GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-07 > 1246904922


From: Didier Vernade <>
Subject: [DNA] R-SRY2627+ dating
Date: Mon, 06 Jul 2009 20:28:42 +0200


Hello list,

R-SRY2627+ is the "R1b" clade presently known as R1b1b2a1b3 at FTDNA
and as R1b1b2a1a2c at ISOGG (2009). It was previously known as R1b1c6
and the project for that clade is still using this old nomenclature :
http://www.familytreedna.com/public/R1b1c6 .
This clade is a sister clade of R-U152+ and R-L21+ mainly localized in
Spain. The map at this location :
http://sd-1.archive-host.com/membres/up/90261920431217746/IberiaR1b2348samples.gif
was posted on dna-forums . The bottom left map shows the positions of
all SRY2627+ in the main academic papers . Other data confirm the high
level of SRY2627+ south and north of the Pyrenees. However SRY2627+
were found elsewhere and around 100 haplotypes from British isles are
kown in public database (if you know more , please speak up ! ). Here is
a spreadsheet with SRY2627+ haplotypes :
http://sd-1.archive-host.com/membres/up/90261920431217746/R1B1C6_08.xls

The following is about dating of that clade. One point to notice is that
DYS490 is splitting the group in 2 parts : DYS490=10 and DYS490=12 .
This marker is a low mutating one and the absence of any DYS490=11 (in
R-SRY2627+) confirms the split. No SNP was found associated with this
split among 23andME SNPs . The 2 DYS490 subgroups are about the same
size among presently known halotypes. I took advantage of the existence
of these 2 subgroups among SRY2627+ to calculate an interclade variance.
I first eliminated haplotypes with know recent family link (keeping only
one) and I work on 56 STR markers ; multicopy markers were eliminated.
The resulting 36 DYS490=12 and 33 DYS490=10 haplotypes are displayed on
that file :
http://sd-1.archive-host.com/membres/up/90261920431217746/R1b1b2a1b3-var.ods
(exist only in openoffice format)

Here is how I calculated the interclade variance :
Each haplotype of the 36 DYS490=12 went through the successive
following steps .
1 - The haplotype is substracted to each haplotype of the DYS490=10
group.
2 - THe sum of the squares on each 33 resulting "substracted" haplotype
is calculated.
3- The 36 colums of 33 "sum of squares" were collected on a file :
http://sd-1.archive-host.com/membres/up/90261920431217746/variance.ods

The sum of all these results is 34388
The formula 2 M G = "Total sum of squares" /N1 N2 is linking the number
of generation G since MRCA for the 2 subgroups to N1 (33) , N2 (36) and
the sum of STR mutation rate M . For an average mutation rate of 1/400
over the 56 STR markers selected M=0.14 . So , we get :
G = 34388 / 2 x 0.14 x 33 x 36 which is G = 34388 / 333 ; G = 103

For an average generation time of G = 25 years the 2 subgroups splitted
2580 years ago ; about 600 BC in round figures. This fits with the rise
of the Iberian culture and early iron age.

I want to insist on the fact that this kind of results is clearly
different from what you get if you just calculate the regular variance.
In academic paers most authors until recently were refering to AMOVA
which is Analysis of MOlecular VAriance and it's based on regular
variance calculation. The results are then and age around 6000 years BC
and I am making this point to make it clear that despite the "sum of
squares" above the generation time is NOT proportional to the square of
deviations to the modal values (while it is for AMOVA) .

A while ago I posted on this list : "R1b1c7 as a calibrating tool" and I
found results fairly close to the values above. For a reason I can't
figure out this calculation approach can probably be approximated by a
linear relationship between deviation to the modals and generations G
since the split of the 2 subgroups ; this split being probably close to
the origin of the haplogroup .
Comments are welcome.

Didier


This thread: