May 21, 2008

Exploring Y chromosome haplogroup frequencies in Greece

The samples:



Note that haplogroup J2-(DYS413<=18) should be fairly close to J-M410, and J2-(xDYS413<=18) should be fairly close to J-M12, since there seem to be few J-M410's with the "long" DYS413 alleles. The correlation analysis:

The only correlation that reaches significance (and at the 1%) level is between the aforementioned J2 groups; this may point to a differential representation of J2a and J2b in Greek sub-populations.

A dendrogram of the 19 populations:



The principal components plot:

As it appears, J2-(DYS413<=18) is positively correlated with PC1, while J2-(xDYS413<=18) is negatively correlated with it. On the other hand PC2 contrasts haplogroup DE on the one hand with mainly R1 on the other; indeed, the correlation between DE and the combined frequency of R1a and P*(xR1a) -which mostly corresponds to R1b and indeed R-M269 in Greece- is almost significant at the 5% level. Ratio of J2*(xDYS413<=18)/J2 in Greece: A north-west "Epirotic" concentration of high ratios is, I think, evident, contrasted with a south-western "Aegean" concentration of low ratios.

12 comments:

Dean said...

I wonder why haplogroups I and R1a correlate poorly. There should be an explanation as to why that is. The weak correlation between I and R1a throws a bit of a wrench in to the Greek-Slavic hypothesis. I and R1a should correlate strongly if there is a Slavic basis of part of modern Greeks, with R1a dominating.

The other strange correlation is between haplogroups I and G2.

Dienekes said...

>> The weak correlation between I and R1a throws a bit of a wrench in to the Greek-Slavic hypothesis.

I and R1a could have entered in Greece in medieval times, because they occur more frequently to the north of it, but they could also have been present in ancient times. Until their internal structure is better resolved we are really not in a position to say much.

>> The other strange correlation is between haplogroups I and G2.

I don't see such a correlation, at least in this data.

Dean said...

There is something bizarre about R1a and I in Greece. Their frequencies are not as clean-cut there as they are in other northern regions, from where medieval invasions could have come, and where R1a is dominant.

Given the skeletal diversity in ancient Greece, it's probably unlikely that there existed only Neolithic haplogroups.

Maju said...

Very interesting recopilation. It is a pity that southern mainland Greece (specially)is poorly represented. Northern Greece and Crete instead have a lot of data.

I see four clades with special importance of their own, to which it may be needed to add R1b (P(xR1a)) that is numerically important too but has not any clear structure.

E3b (DE) is maybe the most important clade in mainland Greece. It is well represented and even dominant almost everywhere, with the exception of Agrinion and Larissa and with some decrease in Macedonia too. It is intresting that Sesklo/Dimini, considered the craddle of European Neolithic, has it so strong. While it may explain the why of Balcanic E3b, it also seems to question the demic migration hypothesis for European Neolithic beyond the Balcans (already quite battered).

Crete seems largelly (but not fully) low in this clade.

J2a1 (J2(DYS413<=18)) is specially important in Crete (except the small locality of Lasithi Plateau that was deserted and later recolonized from the mainland in times of Venetian domination). It is much less important in the mainland instead (except in Larissa).

R1a It shows, rather logically, a strongest presence in Northern Greece, where it could have arrived in at least two migration epysodes: the original Indo-European (Greek) arrival or the Slavic migrations (or both). But Balcanic Slavs are much stronger in I than R1a, so I'd rather tend for the first explanation.

R1a is also important in Rethymnon, a Venetian foundation (though maybe has another better explanation it anyhow seems a localized phenomenon in Crete).

I It is spread around but it is maybe stronger in the North (except Epirus, and specially strong in Serrai), having also some presence in the South/Aegean but irregularly. Being Greece a Balcanic country, I think it's just normal to find relatively high I. Still it does seem to partly (not fully) correlate with R1a in the North (except Serrai). But not in the rest of the country, where it's normally much more important than R1a.

...

Regarding the correlations table I do not see clear their statistical significance (P is lacking), so I'll abstain to comment.

...

The PC graph is interesting instead: with PC1 being basically related to J2a1 (yes or not, the J2* correlation is weak) and PC2 by the R1b/E3b dichotomy.

PC1 basically says Cretan/Larissan or other. But PC2 shows a negative correlation between E3b and R1b that I had not noticed before. Looking at the data again there seems to be something of that but there are exceptions too (Nea Nikomedia, Ioanina and Chios).

My two cents anyhow.

Unknown said...

Are there any real DE Haplogroups found in Greece?
I mean DE not being either E or D Y-DNA Haplogroups.

Dean said...

If haplogroup I in Greece is a result of Slavic medieval invasion, it's possible that many of the Greeks with I were Slavicized Balkanians, and the Slavs, who probably had more R1a, were an elite minority in the people who moved south. This might explain why there is a reverse ratio of I to R1a in the Balkans.

Maju said...

Are there any real DE Haplogroups found in Greece?
I mean DE not being either E or D Y-DNA Haplogroups.


With all likehood not. AFAIK the only DE* has been detected in Nigeria. DE here means E3b (or however the new nomenclature).

This might explain why there is a reverse ratio of I to R1a in the Balkans.

If you look at the raw data, there is not any precise reverse. There are places like Thessaloniki or Rathymnon that have both haplos rather high, and many others (Chios, Patrai, Joanina, Lasithi or Crete-whole) where both are similarly low.

I would rather say that they are not correlated at all, that the statistical significance is not there (though haven't made the maths).

I and R1b seem the less geographically structured among the five main Greek haplos (R1b, R1a, E3b, I and J2a1), not sure why. It may be caused by older age locally (both haplos are generally considered oldest in Europe) or may be a product of irregular immigration from nearby areas rich in both haplogroups.

Maju said...

Btw, I made up a "cakes" map with Dieneke's data (I think much better visually than algorithmically). You may want to check it: http://img519.imageshack.us/img519/7313/greeceydna2xi7.jpg

Dean said...

I think that in every major study, there is more I than R1a in Greece, even if barely more. This is probably a reflection of an older pool of haplogroup I southeastern Europe that was added to by medieval I that came with Slav speakers, who never displaced the ancestors of southern Slavs and Greeks.

But like Dienekes wrote, not much can be definitively stated without understanding haplogroups' internal structures.

Maju said...

I don't see any particular reason to consider Balcanic I "Slavic". There is a lot of I around Europe and only a fracion has Slavic correlations. Even the more "slavic" subclade of I (I-P37.2) is shared with peoples that have no Slavic antecedents whatsoever, like Sardinians and other Western Mediterraneans.

Overall I and its subclades are considered quite old in Europe, not just much older than the genesis of Slavs but older than that of Indo-Europeans and Neolithic. Surely it was affected (carried from here to there) by these demographic and cultural epysodes but I see no reason to see any particularly "Slavic" correlation.

In fact Western Mediterranean I looks likely a Neolithic arrival. Why? Because Cardium Porttery culture ("Mediterranean Neolithic") has its urheimat at the coastal Western Balcans (Dalmatia, Bosnia, Albania) though admittedly also seems to have connections in Greece and even further East in Lebanon).

Being so old and spread around, your search for an Slavic connection is futile.

cacio said...

Maju:

nice maps. Would the white be "others"? (that would include also my own haplogroup, L).

I agree with you that I seems to be ancient and presumably original to the Balkans, unlike R1a. There's a lot of I1b in the Ukraine area, but there is more in the Balkans. I wonder if a better analysis of the internal structure would say something of whether it moved from the Balkans to the Ukraine.

As for the J2's, I think I remember that the King paper almost argued that J2b went through Anatolia and Greece and north from there, while the other J2's (which are dominant in Crete, for instance), almost bypassed Greece and went straight from Anatolia to S. Italy. I would be interesting to evaluate this - a direct Anatolia-Italy route vs through Greece. S Italy has a lot of E3b, though little I1b (or R1a)

Maju said...

Maju:

nice maps. Would the white be "others"? (that would include also my own haplogroup, L).


Thanks. It doesn't add up to 100% in most cases, so the rest (white) would be others, I guess.

L is a rare haplo in Europe though it has been reported in Turkey and Iran, always at low ammounts.

I agree with you that I seems to be ancient and presumably original to the Balkans, unlike R1a. There's a lot of I1b in the Ukraine area, but there is more in the Balkans. I wonder if a better analysis of the internal structure would say something of whether it moved from the Balkans to the Ukraine.

It's an interesting issue but it's in any case one that relates to the Paleolithic probably. My understanding is that I has been in Europe for very long, even competing with R1b for the "honor" of being most ancient (surviving) Y-DNA clade of the continent.

The association of I, at least of I2a, with Neolithic difussion strongly suggests that Balcanic I was there when the Cardium Pottery migrations took place and is not somthing that can be tagged as "Slavic" but rather as SE/southern European. Probably the Illyrian pirates were largely I2a and so the many older peoples without known name that were there before.

As for the J2's, I think I remember that the King paper almost argued that J2b went through Anatolia and Greece and north from there, while the other J2's (which are dominant in Crete, for instance), almost bypassed Greece and went straight from Anatolia to S. Italy. I would be interesting to evaluate this - a direct Anatolia-Italy route vs through Greece. S Italy has a lot of E3b, though little I1b (or R1a)

That's an interesting comment. As far as I can think from the archaeological perspective, Italy recieved several inputs from the Aegean and the Adriatic, begining with Cardium Pottery culture and ending with Etruscans and Greek colonization. But I haven't paid enough attention to the different subclades of J as to make a judgement.

Still from Dienekes' data, J2(xJ2a1) is relatively rare in Greece, being most important in Epirus. This region was not very active AFAIK regarding colonization.

So, even if J2(xJ2a1) is all J2b, it also seems to have "bypassed" Greece largely (except the NW).