search this blog

Saturday, December 12, 2015

The Scythian

Time to have a look at the Scythian steppe warrior from the Mathieson et al. dataset. This is the first Scythian individual to be genotyped.

He comes from the Volga steppes of 380-200 cal BCE and belongs to Y-chromosome haplogroup R1a, which is the dominant Y-haplogroup in Scythian and related remains tested to date.

His genome-wide data puts him closest to Northeast and Northwest Europeans from among present-day populations, rather than West and South Asians, who should, in theory, carry significant Scythian ancestry. We can probably put this down to the complex ancestry of West and South Asians.

Moreover, he can be modeled as a mixture of Middle Bronze Age Potapovka people of the Volga steppes and present-day Nganasans of Siberia. This gels rather nicely with archaeological evidence, which suggests that Scythians were the descendants of Bronze Age Eastern European migrants to South Siberia, who expanded west across the Eurasian steppe during the Iron Age and eventually ended up back in Europe.

Identical-by-State (IBS) similarity

Lithuanian 0.645247
Estonian 0.645233
Latvian 0.645024
Russian_Kostroma 0.644946
Irish 0.644902
Orcadian 0.644792
Norwegian 0.644754
Belorussian 0.644727
Swedish 0.644667
Polish 0.644664
Austrian 0.644639
Danish 0.644587
English_Cornwall 0.644556
Belgian 0.644552
Scottish_Argyll 0.644548

Full output available here

Outgroup f3 shared drift statistics

Estonian 0.313726
Latvian 0.313664
Lithuanian 0.313574
Russian_Orel 0.313346
Finnish_Southwest 0.312997
Orcadian 0.312768
Norwegian 0.312768
Belorussian 0.312676
Russian_Kostroma 0.312669
Swedish 0.312608
Karelian 0.312567
Polish 0.31243
Irish 0.312281
Polish_Estonian 0.312156
Finnish 0.312102

Full output available here

qpAdm mixture model

Potapovka 0.913
Nganasan 0.087
chisq 5.815 tail prob 0.213365

Full output available here

Mathieson et al., Genome-wide patterns of selection in 230 ancient Eurasians, Nature, Published online 23 November, 2015doi:10.1038/nature16152. Genotype dataset available here.

See also...

Genetic origins and legacy of the Scythians and Sarmatians

The Poltavka outlier


mickeydodds1 said...

It is strange that this ancient Scythian chap should show so much IBS similarity with a modern Irish sample.
Ireland, is of course, an isolated island situated right off the westernmost periphery of Europe, and bathed by the Atlantic Ocean. As far away from the Steppe as you can get in Eurasia.
Indeed, it has long been held that the Irish are relict Mesolithic population, who were basically 'passed-by' by the mainstream developments of continental European demographic developments. It has also been long held that Ireland was the 'refuge of last resort' of Mesolithic types forced out to utter westernmost fringes of Europe.

Rami said...

There were a plethora of Indo Iranian groups which were present in S/SC Asia by antiquity. A lot of them had already long admixed with the larger Near Eastern like populations present as well as with Proto Turks in the Altai for some groups. The Tocharians another group were admixed with East Asians in the Bronze Age.
Western Scythians are not the same as Eastern Scythian groups by antiquity.
Most West Asian people do not have any Scythian ancestry. In the case of S/SC Asians its even more complex because you have different Indo Iranian groups appearing at various times.

mooreisbetter said...

I first note as an aside, there is no accepted definition of Scythian, and the ancient sources are deplorable in their application of the term.

I expect many more surprises like this from aDNA. Due to popular demand, I have posted an updated post about How Little We Know About aDNA:

Your comments are welcome. With this finding and others, it looks like 2016 will be a very humbling year.

DMXX said...

I take both mooreisbetter and Rami's points, which are clearly correct. The issue of admixture with other contemporary populations in Eurasia is, I suspect, different from what Davidski's attempting to demonstrate here, which is the firm continuity (and predominance) in the "genetic constitution" of the Iron Age steppe samples with their Bronze Age forebears.

As far as my reading permits, Sintashta is considered by many linguists (and most Russian archaeologists) to be the best candidate for the culture defined by the proto-Indo-Iranians. The Andronovo archaeological horizon is also considered a (slightly) later extension of Sintashta deeper into Asia beyond the trans-Urals region.

We see remarkable genetic unity between Corded Ware, Sintashta and Andronovo (as demonstrated and commented upon by Allentoft et al.), which date to around the Bronze Age. More East Eurasian admixture seems to appear among steppe samples as time progresses. However, both the Iron Age steppe remains (Karasuk, Scythian_IA) remain predominantly CW-Sintashta-Andronovo derived.

Perhaps the pastoralist nomad populations as described in antiquity by the ancient Greeks and Persians were substantially different than the Iron Age samples. Using the Bronze->Iron Age changes as a guide (though no guarantee), we can expect some additional admixture from surrounding populations wherever we may find remains that can be identified as physically "Scythian", with the Bronze Age input still persisting as a majority.
Perhaps again, in some locales, populations adjacent to the steppes did become "acculturated" as Scythians. There's plenty of room for surprises. We'll have to see where the evidence leads us. I too look forward to the developments ahead.

How things developed in West and South Asia is another discussion entirely, since these locations aren't within steppe territory and already had established sedentary populations going back several millennia.

Alberto said...


But is Potapovka the best match? Shouldn't it be Sintashta/Andronovo, with maybe some Afanasievo and East Asian?

Why did the Scythians eventually end up in Europe? Weren't they R1a-Z93?

A bit surprising that it shares so little with Tajiks (more or less Spanish levels). S-C Asians drop quite lower. Sintashta and Andronovo must share even less, I'd suppose. Not great to support a mass migration from these types to S-C Asia.

But if they had little autosomal impact in S-C Asia and little Y-DNA impact in Europe, what happened to them? Maybe they were just too few and diluted into bigger populations at some point?

Coldmountains said...

@ Alberto

I would not take the results too literally. Scythians were a mix of Potapovka-like Indo-Iranians and some Siberians and should cluster because of that with modern Erzya, Moksha, Mishar Tatars and Mordva but here they share not much with them and share less with them than French. But the similarity to north Europeans is obvious and makes of course sense even when I expect Volga Finns and Russians to be on the top of the list. Tajiks have of course more direct genetic ancestry from Scythians than Spaniards and also than all other Europeans but yes modern Indo-Iranians are more mixed with local non-IEs than Europeans who better preserved steppe ancestry

Coldmountains said...


I agree. This Scythian sample was from the Volga region but Scythians lived not just there and in Siberia but in many other places. Scythians north of the Caucasus were heavily mixed with North Caucasians. Scythians in Crimea and south Ukraine with Greeks, Thracians and other folks and I would not be surprised if they were between modern Ukrainians and Crimean Tatars/North Caucasians. Scythians in South Central Asia and the tarim basin were mixed with BMACs and probably genetically also very different but all of them were in the end dominated by R1a and descendants of Sintashta/Andronovo.

Fanty said...

"Indeed, it has long been held that the Irish are relict Mesolithic population, who were basically 'passed-by' by the mainstream developments of continental European demographic developments."

Well, that all based on skull shapes stuff. Rather outdated in the time of genetics. ;)

I recall that the skullshape called "borreby" was originally thought to be a race. Actually something mesolithic central european. As far as I know, the latest idea of the original "borreby" skull is, that its the result of a "recent" (1 or 2 generations ago) mix between a Bellbaeker and what lived in Denmark until then (something late neolithic like Funnelbeaker or so)

And SUDDENLY people talk about that type of skull was "popping out of nowhere" at like 5000 YBP. Really? Why did anyhody ever thought its something mesolithic central european?

Fuck skull shapes. ;)

Coldmountains said...

@ Davidski

Why Scythians get ASI on many calculators? The high CHG makes them looking like being admixed with South Central Asians but they probably were just mixed with some North Caucasian CHG-rich people.

puntDNAL K10 Ancient M374116 Scythian

Admix Results (sorted):

# Population Percent
1 CHG 38.64
2 WHG 32.43
3 ASI 6.77
4 Siberian 6.27
5 ENF 6.13
6 Beringian 2.82
7 E_Asian 2.75
8 Amerindian 2.25
9 Sub-Saharan 1.92

Using 2 populations approximation:
1 50% Russian +50% Pashtun @ 10.286580

Using 3 populations approximation:
1 50% Russian +25% Lezgin +25% Burusho @ 7.501342

Using 4 populations approximation:
1 Lithuanian + Chuvash + Lezgin + Burusho @ 7.152692
2 Lithuanian + Chuvash + Lezgin + Kalash @ 7.264015
3 Russian + Chuvash + Lezgin + Kalash @ 7.462746
4 Russian + Russian + Lezgin + Burusho @ 7.501342
5 Estonian + Chuvash + Lezgin + Kalash @ 7.508308"

Davidski said...

Modeling ancient samples as modern samples is usually not a good idea, since the latter will have admixture from the former, not the other way around.

Grey said...

"It is strange that this ancient Scythian chap should show so much IBS similarity with a modern Irish sample."

If an R1b population from the steppe expanded first - including to Ireland - and then that steppe source population were later displaced by a nearby R1a population then it wouldn't be surprising to find a "Scythian" with a lot of the same ancestry that went into the Irish but with the y dna switched.

Coldmountains said...


This Scythian was not clustering with Irish and had unlikely Bell Beaker R1b ancestry which is the main genetic component of modern Irish. He was a NE European with some elevated CHG and Siberian ancestry. Most similar to modern far east Europeans of the Baltic and Volga region.

Davidski said...

Here are some qpAdm models for the Scythian involving South Asians.

Rob said...


Can you do a similar exercise for the Hungarian "Cimmerian".
Its a good comparison to see what Iron Age steppic groups like like (even the IA HUngarian was hg N).

Davidski said...

I will if they re-sample the Cimmerian so that there are more SNPs available.

This Scythian has three to four times as many markers as the Cimmerian.

Rob said...

OH, so the markers from Gamnba are too low for your K10 ?

Davidski said...

It's just hard to run that sample properly with the markers that are available, and unlike most of the other Gamba genomes it wasn't re-sampled by Harvard.

I'm sure we'll see some good quality Cimmerian and Sarmatian samples from all over Eastern and East-Central Europe next year.

DMXX said...


As per usual, agreed. :)


Good question. There's two scenarios I envisage here. Either Scythian_IA's immediate ancestors were indeed from much further east and returned to western territories even further beyond Sintashta, or East Eurasian admixture diffused its' way into Sintashta during the Iron Age and this fellow represents a back-migration of Sintashta into the Volga-Urals. Highly similar scenarios, just a bit of nuance in explaining the East Eurasian admixture that wasn't seen in the Bronze Age.

Regarding why we'd even suspect a back-migration, Herodotus recounts his favoured origins for the Scythians in Histories that were known in antiquity:

" It is that the wandering Scythians once dwelt in Asia, and there warred with the Massagetae, but with ill success; they therefore quitted their homes, crossed the Araxes, and entered the land of Cimmeria."

The Volga-Urals approximates reasonably with Cimmerian territory (prior to their flight southwards from the Scythians) and Herodotus was alive 100-150 years before our Scythian sample is estimated to have deceased.

So, conflict with neighbouring Asian steppe nomads looks like the historical "best fit" as to how this individual ended up returning to Europe.

Grey said...

Cold Mountains

"This Scythian was not clustering with Irish"

The scenario I described wouldn't lead to them clustering.

Davidski said...

Not sure why you think Z93 is rare in Eastern Europe, since it's actually pretty common around the Volga and Urals where this Scythian was buried.

And we know Z93 didn't arrive there from the east during the Scythian era, but spread from the west to Asia during Poltavka, Sintashta and Srubnaya periods.

Btw, Kurti's point ignored the problem of projection bias. So there's no reason to believe that Scythians will be filling any gaps on PCA plots.

Rob said...

I wasn't suggesting that the Z93 Iron Age Scythians of Europe arrived from Iran; but the flip side is neither did the Iron Age central Asian Sakae derive from Europe (although we don't have yet any samples). My main point is that these Iron Age groups formed in their own vicinities through more circumscribed processes of acculturation and admixture.

Any similarities probably rest on an older Bronze Age (?even older) stratum, and individual exchanges between chiefs and tribes; rather than a large migration from Europe to Asia, or Asia to Europe, in 700 BC

Davidski said...

I don't see why Central Asian Scythians couldn't have been in large part derived from eastern steppe LBA-Iron Age groups, like the Altai Scythians. That's what the PCA suggests in fact.

South Central Asians are being pushed up towards the Scythian IA, Steppe LBA-IA and Altai IA samples.

DMXX said...

Hi Rob,

Your arguments seem preoccupied with the validity of Herodotus as a historical source. Of course, a great deal of his work cannot be treated as historically sound (such as his account for the Arimaspeans). However, some parts clearly do.

I mentioned him as an early example of a back-migration from the Asian to the European steppes before the ancient period (we know for a fact that the Sarmatians displaced the Scythians in the same manner as the Scythians and Cimmerians purportedly did as per Herodotus). That was not the cornerstone for my reasoning, although it coincides nicely with the following.

The basis for my reasoning of a back-migration rests on the East Eurasian admixture that appears largely absent in Sintashta and Corded Ware, which were situated either within or adjacent to the Volga-Urals, yet is clearly present in our Scythian_IA.

This is also why I highlighted the Karasuk and Andronovo remains, as both of these also contain significant (>5%) East Eurasian admixture in quite a few of the individuals. Note these were all sampled north of the Altai.

Unless you're aware of some contrasting genetic evidence I'm not (happy to be updated), the proposition of a back-migration from territories deeper in Asia is the parsimonious explanation for the Scythian_IA and Sintashta-CW autosomal profile mismatch.
Perhaps there was an Iron Age foreign bride program we have no evidence of (not sure what analogue Sintashtoids could have for credit cards :) ).

It's clearly meant as an off-hand comment, but invoking this "genetic hole" between modern Eastern Europe and South Asia is irrelevant to the discussion of what constitutes a "Scythian". Please note there are other explanations for this other than group extinction (e.g. East Eurasian admixture in modern Kazakhs, Uzbeks and Turkmen pulling them away on PCA's).

For those reading the comments, note I've made no mention of R1a1a-Z93 in case this becomes an unnecessary tangent.

DMXX said...

"My main point is that these Iron Age groups formed in their own vicinities through more circumscribed processes of acculturation and admixture.

Any similarities probably rest on an older Bronze Age (?even older) stratum, and individual exchanges between chiefs and tribes; rather than a large migration from Europe to Asia, or Asia to Europe, in 700 BC"

As hefty amount of aDNA is required to validate this, your proposition resembles a hypothesis here.

It's clearly possible (likely to have taken place in certain locales in my opinion)... Just cannot be treated as a unifying explanation until we have sufficient aDNA.
Specifically, we'll need genetic data before putative interactions with steppe vectors took place in certain areas, followed by afterwards, corroborated by any shift in material items coinciding with the "Scythian triad" or other items in keeping with a pastoralist nomad modality.

Until we have the above and the data aligns in your favour... It's a speculative hypothesis, not a unifying explanation.

Grey said...

Might having the best sources of iron weaponry have decided any changes of direction of migratory pressure on the steppe in the IA.

Kurd Dgk said...


Just wanted to share that I had the best fit yet for Kurds today modeling them as Scythia IA, Mota, and Kotias CHG in qpAdm. These are 2 unrelated Kurds from Iraq with ancestry from the Turkey's Kurdistan.

POPULATION Kurd_C2 Chisq Tail Probability
Kotias CHG 69.90% 0.903 92.42%
MOTA 0.80%
Scythian IA 29.30%

POPULATION Kurd_C1 Chisq Tail Probability
Kotias CHG 79.50% 1.2 87.80%
MOTA 2.80%
Scythian IA 17.70%

Kurd Dgk said...

Also, the fixed paths in qpAdm are indicating that Kotias CHG and Kurds are a near clade

Davidski said...

Thanks, those models do make sense to me.

Fred Mason said...

Interesting posts since I am R-Z93>>YP1451 with an Iranian as co-match. My umpteenth great grandfather...likely a Sarmatian guarding Hadrian's Wall has linked me to an eclectic collection of cousins from The Steppes and Caucasus.
We learn a bit more with each new burst of testing...both current and ancient samples.

Rob said...



* "The basis for my reasoning of a back-migration rests on the East Eurasian admixture that appears largely absent in Sintashta and Corded Ware, which were situated either within or adjacent to the Volga-Urals, yet is clearly present in our Scythian_IA"

Indeed, combined SEA and NEA on Dave's new K10 is ~ 11%

As for Herodotus, I;m not suggesting that we throw his books out, but Drew's explanation certainly fits, as Greek historians often made hypothetical connections based on perceived toponymic similarities. Moreover, Herodotus had different ideas about Asia, which he placed in the Near East - where (as i stated) the Scythians and Cimmerians **first appear** in Assyrian and Persian sources, and where archaeological evidence first documents cavalry accoutrements.

" the proposition of a back-migration from territories deeper in Asia is the parsimonious explanation for the Scythian_IA and Sintashta-CW autosomal profile mismatch. "

I'm very open to such as scenario. And as we've seen, big migrations certainly did happen, so we shouldn't retreat into a unrealistic scenarios of 'cultural diffusion' solely.

I agree with you, these Scythians were the first true equestrianized nomads and thus potentially very mobile. But to exactly quantify and qualify the extent and character of such mobility we need 'a hefty amount of aDNA is required to validate this" :)

The 'east Asian' admixture demands explanation, but at present any explanation must be hypothetical. Was it a mini mass-migration ? Was it part of a steady stream of exchange one would expect in a more 'Globalized' Iron Age world ? Did it arrive directly from the Altia, or was it mediated via an intermediary ?

@ Davidski

"I don't see why Central Asian Scythians couldn't have been in large part derived from eastern steppe LBA-Iron Age groups, like the Altai Scythians. "

They could. As above, as our Iron Age aDNA samples begin to accrue, we can shed more light on Scythians, etc

DMXX said...


I'm in agreement with all points. Like you, I don't expect massive population replacements all over the place and do expect a degree of nuance in some of what we're observing here.

Let's see how the coming year shapes up for all of us. I fully accept your point regarding Herodotus, by the way. Although he's one of our primary Western sources, there is such a thing as over-reliance. It's sometimes easy to lose sight of that as a result.

Thanks for the exchange. Hope you're having a good weekend.

Kurti said...


"Btw, Kurti's point ignored the problem of projection bias. So there's no reason to believe that Scythians will be filling any gaps on PCA plots."

Dave I didn't ignore any projection bias.

In fact in my thread more than a year ago I predicted that the some of the "Western Scythian" would cluster as close to East/Northeast Europeans as almost "blonging into this branch, while other Scythians as those close to the Caucasus, or Central Asia would cluster towards South_Central and West Asia. And here we are seeing a Scythian from the Volga river (a Western Scythian by definition) and EVEN if he is closest to "East Europeans on PCAs or aDNA, EVEN he cluster further towards East than any Eastern European today by having Greek levels of CHG and significant ASI admixture. Therefore, as we see on Couldmountains comment, can be modeled as Russian/Pathan and not "Russian or Lithuanian", because he has obviously strong ancestry which is typical for populations of South_Central and West Asia today. That means he has a strong eastern drift.

It would be simply beyond absurd to assume that Eastern or "Southern" Scythians, aka Sakas and Caucasus Scythians would score close to identical to this Volga Scythian.

It appears much more logic to assume that the Scythians who dwelled and occupied a region aa large as modern Russia (by size) were also genetically not so homogenous.

This is the map I presented a year ago. As clearly seen I placed allot of "Western Scythians" next to Northeast Europeans but never among them because I expected and knew even the most Western Scythians would have some "eastern shift" which modern Eastern Europeans would lack, basically out of the logic that they would have absorbed additional Western admixture via mainland Europe (Vikings and Slavic expansion for example). And I knew that even the most eastern and southern Scythians would never plot directly withing any modern South_Central Asian or Caucasians, because I expected additional ASI or Near Eastern ancestry in those regions.

Chad Rohlfsen said...

Yeah, GujaratiDs are pretty high in ASI. They get around 2/3 of what Paniyas score in South Asian clusters. I can do some qpAdm runs with them, like the Kharia one. The key to looking at ASI, as far as I'm concerned, is to first look at the Paniyas and Pulliyars. Look at their rate of Papuan to Austro, then gauge excess ENA in the Austro category. Not all of the ENA is ASI. Some will probably be from admixture from surrounding pops and incoming Scythians which will add some ENA.

Srkz said...

I also performed an analysis of this sample. Long IBD haplotypes distribution is similar to Davidski's IBS analysis in Europe, but differs in Central Eurasia.
Look at the high Kazakh, Bashkir, Uzbek results. Chuvash people (who seems to be a descendants of the Volga Bulgars) got the 1'st overall score. Turkic Balkar people (NW Caucasus) also share many segments with I0247. So i think this guy represents the Iranic tribes that were assimilated by the Turks later.
His East-Asian influence don't looks Nganassan-like, but Tuva- or Mongolian-like. He also shares many segments with Pamiri, Tadjik and Kalash people.

So i agreed, he is probably a descendant of Andronovo/Srubnaya-like people who has mixed with SE Siberians and migrated backwards. And he is the "ancestor" of many modern steppe Turkic people.

Rami said...


Those CHG scores are waaay too high for Gujaratis. Those CHG scores are eating all their South Asia specific Bedouin like ENF .

Those Tajik results are equally messed up. All SC Asians have ASI/EA in varying amounts.

Davidski said...

You're not making much sense there Rami.

CHG, previously "teal", is by far the most important West Eurasian component in South Asia.

This has been known for a while and the study with the CHG genomes confirmed it.

Davidski said...

The Admixture analysis in the Haak paper shows that GujaratiA do have a slither of EEF-related ancestry, but this drops to almost nothing for GujaratiD.

So the results here are consistent with that, because EEF-related ancestry is associated with the ENF/Anatolian Neolithic cluster, which shows up clearly in GujaratiA, but falls to noise levels in GujaratiD.

M. Myllylä said...

IBS-statistics are likely biased due to homozygosity. Southwest Finnish samples don't represent southwestern people in Finland if selected using PCA. This is true because PCA reacts mainly to North Asian and ancient Fennoscandinavian local components, not to the Bronze and Iron Age settlements in Southwestern Finland. So the result holds samples all over Finland showing a combination of most Stone Age genetic remains and genes from later (historical) migrastions from east and west. Unfortunately it is nhot possible to select southwestern samples without making INDIVIDUAL comparisons with identified southwestern samples. Never trust to PCA alone,

Alberto said...


My comment was about the K11 run that Chad posted before. There both Sweden_NHG score almost 20% Anatolia_Neolithic. This makes more sense, because the K8 is based on a much more basal eurasian cluster, and it's unlikely that Sweden_NHG mixed with some ultra-BebouinB-like population to get those 2% and 10% that they show in K8. Actually, even the 20% Anatolian is an underestimate of real admixture from nearby farmers, who already harbored quite higher WHG ancestry themselves, so the proportion with farmers from that time and place is probably closer to 25-30%.

The K8 is a very good run for being unsupervised and mostly in agreement with the previous K11, but I prefer the better resolution on K11 splitting ENF into different populations and origins, though the K8 has the advantage of having a lower baseline for WHG, which reveals better the level of it in Middle Eastern and SC Asian populations. So both are good and complement each other.

Alberto said...


Thanks for that IBD map. It's strange that there is quite a difference between IBS and IBD with this sample. And the latter seems to make more sense in this case, probably.

Usually, Admixture would behave closer to IBS than to IBD, if I'm not mistaken. While IBD is a slightly more specialized method to discriminate further the segments. But in this case Admixture would agree more with the IBD map you posted than with IBS.

Is there any technical explanation about when/why the 2 methods might differ significantly and the advantages/disadvantages of either of them? (question for anyone who might know).

Balaji said...

Ryukendo Kendow,

I am continuing the interesting discussion from the previous thread. As Tobus explained, the number of markers will not bias the D values, however, it will affect the z values. You had commented on the D values for Orcadian being higher than for LBK_EN. I reproduce the numbers below.

Chimp Ust_Ishim Orcadian Loschbour 0.0163 3.035 501362
Loschbour Kotias Ust-Ishim Chimp 0.0305 4.665 404058
Chimp Ust-Ishim LBK_EN Loschbour 0.0151 2.692 279178

There are a relatively large number of markers for all the statistics and so they should be reasonably accurate. Orcadian does indeed have less BEA than LBK_EN. But we know that Orcadian also has CHG. And this could explain why the D statistic for Orcadian is higher than for LBK_EN.

The other interesting topic was the statistics for Indian_Singapore. The numbers are below.

Chimp Ust_Ishim Indian_Singapore Loschbour 0.0134 2.43 368472
Chimp Ust_Ishim Indian_Singapore Kostenki14_UP 0.0048 0.768 405033

I do not know why Loschbour seems to show a basal component in Indian_Singapore but why Kostenk14_UP does not. In any case, Indians in Singapore are heterogeneous. Not all are middle caste Tamils. Many are Muslim Tamils with some Arab and even African ancestry which could explain the high basal component, Davidski probably has genomes of the Kallar from Metspalu et al. Kallar are truly a middle caste Tamil group. See the following about Rowthers.

The topic that I was most interested in is that of establishing that the largest part of ANI is indigenous and not a result of the migration of agriculturalists from the Near East. I think Davidski's Smarter Bear plot is the strongest evidence for this.

No combination of LBK_EN, Satsurblia, Steppe and ASI can create the pattern for South Asians that is seen in the plot.

Shaikorth said...

Balaji, changing the number of markers can cause significant distortions to both Z score and D score, especially when we are talking about hundreds of thousands. Lower number of markers will produce lower scores.

Anatolia_Neolithic Armenian Mota Primate_Gorilla -0.0085 -4.2 480195
Anatolia_Neolithic Armenian Mota Primate_Gorilla -0.0047 -1.858 114965

Srkz said...

"Usually, Admixture would behave closer to IBS than to IBD, if I'm not mistaken. While IBD is a slightly more specialized method to discriminate further the segments. But in this case Admixture would agree more with the IBD map you posted than with IBS."

Yes, IBS is closer to Admixture and usually represents the main Admixture components. In this case the main components are EHG/WHG and CHG/Teal, so the NW Europeans carrying both are in the top of the list, while Central Asians and Volga-Ural people with high amounts of East Asia are falling down. But the proportions of components are more similar to Volga-Ural people in this case, so the least-squares method is shifting I0247's position to the east.

I've remastered the IBD method, now i'm using relatively long and new chunks. In theory, now it represents more fresh ancestry, while other method represents relatively old ancestry.

M. Myllylä said...

Number of markers should should have effect on Z-values, it is sure. It is possible that also D-values change because the accuracy, indicated by Z, is lower. But I am not sure about the direction of change in D-value. The result can be affected by the snp selection. For example modern samples distributed by Estonian BC and belonging to same ethnic groups are much more heterozygous than samples downloaded from 23andMe and FtDna. So there is some kind of bias in selections.

Srkz said...

@M. Myllylä
"For example modern samples distributed by Estonian BC and belonging to same ethnic groups are much more heterozygous than samples downloaded from 23andMe and FtDna. So there is some kind of bias in selections."

What samples do you mean? I never seen any signs of this effect when calculated the heterozygozity levels in modern populations. Maybe this is due to different snp selection?

Kurd said...

@ Balaji

Number of markers does change D value. I see it all the time when I run Dstats. That is why it is important that the sample comparisons be made with a similar number of markers. So basically, the number of positive hits between 2 markers in the comparison is reflected in D.

@ srkz

How can you say anything is IBD with a 5000 year old genome. I would tend to think anything is IBS at that stage. Don't you work with segments that are only 3cM.(just curious)

M. Myllylä said...


"What samples do you mean? I never seen any signs of this effect when calculated the heterozygozity levels in modern populations. Maybe this is due to different snp selection?"

All EBC samples, There is no bias between ethnic groups, only in comparison with some other SNP selections availalbe publicly. But this has definitely a certain effect on dstat results, in comparison with other SNP selections and using ancient samples. I have done tests with many selections up to over million SNPs.

M. Myllylä said...

@Kurd, you are right.

Kurd said...

@ David

As an update, using qpAdm, the Kotias CHG/ Scythia IA / Mota combo produced good fits for all the Kurd samples I have (total 4 ), but for some odd reason it did not produce good fits for the 2 Iranian samples I manage. I have tried all kinds of combos for those 2 to no avail yet. Have you by any chance stumbled on a good combo for Iranians?

Matt said...

@ RK : Matt, thanks for pointing out the discrepancies between the EHG modelling vs the Yamnaya modelling.

The way I am think of the Yamnaya vs EHG modelling is that using Yamnaya rather than EHG sort of forces the modelling to give 2x Yamnaya* compared to what EHG would be added under an equivalent EHG model (to explain relatedness to the Native American outgroups relative to East Asian outgroups). That limits the space for other components and sort of may throw off the ratios of EN to WHG (the others seem to be in pretty close ratios between the Yamnaya and EHG models, actually). So I'd prefer to see them CHG, EHG, WHG and Anatolia_EN considered separately, and if ADMIXTURE's the only way to do that atm...

*(2.2 on average and never far from that)

About the WHG rising again, there is a steady increase in WHG from Yamnaya to Andronovo+Sintashta, perhaps whatever played a role here played a role in the rest of Europe? These changes can partially explain the increases in WHG in Eastern Europe at least.
I don't know if they really "explain" much necessarily (like how, Sintashta->Eastern Europe?) but possibly interesting that the same trend is noticeable there, sure, and thanks for mentioning it.

Dmytro said...

On the issue of the various strands of "Scythian" identity. It is believed that a large portion of the upper class "Paralatae" actually came in from the further east, and that their famous "animal art" was strongly influenced by Northeast Siberian forest populations.== As to the Scythian back and forth movements, that is also now well documented. The classical Scythians of Herodotus almost entirely abandoned the North Pontic steppes ca. 300 BCE, and returned to Central Asia. Two of the units mentioned by Herodotus in his work (the Catiari and Auchatae) were located north of the Syr Daria by 290 BCE, according to Plinus' source Demodamas. Nor were they pushed out by the Sarmatians. There are no Sarmatian burials west of the Don until well on into the 2nd c. BCE. The Skilur Scythian Kingdom based in the Crimea, but stretching from the Danube to the Don, subsequently defeated by Mithradatus Eupator, arose after the major shifts in steppe power politics which occurred ca. 175 BCE or so wwhen the "Hsiong Nu" made their move against the "Yuezhi", with attendant domino effects... I wonder if the Volga Scythian of these comments was actually a Scythian (perhaps), or a Yazig Sarmatian. Love this stuff...

Chad Rohlfsen said...

qpAdm results should be looked at with caution. No one is 80% CHG, 20% Scythian.

Helgenes50 said...

@ Chad
qpAdm results should be looked at with caution. No one is 80% CHG, 20% Scythian.

The Kurds ancestors are probably close to those of the Scythians
Both are supposed to be related to the same Iranian family
What explains this result

Kurd said...

@ Chad

No one said that Kurds are 80% CHG and 20%Scythian, because if that were the case chiaq would be 0, and tail prob would be 100%. The results I got with chisq values between 0.7 and 1.4 and tail probs between 85 and 95% suggest that the Kurd samples I manage, can be substantially be modelled as CHG/Scythian. Can the fit be improved by adding other pops in small percentages. The answer is yes.

Kurd said...

Here are the qpAdm results for the 4 Kurds:

POPULATION Kurd_C2 Chisq Tail Probability
Kotias CHG 69.90% 0.903 92.42%
MOTA 0.80%
Scythian IA 29.30%

POPULATION Kurd_C1 Chisq Tail Probability
Kotias CHG 79.50% 1.2 87.80%
MOTA 2.80%
Scythian IA 17.70%

POPULATION Kurd_E Chisq Tail Probability
Kotias CHG 70.30% 1.198 87.83%
MOTA 2.40%
Scythian IA 27.30%

POPULATION Kurd_SE Chisq Tail Probability
Kotias CHG 24.80% 0.378 94.48%
Yoruba 2.40%
Scythian IA 56.30%
Puliyar 16.50%

Kurd said...

Slight correction

The chisq range is 0.378 to 1.2 and tail prob 87 to 95%

Kurti said...

Kurd said

"As an update, using qpAdm, the Kotias CHG/ Scythia IA / Mota combo produced good fits for all the Kurd samples I have (total 4 ), but for some odd reason it did not produce good fits for the 2 Iranian samples I manage. I have tried all kinds of combos for those 2 to no avail yet. Have you by any chance stumbled on a good combo for Iranians?"

There is more sign of real Scythian presence in Kurdistan than other parts of Iran. It is said by Heredotus that two of 6 major tribes of the Medes might have come from Scythians. He names the Parateceni and Budii, who he connects with the Budini of Scythia. He also mentions Scythian state in Turkish Kurdistan known as "Sagapeni". Take in mind also the Scythian kingdom of Saqqez in the Kordestan province.

Also take in mind in the ~80% CHG most likely the majority is Mede derived CHG. I expect the Medes as a blend of predominantly Iranic nomads who merged with Hurrian kind of people to have been allot heavier in CHG than Volga Scythians.

It is also hard to seperated Scythian ancestry from general Medic because 1. They share allot of ancestry via their ethno-linguistic relationship. 2. Scythians, as eloberated above, were a significant part of the Medes themselves.

So ~20-30% of Scythian like ancestry makes sense.

Rami said...

@ Davidsky NO I am making perfect sense.

Your ignorance regarding S/SC Asians is appalling.
You actually modelled Pania as 65% Kotias.LOL. When the cluster more with Papuans.

ENF in South Asians , one via embedded in Teal/Baloch and the other is even more archaic and distantly related to Bedouin ENF

Synthetically painting a picture to suit your own lurid narrative without an ounce of genome data from the region is incredibly ignorant.

Nirjhar is right!

Kurd said...

@ Ryu

Thanks, I do try to check David's blogs regularly, even though I don't participate much. I appreciate his efforts.

@ Kurti

I have heard many of the things you mentioned from some of my Kurd relatives. With regards to modeling the 4 Kurd samples I have, I tried all kinds of combinations of ancients consisting of LBK, Anatolians, Bedouin, Andronovo, Sintashta, Afansievo, and Karelia HG. None gave the nice fits as Kotias CHG and Scythian IA.

Matt said...

@ Ryu: Thanks for the explanation. I guess your idea here is more that WHG is likely to enter post-EBA groups that are on such a frontier, then they might be more likely to replace or otherwise contribute outsize to the overall population connected to being more pastoralist?

(Hard to know what to make of those ideas of Turchin's or how he has evidenced and proved them. Couple of tangents to those ideas which it is probably not worth to discuss here as O/T but:

One thing that sprung to mind is that if a bunch of quite sedentary local scale cultures merge into a single empire / meta culture in which a given individual or family can easily travel around and migrate through the whole empire and sell and buy land through the whole area, surely everyone has become less sedentary (tied to a single place) than they were before, despite the society being arguably more centralised / large scale / denser?

So how does that work with an idea of sedentary cultures proceeding to become more sedentary, if I'm getting the idea right as it is expressed.
Another is whether what we really know about how reputation economies and tribal elders and codes of law worked before writing and written history in societies with different modes of production. (Whether pastoralist groups like the Mongols increasingly accreted a more reputation based economy over time or instead more elements of codes of law and impersonal dispute resolution, on average, over time?). Are the different modes of production (pastoral vs agricultural) really set on different courses re: dispute resolution and law or the same long term course, with differences in how fast and easily they can get there?)

Chad Rohlfsen said...


The Pitted Ware hunters look like Motala with EN admixture. I don't have the merged set with the Anatolians yet.

result: Karelia_HG SwedenSkoglund_NHG Loschbour LBK_EN1 -0.0376 -5.384 12757 13753 311106
result: Karelia_HG Motala_HG Loschbour LBK_EN1 -0.0584 -10.289 13147 14779 334584
result: Motala_HG SwedenSkoglund_NHG Loschbour LBK_EN1 0.0196 3.473 13391 12876 315186

Chad Rohlfsen said...

MN admixture, I should say.

Krefter said...

Can you guys shorten your post length. It's hard to gather all the information when there are many long posts. thx.

Rob said...

Matt/ Ryu

Are you guys discussing the WHG increasing in the Bronze Age steppe alone or Europe in general over time ?

Davidski said...


You haven't provided any evidence that the West Eurasian admixture in Gujaratis is not overwhelmingly CHG. Your opinion is not evidence.

Btw, I can't remember posting anything about the Paniya at my blog. I post a lot of data in the comments here. Most of it is experimental. I don't have any strong opinions about the genetic structure of the Paniya.

But Paniya don't cluster with Papuans. Even the plot you posted doesn't show that they cluster with Papuans.

And I'm struggling to see how this relates to Nirjhar being right? He seems to think that R1a-Z93 was one of the main paternal markers of the Harappans. That doesn't sound right at all.

Rob said...

@ Srkz

"So i agreed, he is probably a descendant of Andronovo/Srubnaya-like people who has mixed with SE Siberians and migrated backwards. "

Can you please explain how your IBD analysis reveals this ?
IMO all we can say is that by the Iron Age, 'east Eurasian' admixture was present in central-west Eurasia (hardly surprising given that the "steppe highway" was in full swing).

More complex scenarios of back and forth migrations- esp for this specific individual and his body of ancestors require a heavier burden of evidence, which are lacking at this juncture

Rob said...

Very interesting
But the institutions which you mention did not exist in the copper age. The "raiding nomad" form of pastoralism existing as a mirror sedentary states only developed much later- the Iron Age- whether one is talking about the European steppe or that north of China

Rob said...

Very interesting
But the institutions which you mention did not exist in the Copper age. The "raiding nomad" form of pastoralism existing as a mirror sedentary states only developed much later- the Iron Age- whether one is talking about the European steppe or that north of China (the literature is abundantly clear on this outside "pop' publications).
In fact, 'states' did no exist either in the Copper Age, although it was almost there in the Balkans, but rather than continuing toward a trajectory of urbanism and state formation, it 'imploded'.

One also has to be acutely aware of exact chronology, without which, cause and effect are confused, and we are led to false conclusions.

Rob said...


Thanks for your reply, I shall answer in turn to your interesting points.

1) "Judging from the huge crash in neolithic ancestry in the transition to CW in central europe, and the complete replacement of the Y-lines, I would say something like the 'raiding nomad' style probably existed at the time"

I don't think one follows form the other- as you appear to be suggesting that a population can crash *only if* it is attacked by some outside force. Perhaps you're not familiar with the latest in Bayesian radiocarbon modelling, but the Neolithic crash happened before the earliest evidence (genetic and archaeological) of the steppe arrivals - and in some cases almost 1,000 years before (4200 BC vs 3300 BC).

We thus have to be acutely aware of exact chronology, without which, cause and effect are confused, and we are led to false conclusions

2) You suggest: "developments off the steppe allowed for steppe 'states', and developments off the steppe allowed for the 'raiding nomad' lifestyle to be sustainable."

-> This is not born out by the evidence. The "raiding nomad" form of pastoralism existing as a mirror sedentary states only developed much later- the Iron Age- whether one is talking about the European steppe or that north of China (the literature is abundantly clear on this outside "pop' publications).

It really is a matter of being familiar with the archaeological evidence, and not relying on unproven assumptions. I'll quote:

" Attempts by some
scientists (Anthony 1986; Zaibert, et al. 1990, etc.) to interpret
bone articles from Eneolithic sites in Eastern Europe and northern
Kazakhstan (i.e., Derievka, Tjubek, Botai, etc.) as
cheekpieces, and accordingly, to extrapolate the development
of horseback riding from this information, is not absolutely
convincing for a number of archaeological reasons (Bokovenko
1997). In addition, their theories have been developed in the absence
of reliable osteological data (Kosintsev 1999; Levine 1999).
**It was only at the end of the Bronze Age that a sporadic development
in the steppe cultures occurred in which horseback
riding was mastered–and this was probably by shepherds** "
(The Origins of Horse riding and the Development of Ancient Central Asian Nomadic Riding Harnesses N. A. Bokovenko )

In fact, 'states' did not exist, either, in the Copper Age, although it was almost there in the Balkans, but rather than continuing toward a trajectory of urbanism and state formation, it 'imploded'. States only began to take off in the early 4th millennium BC, in the proto-Uruk period, but by then, the Copper Age World of SEE was already extinct.

3) ""The migration period post-Rome is a powerful parallel, as is the period just after the fall of the Han Dynasty; both collapses of the sedentary state resulted in a merger of peoples as the non-sedentaries could no longer support their level of resource extraction from outside, and had to move in."

One must be careful with analogies, especially when inaccurately applied.

4) "a prediction of his theory is that political elites competing with the same metaethnics, i.e. sedentary elites with other sedentaries or nomads with other nomads, tend to try to merge the followers of the vanquished elites with their own, since numbers is such a huge thing in historical states, but competition at the metaethnic frontier tends to be far more vicious and tends to gain a genocidal flavour, as the opposite side's people are 'unusable'. What we know about the IE frontier may be a useful datapoint in support."

Very interesting, I shall look into metaethnics. However, it stands to reason that we should not apply theoretical models, no matter how clever, which contradict actual *hard data* (elaborated above).

Rob said...

@ Ryu

Oh, pardon if I misunderstood, then we're actually saying the same thing - for the most part. Yes, there was obviously a marked sociological upheaval.

But i still question the case you suggest. We must, again, look at the evidence to support our hypotheses - now amended to 'foot raiders'. I'll put it to you that this scenario is *also* difficult to substantiate. I have looked at papers compiled analysing the weapons, fortification structures, evidence of mass graves etc, in late Neolithic Europe - both the steppe and the sown. What appears to be the case is a rise in endemic warfare in late Neolithic (late Copper Age) agricultural world, not constant, but in cycles (c. 4500 BC and again 4200 BC). Undoubtedly, this relates to increasing tensions and competitions amongst local chiefs vying for prestige and control of trade. There is an abundance of evidence supporting a scenario of militirization in late Neolithic society. By contrast, beyond the Dniester, and on the steppe, such evidence is lacking, or at best is a pale shadow. And as you mention, the population density of these proto-pastoralists were much lower. How then are they to have raided and pillaged, with lower man power, lackin even basic copper technology (mostly still using flint(!)), and not even having the advantage of a horse ?

The Germani, on the other, hand, had weapons similar to the Romans (as ie abundantly clear from their burials back in the 'homeland' upon return from service)

* Finally, "we have the Y-chromosomes to explain"

I think we know what happened, though? Population crash in central Europe, the periphery fills the vacuum. Sure, the newcomers might have privileged themselves over any stragglers. By 3000 BC, the demographic inequality might not have been so stark, and in any case, we still have large areas of east central Europe not yet aDNA sampled to complete our evidentiary base.

Rob said...


Back to yours and Matts convo
I wonder if the rise in steppic WHG, but also EEF, (if that's in fact what you were discussing) goes hand in hand with the replaceable of R1b-Z2103 by R1a-Z93 over the Bronze Age.?

Annie Mouse said...

Strangely enough the Irish claim to be of Irish stock. (Fenius Farsaid if the link does not work)

Davidski said...

@Annie Mouse

Strangely enough the Irish claim to be of Irish stock.

You don't say? Who'd have thunk it?

Karl_K said...

This has nothing to do with the Scythian, but about the talk of IBD vs IBS.

My vote for the best ancient genomes to sequence would be the Corded Ware family from Eulau, found in 2005. It is two male children and their father and their mother. The father and children were R1a and the mother and children were K1b.

With those results, you could have four highly phased haploid genomes with 3x independent redundancy to eliminate any errors from degradation, from 2600 BC.

Karl_K said...

And what's up with these Neolithic pits full of arms?

Rob said...

Good spot, Karl

That article is available here

They're from 4300 - 4100 BC, near Bergheim France.
Many other pits of dead have been found in Late Neolithic society ( I think there was a thread on it here not too long ago).

Seems those allegedly peaceful, matriarchal farmers weren't that at all.

Balaji said...

Shaikorth, Kurd

We get more accurate numbers for statistics when the number of markers is higher. But the expected value for the statistic is the same no matter how many markers there are. It would be desirable to have as many markers as possible. But we can still compare D statistics when the numbers of markers are different.

Ryukendo Kendow

You proposed that most Indians could be modeled as CHG+ASI+ASI. Let us look at the Davidski Smarter Bear Plot.

Now let us see how we can get to Gujarati D from Satsurblia (Gujarati D is probably close to the modal Indian). Gujarati D is to the right of Satsurblia , so let us add some ANE. But ANE shares drift with WHG and so in moving to the right, we will also move up, perhaps close to where Kotias is. To try to bring it back down, let us try adding some ASI. But ASI will also move left since ASI will reduce the similarity to ANE.

Populations made by adding ANE and ASI to Satsurblia will always lie close to the line joining Satusurblia and Kotias. But South Asians are not like that. This must mean that the CHG-like component in South Asians, which is the the biggest part of ANI is different from Satsurblia and Kotias.

We have at least three kinds of “basal” - LBK_EN, CHG and ANI. I actually think that there is yet another kind a CHG-like component in the Near East and Southern Europe. That is why Tuscan appears to have more CHG than Orcadian using ADMIXTURE analysis. But D statistics show Orcadian having more Kotias-like ancestry.

"Entities must not be multiplied beyond necessity". Here we have the necessity to do so.

Shaikorth said...

Balaji, the comparison of two D stats where hundreds of thousands of markers are missing is not reliable. To give an example, just recently it was demonstrated here that the Assyrians with 100k markers will look significantly less sub-saharan shifted compared to Neolithic Anatolians than Armenians with 400k markers, but if Armenians are tested with the same 100k markers as Assyrians the shift will look equal, so we can't draw the conclusion that Armenians are more SSA from that.

Kurd said...


'We get more accurate numbers for statistics when the number of markers is higher. '

" But the expected value for the statistic is the same no matter how many markers there are. "
Not true, I run Dstats regularly.

" But we can still compare D statistics when the numbers of markers are different"
Not really, if the difference in number of markers is significant. The comparisons would not be accurate.

Matt said...

@ RK: Just a few more comments on the OT, before I bow out on it:
For example, he more or less proved the historical observation that large empires tend to appear on the edge of steppe/semi-sedentary regions and sedentary ones

I have some trouble parsing what it means to prove that something tends to happen. I guess it means that has proved that the probability will be higher there?

Seems it would be important to deal with the Americas in this instance, as our N=2 sample of an isolated region other than Eurasia (where pastoralism / steppe conditions did not really obtain prior to civilization development).

About whether pastoralists have impersonal dispute resolution, no not really, they don't even really have a government, because its virtually impossible to centralise violence capacity on the steppe, every man can fight so effectively. So the entire society self-regulates through honour, which stipulates disproportionate retaliation for every transgression, undertaken personally by the victim, who needs to keep up his reputation to prevent everything from being stolen.

That surprises me. I would've thought that the whole process would've been mediated by extended kin bands, led by powerful people within the extended family calling the shots on how to resolve disputes, and using their personally loyal men to enforce their will. Sort of like the Mafia. I guess pastoralists tended to be much more "individualistic" or "Wild West" than I expected. Reputation and personal standing I would expect to be important in all pre-literate societies with no material evidence.

Can't find much on the net about law in those societies either - they talk about the Mongolian customary law here and here which covers the 13th century to modern day but I don't really have a great understanding of the dynamics of their society from that, nor how the development and enforcement of law differed there from, say the trajectory in Europe. I also thought the Mongols had a system of nobility and commoners who held power over the place? as per


I'll take it for granted he knows a lot more about murder and personal retribution rates in Ancient Rome, Medieval Europe, the steppe at various times, and the degree to which neutral third parties using a cultural code of conduct were sought to arbitrate disputes on the steppe (vs more agricultural societies) and how these characteristics changed over time, and what trajectories these societies were on (really whether there really were "ultra sedentary" societies, or sedentary societies were simply able to evolve up to advancements that people in pastoral and mobile societies were seeking and pushing towards, but had no path to). Definitely this is not a good place for much discussion about this.


Anyway, back to the Pitted Ware Culture D tests, that's pretty cool. Seems like that shows that the PCA that show some projection towards North Europe relative to Motala might not be so crazy, and that PWC might be a plausible mediator for some WHG ancestry into modern Europe (through successor cultures) although not very much.

I think it's also interesting in light of them not having the same derived SNPs as Motala. So why would they look like Motala + admixture in that context?

Alberto said...

I think that the question of the autosomal structure of South Asians can only be resolved with ancient DNA. I agree with RK that the most simple explanation for now for the base of ANI is CHG + ANE (or let's say it's the best approximation with the current samples).

In any case, and since RK also found that pattern comparing GujaratiA and GujaratiD, maybe some more D-stats can give further clues about it. If someone has the time to run them:

GujaratiD GujaratiA Kotias Chimp
GujaratiD GujaratiA Kotias Mbuti
GujaratiD GujaratiA Kotias Afanasievo
GujaratiD GujaratiA Kotias Afghan_Pashtun
GujaratiD GujaratiA Kotias Anatolia_Neolithic
GujaratiD GujaratiA Kotias Andronovo
GujaratiD GujaratiA Kotias Armenia_BA
GujaratiD GujaratiA Kotias BedouinB
GujaratiD GujaratiA Kotias Greek
GujaratiD GujaratiA Kotias Kalash
GujaratiD GujaratiA Kotias Karelia_HG
GujaratiD GujaratiA Kotias Karitiana
GujaratiD GujaratiA Kotias Ket
GujaratiD GujaratiA Kotias Lezgin
GujaratiD GujaratiA Kotias Loschbour
GujaratiD GujaratiA Kotias MA1
GujaratiD GujaratiA Kotias Okunevo
GujaratiD GujaratiA Kotias Satsurblia
GujaratiD GujaratiA Kotias Sintashta
GujaratiD GujaratiA Kotias Tajik_Ishkashim
GujaratiD GujaratiA Kotias Yamnaya_Samara
GujaratiD GujaratiA Kotias Yoruba
GujaratiD GujaratiA Kotias Chuckchi
GujaratiD GujaratiA Kotias Han
GujaratiD GujaratiA Kotias Papuan
GujaratiD GujaratiA Kotias Georgian
GujaratiD GujaratiA Kotias Srubnaya

Tobus said...

GujaratiD GujaratiA Kotias Chimp -0.0072 -2.619 7178 7283 95406
GujaratiD GujaratiA Kotias Mbuti -0.0088 -3.722 7058 7183 95406
GujaratiD GujaratiA Kotias Afanasievo 0.0042 1.454 6482 6428 91834
GujaratiD GujaratiA Kotias Andronovo 0.006 1.742 3955 3908 56891
GujaratiD GujaratiA Kotias Armenia_BA 0.0035 1.05 4016 3988 57737
GujaratiD GujaratiA Kotias BedouinB 0.0053 2.464 6768 6696 95406
GujaratiD GujaratiA Kotias Greek 0.0069 3.254 6729 6636 95406
GujaratiD GujaratiA Kotias Kalash -0.0004 -0.192 6729 6735 95406
GujaratiD GujaratiA Kotias Karelia_HG 0.0073 2.165 6639 6542 92465
GujaratiD GujaratiA Kotias Karitiana -0.0092 -3.401 6884 7012 95406
GujaratiD GujaratiA Kotias Lezgin 0.005 2.346 6673 6606 95406
GujaratiD GujaratiA Kotias Loschbour 0.0047 1.424 6744 6681 94546
GujaratiD GujaratiA Kotias MA1 0.0075 1.999 4931 4858 67879
GujaratiD GujaratiA Kotias Okunevo 0.0103 3.223 5621 5507 79652
GujaratiD GujaratiA Kotias Satsurblia 0.0016 0.429 4231 4218 75609
GujaratiD GujaratiA Kotias Sintashta 0.0086 1.829 1998 1964 28779
GujaratiD GujaratiA Kotias Tajik_Ishkashim 0.0006 0.247 6753 6744 95405
GujaratiD GujaratiA Kotias Yamnaya_Samara 0.008 3.492 6679 6572 95112
GujaratiD GujaratiA Kotias Yoruba -0.0083 -3.684 7041 7158 95406
GujaratiD GujaratiA Kotias Chukchi -0.0116 -4.594 6879 7041 95406
GujaratiD GujaratiA Kotias Han -0.0176 -7.3 6883 7130 95406
GujaratiD GujaratiA Kotias Papuan -0.0182 -6.712 6889 7145 95406
GujaratiD GujaratiA Kotias Georgian 0.0046 2.229 6627 6566 95406

I don't have the samples to run these 4:
GujaratiD GujaratiA Kotias Afghan_Pashtun
GujaratiD GujaratiA Kotias Anatolia_Neolithic
GujaratiD GujaratiA Kotias Ket
GujaratiD GujaratiA Kotias Srubnaya

Davidski said...

GujaratiD GujaratiA Kotias Chimp -0.0114 -4.441 510071
GujaratiD GujaratiA Kotias Mbuti -0.0113 -4.991 510071
GujaratiD GujaratiA Kotias Afanasievo 0.0044 1.768 501629
GujaratiD GujaratiA Kotias Afghan_Pashtun -0.0021 -0.834 113630
GujaratiD GujaratiA Kotias Anatolia_Neolithic 0.0049 2.24 508715
GujaratiD GujaratiA Kotias Andronovo 0.0081 3.277 507943
GujaratiD GujaratiA Kotias Armenia_BA 0.0008 0.305 342761
GujaratiD GujaratiA Kotias BedouinB 0.0016 0.738 510071
GujaratiD GujaratiA Kotias Greek 0.0037 1.746 510071
GujaratiD GujaratiA Kotias Kalash -0.0032 -1.515 510071
GujaratiD GujaratiA Kotias Karelia_HG 0.0034 1.085 478946
GujaratiD GujaratiA Kotias Karitiana -0.0129 -4.475 510071
GujaratiD GujaratiA Kotias Ket -0.0082 -3.141 113720
GujaratiD GujaratiA Kotias Lezgin 0.002 0.952 510071
GujaratiD GujaratiA Kotias Loschbour -0.0022 -0.61 434671
GujaratiD GujaratiA Kotias MA1 0.0019 0.57 367965
GujaratiD GujaratiA Kotias Okunevo -0.0069 -2.145 258030
GujaratiD GujaratiA Kotias Satsurblia 0.0002 0.065 364176
GujaratiD GujaratiA Kotias Sintashta 0.0085 3.27 466538
GujaratiD GujaratiA Kotias Tajik_Ishkashim -0.0023 -1.007 510070
GujaratiD GujaratiA Kotias Yamnaya_Samara 0.0063 2.824 505969
GujaratiD GujaratiA Kotias Yoruba -0.0113 -5.011 510071
GujaratiD GujaratiA Kotias Chukchi -0.0137 -5.383 510071
GujaratiD GujaratiA Kotias Han -0.0213 -8.581 510071
GujaratiD GujaratiA Kotias Papuan -0.0243 -8.932 510071
GujaratiD GujaratiA Kotias Georgian 0.0028 1.346 510071
GujaratiD GujaratiA Kotias Srubnaya 0.0078 3.39 506404

Davidski said...

Tobus, why are your marker counts so low?

What's your e-mail address?

Alberto said...

@Davidski, Tobus

Thank you to both!

Very interesting. David's stats are pretty clear in the signal they are showing, here ordered from higher to lower:

Tobus' ones don't seem so clear, but they do have very low number of markers. Maybe the biggest discrepancy is with Okunevo that here shows the stronges +ve signal while in David's is quite negative:

But taking David's ones as more reliable the Sintashta, Andronovo, Srubnaya signal looks very clear, while in general the pattern looks consistent.

FrankN said...

@ryu "About whether pastoralists have impersonal dispute resolution, no not really, they don't even really have a government.."
I think you are overlooking an important point: Salt. Indispensible for pastoralists, not only for the animals, but also for conserving meat and dairy as winter provision. So, who controls salt supply, controls pastoralists, and holds an effective and immediate instrument to sanction violation of rules. Moreover, pastoralists tend to regularly convene, typically on their return from summer pastures, for joint slaughtering/pickling, cattle trade, and partner search/marriage. That should also be the occassion for dispute resolution and joint decision making by the elders of the various groups.

A short journey back in time to demonstrate how salt unites cultures or even creates empires:
- Rome owes its initial rise to power to the control of the salt marshes along the mouth of the Tiber.
- Hallstatt: The name speaks for itself - Hall "salt" (literally "the lightly-coloured") + statt "place".
- Halle/Saale: Oldest central-european evidence for industrial saltmaking (mid 4th mill. BC). I think it is not accidentally that major Central European cultures, irrespectively of where they originated, ultimately centered geographically in the Elbe-Saale region: Funnelbeakers, CW, Unetice are prime examples.
- Salzmünde The name says all (it is actually more or less a suburb of Halle, just 10 km north of it).
- Bernburg, Baalberge: Salt mining in Bernburg (also on the Saale, 40 km down from Halle) continues to date. Baalberge is a suburb, 5km south of Bernburg's centre.
- Schöningen: Another ancient salt town.
- Megalithic Brittany: Most likely, the region's salt ponds where already used before they are attested from Roman times. Megalithic Brittany accumulated enormous wealth, evidenced by the high concentration of precious Piemontese Jadeite. Barely neolithicised, and probably more focused on pastoralism than agriculture, it provides the earliest European examples of socially highly diffentiated burials, and engineered large-scale constructions.
- Jericho, possibly the world's most ancient city, with an estimated population of 3,000 by the 8th mill. BC, controlled the access to salt from the Dead Sea.

It isn't about rural population concentration, it is about control of ressources. Salt plays a prime role, because controlling pastoralists means control of a much larger area than controlling cereal farmers. Plus - its much easier to collect pastoralists' surplus, because they come to you with their wealth (cattle), instead of you having to go from farmstead to farmstead.

Tobus said...


I'm just using Sergio's Kotias which is probably why the count is so low - but I'm wondering how yours is so high, the Haak data only has ~300k, no? Email is tobus73 at gmail.

Balaji said...

Ryukendo Kendow,

You are of course right that ASI will decrease similarity to ANE and WHG simultaneously. Therefore if we start with Kotias and mix in some ASI, the movement will be towards Satsurblia. Similarly mixing in ANE will increase similarity to ANE and WHG simultaneously. So we could start at Satsurblia and by mixing in ANE move towards Kotias. We move up and down the Satsurblia-Kotias axis depending on how much ANE and ASI we mix in. This line is some distance from the South Asian trend line.


I have another idea on comparing the “basal” nature of different populations even though we have different numbers of markers available. We had the following D stats.

Chimp Ust_Ishim Brahmin_UP Loschbour 0.0067 1.101 112450
Chimp Ust_Ishim Orcadian Loschbour 0.0163 3.035 501362

We had only 112450 markers for Brahmin_UP but 5011362 markers for Orcadian. Why not calculate D(Chimp,Ust-Ishim;Brahmin_UP,Orcadian)? This will directly be a comparison of the “basal” nature of the two.

Could you calculate the D statistics for the list in the following file?

FrankN said...

@ryu: "..can retard the centralisation of violence capacity and postpone societal pacification indefinitely, c.f. Afghanistan, Somalia, South Sudan which are having difficulty transitioning to stateshttp.
You might want to add Spain (those Basque pastoralists), and 20th century Ireland to the list as well. Or Switzerland, which has emerged from a comparable, multi-lingual, mountain pastoralism culture. Oh, wait ..

I could add hundreds of comments here - my profession is international development, with first-hand experience in Afghanistan and elsewhere, but that would lead us clearly OT.
Let's keep it to this: Cultural patterns, as the "honour and revenge" value system that you have correctly described, and the equally important concept of "hospitality" accompanying and balancing it, plus high local autonomy, anchored in the traditional system of 'elders', need to be considered. Any attempt to build "statehood" against these principles is bound to fail - an experience the British, Russians and now the Americans have been making in (parts of) Afghanistan when trying to implant an unsuitable model that is ultimately grounded in Roman centralism instead of Swiss cantonism. The problem is not with the "pastoralism-derived" culture, it is with outsiders who come and disregard this culture (plus 'central' elites so corrupt that they lose local respect).

Davidski said...


Btw, you wrote Usi-Ishim instead of Ust_Ishim.

FrankN said...

@ryu: What's your evidence for endogamy? I am aware of the Hazara, as a prime example of Mongolian-Central Asian intermarriage. I also remember having read somewhere (here?) about Turkic-speaking people being well admixed. Or check out Hungarian DNA for that matter. The Altaian "sprachbund" phenomenon, as evidence of inter-lingual long distance contact and, most likely, intermarriage, is well documented. The "Steppe" invaders into Europe that we have historical records about (Huns, Bulgars, Avars, Magyars, Mongols) were all multilingual/-ethnic tribal confederations.

I may be missing something here - in that case I am eager to learn from you. But otherwise i suggest you to "get your facts right before flaunting your expertise."

P.S: This isn't meant to get personal. I highly appreciate most of your posts. But this is a case where you (and actually first of all Turcin, whom I am just in the process of reading) have been starting from quite questionable premises, drawing wrong conclusions and unsuitable comparisons that contradict well-founded findings of research on(pre-)history, liguistics, ethnology, and also International Development. If you are interested (and Dave thinks this isn't getting too far OT), I may post a criticism on Turcin's main assumptions here once I am through with his paper.

Taymas said...

Alberto et al,

I'm still trying to wrap my head around all the different formal stats, so forgive me if this is way off, but those D-stats are telling us that GujA has a lot of both Kotias and (Steppe minus Kotias) over GujD, right?

Alberto said...


Yes, basically here we're using Kotias as a kind of baseline, so it's not really important. Both Gujarati A and D have similar levels of Kotias admixture, but GujaratiA shows higher affinity because of being more ANI/West Eurasian, while GujaratiB has more ASI/ENA.

What's more important here is the populations in the 4th position. There we are seeing which one has higher affinity with GujaratiA relative to its affinity to GujaratiD. And there it shows that Sintashta, Andronovo and Srubnaya have the highest relative affinity to GujaratiA. So yes, basically something like GujaratD + X = GujaratiA, and the best match for X is Sintashta in this case.

The pattern was first found by Ryukendo Kendow, and I've checked with the Admixture results and it's also confirmed by them. The "problem" is that this signal is quite clear in GujaratiA, but not so in other populations. Sindhi has a different pattern, as does Brahui or Kalash. So there is no tight correlation between amount of ANI and amount of "Sintashta-like" ancestry. This probably means that there are different layers that make it difficult to know without ancient DNA what is what. But, as long as we can rule out any recent European admixture in the GujaratiA samples, Sintashta seems to be one of the layers present in the subcontinent (though more tests would be needed to really know, since if GujaratiA turns out to be a clear exception then it might mean something else). So, as usual, a very interesting result, but some caution advised in a complex case like this one.

Coldmountains said...


Just because IEs were mixed they don't have to be exogamous. From a genetic point of view tribal strictly endogamous living Pashtuns are much more mixed. You also forget that Pashtuns actually took a lot of non-pashtun women and that among tribal patriarchal societies being endogamous means in most cases just that women are not allowed to take foreign men or their children are not more part of her tribe but men are almost free to take any women they want. Actually Mongol society was much more hierarchical organized than Pashtun society . Pashtun society was/is very decentralized organized and unlike among Mongols there never existed a ruler among which had total control over his people. Pashtun tribesmen never fully accepted central rule.

FrankN said...

@ryu "..or the early Germanics and Vikings, who were very endogamous and tribalistic.."
Well, this guy's family and marriage history, in spite of his Norman descent, looks anything but tribalistic:

Burgundians, Suebians, Lombards seem to have admixed quite easily and quickly with locals in their new homelands. Goths anyway - they were already one of these typical "steppe" tribal federations when they crossed the Danube into the Roman empire. Vandals and Alans (Germanic and "Skythic") wandered and conquered together. The Cimbers and Teutones had a leader named Boiorix, which is clearly Celtic and means "King of the Boii" (those who settled in Bohemia and founded Boinonia aka Bologna). Ceasar describes the Belgians as Celto-Germanic mix. I believe the case of the mixed Danish-Black Forest Bronze Age couple has been presented here by Dave in a separate post not too long ago.

IIRC, last year or so some Swedish Viking age DNA was analysed. Turned out that one woman had a very different profile, pointing to her originating somewhere on the middle/ lower Volga. Her husband was obviously a Varangian merchant who had taken her with him from one of his journeys. Whether she joined voluntarily or was forced - who knows? Anyway, those "endogamous Vikings" had quite a reputation for robbing females on foreign coasts. OTOH, blonde women were quite appreciated and paid high price for in the Bagdhad Califate.

A so knowledgeable and versed person as you are might occasionally be well advised to conduct a little background research before coming up with statements like the one cited above.

FrankN said...

Since it may be pertinent to the ongoing discussion on Near Eastern DNA, let me add that the Baghdad Califate, heavily hit by the Black Death, suffered acute shortage of labour. Varangians, anyway good in human trafficking, were happy to help out with a mix of adventurous Scandinavians (the type that also served with the Varangian Guard in Byzantium) and possibly rather unlucky than adventurous locals that were "collected" along the Dniepr or Volga en route towards the Black/ Caspian sea.
Estimates are up to one quarter of the Califate's total workforce having been "supplied" by the Varangians. Note that the Califate also controlled a number of Central Asian silver mines.

postneo said...

I was saying macro trends in culture and technology mirror population density and "networkedness"

funny you brought up z93 and christianity. Lets now talk about languages... most compiled/ interpreted languages or OS that dominate, spread first from densely populated north eastern US (bell labs etc ..) and now completely dominated by the west coast silicon valley etc. Think about linux though, should we characterize it as a Finnish invasion or remember that its a unix clone from the US with a Finnish veneer.

human languages could have strange trajectories as well. The current sucessful IE descendants may not be representative of its early history.

Bergmans rule is usually applied more towards inter species(polar vs sun bear) and some times sub species . But given our short history, we have physiologically not left Africa yet for such differentiation to take affect. we just wear clothes or burn fuel to maintain an african plastic bubble so to speak. The hominids that were genuinely mutating towards temperate climes e.g Neanderthals did not survive. Eskimos have taken a tiny step perhaps.

The only non african mutations of note are light skin and perhaps lactase persistance. light skin Alleles from asian boundary regions seem to have been artificially amplified and selected in Europe recently because of the need to wear clothes in cold weather, which cuts off sunlight. decrease of sunlight with latitude is really not that dramatic for such rapid change its the clothing that drove it.

Perhaps lactase persistance was amplified in Europe because thermophilic bacteria that break down lactose are sluggish in cold temperatures. But I am not so sure... this ones more murky

@Ryu, FrankN
If you look at dominant cultural trends (sitcom scripts, movies, ) or if you took a slice of cultural artefacts in the US analogous to pot sherds ....say cheap plastic thingies, They do not reflect the rising hispanic demographic in the US. My point is that there were always counter intuitive trends afoot similar to today and meager archeological samples are too coarse to detect these.

Coldmountains said...


I don't know where you got this but Pashtuns mix a lot with their neighbors and ever did so. Pashtun DNA results confirm that and some Pashtuns mixed with Dardics, Turks, Tajiks, Indics and even other ethnic groups. Tribalism among Pashtuns is one reason why they resist central rule but also the egalitarian society. Unlike Baluchs and Mongols which had rulers which with almost full control over their people and which had hierarchical societies Pashtuns tend to disrespect and dislike any kind of ruler which is" higher" than them.

Balaji said...

Thank you Davidski and sorry for mistyping Ust_Ishim. Previously, we had the following.

Chimp Ust_Ishim Brahmin_UP Loschbour 0.0067 1.101 112450
Chimp Ust_Ishim Orcadian Loschbour 0.0163 3.035 501362

These had different numbers of markers but suggested that Ust_Ishim was closer to Brahmin_UP than to Orcadian. Now you have found the following.

Chimp Ust_Ishim Brahmin_UP Orcadian -0.0065 -2.456 9318 9439 133128

This confirms that Ust_Ishim is indeed closer to Brahmin_UP than to Orcadian.

Grey said...


"The only non african mutations of note are light skin and perhaps lactase persistance"

high altitude

"But given our short history, we have physiologically not left Africa yet for such differentiation to take affect."


Matt said...

@ Ryu -

I'm thinking more about applying the same modelling assumptions that he does to Afro-Eurasia as a whole, to the Americas as a whole, without any adjusts to fit, and naturally having the earliest and most developed civilization emerging in Mesoamerica, with the Olmecs first in southern Mexico. Also Norte Chico in Peru.

River environments seem to click for early civilization. One element he discusses is "As a result, Chinese capitals were always located on the Northwestern frontier with the steppe, not in the Yangzi River valley, which is much more centrally located and has much more productive agriculture. In the Mississippian culture, similarly, the greatest plant productivity is in the Southeast (see the map of Gross Primary Production above), yet the first and the largest urban center, Cahokia, is located in the Northwest, on the steppe frontier." But were the more productive areas along that river much more broken and divided territory in the past, and did that itself inhibit civilizational forces.

I'm not really too much in doubt that forces which hold empires together over large land areas tended to favour large areas of open, similar environment (not broken territory, like e.g. very mountainous south of china?). I am not too sure that the origins of civilization really lie much in competition between sedentary societies in response to more mobile (steppe) societies. That seems unlikely.

Anyway, I'm sure this stuff will end up getting a lot of scrutiny over time.

Rami said...


I mentioned Nirjar because he actually calls you out on your colonial era level ignorance regarding S/SC Asians.

In any case till the genomes of S/SC Asian Hunter Gatherers is not found, you cannot make conclusions. The archaeogenetics of S/SC Asia is more complicated than Europe's.

Rami said...

@Coldmountains you got it spot on , you must be Pashtun to know all that.

FrankN said...

@ryu: "Frank, Google "Catholic Church Cousin Marriage Ban".

Have done that. It brought up a lot of things I wasn't aware of before, so thanks for the hint. The conclusion from the research, however, is, that early medieval Germanics weren't endogamous, and may in fact have been instrumental to promoting "the individualistic, nuclear-family W and C Europeans" we see today.

First of all, note that the "Catholic Church Cousin Marriage Ban" was levied by the Council of Agde 505. Except for the Franks, all Germanic tribes by that time were still of Arianic faith. Ostrogoths and Vandals never became Catholic (though the former tolerated Catholicism), Lombards only in 662, Visigoths/Burgundians ultimately in 610 (with some Catholic-Arianic swaying before), Suebans in 572. Thus, initial impact of that ban should have been rather limited (note also that there was hardly any missioning in the eastern part of the Franconian Empire before the mid-7th century).

Nevertheless, Cousin Marriage Bans were decreed by Arianic rulers, most notably the Theoderic the Great (+ 526), and the Lombardian king Rothari (+ 652). Germanic tribal law (e.g. Law of the Anglians and Warnians [in Thuringia], codified 802) contains elaborate regulation on bride price that demonstrates exogenous marriage as the rule. Moreover, there is explicit regulation that the bride price payment for widows falls to the female relatives of the deceased husband (unlike in Roman law where it stayed with the bride's father). For endogamous marriage, where the female relatives of the deceased husband would include his aunt = mother-in-law, and the widow herself, such regulation makes absolutely no sense. There is not any clause in Germanic tribal law addressing bride payment or inheritance in case of endogamous marriage, signifying that it was so rare a practice that it didn't need codifying.

Three cases of cousin marriage are described in Merowingian texts. The first two, both by Gregory of Tours, deal with a Gallo-Roman couple, and an allegoric, possibly invented case from Philippi. This should signify the populations to which Gregory wanted to make his point. The third and most discussed one is Childeric II, ruler of Austria (eastern part of the Merovingian empire, along the Rhine, not in today's Austria, to which that name was transferred much later). He had married his cousin as a political attempt to re-unite the diverging western and eastern parts/lineages of the Merovingian empire. The attempt failed - Childeric's brother Theoderic III, ruler of Neustria (western half) waged war on him. After an initial victory, Childeric and his wife were murdered, and Theoderic ultimately also assumed power in Austria and re-united the empire. Afterwards, Childeric's "incestuous marriage" became a major issue in Theoderic's attempt to legitimize his takeover (and possible fraternicide). The fact that this line of argument was pursued, even though it directly went against Theoderic's own mother and aunt, who had arranged Childeric's marriage, demonstrates how extraordinary and outrageous that endogamous marriage must have been at that time.

Further analysis and documentation is found in the following German language book, from which also the above arguments have been taken.

The "Germanic Endogamism" appears to be an "urban myth" created by some late 19th/ early 20th century scholars, which doesn't uphold against diligent analysis of relevant sources from the period in question.

FrankN said...

@ryu Continuing my previous post

Near Eastern endogamism is a valid point put forward by you that I wasn't aware of so far. Figures range from 20-50%, depending on the local (for an overview on current research see p.273f. in the following link):

However, a recent analysis on Dutch immigrants, while generally confirming substantial (~20%) endogamy in Dutch Moroccan and Turkish communities, cautions:

"The migrant population in The Netherlands mainly comes from isolated rural areas and three quarters of the young marriage partners come from the country of origin, mostly from the same village and/or family. This population has a low education, their socio-economic condition is poor (..) contrary to the population living in the more developed regions of Turkey and Morocco."

So: (a) There is substantial endogamy within various North African, Near Eastern (and South Asian) population, but (b) it yet appears to be unclear whether the feature is restricted to certain, typically rural regions, or applies in general to the countries in question, and (c) further analysis is needed to establish how far back in time this phenomenon extends.
IOW: Endogamy is a relevant issue deserving further follow-up. But with our (or at least mine) current knowledge, it seems still too early for generalisation, and for extrapolating current patterns back in time as explanation of prehistoric admixture patterns.

Alberto said...


Sorry for this off topic here, related to an old debate of ENA in MA1/ANE. But for some future research I'd like to ask you about this possible problem and solution.

D-stats in the form D(EHG, MA1)(X, Outgroup) are +ve for *all* Eurasian populations and ~0 for Native Americans. This probably means that because MA1 is a very old genome, or because of the quality of it or whatever, it shows a decreased affinity to all populations compared to more modern samples. And I think that this could be the reason why the stats D(Loschbour, MA1)(ENA, Outgroup) are ~0.

A way to work around this problem might be to calibrate the stats using another Eurasian sample as an "outgroup". For example, using Kotias (but we could use LBK_EN or whatever):

D(Loschbour, MA1)(Kotias, Chimp/Mbuti)

If that stat let's say has a D value of 0.0100, then if Dai has equal affinity to Loschbour and MA1, when you do:

D(Loschbour, MA1)(Kotias, Dai)

You would expect the stat to also have a value of ~0.0100. So running different stats with different populations in the position of Dai would show which ones are < 0.0100 and which ones are > 0.0100 and will reveal the real affinity of MA1 to different populations. Equally, it might reveal that: D(Karelia_HG, MA1)(Kotias, Karitiana) > D(Karelia_HG, MA1)(Kotias, Chimp/Mbuti), contradicting the find in Haak et al. of Karelia_HG and MA1 being equally related to Native Americans.

I think that this method for calibrating the D-stats could work. What do you think? Do you see any problem with it?

Anyway, this was off topic and only to have it in mind for whenever we revisit the subject (when you have time/will to do it and someone to help out, nothing urgent or important at the moment).

Shaikorth said...

"D-stats in the form D(EHG, MA1)(X, Outgroup) are +ve for *all* Eurasian populations and ~0 for Native Americans. This probably means that because MA1 is a very old genome, or because of the quality of it or whatever, it shows a decreased affinity to all populations compared to more modern samples. And I think that this could be the reason why the stats D(Loschbour, MA1)(ENA, Outgroup) are ~0."

Another reason (which I do think is less likely due to MA-1's age) could be the SNP selection in the Human Origins set affecting results. In Flegontov et al. the f4 statistic (MA-1 Chimp; Karitiana Ket) is non-significant (|Z|<2) using their genome-based merged dataset both with and without transitions. Is this the case using a Human Origins merge?

Shaikorth said...

Correction: the statistic was (MA-1 Yoruba; Karitiana Ket)

Doubt chimp would change much, though if it does Ket should be more likely to share extra ancestry with Yoruba.

rozenfag said...

@ ryukendo kendow
>Mongolia, the only exogamous country of C Asia

Are you kidding? Kazakh and Kyrgyz people are just as exogamous as Mongols.

Onur said...

Are you kidding? Kazakh and Kyrgyz people are just as exogamous as Mongols.

also Turkmens

Grey said...

Marriage practise among populations like the Pashtun may have changed with religion so some admixture might be from before the change.

Seinundzeit said...


For whatever it's worth, Pashtun tribes tend to be very exogamous, when it comes to marrying women. Basically, Pashtun ethnic identity is wholly contingent on the father being a Pashtun, while the maternal side just isn't factored into anything. This is why Pashtuns on the Punjabi-Pashtun ethnic frontier look just like Punjabi people, and why Herati Pashtuns are physically indistinguishable from the Farsiwan of that region, even though an Afghan Pashtun from Khost and a Pakistani Pashtun from Tirah will tend to look very different (in other words, will be very easily distinguishable) from the average Afghan Farsiwan or Pakistani Punjabi. This pattern exists throughout Afghanistan and Pakistan, wherever Pashtuns happen to live alongside other ethnic groups. In addition, ethnic integration ("Pashtunization") is very easy (and common, as I can think of at least 12 concrete examples in both Pakistan and Afghanistan, but I'll mention only a few), as evidenced by the Dardic Pashayi people of Afghanistan, who now consider themselves (and are ofteconsidered by others) to partake of "Pashtun" ethnic identity, despite having no spot on the mythic Pashtun genealogical tree, and despite having been completely monolingual (only speaking their Dardic language) just a few generations ago. Also, I have met many Peshawari Hindkowans who can speak Pashto, and as a result will construe themselves as Pashtun, despite a total lack of tribal ties, despite the fact that they look rather South Asian, and despite the fact that they still can recall their "caste" origins (the South Asian concept of "Jati" is rather alien to Pashtun culture). We also have the example of Laghman province in Afghanistan, where the lines between Pashtun, Tajik, Nuristani, and Dard are rather fluid. Regardless, you are right about cousin marriage, it's very common.

But Pashtun women marrying non-Pashtun men is a rather different proposition. This is considered absolutely unacceptable, in traditional contexts (for a historical example of this, one only has to examine the fate of the "Hindustani Fanatics" who found refuge in Swat during the British occupation of India). Obviously, we are looking at a dynamic that is shot through with socio-political implications. The population which "takes" women, but never "gives" women, is construed as "dominant", while the population that does the opposite is construed as "subjugated" (this brings us to the issue of "Hamsayah" groups, but that is another discussion).

Also, Pashtun tribes (the "nang" ones) are very egalitarian. The closest thing to hierarchy exists in the Pashtun south (Loya Kandahar in Afghanistan + Pakistani Balochistan). But the eastern Pashtuns (in eastern Afghanistan + FATA in Pakistan) are infamous for being very unruly, for being very nervous in regard to any expression of authority, and for living in a state of political anarchy. The "Jirga" is the only real mode of political operation, and tribal maliks have no power to coerce people. At the end of the day, every male "head of a household" is pretty much his own master. Unfortunately, women have no formal rights or power (although, their position in society isn't exactly like that which is depicted in many western sources. In addition, many western anthropologists have no clue about how things work inside households, where women have considerable authority over children, and considerable power to pressurize their husbands to do certain things, like carry out blood feuds), and children have no say in anything. Basically, it's an extreme egalitarianism that is only operative for adult males.

But I imagine this is very OT.

ryukendo kendow said...
This comment has been removed by the author.
Davidski said...

Kotias and MA1 share around 300K SNPs.

I won't be online much after tomorrow, but I got Chad and Tobus to stick Kotias into their Human Origins datasets, so they can help out from now on.

Seinundzeit said...

Hi RK,

Very interesting points, and a fascinating discussion. Although, I should mention that Fukuyama's Hegelian vision of society is, in my view, wholly untenable. Honestly, I like to situate myself among a rather different group of theorists, and my favorite sociology tends to draw on a conceptual tradition which is rather distinct from the one you're utilizing. Nevertheless, it's always refreshing to see concrete and specific socio-political dynamics analyzed from a perspective that is different from my own understanding. A broader conversation on foundational theory is warranted, and I would love to read about how you look at these things.

Regardless, when it comes to Pashtuns, "tarburwali" is a notion that is highly relevant to our discussion. "Tarburwali" is an essential principle in tribal Pashtun society. In fact, "tarburwali" is one of the most important dynamics which underlie and structure Pashtun social life, alongside "melmestia" (hospitality, the guest-host relationship) and "badal" (revenge, literally "exchange", from the slightest of injuries to the most severe of injuries). Basically, "tarburwali" is the notion that "the son of my father's brother is my sworn enemy". In fact, the word "tarbur" means both "enemy" and "cousin" in Pashto/Pukhto. Pashtuns are much more likely to keep feuds with (and are much more likely to kill) their cousins than they are to keep feuds with non-kin (or kill non-kin). I feel that this rather important detail constitutes somewhat of a hurdle, when it comes to conceptualizing power-relations in exogamous versus endogamous societies (in the conceptual scheme which you prefer to use for these questions).

In addition, unlike among Somali tribesmen, anything resembling "collective punishment" is non-existent among Pashtuns. Each individual is responsible for his own actions (in this context, things like murder or kidnapping), whole tribes or families are never implicated.

Also, the fact remains that actual autocratic political organizations are non-existent among eastern Pashtuns. It's almost impossible for any individual (or individuals) to amass enough wealth and social clout for something like that, in places like Khost, Paktia, Mohmand, or Khyber.

rozenfag said...

@ ryukendo kendow
>That is not true, central Asian turks like Kazakhs and Uzbeks

Kazakhs and Uzbeks are very different in that regard. I know that cousin marriages are common among Uzbeks, however I haven't heard about any among Kazakhs, with an exception of a South Kazakhstan, that experienced heavy Uzbek influence. The same goes for Kyrgyz people.

Alberto said...


Yes, I agree. Whatever results we can get just apply to MA1, the single samples of ANE that we have (well, we also have AG2, but that was lower quality and maybe contaminated). So we cannot extrapolate to the theoretical ANE component if MA1 does have some extra ENA admixture/affinity.

Maybe for a quick test, if we consider a safe assumption that MA1 should have higher affinity to Karitiana than Karelia_HG (because of the time and place where it was found. Haak et al. were surprised about the non-significant result and didn't have any good explanation for it), we could try to test that and check how this works by running two single D-stats:

Karelia_HG MA1 Kotias Chimp
Karelia_HG MA1 Kotias Karitiana

(Kotias looks like a good "baseline" for being quite neutral and good quality, but we could use another one if it makes it easier. Same for Chimp if someone prefers Gorilla or Mbuti).

If those stats show difference in the result, then it could mean it works and we can go further with the method.

@Tobus, Chad

If any of you get the chance to run the above 2 stats I would appreciate it. Thanks.

Onur said...

That is not true at all, off memory the Turkmen are the most endogamous of Afghanistan's ethnic groups by some margin, even more so than Pashtuns.

No, cousin marriage is a taboo among Turkmens up to several generations as in Kazakhs, Kyrgyz, Karakalpaks and Mongols and unlike Uzbeks, Uyghurs and Tajiks.

Onur said...


Thanks for the consanguinity stats. The high consanguinity rates in Turkmens must be because, like Kazakhs, Kyrgyz and Karakalpaks, they do not consider descendants of female relatives related.

Chad Rohlfsen said...

I'll check those Dstats after I get home, around 5pm CST.

Matt said...

the most powerful Olmec cities could not control areas just ~30km away, for example, despite the density and homogeneity in the environment, and most villages retained the 'egalitarian huddle' structure not much different from the first Neolithic sites; there is also the complete absence of a standing military force. Such complete stasis in state size and complexity throughout the 1000 years when the culture flourished would have been inconceivable if it occurred in, say, early egypt or China for example.

Similar to Harappa / Indus Valley? Or quite different?

While the opposite expansion from non-steppe-facing polities into the arena, say, from Bengal or from a valley in Sichuan, almost never occurs.

Funnily enough, I was reading today that the Maurya Empire basically expanded over all India from Bihar in Northeast India, which is humid subtropical and borders Nepal and Bangladesh. It's not Bengal though.

The Gupta Empire expanded from the same region.

You could agree with "almost never" though as India tended to fall into smallers kingdom though. I don't know that there was a difference between north and south in cohesive empires (Chola)? Nor northwest and southeast (maximising distance from northwest frontier).

Under Turchin's theory I guess India wouldn't see many large cohesive long lasting empires as it was not very exposed to a "metaethnic frontier"?

Chad Rohlfsen said...

Here are a few stats, in addition to what was asked for.

result: Anatolia_Neolithic Kotias Loschbour Dai 0.0320 6.295 20464 19196 431752
result: Anatolia_Neolithic Kotias Karelia_HG1 Dai 0.0049 0.998 21759 21548 475758
result: Anatolia_Neolithic Kotias Loschbour Primate_Gorilla 0.0220 4.361 20520 19636 405517
result: Anatolia_Neolithic Kotias Karelia_HG1 Primate_Gorilla -0.0017 -0.345 20917 20989 426451
result: Anatolia_Neolithic Kotias MA1 Primate_Gorilla -0.0098 -1.805 15954 16269 331953
result: Anatolia_Neolithic Kotias Dai Primate_Gorilla -0.0067 -1.778 21770 22066 453539
result: Anatolia_Neolithic Kotias Ust_Ishim Primate_Gorilla 0.0013 0.293 21516 21458 452644
result: Anatolia_Neolithic Kotias Paniya Primate_Gorilla -0.0059 -1.319 7013 7096 100587
result: Anatolia_Neolithic Kotias BedouinB Primate_Gorilla 0.0301 9.153 22852 21517 453539
result: Anatolia_Neolithic Kotias Basque_Spanish Primate_Gorilla 0.0351 10.381 23373 21787 453539
result: Anatolia_Neolithic Kotias English_Cornwall Primate_Gorilla 0.0217 6.155 23047 22069 453539
result: Anatolia_Neolithic Kotias Kalash Primate_Gorilla -0.0177 -5.093 21934 22725 453539
result: Anatolia_Neolithic Kotias Georgian Primate_Gorilla -0.0051 -1.499 22415 22643 453539
result: Kotias Atayal Paniya Primate_Gorilla -0.0446 -8.618 7171 7840 101132
result: Kotias Papuan Paniya Primate_Gorilla -0.0165 -2.935 7468 7719 101132
result: Primate_Gorilla Kotias Loschbour MA1 -0.0217 -2.949 14423 15063 301791
result: Primate_Gorilla Kotias Loschbour Dai -0.0574 -10.101 19638 22031 408180
result: Primate_Gorilla Kotias Loschbour Japanese -0.0551 -9.842 19634 21925 408180
result: Primate_Gorilla Kotias Loschbour Nganasan -0.0514 -9.021 19532 21647 408180
result: Primate_Gorilla Kotias Karelia_HG1 MA1 -0.0294 -3.898 14022 14872 313573
result: Primate_Gorilla Kotias Karelia_HG1 Dai -0.0599 -10.654 20194 22768 427575
result: Primate_Gorilla Kotias Karelia_HG1 Japanese -0.0580 -10.478 20180 22664 427575
result: Primate_Gorilla Kotias Karelia_HG1 Nganasan -0.0535 -9.409 19820 22058 427575
result: Karelia_HG1 MA1 Kotias Primate_Gorilla 0.0294 3.898 14872 14022 313573
result: Karelia_HG1 MA1 Kotias Karitiana 0.0248 3.406 15874 15106 345147

Annie Mouse said...

Hmm that did not come out right. :)

Strangely enough the Irish claim to be of Scythian stock. (Fenius Farsaid if the link does not work)

FrankN said...

A. You are bringing up Augustine of Canterburys letter to Rome in 601 AD - this seems indeed to be the most widely misunderstood part of the evidence. The letter addresses three kinds of "incest", namely marrying (a) a cousin, (b) the step-mother (forbidden in Roman law) and (c) the deceased brother's wife. (b) and (c) are well evidenced and were often practiced by the Germanic nobility to avoid succession/inheritance problems, but for (a) there is no evidence aside from the Childeric II case described above. Most notably, Gregory of Tours has strongly criticized Germanic leaders for their "immorality" - e.g. 6 wives of Chlothwic I, 4 wives of Charibert I, in both cases some of them simultaneously (which by itself already speaks against endogamy). However, Gregory doesn't mention any case of cousin marriage among Germanic leaders (though among Gallo-Romans).
Among the three points raised by Augustine, cousin marriage had already been banned by the Council of Agde. He didn’t require further clarification. Hence, including cousin marriage in Augustine’s letter should be understood as strengthening the other two points. The issue of stepmother marriage had shortly before been addressed by Childebert II of Austria’s 596 “Law against Incest” that placed stepmother marriage under death penalty, unless explicitly authorized by the King. Note that that law didn’t mention cousin marriage, presumably because there was already a church ban on it. There have been other bans of stepmother marriage, e.g. by Clothar II of Neustria, and Pippin the Younger, Charlemagne’s father. The latter two were clearly motivated by restricting the power base of potential competitors. Augustine may have had acted under a similar intention on behalf of his sponsor, Aethelbert of Kent. [Ironically, this fired back. Aethelbert’s son, Eadbald, iIn order to consolidate his power base after his father’s dead, in 616 married his stepmother. To be able to do so, he reverted to paganism and expelled the Catholic Church from Kent. In 618, Eadbald allowed the Church’s re-entry, and disposed of his stepmother in order to marry the daughter of a Frankish king (possibly Clothar II of Neustria, see above)].

“a shift from a society where endogamy was widespread”: as apparently in the Maghreb, with a lesser extent in Tunisia that was the centre of the Vandals’ empire. The Vandal’s role in this respect is unclear. Procopius’ description of the integrative character of the Vandal society is confirmed by tombstones of mixed Vandal-Roman couples, or personal names like e.g. Svartifan (Germanic swart =”black” + Berber suffix “-fan”). Salvanius of Massilia praises the Vandal’s enforcement of morality on the provincial Romans, e.g. by forbidding prostitution, which wouldn’t go along well with cousin marriage, an incestuous practice according to Roman law. OTOH, V.Vitens reports that Vandal nobility marrying a low-ranked person, and their offspring, would fall under servitude to King Huneric. The meaning of “low-ranked”, i.e. whether only meaning slaves, or also commoners/freemen, Roman citizens etc., remains unclear. Marriages between Vandal and Alan/ Berber/ old Roman nobility should not have been affected, Huneric himself hat a Roman wife of imperial blood.
Anyway, note that for Sicily, estimates are up to 50% cousin marriage until the late 1950s.

Balaji said...

Chad Rohlfsen,

Thanks for posting those D stats. I found the following particularly interesting.

Anatolia_Neolithic Kotias Kalash Primate_Gorilla -0.0177 -5.093 21934 22725 453539

Previously, Davidski had calculated the following.

Chimp Kalash Kotias LBK_EN -0.0059 -1.503 303595

It appears that by using Anatolia_Neolithic in place of LBK_EN, we get Kalash to show a strong preference for Kotias. Could you calculate the following D stats?

Anatolia_Neolithic Kotias Austroasiatic Kotias Primate_Gorilla
Anatolia_Neolithic Kotias Austroasiatic Primate_Gorilla
Anatolia_Neolithic Kotias Pulliyar Primate_Gorilla
Anatolia_Neolithic Kotias Paniya Primate_Gorilla
Anatolia_Neolithic Kotias Chamar Kotias Primate_Gorilla
Anatolia_Neolithic Kotias Kanjar Primate_Gorilla
Anatolia_Neolithic Kotias Kol Primate_Gorilla
Anatolia_Neolithic Kotias Indian_Singapore Primate_Gorilla
Anatolia_Neolithic Kotias South_Indian Primate_Gorilla
Anatolia_Neolithic Kotias Kusunda Primate_Gorilla
Anatolia_Neolithic Kotias Bengali_Bangladesh Primate_Gorilla
Anatolia_Neolithic Kotias Burusho Primate_Gorilla
Anatolia_Neolithic Kotias Brahmin_UP Primate_Gorilla
Anatolia_Neolithic Kotias Brahmin_TN Primate_Gorilla
Anatolia_Neolithic Kotias Brahui Primate_Gorilla
Anatolia_Neolithic Kotias GujaratiA Primate_Gorilla
Anatolia_Neolithic Kotias Kalash Primate_Gorilla
Anatolia_Neolithic Kotias Makrani Primate_Gorilla
Anatolia_Neolithic Kotias Sindhi Primate_Gorilla
Anatolia_Neolithic Kotias Chuvash Primate_Gorilla
Anatolia_Neolithic Kotias Yamnaya Primate_Gorilla
Anatolia_Neolithic Kotias Andronovo Primate_Gorilla
Anatolia_Neolithic Kotias Orcadian Primate_Gorilla
Anatolia_Neolithic Kotias Tuscan Primate_Gorilla
Anatolia_Neolithic Kotias Lezgin Primate_Gorilla
Anatolia_Neolithic Kotias Georgian Primate_Gorilla
Anatolia_Neolithic Kotias Armenian Primate_Gorilla
Anatolia_Neolithic Kotias Syrian Primate_Gorilla
Anatolia_Neolithic Kotias BedouinB Primate_Gorilla
Anatolia_Neolithic Kotias Iranian Primate_Gorilla

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
FrankN said...

@ryu: German medievalists generally regard Goody as refuted, see linked review of the Ulb paper I posted above condoling him for having “to grapple with overcome theories (Goody, Poly).”

Can't say anything on Lancaster and Whitlock, aside from them not being cited in current German-language research.

A general issue of confusion is the different Roman, high medieval canonic, and Germanic systems to qualify consanguinity. The Germanic is the one we use today. The canonic system used the maximum distance, the Roman system the total distance to the MRCA. So, a (Germanic) cousin 1st degree would be 2nd degree in Canonic, and 4th degree relation in traditional Roman count. This is still confusing many today, and might also have affected Rome-trained early priests coming to Britain. Cf

Roman law forbade marriage to 4th degree, i.e. 1st cousins. However, when Constantine the Great married his son to his half-brothers daughter for dynastical reasons, the law fell in disrespect. It was renewed by Justinian, to be revoked by his son, Aradius, in 405 BC.
The first tightening took place when the Epaon Council of 517 (Burgundian Arian Church) decreed a marriage ban up to 2nd degree cousins (6th Roman grade relatives). The 527 Council of Toledo (Visigoth Arian Church) followed, banning all marriage with blood relatives "as far as the lines of descent are known" (adfiniates liniamenta generis successione cognoscit). Isidor of Seville, maintaining this general principle, had the rule specified in the 624 Seville council to a marriage ban up to 6th degree, as in the Burgundian Church. Further sharpening towards the 7th degree is found in the Visigoth law books of Kings Reccesvinth (+672) and Erwig (+687). The Merowingian Salian law (mid 6th century), in the meantime, had included cousin marriage (not differentiated by mbd vs. fbd, see link!) as offence, leading to confiscation of all property if the marriage wasn't resolved.

Rome only took notice of this regulation when the Bishop of Toledo, Sinredus, after fleeing from the 711 Arab conquest, became a Papal advisor in Rome. Passages of the correspondence between Pope Gregor II and Saint Boniface on the matter, as well as drafts for the 721 Synode in Rome, were directly copied from Visigoth texts. The 721 Synode ultimately settled on a 4th degree (i.e. 1st degree cousins) marriage ban, but several Germanic tribal laws, including the Bavarian codified shortly after St. Boniface's mission, included more explicit regulation.
The ban on marrying 7th grade cousins was decreed by Emperor Henry II in 1003. Positively formulated, Henry II wanted to promote long-distance marriage of the middle nobility in order to strengthen the Empire’s cohesion. Of course, it was also meant to prevent regional accumulation of power that could threaten his position, and specifically intended to weaken his rival, Conrad of Carinthia.
The Catholic Church followed 70 years later, but used the Canonic instead of the Germanic count, thus "only" banning marriage of 6th degree cousins.

The timeline makes obvious that it was not the Catholic church driving the discussion, but the Germanic Arianist community. Gallo-Roman influence may have played a role. However, the initial Burgundian and Visigoth council decisions only affected people of Arian faith, i.e. rather Germanics than the mostly Catholic Gallo-Romans.

The whole "Germanic mbd marriage" thing, btw, appears to be a myth. Among the dozen or so Germanic tribal laws on incest, there is only one not regarding mbd marriage as offence. Actually, one of the four existing copies of the Alemannic Law doesn't list it, the other three do!

As to the Mediterranean, this study from Malte might interest you:

Alberto said...


Thank you very much for all those stats. Some pretty interesting ones now combining Kotias and Anatolia_Neolithic that give a better understanding than the first ones using LBK_EN. For example regarding Basal Eurasian in Kotias vs. Anatolia_Neolithic:

Anatolia_Neolithic Kotias Ust_Ishim Primate_Gorilla 0.0013 0.293 21516 21458 452644
Anatolia_Neolithic Kotias Dai Primate_Gorilla -0.0067 -1.778 21770 22066 453539

It looks like it's pretty much the same, and interestingly (this was seen before, but not so clearly with LBK_EN) the Dai in Crown Eurasian present in Kotias looks stronger than that in Anatolia_Neolithic (related to it being stronger in EHG vs. WHG?).

Also confirms that Anatolia_Neolithic has probably direct input from WHG:

Anatolia_Neolithic Kotias Loschbour Primate_Gorilla 0.022 4.361 20520 19636 405517
Anatolia_Neolithic Kotias Karelia_HG1 Primate_Gorilla -0.0017 -0.345 20917 20989 426451

While Kotias confirms slight preference for EHG over WHG, but non significant:

Primate_Gorilla Kotias Loschbour Dai -0.0574 -10.101 19638 22031 408180
Primate_Gorilla Kotias Karelia_HG1 Dai -0.0599 -10.654 20194 22768 427575

Also as pointed out above by Balaji, while Georgian share the highest drift with Kotias, they also seem to have a good amount of Anatolia_Neolithic that's probably completely missing in the Kalash:

Anatolia_Neolithic Kotias Georgian Primate_Gorilla -0.0051 -1.499 22415 22643 453539
Anatolia_Neolithic Kotias Kalash Primate_Gorilla -0.0177 -5.093 21934 22725 453539

As for the experiment trying to calibrate stats for the depressed affinity of MA1, it doesn't look great. This first one:

Karelia_HG1 MA1 Kotias Primate_Gorilla 0.0294 3.898 14872 14022 313573

Looks like indeed MA1 has an abnormal low affinity to other Eurasian samples. The stat looks to significant to be correct given all the other above. So I was hopeful that it would mean that when using Kotias as the baseline, it would show quite higher relative affinity to Karitiana. But:

Karelia_HG1 MA1 Kotias Karitiana 0.0248 3.406 15874 15106 345147

The difference does not look significant enough. Doing a direct subtraction, D=0.0046, Z=0.492. So it seems that MA1's low affinity to modern populations is real and not an artifact of being old/low_quality/snp_selection, or that the method for calibration just didn't work as expected.

Matt said...

@RK: The imperial nuclei, including the Nandas in Patna, the Guptas in UP, the Delhi sultanate etc. were always the ones directly facing the incursions from the steppe-originated polities such as Indo-greeks/Kushans.

I don't know if it makes a lot of sense to think of the Indo-Greeks as a steppe originating polity. Otherwise, OK, also the Vijayanagara Empire in South India was another one facing a large mega empire (with its origins in intrusion from outside India) in the form of the Delhi Sultanate.

(As an aside, I wonder if this pattern was exacerbated in China by decisions to keep placing the capital and administration in the North even *despite* it not necessarily being the best choice. That in itself would amplify the probability for a new dynasty to arise in the North even if it wasn't objectively a good choice to have the capital in the North. In India that wouldn't happen in the same way).

Here is a classic example of a society with land that is not fully alienated from kinship norms, and a military that is not attached to an impersonal administration. In striking contrast to the internal changes of steppe-facing states like Russia or the Qin who purged their internal elites, and bypassed intermediaries in tax collection all in a very short time in response to intense pressure from the steppe, while the countries further away from them, e.g. Lithuania or Chu, slept on.

Interesting. I wouldn't have ever really thought of Russia as a country which was alienated from kinship norms, or attached to impersonal administration, relative to other countries further West, much more sheltered from the steppe, and the societies I would think of within Europe that were the most alienated from kinship norms, and with the least personalised military, would tend to be the ones furthest from it.

Davidski said...


Also as pointed out above by Balaji, while Georgian share the highest drift with Kotias, they also seem to have a good amount of Anatolia_Neolithic that's probably completely missing in the Kalash.

Unlikely, considering the high affinity between the Kalash and Sintashta, and no chance of ASI in Sintashta.

The EEF-related signal does show up in the Kalash in Admixture, it's just usually very low or much lower than among everyone else, except Yamnaya and Afanasievo.

Alberto said...


"Unlikely, considering the high affinity between the Kalash and Sintashta, and no chance of ASI in Sintashta.

The EEF-related signal does show up in the Kalash in Admixture, it's just usually very low or much lower than among everyone else, except Yamnaya and Afanasievo."

Well, the "high affinity" between Kalash and Sintashta is relative and more easily explained by the CHG and EHG/ANE components that they share. The admixture you linked to is before the appearance of the "teal" component, and before we had CHG genomes, so it's not too accurate. Using the real Anatolia_Neolithic samples and CHG ones, it's clear that the Kalash have very little, if any, Anatolia_Neolithic admixture. The D-stat above comparing to Georgian also confirms this. IBS of Anatolia_Neolithic would also support very low sharing with Kalash. So it has to be very low.

But in the end, the only way of really knowing about the steppe admixture is ancient DNA from the area. I don't know from any BMAC samples being tested, but let's see what Maykop, Kura-Araxes and IVC can show about the people south of the steppe. Those should be coming soon enough, I hope.

Davidski said...

The formal statistics you're talking about don't prove that the Kalash don't have any of this type of ancestry. I'd interpret them to mean that the Kalash have a lower level of it than Georgians, but I wouldn't go further than that.

IBS stats don't really help to settle the issue, although they suggest that the Kalash have a fairly normal affinity to the Anatolian farmers for their part of the world.

FrankN said...

1. "Church itself was debating internally between the 'Germanic' way of reckoning kinship vs its original method". "Romanized" Germanics, i.e. Goths, Lombards, Francs etc. had no problem to a/o adapt the Roman count. The Church debate started only when power had shifted to non-Romanized Germanic territory, i.e. East Franks/HRE, by the late 9th century, long after Augustine and Beda.

2. "association between negative attitude towards cousin marriage and Christianity dates back to St Augustine"
Before we had a/o Burgundian Arians (517), Visigoth (527), Neustria (597).

3. "Germanic Tribal Laws you quoted from all existed *after* christianisation". Nope. They existed long before as oral, memorized tradition (c.f. the many rhymes in the "Sachsenspiegel"). Christianization served their codification. The Law of the Bavarians dates ca. 730-735. St. Boniface established the Bavarian Bishoprics in 739-742 using Celto-Roman, christianised enclaves as a base. There wasn't any time to "create" the Bavarian law - it is a collection of tradition, written down by early monks (Regensburg?).

"The whole point of this is to find out if these tribes were endogamous *originally*.". Exactly. And there is quite some evidence that they were, e.g. from the Nibelung Saga (Grimhild marrying Attila etc.), or with the Vandals, Anglo-Saxon kings, etc.

That isn't to say that there wasn't cousin marriage. Most likely there was, for the advantages listed by Brittle (“more stable marital relationships, greater compatibility with in-laws, lower domestic violence (..), economic benefits of reduced dowry and the maintenance of any landholdings”). In isolated regions, e.g. Orkneys, Norwegian "Fjordland", it could have been difficult to find a non-related partner. But evidence points at a quite exogamous Germanic culture (say, less than 25% cousin marriage).

Much more important is social endogamy: in Germanic society; Nobles marrying nobles, freemen freewomen (“Frau”) etc. To some extent this continues to date in Germany, where parents tend to subtly inquire about a potential spouse's family background; note also yellow press discussion on British/Danish/Swedish Royals marrying "commoners". But a Roman, Celtic, Scythian, Hunnish noble was just as good as a Germanic one, possibly even better in terms of alliance-building.

It wasn't the Catholic Church, but Germanic rulers, who drove the tightening of incest rules, as part of a strategy to centralize power in a highly dispersed political landscape. "Taxes were collected by a patchwork of local elite castes before being passed to any central state." fits the HRE as good, possibly even better than India. In fact, there were hardly any taxes passed on in the HRE. Instead middle/local nobility had to contribute troops in case of war, which they sometimes did, sometimes not. Financial contribution came from territorial archbishoprics (hence the conflict with the Vatican over the right to install Bishops), Imperial cities, and "foreigners", most notably Jews (hence the Vatican's anti-Jewish campaigns to erode the Emperor's tax base). But primarily, an Emperor's power depended on his “Hausmacht”, i.e. troops and taxes from his family territory, and revenue from (silver) mines therein.
In this context, incest regulation arose to prevent middle-nobility's accumulation of power (hence the ban on stepmother and brother's widow marriage), and to increase the likelihood of a territory having no legitimate heir, which would allow the King/ Emperor to assign it to his own property, pass it into bishopric possession, or install a relative as new Duke/ Count. The side-effect of promoting Germanic - Gallo-Roman integration was probably unintended, but such integration would ultimately have occurred anyway outside isolated, mountainous areas.

