search this blog

Saturday, March 4, 2017

On the Caucasus as the PIE Urheimat

From the wrestling pit known as the comments section:

The Caucasus has never been a linguistic spread zone, it is much more like the Pacific Northwest in N America or like the highlands of Papua New Guinea with fragmented 'relict' ethnolinguistic groups all interspersed, similar to Yunnan and montane SW China, NE Siberia around the Okhostkh and Japan, or the foothills of the Himalayas in S Asia. Ethnolinguistic groups, especially relict ehtnolinguistic groups, accumulate there due to the weakness of intergroup competition; the losers of cultural evolution take shelter there. Its no surprise that cultural and genetic diversity gradually accretes in these regions, leading to a complex patchwork that is misleading as to the suitability of these places as source regions for ethnolinguistic and sociocultural expansions. The North European Plain, the valleys of the Great rivers in China, India and Mesopotamia, the fertile highlands of Iran and Anatolia, the Mediterranean basin, the Mississippi basin, and of course the Steppe and Siberia--these are linguistic spread zones, where the sociocultural dynamics are much more winner-takes-all and intergroup competition is much more intense. Indeed, the relict zones tend to be dominated historically by complex societies that exerted control from the spread zones; powerful states emerge in the spread zones more often, with stronger and more articulated forms. So of course the picture is much more homogeneous there through deep time, whether culturally or genetically. So its really quite normal that large areas of the steppe were periodically dominated by single Y chromosome lineages, with sharp boundaries between cultures. The cultures from which the spread originated may have genetic diversity, but the descendant, expansionist cultures may have been subject to intense sociocultural bottlenecks from intergroup competition prior to expansion that create the odd and non-random distributions we see here.

See also...

Women on the move


mooreisbetter said...

"Ethnolinguistic groups, especially relict ehtnolinguistic groups, accumulate there due to the weakness of intergroup competition; the losers of cultural evolution take shelter there."


Was it always such?


Nirjhar007 said...

Okay so , Whats the Maykop aDNA publication status?. Why its taking so long?. Anybody has any clue on this matter?.

bellbeakerblogger said...

I think a very powerful case can be made that the Proto-Indo-Europeans and a majority of its descendant cultures were strongly monogamous in comparison to the rest of humanity. In fact, I believe that is what you see in the burial record of the CWC and BBC when men and women are buried together. Even in ancient times, most of the North Europeans were strongly monogamous (Tacitus) and when polygamy did occur, it tended to be chiefs, and those were more likely political marriages.

That argument:

The spread of languages is usually driven by economics, maybe that's another way of saying what Ryu did.

Nirjhar007 said...

The spread of languages is usually driven by economics

Quite reasonable and hence ,the trade routes become vital to detect, the possible migrations.

Ryukendo K said...

Just to be clear, I am not making the case that there was great polygamy in the regions involved. In fact, I think the opposite was the case: the tribal societies we are talking about do not possess sufficient inequality to create heritable differences in wealth or power such that you can have an arab sheikh whose male relatives number in the thousands due to repeated recent polygamy in the same family (the scenario if you interpret a Y DNA star-like expansion very literally). These societies tended to undergo 'chiefly cycling' where the identity of the elite male was always changing, as charismatic 'big man' competition or success in social pursuits or war raids resulted in gains in prestige. Not until the stratified chiefdoms of the Iron Age do we see a hereditary elite class. Paradoxically, the emergence of complex societies with (some) hereditary inequality is associated in the DNA record with a recovery in the Y-DNA bottleneck (in comparison with mtDNA), not just in Europe, but everywhere in Eurasia where social complexity arose, which is why I think a very special sociocultural configuration existed in the period just prior to the emergence of complex societies in the Metal Ages all around Eurasia, which produced the bottleneck we see in Karmin et al. This will take some modelling and a very interdisciplinary research project to prove.

bellbeakerblogger said...

The part of my comment on polygamy was directed to the above link on a comment made by Amanda S.
But, yes I agree with what you've stated. I think the emerging concept of citizenship where men validate and promote each other over others is something seen in these advanced economies where competition (as you stated) and danger is the strongest.

DL P said...

sounds like someones but-hurt about the new Scythian paper showing strong links to Caucasus..... bummer

Aram said...

It would be nice if Ryukendo clarifies what is his definition of Caucasus. Is it the Great Caucasus range? If yes then he is more or less correct.
But if he includes into Caucasus East Turkey or any other parts of Armenian Plateau or any close neighbourhood of Caucasus then his claims are not backed up by genetic data.
The main player in West Asia after Neolithic was Iran_Chl. What was the main component of Iran_Chl? Then at BA another massive wave by Kura-Araxians. Kura-Araxes started in South Caucasus. I can continue examples in historic times when both Anatolia and Iranian plateaus were depopulated and constantly needed a population replenishment from mountainous zones.
And all this unrelated to PIE umerheit.

Matt said...

In b4: "Ah, so consistently with the logic, the Caucasus *couldn't* have been the source of CHG in Yamnaya. It must have been Central Asia!"

Alberto said...

Rather tangential to the main topic, but related to monogamy vs. exogamy. I went back to Haak et al. 2010 where they analysed mtDNA from LBK sites and got to the correct conclusion that those people came from Anatolia/Near East, due to the higher affinity of LBK mtDNA to modern populations from those areas. The maps from fig. 3 are the fastest way to check it:

But looking at this today it's quite surprising that they got it right, isn't it? Now that we have genome-wide data from LBK, we know that autosomally they're closer to Europeas than to anyone ele, and besides, doesn't most of the European mtDNA come from these early farmers? And hasn't there been big population turnovers in Anatolia and Near East since these ancient Anatolians left to Europe?

So probably a question for Sam/Krefter (or anyone else who might have an answer): why these results? Due to the smallish amount of WHG/EHG mtDNA in modern Europe? Due to selection in Europe acting i a much stronger way on mtDNA than in Anatolia and Near East? Due to a mtDNA turnover in Europe with the arrival of R1 people (which would imply a mass migration of females too? From where?)

Alberto said...


Not sure where the quote is from, but somehow related to it, and to my comment above, and to what we were talking about where might the Yamnaya population in Eastern Scythians come from:

I noticed that in the first Baltic paper from a few days ago (Christ, they published so many papers on the same day that it's difficult to keep track of which is which), there was this comparison of f3 shared drift between Estonian CW samples in the autosome and the X chromosome. It showed a bias of greater shared drift in the X chr. towards European MN pops. So it seems that the method works more or less (assuming this CW was some 70% Yamnaya-like carrying Y DNA R1a and some 30% Europe_MN admixture via females). But at the same time, if that 70% Yamnaya had some 40-45% CHG admixture via females, I would expect CHG to also have a high ratio of shared drift in the X chr, especially compared to EHG (if ultimately EHG brought R1a to the mix). But his wasn't the case. CHG was slightly below EHG.

So while this should be tested directly with Yamnaya (instead of CW) to be sure, it could suggest that the Yamnaya population might have a different origin than a very sex biased recent admixture in the western steppe. Maybe it's an old population with no sex bias admixture (where ever it happened) and what we see in the steppe are just offshoots of this population from small initial migrations -families- what makes them share the same Y-chromosome (basically, founder effects).

No idea, but the resurgence of Yamnaya-like admixture in the eastern steppe could be related to this.

(Just speculating, probably because of my reluctance to accept a model of EHGs taking CHG wives as being very parsimonious).

Rob said...

WRT your first question- Isn't it because Central European LBK farmers didn't leave much of a heritage ? (But instead most European EEF comes from other groups ).

Alberto said...


Could be, but then again, aren't all EEF derived from the same original population?

So for some reason LBK kept the original mtDNA from Anatolia but other groups of EEFs didn't? And ultimately these latter groups spread their mtDNA throughout Europe?

But why those other EEF groups had a different mtDNA from the original Anatolian farmers?

Alberto said...

Or put another way:

Haak at el. 2010 got the correct answer because they were lucky to choose that specific EEF group? If instead they had analysed EEF samples from Iberia or Bulgaria they would have concluded that EEFs were native to Europe (or they came from somewhere else other than Anatolia)?

Maybe. But then it's quite a lucky coincidence.

Rob said...

Yes I see what you mean
Differential founder effects ?
Or perhaps there were 2 or 3 different EEF migrations , one characterised by the Tell settlement building which only reached the Danube by 4500 BC, and thus not yet shown in any study; and perhaps another via the Caucasus

Rob said...

Actually there was this study

Matt said...

@Alberto, re: first question, indeed it is quite surprising they "got it right". If you look at the differentiation of mtdna of LBK, it's stronger with either the pooled European dataset or pooled Near East dataset than either are to each other. Transform the Fsts to a plot with PCoA (Principal Coordinates Analysis) -, it is visually apparent that the LBK is closer to modern NE than modern Europe is, but just barely and the differentiation between modern NE and Europe is very weak compared to Euro HG and LBK. (Indeed the differentiation between modern NE and Euro HG is weak compared to Euro HG and LBK!).

The PCA confirms this is not an artefact of aggregating highly differentiation European and Near Eastern datasets, as the individual countries tend to be closer to each other. LBK set has highly differentiated set of mtdna hapg where some hapg are highly represented.

I can't see an easy way around it, other than to say the LBK they sampled were probably representative of a strongly maternally related subset of the wider early farming horizons in Europe, while others likely had different mtdna, then gene flow occurred after the early neolithic. Plus the later Steppe and HG introgression. Perhaps with the collapse of a cultural pattern that gave importance to matrilineal sharing?

Second question, those f3 are the kind of stats I've been interested in seeing since the Goldberg preprint (which only had X Fst and autosome Fst to compare, which is not quite as useful a measure I don't think, as it seems somewhat confounded by issues of diversity differences between X / autosome).

I do think I follow your argument. As you say, it does seems a bit dicey to try to detect CHG admixed bias in Yamnaya via Estonian CW though, since it's indirect, and with the X Chromosome there are substantially fewer sites than with the autosome, so more possibility for error and non. sig results with ancient dna paritcularly.

I would also say I think you want to be comparing what CHG is like on the autosome vs what CHG is like on the X, not what CHG is like on the X vs EHG is like on the X (if I read you correctly?). As comparing CHG is like on the X vs EHG is like on the X you are confounded by issues of whether CHG vs EHG had higher or lower within group diversity. The issue is the proportion of X sharing vs autosome sharing for the same pop, not directly comparing X sharing between two pops. Plot the two sets of stats against each other and see where CHG is on the line compared to Europe_MN (if it's above / below in the same way and to the same degree it implies the same degree of sex bias).

I'd also really like to see those same stats computed for Bronze Age Europeans vs modern Europeans, to check if the sex biased admixture between MN and Steppe found held true to the present day (i.e. in all the ancestors of present day European people), or whether its a phenomenon of the archaeologically visible Bell Beaker and Unetice cultures, and of population structure in the Bronze Age.

Shaikorth said...


We already know from this paper (fig 7)

that 1000genomes Europeans (all have steppe admixture) have an uniform level of sex bias when compared to WHG and ENF, plausible that it would hold with MN. If the sex bias has vanished after Bronze Aage, we'd have to assume uniform levels of MN-related male-biased ancestry somehow appearing all around Europe in more recent times.

Matt said...

@ Alberto, so here are some plots like that:
With all the populations:

Restricted to pops with >100,000 autosome and >10,000 X SNP overlap:

(removes populations like Iberia_BA which are likely subject to noise from limited SNPS)

Trying to use PCA to clean up differences between f3 value and f3 Z scores:

It seems like CHG is more towards X sharing compared to autosome sharing relative to EHG. But this may be an artefact of CHG->Europe_N relatedness or something anyway. I'm not too sure the SNP coverage on the X is actually good enough to feel very confident with the outcome anyway.

Matt said...

@Shaikorth, OK, trying to understand this Figure 7 and so just taking the NE1 rows in isolation, that shows:

On X Chromosome:

D (Yoruba, NE1, Cac/Arz, CEU) is strongly negative - Cagliari and Arzana Sardinians strongly more related to NE1 on X chromosome relative to CEU.

D (Yoruba, NE1, FIN/GBR/IBS/TSI, CEU) is non-significantly different from 0 for 1000G, including FIN, GBR, IBS and TSI - neither Tuscans, Finns, Iberians are more related to NE1 on X than CEU.

On Autosome:

D (Yoruba, NE1, Cac/Arz, CEU) is mildly negative - Cagliari and Arzana Sardinians are mildly more related on autosome to NE1 relative to CEU).

D (Yoruba, NE1, FIN/GBR/IBS/TSI, CEU) is non-significantly different from 0 - Tuscans, Iberians and Finnish are no more or less related on autosome to NE1 relative to CEU).

Conclusion - Relative to CEU, Sardinians have an excess of relatedness on X to NE1, compared to their autosome, while Tuscans, Iberians and Finnish do not.

Patterns are same, but slightly reduced, comparing Bichon and Loschbour.

So sex biased admixture from a population similar to Tuscans, Iberians, Finnish and CEU (but statistically seems not particularly necessarily any more like any of them than the other, e.g. stats for GBR are not stronger than TSI or IBS) into a Sardinian population related to WHG and EEF?

(Likewise, degree of sex bias in EEF and WHG related ancestry between Finnish, Iberians, Tuscans and British all essentially identical).

Tell me if I'm going wrong here.

Alberto said...


Yes, the reliability of using small amount of SNPs in the X chromosome might be a problem, plus here we're testing indirectly through CW instead of Yamnaya, which again reduces the certainty. But it would be interesting to try directly with Yamnaya.

Not sure how's the best way of testing that. In theory something like:

D(Mbuti, Yamnaya)(CHG, EHG)

First using all the markers and then using only the X chromosome and check for significant differences?

The problem I see with plotting stats and having a line of best fit is that I'm not sure if that line is relevant in this case. It's just dependent on the samples used (with different samples, the line would change), but we're more interested in seeing if each sample has a bias in the shared drift with CW_Estonia in the X chr. vs. autosomes. So probably a 45 degree line would be more informative? Also Z scores are more dependent on the number of markers than D values, so to use Z scores we would have to make sure the number of sites are the same in all the stats.

Shaikorth said...


Yeah, the overall conclusion is that the sex-biased demography in mainland Europeans is similar.

Sardinians X-autosome difference is noticeably higher with NE1 than with Stuttgart. Some regional structure within farmers perhaps, with haplotypes Sardinians prefer NE1 over Stuttgart as a donor when modeled as moderns + ancients which allows the most potential sources. With only ancient donors they get WHG and Iranian Neolithic and prefer Stuttgart, perhaps as a result of lacking Iron Age Italian and MN sources.

Matt said...

@ Shaikorth, well, for the mainland pops, would've really been nice for them to include an outgroup test as well, e.g. Dai from South China or another African population, as that could've tested how far all mainland Europeans are from neutrality.

All we really know from the study is relative to CEU, which is not ideal.

We know that TSI (Tuscans) and IBS (Iberians) are not sex biased relative to CEU, so mainland Italians and Iberians cannot be from a sex biased accumulation of EEF ancestry relative to British (or the North American population with most ancestry from Brits and Danes).

(That is if CEU's ancestors had the sex bias in Bell Beakers, then TSI and IBS can't result from Bell Beakers picking up more EEF ancestry mainly via females in South Europe.)

We know that CEU is not sex biased relative to FIN, so CEU cannot descend from a sex biased accumulation of EEF ancestry by a population with an (EEF vs Steppe) ancestry balance like Finns.

But it would all be much easier with a neutral outgroup!

Matt said...

@ Alberto, you're right, the best fit line as such isn't relevant, in the sense that it is just formed by the average, and tells you where a population is relative to the average (which depends on sample composition!). I really just used PAST3 function which added an automatic best fit without really thinking about it. A best fit lit would be much more interesting if we conditioned it on neutral populations (e.g. Ust Ishim, Papuan, Yoruba, etc.)

What is really relevant is the vertical relative to horizontal position of CHG relative to the same in Yamnaya and EHG. If CHG has more X sharing relative to autosome sharing relative to EHG and Yamnaya, then likely more ancestry is shared via the female line, compared to EHG and Yamnaya.

Matt said...

@Alberto, re: Z vs f3 score, reason I plotted those is the Z should more stable to random differences than the f3 score, which is a big issue when you're looking at the pretty puny numbers of X markers here (20,500 at most, which we would never use for looking for a good measure of relatedness on the autosome).

It's a toss up between these being systematically less significant for lower coverage populations, vs the lower coverage populations being likely to have weird spikes and troughs that aren't real due to lower coverage, and further that this is an issue for all populations here at this depth of coverage.

This may be why f3 outgroup stats which were a more logical method than Fst were not used by the Goldberg paper - you're just throwing away too many sites of information when you limit to the X and noise becomes too predominate for confidence?

Comparisons like the one Shaikorth links where we've got one high coverage ancient (e.g. NE1 or Stuttgart) and then two really high coverage groups of moderns (e.g. 1000 G FIN vs TSI) are so much easier to place faith in. 2x 1000G populations plus Satsurblia or Kotias would be able to tell us whether there was any sex biased relatedness to CHG in present day 1000G Europeans, at least...

Shaikorth said...

"All we really know from the study is relative to CEU, which is not ideal."

It's still pretty informative. If a population has male-based ANE, ENA, SSA or basal sex bias relative to CEU that would show, because all of these would pull the autosome away from WHG and EEF in comparison to CEU.

Matt said...

Yeah, agree it is informative and adds to what we know, just not ideal (compared to what would've been added with a neutral outgroup comparison).

postneo said...

If you go by modern distribution, The reservoir for z93 is in the Pamir, turkmenistan uzbegistan, extending to the Altai. Its also the reservoir were extra yamanaya, ANE rich ancestry survives as opposed to Andronovo.

This also could be where IIr, Sythian and IAr, turkic, Burusho got differentiated and impacted neighboring regions. Moutains are high in linguistic diversity with unrelated groups living in close proximity in equilibrium.

postneo said...

Most entries on this blog tend to forget demographics:

Language spread, DNA and linguistic differentiation cannot happen in vacuum. You need people.

Nirjhar007 said...

Okay so , something interesting folks, see this :

Bones and Borscht: How Neolithic Human Remains from Ukraine Are Enabling the Reconstruction of European Population History
Sponsored by Archaeological Institute of America-Milwaukee Society, UWM Departments of Foreign Languages and Linguistics/Classics, Art History and Anthropology

AIA Society: Milwaukee

Sunday, March 5, 2017 - 3:00pm
Sabin Hall, Room G90
3413 N Downer Avenue
Milwaukee, WI 53211
United States

Recent developments in ancient DNA research have allowed archaeologists to reconstruct human migrations in ways that are reshaping the way we look at prehistory. One of the most remarkable aspects of this new research has been the recognition of a large-scale migration of nomadic pastoralists out of the Pontic-Caspian steppes at the close of the Neolithic and beginning of the early Bronze Age. Many archaeologists and paleogeneticists have gone so far as to suggest this massive movement of people was the mechanism that spread Indo-European languages and established modern European genetic signatures. However, this event remains imperfectly understood. For example, to what extent did Neolithic farmers who neighbored the steppe populations contribute genetically and culturally to this expansion? This question has been the focus of our research at Verteba Cave, Ukraine, one of the only known mortuary sites associated with the farmers of the Late Neolithic. The skeletal and genetic data we have collected from Verteba Cave are beginning to shed additional light on an extremely consequential time period in European population history.

Jordan Karsten is an Assistant Professor of Anthropology at the University of Wisconsin Oshkosh. His research focuses on human osteology, evolution, and European prehistory. In addition to teaching, he also assists different Wisconsin law enforcement agencies with cases requiring skeletal analysis. Jordan earned his PhD in anthropology from the State University of New York at Albany in 2014.

Taymas said...

Why couldn't the steppe-adjacent Caucasus be a source sometimes? Davidski showed that some pop with EEF and WHG pushed onto the steppe post-Yamnaya. Seems like Mongols were not on the steppe originally. 0-1000AD W steppe had a good amount of G and J y-dna, long after the area went very R1.

I get the cultural evolution argument. But regions of cultural-interchange can generate innovations, and where would that start for farmer/steppe? Also the steppe crucible would exist AFTER the major mobility revolution.

HOWEVER Europe suggests agricultural/HG dna exchange is SLOW and there's not much time between agriculture crossing the Caucasus and Samara EL. Also, can't imagine the steppe was as fruitful for HG as the European coasts. If the farmers were the drivers, I would've expected their adna to completely dominate. Farmers also settled good land archipelago-style. Would've taken a long time before their best option was the steppe, no?

Isn't the X-linked pattern predictable? Just assume patrilocality and lopsided population. Not even sure it's necessary: SHG was mixed WHG/EHG, N Caucasus HG could've been EHG/CHG.

So I guess that pre-agricultural N Caucasus was R1a and either EHG or already EHG/CHG. CHG/NLIran crossed the Caucasus and demographically dominated arable areas. Some hunters adopted domesticates and maybe got extra CHG/NLIran. Samara EL isn't PIE (no big expansion, secondary products, wagons). Where exactly, among stock-raising CHG-shifted steppe pops, PIE got going is tricky.

My problem with a later expansion from the south is the lack of any later genetic/cultural turnover from that direction up to a certainly-IE Iron Age western steppe. Later language-family turnover in the steppe came with genetics, while genetics change (E Scythians) even w/o culture. The other problem is the lack of Near Eastern IE evidence until Luwians/Hittites but plenty for other language families.

I don't buy economic/trade arguments. IE is way too big too early. IE also expands into non-IE territory in the historical era (Hitt/Mit/Ita/Cel/Arm/Iran/Ind) and zero indication it was a trade language. IE also loses ground in history and the phenomena were never trade-related. Lastly, linguae francae should be developing in the flood plains too then, no? Yet the first obvious (Aramean) was Iron Age and political (Assyrian deportations, Assyrian/Persian bureaucracy). Too easy to learn a neighbor's language well enough to trade.

Sorry for the monster comment.

Rob said...

@ Taymas

" The other problem is the lack of Near Eastern IE evidence until Luwians/Hittites but plenty for other language families."

The earliest scripts in Anatolia are Hittite.

Taymas said...

@Rob: you narrowed my geography. Across the whole of the near east, scripts go way further back but no IE, despite it undergoing this incredibly massive spread that must've predated Hittite by millenia. Also Hittite wasn't spoken by commoners anywhere in the area, far as we know.

Rob said...

Yes I agree about the Near east- when we are taking about Judea or Mesopotamia.
As to what was spoken by which strata in Anatolia is difficult to discern. Too many speculations abound at this stage; but the only fact we do know is that IE was widespread in Anatolia by M2, comprising of several rather differentiated "Anatolian" languages, which is
somewhat of a rough tomporal guide. Given that there is a (still poorly characterised but clear) cultural shift sometime in the late chalcolitic followed by the rise of hierarchy from 3000 BC onwards, we can begin to deduce when IE arrived; and even guess from where.

Taymas said...

Rob, I'm not exactly sure what you're saying. Fill it out for me? Far as I can tell the M3 cultural shift with the Bronze and the hierarchy looks from the east. Just a few centuries later we've gotten written evidence of NW and NE Caucasian (ish) speakers covering that whole territory (Hatti, Hurrians, etc). As you say, Anatolian is clearly differentiated by M2 but looks like it's occupying *western* Anatolia. No chariots yet. Hittites do their thing a few centuries after M2. Just after that, we've got chariot-specialist Mitanni (opposite branch with none of its relatives nearby yet, clearly coming from the east given the BMAC influence). With the wall of writing-attested non-IE in Eastern Anatolia, the Armenian plateau, and Mesopotamia, we've either got two ends of a pincer or one arm staying still while the other made a giant globe-covering circle. Pincer parsimonious. No?

Rob said...

@ Taymas

The major shift probably happened c. 4500 - 3500 BC.
I really don't get the impression that Anatolian was restricted on western Antolia

NB: The north is bare becuase its mostly mountains & forests.

The attested written, non-IE langauges in M3 are Hattic in SE Anatolia, Sumerian in Mesopotamia & Hurrian also in north Mesopotamia. There are no attested languages in Armenia at this stage. Hattic could be a remnant Neolithic language from Anatolia, but it could also be new.

So at this stage, it remains open from which direction IE entered Anatolai- as teh 2 samples we have from post-Neolithic west Anatolia indicate movement from, both, south Caucasus direction and from NW Black Sea region.

Davidski said...


Davidski showed that some pop with EEF and WHG pushed onto the steppe post-Yamnaya. Seems like Mongols were not on the steppe originally. 0-1000AD W steppe had a good amount of G and J y-dna, long after the area went very R1.

The EEF/WHG-rich groups that moved onto the Caspian Steppe post-Yamnaya were probably from the Pontic Steppe or the forest steppe (Fatyanovo-Balanovo?). They were basically 100% R1a-Z93.

So I guess that pre-agricultural N Caucasus was R1a and either EHG or already EHG/CHG.

In spite of what is often repeated in the comments here, I doubt we'll see any R1a in the North Caucasus or nearby steppe. Certainly not R1a-M417, which is the R1a that really matters. I'm expecting the earliest R1a-M417 to be sitting around the Dnieper and Don.

R1b, on the other hand, is likely in the Mesolithic North Caucasus.

Taymas said...

I don't understand, sorry, would you delineate the narrative you're thinking of?

We have the historical record of the Hittites taking over Hatti, at which point the other Anatolian languages are west of them, and the Mitanni taking over Hurrians. Prior to that we have centuries of written interactions with SE and Central Anatolia which indicate Hatti and Hurrians and no IE.

Pontus: Kaskans.

Yeah, I think there was gene flows across both the Aegean and the Anatolia/Armenian plateaus. We'll need more samples to be sure which gene flows happened at which time and whether slow intermarriage effects or rapid migrations.

But in 1600BC Anatolians were coming from the west and Indians from the east. I'm very skeptical of IEs expanding from anywhere near this region, leaving no trace in the Mesopotamian writing, then pushing back in from both directions post-1600.

Taymas said...


But the European LNBA are way too similar to steppe populations. Doesn't look like the Yamnaya-like steppe slowly became a little more Euro LNBA to me, looks like Euro LNBA backmigrated, no?

Good point, R1b in N Caucasus makes a lot more sense.

Davidski said...

Not sure what happened yet. We might find out in the next big steppe paper.

It appears that Corded Ware in the Baltic became admixed with EEF via bride exchange with Central Euro Corded Ware. Before that it was basically like Yamnaya, but with less CHG.

So the same thing may have happened on the western edge of the steppe, and coupled with that there may have been a migration from the EEF-admixed Pontic Steppe to the Caspian Steppe.

Rob said...

@ taymas
No there is no writen evidence for Hitties coming from west. I had this same conversation with Ariel. All the records state is that the Hitties took over Kanesh, the capital. No direction or origin is indicated, which instead seems to stem from your personal conjecture.

Im not sure the presence of Mittani in northern Mesopotamia dictates
mutual exclusivity in direction of Proto-Anatolian advancement ? Again, it seems largely speculative although possible.

Taymas said...

@Davidski, fair enough.

@Rob, Absolutely it's my personal conjecture. I don't know how we get Hittites from the east, unless in the most very narrow sense (a small, mobile group can come from any direction), when the other Anatolians were west of them. In 1600BC and after, we have a clear situation of Caucasian (ish) speakers to the east of Anatolian. Pre-2000BC not as clear, but by the Mesopotamian writings little reason to the think the Caucasians had moved unless we go another half a millenium back at least. No?

Rob said...

@ Taymas
Which Mesopotamian writings are you referring to and what did they write ?

Rob said...

@ Dave
Doesn't Yamnaya Samara have EEF already ?

Taymas said...

Sorry for the deleted comments, I went and checked my library records. I read Van de Mieroop 2004 and Leick 2002 simultaneously, not sure which would be the best source. Checked wiki to corroborate my memory. It's all Akkadian and trade/diplomacy with both Hatti and Hurrians centuries before M2.

Davidski said...

Doesn't Yamnaya Samara have EEF already?

Something like that in the sense that it's more westerly than CHG. But it usually only shows up in formal tests, and hardly ever in ADMIXTURE, which suggests that it's EEF that doesn't share recent drift with the EEF we have available at the moment.

Rob said...

@ Taymas
Certainly: would it not make sense as the H and H were in SE Anatolia and northern Mesopotamia itself ? The Semitic peoples' colonies arrived in Anatolia only in the second millennium. So we remain uninformed about further in Anatolia and the Caucasus, don't we ?
But it does suggest that Mittani, at least, arrived after M3, we can agree.

Rob said...

@ Dave
Thanks. So if I understood that some CWC from Baltic doesn't have EEF, or EEF -like ancestry, does it mean these Yamnaya -derived CWC People "lost" it, but then some re-acquired it from Central Europeans ?

Davidski said...

No, the early Baltic Corded Ware individuals have the same sort of EEF as Yamnaya, in the sense that they also have that southern admixture that is more westerly than CHG.

But no EEF shows for them in ADMIXTURE, which means they don't have the type of EEF that the German Corded Ware samples have.

Rob said...

Yes I see. So we're looking at a yet unsampled 'Neolithic' population; but we have central European farmers, Anatolian, Armenian, Iranian...
Doesn't leave many other possibilities ?

Davidski said...

We can only speculate till more samples come in, but some of the options I can think of are Balkan farmers not overly closely related to EEF and/or Northwest Caucasus farmers.

Taymas said...

Colonies came right around M2. Written contacts were centuries before. Where/when do you see much of a gap existing between the Kaska, Hatti, and Hurrian areas in the ~2500-1600 period? We're talking an entire branch of the greatest linguistic expansion in history, the gap can't be all that small. Sure, we don't know what was happening in W Anatolia, but that's precisely "Anatolian from the west". As far as beyond the Hurrians in the Caucasus, you'd still have to explain how an entire language group got from there to western Anatolia without seeming to disturb all the non-IEs in between.

Now, we do have an example of a steppe group cutting through the Caucasus without much in the way of a later trace (Cimmerians) but (A) word got around the near east, to say the least, and (B) that's one group, not an entire, pre-differentiated language branch.

Rob said...

I don't understand your point . If PIE arrived in Anatolia in late M4 / early M3; how would anyone in south Mesopotamia note it, when writing only began in mid M3 ?
I also query some of yr basic facts- Kaska are first attested in mid M2, in Hitttie texts. Secondly the "land of Hatti" in Sargons text simply means Anatolia- as it continued to be called for thousands of years, regardless of who rules where and spoke what . Lastly, are you really surprised that Hurrians are the first to appear in Mesopotamian texts, when the very centre was in northern Mesopotamia ?

Aram said...

For those who think that there is weak competition in S Caucasus today.
The list of most militarized countries.

Armenia is the 3rd in the world. Azerbaijan the 8th.
And all this militarism is not just for parades. It is used everyday in the frontline for the last 25-30 years.

Aram said...

The oldest Hurrian text is from Urkesh, NE Syria.

Rob said...

Thanks Aram
Always good to get the view of a local expert :)

Ryan said...

This post is really off the mark. Just as the phylogeny and spread of R1a demonstrate the origin of IE in the Pontic-Caspian Steppe, so does the spread and phylogeny of J2 demonstrate that the Caucasus were a source of their own dispersal. Also, if you think the Pacific Northwest lacked for inter-group competition you don't know much about this part of the world. It had some of the highest levels of conflict of anywhere in the world pre-contact. And in terms of not being the site of any expansions - have you not heard of the Dene?

Ryukendo K said...

@ Ryan

Ryan, J2 indicates a spread from the Caucasus foothills. The North Caucasus Range, plus the inter-caucasus valley, were never such sources.

The Dene did not spread from the Pacific Northwest.

Of course there was fighting--there is fighting everywhere and at all times--but the point is that the fighting did not catalyse an increase in average group size, as indeed took place elsewhere in the Americas. There is intense fighting in the highlands of Papua New Guinea too, but this does not change the fact that the extreme linguistic diversity and deeply branching languages there indicates low rates of group extinction; and therefore the society is experiencing low, not high, levels of intergroup competition.

Taymas said...

Rob, I'm sorry but this isn't being very productive. I'm trying to gather more information and place weights on possibilities, not prove a negative. Let's reverse this:

What is your evidence that the Hatti conquered by the Hittites are not the Hatti of 2300BC?

