Archiver > GENEALOGY-DNA > 2010-02 > 1267392512

From: "Tim Janzen" <>
Subject: Re: [DNA] : low variance MRCA dates for P310cladesinItalyandSEEurope
Date: Sun, 28 Feb 2010 13:28:32 -0800
In-Reply-To: <>

Dear Vince,
I agree with your comments about the fact that the tree structure is
very important. However, I think that it is also important to keep the
sample size in mind as an important variable in all of this. If you are
doing an intraclade coalescence age and your group of haplotypes doesn't
happen to include any samples from an early but existing branch of that
subclade you will obviously be missing one or more haplotypes that would
increase the intraclade coalescence age if they were included in the group
of haplotypes being used for the calculations.
Getting a balance of representative haplotypes from the existing
branches is also an important variable. Let's say we have a subclade with a
true TMRCA of 4000 years. Let's say we have only two haplotypes from an
early branch of this subclade that is say 3000 years old, but we have 100
haplotypes from a relatively recent branch with a true TMRCA of say 700
years. In such a case the 100 haplotypes from the relatively recent branch
will "swamp" the data from the earlier branch and will skew the intraclade
coalescence age to be significantly younger than it would be if only two
haplotypes from the relatively recent branch were included in the
calculations. Getting the proper balance of representative haplotypes isn't
always easy to do.

-----Original Message-----
[mailto:] On Behalf Of Vincent Vizachero
Sent: Sunday, February 28, 2010 1:01 PM
Subject: Re: [DNA] : low variance MRCA dates for

On Feb 28, 2010, at 3:43 PM, Tim Janzen wrote:

> That said, intraclade estimates
> probably don't underestimate the true TMRCA by a lot as long as the
> sample
> size is adequate.

The difference in the two estimates will not depend primarily on
sample size, but rather on the structure of the phylogeny in
question. Ken was making this point in a previous post with a
different subject line, I think. If there are long branches in the
phylogeny, no increase in sample size will compensate for that.

> In any case the term TMRCA has been used on this list for
> a very long time and I don't see any reason not to use it as long as
> we know
> what we are referring to.

Fair enough, but this common usage arose in an era during which there
were no good alternatives to intraclade variance. Interclade
estimation, Bayesian methods, and so on allow us to do a better job of
TMRCA estimation than we used to be able to do and so the need to
specify exactly what we are doing is arguably greater now than in the


This thread: