Tuesday, August 1, 2017
A Bronze Age dominion from the Atlantic to the Altai
The BEAGLE analysis that I foreshadowed a couple of days ago (see here) is finally done. The output is available for download as a matrix of shared genomic tracts in centimorgans (cM) here.
I haven't yet had a chance to look at the results in detail, but I'd say that the outcomes for the three Early Bronze Age (EBA) Afanasievo and Yamnaya individuals make a lot of sense. The high affinity that these individuals show to the Irish EBA samples is not at all surprising, but striking nonetheless. The Afanasievo people, after all, lived in the Altai Mountains deep in Asia, more than 6,000kms from Ireland.
Update 02/08/2017: Interestingly, the graphs below, based on the cM values in my coancestry matrix, suggest that upper caste Indo-Aryan-speaking Brahmins from Northern India share relatively more ancestry with the Afanasievo genome than Iranic-speakers such as Pamir Tajiks, who generally share relatively more ancestry with the younger Andronovo and Sintashta samples. The relevant datasheet is available here.
See also...
Late PIE ground zero now obvious; location of PIE homeland still uncertain, but...
Ancient herders from the Pontic-Caspian steppe crashed into India: no ifs or buts
90 comments:
Read the rules before posting.
Comments by people with the nick "Unknown" are no longer allowed.
See also...
New rules for comments
Banned commentators list
fascinating stuff - Ireland partially sealed off from later expansions by extreme rainfall?
ReplyDeleteSo, the Irish are ultimately - and essentially - Russians.
ReplyDeleteI still can't get my head around that.
@mickeydodds1
ReplyDelete"So, the Irish are ultimately - and essentially - Russians.
I still can't get my head around that."
There's a reason why second generation Croatians and Irish and Swedes and Lithuanians and Poles and Finns - and Russians - all look like 'Americans.'
The same population which carried a large amount of EHG acquired a lot of CHG admixture as they moved east, and the western one acquired less, but it seems the European element was the same
ReplyDelete@Mikeyodds1,
ReplyDelete"So, the Irish are ultimately - and essentially - Russians.
I still can't get my head around that."
That's a bit of an over-interpretation, which is normal when there's interesting data like this. Yamnaya weren't Russians, Irish don't have more Steppe/Yamnaya-like ancestry than other Northern Europeans, at least 50% of Irish ancestry isn't Steppe.
Thanks for this work Davidski.
ReplyDeleteAttempt at a matrix of population averages: https://pastebin.com/xFt3580a
For populations in the matrix without ZZZ_ prefix only, so basically only broadly West and Central Eurasians and South Asians. There were otherwise too many samples in the matrix for the method I am using to calculate averages.
(To get around the problem of each member of population having itself as 1, I counted the highest value twice. Hopefully this did not majorly distort averages).
I'm not totally sure how to use any of this data myself:
Patterns per ancient sample look a bit less consistent and obvious per population, compared to the CHROMOPAINTER coancestry in Martiniano 2017's supplement...: e.g. http://i.imgur.com/gCDb4lN.png / http://i.imgur.com/SzUicnz.png
Still get solid population trees out of them though: http://i.imgur.com/gSor9iu.png
Poles have high affinity mostly with their neighbors but some ancient samples are also quite high:
ReplyDeletehttp://s3.postimg.org/wm8d7iywj/screenshot_251.png
In Martiniano's CHROMOPAINTER coancestry, the RISE47 Nordic Bronze Age has a particular pattern of showing greater coancestry with North East European populations outside Russia and Finland from Poland, Lithuania, Belarus: http://imgur.com/a/HKxB2
ReplyDeleteThat is unusual as most ancients in Martiniano's tend to show either peaks in NW Europe (for Steppe EMBA, Steppe LNBA, Europe EBA, Western Europe Iron Age) or Russians, Finnsand Chuvash (for Russian Iron Age).
Whether or not the NE Europe have less real coancestry with Steppe EMBA, Steppe LNBA, early Bronze Age Europeans etc or just have been through higher expansion / founder effect (which would also reduce coancestry) not so clear.
But in any case, the relative relatedness to the RISE47 sample compared to other ancients seems high.
So perhaps this sample may be an early relative of the Baltic Bronze Age populations from Jones 2017 that have some continuity with Narva and high f3 sharing with NE Europeans.
@Matt
ReplyDeleteLower effective population size seems to dampen Chromopainter coancestry with ancient samples. Orcadians appear to suffer from the same or similar effect. But the overall impact isn't big.
Have you tried running PCA using the cM matrix? You can start with all of the non-ZZZ marked samples; they'll produce a fairly coherent plot.
And then it might be useful to remove various samples to focus on specific things, like the relationship between Steppe_EMBA/MLBA and South Asians.
I'll have a go at that later today or tomorrow. It's a bit of a pain in the ass having to remove the same individuals from the rows and columns, so I'll need to organize things better in future runs.
Actually, I think I can organize the matrix into biogeographical zones. That'll make it easier to run different PCA.
ReplyDeleteDavidski: Lower effective population size seems to dampen Chromopainter coancestry with ancient samples. Orcadians appear to suffer from the same or similar effect. But the overall impact isn't big.
ReplyDeleteYeah, in Martiniano this effect certainly hit Basques, Palestinians, Druze, and Bedouin as well. It's harder to tell if it is relevant in causing less preservation of ancestral haplotype structure in NE Europe or Sardinians, as not obviously outliers relative to expected position as much as the aforementioned (e.g. Orcadians obviously displaced relative Norwegians, Basques relative to expectation, but more subtle, if evident in NE Europe / Sardinian.)
The relative and correspondence values in the CHROMOPAINTER analysis are probably more informative - correspondence analysis on the values places Basques, Orcadians where they "should" be, etc.
I'm finding PCA on this matrix produces some odd results: http://i.imgur.com/Rtg75vK.png. This was covariance PCA, ignores groups.
Correspondence shows some more structure - http://i.imgur.com/EmYTkHa.png
Esp after removing Yemenites and Saudis - http://i.imgur.com/gjhZ0X8.png
PCA with settings of correlation, between group variance mode shows some more structure and a much more normal shape: http://i.imgur.com/bdgzYwG.png. Does seem like even in that case, there is a lot of odd positions for some ancients, and that the structure tends to compress the Near East somehow.
Seem like these are much more hijacked and dominated by isolated populations than, esp. Basque, Zoroastrian, Kalash.
Is this similar to anything you're finding with PCA?
@Matt
ReplyDeleteI agree that founder effect and related issues could be involved with RISE47, but Unetice, Sintashta and Andronovo are E-European shifted in that matrix relative to Yamnaya, Afanasievo and Ireland_BA samples. Plotting something like Icelandic vs Ukrainian_East shows this.
@Matt
ReplyDeleteYeah, there's a lot of IBD sharing between certain ethnic groups in that matrix. The only way to get around this problem is to remove pairs and groups of individuals that share excess IBD.
This can be done by setting some sort of threshold for different types of PCA, or simply by removing problem samples after viewing plots.
For instance, the Basques share too much IBD to include most of them even on the West Eurasian plot, so it might be an idea to remove all of them except maybe 3 of the least related individuals.
Plotting Tajiks (Rushan or Shugnan) against Brahmins shows Afanasievo preference for the latter, especially over Sintashta.
ReplyDeleteYaghnobi will paint a different picture, one that might seem odd geographically but may fit its distinct position within Iranic.
What qualifies as a dominion? Spreading your autosomal DNA? If so Villabruna did just as well, as did CHG.
ReplyDelete@Ryan
ReplyDeleteSurely you're not suggesting that the spread of these more or less Villabruna and CHG related autosomal components can be compared to the massive and sudden explosion of closely related pastoralist groups from the western steppe during the Bronze Age, which in all likelihood carried with it Indo-European languages and very similar culture, and helped to shape the modern world?
This comment has been removed by the author.
ReplyDelete@Matt
ReplyDeleteHere's a more organized cM matrix.
https://drive.google.com/file/d/0B8XSV9HEoqpFQjYzbXVaTWVtLU0/view?usp=sharing
@ Shaikorth, could you screenshot and link the example?
ReplyDelete@Matt
ReplyDeletehttp://i.imgur.com/IeMcwQX.jpg
That looks interesting. What is that exactly?
ReplyDeleteTajik_Rushan vs Brahmin plotted for the samples in averages matrix Matt posted: https://pastebin.com/xFt3580a
ReplyDeleteTook Tajik_Shugnan out because it shares loads of IBD with Rushan (probably related to Rushani being a dialect of Shughni).
@ Davidski said...
ReplyDelete"@Ryan
Surely you're not suggesting that the spread of these more or less Villabruna and CHG related autosomal components can be compared to the massive and sudden explosion of closely related pastoralist groups from the western steppe during the Bronze Age, which in all likelihood carried with it Indo-European languages and very similar culture, and helped to shape the modern world?"
Ryan didn't but I did. We'll see next, when Tyrrhenian Italy and the line through Switzerland to Southern Germany, which brought, I think, all the R-L51 descendants, is tested for aDNA, and above all if R1a will be found within the Villabrunas, and of course the IE languages from there and, in other parts of Italy, Sardinian/Basque and the Etruscans if come with the agriculturalists from Aegean Sea to central Europe or developed between IE and Sardinian/Basque.
Read the paper of Blasco Ferrer. I don't agree that Sardinian language came from the Cantabrian Refugium (as the same Italian genetists like Francalacci agreed about the origin of the mt U5b3) but the other way around. Also that we'll see with aDNA from Italy.
I am saying from ten years, bringing infinite proofs and against all, that R1b1-V88 came from the Italian Refugium, and now...
ReplyDelete"Despite the fact that animal domestication originated in the Near East ∼ 10 ka, and that it was from there that animals such as sheep, goats as well as cattle were introduced into Northeast Africa soon thereafter, contemporary cattle keepers in the Sahel/Savannah belt show uniparental genetic affinities [Y R-V88, mt H1cb1, U5b1b1b] that suggest the possibility of an ancient contact with an additional ancestral population of western Mediterranean ancestry" (Kulichová et al., Internal diversification of non-Sub-Saharan haplogroups in Sahelian populations and the spread of pastoralism beyond the Sahara, Am J Phys Anthropol. 2017 Jul 24. doi: 10.1002/ajpa.23285).
"western Mediterranean" isn't Italy yet, but certainly isn't Middle East anymore.
Holy shit Gioiello, there was no steppe ancestry in Europe west of the steppe until the Late Neolithic/Bronze Age.
ReplyDeleteAnd then suddenly it was all over the place, as far west as Ireland and Portugal, and in samples belonging to R1b-M269 (inc. L51+), which wasn't seen in Western Europe before.
That's what the BEAGLE IBD test that I just ran clearly shows.
@ Davidski
ReplyDelete"Holy shit Gioiello, there was no steppe ancestry in Europe west of the steppe until the Late Neolithic/Bronze Age.
And then suddenly it was all over the place, as far west as Ireland and Portugal, and in samples belonging to R1b-M269 (inc. L51+), which wasn't seen in Western Europe before.
That's what the BEAGLE IBD test that I just ran clearly shows".
Must I be glad that the "shit" is "holy"? Perhaps you know that this adjective has no meaning to me. Of course I respect your opinions and am waiting for more data, but with IBD I saw many people doing everything, as with that part of some women I spoke in my irreverent Italian. Have you ever had anything to do with the most part of such calculators? Do you know Ancestry.Com, La la Land, Gedmatch etc etc? Beyond a few generations IBD becomes IBS and has no meaning. This is my opinion and my bet. Let's wait that Samuel's friends test Italy and nearby beyond what they did so far...
@ Gioiello
ReplyDeleteIf R1b spread out of Italy which is much closer to Spain than the Ukraine, why was R1b (V88) only detected in Spain during the 5000s bC.while R1b were present in the Balkans and Ukraine already as early as 9000 bC. ?
@ Ric Hern
ReplyDelete"@ Gioiello
If R1b spread out of Italy which is much closer to Spain than the Ukraine, why was R1b (V88) only detected in Spain during the 5000s bC.while R1b were present in the Balkans and Ukraine already as early as 9000 bC. ?"
1) First of all the oldest R1b1 found so far is Villabruna, Italy, 14000 years ago.
2) The R-V88 (xV35) found in Iberia is supposed having come from Italy with the Zilhao migration (7500 years ago).
3) As to R-L51-PF7589 Italy has the highest frequency and variance in the map of Argiedude and me. Eastward of Italy the petrcentage is close to 0 %.
These are the facts. East Europe has R1b1-V88 in the Balkans come very likely from the Italian Refugium after the Younger Dryas, and the above all R-L23-Z2105 of Samara from which only a few subclades survived, but not all, because Western Europe has independent subclades. All that could be verified or disproved only from aDNA.
Thanks Shaikorth and Matt for the idea. Please check out the update.
ReplyDeleteBy the way, it's not that I don't trust Matt's work, but I calculated the South Asian averages myself for the graphs that I posted, because Matt did something with his calculation that I didn't quite understand.
Davidski, but did you make your IBD with the charcoals of Chad?
ReplyDeleteWe could name them hence now "chadcoals" for not forgetting the name of this Harvardian scientist...
You sound like Marnie. It's rarecoal, and I do not go to, or work for Harvard. Stick to gardening and poetry. This isn't your kind of hobby.
DeleteChar, glad to see that you are always in tune. Poetry sometime, gardening no more. My garden is a Waste land. Charcoal/rarecoal... I invited many times Davidski to consider Shi Huang and his research, more interesting now with this Neanderthal introgressed from 200000 years old mt of homo s s .....
ReplyDeleteDon't know how MtDNA fits into this picture. I'm just an amateur but MtDNA H6 was in both the Andronovo & Okunev Cultures. Seems H6 might have been in the Afanasievo Culture as well. As far as my MtDNA matches at the country level, Germany is first, then Ireland & England. Perhaps this reflects CWC maternal lines being absorbed by Bell Beakers. The Beakers brought the line to Western Europe & the British Isles. Just a guess and I like the research.
ReplyDeleteSeems humans were not the only ones to carry the Central Asian genetic component to Europe in 3000 BC. According to this article Corded Ware Culture DOGS had an added Central Asian genetic component. http://www.ktvq.com/story/35938685/history-of-dog-tracking-when-dogs-and-wolves-diverged
ReplyDeleteA detail that is interesting for me: Mezhovskaya's position relative to Afanasievo is so consistent that it suggests Mezhovskaya had particular Afanasievo ancestry as opposed to getting that side of ancestry from groups that are closer both geographically and in time, like Sintashta or Andronovo. Either that or they absorbed a late Yamnaya lineage that was closer to Afanasievo than to the part of Yamnaya that is genetically mapped up to date.
ReplyDeleteAt the other hand Tajik_Shugnan breaks this correlation. Do they have a direct Mezhovskaya-like ancestry from Siberia? These Tajik tribes are very variable.
Woah, those figures you added. Can anyone make any sense of the Baloch vs Brahmin one? Particularly with how spread across the map Unetice and Ireland EBA are? Even the different samples from those same groups are way spread out. This makes 0 sense to me. We have all sorts of closely related groups that fan out and have vastly divergent affinities.
ReplyDeleteWould make sense that the steppe ancestry in southern Pakistan is less divergent from that in Northern India than the steppe ancestry in Tajikistan. Thus the patterns become more obscure.
ReplyDelete@Gioiello - Seriously? Rarecoal is short for rare allele coalescent. If you find statistical tools and peer reviewed research funny there's really not much to discuss with you in a productive way. Might as well be talking to Alex Jones about Martian slavers or some other tin foil hat nonsense. Either you base your views on the data or you don't.
