search this blog

Monday, November 7, 2016

Cryptic post-OOA African ancestry in Eurasians (?)


The Max Planck Institute for the Science of Human History is holding the Human Dispersals in the Late Pleistocene - Interdisciplinary Approaches Towards Understanding the Worldwide Expansion of Homo sapiens conference this week. The conference web page and abstract book are here.

The abstract below from Stephan Schiffels is very interesting. It reminds me of some of the discussions that we've had in the comments about potential African ancestry in some Eurasians that might be hiding due to a lack of relevant ancient African samples (for instance, see here).

When and how modern humans left the African continent is still a debated question. Recently, three projects have analysed new genetic data from modern populations in Papua New Guinea and Australia, which has provided new insights on this topic. I will present analyses from one of these publications (Malaspinas et al. 2016), and compare results with findings from the two other projects (Mallick et al. 2016, Pagani et al. 2016). Here, we used MSMC2, a novel computational framework to analyse the distribution of times to the most recent common ancestor along multiple sequences. We find that all non-African populations that we analysed, including Australians, experienced a very similar population bottleneck in the past, consistent with only one out-of-Africa migration for all extant non-African populations. At the same time, we find evidence that some African populations are more distantly related to Australians than to Eurasian populations, and we show that this result is robust to haplotype phasing errors and archaic introgression. We interpret our result as evidence for gene flow between some Africans and Eurasians after the initial split, which is also consistent with results from other population genetic methods. Our analysis suggests that in order to understand human dispersal out of Africa, we need to better understand ancient population substructure within Africa, which is an important direction for future research.

Stephan Schiffels, Analysing Australian genomes to learn about early modern human dispersal out of Africa, Human Dispersals in the Late Pleistocene - Interdisciplinary Approaches Towards Understanding the Worldwide Expansion of Homo sapiens 2016 conference abstract

See also...

Two pronged AMH colonization of Eurasia. Or not

58 comments:

Karl_K said...

Isn't this also just as consistent with an ancient and widespread back-migration from Eurasia into Africa?

Karl_K said...

If it was a back-migration of strictly Basal Eurasians even before the split of the Basal Eurasian populations that went into the various farmer populations, it would not contain Neanderthal or Denisovan admixture.

Karl_K said...

And without ancient genomes, we don't actually know that East Asians do not also have Basal Eurasian ancestry, only that they would have to have much less than the more western populations.

capra internetensis said...

Not much point speculating until we know *which* Africans and *which* Eurasians.

astenb said...

What you said ^ Which Africans.

Samuel Andrews said...

@Karl_K,
"And without ancient genomes, we don't actually know that East Asians do not also have Basal Eurasian ancestry, only that they would have to have much less than the more western populations."

True and we shouldn't restrict our imagination of Eurasian ancestry to Crown Eurasian and Basal Eurasian. Why couldn't there have been 10 different basal Eurasian populations?

bellbeakerblogger said...

I don't know if you saw this one yet...
https://www.academia.edu/29533167/Investigating_kinship_of_Neolithic_post-LBK_human_remains_from_Krusza_Zamkowa_Poland_using_ancient_DNA?auto=download&campaign=weekly_digest

Anon said...

I suppose this is just West Eurasians not East Eurasians too, right?

If that's the case, it was foreseen a long time ago by Cavalli-Sforza, who also had claimed that a 66% East Eurasian 33% SSA fits West Eurasians quite well (in crude statistical terms, not in any real sense obviously).

Also the fact that West Eurasian Fst distance is significantly smaller than all East Eurasians.


Any idea when the full paper is coming out?

Slumbery said...

"We find that all non-African populations that we analysed, including Australians, experienced a very similar population bottleneck in the past, consistent with only one out-of-Africa migration for all extant non-African populations."

Just because there is a "very similar" bottleneck that does not automatically discredit a two-wave model.

Matt said...

That inference based on Fst that was kind of semi-sane back in Cavalli-Szorfa's prime but is pretty hard to look at with present day dna and interpret that way.

When Fst from Yoruba goes:

Karitiana: 0.280, Onge: 0.229, Eskimo: 0.207, Scandinavian Hunter Gatherers: 0.203, WHG:0.201, Ami: 0.198, Natufian: 0.186, EHG: 0.183, Han: 0.178, CHG:0.169, European Early Neolithic: 0.160, Steppe_EMBA: 0.157

It's hard to make this inference that closeness from Fst fits plausibly with any African admixture. Would the Han be more African admixed compared with Natufians? Or with Eskimo and Native Americans and Onge and Ami?

