Could Western Jews (Ash. and Seph.) descend from Aegeans and Levantine admixture?

[eolien]

several questions/comments:

1- I've also read somewhere that a Roman census gave 1M Jews but please take into account that the census concerns only Roman citizens.
2- Where did you get your sefardi samples from?
3- I cannot really comment on your statistical results BUT don't you need some kind of positive and negative controls to make your case stronger? I personally would take some Levantine populations and see how they deviate (for example Cyprus, Malta, Christian Lebanese, Druze)

thank you.

[kingjohn]

nice so you still need some southern french 10% for the sefhardi no matter
what you use in nmonte2 as the distance is shorter
that is cool
i copy paste it those are nice runs by you
regards
Adam

p.s
personally for me it is relevant because i am aschenazi/sefhardi mix

[Erikl86]

several questions/comments:

1- I've also read somewhere that a Roman census gave 1M Jews but please take into account that the census concerns only Roman citizens.
2- Where did you get your sefardi samples from?
3- I cannot really comment on your statistical results BUT don't you need some kind of positive and negative controls to make your case stronger? I personally would take some Levantine populations and see how they deviate (for example Cyprus, Malta, Christian Lebanese, Druze)

thank you.

1 - regarding the census - even if they were not 25% or even 10%, there are endless accounts, both Jewish, and non-Jewish, about huge amount of mainly Greek converts to Hellenistic Judaism.
2 - I got it from the Eurogenes blog itself.
3 - I actually have tried different Levantine admixtures instead of the classical Bronze Age model (although I think it is the most reliable). Check it out:

First of all, I've tried just plain 'ol simple Neolithic Levant + Tuscan + N. European:

[1] "distance%=12.6948 / distance=0.126948"

Ashkenazi_Eastern_Euro

N_Levant_Avg 50.6
IT_Tuscany 35.0
Polish_average 14.3

Got a huge distance of 12.8%.

Then replaced the mixture of N_Levant / CHL_Iran with Samaritans, the closest living people to ancient Canaanites, and got wacko results:

[1] "distance%=0.4041 / distance=0.004041"

Ashkenazi_Jew:average

Samaritan:GSM537032 61.5
French_East:French24434 38.5

Obviously Ashkenazi Jews aren't 60% Samaritans, 40% Rhineland French.

And that result didn't change regardless if I put Tuscans or Aegean Mediterranean admixture.

So, I've added some Caucasian admixture which the Samaritans seem to lack (but should exist, according to my theory, in Jews from early on because of admixture with Philistines, which would have Steppes and Neolithic Anatolian admixture if they are indeed Mycenaean-influenced descendants of Minoans from Crete). I've tried more ancient ones:

[1] "distance%=0.2488 / distance=0.002488"

Ashkenazi_Jew:average

Samaritan:GSM537032 32.4
Italian_Tuscan:NA20509 17.8
Italian_Tuscan:NA20799 17.6
Anatolia_BA:I2499 13.0
Armenia_EBA:I1635 9.7
French_East:French24434 9.6

Surprise surprise - Samaritans got about ~32% - similar to what I got when using Aegean people, and the result is much closer. But, I have two problems: Rhineland French is now way too low for Ashkenazi, about 9.6%. I blame Tuscans, because they might have higher level of Germanic admixture and perhaps direct Bronze Age Anatolian admixture communicated via some assimilated Greek people from south Italy.

Then I've tried to model Mycenaeans, to see if perhaps they were mixed with Levantines which is why they would show up as 50-60% of Western Jews:

[1] "distance%=0.2614"

Mycenaean:average

Minoan_Lasithi,59.2
Minoan_Odigitria,15.8
ArmeniaChL,7.6
Srubnaya_Spiridonovka,5.2
AnatoliaChL,4.8
Srubnaya_Rozhdestveno,3.2
Srubnaya_Barinovka,2.2
Srubnaya_Novoselki,1.2
Srubnaya_Uvarovka,0.8


Srubnaya is Bronze Age Steppes culture which dates some 4000 years ago, around the time I've read the Stepped people immigrated to the Balkans and mixed with Minoans, forming Mycenaean people.

CHL Armenians and CHL Anatolian represent earlier immigration to the Aegean and south Balkan areas.

Then, If I add any pre-Bronze Age Levantine admixture, for example Neolithic Levant or the admixture of CHL Iranians and Neolithic Levantines (which would basically mean Bronze Age Levantines), the distance grows dramitcally, check it out:

[1] "distance%=2.7243"

Mycenaean:average

Minoan_Lasithi,55.2
Minoan_Odigitria,17.8
Srubnaya_Rozhdestveno,4.4
Srubnaya_Barinovka,4
Levant_N,3.6
AnatoliaChL,3.4
Srubnaya_Spiridonovka,3.4
ArmeniaChL,2.8
Iran_ChL,2.2
Srubnaya_Novoselki,1.8
Srubnaya_Uvarovka,1.4

Adding Bronze Age Levantines (N_Levant + CHL_Iran), the distance grows about x10 times, and even then those are pushed down to mere 5.8%.

This means that current Levantine admixture found in modern Greek Islanders is post-Neolithic, post-Mycenaean and was most likely the result of Hellenistic West-East mixture (which also created Western Jews).

Then I've found an article which actually corroborates this small Levantine admixture, coming from Neolithic Anatolians to Minoans and Mycenaeans:

"The Minoans could be modelled as a mixture of the Anatolia
Neolithic-related substratum with additional ‘eastern’ ancestry, but
the other two groups had additional ancestry: the Mycenaeans had
approximately 4–16% ancestry from a ‘northern’ ultimate source related
to the hunter–gatherers of eastern Europe and Siberia (Table 1), while
the Bronze Age southwestern Anatolians may have had ~ 6% ancestry
related to Neolithic Levantine populations. "

So, less than 6%, through Anatolians. (I can't add links, so the name of the article is "Genetic origins of the Minoans and Mycenaeans".).

[eolien]

double post

[eolien]

Perhaps you misunderstood me. I was thinking why not run similar tests on Cyprus, Malta, Christian Lebanese, Druze INSTEAD of Ashkenazi and Sefaradi. I am also of the opinion that the reference samples chosen for these populations could already be selected to give a coherent picture. I for example fall exactly in the middle of Sefardi cluster in G25 PCA but it can be that many Ashkenazi samples would not show such a tight picture. Moreover, you still need to explain why we are close to South italians and not to Greeks et al.

Capture.PNG

[Erikl86]

Perhaps you misunderstood me. I was thinking why not run similar tests on Cyprus, Malta, Christian Lebanese, Druze INSTEAD of Ashkenazi and Sefaradi. I am also of the opinion that the reference samples chosen for these populations could already be selected to give a coherent picture. I for example fall exactly in the middle of Sefardi cluster in G25 PCA but it can be that many Ashkenazi samples would show such a tight picture. Moreover, you still need to explain why we are close to South italians and not to Greeks et al.

Capture.PNG

Oh I see. I'll run some tests then.

As for the explanation why South Italians (and Sicilians especially) are so close to Western Jews, and especially Ashkenazi Jews, rather than Greek Islanders (mainland Greeks are mixed with Slavic people since early Medieval Ages and so Greek Islanders are closer to south Italians than to mainland Greeks in many occasions).

About 37% of Sicilian background comes from Greek colonists which heavily populated southern Italy and Sicily since the 8th century BC. Greek was the majority language in southern Italy and Sicily until Medieval Ages. Greek people weren't expelled, but were Latinized and absorbed by the Latin-speaking people in Italy after the Byzantine Empire lost it's territories in the Italian peninsula.

Modern south Italians and Sicilians should cluster so close to Jews precisely because of my explanation:

Levantine admixture in Italians is probably due to Phoenician colonies in Sicily and South Italy dating some 3500 years ago, and then of course Carthage, which was a Phoenician empire centered in North Africa, controlled Sicily for centuries all the way to 150 BC, the last Punic War.

And later on, there was the Emirate of Sicily from 830-1090, about 260 years (almost 10 generations of Arab-Berber rule).

I would say a lot of Levantine, North African (J2 and E1b1 haplogroups) admixture in Sicily and southern Italy, along side Greek populations. Hence why modern Sicilians would cluster close to Jews. Add to that the fact that until 1800s they were ruled by a Norman kingdom from Northern Europe, which resulted in about 7% of Sicilians carrying North European haplogroups.

Mycenaean Greeks + Near Easterns + small amount of North Europeans, would indeed make them cluster even closer to Ashkenazi Jews than modern Greek Islanders would, and this is what we see.

