Friday, September 2, 2016
Ust'-Ishim man x2
Just wanted to see what would happen if I split Ust-Ishim into two homozygous sequences, and ran him in a Principal Component Analysis (PCA) alongside other fully homozygous individuals and composites, both modern day and ancient. In theory this shouldn't be a problem, considering that his genome is of such an incredibly high quality (~42-fold coverage). The datasheets for the PCA below are available here and here.
This is the only way that I can think of to run a PCA in which genetic drift specific to the Ust'-Ishim population of ~45K YBP western Siberia has a clear impact. Indeed, lots of interesting things on these plots. Note, for instance, that on the second plot the Afanasievo and Yamnaya-Catacomb composites appear to be attracted to the Caucasus Hunter-Gatherer composite. Also note how unusual the Neolithic Iranians (Iran_EN) appear.
Update 02/09/2016: Interactive 3D versions of these PCA are now available at Open Genomes here and here. Below are screen caps of angles that appear to recapitulate the geography of Eurasia. See comments below for details.
See also...
Ancient vs modern day West Eurasian variation
Subscribe to:
Post Comments (Atom)
117 comments:
Yep Nice
You substraced all Neanderthal from Oase 1? Just a wild guess, but the Vindija genome just came online...
So Graph 2 is the sort of plot where we normally have a set of West Eurasians and East Eurasians and Sardinians (among moderns) maximise the main PC differentiating West from East Eurasians.
So crossing up this up with D-stats for West Eurasians (ancient+modern inc. SCA), strongest correlations, PC1 correlates with D(Mbuti,Boncuklu)(Mota,X) at 0.94425 and PC2 correlates with D(Mbuti,El_Miron)(Mota,X) at -0.9524 (only slightly stronger correlation than Villabruna at -0.9524).
PC5 looks like the PC where Ust Ishim is most distinct, though in crossover with D-stats, best correlation with D(Mbuti,Ust_Ishim)(Mota,X) is PC4 for West Eurasians. This PC4 which contrasts Ust-Ishim/Euro HG/East Asian at one end, and Karitiana at the far other end, with "Basal Eurasian" populations intermediate and increasingly far from Ust-Ishim with increasing BEu. So it makes sense this would correlate this way.
PC8 and PC9 seem like sort of meaningless PCs that separate Ust_Ishim 1 and Ust Ishim 2, which I would guess is basically virtual variation?
If you have the weighting of each of these PCs (how much % variance) that would be interesting?
Matt, do you see any interesting correlation with these?
https://docs.google.com/spreadsheets/d/1yrD6VI6c95RslwJeeoD8ELvidfUrFfgUGP2kqi59fQ8/edit#gid=359466717
I'm enjoying the separate arcs forming around the two Neolithics.
This is making (Native American = ANE + East Asian) look doubtful, no? Or am I misinterpreting something?
I'm excited to see where you're going with this. Would it be possible/relevant to include Mal'ta, AG2, and the earlier European HGs?
Cool stuff Davidski, thanks.
Shaikorth, not many populations cross up between them (at least without work manually matching them rather than just matching up same IDs). Only (Abkhasian, Armenian, Basque_French, Belarusian, Bulgarian, Chechen, Croatian, Dai, Druze, Erzya, French, Georgian, Iranian, Italian_Bergamo, Italian_Tuscan, Japanese, Kalash, Karitiana, Lezgin, Lithuanian, Norwegian, Pathan, Russian_Kargopol, Sardinian)
For those that do, strongest correlations:
PC1: Anatolia_Neolithic 0.96, Kotias 0.92, Satsurblia_Cluster 0.92, Oase1 -0.92, Ust_Ishim -0.89
PC2: GoyetQ116-1: -0.63, Malta1 -0.63, AG3 -0.63 (Villabruna -0.43999, El_Miron -0.51909)
PC3: Malta -0.81, AG3 -0.83
PC4: Malta 0.42, AG3 0.47
PC5: Malta 0.64, AG3 0.67
The PC2 loads less with Villabruna on this I think because of different samples, and because Karitiana (not included in the previous correlations) has quite low f3 with Villabruna compared to others, while quite low on PC2.
Please seem my comments on the last post.
I'm working right now on getting the common coverage between WC1 and Bar8, in preparation for whole genome variant calling on both to find "Basal" IBD segments.
If I were to use some genomes as an outgroup for the combination of WC1 and Bar8, should I use Ust'-Ishim in addition to Mota, La Brana 1, and MA1? Any others I should include?
The genomes have to be very high coverage, to cover almost all the regions in common between WC1 and Bar8.
@Taymas
Best I can do for now is to also add Kostenki14. But ideally I'd need a couple of samples of this type to influence the plot, and Kostenki14 is already a fully homozygous sequence, so I can't split it into two.
https://1.bp.blogspot.com/-ASxaPIb_b4w/V8n_CLitABI/AAAAAAAAE18/uzrPlHQbXx04QllvnCiBhHdp8fOYLaJXgCLcB/s1600/PCA.png
https://drive.google.com/file/d/0B9o3EYTdM8lQb2RWNHdNN3ctRWM/view?usp=sharing
https://2.bp.blogspot.com/-jMigi25OZRQ/V8n_HVeN2GI/AAAAAAAAE2A/moDVQuIVYh4PItRYpZTF2FxUjR91gv_LQCLcB/s1600/PCA2.png
https://drive.google.com/file/d/0B9o3EYTdM8lQMkZpWF93dFg1Slk/view?usp=sharing
Dave: I like how Kostenki 14 provides a link (though a bit Iran/CHG-shifted) between UI and the EHG-Motala-Hung_HG cline. Maybe you replace your original PCAs in the opener with the new ones that include Kostenki 14 (and even mark it in red) ?
Would it be possible to include the new Jomon sample(s)? The PCA (Fig 1a) in the Jomon study shows them as substantially shifted to West Eurasians. In fact, they almost come in as BE, or at least as something like 75% ANE (JPT,KHV) - 25% IBS, with a bit of African (Near East?) pull. If that relation holds in your PCA, it would place Jomon somewhere near the junctions of the JPT-Kostenk14-Iberia_MN and Karitiana-UI-Gujarati clines.
Very difficult to say what Kostenki14 is up to on those plots without at least a couple of individuals of the same type.
By the way, the plot based on dimensions 1&4 that I just included in the post is very interesting.
Ust-Ishim Interactive 3-D PC 1-2-3 plots:
3-D Interactive Ust-Ishim PC 1-2-3 (no East Asians)
3-D Interactive Ust-Ishim PC 1-2-3 (with East Asians)
David, here are screenshots for you to post (unless you want a different angle):
Ust-Ishim PC 1-2-3 (no East Asians)
Ust-Ishim PC 1-2-3 (with East Asians)
Isn't there a diploid VCF for Kostenki K14?
Let me also try PC 1-2-4 as well.
@FrankN: I agree the Jomon would be very useful here.
Dave: Would it also be possible to include Paniya in your PCAs? There have been those nMonte trials by Alberto e.a. showing ANE and Paniya to partly be interchangeable. Moreover, at 75% yDNA F and 16% yDNA C, the Paniya are probably as basal as you can get with modern populations when it comes to yDNA phylogeny.
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0050269
[mtDNA-wise, the most "basal Eurasian" population seems to be Socotrians; hardly surprising when looking at the map.]
@OG
Yeah, try 1-2-4 for the plot with no East Asians. That should look pretty sweet.
@Frank
I can only run individuals that have around a million markers available in my dataset.
The 3-D Interactive plots for Ust'-Ishim PC 1-2-4:
3-D Interactive Ust-Ishim PC 1-2-4 (no East Asians)
3-D Interactive Ust-Ishim PC 1-2-4 (with East Asians)
PC 1-2-4 is proving to be especially interesting for determining the degree of "basalness" of a given population.
Here are screenshots with annotations that illustrate this:
Ust-Ishim PC 1-2-4 (no East Asians) showing a vector from Ust-Ishim to the Natufians/Levantine PPNA and the Anatolian Neolithic, flanked by Iran Neolithic and CHG.
Ust-Ishim PC 1-2-4 (with East Asians) showing a vector from the East Asians and Karitiana to the Natufians + Levantine PPNA and the Anatolian Neolithic, flanked by Iran Neolithic and WHG.
Why does adding the East Asians and the Karitiana exchange CHG with WHG, where without East Asians CHG is on the opposite side of the "basal vector" from Iran Neolithic, but with the East Asians, CHG clusters closely with Iran Chalcolithic, about halfway between Iran Neolithic and the Natufians + Levantine PPNB?
Wow, the PC 1-2-4 with East Asians looks just like a map of Eurasia!
(If you rotate things the right way.)
Annotated screenshot of Ust-Ishim PC 1-2-4 with East Asians as a map of Eurasia
That can't be a coincidence, can it? I think not! ;)
OK, I screen capped the plots so that the images had the right sort of borders. See update.
I gotta run. Back tomorrow.
A clearer view of PC 1-2-4 with East Asians as a map of Eurasia, with the cardinal directions shown:
Ust-Ishim PC 1-2-4 with East Asians as a Map of Eurasia
It seems that "North" = less Basal Eurasian, and "South" = more "Basal Eurasian".
Africa would be on the lower left, below the Natufians + Levantine PPNB who have North African admixture.
I bet that the ASI Onge and the Australians will be somewhere off on the lower right, past Iran, but the Australians will be drawn northward by their Denisovan admixture. The ASI peoples such as the Indian Austroasiatic and the Indo-Pacific Kusunda in fact usually cluster closely with Ust'-Ishim.
Notice how Yamnaya (Chalcolithic Steppe, the green diamonds) appear to be EHGs mixed with CHGs, but the Corded Ware and Bell Beakers (Chalcolithic and Bronze Age Europe,light blue diamond and circle) appear to be EHGs mixed with the Anatolian Neolithic.
The Chalcolithic and Bronze Age Iranians, Caucasians, and Levantines appear to be in varying proportions part way between the Iranian and Anatolian Neolithic. What's interesting is that the Chalcolithic Iranian[s] are considerably closer to the Levant and Anatolia than the Early Neolithic Iranian. Is this the influence of the continuing population explosion in Southeast Anatolia where wheat farming and cattl herding originated?
Signficantly as I said, there are Y-DNA G2-P287s and mtDNA K1a, X2, J1, and T2 in Anatolia, Early Neolithic Iran, and Chalcolithic Iran. If we exclude African Y E1b1b1-M35, we can see on the map which Y and mtDNA haplogroups are "Southern" during the LGM and associated with "Basal Eurasian".
@ OG/ Terry
"What's interesting is that the Chalcolithic Iranian[s] are considerably closer to the Levant and Anatolia than the Early Neolithic Iranian. s this the influence of the continuing population explosion in Southeast Anatolia where wheat farming and cattl herding originated"
IMO, its more complicated than that. In the M5-M4 period, there was a lot of settlement flux in Iran, both the central plateau and the Zagros highlands. For example, with the latter, the "pull" of incipient urbanism in Mesopotamia appears to have shifted the highland populace further SW, culminating in agglomeration & intermixture. A similar thing happened in Anatolia, with a distinct phase where settlements west of SE Anatolia appear to be spare & ephemeral.
Open Genomes,
First, thanks for all your efforts, I appreciate it.
With regards to the challenge of trying to identify alleles derived in a ghost population such as Basal Eurasians, the 1st step would have to be to identify all those positions in Iran WC1 or Anatolia N, that have the ancestral alleles, which would make up the majority of their genomes. This can be accomplished by comparing with a few outgroups such as Yoruba. Next, a comparison with Ust, we would be able to identify additional ancestral alleles missed by Yoruba, as well as some derived ones in ENA.
Finally, pick a couple of high coverage genomes that make up the majority of Iran WC1's non basal admixture (something majority ANE), for Anatolia N, this would be a little different (WHG?), to identify those non basal derived alleles.
In the end, at best you would have identified those positions in WC1 with the suspected Basal derived. Many of those positions would be different for Anatolia N, because of how recombination works, but between the 2, you could potentially have more postions with the Basal derived, than using just one of them, but at the end of the day, you would potentially have a partial Basal genome, since neither is 100% Basal. I suppose a few additional positions may be imputed, depending.
OG,
A little clarification on this post ^^^^
The more SSA and ENA genomes you use, the greater the number of positions in WC1 or Bar you will be able to define as either ancestral alleles, or ENA derived alleles.
awesome sauce
David,
How do we define "Basal Eurasian"?
Say, if we extracted the biallelic, non-problematic, common alleles between WC1, Bar8, and your merged Natufian+Levantine PPNB, would that be considered a haploid "Basal Eurasian"?
Filtering out alleles not in common with the Altai Neanderthal doesn't work:
1. We know that the Altai Neanderthal has some kind of early AMH admixture (a good candidate for part of "Basal Eurasian").
2. The earliest "Out of Africa" AMH in the Near East would have had some common Neanderthal Admixture (i.e. from the Carmel Caves Neanderthals) and this would be also symmetrical to all other Eurasians as well.
We also can't use Mota, the Mbuti, or the San to filter out Basal Eurasian alleles:
1. Some "African" alleles would have been lost as the early "Out of Africa" AMH population moved into the interior of Eurasia.
2. There's no evidence that WC1 and Bar8 have additional common post-OOA African admixture, and while the Natufians+Levantine PPNB do have post-OOA African Admixture, this would not be in at least in common with WC1, even though some may be shared with the Anatolian Neolithic (Bar8).
So would the alleles that intersect WC1, Bar8 (The Anatolian Neolithic in general) and the Natufians+Levantine PPNB constitute a haploid "Basal Eurasian", or is it better to leave out the Natufians+Levantine PPNB and just consider the Anatolian Neolithic + the Iranian Neolithic? What if the composite Anatolian Neolithic you are using which has WHG admixture which has eliminated some possible common alleles with WC1? The same would likely be true for the South Asian admixture found in the Iranian Neolithic.
Also consider that pretty much all other Eurasians should have some degree of "Basal Eurasian" admixture as a remnant of AMH Out of Africa. This should even include Ust-Ishim as well.
Have you considered that some of the "ANE admixture" you see in WC1 is really a remnant of a common WHG-EHG-ANE "Basal Eurasian" heritage?
Maybe the right thing to do would be just to find the alleles in common between WC1 and Bar8, instead of composite Anatolian Neolithic and Iranian Neolithic. There are pros and cons to each method.
So what do you get when you find the common alleles between any Anatolian Neolithic and any Iranian Neolithic?
Wouldn't this constitute a set of "Basal Eurasian" IBD segments?
I'm going to try this now for the filtered 1152888 biallelic, "no-problematic" SNPs found in any of the commercial or academic SNP arrays. (This includes both the Human Origins Array as well as the Geno 2.0 SNPs with positions as well as all 23andMe v3 SNPs, and of course the OmniExpress-24 chip used by FF and Ancestry.)
I can also do this for all filtered biallelic, non-problematic SNVs (no indels) in dbSNP 147 from the common coverage regions between the WC1 and Bar8 whole genomes. This of course would extract a much larger set of SNPs, most of which would have been left off the various SNP arrays. These would then in effect be whole genome-based "Basal Eurasian" haplotypes based on ancient samples. I can then extract the frequencies for other samples like the 1000 Genomes. (Does the Simmons dataset have alele frequencies too?)
So if you can help define what is and isn't "Basal Eurasian", we can then start looking for it.
OG,
I would not worry about a possible Basal allele here or there getting filtered out using Archaics or Africans, because the benefit way outweighs the harm, since in return for a couple of basal alleles getting filtered out, you would in return be filtering 1000s of alleles not derived in Basal Eurasians. I assume the goal is to identify Basal Eurasian derived alleles.
The purpose for using Africans is not to filter out any African alleles due to post OOA admixture, but rather to filter out alleles ancestral to Basal Eurasians, since Basal Eurasians just like any other Eurasians carry a huge amount of those.
In fact that is why Dstats work much better than IBS. Unlike IBS, where score is very affected by shared common ancestral alleles, with Dstats we use outgroups such as Mbuti or Chimp to identify ancestral alleles in D (X, Y, Target, Outgroup), which is why IBS will show Africans more similar to Altai (ancestral sharing SSA-Altai) compared to say Europeans, whereas
D( European, Yoruba, Altai, Chimp) will be +ve, thus showing the use of the Chimp outgroup helps Dstats be more accurate than IBS in this example.
If on the other hand the goal is to identify alleles shared by WC1 to the exclusion of others, a quick and dirty way may be to use Plink to extract positions in WC1 that have a MAF > T, where you would try different values for T ( for ex > 85%). I have done this in the past with some success. Those alleles would likely be from relatively recent drift of WC1, and may include some Basal alleles
"while the Natufians+Levantine PPNB do have post-OOA African Admixture"
This is a debated topic. It could be the other way around. Lazaridis found no african admixture in these natufians.
Why should E1b1b not be the male haplogroup for Basal Eurasians? It is clearly associated with West Eurasian or Caucasoid influences in Africa.
@Taymas
"This is making (Native American = ANE + East Asian) look doubtful, no? Or am I misinterpreting something?"
You are not misinterpreting anything. Native Americans is not a product of mixture between ANE and east Asians, but a source population for East Asians and West Eurasians.
@Taymas
Yes, you're missing the fact that there is no ANE sample in the analysis, therefore the analysis can't show the Karitiana as a mixture of ANE and East Asians. Ideally, we'd need a couple of ANE samples to show this well.
Nevertheless, using all nine dimensions the Karitiana are modeled as a mixture of Eastern_HG, Han and Ust-Ishim. But the proportions are off and fit very poor.
@Davidski
"Nevertheless, using all nine dimensions the Karitiana are modeled as a mixture of Eastern_HG, Han and Ust-Ishim. But the proportions are off and fit very poor.'
This means it's not successfully modeled as a mixture of Eastern_HG, Han and Ust-Ishim. But you can keep trying. Now you've added Eastern_HG. They weren't their originally. Oh, yes, and please add Papuans as well. All of them are mixed up in your head as "Karitiana."
Eastern_HG were in all of the datasheets originally. That's why they also appear in the 3D PCA generated by Open Genomes.
@ Davidski
I may read only the Activity stream on Anthrogenica and found that:
Generalissimo replied to a thread Eneolithic aDNA from Lake Baikal in Ancient (aDNA)
I meant that because this is a thesis that used PCR, the R1a-M17 might actually Russian R1a-Z80 from the people who handled the remains.
I wrote on the YFull page of FB and on mine, are you sure that that sample has the mutation M17? It doesn't seem to me, because there are 4 sign of G, the left one more little than the others but similar to other sign for G.
Striking line through Hungary_HG, Motala_HG, Eastern_HG and Ust_Ishim in the first graph.
I wondered what an advanced clustering program would do to this datasheet.
So did a run with mclust. I found 8 clusters and a rest group of (very) ancient pops.
Here is the PCA(1,2):
https://www.dropbox.com/s/qnovwqazoqx4t4m/test4_12.png?dl=0
And this is the PCA(1,4):
https://www.dropbox.com/s/vy177bmzpgpwkt2/test4_14.png?dl=0
I labeled the clusters after their most central member.
The rest group is the green triangles belonging to the large circle having no label.
@Davidsky
Surprisingly, the EM-algorithm does not separate a WHG cluster. A reason might be that the sheet does not have enough low dimensions.
After all 9 dimensions is only a sliver the of the thousands of dimensions of the DNA; do you have the eigenvalues?
Is it possible to get a sheet with considerably more dimensions, say 20 or even 50?
@OG: "We know that the Altai Neanderthal has some kind of early AMH admixture (a good candidate for part of "Basal Eurasian")."
That admix stems from Our of Africa (OOA) 1, 100-150 kya, and there is a good chance OOA 1 never reached West Eurasia (at least not Europe, the Near East is a different issue).
Uniparentally, we should be dealing with yDNA DE here. D surely qualifies as Basal East Eurasian, and the basalness of Jomon aDNA may have to do with the prevalence of D2 among Japanese, especially Ainu and Ryukans. D hasn't yet shown up in European Paleolithic aDNA so far.
E is more complicated, of course. It may well have found refuge from the Lake Toba eruption in the Persian Gulf Oasis, and repopulated the Near East and Eastern Africa from there. But it is widespread enough across Africa to rather Show up as SSA than as BE in statistics.
Clearly BE are yDNA C and F. The former is widespread in European Paleolithic aDNA, though its current distribution is predominantly East Asian, Polynesian and American. That current distribution oould, in spite of C's basalnesss, pull statistics towards the East Eurasian side. F* is predominantly South Asian, with a peak in Paniya.
Subsequently, we had West Eurasian GHIJLT splitting up from F*,and then K2. Technically, K2 should be ragarded as Basal East Eurasian, since it gave rise to all major East Eurasion yDNA haplogroups except D. However, yDNA R complicates the picture for its modern prevalence in Europe. As exemplified by MA1, R* is in D/f-stats drawn to the West Eurasian side. But what about upstream K2b (UI) and P? I guess they may also turn out pretty basal in D/f-Stats.
The mtDNA side is even more complicated in this respect. The phylogeny of M and N is poorely resolved. If there have been survivors of OOA 1, as suggested by the existence and geographic spread of yDNA D, they should carry some of the more basal macrohaplogroups under N/M. Jomon aDNA, predominantly made up by mtDNA D4b2, M7a and N9b (Link, Table 1) might provide some hint in this respect, but ultimately, more (and more ancient) E. Asian aDNA will be required.
http://www.jjarchaeology.jp/contents/pdf/vol002/2-1_034-059.pdf
Otherwise, mtDNA R looks quite basal and as an apt candidate to have joined yDNA CF in OOA II.
