GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2012-02 > 1329101568
From: Bonnie Schrack <>
Subject: Re: [DNA] Out of Africa
Date: Sun, 12 Feb 2012 21:52:48 -0500
> Back to haplogroup A. It would be very useful and educational if you
> kindly answer the following questions, which I numbered for your
> convenience: 1) Do you define haplogroup A (according to the
> ISOGG-2012, for instance) as having SNPs M91 and P97 and the principal
The ISOGG tree now seen on the site has nothing to do with me, as it was
created before all the discoveries of the past year, and before I became
responsible for this haplogroup.
Due to some problems that have had to be dealt with by another person
involved with the ISOGG tree, which I could not have any effect upon,
the actual 2012 tree has still not been posted. I've posted it, in the
meantime, at DNA-Forums in the A haplogroup section, with ongoing updates.
This tree, regardless of whether it's seen at ISOGG or elsewhere, is,
and will be for some while, in the process of having many branches added
and re-defined, so the specific names of the clades should not be taken
overly seriously. What matters is the branching structure, the
positions of the SNPs and tested samples. The names will no doubt be
worked out better when the dust has settled, in a little while.
M91 and P97 are SNPs that are derived in BT and all downstream
haplogroups, and ancestral in A. They were previously used to define A
when its true structure was not known.
> 2) Then, do you believe that other haplogroups descended from
> haplogroup A? Or maybe from haplogroup B?
Yes, all the other haplogroups do descend from haplogroup A; there are
no ifs, ands, or buts about it. I'll go into more detail below.
The complaints about the illogical naming of A clades are
understandable, but it's impossible to do it within the normal rules.
As Cruciani laconically put it, A is not a monophyletic clade. It isn't
one clade at all; it's a collection of distinct clades. We are simply
using the label "A" for all the clades that branch *upstream of BT*.
There are certainly SNPs that are derived in all members of Hg. A, but
these would also be derived in all modern humans, so they are not very
useful for defining A!
Notice that Cruciani, despite having discovered the basic structure, did
not have the nerve to propose new names for any of the branches. That
has been left to us mere mortals!
At a minimum, there are at four major haplogroups within what we're
calling the A haplogroup: A0, A1, A2 and A3 (some of these with major
splits at their base, which could arguably be considered separate
haplogroups). However, even if we renamed these as U, V, W, and X, it
would not resolve the inconsistency that BT is not going to be called
something like A1b3, which it would if we were following more usual,
logical naming conventions.
To re-state the basic structure, the earliest split is between A0 (a new
name) and A1. Then A1 splits into A1a and A1b. A1b then splits into
three branches. Regardless of what names you want to give them, one of
the three branches is BT, the node from which B and all the other
haplogroups descend. First B branches off from it, and then all the
others from CT, the brother branch of B.
You can see that BT is downstream of A1b because the SNPs P108 and V221,
which define that node, are derived in all three branches, A2, A3, and
BT, but ancestral in the other A clades, A0 and A1a. P108 used to be
defined in the reverse direction, like P97 and M91, but now we know that
it's ancestral in the earlier A clades, and derived in most men.
We just received some A2 WTY results, from a San sample donated by the
U. of Arizona, and sure enough, that sample has a the derived T at P108,
and the earlier Armenian A3 WTY had a T as well, but our new A0 WTY for
Jones, has the ancestral C for P108. We also have P108 C results for
A1a members of our project. If anyone checks the WTY result for all the
other haplogroups, P108 will be T.
> 3) If not, not from haplogroup A and not from haplogroup B, but, say,
> from haplogroup BT, then where the haplogroup BT might have appeared?
> In Africa? >From which haplogroup, if not from A? And when, what do
> you think? Thank you. I would appreciate to learn from your insight.
I don't think you really want to learn from my insight, since your ideas
are already formed, but I do have data that, no doubt, you would like to
hear about. No problem, all the data is being shared freely.
> Haplogroup A is the oldest one. However, it is not ancestral to
> non-Africans. Furthermore, it is not ancestral to haplogroup B as well.
You're wrong about this, and it's something so plain and simple, the way
the phylogenetic tree is structured, with a great redundancy of SNPs,
that you will ultimately not be able to wiggle out of it, without
claiming that a bunch of SNPs don't actually exist, or something silly
Or, I suppose, to avoid having "A" be ancestral to BT, you could reject
the name of "A" for all the clades ancestral to BT (A1 and A1b) and call
them something else -- using "A" only to refer to A2 and A3, the brother
clades to BT. But what would be the point of those semantics? Your
real aim is to avoid acknowledging your African ancestry, your descent
from *modern humans* in Africa, which in fact, we all share.
The question of how far back you should include humans in haplogroup A
is immaterial when we're dealing with living humans, who are all clearly
modern (regardless of any claims). But should we include humans from
long ago, that weren't fully modern, in haplogroup A? I don't think so
-- the point of our haplogroups is to find the branching structure of
the lineages of modern humans.
Until proven otherwise, I would expect even the earliest hg. A lineages
to converge within the period of the existence of modern humans in
Africa. No doubt there was occasional admixture from archaic humans who
were living side by side with them -- just as we mixed with the
Neanderthals -- but that need not affect the phylogenetic tree, any more
than it did in Europe, where there are certainly not any Neanderthal
The dating to 142,000 years ago by Cruciani for the A0/A1 split is
significant. Archaic humans need not be invoked in that time frame.
If that split had been dated to 500,000 or 1,000,000 years ago, it might
have been another story.
If you're complaining of Zhivotovsky's and such "population" mutation
rates, which give much *older* dates than genealogical mutation rates,
that's interesting, since by substituting genealogical rates, the dates
of these earliest splits within A would be dated much more recently,
easily within the period of modern humans.