[Erikl86]

I was thinking why not run similar tests on Cyprus, Malta, Christian Lebanese, Druze INSTEAD of Ashkenazi and Sefaradi.

Few thoughts regarding checking for Minoan or Mycenaean admixture in those populations though:

Christian Lebanese should have some Aegean admixture, because the region was heavily Hellenized. I would expect Greek Orthodox Christians to be more Hellenized than Maronites.

Judeans would mix with Philistines because they bordered each other. Philistines disappeared and were absorbed into local Southern shores of the Levant long after the Kingdom of Israel was already abolished by the Assyrians, and when Samaritans began to form their own separate identity, roughly around the 5-6th centuries BC.

Also regarding Samaritans - I am not aware of massive conversion to Samaritanism by Greeks, as opposed to Judaism. So if there is no admixture of Aegean with Samaritans, that would further prove their claim to be descendants of the Northern Kingdom of Israel, rather than Judeans which mixed with immigrants from Assyria as our Bible claims they are.

Cypriots would appear somewhat close to ancient Greeks but also heavily Levantine because there was a very long presence of Phoenicians on that island.

So actually, aside from perhaps Samaritans, which should have some low or nearly no Greek admixture, these are my predictions.

Now for the results:

Let's see about Christian Lebanese:

[1] "distance%=0.0292 / distance=0.000292"

Lebanese_Christian:average

Minoan_Lasithi:I0070 31.55
Iran_ChL:I1661 25.15
Levant_N:I1704 24.50
Iran_ChL:I1665 12.55
Mycenaean:I9041 4.30
Levant_N:I1699 1.65
Iran_ChL:I1662 0.30

Roughly 65% Bronze Age Levantines, 35% ancient Aegean. My bet, these are Greek Orthodox Christian Lebanese which have been sampled

Distance of 0.0292%.

Let's check Samaritans, with the exact same populations as Christian Lebanese:

[1] "distance%=0.5681 / distance=0.005681"

Samaritan:GSM537032

Iran_ChL:I1665 34.5
Levant_N:I1704 33.2
Levant_N:I0867 32.4

No Aegean admixture appeared.

Let's check Christian Israeli Arabs:

[1] "distance%=0.134 / distance=0.00134"

Arab_Israel_1:average

Levant_N:I1704 30.2
Minoan_Lasithi:I0073 26.6
Iran_ChL:I1665 21.8
Iran_ChL:I1662 15.9
Iran_ChL:I1661 5.5

Roughly 73% Bronze Age Levantines, 26% Aegean - mostly Minoan (No Mycenaean).

Next, let's check Druze:

[1] "distance%=0.0289 / distance=0.000289"

Druze:average

Minoan_Lasithi:I0073 21.20
Levant_N:I1704 20.35
Iran_ChL:I1665 19.65
Iran_ChL:I1661 14.40
Mycenaean:I9010 9.15
Iran_ChL:I1662 8.05
Mycenaean:I9006 6.50
Iran_ChL:I1674 0.70

Very similar to Christian Lebanese - Bronze Age Levantine is 63%, Ancient Aegean is 37% (both Minoan and Mycenaean). Distance is 0.0289%.

Now let's check Cypriots:

[1] "distance%=0.0998 / distance=0.000998"

Cypriot:average

Minoan_Odigitria:I9131 24.80
Mycenaean:I9041 21.40
Minoan_Lasithi:I0073 17.10
Iran_ChL:I1661 12.00
Iran_ChL:I1662 8.65
Iran_ChL:I1665 7.25
Levant_N:I1704 6.70
Iran_ChL:I1674 2.10

Mirror image of Druze . 63% Aegean, 37% Levantines. Not surprising at all.

All in all, it seems all non-Muslim Levantines I've examined are mixed with Aegean people to some degree 25-35% roughly, except for Samaritans which yield zero admixture with them. I would say this is probably the result of Hellenistic immigration of Greeks which settled in the Levant, probably mostly from the islands because these were closer (and perhaps even before the Hellenistic period). Palestinian Christians seem to be less mixed with Aegeans than Lebanese Christians, but I suspect that the average is biased by Greek Orthodox Lebanese. Maybe if I will examine only Maronites, it would yield closer results to Palestinian Christians (or even less so).

Western Jews probably are much more Greek because we are the result of already mixed Levantine-Greek ancestry, leaving and marrying Greek women converts in the Hellenistic world.

Now, let's examine Muslims Levantines.

First, Muslim Lebanese:

[1] "distance%=0.7972 / distance=0.007972"

Lebanese_Muslim:average

Iran_ChL:I1674 38.5
Mycenaean:I9010 29.9
Levant_N:I0867 18.9
Levant_N:I1699 12.8

Lebanese Muslims are further away - 0.79% vs. 0.029% from this admixture, and seem to have lower Aegean admixture.

Now let's check Muslim Palestinian:

[1] "distance%=2.3663 / distance=0.023663"

Palestinian:average

Levant_N:I0867 45.2
Iran_ChL:I1674 35.0
Mycenaean:I9010 19.8

Wow - really far away compare to the others. It is obviously mainly Levantine, with some Aegean admixture, but these populations are not enough to get accurate model.

Now I think I know what is missing - sub-Saharan admixture which is found among the Middle East only among Arab Muslims, but not Christians or Jews etc.

This sub-Saharan admixture comes from two sources: south Arabian admixture, and East African female sex slaves.

Let's see, if I add Bedouin to the admixture, if that helps:

distance%=1.5895 / distance=0.015895"

Palestinian:average

BedouinB:average 46.3
Mycenaean:I9010 29.4
Iran_ChL:I1674 24.2

Bedouin improves it, but not a lot.

Unfortunately I don't have sub-Saharan African samples with Basal-rich K7 stats, and I've tried any other combination - Saudis which should have higher sub-Saharan African admixture, even Christian Palestinians. None of those yield less than 1.3%. If I've taken out the Aegean, it would jump to +3%, so it seems to be partially Aegean as well.

Let's try to add Saudi and Bedouin to the Muslim Lebanese to see if it'll help, because Lebanese Muslims are the least admixed with recent sub-Saharan African genes:

[1] "distance%=0.0979 / distance=0.000979"

Lebanese_Muslim:average

Mycenaean:I9010 29.8
Iran_ChL:I1674 29.2
BedouinB:HGDP00607 28.4
Mycenaean:I9006 10.8
Saudi:SaudiA9 1.9

I guess Bedouins, depends on where they were collected, is already mixed with Levantines and Saudis (and probably takes all the Neolithic Levantine component). Obvious south Middle Easterners like Saudi is roughly 2%. This makes sense. See that Levantine and Arabian is 60%, Aegean is 40%. Distance is 0.097%. Basically, they are Lebanese Christians mixed with Arabians, as expected.

[Agamemnon]

I think this exercise is rather futile, using the exact same method one could easily argue that the present-day Lebanese owe the vast majority of their ancestry to an Anatolia_BA source for example (this is more likely to be the case in Cyprus however). In the same vein, it's very likely the Philistines will turn out to be very similar to the Mycenaean samples we have, if that's the case one could also make the argument that Western Jews owe most of their ancestry to the Philistines. On closer inspection, such models are undoubtedly bound to fall apart, which is exactly what I expect to see once we get more ancient data from the Mediterranean.

Also in many of the fits you've obtained for Levantine populations, replacing Iran_Chl and Levant_N with Levant_BA would be more accurate when populations such as the Mycenaeans and the Minoans are thrown in (for Palestinians though, additional African admixture is absolutely necessary to obtain convincing fits).

Edit: I can understand the appeal of such models, after all these ancient Eastern Mediterranean populations are extremely close to Western Jews (I was able to model my father as ~90% Mycenaean for example) and could potentially provide a straightforward explanation for the ethnogenesis of Western Jewry (which was Greek-speaking after all). We are literally grasping at straws with the limited data however, this is just the beginning and I have no doubt that future ancient data from the Mediterranean will make things a lot clearer.

[talljimmy0]

The answer lies in Alexandria. The City of London of the ancient world.

[eolien]

I am really confused. Why are Iran_ChL is considered part of Levant? Why Sidon samples are not used, why some populations prefer Minoan over Mycenaean?

And finally all the conclusions by Erikl86 are already biased because he tries to interpret/justifiy the results by assumed historical events. This is not a valid scientific method. For example Samaritans were known to be a large population in the common era. Therefore we expect many Palestinian Muslims to have origins in Samaritans. The same for Bedouins. Likewise Maronites and Druze are religious groups and not ethnic groups, but whatever their origins they are assumed to remain relatively isolated since the middle ages.