search this blog

Tuesday, September 13, 2016

ISBA7 PalaeoBarn abstracts


The abstract book for this week's meeting is available here. Emphasis is mine. The two abstracts on the genetic shifts in the East Baltic region might look as if they contradict each other, but they don't. What they're suggesting is that the East Baltic was basically home to typical European hunter-gatherers right up until the Late Neolithic, when the Corded Ware people crashed into the area, probably from the steppe via East Central Europe. For more on this topic also see here.

Mitochondrial DNA Analysis of Human Remains from Estonia

Pfrengle et al.

The transition from hunter-gatherer subsistence to farming is one of the most important processes in human history. In Europe, it has been found to be a result of demic diffusion originating from the Near East. The arrival of the first farmers in Europe lead to an increase of genetic diversity as well as genetic admixture of local hunter-gatherer and the migrating farmers. Previous studies investigating European human history using mitochondrial and genome-wide nuclear data from early farmers and hunter-gatherers have provided detailed insights into the process of admixture and replacement throughout the Neolithic period. However this process has been poorly studied in the Baltic region where archaeological research suggests more extensive scenarios. Here we reconstructed the complete mtDNA of 19 individuals from different archaeological sites of Estonia covering the timespan from the Narva Culture to the Corded Ware Culture and determined their mitochondrial haplogroups. The results show that the typical European hunter-gatherer maternal lineages are represented exclusively in all individuals from until the Middle Neolithic. From the Late Neolithic on, haplogroups that are associated with European Neolithic farmers are detected. The results indicate genetic continuity of foraging cultures of Mesolithic and early Neolithic backgrounds and a late demic diffusion into the territory of Estonia associated with people of the Corded Ware culture. In addition, the generated genetic data are used to gain insights into the demography of burial complexes by sex determination and maternal kinship analysis.


The Neolithic Transition at the Edge of Europe

Jones et al.

In Europe, the Neolithic transition marked the beginning of a period of innovations which saw people move from a mobile lifestyle, dependent on hunting and gathering for survival, to a more sedentary way of life based on food production. This new lifeway, which began in the Near East ~11 kya, spread quickly across the continental interior of Europe predominantly through demic diffusion.

While the genetic impact of the Neolithic transition has been well explored in central Europe, its impact on more peripheral regions of the continent has not been as extensively studied. To broaden our understanding of this dynamic phase in European prehistory, we analysed genomes from a 4,000 year temporal transect through the Baltic region spanning from the Late Mesolithic to the Late Neolithic period. We found evidence for connectivity from the Mesolithic to the Neolithic however, we also detected signals consistent with influxes from non-local populations. These influences were distinct from the early farmer admixture which transformed the genetic landscape of central Europe during the Neolithic. Interestingly, dietary stable isotope analyses (δ15N and δ 13C) show that the genetic shifts coincide with diversifications in subsistence strategy. These results suggest that the Neolithic was a period of genetic flux in the Baltic however, the cultural and technological changes observed were largely independent of forager-farmer genetic exchange.


Reconstructing population history in East Asia

Wang et al.

The deep population history of East Asia remains poorly understood compared to that of West Eurasia, due to the lack of ancient DNA data as well as limited sampling of present-day populations especially on the Tibetan Plateau and in southern China. We report a fine scale survey of East Asian history based on genome-wide data from ancient samples in the Amur River Basin, as well as 435 newly reported individuals from 53 populations. Present-day groups can be broadly classified into highly differentiated clusters, corresponding to Amur River Basin, Tibetan Plateau, southern natives and Han Chinese. Populations of the Amur River Basin show a high degree of genetic continuity from seven thousand years ago until today, and are closely related to the strain of East Asian related ancestry present in Native Americans. Tibetan Plateau populations are all admixed, deriving about 5%-10% of their ancestry from an anciently divergent population that plausibly corresponds to the Paleolithic population on the Plateau, and the remaining part from an ancient population that no longer exists in unmixed form but that likely corresponds to expanding farmers from the Middle and Upper Yellow River Basin who also contributed 40-90% of the ancestry of Han Chinese. A total of 10-60% of Han Chinese ancestry derives from southern Native populations, and we show that the type of southern Native ancestry that contributed to Taiwan Island Austronesian speakers is most closely related to present-day speakers of Tai-Kadai languages in southern mainland China.


Contextualizing the Tianyuan genome within present and ancient human genomic diversity

Yang et al.

Recently, many studies have produced an unprecedented number of ancient human genomes, providing insight on human dynamics in many regions, particularly West Eurasia and the Americas. Here, we present genome-wide data from the Tianyuan specimen, dating to ~40,000 years ago. Unlike other ancient genomes studied to date, the Tianyuan genome is the first ancient Upper Paleolithic sample analyzed to have contributed greatly to the East Eurasian ancestral lineage. We compare Tianyuan to several ancient and present day human genomes to better understand both the genetic diversity in the Upper Paleolithic and the similarities and differences between Tianyuan and present day populations. Overall, the addition of genome-wide Tianyuan data provides greater insight into the population history in Eurasia over the last 40,000 years.


Capture of ancient genomic DNA of individuals recovered from a Medieval Alemannic gravesite provides evidence for high mobility of fellowships during the 7th century CE.

O'Sullivan et al.

Whether the historic spread of cultural/language groups such as the Alemanni were migrations or local adoption of culture is still unresolved in archaeology. The Alemanni were a confederation of tribes that inhabited an area, from the third to the 10th century CE, which approximately overlaps with the modern distribution of Alemannic German dialect in Swabia. We present the genomic and isotopic data of eight individuals excavated from a gravesite in Niederstotzingen, Germany of supposed Alemannic origin dated to the 7th century CE. There were two multiple burials at the site suggesting either kinship or fellowship between the individuals. The tombs in the gravesite contained cultural artefacts and weapons indicating close contact of the Alemanni with Longobards and Byzantines. We investigated the genetic affinity of these individuals between each other and to modern West Eurasians. The genetic analysis utilised the targeted enrichment and sequencing of over 1.2 million genetic markers that have known ascertainment. From these data, we found no familial relationship among the individuals in the multiple graves, thus supporting a burial practice based rather on fellowship. All individuals were genetically male. The genetic affinities of the individuals, based on modern genetic distributions, were five Eastern Europeans, two Germans/Austrians and one Southern European. Isotopic data supports that only the Southern European individual was certainly born outside this region. The genetic data appear to correlate with the provenance of the burial artefacts, showing that westward movements and interactions among cultural groups likely occurred in this region during the 7th century CE.


Origins and genetic legacy of the first people in Remote Oceania

Skoglund et al.

The appearance of people associated with the Lapita culture in the South Pacific ~3,000 years ago marked the beginning of the last major human dispersal to unpopulated lands, culminating in the settlement of eastern Polynesia ~1,000-700 years ago. However, the genetic relationship of these pioneers to the long established Papuan peoples of the New Guinea region is debated. We report the first genome-wide ancient DNA data from Asia-Pacific region, from four ~2,900 to ~2,500 year old Lapita culture individuals from Vanuatu and Tonga, and co-analyze them with new data from 356 present-day Oceanians. Today, all indigenous people of the South Pacific harbor a mixture of ancestry from Papuans and a population of East Asian origin that we find to be a statistical match to the ancient Lapita individuals. Most analyses have interpreted the ubiquitous Papuan ancestry in the region today-at least 25%-as evidence that the first humans to reach Remote Oceania and Polynesia were derived from mixtures near New Guinea prior to the Lapita expansion into Remote Oceania. Our results refute this scenario, as none of the geographically and temporally diverse Lapita individuals had detectable Papuan ancestry. These results imply later major human population movements, which spread Papuan ancestry through the South Pacific after the islands' first peopling.


An ancient genomic perspective on the horse domestication process

Librado et al.

The domestication of the horse in the Pontic-Caspian steppes some 6,000 years ago represents one major turning point in human history. With horses, humans could travel for the first time well above their own speed and carry their germs, culture and genes across vast geographic areas. The development of horse-drawn chariots and cavalry also radically changed the history of warfare and was instrumental to the emergence of transcontinental empires. Additionally, beyond the battlefield, farm horses have massively impacted agricultural productivity. The biological changes that accompanied the process of horse domestication are, however, difficult to reconstruct from current patterns of genetic diversity both due to the development of intensively selected and extremely influential breeds during the last two centuries, and the almost extinction of wild horses. Recent developments in ancient DNA research have opened for the characterization of complete genomes, epigenomes and microbiota over long time series. We have applied such approaches to a large panel of horse remains spread across Eurasia and dated to 44,000-200 years ago. This started revealing the genetic structure of horse populations prior to and during early domestication stages as well as the history of genetic changes that accompanied their further transformation in a range of cultural contexts. I will present our latest progress made on an extensive dataset of ancient horse genomes spanning the whole domestication temporal and geographical range.


Mobility between the Aegean and the Levant in the Late Second Millennium BCE: inference from ancient DNA of pigs

Meiri et al.

The Late Bronze and the early Iron Ages (ca. 1450-950 BCE) of the eastern Mediterranean region are characterized by dramatic historical processes. Empires emerged and collapsed, trade connections were established and severed, and at the end of this era socio-political unrest and migration of large groups of people were rife throughout the region. In the 12th century BCE the movements of the so-called "Sea Peoples" affected wide parts of the East Mediterranean. We study the nature of human movements during this period on trade connections, culture and animal husbandry using the ancient DNA of domestic animals, above all pigs. We recently showed that in Israel, European pig haplotypes appeared ca. 900 BCE, and soon after took over the gene pool, with all modern wild boars in Israel carrying European mitochondrial DNA. Here, we broadened the chronological and spatial scopes by studying ancient pig mitochondrial DNA from the southern Levant and Greece. The Near Eastern haplotype Y1 and supposedly Near Eastern haplotype Y2 were discovered in Greece in the mid to late 3rd millennium BCE, while the European haplotypes were found in Israel in the early Iron Age IIA (ca. 900 BCE). We propose that pigs were moved between Europe and Anatolia since the early Bronze Age. Connections between Greece and the southern Levant are observed in the Iron Age, and probably result from the migration of Sea Peoples to the east. These results shed light on networks and movements of people during both times of prosperity and crisis.

127 comments:

Tobus said...

Great to see a full genome sequence of Tianyuan - all the existing inferences are just from a single chromosome.

Hector said...

It seems that the case is closed for Tianyuan. He is East-Eurasian, but probably not East Asian yet. Whatever is his Y haplo Davidski will claim a "moral victory". When he is right(rarely) he gloats. When he is wrong, "close enough". When he is angry,Fu and Yang hate White folks.

Davidski said...

Hector watch out, there's a Eurocentric in your closet, and under your bed...

Hector said...

Gosh I miss Polakko. That sublime expression of supreme loonitude. Where is he , Davidski?

Davidski said...

Hector, you're writing checks that you can't cash.

Give me an example of the so called "loonitude".

Hector said...

"Give me an example of the so called "loonitude""

No, but an example of "irony" is easier.

terryt said...

"The deep population history of East Asia remains poorly understood compared to that of West Eurasia"

Perhaps part of the reason is this: 'watch out, there's a Eurocentric in your closet, and under your bed'. And although Chinese geneticists are doing a tremendous job many are primarily interested in promoting an ancient Chinese identity exactly the same as the present rather than conceding the obviously great change that has happened through human movement. But of course I was especially interested in this:

"Today, all indigenous people of the South Pacific harbor a mixture of ancestry from Papuans and a population of East Asian origin that we find to be a statistical match to the ancient Lapita individuals"

That has been basically accepted for at least 20 years but this paper should silence the many critics. For example I had much heated discussion with Maju over this finding:

"These results imply later major human population movements, which spread Papuan ancestry through the South Pacific after the islands' first peopling".

That movement failed to move much beyond Fiji. This is another recent paper regarding the Lapita people that some will surely find interesting:

http://www.pnas.org/content/113/2/292.full.pdf

Davidski said...

Hector, that's quite a retort. I'm speechless as a result; totally stunned. I'll probably be like this for the next 24 hours, if not more.

Shaikorth said...

It will be interesting to see how Tianyuan's affinities to paleolithic Europeans are. If it looks the same as before, we can assume several "undifferentiated Eurasian" lineages, some of which contributed and some did not. Unfortunately they don't say anything about the non-Tianyuan parts of the EE lineage.

Important dog/wolf related stuff. Suggested that wolves descend from a relatively recent founder effect. Since dog diversity is lower than wolves, this indirectly supports a single not too early origin of modern domestic dogs, if not for a single dog domestication event.

"Contemporary wolves trace their ancestry to an expansion from Beringia following the Last Glacial Maximum
Liisa Loog
The grey wolf was one of the most successful predators since the beginning of Pleistocene and among the very few large carnivores that survived the megafaunal extinctions at the end of the Pleistocene. The grey wolves are also widely believed to have been the source population for all modern domestic dog populations. Today grey wolves play an important ecological role as a key top predator across Eurasia. Nevertheless, very little is known about the demographic history of this large carnivore. We explicitly tested different wolf demographic scenarios, including variat ion in population size and range expansions with origins in different parts of the Northern Hemisphere, by combining a large set of whole mitochondrial genomes, from both ancient and contemporary wolf populations with spatially and temporally explicit population genetic modeling. The demographic scenario best supported by our analyses is a model of a post Last Glacial Maximum expansion of Beringian wolves across Eurasia and North America, which to a large extent replaced the local native wolf populations. The inferred timing of this expansion coincides with previously described demographic turnovers in other species such as aurochs and humans. These results have major implications for the studies of geographic origin of domestic dogs, which up until now have all assumed a static wolf population structure throughout Pleistocene."

Shaikorth said...

Earliest Andean "domestic canid" thought to be a dog seems to have been an Andean fox. Similar to the domestication of leopard cats, looks like it didn't last.

"Genetic Identification of Canid in High Altitude Palaeoindian Peruvian Site
Feuerborn et al.

The Late Pleistocene saw the arrival of humans to the New World accompanied by the first domesticate, dogs. As humans progressed from Beringia to Patagonia they encountered new canid species in North America such as the grey fox and coyote, along with novel South American canid species, including maned wolves, bush dogs, and Andean foxes.Cuncaicha rock shelter site in southern Peru's Pucuncho Basin, is currently one of the oldest known high altitude sites in the world,with occupation beginning approximately 12,400 years ago. A human burial from the site was found with a canid mandible in a shallow pit at the head. Radiocarbon dating of the human and canid remains established contemporary dates at approximately 9,300 to 8,800 calibrated years before present. The specimen was initially identified as a domestic dog based upon the size and morphology of the partial mandible, which would have made the specimen the world's oldest known high altitude domestic dog and the oldest South American dog. Subsequent ancient DNA analysis through shotgun sequencing and mitochondrial enrichment of the specimen revealed the ‘dog' from Cuncaicha shared more similarity with the Andean fox, Lycalopex culpaeus. Computational analysis showed more reads mapping to the Lycalopex culpaeus genome compared with the Canis lupus familiaris genome.The Andean fox has played interesting roles in the lives of humans residing in Tierra del Fuego from archaeological through to historical contexts. Cuncaicha Rock Shelter now joins the list of sites with intriguing interactions between Andean foxes and humans."

And some support for scenarios previously proposed.

"Direct earliest evidence of dog domestication in East Asia

Hu et al.

As one of human best friends and pets, domestic dogs play many roles in human society. However, how, where and when it was domesticated are still controversial. Especially, the dog domestication in East Asia has not well under stood in the zooarchaeological assemblages. In this presentation, the canid remains at the Shuangta Site, Jilin, China dated to the 11000 years ago were carefully studied to differentiate the domesticated dogs from the wolves by the methods of biometric measurements. In addition, stable isotope(C, N) analysis of the animals and human bones were undertaken. Quite large isotopic variations were found in the canid population, indicating some canids might have been controlled intentionally by humans. Therefore, our study provides the direct earliest evidence of dog domestication in East Asia."

Simon_W said...

That Alemannic study comes completely unexpected, but it's fantastic, an old dream of mine coming true: At last some early historical samples from the main area of my own ancestors. Quite a surprise that the majority of them looks rather eastern European. Well, there was influx from the Carpathian basin in the area, most of all from the Quadi, aka Danube Suebi, who had lived around southern Slovakia. But they were a Germanic tribe, so I didn't expect them to be Eastern European-like. My 3/4 Swabian grandmother scored 1% eastern European and 1% Finnish in the FTDNA MyOrigins analysis, which isn't a lot. Apparently the influence of these easterners was quite diluted afterwards. But she does have quite a strong southern/southeastern European admixture. I used to think it was from the Roman era, but since that south European-like individual had migrated in from elsewhere, that influence might be rather Byzantine then.

BTW, the abstract is a little wrong about the geographical extent of Alemannic settlement. It's not just in the Swabian area, but in the whole of southwestern Germany, plus the German speaking part of Switzerland, the Alsace in France, Liechtenstein and westernmost Austria (Vorarlberg).

Simon_W said...

Or maybe the Elbe Germanic ancestors of the Alemanni had picked up a lot of East European-like DNA from the Lusatian Bronze Age substrate in eastern Germany? We don't know what the Lusatian people were like, but given their focus around Poland, it's somewhat likely that they were rather eastern.

Simon_W said...

Well, judging from placenames the actual Slavic settlement in northeastern Bavaria almost extended down to Niederstrotzingen, it faded out somewhat to the northeast of it. But I'm not sure if they were already there in the 7th century, and the site in question didn't show any Slavic cultural influence.

Rob said...

Yes Simon
Slavic settlement came long after the Alemani were formed. I wonder if the "EE" influence is mediated by eastern Germani & Sarmatians

Ir Pegasus said...

Broushaki et al. Early Neolithic genomes from the eastern Fertile Crescent.

Abstract: We sequenced Early Neolithic genomes from the Zagros region of Iran (eastern Fertile Crescent), where some of the earliest evidence for farming is found, and identify a previously uncharacterized population that is neither ancestral to the first European farmers nor has contributed significantly to the ancestry of modern Europeans. These people are estimated to have separated from Early Neolithic farmers in Anatolia some 46-77,000 years ago and show affinities to modern day Pakistani and Afghan populations, but particularly to Iranian Zoroastrians. We conclude that multiple, genetically differentiated hunter-gatherer populations adopted farming in SW-Asia, that components of pre-Neolithic population structure were preserved as farming spread into neighboring regions, and that the Zagros region was the cradle of eastward expansion.
http://science.sciencemag.org/content/early/2016/07/13/science.aaf7943.full

Suevi said...

About 460 AD...
"Now this country of the Suavorum has on the east the Baibaros, on the west the Francos, on the south the Burgundzones and on the north the Thuringos. With the Suavis there were present the Alamanni, then their confederates, who also ruled the Alpine [Alpes] heights, whence several streams flow into the Danube [Danubium], pouring in with a great rushing sound. Into a place thus fortified King Thiudimer led his army in the winter-time and conquered, plundered and almost subdued the race of the Suavorum as well as the Alamannorum, who were mutually banded together."

I wonder whether Suevi/Suavi were genetically Eastern European, Slavic-like...

Simon_W said...

@ Rob

That's also a possibility, something like the Goths. Gothic influence shows up in the archaeological record here and there, but then again the area didn't really belong to the staging points of neither Visigoths nor Ostrogoths.

@ Suevi

Yes, I seriously consider that as well.

And maybe the east European influence didn't really get bred out or strongly diluted, but maybe Southwestern Germans should be modeled as a mix of French-like Gallo-Romans and East Euro-like Alemanni, with some southern admixture on top of that.

Alberto said...

I think that the ancient DNA from the Amur River basin is going to be extremely interesting.

The Laz. 2016 model of East Asians as Onge + ANE was really poor, so this DNA might help to find a much better model (where the Amur River samples might be the "ANE" that they were looking for, while the rest is from southern China).

The overall relationship of this Amur River samples to ANE is going to be interesting too. Maybe it will be possible to model ANE as West Eurasian + Amur River? That's probably reading too much into too little information, but I'm really interested in seeing what those samples really are.

Suevi said...

About 550 AD...
"And in Gaul there flow numerous rivers, among which are the Rhone and the Rhine. But the course of these two being in opposite directions, the one empties into the Tuscan Sea, while the Rhine empties into the ocean. And there are many lakes in that region, and this is where the Germans lived of old, a barbarous nation, not of much consequence in the beginning, who are now called Franks. Next to these lived the Arborychi, who, together with all the rest of Gaul, and, indeed, Spain also, were subjects of the Romans from of old. And beyond them toward the east were settled the Thuringian barbarians, Augustus, the first emperor, having given them this country. And the Burgundians lived not far from them toward the south, and the Suavi [Σουάβοι] also lived beyond the Thuringians, and the Alamani, powerful nations. All these were settled there as independent peoples in earlier times."

These Suavi [Σουάβοι] of Procopius lived beyond the Thuringians only ~80 years before Sclavi Winidi of Einhard raided Thuringians in 631 AD.

About 631 AD...
"In the tenth year of the rule of Dagoberti it was reported to him that an army of Winidorum raided Thoringiam. He left, therefore, with his forces from Mettis and crossed the Ardenna towards Magantiam in order to cross the Rhenum there. In addition to dukes and counts he also had the choicest cohort of brave men from Neuster and Burgundia. There appeared now emissaries of the Saxones in front of Dagobertum asked him to exempt them form the taxes that they [normally] paid to the [Frankish] state. In exchange for that they promised with great zeal and success to defend [against the Winidi] and the Francorum country on the Winidis border to protect. Dagobertus fulfilled this request, after a/the council of the Neustrasiorum and the Saxonibus emissaries gave their promise, in accordance with their custom by hitting [their] weapons, for the entire Saxonibus nation. Though, the promise was not successful, the taxes that they used to pay remained unpaid in accordance with Dagoberti order. Chlothario the Old had required them to pay annual taxes of 500 cows which taxes Dagoberto now exempted them from."

Davidski said...

The Amur River basin ancient samples are pretty much like present-day Ulchi. So in the K7 they'll look something like this.

AG3-MA1 17.09
Andamanese 0.61
Basal-rich 0
Oceanian 0.06
Southeast_Asian 82.21
Sub-Saharan 0
Villabruna 0.02

Alberto said...

@Davidski

If so, then nothing really new. I thought they might represent an unadmixed North East Asian branch, rather than a mixed East Asian + ANE.

The abstract mentions they are "closely related to the strain of East Asian related ancestry present in Native Americans". I guess that by now they should be able to tell the difference between East Asian and ANE. But you never really know with abstracts, so you're probably right about them being similar to Ulchi.

Shaikorth said...

"The Laz. 2016 model of East Asians as Onge + ANE was really poor, so this DNA might help to find a much better model (where the Amur River samples might be the "ANE" that they were looking for, while the rest is from southern China)."

You probably can model Han as Ulchi + Dai but this doesn't invalidate the Onge+ANE model. Dai and other populations of the "southern indigenous cluster" were also Onge+ANE-mixes in the model, they just had something like 1-2% less ANE than Han.

Kristiina said...

Don't you find it intriguing that in Western Europe mtDNA U5, U4 and U2 are considered WHG hunter-gatherer lineages while H, J, T, N1 and K are considered farmer lineages, and Corded Ware is considered a kind of a hunter-gatherer mtDNA revival with lots of U4 and U5, but in the Baltic area Corded Ware introduces the farmer mtDNA.

I had a quick look at Ancestral Journeys site and checked that there is quite a lot of farmer mtDNA on Central European Corded Ware sites. This should mean that Corded Ware assimilated a lot of farmer ancestry and probably also their culture and it is this mixed population that reached the Baltic area. I am particulary interested to know if Corded Ware arrived to the Baltic area from Poland or from Belarus.

Davidski said...

Those Amur River basin genomes can't be any different from the genomes from the Devil's Gate site. They might even be the same samples.

http://smbe-2016.p.asnevents.com.au/days/2016-07-04/abstract/34323

If so, they're just East Asian + ANE, because that's what Ulchi are. Modeling Ulchi as unadmixed East Asians can't work, because they're closely related to Amerindians for specific reasons, including a fair whack of ANE admix.

