GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-12 > 1292262957


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] NW European R1b from Iberia?
Date: Mon, 13 Dec 2010 10:55:57 -0700
References: <0A965AA0-40C8-44B3-9C79-A9EB11938E05@vizachero.com><201012090942.oB99gNdL003003@mail.rootsweb.com><AANLkTi=71WHCmdKtNxFKt2KbkhLfiHqFFMnHXHVonfCL@mail.gmail.com><00ab01cb9a25$c733a870$c2482dae@Ken1><AANLkTi=6YbjmyDW=gkU-JSbLSGvPPUxEXnGoDEdnYxUu@mail.gmail.com><012601cb9a3f$a2287a50$c2482dae@Ken1><AANLkTikF0oWxeXa-xzm4giwZcHS4t=P8kc9y=61FM+O=@mail.gmail.com>


----- Original Message -----
From: "Mike W" <>

> Is there a useful purpose for calculating the TMRCA for the paragroup
> of P312* (or subset of P312 or remainder set or whatever we want to
> call it) when we can easily add all of the more voluminous known
> subclade data (i.e. U152, L21) and get a more complete picture? I'm
> fine with an affirmative answer, but what would be the useful
> application?


[[ I'll start with what would seem most interesting. An interclade variance
between paragroup P312* and subhaplogroup L21 would be an average age back
to the nodes connecting pieces of P312* with the branch line ancestral to
the L21 MRCA. This average would have expectation value younger than P312
MRCA but older than L21 MRCA.

I'd compare that interclade average age with the TMRCA estimate for L21 and
the TMRCA estimate for paragroup P312*. These group TMRCA estimates of
course have the unfortunate feature of requiring an educated guess of
founding haplotypes. But it is fairly straightforward to estimate how much
error one gets in such estimates per mistaken choice of STR founding values.
For some applications one can live with these possible errors, but I suspect
things are now happening so temporally close together in this western Europe
R1b.... tree that these errors will confuse the interpretations. M222 is
exceptional in its young age.

It can most likely be shown that the observable --- InterVar(P312*/L21) -
IntraVar(P312*) - IntraVar(L21) will have a very small statistical sigma.
This means that you may have an absolute statistical error of significant
size in those three age estimates taken separately, their statistical errors
move up or down from true values in sync with each other, so your
statistical confidence in the indicated package of the three age estimates
is much higher. So you get a good sense of the relative timing of things
happening. KN]]



This thread: