search this blog

Saturday, May 9, 2015

The time and place of European gene flow into Ashkenazi Jews


It looks like we're about to see yet another paper on the origins of Ashkenazi Jews. A poster on the topic was presented this week at the Biology of Genomes conference, and is available for download here.


It's a very reasonable effort, perhaps the best one so far. However, I'm of the opinion that the genetic structure of the Near East has changed significantly since the Neolithic. If that's correct, then using modern samples from the Near East to estimate Near Eastern ancestry in Ashkenazi Jews might not work too well.

For instance, let's assume, just for the sake of argument, that ~2,000 years ago the Levant was home to a population genetically almost indistinguishable from present-day Cretans. This might mean that Ashkenazi Jews are much less than 50% European. But we won't know until we see some ancient DNA from the Near East, including from the remains of early Jews.

Citation...

James Xue, Itsik Pe’er, and Shai Carmi, The time and place of European gene flow into Ashkenazi Jews, Biology of Genomes 2015 poster presentation.

139 comments:

Chad Rohlfsen said...

Certainly right. We have to know the Near East reference pops. Some Levantines have a good amount of SSA (8-12%), while some have next to none. Not only migrations, but admixture with African slaves needs to be considered. Pre-Islamic Levantines could be quite Cypriot like.

truth said...

I don't understand this obsession with Ashekanzies that science has, every year there is a few studies on them, there is no mystery we already know everythign about them from a genetic standpoint, wether it's haplogroups or genome-wide.

Maju said...

No, Chad: Palestinians are a quite distinctive population that owes little to neighbors. Palestinians are the real reference for Ancient Levant, at least the Southern Part (Canaan, Israel/Judea, Roman Palestine). Western Jews (incl. Askenazim, Sephardic...) are clearly from a much more northern geography, something like Cyprus or Anatolia. And that is because in the Hellenistic and Roman period (and also in the early Middle Ages to some extent) Judaism was a decidedly proselytistic religion that incorporated many many people from outside Palestine, not unlike their Christian offshoot (although these were more efficient in the mid-run, they were just another proselytistic Jewish sect originally). We can detect a weak thread linking Western Jews to Palestine but mostly they are from the Cyprus-Anatolia area by origin (plus whatever European admixture).

An interesting case is that the, once large, Jewish community in Cyprus precisely was involved in a very cruel conflict in the island in which they eventually lost, being the survivors expelled forever. I do wonder if all or most of the ancestry of modern Jews comes from that very specific community.

Refs.
→ http://www.jewishencyclopedia.com/articles/5169-diaspora
→ http://forwhattheywereweare.blogspot.com/2012/01/spehardi-jews-in-context-of-levant-and.html

Maju said...

Oops, did not subscribe. :(

Sam Hilsen said...

Ashkenazis score 2-3% SSA, a little less than Sephardics. If we double this (assuming as the authors did that they're only 50% Levantine) then we reach 4-5% SSA for the 'original Judaean' population, about what contemporary Lebanese and Syrians score. Thought: the SSA was already there, and contemporary populations aren't that admixed. We know of course of Egyptian and Ethiopian trade and military excursions in the area.

Sam Hilsen said...

I suppose the counter would be 'but Assyrians don't score much SSA.' Still, where did Ashkenazis and other Western Jewish groups get theirs?

Sam Hilsen said...

Maju, explain extensive YDNA sharing between Palestinians and AJ.

Maju said...

Care to provide the specific references, I should explain? Haven't dwelled into that since long ago and I don't recall the details. In fact there's not much clear data on Palestinian Y-DNA, a situation I lament.

It's possible, I guess, that Y-DNA lineages retain more Palestinian origins, while mtDNA ones are rather from other areas (some studies suggested this). If true, this would just show that, against traditional Jewish law, Jewish identity was mostly transmitted from father to son, and that it was easier at some time that women were incorporated into the Jewish community than men. But haven't studied the matter in such detail, really.

Re. your other argument: (1) Western Jews definitely don't look like Syrians/Lebanese (autosomal DNA). They may be slightly more Levantine than Turks/Cypriots but mostly they are like these, (2) Western Jews may have several origins homogenized by a long inbreeding history, including the Syrian and Egyptian diaspora and of course some Palestinian blood too, but these are not the core nor even the 20% of their ancestry.

Sam Hilsen said...

Forgot to say: there are pre-Islamic Levantine groups remaining, like Lebanese (and other) Christians, that score significant SSA, about twice what AJs do.

Sam Hilsen said...

Maju, Judaism was originally transmitted from father to son, and it wasn't until the 3rd century that this was changed. Karaites still go paternally and there are many references both in the bible and other contemporary sources of Jewish men taking non-Jewish wives.

As for the 'Jews don't look like Lebanese' people; well, that's preposterous. First, your perception is inherently flawed, you can't judge on that. Can we please keep the argument genetic? Second, they're half European. Do half Africans look African? Do half Norwegians look Norwegian? No, and half Levantine people tend not to look Levantine, though a few outliers no doubt can pass.

Not only do they share extensively with Palestinians, they do so with other Jews, too.

Your comment is pseudoscientific at best and lends itself to a perception of a distorted and romanticised history that never existed.

Maju said...

Care, Sam, to do a direct ADMIXTURE comparison instead of focusing on a single minor component?

Sam Hilsen said...

Haplogroups are not 'a minor component.' Do you care to substantiate any of your ridiculously absurd claims?

Either way I can't go autosomally as the Ashkenazi are a mixed population and won't have anything in common exactly with any MENA group, so here's IBD sharing:

http://blogs.discovermagazine.com/gnxp/files/2010/06/jewsFST1.png

Maju said...

"As for the 'Jews don't look like Lebanese' people; well, that's preposterous".

It is what the data says. I was initially (mis-)led by Behar's data to think that effectively there was some "Lebanese" or Phoenician link in Western Jews. Wishful thinking caused by insufficient K-level depth: Lebanese are much like Syrians and not much like Western Jews. and that is also apparent in Atzmon's data, as well as my own.

"Can we please keep the argument genetic?"

I'm all the time talking about genetics, autosomal genetics. Have you even read my reference linked above? When I say that they do not "look" like Lebanse, I mean in their genetic signature. The closest population by far are Cypriots, although Turks are also very close and I do not discard that they could be even more related to some Anatolian sub-population like Cilicians, for example (Tarsus had a major Jewish community but there were many others in Asia Minor). So far I haven't been able to test this hypothesis though. But what seems clear is that they are not more related to Lebanese.

Sam Hilsen said...

YDNA connections between all Jewish groups and Levantines: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC18733/
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1274378/
http://www.nature.com/ncomms/2013/131217/ncomms3928/full/ncomms3928.html

Different Jewish groups have shared ancestry:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3032072/

Jews and Palestinians have shared ancestry:
http://link.springer.com/article/10.1007%2Fs004390000426

Sam Hilsen said...

What on earth are you talking about? Are you talking about autosomal similarity (in which case they're close to South Italians, Sicilians, Maltese and Greek Islanders) or are you talking about actual descent? Either way you're badly wrong.

Maju said...

"Haplogroups are not 'a minor component".

I meant the "SSA" component you were all the time talking about.

"Do you care to substantiate any of your ridiculously absurd claims?"

Do you care to try to understand what I'm saying instead of misunderstanding it all the time and knee-jerk reacting based only on your misunderstandings and not on what I'm saying? Do you care to take a look at the reference links?

Sam Hilsen said...

The SSA comment was in response to Chad's comment about SSA. Of course my argument isn't based on that.

Please explain to me why you think, on a genetic level, Jews are descended from or share recent ancestry with Cypriots or Turks.

If the answer is 'because they share admixture components on the DODECAD project' or something along those lines, you have failed to understand autosomal evidence. You have also misread the evidence as, as I've said, they have more in common with other groups.

Your reference link, I'm afraid, would only be relevant if we accepted that a) a 2010 paper were accurate (bear in mind Ashkenazis plot with the people I said they did and b) that the Sephardic population was unmixed. Listen: a half-Norwegian half-Italian will plot with Germans. Does that make them German, or of German descent? No. And that's the mistake you've made. Do you see?

Sam Hilsen said...

The top comment is 100% accurate on your blog and you've totally misinterpreted it. He's telling you you're treating the contemporary populations as if they were exemplar of a single monolithic ancestry group - even the Turks! - and that that's a serious, serious error on your part. You also, however, should not be using autosomal evidence to suggest ancestry, as that's not what it's for. Extreme example: a mulatto will cluster with neither whites nor blacks, but is descended from both, yet not at all from that [presumably East or North African] group that he does plot with. Yet your evidence would say 'and this is proof mulattoes are actually Ethiopian.' This is a serious error, and it is why for mixed groups or groups presumed or suspected of being mixed we use IBD sharing, haplogroup sharing and other evidence to determine shared, recent and significant ancestry.

Krefter said...

@Sam,
"Your comment is pseudoscientific at best and lends itself to a perception of a distorted and romanticised history that never existed."

It's true that Ashkenazi Jews look Central-North European, not anything from the Middle East. This points towards some Central-North European admixture. So, their physical appearance is useful for genetics.

Their SouthWest Asian genetic affinities would be a shock to most people.

Maju said...

Thanks for the links, Sam. The two first papers are extremely old studies (2000, 2001) but it seems apparent that there is no particular relationship with Palestinians. Sadly no Turks nor Cypriots were studied.

The Rootsi studied, which I believe Palisto commented in the Kurdish DNA blog, suggests that R1a among Ashkenazim could be of Kurdish/Assyrian origins (not Palestinian in any case) and related to the Jewish state (Adiabene) that existed for a around a century in Kurdistan. So we are again before a case of convert Hellenistic Jews.

The fourth study is Atzmon 2010, which I have already commented as one of my references in favor of a Turkish or Cypriot origin. It does not deal with Y-DNA.

Finally the fifth study is pay-per-view. I'd ask you to send me a copy by email but being from the year 2000, I strongly doubt it has any interest or is different in any major way from the two first ones, which I have already addressed above.

So, for what I see, other than the Kurdish/Assyrian R1a lineages among Jews, the Rootsi study, there is no modern analysis of Western Jewish Y-DNA. Old ones are confusing but they do not suggest any particularly clear Palestinian connection either.

My most clear reference right now is that Sephardic Y-DNA pool mentioned by Adams 2008 in a study on Iberian genetics. It is indeed an Eastern Mediterranean genetic pool (25% J2, 22% J1, 16% G, 11% R1b, 8% E1b) but does not fit well with the Palestinian references of Fernandes 2011 (pay per view, I'm using Wikipedia's incomplete data in fact), which seem to have much lower J2, and much higher J1 and E1b.

truth said...

@ Krefter

Ahskenazies in general don't look central-european, they look mostly South-East European, similar to greeks or sicilians. See these Rusisan Jews :

http://limmudfsu.org/wp-content/uploads/2014/06/untitled-139.jpg

Sam Hilsen said...

@Krefter: anyone who would be shocked by the partial Levantine origins of Jews must have been living under a rock for much of their life. The layman's idea of a Jew is of some kind of pure Judaean survivor.

@Mujo: read my second comment, it's a more concise explanation of why you're wrong.

Chad Rohlfsen said...

No, Maju. You're wrong to assume anything about the ancients there. Palestinians have been affected by Islamic expansion and slave trade. I'll bet you that ancient Israelites will be half of the SSA that Palestinians are. I don't understand your obsession with that pop, when any Near East discussion pops up.

Maju said...

@Sam: "Please explain to me why you think, on a genetic level, Jews are descended from or share recent ancestry with Cypriots or Turks.
"

→ http://forwhattheywereweare.blogspot.com/2012/01/spehardi-jews-in-context-of-levant-and.html

It is the second link of my first comment, which you obviously haven't yet bothered reading. The main reason I have a blog is not to have to repeat everything once and again (also auxiliary memory).

Of course Atzmon 2010, and to some extent Behar 2010, are also influential in my opinion and converge with it, if not in the phrasing at least in the data.

"Your reference link, I'm afraid, would only be relevant if we accepted that a) a 2010 paper were accurate"...

No, because, besides of referencing those two key studies, I did my own independent research (removing the irrelevant European samples and focusing only on West Asia), whose results are coincident in essence but a tad more clear.

"bear in mind Ashkenazis plot with the people I said they did"...

No, they do not. Western Jews (Ashkenazi and Sephardi alike) cluster with Cypriots (and to lesser extent Turks) in the Atzmon paper. Very clearly so.

"the Sephardic population was unmixed"...

I don't see in any of the studies any significant difference between Sephardic and Ashkenazim. However Sephardic are much less inbred or had a much less dramatic bottleneck than Ashkenazim or Moroccan Jews, so I had to use Sephardic alone because otherwise the peculiars of these two hyper-endogamic populations distorted the analysis. I explain the matter at the beginning.

I guess I could have reduced these samples to single individuals, so they weighted less, but didn't think of it at the time. Feel free to repeat the exercise with whatever variant: it is of course replicable and all the materials are open source/access.

In any case both Atzmon's and Behar's data show very clearly that all Western Jews cluster together very tightly, so Sephardites should indeed be a good representation of the wider population. In fact Sephardites are almost certainly at the origin of the other groups: they are not admixed (nothing suggests that, at least not in any important way): they are ancestral, retaining greater internal diversity.

... "a half-Norwegian half-Italian will plot with Germans".

Not in ADMIXTURE, unless you do something very wrong: he/she will plot as a mix of Scandinavian and Italian, assuming these two populations are represented and their specifics show up (i.e. the normal thing, with the normal method). You're probably thinking in PCA but that's why different methods exist, because PCA has many limitations.

"He's telling you you're treating the contemporary populations as if they were exemplar of a single monolithic ancestry group - even the Turks!"

That's because I don't know how to make individual bars appear in the results (my technical limitations). But normally that doesn't matter because in fact populations tend to behave with some large degree of homogeneity, just check the Behar data if you doubt me. Also feel free to to reproduce the exercise in any improved form you can imagine... I will be very interested for sure.

It's possible that Palestinians show up as different subpopulations and I have been very intrigued about that but, since I don't know how to change the ADMIXTURE plotting into individual bars, I can't do anything to improve that. Another problem is that we don't have regionalized Palestinian samples such as Galilean Arabs, etc. I would bet that Palestinians have several subpopulations but still awaiting for a quality study on them: they only seem to appear as background characters in Jewish-centric studies, yet they are no doubt a more interesting population (and also probably the true descendant of Biblical Jews, at least largely so).

Sam Hilsen said...

Nearly everything you said was wrong.

Ashkenazis plot with Sicilians, and yeah, I read your blog, I even said 'I read your blog' earlier and described the first comment on it, and I told you why you're 100% wrong. You have clearly invested a lot of time on work that is essentially useless because you've misunderstood the premise. Behar's work didn't focus on admixture, and your desperate attempts to make everything fit your worldview when Behar actually agrees with me (take Behar's opinion on similarities between Western Jews, ignore his opinions on everything else, take old Dodecad runs, ignore other evidence, voila, you have a Frankenstein's Monster that supports your ignorance) is really sad. That said, I think Behar also finds what he wants to find, but his work on haplogroups is good.

Admixture works the way I said it works. Neither Sephardics nor Turks existed 2,000 years ago, and no, Sephardics are not the same as Ashkenazics - why are you even posting on this blog if you're going to ignore every single run David comes up with? I'm sorry, but your understanding of population genetics in general is extremely poor.

Maju said...

@Chad: well, you may think a priori whatever but the fact is that Palestinians are very distinctive from other West Asians (so not at all Arab immigrants nor undifferentiated Levantines). Their African component is not that big, no matter how you look at it, I have no idea why you underline that, when it's clearly not the bulk of the issue. It is true that I suspect that most of that African component is Mesolithic and that it corresponds with Lazaridis' "Basal Eurasian" but, even if I'm wrong in that, it's just a small fraction of the overall Palestinian ancestry.

What I would really like to see is for a change a serious genetic study of Palestinians, with regionalized samples, quality Y-DNA, serious comparisons with all their neighbors (Jews included), etc. You can't just deny (who knows why?) the interest of this population, the mystery that it holds, the many answers that it maybe can help produce. I demand serious comprehensive genetic studies of Palestinians, really! So far all we have is a blurry picture of a few, very unspecific samples, mostly used in studies about the Jews (usually the Ashkenazi Jews, as it seems it's all about them).

An interesting element here is that genetic studies on Jews seem to have come to a relative halt after it became evident, for anyone able to scrutinize the data at least, in Behar and Atzmon papers of 2010, that Western Jews are something like Turks. Somehow it's like someone in Tel Aviv thought: all this effort to get the wrong results? Scrap that expenditure!

Maju said...

"I'm sorry, but your understanding of population genetics in general is extremely poor".

I'm not sure why I have to trust more some Sam Hilsen guy fallen from nowhere than my own experience of almost two decades and a very large following on my blog. I won't pretend I'm exempt from error nor that my knowledge is perfect nor that everybody agrees with me all the time (that would be so horribly boring!) but I don't see you able to properly argument your position either. My impression is that you are in denial and that's why you fall to this kind of defensive disqualification. Just relax, take a cold shower, sleep over it and maybe think about performing your own analyses, which I'd be glad to peek on.

truth said...

@ Maju

Those are Turkish Jews, not Turks. Which are pretty much the same as Sephardics.



Sam Hilsen said...

If Truth is right, you've made yet another painfully simple error. Turkish Jews are Sephardic, indeed.

Your two decades are unfortunate because they entrench your mistakes, and make you unwilling to admit to them.

Joshua Lipson said...

Another issue: Why assume only S. Euro and E. Euro as source populations? Is there a proxy for North-Central Europe? Other component-based studies have suggested that AJ are way less than 17% "Slavic", and likely more German-French by comparison. This would also require a lower S. Euro estimate, and correspond well with my intuition that AJ look too Northern European on average to match the Levantine-Italian just-so story of the day.

Matt said...

@ Josh Lipson, the poster suggests they did in fact include British and French as a West group and apparently did not find a contribution from what I can tell; also it seems like from the poster that segments suggests 75% of Ashkenazis' European ancestry is Southern and this is 37% of overall ancestry, so if 25% of Ashkenazis' European ancestry were East European, that is 12% of overall ancestry, not 17%. And the paper only says "the rest mostly East European" so that would probably only mean 8-10% East European ancestry overall.

So indeed less than 17%.

A gene pool like Sephardi + 10% Lithuanian+Polish+Ukrainian, for Ashkenazis (plus some founder effects) doesn't seem too impossible. If anything also seems a little more consistent with the Sephardi->Ashkenazi vector on PCA (e.g. Davidski's West Eurasia PCA) etc, essentially pointing at East Central Europe than extra West European ancestry. 0.25-0.3% inflow per generation if it's an on average rate (and didn't all happen at a later time in history, perhaps almost at once).

Grey said...

My pet theory - based mostly on gut feel i admit - is Jews have two roots: the E1 Levant root and the J2 root from somewhere much further east adjacent to the himalayas somewhere with the J2 arriving in the Levant with a general neolithic population turnover in the middle east and then hooking up with the E1 lot who were already there.

Chad Rohlfsen said...

European Jews plot with Turks only because of their European ancestry. Jews only have 1/4 of the SSA in Palestinians. Islamic expansion and slavery clearly played a role. Iraqi Jews have about the same SSA as Ashkenazis. There was a Nubian influx of genes into Arabia and this went north with Islam. You can see it in the Southern Levant and Mesopotamia, into southern Iran. Ancient Israelites might be 3-4% SSA, but certainly not 8-12% like Palestinians and Lebanese.

Davidski said...

Ashkenazi Jews also have minor East Asian and possibly South Asian admixture, perhaps from contacts via the former Silk Road, which needs to be taken into account in these sorts of genome-wide studies.

A genetic contribution from the Far East into Ashkenazi Jews via the ancient Silk Road

http://www.nature.com/srep/2015/150211/srep08377/full/srep08377.html

Maju said...

@Chad: in ADMIXTURE you can see how they behave and Ashkenazim cluster with Turks and similar populations (Greeks, Cypriots, Armenians) in all studies (plus minor European admixture, yes but beyond the point): they do not cluster with Levant populations (unless you arbitrarily expand the concept of Levant to include Cyprus and Turkey) nor with Europeans other than sometimes Greeks, Sicilians, who are very much like Anatolians.

It's simple: just perform a supervised analysis with Germans, Turks and Palestinians as control populations. I tell you from now that Ashkenazim will score 80-90% Turk, 10-15% German, <5% Palestinian. Or do some of those f3 analyses that never lie, if you really care. Just bouncing heads against the facts is of no help.

Davidski said...

Ashkenazi Jews don't cluster with Turks or Armenians, unless the PCA is really crap.

They cluster with Sephardi and Moroccan Jews, except they're a little more shifted northeast, and also east Sicilians and some southern Italians.

Maju said...

LoL, David: Ashkenazi, Sephardic, Moroccan and Italian Jews are nearly indistinct from each other, they obviously cluster! I have yet to see a PCA where they cluster with South Italians: although it's not the first time I hear that legend of Jewish-Sicilian affinity it does not seem true and I have never seen any specific evidence. The only place where Jews and Sicilians cluster is in some neighborhoods of NYC (spontaneous silly joke).

Anyhow go and discuss with Atzmon: in his PCAs Western Jews systematically cluster with Cypriots and Turks, and please explore the supplemental materials, where there is more and more detailed and yet a bit redundant data confirming this.

And then again I get the same results spontaneously: no meaningful difference between Western Jews and Cypriot/Turks. Why don't you try risking your own analysis? I know you do that all the time with other populations, why not a peek on Jews?

Tobus said...

f3 in descending order:
Ashkenazi_Jew English Yoruba 0.151296
Ashkenazi_Jew Greek Yoruba 0.150034
Ashkenazi_Jew Armenian Yoruba 0.14817
Ashkenazi_Jew Cypriot Yoruba 0.14778
Ashkenazi_Jew Sicilian Yoruba 0.147076
Ashkenazi_Jew Turkish Yoruba 0.145971
Ashkenazi_Jew BedouinB Yoruba 0.13748
Ashkenazi_Jew Palestinian Yoruba 0.137024
Ashkenazi_Jew BedouinA Yoruba 0.132435

DStats in descending order
English Cypriot Ashkenazi_Jew Chimp 0.0082 4.847
Greek Cypriot Ashkenazi_Jew Chimp 0.0054 3.916
Armenian Cypriot Ashkenazi_Jew Chimp 0.0006 0.398
Sicilian Cypriot Ashkenazi_Jew Chimp -0.0013 -0.821
Turkish Cypriot Ashkenazi_Jew Chimp -0.0062 -4.875
BedouinB Cypriot Ashkenazi_Jew Chimp -0.0273 -15.242
Palestinian Cypriot Ashkenazi_Jew Chimp -0.0287 -21.477
BedouinA Cypriot Ashkenazi_Jew Chimp -0.0405 -27.371

Greek Turkish Ashkenazi_Jew Chimp 0.0117 12.862
English Turkish Ashkenazi_Jew Chimp 0.0144 11.728
Armenian Turkish Ashkenazi_Jew Chimp 0.0068 6.05
Cypriot Turkish Ashkenazi_Jew Chimp 0.0062 4.875
Sicilian Turkish Ashkenazi_Jew Chimp 0.0049 4.352
BedouinB Turkish Ashkenazi_Jew Chimp -0.021 -14.406
Palestinian Turkish Ashkenazi_Jew Chimp -0.0224 -25.678
BedouinA Turkish Ashkenazi_Jew Chimp -0.0342 -31.858

Ashkenazi Jews are closer to English (sorry, don't have German), Greek and Armenian than they are Turkish and Cypriot.

Maju said...

That's odd: it's always the same individual, Tobus? Because he/she may be particularly admixed.

Maju said...

How do you judge that the Greek stat (and the Armenian one also) is so close to the English one?

Tobus said...

There are 7 individuals making up the Ashkenazi_Jew "sample".

I don't really know enough history to make real judgement as to why the result are like this - at a guess I'd suggest that a fair whack of post-Roman generic "European" admixture is a likely cause?

Tesmos said...

Davidski, would you say that Sephardic Jews have also minor East Asian admixture or/and rather have more SSA admixture than Ashkenazi Jews on average?

Davidski said...

I don't think Sephardis have any East Asian admixture, and they probably don't have any more Sub-Saharan admixture than Ashkenazi Jews.

truth said...

@ Maju

You are wrong again. The Azmov "Turks" are Jewish Turks, not Turks. Here you have the description of their PCA :

Figure 1.

Principal Component Analysis of Jewish and Non-Jewish Populations

Principal components analysis of Jewish populations in a global (A) and regional context (B, PC1 versus PC2; C, PC1 versus PC3) CNVs (D, PC1 versus PC3). ASH, Ashkenazi Jews; IRN, Iranian Jews; IRQ, Iraqi Jews; SYR, Syrian Jews; ITJ, Italian Jews; GRK, Greek Jews; TUR, Turkey Jews. N. Italian is a combined set comprising Bergamo and Tuscan Italians. In (A), Middle Eastern non-Jewish populations are in blue, Jewish populations are in brown, and European populations are in red.

Maju said...

@"truth": look at the supplementary material, for Chaos' sake! There is a Turkish Jewish sample and there is Turkish Turkish sample.

truth said...

@ Maju

Yes, but the "Turks" on the PCA and the ADMIXTURE analysis both are Jewish turks. You said that Ashkenazies plot with Turks, when it's Jewish Turks.

Maju said...

To be more precise: there is a "Turk and Cypriot" sample, found only (hidden intently?) in the sup. materials. Western Jews cluster with them, particularly with a subcluster of them which for all I know are Cypriots.

THE GRAPH (the two dotted lines and associated legends are my annotations).

Maju said...

Also in Bauchet 2007, a European autosomal study, Ashkenazi Jews cluster with Greeks and Armenians. There are no other West Asian samples though but they also do not seem more extreme than these. Some Italians (Sicilians?) also approximate this Eastern Mediterranean cluster but much more weakly.

Davidski said...

Those plots are useless. You can see where different Jewish groups cluster on the PCA from Haak et al.

truth said...

Come on Maju, lighten up.
That plot from 2007 is obsolete and useless. Very limited plot. You can also see in that same PCA weird things like italians plotting with English or English with Basques, or Polish with English , etc.

The reason Armenians plot with Jews in that PCA is because it lacks other near-eastern samples. But normally Armenians are on a different cline, than Jews.

Karl_K said...

The Haak analysis is pretty clear. Look at the supplemental data. Ashkenazi Jews are like other southern Europeans, but you need additionally some amount of middle eastern, to make up the remainder.

I know they had a bottleneck, but more than 7 samples will be needed for a realistic comparison of admixture, because of the different times and locations involved.

Tobus said...

Perhaps I'm going crazy, but I'm finding it hard to reconcile D-stats with the Haak PCA - I'm expecting to see Greek, Maltese and Sicilian closer to Ashkenzi_Jew than other pops, but:
English Ashkenazi_Jew Greek Chimp 0.0146 8.829
English Ashkenazi_Jew Sicilian Chimp 0.0137 8.088
English Ashkenazi_Jew Maltese Chimp 0.0113 6.637
Armenian Ashkenazi_Jew Greek Chimp 0.005 3.076
Armenian Ashkenazi_Jew Maltese Chimp 0.0041 2.504

I can't seem to find any populations that Ashkenazi are closer to than Europeans: EU, AJ, X, Chimp is always coming up positive. eg:
English Ashkenazi_Jew Syrian Chimp 0.0069 4.217
English Ashkenazi_Jew Turkish_Jew Chimp 0.0085 4.962
English Ashkenazi_Jew Yemenite_Jew Chimp 0.0037 2.217
English Ashkenazi_Jew Iranian Chimp 0.0107 6.504

... doesn't seem to make sense.



Sam Hilsen said...

I'm significantly (3/4) Ashkenazi with the rest Spanish & Portuguese (Sephardic), Mizrakhic, Polish and minor English and I never got a single drop of South Asian on any run. Significant East Asian, though.

Maju said...

@Davidski: Western Jews cluster with Cypriots in Haak et al. too, why are you even arguing? Western Jews clearly belong to a NE Mediterranean population that includes (depending on the study's samples) Cypriots, Turks, Greeks, Armenians and (not entirely) maybe Sicilians too. It does not include neither mainline Europeans nor Levant peoples.

@"truth": 2007 is "obsolete"? Why? Why when it does not contradict but rather confirms what more recent studies find one after the other. It's just another piece of evidence pointing in that very same direction. And when several independent studies (Bauchet, Atzmon, Behar, Haak and my own amateur approach) all point in the same direction, then what's the point arguing? Which autosomal study says otherwise?

@Tobus: If anything we all agree to is that your f3 stats results are perplexing. What happens if you try with a Sephardic sample (for a change)?

andrew said...

I continue to be floored by the fact that so little of the history survives among the most literate population in Europe, other than monks and Roman elites, from the traditional beginning of the diaspora ca. 70 CE until perhaps 500 years ago or less.

How did a people who managed to preserve the Talmud fail to preserve any meaningful part of their own history for so much of that time period?

I suppose that the short answer is Pogroms, but it still seems inadequate.

Krefter said...

@Tobus,

Those D-stats are really unexpected. I wonder if D-stats are only good for comparing people with distant ancestry, not complex mixes of the same old pops. So comparing West Eurasians won't work.

Sam Hilsen said...

Andrew,
the Ashkenazi were a tiny and irrelevant population on their ethnogenesis. We know their history from the turn of the second millennium onwards.

Ryan said...

They probably shouldn't have included Palestinian and Bedouin samples as their "Middle East" proxy, as both have pretty noticeable Sub-Saharan African admixture (particularly the Bedouin).

It would be interesting to see what their results would have been with only the Druze, or a Druze/Samaritan or Druze/Samaritan/Lebanese comparison.

I suspect it would have replicated Behar's work and showed a roughly 50%/25%/25% breakdown of ME/South Europe/Other Europe.

Ryan said...

Maju - didn't you blog about a paper that showed Cypriots as having the most ancestry (I think they were checking mtDNA?) from the Pre-Potter Neolithic A?

I don't think you should assume Cypriot/Jewish affinities are new rather than old.

Nor do I think you should take Assyrian or Kurdish affinities as a negation of Ashkenazi Jews having a great deal of ancestry from people living in Israel ~2500 years ago.

I mean, if we keep pointing to neighbouring populations like the Cypriots, Cilicians, Kurds, and Assyirians, you end up eventually drawing a big circle around Israel. Which is no less persuasive.

For R1a, I'd remind you that Joshua returned from Babylon (supposedly), and presumably brought some DNA along with him and his fellow exiles. I think an older link is more likely though - Hebrew ethnogenesis occurred during the Bronze Age collapse. That seems like a likely time for some Indoeuropean DNA to be introduced.

What is clear from Behar et al, is that Palestinians and Sephardic Jews both derive from the same original source population. Later admixture pushed Palestinians towards Arabia, and Sephardic Jews towards southern Europe. Then further admixture between Sephardic Jews and North/Eastern Europeans (with a minor Khazar component) move Ashkenazi Jews further from their origin in Israel. Druze and Samaritans meanwhile were left mostly untouched.

Tobus said...

@Maju: I don't have a Sephardic sample (at least not one named as such), I'll do some runs with Moroccan and Georgian Jew samples and see what happens.

@Krefter: DStats are essentially just a count of whose alleles match whose, even with lots of admixture they should report the final affinities correctly.

Davidski said...

Sephardi Jews are marked Turkish_Jew in the Haak et al. dataset.

And the weakness or strength, depending on how you look at it, of f4/D-stats is that they work on allele correlations deep in the phylogeny, which means they don't notice recent drift/components.

So the results you're getting are correct, but they don't say anything useful about the recent origins of Ashkenazi Jews.

Tobus said...

Thanks Davidski, I've run through with Turkish_Jew and got the same results - English appear closer to every European, Middle Eastern and Central Asian population than both Ashkenazi and Turkish do.

You are probably right about the D-stat limitations, but still I'd expect "difference between" stats like D(Eu, Jew; X, Chimp) ("where Jews are different to Eu, which X population are they closest to") would score negative for at least *some* modern populations.


Grey said...

@andrew

"How did a people who managed to preserve the Talmud fail to preserve any meaningful part of their own history for so much of that time period? I suppose that the short answer is Pogroms, but it still seems inadequate."

I'd have thought you'd need an event that happened in a lot of places at the same time like the plague.

Grey said...

http://en.wikipedia.org/wiki/Yaghnobi_people

Yaghnobi kids seem to have very big heads.

Maju said...

@Ryan: there was some hint (IBD?) that Cypriots may be more related to EEFs than other West Asian populations but I doubt it is relevant here, else Sardinians would pop up and they don't.

"I don't think you should assume Cypriot/Jewish affinities are new rather than old".

Roman or Western Jews have very different ancestries than other Jews. Rabbinic-Jewish proselytism in the Roman Empire was surely only stopped c. 380 (Theodosius), much as other religions were targeted and persecuted to extinction from that year onwards. So the affinities, the formation of the nuclear Western Jewish population, should date to roughly 300 BCE - 300 CE, when Jewish proselytism was active, successful and tolerated (Christianity is just, in origin, a branch of that process).

"Nor do I think you should take Assyrian or Kurdish affinities as a negation"...

There are no such bulk affinities. Some (but few) Jewish lineages seem to originate from that area and should be linked to Adiabene (1st century CE), particularly as they affect the levites (second priestly caste) ancestry. It is known that in the case of Khazaria, ethnic Khazars were allowed to join the ranks of the levites (but not the cohen), it's probable that the same arrangement was done earlier in Adiabene, judging on the Y-DNA data. But we should not judge the whole by a small part.

"I mean, if we keep pointing to neighbouring populations like the Cypriots, Cilicians, Kurds, and Assyirians, you end up eventually drawing a big circle around Israel".

Actually yo do not: (1) Kurds and Assyrians are not really meaningful (just a few lineages) but, most importantly, (2) it's not "a circle around" but an unidirectional displacement to the Northwest: to Cyprus, Asia Minor and the Aegean, a region which was most important in the Hellenistic "diaspora" (essentially converts, rather than emigrants). This we can also observe in the original spread of the early Christianity, which was not neatly distinct from other Jewish diaspora at the beginning.

"Joshua returned from Babylon"

Babylon is much further to the South, unrelated to Assyrian/Kurds but rather to Chaldeans (would be modern Iraqi Arabs, largely the Sunni ones). But most importantly Iraqi and Iranian Jews seem to be the true descendants from that first diaspora and they are mostly unrelated to Western Jews, having their own very distinctive genetic history.

Anyhow, I insist: those few R1a lineages do not make but a very small part of modern Western Jewish Y-DNA pool, so not really relevant to the matter at hand, they are just a detail, a single tree in the huge forest. Details can be important but they should never obscure the big picture.

"Later admixture pushed Palestinians towards Arabia"...

That's not true. Actually Arabs and Palestinians do not share much: Palestinians are much more diverse, they also seem to have Northern (rather than Southern) admixture (maybe from PPNB). Anyhow both Palestinians and peninsular Arabs are, as we have discussed once and again elsewhere, similar to the West Asian component in EEFs, so their origins can't be argued to be a recent overlay from the South but certainly something old enough in the region to be decisive in the formation of the European Neolithic. Finally all that ideological mythology about Palestinians being Arab immigrants is based on absolutely no historical fact (even Herzl admitted that farmers, the bulk of the population, remained back in the country). The only large scale genocides in that area before the 20th century were the ones undertaken by the Assyrian Empire (affecting only Samaria) and by the Romans after the Great Jewish Revolt, which spared at the very least all Christian Jews (and probably many others, even the Romans could not be that systematic). Neither points to any sort of Arabian replacement, nor does the genetic data either.

Maju said...

@Tobus: "I don't have a Sephardic sample".

There is one in the 1000 Genomes Project data. That's the one I used.

@Davidski: "Sephardi Jews are marked Turkish_Jew in the Haak et al. dataset".

You and I were talking very specifically about Ashkenazi Jews in that study.

LouieJ said...

