Unrepresentative GEDMatch samples from the "East Mediterranean" continuum?

[Sikeliot]

In my view, absolutely. This has been touched on in other threads but it needs to be pointed out, in my hopes that the samples can be better updated.

This is a Eurogenes K15 plot that has been compiled that captures the whole continuum and I see several reasons to believe the samples on that calculator, like many others, need to be broken down, diversified, and improved:

http://i65.tinypic.com/wwnkog.png

Specific examples --

1. Grouping Greek Cypriots and Turkish Cypriots into one average hides Cypriot diversity:
GEDmatch calculators do not separate Greek Cypriots from Turkish Cypriots, but as we see on this plot above, the Greek and Turkish Cypriots are very different. Greek Cypriots shift toward Dodecanese and Calabrians, while Turkish Cypriots are, oddly enough, the primary force shifting the overall Cypriot cluster and sample on the map toward the Levant. Therefore, the Cypriot average, which is a composite, neither accurately captures the average Greek nor Turkish Cypriot. This also contributes to the belief many have that Greek Cypriots are simply Hellenized Levantines, when really we would clearly see if given their own sample that Greek Cypriots are Levant-shifted but still genetically Southeastern European.

My suggestion: separate Greek and Turkish Cypriots into two different averages and the difference will be clear.


2. Using Trapani and Syracuse to represent all of Sicily is inaccurate:
Eurogenes' Sicilian samples consist of two regions on the island: Trapani ("West Sicily") and Syracuse ("East Sicily"), and the Eurogenes K15 plot has a "Sicilian" average that is a composite of the two. Trapani especially has a higher degree of Western European and North African ancestry than the island's average, which provides a very skewed average, and the people of Syracuse are somewhat closer to Apulians with higher Balto-Slavic input than the island's average. This creates a sample that is not representative of the entire island, which was confirmed in the new Cretan study when we see that Trapanese are the only region of the island which doesn't plot with Crete. Syracusans were not sampled.

The above plot has two Sicilian groups: Palermo (green) and Messina/Catania (crimson) as well as Calabrians (yellow). On average, all of these plot southeast of the "Sicilian" (Trapani/Syracuse) average, and it is worth noting that when combined, Messina/Catania and Palermo collectively make up 50% of the island's population. The average does not represent this, especially not when the average person from geographic western Sicily is not scoring the Trapanese average as its top match or even in the top 5 on the calculator.

My suggestion: use Palermo and Catania as West/East Sicily instead.


3. "Central Greece" is a misnomer:
People walk away from GEDmatch with the impression that the "Central Greece" cluster on multiple calculators represents people from central mainland Greece, but it does not. In fact, other than Maniots and Tsakonians, no mainland Greeks even are very close to this population. "Central Greece" is a sample from the Sporades, a small island group in the North Aegean who are closely related to North Aegean islanders (Lesbos, Samos, Chios, Ikaria, etc.) and people from the Cyclades.

Labeling this cluster as Central Greek gives people the impression that anyone scoring this population as a top match is close to people from geographic central Greece, but it is not so. People from actual Central Greece are no different to other mainlanders on average.

My solution: rename this category "North Aegean islanders."


4. MDLP K23 has a "Greek Islander" category that does not match actual Greek islanders .
I do not know where this sample originated but I have yet to see any Greek islander from any island chain actually score it first. It seems, instead, that "Cretan" captures most islanders, along with the South Italian/East Sicilian categories, as a top match. If this sample does not seem to come close to anyone from the region, should it even be included? At the least, it should be specified which island it comes from.

[Greekscholar]

"Central Greece" is the administrative region Athens is a part of which is very confusing. Do we know for sure that sample is from the Sporades? It would make sense based on where individuals plot, but a citation would make me more confident. I am not sure it is a proxy for the NE Aegean islands or not. I will leave that question to those who have seen the samples.

I have to laugh at #4, I have never seen a Greek come close to the "Greek Islander" category either. Crete is a much better reference sample for most islanders. Who knows what that is supposed to represent. My guess was always "some other island" and I just left it at that.

Based on this PCA, I am not even sure "Turkish Cypriot" as a genetic label has much meaning. Is it possible it is a ethnicity that includes lots of different people from the former Ottoman lands who ended up in Cyprus?

[Sikeliot]

"Central Greece" is the administrative region Athens is a part of which is very confusing. Do we know for sure that sample is from the Sporades? It would make sense based on where individuals plot, but a citation would make me more confident. I am not sure it is a proxy for the NE Aegean islands or not. I will leave that question to those who have seen the samples.

I remember being told by multiple people it is a Sporades sample, which makes sense. I read it somewhere but can no longer find it, but what is clear just by viewing the averages and seeing how it plots, is it's not an indigenous mainland Greek sample. I do think it is a good representation of many North Aegean and Cyclades islanders which you see on the plot, as well as a subset of South Italians and Cretans even who shift more toward mainland Greece (Apulians, Syracusans, some Trapanese). So yes, I think it should be relabeled.


I have to laugh at #4, I have never seen a Greek come close to the "Greek Islander" category either. Crete is a much better reference sample for most islanders. Who knows what that is supposed to represent. My guess was always "some other island" and I just left it at that.

The person I have seen come closest to this was a Sicilian from Messina, a Cypriot, and one Dodecanese person but other than that, it is not a representative sample. There is also "Central Sicily" which is not representative either of any people I run through from there... in fact, the person I have seen coming closest to "Central Sicily" is from Ikaria!!


Based on this PCA, I am not even sure "Turkish Cypriot" as a genetic label has much meaning. Is it possible it is a ethnicity that includes lots of different people from the former Ottoman lands who ended up in Cyprus?

This could be true. But either way, they should not be included in the same sample as Greek Cypriots.

[essexboy]

Based on this PCA, I am not even sure "Turkish Cypriot" as a genetic label has much meaning. Is it possible it is a ethnicity that includes lots of different people from the former Ottoman lands who ended up in Cyprus?


This could be true. But either way, they should not be included in the same sample as Greek Cypriots.

It is a tough one as many of the Cypriots have such diverse mixtures that it is difficult to group them all separately but having it broken down more would be a great addition.

[Andrewid]

In my view, absolutely. This has been touched on in other threads but it needs to be pointed out, in my hopes that the samples can be better updated.

This is a Eurogenes K15 plot that has been compiled that captures the whole continuum and I see several reasons to believe the samples on that calculator, like many others, need to be broken down, diversified, and improved:

http://i65.tinypic.com/wwnkog.png

Specific examples --

1. Grouping Greek Cypriots and Turkish Cypriots into one average hides Cypriot diversity:
GEDmatch calculators do not separate Greek Cypriots from Turkish Cypriots, but as we see on this plot above, the Greek and Turkish Cypriots are very different. Greek Cypriots shift toward Dodecanese and Calabrians, while Turkish Cypriots are, oddly enough, the primary force shifting the overall Cypriot cluster and sample on the map toward the Levant. Therefore, the Cypriot average, which is a composite, neither accurately captures the average Greek nor Turkish Cypriot. This also contributes to the belief many have that Greek Cypriots are simply Hellenized Levantines, when really we would clearly see if given their own sample that Greek Cypriots are Levant-shifted but still genetically Southeastern European.

My suggestion: separate Greek and Turkish Cypriots into two different averages and the difference will be clear.



I agree with you that this should be raised.

However, I consider that we must wait for further representative samples of each Cypriot community before making firm conclusions. Y-chromosomal evidence so far,as presented by the Heraclides et al 2017 study, merely shows minor differences between Greek and Turkish Cypriots. Turkish Cypriots show a small percentage of Turkic and North African male lineages absent in Greek Cypriots. This, of course, does not give us the autosomal position. The K15 PCA does indeed show an autosomal difference between the two communities, with the Turkish Cypriots being more Levantine shifted. But how representative are these Turkish Cypriot individuals within the wider community? How many of these individuals are taken from the same location or even family, which may skew the results? We simply don't have enough reliable data to make any definitive conclusions.

If on average Turkish Cypriots do shift further to the Levant than Greek Cypriots, this may be because many Maronites and non-Chalcedonian Christians resident in Cyprus converted to Islam after the 1571 Ottoman conquest. It would make more sense for them to do this as a survival strategy than to convert to Greek Orthodoxy, which had dhimmi status. Some settlers were undoubtedly brought from Anatolia after 1571 but these included Orthodox Christians as well as Muslims. I have even seen Ottoman documents attempting to bring in Jews after the conquest. Some Turkish Cypriot villages trace their descent directly from specific areas of Anatolia. It will be interesting to see how far they have remained endogamous and have a different autosomal profile- though this is much less likely today after decades of greater demographic mobility.

Geneticists in Cyprus are currently also considering the ancestry of Maronite, Armenian and Latin Cypriots. It will be fascinating to see how far Maronites, for example, differ from their Greek and Turkish co-islanders.

[Sikeliot]

I agree with you that this should be raised.

However, I consider that we must wait for further representative samples of each Cypriot community before making firm conclusions. Y-chromosomal evidence so far,as presented by the Heraclides et al 2017 study, merely shows minor differences between Greek and Turkish Cypriots. Turkish Cypriots show a small percentage of Turkic and North African male lineages absent in Greek Cypriots. This, of course, does not give us the autosomal position. The K15 PCA does indeed show an autosomal difference between the two communities, with the Turkish Cypriots being more Levantine shifted. But how representative are these Turkish Cypriot individuals within the wider community? How many of these individuals are taken from the same location or even family, which may skew the results? We simply don't have enough reliable data to make any definitive conclusions.

If on average Turkish Cypriots do shift further to the Levant than Greek Cypriots, this may be because many Maronites and non-Chalcedonian Christians resident in Cyprus converted to Islam after the 1571 Ottoman conquest. It would make more sense for them to do this as a survival strategy than to convert to Greek Orthodoxy, which had dhimmi status. Some settlers were undoubtedly brought from Anatolia after 1571 but these included Orthodox Christians as well as Muslims. I have even seen Ottoman documents attempting to bring in Jews after the conquest. Some Turkish Cypriot villages trace their descent directly from specific areas of Anatolia. It will be interesting to see how far they have remained endogamous and have a different autosomal profile- though this is much less likely today after decades of greater demographic mobility.

Geneticists in Cyprus are currently also considering the ancestry of Maronite, Armenian and Latin Cypriots. It will be fascinating to see how far Maronites, for example, differ from their Greek and Turkish co-islanders.

I am updating the Eurogenes K15 plot with more Sicilians, Greek islanders, and mainland Greeks and if you want to add your Cypriots back on you can do so. It would be interesting to see, with newer samples, how everyone compares.

I am having the plot created to address, mainly, the fact that Trapani and Syracuse are only a subset of Sicilian variation and should not be the only two samples, but Cypriots being added can address the point made about Cyprus also.

But I do definitely think there must be some difference autosomally between Turkish and Greek Cypriots, and it could relate to the maternal side, if y-dna doesn't show much differentiation.

[Greekscholar]

I agree with you that this should be raised.

However, I consider that we must wait for further representative samples of each Cypriot community before making firm conclusions. Y-chromosomal evidence so far,as presented by the Heraclides et al 2017 study, merely shows minor differences between Greek and Turkish Cypriots. Turkish Cypriots show a small percentage of Turkic and North African male lineages absent in Greek Cypriots. This, of course, does not give us the autosomal position. The K15 PCA does indeed show an autosomal difference between the two communities, with the Turkish Cypriots being more Levantine shifted. But how representative are these Turkish Cypriot individuals within the wider community? How many of these individuals are taken from the same location or even family, which may skew the results? We simply don't have enough reliable data to make any definitive conclusions.

If on average Turkish Cypriots do shift further to the Levant than Greek Cypriots, this may be because many Maronites and non-Chalcedonian Christians resident in Cyprus converted to Islam after the 1571 Ottoman conquest. It would make more sense for them to do this as a survival strategy than to convert to Greek Orthodoxy, which had dhimmi status. Some settlers were undoubtedly brought from Anatolia after 1571 but these included Orthodox Christians as well as Muslims. I have even seen Ottoman documents attempting to bring in Jews after the conquest. Some Turkish Cypriot villages trace their descent directly from specific areas of Anatolia. It will be interesting to see how far they have remained endogamous and have a different autosomal profile- though this is much less likely today after decades of greater demographic mobility.

Geneticists in Cyprus are currently also considering the ancestry of Maronite, Armenian and Latin Cypriots. It will be fascinating to see how far Maronites, for example, differ from their Greek and Turkish co-islanders.

I like the hypothesis that endogamy/founder effect at the village level is a reason for the wide range in the Turkish Cypriot samples. This PCA does make it seem like some of the individuals sampled are very close to non-Turkish groups from the Levant and deeper Middle East. Of course, if anyone has their GEDmatch kit numbers, you could run a "Are your parents related" to find out if there is any evidence to support this theory.