You could say something along the lines of "Well, the pattern for the ancient Europeans is due to drift, for recent East Asians and other ENA, lack of admixture, and also drift on top".

But it's simpler at the moment to just interpret it all in different drift and admixture strictly between Eurasian populations, particularly given the formal stats.

Anon said...

@Matt
Well,it is my opinion that Amerindians are so because of their extreme and recent bottleneck;The two Americas are amazingly homogeneous genetically (while also being amazingly heterogeneous in regards to languages though), modulo the newer Siberian admixture in North America.
While Oceanian and other anciently settled islands like the islands of the bay of Bengal are similarly distant due to both small founding population and total isolation.
Above argument also works on other isolated pops like Kalash who.

This also explains why WHG+SHG have such a high Fst with SSA, and all other populations too relatively speaking. This is the result of the massive hits their effective population size took from the ice age.

But I don't see how that would make all modern West Eurasians closer to SSA than any East Eurasian population is, unless you're suggesting they underwent a bottleneck specific to themselves before they diversified and split from each other, and after we split from them circa 45k ya.

Matt said...

You could say that if you wanted to and it could be plausible. It's just not very parsimonious to say that all the differences in fst from Africans are explained by different levels and drift and bottlenecks except the difference in fst from Africans between West and East Eurasians, which is due to admixture.

It's much more consistent and parsimonious to go with the idea that all differences in fst from Africans by all Eurasians are due to differences of degree in bottlenecks and diversity between their ancestral populations, and also the West Eurasian - East Eurasian difference in fst from Africans is simply because the combined collection of West Eurasian ancestors were less bottlenecked.

Modern West Eurasians are less bottlenecked than modern East Eurasians, because they are a composite of populations who were ultimately less far out of a serial bottleneck from Africa.

Though the eurogenes comments section *is* kind of ground zero for people sticking to their theories even when they aren't the most Ockham compliant, and people do tend to be psychologically committed to their theories. So this is more for the benefit of anyone else reading our exchange than because I think I'll change your mind. ;)

Davidski said...

@Karl_K

Isn't this also just as consistent with an ancient and widespread back-migration from Eurasia into Africa?

I guess it might be and maybe I'm confused?

But if so, why didn't Schiffels say it explicitly, as opposed to talking about gene flow between some Africans and Eurasians, and why are some Africans more distant from Australians than from other Eurasians?

To be honest, I don't have an opinion at this stage, and would not be unhappy to see it go either way.

@bellbeakerblogger

Thanks. I did see that paper, but it's kind of old school. Hopefully they get more data.

Anon said...

@Matt
I don't get how you inferred that I was psychologically invested in "my" hypothesis. I am definitely not, I just offered a scenario explaining the discrepancy you commented about.

There is one thing I am quite certain of however.
And that's that there is additional affinity between West Eurasians and SSA that's not present between East Eurasians and SSA. And it's independent of obvious recent admixture, like the slight SSA admixture all around the Mediterranean, the Muslim slave trade, the late Neolithic hybridization in the Horn of Africa, etc.
A few of my reasons are
1. Fst stuff that I've already mentioned, although it's not really important on its own for reasons you and I pointed out.
2. Sharing of E-P2
3.On global PCAs, the first principal component differentiates Eurasians and SSA, and West Eurasians virtually always are a little bit closer to SSA than any East Eurasian pop (eg http://themonkeycage.org/wp-content/uploads/2010/07/genes-thumb.jpg )
3. On TreeMix, West Eurasians very often branch off before East Eurasians, less often, West and East Eurasians share the same branch off node, but w it's never the opposite of East Eurasians branching off before (eg http://themonkeycage.org/wp-content/uploads/2010/07/genes-thumb.jpg http://2.bp.blogspot.com/-VpFMFJly96s/T1zczlkhwFI/AAAAAAAAElQ/_AA49D09W-4/s1600/tree.png , second is with admixture components, so virtual pops). Same with other drift-based neighbour joining methods. (eg https://upload.wikimedia.org/wikipedia/commons/7/77/Neighbor-joining_Tree-2.png)
4. Many times on TreeMix, there is a migration from a Sardinian-like vertex to all Sub-Saharans. When the migration is larger, the "unmixed" inferred SSA population is naturally more drifted form everyone else. Now that we have Levantine aDNA, TreeMix might prefer them to Sardinians, in the same way that the migration node from West Eurasians to Amerindians (ANE) changed when we got aDNA from the steppe. Of course this migration vector doesn't specify neither directionality nor magnitude, but it IMO is very strongly indicative of an additional affinity, and the fact that it's consistently and specifically Sardinian-like (ANE free Med farmer relic) instead of any other West Eurasian pop is all the more indicative of that IMO.
5. Higher IBS sharing between WE-SSA than EE-SSA.
6. The fact that the E3b-rich Natufians appear to be intermediate between Saudis/Arabians and Berbers on global or West Eurasian PCA (albeit with a smaller orthogonal vector of variation that's absent from modern populations). And the fact that Natufian kits on gedmatch consistently come out intermediate between Berbers and Arabians, albeit with varying degrees. In both cases having some additional affinity to SSA that remains uncaptured by formal stats. (the two above facts surprised me by the way, since I thought Berbers were a post-Neolithic "hybrid" of different Neolithic populations, but reality has to be more subtle given these Natufians)
7. This one isn't really meaningful much, but there is also the fact that the first split on ADMIXTURE runs tends to be between East Eurasians and the rest, more often than Eurasians and SSA. Again this one isn't an argument, but still somewhat interesting.

I can't give any detailed account of what happened obviously. But to me, the weight of evidence that there is something additional between West Eurasians and Sub Saharans is overwhelming, and I can't wait for aDNA to elucidate the matter.
I don't claim anything on the complexity or the (bi)directionality of the relation, I just claim its existence. If you deny its existence, I would be interested in your rationale for that because my position isn't set in stone.

Rob said...

@ David Ski

"But if so, why didn't Schiffels say it explicitly, as opposed to talking about gene flow between some Africans and Eurasians, and why are some Africans more distant from Australians than from other Eurasians?"

Is this directed at the pre/ post-Toba dispersal question ?

Matt said...

With your points 3 and 3b, and 5 and 7 (PCA, Treemix branching order, IBS similarity, split order in ADMIXTURE), though, all this is pretty compatible with the Europeans just being from an earlier, lesser bottleneck though (or mixing from different ancient West Eurasian populations that by combining average out closer to less drift and closer to African allele frequencies).
If you go through a lesser bottleneck, then allele frequencies will more resemble Africans, thus closer on PCA, tree models, IBS, ADMIXTURE will differentiate an East Asian component more strongly (it has more varying allele frequencies), etc. It's again not crazy to say that pattern could come about through admixture, just it's also very compatible with the differing bottlenecks model (which is generally simpler).
Treemix is tough for me to interpret since it often produces varying results.
I think the uniparental you point out is more directly suggestive, just very hard to interpret as direct evidence of exchange between Africa and West Eurasia to the exclusion of ENA populations, as uniparental haplogroups can in theory become quite heavily detached from any significant exchange of ancestry over time.

Apologies if I've accused you of being overly attached to a hypothesis that though not crazy, isn't necessarily the simplest, just so many people here often are ;).

I don't think I can prove any flow between some ancient African group and West Eurasians (to the exclusion of other out of Africa) wrong, though I would say that by formal stats if you had admixture between West Eurasians and Africans, they would have to be almost unrelated to present day Africans. That's because if you go with stats like D(Chimp, Yoruba/Mota/Mbuti, West Eurasian, East Asian), which in simplified terms do something like measure affinity to the African population net of genetic drift, using drift from the Chimp outgroup as a control, they basically evaluate to 0 - no differential relatedness net of drift. Except where we have known historical flow from West Eurasia to Africa (recent East Africans, Khoi San, etc.).

So the African group would have to be almost unrelated to any known present day African group or the one adna African group we have (Mota). As Davidski says though, it is possible that "African ancestry in some Eurasians that might be hiding due to a lack of relevant ancient African samples" and there might be some very divergent ancient African group that could've contributed to West Eurasians. Just isn't a very strong hypothesis to me on the current evidence.

Anon said...

I forgot to mention another one.
9. When you project Sub-Saharan on a vector that differentiates West Eurasians from East Eurasians, they have a very clear preference for WE.

Ryan said...

The modern human admixture into Altai Neanderthals was suggested to be basal to San/Yoruba - ie basal to all living humans. Wouldn't a bit of that ancestry contributing to Papuans explain this as well?

Anon said...

@Matt
The Fst argument of bottlenecked populations doesn't work as well for a global PCA, even less for TreeMix and other neighbour joining methods. You can see for example that the bottleneck that skews Fst for Amerindians isn't in effect on a global PCA, or TreeMix where the vertex they branch out of is somewhat intermediate between West Eurasians and East Asians (same with the referenced neighbour joining model of Cavalli-Sforza). On global PCA, those effects are too minutes to be captured by the first few PCs, and on TreeMix, that population specific drift will manifest by a longer edge (eg like for the Kalash in comparison to other south central Asians), but it won't move the vertex that the populations diverge from.

>as uniparental haplogroups can in theory become quite heavily detached from any significant exchange of ancestry over time.
Indeed. But this isn't a problem for me because I doubt the admixture in either is too significant, a relatively slight one would suffice to explain it.

>I don't think I can prove any flow between some ancient African group and West Eurasians (to the exclusion of other out of Africa) wrong, though I would say that by formal stats if you had admixture between West Eurasians and Africans, they would have to be almost unrelated to present day Africans. That's because if you go with stats like D(Chimp, Yoruba/Mota/Mbuti, West Eurasian, East Asian) [...]
Yeah I agree, I can't prove it either, I just have a big hunch for it. Like I said, I'm aware that the D stats for Natufians show that they have no additional affinity to SSA whatsoever. Other methods all disagree with that however, and I don't see how to reconcile those results. If you deny SSA admixture for Natufians, SSA admixture in Berbers becomes all the more mysterious since it seems to have the same character.
Natufians are the reason why I now partially dismiss old D stats showing no complex population history between modern WE and SSA, because I'm convinced there is more to it than that with Natufians. It was enough to convince me otherwise in the past however.

>Apologies if I've accused you of being overly attached to a hypothesis that though not crazy, isn't necessarily the simplest, just so many people here often are ;).
No offence taken.
But in response to "not necessarily the simplest": I don't see how that should influence anything. Occam's razor would dictate that human population history fit a strict phylogenetic tree until the end of the ice age.
Look at these complex models from Reich earlier this year though :
http://www.nature.com/nature/journal/v534/n7606/images/nature17993-sf4.jpg
http://www.nature.com/nature/journal/v536/n7617/images/nature19310-f4.jpg

Our history is anything but simple. Simple models like ANE+WHG+EEF can elegantly approximate the truth, but that's all they are, a mere approximation of an incredibly complex reality.
Appealing to simplicity isn't a refutation of anything.

Davidski said...

@Rob

I'm just wondering whether Schiffel's method has captured unambiguous evidence of bidirectional gene flow between some Africans and almost all Eurasians except Oceanians.

That's what it seemed like to me when I first read the abstract. Now I'm not so sure after re-reading it.

However, the question remains, if Schiffel's method can accurately get around the fact that Oceanians have inflated levels of admixture from archaic humans, then why are some Africans closer to most or all Eurasians relative to Oceanians?

Or is he talking specifically about Australians, and not Oceanians in general? I guess we'll have to wait for the details to see what he's come up with.

Karl_K said...

@Davidski

"But if so, why didn't Schiffels say it explicitly, as opposed to talking about gene flow between some Africans and Eurasians, and why are some Africans more distant from Australians than from other Eurasians?"

He is being vague because that is all that he believes the data says.

There was some kind of gene flow between SOME Eurasians and Sub-Saharan Africans after the isolation of the populations in Papua New Guinea and Australia.

The usage of the word SOME is ambiguous, so we don't know if it is SOME Eurasians, SOME Africans, or SOME of each.

I think that the key to discovering this was the new genomes from Papua New Guinea and Australia suggests that the gene frow was equally related to all other Eurasians.

There are a couple of ways of interpreting that information.

1. Papua New Guinea and Australia had additional admixture from a 'modern human' group just basal to all Africans and Eurasians.

2. TThr single Out-Of-Africa population split almost immediately into two groups. One went on to Papua New Guinea and Australia. The other remained as one population until splitting to become Crown Eurasians, with a significant migraton of this group into Sub-Saharan Africa.

3. There was a Basal Eurasian group that split off on it's own just after the OOA bottleneck (similar yo the timing of the farmer associated Basal Eurasian group). This grop could have expanded into both Sub-Saharan Africa and extensively into Eurasia, but only after the isolation of Papua New Guinea and Australia populations.

Rob said...

Karl/ Dave

Sorry, basic question, but do East Asians & Amerindians have "Basal" ?

Karl_K said...

@Rob

They do not have "Basal" as compared to Ust-Ishim. So if Ust-Ishim did have admixture from a "Basal" population, after Australians became isolated, then we would have no way of knowing about it. That would require a genome from an older population that Ust-Ishim.

Gioiello said...

@ Anon

"Our history is anything but simple. Simple models like ANE+WHG+EEF can elegantly approximate the truth, but that's all they are, a mere approximation of an incredibly complex reality".

German Dziebel said...

@Anon

"Amerindians are so because of their extreme and recent bottleneck;The two Americas are amazingly homogeneous genetically (while also being amazingly heterogeneous in regards to languages though)."

