August 2, 2015

Large monolith found underwater near Pantellaria (Sicily)

A large human-made monolith has been discovered underwater in the Pantellaria shoal, submerged since the end of the Ice Age.

E. Lodolo & Z. Ben-Avraham. A submerged monolith in the Sicilian Channel (central Mediterranean Sea): Evidence for Mesolithic human activity. Journal of Archaeological Science, 2015. Freely accessibleLINK [doi:10.1016/j.jasrep.2015.07.003]

Other source: Pileta de Prehistoria

While it is hard to argue that the monolith is not man-made, as it has three holes of the same size in non-random positions, I would take with a pinch of salt the claim that the would-be standing stone or menhir has been there since the 10th millennium BCE, when the shoal was flooded by seawater. 

Instead I would consider the following scenarios as plausible:
  1. The land could have been at higher absolute altitude in the past and sunk because of local techtonics. It is, we must not forget, a very active geological area.
  2. The monolith could have just sunk when being transported on a ship of some sort between islands. The ship, made of wood and ropes would leave no obvious trace.
So I'd rather imagine the stone to have been produced in the Chalcolithic Megalithic context that has some relevance in the area, very especially the fascinating case of Maltese Megalithism, which spans between 3600 and 700 BCE.

July 14, 2015

Montenegro was part of the Dolmenic Megalithic phenomenon

Just read a most interesting article, with many beautiful images at Old European culture blog: the excavation of a tumulus at Danilovgrad showed it was not a Bronze Age Indoeuropean/Kurgan thing but a true dolmen (trilithon) and many centuries older than expected: c. 2400 BCE. 

There are thousands of similar tumuli awaiting excavation, most in the same rich area of Central Montenegro. This finding puts the Balcanic country (and probably also neighboring regions of the Western Balcans) fully within the Dolmenic Megalithic tradition in the late Copper Age. 

Also an intriguing bronze artifact was part of the grave goods, as well as zig-zag decorated pottery.

July 8, 2015

Zipf's law against 'Genghis Khan' sensationalism

Very interesting new short paper at BioRxiv:

Elsa G. Guillot & Murray P. Cox, High Frequency Haplotypes are Expected Events, not Historical Figures. BioRxiv 2015 (pre-pub, freely accessible) → LINK [doi:]


Cultural transmission of reproductive success states that successful men have more children and pass this greater fecundity to their offspring. Balaresque and colleagues found high frequency haplotypes in a Central Asian Y chromosome dataset, which they attribute to cultural transmission of reproductive success by prominent historical men, including Genghis Khan. Using coalescent simulation, we show that these high frequency haplotypes are expected simply by chance. Hence, an explanation invoking cultural transmission of reproductive success is statistically unnecessary.

Not surprisingly it is, once again, the hyper-sensationalist, hyper-recentist, over-simplifying and evidence cherry-picker geneticist Patricia Balaresque who is the object of these very legitimate criticisms.

The basic argument is very simple: in neutrality conditions haplotype distributions follow Zipf's power law, while a single-founder effect of the type of the alleged Genghis Khan one would never cause that: one lineage would be outstanding, while the rest would show no hierarchy. 

However the authors, to make their argument even more certain, simulated genetic data under the standard coalescent, a neutral model that does not include cultural transmission of reproductive success. As you can expect, the simulations confirmed that what surprised Balaresque and others is just absolutely normal by mere chance: no Genghis Khan effect ever took place.

June 21, 2015

Some improved knowledge of major R1b sublineage S116

While it is far from being the last word on the matter, the new study by researchers of the University of the Basque Country adds some important information to our knowledge of the main European R1b subhaplogroup S116, which dominates much of the continent with a south-western centrality, spanning from Ireland to Italy and from Iberia to Germany.

