The Extended Phenotype

by Richard Dawkins

  But why does there have to be a complete set of genes in every cell of a developing body? It is surely easy to imagine a form of life in which parts of the genome are hived off during differentiation, so that a given type of tissue, liver tissue or kidney tissue say, has only the genes that it needs. Only the germ-line cells, it would seem, really need to preserve the entire genome. It may be that the reason is simply that there is no easy way, physically, to hive off parts of the genome. It is not, after all, as though the genes needed in any particular differentiated region of the developing body are all segregated on one chromosome. We could, I suppose, now ask ourselves why this had to be the case. Given that it is the case, full division of the entire genome at every cell division may simply be the easiest and most economical way of doing things. However, in the light of my parable (Chapter 9) of the rosy-spectacled Martian and the need for cynicism, the reader may be tempted to speculate further. Could it be that the total, rather than partial, duplication of the genome in mitosis is an adaptation by some genes to keep themselves in a position to oversee and thwart would-be outlaws among their colleagues? Personally I doubt it, not because the idea is inherently farfetched but because it is hard to see how a gene in the liver, say, could stand to gain from being an outlaw and manipulating the liver in a way that would be to the detriment of genes in the kidney or the spleen. Following the logic of the chapter on parasites, the interests of ‘liver genes’ and ‘kidney genes’ would overlap because they share the same germ-line and the same gametic route out of the present body.

  I have not provided a rigorous definition of the organism. It is, indeed, arguable that the organism is a concept of dubious utility, precisely because it is so difficult to define satisfactorily. From the immunological or genetic points of view, a pair of monozygotic twins would have to count as a single organism, yet clearly they would not so qualify from the point of view of the physiologist, the ethologist, or Huxley’s indivisibility criterion. What is ‘the individual’ in a colonial siphonophore or bryozoan? Botanists have good reason to be less fond of the phrase ‘individual organism’ than zoologists are: ‘The individual fruit fly, flour beetle, rabbit, flatworm or elephant is a population at the cellular but not at any higher level. Starvation does not change the number of legs, hearts or livers of an animal but the effect of stress on a plant is to alter both the rate of formation of new leaves and the rate of death of old ones: a plant may react to stress by varying the number of its parts’ (Harper 1977, pp. 20–21). To Harper, as a plant population biologist, the leaf may be a more salient ‘individual’ than ‘the plant’, since the plant is a straggling, vague entity for whom reproduction may be hard to distinguish from what a zoologist would happily call ‘growth’. Harper feels obliged to coin two new terms for different kinds of ‘individual’ in botany. ‘The “ramet” is the unit of clonal growth, the module that may often follow an independent existence if severed from the parent plant.’ Sometimes, as in strawberries, the ramet is the unit that we ordinarily call a ‘plant’. In other cases such as white clover the ramet may be the single leaf. The ‘genet’, on the other hand, is the unit which springs from one single-celled zygote, the ‘individual’ in the sense of a zoologist whose animals reproduce sexually.

  Janzen (1977) faces up to the same difficulty, suggesting that a clone of dandelions should be regarded as one ‘evolutionary individual’ (Harper’s genet), equivalent to a single tree although spread out along the ground rather than raised up in the air on a trunk, and although divided up into separate physical ‘plants’ (Harper’s ramets). According to this view, there may be as few as four individual dandelions competing with each other for the territory of the whole of North America. Janzen sees a clone of aphids in the same way. His paper has no literature citations at all, but the view is not a new one. It goes back at least as far as 1854, when T. H. Huxley ‘treated each life cycle as an individual, with all the products from sexual event to sexual event being a single unit. He even treated an asexual lineage of aphids as an individual’ (Ghiselin 1981). There is merit in this way of thinking, but I shall show that it leaves out something important.

  One way to re-express the Huxley/Janzen argument is as follows. The germ-line of a typical organism, say a human, goes through a sequence of perhaps a few dozen mitotic divisions between each meiosis. Employing Chapter 5’s ‘backwards’ way of looking at the ‘past experience of a gene’, any given gene in a living human has a history of cell divisions as follows: meiosis mitosis mitosis … mitosis meiosis. In every successive body, in parallel with the mitotic division of the germ-line, other mitotic divisions have furnished the germ-line with a large clone of ‘helper’ cells, grouped together into a body in which the germ-line is housed. In every generation the germ-line is funnelled down into a one-celled ‘bottleneck’ (a gamete followed by a zygote), then it fans out into a many-celled body, then it is funnelled down into a new bottleneck, etc. (Bonner 1974).

  The many-celled body is a machine for the production of single-celled propagules. Large bodies, like elephants, are best seen as heavy plant and machinery, a temporary resource drain, invested in so as to improve later propagule production (Southwood 1976). In a sense the germ-line would ‘like’ to reduce capital investment in heavy machinery, reduce the number of cell divisions in the growth part of the cycle, so as to reduce the interval between recurrence of the reproduction part of the cycle. But this recurrence interval has an optimal length which is different for different ways of life. Genes that caused elephants to reproduce when too young and small would propagate themselves less efficiently than alleles tending to produce an optimal recurrence interval. The optimal recurrence interval for genes that happen to find themselves in elephant gene-pools is much longer than the optimal recurrence interval for genes in mouse gene-pools. In the elephant case, more capital investment is required to be laid down before returns on investment are sought. A protozoan largely dispenses with the growth phase of the cycle altogether, and its cell divisions are all ‘reproductive’ cell divisions.

  It follows from this way of looking at organisms that the end product, the ‘goal’ of the growth phase of the cycle, is reproduction. The mitotic cell divisions which put together an elephant are all directed to the end of finally propagating viable gametes which succeed in perpetuating the germ-line. Now, holding this in mind, look at aphids. During the summer, asexual females go through repeated generations of asexual reproduction culminating in a single sexual generation which restarts the cycle. Clearly, by analogy with the elephant, it is easy to follow Janzen in seeing the summer asexual generations as all directed towards the final end of sexual reproduction in autumn. Asexual reproduction, according to this view, is not really reproduction at all. It is growth, just like the growth of a single elephant’s body. For Janzen the whole clone of female aphids is a single evolutionary individual because it is the product of a single sexual fusion. It is an unusual individual in that it happens to be split up into a number of physically separate units, but so what? Each of those physical units contains its own fragment of germ-line, but then so does the left ovary and the right ovary of a female elephant. The fragments of germ-line in the aphid case are separated by thin air, while the two ovaries of an elephant are separated by guts but, again, so what?

  Persuasive as this line of argument is, I have already mentioned that I think it misses an important point. It is right to regard most mitotic cell divisions as ‘growth’, ‘aimed’ at the final goal of reproduction, and it is right to regard the individual organism as the product of one reproductive event, but Janzen is wrong to equate the reproduction/growth distinction with the sexual/asexual distinction. There is, to be sure, an important distinction lurking here, but it is not the distinction between sex and non-sex, nor is it the distinction between meiosis and mitosis.

  The distinction that I wish to emphasize is that between germ-line cell division (reproduction), and somatic or ‘dead-end’ cell division (growth). A germ-line cell division is
one where the genes being duplicated have a chance of being the ancestors of an indefinitely long line of descendants, where the genes are, in fact, true germ-line replicators in the sense of Chapter 5. A germ-line cell division may be a mitosis or a meiosis. If we simply look at a cell division under a microscope, there may be no way of telling whether it is a germ-line division or not. Both germ-line and somatic cell divisions may be mitoses of identical appearance.

  If we look at a gene in any cell in a living organism and trace its history backwards in evolutionary time, the most recent few cell divisions of its ‘experience’ may be somatic, but once we reach a germ-line cell division in our backwards march, all previous ones in the gene’s history must be germ-line divisions. Germ-line cell divisions may be thought of as proceeding forwards in evolutionary time, while somatic cell divisions are proceeding sideways. Somatic cell divisions are used to make mortal tissues, organs and instruments whose ‘purpose’ is the promoting of germ-line cell divisions. The world is populated by genes which have survived in germ-lines as a consequence of aid that they received from their exact duplicates in somatic cells. Growth comes about through the propagation of dead-end somatic cells, while reproduction is the means of the propagation of germ-line cells.

  Harper (1977) makes a distinction between reproduction and growth in plants, which will normally amount to the same as my distinction between germ-line and somatic cell division: ‘The distinction made here between “reproduction” and “growth” is that reproduction involves the formation of a new individual from a single cell: this is usually (though not always e.g. apomicts) a zygote. In this process a new individual is “reproduced” by the information that is coded in the cell. Growth, in contrast, results from the development of organized meristems’ (Harper 1977, p. 27 fn.). What matters here is whether there really is an important biological distinction between growth and reproduction which is not the same as the distinction between mitosis and meiosis + sex. Is there really a crucial difference between ‘reproducing’ to make two aphids on the one hand, and ‘growing’ to make one aphid twice as large on the other? Janzen would presumably say no. Harper would presumably say yes. I agree with Harper, but I would not have been able to justify my position until I had read J. T. Bonner’s (1974) inspiring book On Development. The justification is best made with the aid of thought experiments.

  Imagine a primitive plant consisting of a flat, pad-like thallus, floating on the surface of the sea, absorbing nutrients through its lower surface and sunlight through its upper surface. Instead of ‘reproducing’ (i.e. sending off single-celled propagules to grow elsewhere), it simply grows at its margins, spreading into an ever larger circular green carpet, like a monstrous lily pad several miles across and still growing. Maybe older parts of the thallus eventually die, so that it consists of an expanding ring rather than a filled circle like a true lily pad. Perhaps also, from time to time, chunks of the thallus split off, like icefloes shearing away from the pack ice, and separate chunks drift to different parts of the ocean. Even if we assume this kind of fission, I shall show that it is not reproduction in an interesting sense.

  Now consider a similar kind of plant which differs in one crucial respect. It stops growing when it attains a diameter of 1 foot, and reproduces instead. It manufactures single-celled propagules, either sexually or asexually, and sheds them into the air where they may be carried a long way on the wind. When one of these propagules lands on the water surface it becomes a new thallus, which grows until it is 1 foot wide, then reproduces again. I shall call the two species G (for growth) and R (for reproduction) respectively.

  Following the logic of Janzen’s paper, we should see a crucial difference between the two species only if the ‘reproduction’ of the second species, R, is sexual. If it is asexual, the propagules shed into the air being mitotic products genetically identical to the cells of the parent thallus, there is, for Janzen, no important difference between the two species. The separate ‘individuals’ in R are no more genetically distinct than different regions of the thallus in G might be. In either species, mutation can initiate new clones of cells. There is no particular reason why, in R, mutations should occur during propagule formation any more than during thallus growth. R is simply a more fragmented version of G, just as a clone of dandelions is like a fragmented tree. My purpose in making the thought experiment, however, was to disclose an important difference between the two hypothetical species, representing the difference between growth and reproduction, even when reproduction is asexual.

  G just grows, while R grows and reproduces in alternation. Why is the distinction important? The answer cannot be a genetic one in any simple sense because, as we have seen, mutations are just as likely to initiate genetic change during growth-mitosis as during reproduction-mitosis. I suggest that the important distinction between the two species is that a lineage of R is capable of evolving complex adaptations in a way that G is not. The reasoning goes as follows.

  Consider again the past history of a gene, in this case a gene sitting in a cell of R. It has had a history of passing repeatedly from one ‘vehicle’ to another similar vehicle. Each of its successive bodies began as a single-celled propagule, then grew through a fixed cycle, then passed the gene on into a new single-celled propagule and hence a new multicellular body. Its history has been a cyclical one, and now here is the point. Since each of this long series of successive bodies developed anew from single-celled beginnings, it is possible for successive bodies to develop slightly differently from their predecessors. Evolution of complex body structure with organs, say a complex apparatus for catching insects like a Venus fly trap, is only possible if there is a cyclically repeating developmental process to evolve. I shall return to this point in a moment.

  Meanwhile, compare G. A gene sitting in a young cell at the growing margin of the huge thallus has a history which is not cyclical, or is cyclical only at the cellular level. The ancestor of the present cell was another cell, and the career of the two cells was very similar. Each cell of an R plant, by contrast, has a definite place in the growth sequence. It is either near the centre of the 1-foot thallus, or near the rim, or at some particular place in between. It can therefore differentiate to fill its special role in its appointed place in an organ of the plant. A cell of G has no such specific developmental identity. All cells first appear at the growing margin, and later find themselves enclosed by other, younger cells. There is cyclicity only at the cellular level, which means that in G evolutionary change can take place only at the cellular level. Cells might improve on their predecessors in the cell lineage, developing more complex internal organelle structure, say. But the evolution of organs and adaptations at the multicellular level could not take place, because recurrent, cyclical development of whole groups of cells does not occur. It is, of course, true that in G the cells and their ancestors are in physical contact with other cells, and in this sense form a multicellular ‘structure’. But as far as putting together complex multicellular organs is concerned, they might just as well have been free-swimming protozoa.

  In order to put together a complex multicellular organ you need a complex developmental sequence. A complex developmental sequence has to have evolved from an earlier developmental sequence which was slightly less complex. There has to be an evolutionary progression of developmental sequences, each one in the series being a slight improvement on its predecessor. G does not have a recurring developmental sequence other than the high-frequency cycle of development at the single-cell level. Therefore it cannot evolve multicellular differentiation and organ-level complexity. To the extent that it can be said to have a multicellular developmental process at all, that development continues non-cyclically through geological time: the species makes no separation between the growth time-scale and the would-be evolution time-scale. The only high-frequency developmental cycle available to it is the cell cycle. R, on the other hand, has a multicellular developmental cycle which is fast compared with evolutionary time. Theref
ore, as the ages succeed, later developmental cycles can be different from earlier developmental cycles, and multicellular complexity can evolve. We are moving towards a definition of the organism as the unit which is initiated by a new act of reproduction via a single-celled developmental ‘bottleneck’.

  The importance of the difference between growth and reproduction is that each act of reproduction involves a new developmental cycle. Growth simply involves swelling of the existing body. When an aphid gives rise to a new aphid by parthenogenetic reproduction the new aphid, if it is a mutant, may be radically different from its predecessor. When an aphid grows to twice its original size, on the other hand, all its organs and complex structures simply swell. It might be said that somatic mutations could occur within cell lineages of the growing giant aphid. True, but a mutation within a somatic cell line in a heart, say, cannot radically re-organize the structure of the heart. To switch the example to vertebrates, if the present heart is two-chambered, with one auricle feeding one ventricle, new mutations in the mitotic cells at the growing margin of the heart are very unlikely to achieve radical restructuring of the heart so that it comes to have four chambers with a pulmonary circulation kept separate from the rest. In order to put together new complexity, new developmental beginnings are required. A new embryo must start from scratch, without any heart at all. Then a mutation can act on sensitive key points in early development to bring about a new fundamental architecture of the heart. Developmental recycling allows a return ‘back to the drawing board’ (see below) in every generation.