The Blind Watchmaker (40 page)

Macromutations mutations of large effect - undoubtedly occur. What is at issue is not whether they occur but whether they play a role in evolution; whether, .in other words, they are incorporated into the gene pool of a species, or whether, on the contrary, they are always eliminated by natural selection. A famous example of a macromutation is ‘antennapaedia’ in fruitflies. In a normal insect the antennae have something in common with the legs, and they develop in the embryo in a similar way. But the differences are striking as well, and the two sorts of limb are used for very different purposes: the legs for walking; the antennae for feeling, smelling and otherwise sensing things. Antennapaedic flies are freaks in which the antennae develop just like legs. Or, another way of putting it, they are flies that have no antennae but an extra pair of legs, growing out of the sockets where the antennae ought to be. This is a true mutation in that it results from an error in the copying of DNA. And it breeds true if antennapaedic flies are cosseted in the laboratory so that they survive long enough to breed at all. They would not survive long in the wild, as their movements are clumsy and their vital senses are impaired.

So, macromutations do happen. But do they play a role in evolution? People called saltationists believe that macromutations are a means by which major jumps in evolution could take place in a single generation. Richard Goldschmidt, whom we met in Chapter 4, was a true saltationist. If saltationism were true, apparent ‘gaps’ in the fossil record needn’t be gaps at all. For example, a saltationist might believe that the transition from sloping-browed
Australopithecus to
domebrowed
Homo sapiens
took place in a single macromutational step, in a single generation. The difference in form between the two species is probably less thap the difference between a normal and an antennapaedic fruitfly, and it is theoretically conceivable that the first
Homo sapiens
was a freak child - probably an ostracized and persecuted one - of two normal
Australopithecus
parents.

There are very good reasons for rejecting all such saltationist theories of evolution. One rather boring reason is that if a new species really did arise in a single mutational step, members of the new species might have a hard time finding mates. But I find this reason less telling and interesting than two others which have already been foreshadowed in our discussion of why major jumps across Biomorph Land are to be ruled out. The first of these points was put by the great statistician and biologist R.A.Fisher, whom we met in other connections in previous chapters. Fisher was a stalwart opponent of all forms of saltationism, at a time when saltationism was much more fashionable than it is today, and he used the following analogy. Think, he said, of a microscope which is almost, but not quite perfectly, in focus and otherwise well adjusted for distinct vision. What are the odds that, if we make some random change to the state of the microscope (corresponding to a mutation), we shall improve the focus and general quality of the image? Fisher said:

It is sufficiently obvious that any large derangement will have a very small probability of improving the adjustment, while in the case of alterations much less than the smallest of those intentionally effected by the maker or the operator, the chance of improvement should be almost exactly one half.

I have already remarked that what Fisher found ‘easy to see’ could place formidable demands on the mental powers of ordinary scientists, and the same is true of what Fisher thought was ‘sufficiently obvious’. Nevertheless, further cogitation almost always shows him to have been right, and in this case we can prove it to our own satisfaction without too much difficulty. Remember that we are assuming the microscope to be almost in correct focus before we start. Suppose that the lens is slightly lower than it ought to be for perfect focus, say a tenth of an inch too close to the slide. Now if we move it a small amount, say a hundredth of an inch, in a random direction, what are the odds that the focus will improve? Well, if we happen to move it
down
a hundredth of an inch, the focus will get worse. If we happen to move it
up
a hundredth of an inch, the focus will get better. Since we are moving it in a random direction, the chance of each of these two eventualities is one half. The smaller the movement of adjustment, in relation to the initial error, the closer will the chance of improvement approach one half. That completes the justification of the second part of Fisher’s statement.