Rob said...

Kristiina
Your question can be answered with archaeology: CWC arrived to Baltic via Poland, because it's earlier there, and is the classic CWC form c.f. MDC in Belarus

@ SimonW
I think in the final decades of the existence of the west Roman Empire; much of the limitenei were from Pannonia, and it's attended mixes.
But I agree, I'm curious to see what eastern Germani "looked like"

Kristiina said...

Rob, thanks, I am happy about that! :-)

Hector said...

There was another paper on the 7000 years old remains from Russian Far East(actually Wang's paper may have used the result from that paper) and it reported Tungusic people(maybe ulchi) were the closest living descendants and noted Koreans and Japanese as well.

I have quite a few Korean and Japanese geno 2.0 results etc. but none of them had any significant ANE. Granted that it is based on pre-2016 laz. paper which under-reports ANE components among East Asians but I am confident that the "regional continuity" they talked about was not on the ANE component, at least not without more general East Asian components that go along with it.

The Northeast Asian component around Amur that Koreans and Japanese partially inherit probably belongs to "Southeast Asian" component in the more crude classification you guys have set out to use.

The Han component that the jingoistic Chinese researchers refer to(as is usually expected from them) is probably Central East Asian one. The Russian paper stated that Koreans and Japanese were admixed populations involving Northeast Asia(Russian Far East etc.) and more southerly populations from modern day China. Not sure whether the other component is Central East Asian or Southern East Asian.

People of Amur river basin and the adjacent Russian Far East have Y chromosome profiles that are heavy in C3-M217. Very little Q(but you get substantially more as you go north along the coast) and even less R. moderately low O3 and O2b the latter of which may have a Korean or Manchu/Jurchen connection. moderate amount of N's as well. Those who seek the usual "ANE Y markers" will be utterly disappointed.

Davidski said...

@Hector

I have quite a few Korean and Japanese geno 2.0 results etc. but none of them had any significant ANE.

But you don't know how to test for ANE in Koreans and Japanese, so?

These results are in line with ~45% ANE in South Amerindians.

Pop:ID Japanese:HGDP00748
AG3-MA1 7.83
Andamanese 2.7
Basal-rich 0.03
Oceanian 0.7
Southeast_Asian 88.7
Sub-Saharan 0.03
Villabruna 0

Pop:ID Korean:ND13299
AG3-MA1 7.58
Andamanese 1.52
Basal-rich 0
Oceanian 0.2
Southeast_Asian 90.65
Sub-Saharan 0.05
Villabruna 0

Hector said...

"Modeling Ulchi as unadmixed East Asians can't work, because they're closely related to Amerindians for specific reasons, including a fair whack of ANE admix."

That is very unlikely.
"are closely related to the strain of East Asian related ancestry present in Native Americans."

Obviously they are referring to the East Asian part, not ANE part. Otherwise they did not need to make the distinction.

Davidski said...

Ulchi are part ANE you knucklehead.

Have a look at the models in Lazaridis et al. 2016.

Hector said...

"But you don't know how to test for ANE in Koreans and Japanese, so?"

Neither do you. Actually you don't even know how these statistics were derived. LOL. It is actually easy to see through your charades of covering up.

Davidski said...

Actually, I do, using formal stats, Admixture and PCA.

How about you?

Hector said...

If other populations in the reference panel also have ANE components ANE portion will be under-reported. I already made that qualification.

This paper was not referring to the mere genetic distance between Ulchi and NA. NA's East Asian component(naturally excluding ANE one) is close to Ulchi.
If not they just would have said that Ulchi is close to NA overall.

I remember DNA-forums. You will never understand.

Davidski said...

Let's see you do something useful instead of talking crap.

Post some models for Ulchi, Koreans and Japanese. I'll check your models and will post all the output and raw data online.

Hector said...

"Actually, I do, using formal stats, Admixture and PCA."

No, you don't. I have seen you arguing with dienekes. You don't go far beyond using these tools as you would use cook-books. 5 years since the demise of DNA-forums? You could not possibly have learnt all that much with your limited mathematical skills.

Davidski said...

Well let's see your mathematical skills then . Feel free to get in touch with Dienekes to help you out.

God knows, he needs a bit of encouragement, considering all the bullshit he came up with, including about R1a, turned out to be wrong.

Post some models and let's discuss them. I can post all the output and data online, no problems.

Hector said...

"Post some models for Ulchi, Koreans and Japanese ..."

Bha. Not a chance. Just wait until the paper comes out.
You will be weeping as you did when Ust Ishim did not turn out to be K2b mightily crushing your hopes and dreams.

Davidski said...

I'm offering you a chance to embarrass me on my own blog, in front of thousands of people (that's how many visit new threads each day).

I would've thought that you'd jump at the chance?

So the only thing I can think of that would prevent you from taking up my challenge is that you're too stupid and lack the necessary skills. Right?

Hector said...

"I'm offering you a chance to embarrass me on my own blog"

You are not that important actually.

I can earnestly explain how I detected your fraud but that won't be understood by most of your readers here who are your friends and hostile to me for obvious reasons. Over 20 years of my internet career I have learned my lessons.

As for mathematical skills ... if you were not good enough to represent Australia in IMO you are probably outclassed. Not that I did but making the US squad is much much more difficult.

Davidski said...

Enough with the hot air you freak.

You're claiming that Ulchi, Japanese and Koreans don't have any ANE admix.

Show us some models here that demonstrate this, and I'll see whether they can be reproduced.

If I can't reproduce them, I'll explain why in detail, and will post all the output and data.

Nirjhar007 said...

Hector,
You have data from the Paper?.

Davidski said...

Only data he has is out of his ass.

Nirjhar007 said...

Dave, either he wanted to take his 'frustration' out somewhere or he knows what he doing and will back up his words .

Hector said...

"You're claiming that Ulchi, Japanese and Koreans don't have any ANE admix"

Incredible. strawman at its worst.

ANE among the above is unknown. In 2015 it was estimated to be extremely low. In 2016 by the same author it was scaled up a bit. I am not making a judgement one way or another but a reasonable upperbound is not hard to guess. By comparing with the text in the abstract and the Russian paper, it is clear that the bulk of "regional continuity" is NOT due to the ANE component. This is in addition to the straight-forward textual analysis.

Is it so difficult to understand?
It probably is as you have not learned a thing during the 5 years you repackaged yourself.

Davidski said...

Idiot,

I'm not asking you to analyze the presentation abstract. I'm asking you to analyze data and refute my claims of significant ANE in Ulchi, Japanese and Koreans.

Do it and come back here with your results.

Hector said...

"Idiot"

Moron,

"I'm not asking you to analyze the presentation abstract. I'm asking you to analyze data and refute my claims of significant ANE in Ulchi, Japanese and Koreans."

Why would I want to do that?
Hilarious. It is not even your original idea.

Your corny "original" idea was to inflate ANE among East Asians(like 40 percent) just to get rid of Basal components among West Eurasians(or tuck it under the West Eurasian clade). We all know why you want that and you failed already...miserably.

I have no problem with the inflated ANE figure among East Asians, but the removal of basal components among WEurasians is your wishful thinking.

Gioiello said...

But Hector and Agamemnon weren't enemies? And wasn't Achilles who killed him?

Very likely will know the origin of many Jewish Y from "Sea peoples" wild boars. Is it for that that Semites forbid eating pigs?

"We recently showed that in Israel, European pig haplotypes appeared ca. 900 BCE, and soon after took over the gene pool, with all modern wild boars in Israel carrying European mitochondrial DNA. Here, we broadened the chronological and spatial scopes by studying ancient pig mitochondrial DNA from the southern Levant and Greece".

Nirjhar007 said...

Who are you Odysseus ? ;) .

Hector said...

"But Hector and Agamemnon weren't enemies?"

Actually Agamemnon was the ID of a notorious Greek nationalist loon back in the days of usenet, always fighting with Turkish guys. Initially I suspected that Dienekes was he as they were saying similar things like 1D(hammer's old nomenclature equivalent to R1a) is Mongoloid etc. I don't know whether the two are the same to this day. It is odd since Greeks have a lot of R1a's.

Gioiello said...

Actually I descend from Villabruna (I am R1b1a2-L23-Z2110, who was 100% WHG, and my closest relatives in those times were Vespasianus and Hadrianus, who were "pigs" with all their legionaries...

Shaikorth said...

"By comparing with the text in the abstract and the Russian paper, it is clear that the bulk of "regional continuity" is NOT due to the ANE component. "

If we go by the Lazaridis model the non-ANE ancestry in Amur River valley, Native Americans and Southeast Asians alike is like Onge, and these neolithic samples aren't like Onge. If they are most similar to Ulchi it's because of comparable levels of ANE.

Davidski said...

@Hector

So you don't have the ability to take this beyond a simple online screaming match. Shame. But I expected as much. You offer nothing and are a complete waste of time.

Btw, I'd hate to break this to you, but the deep ancestry of East Asians is far from a settled issue. Those abstracts don't resolve anything, and many questions will remain even after the papers are published.

My recent speculation about layers of ANE-related admixture in East Asians based on some TreeMix output is just that, speculation. However, it may well be proven correct at some level when the relevant ancient genomes are sequenced.

On the other hand, the idea that East Asians have by and large significant ANE admixture looks pretty solid. I was indeed the first person to float this in public, via this blog, and that's easy to prove, although it's impossible to say how many people were seeing the same thing behind the scenes before me. It really doesn't matter.

Now piss off.

Nirjhar007 said...

Gioiello,

lol

BTW Your researcher-

http://uu.academia.edu/PeterSchrijver

I find the hattic paper nice . On the Celtic issue, I think you meant the inverted one?.

Gioiello said...

I know an Agamemnon on Anthrogenica (from where I was banned, and glad to write here now) and someone said me that he is Yaniv Erlich, one of the authors of the shit's shit, i.e. DNA.land, pretty stupid as the guy who spoke denying that Ashkenazim had some Khazar intake (with infinite Others). Not all people hates pigs.

Nirjhar007 said...

Hector ,

I agree with Dave here . You have to put some data or numbers on the table or you should just leave..

Gioiello said...

@ Nirjhar007

I thank you very much for the link. I downloaded and printed those papers. I have many to study for the next month. Only a glance to the paper on Hattic. If they were Middle Easterner agriculturalists I am very glad, so I am sure they weren't Southward.

Nirjhar007 said...

Dottore ,

This is his email P.C.H.Schrijver@uu.nl .
http://www.uu.nl/staff/pchschrijver/0

If you feel inspired enough from his works , I think you should make a personal communication with him . You will provide him the genetic insights and I am sure he will appreciate them !.

I myself will contact him . But first I will read his works carefully.

Hector said...

@Shaikorth

http://www.nature.com/jhg/journal/vaop/ncurrent/abs/jhg2016110a.html

It is "open access".
Jomons have a very divergent ancestry.

You are making the same mistake as Davidski; you are conflating different time scales. The affinity Wang refers to is the one sensitive even to population differentiation < 15000 or so. "Onge-like" component can differentiate into various East Asian and Southeast Asian populations. Actually it is certain that they did.

It is clear that the Russian paper saw regional continuity more with Koreans and Japanese than with Chinese and others. If they dealt with only the crudest deep ancestries(ie Onge - ANE) they would not have found sufficient differences between Koreans and Chinese, even southern Chinese. It is very unlikely that ANE is so much greater among Koreans than among Chinese.

Instead the "affinity" is with respect to Onge-like component that differentiated over long time into many regional varieties. In this respect clearly Koreans are likely to be more closely related to the devil's gate specimen than Chinese are. The ANE component, as it appears to be a paleolithic introgression, also was differentiated into regional varieties along with the Onge-like component that went together. However Wang's paper specifically refers to the "East Asian part" of NA.

The reason I gave you the link to the Jomon paper is that even though the Jomons are so divergent, in the figure presented by Davidski they(at least 20 percent up to 40 of Japanese) all appear to go under "Southeast Asian"(or maybe some into ANE if ANE introgression already took place by then). That is due to the different time scale of differentiation.

Davidski said...

@Hector

Because ANE is highly divergent compared to native East Asian components, it has a relatively bigger impact on models.

So to get around this problem you'd need to take into account the ANE admix and focus solely on the East Asian components. However, since the abstract doesn't mention anything about ANE, it's not clear whether they even considered the possibility that East Asians carry ANE. They probably know that Amerindians have a lot of ANE, but this is beside the point.

This is what they say...

Populations of the Amur River Basin show a high degree of genetic continuity from seven thousand years ago until today, and are closely related to the strain of East Asian related ancestry present in Native Americans.

It sounds like they just assumed that the populations of the Amur River Basin don't have any ANE, and ended up with a spurious result.

Why a spurious result? Because if you assume that these East Asians don't have any ANE, and they do, then they'll look more similar to Amerindians than other East Asians, even if the authors were aware that Amerindians carry a lot of ANE.

Do you understand?

Hector said...

Chinese researchers have been claiming that the genetic difference between "northern" East Asians and SE Asians was mainly due to the small extra "European" component among the former.

I think they are "mainly" wrong and their views reflect more of their political agenda but by replacing "European" with ANE, they may have touched on something. Not whole a lot but small something. small because they were actually thinking of very recent admixtures(ie <3000 years) and they are probably wrong in that regard.

It will be interesting to see how Tianyuan is related to Malta boy and other ancient samples.

Shaikorth said...

" "Onge-like" component can differentiate into various East Asian and Southeast Asian populations. Actually it is certain that they did."

It won't in that model because all those East Asian and Southeast Asian populations need ANE in excess of Onge. Onge itself differentiates into something else. Jomon is just an early split, it doesn't have lots of divergent ancestry and doesn't affect the Lazaridis model.

The Southeast Asian shift of Jomon probably means it has South Chinese minority levels of ANE i.e. less than Japanese, which makes sense assuming the later arrivals were Korean-like - in these fits it means more ANE than Japanese (1% more if EHG is used, 2% more if MA-1 is used).

Gill said...

The Tianyuan whole genome sequence should be exciting. I wonder how it will compare to other UP genomes.

Nirjhar007 said...

Waiting is retarded , contact Wang! ;) .

Grey said...

Kristiina

"Don't you find it intriguing that in Western Europe mtDNA U5, U4 and U2 are considered WHG hunter-gatherer lineages..."

I don't know if it's provable but if so i think it will turn out that the northern forager mtdna produced a higher metabolism (using more energy) and so farmer mtdna was favoured in later centuries.

"cold hands, warm heart"

Davidski said...

Modern Europeans are a poor proxy for the ANE in East Asians, so if some East Asians are showing 1-2% European admixture, then this can easily be a proxy for ~10% ANE.

Cover your eyes Hector...

Pop:ID Han_NorthChina:HGDP01287
AG3-MA1 8.69
Andamanese 1.26
Basal-rich 0.68
Oceanian 0.66
Southeast_Asian 88.71
Sub-Saharan 0
Villabruna 0.01

Hector said...

@davidski

It all depends on how competent they are, but if they could isolate East Asian part of NA genome excluding ANE one and compared it with East Asians, populations with higher ANE will come out as more distant from NA, all else being equal(assuming that there are some ANE segments among East Asians that are not present among NA. Even if there is none, there is nothing that will pull ANE heavy East Asians toward NA).

You are thinking of directly comparing NA and East Asians. That is not what the abstract says. Whether they knew East Asians had ANE or not they knew about ANE in NA and they specifically said it was excluded.

capra internetensis said...

@Davidski

The authors include Reich, Laziridis, Patterson, Skoglund, I don't think they are going to miss out on much.

@Hector

Lol Usenet Agamemnon! I remember that guy. He claimed not only that Proto-Indo-European was just Ancient Greek, but that Ancient Greek was the same as Modern Greek.

Nirjhar007 said...

Okay I have contacted the head author . Will let you guys know if he permits!.

Hector said...

@Shaikorth
We are talking about Wang's paper, not Laz' model. We don't know about Wang's concept of genetic affinity yet and how he arrived at his conclusion.

And Tinayuan paper's principal author is Yang not Wang. Fu and Paabo are among the co-authors. That does not mean they took part in the research even though they may have.

Cossue said...

@Simon

The Sueves who settled in Galicia (NW Iberia, cf. https://en.wikipedia.org/wiki/Kingdom_of_the_Suebi) circa 410 used mostly East Germanic names such as Hermeric, Reckila, Maldras, Remismundus, Ariamirus, Theudemirus, Audeca, Anila... And they arrived here in the company of Vandals (an East Germanic people) and Alans (Sarmatians/Iranians).

And Gregory of Tours, writing in the 6th century, identified the Galicians Sueves with the Alemanni: "Suevi id est Alamanni"(History of the Franks, II.2).

Gioiello said...

@ Nirjhar007

In the past I enjoyed in breaking in pieces many papers peer-reviewd, above all about the knowledge that their authors demonstrated upon the STRs values, I wouldn't break in pieces also Schrijvers' papers and put him in the Middle Easterner tumulus too, even thought he seems not having Jewish ancestry, but with Dutches we may not be always sure.
He follows the now discredited Renfrow and about the agriculturalists to Europe he seems linked to old convictions not confirmed from genetics, thinking that these presumed Hattic speaking Anatolians didn't mix with European hunter-gatherers and completely displaced them. Now we have many doubts as to all that, and above all about the Northern Anatolians, who seem more linked to Europeass than to Iranians or Natufians of 8000YBpP when they would have migrated to Europe.
About the linguistic hypothesis that Hattic was similar to Linear A of Crete and would have given some words to IE in Central Europe, not only, but it could be at the origin of the VSO construction of Celtic languages (others with no more proofs thought to an Afro-Asiatic substratum in Western Europe), all that doesn't fit with the true fact that no one speaks in Europe a Hattic language, and, if European descend all from them, we don't understand how that might be happen.
Tumulus.

ren doe said...

@Hector,

Chinese scientists usually prefer to view Chinese origins ultimately coming from SE Asia. For example, all of their studies were geared towards proving NRY haplogroup NO originated in SE Asia.

Another thing is, why don't channel yourself in a more productive way. I agree there might be at times some Eurocentrism in the mainstream anthro-genetic communities. Why don't you come to my site: http://s6.zetaboards.com/man/index/

I will make you a moderator.

Ariele Iacopo Maggi said...

Reality has a eurocentric bias.

Nirjhar007 said...

Reality must not have any bias , otherwise its not reality at all.

alobrix said...

@Simon @Cossue

There is a close genetic relationship between y chromossomes of galicians, austrians(Tyrol) and moravians, so I think an eastern element is probable in the ancient suebi and alemanni.

The ethnogenesis of suebi:

http://arheo.ffzg.unizg.hr/ska/tekstovi/alemanni_suebi.pdf

epoch2013 said...

@Gill

Highly needed too, if we want to completely understand the dynamics roughly 40.000 ya.

Matt said...

Interesting stuff, especially Tianyuan. For one will be cool if we see that slotted into Laz 2016's East Asian model to see if it fits it or defies it!

Tianyuan's population "contributed greatly", so it looks like it wasn't just an Ust Ishim / Oase1 type population with no relatedness to present day individuals. But I guess even after this paper time and sampling would be needed to know if it's like a Kostenki14 population - probably didn't contribute much to present day people, very firmly from a wider clade which did - or a real contributor.

Re; the Amur paper, of course I guess they're starting with the cold places where it's easiest to get dna (though they aren't as interesting as learning what's gone on in China to me). I wonder if high degree of continuity means 90%, 80%, or what.

If you wanted to measure whether the ancient Amur River Basin populations were specifically closer to Native Americans non-ANE part then I guess you could use a D-statistic like

D(Native American, ANE)(Ancient Amur Basin, Other East Asian)

which would be checking the relative relatedness of the non-ANE part of Native American to the whole of Ancient Amur Basin vs Other East Asian (may be confounded by different ANE level between AAB and other East Asian)

or D(Mbuti,Native American)(Ancient Amur Basin or Other East Asian,ANE)

similar principal, checking the non-ANE part of ancient Amur and other East Asian against Native American

or D(ANE,Native American)(ANE_2, Ancient Amur Basin or Other East Asian)

checking the non-ANE part of both against one another.

Kind of wonder how they have confidence in this model where Chinese vary from 90:10 to 60:40 of Yellow River Basin to Southern Native ancestry, when they only have modern samples to go on. (Hopefully more than just ADMIXTURE).

One implication of that abstract is that some Tibetans would be 95:5 of Yellow River Basin to Paleolithic Tibetan ("Tibetan Plateau populations are all admixed, deriving about 5%-10% of their ancestry from an anciently divergent population that plausibly corresponds to the Paleolithic population on the Plateau, and the remaining part" (95-90%)"from an ancient population that no longer exists in unmixed form but that likely corresponds to expanding farmers from the Middle and Upper Yellow River Basin").

So if true, in a sense maybe slightly more Yellow River Basin (early Chinese Neolithic?) than the present day North Chinese themselves (though 90% would still be a remarkably high degree of Neolithic continuity compared to Anatolia).

Chad Rohlfsen said...

To me, it is still plain to see that there are at least two branches here. Some only appear closer to ANE and EHG because, to me, ANE is an early split from East Asian that has significant West Eurasian ancestry from a group not really related more to WHG than UP Euros. Here are a few stats that tell me this has to be true.

If you remember, from the Lazaridis models, some SE Asians such as Yi and She cannot be modeled as Onge plus ANE. Not only this, these pops are significantly closer to Native Americans than the Onge. Also, contradictory to the models as Karitianas as 60-79% Onge, they actually share much more drift with ANE and even EHG than they do the Onge. So, the models are too simplified and lacking an important branch or two. If you'll notice below, it is actually those that break the model and are further from ANE than the Han and such that are closer to Native Americans.

result: Mbuti Karitiana Han Onge -0.0686 -24.744 28761 33000 616926
result: Mbuti Karitiana EHG Onge -0.0362 -8.545 29345 31551 581708
result: Mbuti Karitiana MA1 Onge -0.0417 -8.412 22006 23920 442585


result: Karitiana MA1 Han Onge 0.0680 17.407 21221 18518 442586
result: Karitiana MA1 Han_N Onge 0.0648 15.825 21197 18617 442586
result: Karitiana MA1 Atayal Onge 0.0648 14.682 21123 18551 442586
result: Karitiana MA1 Ami Onge 0.0632 14.794 21101 18592 442585
result: Karitiana MA1 Dai Onge 0.0581 14.425 20961 18657 442586
result: Karitiana MA1 She Onge 0.0701 17.247 21279 18489 442586
result: Karitiana MA1 Yi Onge 0.0633 15.867 21130 18612 442586
result: Karitiana MA1 Daur Onge 0.0664 15.853 21315 18662 442586
result: Karitiana MA1 Oroqen Onge 0.0678 16.431 21382 18669 442586
result: Karitiana MA1 Ulchi Onge 0.0743 18.539 21535 18556 442586
result: Karitiana MA1 Hezhen Onge 0.0672 16.492 21327 18641 442586
result: Karitiana MA1 Mongola Onge 0.0644 15.404 21249 18679 442586
result: Karitiana MA1 Even Onge 0.0199 5.087 20724 19914 442584
result: Karitiana MA1 Nivkh Onge 0.0803 14.424 6230 5304 86557

Barely anything here too..
result: Mbuti Onge Han Karitiana -0.0138 -5.167 28761 29566 616926

Now, back to MA1 as an early split from East Asians with significant early West Eurasian ancestry.... Here is MA1 compared to another West Eurasian of similar age and pull towards Chimp,Africans, and Neandertal.. MA1 should only be compared to age related samples due to the ancient effects.

result: Mbuti Han Vestonice16 MA1 0.0274 5.446 16304 15434 328970

result: Mbuti Onge Vestonice16 MA1 0.0270 4.969 16236 15381 328963

So, not only is MA1 close to "East Asians", but also Onge. So, I feel that a lot of this relationship has to do with the fact that ANE shares more drift with this other branch of East Asians than the Onge, creating an illusion of significant admixture. Not that there isn't some admixture in some East Asians, but it certainly isn't like the models with many pops failing. Not to mention the complete fail in f3s as Onge + MA1 or EHG.

