R1b1b2a1 is almost unique of West Europe

[Typo: in the maps M529, also known as L21, is wrongly written as M259. My apologies]

This is one of the virtues of Myres' paper (that I mentioned yesterday): that a somewhat more clear phylogenetic subdivision is made, emphasizing the difference between West European R1b1b2a and other R1b or R1b1b2, often blurred in previous papers, causing great confusion even to researchers themselves.

A defect is that instead of using a standard name for the defining SNP (L51/S167 per ISOGG) they chose to name it M412. However Argiedude says it is the same marker and I imagine it is. [Update: confirmed: the rs number indicates it is the same SNP].

Another virtue is that some of the substructure of R1b1b2a1 is also mapped, what really covers pretty well the Northern and Italian area of spread of this lineage and even some relatively unmapped areas of SW Europe, specially France.

In any case the apparent structure is curious, so I got the supplementary table S4 (supp. material is freely accessible) and made this map:

Click to enlarge

Notice that, following Argiedude, M412 stands for L51 and M529 stands for L21, what, if confirmed would make the following equivalences:

• R1b1b2a-M412 = R1b1b2a1 (L51)
• R1b1b2a1a2-M529 = R1b1b2a1a2f (L21)

Even if not confirmed, the equivalence should be approximate anyhow.

I'm sorry for the horrible color palette but it's my first attempt to make pie charts with Open Office spreadsheet gadget. Next time I'll do better I hope.

Notice also that I did not use all the samples, in the cases of small countries or less relevant regions I arbitrarily chose and discarded some.

Finally notice that pie charts represent only apportions of R1b1b2 and say nothing of the frequency of the lineage overall, which in most of East Europe and West Asia (excepting Turkey and a few neighbors) is extremely low.

The apparent structure of R1b1b2a1

The most apparent structure is, as we already knew the rather different R1b1b2a1a1 and R1b1b2a1a2 distribution. The first one (color coded brown and light blue) is dominant in the North and rather rare in the South - hence: 'North clade' for short hereafter. The second one (dark green, light green, purple and light orange) is by comparison not just more frequent in the South but also probably more diverse as well - hence: 'South clade'. However it is also found in the North.

Then there are some transitional "remnants": R1b1b2a1a* (L11) and R1b1b2a1* (M412). These should be informative (meaning some extra diversity at their structural levels) in order to infer the history of the haplogroup.

Per the hierarchical distribution seen here and diversity data from older works, the most likely origin of R1b1b2 as a whole is Anatolia.

Then R1b1b2a1* (M412) (yellow) is suggestive of a Mid-Danubian (or Italian or Iberian) coalescence.

R1b1b2a1a* (L11) (middle green) is suggestive of a West-Central European (or SW European) coalescence. More data on the Pyrenean region would clarify this maybe.

And, after this layer, comes the division into the widespread Southern and Northern clades mentioned before.

A reasonable interpretation is that the lineage traveled relatively fast upstream of the Danube (and/or via North Italy onto the SW), branching out then into the two major North/South clades. These two lower level lineages are in fact the two main stars of this demographic expansion.

My bet is that this represents a wave of colonization of Europe (when?) with secondary expansions from SW Europe (Franco-Cantabrian region possibly) and Central Europe (Rhine-Danube region I presume). There are several scenarios that can account for this, essentially Paleolithic (or pushing into the Epipaleolithic).

I don't see clearly how this structure could account for a Neolithic spread, really: no Mediterranean Neolithic pattern is apparent at all and Danubian limited expansion cannot account for any spread to SW Europe, certainly not South of the Loire and certainly not at the frequencies it is found there (nor in Britain/Ireland either). Claiming a Neolithic spread of R1b1b2 with this structure can only be done from a very shallow understanding of Neolithic archaeology and prehistory overall.

The main known demographic expansion we know of in the European Upper Paleolithic is the one after the Last Glacial Maximum, when Magdalenian culture expanded from the Franco-Cantabrian region, both northwards to Central Europe and, later, southwards into Iberia. This led to cultural divergence into the Epipaleolithic, with the important expansion into the newly available areas of the Far North, earlier covered in ice. Within the Epipaleolithic some further cultural flows are detected: from the Franco-Cantabrian region into Iberia (Azilian) and from somewhere in Mid-West Europe into the Southwest (Sauveterre-Tardenoisian).

Also, back to the LGM, Magdalenian techno-culture may have got a NW European ultimate inspirational origin but anyhow mediated by the warmer and richer Franco-Cantabrian region, where the culture flourished properly.

It's difficult to reconstruct in detail but, as far as I can tell, the two main North/South clades must have expanded in the Magdalenian period (one from the Franco-Cantabrian region, the other from Central-NW Europe itself) and also in the ulterior Epipaleolithic. Neolithic does not seem able to account for much but may have helped to shake the board a bit, specially in East-to-West direction.


Frequency maps

Selected frequency maps from the paper
Click to enlarge

Notice that the expansion of the South clade to the Atlantic islands does not invalidate its southern character and probably represents an Epipaleolithic-to-Neolithic spread.

Notice also the large amount of unclassified Southern clade in Iberia. The area around the Pyrenees was not really sampled in this study and therefore it is distorted by neighbors ("South France", looking more like SE France, Valencia and Cantabria).

In order to appreciate better the real thing in this aspect it's probably good to take a look at Cruciani 2010, who did bother to sample near the Pyrenees and gets maybe better (or at least complementary) maps illustrating the same problem.


Update: I superimposed (with complementary colors) the South (red) and North (blue) clades from the frequency maps above. However in order to account for the differences of frequency, I had to lighten the blue shade (North clade) because the scales are different. Take it as an "artist's impression" anyhow:



Update (Aug 27):

Here there is a hopefully better version of the map at the beginning of this post:

Click to enlarge

I put special care in giving each distinct clade an specific color range for easier visualization. All R1b1b2a1 (M412/L51/S167) seems to have coalesced in the Central-to-Western European area but the real expansion seems to have happened after this haplogroup split in two, which I dubbed the North and South clades.

And this is my reconstruction of the haplogroup expansion:

Click to enlarge
Color coded as above

Update (Aug 28):

Take a peek at the comment section, where I briefly discuss molecular clock difficulties and also the only possible Neolithic scenario for R1b1b2a1a2 (South clade): a massive demographic expansion in the context of Megalithism.

Rejecting or confirming this would require greater research in the structure hidden "under the asterisk" in SW Europe. At the moment only two minimally-sized sub-haplogroups are known: Basque/Gascon-specific R1b1b2a1a2b and sub-Pyrenean R1b1b2a1a2c (Gascon, Catalan, etc.). This alone gives highest structural diversity to the Pyrenean region, however most of the South clade remains unresolved (hidden under the asterisk), both in the Pyrenean area as in Iberia proper. And the key issue to solve would be if R1b1b2a1a2 is most diverse at the Pyrenees, what favors a Paleolithic spread scenario, or in West Iberia (and Brittany/West France), what would favor a Neolithic-Megalithic spread scenario instead.

Also it's maybe important to remind here the excellent STR work of Laura Morelli earlier this year, which was discussed in this article.

Importantly, this graph (annotated by me):


The graph is suggestive of the existence of another "West Asian" distinct haplogroup "under the asterisk" (that I labeled "R1b1b2a2?") and a possible Balcanic, rather than Anatolian origin for the R1b1b2 clade.

If so, this would correlate with the high diversity of the (much smaller) brother haplogroup R1b1a in the Italy-West Asia arch (as well as in Central Africa) and would suggest a slightly different origin and scatter for R1b as a whole (ref 1, ref 2).

A couple of new genetic papers I'd love to read

But by the moment just mentioning them briefly.

Both are published by the European Journal of Human Genetics (Nature) and both are mentioned by Diekenes today (link 1, link 2).


African genetic structure shows novel elements

There has been some negligence in mapping the genetic structure of African populations, with most papers taken a West African proxy (typically Nigerians), sometimes enriched with some of the last hunter-gatherers of the continent, to represent the whole complexity of the ancestral continent.

This paper seems to address this lack.

Martin Sikora et al., A genomic analysis identifies a novel component in the genetic structure of sub-Saharan African populations. EJHG 2010. Pay per view.

Abstract

Studies of large sets of single nucleotide polymorphism (SNP) data have proven to be a powerful tool in the analysis of the genetic structure of human populations. In this work, we analyze genotyping data for 2841 SNPs in 12 sub-Saharan African populations, including a previously unsampled region of southeastern Africa (Mozambique). We show that robust results in a world-wide perspective can be obtained when analyzing only 1000 SNPs. Our main results both confirm the results of previous studies, and show new and interesting features in sub-Saharan African genetic complexity. There is a strong differentiation of Nilo-Saharans, much beyond what would be expected by geography. Hunter-gatherer populations (Khoisan and Pygmies) show a clear distinctiveness with very intrinsic Pygmy (and not only Khoisan) genetic features. Populations of the West Africa present an unexpected similarity among them, possibly the result of a population expansion. Finally, we find a strong differentiation of the southeastern Bantu population from Mozambique, which suggests an assimilation of a pre-Bantu substrate by Bantu speakers in the region.

This Mozambican specificity was already spotted accidentally in Patin's paper on Pygmy Genetics last year but himself downplayed its importance because of his focus on Pygmy structure specifically.


West European R1b is distinct

Nothing really new for those who have kept a keen eye on the research of this Y-DNA haplogroup, the most characteristic of West Europe. But I'd still like to know more about the details. I doubt I could concur with the suggested timeline in any case.

Natalie M. Myres et al., A major Y-chromosome haplogroup R1b Holocene era founder effect in Central and Western Europe. EJHG 2010. Pay per view.

Abstract

The phylogenetic relationships of numerous branches within the core Y-chromosome haplogroup R-M207 support a West Asian origin of haplogroup R1b, its initial differentiation there followed by a rapid spread of one of its sub-clades carrying the M269 mutation to Europe. Here, we present phylogeographically resolved data for 2043 M269-derived Y-chromosomes from 118 West Asian and European populations assessed for the M412 SNP that largely separates the majority of Central and West European R1b lineages from those observed in Eastern Europe, the Circum-Uralic region, the Near East, the Caucasus and Pakistan. Within the M412 dichotomy, the major S116 sub-clade shows a frequency peak in the upper Danube basin and Paris area with declining frequency toward Italy, Iberia, Southern France and British Isles. Although this frequency pattern closely approximates the spread of the Linearbandkeramik (LBK), Neolithic culture, an advent leading to a number of pre-historic cultural developments during the past ≤10 thousand years, more complex pre-Neolithic scenarios remain possible for the L23(xM412) components in Southeast Europe and elsewhere.

M412 seems to be a novel SNP not yet reported at ISOGG. S116 (defined as its major subclade) used to describe R1b1b2a2, most diverse around the Pyrenees (unless this paper says the opposite). So I doubt the LBK hypothesis can hold, regardless of frequency.

Genetic flow across the Strait of Gibraltar

Researchers from the University of Geneva try to address the somewhat complex problem of genetic flow across the Strait of Gibraltar, between SW Europe and NW Africa, in a new paper:

Mathias Currat et al., Human genetic differentiation across the Strait of Gibraltar. BMC Evolutionary Biology, 2010. Open access.

They analyze a number of genetic variables such as Y-DNA, mtDNA, blood groups ABO and Rh, and other antigen/antibody systems (HLA, MNS and GM) and produce a number of statistical analysis for them. However they acknowledge that their modeling on continuous, rather than punctual, gene flow, may have some difficulties providing clear-cut answers.

Still they do produce results that are in general favorable for the P scenario (flow since 20,000 years ago), excepting to some extent the Y-DNA, which is probably a secondary sex-biased element overall.

Above: The scenarios are described in page 10 and in the supplementary material. P is "Paleolithic" flow beginning 20,000 years ago (Ibero-Maurusian or Oranian culture) among small populations, N instead is "Neolithic" flow among larger populations. PN is intermediate and PNI is intermediate but also considering the Islamic expansion (though details are not provided).

They also produce results of quite greater affinity within each region (SW Europe and NW Africa) than between them, as expected.


Notice in this graph (and also fig. 3) how autosomal markers' curves and that of mtDNA (MT-HV1, in blue) are very similar, however while mtDNA tends to lesser homogeneity within regions, it strongly tends to higher homogeneity instead across the Strait. This I interpret as meaning greater affinity across the strait in the Paleolithic, before the primarily male expansions reflected in the Y-DNA, altered the scenario.

Sadly no graphs for each of the two regions is provided, what does not allow us to make a more detailed assessment, like whether the Y-DNA expansion happened at both coasts or not, and, if so, if they were fully comparable or somehow different. I say this because the apparent SW European origin of North African mtDNA H (c. 25%) detected by Cherni 2008, and probably other haplogroups like V and K, would be consistent with a more European-like population before the E1b1b1 expansion in the context of Capsian culture (in North Africa) which is probably associated also to the expansion of Afroasiatic languages. However I see no reason to support a similar male-biased expansion in SW Europe.

Let the authors conclude:

Conclusion

While contrasted conclusions were obtained by previous studies based mostly on single genetic loci, our study clarifies the role of the Strait of Gibraltar regarding its permeability to gene flow. Indeed, our multi-locus approach led us to take into account variations between loci when trying to infer past history of human populations around the Gibraltar area. We were thus able to show that the Y chromosome on one side, and HLA-DRB1 on the other side constitute two extreme cases of very strong and very weak (respectively) genetic differentiations between populations across the Strait. The lack of genetic differentiation for HLA-DRB1 is particularly interesting because it can be explained by balancing selection (with a coefficient of selection estimated here to be around 2%). Given the huge worldwide dataset available for this locus, a better understanding on the mechanisms of selection at HLA loci could be very helpful to the study of human evolution, and more generally MHC. Our results obtained for Gibraltar have to be confirmed by further studies in other areas, especially where gene flow between populations is reduced. This work thus constitutes a step forward towards a better characterization of the combined effects of selection and demography on the genetic structure of populations, and especially on their genetic differentiation.

Ancient Danish mtDNA

There is a new paper with some potentially interesting information about hypothetical demographic changes in Northern Europe in recent prehistory.

Linea Melchior et al., Genetic Diversity among Ancient Nordic Populations. PLoS ONE 2010. Open access.

Most of the recovered data is from the last 2000 years and only three sequences are from earlier periods (two from Neolithic and one from Early Bronze). This limits somewhat the conclusions that can be reached from this paper alone but will be useful to complement the previous data and gradually draw a clearer picture of ancient genetics in Europe. Still they make some meaningful findings that I comment below.


Demographic replacement with Urnfield culture in Northern Europe?

The finding of haplogroup U4 and U5a among Neolithic Danes and yet another case of mtDNA U4 in the single Bronze Age successful sample, essentially discards the hypothesis of Neolithic replacement so far north (Bramanti 2009). It still remains as a possibility for Central Europe but this hypothesis is largely reliant on Epipaleolithic and Neolithic Era foragers' data from the Baltic area (essentially U5 and U4 as well) and a limited sample from Paleolithic Swabia.

The haplogroup frequencies for antique and historical Danes are very similar to present. What says that, if there was any replacement, this happened after the Early Bronze Age.

In this sense, I am considering more and more the possibility of a demic replacement in Northern Europe with the Urnfield culture expansion, because a similar situation is apparent in the Elbe basin, where the Corded Ware site of Eulau and previous Neolithic samples show mtDNA apportions quite different from modern (high K in Eulau, high N1a in Danubian Neolithic of East Germany, low H in all), while the nearby site of Liechtenstein, belonging to Urnfield culture (Late Bronze Age), already displays a very modern mtDNA pool, high in H and U.

Notice please that while this may apply to Northern Europe, it is certainly not the case in the South, specially in the Southwest, as well as in Morocco, where mtDNA H is found at modern frequencies more or less at all temporal layers since Late Upper Paleolithic. Sometimes partial data for Northern or Central Europe is happily extrapolated to the whole continent and this is very much incorrect.


The vanishing of haplogroup I

This is even more intriguing, specially because this paper findings (12.5% of haplogroup I) are highly consistent with previous data showing high apportions of mtDNA I in Denmark (all for the last two millennia), when now this haplogroup only amounts to 2.5%.

As I say, the rest of the samples for this period are totally modern but this identity is strangely broken when we consider haplogroup I, which has shrunk dramatically and nobody seems to understand why.

The authors notice that haplogroup diversity appears to have been higher in the past than today, which is surely related, but they are not able to propose a cause for this phenomenon other than drift.

Genetic comparisons of Basques and Jews

This past week I have been really oblivious to anything that was not directly related to the Zionist criminal assault of the Freedom Flotilla. I really did not feel that almost anything else mattered. Sometimes the tragedy of History just takes all the emotional and mental space and there is not really much room for anything else, I guess you can understand that.

However some more or less interesting stuff has been published in the meantime, in particular two papers on West Eurasian autosomal genetics which have been addressed at Dienekes' Anthropology Blog. So it's time to make a brief mention of them:


Basque genetics and other Europeans

Naiara Rodríguez Ezpeleta et al. High-density SNP genotyping detects homogeneity of Spanish and French Basques, and confirms their genomic distinctiveness from other European populations. Human Genetics. Pay per view.

See: Dienekes' relevant post for abstract, some details and his very questionable opinions, as well as for some discussion among readers.




Particularly questionable is his opinion that Basques should still be genetically identical to all other Iberians, based on one paper's Fst data but ignoring a whole array of other papers. These are (at least) Bauchet 2007 (discussed here), Achili 2007 (discussed here), Tian 2009 and Athaniasiadis 2010 (discussed here).

While Iberians are not sampled and plotted, based on the data of Tian 2009, they should show up very close to French and Italians (except the unique Sardinians), slightly above the (0,0) coordinates, in the graph above. Of course, it dependends on which Iberians but I'm talking about the main cluster, not Cantabrians nor Catalans.


Jewish genetics: yet another instance of truth avoidance

G. Atzmon et al. Abraham's Children in the Genome Era: Major Jewish Diaspora Populations Comprise Distinct Genetic Clusters with Shared Middle Eastern Ancestry. American Journal of Human Genetics, 2010. Pay per view (but should be freely accessible in six months).

See also Dienekes' relevant post.

Note: capitalized labels refer to Jewish subgroups. I strongly suspect, based on Fst distances (see second update below), that Palestinian and Bedouin labels are mistaken and should be the other way around.

It really frustrated me once again that Turks (nor Greeks nor Syrians) had been compared with. [note: see update below, this is not true after all and is quite revealing] The most logical hypothesis, based both on historical and the available genetic data, suggests that modern Jews (or most of them) are not so much descendants of historical Jews from Palestine as from a hybrid population that coalesced mainly in Asia Minor in Hellenistic and Roman times.

The above graph does not contradict this hypothesis at all, at least for the main Jewish population of Europe, Turkey and Syria. Irano-Iraqi Jews instead probably had different origins, the same that Yemeni, North African and Ethiopian Jews do as well (not studied here).

The affinity to Adigey (a NW Caucasian people) strongly suggests this, as well as the fact that they seem less Levantine than such an outsider population as are the Druzes (with Anatolian and Egyptian origins, not Levantine). Anyone familiar with European genetics would expect Greeks and Turks to pop up almost exactly where most Jews do.


Comparing with Bauchet-2007

But if you doubt my words, I suggest that you look at Bauchet 2007, which is the only paper so fare that has compared Jews (Ashkenazi) with Greeks and Armenians, resulting in almost total identity in all comparisons (k-means, PC analysis), as well as with South Italians. None of these groups are present in this analysis but they should show up pretty much where the main Jewish cluster is.


PC analysis of West Eurasians (fig. 4-A)

K-means clustering (fig. 4-B)

This paper remains to date one of the main and more clear references to understand European genetics at autosomal level and in particular to understand Basque and Jewish affinities or lack of them.

Together with other materials, some of them listed above, the case is clear:

• Basques are a distinct subgroup of Europeans, though probably akin to some North/NE Iberians and some South/SW French
• Iberians are also a distinct subgroup to the exclusion of Basques, even if they overlap at times (but not more than they overlap with Eastern Mediterraneans)
• Jews seem impossible to take apart from other Eastern Mediterraneans (Greeks, Armenians, South Italians, Adigey) but they seem somewhat distinct from Palestinians, meaning they probably coalesced in Anatolia. The fact that Zionist researchers systematically avoid comparing with Turks, Greeks, etc. probably means that they know this is true but want to hide it.

If you still doubt of the distinctiveness of Basques in relation to Iberians, I would recommend to look at this other PC graph from Achilli 2007, even if it is about mtDNA, where the case is very clear even with a very small Basque sample in an ocean of other European and many different Iberian samples. Iberians do look quite diverse anyhow but none of them really clusters with Basques:

Update: Jewish genetic affinity with Turks and Cypriots revealed in the supplementary material.

Someone has sent me a copy of Atzmon's paper and I must correct my previous statement that no Turks were sampled. In fact there is a small sample of Turks and Cypriots which cluster very closely with the main Jewish group (Euro-Turco-Syrian Jews):


I can't tell at the moment if the subgroup that clusters most closely are Turks or Cypriots or a bit of each but that is all what is in the ESE control group. There are also a couple of NW Europeans who also cluster very tightly and are most likely descendant of recently assimilated Jews (maybe in the time of the Holocaust, when some families found that such was their only hope of survival).


Update: Fst distances

Table 1 is pretty interesting, even if it lacks the Turkish/Cypriot sample, in particular the Fst distances. In it, it seems quite obvious that Iranian and, to a lesser extent, Iraqi Jews are different populations from the main Jewish cluster of this study, that I will call hereafter Western Jews (incl. Turkish, Greek, Syrian, Italian and Ashkenazi subgroups). The most representative subgroup of this "Western Jewry" of Hellenistic-Roman origins are probably Turkish Jews (labeled TUR in the study), followed closely by Greek Jews (GRK) and then Syrian, Ashkenazi and Italian Jews.

In order to simplify, I averaged Fst distances within this group, resulting in the following pairwise comparisons with other populations (bold type):

North Italian - 0.006 (0.004) [0.010]
Palestinians - 0.008 (0.005)
French - 0.009 (0.007) [0.014]
Druze - 0.009 (0.007) [0.009]
Iraqi Jews (IRQ) - 0.010 (0.009) [0.010]
Adygei - 0.010 (0.008) [0.012]
Sardinian - 0.012 (0.010) [0.017]
Bedouin - 0.013 (0.010) [0.009]
Iranian Jews (IRN) - 0.016 (0.014) [0.017]
Basques - 0.016 (0.014) [0.021]
Russians - 0.016 (0.014) [0.021]

Figures in regular brackets (...) are Fst pairwise distances with Turkish Jews (TUR) only, while figures in square brackets [...] are distances with Palestinians.

For reference, the average Fst distance between Western Jewish populations with each other is 0.005 (0.004 of Turkish Jews with the other Western Jews on average).

New paper on Spanish genetics

There's a new paper of some interest on Spanish population genetic structure:

J. Gayan et al. Genetic Structure of the Spanish Population. BMC Genomics. Open access.

The main aim seems to be to provide a Spanish dataset for population genetic research. And so far so good.

However the sampling strategy is awkward to say the least:

In the above map (my creation on the paper's data), red dots indicate sampling locations, while blue areas are regions (autonomous communities) not included in the sample. Most noticeable is that not just the Basque Country has been excluded but also all the surrounding area.

As I say, quite awkward.

Other surely distinctive unsampled areas are the Canary Islands and Galicia.

The whole design of the sample has a Castile-centric bias that is difficult to understand.

But, well, that's what they did. And, once we know that, we can go on to look at the results:

In this graph (fig. 6 annotated by me) we can see how Catalans and Andalusians tend to diverge from neutrality in orthogonal directions. To a less clear extent, the North Castilian samples (Arévalo, Segovia) also diverges somewhat.

We can say that PC1 describes a Catalan-other axis and PC2 an Andalusian-other one. The lack of distinctiveness of some geographically eccentric samples such as the Asturian one (Avilés)may well be caused by the small size of the sample. It is very possible that a PC3/PC4 graph would evidence some distinctiveness that is not apparent here.

Remember that PC graphs are merely bidimensional representations of some of the apparent structure, with all the limitations that this implies.


European comparison

When compared with other populations of European ancestry the PC graph is as follows (annotations by me on fig. 7):


Catalans appear to have some tendency towards Italy and NW Europe, while the less defined eccentricity of Andalusians only seems to tend towards Italy. There are also a couple of Castilian individuals who cluster with NW Europeans, maybe because the North Castile area sampled was the core of Visigothic settlement (just a hunch).


Global comparison

There's not much to say about Spaniards in the global scatterplot (fig. 8), really: all Europeans just cluster very tightly, the same as East Asians (Chinese and Japanese).