ReplyDelete@ Ryan
ReplyDeleteApart "charcoal" for "rarecoal"... "chadcoal" could come only from charcoal and not rarecoal. Take in mind that I am writing in a language which isn't mine, whereas you all probably in the only language you know. I did know very well the meaning of rarecoal, so that I invited many times Davidski to take into account Shi Huang and his school, who are using just very rarely mutating SNPs, but I said that because seemed to me that Chad was wrong in his use of rarecoal from the Harvardians... but I should consider again that question, but it seemed right what I said, seen that Chad didn't protest. Thus if he were right, i.e. if really the Harvardians used that method, nothing against, but then why don't consider Shi Huang, who is the champion of rarerarecoal?
@Ryan
ReplyDeleteMakes perfect sense to me; has for a a while.
What doesn't jive for you?
ReplyDelete@Davidski:
By the way, it's not that I don't trust Matt's work, but I calculated the South Asian averages myself for the graphs that I posted, because Matt did something with his calculation that I didn't quite understand.
You were right to do so. I've had another think about it and there was a slightly simpler method that I could've used that would've calculated the average, where what I did has not quite calculated the average, though the differences should be very small. I wasn't thinking clearly.
@All - if you have not just seen this on twitter - "Genetic origins of the Minoans and Mycenaeans"! Lazaridis 2017... Big news.
@ Matt
DeleteThank you.
@Matt
ReplyDeleteChistmas has come early! Or maybe a tad late actually...
Thanks for the heads-up!
This comment has been removed by the author.
ReplyDeleteFour out of five males belonging to Minoans, Mycenaeans, and
ReplyDeletesouthwestern Anatolians (Supplementary Information section 3)
belonged to haplogroup J, which was rare or non-existent in earlier
populations from Greece and western Anatolia who were dominated by
Y-chromosome haplogroup G2 (refs 1, 2, 17). Haplogroup J was present
in Caucasus hunter–gatherers3 and a Mesolithic individual from Iran 4
Interesting and not outrageous.
Just as a first comment before I shut up and read, looking at the supplement looks like their qpAdm models on p26 are finding:
ReplyDeleteMinoans = CHG:Anatolia_N at 15:85
Myceneans = EHG:CHG:Anatolia_N at ≈5:19:86
So looks like if you model Myceneans as Minoans:Steppe_EMBA then 10:90 (assuming Steppe_EMBA roughly 50:50 EHG:CHG).
(Though the truth will be more complex and possibly have more admixture from different sources, e.g. Balkans / Armenia, etc.)
Abstract does say "However, the Mycenaeans differed from Minoans in deriving additional ancestry from an ultimate source related to the hunter–gatherers of eastern Europe and Siberia, introduced via a proximal source related to the inhabitants of either the Eurasian steppe or Armenia".)
They do also say "We were concerned that the admixture from these three sources could be driven by heterogeneity within the Mycenaean population itself. Mycenaeans do appear to form a tight cluster in PCA (Fig. 1b) and to have similar admixture proportions in ADMIXTURE analysis (Extended Data Fig. 1)." However, after discussion they state "Given the lack of evidence for population structure in this population with the available data, we will treat it as a unit in the remainder of this section. "
More continuity than Iberian scenario in Martiniano? Anyway, back to reading more carefully.
This comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteWell... another nail in Pontic Steppe PIE coffin, but I digress.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDelete@Al Bundy
ReplyDeleteAs they should. Steppe part of the equation, 100% part of the equation.
But the Pontic Steppe was not the original PIE Urheimat.
@Atrior Steppe spread to N.Europe,older languages directly from the Caucasus Iranian Plateau?
ReplyDelete@Al Bundy
ReplyDeleteNot quite.
Also, IE languages to N. Europe DID arrive via Russian Steppe.
This comment has been removed by the author.
ReplyDelete@Al Bundy
ReplyDeleteWhat about IndoIranian?
Exactly.
But what about specifically?
I wonder if somebody let something slip out here ?:
ReplyDeletehttps://www.google.co.za/amp/s/phys.org/news/2017-08-civilizations-greece-revealing-stories-science.amp
"Eastern European Steppe admixture before and after Minoans and Mycenaeans."
This comment has been removed by the author.
ReplyDelete@Al Bundy
ReplyDeleteDon't want to ruin the surprise; which I've stated before and greeted with predictable derision. But so far, the 2017 papers are confirming.
Anyhow, you're in the zone, more or less. It's a big zone though ;)
Yea I wanted to say don't reveal all the cards yet let's what the upcoming papers say.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteR1b isn't from Italy. There is no support for some great Italian refugium. Italy was so sparsely populated a southern group of Magdalenians went right through it and met the group from north of the Alps in Poland.
ReplyDeleteWHG is from the Balkans, which is why they share more ancestry with Near Eastern farmers than UP Europeans. This is common sense.
M269, L23, Z2103/5, L51, are only found in Yamnaya and related people. Simplest explanation is usually the best course to take.
@Al Bundy
ReplyDeleteYea I wanted to say don't reveal all the cards yet let's what the upcoming papers say.
Yes, and the most exciting one is not the one that people might expect.
However it turns out we finally have DNA from Minoans and Myceneans.Amazing.
Yes! Historic day!
Surfacing again (Should really be waiting til Davidski opens the comment thread for this paper tho. I know).
ReplyDeleteTable S2.26 seems to indicate that two working qpAdm models for Myceneans are either:
Anatolia_N:Armenia_MLBA:Steppe_EMBA ≈ 66:30:3
or Minoan_Lasithi:Armenia_MLBA:Steppe_EMBA ≈ 88:1:11
Either model is preferred at about the same fit. Anatolia_N is roughly interchangeable with Greek Final Neolithic (as they state), so I assume they use Anatolia_N because of sample quality.
Essentially seems like it works out like this because Minoan+Armenia_MLBA models would be too CHG rich (for the right level of EHG), while Anatolia_N+Steppe_EMBA models would be too CHG poor (for the right level of EHG).
OTOH, gap in their sampling, so not clear if a Mycenean mainland predecessor would be more like Greek_LN or Minoan. If for example mainland Greece were intermediate in CHG level between Minoan and Greek Final Neolithic, then model would assume the rest as 16% Armenia_MLBA and 7% Steppe_EMBA.
Their discussion effectively indicates models indicate Armenia_MLBA would be more likely to be a mass migration, which would be easier to explain language shift.
Linked to that if they had Balkans BA samples available, and they don't as Mathieson is still in preprint, they could try models including those as a more logical alternative to Europe_LNBA or Steppe_EMBA. Balkans_BA would also probably make a larger contribution to Mycenean, so an easier case for language change.
Their final comment in this section is to this effect: Further sampling of populations north and east of the Aegean may clarify the role that migration may have played in linguistic change in the region..
(Don't take my word for it though, read through the section; many more models present and I may have lost a detail as quick reading through).
Using Greek LN would make Myceneans more steppe-like. They were more Levant-like than Anatolia.
ReplyDelete@Matt
ReplyDeleteTheir final comment in this section is to this effect: Further sampling of populations north and east of the Aegean may clarify the role that migration may have played in linguistic change in the region.
Further sampling is always the best, no question.
That said, it's not dawning on them just yet what's going on...
@Chad - Their comment is:
ReplyDeleteHowever, all the Bronze Age populations also have ancestry related to the Caucasus or Iran, consistent with their shift in PCA (Fig. 1b). This shift began in Anatolia no later than the Chalcolithic (3943-3708 calBCE)16 and was not evident in Greece by the time of the Final Neolithic (4,230–3,995 calBCE) individual from Kleitos14 that resembled (like all other Greek Neolithic individuals) Anatolian farmers (Fig. 1b).
No specific mention of Kleitos having Levant_N ancestry, or not, unlike the BA Anatolians modeled with 9% or Tepecik Ciftlik modelled with 9%. But they don't Kleitos, so I don't know. Seems you would have to have a substantial amount of Levant ancestry in Greek_LN to get much up from 3% and might not even reach 11% even with a substantial amount, and I'm not sure sure if Levant ancestry would even inflate Steppe_EMBA rather than Armenia_MLBA in their model anyway.
I think someone let something slip. Steppe Migration in the Early Bronze Age and then a Early Middle Bronze Age Steppe Migration....
ReplyDeleteNo R1 ?
ReplyDelete@Rob
ReplyDeleteNo R1 ?
Not in any of 5 males sampled:
- J2a1d
- J2a1
- G2e2b2
- J2a1
- J1a
Yeah but only one was Mycenaean
ReplyDeleteStill, it seems that PIE wasn't solely R1
@Rob
ReplyDeleteJ2, the forgotten brother.
@Atriðr - I just don't understand why two members of the same culture would have such wildly different affinities to different populations in South Asia.Why would one Unetice sample be so much closer to Brahmins than another?
ReplyDeleteYeah
ReplyDeletePretty minimal steppe impact.
Probably result of some contact with Thracian Yamnaya.
I don't think it was the driver of PIE . But Ill read the paper, wait for raw data, pray/ meditate, then conclude.
@Rob, looks like very little steppe ancestry in these samples in the models where they have it, so no R (generally frequent in steppe populations, whatever the origin) in 1x Mycenean male is not too surprising.
ReplyDeleteThe Portuguese_BA samples have some R1b of the appropriate clades, but look to have a) more steppe ancestry and b) may have got it from a larger population movement from another part of Europe where R had already undergone a demographic processes putting it at high frequency despite substantial European_MN ancestry.
If Myceneans got steppe from a larger population movement from an intermediate population in the Balkans (as could be more likely than getting it direct!), worth noting that IRC there was absolutely no sign of any high frequency of R lineages in the Bronze Age Balkans from Mathieson 2017 and IRC J was if anything more frequent in those contexts than expectation.
This comment has been removed by the author.
ReplyDeleteGreek r1a is probably slavic in origin. What's the source of greek r1b though?
ReplyDelete@ Matt
ReplyDeleteWe did see some R1 in BA Balkans: a couple of Z2013s Vucedol, and a R1a-Z93 in MBA Thrace.
Yeah I think the Bulk of movement could be form Cotofeni- Usatavo cultures; and from Anatolia.
@ Rob, yeah, I think you're right and I'd trust your memory on this over mine. Main point I had is that didn't seem to be >frequency than steppe autosomal ancestry suggested.
ReplyDelete@ Matt
ReplyDeleteCould this be a hint ? Please read.
https://www.google.co.za/amp/s/phys.org/news/2017-08-civilizations-greece-revealing-stories-science.amp
"Perhaps you know that this adjective has no meaning to me."
ReplyDeletePretty unlikely. Even if you are an Atheist", its unlikely that there is nothing that is "holy" to you. Wich literate means "whole", "Complete", "Perfect" or "healthy" and wich effectively means: "this should preserved in all eternity and never changed" in Germanic and was picked to translate the Latin "sanctus" into Germanic languages.
I bet even the most leftist people would hold some things "holy" and if its the "Human rights" or something alike.
Swat Valleys ''Proto-Kafirs'' will be same .
ReplyDeleteNirjhar007,
ReplyDeleteRather cryptic comment; can you elaborate any further?
To whomever may be interested,
I thought this would be a good opportunity to test the power of PCA data for deriving ancestry proportions.
Mainly, I wanted to see if the Karasuk/Andronovo vs Steppe_EMBA pattern could be replicated with Global_10.
Scaling was applied.
Brahmin_UP
38.1% Iran_M/N
31.9% ASI
29.9% Srubnaya_outlier
distance=0.9914
Kalash
44.05% Iran_M/N + 6.15% Iran_Chal
39.05% Srubnaya_outlier
10.75% ASI
distance=0.4734
Karlani Pashtun, Central Highlands
28.1% Iran_Chal + 22.1% Iran_M/N
26.6% Srubnaya_outlier + 14.4% Karasuk:RISE499
8.9% ASI
distance=0.3186
Tajik_Rushan
25.70% Srubnaya_outlier + 11.70% Andronovo:RISE503 + 11.10% Karasuk:RISE499 + 3.15% Karasuk_outlier:RISE495
35.40% Iran_Chal + 6.35% CHG + 0.60% Iran_M/N
3.55% East-Asian_ENA
2.45% ASI
distance=0.1469
So yes, the same patterns are also evident with this sort of data.
At the end of the day, I'd say that nMonte with PCA is a very effective tool, if one actually attempts models that have some basis in reality.
@All
ReplyDeleteHad some time on my hands, so recalculated the averages here.
1. All Populations AAA-EEE (ancients, Europeans, West Asia, South Asia): https://pastebin.com/iNT3Nqac
2. All Populations AAA-EEE with n>1: https://pastebin.com/Jq9neG7q
Should be the proper average, without the weird workaround I used before for populations with n=1. (That method probably inflated population self similarity a bit). Where a population averages isn't meaningfully possible because n=1, I've put N/A.
(You can't graph in Pastebin where there is an N/A though, so if you use pastebin 1, you'll have to use find and replace with a placeholder number where I've put an N/A, if you want to graph any columns for N=1 populations).
Various graphing of some of these: http://imgur.com/a/lH1Fz.
(Very strong Slavic signal in Greek vs Cypriot. Appears to be strong Yamnaya signal in Brahmin vs Pathan).
Right, I'm going to do AAA, CCC-ZZZ next, leaving out the Europeans BBB.
@ All, and may be more interesting to Shaikorth.
ReplyDeleteOK, took a bunch more time that I thought due to lots to do with dataset, however:
1. All Populations AAA, CCC-ZZ (ancients, West Asia, South Asia, World other): https://pastebin.com/4JuduN8P
2. All Populations AAA, CCC-ZZ with n>1: https://pastebin.com/JYshGAu3
The range and degree of haplotype structure within East Asian, Siberian and Native American populations is huge. Looks comparable to within WHG for many North Eurasians and beyond for Karitiana (to an insane degree), while mainstream, large East Asian populations (Burmese, Han, Korean) not too dissimilar from within large modern European nations.
Graphs: http://imgur.com/a/tAQSh
2. All Populations AAA-EEE with n>1: https://pastebin.com/Jq9neG7q
ReplyDeleteSome random multivariate statistics based on these: http://imgur.com/a/pm43D
1: PCA of the whole matrix.
2: Neighbour joining of the whole matrix
3: PCA after removing the Kalash, Zoroastrian, and Saami who have very long self branches.
4. PCA after removing all ancients from #3.
5. Neighbour joining on data for #4
6. Detrended correspondence analysis on whole matrix.
7. Neighbour joining on output of #6.
Following kind of OT and relevant only to modern genetics (and probably of interest and making sense only to me lol, but felt like it was worth a comment).
ReplyDeleteComparing the Eastern European and Western European populations, I had the preconception that populations from Eastern Europe would share more fastIBD chunks within population.
However, looking at it, that doesn't actually appear to be the case: http://imgur.com/a/Ys2mL
The actual differences seem a bit different in a couple ways:
a) Populations that are a) south of the Baltic and b) west of Russia tend to have pretty similar within population sharing.
So e.g. Belarus, Ukraine, Poland, all seem to have within population sharing that is about the level typical of a large European country.
Populations on the edge into this zone instead tend to have much larger within population sharing. Not sure why this is, but may be due to being smaller populations with more recent expansion?
b) Baltic populations and Slavic and Uralic populations from Russia seem to share a lot of fastIBD with all other Eastern European populations, not just within their own population. Even though, e.g. Belarusians do not share an excessively high number of chunks with other Belarusians, the number shared with Russia North, for example, is very high.
(Also I'd note that in the Balkans, a region Coop et al identified as having a particularly high level of within population IBD sharing, there is a mixed picture; Albanians are pretty high, Croatians are relatively high, but between English (lower) and Norwegians (higher), while Bulgarians and Greeks have little within population IBD.)
Thanks Matt. The averages have apparently come in very useful for some people who are running tests on this output elsewhere.
ReplyDeleteDavid - where does Greek Z2103 derived yDNA fit into all of this?
ReplyDeleteDavid - where does Greek Z2103 derived yDNA fit into all of this?
ReplyDeleteIt's from one or more of the steppe waves into Greece, but I don't know which.