I am Ashkenazi and I have never gotten results for Turks or Armenians on gedmatch or anywhere else. I always cluster with other Jews groups, usually Moroccan Jews and with Samaritans. Sometimes also Druze. I have only gotten results once as being 5% Slavic. I have never gotten an east Asian admixture results and only once did I get 1% SSA results. I think it really can depend on the individual and how mixed they are in particular. Honestly, I was surprised I didn't get a lot more Slavic as people always think I'm Russian, but apparently, I have very little Slavic ancestry.

Maju said...

@Davidski: scrap the former; you were answering to Tobus, not me. Sorry.

@Andrew: To your question: when the main reason for literacy to even exist is religion, the main thing that matters is religious texts. People in the monotheistic past had very limited interest in history, very especially in the Dark Ages. All that mattered to them were legends about prophets, laws and miracles. You can see it in Islam but you do see it also in Christianity and of course in Judaism as well. So unless a prophet, a miracle or a law had to be recorded, there was no reason for them to keep any such record. In the middle ages, often ancient texts were erased to copy (once again) passages of the Bible over them. Only modern technology has allowed us in some cases to recover such precious ancient knowledge, erased by fanaticism and superstition.

Alexandros said...

Interesting how much heated up discussions get whenever a paper on Jewish genetics is published..

I noticed Maju being 'attacked' for mentioning that Jews may be related to Cypriots and/or Anatolians. I do not know about Anatolians, but I may be able to contribute a couple of things on the Cypriots-Jews issue.

Being the administrator of the FTDNA Cyprus DNA project I went through many Cypriot autosomal data, far more than the usual 12 used by the aforementioned studies (Behar et al, etc.). What becomes apparent from our project is that Cypriots are very close to Sephardic Jews in terms of their current genetic profile (i.e. proportion of admixture components) but appear to have higher shared ancestry with Ashkenazi Jews, both in terms of Y-STR haplotype matches and autosomal DNA. The autosomal sharing comes mainly from the myOrigins tool of FTDNA, which although far for being accurate in general, does show shared ancestry between Cypriots and 'Ashkenazi Diaspora' as it is labelled. Many Cypriots have as much as 10-15% of that component. Now this is a very specific component which does not represent current Southwest Asian populations (captured under 'Eastern Middle East'). This common admixture between Cypriots and Ashkenazi Jews can be interpreted in the following 3 ways in my opinion:

1. Recent influx of Ashkenazi Jews in Cyprus (no historical record for this)

2. Ancient exodus of Cypriot Jews (there were thousands of them in Roman times) to Europe, later to become Ashkenazi

3. An ancient Levantine(?) population ancestral to both Ashkenazi Jews and Cypriots.

Choose your most likely explanation. I have mine..

Maju said...

I'm just stating the obvious and so are you, Alexandros. I do get attacks and denials but no good counter-arguments or meaningful counter-data, what is what actually matters. I for one will thank you for your interesting input.

Joshua Lipson said...

On the topic of Western Jews and Cypriots, I won't harp on the distorted conclusions from mtDNA at Halaf, but what about Y E-M123 as a point of ancient congruence? Then again, this is something Palestinians are in on, interestingly to the exclusion (at comparable rates) of Syrians, Lebanese, and Turks.

Ryan said...

Alexandros - Likely a combination of both 2 and 3 I'd think.

Matt said...

I think a lot of the argument here as I perceive it has come from Maju bracketing Turks and Cypriots together when PCA and ADMIXTURE tend to indicate they're of a similar level of difference as Basques and Bulgarians (at least the Turkish samples that are in David's PCA and the Laziridis and Haak studies). Or Dutch and Russians.

This despite the fact that yes, geographically Greek Cypriots and Turks are quite close.

I don't think there's anything objectionable in saying that Ashkenazi Jews may have been almost totally derived from a quite Greek Cypriot like Jewish population of Anatolia from the pre-Turkish Roman period. Although there is likewise nothing too strong to speak for it right now, to me, you will need to look at rare SNP sharing and adna and things like that to begin testing these distinctions, and that needs a lot of genomes etc.

Indeed this is pretty close to exactly what David has said. It's the insistence that Ashkenazis are genetically close to modern day Armenians and Turks that may seem not quite right.

Btw, off topic, Tobus / David, if you're reading, what do the D stats

Chimp Yamnaya Sardinian (Middle Neolithic Spanish / Baalberge)
Chimp Armenian Sardinian (Middle Neolithic Spanish / Baalberge)
Chimp BedouinB Sardinian (Middle Neolithic Spanish / Baalberge)
Chimp Moroccan Sardinian (Middle Neolithic Spanish / Baalberge)

look like?

Ryan said...

@Maju - "there was some hint (IBD?) that Cypriots may be more related to EEFs than other West Asian populations but I doubt it is relevant here, else Sardinians would pop up and they don't. "

Sardinians would be closer to EEFs, but Cypriots closer to the original first farmers source (EEFs already had a bit of WHG ancestry, possibly from Thessaly, no?)


Also Maju - your comments on proselytizing don't really amount to much though - no one is denying south European admixture into Sephardim.

" it's not "a circle around" but an unidirectional displacement to the Northwest: to Cyprus, Asia Minor and the Aegean"

Actually, not the entire Aegean, or Asia Minor. Just the part immediately adjacent to the Levant. And closest to Druze and Samaritans. I find it implausible that Samaritans are crypto-Cypriots.

"Babylon is much further to the South"

Presumably not the entire diaspora was in the capital though.

"That's not true. Actually Arabs and Palestinians do not share much: Palestinians are much more diverse.."

You mistake me. I'm not claiming that any external influences are the dominant component. I'm claiming that there were minor flows from more southern sources into the Palestinian genome that shifts them downwards on the PCA.

Bedouin admixture seems the largest vector. The Subsaharan component is the smoking gun though. I'd suggest more isolated groups like the Druze or Samaritans as better proxies.

Not claiming replacement at all.

" Anyhow both Palestinians and peninsular Arabs are, as we have discussed once and again elsewhere, similar to the West Asian component in EEFs, so their origins can't be argued to be a recent overlay from the South but certainly something old enough in the region to be decisive in the formation of the European Neolithic."

Arabs, Hebrews, Assyrians and the like were all pretty closely related groups if we go back 3,000 years too. Makes things tricky.

You say there's no new data or convincing arguments - what is your explanation for Cypriots, Druze and Samaritans clustering together?

Tobus said...

Matt, no "Moroccan" sample so I used "Moroccan_Jew":

Chimp Yamnaya Sardinian Baalberge_MN 0.0094 2.098 8913 8747 186698
Chimp Armenian Sardinian Baalberge_MN 0.0073 1.697 8940 8810 186810
Chimp BedouinB Sardinian Baalberge_MN 0.0087 1.998 8855 8703 186810
Chimp Moroccan_Jew Sardinian Baalberge_MN 0.009 2.02 8922 8763 186810

Chimp Yamnaya Sardinian Spain_MN 0.0122 4.587 16371 15975 343197
Chimp Armenian Sardinian Spain_MN -0.0006 -0.241 16229 16248 343660
Chimp BedouinB Sardinian Spain_MN 0.001 0.394 16073 16042 343660
Chimp Moroccan_Jew Sardinian Spain_MN 0.0008 0.338 16204 16177 343660

Maju said...

@Matt: in all studies I can remember Cypriots and Turks look very similar, also your typical PCA of West Eurasia would cluster Turks, Cypriots and Bulgarians (a bit to one side these) very far from Basques, so again I don't know what you're talking about again.

"I don't think there's anything objectionable in saying that Ashkenazi Jews may have been almost totally derived from a quite Greek Cypriot like Jewish population of Anatolia from the pre-Turkish Roman period".

That's what it looks like, certainly. The fine detail you ask for in a later sentence has never been properly studied (but look at Alexandros' comment, which clearly goes in that line), but otherwise the affinity of Western Jews (not just Ashkenazim) with these populations of Cyprus/Anatolia (also Armenians, maybe Greeks) is redundant across studies, so for me as certain as it can get at the current level of research.

@Ryan: "Also Maju - your comments on proselytizing don't really amount to much though - no one is denying south European admixture into Sephardim".

I'm not talking of mere secondary admixture, I'm talking of the bulk of the ancestry. And also I'm not talking of just Sephardim but of all Western Jews (Sephardim seem to be just the parent population, the ones who did not go through dramatic bottlenecks in the migration to the West, who still retain a reasonable level of diversity).

"And closest to Druze and Samaritans. I find it implausible that Samaritans are crypto-Cypriots".

Samaritans are, like Druzes, not Palestinian looking (in the genetic signature), so I doubt that they can be attributed true ancient Samaritan ancestry (but probably a small obscure Jewish sect that adopted that ancient identity and migrated to their "Holy Land" thereafter). It's a tiny hyper-inbred population, what makes it extra-difficult to analyze, but the case of Druzes is much more clear: their history confirms that they were (at least partly) religious immigrants from other regions ethnic Greek lands and Low Egypt most likely (what fits their genetic signature well). So judging them as the only true natives is quite forced, a wishful-thinking kind of interpretation that just happens to fit well with the Zionist discourse but not with dispassionate critical wannabe-objective analysis.

Maju said...

@Tobus: Moroccan Jews cannot replace Moroccans but they cluster with other Western Jews, i.e. Cypriot/Anatolian area. It is a very different case to Yemeni Jews, who are indeed similar to mainline Yemenis. Again the 1000 Genomes Project includes not one but three different Moroccan samples (north, center and south).

Matt said...

Maju: in all studies I can remember Cypriots and Turks look very similar, also your typical PCA of West Eurasia would cluster Turks, Cypriots and Bulgarians (a bit to one side these) very far from Basques, so again I don't know what you're talking about again.

Well, actually I went back and had a look at various PCAs and although they are still separable (Cypriots and present day Turks), there is some variability between different on West Eurasian PCAs, and it probably isn't quite as high as between Basques and Bulgarians, so that's right to point out, more like the level of difference between Spanish and Bulgarian

http://i.imgur.com/7SzPiMI.png - Eurogenes

http://i.imgur.com/0KKTxut.png - Lazaridis
http://i.imgur.com/2ypVFLX.png - Haak

and some outlier individuals from populations do almost overlap

They are separable on PCAs which aren't loaded heavily with particular recent drifted populations. Maybe I am having a bit of a shallow view of things here without considering the mtdna / ydna though.

@ Tobus, thanks. I wanted to see if there was any pattern of greater relatedness of Sardinian with Yamnaya, and modern Near Eastern populations, as the Sardinians in some of the ADMIXTURE being run get membership in some Near Eastern and Yamnaya oriented clusters, and plausibly they have some historical admixture (although *relatively* low compared to other Europeans since MN). Pretty unrelated to the main comment thread here.

It looks from those stats that the Sardinians are systematically closer to Yamnaya, Armenian, BedouinB and Moroccan Jews than Baalberge_MN, albeit at a pretty low level of significance. While compared to Spain_MN they are quite a bit closer to Yamnaya (at a higher level of significan than the Baalberge difference) while being very similar in affinity to Armenian, BedouinB and Moroccan Jews.

The level of significance is not high but more pronounced at least than

e.g.

Chimp Iranian_Jew English Icelandic 0 -0.01
Chimp Iranian_Jew English Lithuanian -0.001 -0.651

which you'd described in the LN/EBA comment thread and similar in size to Yamnaya Loschbour Iranian_Jew Chimp 0.0082 2.021, and some are at or close to Z=2.

Maju said...

@Matt: that's surely because the West Eurasian analysis hides the extreme differences between SW Europe and NE Mediterranean populations, these would be expressed in a third dimension but there is no room in a flat plot. They become very visible in Europe-only analyses, where there is invariably a Basque-Eastern Med polarity in PC2 (and that's a major reason why I like to have Europe-only analysis when considering Europe, rather than just West Eurasian ones, a flaw I did criticize for the Haak study, as it dramatically limits understanding).

Instead in West Asian analysis, Turks and Cypriots tend to plot even more closely together. Anyhow, it is not clear who exactly are those "Turks" - I've been told that from Konya but Turkey is large and complex enough to require regionalized analysis. Geneticists should use one sample per 5-25 million people, not one per state. Orcadians vs English? Lithuanians vs. Russians? Basques vs French? It's not apples and oranges, it's like blueberries and watermelons, like marbles and basketballs, like mice and elephants...

Davidski said...

No Maju, the typical genetic map of West Eurasia is typical because it reflects frequencies of the major ancestral components of West Eurasians. See here, for example...

https://drive.google.com/file/d/0B9o3EYTdM8lQSnRhSWppaTUzdkk/view?usp=sharing

The other types of plots you've seen reflect either poor sampling and/or less important genetic substructures caused by recent isolation, expansions and/or drift.

Balaji said...

The notion that the near-Eastern ancestry of the Sephardi and the Ashkenazi Jews is not from the Levant but instead from Anatolia does not seem right. Contrary to Maju's assertion Jews were no where near as proselytizing as the early Christians. Paul made it easy for Gentiles to convert to Christianity by relaxing the dietary restrictions, the need for males to get circumcised and various other rituals. But Jews have kept all these and made conversion into their religion possible only for the most committed. Surely most of the near-Eastern ancestry of Jews is indeed from the Levant.

Conversely Maju's assertion that Palestinian ancestry is overwhelmingly from the Levant also appears questionable. During the time of the Islamic conquest of the Levant, the majority of the population was Christian and remained so for a time after the conquest but now as Dhimmis, second class citizens paying Jizya to their Arab Muslim overlords. The latter would have had four wives each and several concubines so that the next generation would have a higher representation of Muslims. After several generations of this process, the Christians would be reduced to the small minority that we see today and the Muslims would have considerable ancestry from the Arabian peninsula, perhaps as much as 50%. Note that no large-scale genocide is needed to accomplish this. ADMIXTURE analysis for various Levantine populations are available in Harappa World.

https://docs.google.com/spreadsheet/ccc?key=0AuW3R0Ys-P4HdDhib1M5OE1wWENNb2haUFFWZzNBMEE#gid=0

The SSA of Ashkenazi, Sephardi Jews and of Samaritans is 1% and of Palestinians 7%. The African ancestry of the Palestinians would have been almost all via women who served as concubines to Muslims from the Arabian peninsula who settled in Palestine.

Maju said...

I don't think you understood what I said, Davidski (what is odd because we have discussed this before). I'm saying that when you exclude West Asians (which clearly form a distinct cluster relative to Europeans, so it is fully justified) the self organization of the PCA plot changes radically and forms a cross-like shape: the main arm (PC1) is organized just the same as PC2 in Haak or other similar plots: Sardinian vs NE European at the extremes, but the secondary arm (PC2) is something that does not show up in West Eurasian analyses, with a Basque vs East Mediterranean polarity. An example can be seen in Lazaridis, but there are many others. Replace Maltese by Caucasians or Turks and is the same basic structure, only when the West Asian vs Europe distinction takes over a dimension, then the Basque-Sardinian differences collapse and near-vanish, because they don't have room to be expressed properly, as they surely belong to PC3 (which is never shown in flat PCAs).

The isolation argument (mostly a delusional fallacy) is much better applied to Sardinians than to Basques incidentally but, in any case, it is only important in order to explain their extremeness in their respective axes, not the axes themselves. I would still like to know how Gökhem and the other Chalcolithic samples behave in the Europe-only PCA: it seems important.

But we are losing track of the main issue, which is Western Jewish genetics.

Maju said...

@Balaji: "Jews were no where near as proselytizing as the early Christians. Paul made it easy for Gentiles to convert to Christianity by relaxing the dietary restrictions, the need for males to get circumcised and various other rituals".

That is no doubt true and explains (along with maybe other elements) why Christian proselytism succeeded to far larger extremes than the more traditional Rabbinic-Jewish one. However we must understand that Christianity is very clearly at the origin a Jewish sect (there were others in that period) and that its proselytistic attitude has roots in pre-existent Jewish proselytism. In my first comment I linked to a very interesting article of the 1906 Jewish Encyclopedia, which, if you read in full, explains that proselytism and the likely convert origin of most non-Palestinian Jews back in the day. I've read other sources also: there was some major interest in "alternative" religions in the Hellenistic-Roman era and Judaism was one of the most attractive for many. On the other side Judaism, without a state, was struggling for survival. Often converts were treated as lesser members of the community but no doubt that changed as they became second and third generation members. Also we have many instances of nations (tribes or states) converting in bulk to Judaism: Adiabene, Yemen, Khazaria and many African populations (Berbers, Ethiopians were at least partly Jewish before becoming Christians). We don't know all the exact details (the information is not good enough) but the tendency is clearly there. Also there is no way that small Palestine could produce such a huge number of emigrants in such a short period of just a few centuries, more so without being documented anywhere. The reality is that Judaism (along with other Oriental religions like Mithraism or Isianism) acted as "spiritual resevoir" of an Orient that had been defeated several times by the West (Alexander first, Rome later), suddenly losing their historical centrality. Using the Greek area as trampoline these religions jumped to the Western Roman Empire but eventually only Christianity succeeded. All other religions were persecuted, except Rabbinic Judaism, which was tolerated but not allowed to proselytize anymore within the Empire and successor states.

"Surely most of the near-Eastern ancestry of Jews is indeed from the Levant".

That's a supposition that genetics does not confirm but actually rejects.

"[Muslim Arabs in Palestine] would have had four wives each and several concubines"...

That's a very ridiculous claim. Surely the princes and very wealthy men did so but not the vast majority. Anyhow, how would that matter if females also pass their autosomes? Or were also all those imaginary concubines peninsular Arabs. Ah, the fertile oases of Orientalism!

The reality you describe or rather imagine, would be the same not just for Palestine but for Syria, Lebanon, Iraq, Iran, Egypt, etc. Why single out Palestine?

The actual reality of Muslim increase in numbers in all conquered provinces is that, while initially conversion was moderately discouraged, later, after the expansion wave ended, it was instead encouraged. Monophysite Christians (and surely also Jews, in the very same legal and socio-economic situation) found Islam very similar to their previous faith (all them are variants of Judaism, their Christianity was not trinitarian and they did practice circumcission and pork taboo, both pre-Jewish Afroasiatic/Nilotic traditions) and also a slight legal-economic improvement if they converted, and so they did through the centuries. That's how Muslims gradually became dominant in the Levant and Egypt (but a sizable Christian and Jewish minority survived). Similarly many Jews no doubt converted to Christianity where this one was dominant, sometimes under duress but often only just because it was convenient and pretty much indifferent in terms of core faith.

truth said...

@ Maju

1. Cypriots don't even plot with Turks. They are different populations. Turks are like 6-7% mongolid admixed, while Cypriots are not. And Cypriots also have more of levantine-type of ancestry than Turks. The Cypriots are like halfway between Lebanese and Sicilian.

2. I've never seen Jews plotting with Turks (not Turkish Jews, whho are Sephardics).

Alexandros said...

@truth
"The Cypriots are like halfway between Lebanese and Sicilian".

Indeed they are. And in calculators including Greek sub-populations (i.e. MDLP), their closest populations are Cretans, Asia Minor Greeks and Aegean islanders, as well as specific Jewish sub-populations such as Sephardic Jews and Syrian Jews.

Turks are actually relatively far from Cypriots genetically obviously due to their recent Mongoloid admixture, as well as their higher Caucasus-like admixture (vs the Cypriot EEF-like admixture). In David's latest K9 calculator the differences are apparent.

Anyhow, Turks are exceptionally heterogeneous as some have high Mongolic and/or European admixture while others (i.e. Trabzon) have practically 0. I assume when Maju says that Cypriots are genetically very similar to Turks he means the Turks with limited European/Mongolic admixture (i.e. original Anatolians). Even those people are more Armenian-like with different west Asian components compared to Cypriots. In general, Cypriots and Jews are more Mediterranean-like while Turks are more Caucasoid-like.

Maju said...

@Alexandros: I meant that the standard Turkish samples and those of Cypriots perform very similarly in ADMIXTURE. I mentioned several examples along the conversation. Then people began with using a PCA as a shield.

In my experience Greek Cypriots tend to be quite close to Anatolian Turks - they are not identical however and Turks are not properly studied in this sense because they get no regionalized studies, as they deserve (large population with a complex history).

It's interesting what you say about Anatolian Greeks: makes sense. That population is however seldom seen in analyses.

"I assume when Maju says that Cypriots are genetically very similar to Turks he means the Turks with limited European/Mongolic admixture (i.e. original Anatolians)".

I mean the usual Turkish samples we see in studies or the 1000GP, I suspect it's always the same sample. They don't show anything "East Asian" for example in Behar 2010 nor in so many other studies. I guess some calculators may show that for certain individuals or samples (I don't pay much attention to those: generally prefer free running regional ADMIXTURE results - also they seem to tend to exaggerate exotic admixture, surely because their zombie references do not add up to any optimal fit) but in any case it must be pretty tiny, considering haploid genetics (better studied in this particular case).

"Even those people are more Armenian-like with different west Asian components compared to Cypriots. In general, Cypriots and Jews are more Mediterranean-like while Turks are more Caucasoid-like".

Makes sense but the distinction is quite subtle and Cypriots do not tend much to any other West Asian population. If you have to pick the closest West Asian pop. to Cypriots (excluding landless minorities maybe), Turks always get the top position. They are even closer than Balcanic Greeks (although probably not than Anatolian Greeks: I take your word on that because it makes good sense).

I don't want to exaggerate Turkish-Cypriot affinity but there is some serious such affinity in all ADMIXTURE runs, regardless of PCA distortions (PCA is a tool of limited power).

Also (and that's the actual issue) in all studies mentioned Western Jews fall extremely close to Cypriots and next in line are Turks. Lebanese only appear somewhat close in Behar's but Behar did never go deep enough in his analysis, probably too worried about the label "Levant", which he stretched at whim in his conclusions, and the other studies don't show any particular Lebanese affinity.

It'd be indeed interesting to run ADMIXTURE freely with more West Asian populations than the ones I used (for example various regional Turkish pops. or two different Lebanese samples based on religion, Anatolian Greeks, Armenians, regionalized Palestinian samples etc.) but I tell you from now that the results will be similar to what I and so many other studies found: Cypriot/Anatolian affinity overwhelmingly dominates in Western Jews.

Davidski said...

It's very easy to distinguish between Cypriots and the vast majority of Turks in ADMIXTURE runs.

Alexandros said...

@Maju

I see your points and I respect your views. However, I have to strongly disagree on what you mention regarding Turks being the closest West Asian population to Cypriots. This is just not right. Unfortunately, Gedmatch is currently down. I would have shared Oracle results from several Cypriots to help make my point. Basically, Turks are so far away from Cypriots, that Sicilians and South Italians appear in any Oracle list before (i.e. closer to Cypriots) the Turks. Note that the distance from Sicily to Cyprus is around 1500km and from Turkey to Cyprus around 100km.

As I mentioned before, what makes the Turks diverge from Cypriots is basically their higher West Asian (vs EEF) admixture and the presence of East Eurasian admixture. For example, see components below from David's K9 calculator:

Turk (Istanbul19708): East Asian 0.04, Siberian 0.08 (total 12%)

Turk (Aydin18419): East Asian 0.06, Siberian 0.09 (total 15%)

Turk (Aydin18483): East Asian 0.07, Siberian 0.10 (total 17%)

The vast majority of Cypriots score 0% in these components with a maximum of 1%. Therefore, it would have been impossible for Cypriots and Turks to cluster together based on the above. Some Turks (e.g. Trabzon) do have East Eurasian admixture close to 0%, but these people also do not cluster with Cypriots for the reasons I mentioned in my previous post.

Jews on the other hand, do cluster very close to Cypriots, especially if they do not have high North European (WHG-like) admixture.

Balaji said...

@Maju

I agree with many things in what you say about the Islamic conquest of Palestine and what happened after that. However, this is a question of numbers. How much of the Palestinian ancestry is from the Levant before the Islamic conquest and how much from outside the Levant? We know that Peninsular Arab Muslims brought in enough African concubines that presently 7% of the Palestinian ancestry is SSA implying that about 14% of Palestinian maternal ancestry is SSA. A reasonable guess is that a further 14% of the maternal ancestry is from the Arabian Peninsula. Let us also assume that the invading male Arabs who settled Palestine were roughly 14% of the male population. Let us further assume that the invading males had 5 times the reproductive success as the defeated native males. Then the non-native Levantine autosomal percentage would be 0.5*(14%+14%+5*14%)=49%. In other words Palestinians of the present day only have about half of the Levantine ancestry of the natives of the Levant before the Islamic conquest. The Sephardi and Ashkenazi Jews of Israel probably have about the same Levantine ancestry.

Maju said...

OK, Cypriots are somewhat different from Turks (standard sample, who knows in the regions) in the European affinity details. But that European affinity stuff doesn't stick up when you look only at West Asia, what causes that they form a cluster vs their Levant neighbors. But yes it seems that the Cypriot affinity is the most strict one with the current knowledge (which is very ignorant of Anatolian diversity) for Western Jews.

Ryan said...

@Maju: "Samaritans are, like Druzes, not Palestinian looking (in the genetic signature), so I doubt that they can be attributed true ancient Samaritan ancestry (but probably a small obscure Jewish sect that adopted that ancient identity and migrated to their "Holy Land" thereafter)."

That's downright circular. You assume Palestinians have been genetically isolated for the last 2,000, and then use that assumption to confirm the reverse for Samaritans. Nor can you explain the continuous and well attested Samaritan presence for more than 2,500 years.

Look at the ADMIXTURE results for Behar et al. Seriously. It's clear as day that both Palestinians and Samaritans derive from the same group.

You're looking at this backwards - it's not the differences between Samaritans and Palestinians that I'm pointing to - it's the similarities!

"So judging them as the only true natives is quite forced, a wishful-thinking kind of interpretation that just happens to fit well with the Zionist discourse but not with dispassionate critical wannabe-objective analysis."

With due respect, I never said they were the only true natives. If anyone is twisting things to fit their worldview it is you to fit an anti-Zionist discourse. I in fact said quite the opposite of what you claim. I said that the shared ancestry of Samaritans and Palestinians is proof positive that Palestinians are indigenous to Palestine!

Do you have an actual credible explanation for Samaritan/Druze/Cypriot affinities that is reflected in Behar's ADMIXTURE results?

You need to explain why Samaritan ancestry is the dominant component in Palestinians, Cypriots, and Sephardic Jews. Can you?

Maju said...

@Ryan: I sincerely do not care the least about tiny inbred religious sects. In fact I'd gladly ignore the issue of Jewish origins altogether but it's like there is a conspiracy to insert your weirdo sect's genetic pool everywhere! Even Utahns (a Mormon state) have become the standar reference for Europanness (ok it works fairly well but it's a mere fluke). Samaritans! What next?

"You need to explain why Samaritan ancestry is the dominant component in Palestinians, Cypriots, and Sephardic Jews. Can you?"

Samaritan ancestry? I don't see anything in the 3-individual Behar sample that suggests me that they are in any way "ancestral" to anybody but themselves. Anyhow, Behar never went deep enough in the K-means, so I do not see much in that analysis: it's poor.

M. Myllylä said...

Davidski wrote: "It's a very reasonable effort, perhaps the best one so far. However, I'm of the opinion that the genetic structure of the Near East has changed significantly over the millennia. If that's correct, then using modern samples from the Near East to estimate Near Eastern ancestry in Ashkenazi Jews might not work too well"

Speaking about he Near East genetic contunuity I agree with you. But if you mean that the Near Eastern genetic profile matches with Early Neolithic farmers (ENF) I certanly disagree. Considering Near East lookalike genes in Europe we see probably three different gene flows: 1) ENF, which seems to be very old in Europe, 2) real Near Eastern gene flow, which is strongest in Southeast Europe, 3) a mixed gene flow, which came from the East EUropean Plain and included certain Middle Eastern components. It is essential, and also challienging, to make differences between those three Middle Eastern gene flows in reconstructing the Europea history. Someoens speak about gene flows from Anatolian, East European Plain and Near East.

Arch Hades said...

The Christian population of the Levant is obviously more indigenous to the area than the Islamic Palestinians or any Islamic groups. Palestinians are like 10% SSA, not to mention have hyper inflated frequencies of J1 nearly twice that of Christians and Sephardi Jews. Obviously they have seen significant [more than likely IMO 10%+] geneflow from the Arabian peninsula. If anything the Christian populations of the Levant today are much more representative of the original Levantines than the Islamic groups.


I don't personally find Ashkenazi to be very Northern European looking, this new study estimates them to be 10-12% Central-East European and i'd say that's about right. Pigmentation wise they're probably darker than Northern Italians and Northern Spanish and craniometrically I'd bet they're more distinct from Central Euros than Southeastern Euros are. The cranio-facial pattern of Jews is very distinct.

Maju said...

@Arch: I don't see Christian Lebanese more "native" than Muslim Lebanese, I see them "more European". The differences are subtle anyhow and their smell to the patterns created by the Crusades, so that's probably it.

Arch Hades said...

Whether you see it or not, they likely are by a long shot. There's no way christian Lebanese are 20% Western/Central European/'Crusader' any way. I'd bet it no more than 5%.

The Palestinians and Lebanese Muslims meanwhile not only have a lot more outside admixture, they have it from very racially distinct groups like SSAs.

truth said...

The Crusader myth is so stupid. In fact, the Lebanese Christian show less "European" components than their muslim countrymen. Look at K15 :

North-Sea + Atlantic + Baltic + East-Euro :

Christian : 6.12%
Muslim : 11.36%

Maju said...

Not sure what reference are you using, "truth" but "North-Sea + Atlantic + Baltic + East-Euro" seems to miss all the Euro-Mediterranean stuff, i.e. the Italians particularly. Anyhow, I'd like to look at what you suggest, IF you provided a reference - thanks in advance.

truth said...

It's not a reference, it's from Eurogenes K15, but if you want a scientific study there is Haber et al. 2013 see their Table_S3, for the ancestral component labeled as "5" which is the European one, and the averages are :

Christians : 1.96%
Muslims : 4.64%

Maju said...

The European affinities you mention are tiny for either population, but fair enough. Actually if you look at fig S6 (ADMIXTURE data: same as your table at K=10 much easier to read), you see that the "European" (dark green) component vanishes in all West Asian populations at K=10, replaced by a lime green (Arab or Palestinian focused) component. This component or something very similar is in fact a central part of the study. Christian and Muslim Lebanese score identically at that K depth (previous levels only suggest very subtle differences anyhow) and the only difference among them is that Lebanese Muslims have a tiny African admixture that Christians mostly lack: they are slightly more cosmopolitan, but only very slightly so. This implies that there has not been any major immigration to Lebanon since History began to be written (or at the very least Christianization), only tiny interactions.

The problem is that we are comparing apples and oranges: the Hamer study shows that Lebanese are all nearly identical, regardless of religion, and that they ARE NOT similar to Western Jews. Western Jews are most similar to Cypriots but have excess European-like admixture relative to them (Sephardi too), that can't be explained by Anatolian influence. On the other hand, they also have some African admixture (magenta shades, mostly North African) that does not look like originating in Palestine: too high, let's see (North African component):

· Sephardi: 0.4-1%
· Ashkenazi: 0-7%
· Palestinian: 1-5%

My inference is NW African admixture, much like they have accumulated European one along the centuries. It's not West Asian for sure but should not be Egyptian either because Egyptian have high scores in the Tropical African component. So it's NW African almost certainly.

LINK for quick reference: http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003316

Maju said...

Actually Hamer 2013 is most interesting for the matter at hand: it's not a study on Palestinians but it is a study on West Asians (with a Lebanon focus and wider outgroups, and also including the two main Western Jewish populations). It confirms much if not all of what I've been saying, including that the "Arab" (lime green) component is clearly pre-Islamic. It also indicates that African admixture in the Levant is tiny across the board (and should be Egyptian in essence). So thanks for drawing my attention not so much to it (I had it in mind) but to the supplemental material (which I had forgotten and seems quite informative).

Ryan said...

@Maju - "In fact I'd gladly ignore the issue of Jewish origins altogether but it's like there is a conspiracy to insert your weirdo sect's genetic pool everywhere!"

I'm not Jewish FYI. I'm German, Scottish and Welsh.

In terms of Jews inserting themselves into everything, this is a thread about a study on Ashkenazi Jews, so hardly uncalled for here, no?

Samaritans and Druze are useful precisely because they've been endogamous. Yemeni Jews are useful for similar reasons - they appear to be Yemenis who have been isolated from more recent admixture. I'd imagine the Yazidi would be useful samples too.

Obviously endogamy can lead to its own specific drift, which we see in the West Eurasian PCA at K=5, 7, 8 for Druze, Bedouins and Palestinians, but it can provide useful information too.

"Samaritan ancestry? I don't see anything in the 3-individual Behar sample that suggests me that they are in any way "ancestral" to anybody but themselves."

They pretty clearly a good proxy for the majority of Sephardic and Palestinian DNA. As are the Druze, who are better sampled and number in the millions. Most Palestinians around Nablus have Samaritan last names, so even they are a larger group from a genetic standpoint than from a religious one. I'm focusing on Sephardic Jews because I believe you and I agree that they are the primary source population for Ashkenazi Jews.

Take a look at K=8 for the West Eurasian ADMIXTURE run on page four (here http://www.nature.com/nature/journal/v466/n7303/extref/nature09103-s2.pdf). Take a look at the Sephardic Jewish sample and compare it to the Syrian and Lebanese samples. What distinguishes between the two?

The main difference between them is that the Lebanese and Syrian samples have vastly less of the blue component that peaks in Sardinians and is so dominant among French Basques - presumably a Neolithic, South European component. And I think both you and I would agree that this is a signal of European admixture.

What else though? For one, the Lebanese and Syrian samples have a higher ratio of light blue "Georgian" to light pink "Yemeni Jewish" components. That shouldn't be surprising though - both groups are geographically closer to the Caucasus region too. Cypriots share this with Lebanese and Syrian groups as well - they are closer to the Caucasus genetically than Jews are. Again, not surprising if the origin settlers of Cyprus came from the coast of Syria, Lebanon or Anatolia.

Note too that this light pink component is virtually absent among all non-Jewish European populations. Only Jews and Middle Eastern groups like Druze, Cypriots and Yemenis have it.

That's where I think you go awry Maju. Cyprus is linguistically, politically and religiously European, but in terms of both genetics and geography it is part of the Middle East.

Genetic similarity between Cypriots and Jews isn't evidence against a Levantine origin for Sephardic and Ashkenazi Jews - it's evidence FOR it.

Consider the implications of your Cyrpiot origins model. How plausible is it that majority of Sephardic Jewish ancestry is specific to Cyprus (and potentially Crete). The Hellenic and Roman worlds were pretty large after all. Why just Cyprus? You point to a Jewish revolt there, but there were Jewish revolts in Palestine too which you discount out of hand. Why only Cyprus and not Judea too? You assume this, but where is the evidence? The genetic "evidence" you point to only when you already assume your conclusion. In your model, was Cyprus always so Levantine even?

I have a lot of respect for you Maju but I really feel you wander into unfounded territory here.

Ryan said...

Also @Maju - re: Lebanon and Arab/African admixture in general - I more or less agree with your comments, but I'd suggest Yemen as a better proxy for and more likely source of SSA admixture than Egypt.

Maju said...

You're getting me wrong, Ryan. Neither I thought you were Jewish nor I care. I've got very sane interactions with (some) Jewish people and I respect each one regardless of ethnicity, religion or even ideology (with some limits because respect is only for those who respect), for what they actually are or behave as individuals, as human beings.

My complain was about the over-representation of religious minority samples in many genetic studies, minorities which are often of very obscure origins and IMO mostly irrelevant. I may have gone a bit over the board overemphasizing that but you know: I have a temper.

"[Samaritans are] pretty clearly a good proxy for the majority of Sephardic and Palestinian DNA".

Sorry but I don't see that at all in the Behar 2010 ADMIXTURE data. At K=8 Samaritans totally lack the "brown" Palestinian component, which is incidentally present in all Western Jews at low values. They also lack the "dark blue" North European component which is omni-present among Western Jews (at low values except among Ashkenazim, which are somewhat higher). They are something like 60% "pink" Yemeni, 20% "light blue" Georgian and 20% "middle blue" Sardinian.

As for the Druze (Israeli Druze, which are a very small community (130,000 people) and hyper-endogamous at village level) they are dominated by their own "purple" component, virtually absent in any other population.

For one moment I thought we were looking at different things but obviously not from your words: "Take a look at K=8 for the West Eurasian ADMIXTURE run on page four"... Well, that's exactly what I'm doing and I don't see what you say, sorry.

"The main difference between them is that the Lebanese and Syrian samples have vastly less of the blue component that peaks in Sardinians and is so dominant among French Basques - presumably a Neolithic, South European component. And I think both you and I would agree that this is a signal of European admixture".

I wouldn't think so. It's too widespread even among people unaffected by the Crusades like Jordanians. Let's not forget these "components" are actually "affinity tendencies", nothing else. In this case I believe it is a "memory" of the pre-EEF affiliation. Remember that the EEFs (or their modern proxy the Sardinians) had around 50% West Asian (Levantine surely). That Sardinian component is pretty much the same as the EEF component in Lazaridis, etc. Other than Sardinians, no modern Europeans are "purely" in that bloc, so, unless you argue for a pure Sardinian origin of the "European admixture" in, for example Cypriots, it is not acceptable.

"Note too that this light pink component is virtually absent among all non-Jewish European populations. Only Jews and Middle Eastern groups like Druze, Cypriots and Yemenis have it".

Of course but remember that I'm advocating by a Middle Eastern origin of Western Jews but a Northern Middle Easterner one and not a Palestinian one (at least for the bulk of the ancestry). If you mix the Turkish and Cypriot samples (maybe 1:2 apportion), you get something like Western Jews, just add some (but not much) Palestinian ancestry of the low "brown" component type, and voila: your archetypal Sephardi. Maybe you can refine my "recipe" a bit but it's something like that.

...

Maju said...

...

"Genetic similarity between Cypriots and Jews isn't evidence against a Levantine origin for Sephardic and Ashkenazi Jews - it's evidence FOR it".

Not really unless you do as Behar and expand the concept of the Levant at whim. Let's be precise:

1. The historical homeland of ancient Jews was Palestine, not some stretchy "Levant". You could argue for Southern Levant, including Jordan but not an inch more. Anything else is other origin, much as a French is not an Andalusian, even if both are West Europeans, right?

2. Even if you argue for a Syrian-Lebanese (North Levant) origin, Western Jews don't fit that genetic profile: Syrians and Lebanese are much more Palestinian-like than Western Jews.

3. Cypriots are the closest thing to Western Jews but the signature of these seems to require a more complex "founder effect" modeling with a bit of Turkish here and a bit of Palestinian or otherwise Levantine there. This added complexity should be no big deal because (a) we don't have all relevant local samples (how well does that Turkish sample represents for example Cilician Turks? not too well, I bet), and (b) Hellenistic Jews could well have got a variety of origins and not a single one. But in any case we are talking of a main origin in the proximity of Cyprus (Cyprus itself?, Cilicia?) and not in Palestine. Therefore we are talking of essentially Hellenistic converts and not the "historical" Jews of the Bible, be them Moses, David or Jesus.

"Consider the implications of your Cyrpiot origins model. How plausible is it that majority of Sephardic Jewish ancestry is specific to Cyprus"...

Very plausible judging on the genetic evidence. Also we know that a very large and influential community of Jews were expelled from Cyprus after a brutal civil conflict, being banned from ever returning. I've always looked for other more continental origins (hence my insistence on Cilicia as possible origin, another area where Jews were important historically, but they had many communities through all Asia Minor anyhow) but I cannot discard a mainly Cypriot-specific origin at this point.

"You point to a Jewish revolt there, but there were Jewish revolts in Palestine too which you discount out of hand".

Not out of hand but out of genetic evidence! Also the prisoners of the Great Jewish revolt were sold as slaves or executed, the Cypriot Jews were just banished, so it's more similar to other historical expulsions of Jews, which just sent them somewhere else. The historical facts of Western Jewish genesis are, up to a point, obscure but the genetic facts are there for all to view. I stick to these.

"In your model, was Cyprus always so Levantine even?"

Normally I understand the Levant as the same as Arabic Al Shams or Greater Syria. It's not a word used in Spanish anyhow, we just use Middle East or Near East. "Levante" in Spanish is Valencia and Murcia. It doesn't matter because, as Behar demonstrates in fig. 1, it is a flexible term. He chooses to include Cyprus but I never before saw that, so I think it is only in order to justify himself and his undeniable hard facts within the context of Zionist ideology. What matters is: is it a Palestine or Southern Levant origin, as the myth would claim? And the answer is: absolutely not, the main origin is much further North and slightly to the West as well, be it Cyprus, Anatolia or a mix of them.

Just stating the obvious.

Ryan said...

@Maju - Before a respond fully, may I ask - in your view, what would the genetics of Cyprus have looked like 2,000 years ago, and how would they have got that way, and when? And if that's different from now, how did it change?

Maju said...

@Ryan: I see no reason to imagine it'd be substantially different than today. Why would you think that it could be different?

Ryan said...

Also I may be slow to respond... work... may take it to your blog if I take too long if that's alright.

Ryan said...

I don't think it would be different either. But when do you think it's affinities to Lebannon/Syria/Jordan come from?

Maju said...

Mostly Neolithic (not sure if there were two waves, would have to review the literature but I think I have a vague reference for a late Neolithic pop. replacement, not total surely). The affinities are anyhow not just with Lebanon/Syria but rather (or also?) with Anatolia: too large "Georgian" component, too small "Arab" one. If we assume that the North European in Turks is "late" (Indoeuropean for example), that the "pink" in the Levant is old (Levant Meso-Neolithic signature) and that the "Sardinian mid blue" is a "memory" of proto-EEF relations rather than back-flow from Europe, then I'd argue for a mostly Anatolian or at least quite Northern origin of the settlers of Cyprus. What weights most in my reasoning is the "Georgian vs Arab" (light blue vs pink) apportions, which are the two main components that we can find in West Asia and which have also been highlighted in other studies, like Hamer's. The European and African affinity fractions seem to have other implications (immigration or ancient emigration "affinity memory") but are not central to West Asian analysis proper.

Maju said...

To be a bit more precise, Ryan. I think that the pink component (and related ones, included the brown Palestinian and the mid-blue proto-EEF memory surely) represent the most "aboriginal" Meso-Neolithic signature of the Levant population (Kebaran, Natufian, PPNA), which expanded southwards in the Neolithic towards Arabia. Y-DNA J1 is also in this category at least largely. On the other side the light blue component is aboriginally "Highlander" (Zagros-Anatolia, by extension the Caucasus too) and that it expanded with PPNB particularly (which affected Palestine but not Arabia) and maybe other later flows like Halaf-related ones.

Ryan said...

So are you thinking that the "light blue" component in Lebanese/Syrians/Jordanians is intrusive?

Ryan said...

And if it is intrusive, when do you think that intrusion is from?

I wouldn't worry about the minor south European and north European components in these populations. Like you say, there's probably some old affinities there. The slave trade along with the Greeks and Romans probably explain some affinities to.

Personally I'm viewing the light blue / pink fractions as a cline, and one which is relatively old (older than the period of interest at least which would be 2500-2000 years ago). Would you agree?

Maju said...

Natufian and PPNA are present mostly in Palestine with a branch northwards via Syria, reaching at the extreme to that cultural knot that was Göbleki Tepe. So I'd say that the oldest inferable demic layer in Syria (no data so early for Lebanon nor coastal Syria that I know) should be the "pink" one ("green" one in Hamer). The "light blue" one ("blue" in Hamer) should be original from the Highlands (with main cultural center in the Zagros: Jarmo and such). However PPNB spread from North to South, so it's reasonable to think that the Highlander component in Syria and Palestine arrived then, producing the modern mix of both components or at least something similar to that. "Intrusive"? Yes within this context. Very old too anyhow.

"Personally I'm viewing the light blue / pink fractions as a cline, and one which is relatively old (older than the period of interest at least which would be 2500-2000 years ago). Would you agree?"

Yes, I do agree: they belong to the Neolithic, approximately, much as the two European components in the case of Europe. But the apportions of each component are anyhow informative about the possible affinities or lack of them of modern Western Jews, which was my point all the time before this circumlocution triggered by your questions.

Maju said...

Nice maps of West Asian late Prehistory.

Ryan said...

The questions aren't intended to be a diversion, I'm just swamped at work and lacking good internet access, so it's helpful to know where we already agree in the mean time. Hopefully you don't mind.

Re: your earlier comments on what exactly constituents the "Levant" - it's really one of those unhelpfully vague terms isn't it. It originally just referred to the Eastern Mediterranean coast, though now it refers only to the section between Anatolia and Sinai, sometimes including Cyprus, and sometimes not. Either way I think the inclusion of Cyprus here is valid - Cyprus is pretty clearly part of this light blue / pink cline that characterizes the Levant as a whole.

Also, re: Turkish admixture, do you mind if I ignore that component? I agree that Cilicia a likely site of a great deal of admixture, but the Turkic migrations have reshaped its genetic profile enough to be unhelpful here, and including Turkish DNA actually makes your model fit worse and requires more Palestinian to compensate. The Turks have too much light blue and not enough pink. So I'd consider your hypothesis to just be Cyprus + Southern Europe + Palestinian if that's alright?

Maju said...

"So I'd consider your hypothesis to just be Cyprus + Southern Europe + Palestinian if that's alright?"

That's not my "recipe". My recipe is Cyprus+Anatolia+Levant. Maybe 4xCyprus + 2xAnatolia + 1xLevant. The only Western Jews which show significant European admixture are Ashkenazi (and not too much anyhow). Of course Greek admixture could be harder to discern, especially as we don't have in these studies a Greek sample to compare with.

I use "Levant" as equivalent of Greater Syria, Al Shams, excluding Cyprus, Cilicia, etc. That part is possibly Palestinian but could also be Syrian/Lebanese.

As for what you say about migrations reshaping Turkish genetics, it's something anecdotal: Turks are almost identical to Armenians.

Ryan said...

Well, we only have the populations available in that analysis. Cyprus + which ones specifically?

Re: Turkish genetics, I agree it's not a large change, but it's enough to make things fit worse here.

Maju said...

"Cyprus + which ones specifically?"

I'm getting a bit annoyed with these questions, Ryan: Cyprus + Anatolia (Turks) + Levant (whatever, possibly Palestine). I already said that!! It's just an approximate "recipe", I haven't worked out the numbers.

Ryan said...

I'm not trying to be a dick here Maju - I was planning on actually doing some rough calculations based on your "recipe" and mine.

It's just if I stick to that recipe verbatim, it's impossible to get results like Sephardic Jews'. Sephardic Jews all have more "Sardinian" blue than "Georgian" light blue. The reverse is true for Cypriots, Turks and all Levantine groups.

So you need some other group to contribute more "Sardinian" blue somehow without bringing a whole bunch of "Georgian" light blue with it. I open to suggestions of which group.

Then I was going to do some rough numbers assuming the original Jewish source population was similar to Palestinians or Yemeni Jews (ie VERY and on the southern extreme of the light blue/pink continuum) and then again assuming they were genetically similar to Jordanians or Lebanese. That would give you a minimum and maximum for the original Hebrew component of the ancestry under the assumption Cypriot admixture is significant.

It'd need to be 3 populations though (I only have 3 constraints).

Maju said...

@Ryan: You mean "Sardinian" mid blue, right? The fraction in Sephardi T is very similar to that of Cypriots. The proportion of the Palestinian-like components is also very similar to that of Cypriots, maybe a bit larger in Jews. The only difference is that Sephardi Jews have a bit more dark blue (not much), what may just indicate that the Cypriot-like vector is similar but not exactly Cypriot (Cilicia?) or may indicate a slightly more complex admixture ab origin.

But you are right in that the multipliers I figured out in a lazy attempt to reply your question are wrong: it should be something like 80% Cypriot + 20% others, or even 90% Cypriot + 10% others.

Anyhow, figure out your numbers and then come back with your results. No point asking me things I haven't worked out to such detail.

Ryan said...

Maju - Sardinian mid blue, yes, but the bigger issue is that both Cypriots and Turks have too much "Georgian" light blue. You need a population with very little little of the Georgian component too.

I agree that we're using Cypriot as a proxy here, and I think that's sensible.

You need a third proxy population though that is both rich in the Sardinian mid blue and poor in the Georgian light blue. Tuscans? IIRC there is actually a lot of IDB sharing between Tuscans and Ashkenazi Jews at least, so it's as plausible a site for admixture as any, and really just a proxy for a broader Mediterranean.

Without this third proxy, you'd actually need a negative amount of Cypriot ancestry to substract out part of the Georgian component, which obviously isn't possible.

I'm just ignoring the dark blue component by the way. It's so small in the relevant populations that I don't think it tells us much or is worth focusing on, at least for Sephardic Jews. I hope that's alright.

Maju said...

@Ryan: If the Turkish sample is from Konya and the Syrian one from Damascus... there is a huge area in between them, including some large cities of antiquity, often with strong Jewish communities: Tarsus, Antioch, Aleppo and others. We also have the issue of a clear (albeit small) legacy left by Adiabene Kurds (and again no Kurdish sample here). We'd need a much more fine grained sampling of the region and particularly the "Turkish" SE, Syrian North, etc. With the current level of resolution, that's about it.

Also we don't know the exact alchemy being factored in the Western Jewish genesis, which could be quite complex in some aspects, we can just glimpse the best proxies: rather this than than, but I don't think we can jump from the sketch to the blueprint.

LouieJ said...

All this arguing over who is native to the levant is absurd, ok? Both jews and palestinians have native levant ancestry and both have non native ancestry, it's called reality. In thousands of years people change ok? They intermarry, invade, migrate etc. it's human nature. There is no ancient Hebrews today as they were then and even then they were mixing I'm sure. Just like there's no ancient romans today. They were absorbed by other people. Jews are jews because of their ancestry but also by keeping their culture and traditions alive. But reality is we are all mixed somewhere. Sorry but Arabs are not native to the levant either and if the argument being made is that palestinians are not part or even all arab, you are delusional. The fact that both jews and palestinians have ancestry from the levant has nothing to do with who should have the land now. If we are going by genetics clearly both should have it whether you want to admit that or not. I'm sorry to tell you but I'm largely levant in my dna. I don't understand why people feel the need to discredit jews in order to make a political argument in the present. What does one even really have to do with the other?

Maju said...

Nobody is rising the political argument, although it is somehow implicit because of the very discourse of modern Zionism (the early Zionism of Herlz was much less naive about the reality and admitted at the very least that the bulk of ancient Jews from Palestine must have remained in place), which makes unrealistic claims for political reasons. The question of Jewish and Palestinian origins is interesting on its own anyhow and I must say that the veredict of population genetics is quite clear:

1. Western Jews are descendant primarily of populations that are not OT Jewish but rather Hellenistic Jews from further North (converts).

2. Other Jewish populations have their own distinctive origins in various places such as Yemen, Ethiopia, etc. (again converts rather than OT Jews)

3. Palestinians are almost certainly (although the matter requires more research) largely descendants of OT Jews, Cannanites, etc.

LouieJ said...

As a new myself I can tell you whenever I hear someone trying to discredit jews as 'not real jews' but converts, it's almost always a political agenda. Further, as I said I had only 2-5% European genes, the rest was in fact Levantine. Further, if you've ever been to Israel you would know that jews and many palestinians do in fact look similar. Yes, even ashkenazi jews. Not all palestinians are dark skinned, just like not all Lebanese are dark skinned and the reality of that is because the dark skinned ones are at least in part mixed with ARABS who come from Arabia and themselves invaded the levant, not to mention dozens of other places where they treat people just as bad if not worse then how jews treat palestinians. The double standards are also absurd. You would also know that many palestinians themselves will tell you they believe their families only lived there a few hundred years. To say 'all palestinians are the real jews' and 'all western jews are converts' is a flat out ignorant assumption made with yes, a political agenda. I do not have any Turkish or Greek in my dna. I am a Jew. Do I want the so called promised land? Well I lived there and as far as I'm concerned the palestinians can have that dump. That doesn't change my history, unfortunately for you and others who want to pretend others are the 'real jews'.

Maju said...

It's not about "discrediting" and I'm sure you MUST look at the facts behind it on its own merits, letting aside whatever political implications. If nothing else to better understand your own roots. Otherwise it is pretty much as an African-American denying some European ancestry or many other cases of prejudiced denial of facts.

The historical reality is that Judaism was highly proselytistic in the Hellenistic, Roman and even early Middle Ages and only stopped doing that because Christianity and Islam forced them. Christianity is also clearly a form of Judaism and, like it, it was actively proselytizing, with greater success certainly. Initially it would have been very hard to discern an orthodox or proto-rabbinic Jew from a Christian and all known early Christians were also Jews. For both sub-sects a most important area of proselytism was Asia Minor (and nearby regions like Syria or Greece) and post-Jesus Christian holy books show a quick drift from a Palestine center towards Asia Minor and Greece going through Antioch, which was no doubt based on pre-existent Jewish communities, which would have been initially much more open to all this Yahvistic and Messianic stuff than Pagans, particularly after the defeat of the Great Jewish Revolt.

"mixed with ARABS who come from Arabia"

I strongly dispute this (although of course minor admixture can't be discarded and further research would be welcome). First of all Arabs were already in the Levant in Antiquity although restricted to the semi-arid frontier (Nabateans, Ghassanids), second they did not have the numbers to make a major demic impact, third there are no historical sources of any sort indicating any major Peninsular Arab colonization of Palestine (or anywhere else, barring a few war camps and such), fourth the majority of Palestinians (like other populations of the wider region) were already Christians when the Islamic conquest took place, if there was a demic replacement at all, it would have taken place in Roman times, after the Great Jewish Revolt, but the fact that the Acts of the Apostles, which takes place mostly in Palestine, is post-Josephus and shows no sign of meaningful changes rather defeats that argument. But the most important of all evidences is of course genetic: Palestinians show great uniqueness and diversity in their genetic footprint and can't be considered to form a clade with any of their neighbors be these Syrians or peninsular Arabs. Arabness is a linguistic category in the end, as is common with ethnic identities, so when Aramean was replaced by Arab, Palestinians became "Arabs", without any need for admixture, just as did Iraqis, Egyptians or Tunisians.

"To say 'all palestinians are the real jews' and 'all western jews are converts' is a flat out ignorant assumption"...

While I would not use those exact phrases (at the very least I'd replace "are" for "used to be once upon a time"), it is not any assumption but a conclusion based on objective facts. I would ask you to consider the facts themselves rather than stubbornly entrench yourself in a prejudiced view of things.

"I do not have any Turkish or Greek in my dna. I am a Jew".

I can only smirk at that claim. The facts say the opposite, even if nobody denies that "you are a Jew" (present tense based on present day categories).

LouieJ said...

Um first of all Christianity is not a form of Judaism. Second, while Judaism at one time did more widely seek converts I would like to know where you got your information that it was a widespread phenomenon.

Lol I'm sorry but this made me laugh. There's no indication arabs invaded in any major way? Really? Arabs took over enormous areas of territory all over the Middle East and Africa and are to this day raping black Africans every single day. Arabs claim Sudan is an Arab country. Joke!!!! And completely out of touch with reality. If you would like to dispute the huge areas Arabs invaded I'd be happy to tango, but please only post academic sources:

https://cmes.uchicago.edu/page/maps-middle-east

LouieJ said...

Ps you can smirk all you want it doesn't change my dna which I'd be happy to post for you. You have an agenda for making arab history smell rosy when it's anything but in fact the only people who've stolen more land and committed more atrocity is europeans. You have an agenda for claiming palestinians are relatively unchanged historic jews which is utterly absurd given that part if the world had always been a melting pot. I have no agenda because I think the Middle East is a human waste dump full of uncivilized cavemen and I'd be happy not to be associated with it, but I am. And there's a wealth of studies that say jews and palestinians are genetic cousins. There's also a wealth of studies that show palestinians have arab ancestry and ssa because Arabs have been raping Africans for 1400 years, deny it all you'd like and ill just keep posting academic sites and studies which prove otherwise while you get your historic info from Hamas.com

Maju said...

Christianity is very clearly a sect of Judaism. The issue may be less clear with Islam (they claim the Bible has been manipulated but they follow Jewish customs more tightly than Christians, I even once met a Jewish convert to Islam who claimed Islam was the truest Judaism - arguable of course) but with Christianity there cannot be any doubt. And very particularly at the origins.

"I would like to know where you got your information that it was a widespread phenomenon".

I have already mentioned above one of my main sources: the 1906 Jewish Encyclopedia (http://www.jewishencyclopedia.com/articles/5169-diaspora). Naturally I have also read and discussed the issues, with Jews even, in may other occasions: it is a well known historical phenomenon and one can't understand the spread of Christianity or Islam without it. And then there is the undeniable issue of many Jewish convert (non-Christian) states: Adiabene in Kurdistan, Yemen before Mohamed, Khazaria before Kievan Rus destroyed it, Berber principalities, Axum before turning Christian, etc.

"There's no indication arabs invaded in any major way?"

Invaded and settled are different things. Britain invaded India but did not settle it, for example.

LouieJ said...

I'm terribly sorry Mr. But anyone who thinks Christianity is a sect of Judaism can't be taken seriously in a debate anyway. Both Islam and Christianity stem from Judaism, I know in your make believe arab Muslims are special world that's not the case, but it actually is as you both use aspects and figures from the Old Testament and I do believe that Jesus is a prophet in your religion, correct?! So that would mean islam could also be a 'sect' of Judaism, equally absurd. Christianity in many ways is completely and utterly the opposite of Judaism and Judaic belief.

LouieJ said...

Again I only a cept academic sources, anyone can write anything on a website. And 'talking to jews' doesn't make them historians on the history of groups of people. Academic sources please!!

Further there is a reason many Berbers and black Africans hate Arabs and that's because they've done more than invade they have forced their culture onto others and there's a reason why 90% of North African genes are Arab Berber. And that reason is that when people invade other people and live with them for centuries they tend to mix. Sorry to break the news to you.

LouieJ said...

Basically your argument is that only people who never left the Middle East are 'real' historic jews because 3,000 years people never left or really intermarried with anyone else. Then I could say only italians who live in Italy have ancient roman ancestry and anyone else are 'fake romans' and people in Italy though invaded thousands of times since the Roman Empire are essentially still the same as ancient romans with very little mix with other people. Do you realize how absurd that is????? By the way southern italians also have arab ancestry because of: you guessed it arab invasion, but I know the Arabs didnt sleep with anyone they invaded and lived with for centuries, right?! If anything you're an arab nationalist who lives in a world where only Arabs or people who never left the Middle East are really native to there. Absurd but so convenient for you!!

LouieJ said...

Academic journal article on the common origins of palestinians and jews along with historical information on the many differing people who have left imprints on the levant I.e. there's lots of different people)
http://www.ucl.ac.uk/tcga/tcgapdf/Nebel-HG-00-IPArabs.pdf

Maju said...

I pass, man. I'm Pantheistic, my "Jewish prophets" are people like Spinoza, Marx or Mandelbrot. You just stick to your beliefs and stubbornness. This is getting ridiculous because you are not taking me seriously. So I choose not to take you seriously either: my time is precious.

Zaid Almasri said...

Why do some people treat druze as a people rather than a religion.

Druze have several origins including Kurds, Anatoliam Arabs

Ezra Jacob said...

I strongly suspect that much of my father's ancestry were English Jews that were expelled in 1290 during the Expulsion. I suppose many stayed and converted. His autosomal testing shows a strong England match, do you think this could be a reflection of this?

~Ezra Jacob

Ezra Jacob said...

BTW, his ancestors went to Friesland.

Maju said...

What you say makes sense to me, Ezra.

Anyhow, for the rest: I just re-read the entry and comments above and what do people think some 18 months later after realizing (per Lazaridis' work on ancient West Asians) that Bronze Age people in Palestine totally overlap with modern Palestinians (and not with modern Jews)? Anyone changing their mind after that?