PDA

View Full Version : My first models for various South and Northwest Europeans. Any thoughts?



Sikeliot
09-12-2019, 11:44 PM
SOUTHEASTERN EUROPE:

I modeled various Italians, Balkan groups, and Greeks as a mixture of:

GRC_Mycenaean -- proxy for ancient Hellenes
Levant_ISR_Ashkelon_IA2 -- proxy for ancient Phoenicians/Canaanites
Bell Beaker_ITA -- proxy for ancient peoples of Italy
Berber_Tunisia_Chen -- proxy for pre-Arab North Africans
Saudi -- proxy for original Arabian tribes
Assyrian -- proxy for ancient Mesopotamian input
CZE_Early_Slav -- proxy for early Slavic tribes
Welsh -- proxy for Celtic peoples of NW Europe
Norwegian -- proxy for Vikings and early Norman ancestors


to try to proxy for all types of admixture present into these populations (Phoenician, Berber, Arab, Italic, Hellenic, Slavic, etc.), and I got some good fits.

Any populations I should try that I missed?

I have provided each model and some commentary.

My overall observations:

1) Mainland southern Italy, from Abruzzo south to Calabria and everything in between, has Mycenaean as their predominant ancestry, and is more Mycenaean-like than Sicily.

2) Sicilians do not register Mycenaean as their top match, instead East Sicily gets Assyrian, and Trapani gets Italian Bell Beaker followed by a very close second in Assyrian.

3) Trapani has the most diversity of MENA input, and is the only region with notable combined North African and Arabian, and is hardly registering Greek at all. On the other hand, mainland south Italy has almost no additional MENA.

4) Mainland Greece and Albania are a plurality Slavic, and are more Slavic than Mycenaean.

6) All Balkan Slavs are predominantly Slavic, though on a gradient.

7) There is evidence to me of a strong West Asian migration to all of Southeastern Europe, highest in South Italy and Crete, that postdates Mycenaean input, but it is Mesopotamian-like ancestry that is responsible for the West Asian shift of these populations, not Levantine.

SICILIAN WEST (TRAPANI):

Trapani is modeled as primarily a mixture of Italy Bell Beaker and Assyrian. Using Tunisian Berbers and Saudis, about 12% seems distinctively traceable to the Arab conquest. Mycenaean is reduced to a very low 3.6%, the same amount that is registering as Saudi Arabian. Levantine does not register at all, and anything that can be decisively Norman is captured by only 3% Norwegian... instead, early Slavic is preferred for northern ancestry not captured by Bell Beaker Italians.

https://i.imgur.com/fTCf80C.png


SICILIAN EAST (SYRACUSE/RAGUSA):

I call this Syracuse/Ragusa, because I may have located the origins of the Sicilian East sample in a 2015 study and it lists the location as Ragusa... I have, however, heard others report it as Syracusan.

Sicily East also has no Levantine showing up and instead prefers Assyrian. The Mycenaean element is now coming in more strongly at 23%, and the Italian Bell Beaker is lower as is the Berber. The Slavic is lower, but instead of Norwegian we now have some Welsh showing up, about 6.8%.

https://i.imgur.com/aLd9EmI.png

ITALIAN_APULIA:

Apulia has a much higher Mycenaean element than Sicily, at 31%, followed by a sizable Assyrian-like element at 28%. Other MENA inputs are almost nonexistent, which is not surprising as Apulia is the least Near Eastern part of southern Italy. There is also, like East Sicily, a small amount of Celtic-like ancestry proxied by Welsh.

https://i.imgur.com/knST8X6.png

ITALIAN_BASILICATA:

Levant_Ashkelon now starts to appear, and Assyrian decreases which might mean there is overlap between the two components. Italian Bell Beaker is low, and Mycenaean is higher than in Apulia. A similar Welsh component still continues to emerge, alongside Early Slavic.

https://i.imgur.com/Idem3f4.png


ITALIAN_CALABRIA:

Calabria comes out similar to Basilicata, only with higher Mycenaean, lower Northern elements, and some Berber/Arabian like Sicily.

https://i.imgur.com/eU4YuVl.png

ITALIAN_CAMPANIA:

Campania is similar to Calabria only with more Celtic/Slavic type admixture.

https://i.imgur.com/KHxJW46.png

ITALIAN_ABRUZZO:

Similar to Campania, but with more Welsh.

https://i.imgur.com/V0GuxXT.png

ITALIAN_MOLISE:

Similar to Abruzzo, but with Norwegian instead of Welsh.

https://i.imgur.com/EIsjnBF.png

ITALIAN_LAZIO:

Lazio marks the switch from Mycenaean to Italian Bell Beaker as predominant.

https://i.imgur.com/rGMoiYH.png

GREEK (MAINLAND):

Unlike what some others estimated I do not get that mainlanders have more Mycenaean than Cretans, in fact the levels are close... but the predominant element is Slavic.

https://i.imgur.com/9N1YWKq.png

GREEK_CRETE:

Crete has higher Slavic, lower Celtic/Germanic, and no North African/Arabian (both replaced by Levant). Mycenaean and Assyrian levels are close to South Italy.

https://i.imgur.com/Nm7WhHh.png

ALBANIAN:

Similar to mainland Greeks.

https://i.imgur.com/uyiMz09.png

BULGARIAN:

Slavic predominates, followed by Assyrian.

https://i.imgur.com/glSaWbI.png

MACEDONIAN:

More Mycenaean than Bulgarians, but not much difference.

https://i.imgur.com/bnNkn82.png

Romanian:

Similar to the above.

https://i.imgur.com/bceOQw2.png

SERBIAN:

Looks like Balkan Slavs are mostly Slavic... since even Greeks and Albanians are in the 30-40% range.

https://i.imgur.com/RApT7mt.png

BOSNIAN:

https://i.imgur.com/CkILO3M.png

CROATIAN:

Mycenaean now drops out of the estimate.

https://i.imgur.com/grnoKuT.png

Sikeliot
09-12-2019, 11:46 PM
NORTHWESTERN EUROPE:

IRELAND:

I modeled Ireland as:

Welsh -- Celtic Briton proxy
Norwegian -- Viking proxy
England_Saxon -- Germanic conquerors of England
English -- modern English population
English_Cornwall -- see above
Basque_Spanish -- testing to see if Irish have Basque-like ancestry as rumored
German -- other Germanic input not captured by any of the above
French_Brittany -- NW French signal

And I got that the Irish are predominantly matching the English. I do get an 11% Norwegian signal, but clearly the main difference to England is more Welsh/Celtic, less Germanic.

https://i.imgur.com/6l0eBT4.png

ENGLAND:

I modeled English as a mixture of:

Welsh -- Celtic Briton proxy
Norwegian -- Viking proxy
England_Saxon -- Germanic conquerors of England
Irish -- other Celtic ancestry not captured by the above
German -- other Germanic input not captured by any of the above
French_Brittany -- NW French signal
Basque_Spanish -- testing to see if English have Basque-like ancestry as rumored

And with a very good fit, Ireland is a better proxy than Wales, and the Anglo-Saxon input is secondary. This could be in part due to English input into Ireland, but overall it shows that the English are much closer to the Irish than to Germans, Anglo-Saxons, or Scandinavians. But unlike the Irish we DO get a Basque signal!!!

https://i.imgur.com/kWQIaEG.png

SCOTLAND:

Scottish are basically a less German, more Norse version of the English, and slightly more Irish-like too. But the difference to England is small.

https://i.imgur.com/rYbjlxU.png

Sikeliot
09-12-2019, 11:56 PM
WESTERN JEWS:

Similar populations to those used for Balkans and Italy.

Good fit, but again Assyrian comes up instead of Levantine, even the Ashkelon sample. Also, Mycenaean definitely IS NOT the predominant European source, it is Italian Bell Beaker.

https://i.imgur.com/QPVnubd.png

Sephardim:

https://i.imgur.com/0zEopBF.png

Romaniotes:

Romaniotes do, actually, get Mycenaean here.

https://i.imgur.com/cV3az9j.png

Italkim:

https://i.imgur.com/dJPL9BL.png

Syrian Jews:

https://i.imgur.com/EzHcZyb.png

Tunisian Jews:

https://i.imgur.com/0Te8XEL.png

Libyan Jews:

https://i.imgur.com/bPR1OKs.png

Moroccan Jews:

https://i.imgur.com/h7RbqQw.png

Sikeliot
09-13-2019, 12:37 AM
When I use "Armenian" instead of "Assyrian" the SE Europeans favor Armenian over Levantine, and there is simply a replacement of the two components but not much difference in the percent. When I remove both Assyrian and Armenian, the Mycenaean and Levantine both increase for all of the groups and Italian Bell Beaker drops off sharply for South Italians.

The fit decreases noticeably when I remove Assyrian/Armenian and let some of that fall into Mycenaean. So I suspect the better fit is using Assyrian/Armenian to represent post-Neolithic West Asian input, and that Mycenaean input is actually lower than people assumed when they tried modeling Southeast Europeans without Assyrian/Armenian.

With Assyrian/Armenian included, South Italians all still get high Mycenaean except the Sicilians and Cretans and the Sicilians/Cretans have the strongest preference for actually including Assyrian/Armenian rather than excluding them.

xripkan
09-13-2019, 01:18 AM
I think that Ancient Greeks of the Iron Age and later Antiquity were a bit more Indo-european shifted than Myceneans(Doric invasion). If we took them as a sample I think the Slavic element would be somewhat lower. In Western Sicily for example the Slavic-like 12% indicates just some IE ancestry.

Sikeliot
09-13-2019, 01:47 AM
I think that Ancient Greeks of the Iron Age and later Antiquity were a bit more Indo-european shifted than Myceneans(Doric invasion). If we took them as a sample I think the Slavic element would be somewhat lower. In Western Sicily for example the Slavic-like 12% indicates just some IE ancestry.

Agree. But keep in mind the Iberia Empuries Greek sample is still very close to Mycenaeans, so the difference might not have been substantial.

One thing I gather from this is that in southern Italy and Crete, any Arab-era admixture is quite low to nonexistent, and there is only one exception: Trapani/"West Sicily."

Trapani definitely seems to have a different ethnogenesis than the rest of southern Italy. They have the highest "Assyrian" of the South Italian samples, but their Mycenaean is only around 3%. So this would suggest, to me, that the "Assyrian" (really post-Neolithic West Asian) did not arrive with the Mycenaeans in southern Italy or Crete, and is due to other migrations. They are also the only region with significant Berber and Arabian, which is likely Arab-era admixture, together around 13%. I suspect a lot of their Italian Bell Beaker, is due to the repopulation of the region by mainland Italians after the Norman era, but actual Norman admixture is low (probably in that 3% Norwegian). So Trapani comes out as a mixture of people from much further north in Italy, and people from West Asia, North Africa, and to a very small extent, Arabia.

The rest of South Italy as well as the Cretans are much more Mycenaean, and the Abruzzo, Molise, Campania, Apulia, Calabria, and Basilicata models all look really similar, nearly identical. As you move north in southern Italy you also end up with some Celtic-like admixture best proxied by the Welsh.

Sikeliot
09-13-2019, 01:53 AM
I think that Ancient Greeks of the Iron Age and later Antiquity were a bit more Indo-european shifted than Myceneans(Doric invasion). If we took them as a sample I think the Slavic element would be somewhat lower. In Western Sicily for example the Slavic-like 12% indicates just some IE ancestry.

Here is Malta and Cyprus. I expected Malta to be more like Sicily West, but it's different.. much more Mycenaean, much less Italian Bell Beaker, slightly more North African/Arabian, and slightly less Assyrian.

They also get some of the Welsh component, and less Slavic.

https://i.imgur.com/trM6BPV.png

Cyprus too prefers Assyrian as opposed to Levantine.

https://i.imgur.com/fozYmS3.png

Greekscholar
09-13-2019, 02:06 AM
Agree. But keep in mind the Iberia Empuries Greek sample is still very close to Mycenaeans, so the difference might not have been substantial.

One thing I gather from this is that in southern Italy and Crete, any Arab-era admixture is quite low to nonexistent, and there is only one exception: Trapani/"West Sicily."

Trapani definitely seems to have a different ethnogenesis than the rest of southern Italy. They have the highest "Assyrian" of the South Italian samples, but their Mycenaean is only around 3%. So this would suggest, to me, that the "Assyrian" (really post-Neolithic West Asian) did not arrive with the Mycenaeans in southern Italy or Crete, and is due to other migrations. They are also the only region with significant Berber and Arabian, which is likely Arab-era admixture, together around 13%. I suspect a lot of their Italian Bell Beaker, is due to the repopulation of the region by mainland Italians after the Norman era, but actual Norman admixture is low (probably in that 3% Norwegian). So Trapani comes out as a mixture of people from much further north in Italy, and people from West Asia, North Africa, and to a very small extent, Arabia.

The rest of South Italy as well as the Cretans are much more Mycenaean, and the Abruzzo, Molise, Campania, Apulia, Calabria, and Basilicata models all look really similar, nearly identical. As you move north in southern Italy you also end up with some Celtic-like admixture best proxied by the Welsh.

Yes, that is what Agamemnon's PCA indicated as well, which is probably a sign you are on the right track, at least in consistency between runs. I don't know nearly enough about this to say if it the correct way to model, or not, but I appreciate the work. Hopefully you will get some other posters to offer advice or suggestions. For instance, I am interested in why your model and Michalis' are so different.

Sikeliot
09-13-2019, 02:11 AM
Yes, that is what Agamemnon's PCA indicated as well, which is probably a sign you are on the right track, at least in consistency between runs. I don't know nearly enough about this to say if it the correct way to model, or not, but I appreciate the work. Hopefully you will get some other posters to offer advice or suggestions. For instance, I am interested in why your model and Michalis' are so different.

I think this is pretty well established and we see that all of them have a substantially Hellenic base.

What do you make about West Sicily being so different? Do you agree that it looks like it does because of a combination of A) low Greek input and B ) North-Central Italian input superimposed on top of various layers of MENA input?

Greekscholar
09-13-2019, 03:19 AM
I think this is pretty well established and we see that all of them have a substantially Hellenic base.

What do you make about West Sicily being so different? Do you agree that it looks like it does because of a combination of A) low Greek input and B ) North-Central Italian input superimposed on top of various layers of MENA input?

Well, West Sicily was controlled by Carthage, Greeks controlled the east. Certainly you have a great deal of population movement since then, but when you look at the higher North Africa in the west, the higher Greek in the east, and high Italic overall, that would be the simplest conclusion to draw based on your model.

Sikeliot
09-13-2019, 03:34 AM
Well, West Sicily was controlled by Carthage, Greeks controlled the east. Certainly you have a great deal of population movement since then, but when you look at the higher North Africa in the west, the higher Greek in the east, and high Italic overall, that would be the simplest conclusion to draw based on your model.

I think so too. With that said I expect neighboring Palermo and Agrigento might show more Greek input than Trapani has. Trapani might be the least Hellenized province.

Do you have any commentary on the Greek mainland and Cretan samples?

Censored
09-13-2019, 03:38 AM
what is this software

Ruderico
09-13-2019, 06:44 AM
I don't know which type of coordinates you're using, but for Vahaduo I suggest scaled because you can't use a penalty there. Models with unscaled coords and pen=0 can produce odd results.

For pre-arab north Africa you have actual 9th century Guanches in the datasheet, they work well and provide accurate estimations for Iberia. You might also want to add a proper ancient Levantine population there, you might have to try different ones I'm not sure Assyrian alone makes the most sense. There are also ancient British Celts and Germanics in G25 datasheet, which should be chosen over modern proxies
Example:

[1] "distance%=0.8764"

Portuguese

Celtiberian_IA,32.8
Iberian_IA,29.4
Roman_Imperial_proxy,12.4
FIN_Levanluhta_IA_o,10.4
North_African,10
Armenia_LBA,3
Levant_Roman,2




what is this software
Vahaduo, it's a web browser program that works like nMonte, but without penalty, so you're more limited. On the upside it's much faster

Sikeliot
09-13-2019, 11:34 AM
You might also want to add a proper ancient Levantine population there, you might have to try different ones I'm not sure Assyrian alone makes the most sense. There are also ancient British Celts and Germanics in G25 datasheet, which should be chosen over modern proxies
Example:

[1] "distance%=0.8764"

Portuguese

Celtiberian_IA,32.8
Iberian_IA,29.4
Roman_Imperial_proxy,12.4
FIN_Levanluhta_IA_o,10.4
North_African,10
Armenia_LBA,3
Levant_Roman,2




Vahaduo, it's a web browser program that works like nMonte, but without penalty, so you're more limited. On the upside it's much faster

I didn't use ancient British Celts because I was adding those Northern populations to estimate Norman admixture, not anything super ancient. I used Tunisia because Sicilians would be more likely to have Tunisian, rather than Moroccan or Guanche, admixture, and Tunisians may not have ever been completely like Guanches due to their location.

But I tried it and this is what I got:

- Using Assyrians, Armenians, and Lebanese Christians, the Sicilians pick up ARMENIAN and the other two drop out.
- Removing Saudi Arabian decreases the fit, but causes Lebanese Christian to come up, and Armenian to decrease somewhat. Therefore, it makes more sense to use Armenian and Saudi, rather than replace both with Lebanese Christian, and including Saudi Arabian increases the fit.
- With both Tunisia Berber and Guanche in the run, TUNISIAN comes up and Guanche drops out. The fit decreases very slightly when I remove Tunisia Berber and try to just use Guanche.
- CZE Early Slav is a much better proxy for Northern admixture in Sicily than is Norwegian or German, suggesting we might be looking at ancient Indo-European input from early Italic settlers and/or recent Greek input from the mainland, and not Norman.

In all these cases Mycenaean is low in Sicily, but is much higher in all the mainland samples like Basilicata, Apulia, Calabria, and so on.

So essentially I read this as:
- Armenian proxies post-Neolithic West Asian admixture not present in Mycenaeans
- Italian Bell Beaker captures admixture, both ancient and modern, from the northern and central parts of Italy
- Mycenaean represents Magna Graecia Greek settlement
- Tunisian and Saudi capture Arab-era Muslim settlement
- Norwegian captures Norman and/or other Germanic input
and so on.

With my original populations AND your suggested ones:

https://i.imgur.com/riKjeCN.png

Removing my original populations and replacing with your suggested ones:

https://i.imgur.com/YJ4c6Ne.png

Ruderico
09-13-2019, 11:43 AM
That's a lot of Armenian, I'm not sure that's plausible? Is that a modern or a LBA/IA reference? (genuine question, I don't know)
Personally I always try to find a timeframe in which I restrict my samples, and only resort to populations outside that range if there's no contemporaneous reference avaliable

Sikeliot
09-13-2019, 11:48 AM
That's a lot of Armenian, I'm not sure that's plausible? Is that a modern or a LBA/IA reference?

Modern. There is something telling, though, about why mainland southern Italians pick up high Mycenaean, while West Sicilians come up better as a mix of Assyrian/Armenian and Italian Bell Beaker, with a smaller Berber/Arabian element. Sicily actually is different than the mainland, even Calabria.

Historical plausibility or not it is a good fit. And maybe that Armenian could be symbolizing Byzantine migration from Anatolia, which may have had a large impact. And Armenian + Saudi + Tunisian, is a better fit than is Lebanese + Tunisian.

But even Balkan Slavs and mainland Greeks score small amounts of this Assyrian/Armenian so maybe it's post-Mycenaean but related to the Byzantine Empire. And maybe Punic input into Sicily is smaller than we think, and we're only looking at real Arab-era admixture in West Sicily and to a lesser extent East, but not anywhere on mainland south Italy.

Who knows. These are just my models.

Here is what happens when I remove Armenian and simply replace it with Lebanese Christian... I threw Basilicata in there. The fit is still not horrible, but not nearly as good. This did bring up Mycenaean in East Sicily and cause Italian Bell Beaker to drop out, but this doesn't work for West Sicily, which still doesn't get Mycenaean.

So what I also think is we took for granted that Mycenaean was the best way to model Italy/Greece and never added in any West Asians other than Levantines. Maybe this was a mistake, because it works for mainland South Italy, but not Sicily.

https://i.imgur.com/Sb7TOkb.png

Ruderico
09-13-2019, 12:26 PM
That's interesting and makes some sense I think, just remember that sometimes certain samples work really well, but it's hard to understand why. The same thing can happen for any tested population/individual with different references, so my advise is to come up with different models (as in from different time periods) and to check whether or not both saying roughly the same thing, and, if possible, to compare the figures with those of published studies, because ultimately those are the ones that really matter, and that we can rely upon.
I guess the last part might be hard for many cases/populations, for Iberia maybe less so, as I tend to compare North African levels to those published in papers (https://media.springernature.com/lw685/springer-static/image/art%3A10.1038%2Fs41467-018-08272-w/MediaObjects/41467_2018_8272_Fig5_HTML.png?as=webp). This is just my way of doing things.


I'll give an example, if I use Nordic_IA, or one of those northern Szolad samples (SZ30), in my models as the germanic reference my share of it skyrockets to around ~15% and ~25% respectively, which is obviously very implausible as a germanic ancestry estimation, particularly in my case https://i.postimg.cc/Pqbv3nQ6/Ruderico-celt-germ.png since I generally score a big fat 0. Here's a comparison below.
[1] "distance%=1.944"

Ruderico

Celtiberian_IA,80
North_African,8.8
Roman_Imperial_proxy,6
Armenia_LBA,4.2
Yoruba,0.6
Levant_Roman,0.4


[1] "distance%=1.6729"

Ruderico

Celtiberian_IA,51.8
HUN_MA_Szolad,25.6
North_African,8.4
Iberian_IA,7
Roman_Imperial_proxy,4.6
Levant_Roman,2
Armenia_LBA,0.6


The fit even improves dramatically, my suspicion is that SZ30 might have continental Celtic (or some such) ancestry, and that is causing that particular reference to be really good for me, but not for the reason I used it for. Careful not to see similar phenomena in the models, just my 2c

Dimanto
09-13-2019, 01:13 PM
Which sample(s) did you use for Campania? Or do you think the variation within Campania is too small. If I remember correctly, Benevento seemed to be less East Med than the coastal area's and Avellino.

Sikeliot
09-13-2019, 01:53 PM
Which sample(s) did you use for Campania? Or do you think the variation within Campania is too small. If I remember correctly, Benevento seemed to be less East Med than the coastal area's and Avellino.

The Campania sample came from the new study you linked me to.

Dimanto
09-13-2019, 03:39 PM
The Campania sample came from the new study you linked me to.

I see. Why not select your own samples, or is this impossible?

Greekscholar
09-13-2019, 05:45 PM
I think so too. With that said I expect neighboring Palermo and Agrigento might show more Greek input than Trapani has. Trapani might be the least Hellenized province.

Do you have any commentary on the Greek mainland and Cretan samples?

Well, having more ancient Greek DNA appear in modern Greeks is what I would expect to see, but my expectations don't mean much on their own. My question would be how does your model differ from this one.

https://anthrogenica.com/showthread.php?15180-How-genetically-similar-are-Modern-Greeks-to-the-ancient-ones&p=595227&viewfull=1#post595227

If you do a run with the same groups Michalis used, do you get the same results he did? He referred to this as his "devil's advocate model" in another thread, and I am not sure what that means.

The potential overlap between these groups is the biggest tripping point to my understanding. Which samples, if any, really are principle components? Which are a collection of various other principle components? I think after reading many of these threads, the advice to use samples from the same time period is a good one. It doesn't ensure there is no overlap between samples, because groups could share deeper ancestry, but it probably helps.

Sikeliot
09-13-2019, 09:40 PM
Does anyone have suggestions for other input populations to use?

Impaler
09-13-2019, 10:43 PM
This is what I have.

http://i.imgur.com/UoHK2je.png (https://imgur.com/UoHK2je)

Sikeliot
09-13-2019, 11:24 PM
I was able to improve all fits by doing the following:

1. Replacing Norwegian with English as a proxy for Germanic input
2. Replacing Tunisian Berber with Tunisian
3. Replacing Lebanese Christian and Philistine samples with Samaritan

Now, we do see Levantine coming up instead of just Assyrian, so clearly we find that Samaritans better proxy Levantine input into SE Europe.

And the fits come out much better for all groups, especially West Sicily.

And sorry to burst bubbles but European input in Ashkenazim comes up much more Italian Bell Beaker and English even, rather than as Mycenaean ;) And they are much more Levantine than South Italians/Cretans/Maltese, by far.

SICILIANS:

https://i.imgur.com/ORMYN9p.png
https://i.imgur.com/rRh626w.png

SOUTH ITALIANS:

https://i.imgur.com/8UXKLIz.png
https://i.imgur.com/j83MCO2.png
https://i.imgur.com/2LMJeGw.png
https://i.imgur.com/achig2z.png


GREEKS:

https://i.imgur.com/WjbY06t.png

ASHKENAZIM:

https://i.imgur.com/hfxm7zx.png

alexfritz
09-14-2019, 02:52 AM
more italian results with permission of the Sikeliot
Vahaduo(G25 datasheet) based

1italian groups
https://i.imgur.com/WtaYdgF.png
2improved distance lombardy 1adding remedello 2new model
https://i.imgur.com/2gRczHn.png
3italian groups alps 1without hallstatt 2with hallstatt 3without remedello
https://i.imgur.com/12Zt0Up.png

133153 233154 333155

Finn
09-14-2019, 08:00 AM
I was able to improve all fits by doing the following:

1. Replacing Norwegian with English as a proxy for Germanic input
2. Replacing Tunisian Berber with Tunisian
3. Replacing Lebanese Christian and Philistine samples with Samaritan

Now, we do see Levantine coming up instead of just Assyrian, so clearly we find that Samaritans better proxy Levantine input into SE Europe.

And the fits come out much better for all groups, especially West Sicily.

And sorry to burst bubbles but European input in Ashkenazim comes up much more Italian Bell Beaker and English even, rather than as Mycenaean ;) And they are much more Levantine than South Italians/Cretans/Maltese, by far.

SICILIANS:

https://i.imgur.com/ORMYN9p.png
https://i.imgur.com/rRh626w.png

SOUTH ITALIANS:

https://i.imgur.com/8UXKLIz.png
https://i.imgur.com/j83MCO2.png
https://i.imgur.com/2LMJeGw.png
https://i.imgur.com/achig2z.png


GREEKS:

https://i.imgur.com/WjbY06t.png

ASHKENAZIM:

https://i.imgur.com/hfxm7zx.png

Nice model and overviews!

I see that your 'Germanic' category is still somewhat shaky. As a proxy for Germanic you could also try the (outmost) North Dutch, they come close to the old North Sea Germanic people. And Sweden as a proxy for a North Germanic or Scandic population.

English is fare too mixed to be a proxy for Germanic.

I'm curios how this would effect your model.

Ruderico
09-14-2019, 08:42 AM
Nice model and overviews!

I see that your 'Germanic' category is still somewhat shaky. As a proxy for Germanic you could also try the (outmost) North Dutch, they come close to the old North Sea Germanic people. And Sweden as a proxy for a North Germanic or Scandic population.

English is fare too mixed to be a proxy for Germanic.

I'm curios how this would effect your model.

The Levanluhta outlier works pretty well as a germanic reference for the IA

Finn
09-14-2019, 09:28 AM
The Levanluhta outlier works pretty well as a germanic reference for the IA

I doubt, he pops up very high in my mothers auDNA but not in my father's one. My mothers auDNA has a higher pre (or may be proto) Germanic DNA....

I guess this Levanluta outlier is more proto.....I doesn't cover the whole Germanic range.

(see: https://anthrogenica.com/showthread.php?13792-Angle-Saxon-or-Jute&p=600762&viewfull=1#post600762)

Sikeliot
09-14-2019, 11:43 AM
The Levanluhta outlier works pretty well as a germanic reference for the IA

Here we go... I tried to use that instead of English and here is what happened... the fits are still good but slightly decreased. I also tried to replace "Tunisian" with two more isolated Berber samples and see what happened.

The result is an increase in Italic Bell Beaker and Armenian for all the Italians, and the Mycenaean decreases. The "northern" element in far southern Italy and Greece is better captured by Slavic which may also proxy Indo-European input, and not the ancient Germanic sample.

Does this look better to you?


SICILIANS:

https://i.imgur.com/xgkfkbA.png
https://i.imgur.com/000qJ4M.png

GREEKS:

https://i.imgur.com/QrTyp5j.png

SOUTH ITALIANS:

https://i.imgur.com/xGcljqj.png
https://i.imgur.com/Upa5xTP.png
https://i.imgur.com/wDPC8TH.png
https://i.imgur.com/tJSQsUC.png

ASHKENAZIM:

https://i.imgur.com/L0CiuAs.png

Sikeliot
09-14-2019, 11:47 AM
I doubt, he pops up very high in my mothers auDNA but not in my father's one. My mothers auDNA has a higher pre (or may be proto) Germanic DNA....

I guess this Levanluta outlier is more proto.....I doesn't cover the whole Germanic range.

(see: https://anthrogenica.com/showthread.php?13792-Angle-Saxon-or-Jute&p=600762&viewfull=1#post600762)

See the remodels... I may try with North Dutch later.

Finn
09-14-2019, 12:27 PM
See the remodels... I may try with North Dutch later.
ok....

Yes indeed I guess that's indeed what I thought would happen, Levanluhta has lower Steppe and higher SHG/ EHG...compared to 'average' Germanic.

So the Steppe component will be 'filled' by Bell Beaker and Slavic indeed....

But how representative is this? Levanluhta is a kind of (proto-) Germanic outlier.

G25 Basics (ph2ter)

sample": "Custom:AGUser_finn_dad",
"fit": 4.1366,
"STEPPE": 49.17,
"EEF": 35.83,
"WHG": 15,

"sample": "FIN_Levanluhta_IA_o:JK2065",
"fit": 4.5059,
"STEPPE": 40.83,
"EEF": 30.83,
"WHG": 20.83,
"EHG": 7.5,

Finn
09-14-2019, 12:32 PM
Some add, you could also use the early mediaval Germans as a proxy, see how they come close to my Dad in stead of Levanlutha:

"sample": "DEU_MA_Averaged:Average",
"fit": 3.7197,
"STEPPE": 50.83,
"EEF": 35.83,
"WHG": 13.33

alexfritz
09-14-2019, 01:25 PM
in addendum to #25 another round of tinkering
Sardinia

does not improve much beyond EEF+East_Med (as in joseph et al)
except alittle when adding modern italian groups
https://i.imgur.com/CEBODbo.png

33162

alexfritz
09-15-2019, 03:12 AM
some greek groups and some italian groups and sicilians again
2 adding CZE_Early_Slav and DEU_MA to check the respective postWRE/byzantine events
https://i.imgur.com/bhx8DST.png
33195

something mycenaean(??byzantine/doubtful) going on in lombardy/bergamo and not elsewhere in italy except east_sicily and just a guess but samples in lombardy seem not to be from the bergamo area but different areas this time; clear slavic(?southslavic) occurrence in the friuli ?prob tied to the so-called 'hungarian road'
33192

EDIT whatver the avg. DEU_MA repr. it seems quite significant in mediterranean pops. thus replacing DEU_MA with England_Saxon (less ambigious) for postWRE events
https://i.imgur.com/GSAbxfb.png
33196
adding DEU_MA along with England_Saxon (least ambigious) for postWRE events
https://i.imgur.com/BzKBFZ0.png
33201

Sikeliot
09-15-2019, 05:02 AM
some greek groups and some italian groups and sicilians again
2 adding CZE_Early_Slav and DEU_MA to check the respective post-WRE/byzantine events
https://i.imgur.com/bhx8DST.png
33195


something mycenaean(??byzantine/doubtful) going on in lombardy/bergamo and not elsewhere in italy, except east_sicily; slavic(?southslavic) occurrence in the friuli ?prob tied to the so-called 'hungarian road'
33192


Here East Sicily is the region of South Italy with the highest Mycenaean. Makes sense if you see maps of Magna Graecia, as Ragusa and Syracuse were Doric. The Levantine contribution is roughly equal in both sides of Sicily and in Crete, with West Sicily having the small but persistent NW African element.

What I find interesting is South Italy and Crete better modeled by an Anatolian Greek type component rather than Mycenaean, and West Sicily again has low Mycenaean.

Could you try running Ashkenazim?

Kelmendasi
09-15-2019, 11:36 AM
You should add the Croatian Iron Age sample (Illyrian) and Bulgarian Iron Age sample (Thracian).

When using the samples shown for the Albanian result, I get:
Fit - 1.931
GRC_Mycenaean - 45.83%
CZE_Early_Slav - 34.17%
Bell_Beaker_ITA - 16.67%
Assyrian - 3.33%

Sikeliot
09-15-2019, 12:01 PM
Some add, you could also use the early mediaval Germans as a proxy, see how they come close to my Dad in stead of Levanlutha:

"sample": "DEU_MA_Averaged:Average",
"fit": 3.7197,
"STEPPE": 50.83,
"EEF": 35.83,
"WHG": 13.33



Using that ancient German sample removes most of the Early Slavic from Sicilians (which means this could represent Swabian, Norman, and more likely than anything else North Italian input into Sicily after the Norman era), but does not affect Greeks who still come up very much Slavic. South Italians have both, and the German-like ancestry increases as you move north in Italy. Fit is still very good. This is everyone now.

I also tried removing Armenians and trying to capture the West Asian in all these groups with just Georgian and Samaritan but that decreased the fit, so I put Armenians back in.

Nothing I do is getting Ashkenazim to end up Mycenaean, so that along with the finding of uniparental markets on the maternal side matching North Italians, I REJECT the notion of significant Greek input in Ashkenazim.

East Sicily/Ragusa clearly has a lot of Greek ancestry just like southern Italy though.

SICILIAN:

https://i.imgur.com/hL4glID.png
https://i.imgur.com/nRNdQeh.png

SOUTH ITALIAN:

https://i.imgur.com/cwgeWQx.png
https://i.imgur.com/ywulMoS.png
https://i.imgur.com/PwmTOvp.png
https://i.imgur.com/f0BxHo0.png

ASHKENAZIM:

https://i.imgur.com/YC6cFAF.png

GREEK:

https://i.imgur.com/GG1LCMB.png

Sikeliot
09-15-2019, 01:18 PM
Here is Ireland and Scotland with the ancient German sample among others.

Who should I be using to model Celtic input? I have been using Welsh.

https://i.imgur.com/lzRAwkV.png

xripkan
09-15-2019, 08:08 PM
Agree. But keep in mind the Iberia Empuries Greek sample is still very close to Mycenaeans, so the difference might not have been substantial.

One thing I gather from this is that in southern Italy and Crete, any Arab-era admixture is quite low to nonexistent, and there is only one exception: Trapani/"West Sicily."

Trapani definitely seems to have a different ethnogenesis than the rest of southern Italy. They have the highest "Assyrian" of the South Italian samples, but their Mycenaean is only around 3%. So this would suggest, to me, that the "Assyrian" (really post-Neolithic West Asian) did not arrive with the Mycenaeans in southern Italy or Crete, and is due to other migrations. They are also the only region with significant Berber and Arabian, which is likely Arab-era admixture, together around 13%. I suspect a lot of their Italian Bell Beaker, is due to the repopulation of the region by mainland Italians after the Norman era, but actual Norman admixture is low (probably in that 3% Norwegian). So Trapani comes out as a mixture of people from much further north in Italy, and people from West Asia, North Africa, and to a very small extent, Arabia.

The rest of South Italy as well as the Cretans are much more Mycenaean, and the Abruzzo, Molise, Campania, Apulia, Calabria, and Basilicata models all look really similar, nearly identical. As you move north in southern Italy you also end up with some Celtic-like admixture best proxied by the Welsh.

I think that Empuries sample has some Phoenician admixture which adds the East Med element and reduces the IE.

About the results I find interesting that there is a detectable Norman input in Western Sicily. Furthermore Malta seems like Eastern Sicily but with higher North African element. Did you try to change the populations on Cyprous model? I think with a change it will seem much more East Med.

