June 09, 2010

Genome-wide structure of Jews (Behar et al. 2010)

(Last Update: Jun 10)

Another comprehensive new study on Jews (after Atzmon et al. 2010). The paper also has freely available supplementary information online.

On the left, PCA from Supplementary Figure 3, shows clearly at least three different Jewish clusters. Note the main Ashkenazi/Sephardi cluster halfway between Tuscans and Near Eastern populations, a Yemeni Jewish cluster coinciding with Bedouins and Saudi Arabians, and a West Asian cluster encompassing Georgian, Iranian, Iraqi, etc. Jews.

Below is ADMIXTURE analysis in the global context.

There is a ton of information in the above figure, for Jews and non-Jews alike. Some observations:


  • Ethiopians and Ethiopian Jews look identical, between Sub-Saharan Africans and West Asians .
  • Sub-Saharan admixture in Egyptians and Yemenites is quite evident; lack of such admixture in Europe and non-Arabs from West Asia.
  • A little Caucasoid admixture in Mongols
  • Split of Mongoloids into two clusters, which appear to be "northern" and "southern"
  • Central Asian Turkic speakers (Uygur, Uzbek) derived from both Mongoloid sub-clusters; their Caucasoid components are mainly West Asian (light green) rather than north European (dark blue)
  • Non-European components in Russians are resolved into Caucasoid light green and "north Mongoloid" (see above)
  • A little of the "north Mongoloid" component in Turks and some populations from the Caucasus, not much elsewhere in West Eurasia
  • South Asian (green) component in Cambodians
  • Russians and Lithuanians lack south European (light blue) component but have some west Asian (light green)
  • Cypriots are split between West Asia and Southern European components, with minority Semitic (Phoenicians or Syrian Christians?) and northern European ones.
  • French Basque and Sardinians lack West Asian component (light green)

The regional ADMIXTURE analysis is also quite enlightening.


UPDATE I (Jun 10):

What does this study actually tell us about the origins of modern Jews? Are they descended from ancient Jews, and to what degree did they admix with non-Jewish populations either in West Asia or elsewhere?

The smoking gun of an ancestral Jewish gene pool is still missing. Note, for example, the emergence of a "purple" Mozabite cluster in the global ADMIXTURE analysis, or of three distinct Palestinian- Druze- and Bedouin- centered clusters in the regional analysis.

If modern Jews are descended from an ancient Jewish population, we would expect the emergence of such a Jewish-centered component in the ADMIXTURE analysis. Such a component would be centered on Jews but might also spill out to some degree to other populations.

Rather, Jews appear to be variable mixtures of three components (in the regional figure): pink, which is shared by them and Arab speakers; very light blue, which is shared by them and non-Arab West Asians and south Europeans; medium blue, which is centered on southern Europe.

The lack of a Jewish-centered cluster could be either due to a lack of a common core of shared ancestry in various Jewish groups, or to a lack of sufficient resolution in the genetic markers used. There is a common thread among Jewish groups (the pink element), but it is not specific to them.

Nonetheless, we can credit the two new studies with shrinking our universe of viable hypotheses: Ashkenazic Jews don't appear to be either Khazar or converted Slavs/Germans; Iraqi Jews don't appear to have any noticeable Arab-specific ancestry; the Jewish origin of Ethiopian Jews is a fable; Ashkenazic and Sephardic Jews appear to be closely related; and so on.

UPDATE II (Jun 10):

Supplementary Table 4 (pdf) has Y chromosome data for a wide assortment of populations. I find quite interesting the lack of E-M35 chromosomes in Georgian Jews (N=62) and Azerbaijani (N=57) Jews.

UPDATE III (Jun 10):

Notice how different Russians look in the global and regional analyses. In the former, they break down into three components (N/E European dark blue, W/C/S Asian light green, north Mongoloid), while in the latter they appear to have some of the S European light blue.

This should be useful as a cautionary tale to what happens when the full range of parental populations are not present: spurious results can appear.

UPDATE IV (Jun 10):

Consider Figure 2 from the paper:

Jews form three major clusters: one between West Asia and Europe (Ashkenazim and Sephardim); one right in the middle of West Asia (Caucasus Jews and Iranian Jews), and one in the middle of Arabs (Yemenite Jews).

The authors write:
This study further uncovers genetic structure that partitions most Jewish samples into Ashkenazi–north African– Sephardi, Caucasus–Middle Eastern, and Yemenite subclusters (Fig. 2). There are several mutually compatible explanations for the observed pattern: a splintering of Jewish populations in the early Diaspora period, an underappreciated level of contact between members of each of these subclusters, and low levels of admixture with Diaspora host populations.
It is difficult to see how splintering of Jewish populations in the early Diaspora period would result in the observed pattern. In such a scenario, we would expect European Jews to form a separate cluster from Yemenite and Middle Eastern Jews, but we would not expect the differences to be in the direction of the host populations.

It is also not clear why there should be "an underappreciated level of contact" between these subclusters: the fact that they are distinct suggests that there was not much contact, otherwise we would see "intermediate individuals" between the various clusters, which is not the case. Whatever the historical intermarriage across Jewish subgroup boundaries, it must've been low: both the distinctiveness of the three clusters, and the absence of individual variation in ancestral proportions within subgroups suggests that each of the three groups did not admix heavily (recently) with either Jews from elsewhere or non-Jewish host populations.

The evidence as it stands is indicative of relative isolation of three distinct subgroups of Jews in Western Eurasia. What the original makeup of these subgroups was (the Jewish-native mix), prior to isolation, is still up for grabs.

UPDATE V (Jun 10):

It would be tempting to see an Arab-Persian distinction in the neat arraying of West Asian populations in the PCA figure of the previous update with Bedouins and Persian-Caucasian populations occupying the different ends of the spectrum. However, that would be erroneous, I think, as it omits the crucial parameter of Sub-Saharan admixture.

Here is a magnification of the West Eurasian portion of the global PCA (Figure 1 from the paper):


"South" is towards Sub-Saharans and "East" is towards East Asians. Just as the ADMIXTURE analysis suggests, Arabs deviate towards the Sub-Saharan direction. Thus, Persian-Arab distinctions are not necessarily an indication only of differences between these two groups, but also of the presence of variable Sub-Saharan admixture in Arabs.

The global ADMIXTURE indicates why this is the case: Iranians have very little "Semitic" pink and no visible Sub-Saharan admixture, while Arabs have a little Sub-Saharan admixture, which, because of the great genetic distance between Sub-Saharan Africans and West Eurasians, pulls them apart substantially from the Caucasoid cluster.

Indeed, Arabians are intermediate between Caucasoids and East Africans, while the latter are intermediate between Arabians and Sub-Saharan Africans.

Nature doi:10.1038/nature09103

The genome-wide structure of the Jewish people

Doron M. Behar et al.

Contemporary Jews comprise an aggregate of ethno-religious communities whose worldwide members identify with each other through various shared religious, historical and cultural traditions1, 2. Historical evidence suggests common origins in the Middle East, followed by migrations leading to the establishment of communities of Jews in Europe, Africa and Asia, in what is termed the Jewish Diaspora3, 4, 5. This complex demographic history imposes special challenges in attempting to address the genetic structure of the Jewish people6. Although many genetic studies have shed light on Jewish origins and on diseases prevalent among Jewish communities, including studies focusing on uniparentally and biparentally inherited markers7, 8, 9,10, 11, 12, 13, 14, 15, 16, genome-wide patterns of variation across the vast geographic span of Jewish Diaspora communities and their respective neighbours have yet to be addressed. Here we use high-density bead arrays to genotype individuals from 14 Jewish Diaspora communities and compare these patterns of genome-wide diversity with those from 69 Old World non-Jewish populations, of which 25 have not previously been reported. These samples were carefully chosen to provide comprehensive comparisons between Jewish and non-Jewish populations in the Diaspora, as well as with non-Jewish populations from the Middle East and north Africa. Principal component and structure-like analyses identify previously unrecognized genetic substructure within the Middle East. Most Jewish samples form a remarkably tight subcluster that overlies Druze and Cypriot samples but not samples from other Levantine populations or paired Diaspora host populations. In contrast, Ethiopian Jews (Beta Israel) and Indian Jews (Bene Israel and Cochini) cluster with neighbouring autochthonous populations in Ethiopia and western India, respectively, despite a clear paternal link between the Bene Israel and the Levant. These results cast light on the variegated genetic architecture of the Middle East, and trace the origins of most Jewish Diaspora communities to the Levant.

Link

127 comments:

princenuadha said...

this seems to answer my question as to whom is closer to the Italians, the Spanish or the French. As shown, the Spanish are closer to the Tuscans than the French are. I therefore doubt the Italians would be closer to the French than to the Spanish.

Doesn't that result contradict the study in 2008 (europe: genes and geography) which showed the Italians to be slightly closer to the French?

Anonymous said...

Good study. Better than the guff about haplogroups.

Polako has done some of his own admix studies and he had already shown Mongoloids splitting into two groups, the West Asian (Arabian type) admixture in Ethiopians, the presence to the two type of Mongolians, West Asian and South Asian admixture in Hazara and the Uyghur. So it is just confirming what he has done off his own bat. They should have included the two groups of Maasai. Their admix breakup is very interesting.

Pity no Greeks, other than Cypriots.

terryt said...

"Historical evidence suggests common origins in the Middle East, followed by migrations leading to the establishment of communities of Jews in Europe, Africa and Asia, in what is termed the Jewish Diaspora".

But I'm fairly sure that the vast majority of members of that Diaspora moved voluntarily, not forced.

Although the fifth book of the OT was 'miraculaously' discovered during Josiah's reign the Jewish religion preserves elements of much older myths.
I strongly suspect that the religion supported by these ancient myths grew up along the main trading route between the two major population centres, the Nile and Mesopotamian. It probably spread along the Yisre'el Valley and up the Phoenician coast. We know that Phoenician and Hebrew were the same language until two or three thousand years ago. The shared religion and culture meant communication along the trading routes was easier.

In fact I'd go so far as saying that the main elements of the religion spread widely along the gradually expanding trade routes beyond this narrow region. In many places along the routes the religion established itself and the population grew.

Although originally the traders, and the religion itself, had come from the Levant by the time of the Roman Empire its followers had become a diverse bunch. The previous paper indicated that around that time the religious group separated into a Roman network and a Pathan network. The present paper supports an even more diverse origins of the Jewish people.

Maju said...

Superb! This time Behar seems to have got the bull by the horns, as we it's said over here. Bravo!

I just had a first look to the data provided here and cannot yet give a throughout opinion but the clustering with Cypriots in particular of Western Jews (Sephardi, Ashkenazi) is nearly perfect (only exception is the small presence of the North African component among Jews, absent in Cypriots).

There is also more affinity with Turks and Lebanese (Lebanese are a very close match too) than with Palestinians but these are clearly distinct from Bedouins and Peninsular Arabs (including Yemenite Jews in the K-means analysis). In fact Palestinians appear neatly distinct at K=8 in the West Eurasian K-means analysis (while intermediate between Jews and peninsular Arabs in the global graphic).

My question now is: when we talk of Jews (i.e. modern Jews) are we in fact talking of recycled Phoenicians or something like that?

Otherwise, Yemeni Jews look pretty much Yemeni, as we would expect from a community derived from the time when Yemen was a Judaist state (by conversion).

Jews apart, I find noticeable the distinct North African component detected (strongest in Modabites but also important in Moroccans).

You say: "Sub-Saharan admixture in Egyptians and Yemenites is quite evident".

Yes but still very low, specially when compared with the high apportion of mtDNA L, which IMO is of Paleolithic origin in many cases and not a mere product of the slave trade and other historical connections with ultra-Saharan Africa.

"French Basque and Sardinians lack West Asian component".

I'd rather look at the West Eurasian comparison for these sub-regional comparisons matters.

Sardinians appear as almost "pure SW Europeans" (light blue component) but with a small fraction of West Asian blood (pink and maroon components).

Basques appear (again) as very pure Europeans, more West Euro than East Euro but, unlike Sardinians, with both components. The intermediate European (almost 50-50 of each component) is represented well by Orcadians and French.

I decided to call these components West and East (and not North and South, as was my first intent) because the best representative of the dark blue component almost without the light blue one are Chuvash, a Caucasian people. They do have minor West Asian admixture but it is low.

Most interesting. This time I have to bow to Doron Behar's work.

Unknown said...

Unfortunate that Behar et al. did not include Assyrians among the populations sampled. Unless I am missing something, they are very relevant to the topic in question, and have long been neglected in studies of this nature.

Maju said...

@Princeduadha:

The Spaniards in this study are an amalgam of Catalans and Andalusians, who a recent study also showed as somewhat divergent from the main Spanish cluster toward HapMap Tuscans, however Catalans also diverged towards HapMap CEU (NW Europeans from Utah).

Tuscans are not exactly like North Italians, which in the recent Atzmon's paper were extremely close to French (Fst=0.002, the smallest distance in the whole paper except for Turkish and Greek Jews).

For me more interesting is that in this paper French and Spaniards appear as very very close, as I expected based largely on phenotypes, as well as Prehistory, aspects in which Italians, including those from the North, are more divergent. The main difference seems to be in the nature of their West Asian minor components, which is the "Arab" type among Spaniards and the "Caucasian" type among French.

Unknown said...

I'm not as academic as you guys, but this is how I understand these two studies - is this correct?

Modern Jewish communities fall into several distinct and somewhat related groups and are also somewhat related to local non-Jewish populations and middle eastern non-Jewish populations.

The evidence is based on two types of DNA showing slightly different things - autosomal DNA which individuals inherit from both parents and mitochondrial/Y dna.

1.Ashkenazi/Sephardi/Syrian

2.Mizrahi

3.Yemeni

4.Ethiopian

5.India

1. Evidence of Levantine origin and additional enlargement through conversion in the Roman Empire. Subsequent reduction in genepool and later population explosion. Y/Mt DNA shows high percentage of likely levantine origin

2. Evidence of middle eastern origin with intermarriage with local population in caucus/middle east region. Followed by very little marriage outside of community.Y/Mt DNA shows high percentage of likely levantine origin, but hard to differentiate from other local input.

3. A high degree of local intermixing, but with MtDNA possibly consistent with levantine origins. this is consitent with the Yemeni peninsula having experienced widespread Jewish conversion under Dhu Nuwas.

4. Very high degree of mixing with local semitic speaking Ethiopian populations. Little evidence of Levantine origins.

5. High dgree of mixing with local populations but with YDna showing some evidence of Levantine descent. Consistent with community oral history of originating from a very small number of founders.

This is how I understand it anyway....

b.

Zach said...

This highlights a very interesting fact that population replacements happen very sporadically; instead we can call it "population supplements".

In the Old World, already densely populated; we can see the founding genes persisting on. The first agriculturalists may change their language and religion but tend to persist in their land.

It's not as contradictory as it seems, when we hear about the "sack of cities" (Rome, Baghdad come to mind); we forget that until the modern age (and even now that's questionable) cities were not self-sustaining but continually receiving migration from the outlying rural areas.

We aren't the descendants of Athenians, Romans or Imperial Persians but more the rural folk who were their close ethnic kin (the first cousins who stayed behind).

Conquerors very very rarely went to the trouble of destroying their tax base, which until the Industrial revolution were the peasants and trade (which incidentally may be why we had "merchant minorities persist", Arabs, Jews, Armenians)..

The Arabs invaded Iran; urbanised Iran but preserved and enhanced the Qanat networks. That's why surprisingly the Arab-Islamic population influence just drops so unexpectedly (in the 90's Iranian-Persian historiography centred on Arabs invading Iran culturally and genetically).

However the Mongol invasion precipitated the flight of the peasantry who in turn were maintaining the Qanat networks of Greater Khorasan. That's why Uzbekistan (Soghdiana) Turkified and Mongolicized linguistically and genetically. The Mongol may have wiped out the cities but the fear they inspired also led to a loss of the peasantry.

Thus when Jerusalem was sacked (as has been done repeatedly) the local core ethnic population (the fellahs) were not usually the victims, they wouldn't resist.

Zach said...

There is one profound and perplexing mystery; where are the "Palestinian Jews".

The closest we have are Samaritans and historically attested migrants in the 15/16th making an early version of "aliyah".

However this is extremely surprising in light of the fact that all surrounding countries had vibrant and historic Jewish populations (Lebanon, Syria, Egypt).

There was a historic injunction that only Muslims were allowed to live in Arabia but there should have been some continuous Jewish remnant (or at least Arabicized Hebrews) in Palestine; if the Christians and Samaritans were able to persist for so long?

eurologist said...

I decided to call these components West and East (and not North and South, as was my first intent) because the best representative of the dark blue component almost without the light blue one are Chuvash, a Caucasian people.

I would call the light blue northern and eastern Mediterranean, and the dark blue old/ northern European. Northern Europe is not well represented, but without doubt has little light blue and also little light green ("Southern Caucasian").

A similar study excluding regions with significant eastern Asian components would have been useful. I bet the light green would split first (it has an extremely wide east-west range), and then perhaps light blue to distinguish the northern from the eastern Mediterranean.

It is interesting to see that that there is relatively little difference between Egyptians, Palestinians and Jordanians. Also interesting that the peninsular Arab and the Berber components split so clearly at 9 and 10. Probably due to the relatively large number of circum-Mediterranean samples?

Unknown said...

@ZacharyLatif

In "Whose Land? A History of the Peoples of Palestine" by Rev Dr James Parkes - an old but interesting book. He makes the point that the Fellahin stay behind and change religion as you said, this would mean that current populations of Christian and Muslim Palestinians may be descendents of an older Jewish population. Who have merged with incoming Christian and Muslim populations over time.

Last year there were a couple of stories online which you can probably find online which show evidence for "hidden" Jewish-Palestian families. I find these stories are highly politicised though and take them with a grain of salt.

Finally there has been a Jewish presence in Palestine pretty much uninterupted since the Roman expulsion from Jerusalem. It's just that the Diaspora has been a substantially bigger community since even prior to the expulsion and the "native" community has been subsumed by waves of newcomers/returnees.

b

Zach said...

I would just like to clarify beforehand that any of the conclusions reached should not be a reflection on the present conflict raging on and doesn't affect either sides various claim. Its should be more a matter of curiosity and historical inquiry rather than a politicised debate. Its become pretty clear that whatever the conclusion the ancient Hebrews descendants have mixed either in Palestine or in the Diaspora.

We are all mutts except it seems the Europeans, who are anyway set to become more mixed over the next century (interesting to see the genetic testing in the future "testing" the mass immigrations of the 20th century)..

Quick question on the expulsion; there is this "revisionist" history that there was no exile?

Another thought (and this might be a bit more "out there hence my earlier disclaimer") could it be that somehow Judaism spread with the Cohanim/priestly families, who then settled and variously recruited locals. This is a corollary to Razib's thought that Parsi priestly families migrated to India.

Are there any other Jewish lineages apart from the priestly ones (Cohens & the Levites); any "secular" lineages?

Correct me if I'm wrong but are we saying that the Eastern Jews (Mizrahi - Iranian, Iraqi, Yemeni, Caucasian and other Iranic Jews) are mixed with the host populations?

Its becoming clearer that admixture events happened during Pagan/Zoroastrian (pre-Zoroastrians) times and that the Jewish communities crystallized during the advent of mass monotheism. This might make sense in that now we are seeing historically high intermarriage rates in the secular West, which might have mirrored Pagan rome & Achaemanian-Sassanian (?) Persia (after all Esther married the Persian king)..

Dienekes said...

Correct me if I'm wrong but are we saying that the Eastern Jews (Mizrahi - Iranian, Iraqi, Yemeni, Caucasian and other Iranic Jews) are mixed with the host populations?

It's still up for grabs which of the three main Jewish clusters is closer to the pre-Diaspora Jewish population, let alone the pre-Babylonian Exile one.

As I mention in my post, the study doesn't identify a Jewish-specific ancestral component in its ADMIXTURE analysis. This could be either because more markers are needed, or because such a component is truly lacking to a substantial degree, and the three groups are mostly converts.

Maju said...

@Zachary:

"... where are the "Palestinian Jews"".

You mean Ancient Jews from the Biblical, Hellenistic and early Roman times, right?

IMO, the existance of an specific Palestinian component, albeit admixed in most individuals (but "pure" in a large fraction of them) suggests that your "Palestinian Jews" are modern Palestinians who converted to Christianism by grade or force first and then, largely, to Islam.

These would not be essentially different (even if gradually more and more admixed) from the previous Southern Canaanites (or Biblical Canaanites) or from the peoples of the Neolithic cultures that preceded them.

This paper clearly proves (finally!) that Palestinians are not just the same as generical Arabs or Peninsular Arabs, even if somehow related (as one would expect also from ancient Jews and Canaanites, etc.)

"only Muslims were allowed to live in Arabia"

AFAIK, excluding the holy cities of Mecca and Medina, this is not correct. Arabia Peninsula was not always Saudi Arabia. That's a modern development.

I agree with much of the rest of what you say anyhow.

Dienekes said...

IMO, the existance of an specific Palestinian component, albeit admixed in most individuals (but "pure" in a large fraction of them) suggests that your "Palestinian Jews" are modern Palestinians who converted to Christianism by grade or force first and then, largely, to Islam.

These would not be essentially different (even if gradually more and more admixed) from the previous Southern Canaanites (or Biblical Canaanites) or from the peoples of the Neolithic cultures that preceded them.


Clusters don't have dates, and there is no real reason to think that the Palestinian cluster is not fairly recent in origin, and no reason to think that it dates to Neolithic times.

AWood said...

I think there is enough evidence that the Levantine jews comprising Arabian, N. African, Mediterranean, West Asian elements are as close as you can get to the jews of 3000 years ago. The only new element among the Sephardic/Ashkenazi/Moroccan is the N. Euro component. Obviously this was added on later. Whether it was Spaniards, Germans, Italians, Polish, or Russian/Caucasus it's a pretty moot point that you could argue about indefinitely.

Maju said...

@Ben:

As I see it:

Sephardi-Ashkenazi appear most closely related to Cypriots and Lebanese and may well be Greco-Phoenician converts. It's worth recalling that the bulk of traders in the Roman Empire were Greek, Syrian (Phoenician mostly) and Jewish. It is very possible that upon the establishment of a Christian Empire, which absorbed Roman dislike for commerce and related proto-banking activities ("usury"), these traders of the Near East decided more convenient to belong to a tolerated religious minority that did allow and was allowed such activities.

This hypothesis just occurred to me today as I considered the evidence of this paper.

They are Levantine in the broad sense but they don't seem Palestinian at all (nor Druze nor Samaritan nor Negev Bedouin). They are Levantine only in the sense that Lebanese and Cypriots are (and possibly other populations from the Cyprus Gulf area, in Syria or Turkey). Additionally they have some European (or West Anatolian) admixture. That's why asked are they in fact Phoenicians?

Yemeni Jews seem essentially Arab, even the purest of all Arabs maybe (at least in the sense of the pink component). This can well make perfect sense because Yemen was once a Jewish (Judaist religion) state by conversion, much like were Khazars and other less well known realms.

Maju said...

@Eurologist:

"I would call the light blue northern and eastern Mediterranean, and the dark blue old/ northern European. Northern Europe is not well represented, but without doubt has little light blue and also little light green ("Southern Caucasian")".

I was a little confused about Chuvash, which in fact show a lot East Asian component in the global K-means analysis. But the pattern remains: dark blue is highest in NE Europe, light blue in SW Europe. Orcadians are the only representative of NW Europe and a good reference probably for Scots and Norwegians but they are 50-50, just like Romanians for instance (after we exclude the West Asian input).

Anyhow, this analysis does not really get into the depths of internal European structure, which has shown to have more than just two components. In other papers, these populations show a much greater degree of specificity in fact. Maybe we could detect these components at K=15 or whatever but only K=8 seems to low for that in the context of all West Eurasia, where most of the diversity is obviously out of Europe.

We may also be missing specific clusters in other areas.

"It is interesting to see that that there is relatively little difference between Egyptians, Palestinians and Jordanians".

Actually Palestinians show a great degree of specificity at K=8, just like Druzes and Negev Bedouins, which also have distinctive components of their own, and also show various degrees of admixture with other neighboring peoples.

Maju said...

"It's still up for grabs which of the three main Jewish clusters is closer to the pre-Diaspora Jewish population"...

Palestinians very clearly. For me the data of this paper, specially the K-means analysis settles the matter. Modern Jews seem essentially converts from other populations even if often neighboring ones.

Jewish identity is religious, historical and ethnic, no doubt. But it is not an identity that can be tracked (at least for the greatest part) to pre-Diaspora Jews of Palestine. The most direct descendants of these, by far, seem to be Palestinians.

Not unexpected for me at all, really.

Maju said...

@Dienekes:

The Palestinian cluster cluster is clearly distinct and we have absolutely no reason to think that Palestinians (unlike Druzes and maybe Bedouins) have been isolated from their neighbors in an inbreeding dynamic that would produce such cluster.

After the Jewish Wars and specially after Christianization, Palestinians were like any other people: they had maybe a regional identity but were all the time just (Hellenistic, Aramaic) Romans or Arabs. There was no opportunity for this cluster to be formed with such a clear distinctiveness (nor are Palestinians oversampled to cause such an effect either).

Hence the distinctiveness clearly indicates at least a Jewish era origin (and IMO surely much older).

Dienekes said...

The Palestinian cluster cluster is clearly distinct and we have absolutely no reason to think that Palestinians (unlike Druzes and maybe Bedouins) have been isolated from their neighbors in an inbreeding dynamic that would produce such cluster.

Maju, 90% of the time you contradict yourself. On one hand you are saying that the Palestinian cluster is Neolithic in origin, on the other you see no isolation from their neighbors. How is it possible for a population to interbreed with its neighbors for a few thousand years and still develop genetic distinctiveness?

There are Sardinian villages that form distinctive clusters, using your "logic" have they been isolated from each other since the Neolithic?

The Druze form their own cluster, and these are Druze from Israel. So, how come you don't think that it's the Druze who represent native inhabitants, and it's the Palestinians?

There is a very good reason why Palestinians, Druze, and Bedouin form distinct clusters and that is consanguinity, which is prevalent in the Near East. All these clusters are probably of very recent origin (after the advent of Islam), and anyone who thinks they are of "Neolithic" origin, should make an argument for it, e.g., by explaining their paucity in non-Arab populations.

Andrew Oh-Willeke said...

@Dienekes

1. In Update IV, I think you are misreading the phrase "an underappreciated level of contact between members of each of these subclusters."

What the authors are saying, as I read this phase, is that members of subgroups are admixing more with each other than they are with members of other subgroups.

You say: "It is also not clear why there should be 'an underappreciated level of contact' between these subclusters", but what they say is "an underappreciated level of contact between members of each of these subclusters."

The sense of their statement is that it would be even more clear if the word ",respectively.", where added to the end of the sentence.

In other words, Ashkenazi and Sephardi Jews have had more sustained post-diaporan admixture than is commonly realized. Similarly, Caucasian and Middle Eastern Jews have had more admixture with each other than is commonly realized. The subgroups thus represent groups which had continued ongoing ties with each other.

2. "If modern Jews are descended from an ancient Jewish population, we would expect the emergence of such a Jewish-centered component in the ADMIXTURE analysis. Such a component would be centered on Jews but might also spill out to some degree to other populations.

Rather, Jews appear to be variable mixtures of three components (in the regional figure): pink, which is shared by them and Arab speakers; very light blue, which is shared by them and non-Arab West Asians and south Europeans; medium blue, which is centered on southern Europe.

The lack of a Jewish-centered cluster could be either due to a lack of a common core of shared ancestry in various Jewish groups, or to a lack of sufficient resolution in the genetic markers used. There is a common thread among Jewish groups (the pink element), but it is not specific to them."

This is an odd way to read the data. Jewish ethnogenesis has a much younger time depth (around 3500 years ago) than the time depth at which Arab and Southern European populations would have begun to diverge genetically (ca. 8000 years ago at least, if in the Neolithic, and probably as far back as 20,000 years ago if Epipaleolithic).

Given that time depth, what is so surprising about the possibility that the ancient Jewish population was already an admixed people of a predominantly Arab character.

Linguistics and Jewish tradition agree that not all members of the nomadic proto-Semitic community from which the Jews arose became Jews, and that Arabs are the most closely linked fork. Jewish traditions of migrations across the Levant and Egypt, and also period of refuge in Mesopotamia for substantial parts of the population (as well as links to Mesoptamian mythology in Genesis) all point to a continentally admixed formative Jewish people as the expectation.

I'm not aware of anyone (not even Young Earth Creationists) who claim that the ancesteral Jewish people can be traced back to the very dawn of mankind as an original division of human peoples. The Hebrew Bible specifically describes instances of Jewish tribes conquering and assimilating parts of non-Jewish tribes. Instead, the claim generally made is that the Jewish people have been largely ethnically distinct and endogamous in the period after the end of the formation of Jewish states in the Levant.

Also, nothing in the data is inconsistent with a predominantly shares origin of the Arab population and the Jews (indeed everything points to this) that is more diluted in some populations than others, mostly likely at bottleneck points where pressure to allow admixture is greatest and the impact on the subsequent population is most complete. This is consistent with anthropological data that make clear that intermarriage is more accepted in Yemeni Jews than in many Jewish populations, for example.

Andrew Oh-Willeke said...

This data is also as good a point as any to make a point on the concept of race.

The admixture data do not make a very convincing argument for its existence in anything approaching a conventional sense.

There are fine gradations of intermediate populations between every point on the PC axes, and if one were to choose a finer color palatte than the one used, breaking each of the colors into subcolors, it would be even finer.

The clusters show considerable overlap with each other in Figure 2 and Ashkenazi Jew and Caucusian populations conveniently bridge gaps of what would otherwise appear to be isolated populations.

One can say, "Arabians are intermediate between Caucasoids and East Africans, while the latter are intermediate between Arabians and Sub-Saharan Africans."

But, it is probably more accurate to say that there is a more or less continuous cline of genetic variation from Sub-Saharan Africa to Europe associated mostly with geography.

The notion of distinct races suggests that absence of middle ground, but the reality is that the middle ground is heavily populated, and that clusters that are stable enough to be compact and distinct are also much smaller than an traditional conception of race. Conceptions of race that generate compact, distinct clusters (that would consider e.g. "Syrian" to be a race) end up with far more categories (hundreds or thousands) than any conventional use of the term could bear.

Maju said...

@Dienekes:

"Maju, 90% of the time you contradict yourself. On one hand you are saying that the Palestinian cluster is Neolithic in origin, on the other you see no isolation from their neighbors".

No, no.

What I am saying is that there is no knowledge that Palestinians formed an isolated highly inbred community like Druzes. They were not known to have any specificity but the genetic data does show that specificity, so it must be something formed before the Roman era, when there was a very specific, distinct community living there: ancient Jews.

Alternatively (or additionally) one can well argue for an older specificity from the Natufian era, very distinct from the Taurus-Zagros Neolithic area.

Dienekes said...

Andrew, the ancient Jews had a clearly very distinctive religious identity compared to their neighbors, and they also had a degree of political autonomy. So, it's a good guess on my part that they may have developed their own distinctive genetic identity.

The current study does not allow us to determine what the genetic identity of the intersection of the three Jewish subclusters is. Take Mozabites, for example, who appear admixed up to K=9 and distinct at K=10, or Yakuts who appear as mixed Caucasoid-Mongoloid up to K=6 and distinct for higher K.

A population that is admixed may still develop its own distinctive identity (through a blending of the components). This is not apparent in the case of Jews, and this can be explained by either different origins of the three Jewish subgroups (no common origin), or by a lack of resolution in the markers used.

Maju said...

Not sure if all your doubts are clarified in the previous very brief comment, so I'll add some extra clarifications, Dienekes:

"There are Sardinian villages that form distinctive clusters".

Within Sardinians only. Every distinct population should at some depth level form their own cluster. Only if there was lots of continuous interaction between these villages these clusters would not show up.

In a wider comparison all those Sardinians should cluster (as they do in this paper for example) in contrast to other populations, rather than in contrast to each other. Naturally.

Let's keep a sense of proportion.

"The Druze form their own cluster, and these are Druze from Israel. So, how come you don't think that it's the Druze who represent native inhabitants, and not the Palestinians".

The Druze are known to be highly inbred. It's logical they form their own cluster. That is not the case of Palestinians, so we have to look for another explanation, yes or yes?

Whatever the case the Druze component is also absent from Jews, as is the Negev Bedouin one (also from Palestine but also a small isolated community). So modern Jews are still not connected in any specific way to any Palestinian population.

"There is a very good reason why Palestinians, Druze, and Bedouin form distinct clusters and that is consanguinity, which is prevalent in the Near East".

Then why we don't see an specific Saudi cluster or an specific Jordanian cluster or an specific Lebanese cluster or even an specific Jewish cluster. We know that Jews have been more inbred for some 2000 years or more for reason of their ethno-religious distinctiveness but they don't show up as such at all.

We are not talking of the rare small and endogamous community of the kind of Druzes when we talk of Palestinians, actually the opposite: a quite cosmopolitan community, partly because of the religious importance of their homeland and partly for their mainstream religious and ethnolinguistic adscription. I know of no reason to think of Palestinians as some sort of Druze-bis and you will have to prove such claim with at least some conclusive data before I can swallow that.

Dienekes said...

"The Druze form their own cluster, and these are Druze from Israel. So, how come you don't think that it's the Druze who represent native inhabitants, and not the Palestinians".

The Druze are known to be highly inbred. It's logical they form their own cluster. That is not the case of Palestinians, so we have to look for another explanation, yes or yes?


Populations form their own clusters if they are inbred. If they outbreed regularly, then they don't form their own clusters.

Your "other explanation" is ridiculous, as it postulates that Palestinian genetic distinctiveness is of Natufian origin. In short you are replacing a mouse with an elephant, by postulating that a genetic identity has persisted in Palestinians and their ancestors for 10 thousand years.

actually the opposite: a quite cosmopolitan community

On one hand you say that Palestinians are descended from Natufians, on the other hand that they are cosmopolitan.

I leave it to readers of the blog to figure out how a population can be outbred for 10 thousand years and maintain the genetic distinctiveness of the people who lived in that area in the Paleolithic.

Andrew Oh-Willeke said...

"Andrew, the ancient Jews had a clearly very distinctive religious identity compared to their neighbors, and they also had a degree of political autonomy."

The point is that the Jewish identity is known to be younger than the dispersal of sub-continental populations from each other. At a specific time period in history a group of people who previously were just one more group of West Asian-Egyptian nomads becamse "Jews." There isn't anything particularly odd about that group of people being an already admixed group, and yet still being ancestral.

"A population that is admixed may still develop its own distinctive identity (through a blending of the components). This is not apparent in the case of Jews, and this can be explained by either different origins of the three Jewish subgroups (no common origin), or by a lack of resolution in the markers used."

Wouldn't it be more parsimonious to assume that the offshoots of a common ancestor later diverged while isolated?

Unlike so much of population genetics, the Jewish people as such have existed entirely in the historic era. We know far more about the historical events that brought Yemeni Jewish or Caucasian Jewish populations into being than we do about other groups. Instances of mass conversion to Judaism in history are few and far between.

If Jewish people don't share a common origin, one has to come up with a more plausible alternative as to how a group of people suddenly ended up Jewish. It may be possible to have convergent evolution produce similar skin colors or statures or builds. But, two populations can not both become Jewish by convergent evolution.

Zach said...

I'm finally beginning to understand K-analysis but not entirely; getting there piece by piece!

Maju I also independently concluded (assumed) that the Palestinian cluster is post-Islamic. A very simple reason, the Jewish populations should show some trace of this "Hebraic" genome?

There are three separate Jewish clusters; it is hard to believe that they have admixed to such an extent that none of them would have preserved the Pale-Hebrew cluster.

Also another point I thought up independently but Dienekes articulated it well. "Endogamous communities"; the Palestinians, Lebanese and Syrians are not endogamous community however their sub-clans and confessional identities are.

As I've been repeating myself unless you take confessional data the term "Palestinian" & "Lebanese" is simply meaningless.

We know from historical context that Christian Palestinian, Lebanese and Syrians do NOT descend from Arab-Islamic Bedouin; they would be ideal controls for any mixture analysis.

Andrew Oh-Willeke said...

FWIW re the Druze.

There is good reason from the genetic evidence (e.g. of mtDNA types) to believe that the Druze were not drawn from a single coherent ethnicity before the formative period. Druze tradition claims diverse ethnic origins for the Druze as a whole (but not necessarily for particular local communities), and the mtDNA bears this out. For example, there is a Druze community in Israel that is the only one in the world to have X1, X2 and X* mtDNA haplotypes, all in significant amounts. But, this is not found in many other Druze communities.

