Faculty of Biological Sciences, University of Leeds

These pages have been left in this location as a service to the numerous websites around the world which link to this content. The original authors are no longer at the University of Leeds, and the former Centre for Human Biology became the School of Biomedical Sciences which is now part of the Faculty of Biological Sciences.

Evolutionary Developmental Biology Lecture 10

What is it possible/necessary to change?

Phylogeny and ontogeny are very similar: the only difference is that ontogeny goes along a very fixed path - a hens egg always finishes up as a chicken, a mouse's egg always ends up as a mouse but ontogeny, according to Darwin, is a random walk. There is obviously a way of modifying phylogeny - does this mean that there must be a way of changing ontogeny?

So, how do we start to study the problem? First of all what are the differences between animals? In general the differences are anatomical ones, differences in shape. The basic chemicals involved are pretty much the same. Haemoglobin is found in annelid worms and in vertebrates. The base sequence is very similar in each case. Cytochrome, collagen, many molecules are widespread and very similar at most critical points wherever they are found. The basic fixings between cells (desmosomes, gap junctions) are the same. Animals are largely made out of the same stuff, but it is arranged differently. Anatomical differences arise anew in each generation ( we have seen that the early larval forms of many marine organisms are very similar, so much so that we have to wait till they grow up to see if we have a mollusc or a crab: early chick, mouse and human embryos are similar) and also arose through time by evolution.

If Haeckel was right, and animals are all related we ought to be able to make supergroups of phyla, using Bauplans. In fact we can, but we can do this in different ways, because we don't know the 'correct' rules for making these supergroups (OVERHEAD).

Some of the more sensible guesses are: (OVERHEAD)

1. Number of germ layers - diploblasts versus triploblasts

2. Type of symmetry - bilateral versus radial

3. Pattern of embryonic cleavage - some animals eggs divide spirally, some don't (OVERHEAD).

4. Development of the coelom - interior cavity (OVERHEAD)

5. Original opening forms mouth or anus (OVERHEAD)

6. ventral or dorsal nervous system (OVERHEAD)

All these are perfectly respectable, but we don't know which, if any is valid. It is quite good fun playing this game. Nielson (1995), as we have already mentioned, did to good effect and has come up with a tree which explains most things, (OVERHEAD) although five phyla (non of them very large) are problematical. He starts off with the monophyletic phylum as a group and works out relationships from there.

Basically, Nielson contends, his idea about pelagic larvae and settling down on the sea bed at various stages works. This implies protostomes versus deuterostomes is important (OVERHEAD) The weak point, he admits is the derivation of deuterostomes from the four holed trochaea, because there seems to be no very good reason for it to happen that way.

Most respectable animals also have mesoderm. The derivation of this third germ layer is also a problem. Mesoderm can arise (OVERHEAD) from ectoderm (ectomesoderm) in the way we are familiar with in the human head but this is an uncommon way of making it. In flatworms, roundworms, molluscs annelids and arthropods mesoderm comes from a small number of cells set aside, rather like germ cells, early in development (4d mesoderm). In echinoderms it arises as outpushings of the primitive gut in the gastrula. There is nothing to suggest that 4d is ancestral to endomesoderm or vice versa. It looks as if these two types have evolved independently.

Coeloms also seem to have arisen separately in various groups. The logical classification would be acoelomates (coelenterates) versus coelomates (deuterostomes) but awkwardly protostomes seem to have a coelom too. If this is so then either:

the protostome coelom was derived from the deuterostome one


the deuterostome coelom was derived from the protostome one


the coelom evolved twice.

The answer is we don't know, but the first two don't work very well.

So we can set up at least two groups of animals, even if we don't follow Nielsen all the way.

If these animals are all related to each other, as I suggest, and the differences arose by changes in the development of different kinds of animals then we have to start looking at how development might be changed in some detail. What sorts of factors might affect an egg as it develops? We can make a list: (OVERHEAD)

maternal genome

paternal genome

genome of offspring at fertilisation


Early views were that environmental agents affected developing embryos directly - Lamarkism. The classic example of Lamarkism relates to the tails of mice. Lamark said (in essence) if you take a group of mice and cut off their tails, and continually repeat this process, eventually mice will be born tail-less Evolutionary change, recognised by small alterations of type, would thus produced by environmental change, and the change is passed on to subsequent generations.

Probably the first to test this was Geoffroy Saint-Hilaire. Whilst Napoleon was in Egypt in 1799 Geoffroy made a grant application for 600 chicken eggs, an incubator, 112 pairs of pigeons, ostrich eggs, money for measuring instruments and salaries for assistants. What was he doing? He was testing a theory that the sex of birds was determined by egg shape. A little later we find him sealing eggs with wax, removing them from the incubator for a day, soaking them in water and sealing the oviducts of chickens to retain eggs in 'a uterine environment'. What did he find?

Well the sex/egg shape ratio was wrong, but he did find changes, such as retardations caused by cooling which he identified as normal stages of lower vertebrates (recapitulation again). He also found that growth in one element due to an increased blood flow led to decrease in other elements. But on the whole he was impressed by the limited amount of variation he could induce in the common ground plan. The basics were not changed much by the environment - and of course he had never heard of genetics.

So his theory was what we would now call epigenetic, that development is modified by environment, and his conclusions largely negative. The basic plan is conserved: awkward large changes, such as the origin of birds by modification of limb structure were christened saltatory - a great leap forward of unknown cause.

Constraints and the Bauplan.

The Bauplan seems then, fairly fixed, and we can only do so much to change it by modifying the environment. Why should this be? If evolution is random change then all aspects of development should be equally likely to vary. The fact that this is not so worried some people.(OVERHEAD) Whitman said ' but if organisation and the laws of development exclude some lines of variation and favour others there is certainly nothing supernatural in this and nothing which is incompatible with natural selection.'

Mayr suggested that 'some components of the phenotype are built in far more tightly into the genotype than others' and that 'certain components of the phenotype may remain unchanged during phyletic divergence' 'Why is the chordate type so conservative that the chorda is still formed in the embryology of tetrapods and gill arches still in that of mammals and birds?'

These views imply that natural selection is not, in fact, free to act randomly. Some structures are either protected from change, or conservative or inert. In other words there are constraints as to what may vary. Lets look at some of the possible constraints that may exist.(OVERHEAD)

Structural constraints.

Structural constraints are more likely to affect unicellular than multicellular organisms. Being very small may be a problem because of external forces like Browning movement: being very large may be a problem because of the need to transport things from one side of the organism to the other by diffusion for instance. Things like surface area/volume considerations have been eliminated by forming colonies and eventually multicellular organisms. Other mechanical constraints may affect the skeleton: you are not allowed to construct a limb skeleton, say, that will not take the weight of the organism it is supposed to support.

Genetic constraints.

The maximum rate of mutation and the numbers of recombinations formed will both constitute genetic constraints on evolutionary change. We shall see later that some genes are phylogenetically highly conserved, since we find them in many species and higher taxonomic units. These are concerned with fundamental aspects of development and so are somehow prevented from mutating or have a very high priority service contract. We can imagine a priority on any gene product, varying from absolute invariability for the most vital, through partially restrained to free for all. Free for all represents the maximum possible mutation rate: anything less is constrained. Groups of characters may also be associated genetically into suites, such as the suite of genes that makes an upper limb : we recognise these by lack of genetic covariance - i.e. they are all inherited together. The suite of genes governing Aunt Maud's nose shape are less conserved. Pleiotropy (the action of genes in multiple tissues) also acts as a constraint. A particular gene may have two roles in two different organs - it might make Darwin's tubercle on the ear and that little wiggly bit on your little toe. If this is so then you can't have one without the other. It has been argued that the old 'one gene one enzyme' theory i.e. that one bit of DNA makes one protein means that pleiotropism doesn't really exist: this may be so but in practice during development genes behave as if it does.

Developmental constraints

These are limits to variation brought about by the interdependence of developmental processes, and constitute much of the content of the rest of the course. They are recognised by the invariant patterns which persist throughout morphology, and long periods of evolutionary time. Maynard Smith defined them as 'biases on the production of phenotypic variability brought about by the nature of developmental systems'.

Cellular constraints

Limits to rates of cell division, secretion of cell products, cell migration and/or metabolic efficiency are four examples of cellular constraints. They are, of course, associated with developmental constraints, because cells are what is developing.

Metabolic constraints

A tissue which is dependent for its function on a good blood supply, in order to support a high metabolic rate, is limited in its capacity for change. A tissue with a low metabolic rate and no need for a good blood supply has more options. This is probably why cartilage is the most common supporting skeletal tissue in vertebrates and invertebrates, and appears often in ectopic tissue. Vascular tissue, like bone, is limited in its options.

A special case of metabolic constraint, especially in birds and mammals is maternal metabolic constraint: if mum can't cope then the offspring dies.

Functional constraints

As embryos develop, and especially as functions like feeding and respiration begin organ systems become functionally connected. The skeleton of the jaws and branchial apparatus becomes involved with the muscular system: the muscular system becomes involved with the nervous system. The more of these links that develop and the earlier they do so the more difficult it is to uncouple one system in order to modify it. Think of the interrelations of the mesonephros, the metanephros and the male and female reproductive systems and ducts - very constrained by function.

Constraints and the Bauplan

The constraints that we have identified, and no doubt others, have a huge effect on the Bauplan. Basically, they conserve it. Individuals that conform to the Bauplan during development survive: those that do not are eliminated. Spontaneous abortion, failure to implant, embryonic and larval mortality are what ensure that a frog's egg always gives rise to a frog and women have reasonably normal children.

Do we believe in the possibility of constraints and the elimination of variation? Well, there is one good biological example of constraint and perfect conservation - the genetic code. That varies very little and presumably has huge constraints mechanisms built in to preserve it.

So we can now make a statement: (OVERHEAD)

Variation associated with the production of the Bauplan is minimal: constraints play a major role, and selection plays a minor role. The role of selection is in fact negative: those individuals that do not conform are selected against.

We may ask ourselves whether this statement is not the product of a circular argument? Those features that we have selected to be parts of the Bauplan are highly conserved, so of course they are not changed. This is the same circular argument that led Darwin to the selection of the fittest: who else but the fittest will survive? But if we don't take it too literally, and suggest that all constraints are 100% it is a decent working hypothesis: remember that some characters are more constrained than others.


Constraint outwith the Bauplan.

At a second level we have the other characteristics, those which we have said are not part of the Bauplan. We must allow these more freedom by definition (they are not constrained) and allow mutation and selection to work on them. Constraint plays a lesser role here.

One drawback is that this division immediately splits characters into two leagues and necessitates first of all some sort of judgement, presumably applied externally by us, as to whether something is Bauplan or is not and secondly we need a mechanism to apply constraints. Perhaps a more comfortable view is that constraint can vary from 0-100%, and that we admit a character to the Bauplan at, say, 60% constraint. - a purely nominal level. Below this things are fairly easy to change, above this things are more difficult. The more fundamental a thing is, or the more necessary, then the bigger the lock on its chastity belt.

Epigenetic organisation of embryonic development.

We now think we can begin to answer one question, why does a frog's egg always produce a frog?- because of constraint The next question is even tougher: how does an apparently simple structure like an egg give rise to a complicated structure like a frog?

Well, it has to be preformation or epigenesis; either all those structures are there already, and unfold like the homunculus, or they gradually and progressively appear. If you are a preformationist form already exists in the egg. If you are an epigenisist the information necessary to define form exists in the egg. So epigenesis is:

(OVERHEAD) The sum of the genetic and non genetic factors acting upon cells to selectively control the gene expression that produces increasing phenotypic complexity during development.

The first to make a sensible comment on epigenesis was Aristotle, who observed developing chick eggs and saw he progressive appearance of new features, and how they were similar to developing fish and mammals. He therefore came down as an epigenisist. The next man to take this approach, and to improve on Aristotle was Coiter, a Dutchman, 1800 years later.

But Aristotle was in a minority and between Aristotle and Coiter most scientists were preformationists until Harvey and Wolff persuaded the majority otherwise. Wolff saw clearly that parts of the embryo developed from non-equivalent structures: that blood vessels appeared where none were seen before and that the tubular gut developed from a flat plate. Pander's discovery of the germ layers in 1817 provided substance for Wolff's views: structures arose from germ layers. Wolff also recognised, importantly, that the sequence of appearances of parts both depended on pre-existing parts and governed what was to come.

Teratological experiments such as those of Geoffery on eggs proved that the malformed embryos were not preformed: he chose eggs at random and applied treatment which caused the abnormalities. If malformed embryos were not preformed then presumably neither were his control normals.

The big problem however was this: preformationists believed that the pre-existing structure was unfolded: very rational and satisfying, even if there was a minor hiccough in the storage of many generations, like Russian dolls, inside one another. Epigenesis had the major snag of structures arising from non existing structures - a good theory, but one as yet without a mechanism behind it.

In the end it was probably regeneration that clinched the issue for epigenesis. If you cut off an amphibian's limb, or a tadpole's tail, or a Hydra's body it regrows. There is no sense in which the regenerated part was present preformed in the amputation stump. So in one way epigenisis is clearly the winner: embryonic structures are not preformed in the egg. Yet in a sense we now know preformation wins as well - because the instructions for making a frog, rather than the frog itself, lie in the DNA of the frog's egg.

But as well as the DNA the frog's egg contains other things: organelles and cytoplasm. Are they implicated in development? Yes of course they are, because in most organisms development can proceed, at least for a time, without the zygotic nucleus. This varies from the 2-4 cell stage in the mouse, through 4-8 cells in the pig, 8-16 cells in the sheep, to twelve divisions (3-4000 cells) in Xenopus. This however, is a diversion: development never goes the whole way without a nucleus.

How does development work then? We know that zygotes contain copies of all the genes in the genome, and pass these on to their descendants. We also know that not all genes are active in all cells.

Lets now look at genetic and non genetic factors. Although they are likely to be complex and varied, and although we can't quite specify what they are at this stage, if there is any truth in epigenesis we should be able to see development as a hierarchical decision making process. Throughout there should be a series of decisions made by individual cells or groups of cells which are of this form: (OVERHEAD)

Do A/Don't do A

Do A/Do B.

Hierarchical systems have some interesting properties: some of these are common to hierarchies, and not a function of their content and they can evolve faster than non-hierarchical systems of similar size.

This page is maintained by Steve Paxton