Hamilton is here using what he earlier described as ‘the extended meaning of inclusive fitness’ (Hamilton 1964b, p. 25). The conventional meaning of inclusive fitness, the meaning upon which Hamilton himself based his detailed mathematics, is incapable of handling the green-beard effect and, indeed, outlaws such as meiotic-drive genes. This is because it is firmly tied to the idea of the individual organism as ‘vehicle’ or ‘maximizing entity’. Outlaw genes demand to be treated as selfish maximizing entities in their own right, and they constitute a strong weapon in the case against the ‘selfish organism’ paradigm. Nowhere is this better exemplified than in Hamilton’s own ingenious extensions of Fisher’s sex ratio theory (Hamilton 1967).
The green-beard thought experiment is instructive in further ways. Anyone who thinks about genes literally as molecular entities is in danger of being misled by passages like ‘What is the selfish gene? It is not just one single physical bit of DNA … It is all replicas of a particular bit of DNA, distributed throughout the world … a distributed agency, existing in many different individuals at once … a gene might be able to assist replicas of itself which are sitting in other bodies.’ The whole of kin selection theory rests on this general premise, yet it would be mystical and wrong to think that genes assist copies of each other because those copies are identical molecules to themselves. A green-beard thought experiment helps to explain this. Chimpanzees and gorillas are so similar that a gene in one species might be physically identical in its molecular details to a gene in the other. Is this molecular identity a sufficient reason to expect selection to favour genes in one species ‘recognizing’ copies of themselves in the other species, and extending them a helping hand? The answer is no, although a naive application of ‘selfish gene’ reasoning at the molecular level might lead us to think otherwise.
Natural selection at the gene level is concerned with competition among alleles for a particular chromosomal slot in a shared gene-pool. A green-beard gene in the chimpanzee gene pool is not a candidate for a slot on any gorilla chromosome, and nor is any of its alleles. It is therefore indifferent to the fate of its structurally identical counterpart in the gorilla gene-pool. (It might not be indifferent to the fate of its phenotypically identical counterpart in the gorilla gene-pool, but that has nothing to do with molecular identity.) As far as the present argument is concerned, chimp genes and gorilla genes are not copies of each other in an important sense. They are copies of each other only in the trivial sense that they happen to have the identical molecular structure. The unconscious, mechanical laws of natural selection give us no grounds for expecting them to assist molecular copies, just because they are molecular copies.
Conversely, we might be justified in expecting to see genes assisting molecularly different alleles at their own locus within a species gene-pool, provided they had the same phenotypic effects. A phenotypically neutral mutation at a locus changes molecular identity but does nothing to weaken any selection there may be in favour of mutual assistance. Green-beard altruism could still increase the incidence of green-beard phenotypes in the population, even though genes were assisting other genes that were not strict copies of themselves in the molecular sense. It is the incidence of phenotypes that we are interested in explaining, not the incidence of molecular configurations of DNA. And if any reader thinks that last remark contradicts my basic thesis, I must have failed to make my basic thesis clear!
Let me use the green beard in one more instructive thought experiment, to clarify the theory of reciprocal altruism. I have called the green-beard effect implausible, with the possible exception of the sex chromosome special cases. But there is another special case which could conceivably have some counterpart in reality. Imagine a gene programming the behavioural rule: ‘If you see another individual performing an altruistic act, remember the incident, and if the opportunity arises behave altruistically towards that individual in future’ (Dawkins 1976a, p. 96). This may be called the ‘altruism-recognition effect.’ Using the legendary example of Haldane’s (1955) jumping into the river to save a drowning person, the gene that I am postulating might spread because it was, in effect, recognizing copies of itself. It is, in fact, a kind of green-beard gene. Instead of using an incidental pleiotropic recognition character such as a green beard, it uses a non-incidental one: the behaviour pattern of altruistic rescuing itself. Rescuers tend to save only others who have rescued somebody in their time, so the gene tends to save copies of itself (setting aside problems of how the system could get started, etc.). My point in bringing this hypothetical example up here is to emphasize its distinctness from two other, superficially similar cases. The first is the one Haldane himself was illustrating, the saving of close kin; thanks to Hamilton we now understand this well. The second is reciprocal altruism (Trivers 1971). Any resemblance between true reciprocal altruism and the hypothetical altruism-recognition case I am now discussing is purely accidental (Rothstein 1981). The resemblance, however, sometimes muddles students of the theory of reciprocal altruism, which is why I am making use of green-beard theory to dispel the muddle.