What I found intriguing and worth posting this graph is the curious coincidence of the scatter of Kenyan Maasai (MKK) and US African-Americans (ASW). Notice that the LWK sample (Luhya) are also from Kenya but Bantu and they cluster best with Nigerian Yoruba (YRI). Not really sure because it'd need further research but certainly the almost identical disposition of MKK and ASW samples is suggestive of the Maasai (and maybe other Nilotes) being somewhat admixed with West Eurasians. Alternatively the distribution might be reflecting some African-specific differences (just like Indians in the Eurasian context) and its overlap with African-Americans is to some extent an artifact of the limitations of PC analysis.

I do miss a comparison with North Africans, which seems to be a taboo in Iberian and European population genetic studies. However I do detect (and not only here) a slight "African" tendency among some Iberians which may well reflect a greater affinity with North Africans, in turn slightly more akin to ultra-Saharan Africans. This would be an interesting matter to analyze.

Update: I totally forgot to mention that just a few days ago I commented on another paper by a Catalan team that did compare Iberians and North Africans, which may serve for comparison. The results appear wildly different in the PC graph, with Catalans, Basques and Cantabrians clustering on one corner and the other Iberians scattered with a clear tendency towards the Eastern Mediterranean. No apparent North African affinity was detected.

New paper on Mediterranean genetics

We have a new free interesting paper by researchers of the Universitat de Barcelona on the issue of the Mediterranean partial barrier to genetic flow:

G. Athaniasadis et al., The Mediterranean Sea as a barrier to gene flow: evidence from variation in and around the F7 and F12 genomic regions. BMC Evolutionary Biology 2010. Open access.

The F7 and F12 genomic regions refer to a diverse set of neutral and functional polymorphisms located in and around the coagulation factor VII and XII genomic regions (F7 and F12).

Conclusions

As there is no consensus between the two genomic regions regarding gene flow through the Sahara, it is hard to reach a solid conclusion about its role in the differentiation between the two Mediterranean shores and more data are necessary to reach a definite conclusion. However our data suggest that the Mediterranean Sea was at least partially a barrier to gene flow between the two shores.

They compared a set of diverse populations from North West Africa, Iberia and Occitania (South France) and the Aegean area, using also samples from Ivory Coast and Bolivian natives as outgroups.

As the authors admit some uncertainty remains at not sampling some important Mediterranean regions. They mention Italy and the Adriatic but I'd say that a study like this would benefit from including mainland Greeks, other West Asians (Levant specially) and Egyptians, as well as some NW European outgroups (Scottish?).

I do welcome anyhow the sampling of Occitans, a much needed comparison from the viewpoint of Basque and Iberian genetics in particular.

From the study I have selected this PC graph (above) as most representative of their conclusions (fig. 3C). As I like to do, I have made some annotations in various colors: in red locations of the samples (so you don't have to check fig. 1 all the time), in gray the PC axes, in blue their possible meaning (my best hunch).

I have also marked two rather tight clusters (High Atlas and Pyrenees), both with a marked tendency to one pole of the PC2 axis, and two less clearly defined ones (Other North Africans and a mish-mash of Andalusians, Occitans and Cretans).

The High Atlas (or South-SE Moroccan) cluster is the most neatly distinct of all and shows very strong tendency in both PC axes. The African tendency (-PC1)>

The Pyrenean (or North-NE Iberian) cluster shares the same characteristics but with a clear European tendency instead of an African one.

The 'Other North African' cluster is really loose and is only defined because of the differences to the rest of groups, which should have been relativized if Egyptian and Levant samples would have been included most probably. Notice that the +PC2 pole, that I have dubbed 'Neolithic' just indicates an apparent more West Asian tendency, which might or not be Neolithic as such.

The Andalusia-Toulouse-Crete cluster is probably an artifact caused by their neutrality in the West Europe-West Asia (Paleo-Neolithic) axis rather than a true clustering. In any case, the clustering of the Cretan sample with SW Europeans is consistent in all three PC graphs what suggests that, at least in this aspect, Cretans are not so close to Anatolians as geography, history and haploid genetics suggests.

Also, the position of Asturians is very noticeable because they do not cluster at all with Pyrenean peoples but rather seem intermediate between the SW Europeans with a more Medeiterranean tendency (Andalusians, Occitans) and the Turkish sample. Knowing that in other markers, such as Y-DNA, Asturias tends to cluster with other West Iberian regions (Galicia, Leon, Portugal, Extremadura) one wonders if there was demic replacement in that country and, if so, when did it happen. Because it is something that is not obvious at all in the archaeological nor the historical records.

New genetic research on eye color in Europeans

There is a new paper on the genetics of eye color in Europeans that claims to have achieved the highest ever levels of correlation between genes (and other underlying factors such as age, somewhat relevant) and phenotype, reaching 49-52% of accuracy.

Fan Liu et al., Digital Quantification of Human Eye Color Highlights Genetic Association of Three New Loci. PLoS Genetics 2010. Open access.

The originality of the approach is that they did not just quantify "color" but two parameters: hue and saturation. In both cases (table 3) a SNP (rs12913832A) at the HERC2 gene is most important in eye coloration (44.5 for hue, 48.3 for saturation), followed at a fair distance by age (1.2 for hue, 5.0 for saturation). The rest of the factors (other SNPs, gender) only seem to have minor influence (<0.5%),>

Almost half of the underlying factors of eye color still remain obscure.

Finally some good research on R1b1b2!

This is a most important paper for those interested in Y-DNA R1b and specifically European and West Asian R1b1b2:

Laura Morelli et al., A Comparison of Y-Chromosome Variation in Sardinia and Anatolia Is More Consistent with Cultural Rather than Demic Diffusion of Agriculture. PLoS ONE 2010. Open access.

An excellent critique of Balaresque 2009, totally in the line I was favoring myself.

Most important is maybe fig. 2A, where the use of (novel?) STR marker DYSA7.2, in addition to the usual ones, unveils a clear structure within R1b1b2-M269. I have annotated it for easier visualization:

Click to enlarge

Duality

The already quite obvious Europe-Anatolia (A-B) duality of this haplogroup becomes terribly apparent: two different star-like structures connected by a less clear group (C) centered at the Balcans.

I imagine that sooner than later, someone will come up with a defining SNP for the more than likely Anatolian haplogroup. If I'm correct, with the current nomenclature it should be something like R1b1b2a2. However the root of the whole R1b1b2 haplogroup might also be in that cluster (or is it in C?).