The Israeli Druze community that has the X haplotype variation, according to their traditions originates in mountains beyond Mesopotamia. But, another Druze community could easily have come from the South Levant.

So, the label "Druze" to identify a population in a study like this one is a false friend. It implies the existence of a cluster much larger than the sample should reasonably be understood to include.

Maju said...

"Your "other explanation" is ridiculous, as it postulates that Palestinian genetic distinctiveness is of Natufian origin".

You surely know by now that I do tend to favor ancient local genetic origins over replacement (depending on cases but certainly when there's no evidence of replacement), so this should not surprise you.

But I'm also conceding to the possibility that it could be more recent (i.e. Jewish Biblical period or some other period in between). What I do not see is any evidence in this genetic data for replacement (actually the opposite rings true) nor any historical data for any sort of particular isolation of Palestinians since Roman era that could justify a recent formation of such clearly distinctive cluster.

This is the crux of the matter and not necessarily Natufian.

"On one hand you say that Palestinians are descended from Natufians, on the other hand that they are cosmopolitan".

I don't see why these two things are incompatible. I also think that classical Greeks were descendants from Neolithic Greeks (at least essentially so) and they were a very cosmopolitan people.

Cosmopolitanism opens the gates for gradual inflow (and outflow) of genes in significant figures but does not mean that the arrival of a few immigrants automatically changes the whole composition of the population. It just means that they are not highly inbred like Druzes.

There is some of that Palestinian component in other populations and there is even more of foreign components in Palestinians. Still they keep a quite clear distinctiveness, what means that their origins pre-date such admixture processes.

Also their specific component is clearly distinct from what is found among Jews and is not found in any other population at such high frequencies in spite of the wide extensive sampling of all the region. So, if Palestinians are not particularly inbred (fact) and their genetic backbone does not come from anywhere else, it is obvious that they are not immigrants, that they have been there for a very long time (ancient Jews, Canaanites, Neolithic or whatever you find more reasonable).

And that acknowledged, it is also evidence that Jews do not appear to originate from that mainstream Palestinian population with deep local roots but from further North.

Onur Dincer said...

In addition to Lezgins, they should also have studied Azeris as a host population for Azerbaijani Jews. I also wonder the ethnic composition and locations of Iranian samples included in this study (are there Iranian Azeris or any other Turkic speaking populations of Iran among them, for instance?), is there any information about these issues in the full article or in the supplementary materials?

Turkmens too would be interesting to study, I have never seen any autosomal study that includes them. Actually I also have never seen any autosomal study that includes Azeris. I would appreciate any autosomal study that includes at least one of these two ethnicities, any info would be welcome in this regard.

Marnie said...

Anybody else notice that the Sephardim form a tight locus within the Ashkanazim cluster?

eurologist said...

Maju,

Can you explain to me where you see that distinct Palestinian cluster? I just see a continuum.

As I mentioned before, from K=8 to 10, they are very, very similar to Egyptians and Jordanians. In the principal component plot, they are wedged between Saudis and Jordanians, as expected. Like almost any Levantine group, they just seem to be an approximately equal mixture of "original Arabs" (pink: Bedouin/Saudi), northern/eastern Mediterranean (light blue), and southern Caucasian (light green) - with just a dash of Berber (violet) and African (red). Egyptians, as expected, have a bit more African.

I would agree that original Jews would have had a similar admixture, but also think that the ones migrating to Europe would have excepted to have a more northern background (Lebanon, NW Syria). Which means that after ~2000 years ago they would have started already closer to where they are located now in the PC plot.

As to Iranians, historically there is not much Arab to them, as also evident here. So, I am not surprised that they are roughly between Turkish (but without the northern European portion) and their eastern neighbors (southern Indian admixture via Afghanistan and Pakistan and direct exchange). And their Mediterranean component is smaller than their western neighbors, while the southern Arab admixture causes a tiny bit of African influence.

Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks. While that agrees with my lay physiological assessment of the ones I know personally, they are located on the eastern end, geographically. So what is the reason, if any (artifact of first two PCs)?

From paleolithic to modern,

- they are the left-over population from all those migrating into Europe;

- they are derived from the same LGM refuge as much of Eastern Europe, which consequently diffused into much of Europe;

- they are largely post-agricultural newcomers from the Danube, Balkans, and Ukraine.

Or a combination of the above.

But the funny thing is, I can't see their central European affinity at any K level. Europeans have way less or their dominant "southern Caucasian" (light green), and Armenians/Georgians lack the old/northern European (dark blue) almost completely.

Dienekes said...

You surely know by now that I do tend to favor ancient local genetic origins over replacement (depending on cases but certainly when there's no evidence of replacement), so this should not surprise you.

I don't care what you "favor", the current study has absolutely no bearing on whether or not Palestinians have "local genetic origins". Repeating the claim without making the actual argument does not cut it.

What I do not see is any evidence in this genetic data for replacement (actually the opposite rings true) nor any historical data for any sort of particular isolation of Palestinians since Roman era that could justify a recent formation of such clearly distinctive cluster.

You just don't get it. If Palestinians were not isolated since the Roman era and were "cosmopolitan" and intermarried with their neighbors, then you would NOT see such a cluster, whether it was formed in 3,000BC or 10,000BC.

Your logic is flawed: you claim that Palestinians could not have formed a distinctive cluster in the last two thousand years because they intermarried but then you add another eight thousand years of gene flow and claim that they did form that cluster in that longer time span.

It's like seeing a sand castle on the beach, and saying that "it can't be that tall, the waves should have been eroded it since yesterday". Not only are you denying the existence of your senses ("Palestinians do form a cluster"), but you are "solving" the problem of its existence by saying that "it should have been eroded since last week".

eurologist said...

I know this is a bit out of topic, but may shed some light on the general context. As to the analysis up to K=10, I have the following suggestions and comments:

- light green: Northern India/Pakistan, 40,000-12,000 years ago, separated from the south due to drought and different sustenance. Free exchange with Afghanistan, northern Iran, Caucasus, and northern Turkey

- light blue: eastern Mediterranean during and just after LGM - from the southern Balkans (Greece) and the Bosporus along the Levantine coast


Both entered Europe in neolithic times: light blue via Mediterranean agriculture, then with the Phoenicians, then the Greeks, and then the Romans all along the northern Mediterranean shores; light green via the Danubian/northern agricultural expansion.

From that perspective, typically about 15% of Europe is from the "continental, northern route" agriculturalists (but less so in the south), and about 0 - 50% is from southern agriculturalists and later merchants - with a huge gradient to basically zero to the north.

Maju said...

@Zachary:

"Maju I also independently concluded (assumed) that the Palestinian cluster is post-Islamic. A very simple reason, the Jewish populations should show some trace of this "Hebraic" genome?"

Only if they have meaningful ancestry in ancient Palestine and not if, as I think, they are from a totally different background (converts).

So I don't see this as evidence of what you say but the opposite: evidence that modern and Medieval Jews have a different origin. And as it seems to be the case from this paper and also hinted (less clearly) by the recent one of Atzmon, modern Western Jews seem to have originated in the Cyprus-Lebanon area.

I could go on about Y-DNA J1/J2 and all that, but we have discussed that recently in regard to Atzmon 2010 and I do not wish to be repetitive.

Also I just found a little piece of history work by H.G. Wells (A Short History of the World, chapter XXII), hinting at late Roman and Medieval Jewish communities being most likely nothing but convert Phoenicians. It's not evidence but it clearly indicates that at least some had suspected this reality in the past (not me: I was not really aware of the Phoenician possibility until this week and rather favored an Hellenistic Anatolian one, which is maybe similar but not exactly the same thing).

"There are three separate Jewish clusters; it is hard to believe that they have admixed to such an extent that none of them would have preserved the Pale-Hebrew cluster".

Judaism was as proselytist as Christianism, which is nothing but a particularly successful sect of the wider Hebrew religion. While Jesus, Peter, Matthew and such were clearly Palestinian Jews, nobody would dare to claim today that Christians are the genetic descendants of those. And that would also be true in the 4th century, when Christians took over the Empire and suppressed all other religions one after the other, allowing only Rabbinic Judaism to remain.

Similarly it is very probable (and genetics now seem to support that scenario) that the Rabbinic Jews of the 4th century were not either anything but converts. Only since those dates (and specially the 8th century when Islam mimicked Christian religious intolerance also outside the Roman Empire) Judaism became a strictly hereditary sect, deprived of any realistic chance of expansion. But between the beginnings of Yahvistic proselytism (in earliest ancient history but specially in the Hellenistic period) and that closure, many many centuries had passed and many many peoples had been converted.

"As I've been repeating myself unless you take confessional data the term "Palestinian" & "Lebanese" is simply meaningless".

Maybe but as 80-90% of Palestinians are today Muslims and there has been a gradual but steady conversion process from Christianism (and Judaism too), Muslim Palestinians should be quite representative of the real thing.

Lebanese do not seem that different across sects anyhow. Somewhat yes but they all seem to share (except possibly the very special Druzes) a common main local ancestry too, at least that would be the logical conclusion on the latest Y-DNA research (Zalloua, I believe).

I would not emphasize that as you do, because a lot of peoples have such semi-endogamous subdivisions and still they normally appear quite homogeneous in comparison to others because they are neither strictly sub-endogamous (unlike Druze, which are extreme) nor such sub-endogamy can probably be tracked along the millennia either.

"We know from historical context that Christian Palestinian, Lebanese and Syrians do NOT descend from Arab-Islamic Bedouin; they would be ideal controls for any mixture analysis".

Sure. That's a good point. It'd be interesting to find out.

Maju said...

@Marnie:

Beware of PC analysis, which can only depict two dimensions (a simplification). It may mean something or not at all. That's why K-means analysis is such an important alternative way of analysis for autosomal genetics and that's precisely one of the great values of this key paper: that they do not shy away from using it for a very extense array of populations, producing some very interesting results.


@Andrew:

"Wouldn't it be more parsimonious to assume that the offshoots of a common ancestor later diverged while isolated?"

That needs evidence rather than just an appeal to parsimony. Considering that Fst between Western and Iranian Jews (Atzmon 2010) is as high as between Basques/Russians and the same Western Jews, and considering that these rather unrelated European controls are the most distant of all sampled Europeans, we can safely conclude that Iranian and Western Jews are totally different populations within the context of West Eurasia.

The same applies to a less extreme extent to Iraqi Jews in relation to West Jews and surely (no Fst data that I know) to other peripheral Jewish populations, which in all cases are much closer in the K-means analysis to their non-Jewish neighbors than to other Jews.

So we are talking of several very different Jewish populations, spawned from some localized non-Jewish gene pools, regardless that they may have some minor connection between each other and with Palestinians.


@Dienekes:

"the current study has absolutely no bearing on whether or not Palestinians have "local genetic origins"".

The current study is absolutely demonstrative that Palestinians have a strong core of local genetic origins.

If you think otherwise, please point me to that mysterious outsider population from where they derived and which is nowhere to be seen. No other population has any meaningful signature of the Palestinian-specific component (maroon coded), except to a limited extent some Negev Bedouins.

"If Palestinians were not isolated since the Roman era and were "cosmopolitan" and intermarried with their neighbors"...

Why not. If one out of 100 outmarries (a perfectly reasonable assumption in conditions of Medieval comsmopolitanism) it would still bear true. It probably would also bear true even if one out of 10 did, though I find this figure way too high even for a hyper-metropolitan area of the kind of Medieval Constantinople, Baghdad or Venice. It may be high even for today's New York, really.

The USA is a cosmopolitan modern nation, in a hyper-globalized reality. Yet most US citizens marry with others of their kind, even often within ethnic and geographic subdivisions among them.

Cosmopolitanism, relative openness to others, does not automatically mean total dilution into something else. It just means that you are not like Druzes or the most encroached Indian jatis.

I don't get your sand castle metaphore and I think it is because you don't really seem to understand what I'm saying.

Let's see: the castle is there (and I'm the one calling your attention to it) and that means that the waves have not really touched it much, what means that the waves have not been an important factor, what means that Palestinians were not so strongly affected by the Roman and Arabo-Muslim waves.

It's not the castle that is not there: it's the waves which are too weak to reach it mostly.

Dienekes said...

That's why K-means analysis is such an important alternative way of analysis for autosomal genetics and that's precisely one of the great values of this key paper

Whatever this paper does, "K-means" it does not.

I don't get your sand castle metaphore and I think it is because you don't really seem to understand what I'm saying.

I understand perfectly what you are saying and it's nonsense.

If you think otherwise, please point me to that mysterious outsider population from where they derived and which is nowhere to be seen.

It's not up to me to point you to anything, it's up to you to explain how the results of this paper have any bearing whatsoever on the question of whether Palestinians are descended from local populations.

But, this time you have (unfortunately) been caught with your pants down. The Druze sample forms their own cluster and live in Israel. The Palestinians form their own cluster and live in Israel. Now, please explain ON THE BASIS OF GENETIC DATA why you claim that the former are of "Anatolian" origin while the latter are descended from Natufians. How can you determine -as you did- ON THE BASIS OF GENETIC DATA that the Druze are Anatolian immigrants while the Palestinians have been in Palestine since the Paleolithic.

HARD GENETIC DATA only, no opinions or comments about who serves in the Israeli army or not.

Maju said...

@Eurologist:

"Can you explain to me where you see that distinct Palestinian cluster? I just see a continuum".

That's because you are looking at the wrong K-means graph, the global one. Look at the one that includes only West Eurasian populations (incl. North Africans). There is where I'm drawing the data to support my conclusions.

If Palestinians show up as distinct at K=8 (there's nothing deeper in this graph), while other known distinctive populations such as, say, Sardinians or Basques, do not, that is because they are pretty much unique in their genetic background, even within the context of West Eurasia (admittedly the relatively large sample of Palestinians probably helped too).

"...in the PC plot".

I'm not paying too much attention to the PC graph either. The WEA K-means is the really revealing data here. PC analysis can only be roughly equivalent to K=4 depth.

As for the rest, I'd suggest you first look at the Eurasian-only PC graph and revise your conclusions accordingly.

Onur Dincer said...

It would be interesting if they investigated also Greeks (separately investigating Greeks of Greece mainland, of Aegean islands and of Anatolia) and Turkish Cypriots. Also it would be better if they investigated also Turks based on geography of origin (e.g., Anatolian Turks, Balkan Turks; the more the number of sub-groups, the more the details).

Dienekes said...

This paper does not use k-means, so please stop referring to it as such.

Onur Dincer said...

It is noteworthy that Karaite Jews are again (as in Atzmon et al.'s study) missing from this study.

Maju said...

Then what is it, according to you?

Onur Dincer said...

Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks.

It is because of the small but noticeable Mongoloid components (mostly north Mongoloid) among Turks. Because that Caucasoids and Mongoloids are genetically very distant from each other, a small Mongoloid admixture can pull an otherwise Caucasoid population apart from other Caucasoids in global PC plots. This is just one of the many reason why we shouldn't take PC plots so seriously, as they are just overly simplistic and often wrong 2-dimensional representations of genetic distances between populations. If we exclude the Mongoloid components of Turks, Turks appear very similar to other investigated Anatolian and Transcaucasian populations, namely Armenians and Georgians, but with a much more clear north European component. This brings to my mind Greeks, who would also probably have shown more north European component than Armenains and Georgians, if they had been investigated, thus they would probably have appeared genetically closer to Turks than Armeanians and Georgians are. Indeed, Greeks (especially Anatolian Greeks) may have appeared to be the genetically closest population to Turks, if they had been investigated. Azeris and Turkmens are the other mysteries for me.

Katharós said...

Genetics and the Holy “Rassenhygiene”land…

Anyway Herodotus (5th century BC.) mentions 3 groups of people living in the southern Levant, Syrians , Palestinian Syrians “referring to Aramaic speakers”” and Arabs living side by side.

Herodotus Book 3: Thaleia

5. Now by this way only is there a known entrance to Egypt: for from Phenicia to the borders of the city of Cadytis “Gaza” belongs to the Syrians who are called of Palestine, and from Cadytis , which is a city I suppose not much less than Sardis, from this city the trading stations on the sea- coast as far as the city of Ienysos “Khan Yunis, near Gaza” belong to the king of Arabia, and then from Ienysos again the country belongs to the Syrians as far as the Serbonian lake, along the side of which Mount Casion extends towards the Sea. After that, from the Serbonian lake, in which the story goes that Typhon is concealed, from this point onwards the land is Egypt. Now the region which lies between the city of Ienysos on the one hand and Mount Casion and the Serbonian lake on the other, which is of no small extent but as much as a three days' journey, is grievously destitute of water.

Michael said...

A couple of clarifications:

Judaism during the second temple era was not an exclusive sect; multiple sects existed, with different rules governing each sect. The temple was the unifying source for communal control. We do not know fully how conversion or initiation into the community occurred during this time period, nor do we know what percentage of the population were converts.

Arabs had been settling north of the Arabian peninsula since before the 6th century BCE; the Nabataean civilization (Petra, Jordan) was Arab.

Second, see Baruch Kimmerling's book, The Palestinian People for a better overview of the history of Palestine. Though he does not go into ancient history, he does cover the vacuum created in the 6th and 7th century CE when Byzantium and Persia were at war. Two different Arab clans vied for control over Palestine during this time period, and as the conditions got worse, they opened it up for later Arab invaders (The Muslims) to bring peace.

So, the evidence here actually suggests that Western Jews are probably descended from Israelites; the Palestinians are descended from Arabs who also settled in the land.

Finally, with regard to Hellenization, Jews were expelled from Rome for worshiping Zeus Sabazios c138 BCE. See also the Orphic Hymns as translated by Apostolos Athanassakis. One of the hymns to Dionysos refers to the god as Yhyneus and Erikipaios. Neither of these names are Greek. Yhyn appears to be a play on the Hebrew name for God, Yah, hidden in the word for wine (Yayin). Erikapaios might be associated with a name for one kabbalistic aspect of the deity (Arikh Anpin). (Yes, there were Orphic Jewish texts...)

aargiedude said...

Dienekes, where's the European hunter-gatherer cluster? African hunter-gatherer pygmies have a clear genetic cluster. According to you, Europe was inhabited by hunter-gatherers who were almost completely mtdna U4 and U5 (and perhaps U* and U2), while the incoming agriculturalists had virtually or absolutely none of this. Such a huge separation is only seen between the 5 main genetic clusters of the world, such as blacks, Caucasians, East Asians, etc., and these major genetic clusters always appear immediately in these sorts of genetic analyses. The European hunter-gatherers should have formed their own 6th major worldwide cluster. To drive the point further, you've mentioned the ancient mtdna U2 finding in southern Russia as evidence of a long term existence of the mtdna U clade in northern Eurasia. And it's definitely true that U5 has a very high diversity relative to other mtdna haplogroups (within U or not). These people, after tens of thousands of years, should have formed their own genetic cluster. It should definitely appear in an analysis such as this. Where is it?

Maju, I don't think I quite agree with you, though your observation about the Palestinian cluster is totally valid and I do agree with that and it does make me wonder. But if you look at the PC graph in which the populations have been circled and labeled, you can see how Sephardic/Ashkenazi Jews seem to take off from the Levant zone (Palestinians, Syrians, Jordanians) and move in a straight line towards Europe, consistent with a Levantine origin and becoming mixed with Europeans. Imagine if their origin had been primarily Anatolian: their position should have been somewhere between Turkish/Caucasus samples and Europe. Cyprus is an exception, but well, it's just Cyprus, all those Anatolian/Caucasus samples paint the big picture. I also think this is what's going on with the other 2 Jewish groups in this chart. Yemenite Jews seem to be halfway between the Levant zone and where we would reasonably expect Yemenites to exist had they been included, which would be near the lower left section of the Saudi circle. And the Azerbaijani, Georgian, and Iranian Jews also seem to be halfway between the Levant and the Caucasus/Iran. Believe me, I was also thinking Ashkenazi origins would show a big amount of Anatolian ancestry, because of several details of their y-dna make-up, but after going through this chart, and also the worldwide cluster chart, I have to concede that it really does look like Jews (Ashkenazi or not, except the Ethiopian Jews which we already knew they weren't really of Jewish ancestry long before this study came out) are primarily of Levantine ancestry. Another thing that I didn't expect is that it also seems their "host nation" ancestry is notably higher than I'd thought. I would've originally put the non-Levant ancestry of Ashkenazis, Sephardics, and North African Jews at 10% to 15%, but now it's looking more like 1/3 to even 1/2.

Onur Dincer said...

This paper makes the same mistake with the Atzmon paper: omitting Kurds and Zazas. These two populations, especially Kurds, were used in many early Jewish studies and gave clues about their genetic closeness to Jews (in addition to Armenians, Turks, Greeks and some other populations from the Middle East and Europe, which all also appeared genetically close to Jews).

Onur Dincer said...

I also wonder from where and whom Turkish samples were collected.

Also Adygei (Circassian), Lezgin and Russian samples.

Dienekes said...

Dienekes, where's the European hunter-gatherer cluster? African hunter-gatherer pygmies have a clear genetic cluster. According to you, Europe was inhabited by hunter-gatherers who were almost completely mtdna U4 and U5 (and perhaps U* and U2), while the incoming agriculturalists had virtually or absolutely none of this. Such a huge separation is only seen between the 5 main genetic clusters of the world, such as blacks, Caucasians, East Asians, etc., and these major genetic clusters always appear immediately in these sorts of genetic analyses. The European hunter-gatherers should have formed their own 6th major worldwide cluster.

European "hunter-gatherers" do not exist as a population any more; their genes are spread across the European continent.

Clustering algorithms like ADMIXTURE exploit differences between individuals, that is why, e.g., they can uncover a "Pygmy" cluster: there is a bunch of individuals that are differentiated from other Sub-Saharan Africans, and these are the Pygmies.

This is not the case in Europe, where admixture with hunter-gatherers took place thousands of years ago, and no living populations of hunter-gatherers exist today.

Consider a picture drawn with blue and orange pixels. A mechanized procedure will identify two colors, even though "orange" could have been formed by an admixture of various shades of yellow and red in different proportions. However, the existence of such constituents of orange cannot be inferred if there are no yellow and red shades in the picture.

Even if the existence of a latent "European hunter-gatherer" cluster could be inferred, we would not be able to identify it as such, as we don't have a "European hunter-gatherer" yardstick to compare it against.

Dienekes said...

The potential for detecting Anatolian ancestry in modern Jews is limited by the fact that Anatolia is inhabited today by Turks, who are shifted genetically because of their small Mongoloid admixture, and also by the fact that Anatolia is not genetically uniform.

However, it is clear to me that Jews (at least European Jews) are not simply of Anatolian origin, as they have a high J1/J2 ratio compared to Anatolians, as well as an excess of E-M35.

The excess of E-M35 is visible in the "purple" element lacking in Turks, and the J1/J2 excess is visible in the excess of the "pink" element.

Maju said...

"Maju, I don't think I quite agree with you, though your observation about the Palestinian cluster is totally valid and I do agree with that and it does make me wonder".

Enough for me. That's all I meant to point out, as for me it is the smoking gun pointing to the definitive resolution of this debate.

"But if you look at the PC graph in which the populations have been circled and labeled"...

I looked at it and could not see any differences with the basic PC graph plot (supp. material fig 2), posted also as first image by Dienekes. Just that it has been annotated to show where ethnic groups show up.

"...you can see how Sephardic/Ashkenazi Jews seem to take off from the Levant zone"...

I don't see that as clearly as you do. What I see is that they cluster between West Asians and Europeans, partly overlapping with Cypriots (this fact also apparent at Atzmon's supplemental material) and Romanians (some Ashkenazi only).

Anyhow, as I said before, PC analysis is necessarily limited so the K-means analysis is a good complement, specially when so many different populations are involved. And this one (supp. fig. 4b, also posted in the article by Dienekes) happens to be much more revealing.

"Yemenite Jews seem to be halfway between the Levant zone"...

But in the K-means analysis they are the "purest" representative of the pink component that is most common among various Arab populations and to lesser extent Jews.

In the global K-means graph, Yemenite Jews and Yemenites appear almost identical, just that Yemenites have greater apportion of the African component.

"And the Azerbaijani, Georgian, and Iranian Jews also seem to be halfway between the Levant and the Caucasus/Iran".

In the K-means graph, Caucasian Jews seem to confirm this pattern you say (half-way).

I have to concedede on Iranian-Iraqi Jews, who, in fact, in the K-means analysis don't cluster so well with Iranians as I first thought but with other more westernly groups including Cypriots and other Jews. Still the brutal Fst difference (0.016) between Iranian Jews and Western Jews (Atzmon 2010) is a very clear indication of different origin, right?

"I have to concede that it really does look like Jews (...) are primarily of Levantine ancestry".

I agree with this but of clearly North Levant ancestry (Lebanon, Cyprus), not Palestinian. Using the term Levantine in this debate seems to be more confusing than revealing, really.

Maju said...

@Michael:

"So, the evidence here actually suggests that Western Jews are probably descended from Israelites; the Palestinians are descended from Arabs who also settled in the land".

That is not what the Admixture graph (supp. fig. 4b) says. On the contrary, it clearly says that at least to a very large degree Palestinians are different (brown component) from other Arabs (pink component) and Jews too.

"One of the hymns to Dionysos refers to the god as Yhyneus..."

Sounds interesting, as other of what you said. Dyonisos was associated, like Jesus, with victory over Death (resurrection). However in all other aspects, the cults are most different.

...

@Dienekes. I couldn't but notice that the graphs that so much pointless discussion have arisen on the use of the word "k-means" clearly use k-means terminology:

"Supplementary figure 4|ADMIXTURE analysis of the Old World and West Eurasian samples, at K=2 through K=10".

^^

Zach said...

Since there are two comment streams on the same topic I'm just going to use this one.

Unfortunately it is so obvious that the conclusion of any genetic study (any subsequent studies) is going to be turned into political fodder (Palestinians are Bedouin-derived; Jews are converts etc etc).

What this study has inadvertently shows that regardless of the ultimate origins of the Palestinians (which in the grand scheme of things do not matter since identity is what you make of it) there are a separate endogamous population or have been so for almost as long as the Druze? If the Druze form a distinct cluster after 700(?)years of endogamy, we can safely assume the Palestinian population has been a coherent endogamous group for at half a millennia?

Also what's funny about the whole debate about the Palestinian ancestral debate is that the true heartland of Israel is Judea, Samaria & Galilee (the Arab-majority region of Israel).

That means that the Israelis live in "Philistine" and the Palestinians (Bedouin or not Bedouin) live in "Israel".

Also this insistency on "alienising" the Palestinians makes no sense since the Copts (Egypt's indigenous Christians) have the same level of J1s.

http://en.wikipedia.org/wiki/Haplogroup_J1_(Y-DNA)

J1 is generally frequent amongst Arab Bedouins (62%[6]. It is also very common among others such as those of the southern Levant, i.e. Palestinians (38.4%) [7], in Algeria (35%)[7], Iraq (28.2%)[7], Tunisia (31%)[8], Syria (30%), Egypt (20%)[9], and the Sinai Peninsula. The frequency of Haplogroup J1 collapses suddenly at the borders of Arabic speaking countries with mainly non-Arabic speaking countries, such as Turkey (9%) and Iran (9.5%) [10]. It is also highly frequent among Jews, especially the Kohanim caste (46%) [1] [11] .

In North Africa,It spread to North Africa (as identified by the motif YCAIIa22-YCAIIb22; among Algerians 35.0%, Tunisians 31%), J1 first entered Ethiopia with the spread of Semitic speakers [citation needed] Eritrea (11%), Ethiopia (9%), Ethiopia-Amhara (33.3%). J1 also may be found with high frequency in the northern parts of Sudan (J-12f2(xJ2-M172): Arabs 45%, Nubians 41%, Copts 39%, Beja 36%), and present with lower frequency in the region of Darfur (J-12f2(xJ2-M172): Masalit 6%, Fur 6%).[17] Haplogroup J1 may be found in as many as 20% of Egyptian males,[18] with the frequency of this haplogroup tending to be comparatively high in the south of the country.[19]


Also the Cohanim Modal Haplotype is a "Semitic" gene, which means that if anything Cohen Judaised populations abroad. Hence the persistence of only priestly Hebrew names in the Diaspora (who are more Semitic than the rest of the population). At least there should be some tribal/clan reminisces in some Hebraic naming convention.

Also if there was a population replacement; http://en.wikipedia.org/wiki/Palestinian_Arabic

Palestinian Arabic, like all forms of Levantine Arabic, is strongly influenced by Aramaic; which was spoken in the Levant before the arrival of Arabic.

In addition the rural dialects of Palestinian Arabic contain features that appear to resemble their classical Hebrew counterparts.

The clearest example is the second and third person plural pronouns. Hemme (they) resembles Hebrew hēm as against Classical Arabic hum, Aramaic hon and general Levantine Arabic henne. Similarly the suffix -kem (you or your) resembles Hebrew -khem as against Classical Arabic -kum and Aramaic and northern Levantine Arabic -kon.

Dienekes said...

@Dienekes. I couldn't but notice that the graphs that so much pointless discussion have arisen on the use of the word "k-means" clearly use k-means terminology:

I can't help it if you're not smart enough to follow a link.

Onur Dincer said...
This comment has been removed by the author.
Onur Dincer said...

What I don't understand is how mostly most or all of the individuals of an analyzed population appear so uniform in their components in STRUCTURE or ADMIXTURE analyses. They usually appear less uniform (maybe much less) in PC plots. What is happening here?! Is there a process in STRUCTURE and ADMIXTURE analyses that causes them to appear genetically more uniform than they actually are, or are they really genetically so uniform?

aargiedude said...

argiedude:
Dienekes, where's the European hunter-gatherer cluster?

Dienekes:
Clustering algorithms like ADMIXTURE exploit differences between individuals, that is why, e.g., they can uncover a "Pygmy" cluster


Fair enough, you're right, though for curiosity's sake I'd still like to see what would happen if STRUCTURE were run on Mexican samples only, all with roughly the same amount of European ancestry, and see wether or not the program would be able to detect their 2 major ancestries.

.............................

argiedude:
"Maju, I don't think I quite agree with you, though your observation about the Palestinian cluster is totally valid and I do agree with that and it does make me wonder".

Maju:
Enough for me. That's all I meant to point out, as for me it is the smoking gun pointing to the definitive resolution of this debate.


That was badly phrased on my part. I don't quite agree with you about the ancestry of Jews, though your observation about the Palestinians is very good. I totally agree with you that Palestinians have been shown in this study to be overwhelmingly or completely descended from the aboriginal people of the Levant, to the point that their clusters are a perfect blend between their 2 major neighbors: Egypt and Syria/Jordan. And then of course, in the other cluster chart there is the fact that you pointed out about their unique "Palestinian" cluster, which is also very interesting and thus indeed once again points to their continued existence in the region. Dienekes says it could be due to recent genetic drift, like in the study of the Sardinian towns, but we can clearly see that this brown cluster exists in a small but notable minority amongst the exiled Sephardic and Ashkenazi Jews, indicating it existed before their Diaspora, and also possibly indicating that Jews might come more from the northern Levant (Syria) than from Palestine/Israel.

aargiedude said...

eurologist:
Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks.


onur:
It is because of the small but noticeable Mongoloid components (mostly north Mongoloid) among Turks. Because that Caucasoids and Mongoloids are genetically very distant from each other, a small Mongoloid admixture can pull an otherwise Caucasoid population apart from other Caucasoids in global PC plots.


Then why are Russians so close to other Europeans? They supposedly have as much Mongoloid ancestry as Turks, slightly more even.

.............................

Michael:

Arabs had been settling north of the Arabian peninsula since before the 6th century BCE

they opened it up for later Arab invaders

the evidence here actually suggests that Western Jews are probably descended from Israelites; the Palestinians are descended from Arabs who also settled in the land.


I was going to call you a retard but I think it's clear what you really are is a Jewish jerk who believes in the Jewish equivalent of Afrocentrism: that Jews are the true people of the land and Palestinians, locked up in concentration camps for 60 years in the desert, are foreigners anyhow so they should just stop bugging Israel about getting back inside their country and instead go back to where they really came from.

The study doesn't "actually suggest" that Palestinians are descended from Arabs, they look exactly like what you would expect from a people that have always lived in the Levant. Their clusters are a blend between their 2 major neighbors: northern Levant and Egyptians. Exactly what you'd expect from the aboriginal inhabitants of the region. And the study doesn't "actually suggest" that Jews are the true Israelites. Precisely one of the things that surprised me about the results is that it seems Jewish ancestry from the out-of-Levant countries they've been living in is much higher than the 10% or 15% I had presumed till now based on y-dna evidence. It seems Ashkenazis and Sephardics are 1/3 to 1/2 European. So "actually", the study suggests that Palestinians are like the Mayas and Aztecs, while Jews are like Mexicans.

aargiedude said...

Maju:
I agree with this but of clearly North Levant ancestry (Lebanon, Cyprus), not Palestinian.


Further above I came to the same conclusion on my own as I followed through on your explanations about the Palestinian cluster. Not saying that it's definitive, but it's definitely a sizeable piece of evidence.

Still the brutal Fst difference (0.016) between Iranian Jews and Western Jews (Atzmon 2010)

That's a notable difference, yes. But couldn't it well be due to their apparently not insignificant European and Iranian ancestry? And remember that there's an unrecognized issue regarding small sample sizes producing distorted FST results, always to the upside. Comparing roughly the same populations and their FST distances, we can see that Heath's results, with sample sizes in the hundreds, gave notoriously smaller distances than were calculated in other studies that used the much smaller sample sizes of the HGDP dataset. The small sample size fuzz factor (SSSFF) :) is usually 0,020 to 0,050 (always towards increasing the score, the smaller the sample size becomes). So who knows, it could well be that the primary ancestry of European and Iranian Jews is from the Levant and yet they may still achieve an FST between each other of 0,016 due to European/Iranian ancestry, small sample size fuzz factor, and at least in Ashkenazis, the genetic bottleneck issue.

.............................

I have a question regarding these cluster analyses in general. Is there always a "default" "nothing" cluster? For example, in K=2, aren't we really seeing only 1 cluster being detected (usually the African one), with cluster number 2 simply being a grab-bag where "the rest" are dumped, but not necessarily having any particular relatedness except the fact that they don't belong to cluster 1? And if so, wouldn't this also be the case in K=3, K=10, etc.? Shouldn't there be a default, or nothing category, in every one of these K graphs? And assuming I'm right, in the worldwide chart, at K=10, which cluster would be the "everything else that doesn't fit anywhere" cluster? Light blue?

.............................

What I don't understand is how mostly most or all of the individuals of an analyzed population appear so uniform in their components in STRUCTURE or ADMIXTURE analyses. They usually appear less uniform (maybe much less) in PC plots. What is happening here?! Is there a process in STRUCTURE and ADMIXTURE analyses that causes them to appear genetically more uniform than they actually are, or are they really genetically so uniform?

The PC graph is the most fuzzy of all the instruments here. It's taking a fraction of the overall dna, which you can usually see as a percentage on the side of the chart (for example in the PC chart of this study: Eigenvector 1, eigenvalue = 6.1), and using this to paint a picture of how the samples are related. This dna is selected for high regional differences in frequency (aka AIMs, ancestry informative markers). This is good to paint a big picture of how everything fits together, and it only has the drawback that it will be somewhat fuzzier than a cluster chart that uses the whole dna. Plus the AIMs are selected to depict certain regional differences, and are limited to 2 axis because of the 2 dimensional graph. If there exists a 3rd, 4th, or more ancestries in the populations, their AIMs aren't considered and the graph might be unusually distortive in their particular case. For example, on a Euro-only graph Sardinia usually exists somewhat off from the rest of the bloc but doesn't quite transmit the great magnitude of its genetic differentiation. If we built a 3-dimensional PC graph of Europe and chose Sardinian/non-Sardinian AIMs to represent this 3rd vector, then we would probably see Europe more or less flat in this 3rd dimension, and Sardinia jutting out hugely.

aargiedude said...

argiedude:
"Yemenite Jews seem to be halfway between the Levant zone"...

Maju:
But in the K-means analysis they are the "purest" representative of the pink component that is most common among various Arab populations and to lesser extent Jews.

In the global K-means graph, Yemenite Jews and Yemenites appear almost identical, just that Yemenites have greater apportion of the African component.


I take that back, you're right. PC charts are the most blurry of all the instruments here, so if they're in serious disagreement with the cluster analysis, I would go with the cluster analysis instead, and in the case of Yemenite Jews especially I now agree with you. All I said about Jewish ancestry was based on the PC graphs. Nonetheless, Ashkenazi/Sephardic ancestry in the cluster charts seems to be complicated to disentangle. It looks like it could fit into Anatolian as much as Levantine (excluding the blue European ancestry, of course), but Levant seems a somewhat better fit:

1) Notice the purple in Ash/Seph Jews, which isn't found in Turks but is found in the Levant.
2) The Ash/Seph Jewish pink is relatively significant (after removing European ancestry), like in the Levant, but not like in Turkey.
3) Turkey has more green than light blue, the Levant has slightly more light blue than green, and Ash/Seph Jews have hugely more light blue than green. A lot of this Jewish light blue is from European ancestry, but gauging from their other European cluster, dark blue, we can roughly gauge how much of their light blue is of European origin, remove it, and the picture would likely look more akin to the Levantine ratio than to the Turkish ratio.

Subtle differences, I know. It's not exactly a done deal.

Maju said...

Onur: That's because a population, normally, tends to be most similar to itself, because people marry and have children within it most of the time... because they are a population actually and not just a random set of unrelated individuals.

In PC graphs often that is lost because it's roughly like running the algorithms into just K=4. So populations that don't cluster towards any of the extremes (roughly like components or clusters in the Bayesian analysis) tend to be scattered near the central axes.

A good example is Basques in Bauchet 2007: in the PC graph they appear widely scattered between Iberians and NW Europeans, while in the Bayesian graph they appear as a distinct cluster on their own right... just that this cluster only shows up at K=5. Coincidence? Not at all.

Dienekes said...

In PC graphs often that is lost because it's roughly like running the algorithms into just K=4.

Maju, you should probably take a break from commenting about stuff you clearly know nothing about.

The first two principal components of PCA are NOT like "running the algorithm into just K=4". An individual has three degrees of freedom in a K=4 ADMIXTURE run and two degrees of freedom in the first two principal components of PCA.

The number of clusters that emerge in the first two principal components of PCA can vary from K=1 (if one samples from a homogeneous population) to much greater than "4" (there is no theoretical limit, except perhaps the sample size limit)

There are at least 8 clusters here, for example, at least 5 clusters here or here.

Maju said...

@Argiedude:

"That's a notable difference, yes. But couldn't it well be due to their apparently not insignificant European and Iranian ancestry?"

It could be to some extent. You may be right again.

"And remember that there's an unrecognized issue regarding small sample sizes producing distorted FST results, always to the upside".

That's interesting. I recalled in a previous discussion that someone had pointed out this issue on Fst but could not remember who or what exactly was it about.

Anyhow, in Atzmon's paper all samples are of roughly the same figures: 30-50 individuals, so this factor should not be a big deal. Because even if there's a distortion, it'd be a distortion for all.

If it's all because of Western Jewish admixture with Europeans, then the proportion should be like 50% admixture in comparison to Iranian Jews (who have Fst=30 in relation with Basques). I don't think this holds up well on light of other data that suggests that most Jewish ancestry is from West Asia.

"I have a question regarding these cluster analyses in general. Is there always a "default" "nothing" cluster? For example, in K=2, aren't we really seeing only 1 cluster being detected (usually the African one), with cluster number 2 simply being a grab-bag where "the rest" are dumped"...

As I understand it, nope. It's actually: "is this individual more similar to cluster A or B or something in between?" So actually two clusters are generated by affinity.

However I guess it may be quite tricky and the global run in this case is maybe a good example because it clusters West Eurasians with Africans at K=2, something that rings very odd and must be because of the relatively small size of the African sample, dumped by minimal greater affinity with West Eurasians.

"Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks".

Retaking this at the point that Aargiedude left it, I'd say that at K=2 Caucasian and Turks are both about 50-50 between both clusters ("Bedouin" and "Russian", so to say), but at K=3 we see the formation of a "Caucasian" component, which is also the main component of Turks, Iranians and North Levant peoples through the whole run and is also variably present in Europeans at lower frequencies.

Turks, Iranians and North Levantines have however some meaningful participation of the "Arabian" (pink) component, which is not present at Caucasians (excepting Jews).

"Nonetheless, Ashkenazi/Sephardic ancestry in the cluster charts seems to be complicated to disentangle. It looks like it could fit into Anatolian as much as Levantine (excluding the blue European ancestry, of course), but Levant seems a somewhat better fit".

I totally agree. Levant yes but what part of the Levant? Clearly the North (Cyprus and Lebanon) and not the South (Palestine).

Michael said...

Geez, Argiedude, LOL; you are quite a politician. What I'm saying is that there is a definite difference between northern Levantine and Southern Levantine groups and this is demonstrated by the clustering. So, please learn history (seriously), instead of modern politics, and stop calling other people names just because you disagree with somebody. (And don't assume somebody's politics based on a note written off the cuff.)

That said, I agree with you, I should not have used the term Israelite. The reality is that the politics of ethnicity interferes too greatly with all of these studies, whether its Jews and Palestinians, Greeks and Macedonians, or other groups.

Onur Dincer said...

eurologist:
Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks.

onur:
It is because of the small but noticeable Mongoloid components (mostly north Mongoloid) among Turks. Because that Caucasoids and Mongoloids are genetically very distant from each other, a small Mongoloid admixture can pull an otherwise Caucasoid population apart from other Caucasoids in global PC plots.

argiedude:
Then why are Russians so close to other Europeans? They supposedly have as much Mongoloid ancestry as Turks, slightly more even.


I was talking about the global PC plot, not the pan-Caucasoid one (it is actually a West Eurasian PC plot as it omits indigenous North Africans, but I will henceforth call it pan-Caucasoid PC plot for the sake of ease). In the pan-Caucasoid PC plot Turks appear much nearer to where they actually should be: just nearby Armenians, Georgians and Iranians and closer to Europeans than most of the individuals of these three populations are:

http://2.bp.blogspot.com/_Ish7688voT0/TA_tjcMrcnI/AAAAAAAACb0/iMP8YJM3D8E/s1600/pca.jpg

A simpler version of the pan-Caucasoid PC plot:

http://1.bp.blogspot.com/_Ish7688voT0/TBDgV2r3hxI/AAAAAAAACck/sYi1shNB8bc/s1600/westeurasianpca.jpg

In the pan-Caucasoid PC plot Russians, as a European population, of course appear closer to other Europeans than Turks are, as this is a pan-Caucasoid PC plot.

But in the global PC plot (http://www.nature.com/nature/journal/vaop/ncurrent/extref/nature09103-s3.pdf) everything turns upside down because of the distorting effects of the Mongoloid populations as I explained above. And not just Turks, but also Russians are affected by this distortion (similarly because of the small Mongoloid but noticeable Mongoloid component they have). But as Russians have more Mongoloid component than Turks, in the global PC plot Russians are more drifted away from thier actual position than Turks are, thus they appear much closer to Turks than to other Europeans despite being actually very close to other Europeans and very distant to Turks (more so than most of other Europeans are) as is seen in the pan-Caucasoid PC plot (though even pan-Caucasoid PC plot is far from perfectly representing Caucasoid populations' actual genetic distances to each other). Similar problems happen to Arab populations in the global PC plot (of course, less so in the pan-Caucasoid PC plot) because of their small Negroid admixture.

It is unfortunate that the authors of this paper also (besides the pan-Caucasoid PC plot) used roughly the Caucasoid part of the global PC plot when displaying genetic distances between Caucasoid populations in a 2-dimensional visual manner:

http://4.bp.blogspot.com/_Ish7688voT0/TBDp0Qi88iI/AAAAAAAACc0/_WLpsi3NeKI/s1600/magnification.jpg

This confuses the casual reader as it omits the rest of the global PC plot and thus gives the wrong impressions that Turks and Russians are actually so distant from Europeans, and Armenians and Georgians are very close to central Europeans, which are of course false as is clear from the pan-Caucasoid PC plot and, most importantly, the ADMIXTURE analyses, which are much more reliable than any PC plot.

Onur Dincer said...

Onur:
What I don't understand is how mostly most or all of the individuals of an analyzed population appear so uniform in their components in STRUCTURE or ADMIXTURE analyses. They usually appear less uniform (maybe much less) in PC plots. What is happening here?! Is there a process in STRUCTURE and ADMIXTURE analyses that causes them to appear genetically more uniform than they actually are, or are they really genetically so uniform?

Luis:
That's because a population, normally, tends to be most similar to itself, because people marry and have children within it most of the time... because they are a population actually and not just a random set of unrelated individuals.


I don't think most of the populations (actually ethnicities in the current globally popular usage) are genetically so uniform or homogeneous in themselves. Especially populations of the non-Western world (including Balkans and Eastern Europe), which have been ethnicizing and nationizing only in most recent times.

aargiedude said...

Michael:
Geez, Argiedude, LOL; you are quite a politician. What I'm saying is that there is a definite difference between northern Levantine and Southern Levantine groups and this is demonstrated by the clustering. So, please learn history (seriously), instead of modern politics, and stop calling other people names just because you disagree with somebody. (And don't assume somebody's politics based on a note written off the cuff.)


Now I'm feeling slightly stupid. I apologize. I thought you were a ra-ra Israel kind of guy. Sorry 'bout that.

aargiedude said...

eurologist:
Finally, I find it very interesting that Georgians and Armenians in the PC plot are much closer to central Europeans than Turks.

onur:
It is because of the small but noticeable Mongoloid components (mostly north Mongoloid) among Turks.

argiedude:
Then why are Russians so close to other Europeans?

onur:
I was talking about the global PC plot, not the pan-Caucasoid one


Oh, I see, and you're right, those are some very curious results. But seeing that also Iranians, Adygei, and Lezgin are about the same distance out towards the east as are the Turks, what do you think about flipping the argument around and saying that it's the Armenians and Georgians who are the oddball group in the region, lacking the eastward component that seems to be present at the same rate in all other Anatolian/Caucasus/Iranian populations? And which also might indicate that this eastern element is pretty old, not the result of historic population movements.

ashraf said...

"We know from historical context that Christian Palestinian, Lebanese and Syrians do NOT descend from Arab-Islamic Bedouin; they would be ideal controls for any mixture analysis."

Mr Latif, what about Ghassanid and Lakhmid Arabs Christians (st John of Damascus is said to be Arab) and dont forget that Arameans too are an Arabianic speaking earlier wave of Arab bedouins.

The first attested Arabianic language in levant (ie Ugaritic) is by far more closer to Arabic than to Hebrew or Aramaic and other Arabianic(=Semitic)languages(so close that we could name it proto Arabic).

Could the light green component represent the neolithic revolution boosted proto iranohittic associated with the Halafian Culture and the light blue represent the neolithic revolution boosted proto afrasans associated wiyh the natufian culture (both being Lislakh).

Dark blue=>hurro-urarto-caucaso-vasco-pelasgo-preiesubstratoeuropeans?

Dark green=>proto Elamo-Dravidians associated with central Asia/Iran?

But waht about the "purple" and "pink"?

Maju said...

Onur: what I see is that it depends and non-Europeans should not be too different from Europeans in this (after all, political nationalism as we know it has only like 200 years or often much less).

When we talk of "Arabs" is like when we talk of "Latin Europeans" or "Germanics", just that the "imperial" language was introduced many centuries later. That doesn't make Romanians and Spaniards particularly akin, right? In fact it does not make Western Latins specially akin to each other either. It's just a language (and to some extent cultural and legal) layer (including some genetics? possibly but still something minor).

"It is unfortunate that the authors of this paper also (besides the pan-Caucasoid PC plot) used roughly the Caucasoid part of the global PC plot when displaying genetic distances between Caucasoid populations in a 2-dimensional visual manner"

That seems Dienekes' work. I have not seen the original in this case but the "magnification" tag seems self-explanatory.

eurologist said...

Dienekes, where's the European hunter-gatherer cluster? African hunter-gatherer pygmies have a clear genetic cluster. According to you, Europe was inhabited by hunter-gatherers who were almost completely mtdna U4 and U5 (and perhaps U* and U2), while the incoming agriculturalists had virtually or absolutely none of this. Such a huge separation is only seen between the 5 main genetic clusters of the world, such as blacks, Caucasians, East Asians, etc., and these major genetic clusters always appear immediately in these sorts of genetic analyses. The European hunter-gatherers should have formed their own 6th major worldwide cluster.

As I mentioned before, the dark blue is as close as you can get, here - the analysis s heavily distorted by a huge emphasis (and rightly so in this context) of Mediterranean samples. Light green (Southern Caucasian with Afghan/Northern Pakistani/Northern Indian relation) and light blue (Mediterranean, here of all NE regions but mostly eastern, originally) are mostly northern and southern, respectively, agriculturalist and later newcomers.

As to Armenians and Georgians, as I mentioned before, PC3 and up could be important and I would love to see a 3D image - but by definition, only that much. In other words, PC2 results are very meaningful in their own right. And once you remove far eastern, south-Indian and African components, you of course simply remove the Turkishness from the Turks. I agree that in that case, Georgians, and Armenians are just slightly more northern and European than Turks - but that is just stating the obvious.

Also, on the Anatolian Jewish origin I would like to comment that much of historically NW Syria is Turkey, now - but it is really the NW Levant, and as such historically one of the major source regions of later European Jews.

Dienekes said...

As I mentioned before, the dark blue is as close as you can get,

There is nothing that connects the dark blue with Paleolithic Europeans in particular. As I've said above, these clusters don't come with dates.

Onur Dincer said...

Onur:
It is unfortunate that the authors of this paper also (besides the pan-Caucasoid PC plot) used roughly the Caucasoid part of the global PC plot when displaying genetic distances between Caucasoid populations in a 2-dimensional visual manner

Luis:
That seems Dienekes' work. I have not seen the original in this case but the "magnification" tag seems self-explanatory.


No, it is in the original paper. It is in the Figure 1 of the original paper (as Dienekes says when presenting it). You can access it from the freely available "Figures and tables index":

http://www.nature.com/nature/journal/vaop/ncurrent/fig_tab/nature09103_ft.html

Onur Dincer said...

The very small but still clearly noticeable Mongoloid component (they seem to have only the yellow Mongoloid component - at least in a clearly noticeable amount) seems to be easily explained by living near Turkic speaking populations (especially Azeris) and being Muslim, though we should first see a similar genetic analysis of Azeris to be able make a strong case in this issue.

In this paragraph I was specifically talking about Lezgins, but accidentally forgot to mention their name. But the same factors may have affected the genetic makeup of Muslim Adygei/Circassians (vast majority of all Adygei/Circassians worldwide).

I should have written thus:

"The very small but still clearly noticeable Mongoloid component (they seem to have only the yellow Mongoloid component - at least in a clearly noticeable amount) in Lezgins investigated in this study seems to be easily explained by living near Turkic speaking populations (especially Azeris) and being Muslim, though we should first see a similar genetic analysis of Azeris to be able make a strong case in this issue."

I should also have added the small population size of the whole Lezgin ethnicity as a contributing factor.

Onur Dincer said...

But seeing that also Iranians, Adygei, and Lezgin are about the same distance out towards the east as are the Turks, what do you think about flipping the argument around and saying that it's the Armenians and Georgians who are the oddball group in the region, lacking the eastward component that seems to be present at the same rate in all other Anatolian/Caucasus/Iranian populations? And which also might indicate that this eastern element is pretty old, not the result of historic population movements.

I would say that much more investigation is needed to say anything conclusive on this issue. In some genetic studies Armenians show similar amounts of Mongoloid admixture with Turks. So, yes, your alternative is also plausible. A significant proportion, if not all of it, of the Mongoloid admixture in Turks may be pre-Turkic and even pre-historic. A handful of ancient DNA specimens from pre-Turkish Anatolia would say much more on this issue than all of the genetic studies on present-day populations of Anatolia and its geographic neighbors.

Onur Dincer said...

Luis:
what I see is that it depends and non-Europeans should not be too different from Europeans in this (after all, political nationalism as we know it has only like 200 years or often much less).


I have no objection to that, I was talking in a relative sense.

When we talk of "Arabs" is like when we talk of "Latin Europeans" or "Germanics", just that the "imperial" language was introduced many centuries later. That doesn't make Romanians and Spaniards particularly akin, right? In fact it does not make Western Latins specially akin to each other either. It's just a language (and to some extent cultural and legal) layer (including some genetics? possibly but still something minor).

I was talking about intra-ethnic uniformity, not inter-ethnic nor even inter-same language family or sub-family ethnic uniformity.

eurologist said...

There is nothing that connects the dark blue with Paleolithic Europeans in particular.

Except it is European, only...

What else do you think it is?

eurologist said...

OK, sensu strictu I agree with you - but in the context of the question of what larger group may be identified here as old/paleolithic European, the results (as others) are very obvious, so I would argue that any argument against it must come up with strong evidence.

Michael said...

Hi Ashraf,

Thank you for naming the Lakhmids and Ghassanids; they were the tribes mentioned in Kimmerling's book as vying for control of Palestine during the seventh century CE (and yes, they were probably there much earlier.)

Linguists do refer to all Semitic languages as being dialects of Arabic because there are too many cognates and false cognates that make the languages close to each other. However, Ugaritic is actually a Northwest Semitic dialect associated with Hebrew and other Canaanite dialects. These dialects include shifts in the pronunciation of certain consonants at the beginning of the word to the middle and end of the word. Pa'alti (I did) becomes efa'l (I will do) in Hebrew, as opposed to fa'altu and efa'lu in Arabic. As a false cognate, lehem means meat in Arabic, lekhem means bread/grain in Hebrew.

The Biblical account mentions Hurrians and Hittites as recognized members of society; Uriah the Hittite, was the husband of Bathsheba before David. North/South trade, and East/West trade probably had some impact on the indigenous populations.

The minority populations might maintain a slight difference in genetic admixture as a result of separation from the majority population; this would hopefully give a more nuanced approach in population studies, and clarify the history of population movement.

aargiedude said...


Dienekes:
There is nothing that connects the dark blue with Paleolithic Europeans in particular.

eurologist:
Except it is European, only...

What else do you think it is?


Good point! Ha ha, hadn't even thought of that. Further reenforced by the fact that it's present as a small but notable minority in all populations that are geographically touching Europe: Turkey, Caucasus, Uygur, Uzbek. And it makes up at least 50% of the dna of southern Europeans, requiring an odd north to south massive population replacement. Damn, there's even a small amount in Mongols but not in other East Asians! Ok, Dienekes, explain how this all fits in with a massive population replacement of Europeans by Middle Eastern/Anatolian farmers.

aargiedude said...

onur:
as in actual reality Greeks, Albanians, Bulgarians, Makedonians (maybe also Romanians and Yugoslavs) and Italians are genetically closer to Turks than to Russians


Even Romanians, the people geographically nearest to Turkey, are probably closer genetically to the farthest corners of Europe than they are to Turks. The best results we've seen so far are from Heath's study of Europe from 2008. It's one of very few autosomal studies that used hundreds of samples per population, and thanks to that we were able to realize that the sample size does matter, and in the case of very closely related people the margin of error produced by a small sample size can even be greater than the true result. Here are Heath's results for the 4 corners of Europe:

Romania-Spain 0,0023
Romania-England 0,0028
Romania-Russia 0,0030

My own estimates bunching up all HGDP European and Middle Easterner samples into 2 populations, so as to achieve sample sizes approaching Heath's, yielded a distance of 0,0130. And Turkish autosomal genetic clusters indicate that they're more akin to the Middle East than to Europe, though obviously intermediate between either. I would gamble that Turkey's genetic distance to all of Europe combined is somewhere around 0,0050 to 0,0100. So at best, Romanians might be at an equal genetic distance to Turks as they are to the farthest corners of Europe (Spain, England, Russia).

