We have come a long way since Aristotle first wondered why it was that pregnant women did not menstruate, and concluded that babies were made from unshed blood. Aristotle is rightly regarded as the father of biological science, and the history of biology can be seen as a great quest to elaborate on Aristotle’s hypothesis, to understand how babies are made and how it is that all the diverse creatures of the natural world can emerge from tiny, formless eggs – Ex Ovo, Omnia. Cast as such, the story of biology can be told in an unbroken skein, through Harvey and the preformationists who, finding Harvey’s ideas incomplete, created a robust theoretical edifice which stood for longer than the entire history of genetics has so far endured; to the nature-philosophers, who turned the tables yet again, and without whose idealism Darwin’s ideas could not have come into being in the way they did; to Bateson and Morgan, reacting against what they saw as Darwinism’s deficiencies, and setting in train the amazing discoveries of experimental genetics that have marked the past century; right to the present generation of computational biologists whose concepts of the genome as a network seem to resolve the questions that Aristotle asked.
The network concept does seem to embrace all the qualities of the genome I have discussed throughout this book. As a human being – and particularly as a parent – I am as amazed as Aristotle must have been that creatures as intricate and delicate as babies are produced so routinely and with such fidelity. This reliability is a consequence of a genome constructed as a network, which can produce a human being in the most varied and trying of circumstances, at the cost of a small amount of variability in the finished result that we accept as the normal range of variation. As such, the reliability of networks provides, almost as a by-product, an explanation for the existence of variation, a puzzle which perplexed Darwin and Bateson. However, the existence of this variation is something for which we should be grateful, for were the genome to conduct itself with the precision of an engineer, and not tolerating a certain amount of woolliness in its operation, every single variation might be monstrous and no human beings would ever be born.
Such woolliness explains why it is that in its embryonic development every new baby passes through stages reminiscent of the evolutionary history of the species. Babies are not designed from scratch after each conception, but are built according to networks which already exist: the networks that built their parents, their grandparents and all their other ancestors right back to the beginning. These networks were repeatedly damaged and patched up, and accumulated small changes, some of which not just created new individuals but provided the germ of variation that created new species. Change was built into networks which, thanks to their forgiving nature, functioned none the less – building up traces of their histories as they went along. The scars of evolution that mark the genome are matched by those that started with the division of one cell into two identical daughters, and so, after billions of years, created the exquisite marvel that is embryonic development, in which, for example, a human baby goes through a disc-like stage, just like a modern egg-laying reptile, even though the last member of the human lineage that laid eggs lived more than 100 million years ago. But these, and other reminiscences – of lampreys, and sea squirts, and other wonders – are just that: reminiscences, not an orderly parade that might be used to plot evolutionary history with any great accuracy. They are impressions, cartoons, doodles on the chalkboard of life which have not been fully erased.
If networks allow for such laxity, why are there so many distinct species in the world, and how do we know that, say, a red-cheeked bulbul is an entirely different thing from a Western tanager – or, for that matter, that an individual fly of the species Drosophila melanogaster is quite different from a fly belonging to the species D. simulans? The question of why species are, in general, so distinct from one another has long been perplexing. Networks are so forgiving, yet the world’s creatures are not all members of one huge, heterogeneous, continuously variable continuum. To put it in more concrete terms, human beings come in a wide variety of shapes and sizes, but they are all equally human, and the world is not full of quasi-human creatures that are more or less like chimpanzees or other creatures. This, in essence, is the demand that Bateson made of the bio-metricians: if variation within a species was limited to extremely small gradations of traits, as the biometricians insisted, why were differences between species so clear to see?
I suggest that the network concept of the genome might provide the answer that Bateson sought. Although the output of a network can vary, the pattern of connections that comprises the network does not actually change within a given species. All humans, for example, are products of the same network, whose overall output depends on how the various genes and modules are wired together. As Garrett Odell and his team at Seattle showed with the segment-polarity module, segmentation can be produced in an enormous range of situations, provided the network itself is wired up correctly. Even if genes in the network were damaged beyond repair, as in the fruit-fly monsters created by Nüsslein-Volhard and Wieschaus, segments might still be produced.
But what if a mutation resulted in a change in the pattern of connection itself? The change need not be very great: for example, a mutation that created an operator sequence where none existed before might allow repressors or other factors to bind there. The result would be the creation of a network qualitatively different from the one that had existed. As the Seattle group showed, networks are very sensitive to such changes in connectedness. If they are not connected correctly, they are very likely not to work at all. Any such change would therefore have to be small, and perhaps not initially of much significance. After all, the creation of a new operator sequence would be of little consequence if no repressor yet existed to bind to it, or if such substances that might do so were better at binding elsewhere. However, the result of such a change would be to introduce an element of instability into the network. It is a property of networks, and distributed systems, that they are inherently good at working round such problems. Mutations that introduce changes in the connection pattern in a network might soon be followed by a host of other small, compensatory changes in quick succession to give the new network some kind of equilibrium. Given the right circumstances, such changes might form the basis of a new species.
This idea, that networks might switch rapidly from one so-called ‘metastable’ state to another, is nothing new in science, especially in physics and mathematics. To apply it to the genome might seem speculative in the extreme. Nevertheless, the idea does seem to fit with what we currently know about speciation – that is, the genetics of the origin of species. Speciation is all about sex and reproduction. It can happen only if the creatures in one population stop breeding with those in another, for if they continued to do so their genes would stay forever mixed, divergence could not happen, and we would indeed all be part of a single, heterogeneous continuum with people at one end, flies at the other, and the ghost of Franz Kafka somewhere in the middle. In the real world, however, creatures go to enormous lengths to avoid misalliance, especially if a pair of different creatures look like each other. In groups of animals in which there are many distinct but very similar species, males or females evolve very exaggerated and specific sexual features to ensure that members of one species do not mistakenly mate with members of another. Many species of spider, for example, are identical but for the elaborate genitalia of the male which, like a key, will only fit into the genitalia of the female of the same species. Female fruit flies will swoon only before males who perform a specific mating ritual with the requisite precision. If the process of speciation depends on genetic variation, the genes involved will be associated with traits connected with sexual anatomy and behaviour. Theories of speciation based on collections of genes – rather than their genomic totality – are good at explaining how divergence proceeds once it has begun, but are less successful at describing the very first changes that lead to divergence. And the still highly controversial question remains – whether speciation consists of many small changes in a large suite of genes, or a smaller number of changes in a few, key genes.
I think this might be the wrong question, because it concentrates too closely on the behaviour of genes in isolation instead of looking at the behaviour of the whole genome. I suspect that speciation might involve changes in the interaction patterns of genes or modules associated with sexual behaviour. Because all modules are connected to some degree, these genes or modules would have close links with other genes or modules specifying other body parts or functions. Because sexual isolation is crucial to speciation, it is likely that such links are close: changes in, say, networks specifying sexual anatomy would have rapid knock-on effects on other parts of the body. Changes involving sex genes would have a domino effect, opening the way for natural selection to prompt compensatory changes in other, more remote parts of the network. Such changes of wiring might have the effect of amplifying any small, external change in behaviour, as well as any accompanying changes in form or function. Again through natural selection, these changes would feed back on the genome, providing the opportunity for many other changes to occur. Eventually the network would achieve a new stability, a new equilibrium – and a new species.
If this seems speculative, it is – and deliberately so. But in fact, a tentative model of such connections already exists and has already been mentioned – the case of the fruit-fly gene bric-a-brac. This gene is responsible for the gender-specific pigmentation of fruit-fly abdomens, which has a role in the flies’ sexual behaviour and choice of mate. At the same time the gene is part of a regulatory network – involving the Hox genes – that shapes the abdomen itself, which suggests a connection between the courtship behaviour of flies and the shape of their bodies.1 Once a new species is formed, perhaps in the wake of one small change, other compensatory changes might soon follow so that it becomes progressively more difficult for a member of a new species to mate with a member of the ancestral species and produce viable offspring. As species accumulate changes immediately after the split, hybrids first become weak, then sterile mules, then inviable monsters that die in the womb, then cease to exist.
All of the above is advanced in the spirit of tentative speculation, but it does seem that the network view of the genome has the potential to explain an enormous amount about its past history and current behaviour. It is certainly a better model of reality than the clichéd description of DNA as a ‘blueprint’. However, we must guard against proclaiming that the network view is the final and definitive answer to how form is created from the formless. If the history of scientific enquiry teaches us one thing, it is that we should never assume that progress is either uniform or proceeds in one direction, such that our predecessors were always wrong, and the further we delve into the past the more wrong we find they were. The history of biology, told here as the search for the agency that created form from the formless, shows that such a patronizing view of the past is in fact worse than wrong, for it blinds us to the work of great minds, all of whom made important contributions to the way we think now. Like the genome and the process of embryonic development it engenders, the development of our own thought bears the scars of its own evolution without which it might be very different – perhaps incomprehensibly so. Either that, or it might not exist at all. We owe a debt of respect to our predecessors, even though their thoughts might seem to us misguided or even bizarre.
The story of how our genome came to be understood can be seen as a simple quest. Alternatively, it may be seen as having been prey to successive waves of intellectual revolution. As Harvey’s experiments overturned centuries of orthodoxy based on Aristotle, the work of the preformationists overthrew Harvey’s epigenesis, and preformation succumbed in its turn to a resurgent epigenesis that became nature-philosophy, which produced what would eventually become modern embryology. In the meantime, Darwin’s work was a reaction to the idealism of nature-philosophy in which variation was seen as a hindrance to understanding, rather than a phenomenon demanding explanation in its own right. Unable to achieve a complete understanding of the sources of variation, however, Darwinism was eclipsed, if not overturned, by the new genetics of Bateson and Morgan, and was not rehabilitated until these early geneticists had all but left the scene.
That rehabilitation, by Morgan’s student Dobzhansky and others, created population genetics – an understanding of evolution based on collections of genes pictured as individual agents, often in competition with one another. This world-view is perhaps most eloquently expressed by Richard Dawkins in The Selfish Gene, and has successfully explained much about the way nature works. But a new view is now emerging which claims much more. The network view offers fresh approaches to enduring problems such as speciation, which population genetics has found very hard to explain fully. In addition, the new view suggests opportunities for discussing the grand sweep of evolution and the minutiae of individual development within the same theoretical framework – something that population genetics has very largely failed to achieve.
One of the great strengths of population genetics is its universality. Evolution is discussed in terms of the ebb and flow of different genetic variants, or alleles, among populations of individuals. When put so simply, the findings of population genetics can easily be expressed in mathematical terms, allowing scientists to devise predictive theories of evolution of disarming elegance and enormous power. The equations of population genetics are not tied to the life or history of any particular organism, and are therefore universal. Among many other things, they can explain why bees are so faithful to their queen; why female birds in apparently monogamous relationships cuckold their mates with such abandon; how mice discriminate between their own kin and a potential mating partner; and how cannibalistic salamanders select their next victim. I fully expect that when we encounter life forms on other planets, the social behaviour of aliens a dozen light years away should be expressible by the same formulae that can be used to summarize the courtship of two flies on a cow pat in the field next door.
Curiously, it is in this very universality that the weakness of population genetics lies. Because it can apply to any population of organisms, anywhere and at any time, it is not very good at expressing the particular circumstances of the evolutionary history of which we are a part. In particular, it is impossible to extrapolate from the behaviour of genes in populations to the evolution of species over geological time, because one can never assume that the conditions in which organisms are evolving in one set of circumstances will necessarily apply in a different situation millions of years later. As soon as this assumption is made, however, one strays into the realm of fable – meeting stories in which, for example, an overwhelming and unswerving selective need, maintained over millions of years in all circumstances, drove a population of reptiles to evolve feathers and become birds, or coerced the ancestors of humans to adopt an erect posture, freeing their hands for carrying food and nursing babies. Such tales are neither better nor worse than Lamarckian besoin or nature-philosophical urges towards cosmic perfection, and Bateson – rightly – found them deplorable. Modern Creationists have been very quick to exploit this tendency for population genetics to be extrapolated into realms that it was not designed to address. Creationists will accept the ‘microevolution’ of adaptation within populations, while denying that the natural world above this relatively trivial level need be explained by a kind of ‘macroevolution’, especially as population geneticists have failed to describe macroevolution convincingly.
None of this is meant to imply that the population genetics that explains so much about the behaviour and constitutions of animals and plants, and has served us so well for nearly a century, is either misleading or wrong. However, it may very well be that some aspects of evolution, such as speciation and macroevolution, might be answered more successfully by taking a more rounded approach – that is, with reference to the behaviour of the genome as an integrated entity, rather than as a collection of genes: as a genome whose complexion is sensitive to its own history. It is possible that the network view as I have sketched it here has the potential to become a more complete, inclusive description of reality in which population genetics describes microevolution – that is, relatively small genetic changes and events within species – better than the more dramatic descriptions based on macroevolution, the appreciable changes of form over measurable intervals of geological time. By way of analogy, everyone recognizes that Einstein’s relativity offers a better account of gravity than Newton’s laws, because it encompasses, within the same theoretical framework, both everyday events and the more extreme phenomena for which Newtonian descriptions are inadequate, such as the motion of massive objects travelling close to the speed of light. In most circumstances, however, Newtonian rules will apply: despite relativity, apples still fall from trees the way they always did.
More fundamentally, however, the network view could represent a reaction to the reductionist tendency of much in modern science that seeks to apprehend structures in terms of their component parts, rather than as complete entities. In contrast, a view that connects the individual development of a human baby with the evolution of all life over billions of years, linking the microcosm with the macrocosm in true nature-philosophic style, is something Goethe might have appreciated. Only time will tell whether the network view will be seen as something set in opposition to more reductionist styles of genetics, in the way that Bateson’s genetics was a reaction to Darwinian evolution before their eventual reconciliation; or that preformation was couched in opposition to epigenesis, before it was discovered – in the relationship between the genome and embryonic development – that there is merit in both views.
I rather hope that, at least to begin with, the network view of the genome will come to be seen as something opposed to the reductionist programme, because it is otherwise impossible to grasp the scientific implications of the emergence of humanity if the human genome is considered simply as a collection of genes, or a list of bases, rather than an entity with a unique history and unitary properties all its own, properties that cannot be predicted from single genes or sequences of DNA. After all, we cannot point to any one sequence in the human genome and find a homunculus, nor will we ever be able to spot, in a forest of sequence differences, those that led to the spark of sentience. The differences are everywhere, and yet they are nowhere – consequences of the function of the network as a whole, not of any one of its parts that we can identify.
In which case, to state that the draft sequence of DNA will tell us what it means to be human is to overstate the case by a large margin. We have come a long way since Aristotle, but we have a great deal left to do, and to learn. The great adventure has only just begun.