GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-02 > 1266110037


From: David Ewing <>
Subject: Re: [DNA] Variance Assessment wrt back and parallel mutations
Date: Sat, 13 Feb 2010 18:13:57 -0700


Thank you, Anatole, for once again spending time on the Ewing data and
trying to help me understand how to think about this.

I understand that the first step in your method is to compose a "family
tree" based strictly on the Y-DNA STR results, and that it is important not
to do this "just by eye." You did this for me, but told me only of the
members of two of the four principal branches you found and you did not tell
me what method or criteria you used. This is good, because it has forced me
to try reproduce what you did.

First, I consolidated all of my Ewing R:M222 data into a single table and
"ungrouped" them with respect to my Group assignments. The "combined" R:M222
Ewing table can be found at
http://dl.dropbox.com/u/431003/CombinedEwingM222Table.xls.
At the bottom of that table, I have made tables of what I take to be the
four branches you identified, but in the 3rd and 4th branch tables I
included for comparison some men I don't understand why you assigned to the
1st and 2nd branches and highlighted their ID codes in blue. There is new
data on one man, MD, that you did not have to work with.

I added a column and entered for each haplotype the genetic distance (hybrid
mutation model) from the Ewing modal and sorted the table on that parameter.
The four haplotypes at the greatest distance from the Ewing modal (ranging
from 6 to 8) appear in your list of nine haplotypes in the "oldest" branch
you identified: in descending order of genetic distance, HM, TD, RA2 and EL.
So far, this squares very well with what I might have expected, as these
four haplotypes are all in my Group 3, which is defined as "R:M222
haplotypes that are more than GD 5 from the Ewing modal." HM & TD are known
to bear the relationship of 2nd cousins once removed to one another, and
constitute Ewing Group 3a. RA2 and EL are thought to be relatives based on
circumstantial evidence from naming patterns and geographic proximity, but
we do not know their actual conventional relatedness. They constitute Ewing
Group 3b. Group 3a and Group 3b appear not to be especially closely related.
Interestingly, the Group 3b haplotypes are a very close match for the modal
of the R:M222 McLaughlins of Donegal, and we think they probably have a
biological ancestor who was a McLaughlin. The common ancestor of our Group
3a, Group 3b and the rest of the R:M222 Ewings was some R:M222 individual
who lived long before surnames came to be used, and I have no reason to
doubt your estimate of 1475+/-260 years ago--which is along about the time
Niall of the Nine Hostages was supposed to have lived. (Those interested in
Relationship Diagrams can find the two unrelated Group 3 kindreds at
http://dl.dropbox.com/u/431003/Group3RelationshipDiagrams.pdf)

But if I read you rightly, you have three more men definitely in this group,
a fourth with one question mark and a fifth with eight question marks. EF
and EG2 are genetic distance 2 from the Ewing modal; RL is genetic distance
3; WC? and EN???????? are also genetic distance 3. We have eleven men who
are genetic distance 4 or 5 from the Ewing modal who you included in the
more closely related groups, so plainly genetic distance is not the
criterion you are using to identify this most distantly related group. I
asked myself what else distinguishes EF, EG2, RL, WC and EN. They are all
off-modal at DYS 439. Four of them have 12 repeats and one 14 repeats, but
there are other men in the project off-modal at this marker who you did not
include in this group, and all of the Group 3 men match the modal at this
marker, so that cannot be it. For the moment, I am stumped. Let's see if I
can back into an understanding by considering the other branches you found.

(Let me remark parenthetically at this point that while there is no doubt
that some of the conventional genealogy we have to work with is mistaken,
some of it is also correct, and for very close relatives, it is almost
certainly correct. This is a "parenthetical remark" because it is not
intended as a refutation or criticism for your method. For now, I am only
trying to understand what you are doing. Still, there is little doubt that
EN's 3rd cousin is RC (
http://dl.dropbox.com/u/431003/Group2bRelationshipDiagram.pdf)
WC's 3rd cousin is SC (
http://dl.dropbox.com/u/431003/Group1cRelationshipDiagram.pdf)
RL's 4th cousin is GW (
http://dl.dropbox.com/u/431003/Group1aRelationshipDiagram.pdf)
EF & EG2 also have conventional genealogies connecting them with other
project participants, but the relationships are distant enough that we
cannot make as strong a claim for their correctness.
EF (http://dl.dropbox.com/u/431003/Group1bRelationshipDiagram.pdf)
EG2 (http://dl.dropbox.com/u/431003/Group1aRelationshipDiagram.pdf)

So, how about the branch you found that has a common ancestor 300+/- 150
years ago? This branch includes RA, RB, DN, GW, LE, AL and RF. All seven are
off-modal at DYS 576 = 19. TD and RL in the previous branch also have DYS
576 = 19. TD is quite different from the others in this second branch, but
RL is quite similar, and I do not understand why you have not included him
here. (Also, parenthetically, we know he is GW's 4th cousin.) What is more,
two of the men you have placed in this group have DYS 391 = 10 (AL & RF). As
you know, about 1/3 of the men in the closely related group of Ewings (ie,
those within GD 5 of the Ewing modal) have DYS 391 = 10. I have used that
marker to distinguish Ewing Group 2 from Ewing Group 1, and in previous
discussions with you I had thought you also recognized that this marker
distinguished a branch. I am afraid I cannot understand why you included
these two men in this second branch rather than in the third branch with the
other men who have DYS 391 = 10. The same thing goes for EN, who also has
DYS 391 = 10 and was in your "oldest branch." As for those in this second
branch that have DYS 391 = 11, we have conventional lineages suggesting that
the common ancestor lived prior to 1650, and evidence that the earliest
known ancestors of these individuals were living very nearby one another and
were probably closely related, so your estimate is very likely to be
correct. Still, you have excluded from this branch a large number of
individuals for whom there is good documentary evidence for relationships
with individuals you included in the branch--not that I was expecting you to
take conventional genealogy into account for this exercise.

I regret very much that I cannot figure out exactly what you have done. I
had divided my data into three Groups based strictly on Y-DNA criteria with
no "by eye" judgments except in the case of one problematic haplotype (GR).
My Groups were defined as follows:
Group 3 -- all Ewing R:M222 haplotypes that are greater than GD 5 from the
overall Ewing modal.
Group 2 -- all Ewing R:M222 haplotypes that are GD 5 or less from the
overall Ewing modal and have DYS 391 = 10.
Group 1 -- all Ewing R:M222 haplotypes that are GD 5 or less from the
overall Ewing modal and have DYS 391 = 11.
I did assign GR to Group 1b based on his conventional genealogy
notwithstanding that he has DYS 391 = 10, but I accept that he may very well
be mistaken about his lineage, and for the present purpose agree that he
belongs in Group 2. In every other case the criteria were strictly applied.
(Actually, though I did not use it to assign Group membership, there is
another criterion that is met. All of the men in Groups 1 & 2 match at least
four of the five distinctive Ewing markers that are off-modal from the
R:M222 modal, and no man in another group does so.)

I have not been able to discern the method you used to identify the four
branches you found in these data. Your branches somewhat approximate my own,
except that you have an "extra" branch and have assigned twelve individuals
differently than I had done.
The "oldest branch" you found includes four haplotypes that are quite
dissimilar to the other Ewing haplotypes and very clearly have a much
earlier common ancestor, but also includes five others who seem not to
differ systematically from the large group of closely related Ewings.
The second branch you identified seems to be based on the presence of DYS
576 = 19, but RL also has this and was included rather in the first branch
for reasons that are obscure to me. Further, there are six haplotypes with
DYS 576 = 17, but you seem not to have made anything of that.
The third and fourth groups you identify appear to correspond exactly with
my Group 2 and Group 1, respectively, except that you put a few of them in
the previous two groups.

I think I should get clear about your logic for identifying the branches
before going too much further, but let me ask you this question. You have
men with DYS 391 = 10 in three of your branches. In two of these there are
also men who do not have DYS 391 = 10. Is this not tantamount to suggesting
that there have been parallel mutations at DYS 391 = 10?

And finally, a comment about anathema: the dreaded back mutation. My Ewing
Groups 1 & 2 are so similar that I think you cannot dispute that they have a
common ancestor. Indeed, the only way we can systematically distinguish them
is on the basis of DYS 391, which is how we defined the two groups. You also
found two branches that are comprised by almost all of the same men as my
Groups 1 & 2, assigned in the same way. The R:M222 modal has DYS 391 = 11. I
have long been troubled by the fact that the variance in Group 2 is greater
than the variance in Group 1, suggesting that Group 2 has an earlier common
ancestor. Indeed, your calculations suggest the common ancestor of your
third branch (which is almost the same as my Group 2) is 625+/-110 ybp, and
the common ancestor of your fourth branch (which is almost the same as my
Group 1) is 450+/-70 ypb. Since the MRCA of R:M222 had DYS 391 = 11, there
must have been a mutation at DYS 391 from 11 to 10 in the line leading from
the MRCA of R:M222 to the MRCA of your third branch (my Group 2). I take
this to mean that the common ancestor of the two groups had DYS 391 = 10,
no? This means that the founder of the more recent fourth branch (my Group
1, all of whom have DYS 391 = 11) must have had a mutation at DYS 391 from
10 to 11. But that is what the R:M222 ancestor had. Are we not forced to
conclude that there was a mutation from DYS 391 from 11 to 10 and then BACK
to 11?

My focused question at the end of this long email: What is your algorithm
for identifying branches in a family tree before applying your MRCA
methodology?

Warm Regards,

David Ewing


This thread: