The Extended Phenotype: The Long Reach of the Gene (Popular Science) (36 page)

Lorenz distinguished two kinds of theory for the origin of this kind of fit between organism and environment, but both his theories (natural selection and reinforcement learning) are subdivisions of what I am calling the selection theory. An initial pool of variation (genetic mutation or spontaneous behaviour) is worked upon by some kind of selection process (natural selection or reward/punishment), with the end result that only the variants fitting the environmental lock remain. Thus adaptedness improves by selection. The instruction theory is quite different. Whereas the selective key-maker begins with a large random pool of keys, tries them all in the lock, and discards those that don’t fit, the instructive key-maker simply takes a wax impression of the lock, and makes up a good key directly. The instructively camouflaged animal resembles its environment because the environment directly imprints its appearance on the animal, as elephants merge into the background because they are covered by its dust. It has been alleged that French mouths eventually become permanently deformed into a shape suitable for pronouncing French vowels. If so, this would be an instructive adaptation. So, perhaps, is background resemblance in chameleons, though of course the
capacity
to change colour adaptively is presumably a selective adaptation. The adaptive changes in physiology to which we give names like acclimatization and training, the effects of exercise, use and disuse, are probably all instructive. Complex and elaborate adaptive fits can be achieved by instruction, as in the learning of a particular human language. As already explained, it is clear that in Steele’s theory the adaptedness comes not from instruction but from selection, and genetic replicator selection at that. My world-view will be overturned if somebody demonstrates the genetic inheritance, not just of an ‘acquired characteristic’ but of an instructively acquired adaptation. The reason is that the inheritance of an instructively acquired adaptation would violate the ‘central dogma’ of embryology.

Oddly, my belief in the inviolability of the central dogma is not a dogmatic one! It is based on reason. I must be cautious here, and distinguish two forms of central dogma, the central dogma of molecular genetics and the central dogma of embryology. The first is the one stated by Crick: genetic information may be translated from nucleic acid to protein, but not the other way. Steele’s theory, as he himself is careful to point out, does not violate this dogma. He makes use of reverse transcription from RNA to DNA, but not reverse translation from protein to RNA. I am not a molecular biologist, and so cannot judge the extent to which the theoretical boat would be rocked if such reverse translation were ever discovered. It does not seem to me obviously impossible in principle, because the translation from nucleic acid to protein, or from protein to nucleic acid, is a simple dictionary look-up procedure, only slightly more complex than the DNA/RNA transcription. In both cases there is a one-to-one mapping between the two codes. If a human, or a computer, equipped with the dictionary, can translate from protein to RNA, I don’t see why nature should not. There may be a good theoretical reason, or it may just be an empirical law that has not yet been violated. There is no need for me to pursue the matter because, in any case, a very good theoretical case can be made against violating the other central dogma, the central dogma of embryology. This is the dogma that the macroscopic form and behaviour of an organism may be, in some sense, coded in the genes, but the code is irreversible. If Crick’s central dogma states that protein may not be translated back into DNA, the central dogma of embryology states that bodily form and behaviour may not be translated back into protein.

If you sleep in the sun with your hand over your chest, a white image of your hand will be imprinted on your otherwise tanned body. This image is an acquired characteristic. In order for it to be inherited, gemmules or RNA viruses, or whatever agent of reverse translation is postulated, would have to scan the
macroscopic
outline of the hand image and translate it into the molecular structure of DNA necessary to program the development of a similar hand image. It is suggestions of this kind that constitute a violation of the central dogma of embryology.

The central dogma of embryology does not follow inevitably from common sense. Rather, it is a logical implication of rejecting the preformationist view of development. I suggest, indeed, that there is a close link between the epigenetic view of development and the Darwinian view of adaptation, and between preformationism and the Lamarckian view of adaptation. You may believe in inheritance of Lamarckian (i.e. ‘instructive’) adaptations, but only if you are prepared to embrace a preformationistic view of embryology. If development were preformationistic, if DNA really
were
a ‘blueprint for a body’, really were a codified homunculus, reverse development—looking-glass embryology—would be conceivable.