@FrankN
Based on the sequencing of the Altai Neanderthal, it does seem that there was a possible pre-Toba AMH population somewhere in Eurasia, that admixed with the ancestors of the Altai Neanderthal, but not the other Neanderthals.
The question of the origin of Y-DNA DE-CF remains unresolved. However, the tMRCA of DE according to YFull is 62,500 ybp:
YFull YTree v4.07 CT phylogeny and tMRCAs
This of course is too late for Toba. Interestingly, while the tMRCA of E is 54,200 ybp, the tMRCA of D is 46,800 ybp, and separately, the tMRCA of C is 48,800 ybp, GHIJK 48,500 ybp, K 45,500 ybp, and K(xLT) 45,400 ybp.
The same event apparently caused the simultaneous expansion of D, C and F, and the immediate subclades of F, but not E, which expanded earlier. There was an initial warming around 58,000 ybp, but the H3 Event at c. 48,500 ybp was a dramatic sudden warming which may have enabled post-Toba AMH to expand throughout Eurasia.
It may not seem parsimonious, but it is possible that D, C, and F co-migrated at the very same time and having arrived in the Near East sometime after 58,500 ybp, all began to expand at the H3 Event. In that case, E would have been left behind somewhere in East Africa (the Ethiopian Highlands?) and underwent a different process of expansion several thousand years later.
The fact is that Mota shows absolutely no sign of Eurasian (i.e. Neanderthal) admixture and Mota is Y E1b1a2. There's no way around this. Even among modern populations, many predominantly E1 and E2 populations show no sign of Eurasian admixture.
The format of E-Z830 is 23,800 ybp, just before the start of the LGM at 23,000 ybp,and the tMRCA of the "Eurasian" branches of E-Z830 is 19,200 ybp, right at the height of the LGM:
YFull YTree v4.07 E-Z830
(Notice that E-Z1267 and its subclade E-M293 are exclusively African, not Eurasian.)
It does seem that the E-M123 (E-M34?) found among the Natufians of Mt. Carmel entered the Near East just after the end of the LGM, or at latest, at the time of the Bolling Interstadia / H1 at 14,700 ybp.
The Natufian "merged" sample in Gedmatch has a rather substantial Eurogenes K36 "North African" component that is entirely lacking in "Basal Eurasian" WC1:
M041601 Natufian Merge "North African" by chromosome:
North_African 25.1 21.3 13.7 19.4 22.0 15.8 23.3 13.9 5.7 13.7 17.0 13.1 10.6 17.2 16.9 20.0 19.1 12.5 13.2 16.2 0.7 12.2
M392829 WC1 Wezmeh Cave, Iran 7455-7082 calBCE (9465-9092 ybp) has absolutely zero North African.
Note that the sister clades of E-Z830 in E-M35 are predominantly African, with the exception of E-L618 which has a tMRCA of only 7700 ybp.
So based on the Y-DNA combined with the lack of non-MN mtDNA clades among early Eurasians, E-M35 or E-Z830 cannot be the source of "Basal Eurasian" because the ancient samples who carry E-M123 (E-M84) and E-M78 also have this additional "North African" component lacking in other higher percentage "Basal Eurasian" samples.
About "Basal East Asian". The chromosome 21 sequence of Tianyuan already showed at 40,000 years ago an emerging "East Asian" component:
DNA analysis of an early modern human from Tianyuan Cave, China
That's about as "Basal East Asian" as one could get, but this component is not symmetrical to all Eurasians, only to East Asians. It's by definition, not "Basal Eurasian".
The "Basal Eurasian" component has to be found elsewhere, in the Near East, not East Asia.
You can't use modern clusters for ancients. E is Eurasian. There's no way around that. There's no evidence of SSA in Natufians, but likely Eurasian in all Africans. Just wait until Ibero-Maurusians come out.
BTW, the so-called F* among the Paniya is actually a subclade of H1. The Lahu KVH Vietnamese, and Malays all have a slighly pre-GHIJK clade of F (which includes F2) but this is not found in South Asia.
Both Ust'-Ishim and Oase1 were Y-DNA NO / K2b-M2335*, and Oase1 was a pure mtDNA N*, derived for 3 out of 5 mtDNA N SNPs. Ust'-Ishim was R* with one additional contro-region SNP. At that point, neither were "Basal Eurasian" and neither were the extinct mtDNA Ms among the European Aurignacians.
Neither mtDNA N*,R*, or M* per se can be associated with "Basal Eurasian" although they predated the breakup of "Basal Eurasian" into separate components.
Both Y-DNA G and H2-M282 split from their respective clades 48,500-45,000 years ago which is right around the time that "Basal Eurasian" separated from its specific Eurasian descendant populations.
The tMRCAs of mtDNA clades are very difficult to date, because ancestral haplotypes can persist for thousands of years alongside derived haplotypes. We have no idea about the real tMRCAs of mtDNA K1, X2, J1 and T2, but we do know that Tianyuan was already in mtDNA haplogroup B at 40,000 ypb, only about 1500 years after its "ancestral" N* Oase1 haplotype at 41,500 ybp, which was still in existence for thousands of years after Ust'-Ishim at c. 45,000 ypb. (Even if, based on the Y-DNA, Ust'-Ishim was actually later, it would still be no later than Oase1.)
@ Frank N
Off Topic; if you have time, have a read of : "The Idea of Eneolithic" by S Kadrow and let me know what you think
@OG: Even though there has been some adjustment, TMRCA estimates for yDNA still essentially go back to Karafet, Underhill e.a. (2008)
http://genome.cshlp.org/content/early/2008/04/02/gr.7172008.full.pdf+html
If you read that study carefully, you will note on p. 7 (enhancement is mine):
"To provide estimates of the age of the nodes, we chose to fix the time to the most recent common ancestor of CT (defined by P9.1, M168, and M294) at 70 thousand years ago (Kya), which is consistent with previous estimates from genetic and archaeological data (Lahr and Foley 1998; Hammer and Zegura 2002; Macaulay et al. 2005), and is the chronological approximation given in Jobling et al. (2004) (p250) for the first major human out-ofAfrica dispersals. We estimated the times for intermediate nodes by using a linear interpolation. The age estimates in years should be viewed with caution because we do not know if the calibration date chosen above is accurate."
Well, we now have evidence that AMH admixed with Altai Neandertals much earlier. This should give every reason to review the basic assumption underlying those yDNA TMRCA estimates, especially as we know that mtDNA TMRCAs temd to be much earlier than those for yDNA.
Otherwise, thx for the Tianyuan link that had escaped me. Note here that Tianyuan has been determined as mtDNA R*, carrying "four substitutions (5,348, 5,836, 11,257, 16,293) [..} ancestral to present-day haplogroup B". Tianyuan pre-B is clearly Basal East Eurasian (plus Amerindian), but its ancestral R* might stil go for BE.
Might I ask from where you know that the Paniya F* is in fact a subclade of H1?
@huijbregts
https://drive.google.com/file/d/0B9o3EYTdM8lQZG9xNWswNnYwVjg/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQNU1MX3JMX29PNlU/view?usp=sharing
But methinks you need more individuals in such analyses to get strong cluster assignments with Mclust.
I managed to recreate the "fateful triangle" with Ust-Ishim. Not sure why though?
https://3.bp.blogspot.com/-ugIau5__qIU/V80e0Y1fivI/AAAAAAAAE3U/263_3fJucIU6ymA9H18auFQXSsPrAn1tgCLcB/s1600/Triangle_PCA.png
https://drive.google.com/file/d/0B9o3EYTdM8lQRDlyQWxEb095WmM/view?usp=sharing
@Davidski
"I managed to recreate the "fateful triangle" with Ust-Ishim. Not sure why though?"
Maybe because Ust-Ishim has no WHG or Basal DNA.
@Davidski
Thank you for the files.
Mclust is a powerful program (if you have explored the options).
In the case of your Dim=9 file, the solution of the mclust problem was simple: average within the pops.
The PCA(1,2) with the 7-cluster solution is here:
https://www.dropbox.com/sh/8xqciutho6pv8ak/AADJlnOM9T-cWpHwOvUIJpLfa?dl=0
Here WHG is a conspiciously distinct cluster. The modern Europeans have their own cluster, distinct from the modern Caucasus pops, etc.
There is also a rest cluster of the red squares, which I have not labeled. It is huge, has a low density and a lot of satellites; so it reminds me of the planet Jupiter.
Notice Usht-Ishim at the far right; it is less extreme than in your fateful triangle.
@FrankN
The calibration chosen by Karafet et al 2008 turns out to have been quite accurate, but it is *not* the basis for present estimates. A good review of various calibrations from Y-Full, with a detailed explanation how they obtained the rate they use, can be found at Defining a new rate constant for Y-chromosome SNPs based on full sequencing data.
Presently most estimates are based on radiocarbon-dated, high quality ancient genomes, notably Ust' Ishim man, as reported in Fu et al 2014, "Genome sequence of a 45,000-year-old modern human from western Siberia", and Anzick-1 from Rasmussen et al 2014, "The genome of a Late Pleistocene human from a Clovis burial site in western Montana".
The modern human admixture into the Altai Neanderthal is *not* related to modern Eurasians, basal or otherwise. It was suggested to have split from the modern human lineage around the same time as San or even earlier. So this is not at all a good calibration point for modern Eurasian DNA.
Present estimates for mitochondrial haplogroup TMRCAs are not systematically earlier than for Y haplogroups (the only exception that comes to mind is Rieux et al 2014); in fact they are quite similar though of course with large confidence intervals. Posth et al, using a large number of Pleistocene ancient genomes, calculated the TMRCA of M and N at 55-45 and 55-47 thousand years ago, coinciding well with the estimated TMRCAs of C, D, E, and F. When we get back before Ust' Ishim man, or when dealing with African-specific lineages (e.g. see the recent South African paper from Barbieri et al), we have no ancient DNA to calibrate with, so the uncertainty going back to the MRCA of CDEF is greater, but our current estimates have a pretty solid basis.
@Capra - well explained.
Thank you for pointing out that the early Eurasian "AMH admixture in the Altai Neanderthal is *not* related to modern Eurasians, basal or otherwise."
First off, has anyone thought of extracting this small percentage of "Early Eurasian AMH" and plotting it with all other populations, ancient and modern? If we can identify the SNPs, it should be possible, right? David?
