like nose to an elephant-like trunk would be an example, if lengthening were all that happened. It is less obvious that a drastic change of this kind necessarily constitutes a leap into the ocean of impracticality or death.

  I promised that I'd return to ‘Boeing 747’ and ‘Stretched DC8’ macro-mutations. Remember Sir Fred Hoyle's debating point about {100} junkyards and 747s? He is reported to have said that the evolution, by natural selection, of a complicated structure such as a protein molecule (or, by implication, an eye or a heart) is about as likely as a hurricane's having the luck to put together a Boeing 747 when whirling through a junkyard. If he'd said ‘chance’ instead of ‘natural selection’ he'd have been right. Indeed, I regretted having to expose him as one of the many toilers under the profound misapprehension that natural selection is chance. Any theory that expects evolution to put together a new, complex machine like an eye or a haemoglobin molecule, in a single step from nothing, is asking too much of chance. On this theory, natural selection has hardly any work to do. All the ‘design’ work is being put in by mutation, a single large mutation. It is this kind of macro-mutation that deserves the metaphor of the 747 and the junkyard, and I call it a Boeing 747 macro-mutation. Boeing 747 macro-mutations do not exist and they have no connection with Darwinism.

  Turning to my other airliner analogy, the Stretched DC8 is like an Ordinary DC8 only rather longer. The fundamental design of the DC8 is all there, but an extra length of tubing has been let into the middle of the fuselage. There are also more seats, more luggage lockers and more of all the other things that repeat down the length of a plane. Equally obviously there are extra lengths of the cables, tubes and carpets that run the length of an airliner's fuselage. Slightly less obviously, there surely must be numerous consequential modifications to other parts of the plane, necessitated by the new task of lifting a greater length of the fuselage off the ground. But, fundamentally, the difference between the DC8 and the Stretched DC8 comes down to a tingle macro-mutation: the fuselage is abruptly and suddenly much longer than its predecessor. There was not a gradual series of intermediates.

  Giraffes have evolved from an ancestor rather like a modern okapi (Figure 3.3). The most conspicuous change is the elongation of the neck. Could this have come about in a single, large mutation? I hasten to say that I am sure it didn't. But that is another matter from saying that it couldn't. A Boeing 747 mutation
like a brand-new complex {101}

  Figure 3.3 Steps to a long neck. Okapi, Okapia jobnstoni, which may be similar to an ancestor of modern giraffes, with giraffe, Giraffa camelopardalis reticulata. {102}

  eye — complete with iris diaphragm and refocusable lens, springing from nothing, like Pallas Athene from the brow of Zeus — that can never happen, not in a billion billion years. But, like the stretching of the DC8, the giraffe's neck could have sprung out in a single muta-tional step (though I bet it didn't). What is the difference? It isn't that the neck is noticeably less complicated than the eye. For all I know it may be more complicated. No, what matters is the complexity of the difference between the earlier neck and the later one. This difference is slight, at least when compared with the difference between no eye and a modern eye. The giraffe's neck has the same complicated arrangement of parts as the okapi (and presumably as the giraffes own short-necked ancestor). There is the same sequence of seven vertebrae, each with its associated blood vessels, nerves, ligaments and blocks of muscle. The difference is that each vertebra is a lot longer, and all its associated parts are stretched or spaced out in proportion.

  The point is that you may only have to change one thing in the developing embryo in order to quadruple the length of the neck. Say you just have to change the rate at which the vertebral primordia grow, and everything else follows. But in order to make an eye develop from bare skin you have to change, not one rate but hundreds (see Chapter 5). If an okapi mutated to produce a giraffe's neck it would be a Stretched DC8 macro-mutation, not a 747 macro-mutation. It is therefore a possibility which need not be totally ruled out. Nothing new is added, in the way of complication. The fuselage is elongated, with all that that entails, but it is a stretching of existing Complexity, not an introduction of new complexity. The same would be true even if the giraffe had more than seven segments in its neck. The number of vertebrae in different species of snakes varies from about 200 to 350. Since all snakes are cousins of each other, and Miice vertebrae cannot come in halves or quarters, this must mean that, from time to time, a snake is born with at least one more, or one fewer, vertebra than its parents. These mutations deserve to be called macro-mutations, and they have evidently been incorporated iti evolution because all these snakes exist. They are DC8 mutations because they involve the duplication of existing complexity, not the 747 invention of new complexity. {103}