Chad Rohlfsen said...

Another factor which can make non-Onge people look more shifted to EHG is smaller archaic ancestry than the Onge. Some of these are quite small, but with how divergent extra archaic is, you can get someone sharing more with EHG or any other Eurasian than the Onge, without actually having EHG ancestry.

result: Mbuti Neandertal Onge Ami -0.0016 -0.445 20188 20251 531192
result: Mbuti Neandertal Onge Atayal -0.0008 -0.231 20189 20223 531193
result: Mbuti Neandertal Onge Dai -0.0038 -1.162 20102 20256 531193
result: Mbuti Neandertal Onge Daur -0.0026 -0.783 20255 20361 531193
result: Mbuti Neandertal Onge Even -0.0054 -1.719 20455 20676 531191
result: Mbuti Neandertal Onge Han -0.0048 -1.501 20121 20313 531193
result: Mbuti Neandertal Onge Han_N -0.0044 -1.329 20161 20338 531193
result: Mbuti Neandertal Onge Hezhen -0.0060 -1.863 20123 20365 531193
result: Mbuti Neandertal Onge Oroqen -0.0033 -0.996 20205 20340 531193
result: Mbuti Neandertal Onge Ulchi -0.0025 -0.771 20236 20337 531193
result: Mbuti Denisovan Onge Ami -0.0019 -0.613 23370 23458 616393
result: Mbuti Denisovan Onge Atayal -0.0023 -0.736 23360 23469 616394
result: Mbuti Denisovan Onge Dai -0.0022 -0.748 23309 23410 616394
result: Mbuti Denisovan Onge Daur -0.0022 -0.752 23488 23593 616394
result: Mbuti Denisovan Onge Even -0.0038 -1.406 23779 23961 616392
result: Mbuti Denisovan Onge Han -0.0042 -1.478 23319 23515 616394
result: Mbuti Denisovan Onge Han_N -0.0044 -1.533 23355 23561 616394
result: Mbuti Denisovan Onge Hezhen -0.0049 -1.729 23368 23599 616394
result: Mbuti Denisovan Onge Oroqen -0.0044 -1.460 23403 23608 616394
result: Mbuti Denisovan Onge Ulchi -0.0025 -0.869 23459 23576 616394

Those with higher archaic ancestry also push away from Africans, like the ancient UP samples, also making them look minimally closer to Ust-Ishim, without actually having more ancestry from Ust_Ishim than anyone else. It all follows the same pattern.

result: Mbuti Neandertal Onge Ust_Ishim 0.0183 3.025 21514 20743 530072
result: Mbuti Denisovan Onge Ust_Ishim 0.0191 4.000 25067 24127 615026

Davidski said...

But you're using dead branches from Europe with elevated Neanderthal admixture to come to this conclusion.

It seems like Siberia was an expansion hub just after the Ice Age, and if it had such a profound impact on West Eurasia, then why not East Eurasia?

Why are you assuming that the Onge can't have ancestry from Upper Paleolithic Siberia?

Karl_K said...

"Why are you assuming that the Onge can't have ancestry from Upper Paleolithic Siberia?"

This is an excellent question. People assume that some particular population is unadmixed, but without really knowing. It is a hypothesis, not a fact.

Nirjhar007 said...

Dottore ,
About the case of Hattic .The language can have some connection with Afro-Asiatic:
https://www.academia.edu/11311781/HATTIC_BETWEEN_WEST_CAUCASIAN_AND_AFRO-ASIATIC

About Europe, I think it is possible that Hattic languages spoken by Early European Farmers have disappeared leaving only some words or grammatical features. Schrijver suggests that Minoan is related to Hattic, and that language disappeared, submerged by Greek. There are several languages that have disappeared in Europeanhistory, even more probably in prehistory!.

Chad Rohlfsen said...

Possible, but stats place Onge outside the relationship with Native Americans very significantly.

Chad Rohlfsen said...

The stats for Vestonice and MA1 create interesting possibilities.

MA1 and Native Americans could be part of the same mixing event, making them much closer to each other than to the Onge. With Han not much closer to ANE than Onge, nor much closer to the Onge than Native Americans, but way closer to Native Americans means we likely need another group regardless. Still, some East Asian in ANE would make more sense of those stats above.

ren doe said...

That there was a proto-Amerind/early Siberian population, related to but distinct from East Asians, and that Mal'ta was a result of gene-flow (you can either see it as early Easr Eurasian with West Eurasian admixture or West Eurasian with East Eurasian asmixture) has beeen my idea since the Mal'ta findings came out. At first it was because ANE didn't make any sense in the context of archaeology, physical anthropology, and phylogeography, and then more studies came out that made the whole model ever more complicated to swallow. I raised it several times (on Dienekes, on Anthrogenica as "ANE as an Admixture"), hoping someone with better technical knowledge would look at it more critically, but my posts been mostly ignored.

Thanks to Chad, now we have some hard technical arguments to go on. (I've noticed Davidski casually mention into this direction as well.)

Chad, what about a secondary pulse coming from southern Central Asia (NRY haplogroup P, mtDNA haplogroup X2, R1b) that has some sort of South Asian connection, since P is nested within K2b, and likely came out of South Asia. Would that explain the position of Mal'ta and its Onge connection better?

Gioiello said...

@ Nirjhar007
What you say is an unlikely possibility, because one thing is the fact that many languages disappeared for the known vicissitudes, and another is to think that all the European population derived from agriculturalists coming from Anatolia and not mixed with hunter-gatherers (what we know now isn't true) would have lost their numerous dialects after thousands of years without leaving any rest.
I know that Indo-Europeans from Eastern Europe displaced many languages in India, but not only Dravidian languages survived, but infinite others of the Sino-Tibetan, Austronesian stock survived, even isolated like Burushaski etc. Nothing in Europe linked to Hattic.
About the VSO construction, but not of all Celtic languages but only of the insular ones, whereas many stupid bloggers vomit words without meaning, and if are Afro-Asiatic nationalists think to a Berber substrate, etc., or educated linguists like Schrijver think to the improbable Hattic, the solution was found from a true scientist and great linguist, Vendryes, and through the "Wackernagel's law" already more than a century ago.

Gioiello said...

@ Nirjhar007

Nirjhar, but why, if one is searching for the language spoken from the Anatolian agriculturalists who mixed with European hunter-gatherers, without replaicing them, rather than Hattic didn't Schrijver think to Etruscan?
Of course it is unlikely that Etruscan came from Asia Minor as Herodotus said, better Dionysius of Alicarnassus who thought Etruscans having come from North with Villanovians and linked with Rhaetians and perhaps Camuns, but that they have come from Anatolia some thousands of years before it isn't unlikely.
Firstly because from a genetic point of view also Barbujani et alii, after having followed the theory of Herodotus, said that the link between Etruscans (and in part Tuscans as I am) and Anatolians there was, but old more than 5000 years, and Alfredo Trombetti thought that Etruscan was intermediate between Indo-European and Caucasian, and that may have happened in Central Europe or in the Alpine zone but also in Northern Anatolia, which had genetic links with Europe in any meaning.
If Schrijver had thought to Etruscan rather than Hattic, I'd have taken into account his theory...

Gioiello said...

The language of Lemnos before Greek conquest of 510 BC is certainly linked to Etruscan, but not so closely, thus it may well be a remnant of that migration.

Matt said...

Chad:

To me, it is still plain to see that there are at least two branches here.

So, the models are too simplified and lacking an important branch or two.

Whatever model, there, at some point, have to be two separate branches leading up to East Asians and Onge (and another to Papuan).

And no question the model will be simpler than reality.

The question though, for the branches leading to East Asians and Onge, is do they:

a) Drift together, as a single branch, after splitting off from West Eurasian (WHG), before diverging

(Onge and proto-East Asian are a clade compared to West Eurasian)

b) Split from one another before West Eurasian (WHG) breaks off from either Onge or the branch leading to East Asian

(Onge and proto-East Asian are not a clade compared to West Eurasian, and one of them is a clade with West Eurasian, and the other is basal like Ust Ishim)

If it's a), then Lazaridis 2016's model is unchanged. We're talking about undetailed internal structure to the ENA clade in Lazaridis's model, which is totally consistent with their model.

If it's b) then that's implies some pretty strong statistics of the form:

D(Mbuti,Kostenki14)(East Asian, Onge) or D(Mbuti,GoyetQ116)(East Asian, Onge)

and fails without strong statistics there. (We don't have these from what I know.)

(However, what makes ENA populations closer to MA1 than K-14 and other UP Europeans like GQ116 could definitely be admixture to MA1 from some population on the ENA branch.... and Fu's paper would allow for and mentions the possibility of such.)

Shaikorth said...

Onge do not have elevated archaic ancestry compared to other East Eurasians, regardless of whether we talk about Neanderthal or Denisovan. This was confirmed with SGDP sequences. So if some genotype sets give stats indicating otherwise we have grounds to dismiss them.

https://genetics.med.harvard.edu/reich/Reich_Lab/Welcome_files/2016_CurrentBiology_Sankararaman_DenisovaMap.pdf

Davidski said...

Stats...

Mbuti Kostenki14 Han Andamanese_Onge -0.0042 -1.3
Mbuti GoyetQ116-1 Han Andamanese_Onge -0.0066 -1.914

Mbuti Kostenki14 Dai Andamanese_Onge -0.0051 -1.566
Mbuti GoyetQ116-1 Dai Andamanese_Onge -0.0068 -1.917

Matt said...

Thanks for that.

Actually probably more importantly, if either Han or Onge were an outgroup to a Han-West_Eurasian or Onge-West_Eurasian clade (in most of their ancestry), then:

D(Mbuti,Onge)(Han,Kostenki14), D(Mbuti,Onge)(Han,GoyetQ113), D(Mbuti,Han)(Onge,Kostenki14), D(Mbuti,Han)(Onge,GoyetQ113)

would show it, in terms of the stats being close to zero.

Davidski said...

Don't have GoyetQ113 in this dataset...

Mbuti Andamanese_Onge Han Kostenki14 -0.0674 -18.45
Mbuti Andamanese_Onge Han GoyetQ116-1 -0.0552 -13.481
Mbuti Han Andamanese_Onge Kostenki14 -0.0632 -15.997
Mbuti Han Andamanese_Onge GoyetQ116-1 -0.0486 -11.856

Gioiello said...

@ Nirjhar007

"This suggests that the Proto-Celtic homeland is more plausibly situated somewhere in Northern Spain, southern France or perhaps northern Italy rather than in Central Europe north of the Alps" (Peter Schrijver, Pruners and trainers of the Celtic family tree: the rise and development of Celtic in the light of language contact" p. 200).
Schrijver is a great linguist and I don't know when he began to think that, but I am saying that (and much more) since I formulated my theory of an "Italian Refugium". It shouldn't be difficult to see as how all that coincides with what I am saying about the expansion of Bell Beakers, and that BB derive from Cardials and Cardials derive from Villabruna.

Alberto said...

Strangely (?), these stats:

Mbuti Kostenki14 Han Andamanese_Onge -0.0042 -1.3
Mbuti GoyetQ116-1 Han Andamanese_Onge -0.0066 -1.914

Seem to be just as significant this one:

Mbuti MA1 Han Onge -0.0061 -1.691 442585

So if this means that:

MA1 GoyetQ116-1 Han Andamanese_Onge ~0 ~0

I don't know if that would be compatible with the model proposed.

Nirjhar007 said...

Dottore,
About what is unlikely, I don't find a strong reason, the only non-IE language survived in Europe is Basque, apparently not related with Hattic nor with Etruscan. But we know of many other languages like Ligurian, Iberian (probably ancestor of Basque), Tartessian, the language of the Picti in Scotland, not to speak of the many IE languages disappeared like the Italic languages and Gaulish.
If the IEs arrived in the 3rd millennium BC, they had almost 5 millennia to make disappear the other languages except the isolated Basque, while non-IE Finns and Magyars arrived later. South Asian geography can probably explain the survival: Burushaski is isolated in Hunza valley of the Karakorum mountains, higher than all European mountains.
Dravidian languages belong to the Deccan, which was already heavily populated, and Austro-Asiatic was spoken in the forests by tribals.
In Europe there are no tribals remained, except maybe the Sami of Lapland, while in India there are many.
So, we cannot compare these different situations. Of course Etruscan/Tyrrhenian can be a language connected with EEF, but probably not the only one. I know the question of the age of arrival of Etruscans, derived from mtDNA.
We should check also Y DNA. Maybe, the population arrived more than 5000 years ago is connected with Rinaldone culture, which lasted for centuries and occupied a similar region as Etruscans (and possibly was IE, as some cultural aspects suggest), but the metal seekers who arrived in the late bronze age (1200 BC, the age of Sea People) were the
bringers of Etruscan language.

Gioiello said...

@ Nirjhar007

I thank you for your response, but very likely about Ligurian, Tartessian for not saying Italic languages things aren't as you say. Perhaps you know much more the situation of India.

Simon_W said...

What I find strange: Why would a male of south European origin go to live with the barbarians, and why would he decide to follow a Germanic leader, or alternatively (if he was one of the leaders) why would they accept him as a leader, in a relationship so close that they would even get buried together? I mean, this wasn't the EU back then, there was no freedom of movement and residence, and no multicultural society, but one of tribals. One possible answer might be: That southern guy might have been an Italian Longobard of mixed extraction who nonetheless considered himself Longobard.

Simon_W said...

When the Longobards had to leave their Pannonian dwelling places to the Avars in 567 AD, apparently not all of them went to Italy. There is archaeological evidence for the migration of some of them up the Danube. Their influence is discernible about up to Ulm. And some graves contained people with fully Longobard equipment. For example a woman in Dischingen, which is very close to Niederstotzingen. So maybe the majority of the people there was rather Longobard, and maybe the latter had acquired strong East European admixture on their migration through east central Europe. I think I'll have to check a craniometric cluster analysis I've once seen, I seem to remember vaguely that the Pannonian Longobards were atypical for Germanics, but I'm not sure. I'll check it in the next few days.

There's also a chronological oddity about the Longobard influence in these graves in Niederstotzingen. As mentioned, the migration of Longobards up the Danube was immediately after 567 AD. But the graves in Niederstotzingen date to the 7th century, according to Wikipedia to its latter half. So do they perhaps mean Longobard influence from Italy rather than an echo of the Pannonian Longobard immigration a century before?

Simon_W said...

The Byzantine cultural influence in Merovingian era graves of southern Germany is usually explained as stemming from Germanics who had served in Byzantine armies and thereby acquired the Byzantine equipment, so it's not completely expected to find a person of south European extraction there. Hence my suggestion with an Italian Longobard.

Matt said...

@ Davidski, thanks, doh that was the sample I was thinking of.

@ Alberto:

Yes it seems like, while, D(Mbuti,Onge/Han)(MA1,Kostenki/GoyetQ116-1) is significant - more for Kostenki, indicating shift of MA1 towards the ENA clade and presumptively admixture with that clade is some form...

The D(Mbuti,MA1)(Han,Onge) vs D(Mbuti,Kostenki/GoyetQ116-1)(Han,Onge) is not significantly different, which is a strike against MA1 related population differently exchanging genes with either Han / Onge ancestors, rather than a common ancestor to them or a early branch off from their clade.

(Or the reverse process in West Eurasia).

Lazaridis 2016 S11 do note some issues / questions with MA1 in their supplement, though I don't have time atm to pick through that and what those exactly were.

One thing I do wonder about their model is they use the outgroups:

M5: Mbuti, Ust_Ishim, Papuan, Kostenki14, Switzerland_HG

Seems like would really be good at distinguishing WHG related ancestry from ancestry that is more generically part of the broader Euro-Siberian hunter gatherer clade. I wonder if this is any good at finding "Ancient North Eurasian" clade ancestry specifically.

It would be easier to have confidence in the model if they / we check if the model can replicate with using other West Eurasian UP type populations in place of MA1, and/or use MA1 in the outgroups (Right pops) while using AG3 and EHG in the model (Left pops).

Alberto said...

@Matt

Yes, there was this higher affinity than expected between MA1 and Onge, so maybe that could explain those stats (kind of another coincidence: the exact amount of Onge admixture in MA1 to balance the ANE admixture into Han). But I guess that should be easy to test by using AG3. And in general, stats like D(AG3, GoyetQ116-1)(X, Onge), where X are Siberian and East Asian populations could show who has real ANE admixture.

What looks like a more parsimonious explanation of why Han (and East Asians in general) are closer to West Eurasians would be if Onge has some kind of Basal Eurasian admixture (which could also include archaic) that decreases affinity to West Eurasians. I think that Onge behaves a bit as intermediate between East Asians and Papuan/Australians, so probably a better model would be Onge as Atayal + Australian (kind of as this PCA shows https://3.bp.blogspot.com/-ffTGrGcXJms/V8-e3AwKSLI/AAAAAAAAE3w/0taY9NPA81gzAFdEbdZZ88HQ3AGWV9K8ACLcB/s1600/Karitiana.png).

Simon_W said...


My memory was wrong, according to a Penrose analysis by W. Bernhard 1986, the Pannonian Longobards were craniometrically not atypical for Germanics. As FrankN already mentioned in another thread, this Longobard series from Hungary and Austria was very close to a sample from Haithabu, 800 – 1100 AD.

Moreover I forgot that the Longobards in Italy were not allowed to mix with the Italian locals until after the 8th century. Hence they cannot have had southern admixture.

But I found something else that makes a lot of sense: The Italian Longobards were craniometrically rather distinct from the Pannonian Longobards and indeed from all other Germanics! In a Penrose analysis conducted by Rösing and Schwidetzky 1977 they exclusively clustered with West Slavs and early Magyar Hungary! This means at least their cranial metrics suggest they were rather eastern European. Presumably what happened was this: Right before they invaded Italy, many non-Germanic people must have joined them, e.g. Sarmatians, Gepids and perhaps Romanized Pannonians.

And if at a later date, some Italian Longobards went to southern Germany, maybe to meet their compatriots there, they may have had some Romance Italians in their company.

As for the early Goths of the Chernyakhov culture and the migration age Germanics of Czechia and Slovakia, they were similar to each other and formed an own subcluster that was distinct from the western Germanics like the Alemanni, Bavarii, German Franks etc. Unfortunately I can't tell from the analysis how close they were to Slavic people. At any rate, their facial index was rather high, i.e. narrow faced, which isn't too typical for early Slavs. The Germanic subcluster closest to the cluster of the Chernyakhov culture and the Germanics of Czechia and Slovakia were the Merovingian era Thuringi and the Varini of Mecklenburg who formed an own subcluster, no wonder since the Varini according to some sourced participated in the ethnogenesis of the Thuringi. In the end I would guess that this cluster of early Goths, Thuringi etc was less East European than the Italian Longobards. But that's just a hunch.

And at any rate the cranial series of the Alemanni from southwestern Germany, according to both cited Penrose analyses, were closest, and indeed very close, to the Bajuvari and the Franks from Germany, the next closest series being the Saxons from northern Germany. So in all likelihood not very eastern European folks!

Simon_W said...

EDIT: Gepids were Germanics of course, but presumably with strong eastern admixture.

Matt said...

Alberto: What looks like a more parsimonious explanation of why Han (and East Asians in general) are closer to West Eurasians would be if Onge has some kind of Basal Eurasian admixture (which could also include archaic) that decreases affinity to West Eurasians.

Well, I think we're talking about adding an extra layer that diverged at around the same time as Ust Ishim but after the classic Basal Eurasian, so there are a couple reasons why that wouldn't be more parsimonious than an ANE edge to East Asians:

1) Adding an extra ghost population with an admixture edge seems less parsimonious than just adding an admixture edge.

2) That wouldn't explain differential relatedness of East Asians to EHG, AG3, CHG vs other West Eurasians compared to other West Eurasians (if that is in fact that case and it appears to be)

Visually - http://i.imgur.com/pa9nRlp.png. In the model I've labelled "Alternative?" the East Asian group would all be a clade with respect to the different nodes in the West Eurasian clade.

Alberto said...

@Matt

I don't think that #1 is such a problem. In that tree you'd need to place Papuan/Australian somewhere and make them fit with either just archaic admixture as the "basal" component or with some form of Basal Eurasian. So there's no real need for adding a ghost population for Onge, only to fit Oceanians and then place Onge as a mix of East Asian and Oceanian. I think that such model would probably work with D-stats and maybe qpAdm.

#2 is more complex. I think that indeed ANE is closer to ENA than West Eurasians are, though it's not always clear (different samples behave differently, it seems). But the problem is not solved by an edge from ANE to East Asian, since that leaves Onge and Oceanians out, and if ANE is closer to ENA it's to all of them, not just to East Asians (or so I think). So you'd need an edge from ANE to the base of the ENA branch. Or, probably better, an edge from the ENA branch to ANE. That might make all the stats be in agreement.

Alternatively, ANE could split not from the West Eurasian branch, but from the branch that leads to ENA, just later than the West Eurasian branch and then get an edge from the West Eurasian branch.

I don't really know which model would work better, or if it's even possible to really find one that would fit with all the stats. Maybe Tinyuan and some other upcoming ancient DNA can help to explain how things went.

Chad Rohlfsen said...

MA1 needs admixture from a UP West Eurasian source, after diverging from other East Asians. Similar story as Native Americans. It makes sense considering it's in Siberia.

Matt said...

Alberto: But the problem is not solved by an edge from ANE to East Asian, since that leaves Onge and Oceanians out, and if ANE is closer to ENA it's to all of them, not just to East Asians (or so I think).

But the thing is that's not quite what the stats from SI 11 of Laz 2016 say.

ANE is closer to all East Eurasians than Kostenki14 and European Paleolithic as you say; however it also should be relatively closer to East Asians than any other ENA pop - Onge / Papuan (taking the SI11 of Laz 2016 as accurate).

So it would seem you need both

a) some form of preferential admixture edge between both all ENA and ANE (which, btw, the simple ADMIXTUREGRAPH models from Laz 2016 actually seem incomplete in lacking), which could be of either of the forms you describe (ANE->all East Eurasians, or proto-all East Eurasian->ANE, or in reality perhaps more realistically but less parsimoniously both directions)

and then b) between the ANE product and only the East Asian->Siberian->Native American cline.

It seems like you can't just have the former or the latter.

(Something along the lines of this: http://i.imgur.com/XkQEULy.png)

Alberto said...

Ok, so one problem here is the difference between different ANE samples in their relationship to East Asians and Onge. I think that there is some strangeness in Karelia_HG in that it either has Han-related admixture or a strangely reduced Onge affinity. So it would be necessary to define what is the "standard" ANE to be able to place it in the tree.

For example, if a stat like D(Karelia_HG,Samara_HG)(Han,Onge) turns out significant (and I think it might), that's a problem for testing the models. In those stats in Table S11.1, there is:

f4(EHG, MA1; Han, Onge) 0.00121 3.5

Which already shows the problem. So the question is if MA1 is an outlier ANE sample having higher Onge affinity than others, or if Karelia_HG is the outlier having lower Onge affinity than others, or both are true (as compared to Samara_HG and AG3 that might be more neutral).

Kostenki14 (and Natufian) seem to have a Papuan and Onge shift (probably not present in GoyetQ116-1 or Vestonice16, not in WHG).

So I guess that all samples should be run in some D(X,Y)(Han,Onge) form to check for outliers and choose the "standard" ones for making the calculations for building a tree. Otherwise we'll always get contradictory results.

Matt said...

Yes, correct that different ANE heavy samples seem show varying things.

Though note Alberto, if we take:

f4( EHG MA1 Han Onge): f4: 0.00121, Z: 3.5
f4 (EHG Kostenki14 Han Onge): f4: 0.00233, Z: 5.7

then that implies

f4 (MA1 Kostenki14 Han Onge): f4: 0.00112, Z: 2.2

that is below significance, but still in the direction of a consistent signal for preference by ANE for Han over Onge. (Is this inconsistent with any of Davidski's stats upthread?).

