Wolfang Haak’s team is again providing interested, yet unclearly representative, ancient mtDNA data from the Elbe region:
Before I proceed discussing the findings, which are interesting, I must warn of two caveats:
One concerns the use by Haaks et al. of only the HVS-I fragment of the mtDNA, which in many cases is inconclusive for haplogroup identification. In this sense, Bramanti
made a much better job in 2009 by checking also the control region for more clearly defining mutations which in some cases were decisive in clarifying haplogroup adscription. I had hoped this example of carefulness would set precedent, but seems not.
The other is the concentration of the new and old samples in a very small area of the Middle Elbe, which is totally unrepresentative of the wider Danubian Neolithic. This was a major issue in the previous research (2005
) and is aggravated by the new data coming only from Derenburg. I would highly commend trying to get more samples from further south, specially Hungary, Austria, Moravia and the Rhine basin, where the bulk of Danubian Neolithic (LBK) existed. Alternatively I’d commend to compare the Elbe group and the other Danubians separately, in order to discern if there are more or less important differences, as there is a very serious risk of the Elbe group being not really representative for the whole Danubian Neolithic peoples.
On the positive side, I applaud publishing this new paper as open access and I rather like the comparison with modern populations they made.
The new 22 Derenburg mtDNA samples, which make 27 after adding the five ones from the previous study, include three individuals each (13.64%) from each of the following haplogroups: H, HV, J, T, N1a and K. Additionally two individuals (9.09%) belong to W, while one is V and another U5a (4.55% each).
Y-DNA was successfully sequenced from three individuals, which make now the oldest Y-DNA sequences anywhere in the World. They are two F(xG,H,J,I,K) and one G2a3, very much marginal in the modern Y-DNA pools, confirming the outlier nature of these farmers and their likely low impact on modern Europeans, even in Central Europe itself.
Comparisons and analysis
Most interesting are the PCA analysis (fig. 2) and the genetic distances with modern populations, expressed on a map in fig. 3.
|PCA (fig.2) annotated by me
The most obvious conclusion is that Danubians, specially the hyper-mega-super-ultra-oversampled Elbe group, are total outliers in the context of all modern populations, throwing downhill the hypothesis of Neolithic replacement even for the Central European case.
The most similar populations are Highland West Asians (Anatolia-Caucasus) by PC1 and some Central Europeans and French by PC2.
It is also notable that an also quite suspicious “hunter-gatherer” aDNA pool (mostly from the Baltic area and in many cases actually peripheral Neolithic), labeled HG and dominated by U5 and U4 haplogroups, is much closer to modern Europeans than the LBK peoples. However their best modern match are Mordvins
(MOR), suggesting these may represent best the ancient NE European pre-Neolithic pool, at least mtDNA-wise.
|Fig. 3 distances of all LBK samples (A) and Derenburg (B) to modern populations
Warning update (Nov13): it is not the least clear how these maps come from the raw data. I can’t find any way in which Anatolia-Caucasus become closer than other regions.
Here it is interesting that the greatest affinities of Danubians are with Anatolia and Picardie, while they have low affinity with Northeasternmost Europeans (Finnic peoples) and then with Ibero-Aquitanians, Welsh-Cornish, Baltics (incl. Belarus), as well as Bosnians. They also have low affinity to the peoples of Arabia and Palestine.
The Derenburg sample looks most akin to modern Iranic peoples such as Kurds, retaining to some extent the Picardie connection as well.
This is overall strongly suggestive of a genuine founder effect in at least some Danubian farmer populations coming ultimately from the Taurus-Zagros area, rather than the Balcans or Hungary.
However their extremely low influence in modern populations still needs of a good explanation.
It is notable that the lineages that best define the position of Danubians
in the PCA are N1a, X, W and K. I was already quite persuaded that X and W, as well as the rare N1a, were of Neolithic arrival in Europe, but I was unsure about K. This seems to confirm, I understand, that K
, the best-surviving Danubian
-specific haplogroup, is of Neolithic origin and dispersal.
Haplogroup K has in general low frequencies nowadays (c. 6%) but is more concentrated in some scattered regions:
The other relative success history is T
, possibly T2. T is widely distributed but T2 shows an star-like structure sign of a sudden expansion, possibly in Neolithic times. K subclades also show sign of sudden expansion, in my opinion, about the same recent period.
As for the other Danubian
is thinly distributed, being most common in Northern Pakistan, finally N1a
is also very rare, with largest frequencies among Peninsular Arabians, some Croatian islands, mountain Ruthenians and Volga populations.
For N1a specifically, please notice that renowned geneticist M. Palanichamy concluded that LBK’s N1a variant is a native European clade, not a West Asian one.
The other lineages found are more normal among modern populations: HV
(xH,V) suggests West Asia as does X
. Instead H
are most common and probably original from Europe.
For further references in aDNA, visit Jean Manco’s excellent page. See also my (not yet updated) map series on European ancient mtDNA.
Important update (Nov 11):
Eurologist in the comments section makes what seems to be an excellent point: Haak and colleagues have, somewhat arbitrarily, divided the haplotypes in “informative”, “non-informative” and “unique” (table S4), on what they found their conclusions and the map of affinities posted above (fig. 3).
The unique clades are those that have no matches, fair enough, but the non-informative ones are the ones that have lots of matches. This is problematic to say the least, because they are happily excluding 59.5% of all matching haplotypes.
So they build the affinities map only on 40.5% of all matching sequences and this is surely not acceptable. Specially when Anatolia, Zagros and Caucasus happen to achieve only a poor 19.8% match overall, while several European regions reach well above 30%.
[Note: had to modify what follows because it contained errors].
When we ignore this arbitrary distinction, the best matching region is Wales+Ireland at 39.8%, followed by:
- >35%: Scandinavia
- 30-35%: Scotland, North-Central England, Portugal, Basques/North Spain, rest of Spain, N. France, mainland Italy, Italian islands, central Germany, South Germany/Austria/Switzerland, Poland and Volga Finns.
- 25-30%: Iceland, Shetland, other Scottish Islands, West/East England Finnland/Karelia, Low Germany, Baltic countries, South Russia Ukrainie, Belarus, Czech/Slovakia, Croatia/Slovenia, HungariyRomania and Balkans (minus Greeks).
- 20-25%: North Russia, Volga Turks, Greece and Morocco.
- 15-20%: all West Asia and Caucasus samples.
So is it the other way around to what Haak et al. claim? Probably the answer is complex rather than simple, as the founder LBK peoples must have been themselves. But it is clear that there are good reasons to think that aboriginal European lineages were already part of the Danubian Neolithic at its very foundations.
Best regional matches by ancient LBK lineage (red: “non-informative”, green: “informative”):
- J (069-126) (n=4): Wales/Ireland (30/500)
- H (311) (n=4): Volga Finns (24/500)
- H (CRS) (n=4): Italian islands (112/500)
- H (093) (n=1): W/E England and Scottish islands (7/500 each)
- V (298) (n=2): Basques/North Spanish (32/500)
- T2 (126-294-296-304) (n=3): Baltic countries (19/500)
- T (126-294-296) (n=1): Central Germany and Caucasus (10/500 each)
- T (126-294-304) (n=1): Cezch/Slovaks (7/500)
- W (223-292) (n=1): Finland/Karelia (28/500)
- K (093-224-311) (n=2): Caucasus (10/500)
- K (224-311) (n=2): Iceland (30/500)
- K (224-249-311) (n=2): North Russia (1/500)
- W (093-223-292) (n=2): Baltic countries and Anatolia (1/500 each)
- T2 (093-126-294-296-304) (n=1): Arabs/Cypriots (2/500)
- T2 (126-147-294-296-297-304) (n=1): Scotland (3/500)
- T (126-189-294-296) (n=1): Croatia/Slovenia (4/500)
- T (126-292-294-296) (n=1): Czechs/Slovaks (3/500)
- U5a (093-256-270) (n=1): Iran (3/500)
- N1a (147-172-223-248-355) (n=1): Czech/Slovaks (1/500)
- N1a (147A-172-223-248-320-355) (n=2): Iran (2/500)
- N1a (092-129-147-154-172-223-248-320-355) (n=1): Central/North England, Scandinavia and South Russia (1/500 each)
Additionally there are other four lineages (2 N1a, 1 H and 1 U3) with no known modern matches (“unique”).
Update (Nov 12):
Eurologist notices the following, comparing with Palanichamy’s paper (mentioned above):
Here are some more matches or close relations when comparing to the data listed in Palanichamy:
N1a (147A-172-223-248-320-355 (n=2): Iran (2/500): also a match each in France and US of European descent
N1a (086-147-172-223-248-320-355) – exact match with Portugal, and 1 Austria and 4 Switzerland very close
355) is relatively close to
(92-129-147A-154-172-223-248-320-355) – Italy, Norway, Germany, Russia