But the blueprint metaphor of the textbooks is dreadfully misleading, for it implies a one-to-one mapping between bits of body and bits of genome. By inspecting a house, we may reconstruct a blueprint from which somebody else could build an identical house, using the same building technique as was used for the original house. The informational arrows from blueprint to house are reversible. The relative positions of the ink lines in the blueprint and of the brick walls in the house are transformable, one into the other, by a few simple scaling rules. To go from blueprint to house, you multiply all measurements by, say twenty. To go from house to blueprint, you divide all measurements by twenty. If the house somehow acquires a new feature, say a west wing, a simple, automatic procedure could be written down for adding a scaled-down map of the west wing to the blueprint. If the genome were a blueprint with a one-to-one mapping from genotype to phenotype, it would not be inconceivable that the white imprint of a hand on an otherwise tanned chest could be mapped on to a sort of miniature genetic shadow of itself, and so inherited.

But this is utterly alien to everything we now understand about the way development works. The genome is not, in any sense whatsoever, a scale model of the body. It is a set of instructions which, if faithfully obeyed in the right order and under the right conditions, will result in a body. I have previously used the metaphor of a cake. When you make a cake you may, in some sense, say that you are ‘translating’ from recipe to cake. But it is an irreversible process. You cannot dissect a cake and thereby reconstruct the original recipe. There is no one-to-one, reversible mapping from words of recipe to crumbs of cake. This is not to say that a skilled cook could not achieve a passable reversal, by taking a cake presented to him and matching its taste and properties against his own past experience of cakes and recipes, and then reconstructing the recipe. But that would be a kind of mental selection procedure, and would in no sense be a
translation
from cake to recipe (a good discussion of the difference between reversible and irreversible codes, in the context of the nervous system, is given by Barlow (1961)).

A cake is the consequence of the obeying of a series of instructions, when to mix the various ingredients, when to apply heat, etc. It is not true that the cake
is
those instructions rendered into another coding medium. It is not like a translation of the recipe from French into English, which is in principle reversible (give or take a few
nuances)
. A body, too, is the consequence of the obeying of a series of instructions; not so much when to apply heat as when to apply enzymes speeding up particular chemical reactions. If the process of embryonic development is correctly set in motion in the right environment, the end result will be a well-formed adult body, many of whose attributes will be interpretable as consequences of its genes. But you cannot reconstruct an individual’s genome by inspecting his body, any more than you could
reconstruct William Shakespeare by decoding his collected works. Cannon’s and Gould’s false argument of p. 116 is validly adapted to embryology.

Let me put the matter in another way. If a man is particularly fat, there are many ways in which this might have come about. He might have a genetic predisposition to metabolize his food particularly thoroughly. Or he might have been overfed. The end result of an excess of food may be identical to the end result of a particular gene. In both cases the man is fat. But the routes by which the two causal agents produced their common effects are totally different. For a man who is artificially stuffed with food to pass on his acquired fatness to his children genetically, some mechanism would have to exist that sensed his fatness, then located a ‘fatness gene’ and caused it to mutate. But how could such a fatness gene be located? There is nothing intrinsic in the nature of the gene that makes it recognizable as a fatness gene. It has its obese effect only as a result of the long and complex unfolding sequence that is epigenetic development. The only way, in principle, to recognize a ‘fatness gene’ for what it is, is to allow it to exert its effects on the normal processes of development, and this means development in the normal, forward direction.

This is why bodily adaptations can come about by selection. Genes are allowed to exert their normal effects on development. Their developmental consequences—phenotypic effects—feed back on those genes’ chances of surviving, and as a result gene frequencies change in succeeding generations in adaptive directions. Selective theories of adaptation, but not instructive theories, can cope with the fact that the relationship between a gene and its phenotypic effect is not an intrinsic property of the gene, but a property of the forward developmental consequences of the gene when interacting with the consequences of many other genes and many external factors.

Complex adaptation to an environment may arise in individual organisms through instruction from that environment. In many cases this certainly happens. But, given an assumption of epigenetic, not preformationistic embryology, to expect such complex adaptations to be translated into the medium of the genetic code, by some means other than the selection of undirected variation, is a gross violation of all that I hold rational.

There are other examples of what looks like true Lamarckian ‘instruction’ from the environment being inherited. Non-genetic anomalies that appear, or even that are surgically induced, in the cortex of ciliates, may be directly inherited. This has been demonstrated by Sonneborn and others. By Bonner’s account, they cut out a small portion of the cortex of
Paramecium
and reversed it. ‘The result is a
Paramecium
with part of one row of basal bodies in which the fine structure and details are all pointed 180° away from the rest of the surface. This anomalous kinety is now inherited; it appears to be a permanent fixture of the progeny (which have been carried through 800 generations)’ (Bonner 1974, p. 180). The inheritance appears to be non-genic
and is obviously non-nuclear. ‘… the cortex is made up of macromolecules that assemble in a particular pattern, and … this pattern, even in a disturbed state, is directly inherited… . we have a large and exceedingly complex cortex whose pattern of fitting together is a property of the macromolecules at the cortex and is not directly under nuclear control. Over what must have been a long time and a vast number of cell cycles, a surface structure evolved. The structure itself had properties such that its immediate form is independent of the nucleus; at the same time, it is totally dependent on the nucleus, we presume, for the synthesis of its specifically shaped building blocks’ (Bonner 1974).

As in the case of Steele’s work, whether we regard this as the inheritance of acquired characteristics depends upon our definition of the germ-line. If we direct our attention to the individual body, a surgical mutilation of its cortex is clearly an acquired characteristic having nothing to do with the nuclear germ-line. If, on the other hand, we look at underlying replicators, in this case perhaps the basal bodies of cilia, the phenomenon falls under the general heading of replicator propagation. Given that macromolecular structures in the cortex are true replicators, surgically rotating a portion of cortex is analogous to cutting out a portion of chromosome, inverting it, and putting it back. Naturally the inversion is inherited, because it is part of the germ-line. It appears that elements of the cortex of
Paramecium
have a germ-line of their own, although a particularly remarkable one in that the information transmitted does not seem to be encoded in nucleic acids. We should definitely predict that natural selection might act directly on this non-genic germ-line, shaping surface structure for the adaptive benefit of the replicating units in the surface itself. If there is any conflict between the interests of these surface replicators and nuclear genes, the resolution of the conflict should make a fascinating study.

This is by no means the only example of non-nuclear inheritance. It is becoming increasingly clear that non-nuclear genes, either in organelles such as mitochondria or loose in cytoplasm, exert noticeable effects on phenotypes (Grun 1976). I had intended to include a section called
The Selfish Plasmagene
, discussing the expected consequences of selection acting on cytoplasmic replicators, and the likely results of conflicts with nuclear genes. However, I had got no further than some brief remarks on ‘selfish mitochondria’ (now in
Chapter 12
), when two papers arrived (Eberhard 1980; Cosmides & Tooby 1981) which, independently, say everything I might have said and much more. To give just one example, ‘The migration of egg mitochondria to cluster around the egg nucleus, so as to favor their inclusion in the “neocytoplasm” of the proembryo in the gymnosperms
Larix
and
Pseudotsuga
… may result from competition for inclusion in the embryo’ (Eberhard, p. 238). Rather than extensively duplicate what is in them I would prefer simply to recommend readers to consult both these excellent
papers. I will only add that both papers are good examples of the kind of discussion that I believe will become commonplace, once the replicator replaces the individual organism as the fundamental conceptual unit in our thinking about natural selection. One does not have to be clairvoyant to prophesy, for example, the rise of a flourishing new discipline of ‘prokaryotic sociobiology’.