So we're not talking switchovers of preference for Han vs Onge here among different ANE, rather variance in the strength of the signal towards Han.

So implies some substructure among ANE in terms among themselves and with geneflow with East Asians. Possibly as my default hypothesis MA1 is an earlier branch off that has does not fit the phylogeny of the recent ANE, that contributed to others, as well (could be other ideas that explain this).

Chad Rohlfsen said...

Here's one fit for MA1 with She and Vestonice.

left pops:
MA1
Vestonice16
She

right pops:
Papuan
Mbuti
Ust_Ishim
Onge
GoyetQ116-1

best coefficients: 0.739 0.261
ssres:
0.000058790 0.000008407 -0.000284784 -0.000444535
0.221321681 0.031648343 -1.072100987 -1.673503681

Jackknife mean: 0.738426746 0.261573254
std. errors: 0.040 0.040

error covariance (* 1000000)
1576 -1576
-1576 1576


fixed pat wt dof chisq tail prob
00 0 3 0.944 0.814896 0.739 0.261
01 1 4 34.840 5.01006e-07 1.000 -0.000
10 1 4 370.980 0 0.000 1.000

Alberto said...

@Matt

There is an inconsistency there with the stats above. Probably because you actually took the stat with Kostenki and Papuan instead of Onge. The correct one with Onge is:

f4 (EHG Kostenki14 Han Onge): f4: 0.00140, Z: 3.9

Which would imply:

f4 (MA1 Kostenki14 Han Onge): f4: 0.00019, Z: 0.4

Which is insignificant. And using GoyetQ116-1 instead of Kostenki14 probably even lower. So it's very sample dependent and we'd need to know which samples are outliers to prune them out and run reliable stats.

@Chad

If you have the chance, could you run a few stats to check this?

Karelia_HG Samara_HG Han Onge
Karelia_HG Samara_HG Han Karitiana
MA1 AG3 Han Onge
MA1 AG3 Han Karitiana
AG3 Karelia_HG Han Onge
Kostenki14 GoyetQ116-1 Han Onge
Kostenki14 Vestonice16 Han Onge
Villabruna Loschbour Han Onge

I think that ANE (as in MA1, AG3) can indeed be a mix of 2 branches, though I don't think that we yet have the correct sample to test such model (She kind of works, but it cannot be the exact branch. Maybe with Tinyuan we gain some insight into all of this).

Matt said...

@ Alberto, ah true. Also, the stat f4(EHG MA1 Han Papuan) is not in their table, so perhaps that was non significant. Possibly that is another Chad could cross check with an equivalent D-stat.

@ Chad: How does this model work if you drop Onge as pleft and include it as pright instead of She, as it is more plausibly maximally unadmixed?

Shaikorth said...

Lazaridis 2016 had some tests related to this. EHG as Han + WHG looks like a bad fit, Han are too close to Onge on the ANE-Onge cline.

http://oi66.tinypic.com/280qgsh.jpg

EHG+Onge tends to provide higher P-values for fits in East Eurasia than EHG+She (tables S11.2-3).

Matt said...

@ Shaikorth, though in those models where EHG as Han + WHG failed, the left contains Bichon. Bichon would be highly sensitive to using WHG to model a population with the bulk of ancestry from another West Eurasian related clade (ANE).

It may be akin to saying that using She in the left outgroup breaks the Onge+ANE model for Han - naturally it wouldn't work under those conditions.

Shaikorth said...

They likely view it more comparable to modeling Han with She + X and Onge or Papuan in the outgroups, since in their model WHG is modeled as Bichon + EHG or something like it. If She was a leftpop in Onge+ANE model, a rightpop mix would be in the outgroups and that would more clearly skew the model.

Chad Rohlfsen said...

left pops:
MA1
Vestonice16
Onge

right pops:
Papuan
Mbuti
Ust_Ishim
GoyetQ116-1

best coefficients: 0.777 0.223
ssres:
0.000185017 0.000143288 -0.000100967
1.257359135 0.973772164 -0.686160170

Jackknife mean: 0.776462190 0.223537810
std. errors: 0.046 0.046

error covariance (* 1000000)
2133 -2133
-2133 2133


fixed pat wt dof chisq tail prob
00 0 2 0.222 0.894987 0.777 0.223
01 1 3 19.060 0.000265664 1.000 -0.000
10 1 3 291.639 0 0.000 1.000

With She in pright.

left pops:
MA1
Vestonice16
Onge

right pops:
Papuan
Mbuti
Ust_Ishim
GoyetQ116-1
She

best coefficients: 0.730 0.270
ssres:
-0.000141330 -0.000215302 -0.000858844 -0.000809391
-0.234006407 -0.356485409 -1.422028440 -1.340147108

Jackknife mean: 0.729124120 0.270875880
std. errors: 0.039 0.039

error covariance (* 1000000)
1496 -1496
-1496 1496


fixed pat wt dof chisq tail prob
00 0 3 4.381 0.223108 0.730 0.270
01 1 4 42.518 1.30273e-08 1.000 -0.000
10 1 4 364.349 0 0.000 1.000

Chad Rohlfsen said...

result: Karelia_HG Samara_HG Han Onge 0.0120 2.629 11664 11387 295577
result: Karelia_HG Samara_HG Han Karitiana -0.0047 -0.918 11359 11468 295580
result: MA1 AfontovaGora3 Han Onge -0.0138 -2.355 4527 4654 124179
result: MA1 AfontovaGora3 Han Karitiana 0.0333 4.941 4730 4426 124180
result: AfontovaGora3 Karelia_HG Han Onge -0.0141 -2.758 6232 6410 164965
result: Kostenki14 GoyetQ116-1 Han Onge 0.0018 0.407 17462 17400 401307
result: Kostenki14 Vestonice16 Han Onge -0.0001 -0.013 17841 17843 423380
result: Villabruna Loschbour Han Onge 0.0041 0.938 18764 18610 519647
result: Mbuti Neandertal Han Onge 0.0048 1.501 20313 20121 531193
result: Mbuti Neandertal Karelia_HG Samara_HG 0.0045 0.658 8597 8521 252275
result: Mbuti Neandertal MA1 AfontovaGora3 -0.0058 -0.549 3365 3404 106015
result: Mbuti Neandertal MA1 Karelia_HG -0.0158 -2.368 12595 12998 356726
result: Mbuti Neandertal MA1 Samara_HG -0.0210 -2.558 6526 6805 187549
result: Mbuti Neandertal Karelia_HG AfontovaGora3 0.0159 1.712 4824 4673 140650
result: Mbuti Neandertal Samara_HG AfontovaGora3 0.0152 1.344 2846 2761 83955
result: Mbuti Ust_Ishim Han Onge 0.0011 0.312 29801 29737 615311
result: Mbuti Ust_Ishim Karelia_HG Samara_HG 0.0083 1.227 12787 12578 294886
result: Mbuti Ust_Ishim MA1 AfontovaGora3 -0.0191 -2.308 4913 5104 123903
result: Mbuti Ust_Ishim MA1 Karelia_HG -0.0193 -3.027 18457 19185 413133
result: Mbuti Ust_Ishim MA1 Samara_HG -0.0183 -2.448 9682 10044 218810
result: Mbuti Ust_Ishim Karelia_HG AfontovaGora3 0.0045 0.614 6968 6905 164581
result: Mbuti Ust_Ishim Samara_HG AfontovaGora3 0.0041 0.457 4147 4113 98686
result: Mbuti Onge Karelia_HG Samara_HG 0.0046 0.857 12812 12696 295575
result: Mbuti Onge MA1 AfontovaGora3 -0.0223 -3.160 4915 5139 124178
result: Mbuti Onge MA1 Karelia_HG -0.0123 -2.377 18724 19191 414013
result: Mbuti Onge MA1 Samara_HG -0.0108 -1.891 9788 10002 219308
result: Mbuti Onge Karelia_HG AfontovaGora3 0.0026 0.418 6995 6959 164963
result: Mbuti Onge Samara_HG AfontovaGora3 -0.0067 -0.890 4130 4186 98933
result: Mbuti Han Karelia_HG Samara_HG -0.0063 -1.318 12730 12890 295578
result: Mbuti Han MA1 AfontovaGora3 -0.0095 -1.477 5032 5129 124180
result: Mbuti Han MA1 Karelia_HG 0.0040 0.856 19190 19036 414019
result: Mbuti Han MA1 Samara_HG -0.0059 -1.094 9917 10033 219312
result: Mbuti Han Karelia_HG AfontovaGora3 -0.0100 -1.741 6956 7097 164965
result: Mbuti Han Samara_HG AfontovaGora3 -0.0015 -0.224 4184 4197 98934

result: Onge MA1 Vestonice16 GoyetQ116-1 -0.0104 -1.347 20412 20839 377610
result: Onge MA1 Vestonice16 Kostenki14 -0.0100 -1.408 23163 23629 437971

Alberto said...

@Chad

Thanks for all those stats.

For what I see, a bit as expected, Karelia_HG has a slightly increased affinity to Han (probably easily explained by small admixture), while MA1 has increased affinity to Onge (due to some old connection?).

Without more samples it's difficult to know what's the most realistic option, though I'd say that using AfontovaGora3 as the standard ANE might be the best option for now when running D-stats for testing the model for East Asians as Onge + ANE (since AG3 is also the best sample for Native Americans and for Karelia_HG itself, but without showing any special affinity to Han or Onge). Maybe something like:

AfontovaGora3 Vestonice16 Itelmen Onge
AfontovaGora3 Vestonice16 Nganasan Onge
AfontovaGora3 Vestonice16 Ulchi Onge
AfontovaGora3 Vestonice16 Han Onge
AfontovaGora3 Vestonice16 Ami Onge
AfontovaGora3 Vestonice16 Atayal Onge

Could show how feasible the model is from D-stats point of view?

Chad Rohlfsen said...

result: AfontovaGora3 Vestonice16 Itelmen Onge 0.0552 8.605 6812 6099 147188
result: AfontovaGora3 Vestonice16 Nganasan Onge 0.0390 6.383 6682 6180 147188
result: AfontovaGora3 Vestonice16 Ulchi Onge 0.0199 3.640 6471 6219 147188
result: AfontovaGora3 Vestonice16 Han Onge 0.0102 1.933 6358 6230 147188
result: AfontovaGora3 Vestonice16 Han_N Onge 0.0042 0.759 6341 6289 147188
result: AfontovaGora3 Vestonice16 Ami Onge 0.0077 1.325 6345 6248 147188
result: AfontovaGora3 Vestonice16 Atayal Onge 0.0080 1.295 6345 6245 147188
result: AfontovaGora3 Vestonice16 Miao Onge 0.0107 1.931 6363 6228 147188
result: AfontovaGora3 Vestonice16 She Onge 0.0087 1.497 6348 6238 147188
result: AfontovaGora3 Vestonice16 Dai Onge 0.0115 2.110 6370 6225 147188
result: AfontovaGora3 Vestonice16 Thai Onge 0.0081 1.491 6377 6275 147188
result: AfontovaGora3 Vestonice16 Kinh Onge 0.0090 1.581 6353 6239 147188
result: AfontovaGora3 Vestonice16 Mamanwa1 Onge 0.0013 0.186 2244 2238 35497
result: AfontovaGora3 Vestonice16 Mamanwa2 Onge 0.0095 1.226 2275 2232 35497
result: AfontovaGora3 Vestonice16 Borneo Onge 0.0111 1.564 2263 2213 35497
result: AfontovaGora3 Vestonice16 Semende Onge 0.0131 1.823 2273 2214 35497
result: AfontovaGora3 Bichon Itelmen Onge 0.0506 8.211 7647 6910 168821
result: AfontovaGora3 Bichon Nganasan Onge 0.0335 5.683 7501 7015 168821
result: AfontovaGora3 Bichon Ulchi Onge 0.0198 3.635 7307 7023 168821
result: AfontovaGora3 Bichon Han Onge 0.0033 0.627 7135 7089 168821
result: AfontovaGora3 Bichon Han_N Onge 0.0029 0.553 7150 7109 168821
result: AfontovaGora3 Bichon Ami Onge 0.0062 1.077 7167 7079 168821
result: AfontovaGora3 Bichon Atayal Onge 0.0022 0.371 7129 7097 168821
result: AfontovaGora3 Bichon Miao Onge 0.0021 0.373 7122 7093 168821
result: AfontovaGora3 Bichon She Onge 0.0057 1.015 7162 7081 168821
result: AfontovaGora3 Bichon Dai Onge 0.0055 1.056 7139 7060 168821
result: AfontovaGora3 Bichon Thai Onge 0.0030 0.571 7153 7110 168821
result: AfontovaGora3 Bichon Kinh Onge 0.0016 0.284 7129 7107 168821
result: AfontovaGora3 Bichon Mamanwa1 Onge 0.0024 0.325 2473 2461 39503
result: AfontovaGora3 Bichon Mamanwa2 Onge 0.0078 1.041 2500 2461 39503
result: AfontovaGora3 Bichon Borneo Onge 0.0083 1.224 2478 2437 39503
result: AfontovaGora3 Bichon Semende Onge 0.0082 1.122 2481 2441 39503

Alberto said...

@Chad

Thanks!

So with AG3 the stats don't invalidate the model. But it's still all a bit too sample dependent to really know. I don't think that Atayal or Kinh can have above noise levels of ANE, and from that baseline to Han there's no more than 1-2% ANE difference.

(For the part of Onge being the right admixing population for the ENA side of East Asians I guess we all agree that it's not a good proxy for a putative proto-East Asian population without any ANE, but that's not so important for the model itself because we know we'd need an ancient sample to represent that population. I was more concerned about the possibility of significant ANE admixture in all East Asians, because for that we should have the right samples to test it. And in that I'd say it's inconclusive: possible, but not significant enough).

Matt said...

@ Chad, thanks for the models. It looks like the estimated West Eurasian+ENA combination to model MA1 is the same whether or not Onge is in the outgroup or used as the ancestor population and whether She is in the outgroup or not...

@ Davidski and others:

Off topic: The following is kind of trivial and may be a bit tl;dr but thought you might be interested:

When the Laz 2016 paper came out, I had a go at using the MDS and PCA functions in the paper on the Fst data in the paper (http://biorxiv.org/highwire/filestream/16360/field_highwire_adjunct_files/3/059311-4.xlsx), so see if they reproduce the "West Eurasian" plots.

Turns out that using either PCA or using MDS on Euclidean distance doesn't do that and produces these horseshoe shaped plots instead, dominated by pulling high Fst populations together, e.g.

PCA: http://i.imgur.com/Ph0v7t7.png
Euclidean MDS: http://i.imgur.com/oubeYuj.png

You've just got an Axis 1 splitting generally high Fst poplations on the one end (e.g Natufian, WHG) from low Fst on the other (generally moderns, particularly South Central Europe)

But I found today there is actually an MDS function that does reproduce the West Eurasia PCA from this Fst data.

If you take the Fst matrix then run it through MDS on a Rho correlation distance, it produces this:

Rho MDS: http://i.imgur.com/C8Qatv2.png (zoom on Europe: http://i.imgur.com/Xo7SBhA.png)

Kind of remarkable to me because it seems to replicate almost exactly the kind of PCA we normally see, just from the Fst alone. Though the ancients may be compressed towards moderns...

(Dimension 3 in that Rho MDS seems to produce an ancient-moderns contrast with strange patterns, particularly contrasting to Arabian populations, so that may be an indication the ancient FSTs are affected by some phenomenon splitting them away: http://i.imgur.com/41wdySP.png).

Here's a neighbour joining clustering based on the same measure (Rho distance for Fst): http://i.imgur.com/wlKsah1.png. It actually works, unlike clustering on the euclidean distance...

Like I say, kind of trivial, as it's not a method you need, but may be of interest to any others here who want to download PAST3 and stick an FST matrix in.

Another thing that might interest some, I ran this over the subset of the Fst matrix that was Europe only, and you get a figure that very much agrees with the fine detail of Europe - http://i.imgur.com/WDBgEp6.png with a nice separation between the different regions. However, this did seem more unstable over 3 dimensions, over choosing what to put in D1 vs D2 vs D3, which might be to do with the algorithm over the small set of correlations involving small differences.

Also ran the same measure on the Fst from Haak et al, a more restricted set but including Sindhi. Still some relation to reality but didn't seem to turn out quite so well: http://i.imgur.com/tmTzOIr.png. For one it seems like the lack of CHG references seems to cause it to place the WHG->EHG cline oddly, due their attraction to Eastern Europe.

Back to Lazaridis's stats, using the Correlation distance measure on MDS also seems to reproduce the West Eurasia PCA pretty well, and actually to give more distance to the ancients: http://i.imgur.com/mekuh36.png. On the other hand, less fine structuring among recent people, because the ancients have more weight... I guess that is an inevitable tradeoff.

Finally neighbour joining on the full set of Fsts from Lazaridis 2016, including the non-West Eurasian populations, using Euclidean then the correlation distance measures: http://i.imgur.com/Qv2WKQq.png

(Posted this earlier on the upthread, but either got swallowed or edited out. If edited out, won't repost again).

Matt said...

@ Chad, it looks like those models you've run don't really vary whether She or Onge is in the outgroup or in the right...

@ Davidski and others:

Off topic: The following is kind of trival and may be a bit tl;dr but thought you might be interested:

P1 of 2

When the Laz 2016 paper came out, I had a go at using the MDS and PCA functions in the paper on the Fst data in the paper (http://biorxiv.org/highwire/filestream/16360/field_highwire_adjunct_files/3/059311-4.xlsx), so see if they reproduce the "West Eurasian" plots.

Turns out that using either PCA or using MDS on Euclidean distance doesn't do that and produces these horseshoe shaped plots instead, dominated by pulling high Fst populations together, e.g.

PCA: http://i.imgur.com/Ph0v7t7.png
Euclidean MDS: http://i.imgur.com/oubeYuj.png

You've just got an Axis 1 splitting generally high Fst poplations on the one end (e.g Natufian, WHG) from low Fst on the other (generally moderns, particularly South Central Europe)

But I found today there is actually an MDS function that does reproduce the West Eurasia PCA from this Fst data.

If you take the Fst matrix then run it through MDS on a Rho correlation distance, it produces this:

Rho MDS: http://i.imgur.com/C8Qatv2.png (zoom on Europe: http://i.imgur.com/Xo7SBhA.png)

Kind of remarkable to me because it seems to replicate almost exactly the kind of PCA we normally see, just from the Fst alone. Though the ancients may be compressed towards moderns...

Matt said...

P2:

(Dimension 3 in that Rho MDS seems to produce an ancient-moderns contrast with strange patterns, particularly contrasting to Arabian populations, so that may be an indication the ancient FSTs are affected by some phenomenon splitting them away: http://i.imgur.com/41wdySP.png).

Here's a neighbour joining clustering based on the same measure (Rho distance for Fst): http://i.imgur.com/wlKsah1.png. It actually works, unlike clustering on the euclidean distance...

Like I say, kind of trivial, as it's not a method you need, but may be of interest to any others here who want to download PAST3 and stick an FST matrix in.

Another thing that might interest some, I ran this over the subset of the Fst matrix that was Europe only, and you get a figure that very much agrees with the fine detail of Europe -http://i.imgur.com/WDBgEp6.png with a nice separation between the different regions. However, this did seem more unstable over 3 dimensions, over choosing what to put in D1 vs D2 vs D3, which might be to do with the algorithm over the small set of correlations involving small differences.

Also ran the same measure on the Fst from Haak et al, a more restricted set but including Sindhi. Still some relation to reality but didn't seem to turn out quite so well: http://i.imgur.com/tmTzOIr.png. For one it seems like the lack of CHG references seems to cause it to place the WHG->EHG cline oddly, due their attraction to Eastern Europe.

Back to Lazaridis's stats, using the Correlation distance measure also seems to reproduce the West Eurasia PCA pretty well, and actually to give more distance to the ancients: http://i.imgur.com/mekuh36.png. On the other hand, less fine structuring among recent people, because the ancients have more weight... I guess that is an inevitable tradeoff.

Finally neighbour joining on the full set of Fsts from Lazaridis 2016, including the non-West Eurasian populations, using Euclidean then the correlation distance measures: http://i.imgur.com/Qv2WKQq.png

Shaikorth said...

Matt, can you make any sensible PCA or MDS with Broushaki et al. fits with ancients as sources? Those have CHG. Perhaps there's also a way to deal with high self-copying, though in Europe only Orcadians, Sardinians and French Basques exceeded 10%.

Matt said...

@ Shaikorth, hey I assume you mean the haplotype matching ancestry proportions with donors as Ust'-Ishim, KK1, Loschbour, WC1, NE1, LBK, Mota, yoruba, han, SelfCopy? I'll have a go since you asked...

It doesn't work well in straight up MDS or PCA measures with the whole world data. Populations are generally placed near to one other similar populations, whichever distance you use (Correlation / Euclidean), but there's no real long range structure captured.

Firstly, that's for the same reason that a supervised ADMIXTURE that was set to have two weakly distinct EEF components, a CHG component, a WHG component, an Iranian component, a couple of African components and Han, then a couple of catchalls. The components wouldn't really represent the world very well, and there'd be no information there that two populations, for example, who each have 50% of each of the EEF components would tend be closer than a population with 50% of Yoruba.

Secondly, because there's no way in the data for any of the algorithms to know that the Self Copy or Ust Ishim fraction represents something different for each population. You could change that by adding lots more columns, so there's a column "Karitiana Self Copy" if Karitiana self copied, "Papuan Self Copy" if Papuan self copied, etc. But that wouldn't really help unless you messed with it manually to merge those into "Amerindian Self Copy", "Oceanian Self Copy", etc, which is sort of circular.

If I restrict to the West Eurasians (those with high haplotype donation from KK1, Loschbour, WC1, NE1, LBK), retype the donors so that they have 100% donation from self (e.g. Loschbour gets 100% Loschbour, not 97% NE1 and 3% Han), then merge the two EEF donors, then run it in Correlation distance MDS, then you can get something recognisable: http://i.imgur.com/fqTqRCK.png. But it doesn't look great (for instance the distance between KK1 and WC1 is very stretched by KK1's high donation to Caucasus and to a lesser extent Northern/Eastern Europe, while Loschbour seems far too close).

If I skip the step of merge the EEFs, then it's not too different: http://imgur.com/Zl0UTdZ but with an odd placement of NE1 in a "steppe" position due to NE1 acting as a donor to South-Central Eastern Europeans (Hungarians, Croatians, Bulgarians, Romanians) and Turkey, who are pulled off clines towards it, while LBK acts more as a donor to Western Europe. There is presumably a bit of detectable haplotype sharing with NE1, a Hungarian EEF vs the LBK_EN from Germany, still to this day (a Cardial Culture farmer might be an even better donor to many Western Europeans).

Rotated versions of above: http://i.imgur.com/c5cjjv2.png and http://i.imgur.com/SzKYWDz.png.

Correlation based PCA works a bit as well: http://i.imgur.com/MQW1lAz.png (NE1 and LBK columns separate), http://i.imgur.com/MQW1lAz.png (NE1 and LBK columns merged) with particularly off placement for higher self copy Sardinians, Orcadians, Basques who are pulled towards other high self copiers and generally more influence from the non-West Eurasian donors and Self Copy.

(Probably more detail than you were looking for!).

Shaikorth said...

You could try Eurasians + Africans with selfcopy <10%.

The idea I get from these stats is that Loschbour is acting as a surrogate for Villabruna-EHG, if there was a proper EHG sample it'd be more distant. Near Eastern in North Europe mostly goes to Kotias.

India and southern Pakistan getting just Ust-Ishim and Iran_N is sort of interesting because at least Andronovo/Sintashta type steppe ancestry should include Loschbour using these sources, and indeed Tajiks & surroundings have that...