If anyone has a problem identifying these SNPs, the Reich Lab should be responsible for providing a VCF with these SNPs alone. Of course, the variants would have to overlap somewhat with (African?) modern humans today, so there should be rsIDs for at least some of these SNPs. The Altai Neanderthal paper was a peer-reviewed paper which made this claim, so they are obligated to provide this raw data to those who, ask for it, especially since others might try to replicate their finding to validate it. (Remember the Mota "LBK-like ancestry in Africans" debacle where a simple miscalculation made it seem that *every* African population had at least 4% Eurasian admixture. External validation of such claims is very important.)
Who would like to contact the Reich Lab for this data?
@Nick Patterson (Broad)?
About "Basal Eurasian" and African alleles:
Why isn't it valid to include predominantly African alleles in "Basal Eurasian"?
Think about this:
1. *Yes*, WC1 apparently does not have any post-OOA African admixture.
2. Apparently, Bar8 does have some post-OOA North African admixture (almost certainly via the Natufians and Levantine PPNB, something important in and of itself, but this will not be in common with WC1, who has no such admixture.
M711494 Bar8 Anatolian Neolithic, Barcin, 6212-6030 BCE (82222-8040 ybp):
Eurogenes K36:
North_African - 8.5 5.3 - 2.5 1.3 2.6 - 0.2 2.2 - 2.8 2.8 8.7 2.1 - 5.2 0.7 5.9 8.2 3.1 24.4
Consider this scenario:
An isolated initial OOA population lived in the Near East for several thousand years. (At most, say 58.5 kya to H3 at 48.5 kya = 10,000 years). At 48.5 kya, one or more small Near Eastern AMH groups begin to move out into Eurasia. Ust'-Ishim and Oase1 represent at least one of these groups, mtDNA N* and R* respectively, and Y-DNA K2b-M2335*.
As these smaller groups move out into the eastward along the Indian Ocean coast and into the interior of Eurasia, they progressively lost "pure ancestral African" AMH alleles, either by archaic admixture or by adaptation and selection. So there would be reduced set of ancestral African alleles through drift among all early Eurasians who moved out into Eurasia. The PC plots show this to be consistently the case for Ust'-Ishim even at 45 kya, because he no longer plots anywhere near Africans or Near Easterners, but rather somewhat close to South Asian Austroasiatic peoples.
So to merely be "symmetrical" to all other Eurasians, it isn't necessary to have novel derived alleles but also ancestral African alleles that where lost from everyone else through drift. This would especially show up if a few Africans without any trace of Eurasian admixture (e.g. Mota, and maybe San and Mbuti) are included, although ancient samples like the Natufians and their Levantine descendants who have some (North) African admixture may confuse the issue.
One definition of "Basal Eurasian" is:
"The set of Eurasian alleles which show no sign of Neanderthal (or Denisovan) origin."
This definition does not take into account any possible Neanderthal admixture into the initial OOA AMH population in the Levant c. 56,000 years ago. (Qafzeh Cave). If this was a small group without any population structure, the Neanderthal admixture would be universal among their immediate descendants, regardless of any additional Archaic admixture later on by groups that left the Near East.
These Neanderthal alleles would also count as "Basal Eurasian" since they would be widespread in nearly all Eurasian populations.
Another definition of "Basal Eurasian" would be:
"The set of Non-African derived Eurasian alleles."
This does not take into account ancestral African alleles found in the initial OOA population lost through drift in all other coastal and interior Eurasian populations.
So neither the definition of "non-ancestral African / derived Eurasian alleles" nor "non-Archaic alleles" works as a definition of "Basal Eurasian".
Ust'-Ishim does appear to retain some (seemingly) ancestral African alleles that would later lost in Eurasian populations through drift and selection, unless these are just Neanderthal alleles that merely appear in the calculators to be "African" because of the ascertainment bias that favored alleles which are polymorphic among present day day populations, such as those SNPs selected for the Affy Human Origins Array chip. (Neither Ust'-Ishim nor the Altai Neanderthal were included in the set of genomes used to ascertain the SNPs, but the San, Mbuti, Denisovan, and Australians were.)
Again, if we can't rigorously define "Basal Eurasian" and agree on a definition, how do we know what to look for?
"Basal Eurasian" becomes another kind of "Teal people", just another mathematical principal component that generates a mystery, without explaining precisely what it is without further analysis.
If anyone wants to solve the mystery of the initial population of Eurasia by (modern) AMH, they should be very interested in what is and isn't "Basal Eurasian" first, right?
If Basal Eurasians as a real ancient population ever existed, the most likely scenario is, that they were part of the AMH which isolated themselves from the ancestors of modern Subsaharans and the other Eurasian populations for a significant period of time. BE most likely resided in Northern Africa, Eastern Africa or (Southern) Arabia. From whereever they came, there might have been more than one migration into different parts of Eurasia. The latest might have had different additional influences than the first.
We need to find and sample this ancient population. You can't make a good call about their yDNA with what we have now. But G and E1b1b are the most likely candidates.
OG,
The last Mota paper with Eurasian was the corrected one. The one without Eurasian admixture was the incorrectly processed one. Secondly, K36 is based on modern clusters. There is no Natufian cluster in that run so it compensates for extra basal by using "North African". There isn't any formal stat that puts SSA in Natufians. They are no closer to Africans than any ancients, including Mesolithic Europeans.
@ Davidski:
I managed to recreate the "fateful triangle" with Ust-Ishim. Not sure why though?
At the risk of being obvious, the Ust Ishim sample here seems quite distinct in the PC3, particularly from the Iran, South Asian and Caucasus populations its closer to in PC2 (in an basically reverse pattern to PC2)-
http://imgur.com/O4aXT4g
http://imgur.com/OncFEkO
If I assume that the PCs 1,2,3 have the weights of the variance 45%, 28% and 22% respectively (and the others negligible), then you get a quite sensible position in neighbour joining clustering, with it forming the outgroup to the Euro HG clade (which would make sense as a "Crown Eurasian" with otherwise no particular relation to any any of the moderns):
http://imgur.com/ZsmlgW9
Seems OK, though I'm not sure it *exactly* tracks with the D-stat (though may not if there is different drift among the branches, which is not as relevant to the D-stat).
If you assume 55%, 30%, 10%, or have virtually all the variance sitting in PC1 and PC2, like 57%, 40% then it's a less sensible position (less consistent with D-stats), sitting with Iran_EN.
http://imgur.com/IxF9qYP
http://imgur.com/5DK5NDz
I'd assume something like Huijbregts says, where it takes that position in the PC1 and PC2 because the PC2 is measuring more of a lack of specific drift shared with the ancient Near East and Basal Eurasian and also WHG, then what makes Ust_Ishim relative more similar to the other "Crowns" (and also ancient Near East) and different from the Iran_EN+CHG (or ANE) ancestral guys is in its own PC at PC3. If you wanted to use distance in these PCs from Ust_Ishim to try and calculate any Basal Eurasianness, I think including all of PC1-PC3 would be quite important.
Not sure if this is correct without knowing what the weight of each of the PCs is though.
@Chad and everyone, about "E" being Eurasian:
Regardless of whether this seemed plausible at one point, there's strong evidence against it that's hard to refute:
1. Neanderthal and Denisovan alleles can now be reliably detected. While all Eurasian populations carry at least some of those alleles, there are African populations who are virtually exclusively Y-DNA E that are completely lacking in Eurasian Neanderthal and Denisovan alleles.
2. Mota was E1b1a2 and shows no sign of Eurasian admixture.
3. Many populations who have a very large majority of Y-DNA E have no sign of Eurasian admixture, even excluding Neanderthal alleles.
4. The most basal clades of Y-DNA E, E2 and E1a, are at their root, exclusively African.
5. The tMRCA of DE is 68,500 ybp, which is approximately the time of the Toba Caldera at 71,500 ybp. The tMRCA of E within Africa (i.e. the E1 / E2 split) is 54,200 ybp, not long after the first evidence of a surviving OOA population in the Levant and an initial pulse of Neanderthal admixture.
6. The tMRCAs of the earliest exclusively non-African E clades in E-Z830 / E-M123 / E-M84 is during or right after the LGM at about 19,200 ybp, not earlier.
7. While it does seem that mtDNA U6 was present among the Aurignacian population of Europe, there's no evidence that E originated or co-migrated with U6 from Europe to North Africa. Y-DNA E of course is much more widespread in Africa than U6.
8. As Capra pointed out, Posth et al. (2016) calculated a tMRCA for mtDNA M at 55-45 kya. mtDNA M1 is East and North African and this could correspond to an initial OOA migration from a population, some of whome later survived in the Ethiopian Highlands.
9. Because Oase1 was derived for only 3 out of 5 mtDNA N* SNPs, a lineage that's entirely extinct today, it seems very likely that N* with these 3 SNPs came into existence not earlier than about 55,000 ybp, or 13,500 years before Oase1 at 41,500 ybp. That's in fact being very generous and the tMRCA of full N* may just correspond to the tMRCA of Y-DNA F* at 48,500 ybp.
10. The Iberomaurusian culture existed in North Africa during the LGM, starting 23,000 ybp, so if they had haplogroup E-M35 that would just prove the existence of the ancestors of Eurasian E-M35 clades in North Africa during the LGM. One presentation at the ISABS 2013 Conference claimed to have found through HVR1-only mtDNA sequencing haplogroups U6, H, JT and V among the samples from Taforalt in Morocco (23,000-10,800 ybp) and Afalou in Algeria (11,000 to 15,000 ybp), but notice that the later date for Taforalt is over 10,000 years after the LGM. This study was never published for some reason even after three years perhaps because it never passed peer review.
All of this adds up to pretty overwhelming evidence of an African origin for Y-DNA E, and a much later post-LGM migration to the Levant, rather than the other way around.
BTW, I never said the Natufians had Sub-Saharan African admixture, only North and Northeast African admixture. There's no reason that these cannot be entirely native to (North) Africa.
Iranian Neolithic WC1 and the Northwest Anatolian Barcin population show no sign of the presence of any E-M35. WC1 shows no sign of any kind of African admixture whatsoever, North African or Sub-Saharan. There is certainly WHG admixture among North African Berbers today, but WC1 has no sign of WHG admixture either.
If WC1 shares any such thing as "Basal Eurasian" with Bar8, it just can't be the North African admixture brought in by Natufian E-Z830.
So how do we define "Basal Eurasian" again?
Howevere, Natufian's relative fst to africans and non-basal eurasians looks different from Iran_N's, in the same way Bedouins differ. Euri/anatolian neolithic doesn't show this.
We could take this formal stat and conclude Mbuti shows no similarity to Yoruba over Anatolia_Neolithic.
Mbuti Gorilla Yoruba Anatolia_Neolithic -0.0014 -0.7 530833
African variety is huge and Natufian sequences don't have the best coverage. More DNA needed.
The Mota / Llorente et al. (2015) study's conclusion that every African population had at least 4% Eurasian admixture was retracted:
DNA Study of First Ancient African Genome Flawed, Researchers Report - New York Times Feb 4, 2016
Error found in study of first ancient African genome
Finding that much of Africa has Eurasian ancestry was mistaken.
Nature 29 Jan. 2016
Incompatible software
Manica says that the error occurred when his team compared genetic variants in the ancient Ethiopian man with those in the reference human genome. Incompatibility between the two software packages used caused some variants that the Ethiopian man shared with Europeans (whose DNA forms a large chunk of the human reference sequence) to be removed from the analysis. This made Mota man seem less closely related to modern European populations than he actually was — and in turn made contemporary African populations appear more closely related to Europeans. The researchers did have a script that they could have run to harmonize the two software packages, says Manica, but someone forgot to run it.
Pontus Skoglund, a population geneticist at Harvard Medical School in Boston, Massachusetts, says that he was surprised by the claim that as much as 6–7% of the ancestry of West and Central African groups came from the Eurasian migrants. But after obtaining the Mota man’s genome from Manica’s team, he and his colleague David Reich carried out their own comparison and found no evidence for that conclusion. They informed Manica’s team, who then discovered the processing error.
“Almost all of us agree there was some back-to-Africa gene flow, and it was a pretty big migration into East Africa,” says Skoglund. “But it did not reach West and Central Africa, at least not in a detectable way.” The error also undermines the paper’s original conclusion that many Africans carry Neanderthal DNA (inherited from Eurasians whose ancestors had interbred with the group).
@ Chad, I think this is correct, but just for the sake of being the Devil's Advocate here, the stats they've used to check this are of the form, f4(Natufian, Other Ancient; African, Chimp) - so what kind of strength do you get on analogous stats for people today, like f4(Libyan / Egyptian, Druze / Sardinian; African, Chimp), where we actually know one of the populations as detected by ADMIXTURE to have African ancestry?
When Davidski has run the "double outgroup" stats, the pull towards Yoruba seems quite weak even for populations like Moroccan. There could be some kind of confound here.
If the population of Africa was quite structured maybe you could have a population who would look African to ADMIXTURE in the sense of "didn't go through the Eurasian bottleneck" but who were not phylogenetically closer to any of the present day Africans than Eurasians are.
Whether haplogroup E originated in Eurasia geographically is one thing, whether it originated in a population genetically closer to Eurasians than to modern Subsaharan Africans another.
A Northern African population very close to Eurasians, but isolated from the ancestors of modern Subsaharan Africans might have been part of the Basal Eurasian puzzle. We need samples and there is no way around that.
If WC1 shares any such thing as "Basal Eurasian" with Bar8, it just can't be the North African admixture brought in by Natufian E-Z830.
How can you be sure that this "North African admixture" (only proven by Eurogenes K36?) was brought in by E1b1b males found in Natufians?
In "post Natufian times" there were much more migrations from the Fertile Crescent into North Africa than the other way around. So even if some sort of relatedness of Natufians and modern North Africans is for real, you don't know the direction of the gene flow.
Maybe you are right, but the case is not closed yet.
OG,
M1 is younger than Eurasian M, from what I've seen. There is also WHG related ancestry in Iranians. They aren't much closer to ANE than WHG.
OoA is also well after Africans and Eurasians split. CT is Eurasian, therefore, DE is. The split date is usually put out close to 90-100k years ago. BE is said to diverge closer to 60kya. Just because they were in Africa doesn't mean they were Sub-Saharan. U6 and M1 are wide spread. Even in West Africa.
Basal Eurasian has no African alleles. It is no closer to Africans, but split off before Ust-Ishim. These folks getting SSA components in Admixture is about not fitting modern clusters as well as some ancients having longer Neandertal segments and more admixture.
With Natufians, Admixture drops BedouinBs to 3% SSA. This was the lower range that was expected, IIRC. Being about as close to SSA as Bedouins can then be caused by Bedouins having more Neandertal than Natufians, confounding the stat.
Amerindians were first modeled as a highly drifting offshoot of East Asians, then as a mix of East Asians and West Eurasians, and now as a mix of ANE, EHG, East Asians and Papuans. How do you differentiate between a population that branched off before others and kept admixing within itself (between its subpopulations) and a population that's a product of admixture between those "other populations"?
Amerindians are basically a mixture of East Asians and Upper Paleolithic Siberians. This is easy to show with tests that can get around their strong recent genetic drift.
So now you're stuck with having to prove that the Americas were home to deeply structured human populations during the Upper Paleolithic, that then mixed and migrated to Eurasia, and/or mixed as they migrated to Eurasia.
These populations have to include the ancestors of East Asians plus Onge and Papuans, the ancestors of at least a couple of different types of Upper Paleolithic Siberians represented by Ust-Ishim and MA1, and the ancestors of various types of European Upper Paleolithic foragers.
This is idiotic. There's nothing to discuss.
"Amerindians are basically a mixture of East Asians and Upper Paleolithic Siberians. This is easy to show with tests that can get around their strong recent genetic drift."
This is baloney.
Amerindians are the most structured of all modern human populations. This has been proven by all the tests. I don't really need to do anything.
"These populations have to include the ancestors of East Asians plus Onge and Papuans, the ancestors of at least a couple of different types of Upper Paleolithic Siberians represented by Ust-Ishim and MA1, and the ancestors of various types of European Upper Paleolithic foragers."
That's what we would expect from a population that expanded to colonize the rest of the world from a single isolated location. Your model whereby all of those folks, after having settled in their respective regions, began streaming toward the New World is sheer nonsense.
"This is idiotic. There's nothing to discuss."
This is because you have no logic, or method, or real data. You are just an amateur playing computer games.
Amerindians are not deeply structured you clown.
They're basically a mixture of East Asians and Ancient North Eurasians, with strong local founder effects and drift, resulting in high Fst genetic distances between some of their populations.
You can have high Fst between populations that are basically made up of the same ancient ancestral groups. This is not a contradiction.
Just because this was misinterpreted in some old paper that got stuck in that thick senile head of yours doesn't make it relevant.
Hi Davidski. Could you to make D-Stats to followed sets
Iran_EN Iran_ChL Eastern_HG Mbuti
Caucasus_HG Iran_ChL Eastern_HG Mbuti
Iran_EN Iran_ChL Caucasus_HG Mbuti
Caucasus_HG Iran_EN Eastern_HG Mbuti
Iran_EN Iran_ChL Caucasus_HG Mbuti
Yamnaya-Catacomb_Kalmykia Yamnaya_Samara Eastern_HG Mbuti
Yamnaya-Catacomb_Kalmykia Yamnaya_Samara Iran_ChL Mbuti
Yamnaya-Catacomb_Kalmykia Yamnaya_Samara Iran_EN Mbuti
Yamnaya-Catacomb_Kalmykia Yamnaya_Samara Caucasus_HG Mbuti
Eastern_HG Yamnaya_Samara Iran_ChL Mbuti
Eastern_HG Yamnaya_Samara Iran_EN Mbuti
Eastern_HG Yamnaya_Samara Caucasus_HG Mbuti
Caucasus_HG Iran_ChL Yamnaya_Samara Mbuti
Iran_EN Iran_ChL Yamnaya_Samara Mbuti
Caucasus_HG Iran_EN Yamnaya_Samara Mbuti
I want to check my suspicions that it is not Iran_Chl influenced on Yamnaya, and EHG-liked influenced on Iran_Chl. The absolute distance between the Eastern_HG and Caucasus_HG, Iran_EN, Iran_ChL decreases with time.(?)
@David
Amerindians are not deeply structured you clown.
They're basically a mixture of East Asians and Ancient North Eurasians, with strong local founder effects and drift, resulting in high Fst genetic distances between some of their populations.
You can have high Fst between populations that are basically made up of the same ancient ancestral groups. This is not a contradiction.
Yes. Consider the Kalash, who basically have the same ancestral components as their neighbors, but because of their high drift they show relatively high Fst distances to their neighbors and acquire their own component in ADMIXTURE analyses.
@Pegasus
It's pretty clear that Yamnaya doesn't have any ancestry from Chalcolithic Iran, because it doesn't have any mtDNA from the south Caspian.
https://1.bp.blogspot.com/-YquIP_NuqEA/V8z9vO5RkOI/AAAAAAAAE3E/sAmtYvdSCIEYQqGW3I0l8IK8AURGk-MLQCLcB/s1600/Steppe_EMBA_vs_Iran_ChL%252BN.png
Here are your D-stats...
Iran_Neolithic Iran_Chalcolithic Eastern_HG Mbuti -0.0172 -4.467 754163
Caucasus_HG Iran_Chalcolithic Eastern_HG Mbuti 0.0126 3.57 930989
Iran_Neolithic Iran_Chalcolithic Caucasus_HG Mbuti -0.0084 -2.229 813500
Caucasus_HG Iran_Neolithic Eastern_HG Mbuti 0.0286 6.654 792198
Iran_Neolithic Iran_Chalcolithic Caucasus_HG Mbuti -0.0084 -2.229 813500
Yamnaya_Kalmykia Yamnaya_Samara Eastern_HG Mbuti -0.0099 -3.879 979362
Yamnaya_Kalmykia Yamnaya_Samara Iran_Chalcolithic Mbuti -0.0083 -4.171 974318
Yamnaya_Kalmykia Yamnaya_Samara Iran_Neolithic Mbuti -0.0051 -2.085 835411
Yamnaya_Kalmykia Yamnaya_Samara Caucasus_HG Mbuti -0.0069 -2.964 1071111
Eastern_HG Yamnaya_Samara Iran_Chalcolithic Mbuti -0.0214 -7.643 924154
Eastern_HG Yamnaya_Samara Iran_Neolithic Mbuti -0.0247 -6.906 788500
Eastern_HG Yamnaya_Samara Caucasus_HG Mbuti -0.0262 -7.628 1016162
Caucasus_HG Iran_Chalcolithic Yamnaya_Samara Mbuti 0.017 6.289 1004360
Iran_Neolithic Iran_Chalcolithic Yamnaya_Samara Mbuti -0.0165 -5.597 810921
Caucasus_HG Iran_Neolithic Yamnaya_Samara Mbuti 0.0322 9.729 853074
@German,
Amerindians have very low mtdna and y DNA variation. Africa has the highest. This is a big reason why Africa is considered the location humanity originated.
Amerindian mtdna/T DNA is nested within Eurasia diversity, because Amerindians are from Eurasia(and ultimately Africa). Their mtdna is clearly east Asian. Their y DNA is a brother to what MA1. This is consistent with their autosomal DNA.
Nothing can explain Ametidian DNA if they never left America and Eurasians orignatws in America. You're arguments are as reasonable as.someone who argues the sky is green. Your artogence added on top of that makes you annoying. Either you're stupid or being forced to post these things by threat of death. Either way something screwef up is going on with you.