"the bottleneck that skews Fst for Amerindians isn't in effect on a global PCA, or TreeMix where the vertex they branch out of is somewhat intermediate between West Eurasians and East Asians (same with the referenced neighbour joining model of Cavalli-Sforza). On global PCA, those effects are too minutes to be captured by the first few PCs, and on TreeMix, that population specific drift will manifest by a longer edge (eg like for the Kalash in comparison to other south central Asians), but it won't move the vertex that the populations diverge from."

Nice observations...But Amerindians can't be product of "extreme and recent" bottleneck as their "component" is ascertained via ancient DNA at 24,000 YBP already. It means the pattern you're observing comes from a different process. Amerindians are homogenous because they are isolated from the rest but also because they have been admixing with each other over a very long period of time. And they appear intermediary between East and West Eurasians not because they are product of recent admixture between two fully formed regional Old World populations but because they predate the split between those but they kept admixing internally since West and East Eurasians have gone their separate ways in the Old World. Hence, the high Fst values that Amerindians show.

@Karl _ K

"1. Papua New Guinea and Australia had additional admixture from a 'modern human' group just basal to all Africans and Eurasians. 2. TThr single Out-Of-Africa population split almost immediately into two groups. One went on to Papua New Guinea and Australia. The other remained as one population until splitting to become Crown Eurasians, with a significant migraton of this group into Sub-Saharan Africa."

Recall that Amerindians are also not uniform in their relationship to PNG and Australia. Some Amerindian populations are closer to them than others.

Karl_K said...

Obviously Dziwbel has hit the nail on the head.

"Recall that Amerindians are also not uniform in their relationship to PNG and Australia. Some Amerindian populations are closer to them than others."

There must have been a popuation just basal to all other modern humans, that lived in the Amazon (where fossils can't form and they didn't use stone tools).

From there, they went to Eurasia and Africa and Australia ~200,000 years ago and then the Eurasians all died out, but then a small group of Africans back migrated to Eurasia. When some of them got to South America and Australia, they admixed with that original population.

Case closed.

FrankN said...

@Matt, Anon e.a.: The Conference Book contains further abstracts, which provide a wider and quite interesting perspective on possible African-West Eurasian genetic interaction (emphasis is mine):

Eleanor Scerri, University of Oxford, UK:
Inferring Late Pleistocene hominin demography in the Green Sahara and surrounding regions:
Mounting evidence, including some of the oldest examples of technological regionalization and ‘symbolic’ material culture with early Homo sapiens fossils, is highlighting the North African Middle Stone Age (NAMSA, ~250-25 thousand years ago [ka]) as a key area for human origins research. The geographic situation of North Africa and an increased understanding of the wet/dry climatic pulses of the Sahara Desert also suggest that North Africa played a strategic role in continental-scale evolutionary processes by modulating human dispersal and demographic structure. (..). Determining how well this line of evidence matches the biological records from fossil and genetic data, and to what degree it emphasizes different aspects of the evolutionary process represents a promising new avenue of research.

Shannon McPherron, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany
The age of the Homo sapiens fossils from Jebel Irhoud (Morocco) and the origins of the Middle Stone Age:
Jebel Irhoud (Irhoud), Morocco, contains stratified archaeological deposits best known for yielding abundant late Pleistocene hominin remains associated with a Levallois based Middle Stone Age stone tool assemblage. Taxonomically these fossils have been generally considered primitive forms of Homo sapiens (..). New excavations (..) resulted in the discovery of additional hominin fossils associated with Middle Stone Age assemblages (..). In addition to new TL dates for the sequence, based in part on new dosimetric data, we also recalculated the uranium series/electron spin resonance age for a tooth from the Irhoud 3 hominin mandible. (..) The fossils and stone tools are substantially older than previous age estimates. Here and in a companion presentation, these results and their implications for the emergence of our species and of the Middle Stone Age are discussed.

Johannes Krause, Max Planck Institute for the Science of Human History, Jena, Germany
Genomic history of Upper PaleolithicEuropeans
Here we have analysed genome-wide data from 51 modern humans remains that span around 40,000 years of Eurasian prehistory. (..) Whereas the earliest modern humans in Europe did not contribute substantially to present-day Europeans, all individuals between ~37,000 and ~14,000 years ago descended from a single founder population which forms part of the ancestry of present-day Europeans. A ~35,000-year-old individual from northwest Europe represents an early branch of this founder population which was then displaced across a broad region, before reappearing in southwest Europe during the last ice age ~19,000 years ago. During the major warming period after ~14,000 years ago, a new genetic component related to present-day Near Easterners appears in Europe.

