Humans and apes

[A.P. Jason de Koning]

> This would imply that a DNA sequence is not susceptible to corruption by
> selection. Isn't this wishful thinking? If a gene has a product, such as
an

I did not suggest that selection does not happen at the molecular level. I
only claimed that (positive) selection must be less frequent at the
molecular level than on morphology.  My area of expertise is in fact in
detecting positive selection at the molecular level, so I certainly don't
mean to suggest it never happens!

To clarify, there are two parts to my argument:  1) That most observable
interspecific variation at the molecular level is "neutral" (sensu the
Kimura/King and Jukes Neutral Theory of Molecular Evolution).  For
protein-coding DNA there are 3 relevant points:  a) The genetic code is
highly degenerate (there are many 'silent' substitutions possible);
therefore many mutations have no effect on the presumed target of selection
(encoded proteins);  b)  Rates of amino-acid replacements are almost always
lower than expected given the underlying nucleotide substitution rate,
indicating that a large proportion of mutations never reach fixation due to
negative (conservative) selection; c) The majority of amino-acid
replacements that we do observe in interspecific comparisons (that did reach
fixation) behave as though they were silent.  These amino-acid replacements
appear to have little or no fitness consequences for structural proteins and
enzymes (at least in the short term); experimental mutagenesis of proteins
in the lab bare out this conclusion.  We can consequently use characters
that are known to be silent with respect to the target of selection (the
phenotype).  We can also test whether molecules are evolving neutrally using
model based inference and rate tests.  Silent and non-silent partitions
often yield identical trees with statistical certainty.

The take home message is that even if 'most' observed molecular variation is
not neutral, a lot of it is, and we can both choose to use only (what
overwhelmingly appear to be) neutral characters, and test hypotheses about
the neutrality of characters that we want to use.  (Perfectly no, but we are
getting better at it).  Parallelism and convergence must be detected
comparatively, using fundamentally the same approaches as with morphology.
We know of several interesting cases of phylogenetic reconstruction being
misled by molecules that demonstrate high levels of convergence, but these
cases are apparently unusual (having looked at thousands of proteins, in
comparisons across a vast range of taxonomic levels).  One explanation for
the rarity of  convergence in protein evolution is that protein structures
and functions can be represented by many different amino-acid sequences;
that is, the sequence to structure mapping is highly degenerate.  Hence many
substitutions are unseen by selection.

The second part of my argument is simpler (so it seems): 2) That behavior
and morphology evolve mostly under selective pressure for adaptation.  I
imagined this to be highly uncontroversial--rejecting this idea would be to
say that we think morphology evolves independently of fitness.  Certainly it
may in some cases, but I think the majority of evidence is to the opposite.
Does any one else have something to say about this?

The way geneticists (correctly, IMHO) envision the interesting types of
organismal adaptation are from the top down: selection acts on the genetic
material via the phenotype (which includes behavior, life history,
morphology, and so on).  Since the construction of the phenotype from the
genotype is highly hierarchical, with many redundancies and mechanisms that
buffer disruption by simple mutations (known empirically and
experimentally), many aspects of the genotype accumulate mutations freely.

> effects). There is certainly growing evidence for the eukaryotes acquiring
> certain genes from their prokaryotic symbionts (in extreme cases these
> "symbionts" are now mitochondria or chloroplasts). It is unclear to me how

Right, but we can detect these events because such transfers usually occur
between deeply diverged groups (e.g., vertebrate host, invertebrate
parasite; prokaryote to eukaryote, etc.)  It is obvious that your gene tree
is not the species tree when you have cow grouping with fruit fly...

Cheers,
- Jason