It is surely important to notice that this presumptive R1b1b2a2 haplogroup (at least a clear haplotype star-like structure) also experienced quite apparently a rapid expansion of its own, mostly in West Asia and the Balcans, which is in turn surely related to the expansion of R1b as such through the Mediterranean, Africa (R1b1a) and Central Asia (R1b1b1).

R1b expansion pattern

We may well say that R1b expanded quite dynamically but not with total simple continuity but in colonization "bouts". The first bout is probably best represented by R1b1a scatter in the NE Mediterranean arch (two subclades) and around Central Africa (maybe with a Sudan origin, two subclades too). The second bout is surely a minor one towards Central Asia represented by R1b1b1 (typical of Uyghurs, an odd Chalcolithic founder effect in a previously desert area). The third bout is the spread of R1b1b2-M269 in West Asia and Europe, which represents the bulk of the R1b worldwide in raw numbers.

This third bout is actually two, as is evident in this paper (but was also evident before for those who can "read between lines" at least) but they may have happened in parallel (in which case my annotation above is more strictly correct) or they may have happened sequentially (in which case, lineage B would be R1b1b2a*, except for one of the branches that is actually the root: R1b1b2*).

A quite plausible reconstruction of the spread of R1b (c. 50-30 Ka ago?)
Click to expand

Synthesis

Whatever the case, two things are clear:

1. There is a neat distinction between R1b1b2(xR1b1b2a1) and R1b1b2a1.

2. R1b1b2a1 had its own distinct and fundamentally unique expansion in West Europe, sensu lato.

Also notice that, as happens in previous versions, there are many branches hanging from the R1b1b2a1 central node (modal haplotype also), which appears to contradict the known subhaplogroup structure that only recognizes two branches and which seems to be quite imperfect as of now.


Tempo

As you know I am not any fan of the molecular clock but it's worth mentioning the estimates of this paper for they totally contradict the Neolithic model: the pool of Sardinian and Anatolian R1b1b2 gets an estimate age of 32.6 Ka, the Sardinian one has an age of 27.0 Ka and the Anatolian of 19.6. But the regional ages may be misleading because they include members of both clades, however the overall age should be roughly correct within the molecular clock paradigm.

This would fit with a Gravettian expansion but could also fit with an even older Aurignacian one. It makes some good sense, specially as we are talking of a single expansion that replaced all what was there before, which could only happen at very low population densities, and also because it is a single expansion and not two. It also makes good sense with the likely pattern of expansion of its likely mtDNA companion haplogroup H (also V probably).

As the authors admit, the Magdalenian late glacial recolonization (and noticeable demic expansion) is also within the scope. However I find difficult to explain R1b1b2a1 in Italy within this context, as Italy remained Epigravettian until the very Neolithic. It is also much easier to replace all when there's almost nothing to replace, as was surely the case after the Campanian Ignimbrite supervolcano eruption of c. 40 Ka ago.


Other lineages

The paper is also of interest for other haplogroups such as E, G and J. However I fail to see anything clear surely because of the excessive emphasis in exploring Sardinian-Anatolian connections without a wider scope.


See also: for a completely opposite editorial opinion and likely heated discussion at Dienekes'.

R1b1b2 and R1b1b2a1 distinct STR diversity

I'm borrowing here the work of Aargiedude (again) because it's such a priceless addition that it really deserves to be paid some attention. If professional geneticists would be half as serious as this amateur, we'd know a lot more and a lot better about our the population history of humankind by now.

These two maps represent the haplotype diversity of R1b1b2a1 (ht15) and R1b1b2* (ht35) respectively:

Click to expand

Aargiedude's own observations at Dienekes' blog follow:

I've finished estimating ht15 and ht35 diversity, it's taken me 2 days to do this, and I think the results are amazing. The diversity clines of ht15 and ht35 are almost polar opposites. I think these results seriously call into question the conclusion of the Balaresque study, which is what prompted me to look into this. There's no surfing-on-an-expanding-wave phenomenon occuring with ht15. It has its lowest variance in the supposed origination point: Anatolia.

Instead, as Maju pointed out brilliantly, the study's tree diagram of their R1b1b2 haplotypes seems the result of 2 separate events, not a single wave diffusion. And he was absolutely right.

The diversity of ht35 doesn't form a gradually decreasing cline. It seems to be uniformly similar from Iran to west Iberia, or at least up to Italy, because there are issues with the validity of the North African and Iberian data (small sample size in one case and confusion with ht15 samples, in the other). Its cline seems to be more north-south than diagonally from southwest to northeast. East European countries have the same ht35 diversity as West Europe, with the special consideration of the west Iberian results.

Some technical details to keep in mind. Ht15 can be differentiated from ht35 by barely 2 markers: 393 and 461. Few studies test 461, so most of the samples I used were chosen on the basis of 393 alone. But about 3% of ht15 and 10% of ht35 have the "wrong" value, becoming confused with the other group. This usually doesn't matter, exceot in countries where there is an overwhelming ratio difference between the frequencies of both groups, such as in Iberia, France, Britain, Netherlands, Anatolia, and the Levant. In these extreme cases, I've included, where possible, 2 pair of results. The top pair uses samples predicted as narrowly as possible, by using both 393 and 461. The bottom pair is the standard prediction using just 393. The top pair should be more accurate, but they tend to lack in sample size, so then again, maybe not. Notice in the case of Iberia, that the less restrictive result changes drastically from the more restrictive result, and results in identical values to Iberia's ht15 diversity estimate, suggesting most of the samples are in fact ht15 samples that are being confused for ht35 because they had a mutation in 393 to the modal value of ht35 on that marker. Curiously, this didn't happen in France, where I was only able to use the less accurate method (393 alone), and yet the result is notably low and different from France's ht15 diversity. I'd seriously take North Africa's high ht35 result (0,30) with a military-issue teaspoon of salt, it's just 5 samples. On the other hand, it's notoriously high ht15 diversity (0,28) is pretty solid.

To recap, Baralesque and all geneticists are stuck in a time warp, they're back in 2003, thinking R1b is just R1b. What a waste, after going through all the effort of collecting and processing the samples, to not have had the sense to test for a few extra key mutations that define some major subdivisions of R1b1b2 and are well known for more than 5 years. The conclusions they reached would then have been very different.

The Balaresque paper was discussed at Leherensuge a few days ago.

R1b1b2a1: Neolithic or what?

