Re: antiquity of segmentation

From
Date

On Wed, 14 May 1997, Bill Shear wrote:

> Last week a very exciting and interesting article and commentary appeared
> in SCIENCE showing the possible homology of segmentation genes in
> arthropods and chordates. DiRobertis, in a commentary, proposed that
> segmentation therefore would have to have been a property of the common
> ancestor of protostomes and deuterostomes.  He went on to list a number
> other properties this hypothetical ancester would have to have had,
> suggesting that it would have been much more complex than originally
> envisioned.

Well, just some rambling thoughts on this.

This work is obviously very interesting, but I wonder what the actual
significance of it is.  It depends on two things: i) whether or not gene
evolution is (de)coupled with morphological evolution, and ii) what one
means by 'segmentation'. 
Supposing (for example) that the segmentation genes were involved in doing
something else before they started specifying segmentation: like building
the nervous system or something.  In this case, the genes could have been
independently co-opted in protostomes and deuterostomes.  If there is a
shift in gene function through time, then one can have 'segmentation
genes' without segmentation.  It is clear, for example, that various sorts
of genes are turned on and off at different times in development, and
perform different roles each time, which would lend some support to this
idea.  It also seems likely that there is 'homology' between genes
involved in building eye structure and limb struture throughout a whole
swathe of animals.  Are we to conclude from this that the last common
ancestor of insects and humans therefore had legs, segmentation and eyes
therefore?  This would create certain problems in conventional ideas of
who these animals are related.  The alternative is the idea of a 'genetic
toolbox': when you want to build an outgrowth, or segment, or make an eye,
you always select the same genetic tool from what's available.  

The second problem is that 'segmentation' itself is rather a vague word.
Are onychophorans, for example, segmented?  Or kinorhynchs?  Or
tapeworms? Lots of animals have some sort of serial repetition without
necessarily being 'segmented' in the way that an annelid or arthropod is:
indeed, even annelids and arthropods are probably not segmented in the
same way as each other.  

So, one can envisage the last common ancestor of prot's and deut's
possessing segmentation genes, but not necessarily expressing them in the
same places as they are expressed in segmentated animals today; and
perhaps not doing the same job either.  This might strike one as being
unparsimonious, in that it might seem easier to argue that in the absence
of contrary evidence one should assume they were doing the same job.  But
there is contrary evidence, first from looking at character state
distributions in large-scale metazoan phylogenies, which would tend to
imply that features such as the coelom and 'segmentationŠ'are derived more
than once, and second (one might controversially argue) from the fossil
record: in the case of the arthropods one can make a case that full-blown
arthropod segmentation is a derived feature within the arthropod
stem-group.  But I suspect not everyone would agree with this...


Graham  Budd