Gredosia component and R1b-P312

[JeanL]

Ok, likely this is highly speculative given that ADMIXTURE components are highly variable, and they have a lot of dependence on the population dynamics, and on the whether the run is supervised or unsupervised, population size etc.  Nonetheless I found this really interesting, when looking for European populations that have a low (i.e. <5%) Caucasus component in the Dienekes K12a ADMIXTURE run one comes across the following set of populations:

British_D  3%
Irish_D 1.5%
Finnish_D 1.5%
Swedish_D 1.1%
Norwegian_D 1.2%
British_Isles_D 2.9%
French_Basque 0.1%
Orcadian 0.4%
Orkney_1KG 0.4%
Kent_1KG 3.1%
Cornwall_1KG 2.4%
Argyll_1KG 1.3%

Now save for the Finnish_D population who appear to have a lower Gredosia component (0.3%) all other populations seems to have a very high Gredosia component compare to the Caucasus component, this component appears as follows:

British_D  8.4%
Irish_D 9.5%
Swedish_D 6.1%
Norwegian_D 6.4%
British_Isles_D 7.8%
French_Basque 7.6%
Orcadian 9.7%
Orkney_1KG 9.8%
Kent_1KG 8.4%
Cornwall_1KG 8.8%
Argyll_1KG 10.6%

In contrast European populations with elevated or moderate Caucasus component seems to display a low Gredosia component. For example Greek_D have 37.5% Caucasus, yet only 1.7% Gredosia, neighboring Bulgarian_D shows 28.4% Caucasus, while showing 1.3% Gredosia, even Balts like Lithuanians_D who show 6.6% Caucasus, only show 0.2% Gredosia.

There is much more that can be observed in the spreadsheet found here: https://docs.google.com/spreadsheet/ccc?key=0ArJDEoCgzRKedGdRbkxKMDdlZkJWc21tdkpldWxwVmc#gid=0

The main point here is that there seems to be a west-east correlation for the Gredosia component inside of Europe. This component appears to peak in R1b-P312 rich populations, the correlation is not perfect, but it sure is there. Whether this component is a real one or the byproduct of the sets of populations used in the run is yet a story to uncover. I want to hear some opinions on the matter. Mind all of you, this was just an observation, there are many assumptions that are taken when doing this sort of ADMIXTURE run, however with other things equal, the observation showed that it is rather interesting that there seems to be an autosomal component that shares a somewhat similar geographical distribution inside of Europe as R1b-P312.

PS: It seems Spanish Basque also show the same peculiarity. Looking at this picture from Dienekes blog: http://2.bp.blogspot.com/-9zbUNdodooc/TutQR_n5AZI/AAAAAAAAEXw/zVGNSQwftgE/s1600/_12.png

Not only do Spanish Basque have the highest share of the Gredosia component(6-8%), but they also have the lowest Caucasus component(2-3%).

[rms2]

I've got to admit I read Dienekes' blog, and his Dodecad and other autosomal admixture stuff baffles me. I cannot say I understand it at all, except in some kind of broad, general way.

Gedrosia was roughly what is now Baluchistan, right?

It is surprising that the Basques would get a low Caucasus result, given the theories about the connection between the Basque language and Caucasian languages. But that could just mean the Basques have been so thoroughly Europeanized over the millennia that their Caucasian origins have been thoroughly effaced.