Sikeliot
09-15-2019, 08:50 PM
I think that Empuries sample has some Phoenician admixture which adds the East Med element and reduces the IE.

About the results I find interesting that there is a detectable Norman input in Western Sicily. Furthermore Malta seems like Eastern Sicily but with higher North African element. Did you try to change the populations on Cyprous model? I think with a change it will seem much more East Med.

Malta there is probably a good proxy for central Sicily (Agrigento, Palermo, Caltanissetta, etc) which I wish we had a sample from. NE Sicily is probably best proxied by Calabria. I don't see Norman in western Sicily, I see evidence of Italian repopulation on top of a MENA base.

What do you make of Slavic predominating over Mycenaean in Greeks?

I didn't try the changes with Cyprus. Here, though, are a few of the samples using Minoan instead of Mycenaean...


https://i.imgur.com/jO3uv6q.png
https://i.imgur.com/xgHb8zZ.png
https://i.imgur.com/KmePHSm.png

Sikeliot
10-20-2019, 02:40 PM
Here are West Sicily, East Sicily, Crete, and Greece using ALL the populations in the G25 spreadsheet. It lead to a mixture of ancient and modern so how do I interpret some of these?

East Sicily:


Target: Sicilian_East
Distance: 0.1010% / 0.00101043
Aggregated
9.4 Anatolia_Kumtepe_N
8.2 HUN_MA_Szolad_o1
7.0 DEU_MA_o
6.2 EGY_Hellenistic
5.6 Anatolia_Barcin_C
5.0 Spanish_Asturias
4.8 HUN_Tisza_LN
4.6 RUS_Maykop_Novosvobodnaya
4.4 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
4.4 SWE_IA
4.0 Anatolia_Kaman-Kalehoyuk_MLBA
3.4 English_Cornwall
3.2 Levant_Canaanite_MBA
2.8 USA_colonial_period
2.6 IRN_Belt_Cave_Meso_low_res
2.4 Baltic_LTU_Late_Antiquity_low_res
2.4 GRC_Minoan_Odigitria_low_res
2.2 Yemenite_Mahra
2.0 BGR_Varna_En3
1.6 UKR_N_o
1.4 Anatolia_Kumtepe_N_low_res
1.4 DEU_Welzin_BA_outlier3
1.4 Levant_ISR_Ashkelon_LBA
1.4 MKD_N
1.0 RUS_Darkveti-Meshoko_En
0.6 CZE_Bilina_N_Baalberge
0.6 England_Saxon
0.6 Georgian_Imer
0.6 Iberia_Central_CA_Afr
0.6 IRN_Hajji_Firuz_BA
0.4 Icelandic
0.4 IND_Roopkund_C
0.4 Syrian_Jew
0.4 Yemenite_Al_Jawf
0.2 AFG_Darra_i_kur_MBA_low_res
0.2 Bell_Beaker_HUN_EBA_o
0.2 Berber_Algeria
0.2 Ethiopian_Afar
0.2 Greek_Trabzon
0.2 Gupta
0.2 HRV_EBA
0.2 HUN_MA_Szolad_o2
0.2 Juang
0.2 Levant_LBN_MA_Mixed
0.2 Saharawi
0.2 Spanish_Canarias
0.2 Yemenite_Ma'rib

West Sicily:

Target: Sicilian_West
Distance: 0.1013% / 0.00101322
Aggregated
11.6 Anatolia_Barcin_C
7.2 Spanish_Navarra
6.8 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
5.4 Levant_LBN_MA_NE
5.0 HUN_Tiszapolgar_ECA
4.8 Baltic_LTU_Late_Antiquity_low_res
4.8 Levant_ISR_Ashkelon_LBA
4.4 KAZ_Ak_Moustafa_MLBA1
3.6 DEU_MA_o
2.8 Italian_Liguria
2.8 Tunisian_Jew
2.6 Spanish_Castilla_Y_Leon
2.6 UKR_Globular_Amphora
2.4 IRN_Belt_Cave_Meso_low_res
2.4 RUS_Saltovo-Mayaki_low_res
2.0 DEU_Welzin_BA_outlier2
2.0 SRB_N
2.0 UKR_N_o
1.8 EGY_Hellenistic
1.8 HUN_MA_Szolad_o1
1.8 Iraqi_Jew
1.6 CZE_Bilina_N_Baalberge
1.6 Yemenite_Al_Jawf
1.4 BGR_Varna_En3
1.4 CZE_N
1.4 IRL_MN
1.4 Saharawi
1.2 Basque_French
1.0 HUN_Protoboleraz_LCA
0.8 Iberia_Central_CA_Afr
0.8 Spanish_Pais_Vasco
0.6 Anatolia_Kumtepe_N
0.6 Spanish_Alacant
0.4 Berber_Algeria
0.4 Brahmin_Uttar_Pradesh
0.4 Egyptian
0.4 Icelandic
0.4 KEN_Pastoral_N_o
0.4 Samaritan
0.4 Tunisian
0.4 Uttar_Pradesh
0.2 AFG_Darra_i_kur_MBA_low_res
0.2 Berber_MAR_ERR
0.2 Corded_Ware_CZE_o
0.2 HUN_LBK_MN
0.2 HUN_Sopot_LN
0.2 Iberia_Northeast_RomP
0.2 Moroccan_Jew
0.2 Mozabite
0.2 Roma
0.2 TKM_Gonur3_BA
0.2 TKM_Sumbar_LBA
0.2 Yemenite_Amran

Crete:


Target: Greek_Crete
Distance: 0.1447% / 0.00144749
Aggregated
16.8 Kura-Araxes_ARM_Kalavan
10.4 GRC_N
8.0 IRN_Hajji_Firuz_IA
7.6 UKR_Chernyakhiv_Shyshaky
7.2 BGR_IA
6.8 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
5.6 Levant_ISR_C
5.2 IND_Roopkund_B_o
4.4 Levant_PPNC
4.0 Cossack_Ukrainian
3.8 Levant_ISR_Ashkelon_LBA
3.2 Bell_Beaker_England_EBA
2.4 RUS_Saltovo-Mayaki_low_res
2.4 UKR_Trypillia
2.0 Levant_JOR_EBA
1.8 DEU_Welzin_BA_outlier2
1.8 Lak
1.6 SRB_N
1.2 KAZ_Golden_Horde_Euro
0.8 HUN_MA_Szolad_o1
0.6 HUN_ALPc_III_MN
0.6 Italian_Northeast_o
0.4 Levant_PPNB
0.4 USA_colonial_period
0.2 Iberia_Northeast_Empuries2
0.2 Levant_Natufian
0.2 MKD_N
0.2 ROU_N
0.2 Turkish_Northwest

Greek:

Target: Greek
Distance: 0.0292% / 0.00029231
Aggregated
12.2 Macedonian
10.4 Italian_Northeast_o
7.0 KAZ_Golden_Horde_Euro
6.6 GRC_Minoan_Lassithi
5.0 Anatolia_IA_low_res
4.4 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
4.2 GRC_N
3.8 Anatolia_Kumtepe_N
3.6 Kura-Araxes_ARM_Talin
2.8 DEU_MA_o
2.8 HRV_Sopot_MN
2.4 IRN_Hajji_Firuz_IA
2.4 Levant_LBN_Roman
1.8 HUN_LBK_MN
1.8 Lithuanian
1.8 MKD_N
1.6 DEU_Welzin_BA_outlier2
1.4 Anatolia_Kaman-Kalehoyuk_MLBA
1.4 Cossack_Ukrainian
1.2 ARM_LBA
1.2 HUN_Avar_Szolad
1.2 Lak
1.0 Armenian
1.0 BGR_Varna_En3
1.0 IND_Roopkund_B
1.0 Kaitag
1.0 UKR_N_o
0.8 Baltic_LTU_BA
0.8 Bosnian
0.8 HUN_MA_Szolad_o1
0.8 Moldavian
0.8 UKR_Srubnaya_MLBA
0.6 Corded_Ware_Proto-Unetice_POL
0.6 Iberia_Northeast_Empuries2
0.6 Levant_ISR_MLBA
0.6 Levant_LBN_MA_NE
0.6 RUS_Srubnaya_MLBA_o
0.4 AFG_Darra_i_kur_MBA_low_res
0.4 ARM_Areni_C
0.4 Baltic_EST_MA
0.4 Corded_Ware_Baltic_early
0.4 Lebanese_Druze
0.4 Levant_PPNC
0.4 ROU_N
0.4 SRB_N
0.4 Yamnaya_KAZ_Mereke
0.2 Baltic_LVA_BA
0.2 Chechen
0.2 Georgian_Jew
0.2 GRC_Peloponnese_N
0.2 Iranian_Jew
0.2 Italian_Apulia
0.2 Latvian
0.2 Lebanese_Muslim
0.2 RUS_Petrovka_MLBA
0.2 RUS_Saltovo-Mayaki_low_res
0.2 Russian_Voronez
0.2 Samaritan
0.2 Scythian_MDA
0.2 Surui
0.2 TKM_Parkhai_MBA
0.2 Turkish_Azer
0.2 USA_colonial_period

Genetique
10-20-2019, 02:49 PM
The différences are blatant, East Sicily scores tiny percentages in a lot of Middle-Eastern/Berber populations, West Sicily much less and more in europeans populations like Spanish. East Sicily has 6,2% of Egypt versus 2% for West Sicily.

Whereas Sicily has low Minoan admixture, Grece seems to have more Slavic (we already know that), Minoan and Caucasus/Anatolian related populations. Crete is like Greece but much more Caucasus

For East Sicily I have bolded the berber/arabian related populations

9.4 Anatolia_Kumtepe_N
8.2 HUN_MA_Szolad_o1
7.0 DEU_MA_o
6.2 EGY_Hellenistic
5.6 Anatolia_Barcin_C
5.0 Spanish_Asturias
4.8 HUN_Tisza_LN
4.6 RUS_Maykop_Novosvobodnaya
4.4 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
4.4 SWE_IA
4.0 Anatolia_Kaman-Kalehoyuk_MLBA
3.4 English_Cornwall
3.2 Levant_Canaanite_MBA
2.8 USA_colonial_period
2.6 IRN_Belt_Cave_Meso_low_res
2.4 Baltic_LTU_Late_Antiquity_low_res
2.4 GRC_Minoan_Odigitria_low_res
2.2 Yemenite_Mahra
2.0 BGR_Varna_En3
1.6 UKR_N_o
1.4 Anatolia_Kumtepe_N_low_res
1.4 DEU_Welzin_BA_outlier3
1.4 Levant_ISR_Ashkelon_LBA
1.4 MKD_N
1.0 RUS_Darkveti-Meshoko_En
0.6 CZE_Bilina_N_Baalberge
0.6 England_Saxon
0.6 Georgian_Imer
0.6 Iberia_Central_CA_Afr
0.6 IRN_Hajji_Firuz_BA
0.4 Icelandic
0.4 IND_Roopkund_C
0.4 Syrian_Jew
0.4 Yemenite_Al_Jawf
0.2 AFG_Darra_i_kur_MBA_low_res
0.2 Bell_Beaker_HUN_EBA_o
0.2 Berber_Algeria
0.2 Ethiopian_Afar
0.2 Greek_Trabzon
0.2 Gupta
0.2 HRV_EBA
0.2 HUN_MA_Szolad_o2
0.2 Juang
0.2 Levant_LBN_MA_Mixed
0.2 Saharawi
0.2 Spanish_Canarias
0.2 Yemenite_Ma'rib

Sikeliot
10-20-2019, 02:56 PM
The différences are blatant, East Sicily scores tiny percentages in a lot of Middle-Eastern/Berber populations, West Sicily much less and more in europeans populations like Spanish. East Sicily has 6,2% of Egypt versus 2% for West Sicily.

Looked to me like West Sicily has more Levantine though. That East Sicily sample is much more representative island-wide than is the West Sicily sample though.

What stood out to me is the mainland Greek sample not scoring really anything Afroasiatic. The difference to Sicily or Crete is not JUST the Slavic.

Genetique
10-20-2019, 03:02 PM
Looked to me like West Sicily has more Levantine though. That East Sicily sample is much more representative island-wide than is the West Sicily sample though.

What stood out to me is the mainland Greek sample not scoring really anything Afroasiatic. The difference to Sicily or Crete is not JUST the Slavic.

Yep West Sicily has more Levantine. I've just seen also suprising populations like Roma, South-Asiatic DNA traces ?

It makes sense that Mainland Greece don't scores any afroasiatic, only Sicily was ruled by Arabs and Phoenicians.

Crete appears very different… They shift to the east towards Caucasus but not towards south.

Sikeliot
10-20-2019, 03:06 PM
Yep West Sicily has more Levantine. I've just seen also suprising populations like Roma, South-Asiatic DNA traces ?

It makes sense that Mainland Greece don't scores any afroasiatic, only Sicily was ruled by Arabs and Phoenicians.

Crete appears very different… They shift to the east towards Caucasus but not towards south.

Correct. From what I remember it seems Crete is the closest to CAMPANIA out of all the Italians.

Genetique
10-20-2019, 03:14 PM
Correct. From what I remember it seems Crete is the closest to CAMPANIA out of all the Italians.

So Campania would form a cluster with Crete, East but North of Sicilian cluster.

Dimanto
10-20-2019, 03:21 PM
Yep West Sicily has more Levantine. I've just seen also suprising populations like Roma, South-Asiatic DNA traces ?

It makes sense that Mainland Greece don't scores any afroasiatic, only Sicily was ruled by Arabs and Phoenicians.

Crete appears very different… They shift to the east towards Caucasus but not towards south.

My Neapolitan uncle also scores some Levantine on 23andme and we have no Sicilian ancestry. How do you trace Levantine ancestry?

Sikeliot
10-20-2019, 03:21 PM
So Campania would form a cluster with Crete, East but North of Sicilian cluster.

IMO yes.

Here are the mainland southerners. Calabria has high Levant, higher than Sicily (and IMO this sample may also be close to Palermo, Messina, etc) but the others do not.

Calabria:



Target: Italian_Calabria
Distance: 0.0790% / 0.00078983
Aggregated
7.8 Anatolia_Ovaoren_EBA
7.8 BGR_Krepost_N
7.2 Levant_LBN_Roman
7.0 Italian_Liguria
6.0 HRV_Starcevo_LN
5.6 HRV_MBA
5.2 Levant_LBN_MA_NE
4.8 SRB_N
4.2 Iraqi_Jew
4.0 Anatolia_Isparta_EBA
3.0 Georgian_Laz
2.8 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
2.8 Corded_Ware_CZE_o
2.6 Shetlandic
2.2 Sorb_Niederlausitz
1.8 AFG_Darra_i_kur_MBA_low_res
1.8 England_Roman_Near_Eastern_o
1.6 KEN_Pastoral_N_o
1.6 USA_colonial_period
1.4 Iberia_Northeast_Empuries2
1.4 ITA_Collegno_MA_o2
1.4 Swiss_French
1.2 Anatolia_Barcin_C
1.0 IRN_Hajji_Firuz_IA
1.0 Nordic_LN_low_res
0.8 Italian_South
0.6 Anatolia_Kumtepe_N_low_res
0.6 GRC_Mycenaean
0.6 Italian_Northeast_o
0.6 RUS_Darkveti-Meshoko_En
0.6 TKM_IA
0.6 TKM_Sumbar_LBA
0.4 Armenian
0.4 Baltic_LTU_Late_Antiquity_low_res
0.4 BGR_IA
0.4 DEU_MA_o
0.4 GRC_N
0.4 HUN_LBA
0.4 Italian_Bergamo
0.4 Spanish_Murcia
0.4 TKM_Gonur3_BA
0.4 UZB_Sappali_Tepe_BA
0.4 Yamnaya_KAZ_Karagash
0.2 Algerian
0.2 Anatolia_Pinarbasi_HG
0.2 Berber_MAR_TIZ
0.2 Chechen
0.2 Dai
0.2 GRC_Minoan_Lassithi
0.2 GRC_Minoan_Odigitria_low_res
0.2 IND_Roopkund_B
0.2 Iranian_Zoroastrian
0.2 Irish
0.2 ITA_Collegno_MA_o1
0.2 Italian_Northeast
0.2 Italian_Umbria
0.2 Kura-Araxes_ARM_Kalavan
0.2 MAR_Iberomaurusian
0.2 Saharawi
0.2 Somali
0.2 Spanish_Soria
0.2 TKM_Parkhai_MBA
0.2 UKR_Chernyakhiv_Shyshaky


Campania:



Target: Italian_Campania
Distance: 0.0126% / 0.00012598
Aggregated
5.0 Georgian_Jew
4.0 Anatolia_Isparta_EBA
4.0 SRB_N
3.8 HUN_MA_Szolad_o1
3.6 DEU_MA_o
3.4 Iraqi_Jew
3.4 Moroccan_Jew
2.8 Levant_LBN_Roman
2.0 Italian_South
2.0 Italian_Umbria
2.0 Shetlandic
2.0 TZA_Zanzibar_Euro_outlier
1.8 England_Saxon
1.8 GRC_Peloponnese_N
1.8 Iberia_Northeast_Empuries2
1.8 Levant_LBN_MA_NE
1.8 SVK_Poprad_MA
1.6 BGR_Krepost_N
1.6 Georgian_Imer
1.6 Levant_Canaanite_MBA
1.4 HUN_Protoboleraz_LCA
1.4 Italian_Liguria
1.4 Levant_PPNC
1.2 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
1.2 GRC_Minoan_Lassithi
1.2 IND_Roopkund_B
1.2 Irish
1.2 Kura-Araxes_ARM_Kalavan
1.2 TKM_IA
1.0 BGR_C
1.0 GRC_N
1.0 HRV_Starcevo_LN
1.0 HUN_Sopot_LN
1.0 ITA_Collegno_MA_o1
1.0 Levant_JOR_EBA
1.0 UKR_N_o
0.8 Berber_Tunisia_Chen
0.8 English_Cornwall
0.8 GRC_Minoan_Odigitria_low_res
0.8 IND_Roopkund_B_o
0.8 Levant_ISR_Ashkelon_LBA
0.8 Romaniote_Jew
0.8 Spanish_Navarra
0.8 TKM_Gonur3_BA
0.8 UKR_Sredny_Stog_II_En
0.6 Bosnian
0.6 Cypriot
0.6 HRV_MBA
0.6 Iberia_Northeast_c.6-8CE_ES
0.6 Icelandic
0.6 Italian_Abruzzo
0.6 Italian_Aosta_Valley
0.6 Romanian
0.6 Tabasaran
0.6 Yemenite_Mahra
0.4 AFG_Darra_i_kur_MBA_low_res
0.4 Armenian
0.4 ARM_MBA
0.4 Corded_Ware_Baltic_early
0.4 CZE_MN
0.4 Dutch
0.4 Greek_Trabzon
0.4 Iberia_Northeast_RomP
0.4 IRN_Hajji_Firuz_IA
0.4 Italian_Marche
0.4 Italian_Northeast
0.4 KAZ_Golden_Horde_Euro
0.4 RUS_Maykop_Novosvobodnaya
0.4 Sorb_Niederlausitz
0.4 Swiss_Italian
0.4 TKM_Parkhai_EBA
0.4 Tunisian_Jew
0.4 UKR_Srubnaya_MLBA
0.4 USA_colonial_period
0.4 Welsh
0.2 Anatolia_Kaman-Kalehoyuk_MLBA
0.2 Anatolia_Kumtepe_N
0.2 Avar
0.2 Baltic_LTU_Late_Antiquity_low_res
0.2 Berber_MAR_TIZ
0.2 Czech
0.2 England_Roman_Near_Eastern_o
0.2 French
0.2 Greek_Crete
0.2 HRV_Impressa_N
0.2 HRV_Vucedol
0.2 HUN_Balaton_Lasinja_CA
0.2 Iberia_Central_CA
0.2 Iberia_Southeast_c.3-4CE
0.2 Iranian_Mazandarani
0.2 IRN_Hasanlu_IA
0.2 ITA_Collegno_MA
0.2 Italian_Jew
0.2 Italian_Lazio
0.2 Italian_Lombardy
0.2 Italian_Northeast_o
0.2 Italian_Tuscany
0.2 Kubachinian
0.2 Kura-Araxes_ARM_Talin
0.2 Kurdish
0.2 Lak
0.2 Lebanese_Druze
0.2 Levant_ISR_C
0.2 Levant_LBN_MA_Mixed
0.2 Levant_PPNB
0.2 MAR_LN
0.2 Mozabite
0.2 ROU_N
0.2 RUS_Saltovo-Mayaki_low_res
0.2 Samaritan
0.2 Slovakian
0.2 Spanish_Soria
0.2 SRB_Starcevo_N
0.2 Syrian_Jew
0.2 TKM_Parkhai_MBA
0.2 Turkish_Trabzon
0.2 Wales_CA_EBA
0.2 Yamnaya_KAZ_Karagash


Basilicata:


Target: Italian_Basilica
Distance: 0.0415% / 0.00041483
Aggregated
7.8 DEU_MA_o
6.8 SRB_N
6.0 Italian_Liguria
5.6 UKR_Chernyakhiv_Shyshaky
5.4 Shetlandic
4.8 IND_Roopkund_B_o
4.6 EGY_Late_Period
4.0 GRC_N
3.6 GRC_Minoan_Odigitria_low_res
3.6 UKR_Srubnaya_MLBA
3.2 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
3.2 Samaritan
3.0 Anatolia_Kumtepe_N
2.8 Kura-Araxes_ARM_Talin
2.8 UKR_Chernyakhiv_Legedzine
2.6 ITA_Collegno_MA_o1
2.2 TKM_Gonur3_BA
2.0 BGR_Krepost_N
1.8 Ostrogothic_Crimea_ACD
1.8 Swiss_Italian
1.4 BGR_C
1.4 Greek_Trabzon
1.4 Iberia_Northeast_Empuries2
1.4 Iraqi_Jew
1.4 Levant_PPNC
1.2 BGR_IA
1.2 HUN_MA_Szolad_o2
1.2 Kura-Araxes_ARM_Kalavan
1.0 HRV_IA
0.8 Tunisian_Jew
0.6 England_Roman_Near_Eastern_o
0.6 Kaitag
0.6 Levant_LBN_MA_NE
0.6 RUS_Maykop_Novosvobodnaya
0.6 Syrian_Jew
0.6 TKM_IA
0.4 AFG_Darra_i_kur_MBA_low_res
0.4 Bell_Beaker_England_EBA
0.4 Chechen
0.4 Corded_Ware_Baltic_early
0.4 CZE_Early_Slav
0.4 MKD_N
0.4 UKR_Trypillia
0.2 Abkhasian
0.2 Anatolia_Kumtepe_N_low_res
0.2 Armenian
0.2 Baltic_LTU_Late_Antiquity_low_res
0.2 England_LBA
0.2 Georgian_Jew
0.2 Greek_Central_Anatolia
0.2 IRN_Shahr_I_Sokhta_BA1
0.2 ITA_Collegno_MA_o2
0.2 KEN_Pastoral_N
0.2 KEN_Pastoral_N_o
0.2 Kurdish
0.2 Norwegian
0.2 PAK_Arkotkila_IA
0.2 RUS_Kubano-Tersk
0.2 TKM_Parkhai_LBA_o
0.2 Turkish_Kayseri
0.2 USA_colonial_period



Apulia:


Target: Italian_Apulia
Distance: 0.0268% / 0.00026791
Aggregated
8.2 Italian_Liguria
6.8 HUN_MA_Szolad_o1
4.2 Levant_LBN_Roman
4.2 UKR_Srubnaya_MLBA
3.8 Armenian
3.0 HRV_IA
3.0 Iraqi_Jew
2.8 Samaritan
2.6 BGR_C
2.6 GRC_Minoan_Lassithi
2.6 Iberia_Northeast_Empuries2
2.6 IND_Roopkund_B
2.4 Anatolia_Kumtepe_N
2.4 DEU_MA_o
2.4 Italian_Aosta_Valley
2.2 ITA_Collegno_MA_o1
2.0 England_Roman_Near_Eastern_o
1.8 BGR_Krepost_N
1.8 Shetlandic
1.8 TZA_Zanzibar_Euro_outlier
1.6 Anatolia_Kaman-Kalehoyuk_MLBA_low_res
1.6 SRB_N
1.4 Georgian_Jew
1.4 GRC_N
1.4 Greek_Trabzon
1.4 IND_Roopkund_B_o
1.4 MKD_N
1.4 UKR_N_o
1.2 Georgian_Imer
1.0 IRN_Hajji_Firuz_IA
1.0 Levant_PPNC
1.0 SVK_Poprad_MA
1.0 TKM_Gonur3_BA
0.8 ARM_LBA
0.8 Baltic_LTU_Late_Antiquity_low_res
0.8 BGR_IA
0.8 CZE_N
0.8 KAZ_Golden_Horde_Euro
0.8 RUS_Saltovo-Mayaki_low_res
0.8 Turkish_Trabzon
0.6 Anatolia_Isparta_EBA
0.6 Italian_Northeast_o
0.6 Levant_ISR_C
0.6 Lithuanian
0.6 Romanian
0.6 Serbian
0.6 Welsh
0.4 AFG_Darra_i_kur_MBA_low_res
0.4 Berber_Tunisia_Chen
0.4 Bosnian
0.4 EGY_Hellenistic
0.4 HRV_Vucedol
0.4 HUN_Tisza_LN
0.4 Italian_Marche
0.4 Italian_South
0.4 Kaitag
0.4 KAZ_Ak_Moustafa_MLBA1
0.4 Kura-Araxes_ARM_Kalavan
0.4 Lak
0.4 Macedonian
0.4 Moroccan_Jew
0.4 UKR_Trypillia
0.4 Wales_CA_EBA
0.2 Anatolia_Kaman-Kalehoyuk_MLBA
0.2 Anatolia_Kumtepe_N_low_res
0.2 Chechen
0.2 Czech
0.2 Greek
0.2 HRV_MBA
0.2 HRV_Starcevo_LN
0.2 HUN_Protoboleraz_LCA
0.2 Irish
0.2 Italian_Umbria
0.2 Ostrogothic_Crimea_ACD
0.2 Sephardic_Jew
0.2 Spanish_Alacant
0.2 TKM_IA
0.2 Tunisian_Jew
0.2 UKR_Chernyakhiv_Shyshaky
0.2 USA_colonial_period
0.2 Yamnaya_KAZ_Karagash
0.2 Yemenite_Mahra

Genetique
10-20-2019, 03:33 PM
IMO yes.

Here are the mainland southerners. Calabria has high Levant, higher than Sicily (and IMO this sample may also be close to Palermo, Messina, etc) but the others do not.

Calabria:




Campania:




Basilicata:




Apulia:



Calabria samples come from Reggio, but what's interesting is that they aren't the same to those of East Sicily. Very low berber/arabian related populations but high Levantine and Caucasus = Byzantine Empire ! They seem although to have more italian/slavic populations.

Basilicata has 4.6% Egypt, interesting.

Sikeliot
10-20-2019, 03:56 PM
Calabria samples come from Reggio, but what's interesting is that they aren't the same to those of East Sicily. Very low berber/arabian related populations but high Levantine and Caucasus = Byzantine Empire ! They seem although to have more italian/slavic populations.

Basilicata has 4.6% Egypt, interesting.

The Calabria sample is, I believe, Catanzaro but I might be wrong. East Sicily is representative of Ragusa area not Messina.

Genetique
10-20-2019, 04:08 PM
The Calabria sample is, I believe, Catanzaro but I might be wrong. East Sicily is representative of Ragusa area not Messina.

Ok interesting ! Which part of Catanzaro ? I think it's Catanzaro, Catazanro (the city itself and its surroundings).

Sikeliot
10-20-2019, 04:14 PM
Ok interesting ! Which part of Catanzaro ? I think it's Catanzaro, Catazanro (the city itself and its surroundings).

Not sure.

But i do know we see that sample is the most distinct of all the mainland southern ones and is closer to Sicilians.

Genetique
10-20-2019, 04:20 PM
It can be also Southern Catanzaro it would make sense with the results. Inland Northern Catanzaro would be likely similar to Cosenza I think.

Genetique
10-20-2019, 04:29 PM
https://cdn.discordapp.com/attachments/492015060932362242/635514190039351308/unknown.png

https://cdn.discordapp.com/attachments/492015060932362242/635514341583749180/mappa-ndrangheta-provincia-catanzaro.png


As you see most of Catanzaro is plain or low mountains, when I talk about Northern Catanzaro it's the area that I draw in red, excluding coastal or nearby coastal villages (like San Mangone d'Aquino or Nocera Terinese).

Ajeje Brazorf
10-20-2019, 04:59 PM
Here are West Sicily, East Sicily, Crete, and Greece using ALL the populations in the G25 spreadsheet. It lead to a mixture of ancient and modern so how do I interpret some of these?

East Sicily:


West Sicily:


Crete:


Greek:


Calabria:


Campania:


Basilicata:


Apulia:


You already know the answer: a model with so many populations is a mess, leads to overfitting and makes little sense.


What stood out to me is the mainland Greek sample not scoring really anything Afroasiatic. The difference to Sicily or Crete is not JUST the Slavic.

Despite their Slavic ancestry, mainland Greeks also received a further Levant/Caucasus contribution that Ancient Greeks lacked, probably of Anatolian origin.

[1] "distance%=3.1920"

Greek_Thessaloniki

Anatolia_Barcin_N,50.8
Yamnaya_RUS_Samara,33.7
Levant_PPNB,5.8
Kura-Araxes_ARM_Kaps,5.8
WHG,2.5
IRN_Shahr_I_Sokhta_BA2_I8728,0.5
Han,0.3
IRN_Ganj_Dareh_N,0.3
Dinka,0.2

[1] "distance%=3.4077"

Greek_Peloponnese (n=2)

Anatolia_Barcin_N,50.1
Yamnaya_RUS_Samara,33.5
Levant_PPNB,6.5
Kura-Araxes_ARM_Kaps,5
IRN_Ganj_Dareh_N,2.8
WHG,1
Han,0.6
IRN_Shahr_I_Sokhta_BA2_I8728,0.5

[1] "distance%=2.2249"

Iberia_Northeast_Empuries2_I8215

Anatolia_Barcin_N,68.6
Kura-Araxes_ARM_Kaps,16.6
Yamnaya_RUS_Samara,14.6
Han,0.2

Sikeliot
10-20-2019, 05:28 PM
Despite their Slavic ancestry, mainland Greeks also received a further Levant/Caucasus contribution that Ancient Greeks lacked, probably of Anatolian origin.

[1] "distance%=3.1920"

Greek_Thessaloniki

Anatolia_Barcin_N,50.8
Yamnaya_RUS_Samara,33.7
Levant_PPNB,5.8
Kura-Araxes_ARM_Kaps,5.8
WHG,2.5
IRN_Shahr_I_Sokhta_BA2_I8728,0.5
Han,0.3
IRN_Ganj_Dareh_N,0.3
Dinka,0.2

[1] "distance%=3.4077"

Greek_Peloponnese (n=2)

Anatolia_Barcin_N,50.1
Yamnaya_RUS_Samara,33.5
Levant_PPNB,6.5
Kura-Araxes_ARM_Kaps,5
IRN_Ganj_Dareh_N,2.8
WHG,1
Han,0.6
IRN_Shahr_I_Sokhta_BA2_I8728,0.5

[1] "distance%=2.2249"

Iberia_Northeast_Empuries2_I8215

Anatolia_Barcin_N,68.6
Kura-Araxes_ARM_Kaps,16.6
Yamnaya_RUS_Samara,14.6
Han,0.2


Interestingly enough yes, you're correct.

I ran Sicilians through too and this model actually makes them very similar despite east/west. And compared to the Empuries sample, there is much more Levantine but also higher Steppe even for Sicily.

Target: Sicilian_West
Distance: 2.5478% / 0.02547849

38.2 Anatolia_Barcin_N
20.4 Yamnaya_RUS_Samara
19.0 Levant_PPNB
12.6 Kura-Araxes_ARM_Kaps
5.4 WHG
2.6 IRN_Shahr_I_Sokhta_BA2
1.8 IRN_Ganj_Dareh_N

Target: Sicilian_East
Distance: 2.3450% / 0.02345014

39.8 Anatolia_Barcin_N
20.0 Yamnaya_RUS_Samara
18.0 Kura-Araxes_ARM_Kaps
17.0 Levant_PPNB
2.8 WHG
2.4 IRN_Shahr_I_Sokhta_BA2

Kelmendasi
10-20-2019, 07:59 PM
When ran on Vahaduo:

Distance: 2.0475%
46.4 - GRC_Mycenaean
37.2 - CZE_Early_Slav
15.0 - Bell_Beaker_ITA
1.4 - Assyrian

Greekscholar
10-20-2019, 08:59 PM
Interestingly enough yes, you're correct.

I ran Sicilians through too and this model actually makes them very similar despite east/west. And compared to the Empuries sample, there is much more Levantine but also higher Steppe even for Sicily.

Target: Sicilian_West
Distance: 2.5478% / 0.02547849

38.2 Anatolia_Barcin_N
20.4 Yamnaya_RUS_Samara
19.0 Levant_PPNB
12.6 Kura-Araxes_ARM_Kaps
5.4 WHG
2.6 IRN_Shahr_I_Sokhta_BA2
1.8 IRN_Ganj_Dareh_N

Target: Sicilian_East
Distance: 2.3450% / 0.02345014

39.8 Anatolia_Barcin_N
20.0 Yamnaya_RUS_Samara
18.0 Kura-Araxes_ARM_Kaps
17.0 Levant_PPNB
2.8 WHG
2.4 IRN_Shahr_I_Sokhta_BA2

I'm not sure exactly what to make of an "all sources" model. Empuries_2 is very close to Mycenaean, the admixture from both the east/south and north do not seem to have taken place (in that community and its founder community) in the time period between the samples..........which is pretty substantial, over 1,000 years I think. In my experience, modern Mainland Greeks still need a Levant/Anatolian source population for best results.

Sikeliot
10-20-2019, 10:05 PM
I'm not sure exactly what to make of an "all sources" model. Empuries_2 is very close to Mycenaean, the admixture from both the east/south and north do not seem to have taken place (in that community and its founder community) in the time period between the samples..........which is pretty substantial, over 1,000 years I think. In my experience, modern Mainland Greeks still need a Levant/Anatolian source population for best results.


We see this in Sarno et al 2017 also. While mainland Greeks do not require nearly as much Near Eastern as Sicilians, Cretans, and the Dodecanese, they still do require some. I wonder if this is spillover from the islands or Anatolia.

Constantine
10-21-2019, 12:37 AM
You should add the Croatian Iron Age sample (Illyrian) and Bulgarian Iron Age sample (Thracian).



Samples like this would "tidy things up," since the "Slavic" here is suspect.

Magnetic
10-21-2019, 07:08 AM
the problem is that some components will be alreaey in another component . for example when you use "assyrian" the scores for levant ishkelon will be very low because the "assyrian" will already have ashkelon in it . people often do this mistake with their modelling . like using Areni C and Barcin at the sane time ending up getting a low score for barcin because Areni already having heavily barcin in it etc

Kelmendasi
10-21-2019, 01:58 PM
Samples like this would "tidy things up," since the "Slavic" here is suspect.
My breakdown when including BGR_IA and HRV_IA

Distance:1.9778%/0.01977785
36.2 - GRC_Mycenaean
32.8 - CZE_Early_Slav
14.0 - BGR_IA
10.0 - HRV_IA
7.0 - Bell_Beaker_ITA

Constantine
10-21-2019, 10:47 PM
My breakdown when including BGR_IA and HRV_IA

Distance:1.9778%/0.01977785
36.2 - GRC_Mycenaean
32.8 - CZE_Early_Slav
14.0 - BGR_IA
10.0 - HRV_IA
7.0 - Bell_Beaker_ITA

I was thinking replace CZE_Early_Slav with something less ambiguously Slavic, something closer to the likely Slavic homeland (around/northeast-ish Ukraine) and insert pre-Slavic Balkan stuff.

Interesting results tho.

Greekscholar
10-21-2019, 11:05 PM
I was thinking replace CZE_Early_Slav with something less ambiguously Slavic, something closer to the likely Slavic homeland (around/northeast-ish Ukraine) and insert pre-Slavic Balkan stuff.

Interesting results tho.

Yes, I prefer using Baltic_LTU_BA, or Baltic_EST_BA instead of CZE_Early_Slav. I feel the CZE sample is too recent in time and already contains admixture from the Balkans and overlaps with the much younger metal age populations. Just my opinion though.

Constantine
10-21-2019, 11:39 PM
Yes, I prefer using Baltic_LTU_BA, or Baltic_EST_BA instead of CZE_Early_Slav. I feel the CZE sample is too recent in time and already contains admixture from the Balkans and overlaps with the much younger metal age populations. Just my opinion though.

Yeah. There's no way CZE_Early_Slav is accurate (when we consider it Slavic). It's way too high for the entire Balkans, pretty much erasing anything non-Med.

The fact that is shows up as double digits in parts of Italy is probably the biggest red flag.

Sikeliot
10-22-2019, 01:28 AM
Yeah. There's no way CZE_Early_Slav is accurate (when we consider it Slavic). It's way too high for the entire Balkans, pretty much erasing anything non-Med.

The fact that is shows up as double digits in parts of Italy is probably the biggest red flag.


What about these as a model for Sicilians? I got a very good fit.

Sicilian_West
Distance: 1.0896% / 0.01089560

23.2 Italian_Lombardy
21.2 Greek
19.0 Greek_Central_Anatolia
15.2 Basque_Spanish
8.2 Tunisian
7.6 Lebanese_Christian
3.2 Yemenite_Ma'rib
1.4 Yemenite_Amran
1.0 Yemenite_Al_Jawf

Sicilian_East
Distance: 1.1830% / 0.01182966

50.6 Italian_Lombardy
30.6 Greek_Central_Anatolia
6.4 Yemenite_Ma'rib
5.4 Tunisian
3.2 Lebanese_Christian
3.0 Greek
0.8 Yemenite_Dhamar

Constantine
10-22-2019, 02:04 AM
What about these as a model for Sicilians? I got a very good fit.

Sicilian_West
Distance: 1.0896% / 0.01089560

23.2 Italian_Lombardy
21.2 Greek
19.0 Greek_Central_Anatolia
15.2 Basque_Spanish
8.2 Tunisian
7.6 Lebanese_Christian
3.2 Yemenite_Ma'rib
1.4 Yemenite_Amran
1.0 Yemenite_Al_Jawf

Sicilian_East
Distance: 1.1830% / 0.01182966

50.6 Italian_Lombardy
30.6 Greek_Central_Anatolia
6.4 Yemenite_Ma'rib
5.4 Tunisian
3.2 Lebanese_Christian
3.0 Greek
0.8 Yemenite_Dhamar

Not really sure what to make of it, but it seems to mostly line up historically. I'm not well versed in the samples. But Yemenite?? LOL

Sicily, to me, just seems to have more Neolithic DNA than Greece by a significant margin. Same for Greek islanders. I wonder if a significant portion of these MENA (including Greek_Central_Anatolia) percentages occur due to high-level sharing of this Neolithic base component. Not sure how the algorithm works (yet).

Sikeliot
10-22-2019, 02:20 AM
Not really sure what to make of it, but it seems to mostly line up historically. I'm not well versed in the samples. But Yemenite?? LOL

Sicily, to me, just seems to have more Neolithic DNA than Greece by a significant margin. Same for Greek islanders. I wonder if a significant portion of these MENA (including Greek_Central_Anatolia) percentages occur due to high-level sharing of this Neolithic base component. Not sure how the algorithm works (yet).

I tried to address this in the above model. I was able to isolate Berber, Arabian, and Levantine ancestry which add up to 15-20% of Sicilian DNA (similar to what Sarno et al 2017 calls "Near Eastern" and is about 15-20%) and is likely post-Neolithic, due to Phoenicians and Arabs. The "Greek Central Anatolia" could definitely be partially Neolithic as well as ancient Greek and has Caucasian affinities. Neolithic ancestry might also be embedded in Greek, Basque, and Italian Lombardy.

Mingle
10-22-2019, 03:19 AM
Yes, I prefer using Baltic_LTU_BA, or Baltic_EST_BA instead of CZE_Early_Slav. I feel the CZE sample is too recent in time and already contains admixture from the Balkans and overlaps with the much younger metal age populations. Just my opinion though.

Just did that:

https://i.imgur.com/Up7IAHk.png

But if the Baltic BA is 24% in Albanians and 14% in Greeks, then the Slavic percentage would be higher bringing us back to the 30%+ and 20%+ ranges respectively which isn't too different from what you get with CZE Early Slav or HUN Avar Szolad.

So we need some component to account for a wave(s) of Pre-Slavic non-Medish ancestry, if there is any. Otherwise, it looks like the Slavic may be that high (I think its too high but that's what it says).

Also, I think that when using pen=0.001 instead of the Vahaduo pen=0, the Baltic could possibly decrease by a small bit. The Greek sample above got 9.17% Baltic for instance on pen=0.001. Not a huge difference but in other cases changing the penalty may be significant.

Sikeliot
10-22-2019, 03:42 AM
Using Arabian as an element for Sicilians is actually good because it DECREASES total MENA to a reasonable level of 15-20%, otherwise we end up with extremely inflated estimates of nearly 40% with much of that not being recent at all but really being ancient Anatolian which is present in all of SE Europe, including Italy and Greece and would have been part of any Italian or Greek migration.

So if we control for that and try to use modern groups we find that about 15-20% MENA is reasonable, a mixture of Berber, Arabian, and Levantine. The rest of the ancestry can be accounted for by using different sorts of Italians and Greeks, plus a small amount additional West Med in Trapani.

Greekscholar
10-22-2019, 03:44 AM
Just did that:

https://i.imgur.com/Up7IAHk.png

But if the Baltic BA is 24% in Albanians and 14% in Greeks, then the Slavic percentage would be higher bringing us back to the 30%+ and 20%+ ranges respectively which isn't too different from what you get with CZE Early Slav or HUN Avar Szolad.

So we need some component to account for a wave(s) of Pre-Slavic non-Medish ancestry, if there is any. Otherwise, it looks like the Slavic may be that high (I think its too high but that's what it says).

Also, I think that when using pen=0.001 instead of the Vahaduo pen=0, the Baltic could possibly decrease by a small bit. The Greek sample above got 9.17% Baltic for instance on pen=0.001. Not a huge difference but in other cases changing the penalty may be significant.

If you swap out Baltic for that CZE_Early_Slav in your model, does it cut into the Mycenaean percent? That is what I tend to find, but it is possible the HRV and BGR samples could prevent that.

I did a few runs with the Neolithic/Copper Age sources I previously posted here, https://anthrogenica.com/showthread.php?18600-G25-coordinates-results&p=612801&viewfull=1#post612801

swapping out Yamnaya for the Baltic Bronze Age samples and the percent was nearly the same. When I substituted in CZE_Early_Slav, the percent increased significantly and ate into the other sources. I will try to post a few tomorrow.

Mingle
10-22-2019, 04:09 AM
If you swap out Baltic for that CZE_Early_Slav in your model, does it cut into the Mycenaean percent? That is what I tend to find, but it is possible the HRV and BGR samples could prevent that.

I did a few runs with the Neolithic/Copper Age sources I previously posted here, https://anthrogenica.com/showthread.php?18600-G25-coordinates-results&p=612801&viewfull=1#post612801

swapping out Yamnaya for the Baltic Bronze Age samples and the percent was nearly the same. When I substituted in CZE_Early_Slav, the percent increased significantly and ate into the other sources. I will try to post a few tomorrow.

You're right. The Early Slavic sample tends to eat from both the Mycenaean and Bulgaria MLBA sample.

But you have to take into account that the Slavs that came from Central Europe to the Balkans were not a "purely Baltic/northern people". They would have been more southern than the Baltic BA sample. Whatever amount of "Baltic BA" modern Balkan populations get, their actual Slavic percentage will have to be higher than that since the Slavs were more southern than the Baltic BA sample.

When taking into account that the Slavs were genetically more southern than the Baltic BA sample, then the percentages given by CZE Early Slav make some sense. I still think the CZE Early Slav gives high numbers, but so does the Baltic BA.

I'm just making numbers up now to illustrate my point, but it goes like this - if Greeks are 10% Slavic, they should get 5% Baltic; if Albanians are 20% Slavic, they should get 10% Baltic; if Bulgarians are 30% Slavic, they should get 15% Baltic; and so on.

So considering that, the Baltic numbers given by the calculator I posted are also high. So either another proxy for an earlier pre-Slavic northern component are needed or the Slavic percentage among modern Balkanites really could be that high.

By the way, Italians getting high CZE Early Slav could just be overfitting with their Germanic ancestry. I wouldn't think the calculator is bad just because Italians get high CZE Early Slav. If Central Anatolia Greeks got semi-significant ancestry from CZE Early Slav, then that would really say a lot.

pegasus
10-22-2019, 04:56 AM
Interestingly enough yes, you're correct.

I ran Sicilians through too and this model actually makes them very similar despite east/west. And compared to the Empuries sample, there is much more Levantine but also higher Steppe even for Sicily.

Target: Sicilian_West
Distance: 2.5478% / 0.02547849

38.2 Anatolia_Barcin_N
20.4 Yamnaya_RUS_Samara
19.0 Levant_PPNB
12.6 Kura-Araxes_ARM_Kaps
5.4 WHG
2.6 IRN_Shahr_I_Sokhta_BA2
1.8 IRN_Ganj_Dareh_N

Target: Sicilian_East
Distance: 2.3450% / 0.02345014

39.8 Anatolia_Barcin_N
20.0 Yamnaya_RUS_Samara
18.0 Kura-Araxes_ARM_Kaps
17.0 Levant_PPNB
2.8 WHG
2.4 IRN_Shahr_I_Sokhta_BA2

It seems the Normans had more impact than I thought but I think as others have stated, more late IA sources are still needed.


Target: Sicilian_East
Distance: 1.1135% / 0.01113491
Aggregated
45.2 Levant_ISR_Ashkelon_IA1
31.4 Iberia_Northeast_Empuries2
20.8 SWE_Viking_Age_Sigtuna
2.6 Corded_Ware_CZE


Target: Sicilian_East
Distance: 1.1253% / 0.01125309
Aggregated
39.2 Iberia_Northeast_Empuries2
22.4 SWE_Viking_Age_Sigtuna
19.4 Levant_ISR_Ashkelon_ASH066
15.2 Levant_ISR_Ashkelon_ASH2-3
3.8 IRN_Seh_Gabi_C

Coldmountains
10-22-2019, 05:38 AM
Just did that:

https://i.imgur.com/Up7IAHk.png

But if the Baltic BA is 24% in Albanians and 14% in Greeks, then the Slavic percentage would be higher bringing us back to the 30%+ and 20%+ ranges respectively which isn't too different from what you get with CZE Early Slav or HUN Avar Szolad.

So we need some component to account for a wave(s) of Pre-Slavic non-Medish ancestry, if there is any. Otherwise, it looks like the Slavic may be that high (I think its too high but that's what it says).

Also, I think that when using pen=0.001 instead of the Vahaduo pen=0, the Baltic could possibly decrease by a small bit. The Greek sample above got 9.17% Baltic for instance on pen=0.001. Not a huge difference but in other cases changing the penalty may be significant.
Baltic_Ba is a terrible source. Proto-Slavs and not even Iron Age Balts resemble them anymore. They are related and share the same genetic drift but this Bronze Age samples are way to HG shifted to resemble Slavs especially the ones who migrated to South Europe. Slavs had also more steppe ancestry than this Balts. They are maybe some distant "cousins" of Slavs but definetly not ancestral to them
Avar_Hungary_Szolad_2 is the best proxy for Proto-Slavic ancestry we have for now.

dosas
10-22-2019, 05:38 AM
The Dorians carrying more Steppe is not based on any actual data, from what I have seen, and, correct me if I am wrong, is more or less speculation based on myths. Sikeliot is right that Classical Age Empuries2 (post-dating 'the descent of the Herculeans') is incredibly close to GRC_Myc.

NarLFC
10-22-2019, 06:12 AM
When I use "Armenian" instead of "Assyrian" the SE Europeans favor Armenian over Levantine, and there is simply a replacement of the two components but not much difference in the percent. When I remove both Assyrian and Armenian, the Mycenaean and Levantine both increase for all of the groups and Italian Bell Beaker drops off sharply for South Italians.

The fit decreases noticeably when I remove Assyrian/Armenian and let some of that fall into Mycenaean. So I suspect the better fit is using Assyrian/Armenian to represent post-Neolithic West Asian input, and that Mycenaean input is actually lower than people assumed when they tried modeling Southeast Europeans without Assyrian/Armenian.

With Assyrian/Armenian included, South Italians all still get high Mycenaean except the Sicilians and Cretans and the Sicilians/Cretans have the strongest preference for actually including Assyrian/Armenian rather than excluding them.

Are these the Armenian samples on G25 that you are using? Because if so, then they are flawed.

Mingle
10-22-2019, 06:20 AM
Baltic_Ba is a terrible source. Proto-Slavs and not even Iron Age Balts resemble them anymore. They are related and share the same genetic drift but this Bronze Age samples are way to HG shifted to resemble Slavs especially the ones who migrated to South Europe. Slavs had also more steppe ancestry than this Balts. They are maybe some distant "cousins" of Slavs but definetly not ancestral to them
Avar_Hungary_Szolad_2 is the best proxy for Proto-Slavic ancestry we have for now.

My assumption was that using CZE Early Slav and HUN Avar Szolad was completely bloating their Slavic ancestry. So I decided to use Baltic so that we could set a more realistic baseline and then work our way from there.

But after using Baltic and seeing the Balkanites get it at such high numbers, the numbers from Early Slav and Avar Szolad don't look too bloated anymore.

Anyways, I took your suggestion and used that specific Szolad sample:

https://i.imgur.com/Dw28vfh.png

Its slightly surprising how none of them go after HRV MBA, even Croatia.

Also, I tried to use Bronze Age samples to avoid any possible overfitting with modern samples, but when using IA samples, the result wasn't much different:

https://i.imgur.com/57zEDaI.png

Coldmountains
10-22-2019, 08:05 AM
My assumption was that using CZE Early Slav and HUN Avar Szolad was completely bloating their Slavic ancestry. So I decided to use Baltic so that we could set a more realistic baseline and then work our way from there.

But after using Baltic and seeing the Balkanites get it at such high numbers, the numbers from Early Slav and Avar Szolad don't look too bloated anymore.

Anyways, I took your suggestion and used that specific Szolad sample:

https://i.imgur.com/Dw28vfh.png

Its slightly surprising how none of them go after HRV MBA, even Croatia.

Also, I tried to use Bronze Age samples to avoid any possible overfitting with modern samples, but when using IA samples, the result wasn't much different:

https://i.imgur.com/57zEDaI.png

You should replace Welzin either with Deu_Ma or some early Germanic. Welzin is rather Hungary_Ba-like and probably did not directly contribute to most modern pops.

Sikeliot
10-22-2019, 11:51 AM
It seems the Normans had more impact than I thought but I think as others have stated, more late IA sources are still needed.


Target: Sicilian_East
Distance: 1.1135% / 0.01113491
Aggregated
45.2 Levant_ISR_Ashkelon_IA1
31.4 Iberia_Northeast_Empuries2
20.8 SWE_Viking_Age_Sigtuna
2.6 Corded_Ware_CZE


Target: Sicilian_East
Distance: 1.1253% / 0.01125309
Aggregated
39.2 Iberia_Northeast_Empuries2
22.4 SWE_Viking_Age_Sigtuna
19.4 Levant_ISR_Ashkelon_ASH066
15.2 Levant_ISR_Ashkelon_ASH2-3
3.8 IRN_Seh_Gabi_C



The Swedish is coming up high because you have no Italic reference population or Steppe group. It is certainly not all or mostly Norman. If you put Bell_Beaker_ITA in there it'd cut through that Swedish.

Which one above is Sicilian West? You labeled them both East. I'd assume probably the first but I could be wrong.

Alain
10-22-2019, 12:31 PM
You should replace Welzin either with Deu_Ma or some early Germanic. Welzin is rather Hungary_Ba-like and probably did not directly contribute to most modern pops.
I think Balts have a higher steppe share than Slavs as they are the earliest Indo-Europeans in Europe and to what can you count Welzin Hungary BR2?

Greekscholar
10-22-2019, 04:35 PM
Baltic_Ba is a terrible source. Proto-Slavs and not even Iron Age Balts resemble them anymore. They are related and share the same genetic drift but this Bronze Age samples are way to HG shifted to resemble Slavs especially the ones who migrated to South Europe. Slavs had also more steppe ancestry than this Balts. They are maybe some distant "cousins" of Slavs but definetly not ancestral to them
Avar_Hungary_Szolad_2 is the best proxy for Proto-Slavic ancestry we have for now.

The BA_Baltic samples being used are still very close to modern day Baltic people. By the time you reach 700AD, the age of that CZE_Early_Slav sample, there would have been significant mixing with the people already living in central Europe. That sample plots in the heart of the modern cluster of Germanic and Slavic people. Do you believe that sample is reflective of ancestral bronze age Slavs? Or the genetic make-up of dark age Slavs after already mixing with other populations?

https://anthrogenica.com/showthread.php?18600-G25-coordinates-results&p=611783&viewfull=1#post611783

When I model CZE_Early_Slav, the sample seems to require either Mycenean or BGR_IA (both very close as you can see.) The CZE_Early_Slav sample is likely reflective of the genetics of the people who moved into Greece during the Dark Ages, but the overlap in ancestry by 700AD with known ancestral Greek groups (Mycenaeans) makes it difficult to know how much ancestry modern Greeks got directly from these migrants, versus how much is a proxy for Slav+Thracian+other northern groups, or at worst, the overlap with Mycenaean is crashing that known component too low by directly eating into it.

I don't know the deeper answers. I just get worried when samples are from very different time periods and have obviously overlapping ancestry.

Greekscholar
10-22-2019, 05:11 PM
Oh, and thank you both for the modeling with Hun_Avar. I had used it before, but not paid too much attention to it. I will give it a more in-depth look.

Greekscholar
10-22-2019, 09:42 PM
OK, here is a basic 4-part model for Greeks only. CZE and Avar function about the same in this model, Baltic is much different. The exact difference is in the amount of Empuries ancestry in the model. Fits are all about the same.

https://imgur.com/N0hnqVb.jpeg

Here is another model adding in BGR_IA and HRV_IA. In this model, the CZE and Avar samples eat into the BGR_IA(not Empuries) significantly compared to Baltic. Again, fit is about the same for all three. But, if you add up the percents for all three, you still get about the same totals. So, I think we could be looking at a situation where that sum total represents the historic admixture from northern populations, regardless of how you choose to model it. In this case, only using the Baltic sample will under represent the genetic contributions from these people.

https://imgur.com/xl2GrzA.jpeg

The other possibility is that BGR_IA is too close to Empuries/Mycenaean and the model is unable to tell them apart. In this situation, the model is overfitted with too many similar Iron Age populations and the genetic contribution from these northern populations is over represented at the expense of Ancient Greek, and the Baltic sample is the best reflection of the actual amount of migration from the Slavic people. I don't know which is historically correct, and don't have a "rooting interest" in one result or the other. As mentioned, I just get concerned when I see fit change minimally, but percents change rather dramatically.

Sikeliot
10-23-2019, 04:06 AM
OK, here is a basic 4-part model for Greeks only. CZE and Avar function about the same in this model, Baltic is much different. The exact difference is in the amount of Empuries ancestry in the model. Fits are all about the same.

https://imgur.com/N0hnqVb.jpeg

Here is another model adding in BGR_IA and HRV_IA. In this model, the CZE and Avar samples eat into the BGR_IA(not Empuries) significantly compared to Baltic. Again, fit is about the same for all three. But, if you add up the percents for all three, you still get about the same totals. So, I think we could be looking at a situation where that sum total represents the historic admixture from northern populations, regardless of how you choose to model it. In this case, only using the Baltic sample will under represent the genetic contributions from these people.

https://imgur.com/xl2GrzA.jpeg

The other possibility is that BGR_IA is too close to Empuries/Mycenaean and the model is unable to tell them apart. In this situation, the model is overfitted with too many similar Iron Age populations and the genetic contribution from these northern populations is over represented at the expense of Ancient Greek, and the Baltic sample is the best reflection of the actual amount of migration from the Slavic people. I don't know which is historically correct, and don't have a "rooting interest" in one result or the other. As mentioned, I just get concerned when I see fit change minimally, but percents change rather dramatically.

Using your last model here is how Sicilians come out, along with Apulia and Calabria. It is a very good fit for Apulia, and not such a great fit for West Sicily.

West Sicily comes out with inflated Levant compared to the others because it's the only MENA reference being used, and they look extra Baltic because we have nothing to capture later Italic or Norman ancestry there. Apulia does not have much from either migration so it remains a decent fit.

East Sicily this time has the highest ancient Greek-like input along with Apulia.


Italian_Apulia
Distance: 1.0459% / 0.01045887
40.0 Iberia_Northeast_Empuries2
20.8 BGR_IA
15.4 Levant_Canaanite_MBA
12.2 Baltic_LTU_BA
11.6 ARM_LBA

Italian_Calabria
Distance: 1.8282% / 0.01828220
34.6 BGR_IA
24.6 Levant_Canaanite_MBA
21.6 Iberia_Northeast_Empuries2
12.6 ARM_LBA
6.6 Baltic_LTU_BA

Sicilian_West
Distance: 2.5051% / 0.02505070
37.2 Iberia_Northeast_Empuries2
28.4 Levant_Canaanite_MBA
21.4 Baltic_LTU_BA
13.0 BGR_IA

Sicilian_East
Distance: 1.8110% / 0.01810953
53.2 Iberia_Northeast_Empuries2
25.0 Levant_Canaanite_MBA
17.0 Baltic_LTU_BA
4.6 ARM_LBA
0.2 BGR_IA

Alain
10-23-2019, 04:19 AM
It seems that during the Avars the Greeks got a Slavic admixture to the south, decreasing understandable

xripkan
10-24-2019, 03:03 PM
The BA_Baltic samples being used are still very close to modern day Baltic people. By the time you reach 700AD, the age of that CZE_Early_Slav sample, there would have been significant mixing with the people already living in central Europe. That sample plots in the heart of the modern cluster of Germanic and Slavic people. Do you believe that sample is reflective of ancestral bronze age Slavs? Or the genetic make-up of dark age Slavs after already mixing with other populations?

https://anthrogenica.com/showthread.php?18600-G25-coordinates-results&p=611783&viewfull=1#post611783

When I model CZE_Early_Slav, the sample seems to require either Mycenean or BGR_IA (both very close as you can see.) The CZE_Early_Slav sample is likely reflective of the genetics of the people who moved into Greece during the Dark Ages, but the overlap in ancestry by 700AD with known ancestral Greek groups (Mycenaeans) makes it difficult to know how much ancestry modern Greeks got directly from these migrants, versus how much is a proxy for Slav+Thracian+other northern groups, or at worst, the overlap with Mycenaean is crashing that known component too low by directly eating into it.

I don't know the deeper answers. I just get worried when samples are from very different time periods and have obviously overlapping ancestry.

You are right. Czech_Early _Slav is modelled at G25 as a combinaion of Baltic BA (proto-Slavs), Scythians of Moldova(a mix of Thracians and Scythians) and Croatia Iron Age which is considered an Illyrian sample. Probably the people who came to Greece were not Proto-Slavs but similar to this sample. So this sample covers most of the northern ancestry Greeks in general received. If you use only Baltic BA you do not cover the Northern Balkan ancestry which is included in ancient Greek reference in this case. I noticed that I score the same ancient Greek percentage either in one case (Czech Slav) or the other (Scythians Moldova + Baltic).
For some reason I have score high Northern Thracian and Scythians but no Illyrian either HRV IA or Vucedol. Maybe the Tharcian-Scythian mix comes from my Thessalian ancestry.