Developments twin arrows

It's now time to bring development back into the picture in a more explicit way. As we noted in the previous chapter, internal selection can occur based on fitness differences between adults, so there is no logical necessity for this process to be linked to development other than for the obvious reason that adults arise through the developmental process. But there is another angle on all this, which does indeed suggest that internal selection and development may have a particularly close relationship.

By the way, you will notice that I have debunked both 'internal' and 'external' selection in their pure forms, yet here I am starting to use these terms again as if I don't believe my own argument for a continuum. I must therefore stress that I do believe my argument, but I also need to be able to write in a relatively straightforward, non-cumbersome way. So phrases like 'selection in which 75 per cent of the fitness difference is due to a difference in internal integration and 25 per cent due to the interaction with the environment' are not the best way forward. So from here on, please interpret 'internal selection' as meaning predominantly internal, and likewise for 'external'.

Suddenly it begins to matter which type of creature we are dealing with. Although evolutionary theory, like all science, should be as general as possible, and although the idea of selection (internal and/or external) can be applied to any type of creature, there are some things in relation to which we have to be careful not to overgeneralize. And what follows is one of them.

We're mammals so let's start with a mammal embryo. It could be our old friend the microscopic horse; alternatively it could be the human embryo that, on a probabilistic basis, is growing inside a small proportion of my audience as they read these words. Equally, it could be a mouse or a deer embryo. All that matters is that it is a placental mammal.

What all placental mammal embryos have in common is protection from the elements. In this, they differ from many other types of animal, whose earliest developmental stages are 'out there', exposed to the vagaries of the physical environment, predators, and other threats. This is true, for example, of marine 'trochophore' larvae.

The superb protection from the elements afforded by the mammalian womb has a major effect on the balance of internal and external selective forces that act on the embryo. In this sort of situation, internal selection reigns supreme. As we know, many mammalian embryos die before they are born; human miscarriages, for example, are all too common, and doubtless the same is true of our various mammalian cousins. But this mortality is not caused by intrauterine predators. Nor can it be caused by harsh winters, except indirectly when the death of the embryo is caused by the death of the mother. What kills mammalian embryos, therefore, must be problems within the developing embryo itself, though problems with the umbilical connection to the mother are sometimes also a factor.

So we are back to the business of building bodies. In the case of the mammal embryo, and indeed of any other embryo in the true sense of the term (i.e. as distinct from a larva or a juvenile), building a body is its main job. Of course, it has to function well enough to stay alive in the process. Blood, for example, must flow to all the developing tissues. But there is no need to find a suitable microhabitat in a heterogeneous and mostly unsuitable environment. The mother has already provided it with the most secure microhabitat available. There is no need either to search for food. Again, the mother provides. So in this unique, highly protected situation, most of the selection that goes on is based on differences in organismic integration.

But what do we mean by 'organismic integration' in an organism whose very essence is a state of change? This is a very different scenario from the joint between the wing and the thorax of a fly that we considered earlier. That was integration of the adult organism. This adult, or indeed any other, has a very different - we could almost say 'opposite' - range of problems to deal with than an embryo. It does have to deal with the external environment and all the threats that that entails. But it does not have to transmute gradually into some very different kind of creature. Although the adult does indeed have to make another 'creature' - the sperm or egg that will, when they meet, form the next generation - this is a very different task from that faced by the embryo. An adult confines this process to its reproductive tract; the rest of the adult body is built for survival. And it doesn't change much over time. If a wing/thorax joint works well when the fly emerges from its pupal case and first takes to the air, it will probably still work well, injuries aside, later on in its life.

The embryo, in contrast, has a much lesser job to do on the survival front but a much greater job to do in terms of building a body. This is not something confined to a particular part of the embryo; rather, the whole embryo is, throughout its entire existence, transforming itself into something else. The embryo is the epitome of the phrase 'life flows'. Indeed, one way to picture an embryo is as a trajectory through multicharacter hyperspace. Even this is too simple because it is not just character values - like brain size - that vary as the embryo develops. Characters that were not there initially gradually come into being. We go from no brain to proto-brain to small brain to bigger brain.

Integration, in this sort of situation, comes in two forms. At any precise moment, the embryo must be sufficiently well integrated to function; to stay alive; to avoid the fate of miscarriage. But, when viewed over an extended period of time, the embryo must remain integrated in a different way. As well as the early embryo and the later embryo each having their own 'instantaneous' integration, there must also be a sort of 'temporal integration', so that the trajectory from one to the other works. And although I have set this up as a two-point comparison, this is just a starting point for thinking about the situation. In reality, every 'stage' must be compatible with every other.

This compatibility has a directionality to it. An earlier stage makes a later stage but not vice versa. So if there is any change in an earlier stage as a result of mutation in a gene that helps to control what happens at that stage, then not only must the instantaneous integration at the point when the mutation first takes effect be maintained, but also, if the embryo is to continue on its way later, the temporal integration between stages must not be lost. But there is a lot in this phrase 'on its way'. On its way to where? Well of course it is on its way to later embryonic stages, to juvenile stages and ultimately to the adult. But these may vary to some degree. After all, early embryos are not evolutionarily immutable. We know this because the early embryos of different mammal species are often similar but never identical. And the same applies to other groups.

All this simply means that embryos have to either stay the same or else change in a way where integration is maintained or enhanced. If integration is lost, the embryo dies. If integration is reduced, the embryo may live and continue to develop; but, other things being equal, the mutation causing the reduction will be removed from the population by natural selection. This last point is important, because we should never forget the probabilistic nature of evolution. This is a point that is deeply embedded in the subconscious of population geneticists. But developmental biologists tend to think in a different way: about changes, for example, that make the embryo 'inviable'. There is a risk, if we think in terms of viable and inviable embryos, of treating the effects of mutations as all or nothing, which certainly does not reflect reality. What we need to do is to unite a time-extended developmental view of organisms with a probabilistic view of the nature of evolutionary changes. We must take the strengths of each discipline and put them together. Only then

figure 19 In evolution, later developmental stages typically become more different from each other than earlier ones for two reasons: the simple asymmetry of causality where early stages can affect later ones but not vice versa (top); plus the fact that it is easier to modify later stages (bottom). Vertical arrows indicate times of initial developmental expression of genes that alter in evolution; shaded triangles indicate increasing cumulative developmental divergence between two evolutionary lineages; larger triangle includes both 'asymmetry' and 'ease' effects.

will a form of evo-devo emerge that is truly mechanistic in both time frames.

Let's dissect this embryonic directionality, in which earlier stages make later ones, from an evolutionary point of view (Figure 19). One key question here is: how does the probability of a mutation in a developmental gene being advantageous or disadvantageous vary with the time of first expression of the gene concerned in the growing embryo? Do mutations in early-acting genes have greater, equal or lower probabilities of destroying that all-important temporal integration of the overall trajectory? Or, if you prefer to put it in a way that relates more to evolutionary change than to evolutionary conservation: do mutations in early-acting genes have a greater, equal or lower probability of causing improvements?

The conventional wisdom is that the earlier developmental reprogramming takes place, the more likely it is to be detrimental. If you change some fundamental early step, like the formation of the anteroposterior axis of the embryo, you are asking for trouble. Too many downstream processes are likely to be disrupted. The chances are that the embryo will die. But let's move forward in embryonic time to a much later stage when digits are being formed. Here, changes are not so likely to be lethal. Some humans have extra fingers and toes -'polydactyly'. This is the result of a gene mutation that deflects the process of digit formation right back at its origin in the first couple of months of embryonic development - still 'late', though, in our comparison. But the new trajectory works reasonably well. If the developing digits were used to make something else further downstream, and their exact number was crucial, then the polydactyly mutation might be lethal. But this is not the case.

Although polydactyly is a genetic and developmental phenomenon, it is not an evolutionary one, at least not in our recent past. There are no human races whose members typically have six fingers on each hand or six toes on each foot. Instead, individuals are affected only very occasionally. But in the grand scheme of vertebrate evolution, changes in the number and structure of digits have been common. Some of the earliest tetrapods, it appears from the fossil record, had eight toes on each foot. And many modern mammals have a reduced number of digits - for example the horse. Also, the difference between a webbed foot and a 'free-toes' foot is probably due to mutations that affect the process of programmed cell death that normally occurs in the interdigital regions. So evolution can and does frequently modify later embryonic processes on which not too much depends in the way of generating other things further downstream. But in contrast, evolution almost certainly finds it harder to modify those crucial early stages on which all later events depend.

The story so far, if true, would mean that if we took two species of mammal - say a human and a horse - and compared their embryos at various points through developmental time, we would find that they looked most similar at the earliest stages and most different at the latest ones. And this is indeed what happens. This is what Karl Ernst von Baer observed in the early nineteenth century.1 It was given an evolutionary interpretation by Darwin himself,2 who asked how we could explain the fact of 'embryos of different species within the same class generally, but not universally, resembling each other'. He answered his own question by saying that it was explicable on the basis of 'descent with modification', and with many of those modifications only taking effect part-way through development, thus leaving the embryos more similar than the corresponding adults. 'Thus', he concludes, 'community of embryonic structure reveals community of descent.'

This von Baerian picture of diverging embryos, which Darwin accepted and gave evolutionary meaning to, has been questioned in recent years from several different points of view. In none of these cases do I think that our picture should be radically altered; that is, I believe that von Baer was broadly correct in his observations and Darwin broadly correct in his interpretations of them. However, the criticisms that have been made are instructive in certain ways, as we shall see. The central point in all the criticisms is that the standard picture of von Baerian divergence is too simple. Let's now look at the various aspects of this issue.

First, the very earliest embryonic stages may actually vary more, not less, than the immediately following ones. This leads to a bundle of developmental trajectories of different species within a class or phylum resembling not an ice-cream cone but rather a very asymmetric egg timer with its point of constriction close to its base. For many comparisons, this is a more accurate picture, and it is often related to the different kinds of egg environment that different embryos start in; for example eggs with very different amounts of yolk. A mammalian embryo supplied with nourishment from the placenta can, in its earliest stages, be quasi-spherical. In contrast, a bird embryo, which shares its eggshell home with a vast quantity of yolk, may have to start off as something more akin to a little disc curved across a small part of the yolk's periphery.

Second, von Baer and other nineteenth-century embryologists only looked at the embryos of a rather small number of species; and these constituted a non-random sample of the animal kingdom. The more species you look at, the more complex the overall picture becomes. How, for example, do you apply von Baerian divergence to the case of groups with indirect development, like flies? My feeling on this is that with regard to embryonic development - that is, the formation of little larvae within flies' eggs - von Baer's picture is still reasonably accurate. However, it can't be so easily applied to that later storm of developmental activity that breaks out during metamorphosis.

Third, it has been suggested by Leiden-based biologist Michael Richardson3 that one of the nineteenth-century embryologists (Haeckel) 'doctored' his drawings, in the same way that Mendel is thought to have 'doctored' his ratios of pea plants. Well, even if this is true, which has been disputed, I don't believe that the doctoring was sufficiently extensive that the basically divergent picture should be discarded. It may be a little messier than some of the drawings suggested, but the cone (or egg timer) remains.

Fourth, we should be cautious, as ever, in inferring a mechanism from a pattern. Let me play devil's advocate for a moment. Suppose that my earlier assertion about advantageous mutation being 'easier' in later developmental stages than earlier ones is false. Suppose instead that 'evolvability' remains constant over developmental time. If this 'null model' were true, we could still end up with a pattern of von Baerian divergence among the ontogenetic trajectories of related species, simply because later stages will be affected by all changes whereas earlier ones will only be affected by a subset of them. However, although this simple and obvious asymmetry could, on its own, explain what we see, I don't believe for a moment that it does; rather I think that it is only part of the picture - see Figure 19. That is, I don't believe my own devil's advocacy. I suspect that the probability of beneficial reprogramming of development is strongly correlated with developmental time (with some complications that we'll get to later); but rigorous quantitative testing of this suspicion remains in the future.

Finally, biologists who work on marine groups characterized by a small planktonic larva - whether the molluscan trochophore or the echinoderm pluteus - tend to be less inclined to support von Baerian divergence as a general phenomenon than those who work on the embryos of terrestrial direct developers such as mammals or birds. This is quite understandable, of course. These small marine larvae should not be expected to behave in the same way as mammalian embryos. There are two good reasons for asserting this. First, because these forms are larvae, they need to make their own way in the world. They float around in the ocean rather than in the womb. Thus external selection will play a greater role here than in true embryos. Second, in at least some cases, only a small part of the larva is used to make the adult. So, just as the adult fly is made from its larva's imaginal discs, in many echinoderms the adult is made from just a small part of its larva - a part known as the 'rudiment'. So internal selection will play a less important role in the reprogramming of the development of most parts of the larva. This means that major changes in early development that occur in the divergence of different species, such as the evolutionary switch from a larva with feeding 'arms' to one without (because it has a supply of yolk) are to be expected. This particular form of developmental reprogramming has occurred in several separate evolutionary lines of sea urchins, as has been noted by the American biologists Rudolf Raff, Greg Wray and their colleagues.4

So it all comes back to the problem of deciding upon the appropriate level for generalization. In my view, if we get this right, a lot of apparent disagreements disappear. If we attempt to construct a 'law' that is applicable to all multicellular organisms, we will probably fail. Certainly neither von Baer's nor Haeckel's 'laws' are as general as that. Rather, we need to strike a balance between science's quest for generality and simplicity on the one hand and the messiness that actually prevails in nature on the other. I wish that I could remember who gave the wonderful advice 'seek simplicity, but distrust it'. This, to my mind, is exactly how biologists should proceed. We all try to build simple general theories and apply them to the widest range of taxa and phenomena imaginable. But then we look, one by one, at these taxa and phenomena and ask whether our generalization is justified. If not, we try one level down, and so on. Sometimes it turns out that our would-be universal law has not lost much generality because it is indeed OK one level down from where we started. In other cases, nature is less kind and we end up with a statement that is only true for a particular family (say) or for a particular short stretch of the whole of developmental time.

Let's see if this hierarchical view of the generality of theories can be made to work in the evo-devo realm. The following attempt to do this will take the form of a series of statements with, in each case, a comment on the level of generality at which I believe it is possible to apply the statement concerned.

'Adaptive evolution of development is caused by the three processes of mutation, developmental reprogramming and natural selection.' True of all multicellular creatures.

'The direction of evolutionary change is caused by the interplay between mutational/developmental bias and selection, and is also influenced in unpredictable ways by historical contingency.' True of all multicells, but probably with considerable variation from case to case in the relative importance of these direction-influencing factors.

'There are important, albeit stochastic, patterns in the 'ease' of achieving selectively advantageous reprogramming; these take the form of trends over the course of developmental time, but not necessarily simple ones.' Personally, I believe that this also applies to all multicells.

'There is one particular pattern that is found in cross-taxon comparisons, namely similar early stages giving way to progressively more different later stages as development proceeds.' It seems to me that this pattern is found, often with the egg-timer complication superimposed on it, in some big chunks of the living world but not others. The 'chunk' to which it does apply is not neat and tidy. It is perhaps truer of vertebrates than invertebrates, truer of direct developers than indirect, and truer of terrestrial organisms than aquatic ones.

'This pattern is caused by variation in the 'ease' of achieving changes, as discussed above, and by the simple asymmetry that early changes affect most stages whereas late changes only affect late stages.' This has the same level of generality as the statement immediately above; but we must add the proviso that the relative contributions of these two mechanisms to the observed pattern have yet to be determined.

'Evolutionary changes in the course of development lead in the direction of increasing complexity.' This is a good example of a statement that can be applied to some lineages but not to others, and that even in the former is only visible over very long periods of time. Many lineages do not increase or decrease in complexity much of the time -

rather, they diversify within a broad level of complexity.

These few examples help to illustrate what I mean by levels of generality. You can, no doubt, come up with additional assertions that you might wish to make about the evolution of development, and see if you can arrive at a defensible view of where they fit, in terms of which is the highest generality level at which they seem to be true.

It's probably worth making the effort to do this. Personally I don't want to list any more examples here. But I do want to sound a note of caution.

My friend and colleague Alec Panchen has said, in his book Classification, Evolution and the Nature of Biology,5 that rather than looking for universal laws, as physicists do, biologists should look for what he calls 'taxonomic statements'. In other words, he is urging caution about extrapolating something that is true within one level (or rank) of taxon to a higher level where it may have exceptions. This is very sensible, and, in one way, I have no wish to argue with it. But I do need to make the point that in relation to generalizations about the evolution of development, descending the taxonomic hierarchy may not always be the right route towards restricted generality, when it becomes apparent that restriction is necessary. Other forms of restricted generality are possible. So, for example, if 'generalization X' applies to all species of direct developer and no species of indirect developer, then, given the irregular way in which direct development is scattered across taxa, we do not end up making a 'taxonomic statement' at all, at least in the normal sense in which that phrase is intended. After all, whatever this mythical 'X' is, it will apply to birds and mammals but not to most amphibians. Yet, going way out from our vertebrate starting point, it will apply to landsnails and slugs, but not to their aquatic relatives. So it is a 'statement of restricted generality', but not a 'taxonomic statement' in the sense of applying only to a particular clade.

100 Pregnancy Tips

100 Pregnancy Tips

Prior to planning pregnancy, you should learn more about the things involved in getting pregnant. It involves carrying a baby inside you for nine months, caring for a child for a number of years, and many more. Consider these things, so that you can properly assess if you are ready for pregnancy. Get all these very important tips about pregnancy that you need to know.

Get My Free Ebook

Post a comment