But now, suppose we move the microscope tube a large distance equivalent to a macromutation - also in a random direction; suppose we move it a full inch. Now it doesn’t matter which direction we move it in, up or down, we shall still make the focus worse than it was before. If we chance to move it down, it will now be one and one-tenth inches away from its ideal position (and will probably have crunched through the slide). If we chance to move it up, it will now be ninetenths of an inch away from its ideal position. Before the move, it was only one-tenth of an inch away from its ideal position so, either way, our ‘macromutational’ big move has been a bad thing. We have done the calculation for a very big move (‘macromutation’) and a very small move (‘micromutation’). We can obviously do the same calculation for a range of intermediate sizes of move, but there is no point in doing so. I think it really will now be sufficiently obvious that the smaller we make the move, the closer we shall approach the extreme case in which the odds of an improvement are one-half, and the larger we make the move, the closer we shall approach the extreme case in which the odds of an improvement are zero.

The reader will have noticed that this argument depends upon the initial assumption that the microscope was already pretty close to being in focus before we even started making random adjustments. If the microscope starts 2 inches out of focus, then a random change of 1 inch has a 50 per cent chance of being an improvement, just as a random change of onehundredth of an inch has. In this case the ‘macromutation’ appears to have the advantage that it moves the microscope into focus more quickly. Fisher’s argument will, of course, apply here to ‘megamutations’ of, say, 6 inches movement in a random direction.

Why, then, was Fisher allowed to make his initial assumption that the microscope was nearly in focus at the start? The assumption flows from the role of the microscope in the analogy. The microscope after its random adjustment stands for a mutant animal. The microscope before its random adjustment stands for the normal, unmutated parent of the supposed mutant animal. Since it is a parent, it must have survived long enough to reproduce, and therefore it cannot be all that far from being well-adjusted. By the same token, the microscope before the random jolt cannot be all that far from being in focus, or the animal that it stands for in the analogy couldn’t have survived at all. It is only an analogy, and there is no point in arguing over whether ‘all that far’ means an inch or a tenth of an inch or a thousandth of an inch. The important point is that if we consider mutations of ever-increasing magnitude, there will come a point when, the larger the mutation is, the less likely it is to be beneficial; while if we consider mutations of ever-decreasing magnitude, there will come a point when the chance of a mutation’s being beneficial is 50 per cent.

The argument over whether macromutations such as antennapaedia could ever be beneficial (or at least could avoid being harmful), and therefore whether they could give rise to evolutionary change, therefore turns on
how
‘macro’ the mutation is that we are considering. The more ‘macro’ it is, the more likely it is to be deleterious, and the less likely it is to be incorporated in the evolution of a species. As a matter of fact, virtually all the mutations studied in genetics laboratories - which are pretty macro because otherwise geneticists wouldn’t notice them - are deleterious to the animals possessing them (ironically I’ve met people who think that this is an argument
against
Darwinism!). Fisher’s microscope argument, then, provides one reason for scepticism about ‘saltation’ theories of evolution, at least in their extreme form.

The other general reason for not believing in true saltation is also a statistical one, and its force also depends quantitatively on
how
macro is the macromutation we are postulating. In this case it is concerned with the complexity of evolutionary changes. Many, though not all, of the evolutionary changes we are interested in are advances in complexity of design. The extreme example of the eye, discussed in earlier chapters, makes the point clear. Animals with eyes like ours evolved from ancestors with no eyes at all. An extreme saltationist might postulate that the evolution took place in a single mutational step. A parent had no eye at all, just bare skin where the eye might be. He had a freak offspring with a fully developed eye, complete with variable focus lens, iris diaphragm for ‘stopping down’, retina with millions of three-colour photocells, all with nerves correctly connected up in the brain to provide him with correct, binocular, stereoscopic colour vision.