Laura Valverde, María José Illescas et al., New clues to the evolutionary history of the main European paternal lineage M269: dissection of the Y-SNP S116 in Atlantic Europe and Iberia. European Journal of Human Genetics, 2015. Pay per view (as usual supp. materials are freely accessible) → LINK [doi: 10.1038/ejhg.2015.114]

Preliminary status of the research

The improvement of the knowledge of this major European lineage had been stuck, as far as I know, since the various studies published in 2010, notably Myres et al. (discussed HERE). At risk of repeating myself, I will again display here some of the maps derived from the data of that key paper, as they are very useful references for discussing the new one:

Frequency of R1b subclades relative to overall R1b (per Myres 2010)
note: M529 is wrongly labeled M259
Composite image showing the overall frequency of R1b-S116 (red) and R1b-U106 (blue)
according to Myres et al. 2010

The new data

The new study is not as comprehensive in their sampling as that of Myres, so it heavily relies in previously published data, which is enough for the already studied subclades. The resulting maps are however somewhat different from Myres, because a lot newer Basque and Iberian data is present here (Myres did not sample Basques, whose frequencies and diversity for S116 are outstanding):

Fig. S1

They do however consider a third major sublineage of S116, defined by the mutation DF27, which is strongest among Basques and other SW Europeans. They also considered a sublineage described by the SNP L238 but only could find a single individual carrying it (a Breton from Brest), so it should be considered as part of the wider S116* paragroup and not relevant on its own. 

These are the results for DF27 and S116*, i.e. S116(xM529,U152,DF27):

An important detail is that, after excluding the three major subhaplogroups, the remaining S116* seems concentrated in Ireland and the Basque Country. However we should await for new information that could come from France, Southern Germany or even parts of England in the future (these areas showed some notable S116* in Myres 2010 but DF27 was not excluded then).

In this sense it must be mentioned that a sublineage of DF27 (SRY2627) has been known since more a decade ago, before even the modern nomenclature arose in 2001 (Rosser 2000 called it Hg22), and was indeed spotted not just among Basques but also among some Bavarians. So I would not dare to exclude at least some presence of DF27 further northeast than what this study shows. However it is indeed clear that its primary distribution is in Iberia and particularly among Basques. 

It is also important to underline that the maps may be a bit misleading because of the existence of two different Basque samples: a rural one (mostly with Basque surnames) with overall R1b and also S116* frequencies similar to the Irish and a urban one (of more mixed ancestry) with somewhat lower frequencies.

From the supplementary material I gather that Basques have the following frequencies (green: rural Basques, blue: urban Basques):
  • S116*: 16%, 8%
  • DF27: 71%, 51%
  • M529: 2%, 3%
  • U152: 3%, 1%
Otherwise the frequency of S116* is most notable among the Irish (18%) and Central-East Iberians (8-12%) but lower in Brittany (6%), Cantabria (6%) and Portugal (4%), being absent in Galicia and Asturias.

The frequency of DF27 is highest among Basques (63% on average, 71% for the rural sample) and then similarly high across Iberia (40-48%). It reaches 17% among Bretons and just 1% among the Irish. It must be noted that when you apportion DF27/S116, the result is similar through all Iberia (72-80%), Basques included (hat tip to Jean).

The frequency of M529 is very high among Irish (54%) and Bretons (52%) but under 5% everywhere else, except the following: 6% in both Asturias and in Cantabria, 7% among Galicians. The lineage is present in all sampled populations except Alicante and Andalusia. Therefore, inside Iberia, it shows some NW-SE clinality.

The frequency of U152 is under 10% across the board but also found in all sampled populations. The lowest ones are urban Basques (1%) and the highest ones Galicians and Asturians (9% and 8% respectively).

Note: figures above corrected (Jun 22) because there was a confusion between Cantabrians and Galicians in the first version of this entry. Thanks to Cousso for noticing.

Some conclusions

In the original entry I wrote here that I had found very striking a very sharp contrast between Basques, on one side, and Cantabrians and Asturians on the other. This was wrong because I committed an error in parsing to notes and confused therefore Cantabrians with Galicians. Hence the "sharp contrast" at the Western edge of the Basque Country is not so sharp after all, because Cantabrians act as buffer. There is still a curious contrast between Basques and the old Gallaecia province (later also Suabian Kingdom and Kingdom of Asturias-León), that is: Galicians+Asturians. These show the highest peninsular frequencies of M529 and U152, while Basques only have low frequencies of them; even more significantly maybe Basques have one of the highest S116* frequency in Iberia (just below the Irish in the overall sample), while Galicians and Asturians have none of it. In any case all this is reminiscent of the overall genetic contrast between West Iberia and the rest of the peninsula mentioned in other occasions, contrast that affects many haplogroups but not all.
[Paragraph edited: Jun 22].

The other conclusion is that, while we must await for further data, particularly from the French state and also from Southern Germany, the combination of this new data with that of Myres 2010 only ratifies me in my previous conclusions, which are:
  • R1b overall originated in West Asia, expanding in several directions from that region.
  • R1b-M269 (the main West Eurasian subclade) expanded from either the Balcans or Highland West Asia (Iran?)
  • The subsequent expansion in Europe (L23, M412 and L11 stages) is not too clear but could well have followed a double Central European and Mediterranean routes. More research is needed for these transitional stages between the Balcanic/West Asian phase and the Western European ones. 
  • Once in Western Europe, two L11 sublineages experienced parallel expansions:
    • U106 probably expanded from the Netherlands or Frisia (or maybe Doggerland in a Paleolithic scenario). Detailed research awaits however.
    • S116 surely expanded from somewhere in what is now France (possibly towards the Atlantic, judging on where S116* is most common), with three main subclades, each one following its own pattern of expansion:
      • M529 towards the Northwest (Brittany, Britain, Ireland...)
      • U152 towards the East (most notable in Switzerland and Italy, but also important in France, Germany and Britain, with offshoots of plausible Celtic transport in the Balcans and even Anatolia).
      • DF27 mostly to the South, peaking among Basques but also important in much of Iberia. It remains to be discerned how important it is in other European regions.
I put these notions on a map. It must be considered a rough sketch, a working hypothesis, because there is not enough data to be reasonably certain about all the details:

I would not dare to give tempos here. The sketched pattern of expansion can be equally consistent with a Neolithic or a Paleolithic modeling. The important pivotal role of France and the Netherlands could weight in favor of a Paleolithic model but it is true that aDNA and certain prehistoric reconstructions could allow for the French role (at least) to fit within an Atlantic Neolithic (Megalithism + Bell Beaker) theory for the expansion of S116 and I see no reason why the Netherlands could not have also played a similar role in NW Europe.

Thanks to Jean and Mike for the heads up.

Update (Jul 4): two "forgotten" papers of relevance:

1. George B.J. Busby et al., The peopling of Europe and the cautionary tale of Y chromosome lineage R-M269. Royal Society Proceedings B, 2011 → LINK.

Must read: demolishing (and well deserved) criticism of Balarasque 2010 and to some extent also of Myres 2010. They totally dismiss STR-based age estimates as wrong, misleading.

2. Rosa Fregel et al., Demographic history of Canary Islands male gene-pool: replacement of native lineages by European. BMC Evolutionary Biology, 2009 → LINK.

The anciente Guanche mummies' Y-DNA pool includes 10% R1b-M269. Considering that the islands were colonized c. 1000 BCE, I can only imagine that the Steppe Horde will find some way to blame the forgotten squire of Herakles for that. Or something...

Thanks to Georg for mentioning: that's the kind of feedback I love.

June 16, 2015

Alentoft 2015: more ancient DNA from beyond the Rhine

I've got some "friendly spam" insisting that I write something on this paper:

Morten E. Alentoft et al., Population genomics of Bronze Age Eurasia. Nature 2015. Pay per view (supp. materials are freely accessible though) → LINK [doi:10.1038/nature14507]

On the positive side, I think the study is much better quality than Haak's, with much more extensive and good quality supplementary materials, many of which are interesting. On the negative side, it totally lacks of any Western European samples, excepted the already known Epipaleolithic ones, being therefore pretty much useless to the understanding of the formation of the modern European genetic pool, except in a negative sense (stuff that is still clearly missing with the Central and Eastern references we have).

This huge blank in Western Europe, who nobody seems willing to fill up, would be quite apparent in the locator map except that it is poorly dissimulated by cutting a good chunk of Europe out of it:

from fig.1
What do we get in this paper then? Lots of info about Central Europe, notably the Carpatho-Danubian region, also about the Greater Caucasus, Siberia and historical Denmark. Add to that a couple of novel sequences from Bronze Age Montenegro and one from Estonia. 


It's interesting in its own way but I'm still demanding Western European ancient DNA. In fact this study (piling up with others) clearly evidences that the Western European, particularly Atlantic aDNA must be key to the understanding of the formation of European peoples. For example:

Something is amiss, right? Well, that is the LCT allele. Bell Beaker Blogger spotted it very well but he could not explain it. So I told him: There are no samples from Western Europe. That's why! 

Later I added:
Actually Chalcolithic Basques from the Ebro banks already had more rs4988235-T than any of the samples shown here: 27% overall with as much as 31% in San Juan Ante Porta Latinam.

Notice anyhow that T allele frequency is not equal to "lactase persistance" because, on this allele alone, it may be distributed unequally through individuals (and that's precisely what we see in the Chalcolithic Basque samples, suggesting two different populations). Also there may be other alleles producing the same effect, just that they are not that famous or even known at all.

I must say that the recent finding of similar mtDNA pools to those found in ancient Basques and resembling modern ones in the Seine basin, does reinforce the notion of Western Europe being key, not just for this allele but in general for the conformation of modern genetic pools in much of Europe.

It would seem (Mälstrom 2007) that the Gokhem people (Megalithic SW Swedes) also had some notable LCT frequencies, which are missing everywhere else (in Central-East and Mediterranean Europe, that is) before modernity. Hence the European LCT allele must have expanded from the Atlantic.


It's not the only this particular trait, mind you. For example, another detail that Bell Beaker Blogger spotted in this paper's data, is that the first known modern S116-derived is in a Bell Beaker individual. Not in Yamnaya, not in Corded Ware, not in Unetice... but in the Bell Beaker carriers that (culturally at least) came from the Southwest.

Specifically it is R1b-U152 (alias S28), which has sometimes been dubbed the "Celtic" subclade because of its distribution across the Alps, being particularly important in Switzerland and North Italy (including non-Celtic regions like Tuscany, Piamonte and Liguria, as well as Corsica). It has also some notable presence in much of France, SW Germany and Belgium. 

I already argued back in 2010 that R1b-S116 must have expanded from Southwestern Europe, possibly Southern France. No need to extend myself on this matter because there has been nothing new in all these five years (sadly enough). Just attach some old maps here for your convenience:

Relative prevalence of R1b-M269 subclades
note: M529 is wrongly listed as M259

Frequencies of the main European R1b subclades: S116 (red) and U106 (blue)


Hidden deep in the supplemental materials there is sup. fig. 6, which is an ADMIXTURE analysis of ancient and modern sequences. I simplified it by removing modern Asian and African samples as well as low K scores, all them pretty much irrelevant, adding more clear labels and rotating it:

The most interesting aspect is that this analysis pretty much gets rid of the exaggerated Yamna influence attributed by Haak, bringing it down to much more acceptable levels. Even Corded Ware peoples were, it seems now, only weakly related to Yamnaya and much more strongly related to Paleo-European hunter-gatherers (tan) and Mediterranean early farmers (yellow). Some of the Caucasus (teal) component was anyhow already present in pre-Indoeuropean farmers from Hungary and it is only the minor Siberian element (blue shades) what really marks the distinctiveness of Indoeuropeans. 

Notice that in this analysis, early European farmers like Stuttgart are not single color but dual: they appear as almost exactly a 50-50 mixture of Paleo-Europeans (tan) and Mediterranean farmers (yellow). The only (almost) true yellow reference are Naqab Bedouins, pointing again to a PPNA origin of the migrant farmers, who, after admixture with European aborigines, surely in Thessaly, spread the Neolithic through the subcontinent. 

When, at K=19, the Naqab Bedouins form their own distinctive component (pale yellow) the yellow component suddenly expands in all other samples at the expense of the rest. It changes meaning at that point, beware: it has become the Sardinian component, although one can well argue that this component is very much dominant among early European farmers (but always mixed with some Paleo-European, some Caucasus, some Bedouin even), there is not any single sample that is clearly dominated by it at >80% frequencies (visual estimate), only Sardinians score that high. Admittedly it'd be interesting to re-run this without Sardinians and see what happens. 

From the Basque origins viewpoint, I find notable that (again) the Basque and Swedish farmer sequences are similar all along. The latter have more yellow component however, implying that Basques are even more Paleo-European than they were. 

In contrast, in extended data fig. 6, Basques appear as less related to WHG than our immediate neighbors (Spanish and French), what is probably an artifact of Indo-European admixture in the latter, as Indoeuropeans were no doubt largely Paleo-European and the Siberian elements they carried may also weight in that same direction. This is also visible in the ADMIXTURE graph above: French clearly carry some more tan Paleo-European, in addition to the Caucasian teal (and in some K-values also minor Siberian blue shades).

In general we see an apparent increase of the Paleo-European component as we move away from the Neolithic but while in the French case this can be partly attributed to Indoeuropean flows from, ultimately, Eastern Europe, in the Basque case that is not the case. There must be another source of that excess Paleo-European element and that source must necessarily in Atlantic Europe. 

Time to do proper research, time to sample the ancient Far West.

May 7, 2015

Neolithic mtDNA from the Seine basin

Finally we get some ancient DNA from the French demarcation, which should be one of the focus of the research, because of the importance of the territory of the French state in European paleo-history since the depths of the Paleolithic.

This data set is, in spite of its limitations, most important because it seems to support the notion of Megalithism being an important factor in the formation of European populations as we know them.

I strongly recommend reading the whole paper because it does not only deal with the genetic aspects but also offers excellent background on the archaeological context of the region to which these (non-monumental) burials belong to.

Maïté Rivollat et al., When the Waves of European Neolithization Met: First Paleogenetic Evidence from Early Farmers in the Southern Paris Basin. PLoS ONE 2015. Open accessLINK [doi:10.1371/journal.pone.0125521]


An intense debate concerning the nature and mode of Neolithic transition in Europe has long received much attention. Recent publications of paleogenetic analyses focusing on ancient European farmers from Central Europe or the Iberian Peninsula have greatly contributed to this debate, providing arguments in favor of major migrations accompanying European Neolithization and highlighting noticeable genetic differentiation between farmers associated with two archaeologically defined migration routes: the Danube valley and the Mediterranean Sea. The aim of the present study was to fill a gap with the first paleogenetic data of Neolithic settlers from a region (France) where the two great currents came into both direct and indirect contact with each other. To this end, we analyzed the Gurgy 'Les Noisats' group, an Early/Middle Neolithic necropolis in the southern part of the Paris Basin. Interestingly, the archaeological record from this region highlighted a clear cultural influence from the Danubian cultural sphere but also notes exchanges with the Mediterranean cultural area. To unravel the processes implied in these cultural exchanges, we analyzed 102 individuals and obtained the largest Neolithic mitochondrial gene pool so far (39 HVS-I mitochondrial sequences and haplogroups for 55 individuals) from a single archaeological site from the Early/Middle Neolithic period. Pairwise FST values, haplogroup frequencies and shared informative haplotypes were calculated and compared with ancient and modern European and Near Eastern populations. These descriptive analyses provided patterns resulting from different evolutionary scenarios; however, the archaeological data available for the region suggest that the Gurgy group was formed through equivalent genetic contributions of farmer descendants from the Danubian and Mediterranean Neolithization waves. However, these results, that would constitute the most ancient genetic evidence of admixture between farmers from both Central and Mediterranean migration routes in the European Neolithization debate, are subject to confirmation through appropriate model-based approaches.