@Davidski
"resulting in high Fst genetic distances between some of their populations."
They have world-highest Fst. They follow Neandertals and Denisovans in the level of substructure and homozygosity. Then everyone else. Their Fst was probably even higher 10,000 years ago.
"Just because this was misinterpreted in some old paper that got stuck in that thick senile head of yours doesn't make it relevant."
It's al over the "new" papers. My good old reading skills are just better than yours. And I'm likely younger than you, pops!
"They're basically a mixture of East Asians and Ancient North Eurasians."
Saying it 100 times doesn't mean proving it. Try again. How can you derive a population with low heterozygosity from two (!) populations (East Asians and West Eurasians) with high heterozygosity?
@Samuel Andrews
"Africa has the highest. This is a big reason why Africa is considered the location humanity originated."
There are two kinds of diversities: intergroup diversities are highest in America, intragroup - in Africa. Autosomally. The reason Africa is "diverse" is because Eurasians who colonized it absorbed an "archaic" substrate. There were no archaics in America, so Amerindians are not as "diverse" as Africans. Learn some basics and then use some smarts!
@Onur
"Consider the Kalash, who basically have the same ancestral components as their neighbors, but because of their high drift they show relatively high Fst distances to their neighbors and acquire their own component in ADMIXTURE analyses."
Why can't it be an old drifting populations that just stayed drifted, not became drifted recently? in the Pleistocene, populations were small and had high Fst. Kalash is a relic. Why not consider this possibility?
Kalash is a relic. Why not consider this possibility?
Because it's not true...duh. They're a mixture of Neolithic farmers from Iran and Bronze Age herders from the Eurasian steppe.
This is as easy to demonstrate as the fact that Amerindians are a mixture of East Asians and Ancient North Eurasians.
You're stuck with this reality. There's no way around it.
@Davidski
"They're a mixture of Neolithic farmers from Iran and Bronze Age herders from the Eurasian steppe."
Tons of populations can be described in those terms. What makes Kalash special is that, even on one of your charts in this post, they are shifted toward the Karitiana pole compared to other central and South Asians. Is it recent drift that makes them more Karitiana-like?
"You're stuck with this reality. There's no way around it."
You are good at proclaiming and then repeating. That's not how science works. Try different models and you may find a better one to explain reality. You don't need to get stuck with just one model.
You can get low heterozygosity from a small founder population. This is common in hunter-gatherers and mixed pops, like the Boncuklu Anatolians. Of course Native Americans are mixed. Full-blood Native Americans are only YDNA Q and C. Where's the G, T, L, R, I, K, etc? MtDNA is even worse, as strictly East Asian. You're a crackpot that is the butt of jokes, all around. Get a clue! I swear, you have to be a troll. No one can seriously be this stupid.
German
So what do you make of this paper by Reich ?
https://genetics.med.harvard.edu/reich/Reich_Lab/Welcome_files/SkoglundReich2016_Americas.pdf
@Chad Rolfsen
"MtDNA is even worse, as strictly East Asian."
Sure thing, if you call West Asia where hg X is found "East Asia". And also if you call Oceania, where hg B is found, "East Asia." mtDNA hg A is very rare in East Asia, but it's all over America. The only reason you refer to it as "East Asian" instead of "Amerindian" is because you are ignorant and unintelligent.
"Of course Native Americans are mixed. Full-blood Native Americans are only YDNA Q and C. Where's the G, T, L, R, I, K, etc?"
Amerindians don't have any principal "East Asian" Y-DNA markers such as N and O. Those are all over northern and southern East Asia but they are not in America. But you do raise a good point: if America was so mixed (East Asia, West Eurasia, Papua New Guinea, etc.) and colonized recently, then we would have seen N, O, G, T, L, R, I, K there just like we see it in the Old World. But we don't.
So think again! The only reason I don't single you out as troll is because there are too many of you around the web - illogical, undereducated, useless...
I met Pontus. And I like him, although he's not an anthropologist - they rotated him out of Drosophila studies. I don't mind people looking at things the "old" way - they in fact bring up great new data, so I always read them with genuine interest. I agree with them that whole genome analysis is better than mtDNA or Y-DNA. But when they say "admixture between East Asians and West Eurasians happened before diversification in America," they imply that this admixture happened in Siberia but in Siberia we don't have a better candidate for this admixed population than any Amerindian population. So "admixture" must have happened in America. If modern Amerindians are admixed, this admixture must have involved ancient New World subpopulations, not "West Eurasians" and "East Asians." It's a terminology problem that has huge implications for the interpretation of events. Populations that stay behind have a higher chance of readmixing after their divergence than populations that colonized a new territory and continued to separate into increasingly more divergent populations beyond repair. The latter are the "East Asians" and the "West Eurasians" in Raghavan/Reich/Skoglund's model. The reason they are so distinct in PCA plots is because they migrated in two opposite directions. The reason why Amerindians are intermediary between them on PCAs (but still divergent from Ust-Ishim) is because it's from America that "East Asians" and the "West Eurasians" moved to their respective Old World regions and they carried with them diversity that had already been present in ancient New World subpopulations.
German, you're exceptionally woeful at population genetics.
I've never seen anyone this utterly hopeless.
Ohhh... X2. I did forget about that 3% of Native Americans. Too bad X2 in Native Americans isn't ancestral to other X2. They all only split about 20kya too. But, again, where is the rest of CF or DE in Native Americans? We're all dying to know. Boy, if only I could show something to refute your claims, like stats....
Oh, wait... right here..
So, Native Americans are the ancient source, but not mixed between MA1 and a Han like population? Really? Explain why Native Americans are closer to both Han and MA1 than they are to the 45kyo Siberian, Ust-Ishim? How come Native Americans are no closer to Ust-Ishim than Han or MA1, or even Loschbour?
result: Chimp Karitiana MA1 Han 0.0325 5.913 24006 22494 432449
result: Chimp Karitiana MA1 Ust_Ishim -0.0943 -13.537 20903 25255 431519
result: Chimp Karitiana Han Ust_Ishim -0.1272 -25.542 28296 36543 601303
So, Native Americans didn't breed into Africans, but descendants of Native Americans, modern Eurasians did? Really? Good luck finding a Eurasian significantly closer to Africans..
result: Chimp Yoruba Natufian Karitiana 0.0044 0.951 12149 12042 271523
So, if Native Americans are the ancestors of everyone, why aren't they equally related to everyone?
result: Chimp Ust_Ishim Loschbour Karitiana 0.0035 0.609 30237 30027 595647
result: Chimp Han Loschbour Karitiana 0.1103 26.650 34785 27873 596895
Oh, and Papuans aren't related to East Asians? Really?
result: Chimp Papuan Karitiana MA1 -0.0484 -8.062 19900 21923 432448
result: Chimp Papuan Karitiana Han 0.0109 3.403 29061 28436 602688
result: Chimp Papuan Han MA1 -0.0547 -10.093 21089 23530 432448
result: Chimp Papuan Han Loschbour -0.0586 -13.516 29237 32877 596893
result: Chimp Papuan Han Natufian -0.0955 -19.140 13023 15773 271522
Most damning of all, Native Americans themselves to be a mix of a Han and MA-1 like population, with only slightly more Han related ancestry.
result: Chimp Karitiana MA1 Han 0.0325 5.913 24006 22494 432449
result: Chimp Han MA1 Karitiana 0.1118 22.627 24006 19179 432449
result: Chimp MA1 Han Karitiana 0.0796 18.960 22494 19179 432449
Look at that. Related to Han, to the exclusion of MA1, but also related to MA1, to the exclusion of Han. Doesn't sound like a pop ancestral to MA1 and Han, but one mixed with both. But, of course you will just say things like unintelligent, rather than actually being able to make a coherent argument against this. Please, offer us your scientific data that refutes this. By the way, fringe archeology papers are not "scientific proof". You need evidence to back your claims, so let's see it!
@Chad
"Too bad X2 in Native Americans isn't ancestral to other X2. They all only split about 20kya too."
All of X and Amerindian A2 share a transition first identified by Maca-Meyer 2001. This makes the presence of both X and A in the New World (shared by West Asians in the case of X and East Asians in the case of A but not by both) understandable. It's a situation exactly parallel to the closer relationship of East Asians and West Eurasians to Amerindians, but not to each other.
"So, if Native Americans are the ancestors of everyone, why aren't they equally related to everyone?"
Why would they be "equally related to everyone"? Some Africans are equally related to everyone only because they carry an archaic substrate (under my model). If a population stays behind and its subpopulations admix into each other, then it all depends on how they admix: some populations will end up being closer to a third population than others.
"How come Native Americans are no closer to Ust-Ishim than Han or MA1, or even Loschbour?"
They share more of Neandertal chunks in UI than East Asians or West Eurasians do. Oase is closer to Amerindians than to East Asians or West Eurasians. Altai Neandertal is closer to Amerindians than it is to East Asians and West Eurasians. MA-1 is closer to Amerindians than to anyone else. Tianyan shares more nucleotides with Karitiana than with anyone else. There's something here, don't you think?
"Look at that. Related to Han, to the exclusion of MA1, but also related to MA1, to the exclusion of Han. Doesn't sound like a pop ancestral to MA1 and Han, but one mixed with both."
Oh, and in your model Han and MA-1 are unrelated populations that descended from God in two separate breaths?
@Davidski
"you're exceptionally woeful at population genetics."
Again, prove it, don't say it! You doubt that isolated, ancestral populations continue to admix after daughter populations migrated out? You believe that ancestral populations who remained in close proximity to each other (by definition) kept diverging without gene flow, while daughter populations who diverged (by definition) must admix with each other in some remote corners of the world?
You doubt that isolated, ancestral populations continue to admix after daughter populations migrated out?
What evidence do you have that the ancestors of European Upper Paleolithic populations, Siberian Upper Paleolithic populations, and East Asians (including Onge and Papuan) lived in the Americas during the Upper Paleolithic?
And what evidence do you have that they migrated out?
Also, how do you put that evidence, if you actually have any, into a meaningful context?
Where did they live more or less? Why and how did they migrate out? Where did they mix to produce the ancient Eurasian genomes that we have available? Why didn't most of them, apart from the Ancient North Eurasians and East Asians, leave any significant evidence of their presence in the Americas in modern genomes?
Your theory is, at the very best, fact free. Why don't you try and put some meat on those bones?
@Davidski
"Your theory is, at the very best, fact free. Why don't you try and put some meat on those bones?"
My theory is the only one out there that's built on a solid multi-field, anthropological foundation. It's principal weakness is its explicit disregard for the way people have been thinking about "the peopling of the Americas" from the times of the missionaries. It's main assumption is that archaeology builds its evidence over a long period of time - not over 200 years.
Let's begin by having you read my book and my site. Why did I write them you think - to later copy-paste everything in your comments section?
The principal weakness of your theory is not that you don't have any evidence for it yet. It's that all of the evidence available totally contradicts your theory.
And to say that you have a poor grasp of the relevant population genetics data is being very generous indeed.
You're not even operating in the real world here.
@Davidski
"all of the evidence available totally contradicts your theory."
Show me at least one archaeological culture in Northeast Asia that can explain the most ancient Paleoindian cultures. We've been looking for them for 200 years. Nothing has been found. So there's no archaeological evidence suggesting that Amerindians came from Asia. It's a fact.
"You're not even operating in the real world here."
Again - statements without proof. The fact is that I have two doctorates and you have zero. That's what a fact looks like. Now go from there an correct your theory about who has a grasp of population genetics.
The fact is that I have two doctorates.
But you're also totally full of shit. So your doctorates mean nothing.
Really, Davidski? I thought you had one (at least)! But I don't understand why we can not discuss with more respect (perhaps I am too a little respectful, but only when I say the truth I hope). I knew Dziebel on the Dienekes blog, and, even though I am not for an Out Of America, so far (only for an out of Italy you know, but for only these last 15000 years), I thought that his ideas merited to be discussed as any other. It seems to me that the paper of Mr Reich and his alius has many caveats about his theories, those that for you are gold. But isn't him the king of the kurganists? But isn't he searching for the ancestors of the Villabrunas amongst the black Natufians? And what has he to say about other compatriots of his who said that Ashkenazim are 70% European (45% Italian), except lowering after the percentage to 60%? Are the creators of DNA.land amongst his fellows? You know that I inquire only the Y and the mt, and I'll say which are those percentages through them. Costa et al., so long after me, said that the mt is even 90% European. I am doing my best also about the Y, also about hgs. J and E. I have some Reichs possible ancestors of his: one is R-M269* (overwhelmingly in Sardinia /Italy), another is an R1a-Levites (those that Behar thought Old Jewish, aftern said that he was wrong, and now has promised coming to his first step), another a modern European R-M269, and very likley an J and an E. If he send me his data, I may do an expertise... for free.
David your response is useless. why don't you counter with some logic and data even 1 or 2 sentences or keep silent.
Ok David I had not read more recent posts when I wrote that but the trend seems similar more insult centric with little meat.
Chad is countering
First of all, I have zero tolerance for being insulted on my own blog. Anyone who does this is going to get abused by me. I couldn't care less whether you or anyone else finds this acceptable.
Secondly, I have zero tolerance and no time for pseudo-science and hallow arguments. There's absolutely no genetic evidence backing an Out of America theory, so what's to discuss and debate?
@ Davidski
Of course no one of us believes in an Out of America, but the same theories of Reich changed during the time, and the last paper posted above has many interesting caveats. Who studies Native American cultures as Dziebel by an anthropological point of view may have many doubts that things went as was firstly said, all in these last 12 or 13000 years. Now the same Reich hypothesizes a long stop in the Beringia, many waves of migration, perhaps a little group through South Pacific. If you study Native American languages you don't believe any more to your genetic convictions: American languages are more varied than the African ones. You should think that A00 was unable to speak and that language was brought from the back migration from Asia.
@German Dziebel
Why can't it be an old drifting populations that just stayed drifted, not became drifted recently? in the Pleistocene, populations were small and had high Fst. Kalash is a relic. Why not consider this possibility?
The Kalash are only relic in the sense that they lack the small Islamic-era admixtures seen among their Muslim neighbors. But their major components are the same as the ones of their Muslim neighbors and have a similar distribution. They, like their Muslim neighbors, are characteristically post-Bronze Age and thus post-Steppe migrations in their genetic makeup.
@Gioiello
Mr Reich and his alius has many caveats about his theories, those that for you are gold. But isn't him the king of the kurganists? But isn't he searching for the ancestors of the Villabrunas amongst the black Natufians?
When did Reich ever attribute the ancestry of the Villabrunas to Natufians? Also, why do you call Natufians black? Natufians have no discerbable Sub-Saharan ancestry, they are perfectly within the Caucasoid genetic variation.
@Gioiello
If you study Native American languages you don't believe any more to your genetic convictions: American languages are more varied than the African ones. You should think that A00 was unable to speak and that language was brought from the back migration from Asia.
Languages have very little relevance when we are dealing with time depths of tens of thousands of years. Also, do note that the Neolithic and post-Neolithic migrations erased most of the pre-Neolithic linguistic diversity of Africa. The Bantu expansion is the biggest one of those migrations, but not the only one.
Onur, of course I am a literate, a polemist, thus take my writing with a grain of salt. A literate speaks for metaphors and a polisemic language: 1) Reich, in an interview, said that he didn’t know where Villabruna came from, perhaps the Balkans… but that he was in Italy 14000 years ago didn’t mean anything? I wrote that Villabruna has threefold the age of Abraham, if we believe the Bible, if we believe documents perhaps fourfold. 2) Natufians weren’t Africans, we know, but look at the bedouin posted by Passariello (Passa on Anthrogenica), who, being an Italian hg. E, thinks to be closer to Natufians than to me. I didn’t convince him in saying that his E (perhaps E-V65, but actually M78+ V12- for 23andMe) may be in Italy even before my R1b.
About African languages you are right only in part. Even if we erase the Bantu expansion, Africa hasn’t the complexity of America I think and all the Southern Africa where Bantu expanded spoke clicks languages that from Wa-Sandawe and Hadza to South didn’t seem so differentiated even though spoken in a vast territory, so we may doubt about their ancientness, certainly not to compare to Y A00 and mt L0 etc.
Gioiello, Reich thinks the ancestry of Villabruna and all other WHG might have largely come from somewhere in the east due to their closer genetic connection to Near Easterners relative to older Western and Central Europeans. Due to their lack of Basal Eurasian ancestry, he thinks that ancestry is more likely to have come from the Balkans than somewhere in the Near East. These are all valid points and I agree with Reich on them.
As for Passariello, I know him from Facebook, he is a nice guy. I think he is talking about Y-DNA lines and not about autosomal genetics when he is talking about being closer to Natufians than to you. If so, he is right since you, being R1b, are much more distant to him, who is E, on Y-DNA than the average E Natufian is. The Natufian E is from the Caucasoid clade of E and Bedouin E is also overwhelmingly Caucasoid E. Autosomally also Bedouins are within the Caucasoid genetic variation despite their small Sub-Saharan admixture.
Gioiello, there is no established connection between the click languages of Eastern and Southern Africa. Using clicks is not enough to regard them as related.
No sane person is saying that any modern language family comes from deep Upper Paleolithic times. In fact, I have been saying that languages are unreliable tools in tracing ancestries from Upper Paleolithic times.
Onur, you know that my analyses took into account above all the Y and the mt, and not the autosome. There is a long background about that, from when I did even deCODEme and 23andMe (the unique autosome I believe in and use: through it I found the parents to many adopted persons), till when I was found more than 20 % Ashkenazi and many believed I hurried to a Bar Mitzva. Of course I have always said that people went and come from everywhere, but perhaps I was the only one to say that R1b was in Italy before the Younger Dryas (and the fine has to come yet as we say in Italian). Of course Villabruna was a WHG and run the Siberian corridor, but he was in Italy 14000 years ago, his clan wintered in Italy during the Younger Dryas, migrated to Samara, Anatolia and Middle East. M269 and L51 (not speaking of M335, V88, M18, V35) were in Italy, I think. I think it is enough for speaking of an "Italian Refugium".
Of course some form of language began in Palaeolitic. Very likely Villabruna spoke a form of Nostratic and the link with Uralic and Altaic etc. was due to the Siberian corridor. You know that I think that Indo-European comes from there and that satem languages were brought to Samara from R-L23 and Anatolian languages from the migration to Anatolia. All the centum languages started from Italy or the Alps or Central Europe and very likley they are older than people thinks. This is my hypothesis, of course not demonstrable, but aDNA will say much about that too.
Wikipedia says: "The Sandawe language is a tonal language with clicks, apparently related to the Khoe languages of southern Africa". That Wa-Sandawe and Hadza were related with Khoe languages was demonstrated from Alfredo Trombetti more that a century ago. Wikipedia writes also: "In the late 20th century Hadza was included in a proposed Khoisan language family, largely on the basis of its use of clicks, but this classification is no longer accepted".
Re: the current Lazaridis model, EHG is a mix of WHG and a population on an Onge-Han cline beyond Han, we could call this ANE-related. East Asians and Native Americans in varying proportions are mixes of this ANE-related population and Onge, which would make them the closest thing we have for proper ancestral populations there. Of course there may be more extreme ANE than ANE that contributed to EHG, and a more extreme Onge-like population than Onge yet to be found which would make EHG's ANE and Onge mixes too...
http://www.nature.com/nature/journal/v536/n7617/fig_tab/nature19310_SF7.html
@Gioiello
You know that Joanna Nichols argued that the settlement of America must be much older than 15 000 years, because of the diversity of languages, but she did not say that the settlement of Africa must be young. Africa is mostly covered by not-too-old spreads, which she argues just from linguistics, but it is confirmed by genetics and archaeology.
You can have a Y haplogroup which diverged 300 000 years ago and no one will notice, there can be many different alleles in your village, but you cannot speak many different languages. The Bantu are mostly in subclades of E-V43, but we can find many other haplogroups at low frequency. But everyone speaks the language of V43. (Of course, now, in the New World, almost everyone speaks the language of M269, since the conquerors with ships and horses and iron came 500 years ago.)
We must also compare apples to apples: Greenberg, who put all of Africa into four families, put all of the New World into three. Using the method of Starostin (who unites all of Khoisan) Leschiner put the origin of all Amerindian at 11 900 BC. If we use the criteria normal for the New World instead we will not unite Khoisan, or Niger-Congo or Nilo-Saharan, they will all be many different families.
German,
The Neandertal chunks mean nothing when Native Americans are no closer to Ust Ishim than anyone else. So, it's irrelevant.
So, again, bring forth genetic evidence that Native Americans are anything but a mix, like all of us. Also, I'm still waiting on you to provide evidence of other CF and DE lineages in full-bloods.
Teaser alert! You can't do either. Your ideas don't stand up to genetic evidence. Changing views based on evidence is science. Blindly holding to horseshit is belief. Beliefs aren't based on facts, like your views. Perhaps that horseshit god comment should be forwarded to you.
So, how about providing scientific data instead of just insults to someone that is an evolutionist and atheist, by the way.
Christ sake. We don't even need aDNA or PCAs
There's no sign of human habitation in America before 14 kya
Let's move on
Here’s an interesting TreeMix output; which is reported as the bootstrap probability (%) of gene flow to be 86%.
It doesn't show Amerindians to be the recipient of Mal’ta MA1 admixture.
http://www.nature.com/jhg/journal/vaop/ncurrent/fig_tab/jhg2016110f4.html#figure-title
I wouldn't call it interesting. I'd say it's a pretty poor representation of the fact that Karitiana and Europeans share ancestry that is not East Asian or even exactly like MA1, but more like AG3.
@ capra internetensis
I thank you very much for your inputs. You are a linguist. I am not. I am 68 an retired. Perhaps it is too late for taking again my linguistic interests. I did other during my life. I am from a peasant family. Bad studies. Days in the bars. I hadn't an academic carreer. No communist, no believer. I've been for forty years a teacher of Italian, Latin, History, Geography in the High School. No time for deep studies... but I wrote 4 books of poetry and about critics. Only in these ten years genealogy and genetics as an amateur. But I wrote a paper (not published) before my thirties: Chinese and Sino-Tibetan and Indo-European for "6" from *dhlekhl- that the great Romano Lazzeroni of the "Scuola Normale Superiore" of Pisa read. You'll understand. Of course I got all the works of Alfredo Trombetti. About clicks languages he thought they belonged to the camitic stock, thus come from North through the Rift.
@Niineta, OTOH we have several TreeMix tests showing the ANE flow into Natives.
http://oi63.tinypic.com/33yjpqx.jpg
Even if the ANE isn't exactly like MA-1 it's there.
Shaikorth said... @Niineta, OTOH we have several TreeMix tests showing the ANE flow into Natives.
http://oi63.tinypic.com/33yjpqx.jpg
Even if the ANE isn't exactly like MA-1 it's there.
Of course you do. . . yet TreeMix produced two very contradictory results for the same population.
So it would appear that ANE in Amerindians is more of a result of who decides which direction the gene flow occurred.
The original report of gene flow into Amerindians was to show Amerindians derived 1/3 of their ancestry from West Eurasians; which was a load of crap. So how much validity does the direction of gene flow have now?
Transferring the gene flow from another population; doesn’t correct the false assumption the model was based on.
Modeling Amerindians as a conglomerate of Asian populations seems to be problematic. Maybe that's a clue.
No one has decided which direction the migration edge goes, these are not supervised tests. The study which has the TreeMix run you linked mentions that the gene-flow should be from MA-1 to Karitiana too, not vice versa.
http://www.nature.com/jhg/journal/vaop/ncurrent/full/jhg2016110a.html
"The directionality of this gene flow event was as expected, however, the inferred gene flow from Karitiana to Mal’ta MA1 (Supplementary Results S2; Supplementary Figures S16b-j and S17c-j), was in the reverse direction to what was reported by Raghavan et al.23 who used a much larger sequence data. This result might have been caused by using a relatively small SNP data set. We therefore used only modern human data, and reran TreeMix. We used a much larger 0.7 million SNP loci data. However, the resulting tree showed some anomalous gene flow directions; from Papuans to Denisovans (bootstrap probability=98%) and CEU to Papuans (bootstrap probability=86%), whereas the tree topology was consistent with that of Figure 4 (see Supplementary Figure S19). The reason for this anomaly may be that some filtering steps between our analyses and Reich et al.16 and Meyer et al.19 are different, and/or homozygous diploid genotypes in individual genome (that is, Denisovan, Papuan and so on) were used instead of their original genotype, though gene flow from Mal’ta MA1 to Karitiana correctly appeared in both all sites and transversion only when we add Mal’ta MA1, Karitiana and Ust’-Ishim to the large SNP loci data (data not shown)."
@Niineta
So it would appear that ANE in Amerindians is more of a result of who decides which direction the gene flow occurred.
Nope.
TreeMix is just for exploring data and illustrating models that also have backing from direct formal stats.
Look at these D-stats. Karitiana are very clearly a mixture of something like Onge or Han with something like MA1 just like my PCA in the latest post show. Most of the Z scores at the end here are huge and in favor of Karitiana.
Mbuti Onge MA1 Karitiana 0.0393 9.646
Mbuti MA1 Onge Karitiana 0.0839 18.34
Mbuti Karitiana MA1 Onge -0.0448 -9.717
Mbuti Han MA1 Karitiana 0.104 26.254
Mbuti MA1 Han Karitiana 0.0799 20.69
Mbuti Karitiana MA1 Han 0.0243 5.268
And for all intents and purposes, MA1 is a West Eurasian forager with no Amerindian input, because he has no obvious East Asian input, which, as per above, Amerindians do.
Mbuti Onge MA1 Loschbour -0.0066 -1.365
Mbuti MA1 Onge Loschbour 0.0614 11.812
Mbuti Loschbour MA1 Onge -0.068 -12.529
Mbuti Han MA1 Loschbour -0.0089 -2.034
Mbuti MA1 Han Loschbour 0.0502 9.748
Mbuti Loschbour MA1 Han -0.059 -11.329
So which way did the gene flow go and which of the TreeMix graphs are correct? Awaiting eagerly your reply.
Davidski said... “So which way did the gene flow go and which of the TreeMix graphs are correct? Awaiting eagerly your reply.”
By Shaikorth’s report, the TreeMix output indicated gene flow from Karitiana to Mal’ta MA1; which was rejected.
"The directionality of this gene flow event was as expected, however, the inferred gene flow from Karitiana to was in the reverse direction to what was reported by Raghavan et al.23 who used a much larger sequence data.”
They then created an acceptable direction (from Mal’ta MA1 to Karitiana) but to the detriment of the gene flow direction of the other population’s. (eg. Papuans to Denisovans and CEU to Papuans).
The only difference with this report and the Japanese report is the Japanese did not reject the direction of the gene flow. Since both obtained the same results; who is wrong?
Davidski said... “And for all intents and purposes, MA1 is a West Eurasian forager with no Amerindian input, because he has no obvious East Asian input, which, as per above, Amerindians do.”
A number of studies have reported Amerindians to have input from Southeast Asia. Both Amerindians and East Asians have a connection to Southeast Asia but no study has ever reported Amerindians as the descendants of East Asians.
The only population reported to have East Asian input are the Artic (Inuit) populations; who are unrelated to the Amerindians. In David Reich study (2012), Reich had to rename the Amerindians “First Americans” to show there were two separate populations in the Americas and only one had received input from East Asians.
You can juggle your figures all you want; but there is no study which has reported East Asian input into Amerindians as you claim.
Lazaridis et al. 2016 models both East Asians and Native Americans as mixes of ANE and Onge in the supplements, only proportions of these vary.
This is Skoglund & Reich's model of the ancestral populations of Native Americans (doesn't contradict the Lazaridis model because of how Han are modeled there)
https://twitter.com/pontus_skoglund/status/743146170122375168
Coincidentally Broushaki et al. finds Karitiana, Surui etc. can be modeled with significant Han based on their total variation distance from high coverage ancient genomes and modern Han & Africans.
http://science.sciencemag.org/highwire/filestream/681447/field_highwire_adjunct_files/4/Table_S25.xlsx (subtable 3)
@Niineta
You seemed to have missed these D-stats?
Mbuti Onge MA1 Karitiana 0.0393 9.646
Mbuti MA1 Onge Karitiana 0.0839 18.34
Mbuti Karitiana MA1 Onge -0.0448 -9.717
Mbuti Han MA1 Karitiana 0.104 26.254
Mbuti MA1 Han Karitiana 0.0799 20.69
Mbuti Karitiana MA1 Han 0.0243 5.268
Mbuti Onge MA1 Loschbour -0.0066 -1.365
Mbuti MA1 Onge Loschbour 0.0614 11.812
Mbuti Loschbour MA1 Onge -0.068 -12.529
Mbuti Han MA1 Loschbour -0.0089 -2.034
Mbuti MA1 Han Loschbour 0.0502 9.748
Mbuti Loschbour MA1 Han -0.059 -11.329
I totally forgot about this discussion. Sorry, work must have carried me away.
@Capra
"You know that Joanna Nichols argued that the settlement of America must be much older than 15 000 years, because of the diversity of languages, but she did not say that the settlement of Africa must be young."
Joanna has never really tackled African diversity. That's not her focus. Cavalli-Sforza (and several geneticists after him), on the other hand, did map his early gene trees onto Greenberg's and Ruhlen's linguistic macrophyla, and Greenberg did gear his Amerind classification to the known archaeological horizon of the peopling of the America. Megalocomparison in linguistics has not been accepted. This means Cavalli-Sforza found similarities between gene trees and flawed linguistic classifications.
"If we use the criteria normal for the New World instead we will not unite Khoisan, or Niger-Congo or Nilo-Saharan, they will all be many different families."
Not as many as in the New World. Post-Greenbergian African splitters count 20 stocks in Africa. Contrast this with 140 stocks established by splitters in America. Also America scores much higher than Africa on the number of isolates. Plus factor in all the extinctions of poorly documented languages in the New World since-1492. And America has known plenty of large-scale language expansions, including foraging ones (Athabascan). So it's (speculative) probability of language loss is much higher than that of Africa.
@Gioello
Bingo on African vs. New World linguistic diversity.
@davidski
"And for all intents and purposes, MA1 is a West Eurasian forager with no Amerindian input, because he has no obvious East Asian input, which, as per above, Amerindians do."
Poor logic. If West eurasians and East Asians split from Amerindians and diverged (as out-of-America postulates), why would MA-1 have any East Asian component? They diverged, remember. Divergence means loss of similarity over time. East Asians, too, have Amerindian input but they diverged from West Eurasians and have no similarity to them outside of their Amerindian heritage.
@Capra and Gioiello
I think any linguist (Nichols or else) would agree with one thing: historically the linguistic landscape we see in Africa and Europe (few large language families dominate) evolved (via different processes) from a linguistic landscape similar to the one found in America and Papua New Guinea (many small language families and isolates), not the other way around.
There's a clear genetic parallel to the linguistic diversity landscapes: intergroup diversities are highest in Amerindians (followed by Papuans and preceded by Neandertals and denisovans). Small isolated demes would result in precisely this kind of genetic diversity and in the kind of linguistic diversity (many small language families and isolates) we discussed. What we see in Africa both in terms of high intergroup diversity and low linguistic diversity are both derived states for humans.
@ German Dziebel
Then what you say now disproves what I thought you thought before, i.e. that a great diversity of the languages families should demonstrate an older origin.
All,
BTW, here's a TreeMix that shows a migration edge from Karitiana to Malta and modern West Eurasians, not the other way around, with bootstrap values at 86%.
http://www.nature.com/jhg/journal/vaop/ncurrent/fig_tab/jhg2016110f4.html#figure-title
@Gioiello
How so? I thought what I wrote supports precisely the conclusion that a great diversity of languages (and great intergroup genetic diversity) demonstrates an older origin. (High intragroup genetic diversity, on the other hand, has nothing to do with ancient origin.)
Post a Comment