In the biomorph model we assumed that this kind of multidimensional improvement could not occur. To recapitulate on why that was a reasonable assumption, to make an eye from nothing you need not just one improvement but a large number of improvements. Any one of these improvements is pretty improbable by itself, but not so improbable as to be impossible. The greater the number of simultaneous improvements we consider, the more improbable is their simultaneous occurrence. The coincidence of their simultaneous occurrence is equivalent to leaping a large distance across Biomorph Land, and happening to land on one particular, predesignated spot. If we choose to consider a sufficiently large number of improvements, their joint occurrence becomes so improbable as to be, to all intents and purposes, impossible. The argument has already been sufficiently made, but it may be helpful to draw a distinction between two kinds of hypothetical macromutation, both of which
appear
to be ruled out by the complexity argument but only one of which, in fact,
is
ruled out by the complexity argument. I label them, for reasons that will become clear, Boeing 747 macromutations and Stretched DC8 macromutations.

Boeing 747 macromutations are the ones that really are ruled out by the complexity argument just given. They get their name from the astronomer Sir Fred Hoyle’s memorable misunderstanding of the theory of natural selection. He compared natural selection, in its alleged improbability, to a hurricane blowing through a junkyard and chancing to assemble a Boeing 747. As we saw in Chapter 1, this is an entirely false analogy to apply to natural selection, but it is a very good analogy for the idea of certain kinds of macromutation giving rise to evolutionary change. Indeed, Hoyle’s fundamental error was that he, in effect, thought (without realizing it) that the theory of natural selection
did
depend upon macromutation. The idea of a single macromutation’s giving rise to a fully functioning eye with the properties listed above, where there was only bare skin before, is, indeed, just about as improbable as a hurricane assembling a Boeing 747. This is why I refer to this kind of hypothetical macromutation as a Boeing 747 macromutation.

Stretched DC8 macromutations are mutations that, although they may be large in the magnitude of their effects, turn out not to be large in terms of their complexity. The Stretched DC8 is an airliner that was made by modifying an earlier airliner, the DC8. It is like a DC8, but with an elongated fuselage. It was an improvement at least from one point of view, in that it could carry more passengers than the original DC8. The stretching is a large increase in length, and in that sense is analogous to a macromutation. More interestingly, the increase in length is, at first sight, a complex one. To elongate the fuselage of an airliner, it is not enough just to insert an extra length of cabin tube. You also have to elongate countless ducts, cables, air tubes and electric wires. You have to put in lots more seats, ashtrays, reading lights, 12-channel music selectors and fresh-air nozzles. At first sight there seems to be much more complexity in a Stretched DC8 than there is in an ordinary DC8, but is there really? The answer is no, at least to the extent that the ‘new’ things in the stretched plane are just ‘more of the same’. The biomorphs of Chapter 3 frequently show macromutations of the Stretched DC8 variety.

What has this to do with mutations in real animals? The answer is that some real mutations cause large changes that are very like the change from DC8 to Stretched DC8, and some of these, although in a sense ‘macro’ mutations, have definitely been incorporated in evolution. Snakes, for instance, all have many more vertebrae than their ancestors. We could be sure of this even if we didn’t have any fossils, because snakes have many more vertebrae than their surviving relatives. Moreover, different species of snakes have different numbers of vertebrae, which means that vertebral number must have changed in evolution since their common ancestor, and quite often at that.

Now, to change the number of vertebrae in an animal, you need to do more than just shove in an extra bone. Each vertebra has, associated with it, a set of nerves, a set of blood vessels, a set of muscles etc., just as each row of seats in an airliner has a set of cushions, a set of head rests, a set of headphone sockets, a set of reading-lights with their associated cables
etc.
The middle part of the body of a snake, like the middle part of the body of an airliner, is composed of a number of
segments
, many of which are exactly like each other, however complex they all individually may be. Therefore, in order to add new segments, all that has to be done is a simple process of duplication. Since there already exists genetic machinery for making one snake segment - genetic machinery of great complexity, which took many generations of step-by-step, gradual evolution to build up new identical segments may easily be added by a single mutational step. If we think of genes as ‘instructions to a developing embryo’, a gene for inserting extra segments may read, simply, ‘more of the same here’. I imagine that the instructions for building the first Stretched DC8 were somewhat similar.