Finally, the map on the Abstract Book’s cover is worth a look. Note the “red belt” stretching from Nigeria to Algeria/ Tunisia and beyond…

astenb said...

@FrankN - Which conference book is this?
Link?

FrankN said...

@BBB, Dave.:, re
https://www.academia.edu/29533167/Investigating_kinship_of_Neolithic_post-LBK_human_remains_from_Krusza_Zamkowa_Poland_using_ancient_DNA?auto=download&campaign=weekly_digest

Yep, old school, only mtDNA, at first sight. But on a closer look, the mtDNA they have found is quite interesting:

K1a4, K1c: K1(a) is a typical EEF hg and as such unsurprising. However, while found in Mesolithic Greece, K1c so far hadn’t been known from Early Neolithic contexts, so some (e.g. Maciamo) believed it to be associated with EHG. Apparently, its presence in Central Europe precedes Steppe_LNBA influence, and later finds (Hung_EBA/Mako, BB_Germ RISE563) may signal EEF continuity.

H3d: H3 is a typical Iberia_EN / (Epi-)Cardial hg that hasn’t yet been reported from LBK, but forms part of the “Mediterranean element” within the Paris Basin RRBP (4.900-4.500 BC) as analysed in Rivollat 2015.
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0125521#sec001
How did it make its way into Poland? Did post-RRBP Michelsberg reach out further east than assumed so far. Or did H3 already arrive with Rössen that also had its origins quite west of the Rhine? Whatever the case – Mediterranean (Cardial) mtDNA in Polish Lengyel is quite a surprise!

U5b2a1a has previously only been reported from Benzigerode (Bernburg Culture), but U5b2 of course is typical WHG (e.g. U5b2a2 from Blätterhöhle, 9 mBC). Lorkiewicz 2015 already reported 7 mtDNA results for Polish Lengyel (Brześć Kujawski), including one U5a (WHG/ SHG/ EHG). So, we now have 2/11 = 18% UHG mtDNA within Polish Lengyel, which is quite above the level we know from LBK. 11 samples, of course, aren’t enough for drawing solid conclusions. Nevertheless, it appears that the resurgence of HG (mt)DNA in the CE plain may already have commenced during the late EN, earlier than assumed so far.

@Astenb: Link in Dave's opener!

Karl_K said...

So it could actually be that there were multiple OOA events, but only one was bigly successful, and that one reached Sahul without admixture with any previous OOA 'modern humans', while the Eurasians absorbed the smaller earlier OOA populations that were more related to modern SSA populations.

German Dziebel said...

@Karl_K

"From there, they went to Eurasia and Africa and Australia ~200,000 years ago and then the Eurasians all died out, but then a small group of Africans back migrated to Eurasia. When some of them got to South America and Australia, they admixed with that original population.

Case closed."

You either publish slews of boring scenarios that put me to sleep, or wild fantasies that have no basis in reality. Science, however, is in the middle. Try again.

Karl_K said...

Sorry. Case Closed.

capra internetensis said...

@Karl_K

If Eurasians mixed with the Early OoA remnants and Sahulians did not, then I wouldn't really expect the level of admixture to be consistent across Eurasians.

More broadly the major element of Sahulian ancestry doesn't form a branch against Eurasians in general i.e. they are East Eurasians carrying sub-branches of the usual haplogroups. It doesn't look like they were isolated from the main stream of Eurasians, which suggests that they got their weirdness by mixing with a weird minority element, not by missing out on something all other Eurasians had.

German Dziebel said...

@Karl_K

"Sorry. Case Closed."

I thought you just opened it. I agree keep it closed.

German Dziebel said...

@Capra

"It doesn't look like they were isolated from the main stream of Eurasians, which suggests that they got their weirdness by mixing with a weird minority element, not by missing out on something all other Eurasians had."

Wouldn't this work for Africans as well? It's just their "weird" element is much larger than the "weird" element in Papuans. And different, obviously. The more archaics along the way, the "weirder" you get.

capra internetensis said...

@German

The Out-of-Africa-related element in Africans is more in the nature of a sister branch - Y hg E and mt hgs L3a-x - where Sahulian is a daughter - Y hgs C1b2 and MS, primary or shallowly-nested branches of mt hgs M and N.

But this is indeed not the source of African "weirdness" - the cause is divergent human ancestry. It's just that the vast majority of this ancestry is no more "archaic" than a Kalahari Bushman - or less so. This can be distinguished from a small amount of really archaic ancestry by haplotype analysis - and here autosomal haplotypes corroborate what we already know from uniparental ones.

My guess is that the Out-of-Africa-related element spread across the greater Sahara region during a pluvial period, probably around the time of the Out-of-Africa (or Back-to-Africa) migration, and then contributed to North Africans (-> Eurasians) and to Sahelo-Sudanian Africans (-> equatorial and southern Africans). But the strange Natufian results cast doubt on this.

German Dziebel said...

@Capra

Remind me please what "strange Natufian results" you have in mind?

German Dziebel said...

@Capra

"But this is indeed not the source of African "weirdness" - the cause is divergent human ancestry. It's just that the vast majority of this ancestry is no more "archaic" than a Kalahari Bushman - or less so. This can be distinguished from a small amount of really archaic ancestry by haplotype analysis - and here autosomal haplotypes corroborate what we already know from uniparental ones."

By looking at the dispersion of uniparental haplotypes during the post-1492 migration of African slaves to the New World we can set up an expectation for an out-of-Africa migration 50-60,000 years ago. And what we see in post-1492 times is that all of those "pre-L3" (mtDNA) and pre-CT (Y-DNA) haplotypes are easily found in the black migrants to the New World. We don't see the same picture along the putative migration routes out of Africa that were followed by modern humans 50-60,000 years ago. None of those "pre-L3" (mtDNA) and "pre-CT" (Y-DNA) haplotypes are found in East Asia, West Eurasia or the Sahul. Hence, there was no out-of-Africa migration 50-60,000 years ago. African divergence must be explained by archaic admixture in Africa.

capra internetensis said...

Whoops, sorry I fed the troll.

Thanks for posting those North African abstracts, FrankN.

German Dziebel said...

@Capra

"I fed the troll."

Bon appetit!

Karl_K said...

@capra

I think we will have to wait and see.

It looks like something interesting may have happened to produce a signal, but there are too many unknowns right now to determine exactly what happened, or even discuss it without loony theories overtaking the conversation.

terryt said...

@ Capra:

"which suggests that they got their weirdness by mixing with a weird minority element, not by missing out on something all other Eurasians had".

That is a scenario I have suspected for about 10 years. The 'Denisova' element could easily be explained by such a situation, but there could be some other explanation as well. I would really like to know if thete was any real difference in the relative proportions of Denisova in New Guinea and Australia.

Ryan said...

Didn't Reich have a paper that showed Papuans deriving their Denisovan component from an extinct group with far higher Denisovan ancestry? Something like 25-50% Denisovan 50-75% modern human? That would line up with these results wouldn't it.

The wierd Mbuti or Biaka to Papua migration edges that David and Razib have crop up in many of their treemix runs also seem to support this paper.

capra internetensis said...

@terryt

"The New Guineans and Australians are estimated to have indistinguishable proportions of Denisovan ancestry (within the statistical error)...."

@Ryan

Reich et al 2011 modelled the common ancestor of Papuans and Australians as a ~50%-50% mix of an Onge-like population and older Denisovan-admixed population. The Denisovan-admixed population had ~7% Denisovan and ~93% a modern human population that split off before Chinese, Onge, and Malayasian Negritos but after Yoruba, i.e. some basal Eurasian or East Eurasian group.

Pretty old paper though, I don't know how well it holds up.

Ryan said...

@capra - Thanks. I'd say the paper is holding up well so far lol.

epoch2013 said...

It seems pretty important to known what African populations show affinity to exactly what West-Eurasians. If it is all but Mbuti, I'd say the Papua may have an early wave admixture, like recently suggested. The Kuhlwilm paper stated that the human admixture in Altai neanderthals was basal to everyone. If that admixted in humans east of the Wallace line it could explain these results. In that case Mbuti, who split off the earliest, would share a tad ancestral with the earliest OoA bunch that others don't.

terryt said...

@ Capra:

I presume you meant to say "New Guineans and Australians are estimated to have proportions of Denisovan ancestry indistinguishable from each other". Thanks for that information.

@ Ryan:

"The Kuhlwilm paper stated that the human admixture in Altai neanderthals was basal to everyone. If that admixted in humans east of the Wallace line it could explain these results"

I have long suspected that the population that gave rise in large part to Australian/New Guinean had moved east north of the Tibetan Plateau. In other words it carried the Denisovan from the Altai region.

epoch2013 said...

O I now read that they mention Eurasian rather than West-Eurasians. That makes it more odd, as it suggests evenly distributed admixture. But when and where would that have occured?

epoch2013 said...

@Anon

My ideas on this: The slight affinity between West-Eurasians and SSA could be Basal Eurasian. While BE is more related to non-SSA, it shares an amount of ancestral with SSA that it doesn't with Eurasians because it split of earlier. Hence, all non-BE Eurasians, called Crown-Eurasians if I understand correctly, share some drift or mutations or whatever that they don't share with BE.


When you add a Neanderthal to ADMIXTURE it almost entirely shows African. However, Africans do not have a closer relation to Neanderthals than non-SSA's. I think that is something similar.

epoch2013 said...

No. Crown-Eurasians aren't all non-BE Eurasians. They are the next split: Ust'Ishim and so on. My bad.

Hector said...

capra internetensis cracked me up.
" The Denisovan-admixed population had ~7% Denisovan and ~93% a modern human population that split off before ->Chinese<-, ..."

So then where were the paleolithic, say,Luxembourgers at the time? When did they split?

epoch2013 said...

Aha, Razib Khan has more: http://www.unz.com/gnxp/the-first-settlement-and-last-glacial-maximum/

So Dinka and Yoruba, but not San, show affinity.

Note, more recent Holocene gene flow form Eurasians into Sub-Saharan populations is clear in Nilotic populations, such as the Masai. This is something different. Perhaps it is older gene flow back into Africa, or, perhaps it is evidence of substantial African gene flow into Eurasia, possibly during the Pleistocene.

capra internetensis said...

@Hector

WHG wasn't on that admixturegraph.

In the graph from the most recent paper (the SGDP one) Australasians are a sister group to Onge; Ami and Dai form a sister branch to Onge and Australasian; and Kostenki14 branches off above that. In this one there is no earlier Denisovan-admixed population.

But it won't surprise me if everything changes once we get ancient genomes from other parts of the world.

terryt said...

@ Capra:

"Australasians are a sister group to Onge; Ami and Dai form a sister branch to Onge and Australasian"

The trouble with using "Australasians' as a group is that it involves at least two very distinct populations. Wallace's Line has been crossed by three different Y-DNA lines, the last being very heavily influence by an 'East Asian' genetic contribution from southern China just 4-5000 years ago. It is doubtful Onge have much of that last but until the different populations that crossed the line have been isolated we can have no idea of what was happening on the mainland. There was obviously a lot of to and fro and mixing and so, yes, everything will change once we get ancient genomes from that part of the world.

Shaikorth said...

Epoch,

There were some D-stats (French East Asian SSA Chimp) in Wong et al that indicated Dinka and Yoruba were equidistant to W and E Eurasia, so their Eurasian admixture, if it's there, could be something that's neither.

others,
re: earlier stuff about global PCA's, remember that the population at the EE end of the triangle can also vary depending on the number of samples (and the dimensions incorporate drift), if there's enough East Asians it's them, if there's enough Native Americans it's them and so on.

http://oi64.tinypic.com/rr5li9.jpg

It perhaps "should" be Onge if we want the group which is truly most distant from others to end up there.

epoch2013 said...

@Shaikorth

This is the paper: http://www.nature.com/nature/journal/v538/n7624/full/nature18299.html

They state: Split between Euarsians and Sahuls round 58 ky BP. Geneflow since then, suggesting the groups weren't separated. Split between East- and West-Eurasians 42 ky ago.

"MSMC analyses suggest that the Yoruba/Australo-Papuans and the
Yoruba/Eurasians cross-coalescence rates are distinct, implying that the Yoruba and Eurasian gene trees across the genome have, on average,
more recent common ancestors (Extended Data Fig. 4c, Supplementary
Information section S08). We show through simulations that these differences cannot be explained by typical amounts of archaic admixture
(< 20%, Extended Data Fig. 4d). Moreover, the expected difference in
phasing quality among genomes is not sufficient to explain this pattern
fully (Supplementary Information section S08). While a similar
separation in cross coalescence rate curves is obtained when comparing
Eurasians and Australo-Papuans with Dinka, we find that, when
comparing Australo-Papuans and Eurasians with San, the cross coalescence curves overlap (Extended Data Fig. 4c).
"

Supp info: http://www.nature.com/nature/journal/v538/n7624/extref/nature18299-s1.pdf

Poise n Pen said...

So much doublethink here.

If OOA is true then it's ALL african DNA. But it's not so it doesn't matter. When you have some ancient samples of modern eurasian DNA in africa before it came here then wake me up but I'm not holding my breath.

capra internetensis said...

Keep at it, Poise n Pen! One more content-free drive-by posting and I'm sure we'll all shake loose from our liberal conspiracy brainwashing and see the light.