What is your evidence that the Kaskians were considerably distant from the 1500BC location despite the relatedness to Hatti and NW Caucasian?

What is your evidence that the Caucasian-ish families (Hatti\Kaskian\NWCaucasus and Hurrian\Urartian\NECaucasus) had an archipelagic spatial pattern instead of a bloc?

What is your evidence that there was any IE in Eastern Anatolia, the Anatolian plateau, or the Caucasus prior to 1600BC?

Rob said...

To finish:

This is a pertinent conclusion from Bryce's book on Hittites;;

"Briefly, the overall situation can be summarized thus:

1. None of the evidence at our disposal points unequivocally to a
major influx of newcomers into Anatolia, whether Indo-European or
otherwise, during the third millennium.

2. We are unable to determine with any certainty the predominant
ethnic character of the populations who inhabited the Early Bronze Age
kingdoms. It is possible, though not provable, that the dominant culture
of central Anatolia in the Early Bronze Age was that of a non-Indo-
European population whom we call the Hattians. This does not exclude
the possibility that there were already peoples of Indo-European origin in the same region during this period.

3. We do not know when Indo-European groups first appeared in Anatolia, whether a century, a millennium, or several millennia before their first attested appearance in written records. Archaeological evidence has not provided us with conclusive evidence as to the date of an Indo-European arrival."

Rob said...

4. We can however be certain of an Indo-European presence in
central Anatolia by the end of the third millennium, since Indo-European
personal names appear in the records of Assyrian merchants who
set up trading colonies in the region early in the second millennium.

Ryan said...

@Ryukendo K - the Dene did not spread from the Pacific Northwest? Huh? Where do you think they spread from - Mars? No, they spread from SE Alaska. Algic probably spread from the Columbia Plateau in Oregon/Washington. Those two language families cover most of North America between them!

"Ryan, J2 indicates a spread from the Caucasus foothills. The North Caucasus Range, plus the inter-caucasus valley, were never such sources."

My point stands. No languages spread from the middle of a glacier either. The Caucasus foothills are part of the Caucasus (and frankly we don't know enough going back before that to trace things further).

"Of course there was fighting--there is fighting everywhere and at all times--but the point is that the fighting did not catalyse an increase in average group size, as indeed took place elsewhere in the Americas. There is intense fighting in the highlands of Papua New Guinea too, but this does not change the fact that the extreme linguistic diversity and deeply branching languages there indicates low rates of group extinction; and therefore the society is experiencing low, not high, levels of intergroup competition.""

No, the Pacific Northwest wasn't typical in its levels of conflict. It was notorious among the Americas - higher than anywhere else. And the before small pox hit Puget Sound had a higher population than the State of New York.

You're getting things backwards. What makes these areas unique isn't a lack of competition, but rather on the other side of the ledger - geography that's excessively conducive to groups fissioning in two and forming new groups. More languages survive in absolute terms but not as a percentage.

Take Algic for example. Yurok and Wiyot are separated from Algonquian by a thousand miles of mountains. They didn't fly all the way the way there - the groups that were intermediate to them were just wiped out and assimilated. Their homeland was taken from them, and now we don't even know their names.

capra internetensis said...

Landscapes with many geographical barriers divided into relatively isolated and defensible territories may be less suited to form large politically united groups, but in some environments they may support a lot more people than in the surrounding regions, e.g moist highlands among arid lowlands, and be able to colonize by demographic weight.

I expect the relative weight of these factors would vary with the circumstances, and that the refuge area could indeed be the source of the spread from time to time.

Ryukendo K said...

@ Ryan

Ryan, this is a stupid debate, you're not thinking about this problem clearly. We don't just suspect that the rate of intergroup competition in these areas is low relative to elsewhere, despite high rates of killing--we know this.

But lets say you are right--if these areas are so diverse because of rapid generation of new languages, but then these areas also do experience powerful intergroup competition, so 'less languages survive here as a percentage' than elsewhere, you're also saying most ethnolinguistic groups in these areas end up dead--if thats the case, why don't all major sociocultural expansions take place from these regions to the spread zones? Since the both the number of new ethnolinguistic groups generated is high (conducive to 'fissioning'), which results in high cultural variation and innovation, and the selection pressure on them is high (% of surviving languages is low). Why don't powerful tribal groups expand out from the Oregon coast and Papua New Guinea? Why is it that the powerful and homogeneous sociolinguistic groups exist in the great lakes region and mainland SE Asia instead?

These non-spread zones don't just have high linguistic diversity--they possess languages that have diverged for 10000s of years from each other, multiple languages squeezed closely together geographically, but with no discernible genealogical relation to each other. What does this suggest about the rate of group survival in these regions relative to the spread zones? Think.

Ryukendo K said...

Furthermore, the Dene expanded from Alaska, not the Pacific Northwest, like I said. I do not think that the Algics expanded from the Columbia plain--this is at best a speculative claim in the literature, raised as a hypothesis and then repeated. The pattern in N American sociocultural history is for the coastal regions to receive flux from the plains, not the other way round. For what its worth, the Algonquins expanded from the great lakes region.

Ryukendo K said...

@ Capra

Capra, the PNG highlands and the NW Pacific Plains share the characteristic of having high population density and high warfare, but little sign of state formation or even of a powerfully expansive group that we get in Eurasia post neolithic, e.g. BB, CW, Shang Sinitics etc. I think there is, in general, an excessive focus on economic and demographic issues in studying ancient cultural expansions, and too little focus on ideological and sociological factors. Of course, subsistence strategy plays a big role, but culture, ideology and mode of social organisation are more directly responsible for the size of the group, its ability to unify disparate interests, and to demand sacrifice and coordination among group members during times of crisis, especially in war. If such mechanisms do not develop, then an area can be engaged in intense warfare over millenia post neolithic but still have village-level or even kinship-based organisation and fail to produce an expansionist group or a complex society. Therefore I think that the intense intergroup competition actually occurred outside of these areas, in the places where large-scale societies existed today. The violence of PNG highlanders or Pacific NW groups may shock us now, but perhaps thats because we have no point of reference. If we had immediate access to that period of prehistory where chiefdoms, states and linguistic expansions were taking place, I'm sure we would gain some perspective. After all, Karmin et al is giving us some suggestive clues.

Just my two cents.

capra internetensis said...


Are you suggesting that *all* language spreads are due to the formation of large scale societies, somehow?

Pacific Northwest coastal groups, despite being kinship-based, are certainly known to have expanded their territories. Typically along the coast, but the Tlingit also expanded inland at the expense of the Dene.

Ryukendo K said...

@ Capra

No, thats not what I'm suggesting. The IEs did not spread due to that. I am suggesting that cultural developments favouring expansion of groups tend to emerge under conditions of intense intergroup competition, and the development of complexity is one such set of innovations.

Taymas said...

@Rob, you finally give me something to work with and then want to end the discussion, grant yourself the final word, and move on to name-calling? I'm disappointed. I'll ignore the provocation and treat your ideas with respect because that's the kind of conversation I like to have.

You believe that Hurrian pushed north during MBA. Perfectly plausible. However, of their language relations, it certainly looks like they're the southern-most. While pre-Urartians might've been on the Armenian plateau, a Hurrian push north could've been their very origin. But, you still need room on the space-time map for the NE Caucasians then. Unless you think the relation between the families was because Hurro-Urartians kept pushing north and had an influence on NE (not NW but I know what you meant) Caucasian post-BA. OK so by making several assumptions, we've created a low-odds spacetime gap in the BA Armenian Plateau. Unfortunately, this doesn't actually tell us who was there. No particular reason to favor any single one of the neighboring (in space or time) language families than any other, under this framework. I'm still left with very low odds for IE on the EBA Armenian plateau.

More likely "Land of Hatti" stuck because there were still Hatti there, as we know. The Land outlasted the Speakers a little while, probably, but I'm quicker to buy a few Cs of inertia than a few Ms. If the Land stuck despite a pre-Hittite turnover, why think IEs had anything to do with it? Where's the IE influence on Hattic?

Yes, in fact I'd say it's highly likely the Hittites were quite close to Kanesh just prior to conquering it. But I think it's far more likely they were closely west, the direction of all the other Anatolian speakers. I think you're under the impression I'm arguing for a very recent arrival of the Hittites from the steppe. I'm not.

Your argument regarding the Germanics is quite apt. That's exactly the argument I'm making regarding the Anatolians (including the Hittites), the Hatti, the Hurrians, and the Kaska. Which gives us a BA with lots of IE in W Anatolia and lots of non IE east of Kanesh, and thus unlikely Anatolian arrived from the east in this timeframe. You will have to forgive me my typo: "Eastern Anatolia, the Anatolian plateau, or the Caucasus" was clearly meant to be "Eastern Anatolia, the Armenian plateau, or the Caucasus". I'm not usually intentionally redundant.

I'm totally cool with the Bryce quote, don't see how it contradicts anything I've stated.

If you don't wish to continue the conversation, that's totally fine and I'll leave you the last word.

Matt said...

@ Capra - "Landscapes with many geographical barriers divided into relatively isolated and defensible territories may be less suited to form large politically united groups", indeed because maybe intergroup competition is more intense, and intergroup cooperation and intermarriage which dissolves language barriers less frequent?

(There are some examples in history of mountainous regions from which empires spread - Greece and the Andes at least spring to mind. Though perhaps they are not as many as they should be?

IRC the British Empire, though not something I have much knowledge on, often preferred to recruit soldiers from hill and mountain groups were preferable, though whether due to perceived effectiveness or ethnic separation from majority populations used as enforcement...).

On blog post, at least for genetic diversity, and at least on the y (probably also in autosomal haplotype structure?), seems to accumulate in the Caucasus, but not so much more or less in contrast with the Near and Middle East, which I would think to be spread zones of marching empires par excellence? Linguistic diversity is a better case, but the Near and Middle East look comparably diverse on y-dna to the Caucasus.

(It's probably a bit unfair of me to comment on this, as Ryukendo has already clarified that y-dna lineages do not dominate even "spread zones" except under quite unusual conditions).

Ryukendo: Why don't powerful tribal groups expand out from the Oregon coast and Papua New Guinea?

This said, doesn't it look like the last major expansion - Polynesians - was driven by a group of males from Papua (star like y-dna expansion) who conquered Austronesian speakers and adopted their language? Potentially driven by a more effective way of making war?

Ric Hern said...

Ezero Culture...

Ric Hern said...

The Hamangia Culture maybe ? I read that it was closer related to Anatolian Farmers than to the neighbouring Early Neolithic Cultures.It looks like they kept to the Coastline and maybe spread this way all the way to the Caucasus ?

Ric Hern said...

Rob said...

@ Taymas
I'm happy to continue the conversation when you figure out where west Anatolia is :)

Srivatsan Kannan said...

Hi, I'm a brahmin from TamilNadu,India. Recent gene test on me shows the haplogroup M172-PF5174. Can someone throw light on the origin on this haplogroup? Is it common here in South India among all population?