Seinundzeit said...


I'm sorry for taking so long to respond. I was finally able to read the stuff you linked and mentioned. Anyway, exceedingly interesting material, thanks!

Chad Rohlfsen said...

My list doesn't have all of the pops requested, but here is what I have...

result: Anatolia_Neolithic1 Kotias Loschbour Dai 0.0319 6.187 20463 19199 432080
result: Anatolia_Neolithic1 Kotias Karelia_HG1 Dai 0.0068 1.321 21847 21550 475882
result: Anatolia_Neolithic1 Kotias Loschbour Primate_Gorilla 0.0272 5.112 20600 19511 405798
result: Anatolia_Neolithic1 Kotias Karelia_HG1 Primate_Gorilla 0.0052 0.992 21086 20868 426566
result: Anatolia_Neolithic1 Kotias MA1 Primate_Gorilla -0.0079 -1.438 15931 16184 332196
result: Anatolia_Neolithic1 Kotias Dai Primate_Gorilla -0.0015 -0.385 21899 21965 454082
result: Anatolia_Neolithic1 Kotias Atayal Primate_Gorilla -0.0028 -0.654 21858 21980 454082
result: Anatolia_Neolithic1 Kotias Han Primate_Gorilla -0.0026 -0.683 21879 21994 454082
result: Anatolia_Neolithic1 Kotias Ust_Ishim Primate_Gorilla 0.0038 0.786 21562 21401 453183
result: Anatolia_Neolithic1 Kotias BedouinB Primate_Gorilla 0.0343 10.006 22972 21449 454082
result: Anatolia_Neolithic1 Kotias Basque_Spanish Primate_Gorilla 0.0374 10.513 23447 21757 454082
result: Anatolia_Neolithic1 Kotias English_Cornwall Primate_Gorilla 0.0255 6.968 23162 22009 454082
result: Anatolia_Neolithic1 Kotias Kalash Primate_Gorilla -0.0132 -3.571 22044 22632 454082
result: Anatolia_Neolithic1 Kotias Georgian Primate_Gorilla -0.0011 -0.312 22532 22581 454082
result: Anatolia_Neolithic1 Kotias Austroasiatic_Munda Primate_Gorilla -0.0014 -0.304 6950 6970 99622
result: Anatolia_Neolithic1 Kotias Paniya Primate_Gorilla -0.0001 -0.015 7054 7055 100862
result: Anatolia_Neolithic1 Kotias Bengali_Bangladesh Primate_Gorilla -0.0089 -2.550 21893 22285 454082
result: Anatolia_Neolithic1 Kotias Burusho Primate_Gorilla -0.0088 -2.604 22039 22429 454082
result: Anatolia_Neolithic1 Kotias Brahui Primate_Gorilla -0.0068 -1.990 22107 22409 454082
result: Anatolia_Neolithic1 Kotias GujaratiA Primate_Gorilla -0.0058 -1.636 22132 22389 454082
result: Anatolia_Neolithic1 Kotias Kalash Primate_Gorilla -0.0132 -3.571 22044 22632 454082
result: Anatolia_Neolithic1 Kotias Makrani Primate_Gorilla -0.0048 -1.467 22084 22300 454082
result: Anatolia_Neolithic1 Kotias Sindhi Primate_Gorilla -0.0101 -3.041 22021 22468 454082
result: Anatolia_Neolithic1 Kotias Chuvash Primate_Gorilla 0.0102 2.896 22592 22134 454082
result: Anatolia_Neolithic1 Kotias Yamnaya_Samara1 Primate_Gorilla -0.0138 -3.411 21903 22517 449348
result: Anatolia_Neolithic1 Kotias Andronovo Primate_Gorilla 0.0048 1.205 22514 22298 453014
result: Anatolia_Neolithic1 Kotias Orcadian Primate_Gorilla 0.0229 6.650 23052 22020 454082
result: Anatolia_Neolithic1 Kotias Italian_Tuscan Primate_Gorilla 0.0338 9.672 23307 21781 454082
result: Anatolia_Neolithic1 Kotias Lezgin Primate_Gorilla -0.0001 -0.026 22512 22516 454082
result: Anatolia_Neolithic1 Kotias Georgian Primate_Gorilla -0.0011 -0.312 22532 22581 454082
result: Anatolia_Neolithic1 Kotias Armenian Primate_Gorilla 0.0179 5.177 22913 22106 454082
result: Anatolia_Neolithic1 Kotias Syrian Primate_Gorilla 0.0229 6.682 22726 21707 454082
result: Anatolia_Neolithic1 Kotias BedouinB Primate_Gorilla 0.0343 10.006 22972 21449 454082
result: Anatolia_Neolithic1 Kotias Iranian Primate_Gorilla 0.0097 2.852 22596 22162 454082

Matt said...

You can compare the political organisation of Russia to Poland-Lithuania. The court of Russia centralised very early, and an appointed bureaucracy in the form of the 'dvoryans' and boyars appeared very early as well, never had an equivalent in India, for example, and something like that developed much later in the rest of Europe, which were still having Fronde as late as 1600s, 100 years after the extreme centralisation and gutting of the nobility by Ivan the Terrible.

You could compare yes, but also to Sweden, the Holy Roman Empire, England, Spain, etc.

In terms of Russia of the day, it seems true that you do have a centralisation of a bureaucracy under an autocrat. To some degree historians seem to float the idea this was a result of insanity or arbitary rule, although this possibly was easier on a threshold, with the state poised against the Tatars, so does not contradict the idea of societies scaling towards a centralised autocracies of some kind when faced with an external threat. The historical narrative does not really seem to be that this occured in response to a threat from the steppe really. (If anything the comments seem to be that the policy of gutting the nobility (oprichnina) weakened the state's military capacity.) Otherwise though, Russia post-Ivan does not seem like an example of a marcher state with a more effective and efficient military and government which then turns around and dominates the centre (if that is what the comparison to the Qin state was suggesting?). It seems more like a Gunpowder Empire(?), one which was less efficient and cohesive as a society relative to those back west (despite the centralisation and despite the bureaucratisation) which turned its expansion outwards towards Siberia, where the local groups did not have European firearms or the base to sustain them.

A striking number of states nucleated in the northwest of China in specific area, the Wei river valley and surrounds; the Erlitou Shang, Zhou, Han, Wei, and Tang dynasties all originated there in the time of divisions, so its generally the state in that position which unites the rest of China. This has little to do with the position of the capital.

My thinking was that if the machinery of governance for the whole country was systematically placed in the same region, again and again, that alone would provide impetus for groups in that region to seize control of the governance of the country, even if was never rational or guided by deep cultural evolution to place that machinery of governance there. If that doesn't apply though, it doesn't apply.

The Delhi sultanate cannot really be considered a steppe state, as it went through five dynasties on the same territory.

Well, I did say that the origin was initially outside India, specifically, not that they were a steppe state. So that seems like an odd comment in response.

Balaji said...

Chad Rohlfsen,

Thank you for providing these statistics. Could you comment on the following differences?

result: Anatolia_Neolithic Kotias Kalash Primate_Gorilla -0.0177 -5.093 21934 22725 453539
result: Anatolia_Neolithic Kotias Georgian Primate_Gorilla -0.0051 -1.499 22415 22643 453539


result: Anatolia_Neolithic1 Kotias Kalash Primate_Gorilla -0.0132 -3.571 22044 22632 454082
result: Anatolia_Neolithic1 Kotias Georgian Primate_Gorilla -0.0011 -0.312 22532 22581 454082

Could it be that Anatolia_Neolithic and Anatolia_Neolithic1 are different?


Thanks for writing about the implications of the above statistics. The people who spread agriculture to Europe were like Anatolia_Neolithic. The rest of the Near East also has much affinity to Anatolia_Neolithic. Anatolia_Neolithic also spread into the Caucasus diluting the CHG there. But in South Asia there is not enough of it for it to have been how agriculture arrived. Rather agriculture must have been an indigenous development.

Chad Rohlfsen said...

They are different. I re-ran them. Anatolia_Neolithic1 is the 5 samples that have over 95% coverage.

Haryana Singh said...


What you describe sound very similar to caste system, at least the one in Haryana. I guess hamsaya are low caste or not considered part of pathans despite speaking pashto. This is exactly what caste system is like. Pathans not giving away women but have no problem taking them from others. Again 100% similar to caste system in India/haryana where high caste behave in similar way.

FrankN said...

A solid analysis based on "hard facts" should start with the economic basics. ENF agriculture allowed for a settlement size of 50-100 people. MN innovation (dairying, ard/plough) pushed the limit up to 500 people. Temporarily, up to 1200 people could be supported if all agricultural land was used simultaneously, but this wouldn't have been sustainable due to soil exhaustion, causing the settlement to be given up after at max. 33 years. Cf. the excellent discussion for CT:

From my own work in Indonesia, I feel the SEA Neolithic might have supported somewhat higher numbers, depending on the productivity of early rice/millet varieties, intercropping/multiple harvest (e.g. rice-maniok alternation), the impact of fowl domestication, and natural refertilisation by river floods/ volcanic material. For the (proto-)historic period, this has certainly been the case, but I haven't seen analysis of the SEA Neolithic yet. In general, I would assume the specifics of the SEA Neolithics rather resulting in a more dense and stable rural settlement pattern, supported by investment into irrigation/ terracing, than in larger individual settlements.

Thus, establishment of (proto-)cities requires access to regional food surplus (termed "support by satellite farming villages" in the CT analysis above). My suspicion for CT is rather transhumating pastoralism than "satellite farming" - this would a/o explain the large empty central area within their "Mega-Cities" (winter cattle maintenance). Interestingly, a similar layout has been found in the LBK settlement of Nieder-Mörlen NE of Frankfurt, which was exceptionally large (500-1,500 inhabitants) for LBK standards, controlled the Bad Nauheim salt springs (industrially exploited at least since La Tene and Roman times), and has provided multiple finds related to the La Hoguette Culture (Western Alps) and to Starcevo/Vinca.

Economic Geography has shown cities typically developing at the border of different ecological zones, as a function of respective trade, e.g. wood, stones, ore from mountains; cereals from loess plains; reed, birch bark/ tar, beaver fur from marshland; etc.) Developed pastoralism, dairy-based and each autumn culling male offspring and older cows, produces a surplus of leather/hides and horn/bones (as raw material for toolmaking). This can be exchanged against cereal surplus, plant fibres (roping!), plus firing material/fired products (ceramics, metal) in case of steppe pastoralism, and/or salt (mountain pastoralism).
Inside the farming core, as densely populated as it typically is, OTOH, there is little to exchange with other farmers. Thus, urbanisation tends to occur much later, mostly as secondary centres, when steppe/mountain/coastal products are being traded further up-/downstream into the farming core.

This explains why so many "capitals", be it Beijing, Cahokia, or Gizeh/Cairo, formed at the interface of farming and steppe/savanna pastoralism instead of within the farming core. Mesopotamia is also instructive: It hardly developed a cultural union along Euphrates/ Tigris. Instead, we until today have north-south differentiation (Kurds/Shiites/Sunnites); such general structure can be traced back to Akkadians/ Sumerians and beyond. Obviously, multiple cities developed along Euphrates/ Tigris, each flourishing from trade between river-based farmers and the surrounding mountain/steppe pastoralists, but having little to offer to the next city further up-/downstream. The same applies to the Upper/Middle Nile.

FrankN said...

"A striking number of states nucleated in the northwest of China in specific area, the Wei river valley and surrounds"
Indeed. A fertile floodplain surrounded by mountains, with high potential to developing trade hubs between farmers and mountain pastoralists. Densely populated at least from 1000 BC (Western Zhou dynasty) onwards, no steppe nearby (Xi'an - Zhongwei approx. 700 km). This essentially kills most of Turchin's musing, because his "statistical evidence" is none.

Instead of being “steppe”, the key feature of the Wei Valley seems to be its position on a long-range trade route, namely the Silk Route. I would assume this route, as similar others, to have developed from a later (LN to Bronze Age) interconnection of previously unrelated centers of agricultural-pastoralist exchange. Obviously, it was pastoralists and not farmers creating the connection: Sometimes steppe/savanna pastoralists (Wei Valley to BMAC, Nile to Niger, Yemen to Mesopotamia/ Levante), in other cases mountain transhumance (Tibet, Hindukush, Caucasus/ Zagros, Carpathians, Balkans, Alps, Pyrenees, also Andes?), but aquatic foraging should neither be neglected (Baltic, Mediterranean, Bug/Dniepr “highway”, insular SEA, Caribbeans, American Pacific coast etc.).
Once connected to a long-range trade system, further expansion downstream is the logical next move. Farmers will rarely do it – population densities downstream, in the coastal plain, should be even higher than in the mountain valley. So, its pastoralists again, who ultimately create the coast-to coast connection*.

Along the chain, the distribution of margins needs to be negotiated, and there is always the risk of somebody establishing a parallel trade route that takes intermediaries out of business. Prime example here is of course the Portuguese sea route around Africa to SEA/ China, which took the Hanseatic League, Venice, Russia, Turks, Arabs, Central Asians and others out of the lucrative silk and spice trade, but there are many more (e.g. the Greek establishment of Messily, to the benefit of Gauls, but detriment of Veneti, Hallstatt/ Raetians, Germanics, setting in motion various migrations that culminated in the Helvetian’s exodus used by J. Ceasar to legitimate his conquest of Gaul).
In fact, the Silk Route may follow three corridors: (a) the central one, via the Yangtze/ Wei Valleys and Central Asia, (b) a southern course, from Yunnan along Brahmaputra, Ganges and the Upper Indus into the Iranian Plateau and/or to the Persian Gulf, and (c) a northern, “Steppe” road via Mongolia and the Altai to the lower Volga (Samara/ Astrakhan), where it splits into a “Varangian branch” to Novgorod and the Baltic Sea, and a “Greek branch” to the Pontic Sea and into the Mediterranean.
It would be interesting to analyze whether the power shifts described by Turchin relate to shifting trade volumes along these corridors. In any case – the best way to control margins, and avoid being taken out of business via trade shifting to another route is controlling the most central nodes of the network, which in East Asia appear to be Beijing and Xi’an (Wei Valley).

The best European analogies to the Wei Valley I can think of would be Bohemia and Transdanubia, at central upstream positions of the Danube – Elbe/Oder link. LBK was still predominantly farmers. But most of what emerged afterwards along that road – Lengyel, Baden/Boleraz, Unetice/Nordic BA, Danubian Cremation/ Urnfield- included substantial pastoralist elements.

*) Eastern Europe and Central North America, with their extensive river networks and low-lying watersheds (e.g. St. Lawrence/ Great Lakes to Mississippi) are exceptions more prone to aquatic foragers; South America (Amazonia/Andes) is a special case involving both pastoralists and aquatic foragers.