The studied sample comes from Gurgy (NW Burgundy, near Auxerre) and is very large: 55 successful SNP-defined haplogroups, 39 HVS-I sequences, including 27 distinct haplotypes. The burials are dated to the 6th millennium BCE, when the area was reached by Neolithic. The following haplogroups were found (table S1):

  • HV: 22, of which:
    • V - 2 (4%)
    • H - 20 (36%), of which:
      • H* - 6
      • H1 - 10
      • H3 - 4
  • U - 20, of which:
    • U* - 3 (5%)
    • U4 - 1 (2%)
    • U5 - 5 (9%)
    • K - 11 (20%)
  • JT - 8, of which:
    • J - 6 (11%)
      • J* - 4
      • J1 - 2
    • T - 2 (4%)
  • X - 2 (4%)
  • N1a - 3 (5%)

For some reason the total I get from table S1 is 51 individual haplogroups instead of the 55 expected ones. I have double and triple-checked and can't find the four missing sequences, sorry. Count corrected (May 9): there were indeed 55 sequences (my bad).

In any case the mtDNA pool is surprisingly "modern" with most haplogroups in very similar frequencies from what we would find in present day Western European populations. This is not at all like what was found in Germany's Neolithic, at least initially, characterized by low frequencies of H and high frequencies of presently rare haplogroups like N1a, being instead more similar in its "modernity" to what has been found in the Basque Country (see HERE for a quick reference).

This suggests that we have in Neolithic Europe the following regions, judging on the "modernity" of their mtDNA pools (only):
  1. Central Europe (Germany, Hungary): low H, clearly "pre-modern"
  2. Mid-Western Europe (France, Basque Country): normal H and roughly also other lineages, almost "modern"
  3. Portugal: seemingly very high H, "hyper-modern"

And this strongly hints again, along with the early presence of lactose tolerance among Chalcolithic Basques and the massive consumption of dairies among British farmers (of North French origin) to an Atlantic Neolithic origin of at least the bulk of the genetic pool of modern Western Europeans.

Sadly for the fans of patrilineal genetics, no single Y-DNA sequence could be produced.

Mostly Danubian origins?

Paradoxically these early farmers from Burgundy do not seem strongly related to the South, at least not judging by the North Iberian data used as reference but rather to the Danubian Neolithic peoples of Germany instead. This is most apparent in the haplotype graph of fig. S7:

Fig. S7 - Median-joining network

We can see that most Gurgy haplotypes (red) cluster with Danubian Neolithic ones (green) rather than with North Iberian (blue) or Paleolithic ones (purple). Basically there is only one clear exception: an H lineage that seems indeed more related to the South than to Central Europe but the red-green exclusive connections are much more common. 

However when analyzed statistically, the Gurgy population appears intermediate between the Central European and North Iberian ones. For example:

Fig 1. Principal Component Analysis (PCA) on the ancient mtDNA dataset.

Not sure if this is yet another warning about the limitations of statistical analysis or instead suggests that there is more importance of the southern ancestry but that it has a different origin in Occitania (SE France) that is not being taken into account. 

My best hunch is that the statistical result is product of the relative affinity to Basque Neolithic (the "modernity" of the overall pool as discussed above, Basque samples are by far the most numerous of Northern Iberia) combined by a more direct affinity with the German Neolithic in the detail of the sequences. 

Closer to Chalcolithic than to Early Neolithic populations

There is no haplotype structure to consider here but the statistical analysis that the authors perform does find that the Gurgy population was, oddly enough, closer to later populations in both Germany and Iberia than to their contemporaries. If this could be confirmed, we would have a candidate population for the origin of the changes that affected Europe (at least Central Europe) in the early Chalcolithic (prior to the Indoeuropean invasions). 

Fig 4. Pairwise FST distances.

It is apparent in the above figure that FST distances of the Gurgy population are much shorter (hence probably more related) with late Neolithic and Chalcolithic populations of both Central and Southwestern Europe than with any population that pre-dated the abandonment of the necropolis. Among these however it is Karsdorf the one most closely related, reinforcing the notion of a mainly Danubian origin, albeit a bit peculiar one (Derenburg, Halberstadt and the average "PRE_Central_F" are not particularly close).

But the most interesting part is surely the much greater affinity to the populations after the 4000 BCE chronological divide, which is also the baseline of the expansion of the Megalithic phenomenon. This matter requires more detailed analysis but it does suggest that the Franco-Basque area could well have been important in the formation of Chalcolithic and therefore modern European populations in the Western half of the subcontinent. A vehicle for this demographic "reform" should have been Megalithism, no doubt.

But we do need more data, sure we do.