Archiver > GENEALOGY-DNA > 2009-12 > 1260336587

From: (John Chandler)
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent
Date: Wed, 9 Dec 2009 00:29:47 -0500
References: <><A93225B377724B83BF5250F368878955@anatoldesktop><>
In-Reply-To: <>(message from David Faux on Tue, 8 Dec 2009 08:55:30 -0800)

David wrote:
> Alas, time - related environmental
> changes mean that correction factors need to be introduced, and from time to
> time re-evaluated. Is this not what Zhivotovsy is doing?

Certainly not. I think it would be helpful here to review the basis
of the statistical dating technique. It can be shown that a
fixed-size population with every member having in due course exactly
one offspring will experience a linear growth in the variance of its
genetic markers subject to mutation. The rate of growth in the
variance is simply the mutation rate duly averaged over the markers
analyzed. If you know the current variance and the rate of increase,
you can easily calculate the date at which the variance must have
started out at zero.

Unfortunately, populations are not like that, and real populations do
not have linearly increasing variances. I repeat: in real life, the
variance DOES NOT grow linearly with time. The intraclade dating
technique is flawed at its very foundation, and all it can really do
is place a lower bound on the age because the rate of growth is
bounded above by the (measurable) mutation rate. In fact, the rate of
variance growth changes with the size of the population, and the
population growth rate, and various socialogical and economic factors.
In short, every population is unique, and every epoch is unique as
well within a population's history. What Zhivotovsky et al. did was
offer a "one-size-fits-all" estimate assuming that all populations at
all times are alike -- an assumption that is simply wrong. Indeed, Z
went on to publish some simulations that showed he could reproduce the
lowered average growth rate over various time periods by assuming
particular values for the (simulated equivalent of the)
above-mentioned factors. No sweat. The problem is that, in real
life, those factors are not measurable for populations in the past,
and so we cannot calculate what the variance growth rate should have
been at any given place and time. Again, all we can say is that the
rate is never higher than the mutation rate.

> Why would one assume that the mutation rate of this shrinking
> violet was the same 3000 years ago as it is today? No one has observed
> father - son mutation rates from Bronze Age times as they have the
> bi-layered core samples from my miromictic lake.

I hope you can now understand why a change in the variance growth
rate need not imply a change in the average mutation rate. You've
worked very hard to convince yourself that the mutation rate is
changing, but it was a wasted effort.

John Chandler

This thread: