The Greatest Show on Earth

by Richard Dawkins

  A Mendelian gene is an all-or-nothing entity. When you were conceived, what you received from your father was not a substance, to be mixed with what you received from your mother as if mixing blue paint and red paint to make purple. If this were really how heredity worked (as people vaguely thought in Darwin’s time) we’d all be a middling average, halfway between our two parents. In that case, all variation would rapidly disappear from the population (no matter how assiduously you mix purple paint with purple paint, you’ll never reconstitute the original red and blue). In fact, of course, anybody can plainly see that there is no such intrinsic tendency for variation to decrease in a population. Mendel showed that this is because when paternal genes and maternal genes are combined in a child (he didn’t use the word ‘gene’, which wasn’t coined until 1909), it is not like blending paints, it is more like shuffling and reshuffling cards in a pack. Nowadays, we know that genes are lengths of DNA code, not physically separate like cards, but the principle remains valid. Genes don’t blend; they shuffle. You could say they are shuffled badly, with groups of cards sticking together for several generations of shuffling before chance happens to split them.

  Any one of your eggs (or sperms if you are male) contains either your father’s version of a particular gene or your mother’s version, not a blend of the two. And that particular gene came from one and only one of your four grandparents; and from one and only one of your eight great-grandparents.*

  Hindsight says this should have been obvious all along. When you cross a male with a female, you expect to get a son or a daughter, not a hermaphrodite.† Hindsight says anybody in an armchair could have generalized the same all-or-none principle to the inheritance of each and every characteristic. Fascinatingly, Darwin himself was glimmeringly close to this, but he stopped just short of making the full connection. In 1866 he wrote, in a letter to Alfred Wallace:

  My dear Wallace

  I do not think you understand what I mean by the non-blending of certain varieties. It does not refer to fertility. An instance will explain. I crossed the Painted Lady and Purple sweet peas, which are very differently coloured varieties, and got, even out of the same pod, both varieties perfect but none intermediate. Something of this kind, I should think, must occur at first with your butterflies . . . Though these cases are in appearance so wonderful, I do not know that they are really more so than every female in the world producing distinct male and female offspring.

  Darwin came that close to discovering Mendel’s law of the non-blending of (what we would now call) genes.* The case is analogous to the claim, by various aggrieved apologists, that other Victorian scientists, for example Patrick Matthew and Edward Blyth, had discovered natural selection before Darwin did. In a sense that is true, as Darwin acknowledged, but I think the evidence shows that they didn’t understand how important it is. Unlike Darwin and Wallace, they didn’t see it as a general phenomenon with universal significance – with the power to drive the evolution of all living things in the direction of positive improvement. In the same way, this letter to Wallace shows that Darwin got tantalizingly close to grasping the point about the non-blending nature of heredity. But he didn’t see its generality, and in particular he failed to see it as the answer to the riddle of why variation didn’t automatically disappear from populations. That was left to twentieth-century scientists, building on Mendel’s before-his-time discovery.†

  So now the concept of the gene pool starts to make sense. A sexually reproducing population, such as, say, all the rats on Ascension Island, remotely isolated in the South Atlantic, is continually shuffling all the genes on the island. There is no intrinsic tendency for each generation to become less variable than the previous generation, no tendency towards ever more boringly grey, middling intermediates. The genes remain intact, shuffled about from individual body to individual body as the generations go by, but not blending with one another, never contaminating each other. At any one time, the genes are all sitting in the bodies of individual rats, or they are moving into new rat bodies via sperms. But if we take a long view across many generations, we see all the rat genes on the island being mixed up as though they were cards in a single well-shuffled pack: one single pool of genes.

  I’m guessing that the rat gene pool on a small and isolated island such as Ascension is a self-contained and rather well-stirred pool, in the sense that the recent ancestors of any one rat could have lived anywhere on the island, but probably not anywhere other than on the island, give or take the occasional stowaway on a ship. But the gene pool of the rats on a large land mass such as Eurasia would be much more complicated. A rat living in Madrid would derive most of its genes from ancestors living in the western end of the Eurasian continent rather than, say, Mongolia or Siberia, not because of specific barriers to gene flow (though those exist too) but because of the sheer distances involved. It takes time for sexual shuffling to work a gene from one side of a continent to the other. Even if there are no physical barriers such as rivers or mountain ranges, gene flow across such a large land mass will still be slow enough for the gene pool to deserve the name ‘viscous’. A rat living in Vladivostok would trace most of its genes back to ancestors in the east. The Eurasian gene pool would be shuffled, as on Ascension Island, but not homogeneously shuffled because of the distances involved. Moreover, partial barriers such as mountain ranges, large rivers or deserts would further get in the way of homogeneous shuffling, thereby structuring and complicating the gene pool. These complications don’t devalue the idea of the gene pool. The perfectly stirred gene pool is a useful abstraction, like a mathematician’s abstraction of a perfect straight line. Real gene pools, even on small islands like Ascension, are imperfect approximations, only partially shuffled. The smaller and less broken-up the island, the better the approximation to the abstract ideal of the perfectly stirred gene pool.

  Just to round off the thought about gene pools, each individual animal that we see in a population is a sampling of the gene pool of its time (or rather its parents’ time). There is no intrinsic tendency in gene pools for particular genes to increase or decrease in frequency. But when there is a systematic increase or decrease in the frequency with which we see a particular gene in a gene pool, that is precisely and exactly what is meant by evolution. The question, therefore, becomes: why should there be a systematic increase or decrease in a gene’s frequency? That, of course, is where things start to get interesting, and we shall come to it in due course.

  Something funny happens to the gene pools of domestic dogs. Breeders of pedigree Pekineses or Dalmatians go to elaborate lengths to stop genes crossing from one gene pool to another. Stud books are kept, going back many generations, and miscegenation is the worst thing that can happen in the book of a pedigree breeder. It is as though each breed of dog were incarcerated on its own little Ascension Island, kept apart from every other breed. But the barrier to interbreeding is not blue water but human rules. Geographically the breeds all overlap, but they might as well be on separate islands because of the way their owners police their mating opportunities. Of course, from time to time the rules are broken. Like a rat stowing away on a ship to Ascension Island, a whippet bitch, say, escapes the leash and mates with a spaniel. But the mongrel puppies that result, however loved they may be as individuals, are cast off the island labelled Pedigree Whippet. The island itself remains a pure whippet island. Other pure-bred whippets ensure that the gene pool of the virtual island labelled Whippet continues uncontaminated. There are hundreds of man-made ‘islands’, one for each breed of pedigree dog. Each one is a virtual island, in the sense that it is not geographically localized. Pedigree whippets or Pomeranians are to be found in many different places around the world, and cars, ships and planes are used to ferry the genes from one geographical place to another. The virtual genetic island that is the Pekinese gene pool overlaps geographically, but not genetically (except when a bitch breaks cover), with the virtual genetic island that is the boxer gene pool and the virtual island t
hat is the St Bernard gene pool.

  Now let’s return to the remark that opened my discussion of gene pools. I said that if human breeders are to be seen as sculptors, what they are carving with their chisels is not dog flesh but gene pools. It appears to be dog flesh because the breeder might announce an intention to, say, shorten the snouts of future generations of boxers. And the end product of such an intention would indeed be a shorter snout, as though a chisel had been taken to the ancestor’s face. But, as we have seen, a typical boxer in any one generation is a sampling of the contemporary gene pool. It is the gene pool that has been carved and whittled over the years. Genes for long snouts have been chiselled out of the gene pool and replaced by genes for short snouts. Every breed of dog, from dachshund to Dalmatian, from boxer to borzoi, from poodle to Pekinese, from Great Dane to chihuahua, has been carved, chiselled, kneaded, moulded, not literally as flesh and bone but in its gene pool.