Remember Karafet's study of Indonesia? I made the observation that there was an extremely sharp y-dna cline right along Wallace's Line. That sort of genetic separation is what you would expect as a result of paleolithic interactions, seriously limited by geographical barriers, not by Neolithic/historic population movements, who could hardly care less of such a boundary as they expand throughout Southeast Asia in their boats. The same thing is at play in Europe. The biggest y-dna and autosomal divides are always along the water divides: Iberia-Morocco, Romania-Turkey, Italy-Sardinia. The width of the waterway isn't that significant, the important thing is that the contact point between the 2 regions narrows down, restricting geneflow and gradually producing different genetic clusters on either side, where geneflow is relatively unrestricted amongst themselves. You might point out that England seems to be very close to mainland Europe, in fact its y-dna seems almost indistinguishable from the Netherlands, but this argument backs up my point oif view even further: the North Sea was open land during the paleolithic, and it was no narrow bridge, either, it was a very wide walkway. And yes, I'm aware that my point of view seems like that of a martian from another dimension, when you all are talking about Neolithic this and that, and even trying to determine which haplogroups correponded to which population movements.

aargiedude said...

Here's something odd:

In the global cluster chart, Sardinia has about 40% light blue in K=8, which goes up to 55% in K=9, and 65% in K=10. But the new clusters added in K=9 and K=10 don't affect Sardinia in any way, they hardly exist. Sardinia is always a combination of just light blue and dark blue.

Why would the addition of new clusters that are virtually inexistent in Sardinia cause the STRUCTURE program to change so drastically the manner in which it interpreted the existence of the other 2 older clusters?

terryt said...

"That means that the Israelis live in 'Philistine' and the Palestinians (Bedouin or not Bedouin) live in 'Israel'".

I have long found that amusing.

"it's the Armenians and Georgians who are the oddball group in the region, lacking the eastward component that seems to be present at the same rate in all other Anatolian/Caucasus/Iranian populations?"

Mountain people tend to remain isolated and are able to ward off invaders more effectively than are plains-dwelling people. Perhaps that's the reason Armenians and Georgians managed to remain separate from the invaders.

"And which also might indicate that this eastern element is pretty old, not the result of historic population movements".

Probably true.

"North/South trade, and East/West trade probably had some impact on the indigenous populations".

And probably helped spread that indigenous population along the trade routes. Interesting that most people here seem to agree the Jewish homeland is actually north of Judah, towards the ancient Israel. We know the Phoenicians were traders and that they had strong connections to King Ahab's Israel, centred in the Yisre'el valley.

Polak said...

You guys should stop thinking of these Russians are representatives of the large ethnic Russian population of western and central European Russia.

These are the HGDP Russians from Vologda, and after carefully checking out their data, I have to tell you they're very different from me, while all the western and central Russians I'm sharing data with are very similar to me and other Poles.

So this Vologda bunch look like descendants of North Balts, Baltic Finns, some Slavs, and maybe even some other eastern groups like Samoyeds. Their North/East Asian affinity is absolutely huge compared to that of other Russians.

BTW, what I found interesting in this study was the breakdown of the genetic substructure around the Baltic. It seems to me that this area totally lacked the Western/Central European medium blue component which probably arrived with Balto-Slavic and Germanic expansions. The Belorussians have a bit of it, while most of the Lithuanians and Vologda Russians don't appear to carry any.

So Lithuanians appear to be a good proxy set for pre-Indo-European Northeast Europeans. The Vologda Russians aren't as good, as they seem to have been heavily affected by Uralics, and maybe Turkics.

ashraf said...

I think those classifications based on some "common sound changes" are not so important(you can see some of these corresponding sound changes in some of the Arabic dialects j<=>y, k<=>tsh),also those kinds of classifications had shown their flaws as other language families are concerned(for example the classification of Iranic languages).

I think waht is by far,more important than the "sound changes" is the grammar and in the case of Ugaritic its grammar is by fare closer to Arabic than to Aramean-Hebrew.

Besides, Ugaritic (identically as Arabic) has retained 28 of the 29 proto Arabianic phonemes.

Please look here below:
http://www.mail-archive.com/ancient_hebrew@yahoogroups.com/msg00059.html

"We find this phenomenon even causes entirely different words to be merged into
one word in Hebrew, so for instance the Semitic root for plow is H-r-th. So in
Ugaritic we have H-r-th, in Arabic Haratha, whilst in Hebrew we have Kharash
(the Haa has merged into Khaa and the thaa has merged into shin). But wait,
Hebrew has two meanings for the root Kharash, the other meaning is to be
silent. This is because there's another Semitic root Kh-r-s which means to be
silent. In Hebrew sin and shin have also switched places, so we have in Hebrew
Kharash for being silent also, which in Arabic is Kharasa."



"The second issue is that of grammar. The Hebrew language has almost completely
lost the case system, which only remains in vestiges of some words. Arabic is
the only surviving Semitic language which still retains the proto-Semitic case
system. Likewise for the dual number, which all Semitic languages lost, except
for Arabic and Ugaritic again"

Even without reading books on Semitic languages it's sufficient to take a look at wikipedia and compare Ugaritic, Arabic, Aramean and Hebrew.

ashraf said...

Addenda
Hello/Hala mr Michael (Mankael in Arabic)

Even without reading books on Semitic languages it's sufficient to take a look at wikipedia and compare Ugaritic, Arabic, Aramean and Hebrew to see that Ugaritic is closer to Arabic than to Aramean and ancient Hebrew(look also at the moods, verb patterns, numerals and verb conjugation of Arabic-Ugaritic and how they make a cluster distinct of Hebrew and ancient Hebrew).

Dienekes said...

Good point! Ha ha, hadn't even thought of that. Further reenforced by the fact that it's present as a small but notable minority in all populations that are geographically touching Europe: Turkey, Caucasus, Uygur, Uzbek. And it makes up at least 50% of the dna of southern Europeans, requiring an odd north to south massive population replacement. Damn, there's even a small amount in Mongols but not in other East Asians! Ok, Dienekes, explain how this all fits in with a massive population replacement of Europeans by Middle Eastern/Anatolian farmers.

As I have said before, there is nothing in this cluster that would date it to the Paleolithic (except your irrational attachment to your pet theory).

What we do know about it is that it is centered in Europe and extends to Mongolia and India. Thus, it could not have formed after the Indo-Iranian unity (c. 2000BC), but when it formed before that is anyone's guess.

I have already explained to you once that you can't identify a cluster as "hunter-gatherer" using ADMIXTURE unless you have a "hunter-gatherer" population to compare against. You seemed to get it the first time.

I see absolutely no reason why the cluster could not have formed by admixture of early Neolithic farmers with the remnants of pre-farming populations.

Maju said...

"Why would the addition of new clusters that are virtually inexistent in Sardinia cause the STRUCTURE program to change so drastically the manner in which it interpreted the existence of the other 2 older clusters?"

The components, as Polak explained well in a different discussion months ago, only mean fraction of genetic affinity (or more precisely: affinity likelihood). They may totally vanish at greater depths in any population, as some other cluster shows up, for example a Sardinian-specific cluster.

Probably as the apportion of other individuals included in the component varies, so does the cluster within Sardinians.

And, btw, it's ADMIXTURE. Similar to STRUCTURE but not exactly the same (the main difference seems to be that it's a lot faster processing the data).

Gioiello said...

Argiedude writes: "Even Romanians, the people geographically nearest to Turkey, are probably closer genetically to the farthest corners of Europe than they are
to Turks".

Probably a little bit of History would be useful. Rumanians are the descendants of Italians and people of the Rhine Valley (Romanized Celts) after the Trajan's
conquest of Dacia. They have maintained their Latin language (Rumanian is 90% Latin)
and probably it wouldn't be happened if they wouldn't have had a Latin origin.
After the 3rd century they developped not in Dacia but on the right bank of Danube.
Of course during the centuries many other peoples mixed with them, but, if we judge
by Arumanians, there are clearly Italian Y and mtDNA among them (as I have pointed out in the past also on this forum).

ashraf said...

Could we say that the "yellow" component=>"Altaic"?
The orange component=>"Macro-Sinic"?
Dark Green=>"Dravidic"(but in reality more likely "Dravidized" "original" Indians)?
But what the pink and purple compoent could stand for?
Red=>Bantus?
Brown=>pre Bantu African languages?

Remains that the purple component looks very enigmatic as it appears north African but not Egyptian nor Ethiopian=>most likely pre Berber substratic (as Moroccans being far west in north Africa are most likely Berberized pre Berbers??

The lack of light blue and the small presence of light green amongst Russians is against the hypothesis of an Ukrainian steppe for the proto indo-hittite (proto
iranonessic)?

The lack of light blue amongst Ethiopians is against the hypothesis of Ethiopia being the proto afrasanic homeland?(but Egypt could be a good candidate)

Also how could we explain that majoritly indo-european speaking Iran and majoritly Altaic speaking Turkey and majoritly Arabianic speaking Palestinians and Druzes are so close at (k=10) keeping in mind that Indo-Iranians appeared circa 1400th century bc in middle east (first attested Indo-Iranians in the history).
Arabians/Semites appeared circa 2800th century
And Turks appeared circa 8th century ce.

But the bulk of the peopling of middle east (and also Europe&north Africa) is most likely due to the neolithic revolution after the discovery of agriculture long before the date of proto Arabianic and proto Iranonessic (proto indo-european) but perhaps just fit with proto norafrasan/proto afrasan or perhaps proto lislakh or proto nostratic(remembering the huge amounts of common afrasan/indo-european vocabulary-especially for animals and agriculture with intrinsec[either afrasan either indo-european]
etymology and structure
-and similar morphology)

Zach said...

few things don't add up - Palestine was not a rich province in the Ummayad empire.

The Arabs were drawn to the wealthiest provinces; why is the Arab contribution to Syria (particularly) and Iraq so low?

The Bedouins were settling Egypt (where a distinct population) and joining the raids in North Africa.

The almost complete population replacement advocated for Palestine but one that doesn't extend to Lebanon (understandable mountainous), Syria (nonsensical Damascus was the first Arab capital), Iran (the wealthiest province of the empire).

Arab invasions seem to have had a minimal impact on Iran and a 1000 years of Turko Mongolian conquest barely affected the Iranian gene pool (Iran hasn't had a native dynasty since the Samanids; the Pahlavis were Turkish speaking).

The population replacement theory depends on (a.) "limitless flow" of Bedouins (and other nomadic tribes). Turko-Mongolian replacement has been completely disproved for Azerbaijan for instance, which is a good corollary for the Arab invasions.

(b.) that the Arab conquerors/Bedouin settled to become the peasant class.


It makes sense that owing to its closeness to the Hijaz Palestine would be the recipient of Arabian migration just as Al-Jazirah of Iraq has a history of accepting immigration from the Nejd.

However has that substantially affected the "core ethnicity" of the Palestine region; my thoughts would be no more or no less as it has Mesopotamia, where some regions were lightly Arabized (south iraq and north iraq) and others received large influxes of Bedouins (central Iraq - Baghdad).

Anonymous said...

@ Ashraf : "The lack of light blue and the small presence of light green amongst Russians is against the hypothesis of an Ukrainian steppe for the proto indo-hittite (protoiranonessic)?"

Only because you understand it as a massive population movement coming directly from the north of the black sea to Anatolia, which it was probably not.
It was rather a minority invading the south-eastern Europe, mixing there maybe for more than 1,000 yrs and then moving to Anatolia during bronze age.

The first Kurgan wave that wreaked havoc in Bulgaria (destruction of Karanovo VI) around 4,400 BC, could be that coming of the ancestors of Hittites and Luwians (The date works well with the archaisms of Anatolian languages, it could easily be pre-proto-indo-european in the Kurgan theory context), and the destructive population movement in Anataolia coming from the west around 2,700 BC could be their arrival in Anatolia, for instance.

ashraf said...

Perhaps the light yellow corresponds to macro-Altaic (Altaic+Nivkh+Austro-Asiatic...)
mr Latif
The difference betweeen the Turkicization of Anatolia and Azerbaidjan and the Arabisation of Egypt and Iraq is that:
Turks belonged to a distinct linguistical, phenotypical and haplotypical pool than Azerbaidjanis and Anatolians whereas Arabs belonged to the same linguistical and to a great extent the same haplotypical and phenotypical pool.
But still, I think and it is most likely that there was an important Arab input to Egypt, Iraq, Levant(but it's hard to determine its importance 20-30-40-50%? because Arabs and Arabised Arabians belonged to the same ethnico-genetical pool-if we dont count the African input in muslim Arabs)
There was also pre Islamic Arabic states as north as Musul in Iraq (kingdom of Arbaya-Hatra).
Why should Arabian Arabs be so distinct from proto Arabian Ugarites?
A part of Arabian Arabs (Yathrab, Mecca...)and Yemeni Arabs were not bedouins.

mr Wagg.
According to the monumental book "Indo-Europeans and Indo-European languages" Hittite speaking Anatolians came from south eastern Anatolia (Halaf culture-Göbeklitepe) and that make sens as if Anatolians came would have came from west, the western fringes of Anatolia and the north-western Anatolian region should have been Anatolian indo-european speaking but that was not the case as the region of north-west Anatolia was Caucasic Hattic speaking and the Western fringes of Anatolia were Ahiyawanic speaking.
look to this map below

http://dnghu.org/anatolian-languages-map.jpg

Plus, some toponymes of old Anatolia are Indo-European suggesting a long presence of indo-europeans before they entered history (as for example the name of the river Marasawanta ie the Greek Halys and the Turkish kizilirmak).
Also you have to explain the Sumerian, Hurro-Hattic, Kartvelian, Egyptian, Semitic and north-western Caucasic loans into proto Indo-European.

After reading that book, proto indo-europeans seems to have a "middle-eastern" culture, mythology and mentality with parallels in Sumerian, Kartvelian, Hurrian, Semitic and Egyptian mythology.
For example proto Indo-Hittites were familiar with such modern arab and kurd traditions as the:
"berdel"=a family take a bride from another family and giving them a bride "in counterpart"
Also hijacking brides...

If proto Indo-Hittite homeland was the Balkan we would not have Sumerian, Egyptian, Semitic, Hurrian and north eastern Caucasic loans in proto indo-hittite.
We would also have more hg's I in Iran/Tarim/Afghanistan...

Also the flaura and fauna of proto Indo-European can not match with Balkans (Balkans would be a sort of a second homeland).

There are also many lexical paralleles between proto Indo-Hittite and Arabianic(as well as common numerals, basical verbs, animals, plants and metals...).
hem/dam(article d+am)=blood
human/adam=human
humus/adama=ground

ashraf said...

If we go to make a parallel between mythological stories (which,in many times, are the "flashbacks" of very ancient events??)
Could we say with some fantasy that:
1/The pink component in Arabs correspond to the so called "perished Arabs/Arab i baide"[of course it does not mean that those perished Arabs were Arab speaking but perhaps a very ancient folk or human group as very ancient gatherers or Mountainous folk??...]?
2/The light blue component in Arabs correspond to the Ariba Arabs[here too mosl likely a sort of mythological metaphor=>desert herders, tradesmen??...]?
3/The light green component in Arabs corresponds to the "arab musta'riba" (Arabised Arabs)[farmers...??]?
The story of Cain/Qabil killing his brother Abel/Habil could be a mythological metaphor connected somehow with homos neanderthalis retreating back and then perishing letting modern humans colonise the earth!
I dont count Sumerians as from my readings Sumerians seem to have been a recently incoming small folk from Iran (perhaps connected with Tibetan language)??

Gioiello said...

Ashraf writes: "There are also many lexical paralleles between proto Indo-Hittite and Arabianic(as well as common numerals, basical verbs, animals, plants and metals...).
hem/dam(article d+am)=blood
human/adam=human
humus/adama=ground"

Dienekes, please, stop all these foolishnesses!

Michael said...

Ashraf,

Everything you mention is valid; however, in NW Semitic languages, there are approximately 7 letters (bgdcfrt) where there is a pronunciation shift based on the location of the letter (at the beginning it is pronounced one way, in the middle and at the end it is pronounced differently). This does not occur in Arabic or other SW Semitic languages.

Something that has not been addressed here is demographic shifts caused by cultural differences. Christianity allows only one living spouse per individual; Islam allows up to 4 wives per man. The conquest of Palestine by the Muslims would place the Muslim population at the top of the social scale, probably making them the wealthiest community. In an agrarian society, more children increases the number of hands working the land, so more wives increases the potential for more children. In a clan based society, more children increases the number of marriages with other clans, and improves the political situation for the family should there be inter-clan hostilities.
Over the course of a handful of generations, they would become the majority of the population in a region that had undergone massive population shifts. The Lebanese, Syrians, and Iraqis had larger populations, and did not experience the same kind of warfare as occurred in Palestine during the first to seventh centuries. Major wars decimated both Jewish and Samaritan populations and Christianity became the dominant religion. When the Muslims arrived to Palestine, they became the political leadership, and after several generations, their descendants would become the majority. With conversions of local populations, they would gain some of the genes, but the various ethno-religious communities need to be studied to determine whether they reflect Southern Levantine, Northern Levantine, or Southern European origins.

The major questions are, what percentage of the population actually engaged in multiple marriage, and historically, did it have a major impact on society?

Onur Dincer said...

Ashraf, by Ahiyawanic, do you mean Achaean/Mycenaean language (Bronze Age Greek), which was spoken at that time probably only in Greek mainland (probably only southern parts) and some of the Aegean islands (maybe also some coastal western Anatolian cities), or do you mean a non-Greek, and maybe also non-Indo-European, western Anatolian and/or Aegean island language?

Also in what meaning are you using the term "Indo-Hittite"? Are you using it synonymously with the term "Indo-European"?

