search this blog

Friday, August 12, 2022

Mediterranean PCA update

I updated my Principal Component Analysis (PCA) of Mediterranean populations with most of the ancient Jewish samples from the Waldman et al. preprint. To view intereactive versions of the plots paste the data from here into the PCA DATA field here and press PLOT PCA. The ancient Jews are labeled DEU_MA_Erfurt.

See anything interesting? I'm again seeing more complexity than claimed by Waldman et al., but what would I know anyway?

See also...

My take on the Erfurt Jews

Greeks in a Longobard cemetery


1 – 200 of 224   Newer›   Newest»
StP said...

Today my favorite family tree of the European populations haplotypes:

Revealing the recent demographic history of Europe via haplotype sharing in the UK Biobank
Edmund Gilbert et al., June 13, 2022
Edited by Marcus Feldman, Stanford University, Stanford, CA; received October 26, 2021; accepted April 21, 2022 119 (25) e2119281119

Michalis Moriopoulos said...

Thanks for this update! Glad to see it resurrected. I've been curious to see what it would look like now that we have many more samples since the Roman (or maybe Lombard paper) where I believe you introduced this idea. Should have some more with Southern Arc and Danubian limes that can be used here, too. Looks like it really does a good job of separating out the Jews from the other East Meds.

Davidski said...

I beat Feldman at al. by almost a decade.

a Gorilla said...

Your Cypriot samples do not represent the average Greek Cypriot. There are 659,000 Greek Cypriots in Cyprus and 5000 Maronites (Lebanese).

2 out of 8 of the Cypriot samples are Maronites (Lebanese) and 1 is mainland Greek mixed.

I've seen many of individual Cypriot samples and many other from papers and not a single one of them was like those outliers. 3 out of 8 of the Cypriot samples are not Greek Cypriot.

Davidski said...


Davidski said...

@Varkoume Ipervolika

These Cypriot samples are from Harvard's Human Origins dataset.

So if you can get in touch with the people who put it together and get them to change the labels accordingly, or get new Cypriot samples, I'll then update my PCA.

Aram said...

Thanks for the PCA.

This will be an offtop. But I think it will be interesting.

The new I2c2 (new name I2a2b) LBA/EIA sample is now available in Yfull.

The oldest I2c2 were found in European Neolithic cultures in Hungary , Romania etc. Chances are high that it was present in Cucuteni also. Most important is the fact the I2c found in Anatolia are negative to I2c2. Thus there is little doubt now that Lchashen culture I2c2 is derived ultimately from European Neolithic Farmers. Though an Eastern European forager origin can't be ruled out.

Simon Stevin said...


Hey David what’s the autosomal composition of SA6010 (Y-DNA: R1b-V1636*)? The Reich Lab changed his label from “Yamnaya_Caucasus” to “Russia_Caucasus_Eneolithic.” They claim he’s a relative of the Steppe Maykop sample SA6001, and that’s why they changed SA6010’s direct dating to that of SA6001’s. Is SA6010 basically Vonyuchka/Progress/Khvalynsk-like, or does he display a Steppe Maykop affinity the other WSH samples lack? What do you think about these changes by Harvard?

Davidski said...

Yeah, that's probably the correct decision, because that sample doesn't really look like a Yamnaya sample.

This also means that there's no genealogical relationship between Maykop and Yamnaya, which makes sense.

Assuwatama said...

Teach, any comments on maier et al 2022 findings

"..a model that was shown to be fitting for all Indus Periphery individuals modeled one by one by Narasimhan et al. (Ganj Dareh Neolithic + Onge (ASI) + West Siberian hunter-gatherers (WSHG)) was rejected for the grouped individuals with a p-value = 0.0044. In contrast, a model “Indus Periphery = Ganj Dareh Neolithic + Onge (ASI) + WSHG + Anatolia Neolithic” was not rejected based on the p>0.01 threshold used in Narasimhan et al. (p-value was marginal but passing at 0.03) and produced plausible admixture proportions for all four sources that are confidently above zero: 53.2 ± 5.3%, 28.7 ± 2.1%, 10.5 ± 1.3%, 7.7 ± 2.9%, respectively."

Assuwatama said...

From your Rakhigarhi blog

IRN_Ganj_Dareh_N 0.674±0.015
Onge 0.237±0.014
RUS_Tyumen_HG 0.090±0.012
chisq 14.877
tail prob 0.0212326
Full output"

I have absolutely no idea how come there was no Anatolian in 2019 but 3 years later there magically appears 5-10% Anatolian.

This potentially nullify everything that was said in 2019.

Rob said...

@ Aram
That I2c is “Neolithic” but in all probably arrived in Caucasus much later. So there is a north Balkan - Armenian link of sorts

Simon Stevin said...


Pardon me asking, but what does SA6010 look like in G25 or qpAdm?

Davidski said...

Very similar to Eneolithic steppe (Progress).

Michalis Moriopoulos said...

Thanks for the file, Dave. There's already a sample in the upcoming Anglo-Saxon paper (some people have converted the bams to K36) that will benefit from this analysis once its genome is available. One of them is an East Med sample from a period in Germany well after the fall of the Roman Empire and people on AG are arguing about whether it's likely to be Jewish or not.

Urki said...

Curiously the Spanish Clusters are closer to the north Italians than to the south Italians (les med easterners flow into Iberia?)

Davidski said...

Iberia is also closer to North Italy than to South Italy.

Aram said...


Yes. Later but no later than LBA. 1500bc. It's TMRCA is 3900 years.
Reich recently said that in Neolithic Armenia there is an increase of Levantine / Natufian shift, which confirms that farming was introduced in modern Armenia from Urfa region/ North Mesopotamia rather than NW Anatolia. Which further reduce the chance that I2c2 came in Neolithic.
In any case the two samples of I2c from NW Anatolia are _negative_ for I2c2-SK1271 level.
While those samples from Europe are positive


I3535 HAJE10a I2a2b* I-L758>M170>P215>CTS2257>L596>SK1271 I I2L596 Kolgeh 265599 0.287 0.355 PASS n/a LipsonNature2017 Direct: 95.4%; IntCal20, OxCal v4.4.2 Bronk Ramsey (2020); r:5; Atmospheric data from Reimer et al (2020) 5217-4997 calBCE (6170±40 BP, Poz-83632) -5118 -5217 -4997 Alfold_Linear_Pottery ALPc_III_MN Hungary_MN_ALPc Hajdúnánás-Eszlári út Hungary NA 4514

I2384 HAJE7a I2a2b* I-L758>M170>P215>CTS2257>L596>SK1271 I I2L596 Kolgeh 683331 1.569 .. PASS (X-contam=0.013; mtcontam=0.95) n/a LipsonNature2017 Direct: 95.4%; IntCal20, OxCal v4.4.2 Bronk Ramsey (2020); r:5; Atmospheric data from Reimer et al (2020) 5305-5046 calBCE (6220±40 BP, Poz-83631) -5164 -5305 -5046 Alfold_Linear_Pottery ALPc_III_MN Hungary_MN_ALPc_published Hajdúnánás-Eszlári út Hungary 0 0 4511

I11933_I11934_I11935 515, I11934,516, I11935,517*/ I2a2b* I-L758>M170>P215>CTS2257>L596>SK1271 I I2L596 YLeaf+pathPhynder 726484 HarneyCheronetbioRxiv2020 7,280-7,035 ybp -5207.5 -5330 -5085 Balkans_N Hungary Neolithic Polgár-Ferenci-hát, Hungary, 715/1022 Bickle and Whittle, 2013; Anders, 2017 Hungary 11001

I11902_I11903_I11904 s11902,477, I11903, 478,I11904,*/ I2a2b* I-L758>M170>P215>CTS2257>L596>SK1271 I I2L596 YLeaf+pathPhynder 732856 HarneyCheronetbioRxiv2020 6,300-6,050 ybp -4225 -4350 -4100 Balkans_CA Romania Chalcolithic Urziceni, Romania, Grave 27 Virag et al., 2006 Romania 11006

I2789 GEN_61 I2a2b I-L758>M170>P215>CTS2257>L596>SK1271>BY421 I I2L596 Kolgeh 564162 0.764 0.173 PASS n/a LipsonNature2017 Context: Archaeological period 3800-3600 BCE -3700 -3800 -3600 Balkans_LCA Protoboleráz horizon Hungary_LateC_Protoboleraz Abony, Turjányos-dűlő Hungary 0 0 4574

I3575 sepultura 44 individuo 1 I2a2b I-L758>M170>P215>CTS2257>L596>SK1271>BY421 I I2L596 Kolgeh 284048 0.314 0.115 PASS n/a OlaldeScience2019 Context: Archaeological period 400-600 CE 500 400 600 Iberia_EMA SE_Iberia_c.5-8CE Spain_Visigoth_Granada El Castillón, Montefrío, Granada, Andalusia Spain NA 9424

BERG103-2 I-Y1 I2a2b1h~ I-L758>M170>P215>CTS2257>L596>SK1271>BY421 I I2L596 BrunelPNAS2020 3943-3710 calBC -3826.5 -3943 -3710 France_MN2 BORS-Michelsberg Bergheim Saulager France

Davidski said...

Preliminary G25 coords for the new Anglo-Saxon samples...

Matt said...

Thanks Davidski.

Alright guys, using the population labels for the sample IDs (h/t teepean at anthrogenica), here's a datasheet (scaled) with some population IDs I've added:

Since the time of the samples is unclear, I've just gone with Country_Code+EMA2022+Region

Here's a West Eurasia PCA of all the samples:

The weird guys who are Mediterranean (of various sorts) are from the Germany North Rhine set and are IND010, IND003, IND006, IND016, IND009, IND011. There are some other guys who are odd from the England Lincolnshire and Yorkshire sets: S20648, S3056, S3011. There are probably some other more subtle outliers but these are the obvious ones.

Matt said...

Anglo Saxon/EMA Paper Preliminaries - Top Modern Day Population Matches for All Samples:

Comparing 245 England EMA to 247 England IA (by luck a complete balance in sample number), in terms of distance to:

Present-day England:
Present-day Wales:
Present-day Denmark:

(Preference for English vs Danish:

Oddly it looks like Iron Age "English" (British in England) are closer to English people today than are the EMA sample set, albeit these are preliminary coordinates not based on genotypes! The Danish are definitely closer to the EMA English than the Iron Age samples.

It does still seem odd, the idea that a Norwegian-like plus a French-like explains the bounceback to an England_IA like profile, but perhaps this is possible? The strong matching of samples to Norway and Denmark seems to leave not so much alternative?

A three-way model with FRA_Grand_Est, GBR_England_IA and GBR_EMA_England samples looks roughly like the paper is reported to have:

Rob said...

after quick look these A-S are heavily Danish_IA like, wiht some local (Britis_IA) admixture.
not sure about those pre-thoughts of continental Celtic-like admixture are needed

Rob said...

@ Matt
Right, you mean modern English require French-like input. Could be Normans and other continentals post Anglo-Saxon

Matt said...

@Rob, yes, all that's very plausible.

Here's what I get when I plot some sample averages:

It looks on North Europe PCA like you could model GBR_EMA2022_England best as between the Germany Mecklenburg samples (the most north shifted DEU set of samples, by far) and GBR_England_IA, but not the other DEU/NLD/DNK EMA populations with more southern and eastern shifted ancestry.

The average of NOR_SWE_DNK_IA works as well but this has small sample size.

This probably needs sifting out the outliers to get better results (e.g. I'm not sure if some of the population averages from Germany are south shifted due to the presence of outliers? DEU North Rhine certainly is).

Note that the samples here from this Early Middle Ages paper show the England average with 245 samples to be exactly like the average of the 8 Anglo-Saxons we've had for years! That bodes well for accuracy of these G25 positions.

Presuming that Early Middle Ages Mecklenburg / DNK_NOR_SWE_IA represents the incoming population, then England_EMA is about 65:35 between that and England_IA.

If we allow both Mecklenburg/Scandinavian_IA and England_IA/Irish_EMA, then England_EMA is about 50:50 between these two sets.

Would note though that the DNK_EMA set here from Copenhagen is significantly south and east shifted compared to the Iron Age samples, and much more like the Viking Age, so there is some understanding needed of why that is the case.

I think this will be clearer once we have Copenhagen's big Iron Age Danish set from "Population Genomics of the Stone Age" out so we can use that as a base; to me it looks like there are some problems with using these Medieval Age / Viking Age Danes in that they are shifted towards the Baltic and the south compared to the limited small Iron Age Scandinavia set, while the IA set has the problem of only getting decent sample size by aggregating a large area of space and time. The sites in Germany also possibly have some problems, in terms of just using their averages.

I don't think that North-Central European is a monolith (at least not judging at the region level) so it will need some looking at samples still to work out what the most plausible directly ancestral population is and hence what the most plausible admixture model is... Another alternative finer site breakdown that the big regional averages I've used might help too.

(We also might get a better understanding of this if some excellent person were to convert the data for the large set of Iron Age French that's here: . Or we could just wait until later in the month when Reich Lab will update the Allen Ancient DNA file).

Matt said...

One thing more to not, according to the leaked presentation showing sampling of the paper, the samples from Mecklenburg (North Germany) should be from Haven, which is early in their sample sequence (pre-Migration Period):

So maybe this explains why these would be a good ancestor population here. Many of the others look post-Anglo Saxon migration/later in the first millennium CE. A lot of this might sort itself out once we have the dates from the published paper.

Garvan said...

@Matt + All.
Do you know when or if we will learn more about the age of the samples? Or more details of burial? The Irish samples fall into two clusters + two outliers on Euclidean distances in PAST. It would be exciting to identify or even get hints on first or second generation Norman burials.

Rob said...

@ Matt , thanks for the summary & graphs

Simon_W said...


"according to the leaked presentation showing sampling of the paper, the samples from Mecklenburg (North Germany) should be from Haven, which is early in their sample sequence"

Makes sense; these samples don't look like they could be Slavic. But after the migration period the Slavic Obotrites were living there and the German colonisation began as late as the 13th century. If these are pre-migration age, they must be the Varini. Not exactly Angles or Saxons, but like them living north of the Elbe - their eastern neighbours so to say.

Simon_W said...

Among the samples from Lower Saxony, ADN has the most appreciable share of moderately southern shifted individuals (Belgian-like, Alsatian-like etc.). I suppose ADN means Anderten, a part of Hannover. The burial site of Anderten has been dated to the 7th-9th century. The pagan Saxons were subdued by Charlemagne between 772 and 804, so that could be the explanation.

Simon_W said...

Picked up from the Eupedia forum: IND seems to be the Merovingian burial site of Inden-Altdorf, essentially Frankish in character and in usage from the 6th-8th century. Judging from Matt's analysis, none of the individuals are Scandinavian-like. The northernmost ones are Dutch- and Afrikaner-like, and one is Orcadian-like. Overall not a very northern-like sample, and then these strinking southern outliers there, interesting.

Simon_W said...

Some more info about the samples from Germany and the Netherlands & tentative date inference from what I gathered by googling:
ADN = Hannover-Anderten, 7th-9th c.
DRU = Drantum (Emstek), carolingian
DUN = Dunum, East Friesland, 8th-10th c.
GRO = Groningen, 7th c. and later
HID = Hiddestorf (Hemmingen), around 530
SWG = Schleswig-Rathausmarkt, 12th-13th c.
ISS = Issendorf (Stade), 5th-6th c.
LBU = Liebenau, 4th-9th c.
MDM = Midlum
ZET = Zetel (East Friesland), 7th-9th c.

Tom said...

Waldemar posted this on Anthrogenica

Y-chromosome analysis of Goths from the Maslomecz group cemeteries in southeastern Poland

The Goths were an important part of European history from the 1st to the 8th century AD. For many years, various hypotheses have emerged among researchers regarding their origin and dispersal. Unknown is also the scale of assimilation of the local people they encountered during their migrations across Europe. The aim of this study was to determine the origin and genetic structure of the male lineages of the Goths from the Masłomęcz group who inhabited the Hrubieszów Basin (southeastern Poland) from the 2nd -4th century AD using Y chromosome. The material for analysis consisted of skeletal fragments collected from 43 individuals. Samples with the highest endogenous DNA content were sequenced directly to low genome coverage. For the others, after sex determination, we performed targeted enrichment with a custom panel of 10k Y chromosome SNPs. This allowed us to determine the Y chromosome haplogroup of 18 individuals.

A total of 14 individuals (78%) represents the Y chromosome haplogroups most closely related to the Scandinavian population. Thirteen individuals were classified into subclades of haplogroup I1, four to haplogroup R1a and one to haplogroup J2b. Haplogroup I1 currently occurs mainly among people living in Scandinavia. Ancient DNA analyses showed show that I1 has been present in the Scandinavian population since at least the Bronze Age. One of the individuals belonging to haplogroup R1a-Z284 belongs to a subclade found almost exclusively in the ancient and modern Scandinavians. The remaining four individuals belong to haplogroups R1a and J2b most probably represent the effect of assimilation of local people, met by Goths during their numerous war expeditions and settlement expansion.

Could the local Z280>CTS1211 actually represent a local Slavic population or did they belong to a pre-Slavic group?

Dave the Slothtopus said...

How did these [Saxon Paper] come so fast when the supplementary info isn't even out yet, but the 100AD - 400AD Klosterneuburg guys still aren't available?

Dave the Slothtopus said...

Re: My previous question about the Klosterneuburg guys -- Mind you I'm not angry, just curious about the mechanics of the whole thing and what type/suffix sort of files are needed for different kinds of analysis.

Rob said...

The “Polish Goths” ydna data makes sense
- R1a-Z284 is deeply related to BAx & seems to have persisted as a minority lineage in ‘Germanic’ groups
- I1 highly linked to Nordic/ Germanic pops . IMO represents a northwest European introgression in late / post BB phase
- R1a-Z280 : post -Luzatian groups which survived in central -south Poland
- J2b often mislabelled as proto-Illyrian, in fact is an “east Alpine pre-IndoEuropean” lineage . Presence in southern Poland by Kate Iron Age not too surprising

Davidski said...

I'm pretty sure that BAC was one of the main building blocks of Proto-Germanic, which is probably why Germanic shows such a close relationship to Balto-Slavic.

It's not yet clear why there was a I1 founder effect in Bronze Age Scandinavia.

Rob said...
This comment has been removed by the author.
Rob said...

Yes, one of ~ 3
The main shift in Scandinavia after the BAx period is toward the south-west, geographically speaking. To account this you first got the arrival of late Bell beakers into Scandinavia.
Additional connections between NWE & Scandinavia include gallery graves. There is an I1 individual in Neolithic France, & it might occasionally pop-up in other late Neolithic individuals from Northwest Europe

Matt said...

@Simon_W, forgot I had (in the file uploaded by teepean of anthrogenica, where I got the country and region from) a list of which specific areas/sites each three letter sample ID is associated with:

ADN - Lower Saxony - Hannover-Anderten
BUK - Kent - Dover Buckland
DRU - Lower Saxony - Drantum
DUN - Lower Saxony - Dunum
GRO - Groningen - Groningen
HAD - Cambridgeshire - Hatherdene Close
HID - Lower Saxony - Hiddestorf
HVN - Mecklenburg-Vorpommern - Haeven
IND - North Rhine-Westphalia - Alt-Inden
KIL - Roscommon - Kilteasheen - The Bishop’s Seat
KPN - Zealand - Sct. Clements - Copenhagen
LAK - Suffolk - RAF Lakenheath
OAI - Cambridgeshire - Oakington
POH - Kent - Polhill
SED - Norfolk - Sedgeford
SWG - Schleswig-Holstein - Schleswig Rathausmarkt
EAS - Kent - Eastry Updown
ELY - Cambridgeshire - Ely
ISS - Lower Saxony - Issendorf
LBU - Lower Saxony - Liebenau
MDM - Friesland - Midlum
SRS - Lower Saxony - Schortens
ZET - Lower Saxony - Zetel


Interesting info re; the nature of those sites (esp. Alt-Inden!).


Also, couple datasheets that might be useful:

1. A simple datasheet that lists the closest modern-day population per sample:

Summarises the above imgur links where I did this, but a spreadsheet format may be easier to look at this in a spreadsheet program of your choice, for filtering etc.

2. A more complicated datasheet that has the same as the above, but with some extra columns for two Vahaduo models I ran:

These are:

Model A which models in terms of deep ancestry proportions (Barcin, Yamnaya, etc), to detect outliers.

Model B where, if Model A showed the close enough deep ancestry for it, I modelled the rest of the samples with the average for IRL_EMA2022_Bishops_Seat_n=33 and DEU_EMA2022_Mecklenburg_n=8. This on the theory that these two averages are pretty close in deep proportions and should maximize the Celtic-Germanic specific kind of drift in this set as far as I can tell. Not a literal model at all, but meant to be another tool to quickly sort out which samples look more Continental North European rather than Insular Northwest European. Caution is this may still be influenced by deep ancestry proportions; it's just a quick look method.
Both of these models are not gospel when it comes to proportions at all, but should be a useful enough quick way to identify the more variable samples without having to eyeball them all on PCA. Just as a beginning for more nuanced exploration.

(E.g. from that sort of data, looking on PCA and setting filters - - might be able to identify which samples from England look closer to "most Saxon like"?

Quick West Eurasia plot of the samples that look like outliers from the general set: . IDN010 and S20648 seems particularly interesting as not very close to any particular modern population today as far as I could tell.)

History nerd said...

One hypothesis is a random effect of the clans controlling the bronze trade networks. The trade generated massive wealth for these clans, as we see in archaeology. From historical sources, these elites were closely related in a uniform culture. This contributed to the expansion of I1 in a relatively small population. Could it be substantiated by more samples and IBD sharing methods?

EastPole said...


“I'm pretty sure that BAC was one of the main building blocks of Proto-Germanic, which is probably why Germanic shows such a close relationship to Balto-Slavic.”

Germanic does not show a close relationship to Balto-Slavic:

We don’t know what languages were spoken by R1b groups of BAC, BB which were the main building blocks of Proto-Germanic. I only think that R1a groups of CWC were probably Indo-Slavonic speaking.

There was probably some Balto-Slavonic influence on Proto-Germanic languages according to some linguists:

It is not clear when it happened and how. Maybe Unetice, or CWC or..maybe something later like Suavi etc.:

DM said...


I think that was yet another pre-I1 but there is no BAM file to confirm or deny it either way. It doesn't seem to have left any trail because no I-M253 has been found outside the Nordic countries in samples from the Bronze Age yet. I read a theory that given the one pre-I1 found in SHG, I-DF29 may have come from a secluded SHG pop. All branches and the oldest samples are rooted in Sweden/Denmark so subclade diversification must have happened there. Will we ever find out?

Matt said...

The abstracts from EAA 2022 uploaded

Seems at least one has a poster showing full paper results - - "Population history of Early Medieval Ukraine"

Rob said...

looks like one of the 'Alans' from Saltovo has significant East Asian admixture

Copper Axe said...

@Rob and Davidksi,

If you look at Allentoft's pre-print, in the supplementary they model their new samples using genomed pooled together grouped together by way of IBD sharing. Relative to the BAC samples, the "I1 cluster" (4000-3000bp) have quite a bit more ancestry from the "Europe_NW" IBD cluster linked to high steppe SGC/Beaker derived people, they dont have ancestry from the more southern west/central European IBD cluster however. Also a bit of an interesting profile with steppe ancestry that isn't very high and some extra european HG ancestry (atleast on G25, Anglesqueville disputed this by way of qpadm iirc, modeling the samples from this period purely as BAC+TRB).

It would be interesting to see if the reverse is also seen in continental Germanic samples prior to the migration era to a similar degree, e.g an increase of ancestry from the Battle Axe associated IBD cluster in the iron age relative to the bronze age.

Copper Axe said...

"looks like one of the 'Alans' from Saltovo has significant East Asian admixture"
Is it actually an Alan or rather a Turkic individual (or mixed)? Because the Saltovo-Mayaki culture contained several ethnolinguistic groups. The Alans of the S-M culture are generally seen as Alans from the Caucasus mountains rather than the Alans of the N. Caucasian steppe areas, as those areas had been depopulated for a period (due to the Khazars?), and the "mountain Alans" were resettled by the Khazars in areas such as those around the Don, but also in Tatarstan apparently according to an archaeology article I came across once.

Matt said...

@Davidski/Simon, following might be of interest in terms of how these EMA paper samples are projecting on Global25.

It seems to me like there may be some problems or instability with how these a projecting on the dimension PC1 in G25, and some of the samples may behave OK on Europe-wide PC but be too compressed towards Africans on PC1.

This may be influencing why Afrikaner is getting picked as closest population for some samples, since it resembles a Dutch population with small African/Cape Coloured admixture.

For some evidence, here's using Vahaduo's distance difference method (Difference AC-BC) on some population averages from this set, compared to modern populations:

Where I compare the distance difference between England_EMA (this paper - 245 samples) and England_IA (Patterson and Isakov - 247 samples), then you get a very sensible result where the populations peaking being relatively closer to EMA are Scandinavian and Baltic while those closer to IA are Basques and SW French.

Similarly, the comparison between England_EMA and Denmark_EMA is sensible enough, with Denmark_EMA, albeit based on 14 samples only, showing an affinity to Southern Slavic people.

However, some other comparisons are less stable and logical; England_EMA vs IRL_EMA (33 samples) peaks at Iranians being closer to IRL_EMA, which makes little sense, the comparison of England_EMA vs Groningen_EMA shows England_EMA closer to Native Americans, and against Mecklenburg_EMA and Lower_Saxony, the Mecklenburg and Lower_Saxony samples are closer to Africans. I'm not sure this can be explained by outliers, as it seems that when I look at the samples that aren't outlying on the Vahaduo PCA, they look to have compressed values on PC1.

E.g. values like 0.105 or 0.113 which are not normal for North Europeans without African admixture.

It looks like these problems average out in the big average for the England_EMA samples and compare it to the big average for England_IA, but are shaky when it comes to averages based on even 49 samples.

I don't think this would change who the outliers are, but things like this may cause attraction to low levels of non-European ancestry in models.

Here's a quick check using Vahaduo's modelling that shows some samples picking up African admixture that's probably an error: (stuck this list in a pastebin here: )

If it's not fixable by any obvious means, we may need to wait for the genotypes to really be sure?

Davidski said...


These samples may not have been genotyped accurately or they might be unusually noisy due to a lack of UDG treatment.

So yeah, we have to wait for the official genotypes and see how they behave.

Rob said...

@ psikh

There was said to be a pre-I1 in Staro Forvar Pitted Ware individual ~ 3000 BC. However, all other TRB, Pitted Ware, HGs from Denmark, Swden and Norway are I2a..
Also there was a reported I1 in an LBK individual from Hungary, but based on RFLP methods and not carbon dated, so not sure how reliale that is.
Overall, however, data points to SW European *deep* origin of I1, but extant I1 spread from the Nordic region. It seems I1 wasnt particularly prevalent before the NBA, and it has nothing to do with 'steppe migrations' (as some people say) because it was uncommon even in the Mesolithic.

As others have pointed out, the Allentoft paper (in fact we discussed on Eurogenes several years earlier), the LN-BA Nordics show a different kind of introgression than SHG/ WHG (as seen in BA Baltics, for ex), thus i favoured a NW European source for I1 over ''local HGs'; but anything remains possible

Curiously, as Iain McDonald has sketched out, R1b-U106 probably arrived via central Europe, so what is termed 'NW European' is somewhat qualitative, and there are limits to the IBD method

@ Copper Axe
Yep might be a Kazar

Puree said...

In relation to the 'Middle Eastern' portion of the Erfurt samples, it would be nice to have more Tigris-Euphrates G25s for the period from 500 BCE- 1000CE, since this is an area where the Jewish Diaspora thrived prior to migrating elsehwere. Davidski recently posted some Syrian samples dated to about 728 CE which are a great addition to the dB (SYR005 and SYR013), but I believe more would be helpful vis a vis Erfurt, and European phylogeography more generally.

I believe the site potential is there because I recently came across a researcher at the U of Warsaw who has published several papers involving various Mid-Euphrates and other interior Middle Eastern sites (for example Tel Ashara and Tel Masaikh) which include sample sub-sets from Old Babylonian (1800-1600BCE), Neo-Assyrian (600-1200CE), and Early Islamic (600-1200CE, in this latter group are samples R3359 and R3361 from about 700 CE and 1200 CE at Tel Masaikh, which were discussed in this paper, but unfortunately the work is about paleopathology, carbon and nitrogen ratios, dental conditions, etc. (;; I wonder if maybe there could be more BAM files made available through this work somehow. Soltysiak's work shows there is an untapped potential for autosomal sampling of ancient Tigris-Euphrates remains.

DM said...


The Stora Forvar pre-I1 sample is Mesolithic not Pitted Ware. You can read about it in the Allentoft supplementary. I'd agree that anything is still possible. As Copper Axe pointed out the LNBA Scandis don't have any ancestry from the SW European IBD cluster so a late entry might not be very probable.

weure said...

@Eastpole "maybe something later like Suavi etc.:"

That's the clue. The bunch of Saxon pirates that began the rule the world around the North Sea coasts were mostly rooted among the most northern outlier Suebi, from the "bottleneck" (nowadays Schleswig/ border Denmark) like the Angli, Warini etc bounded by the same Nerthus cult (see Tacitus) and a language that founded English and Frisian.

Not only in England but also in the Netherlands we see that the 'Saxon genetic cline' is still very recognizable and is differentiated from the Frankish one.

Rob said...

@ psikh
Yep by deep ancestry I meant Palaeolithic or something. Obviously the post-bottleneck I1 is from somewhere near if not in Scandinavia

Rob said...

to add, for some reason I had that SF dated to ~ 3000 BC, but it is in fact ~ 5200 BC. This makes a big difference, as the chronologically closest I1 is in continental Europe. The I2a-rich PWC expanded across eastern Scandinavia post 3000 BC, with a distnctive CCC adstrate in GW-ancestry. Definitive links between France & Scandinavia in the immediate pre-NBA period (Gallery graves, Grand-Pressigny flint, etc)

archlingo said...

Dear bloggers,
it is very sad that the manifold competence (not only) in this blog exhausts itself exclusively in empty, arrogant and blasé assertions, predominantly, without the slightest effort around any concrete reasoning, means: Time. Place, aDNA. No one is served by pure opinions. And neither with mutual unsubstantiated insults. I find that very unfortunate.

Arza said...

Three Assays for in-Solution Enrichment of Ancient Human DNA at More than a Million SNPs

In-solution enrichment for hundreds of thousands of single nucleotide polymorphisms (SNPs) has been the source of >70% of all genome-scale ancient human DNA data published to date, making it economical to study ancient samples with low proportion­­­s of human DNA, and increasing the rate of conversion of sampled remains into interpretable genetic data. So far, nearly all such data has been generated using a set of bait sequences targeting about 1.24 million SNPs (the ‘1240k reagent’), but synthesis of the reagent has been cost-effective for only a few laboratories. In 2021, two companies, Daicel Arbor and Twist BioSciences, made available assays that target the same core set of SNPs along with supplementary content. We test all three assays on a common set of 27 ancient DNA libraries, and show that all three are effective at enriching many hundreds of thousands of SNPs. For all assays, one round of enrichment produces data that is as useful as two. In our testing, the “Twist Ancient DNA” assay produces the highest coverages, greatest uniformity on targeted positions, and almost no bias toward enriching one allele more than another relative to shotgun sequencing. A caveat is that the 1240k and Twist Ancient DNA results were based on protocols we optimized in our lab, whereas for Daicel Arbor we used the manufacturer’s recommended protocol at the time of our testing, and thus our results for Daicel Arbor should be interpreted as a lower bound on performance that leaves substantial room for further optimization. We also identify hundreds of thousands of targeted SNPs for which there is minimal allelic bias when comparing 1240k data to either shotgun or Twist data. This makes it straightforward to co-analyze the large datasets that have been generated using 1240k and Twist capture, and shotgun sequencing approaches.

Contains unpublished samples, including UP from Georgia

Arza said...

Also new dataset is coming:

Upcoming versions:
V52.2: expected Aug 2022

Matt said...

@Arza and all, this is the preprint we discussed earlier in the year here (, so I'll just reproduce my comment from that:

"@all, this new paper "Three Reagents for in-Solution Enrichment of Ancient Human DNA at More than a Million SNPs" - - (h/t anthrogenica) from Nadine Rohland and Shop Mallick and others, is following up from a talk Reich gave at ASHG 2020 (

It discusses the use of two alternative reagents which can be more cost effective and reduce biases between co-analysing shotgun samples and in the case of one, which would reduce biases in inflating differentiation between African samples with each other and non Africans.

This latter one is possibly being used to reduce African biases in the paper Arza mentioned as having an abstract on the ENA - "Ancient DNA and deep population structure in sub-Saharan African foragers" - ?

By the way, the data from this paper that Arza mentioned is published on the ENA (or most of it is), so if any brave souls want to map it to G25, now is the time!

The new paper by Reich group "Three Reagents for in-Solution Enrichment of Ancient Human DNA at More than a Million SNPs" provides a list of the 27 libraries (from 26 samples) they've tested this on, and I've taken out the IDs and checked which ones are published -

Slightly under half are unpublished. Excitingly one of the samples used is Dzudzuana2, so that means they probably intend to publish the Dzudzuana samples soon!"

I was clearly premature in thinking they'd publish a new version of the Allen database including Dzudzuana... 7 months later...

If the sample S2949 can be converted, that is the critical sample!

The others which are unpublished are: S20720, S20721, S21299, S20703, S8432, S2818, S11857, S4795, S1965, S4532, S1961. (The PCA positions are suggestive that some of these may be related to the steppe ancestry?)

I'm guessing these will be in the new version of the Allen database which will most likely be dropped with the new Near East (and hopefully Early Middle Ages?) papers on Friday, but if any enterprising soul wants to get a jump on it...

(Arza, thanks for posting that up! Hope all is still well with you and we can have some good conversations when more data relating to historical and Iron Age East-Central Europe comes through.)

Wise dragon said...

Hi Davidski, one poster on Eupedia has leaked from Reich Southern Arc Paper this info. 16 new samples from Hasanlu in the northwest of Iran, all of them R1b with no EHG ancestry, and thus with no Steppe ancestry. The samples are all R1b Z2103.

Interesting finding. Plus on 26 August all 3 papers will be released.

Matt said...

Re; EAA abstracts, this one:

"New radiocarbon dates from the Usatovo culture suggest a re-evaluation of the culture’s chronology"

"Usatovo is one of the kurgan cultures of Eneolithic southeast Europe, occupying a strategic crossroads between the eastern Balkans and the Ponto-Caspian steppe. Usatovo is considered to be related to the farming cultures of southeastern Europe as well as to the pastoralist groups of the North Pontic. Usatovo currently dates to the second half of the 4th millennium BCE, based on C14 measurements obtained on pottery and charcoal using the scintillation method, as well as through ceramic typologies. Recent AMS radiocarbon dates obtained on human remains from the Usatovo site of Mayaki fall within the second half of the 5th millennium BCE. Stable isotope values (δ13C and δ15N) obtained on human remains are atypical for farming or hunter-gatherer communities of southeast Europe and imply a heavy input from freshwater and/or marine resources in diet. Stable isotope ratios suggest the presence of Reservoir Effect (RE) potentially offsetting the dates from the human remains, although its quantification is difficult to establish due to the lack of contextual faunal remains. Even with the RE taken into account, the discrepancy between the scintillation C14 dates/ceramic typology and absolute AMS dating may still remain. Both anthropological and ancient DNA data are consistent with Usatovo being an admixture of farmer and steppe ancestries. The timing of this admixture may help refine the Usatovo chronology."

Reich and Patterson are contributing authors. I think this will possibly include the publishing of the genetic family relationship between the Steppe Maykop person (second half of the 4th millennium BCE) and the Usatovo person.

Davidski said...

@Wise dragon

It's bullshit.

Those new Z2103 Hasanlu samples are from the Iron Age and they do have some steppe ancestry.

Also, there's steppe ancestry in Bronze Age western Anatolia.

Rob said...

The “emperors” at shortmanPedia have some bizarre theories

Assuwatama said...

Do you have any sample from Afghanistan 2900-2400bce?

Matt said...

England EMA paper again:

1) The individual sites plotted on G25 PCA:

There is quite a strong cline variation for all sites, including Continental North Europe. So it would be difficult to use any as a single source for anything I guess. The CNE sites don't look less dispersed than the England ones, though I have not run anything formally to check dispersal (ST.DEV / distance from average).

As mentioned the Haeven site comes close to being a tight cluster it seems.
Though interestingly samples HVN003 and HVN002 seem to show some Balto-Slavic related shift, so possibly this fits with the potential dating at pre-Migration Age and their location where Simon notes, "But after the migration period the Slavic Obotrites were living there and the German colonisation began as late as the 13th century.". Although this is relative to the being extremely "North-Western" on that PCA to begin with compared to the other sites.

2) I used the data in PAST4 to make a simulated population at the end of the cline of the English_EMA samples: (pastebin of coordinates:

It's designed to represent an average of one end of the cline, though it can't encompass all the samples. It has to be placed NW on North-Europe PCA of say, present-day Norwegian average, to work.

Perhaps the genotype files will be more accurate, have better projection and smooth some of this out.

Dave the Slothtopus said...

"As mentioned the Haeven site comes close to being a tight cluster it seems.
Though interestingly samples HVN003 and HVN002 seem to show some Balto-Slavic related shift, so possibly this fits with the potential dating at pre-Migration Age and their location where Simon notes,[snip]. Although this is relative to the being extremely "North-Western" on that PCA to begin with compared to the other sites."

Funny that HVN002 is an I1 but HVN003 is a P312>L21>DF13. The Haeven samples appear to be some of the oldest samples according the charts we got a peek at from a lecture, and I think maybe the Warini would have held this territory at the time. So it's interesting that "North-Western" extends to there in the pre-Slavic, pre-Saxon Migration eras. Maybe a peek into the pre-migration continental tribes. Note that DF19, DF13 and U106 also show up in the pre-400 CE Klosterneuburg auxiliary samples (but we don't have their autosomal makeup, yet).

Mecklenburg-Vorpommern_Häven:HVN001 [P312>DF19>Z302>>R-Y23278*, mtDNA HV0f]
Mecklenburg-Vorpommern_Häven:HVN002 [I1>>CTS11775, mtDNA H2a2a1]
Mecklenburg-Vorpommern_Häven:HVN003 [P312>>L21>DF13>>R-Y3646, mtDNA H1af1b]
Mecklenburg-Vorpommern_Häven:HVN004 [P312>>L21>DF13>>R-Z2961, mtDNA J1c]
Mecklenburg-Vorpommern_Häven:HVN005 [U106>>Z159>>R-Y6451, mtDNA H2a2a1b]
Mecklenburg-Vorpommern_Häven:HVN006 [U106>>Z9>>R-S5627, mtDNA U2e1]
Mecklenburg-Vorpommern_Häven:HVN007 [I1>>Z2844, mtDNA H10e]
HVN008: I1>>I-Y15031, mtDNA I4a1
HVN009: Female?, mtDNA U2e1f1
Mecklenburg-Vorpommern_Häven:HVN010 [I1>>>I-Y6375, mtDNA H2a2a1]

Tigran said...


Will we ever get confirmation on R1 coming from an East Eurasian population possibly from SE Asia?

Alistair David Brine said...

Any thoughts on this?

Tajiks are strange.

Ric Hern said...

@ Tigran

How did you get to the possibility that R1 is from Southeast Asia ? You surely know by now about MA1 and Yana RHS ?

Gaska said...

Anyone want to bet with me which were the male lineages directly involved in the spread of PIE?

What markers have been documented by genetic analysis published 5,500-3,500 BC in the Fertile Crescent, Caucasus, Anatolia and the steppes? Of course not counting those that only exist in the imagination of professional or amateur geneticists

Someone told me that Reich is very sure of what he is saying, will it be true? in a few days the mystery will be solved.

T.Gamkrelidze died a little more than a year ago, I am sure he would have liked to discuss the papers to be published.

vAsiSTha said...

"Also, there's steppe ancestry in Bronze Age western Anatolia."
Which sample?

Gaska said...

Being a non-Indo-European R1b-Df27, I have no horse in this race but it is clear that the positions are going to remain irreconcilable unless someone (once and for all) provides conclusive evidence one way or the other. Genetic continuity in male markers is an indispensable condition for linking ancient populations and speakers of a given language.

Has Harvard found the uniparental link between the eastern wing of the southern arc and the steppes? One or several migrations northward from the Caucasus? When exactly did these migrations take place? Does it coincide with the time of PIE formation?

Have Gimbutas' fans found evidence of steppe migrations south of the Caucasus?
On what dates did they occur?
What were the male markers involved?
What about Anatolia and Greece? Why are there only J2 Indo-Europeans there?

Too many questions, too few answers, too much speculation
Steppe ancestry south of the Caucasus?
CHG & Levantine & Iranian ancestry north of the Caucasus?

No region of the world (or population) is immune to genetic admixture, geography and genetics are linked, and therefore it is evident that autosomally the North and South Caucasus have been linked since time immemorial (CHG in the steppes since neolithic times, EHG in the South Caucasus since the mesolithic). In my opinion autosomal components are not going to solve the mystery, the key is in the migrations of uniparental markers. The leaks suggest that there are people trying to resurrect Maykop.

In any case, maybe these papers will serve to put an end once and for all to the moronic stuff that P312 and the BB culture spoke an Indo-European language.

vAsiSTha said...

Lol Sure, you have samples which Reich doesn't know of yet.

"A striking signal of steppe migration into the Southern Arc is evident in Armenia and northwest Iran where admixture with Yamnaya patrilineal descendants occurred, coinciding with their 3rd millennium BCE displacement from the steppe itself. This ancestry, pervasive across numerous sites of Armenia of ~2000-600 BCE, was diluted during the ensuing centuries to only a third of its peak value, making no further western inroads from there into any part of Anatolia, including the geographically adjacent Lake Van center of the Iron Age Kingdom of Urartu."

Also, Anatolian Languages came from the east. Ref Petra Goedegebuure | Anatolians on the Move: From Kurgans to Kanesh

Wrt Hasanlu (and the chariots founds on cups there), there is a nearby (within 100km) Hajji Firuz BA sample from 2400bce, which is part Yamnaya and close to Yamnaya_Ozero_outlier.

Hajji Firuz IA (1100bce) can be modeled as Hajji_C + Hajji_BA + BMAC. It shows no extra steppe ancestry other than 2400bce Hajji_BA. Hajji_IA also shows 30% BMAC like ancestry which is the only new external ancestry seen in this sample from the BA period. Result file

I am sure 2400bce yamnaya is not responsible for the chariots seen at Hasanlu. Then you have the Alalakh well lady from BMAC showing bmac ancestry was travelling west throughout this time.

Matt said...

@Dave; btw there was a "leak" on Anthrogenica that broke down more what they mean when they say "CNE" ancestry. It seems to mean a particular group that included these Haeven samples - . Other later samples from Germany/North Germany and Denmark are then modelled as admixed.

There are some differences between how north shifted these samples are, but I would caution that the same is true of the samples from Britain -

So it may not be so simple as we can model England as taking any of the continental groups and mixing with SE England Iron Age and may need to take into account population structure there and possible rotation. (E.g. the story is that the British called in Anglo-Saxons to defend them from Picts and Scotii, which implies some significant north to south migration within Britain is also possible?).

Davidski said...


I've seen Anatolian samples from Reich's paper as well as from other upcoming papers.

And my position is that proto-Anatolians migrated into Anatolia from the steppe via the Balkans.

vAsiSTha said...

As i said, only problem is that anatolian languages are attested in the east first. Lol.

Davidski said...

Not sure what you mean by "east", but obviously western and central Anatolia is where the Luwian, Hittite and other related peoples were based, and the Hittites spread east from there.


Gaska said...

I have also seen some samples and it seems evident that we can rule out a direct migration from the steppes to Anatolia between the Neolithic and the Iron Age. This leaves the Indo-European languages of Anatolia without reference because linguists agree that it is the oldest branch of IE. In my opinion the Balkanic solution could work, but you would have to clarify

1-Origin of those migrations in the steppes-Yamnaya>Balkans>Anatolia?????
2-Time of those migrations-Chalcolithic? Bronze Age?
3-Which markers-R1b-V1636, R1b-Z2103, I2a-L699, Q1b2a?

It will not be easy to defend your position because the genetic and cultural relationship between Anatolia and the Balkans was very intense and long lasting in both directions.

Davidski said...

I don't think it'll be very difficult to defend my obviously mainstream position considering the ancient DNA that is on the way.

StP said...

@Gaska said...
Anyone want to bet with me which were the male lineages directly involved in the spread of PIE?

This figure might be the answer:

Gaska said...

Well, I was going to help you with some interesting leaks, but you seem pretty confident so I guess you won't need any extra information. Remember Lazaridis' theory of Greeks from the east i.e. Anatolia ?, maybe some samples will help confirm or rule out this theory. Interesting times ahead.

Davidski said...


Remember Lazaridis' theory of Greeks from the east i.e. Anatolia?

Zero chance.

Assuwatama said...

Archaeology doesn't support Sintastha-Andronovo connection with Indo-Iranians either.

I don't know how to paste these images here. Copy functions in the pdf file not working.

You can read it here.

Ancestral clad to Indian R1a possibly came from ME or nearby and not Andronovo.

Davidski said...

Ancestral clad to Indian R1a possibly came from ME or nearby and not Andronovo.

Haha. Bullshit.

Davidski said...

Fatyanovo has the earliest Z93 to date you crackpot.

Nothing like that in the Near East.

Assuwatama said...

There is Y2 in middle East with 2500bce formation date with 2200bce TMRCA. (Modern sample)

Rob said...

Goedegebuure is cool, but Kloekhurst, Darden, Yakubovich, etc have different opinions. So Im not sure why her views are particularly special

Yes, the question of IE in Anatolia isn;t straightforward, but geneticists first need to develop the fundamentals in terms of broader scale populations movements in place of vague notions of continuity based on distal admixture analyses.

vAsiSTha said...

"Not sure what you mean by "east", but obviously western and central Anatolia is where the Luwian, Hittite and other related peoples were based, and the Hittites spread east from there."

Lol, youre a bullshittologist.

"Stefanini (2002: 788-789) pictures the following development of the Indo-European languages in Anatolia of the third millennium BCE. The Indo-Europeans that later became the Anatolians crossed the Caucasus7 and were therefore first exposed to Hurrian type (i.e., East-Caucasian) languages. At the beginning of the third millennium they entered Eastern Anatolia, spreading out on a substrate of Hattian type (i.e., WestCaucasian) to the North, and again on a Hurrian type substratum to the South."

The oldest Anatolian names (2400bce) come from tablets in syria (Ebla) which mentions anatolian names to to its north from the town of 'Armi' which is in east anatolia. (ref youtube: Petra Goedegebuure | Anatolians on the Move: From Kurgans to Kanesh).

Again, read up more, this bias of seeing steppe in everything is clearly addling your brain.

Assuwatama said...

Victor Sarianidi

"ME tribes started migrating 2500bce and ended up north in BMAC and Arabian shores in south."

Guys in Oman, Kuwait and UAE have ancestral clads to Indian l657. Both not present in steppe modern and ancient samples.

Exceptions: among those with Indian heritage.

Davidski said...

They're just hypotheses about how proto-Anatolians may have entered Anatolia from the steppe.

But the Caucasus route from the steppe is not mainstream

The mainstream theory is the Balkan route. And that's because Anatolian speaking sites are clustered in western and central Anatolia.

And the Anatolian names at Ebla are supposed to be those of traders from somewhere else and they may or may not be Anatolian. It's just another hypothesis.

Davidski said...


The oldest Z93 is in Fatyanovo you moron.

vAsiSTha said...

"They're just hypotheses about how proto-Anatolians may have entered Anatolia from the steppe."

Lol. There are other regions of the world which are not steppe.

Anyway, so you're claiming that you know for sure of steppe presence in west anatolia prior to 2400bce? What is that - like 2% steppe?

And the whole region of central anatolia till 1500bce is devoid of steppe ancestry but suddenly anatolian names end up east?

Assuwatama said...

Jaat L1a2 is found in Turkmenistan 4600bce with CHG+IranN autosomal profile.

Modern L1a2 Jaats are 40% steppe, some BMAC + Harappa.

Teach you sound like this Jaat and that Jaat spoke indo-Aryan.

Assuwatama said...

I am not aware of the European side of things but Basques are R1b, right? Aren't IE. Etruscans and Latins were identical. One wasn't IE.

Z94 migrating into area inhabited by linguisticaly different people has to learn the local language to survive. His descendants will soon loose their tongue and in 200 years their autosomal profile too will disappear.

These developed India specific mutations and entered India at a later date with different autosomal profile.

BTW some in your group are trying to explain the lack of steppe in Anatolians via exact same means.

Kouros said...


Are the new samples on Reich site include upcoming ones? Merci


I am friends with Dr Lhuillier she works extensively on BMAC sites by 1400 BC, it is a more societe guerre , Steppe nomade have carte blanche rule over many settlement and integrate but imposing their languages. Arabs did not get R1a from Oxus people but admixed Steppe Nomade/Oxus population who migrate to Persia and Sud Asia mix and these people trade / travel to Arabia. Kuwait has large Ajami / Persian population. I seldom comment but the posts you make are ignorant and puerile.

Davidski said...

I doubt that any of the samples at the Reich Lab website are from papers or preprints that haven't been released yet.

Usually what happens is that the samples are uploaded when a paper comes out, and occasionally after a preprint comes out.

Gaska said...


Maybe R1a-M417 and the CWC is linked to the diffusion of certain branches of Indo-European (Balto-Slavic) in mainland europe, but it has nothing to do with Western Europe, Italy, Greece, Anatolia etc...By the way Polish is for me almost as difficult as Euzkera

Assuwatama said...

Why would Arabs get it from BMAC or steppe?

Y2/Y3/l657 wasn't in steppe.
Steppe z2124 is a brother clad not father clad. Persians aren't l657 either.

Y2 formed before sintastha.

Fire-Soma temples predate 2000bce. Funeral practices in BMAC shows Zoroastrian features.

You are talking about 1400bce....BMAC show Indo-Iranian features 2200bce.

Posted pics here.
I believe you have no idea what you are talking about.

Assuwatama said...

"Architectural similarities between Northern Mesopotamia - Agean - Mitanni - BMAC has been observed.

From 2500bce ME tribes started migrating from somewhere around south of Caucasus and spread eastward."

Victor Sarianidi

Assuwatama said...

Another interesting pointer.

Swat valley has 2 R1a none l657 type. one of them is Z94 dated 300bce (need to recheck though). Mostly E1b1b, R2, L1a and Q1b.

Indo Aryan Gandhari Prakrit was spoken in this region.

Fast forward medieval and modern times and this region is inhabited mostly by Pashtuns who are abundant in steppe autosomal ancestry and high percentage of R1a + speak Iranian language.

Swat valley has huge BMAC ancestry. Victor long back showed it's connection to BMAC. Archaeogenetics wasn't even a thing back then.

Rob said...

@ Vasistha

There have been bilateral contacts between sutheast Europe & Anatolia since the Paleolithic. Why would the bronze age be any different ? Iran/CHG-infused groups reached Greece & the Aegean in the late Neolithic and soon after there were movements in the other direction. It's hardly rocket science.

Gaska said...


The European side of things is very different from the South Asian side of things-In the more developed regions of Western Europe NON Indo-European languages survived for hundreds of years despite the Roman conquest, thanks to the greater cultural trade exchange with Phoenicians and Greeks and their alphabets. Thus we know Iberian, Tartessian, Etruscan, Raethian and Basque-Aquitanian, all of them spoken by peoples that were and are overwhelmingly R1b-P312. If other western regions had had such cultural contacts, we would surely know other non-Indo-European languages in Britain, Ireland or France-IMO, the Indo-European languages in Western Europe have nothing to do with the Chalcolithic or Bronze Age but with the Celtic migrations (Urnfelder, Hallstatt, La Tene) of the Late Bronze Age and Iron Age. Therefore neither R1b-P312 nor the BB culture have anything to do with Indo-European (Germanic, Italo-Celtic etc)

Regarding South Asia, it seems evident that R1a is linked to the EHGs and therefore the markers that appear in Iran, India etc must be linked to steppe migrations that brought to those regions certain branches of IE. The only way you have to prove otherwise is to find that marker in South Asia old enough that R1a-EHGs are no longer important or find south asian populations that are overwhelmingly R1a and that speak (or spoke during the Iron Age) a non-Indo-European language-Do such populations currently exist in India?.

And regarding Anatolia, the chances that the Hittites have their origin in the Balkans or in Caucasus-Syria-Iran-Mesopotamia-Central Asia etc... are zero in all cases.

Copper Axe said...


Because you must be bored out of your mind with these delusional comments here about R-Y3 coming from Arabia or whatever, check this out:

This will contain the Seima-Turbino samples from Rostovka you mentioned a long while back. You know those lovely discussion threads on Anthrogenica with your boy Jaska :)

IIRC you mentioned some WSHG-pops, steppe_mlba pops and one outlier thats mostly/completely Kra001-like. Thoughts on this new academic zeitgeist of strongly connecting the StP to the spread of Uralic languages? There was another article last year which had a nonsensical proposal with Proto-Uralic being some kind of lingua-franca of the Seima-Turbino traders or something.

I dont have an issue with the proposal that Indo-Iranian/Proto-Uralic contacts took place within this context but the spread of STP sites seems too early and widespread to be an archaeological representation of the spread of Uralic peoples, particularly those in Europe.

ALDER/DATES results from both academic works and by amateurs such as Altvred have shown that the west/east mixing in Uralic people postdated the STP by several centuries/nearly 1000 years, which is hard to match with the obviously west-Siberian presence in the STP, and not to mention the steppe derived people around the Urals and the Altai at this time. Whatever their migration route was it must've largely avoided these populations at least in terms of genetic mixing. Which isn't what you'd expect from a demographically small forager population in intense trading contact with their neighbouring peoples. Or even being in a position of supremacy (Hungarian conquerors but 3000 years earlier I guess) as some people like to suggest. Not to mention there not being much evidence for steppe_mlba input in the most western Uralics, or WSHG in most of European Uralics.

Meanwhile we see plenty of mixing between steppe derived and west/east siberian derived people during the 3rd and 2nd millenium BC whenever they were proximate to one another.

Copper Axe said...

By the way is there a reason Yakutia_IA isn't on G25? There has been a discussion on AG for a while whether Nganasans are actually genetically Uralic or just superficially appear to be so, which got reinvigorated by a model with Nganasans mostly derived from Yakutia_IA. On G25 however all non-Uralic Northeast Asians seem to heavily prefer Ymyyaktakh over Kra001 which is the opposite for Uralics and the Ket, who genetically are significantly Uralic anways. Nganasans on G25 work as Kra001+Yymyyakhtakh in a near 1-to-1 ratio for their northeast asian ancestry, and I would like to see whether that Yakutian sample scored similarly. The sample had N-P43 by the way which is common in Siberian Uralic peoples but also other modern and (presumeably) ancient populations.

Simon Stevin said...


Hey Dave, does Iran_N have EHG rather than ANE? Some of the usual suspects over at AG are it again. It looks like they’re trying dilute as much steppe as they can in BA-IA Iran, by trying to show that EHG was already there in Iran_N (they’re also probably trying to prove an Iranian origin for R1a-M198 and R1b-M269). Why is it that they think steppe ancestry can be supplemented by EHG via Iran_N? Would their fits be poorer/less likely due to the fact that there are slightly differing SNP frequencies in WSHs, when compared to ICHGs and EHGs? I believe neither population matches the exact genetic profile of WSHs, and various models would show which source is the most likely one (WSH vs Iran_N with EHG) correct?

Kouros said...

@Assuwatama , David is correct , you are an imbecile.

Sarianidi expire in 2013, we French have taken the work there specifically Lhuillier, Farrige research and dig a lot on early Indus sites in Pakistan in the 70s. Lhuillier document 5 pastoralist sites surrounding Gonur Tepe and other Margiana site. Ultimately Eastern Corded ware is location where R1a and Steppe ancestry diffuse into Kazakh Steppe then into BMAC. Almost 200 Agrarian population from Central Asia , not 1 R1a but just north of Turkmenia you have populations with 100% R1a. Also linguistics , Fatyanovo which David mentioned is near Baltic region and it well known Baltic language and Sanskrit have many cognates. Fatyanovo is Z93 since 4800 years ago before any of those subclades you listed.

Davidski said...


Iran_N doesn't have any EHG.

There's definitely steppe ancestry in BA-IA Iran. It gets diluted by the IA quite a bit, but this is irrelevant.

Many people would like to believe that the R1b-Z2103 in BA-IA Iran isn't from Yamnaya/Catacomb, but it obviously is, and this fact won't be contradicted by any of the new data in the upcoming Reich Lab paper.

Davidski said...

@Copper Axe

What are the individual codes for Yakutia_IA and where are these samples from?

Copper Axe said...

Yak030 from this article:

Davidski said...

Just had a look at Yak030, and the Human Origins version that I have only has 31616 SNPs.

That's not enough for the G25.

It's probably possible to get to around 60K SNPs for this sample if the 1240K version is available anywhere, but that's still low coverage, and not really worth getting too excited about.

Rob said...

Iran_N does have ANE admixture (but not EHG), + tad extra ENA-shifted ANE

@ Copper Axe
Yeah I think S-T is too broad to be FU on the one hand, and the Kra-1 related admixture does not completely overlap with S-T on the other.

Assuwatama said...


hope this long post gets accepted.

BMAC Y haplogroups

In 2018, Narasimhan and co-authors analyzed BMAC skeletons from the Bronze Age sites of Bustan, Dzharkutan, Gonur Tepe, and Sapalli Tepe. The male specimens belonged to haplogroup E1b1a (1/18), E1b1b (1/18), G (2/18), J* (2/18), J1 (1/18), J2 (4/18), L (2/18), R* (1/18), R1b (1/18), R2 (2/18), and T (1/18).

BMAC Temples

Thus, a fire temple in Djarkutan (northem Bactria) and two temples in Dashli-3 (southem Bactria) manifest common principles of ancient Bactrian architecture. A fre temple and three other temples excavated in Margiara were dedicated both to the fire cult and the cult of the hallucinogenic beverages of the soma-haoma type. In Gonur, the capital of Margiana, a fortress with a centrally located palace was fully excavated, and the obtained material allows us to come to a prelimninary conclusion that individual buildings in Margiana, both secular and religious, have certain uralogies with corresponding buildings in other regions of the Near East. In spite of some specific features of the monumental architecture of Margiana and Bactria, it can still be compared with the architecture of northen Mesopotamia and the Aegean world. The parallels between the architecrural blocks under comparison are few in number but they are so representative and clear that one cannot ignore them. First of all we should mention the Propylaea (fig. l) of the northern Gonur palace, where two passages and rabbets were directly analogous to the ones on the second floor of the Knossos palace in Crete (Evans 1928, fig. 448). It should be noted that rabbets are consiclered a definite element of monumental architecture (Roaf 1998: 57-58) and the "burnt building" in Hasanlu was defined as a temple exactly because its central entrance was decorated with such rabbets.

Assuwatama said...

In the palace and fire temple at nofthem Gonur (Sarianidi 1998a) that belonged to ttre earliest period in ttre history of the BMAC, i.e. to the last ages of the third millennium BC, there were found so-called "blind windows" or altar niches of a kind. Especially representative is the sacred place (room t00) in the fire temple with its five "blind windows" and a double chambered hearth of a special construction (Sarianidi 1998a, fig. 63). The same kind of sacred place with several "blind windows" was also found in the palace of northem Gonur (room 185), where a similar, though much bigger, double-chambered hearth was placed on a cenüally located low platform (Sarianidi 2000). These two double-chambered fue places were not used for cooking but for sacriFrcial ceremonies. They were constructed in such a way that fire reached the cooked meat, a condition that directly corresponded to the Zoroastrian sacrificial traditions with their roots in "Ilanian paganism" (Boyce 1989: 153). Each heafh was divided inside by a low partition into two unequal chamben; the smaller was used as a furnace and the bigger as a place for cooking sacrificial food (Sarianidi 1998a: l7l). Both the sacred rooms of Gonur (100 and 185) used for sacrificial ceremonies have direct parallels in the monumental building of Tillya Tepe (fig. 5) in eastem Anatolia. The latter belonged to a much later period and also had several "blind windows" and special hearths made in the floor (French 1983, pl. t).

The Mitannian temple in Tell Abraq, with its altar in the form of a "blind window", occupies an intermediate chronological place between the monuments of northem Mesopotamia and Margiana (Oates 1987) and gives a certain support to this assumption. The elite of the Mitannia¡r society was known to consist of IndoIranians, the fact that to a ceftain extent allows one to presume the Aryan character of the Tell Abraq temple. Such supposition can lead one to the conclusion that parallels between the "blind windows" in the fire temple and the palace at northem Gonur (as well as at the Bactrian Dashli-3 site), on the one hand, and the corresponding altar in Tell Abraq, on the other, reflect - besides the architectural similarity - common cult ceremonies and rituals that in the long run can somehow be connected with Indo-Iranian religious beliefs.

In short, there are certain similar features between separate monumental architectural blocks of northem Mesopotamia and partly of the Aegean world, on the one hand, and the Bactria-Margiana complex, on the other. Moreover, the .,strong influence of the Aegean world" on the eastem Meditenanean and especially on Syrian monumental a¡chitecture has been referred to (Moortgat 1969:79 Plommer 1956: 92\,and now this influence is, to a certain extent, noticed as far away as in Centrat Asia.

Assuwatama said...

BMAC Funerals

The chamber tombs at the Gonur necropolis can be looked upon as house models used by the Indo-European peoples, lndo-Iranians in particular (Gimbutas). In the Greek literature this type of grave is called the "House of Hades"; in the S.gveda, correspondingly, it is called the "House of Yama" (Jones-Bley 1997: 194200; Hansen 1980: 35-37). Alongside brickJayed tombs, there existed simple ones in which the walls had only a thick clay plaster instead of brick-laying. Most graves at the Gonur necropolis were plundered soon after the funeral and only ZOVo of the necropolis (including all chamber tombs) remained untouched. In many chamber tombs it should be noted that bone crumbs and small bones (teerh, finger parts) - instead of full skeletons - were found on the floors. This proves that some time after the burial the relatives gathered the bones of the dead and reburied them in some other place. This assumption was additionally supported by the fact that the level of floors dug in the sand soil were 15-20 cm below the brick walls. This can probably be explained by the repeated sweeping of sand floors after each removal of a decomposed dead body.

Assuwatama said...

According to the Videvdat, the tradition of building temporary tombs was typical of Zoroastrian funeral rites. Such temporary tombs or kata "may be considered as a kind of morgue of the earlier period" (Hismatullin & Krjukova 1997 219), and it seems very likely that the Indo-Iranians used the practice of reburying their dead. Since Margiana represents "Iranian paganism" (Grene & Gnoli), the temporary graves at the Gonur necropolis rcflect ancient Iranian funeral traditions that existed in the Iranian world long before the origin of the Zoroastrian religion. In this respect, the "complex of funeral rituals" in the westem wing of the Gonur palace (fig. I I ) is of special interest. All rooms of the complex are joined together by common passages, while all walls and floors (except in room 55) were completely covered with white plaster, often in several layers. It is quite probable that the rooms of the "complex of funeral rituals" were whitewashed not only inside but outside as well, since traces of plaster were found on the facade of these rooms. The central passage almost in the centre of the facade wall leads to a vast room with smaller rooms on both sides. Room 48 stands out among them. It was decorated with a great number of recessed niches and with a round hearth in the cenre of the floor, which bears slight traces of fire on the inner brick layer. Opposite the hearth there is a fire-place with traces of flre inside it. The traces of fire in the central hearth indicate the special purpose of this room.

With its round pit and clear traces of water on its bottom, room 8l deserves special attention. A small outlet with th¡ee ceramic tubes is seen in the wall next to the pit. The whole picture gives one an impression of some kind of a drainage system used for letting out waste water (fig. l2). Two similar pits, also with traces of water, are in line in room 163. Shallow pits filled with bumt animal bones indicate that this was part of the "complex of funeral rituals". The plan of this complex reflects certain Zoroastrian funeral rituals, and one may have sound grounds to believe that the complex was most probably used for washing corpses. Though the fact of corpse washing is not mentioned in the Videvdat, later Zoroastrians were known to practise it and this tradition is still alive nowadays. If our assumption is true, we may suppose that the above-mentioned pits with water spots, and especially the excavated drainage system, may testify to the fact that the room was actually used for washing corpses. The number of excavated pits (hree altogether) fully conesponds to the funeral rituals of late Zoroastrians who used three pits for this purpose: for men, women and children correspondingly. According to Zoroastrian rituals, the dead person was brought into the "house of the dead" through one entrance and taken out through a special hole. Our excavations seem to illustrate this, since in room 45 the corpse was brought into tlrc "complex of funeral rituals" through a central entra¡rce and taken out through another door. Iranian Zoroastrians also had a special room for keeping their funeral equipment in the "house of the dead".

Assuwatama said...

The Russian and American geomorphologists came to a unanimous conclusion that in the second half of the third millennium Bc the solar activity caused a xerothermic period in the Near East that affected the territory from Greece to Indus and partly caused the collapse of the ancient Egyptian kingdom (Bell l97l; 1975; Klimenko & Prusakov 1999). It is very important to notice that the xerothemic period influenced the teritories of the Near East unevenly: some areas (from Anatolia up to the Iranian plateau) seemed to remain untouched while some other areas suffered from a considerable lack of water and the local tribes were forced to migrate in search of new fertile oases. Asia Minor and Mesopotamia were already densely populated when this tribal rnigration started and so the people set forward in a general eastem direction searching for a new homeland. At last they reached the fertile oases of Central Asia and decided to settle and colonise the historical lands of Bactria and Margiana (Sarianidi 1998a). This vast tribal invasion embraced a huge tenitory of the BMAC fro¡n the southem coast ofthe Caspian Sea up to the northem coast of the Arabian Sea and Baluchistan, also including some areas of the Indian subcontinent. At present, monuments of the BMAC type are known in the Kandahar area (southern Afghanistan) and Quetta (Pakistan) in the vicinity of the Indus Valley (personal communication of C. Lamberg-Karlovsky). This new materialcompletely supports the theory of the anival of the Near Eastem tribes in Central Asia and farther up in the Indian subcontinent (Sarianidi 1998a: 150-166). Until lately it was believed that the recently anived tribes settled only in the territory of Bactria. The latest excavations at the Gonur necropolis, however, give grounds to suppose that they reached as far as the Swat Valley. The excavations by Italian and Pakistan archaeologists, who found graveyards of the Gandhara culture in this valley, offer ¿ur additional support to the idea of the vast tribal settlement. The funeral ceremonies of this culture included flexed burials, cremation and fractional burials. The latter presuppose the removal of soft tissues from dead bodies before burying the bones and the existence of temporary tombs with the consequent way of burying. All these types of burials, except for cremation, were found at the Gonur necropolis. The fractional burials at the Gonur necropolis were represented by a kind of "dakhma" (figs. 13 & 14) found in the ruins of a neglected palace. It contained the mixed up bones of more than ten corpses placed on a special layer of ashes (Sarianidi 1998a: 7 l:72).

There is more but I would stop right there.

StP said...

@Gaska said...
Maybe R1a-M417 and the CWC is linked to the diffusion of certain branches of Indo-European (Balto-Slavic) in mainland europe, but it has nothing to do with Western Europe, Italy, Greece, Anatolia etc...By the way Polish is for me almost as difficult as Euzkera

The source of the various kentum dialects was not the developed satem of the family R1a, but the original PIE of the people R1a or R1b

A said...

Reminder that we already have Mitanni royal DNA.

Amenhotep III's mother Mutemwiya was the daughter of the Mitanni king Artatama I. She married Amenhotep III's father Thutmose IV as part of an alliance between Egypt and Mitanni. Amenhotep III had mtDNA H2b, inherited from his Mitanni mother.

"H2b is a minor branch. It contains several ancient samples from Russia, all basal to the rest of the branch, including one individual from the Yamnaya culture and one from the Late Bronze Age Srubnaya culture, both from the Pontic-Caspian Steppe region, and five other Bronze Age samples from east of the Volga river: three from Sintashta and two from Krasnoyarsk. Also in a basal position, there are three modern Russian samples (two from the Altai region) and one Danish sequence. Interestingly, while the vast majority (70%) of H2 modern sequences in our dataset are of European origin, H2b displays a strong South Asian component, with seven samples from Pakistan, India and Sri Lanka. The newly published Sintashta and Middle Bronze Age Krasnoyarsk (Russian) sequences (Narasimhan et al.2018), together with the previously released Yamnaya and Srubnaya, span a period from 5 to 3.5 ka. These, plus the modern South Asian sequences, support our earlier suggestion that H2b was involved in movements east and southwards from the Pontic-Caspian region into South Asia, by documenting its progress eastwards across the Eurasian Steppe. The Sintashta Culture in the Ural Mountains, or a “Sintashta-derived” culture (such as the Andronovo), is thought to have expanded eastwards into Central Asia 3.8 ka, reaching South Asia"

(Silva et al. 2019, Untangling Neolithic and Bronze Age
mitochondrial lineages in South Asia)

epoch said...


"The oldest Anatolian names (2400bce) come from tablets in syria (Ebla) which mentions anatolian names to to its north from the town of 'Armi' which is in east anatolia"

So? It is hardly proof that Amri - whose location is unknown - as such spoke an Anatolian language. Or even partly used it. There is just the mentioning of some names which could be explained as Anatolian.

There are archaeologists that theorize that there was a great caravan route between Troas and North Syria.

Whatever it is, there is very serious evidence of intensive cultural connections between Troy and Mesopotamia. So Anatolian names of traders popping up in Syrian documents is to be expected with an over the Balkans scenario.

gsueso said...

Sorry if this is a stupid question, but what are these samples exactly? The normal scale and unscaled samples produce different results.

Assuwatama said...

By sequencing 727 ancient individuals from the Southern Arc (Anatolia and neighbors in Southeastern Europe and West Asia) over 10,000 years, we contextualize its Chalcolithic and Bronze Ages (~5000-1000 BCE), when extensive gene flow entangled it with the Eurasian steppe. At least two streams of migration transmitted Caucasus and Anatolian/Levantine ancestry northward, contributing to Yamnaya steppe pastoralists who then spread southwards: into the Balkans, and across the Caucasus into Armenia, where they left numerous patrilineal descendants. Anatolia was transformed by intra-West Asian gene flow, with negligible impact of the later Yamnaya migrations. This contrasts with all other regions where Indo-European languages were spoken, suggesting that the homeland of the Indo-Anatolian language family was in West Asia, with only secondary dispersals of non-Anatolian Indo-Europeans from the steppe.

Samuel Andrews said...

@A, very cool. Thanks for sharing.

Samuel Andrews said...

Davidski are the Gerber 2022 Bronze Hungary samples available yet?

Simon Stevin said...

Do EEFs have some Natufian ancestry? This is something I’ve been trying to find a concrete answer on for a while. The E1b Y-DNA lineages found in EEFs may point to an ancestral connection with the Natufians/Levantine HGs and para-ANA/Taforalt-like groups.

Tigran said...

@Ric Hern
Modern K2b diversity is in SE Asia. Tianyuan from north china is the ancestor of ANE and founder of Europe.

Rob said...

@ epoch

Yeah I’ve been thinking the same thing

Modas Califa said...


Mr. Tigran, I notice that you constantly repeat this same comment (I've seen you writing the same thing for about 2 years), and it seems that you think this is some sort of trolling. There would be no demerit if all the ancestors of Europeans came from Asia or any other part of the world (even because our species arose in Africa). But the fact is, they didn't. The ANE lineage is mostly western Eurasian, that is, it is primarily related to other European Paleolithic groups.

Rob said...

not even the ENA component 'came from SEA', it is from central Asia. SEA & south China were population sinks.

Matt said...

@Simon Stevin: - "We report the first ancient DNA from the world’s earliest farming cultures of southeastern Anatolia and northern Mesopotamia, as well as the first Neolithic period data from Cyprus and Armenia, and discover that it was admixture of Natufian-related ancestry from the Levant—mediated by Mesopotamian and Levantine farmers, and marked by at least two expansions associated with dispersal of pre-pottery and pottery cultures—that generated a pan-West Asian Neolithic continuum. "

So it seems so.

Re; the other version of the "Southern Arc" abstract - "At least two streams of migration transmitted Caucasus and Anatolian/Levantine ancestry northward, contributing to Yamnaya steppe pastoralists", guess that fits with the guess that the above referring to "Yamnaya had two distinct gene flows, both from West Asia", they are referring to a Caucasus_Eneo/Maykop like wave and a prior wave without Levantine/Anatolia_N related ancestry. The models with Maykop, EHG, and Steppe_Eneolithic worked OK for Yamnaya for me (30%, 26%, 44% respectively). Be interesting to see if they have some specific evidence against the theory that Yamnaya's EEF from Europe or it's more like their qpAdm models just work better this way.

Rob said...

@ Matt
I don’t think Harvard have much of a handle on matters these days; so they’re hardly worth quoting
I have time for their Spanish collaborators of course, but not some of the headline personas

Assuwatama said...

@A That argument doesn't hold water.

Wives of Mitanni have non-Aryan names. They possibly were born from local Hurrian women taken as wives by Mitanni Royals.

Gilukhipa, Tadukhipa, Mutemwiya etc.

"The H2, H6 and H8 haplogroups are somewhat common in Eastern Europe and the Caucasus. They may be the most common H subclades among Central Asians and have also been found in West Asia."

Assuwatama said...

Mitanni Aryan names


Rig Vedic Aryan names


Iranian names


Rob said...

how much more of Assuwatama's OIT screen-dumps do we need to see ?

Gaska said...

Taking into account that the steppe theory is in continuous evolution since 15 years ago adapting to the new genetic discoveries that are being made, I suppose someone could explain me what is the current interpretation of the role played by the R1b-L51>L151>P312 marker in the genesis and spread of Indo-European languages

At the moment we know that it has absolutely nothing to do with Tocharian, Indo-Iranian, Balto-Slavic, Germanic, Armenian, Albanian, Greek-Mycenaean, Anatolian, Hittite etc.... And when it appears in Western Europe it is linked to non-Indo-European languages. It seems a joke that so many people still consider this lineage as essential to explain the diffusion of IE in mainland europe. Maybe some of you have a reasonable interpretation of the situation.

Perhaps Harvard has found this marker in the eastern wing of the southern arc ???, and should then find a reasonable explanation for its non-existence in Iran, India, Anatolia, Levant, the steppes, the Balkans, Carpathian Basin, Baltic countries, Scandinavia etc. .... or maybe a lot of people should start thinking that maybe (I'm just saying maybe) they've jumped the gun.

Matt said...

@Rob, I don't think the Natufian ancestry expansion is unlikely though. Didn't you argued for something similar when it came to Barcin vs Pinarbasi HG in Anatolia? That the Neolithic population has substantial gene flow from Levant?

The "two waves to Yamnaya" doesn't seem genetically unlikely to me either. The linguistic interpretation has the problems we've discussed ad nauseum here. (One thing I think more about is is I find it unparsimonious really; you're having to come up with two reasons for the separate expansions, rather than the simple explanation that the mobility derived on the steppe drove the entire expansion. Post-neolithic language expansions seem to be better explained by mobility technology than crops. There's a "greater average mobility drives opportunistic expansion" explanation for IE under the steppe theory but in a two-phase West Asian and steppe theory you have to explain these separate developments).

@Gaska, no one said it's essential, they say that it's found earliest, so far on the steppes and Western Russia (more or less, please don't pedant here about Lithuanian results) and they say it's found in a much wider range after the expansion of steppe ancestry, which is highly likely to coincide with the expansion of Indo-European languages. None of these points can be realistically contested. I'm not sure what you're trying to do by making some claim that this patrilineage is said by some to explain or be necessary for IE expansion; no one is saying this.

Copper Axe said...

"The models with Maykop, EHG, and Steppe_Eneolithic worked OK for Yamnaya for me (30%, 26%, 44% respectively)."

Being realistic, Isn't an issue here that Maykop and Yamnaya are practically contemporary (while Yamnaya-like profiles are for sure older than 3300 BC)?

Also what do others here make of such a model? Does it pass the tests such as uniparentals etc?

Davidski said...

No real evidence of Maykop contributing to Yamnaya even if some qpAdm models suggest that it did.

Yamnaya is an offshoot of Sredny Stog, which existed well before Maykop.

Rob said...

@ Matt

yes I agree about Natufians, but in reality this all began from northeast Africa

W.r.t the '2 waves': some of the genetic constructs are so wishy-washy, imprecise & biased that they have no value to anybody who is a genuine & competant scholar of Eurasia. Of course, the data itself is valuable, but most scholars who are open to archaeogenetics cant use ADMIXTOOLS for ex, and are left to simply 'trust' what is said.

First of all, there are in fact at least 4 waves

1. a ~ 6000 BC of CHG-related hunter-gatherers reaching the Volga. These mixed with EHG, and then moved back south toward the Caucasus ~ 4500 BC- the so-called 'steppe Eneolithic'
So here we have false construct # 1 . ''Steppe Eneolithic' does not constitute a migration from the Caucasus, Iran or whichever

2. minor EEF-related admixture into the western fringe of the steppe. This is clear & obviuos. Why would they call this 'west Asian' instead of EEF ? Seems weasel-ish

3. arrival of Meshoko-Darkveti groups; CHG rich early mountain farmers. A fascinating group, but not really relevant

4. Arrival of Majkop after 3800 BC. Limited interaction with the steppe, e.g. Ozera outlier. maybe they brought new metallurgical & shepherding techniques, but they ultimately disappeared even from the mountain fringe itself. They were so culturally distinct & separate, moreover separated by a 'buffer' of ''steppe majkop'', that any language transmission now seems like extreme special pleading.

So what are they referring to here ? Where is the southern arc genesis ? There is no sample or statistic that Harvard can bring forwrad that would salvage their claim.

Then there is the issue of West Asia itself. Reich's claim that Anatolia was impermeable to outside influences is simply a joke. Then everybody and everythign between central Anatolia and northern India is "Iran_N-related'. What are we even talking about ? There's no definition.

As I said, its not even about PIE. It's about not selling the data short by fabricating caricatures and false narratives.

Gaska said...


If they say that it has been found before in the steppes and western Russia they are wrong, but in any case, if nobody is saying that it is essential to explain neither the origin nor the spread of IE, that means that you (and the rest) are of the opinion that it played a marginal role. Do you have a theory about which branches of Indo-European spoke that lineage?

Regarding their link to steppe ancestry, you (and they) have to remember that women played a very important role in this matter because exogamy is the best survival strategy that mankind has had since time immemorial.

Matt said...

I just used Maykop was just the highest quality and largest of the Caucasus like profile; the Darkveti set is infeasible in qpAdm because they're low coverage or relatives. See what the Neolithic Armenia set in the Reich Lab's paper has to offer.

Davidski said...


Neolithic Armenians are similar to those Azeri LN samples, but with a bit more Anatolian ancestry.

Matt said...

@Rob, it doesn't seem so obvious that transmission of EEF is from the west; we have the replacement of the Ukraine_N like people with the Srendi-Stog type genotype, which comes presumably from the east, but we don't really have intermediate samples evidencing that spread in of EEFs from the west yet, yet.

Matt said...

I don't think there's a lot more to say about until we see the form of the argument they make anyway.

Copper Axe said...

If the transmission of ANF came from Caucasian populations, shouldnt progress/vonyuchka and the non-outliers at steppe Maykop have this ancestry too? Or Khvalynsk even?

It would be odd for these populations to largely avoid this mixing (aside from the outliers ofc) while the near contemporary Sredny Stog/Yamnaya populations having about 30% despite likely having an origin point that was geographically more diatant to Maykop than the aformentioned samples/populations.

Rob said...

@ Matt

“it doesn't seem so obvious that transmission of EEF is from the west; we have the replacement of the Ukraine_N like people with the Srendi-Stog type genotype, which comes presumably from the east, but we don't really have intermediate samples evidencing that spread in of EEFs from the west yet, yet.”

No that’s not quite true. Again, this narrative stems from the lack of understanding by geneticists practising “statistical homeopathy “

The replacement of the Ukraine N type (DDII) population - which reached all the way toward the Tisza and Balkans- was due to the expansion of the farmers, not easterners . Again this is clear, as Tripolje shifted all the way toward the Dnieper by 4000 Bc
However some (I2a2 rich) remained along the lower Dnieper but shifted toward the Volga-Caspian network (? As a counter-reaction) and hence were autosomally inundated into EHG/ CHG

Then GAC in turn replaced CT. Yamnaya is optimally represented as Khvalynsk/ Progress + GAC or late CT (eg Gordinevtsi). This is also clear in qpAdm and archaeologically.

Gaska said...

Rob is right, there is evidence of the arrival of EEF from the west because their WHG percentage is very high (in fact they can be modeled perfectly with samples from the Iberian Chalcolithic that have WHG percentages similar to GAC).

Gaska said...

And he is also correct in saying that Anatolia was not impermeable to outside influences because no region of the world is, not even the Pontic steppes, which are open spaces easy to cross. In my opinion, sooner or later, samples with steppe ancestry will appear in Anatolia (the ones we currently have are not worth discussing), but they will only have genetic-linguistic value if they were abundant and linked to uniparental male steppe markers (and this, in view of what we have seen, seems impossible at present)-

Regarding Maykop, I do not agree with Rob because although apparently their genetic influence on Yamnaya seems minimal, it is true that they share some uniparental markers (both male and female)-This may not be enough to cause a language shift, but the cultural and technological influence may be great enough to do so. Perhaps Harvard will provide the solution-It is funny to see the gymnastic mental exercises of some people who accept without problems that the supposed R1b-P312 conquerors in Iberia renounced their Indo-European mother tongue in spite of their technological superiority and yet they do not want to hear about the Yamnaya-Z2103 abandoning their language when faced with the influence of a technologically and economically superior culture like Maykop, even though the genetic exchange was scarce

Rob said...

Majkop might be relevant for some eastern groups, Catacomb, the steppe-Caucasus hybrid groups, few outliers on lower Don etc. But it a tenuous position to claim a model which points to a sweeping influences from one southern groups. Everything points to a longue duree development among local HGs, the earliest & most sustained contacts have been with European farmers.
The evidence is unequivocal

Gaska said...


yes, Majkop may not be the solution.

I don't know if I understood you correctly but if you disassociate CHG and South Caucasus from the PIE homeland and it is only a matter of the EHGs, then nothing prevents the rest of European HGs to speak that language too, because all of them-Iron Gates, Western & Baltic HGs are genetically closely related. Why would some speak PIE and others not? Maybe the solution is the contact zone with the farmers of old Europe because the agricultural terms also exclude the HGs.

And if Ukraine were the common homeland, then you would still have to explain the dispersion of the different branches in Greece, Anatolia etc…. which to date seems impossible (unless you are in favor of declaring a region as Indo-European in case of finding one sample in a thousand that has 1% of steppe ancestry). It has always seemed to me absurd to link an autosomal component with a language, but some are coming to defend ridiculous positions (Anthony would be an example).

Gaska said...


What would you think if Reich had been able to prove that the origin of Z2103 is south of the Caucasus and that those men mixed in the steppes with EHG-women to give rise to the Yamnaya culture and with Anatolian women to form the Hittite culture?

Tomorrow we will know if there is convincing evidence

Simon Stevin said...


Thanks for the link; there’s also this from the paper “Late Pleistocene human genome suggests a local origin for the first farmers of central Anatolia” (March 2019):

“We find high genetic continuity (~80–90%) between the hunter-gatherers and early farmers of Anatolia and detect two distinct incoming ancestries: an early Iranian/Caucasus related one and a later one linked to the ancient Levant. Finally, we observe a genetic link between southern Europe and the Near East predating 15,000 years ago. Our results suggest a limited role of human migration in the emergence of agriculture in central Anatolia… In contrast, we find that the later ACF individuals share more alleles with the early Holocene Levantines than AAF do, as shown by positive D(ACF, AAF; Natufian/Levant_N, Mbuti) ≥ 3.8 SE (Fig. 2b, Supplementary Figures 4, 5, and Supplementary Data 3)…Anatolian Aceramic farmers inherit about 10% of their genes from a gene pool related to the Neolithic Iran/Caucasus while later ACF derive about 20% of their genes from another distinct gene pool related to the Neolithic Levant.

Accordingly, we find that Iron Gates HG can be modeled as a three-way mixture of Near-Eastern hunter-gatherers (25.8 ± 5.0 % AHG or 11.1 ± 2.2 % Natufian), WHG (62.9 ± 7.4% or 78.0 ± 4.6%, respectively) and EHG (11.3 ± 3.3% or 10.9 ± 3%, respectively); (χ2p = 0.308 and χ2p = 0.589 respectively; Supplementary Tables 4 and 9).”

Simon Stevin said...

@ Davidski and all

Looks like parts of the “Southern Arc” paper have been released, finally…

Moesan said...


On my poor readings I thought the first metallurgical technics and models into western Steppes came from Balkans through Cucuteni-Tripolye cultures, along with some races of cattle. Only later came the influences from Caucasus. CWC in Ukraina took lately some kind of advantage of the Balkans-Tripolye matallurgic skills. So one (importatn) part of the EEF or ANF-like admixture in North Black Sea area could easly come from East-Central Europe whatever the part of this component in Late nothern Caucasus groups. I think it has been admitted that a part of the firstly sedentary Tripolye people shifted towards a more stock-breeding half nomadic groups.
Sorry for my short English.

Wise dragon said...

Hi Davidski, I read today the Lazardis paper(pdf): The genetic history of the Southern Arc:
A bridge between West Asia and Europe

Lazaridis concludes 2that All ancient Indo-European speakers can be traced back to the Yamnaya culture, whose southward expansions into the Southern
Arc left a trace in the DNA of the Bronze Age
people of the region." He notes that "The ancestry of the Yamnaya was, by contrast, only partly local; half of it was West Asian, from both the Caucasus and the more southern Anatolian-Levantine continuum."

So, according to Lazaridis Yamnaya had Levantine/Natufian admixture too.

What do you think?

Rob said...

@ Moesan
You make perfect sense.
Again, I do not deny important influences from Caucasus. I was calling for understanding this important zone 10 years ago.
But this was a 'flash-in-the-pan' Which is why Philip Kohl called it a 'singularity' lasting few hundred yeras.

@ Gaska
If Z2013 is from the Caucasus, then absolutely - we can state that the south Caucasus played a very important role in forming Yamnaya
But the chances of that are almost Zero. R1 is associated with ANE from central-western Siberia. It was in EE by 20,000 BP (my educated guess)
Even though CHG itself has ANE admixture, this was an exogenous intrusion

Rob said...

This proves my contention - early dialects of PIE somehow relate to I2a rich hunter-gatherers from east-central Europe. Of course, Im not saying IE is a WHG language or anythign sillyliek that.
A lot of people were giving me cheek for this , but its established.

Rob said...

@ Wise Dragon

thanks for sharing,

-> 'no steppe ancestry in Anatolia"

TUR_Aegean_Izmir_Yassitepe_MBA I2a1b1a2a1 (Dnieper Donets, aLPC, Yamnaya, Bell Beaker)

TUR_Aegean_Bodrum_Halikarnassos_Anc1 - I2a1b1a2. Still there in Roman period

A lot of people were giving me cheek for my theory: early brances of IE appear to be related to I2a-rich groups from east-central Europe. A clear link between Dnieper, Balkans & Anatolia
There you go

-. It (R1b-V3616) is also found in the piedmont of the North Caucasus at Progress-2 (17), the open steppe at Khvalynsk II (9), and the Single Grave Culture of Northern Europe (Gjerrild) (33). The individuals from Armenia and Arslantepe lack any detectible Eastern huntergatherer autosomal ancestry (Fig. 6C), which is maximized in the Khvalynsk individuals, an observation that provides some evidence for a southern origin for the R-V1636 haplogroup

haha yeah right.

Rob said...

I have to apologise to the Meshoko folk . They appear to be a good source of 'west Asian ancestry' in Yamnaya.

Caucasus_Eneolithic (Meshoko)

right pops:

Note the extensive pRIght list of WA-related outgroups.
Majkop was rejectd as a source, F2 stats called for somethign higher in CHG

best coefficients: 0.431 0.569
SEs : 0.015
Tail Prob 0.23

Davidski said...

OK, so if we ignore much of the paper and especially those bizarre claims about R-M269 and R-V1636, as well as about the Indo-Anatolian homeland, then this is a really groundbreaking moment for us.

The samples actually corroborate the linguistic hypothesis that the PIE homeland was on the steppe.

Very cool moment.

Copper Axe said...


Did you see the sample with R-M417 from Romania?

I6184/ROM23, Grave 18 (premolar), genetically male.
This grave belongs to Starčevo-Criş communities, and it was identified 1t 10 m
west from house 11. There was identified a single individual, in a crouched position, on
the left side, oriented on directions close to east. Funerary inventory was not found in
this grave. Instead, numerous ceramic fragments have been identified over and under the
body of the deceased, which represent a type of funeral treatment specific to the
Starčevo-Criş communities. The skeleton belongs to a mature male, 40-50 years old
(Popusoi 2005; Lazar 2021).

Dates to 5600 BC (6000-5300 BCE).

Copper Axe said...

Looking at Bulgaria, gonna have to give Rob some credit because there is a whole bunch of I2 there in the chalcolithic. The sample with Q-BZ1499 has unclear dating (3000-1300 BC), hopefully his genetic profile will solve the situation. Also I'm personally interested in I20186 (R-Z93 and U5a1a1), I wonder if its a Cimmerian/Scythian or a LBA migrant from the caspian steppes/central asia, we have a bunch of those, those Moldovan samples for example as well as "Cimmerian" MJ31 from the Belozerska culture.

Copper Axe said...

The usage of CHG over Iran_N, despite them not acknowledging CHG for years is kinda odd right? But it makes sense if you consider CHG is a tad bit closer to EHG. Almost as if you could make small amounts of EHG disappear when using CHG instead of Iran_N...

Rob said...

Thanks Axeman. FFCourt and BB blogger had made similar observations I think

gamerz_J said...


So what is to be made of their argument that Yamnaya have Anatolia/Levant_N over EEF for their "2nd West Asian" wave? I am assuming here they mean something like Caucasus_En or from the surrounding region, however is this not contradicted by WHG uniparentals among some Steppe samples?

I am not really familiar with archaeology so maybe this is plausible

They also inferred a very high amount of Natufian ancestry for Anatolian farmers, even at the Barcin site (though their models were not very clear on that) so I am guessing the argument is that Natufian/Levant_N ancestry spread agriculture mostly by demic diffusion.

Davidski said...


Basically, this paper isn't worth reading. But the samples are interesting.

I'm going to clear my weekend schedule and get to work on them.

The G25 coords should be available within a day. I'll post them here first.

Rob said...

@ gamerz J

To my surprise in my above screen dump, and without having access to any new samples, Cauc.Eneolithic / Meshoko-Darkveti gives best fits for Yamnaya (second wave), in addition to the CHG inherent in Khvalysnk (ie the Volga-Caspian pottery network, or whatever we may wish to call it)
The presence of I2a2 in Yamnaya groups is not contradiction to this. They were a relative minority, and uniparentals can take on new G-w ancestry within a few generations.
Then for CWC and western Yamnaya groups, EEF-related admixture comes very important too

So yes, I agree with their 2 waves deduction, but the overall conclusion is problematic becasue they ignore or are unfamilar with uniparental data, archaelogy and all other modes of collateral evidence necessary to be form a well=shaped conclusion

Barcin Farmers would be AHG + northern Fertile crescent. Whilst the latter would share ancestry with Natufians, direct natufian contribution seems lacking as there is no Y-hg E-M78 in Anatolia, but does appear in the Adriatic and Iberian Farmers..somehow

Rob said...

Interestinc Cyprus PPN are H2. Woould have expected some Y-hg E there

Modas Califa said...

If I understand correctly, what they propose is that the Indo-European family is a subfamily of a broader group that they call "Indo-Hittite". The Indo-Hittite family arises from the South Caucasus and somehow influences the Yamnaya culture. Within linguistics, is there any evidence for this?

Wise dragon said...


And 99% of Indo-European speakers stem from
Corded Ware ancestors. It is only three small groups:
Greeks, Armenians, Albanians who go up to the
Yamnaya not via Corded Ware intermediaries. Many
others were wiped out linguistically, e.g. Tocharians
and most Paleo-Balkan speakers

CHG Chad said...

Davidski is right that the paper is overated. But it contains a lot of samples to play. David if you have any info about the Samsun outlier with the EHG/Steppe and East Eurasian admixture plss share it. He was Scythian-Sarmatian???

Davidski said...

@Modas Califa

The Indo-Hittite family arises from the South Caucasus and somehow influences the Yamnaya culture. Within linguistics, is there any evidence for this?

No, it's bullshit.

Copper Axe said...

My issue with Meshoko being the sole/main donor of ANF in Yamnaya/CWC/Sredny Stog is the same with Maykop tbh, I fail to see how double digit amounts of eneolithic Caucasus ancestry became ubiquitous amongst Sredny Stog populations, while steppe_en samples across the 4th millenium b.c, contemporary to SS, Meshoko and Maykop, and geographically closer to the Caucasian populations don't have it. I know a future sample will but its on an individual basis rather than across the whole population (which shows there wasnt some kind of genetic barrier between the two populations). Same goes for Steppe Maykop, this stream of ancestry is only present by way of outliers, and this is easy to show by way of uniparentals. Unless there were some recent discoveries I'm unaware of Meshoko only shows up halfway during the 4th millenium BC.

