Re: Haldane's Dilemma and quantitative genetics

Wirt Atmar wrote:
ErikW writes:
..... ... ...
Its just that thing about neutral mutations and population structure I
don't agree with. I detect population structure because I can detect
neutral molecular variation. One of two neutral variants will fix
sooner or later even across gene flow barriers, provided reproductive
isolation doesn't happen before that. I mean, I see the (presumably)
neutral variants so how can your argument be correct?

The question you're asking, I believe, is: "why am I seeing neutral
mutations in my populations if you're claiming they're so difficult to
establish?" Let me answer that question in a moment. First please allow
me to reiterate my claim.
... ... ...
A novel neutral allele begins
life just an iota off of the 0% axis, and as in any gambler's ruin
markovian process, must play against the "house," and is therefore
enormously more likely to go extinct than fixate, being initially just
a hair's breath away from the extincting absorbing barrier.

But in a geometrically distributed population, the situation is much
worse still. In Walsh's graph, the probability of increasing or
decreasing an allele's frequency is 50-50. But that's not true in a
geometrically distributed set of demes that are communicating with one
another. The probability isn't 50-50, but perhaps more akin to 80-20 in
favor of the "house" (the standard allele winning at each generation),
with the precise ratios being determined by the geometries.

Not only does the neutral allele ride right on the edge of extinction
initially, its odds of surviving to the next generation are very poor.

ErikW asks "I see the (presumably) neutral variants so how can your
argument be correct?"

Wirt says "the odds of a neutral allele surviving are very poor". On
the face of it, this does not explictly satisfy Erik's unease, indeed
appears to contradict him. But it does not.

As I posted previously [30/06/2006 16:31]:
<<<
Is this again just a word-confusion? It is perfectly possible for any
*specific instance in one individual* of a neutral change, at the
molecular level, to be 'almost impossible' to go to fixation within a
population; whilst *so many* such instances arise in many/every
individual[s] in the population that, at the population level, every
generation several of these instances are fated to go to fixation.

From rather abstract reasoning I would expect there to be of the order
of 10 novel neutral mutations, at the molecular level, in every single
human -- can anyone help me with an actual authoritative figure here?
If Wirt is saying that for each one of these, it is 'almost impossible'
for it go to fixation, I can agree. But enough 'almost impossibles' add
up to a near certainty, and I am not sure whether Wirt's claim is also
held by him to be true at the population level. Do his calculations
assume that there is only 1 novel neutral mutation in a population at
one time, or do they assume that every individual has novel neutral
mutations?


Can I check whether Wirt accepts or denies that his statements are
completely compatible with there being several neutral mutations
appearing *and going to fixation* every generation -- much as ErikW
says that he sees? If he accepts this, then the apparent contradiction
between ErikW and Wirt is not a real contradiction.

Inman Harvey
--
Inman Harvey >> Evolutionary and Adaptive Systems
Group <<
>> COGS/Informatics, Univ. of Sussex, Brighton BN1 9QH,
UK << inmanh@xxxxxxxxxxxxxxx
http://www.cogs.susx.ac.uk/users/inmanh/ <<


.