timi at mendel.berkeley.edu (Tim Ikeda) writes:
>Agneta.Guillemot at historia.umu.se (Ludvig Mortberg) wrote:
>> If you can't identify any reliable morphological traits try sequencing
>> a couple of genes with broad outgroups. What to sequence is hard to
>> decide. Go for something linked to morphology or metabolism.
>Hmm... Sequence data that is directly linked to morphology?
>I don't think that is much of an option at present given the
>uncertainty in mapping the genotype to the phenotype or gross
>morphology. As Ed Rybicki said, this is also not a lot of help
>for bacteria (ie. most of the organisms on the planet) or viruses.
>Metabolic or structural genes seem a better bet and that is part
>of the reason why they have been used.
And even if you could get sequence directly related to morphology, why
would you want to? These are *exactly* the sorts of features that
convergent evolution makes meaningless. What does it mean in a an
evolutionary sense that a certain bug is a rod? Are all bacilli
monophyletic? Even most metabolism related genes aren't very
informative. So a bug ferments lactose. So? It could be a recent (in
evolutionary terms) adaption. The superiority of ribosomal RNA
analysis is that selection does not interfere with evolutionary
information.
While molecular phylogeny is at its most useful with microbes, I'd
have to say *I* trust it more even with higher organisms than
morphology -- Does Ludvig *really* think that creating a tree based on
the number of hairs found on a insect's leg is more valid than basing
it on ribosomal RNA? Why should the number of hairs be unaffected by
selection? If it isn't, then how do you tell convergent evolution from
real data?
