> I am sure you do (stand by what I said) but should you at least not try
> to address Kevin's arguments?
I have not for several reasons:
1. To begin with, I am a statistician, not a biologist. I am not expert
enough in biology to address the various nuances of meaning contained in
the word "mutation." Kevin, and perhaps biologists in general, currently
differentiate between "mutation rate" and "substitution rate." Fine. He
argues that mutation rates within a given gene are largely constant
(comparing it to the decay of a given isotope), but the rate varies
sometimes enormously from one gene to another (comparing it to variability
among isotopes). Interesting analogy. Beyond that, he differentiates
between the occurrence of a mutation and whether or not it becomes fixed.
2. While perhaps a nice lecture for Biology 101, this seems to me to raise
a lot of issues that go beyond the clear intent of my remark.
3. I do not have the time to devote to this subject in a way that perhaps
it deserves.
4. Glenn has already made my point (better than I did) when he wrote:
> Yesterday I got a chance to go to the library and peruse the last 6
> months of Science Magazine. There is an interesting article on the
> inconstancy of mtDNA as a clock. (Evelyn Strauss, "Can Mitochondrial
> Clocks Keep Time?" Science 283(1999):1435-1438). There are several items
> which this article discusses in relation to the clocks:
>
> "It's now clear that in many cases, the main assumption underlying
> molecular clocks doesn't hold up: Clocks tick at different rates in
> different lineages and at different times. And new work on the biology
> of mitochondria suggests that their evolution may be more complicated
> than researchers had suspected." Evelyn Strauss, "Can Mitochondrial
> Clocks Keep Time?" Science 283(1999):1435-1438, p. 1435
I do know something about statistical modeling. To begin with, one assumes
things that are surely false in an absolute sense in order to keep the
model simple enough to work with. There is a constant battle between
making a model simple enough to work with and complicated enough to be
realistic. In the present case, the problem is more severe because one is
trying to model a process that is very poorly understood. So what is
realistic is unclear. Still one models, as one must in science, and sees
what happens. What comes out might be insightful or garbage, or someplace
in between.
What we have seen in the case of human mtDNA, foisted on us by what Glenn
calls the "bullying Out of Africa School," is a model based on severely
flawed assumptions. The worst of these is surely the assumption (once
taken to be an indisputable fact) that human mtDNA is passed along with no
paternal influence. If I might insert a personal note at this point: I can
recall having had various conversations with my cousin Elwyn Simons and
his wife Friderun Ankel-Simons, long before her paper with Cummins (1996)
was written, questioning the conventional wisdom on this matter. At the
time, I raised the "paternity issue" on the reflector (evolution) forum,
and Terry responded, quite properly, with what was then accepted dogma. As
Glenn has pointed out, that dogma has been overturned, and, with it, much
of the statistical modeling surrounding the existence of "Eve" has been
trashed. (I am not referring to the Eve in Genesis.)
It is my belief that a second serious flaw in the modeling is the
assumption of a constant mutation rate. (Even if there were a constant
rate, one would need to have a decent estimate of this rate, but that is a
secondary issue, and no issue at all if the rate is not constant.)
Can I prove that I am right. Of course, not. But, as Glenn has pointed
out, there is literature which supports the view that the rate is not
constant. No doubt there is other literature which asserts the opposite. I
would further argue that it stand to common-sense reason, given the
complexity of the matter, that fixed mutations do not (can not?) occur
across hundreds of thousand of years at a constant rate. The burden should
be on those who believe it is constant. And that is a heavy burden.
As I said, I do not have time to pursue this matter. But let me suggest
some papers which might be of some relevance:
a. a paper entitled "Mutation rate varies among alleles at a
microsatellite locus: Phylogenetic evidence," appearing in the _Proc. Nat.
Acad. of Sci._, Dec. 1996.
b. "Mutation rate heterogeneity and the generation of alleles diversity
at the human minisatellite," appearing in _Human Molecular Genetics_, Nov.
1996. (One of the authors is C.A May, who is well known for his
statistical modeling of evolutionary processes.)
c. "Buffon's needle statistics are not applicable to calculate the
mutation ate of the number of structural genes in the human genome,"
appearing in _Experimental Hematology_, Sept. 1996. (Whether directly
relevant to the present subject or not, the title is surely intriguing.)
d. "Mutation rate: a simple concept has become complex," appearing in
Environmental and Molecular Mutagenesis_, 1998.
e. "Elevated mutation rate in mutT Bacteria during starvation," appearing
in _Journal of bacteriology_, May 1996.
f. "Sex differences in mutation rate in higher primates estimated from AMG
intron sequences," appearing in _Journ. of molecular evolution_, April
1997.
Gordon Simons