Hopefully some of those early Bulgarian samples like the R1b Smyadovo sample have decent enough coverage to look at their autosomes, it will be interesting to see if they already have ANF as its date is contemporary to the first few centuries of the Meshoko culture.

Matt said...

I don't think they've really got the goods on confirming that the wave that enriched Anatolian in Yamnaya/Sredny is from WA rather that SE Europe. But also not the reverse. Papers from Ghalichi Ayshin and the Patterson paper that has been slightly leaked out will hopefully provide more confirmation one way or the other via finding the presence of admixed individuals from one direction or t'other.

I don't think Laz should have given this that level of prominence in the abstract, there doesn't appear to be enough weight, though I'll read their models more with interest.

Copper Axe: "The usage of CHG over Iran_N, despite them not acknowledging CHG for years is kinda odd right? But it makes sense if you consider CHG is a tad bit closer to EHG. Almost as if you could make small amounts of EHG disappear when using CHG instead of Iran_N..."

You could say vice-versa too. "CHG is real, distinct and important for West Asia. Wait, if it makes steppe ancestry go, maybe it's not!".

The real check that's neutral to all this bsing back-and-forth between the two sides is asking if they've done models with proximal sources rather than distal ancestors. Can they explain things in Turkey with proximal neolithic Aremenia, eneolithic Caucasus without resort to steppe? If neoArm+neoAnatolia works for ex, then that's conclusive.

Wise dragon said...

I personally don't buy this Yamanya were part Natufian/Levantine part which is being pushed by Lazaradis. There are many people who literally worship geneticists and think they are always right and everybody who questions them is either an idiot or has an agenda.

Matt said...

@Copper Axe, it's a very good point on lack of any DVM related in sampled steppe_eneo, but also there are few samples at the moment and it's not unusual for us to encounter some unadmixed HG in Europe even when we know admix was happening in a region more or less contemporary. If Ayshin Ghalichi or someone else can pull out a wider set of samples in this region and time, including from a wider range of sites and the western region, then if there's nothing that will strengthen the case that it couldn't possibly be mediated via Caucasus. Or alternatively if archeological nothing usable is found while we have pre-SS admixture trail from the west, that will probably push it the other way.

Rob said...

w.r.t CHG vs Iran_N, you would have to run an extensive battery of F4s instead of chosing arbitrarily or personal preference. These have shown that CHG should be used for Caucasus populations, and Iran_N for most but not all (!) Turan & Iran region Pops. Ill elaborate later

@ Copper Axe

Meshoko being a major source for Yamnaya makes sense though (of course in addition to the CHG in the 'Neolithic phase'). Of course, CWC needs EEF in addition


best coefficients: 0.341 0.361 0.239 0.059
TP 0.54

RE: ROM 023 - the contextual date inference must be wrong. Kris/ eastern Starcevo sites were reused by later groups as burials.

Rob said...

Apparently there are Yamnaya like genomes from 4000s Bc
Majkop only reallly gets going after 3600. Too late

gamerz_J said...


Does Meshoko provide a better fit that Trypillia farmers? I am not sure that's settled. And what if this "Levantine" we are seeing is due to some type of Iran_N in Progress (or Iran_N-like) or a more basal (perhaps less EHG) version of CHG?

Regarding Natufians and Barcin, well it's obvious Barcin has some Natufian-like ancestry, but imo a lot less than what they suggested (up to 43%!!) unless by Natufian they mean Kebaran-like... Feldman et al (2019) had a lot more Pinarbasi for example and Lazaridis himself had ANF as mostly Dzudzuana-like. Natufians had additional affinities with Taforalt. Barcin should almost certainly have European HG ancestry as well

I think most of West Asia by the Late Neolithic at least, had both Iran_N and CHG. Hence why it's hard to differentiate between them. In one of their papers they understood that in the other they focused on CHG a lot more than they should have and I think that's what CopperAxe is criticizing.

"I don't think they've really got the goods on confirming that the wave that enriched Anatolian in Yamnaya/Sredny is from WA rather that SE Europe"

Probably the best way to check this is if Yamnaya have WHG and lack Levantine. Not sure what's the best method to do that, but if they do have WHG and lack Levant it's 100% from Europe, if they don't have WHG and have Levant it's ofc 100% from West Asia. In this paper they argue for the latter scenario but I noticed they found higher amounts of Levantine ancestry across all pops they tested, compared to earlier papers at least.

Copper Axe said...


"You could say vice-versa too. "CHG is real, distinct and important for West Asia. Wait, if it makes steppe ancestry go, maybe it's not!"."

Not really because it is Harvard that for years opted to not mention CHG and use Iranian-related. We so have significantly more Iranian neolithic samples than CHG, that span a larger geographical distance (Gabj_dareh to Hotu for example). Both of these should be relevant for their proximate neighbours, which for CHG is the steppes, eastern Anatolia and western Iran perhaps. For Iran_N its the entire middle east, central Asia and South Asia.

Copper Axe said...

"Unless there were some recent discoveries I'm unaware of Meshoko only shows up halfway during the 4th millenium BC."

Correction: I meant halfway 5th millenium BC (4500 BC), my bad.

Matt said...

Re modelling choices, in the neolithic paper's supplement :

"Caucasus hunter-gatherer vs. Iranian Neolithic ancestry

CHG and IRN_Ganj_Dareh_N often appear as interchangeable sources for Neolithic populations and we wanted to determine if we can differentiate between the two. We thus fit the two 3-way models: (CHG or IRN_Ganj_Dareh_N) + Pınarbaşı + Natufians, i.e., setting one of the two populations as a source and including the other in the right set of outgroups (Table S 4). For many Test populations both models cannot be rejected, and so for them the “CaucasusIran” influence does not clearly stem from either the CHG or Neolithic Iran. For two populations (AZE_N and Mesopotamia_PPN) which neighbor both Iran and the South Caucasus both models are rejected, suggesting that while these populations clearly have some “Caucasus-Iran”-related ancestry as suggested by the modeling of Table S 2, they cannot be modeled with only one of the two to the exclusion of the other: a possible interpretation is that these geographically intermediate populations possess ancestry related to both their Caucasus and Iran neighbors. For the Barcın Neolithic, the CHG model is rejected (1.13E-03) while the Iran one narrowly accepted (0.0142). This population can be modeled to derive some of its ancestry from the east of NW Anatolia, and thus potentially from both CHG- and Iran-related sources. Finally, for the South Caucasus Neolithic at Aknashen, the CHG model is not rejected (p=0.46) while that with Iran as a source is (p=3.23E-04), suggesting that in the South Caucasus the Neolithic population can be modeled with CHG ancestry alone."

Aknashen is one of their two Armenia Neo samples (the other they suggest is a Levant ancestry rich potential outlier).

Matt said...


"I think most of West Asia by the Late Neolithic at least, had both Iran_N and CHG. Hence why it's hard to differentiate between them. In one of their papers they understood that in the other they focused on CHG a lot more than they should have"

I think you can talk about the CHG and Iran_N and why they choose to have adopted it now, and we criticise that if there's no good modelling reason to do so e.g. qpAdm dies when trying to use both of something - if that's the case then just using CHG to get a stylised model is kind of defensible. And its fair for you and CA to question this and we should check the papers to look for a reasonable explanation.

But ultimately what really matters is whether there is a discontinuity at any point that requires EHG and can't be met by previous populations in Anatolia and West Asia that are already required to be connected. The more abstract distal modelling choices are not really the main important thing. If we ended up saying "You can model LBA Anatolia with earlier HG+Neolithic+CA sources in West Asia that already have geneflow and some reasonable assumptions about unsampled populations on these clines... Or you can choose some random distal model which spikes Iran_N and gets steppe/EHG in there" it would be very hard for us to justify anything like the latter for'ex.

Tigran said...


I would love it if your theory on I2 being the PIE lineage was right. But wasn't I2 a minority in the steppes never reaching more than 20% in any time period? Also if IE is related to R1 carriers do you think it traces back to the Sunghir or Tianyuan like ancestry in ANE/EHG?

Davidski said...

Yeah, the PIE homeland was in southern China.

Well done. Go publish a paper.

John Smith said...

Hi Davidski, in this sheet you can find the date and coverage for each sample of the Southern Arc paper.

What I wanted to ask you is, will you be able to do the coordinates for most of these samples? I think many of them have too little coverage.

Romulus the I2a L233+ Proto Balto-Slav, layer of Corded Ware Women said...

5500 BCE R1a from Romania and already autosomally a perfect match for CWC.

Davidski said...

I6184/ROM23 must be wrongly dated. No C14 date, so not too surprising.

Distance to: I6184
0.03159564 RUS_Fatyanovo_Moscow_BA:HAN002
0.03717467 Corded_Ware_CZE_early:VLI011
0.03774062 Corded_Ware_CZE_early:RDV001
0.03776144 KAZ_Dali_MLBA:I3448
0.03852156 Corded_Ware_CZE_early:TRM003

Davidski said...

OK, can someone pull out the relevant Arc samples and label them? I don't know what the hell I'm looking a here?

Simon Stevin said...

Don’t know if this will help, but Ted Kandell made a preliminary list of all samples:

Nathan Paul said...

Not sure what difference all these discussions make.other than Abba Dabba dooo. F evolved in south Asia /south west Asia. K2a2 evolved in east Asia. K2B2 evolved and spread in west/east Europe. we get it.

Matt said...

@Davidski, here's a quick automatch between the S Arc supplementary datasheet and the scaled datasheet you uploaded.

Quick and Dirty Match:

I can't be totally confident its got all of them, because the Reich lab seem to have a habit of using extensions like .SG, .All, etc. in their anno/ind files that they don't use in the supplementary information, which makes things difficult.

But it looks like all the ones in their supplement that didn't match (175/777) had less than 105k SNPs, so I assume that that just means that you weren't able to get them on.

Many of the remaining samples in that datasheet matched the latest Human Origin anno. The 972/4543 that didn't match look to have not matched automatically largely because of these .SG, _published, etc annotations they put on the label that are annoying to filter in and out to get them all matching.

Matt said...

Oops actually posted too fast, this is a better link for my quick and dirty automatched datasheet:

Matt said...

Quick set of Vahaduo distances showing which SArc samples are closest to various modern and ancient averages:

Matt said...

Btw, hate to be doing this when it's what the paper's about, but what do you guys think of the Serbia Iron Gates Middle Bronze Age WHG rich outlier, I5243

"Balto-Slavic drift" or just HG rich?

Date 2458-2238 calBCE, y-dna I2a1b1a1b1b1~, extremely generous SNP count 837553.

Arza, Ambron?

Rob said...

@ gamerz J

''Does Meshoko provide a better fit that Trypillia farmers? I am not sure that's settled. And what if this "Levantine" we are seeing is due to some type of Iran_N in Progress (or Iran_N-like) or a more basal (perhaps less EHG) version of CHG?''

There's no direct Levantine there. Harvard have just a strange way of wording things combined with the fact they often can't see the forest for the trees

Meshoko worked better than Trypillia in my current set up. Distantly, Meshoko has 'Levantine' ancestry because all these groups, from LBK to Copper Age Iran have Levantine due to the expansion of PPN which came from Fertile Crescent

@ Tigran

Ill explain for the purpose of others if anything.
I have never stated that PIE comes from I2 but i did say that some ''early branches'' (so-called) appear to correlate with a specific subclade of I2a. In reality, these early branches are flat - they are not chronologically that much earlier than Yamnaya or CWC. The implication is that 'PIE' was already structured before dispersal, most likely due to the different networks within the broader HG cline which stretched from the Danube to the Urals. But given that i2a keep popping up, it does ground it toward the western ares of the steppe, despite the illusions people have come up due to their own misunderstanding of genome-wide data.

Sunghir & Tianyuan have nothing to do with PIE, neither does China.

@ Matt

''CHG and IRN_Ganj_Dareh_N often appear as interchangeable sources for Neolithic populations''

They're distinctive although similar. Using them laisez-faire will skew the result incorrectly. E.g. if IranN is used, the result will require extra EHG which isnt necessarily there. in order to choose correctly, need to be guided by F4s, Y-DNA, 'historical plausability', etc

Davidski said...


Working out fine scale admixtures with distal models is bullshit.

Anyway, do you have a link to the sample descriptions? Like an excel or something?

Copper Axe said...

By the way who remembers the R-M269 Smyadovo sample (4600-4000 BC) from Mathieson's 2018 article?

Target: 337:I2181
Distance: 5.7166% / 0.05716584
57.6 AUT_LBK_N
29.0 Yamnaya_RUS_Samara
13.4 UKR_N

I think most of that Ukr_N is from the farmer side however. Kinda expected him to have more steppe ancestry but oh well.

Davidski said...

@Copper Axe

Since you're lurking, can you tell me which of the TUR samples here are closest in space and time to some well known Hittite and Luwian sites?

vAsiSTha said...

Arm_Akneshan_N is simply Hajji_Firuz_N.
Arm_Masis_Blur_N is simply Azerbaijan_LN.

Hasanlu, MBA, LBA * IA samples are very interesting.

Have run various rotating models on my blog.

Romulus the I2a L233+ Proto Balto-Slav, layer of Corded Ware Women said...

That EEFs could contribute language with autosomal only and no Y chromosomes is unthinkable but for people south of the caucasus to transfer language to the yamnaya with only autosomal and no Y chromosomes is apparently some new genius hypothesis. It's weak. Only possible way I see IE from south of the caucasus is through western anatolia where J2 actually shows up.

Tigran said...


That makes sense and is pretty convincing. Does PIE have an original lineage though?

«Oldest ‹Older   1 – 200 of 224   Newer› Newest»