In true reciprocal altruism, the ‘altruist’ stands to gain in the future from the presence of the individual beneficiary of his altruism. The effect works even if the two share no genes, and even (contra Rothstein 1981) if they belong to different species, as in Trivers’s example of the mutualism between cleaner fish and their clients. Genes mediating such reciprocal altruism benefit the rest of the genome no less than they benefit themselves, and are clearly not outlaws. They are favoured by ordinary, familiar natural selection, albeit some people (e.g. Sahlins 1977, pp. 85–87) seem to have difficulty in understanding the principle, apparently because they overlook the frequency-dependent nature of the selection and the consequent need to think in game-theoretic terms (Dawkins 1976a, pp. 197–201; Axelrod and Hamilton 1981). The altruism-recognition effect is fundamentally different, though superficially similar. There is no need for the altruism recognizer individual to repay a good turn done to himself. He simply recognizes good deeds done to anybody, and singles out the altruist for his own later favours.
It would be impossible to give a sensible account of outlaws in terms of individuals maximizing their fitness. That is the reason for giving them prominence in this book. At the beginning of the chapter I divided outlaws into ‘allelic outlaws’ and ‘laterally spreading outlaws’. All the suggested outlaws we have so far considered have been allelic: they are favoured over their alleles at their own loci, while being opposed by modifiers at other loci. I now turn to laterally spreading outlaws. These are outlaws unruly enough to break away altogether from the discipline of allelic competition within the confines of a locus. They spread to other loci, even create new loci for themselves by increasing the size of the genome. They are conveniently discussed under the heading ‘Selfish DNA’, a catch-phrase which has recently acquired currency in the pages of Nature. This will be the subject of the first part of the next chapter.
9 Selfish DNA, Jumping Genes, and a Lamarckian Scare
This chapter will be a somewhat miscellaneous one, gathering together the results of my brief and foolhardy incursions into the hinterlands of fields far from my own, molecular and cell biology, immunology and embryology. The brevity I justify on the grounds that greater length would be even more foolhardy. The foolhardiness is less defensible, but may perhaps be forgiven on the grounds that an equally rash earlier raid yielded the germ of an idea which some molecular biologists now take seriously under the name of Selfish DNA.
Selfish DNA
… it appears that the amount of DNA in organisms is more than is strictly necessary for building them: a large fraction of the DNA is never translated into protein. From the point of view of the individual organism this seems paradoxical. If the ‘purpose’ of DNA is to supervise the building of bodies, it is surprising to find a large quantity of DNA which does no such thing. Biologists are racking their brains trying to think what this surplus DNA is doing. But from the p
oint of view of the selfish genes themselves, there is no paradox. The true ‘purpose’ of DNA is to survive, no more and no less. The simplest way to explain the surplus DNA is to suppose that it is a parasite, or at best a harmless but useless passenger, hitching a ride in the survival machines created by the other DNA [Dawkins 1976a, p. 47].
This idea was developed further and worked out more fully by molecular biologists in two stimulating papers published simultaneously in Nature (Doolittle & Sapienza 1980; Orgel & Crick 1980). These papers provoked considerable discussion, in later issues of Nature (symposia in Vol. 285, pp. 617–620 and Vol. 288, pp. 645–648) and elsewhere (e.g. BBC radio discussion). The idea is, of course, highly congenial to the whole thesis being advanced in this book.