Btw, I recommend you to just use the term "Semitic" instead of "Arabianic" for the sake of clarity (as probably no one else but you uses the term "Arabianic" in the whole world :D).

AWood said...

"but the various ethno-religious communities need to be studied to determine whether they reflect Southern Levantine, Northern Levantine, or Southern European origins."

At what point do they become South European? Keep in mind these admixture graphs represent deep ancestry that would infer significant mass migration of men and women to change drastically. Are you talking specifically of the diaspora jewish gene pool, or local groups in the Levant?

ashraf said...

"Dienekes, please, stop all these foolishnesses"

This is not mine but Gamkrelidze and Ivanov's (I only added "dam/hema"=blood) I think the connection between conceptions as blood, human and ground is very clear (and this theme is well present in the Arabianite and Iranonessite mythologies and traditions and also proverbs expressions and semantics)
We have also pie hema+hus (house) ps khayma+hush (house) and pie hes (to be) and ps es (to be) wich are perhaps also connected with this semantic system.

Indo-Hittite consist of 2 branches =Indo-European+Anatolia but as Iranic is older than Indic (and closer geographically to the ih homeland) and as Hittite is misnomer (more correct is nesite)=>iranonesic (ic for language).

With the same reasoning, Arabianic (or perhaps other terms such as Ethioarabic...)would refer to the Semitic languages of Arabian peninsula (similar to the label Iranic or Anatolic) since (if I'm not mistaken)Semitic is the only language family which is named after a single mythological person (Sem) whereas the majority of other language families are named after geographical locations.

The name "nostratic" too is somehow controversial as it's suggest an opposition ("nostra"=ours vs "vostra"=yours).

But of course Semitic and Indo-Hittite are the worldwidely accepted appelations.

Yes in the Gamkrelidze's book they wrote about a western neighbor of the Hittites name ahiyawa and they speculate about ahiyawa being possibly related to the akheans.

As for the figures I think "light green" color makes much more sens as the "indo-hittite maker" than the dark blue component since that component is very weak in the strong oldest indo-hittite communities (Turkey,Iran,Armenia,Cyprus,India,Afghanistan...) and is strongly represented amongst Basks.

AWood said...

"As for the figures I think "light green" color makes much more sens as the "indo-hittite maker" than the dark blue component since that component is very weak in the strong oldest indo-hittite communities"

--Rather than associate a specific component with language, it might be better to associate the light green with emerging agriculture and it's spread both east and west. I'm skeptical that it was entirely responsible for the spread, probably just the LBK and Danubians. It's likely agriculture was spread to the Atlantic without that component, or in very small quantity. The lack of light green among Basques and Sardinians may be the result of being fairly isolated communities rather than never interacting or absorbing these people at all.

Anonymous said...

@ Ashraf : Too many disputable and unproved allegations to reply to all so let's make it short.

just two things :

a/ your comment that "Iranic is older than Indic" is wrong.

b/ The proto-indo-european language can't be the result of a neolilthic wave coming from Anatolia because the proto-indo-europeans not only knew the wheel but they also knew a metal (latin aes, sanskrit ayas, Gothic aiz) - two things that don't work well with neolithic - and the horse (ubiquitous root among the indo-european languages - this animal was also obviously quite important culturally to them) while this animal was not present in Anatolia, south-west Asia or south Asia untill rather late bronze age.

Michael said...

Aaron,

This is with regard to both Christian and Jewish populations of the Levant. Alexander built ten cities in the Levant, referred to as the Decapolis, which had Greek colonies for trade and spreading Greek culture throughout the Middle East.

The Orthodox and Catholic churches only gained political power in Rome during the reign of Constantine. With the various Councils (Nicaea and Chalcedon), various sects left the empire for Persia. Greek philosophy was translated into Parsi. After the founding of Baghdad, the Muslims asked for translations of Greek, Roman, and Persian philosophy, science, and history, and established translation schools made up primarily of Christians.

Southern European is a possibility, and it could be from a wide variety of sources. It would probably be relatively small in proportion.

Anonymous said...

@ Ashraf : and BTW you misunderstood me. I never said proto-indo-european appeared in the Balkans so your Y-DNA hg I stuff in Xinjiang and Afghanistan is irrelevant.
The Afro-asiatic/hurrian/ etc... loanwaords in Anatolian you mentionned are OK with the Kurgan hypothesis.
Also the more massive prooves of the west destructive population movement of 2,700 BC is precisely in the area of the Anatolian languages (like near Konya where it was particularly visible in the archeology IIRC).
Also the earliest evidences of Anatolian languages are quite late, so many things could have happened in the geopolitics of Antolia before written documents in the region.
All this don't say much.

aargiedude said...

Dienekes
As I have said before, there is nothing in this cluster that would date it to the Paleolithic (except your irrational attachment to your pet theory).

I have already explained to you once that you can't identify a cluster as "hunter-gatherer" using ADMIXTURE unless you have a "hunter-gatherer" population to compare against. You seemed to get it the first time.


I asked where the European paleolithic cluster was and you made a good argument that it shouldn't show up because it's become too heavily diffused and doesn't stand out in any population as a necessary reference for the STRUCTURE program to recogonize it. Fine. I haven't suddenly forgotten that. I'm talking about something else here, and that is the fact that most of Europe belongs to a cluster that exists as a minority elsewhere, and in a pattern that strongly suggests it was born in Europe and then diffused to all the bordering regions that touch on Europe.

I see absolutely no reason why the cluster could not have formed by admixture of early Neolithic farmers with the remnants of pre-farming populations.

Really? Is that how things work? So sometimes the mixing of populations will result in the appearance of a brand new cluster different from the parental clusters, and other times the ancestral clusters will persist and we will be able to detect the proportions of the ancestral populations that gave birth to the new population. Is that it? Really? And under what circumstances do either case tend to occur? Oh, let me guess, when you need them to for your pet theory of total population replacement by farmers. Right?

The dark blue cluster doesn't show any particular connection to the farmers that replaced all of Europe. Anatolia has a sizeable minority amount similar to other regions directly contacting Europe. And further south, in the Levant, the dark blue diminishes to virtual inexistence. It looks like the direction of flow of the dark blue cluster is from Europe towards the Middle East. The only way this can be reconciled in favor of the farmer replacement theory is to assume that the cluster didn't exist in either the paleolithic Europeans or the farmers, and it came into existence after this supposed event.

PS: In the global cluster chart, we can see, as usual, that the first 3 clusters to appear are sub-Saharans, West Asians, and East Asians. Cluster #4 is the Indian sub-continent, which is expected, as I'm sure we'll all agree that amongst West Asians, Indians are probably the most genetically divergent from the rest of the pack. What cluster appears next? European dark blue. Number #5 spot. Or put another way, of the 7 West Asian clusters, the European cluster is the 2nd to appear. The order of appearance, of course, tends to be correlated with the magnitude of its differentiation, that's why the within-Caucasian clusters only start to appear after the inter-Continental clusters have all shown up.

...............................

Thanks, Maju. [your observations on why the cluster ratios would change between K=8 and K=10, and for noting that the specific program they're using is actually ADMIXTURE.] And yeah, I see now why that ratios would change, makes perfect sense.

Kepler said...

"I think waht is by far,more important than the "sound changes" is the grammar and in the case of Ugaritic its grammar is by fare closer to Arabic than to Aramean-Hebrew."

Are you sure? I think that is not the case. Hebrew is part of the Northwest Semitic language branch.
Arabic is not.
http://en.wikipedia.org/wiki/Northwest_Semitic_languages


I am not sure we should be so sure about population's closeness by looking at the similarity of some linguistic features...specially if we are based on rather limited corpora.

Dienekes said...

I'm talking about something else here, and that is the fact that most of Europe belongs to a cluster that exists as a minority elsewhere, and in a pattern that strongly suggests it was born in Europe and then diffused to all the bordering regions that touch on Europe.

"Born in Europe" does not mean "Born in Europe in the Paleolithic or among European hunter-gatherers", as I have already explained.

In fact, as I have also explained, the fact that it is not well-represented outside Europe, or only in places where it can be explained by late prehistoric population movements is a fairly good argument that it is not very old. Because, if it was very old, then it would spill out over Eurasia and even Africa, as it would have tens of thousands of years to do so.

Really? Is that how things work? So sometimes the mixing of populations will result in the appearance of a brand new cluster different from the parental clusters, and other times the ancestral clusters will persist and we will be able to detect the proportions of the ancestral populations that gave birth to the new population. Is that it? Really? And under what circumstances do either case tend to occur? Oh, let me guess, when you need them to for your pet theory of total population replacement by farmers. Right?

Yes, that is how things work. The blending of different components through a period of endogamy produces a new cluster, as the constituent elements are blended together and do not contrast each other.

Some good examples of this in the global PCA are Mozabites who form their own "purple" cluster out of Caucasoid and Negroid constituent elements, while Mongols do not appear to be homogeneous and do not form their own cluster. In the regional PCA we see that Bedouins do not form a homogeneous population, but are split into two quite distinct groups, one of which is distinctive, while Sardinians form a uniform population out of European and West Asian elements.

Finally, you are able to detect the ancestral proportions that gave rise to the new population, as I have also already explained, when you have relevant reference points to compare against, in this case, a European hunter-gatherer population. When you don't have such a reference point, you are simply foolish to interpret any dateless cluster as representing European hunter-gatherers.

The dark blue cluster doesn't show any particular connection to the farmers that replaced all of Europe. Anatolia has a sizeable minority amount similar to other regions directly contacting Europe. And further south, in the Levant, the dark blue diminishes to virtual inexistence.

This can be due to the fact that it emerged in Europe AFTER the Neolithic, and not BEFORE the Neolithic. There is no inconsistency there, and, as I have explained several times now, its paucity outside Europe makes better sense if it is of recent rather than older vintage.

PS: In the global cluster chart, we can see, as usual, that the first 3 clusters to appear are sub-Saharans, West Asians, and East Asians. Cluster #4 is the Indian sub-continent, which is expected, as I'm sure we'll all agree that amongst West Asians, Indians are probably the most genetically divergent from the rest of the pack. What cluster appears next? European dark blue. Number #5 spot. Or put another way, of the 7 West Asian clusters, the European cluster is the 2nd to appear. The order of appearance, of course, tends to be correlated with the magnitude of its differentiation, that's why the within-Caucasian clusters only start to appear after the inter-Continental clusters have all shown up.

That is nonsense. According to your "logic" the greatest differentiation is between Afro-Europeans and East Asians who split off first, while we know that it is between Africans and Eurasians.

Anonymous said...

@Ashraf, it seems quite clear that you are completely ignorant of Semitics. You conflate issues, and conflate Israeli hebrew(actually 1 DIALECT of Israeli hebrew) with biblical hebrew. Tha and Shin have not been merged, rather there was a shift into hebrew, the Sound tha existed still as Taw without daghesh. 7et and Khaf only merged in some dialects, and is not correct, it has nothing to do with ancient, even today half of Israelis pronounce them distinct, (same with alef and 3ain) the Merging of Taw with 6eth, and waw with veth is a new thing, Arabic also lost sounds like P which is a frequent sound in Ugaritic, and in other semitic languages besides Arabic.

Maju said...

"According to your "logic" the greatest differentiation is between Afro-Europeans and East Asians who split off first, while we know that it is between Africans and Eurasians".

That's probably because the African sample is comparatively much smaller, what pushes them towards the West Eurasian cluster, surely on the minor admixture that exists between both clusters.

It's odd but it's not the first time that such genetic-statistical data has produced West Eurasian-African affinity results. I recall that in 1996, Cavalli-Sforza already suggested that West Eurasians had some African admixture. The opposite is also surely true as well.

It emphasizes in any case how sample sizes do matter a lot.

ashraf said...

mr Kepler
Rather than looking at wikipedia article, I recommand you to read a book about Semitic languages.
This the last thing I could say about that topic.
As I said those classifications are controversial.
Some times ago Arabic was classified as a south Semitic language with Ethiopic in Wikipedia, now it's classified as western Semitic.
But still you can compare Arabic-Ugaritic with Hebrew-Aramaic there in Wiki.
Also could you explain what are the shared northwestern Semitic features that Arabic lacks?(after reading much about Semitic languages my impression is that shared Ugaritic-Arabic features lacking in Hebrew-Aramean are more important and numerous)

"a/ your comment that "Iranic is older than Indic" is wrong."

Iranic should be older than Indic
because it's by far more diverse.

"b/ The proto-indo-european language can't be the result of a neolilthic wave coming from Anatolia because the proto-indo-europeans not only knew the wheel but they also knew a metal"

I never said proto ie is neolithic (albeit pre proto ie would be so).

c/"The Afro-asiatic/hurrian/ etc... loanwaords in Anatolian you mentionned are OK with the Kurgan hypothesis.
Also the more massive prooves of the west destructive population movement of 2,700 BC is precisely in the area of the Anatolian languages Also the earliest evidences of Anatolian languages are quite late"

Those loanwords are not in Anatolian but in proto indo-hittite, if you read Gamkrelidze's book you will see that many of those loans are lacking in Anatolian languages.
The Anatolian branch is not late, as it's the first attested indo-hittite language(of course other old ie languages in Anatolia, for many resons, did not "afford" to left a written record of their languages)
Anatolian is much older than 2700 bc since when first appeared in written records there was already many distinct dialects (Luwian, Palaic, Hittite...)
I dont understand why massive destruction movements (if occured) should necessarily bring languages (for example the sea peoples and the mongols in middle east/Europe did not brought one).
As for the matter of the wheel and horse, it is well explained in that book (if I remember well first wheels and chariots were found in Sumerian sites and horses were present in western Asia, besides words for Horse and Wheel were loans/cognates from Sumerian/Akkadian and other languages I forgot[I will try to write that stuff later])

For the Kuragns, if I remember well in Gamkrelidze's book it's said that they came from western Asia/Anatolia and are perhaps responsible of the dispersal of the Indo-Iranian languages.

ashraf said...

I consulted the book here some notes:
The most important proof against(aside from the demographic problem)an Ukrainian steppe for proto indo-european is the words for flaura, fauna, myths and tools in proto indo-european.

"The word for "plow" in proto indo-european is "harthro"(nearly identical to the Arabianic one)and there was not agriculture that time in Ukrainian steppes."

As you wrote, "the pie word for "copper/bronze/iron" is "haye/os", in the book it's written that the earliest specimens of copper come from the middle east beginning with çatal hüyük."

"pie for copper "reudh" is connected with Sumerian one "urudu".
Copper mettalurgy is first attested in mesopotamia (Hassuna and Ubaid cultures)."

Oddly enough, "one word for silver is shared by some ie languages(silpar), semitic and ... Bask."

On the other hand "the pie word for silver (hark) is shared by Caucasic languages."

This for metals as for the wheel:
"Pie is "kel"(rotate,wheel)[it rembers me Arabianic kur=rotate,circle] and "keklo"(wheel,wheeled carriage).
Another pie word for cart is "roto"[it remembers me Arabianic arrada=war chariot]
Those words have cognates in Hebrew (gilgal), Arabiac ('agal) Aramean (galgal), Georgian (Gorgal), Sumerian (gigir).
"
As for the horse:
"The Przwalski horse found in central and eastern esia and surviving to this day in scattered groups near the Gobi desert in Mongolia cannot have been that ancestor because according to recent data it differs
genetically from the domestic horse the przewalski horse has 66 pairs of chromosomes while the domestic horse has 64 hence the original area of the
perzewalski horse is ruled out as the center of domestication of the horse further evidence against an asian center of domestication is the lack of domesticated
horses in eastern Asia in particular in china until the yin period at which point the horse is introduced from the west evidently under west asiatic cultural influence.
the onager was found in the broad steppe zone north of mesopotamia seventh millennium bc
in the southern near east the domesticated horse is attested for the fourth
millennium bc in culture sites in mesopotmia elam Susa and adjacent
areas of ancient iran where ancient horses are depicted on vases and statuettes
from hafaj near baghdad
an analogous picture can be reconstructed for asia minor at an even earlier
period.
domesticated horse bones were recntly found at demirci Höyük yankkaya and norshintepe in eastern anatolia they are found in Bronze
age strata as early as the second half of the fourth millennium bc
the earliest written evidence for horses in asia minor is found in old
assyrian tablets from Kültepe karum Kanish which make frequent mention of "rabf sisawm"=chief in charge of the horses and also sometimes mention the use of
horses sisum for transport
that ancient Mesopotamia is one possible area of horse domestication follows from
the fact that traces of horses going back to the seventh millennium bc are found in this area.
simlar remains found in paleolithic caves in palestine evidently are those of wild horses"

Here "cognates with pie "ekwo" (horse):
Hurrian essi,Akkadian sisu,Ugaritic ssw,Aramean susya,Hebrew sus,Egyptian ssmt, Abkhaz acy,Avar cu,Georgian acu,Akhvakh icwa"

Onur Dincer said...

I don't know if anyone noticed, but in this study Palestinians appear where they should have been in Atzmon's study: between Bedouins and Druze. In Atzmon study Palestinians appear more distant to Europeans than Bedouins are, but logically reverse should be true, as is the case in this study. But this time in this study Bedouins appear more distant to Europeans than Saudis are. Logically reverse should be true. May that be because of genetic drift as Negev Bedouins are a small population?

ashraf said...

For the Ugaritic problem, I recommand you to read the book of Edward Lipinsky "Semitic languages outline of a comparative grammar".
The author explains the difficulty of classifiying Semitic languages and states that Ugaritic shares archaic Semitic features with Arabic and Eblaite that are not present in Canaanite (Hebrew) and Aramean.
Though he classifies Hebrew (Canaanite), Aramaic Ancient north Arabic and Arabic together under a western Semitic branch and classifies Amorite, Paleosyrian
(Eblaite, "Mariic"),Amorite and Ugaritic together under north Semitic branch.
But remains that the fact that Arabic and Ugaritic inherited 28 of the 29 proto Semitic phonemes (compared to only 22 in Canaanite-Hebrew) and maintaned the case system, dual and moods distinction (in opposite of Canaanite-Hebrew&Aramaic) should give a clear idea about the closeness of Arabic-Ugaritic.

Anonymous said...

@ Ashraf :

"Iranic should be older than Indic because it's by far more diverse."

It doesn't mean a thing and the earliest written tracks (rigvedic sanskrit) are Indic anyway (or close : Mitanni). Many specialists consider Indic closer to proto-indo-iranian, if I'm not mistaken. At worst, it's not possible to know it, so let's stick with what we actually know.

"the pie word for "copper/bronze/iron" is "haye/os", in the book it's written that the earliest specimens of copper come from the middle east beginning with çatal hüyük."

I've read things different about it but even if true, the technology and word could have travelled together, but another good question is how frequent is this root in non-indo-european languages of Anatolia ... I am not aware it is present, and as such the root should be treated as IE specific and from outside of Anatolia.
Anyway that's post-neolithic which was the point of my post : a massive IE spreading from Anatolia during Chalcolithic/bronze age don't work well, unlike the Kurgan hypothesis. And so far the visible R1a1a + west Eurasian MtDNA hgs migration to the west and to the east (up to mongolia and south Siberia) supports the Kurgan theory much better I think.

"I never said proto ie is neolithic"

Then how did it spread from north-western-europe up to east India and north-western china? Anything in Archeology to support such a phenomenon?
On the other hand the intrusive population movements from the north of the black sea, both west and east, are visible in the expected timeframe (chalcolithic/bronze age).
An anatolian language didn't magically appear throughout Eurasia without leaving specific tracks in archeology. Neolithic could have explained it but the vocabulary goes against that idea. After, it doesn't seem likely.

"Those loanwords are not in Anatolian but in proto indo-hittite, if you read Gamkrelidze's book you will see that many of those loans are lacking in Anatolian languages."

I don't have the book in front of my eyes and I don't know the opinion of the other specialists, so... but anyway migrations from Anatolia spreading agriculture and some cultural element in Europe is accepted by basically everyone so nothing say that these loanwords from specific vocabulary aren't from such a (former) population substrate and/or cultural exchanges concerning new technologies, if corroborated.

"The Anatolian branch is not late, as it's the first attested indo-hittite language"

That's what I mean by "late" : around +/- 1,500 BC, well into "late" bronze age, i.e. way way after its origin.

"Anatolian is much older than 2700 bc"

No kidding.

"I dont understand why massive destruction movements (if occured) should necessarily bring languages"

It's a proof a of a disruptive intrusion ans as such a good candidate for such a phenomenon.... It's not a proof per se, of course.

Anonymous said...

@ Ashraf :

"The most important proof against(aside from the demographic problem)an Ukrainian steppe for proto indo-european is the words for flaura, fauna, myths and tools in proto indo-european."

I disagree. Like many peoples I think they all work well with the north of the black sea.

"The word for "plow" in proto indo-european is "harthro"(nearly identical to the Arabianic one)and there was not agriculture that time in Ukrainian steppes."

There was agriculture in the north of the black sea in the concerned timeframe and it didn't originate there, so loanwoards are to be expected.

about Wheel and reudh roots : It could have been borrowed. Noone is claiming that it originated north of the black sea. I was using it as a timeframe marker.

"if I remember well first wheels and chariots were found in Sumerian sites and horses were present in western Asia"

Apparently both allegations are wrong. Concerning the wheel I already provided some infos the last time, one or two months ago.
some infos concerning the horse domestication : http://en.wikipedia.org/wiki/Domestication_of_the_horse (sources at the end)

"Here "cognates with pie "ekwo" (horse):
Hurrian essi,Akkadian sisu,Ugaritic ssw,Aramean susya,Hebrew sus,Egyptian ssmt, Abkhaz acy,Avar cu,Georgian acu,Akhvakh icwa" "


That's perfectly logic. If it was brought "late" (that is bronze age) from the steppes, the IE word had to pervade all these languages.
Given the geographic origin of the horse, that's more a proof of IE transmission than anything else.

"For the Kurgans, if I remember well in Gamkrelidze's book it's said that they came from western Asia/Anatolia"

You must be kidding me. There are a few Kurgans in the north-east of Anatolia IIRC, but they are more recent than the oldest ones found north of the black sea. Archeology shows they spread from the north to the black sea to east of Europe and Asia. Think of Merheleva Ridge in east Ukraine that is 4,000 BC, for instance. ( http://en.wikipedia.org/wiki/Mergeleva_Ridge )

PS : since i believe we are off-topic and that it's obvious we won't agree, I probably won't reply the next one (don't take it personnally) unless there is something that I consider really important to be adressed, in which case I will.

ashraf said...

For the Mitanni the authors explain how the Mitanni is a distinct(from earlies iranic and eraliest indic)and how it has some very archaic features.

For the diffusal of ie languages and the kurgan.

"the movement of ancient indoeuropean tribes to the west northwest and southwest which had begun in the midthird millenium bc became especialy intensive toward the end of the third millennium bc this is the date to be assigned to the corded Ware and battle axe cultures in europe"

"we can also trace the trajectory of horsdrawn carts from their point of origin in the near east eastward to centralasia and farther to the northeast using the petroglyphic depictions of such carts and horses"

"the sintashta burial ground southeast of urals in the village rymninskij is associated with ancient indoeuropans based on presently available archeological data a cultural connection can be assumed between northern iran and the central asian area to the caspian sea and beyond to the west"

"certain detals of ritual like the
rows of bulls heads bear a striking analogy not to iranian materials but to the near east beginning with çatal höyük"

"the spatial and temporal correlation of the secondary indoropean homeland with the kurgan culture of the ural volga steppes in the third milennium bc
the secondary homeland was settled in the third millennium bc not only by the ancient european tribes as a dialect grouping but also by aryans and possibly other indoeuropan tribes this region at this time is the area of what
is known as the kurgan culture (also pit grave culture russian drvnejmnaja culture)this culture was spread from the northern black sea area to the volga steppe and the aral sea"

"the features of the culture reconstructed from material remains are compatible with the culture reconstructed for indoeuropean on linguistic evidence
the kurgan culture of the third milennium bc is characterized by the presence of stockbreeding and agriculture wheeled carriages use of the domestic horse as a draft animal developed copper and then bronze metalurgy and construction of fortresses in high places
also characteristic of the culture
are the distinction of social classes the presence of tribal leaders and a special
class of warriors a significant number of religious symbols(solar chariot and others)and burial with cremation in a few instances"

"a crucial factor in light of the interpretation we propose for the kurgan culture is the fact that it shows connections with the near eastern world"

Dienekes Pontikos said...

Irrelevant comments should no longer be posted on this thread.

terryt said...

"According to your 'logic' the greatest differentiation is between Afro-Europeans and East Asians who split off first, while we know that it is between Africans and Eurasians".

This apparently preposterous suggestion may well be the case.

"It's odd but it's not the first time that such genetic-statistical data has produced West Eurasian-African affinity results. I recall that in 1996, Cavalli-Sforza already suggested that West Eurasians had some African admixture".

In fact Cavalli-Sforza's map of the first principal component for the world shows something very much along the lines of Dienekes' comment. The modern population through Central Asia appears to be basically a hybrid between his two genetic poles. I can't find the map on the Net but it's in his book "History and Geography of Human Genes".

The pattern of separation Dienekes is so incredulous about is easily explained if it can be shown there was no single OoA. Perhaps there were two, at least? What does the evidence say?

Many claim a distinction between eastern and western haplogroups within Eurasia. And arguments continue as to whether the OoA occurred as recently as 40-50 K, or way back before Toba, at least 60-70 K, or even older. And some claim mtDNA haplogroups M and N represent separate movements from Africa.

Maju, for one, is is certain that humans rapidly separated into Indian and African versions. And this eastern version moved rapidly into SE Asia, and then into Australia. He also claims that another expansion from India moved to the northwest, and then into Europe. Recent research shows that elements of this expansion did actually move back into Northeast Africa. Perhaps by then the northern population had become a hybrid between the eastern group and a second OoA. This would certainly accomodate the possible 'greatest differentiation is between Afro-Europeans and East Asians who split off first'.

terryt said...

"In fact Cavalli-Sforza's map of the first principal component for the world shows something very much along the lines of Dienekes' comment".

I still can't find Cavalli-Sforza's map, but some may find my version of it interesting. It's near the beginning of this article:

http://humanevolutionontrial.blogspot.com/2009/06/human-evolution-on-trial-into-australia.html

eurologist said...

The dark blue cluster doesn't show any particular connection to the farmers that replaced all of Europe. Anatolia has a sizeable minority amount similar to other regions directly contacting Europe. And further south, in the Levant, the dark blue diminishes to virtual inexistence. It looks like the direction of flow of the dark blue cluster is from Europe towards the Middle East. The only way this can be reconciled in favor of the farmer replacement theory is to assume that the cluster didn't exist in either the paleolithic Europeans or the farmers, and it came into existence after this supposed event.

I agree with you.

Europe is expected to have its own cluster - and it has. There is nothing neolithic about it. At most 15% to 30% of central to northern Europe are neolithic (depending on region), and this study confirms that.

Dienekes said...

Europe is expected to have its own cluster - and it has. There is nothing neolithic about it. At most 15% to 30% of central to northern Europe are neolithic (depending on region), and this study confirms that.

Your opinion is worthless if it is not backed up by evidence, and there is no reason to place the origin of the blue cluster at any particular time period, and certainly not to the Paleolithic.

In fact, if we look at Africa many of the discovered clusters have strong linguistic associations, thus making them ineligible for a Paleolithic origin, as there is an upper limit to inference on language relationships. The same is true for Asia.

In short there is no reason to ascribe the dark blue cluster to Paleolithic Europeans, and I will not repeat the arguments why that is unlikely to be the case.

terryt said...

Someone else has come up with the same map as I did, only it's in pretty colours, so it's easier to understand:

http://www.google.co.nz/imgres?imgurl=http://www.wwnorton.com/college/anthro/evolve4/ch/15/15_1.gif&imgrefurl=http://www.wwnorton.com/college/anthro/evolve4/ch/15/welcome.shtml&usg=__u7tS_YhyRNt0Qrcq7IKmjYzWi8g=&h=165&w=300&sz=7&hl=en&start=42&um=1&itbs=1&tbnid=iyWO6IJO7l1o3M:&tbnh=64&tbnw=116&prev=/images%3Fq%3D1st%2Bprincipal%2Bcomponent%2Bthe%2Bworld%2Bcavalli-sforza%26start%3D36%26um%3D1%26hl%3Den%26sa%3DN%26ndsp%3D18%26tbs%3Disch:1

Maju said...

Eurologist: I'd say that the European blue components correspond grosso modo with the components at K=2 in other studies. The depth of the analysis here is not enough to show anything else and we do know that the vast majority of West and North European ancestry does not belong to them but to distinctive regional clusters (see Behar 2007 specially).

So the dark blue component it's most likely to represent the blue component in Behar 2007 and the PC2 of Cavalli Sforza's 1996-97 materials, strongest in NE Europe, specially among Finnic peoples. It always shows up as very strong when Europe is considered as a whole, along with an Aegean component but, when we dig further, both become minoritary in the West. The counterpoint is that there is not one single Western component but probably three, representing three different populations, which IMO should have (Epi-)Paleolithic roots.

horacioh said...

Ethiopian and Ashkenazim are both true jews in my hypothesis, the "L2" mtDNA marker is present in the two populations also the derived and sibling mtDNA Hg "M" and "N", as well as the Y markers Hg E3b and 4s too, all of this from East Africa and so. They belong respectively at one of the three nucleous or center jewish ancient populations, that evolving the called "Syrian-European nucleous"(helenistic and Roman times).
The oldest center -Ethiopians belong these- were that developed in Napata and Elephantine (Kush) and whose nucleous or center was after Alexandria, and I called "Coptic Nucleous" derived in two bias, and split forwards the North via Europe -intermixed with the Syrian Europe nucleous- or the South, via Nile and the Horn Of Africa.
The "Babilonian and Persian nucleous" is other of the above three mentioned centers and included Bukara, Iranian and Iraki mainly.
All of this Nucleous take Judaea and Israel like a axis and pendulo.
Another fourth Nucleous or center I call "East Europe" -not mainly conected with ME-, is not ancient like the three others and was the Jewish Khazar Empire stiring into Askenazy current population and others. All of this events were naturaly intrajewish asimilations in all jews current populations.
The Ashkenazim hyperhaploydia is explained by the superposition and overlay of diverse fount or source population , that are all of this of Jewish origin (that consider converted into intraJewish assimilations) , one coming from the “Syrian European nucleous” – that Sephardic as well as preAshenazim bring inside -. The other convergence were the “Coptic Jewish nucleous”, coming from Alexandria, the main and largest Judaic center in ancient times – the buried and graves in Jewish graveyards and catacombs of Tuscan, and Alsace as too Rhineland cities take a lot of Egyptian ornaments and display figures from these, as well as Y and mtDNA markers - . The great Jews migration from Egypt beginning after the Muslim invaders from Arabia in the VII AE century. The “Babylonian and Persian nucleous” take place and contacts newly with and when the “preAshenazim second fase” were migrating to the East Europe. A remarkable contact was with the fourth “East Europe Jews nucleous”-not related or little related with ME-, with the descendant of the Jews Khazarians ones, spreading every where and carrying a lot of East Europe and Eurasian markers. That happen between the XI and XII century AE.
The Tuscan host populations come from Anatolia like infers mtDNA markers, and others, yet present today – a thread Etruscan link - and are so common in South East Basin like Albanian, Grecian, Tunisian and Anatolian , as well as the entirely Italy and some South France spots, practical absent in central or North Europa or East Asia. If we compare Ashkenazy jews with this South European Tuscan population will see a more European genes pool coming from Europe than if we compare with Central and North European population.


Dr Hector H. Otero C.
Argentina.

See too:
http://www.nature.com/news/2010/100603/full/news.2010.277.html
See, Comment 11149 and 12952 with table 1 -partial-, remember that the sibling Hg "M" and "N" -from "L3"- too correspond to East African origin, and are not included at all-see complete table 1, in reference-.

Shimon Abekassis said...

As far as I know the ancient Israelites were themselves a mosaic of distinct groups that settled in (or collonized) the Levant, namely Judaens, Amonites, Moabites, Philistines, Edomites, Midianites, Nabateans, Jebusites, Hittites, Arameans, Hurrians, Assyrians, Babylonians, Persians, Macedonians and Romans.

We know Philistines (who were absorbed by the Israelites after being brutally defeated by the Persians)were orignal from Southern Europe, which partially explains the connection to Cypriotes and other Southern Europeans

Hurrians on the other hand were original from the Caucasus region and (according to the Tikunani Prism) were highly linked to the semi nomadic warriors known as Habiru. Most of the names found in it are Hurrian. A few others are Kassite and Babylonian.

These Habiru are described by Egyptian sources as rebels of Canaan and allies of the Hittites, a people of Anatolian/Caucasian origins that conquered and dominated a large portion of the Levant for centuries. This conflict between them and the Egyptians may very well be the source of the inspiration for the Exodus narrative found in the Torah.

Having that said, it seems ancient Israelites had already some degree of North African, Northern Arabian, Mesopotamian, Anatolian, Caucasian origins before they began settling in Southern Europe.

Many Israelites were traders and as a result of several periods of violence and poverty, it's only natural that part of them began migrating to more prosperous or peaceful areas such as Rome or Southern Arabia. Others, though, were deported to other neighbouring regions during the Assyrian and Babylonian occupation periods, where they intermmaried with local populations.

Admixture with other populations was in fact quiet frequent during certain periods of Israelite History. Seems like a natural strategy to make Israelite communities flourish outside Israel. It could also be a sign of how small they were in the beggining.

By comparing Mitchondrial DNA and Y-DNA analysis, it seems there was a shortage of Israelite females in Europe and certain areas of North Africa and the Middle East where Israelite communities had been established, hence it was only normal Israelite males would seek foreign women to build families.


Ashkenazim are perhaps the most notable example in the sense. They're a mosaic of ancient Israelites and Southern Europeans with some infusion as well from Central and Eastern Europe and the Caucasus.

The so called Palestinians of Israel on the other hand are descendants of ancient Israelites who were gradually converted to Christianism and Islam during the Roman, Arabian and Ottoman occupation periods and also migrants (mostly of Egyptian, Arabian, Babylonian and Assyrian origins) that kept settling in Eretz Israel.

For the next studies I believe it would be very interesting to see a detailed comparison between these "Palestinians", Samaritans and also Jews from Portugal, Spain, Morocco, Algeria, Tunisia, Lybia, Egypt, Yemen, Jordan, Lebanon, Syria, Turkey, Iraq and Iran, along with non Jews from these same countries just to figure out better how related they really are. A distinction in the samples used between between "Palestinians" of Gaza, West Bank and Jordan would be nice as well.

Kevin Brook said...

onur observed "It is noteworthy that Karaite Jews are again (as in Atzmon et al.'s study) missing from this study [Behar et al. 2010, "Genome-wide structure of Jews"]." Here is a remedy to that:

Karaite Jewish Ancestry Explored in DNA Study

On August 28, 2013, Khazaria.com released phase 2 of the results of a genetic study titled "The Genetic Signatures of East European Karaites" at http://www.khazaria.com/genetics/karaites.html Phase 1, which had been less comprehensive, at lower test resolutions, and without maternal lineage testing, had been published in the second edition of Kevin Alan Brook's book "The Jews of Khazaria" in September 2006. This is the first genetic study of people who practice the Karaite form of Judaism. Previous studies had included Rabbinical Jewish populations and the Samaritans. The results of this new study complete the picture of the ancestry of the worldwide diaspora of the people of Israel.

Key findings:

1. The East European Karaite have dominant Middle Eastern (Southwest Asian) elements and frequently match Ashkenazi Jews, Sephardi Jews, Mizrahi Jews, Egyptian Karaites, and non-Jewish peoples of Southwest Asia (the Middle East) and the Caucasus (especially the South Caucasus). They also sometimes matched Europeans.

2. Karaite Cohens are sometimes related to Ashkenazi Cohens.

3. The presence of the Y-DNA haplogroup Q1b1a (Q-L245) in Ashkenazi and Karaite samples is not indicative of Khazar ancestry but rather of Southwest Asian ancestry.

The analytical portion of the article reaches this conclusion:

"Overall, East European Karaites are largely a Middle Eastern people descended from the Israelites, but like other Jewish populations they are a mosaic - the descendants of several ethnic groups that joined this specific stream of Judaism during different periods. Aside from their Israelite component, our DNA study has led us to conclude that they also descend in part from ethnic groups that lived in the Byzantine Empire and in Asia. They may have small amounts of Western European and Caucasus region ancestries (the Byzantine Empire at times included portions of those regions)."

The researchers provide each tested Karaite's Y-DNA and/or mtDNA haplogroup names, the HVR1 and HVR2 mutation values for their mtDNA, and details about their matching individuals including how closely they match and, in some cases, data related to how recently their most recent common ancestor lived. The names and original kit numbers of the tested Karaites are kept confidential.