I found at Dienekes the reference of a new paper on R1b1b2, focused on demonstrating (quite forcibly) that the lineage is Neolithic and not Paleolithic. Dienekes is, of course, happy that they chose to use his favorite (but rather disliked in the field) molecular clock methodology: the one based on the pedigree mutation rate, which make all haplogroups look extremely recent.

Patricia Balaresque et al. A Predominantly Neolithic Origin for European Paternal Haplogroups. PLoS Biology 2010. Open access. I'm provding a link to PubMed Central because at the time of writing this the original link remained broken.

I am not really happy: the pedigree rate can't be used for ages older than 5000 years (and all dates reported in this paper are clearly older) and, anyhow, the highly hypothetical molecular clock methodologies are anything but helpful when they become the main theme of a "research" paper.

I am not happy either because instead of using the already known SNP-based phylogeny of the haplogroup, they choose to treat the whole haplogroup as a single amorphous clade, when it is clearly structured. This isn't very helpful either.

Finally I am not happy either because they treat (again) the process of Neolithic spread in Europe as a single phenomenon, when it is in fact a complex array of various cultures, notably two different main vectors: one via the Morava-Danube and another via the Mediterranean coast. Both with origins not in Anatolia directly but in the Balcans. They ignore all these archaeological facts rather insultingly.

However for those who like to dig in the raw data, instead of just jumping to the too often biased and misleading conclusions, the paper still has some interest.

Notably I found figure 3 (haplotype structure) quite interesting. Here you have it with a crucial annotation for better understanding:

Click to expand

The crucial annotation is of course marking (with a dark red line) what is not part of R1b1b2a1, which is, as you can see a single branch of the star-like structure, but, unlike the others, is clearly not part of the fundamentally European haplogroup R1b1b2a1. Part of it is at the root of R1b1b2 and R1b1b2a but the rest of sub-branches must be derived, representing a distinct process centered in Turkey and nearby areas.

I dwelt on this matter in a previous post, so I'm not going to go all over it again here. Just for a quick reference a copy of the graph I posted then, showing major haplotypes and their relation with the various layers of the haplogroup:

Click to expand. Based on Alonso-2005.
DYS are 19-390-391-392-393.

As you can easily see most of the Turkish diversity belongs to the R1b1b2(xR1b1b2a1) part of the haplogroup structure. And sure I don't doubt that Anatolia or somewhere nearby is at the ultimate origin of R1b1b2. But there is a sharp distinction between that and what we find in Europe, which almost exclusively belongs to R1b1b2a1, a very specific sublineage.

And a sublineage that is very much ramified in a star-like structure, implying rapid demic expansion. When? That is not really the crucial issue as I see it. "Where?" should be the first question and a question that no paper has yet dealt with from the viewpoint of R1b1b2a1 on its own right.

In the past the lack of knowledge of the structure of the haplogroup may have served as excuse but not anymore. In fact any self-respecting geneticist should look at that SNP-based structure before dealing with STR-based haplotypes and take good notice of the distinctions.

And, if not, why not to include R1b1b1 (Central Asian) or even R1b1a (Italy and Africa mostly) and other R1b*? It is an arbitrary choice, poorly justified.

But, well, what do we get from this data set after we scrap off the extreme bias? For those who enjoy dealing with haplotypes in detail there is a long list in the supplementary material, which duly processed may provide very useful information.

I have not the time nor the resources to do that, so I have done something much simpler but also very informative: count the haplotypes by region as defined in figure 3 (above). Sadly France (incl. Basques), Germany, Netherlands, Denmark, England (incl. Cornwall) and Ireland are all dumped together in the category "other" ("West Europe" hereafter). Unlike the authors I do make the important distinction of what is R1b1b2a1 and what is not.

Follows the number of haplotypes by region and phylogenetic category (manual count so subject to minor error maybe):

R1b1b2(xR1b1b2a1):

• Turkey: 37
• West Europe: 20
• Iberia: 8
• Balcans: 2
• Italy: none

R1b1b2a1:

• West Europe: 198
• Iberia: 90
• Turkey: 13
• Italy: 11
• Balcans: 6

What does this say? That even between perfectly comparable regions such as Turkey, Italy and Iberia, the highest diversity for R1b1b2a1 is in the West. If you look again at figure 3 you'll notice that most Turkish haplotypes of this clade are derived from European ones, what implies back-migration after the formation and spread of R1b1b2a1, which must have happened in Western or Central Europe.

When did this happen? I am not sure but I have some things clear:

The structure of R1b1b2 does not correspond at all with what one would expect from a demic spread from the Balcans (not Anatolia directly) through two clearly distinct pathways, one to the Danube and Central Europe and the other through Italy to SE France and Iberia. Neither the Balcans nor Italy look particularly central nor we see two differentiated founder effects but only one.

Also the distribution of R1b1b2a1 in a cline that is, as the authors of this paper shamelessly admit, totally the inverse of what one could expect of a demic spread from SE Europe.

The structure of R1b1b2a1 in fact strongly suggests a spread from somewhere in the region described as "Other" and or Iberia, i.e. in Magdalenian Europe. This is in full agreement with the generally accepted theory that R1b1b2a1 spread from the Franco Cantabrian region after the Last Glacial Maximum, along with Magdalenian culture. It could have other explanations (Epipaleolithic flows, older Upper Paleolithic cultural dispersals like Gravettian or Aurignacian) but it just cannot fit within a Neolithic frame. No way!

How did it back-migrate to Anatolia? Possibly with the people who carried the rock art fashion to southern Turkey (Beldibi), which may have an offshoot also at Egypt, where some R1b1b2 is also found, as well as related haplogroup R1b1a. The exact process is still somewhat uncertain anyhow.

Sadly enough the authors have missed an opportunity to analyze the regional structure of this haplogroup in Europe. Hopefully someone else will eventually do it, helping to clarify the matter. The raw data is anyhow there for whoever wants to do it.


Update: a much more realistic geographic analysis of the diversity at the two different phylogenetic levels (by Aargiedude) can be found at this new post.

R1b1 origin: Italy or West Asia?

One thing I have been chewing on since I read the recent Cruciani paper on R1b1a-V88 is that, at least in that study, the region where R1b1-P25 has highest basal diversity (by haplogroups) are Italy and West Asia. Both regions have R1b1a, R1b1b and some other R1b1* (3/1173=0.25% in Italy and 1/328=0.3% in West Asia).

Another R1b1* was observed in the East Asian sample but this region lacks R1b1a.