The facts are as follows. The total amount of DNA in different organisms is very variable, and the variation does not make obvious sense in terms of phylogeny. This is the so-called ‘C-value paradox’. ‘It seems totally implausible that the number of radically different genes needed in a salamander is 20 times that found in a man’ (Orgel & Crick 1980). It is equally implausible that salamanders on the West side of North America should need many times more DNA than congeneric salamanders on the East side. A large percentage of the DNA in eukaryotic genomes is never translated. This ‘junk DNA’ may lie between cistrons, in which case it is known as spacer DNA, or it may consist of unexpressed ‘introns’ within cistrons, interspersed with the expressed parts of the cistron, the ‘exons’ (Crick 1979). The apparently surplus DNA may be to varying extents repetitive and meaningless in terms of the genetic code. Some is probably never transcribed into RNA. Other portions may be transcribed into RNA, but then ‘spliced out’ before the RNA is translated into amino acid sequences. Either way, it is never expressed phenotypically, if by phenotypic expression we mean expression via the orthodox route of controlling protein synthesis (Doolittle & Sapienza 1980).
This does not mean, however, that the so-called junk DNA is not subject to natural selection. Various ‘functions’ for it have been proposed, where ‘function’ means adaptive benefit to the organism. The ‘function’ of extra DNA may ‘simply be to separate the genes’ (Cohen 1977, p. 172). Even if a stretch of DNA is not itself transcribed, it can increase the frequency of crossovers between genes simply by occupying space between them, and this is a kind of phenotypic expression. Spacer DNA might, therefore, in some sense be favoured by natural selection because of its effects on crossover frequencies. It would not, however, be compatible with conventional usage to describe a length of spacer DNA as equivalent to a ‘gene for’ a given recombination rate. To qualify for this title, a gene must have an effect on recombination rates in comparison with its alleles. It is meaningful to speak of a given length of spacer DNA as having alleles—differing sequences occupying the same space on other chromosomes in the population. But since the phenotypic effect of spacing out genes is a result solely of the length of the stretch of spacer DNA, all alleles at a given ‘locus’ must have the same ‘phenotypic expression’ if they all have the same length. If the ‘function’ of surplus DNA is ‘to’ space out genes, therefore, the word function is being used in an unusual way. The natural selection process involved is not the ordinary natural selection among alleles at a locus. Rather, it is the perpetuation of a feature of the genetic system—distance between genes.
Another possible ‘function’ for the non-expressed DNA is that suggested by Cavalier-Smith (1978). His theory is encapsulated in his title: ‘Nuclear volume control by nucleoskeletal DNA, selection for cell volume and cell growth rate, and the solution of the DNA C-value paradox.’ He thinks that ‘K-selected’ organisms need larger cells than ‘r-selected’ ones, and that varying the total amount of DNA per cell is a good way of controlling cell size. He asserts that ‘there is a good correlation between strong r-selection, small cells and low C-values on the one hand and between K-selection, large cells and high C-values on the other’. It would be interesting to test for this correlation statistically, taking account of the difficulties inherent in quantitative comparative surveys (Harvey & Mace in press). The r/K distinction itself, too, seems to arouse widespread doubts among ecologists, for reasons that have never been quite clear to me, and sometimes seem unclear to them too. It is one of those concepts that is often used, but almost always accompanied by a ritualistic apology, the intellectual equivalent of touching wood. Some objective index of a species’s position on the r/K continuum would be needed before a rigorous test of the correlation could be undertaken.
While awaiting further evidence for and against hypotheses of the Cavalier-Smith variety, the thing to notice in the present context is that they are hypotheses made in the traditional mould; they are based on the idea that DNA, like any other aspect of an organism, is selected because it does the organism some good. The selfish DNA hypothesis is based on an inversion of this assumption: phenotypic characters are there because they help DNA to replicate itself, and if DNA can find quicker and easier ways to replicate itself, perhaps bypassing conventional phenotypic expression, it will be selected to do so. Even if the Editor of Nature (Vol. 285, p. 604, 1980) goes a bit far in describing it as ‘mildly shocking’, the theory of selfish DNA is in a way revolutionary. But once we deeply imbibe the fundamental truth that an organism is a tool of DNA, rather than the other way around, the idea of ‘selfish DNA’ becomes compelling, even obvious.
The living cell, especially the nucleus in eukaryotes, is packed with the active machinery of nucleic acid replication and recombination. DNA polymerase readily catalyses the replication of any DNA, regardless of whether or not that DNA is meaningful in terms of the genetic code. ‘Snipping out’ of DNA, and ‘splicing in’ of other bits of DNA, are also parts of the normal stock in trade of the cellular apparatus, for they occur every time there is a crossover or other type of recombination event. The fact that inversions and other translocations so readily occur, further testifies to the casual ease with which chunks of DNA may be cut out of one part of the genome, and spliced into another part. Replicability and ‘spliceability’ seem to be among the most salient features of DNA in its natural environment (Richmond 1979) of cellular machinery.
Given the possibilities of such an environment, given the existence of cellular factories set up for the replication and splicing of DNA, it is only to be expected that natural selection would favour DNA variants that are able to exploit the conditions to their own advantage. Advantage, in this case, simply means multiple replication in germ-lines. Any variety of DNA whose properties happen to make it readily replicated will automatically become commoner in the world.
What might such properties be? Paradoxically, we are most familiar with the more indirect, elaborate, and roundabout methods by which DNA molecules secure their future. These are their phenotypic effects on bodies, achieved by the proximal route of controlling protein synthesis, and hence by the more distal routes of controlling embryonic development of morphology, physiology and behaviour. But there are also much more direct and simple ways in which varieties of DNA can spread at the expense of rival varieties. It is becoming increasingly evident that, in addition to the large, orderly chromosomes with their well-regimented gavotte, cells are home to a motley riff-raff of DNA and RNA fragments, cashing in on the perfect environment provided by the cellular apparatus.
These replicating fellow-travellers go by various names depending on size and properties: plasmids, episomes, insertion sequences, plasmons, virions, transposons, replicons, viruses. Whether they should be regarded as rebels who have broken away from the chromosomal gavotte, or as invading parasites from outside, seems to matter less and less. To take a parallel, we may regard a pond, or a forest, as a community with a certain structure, and even a certain stability. But the structure and stability are maintained in the face of a constant turnover of participants. Individuals immigrate and emigrate, new ones are born and old ones die. There is a fluidity, a jumping in and out of component parts, so that it becomes meaningless
to try to distinguish ‘true’ community members from foreign invaders. So it is with the genome. It is not a static structure, but a fluid community. ‘Jumping genes’ immigrate and emigrate (Cohen 1976).
Since the range of possible hosts in nature at least for transforming DNA and for plasmids such as RP4 is so large, one feels that at least in Gram-negative bacteria all populations may indeed be connected. It is known that bacterial DNA can be expressed in widely different host species … It may indeed be impossible to view bacterial evolution in terms of simple family trees; rather, a network, with converging as well as diverging junctions, may be a more appropriate metaphor [Broda 1979, p. 140].
Some authors speculate that the network is not confined to bacterial evolution (e.g. Margulis 1976).
There is substantial evidence that organisms are not limited for their evolution to genes that belong to the gene pool of their species. Rather it seems more plausible that in the time-scale of evolution the whole of the gene pool of the biosphere is available to all organisms and that the more dramatic steps and apparent discontinuities in evolution are in fact attributable to very rare events involving the adoption of part or all of a foreign genome. Organisms and genomes may thus be regarded as compartments of the biosphere through which genes in general circulate at various rates and in which individual genes and operons may be incorporated if of sufficient advantage …’ [Jeon & Danielli 1971].