  It isn’t all done by carving. Many of our familiar breeds of dog were originally derived as hybrids of other breeds, often quite recently, for example in the nineteenth century. Hybridization, of course, represents a deliberate violation of the isolation of the gene pools on virtual islands. Some hybridization schemes are designed with such care that the breeders would resent their products being described as mongrels or mutts (as President Obama delightfully described himself). The ‘Labradoodle’ is a hybrid between a standard poodle and a Labrador retriever, the result of a carefully crafted quest for the best virtues of both breeds. Labradoodle owners have established societies and associations just like those of pure-bred pedigree dogs. There are two schools of thought in the Labradoodle Fancy, and those of other such designer hybrids. There are those who are happy to go on making Labradoodles by mating poodles and Labradors together. And there are those who are trying to initiate a new Labradoodle gene pool that will breed true, when Labradoodles are mated together. At present, second-generation Labradoodle genes recombine to produce more variety than pure-bred pedigree dogs are supposed to show. This is how many ‘pure’ breeds got their start: they went through an intermediate stage of high variation, subsequently trimmed down through generations of careful breeding.

  Sometimes, new breeds of dog get their start with the adoption of a single major mutation. Mutations are the random changes in genes that constitute the raw material for evolution by non-random selection. In nature, large mutations seldom survive, but geneticists like them in the laboratory because they are easy to study. Breeds of dog with very short legs, like basset hounds and dachshunds, acquired them in a single step with the genetic mutation called achondroplasia, a classic example of a large mutation that would be unlikely to survive in nature. A similar mutation is responsible for the commonest kind of human dwarfism: the trunk is of nearly normal size, but the legs and arms are short. Other genetic routes produce miniature breeds that retain the proportions of the original. Dog breeders can achieve changes in size and shape by selecting combinations of a few major mutations such as achondroplasia and lots of minor genes. Nor do they need to understand the genetics in order to achieve change effectively. Without any understanding at all, just by choosing who mates with whom, you can breed for all kinds of desired characteristics. This is what dog breeders, and animal and plant breeders generally, achieved for centuries before anybody understood anything about genetics. And there’s a lesson in that about natural selection, for nature, of course, has no understanding or awareness of anything at all.

  The American zoologist Raymond Coppinger makes the point that puppies of different breeds are much more similar to each other than adult dogs are. Puppies can’t afford to be different, because the main thing they have to do is suck,* and sucking presents pretty much the same challenges for all breeds. In particular, in order to be good at sucking, a puppy can’t have a long snout like a borzoi or a retriever. That’s why all puppies look like pugs. You could say that an adult pug is a puppy whose face didn’t properly grow up. Most dogs, after they are weaned, develop a relatively longer snout. Pugs, bulldogs and Pekineses don’t; they grow in other departments, while the snout retains its infantile proportions. The technical term for this is neoteny, and we’ll meet it again when we come on to human evolution in Chapter 7.

  If an animal grows at the same rate in all its parts, so that the adult is just a uniformly inflated replica of the infant, it is said to grow isometrically. Isometric growth is quite rare. In allometric growth, by contrast, different parts grow at different rates. Often, the rates of growth of different parts of an animal bear some simple mathematical relation to each other, a phenomenon that was investigated especially by Sir Julian Huxley in the 1930s. Different breeds of dog achieve their different shapes by means of genes that change the allometric growth relationships between the parts of the body. For example, bulldogs get their Churchillian scowl from a genetic tendency towards slower growth of the nasal bones. This has knock-on effects on the relative growth of the surrounding bones, and indeed all the surrounding tissues. One of these knock-on effects is that the palate is pulled up into an awkward position, so the bulldog’s teeth stick out and it has a tendency to dribble. Bulldogs also have breathing difficulties, which are shared by Pekineses. Bulldogs even have difficulty being born because the head is disproportionately big. Most if not all the bulldogs you see today were born by caesarian section.

  Borzois are the opposite. They have extra long snouts. Indeed, they are unusual in that the elongation of the snout begins before they are born, which probably makes borzoi puppies less proficient suckers than other breeds. Coppinger speculates that the human desire to breed borzois for long snouts has reached a limit imposed by the survival capacity of puppies trying to suck.

  What lessons do we learn from the domestication of the dog? First, the great variety among breeds of dogs, from Great Danes to Yorkies, from Scotties to Airedales, from ridgebacks to dachshunds, from whippets to St Bernards, demonstrates how easy it is for the non-random selection of genes – the ‘carving and whittling’ of gene pools – to produce truly dramatic changes in anatomy and behaviour, and so fast. Surprisingly few genes may be involved. Yet the changes are so large – the differences between breeds so dramatic – that you might expect their evolution to take millions of years instead of just a matter of centuries. If so much evolutionary change can be achieved in just a few centuries or even decades, just think what might be achieved in ten or a hundred million years.

  Viewing the process over centuries, it is no empty fancy that human dog breeders have seized dog flesh like modelling clay and pushed it, pulled it, kneaded it into shape, more or less at will. Of course, as I pointed out earlier, we have really been kneading not dog flesh but dog gene pools. And ‘carved’ is a better metaphor than ‘kneaded’. Some sculptors work by taking a lump of clay and kneading it into shape. Others take a lump of stone or wood, and carve it by subtracting bits with a chisel. Obviously dog fanciers don’t carve dogs into shape by subtracting bits of dog flesh. But they do something close to carving dog gene pools by subtraction. It is more complicated than pure subtraction, however. Michelangelo took a single chunk of marble, and then subtracted marble from it to reveal David lurking inside. Nothing was added. Gene pools, on the other hand, are continually added to, for example by mutation, while at the same time non-random death subtracts. The analogy to sculpture breaks down here, and should not be pushed too tenaciously, as we’ll see again in Chapter 8.

  The idea of sculpture calls to mind the over-muscled physiques of human body-builders, and non-human equivalents such as the Belgian Blue breed of cattle. This walking beef factory has been contrived via a particular genetic alteration called ‘double muscling’. There is a substance called myostatin, which limits muscle growth. If the gene that makes myostatin is disabled, muscles grow larger than usual. It is quite often the case that a given gene can mutate in more than one way to produce the same outcome, and indeed there are various ways in which the myostatin-producing gene can be disabled, with the same
effect. Another example is the breed of pig called the Black Exotic, and there are individual dogs of various breeds that show the same exaggerated musculature for the same reason. Human body-builders achieve a similar physique by an extreme regime of exercise, and often by the use of anabolic steroids: both environmental manipulations that mimic the genes of the Belgian Blue and the Black Exotic. The end result is the same, and that is a lesson in itself. Genetic and environmental changes can produce identical outcomes. If you wanted to rear a human child to win a body-building contest and you had a few centuries to spare, you could start by genetic manipulation, engineering exactly the same freak gene as characterizes Belgian Blue cattle and Black Exotic pigs. Indeed, there are some humans known to have deletions of the myostatin gene, and they tend to be abnormally well muscled. If you started with a mutant child and made it pump iron as well (presumably the cattle and pigs could not be cajoled into this), you could probably end up with something more grotesque than Mr Universe.

  Political opposition to eugenic breeding of humans sometimes spills over into the almost certainly false assertion that it is impossible. Not only is it immoral, you may hear it said, it wouldn’t work. Unfortunately, to say that something is morally wrong, or politically undesirable, is not to say that it wouldn’t work. I have no doubt that, if you set your mind to it and had enough time and enough political power, you could breed a race of superior body-builders, or high-jumpers, or shot-putters; pearl fishers, sumo wrestlers, or sprinters; or (I suspect, although now with less confidence because there are no animal precedents) superior musicians, poets, mathematicians or wine-tasters. The reason I am confident about selective breeding for athletic prowess is that the qualities needed are so similar to those that demonstrably work in the breeding of racehorses and carthorses, of greyhounds and sledge dogs. The reason I am still pretty confident about the practical feasibility (though not the moral or political desirability) of selective breeding for mental or otherwise uniquely human traits is that there are so few examples where an attempt at selective breeding in animals has ever failed, even for traits that might have been thought surprising. Who would have thought, for example, that dogs could be bred for sheep-herding skills, or ‘pointing’, or bull-baiting?