The Italian R1b1* seems to belong (by haplotype) to two different subclades, what makes Italy provisionally a good candidate for the origin of R1b1. However notice that the Italian sample is much larger than the West Asian one, almost in direct proportion to the number of R1b1* individuals found.

Italy (incl. Corsica and Sardinia) also has high R1b1a diversity (by subhaplogroups: 2/4 basal sublineages, but lacks R1b1a*). This would suggest that the origin of R1b1a actually lays towards Africa, where basal diversity seems somewhat higher.

Of course it's a very tricky issue. Take with a grain of salt.

African R1b is distinct single haplogroup

I missed the relevant post at Dienekes and I'm not subscribed to the relevant magazine. Hence I'm posting on this matter somewhat late.

Whatever the case, a new subhaplogroup of Y-DNA R1b has been finally defined including all or most of African R1b1*. The defining SNP has been named V88. This lineage is found specially at the Nile area (more common in Sudan than Egypt and more in Upper Egypt than Lower Egypt) and in the Chad basin (Chadic speakers).

The authors conjecture an age of 9200-5600 BP for the haplogroup but I'd say that it should be rather from c. 16,000 BP, when rock art in the style of Europe and Anatolia (other regions high in R1b) is found in Upper Egypt. As always, molecular clock age estimates should be taken with extreme caution if not total disregard: it's much closer to pseudoscientific speculation than to empirical science.

Fluvio Cruciani et al. Human Y chromosome haplogroup R-V88: a paternal genetic record of early mid Holocene trans-Saharan connections and the spread of Chadic languages. European Journal of Human Genetics 2010. Behind a paywall.

___________

Update (Jan 13):

I finally put my hands on the paper (thanks to Vincent) and is quite curious. Anyhow it is worth mentioning first of all that the new phylogeny has already been added to ISOGG.

Second, it is also worth mentioning that this paper does not deal with Sudanese R1b1*, which is probably part of the new haplogroup but still awaiting confirmation.

Third, it is most interesting that, while the new R1b1a (V88) is most frequent in Central Africa (and secondarily North Africa, specially among Siwa Berbers, Egypt), two out of four basal sublineages are exclusively found in Southern Europe.

In detail:

• R1b1a* (V88) is common in Central Africa (0-95.5% depending on the ethnicity), with some presence in North Africa (0-23.7%, this last among Siwa Berbers) and also found in West Asia and the Balcans at very low levels (0.3 and 0.2% respectively of regional composite samples).
  
• R1b1a1 (M18), formerly R1b1a, is located in Corsica (0.7%), though is also known (older materials) to exist in Sardinia at high frequencies and, if my memory is correct, SE France as well.
• R1b1a2 (V8) is a private lineage found only in one Tali individual (4.5%, n=22). The Tali are a Niger-Congo (Adamawa) speaking population of North Cameroon that also shows 9.1% (two individuals) with R1b1a*.
• R1b1a3 (V35) is a very small lineage restricted to Italy (two individuals, one belonging to sublineage R1b1a3a-V7), where some R1b1* (P25) was also observed (3/1173).
• R1b1a4 (V69) makes up the largest subhaplogroup of R1b1a and is found with about the same distribution as R1b1a*, that is in Central (0-62.5%) and North Africa (0-4.9%).
  

Conclusions? Few and cautious. The haplogroup shows a striking distribution in two branches: one in Africa (R1b1a* and R1b1a4 primarily) and the other in southern Europe (R1b1a* at very low levels with R1b1a1 as main subclade almost circumscribed to Sardinia). Based on its European distribution, apparently unrelated to Africa, it should have a timeline dating to at least Neolithic times. It might be older in Africa (see above), though I'd be willing to consider a Neolithic timescale if a consistent archaeological pattern is provided as support. A Nile area origin is very likely for this continent (but see the haplotype structure, with Central African R1b1a* at the very center - yet looks as an artifact because of the many lateral branches leading to R1b1* and other control subclades).

______________________________


Adendum (Jan 15):

Ebizur has posted some nice complementary information for this lineage, specifically the M18 subclade, in the comments section. I find it so informative that I can't but copy here:

Haplogroup R1b1a1-M18 has previously been observed in Sardinia and Lebanon.

Peter A. Underhill, Peidong Shen, Alice A. Lin et al. (2000), "Y chromosome sequence variation and the history of human populations," Nature Genetics, Volume 26:

Sardinia
1/22 = 4.5% Haplotype 2(=A3b2-M13/M63/M127)

1/22 = 4.5% Haplotype 32(=E1b1b1c1-M34(xE1b1b1c1a1-M136))
4/22 = 18.2% Haplotype 35(=E1b1b1a-M78(xE1b1b1a3a-M148))

11/22 = 50.0% Haplotype 50(=I2a1-M26(xI2a1a-M161))
1/22 = 4.5% Haplotype 56(=J2a4b-M67(xJ2a4b1-M92, J2a4b2-M163/M166))
2/22 = 9.1% Haplotype 71(=F-M89(xH(1?)-M52/M69, I-M170, J2-M172, K-M9, M62))

2/22 = 9.1% Haplotype 100(=R1b1a1-M18)

Daniela Contu, Laura Morelli, Federico Santoni et al., "Y-Chromosome Based Evidence for Pre-Neolithic Origin of the Genetically Homogeneous but Diverse Sardinian Population: Inference for Association Scans," PLoS ONE, Issue 1, January 2008:

Cagliari (southern Sardinia)
3/187 = 1.6% R1b1a1-M18

Sorgono (central Sardinia)
5/103 = 4.9% R1b1a1-M18

Tempio (northern Sardinia)
0/86 = 0.0% R1b1a1-M18

Pierre A. Zalloua, Yali Xue, Jade Khalife et al., "Y-Chromosomal Diversity in Lebanon Is Structured by Recent Historical Events," American Journal of Human Genetics 82, 873–882, April 2008:

Lebanon
5/914 = 0.55% R1b1a1-M18
(seems to have been found in 3/104 = 2.9% Lebanon Druze, 2/432 = 0.46% Lebanon non-Druze Muslim, and 0/378 = 0.0% Lebanon Christian).

Anyway, according to presently available data, the maximum frequency of R1b1a1-M18 is found in the same population in which the maximum frequency of haplogroup I2a1-M26 is found: that of the central highlands (i.e. the so-called "Barbagia") of Sardinia. Barbagia is one of the most sparsely populated areas in Europe.

Update: I uploaded the paper HERE.