K33
07-01-2020, 06:14 AM
A quick search of this forum reveals that German Dziebel has been mentioned in passing as something of a laughing stock here. Dziebel was uncompromising in his assertion that modern humans originated in the Americas, and advanced a number of speculative and some rather dubious theories (http://anthropogenesis.kinshipstudies.org/anthropology-of-human-origins/out-of-america-family-of-hypotheses/)in support of this proposal: one of these for example championed modern humans’ direct evolution from New World monkeys instead of African primates. Now, another of his theories speculated that humans evolved directly from "East Eurasian hominids" who had crossed into the New World:
Between 200,000 and 100,000 years ago, a subset of this original [East Eurasian] hominid population migrated to the New World (via the Bering Strait Land Bridge), where speciation into modern humans occurred. While, under out-of-America II, the unique social behaviors shared between modern humans and New World primates (pair bonding, paternal investment, cooperative breeding and speech) are interpreted as homoplasies, the fact that such key aspects of human social and cognitive behavior are shared with Platyrrhines suggests that the immediate ancestors of modern humans were exposed to the same New World environment as the New World monkeys.
Buckle up. In this thread I will present comprehensive evidence for a modified version of this theory, except instead of some archaic “Eastern” hominid, my contention is that it was homo sapiens who trekked from Africa through Asia and into America by around 100kya-- and later back-migrated into Eurasia and Africa. The Apidima-1, Misliya-1, Skhul & Qafzeh, and Al-Wusta remains all demonstrate human presence outside of Africa in the 215-100kya time-frame. Further east, Chinese researchers have discovered what they describe as 47 fully modern looking human teeth in South China dated 80-120kya (https://www.sciencemag.org/news/2015/10/trove-teeth-cave-represents-oldest-modern-humans-china), showing that humans made it at least this far. The Tongzi teeth (also from South China) were dated 172-240kya (https://www.sciencedirect.com/science/article/abs/pii/S0047248418303464#fig3), although their provenance is more dubious. There seems to be a definite homo sapiens presence in North India at least 80kya (https://www.thehindu.com/sci-tech/science/early-humans-lived-in-northern-india-80000-years-ago/article30924041.ece) and some type of advanced hominid in Australia as early as 120kya (http://ancientnews.net/2019/03/11/archaeology-places-humans-in-australia-120000-years-ago/). And there is one very suggestive hint found in ancient DNA: the Altai Neanderthal has modern human admixture that introgressed approximately 100kya (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4933530), while the more westerly Croatian Vindija Neanderthal does NOT. Hitherto many (including myself) have just regarded most of these “early” OOA populations as genetic dead-ends…
It’s true, we don’t have definitive evidence of AMH in the Americas anywhere near deep in history as I’m suggesting (100kya or so), but I would posit this is partly due to lack of initiative. Only a century ago the “consensus” was that the Americas were populated no earlier than 5kya. Only in the 1960s or so was the date pushed back to 10-15kya or so with the acceptance of the Clovis Culture. But in recent DNA papers to make sense of American diversity, this date has been pushed back to 20kya or or 25kya—even though the Bering Land Bridge was closed between 40-15kya (https://www.youtube.com/watch?v=USIAcXfv39k). And how much “deep” archaeology has really been undertaken in the Americas compared to Africa or Europe? For that matter, even East Asia has been mostly terra incognito-- the first Denisovan was discovered there only in 2010, vs. the mid-19th century for European Neanderthals.
In terms of affirmative evidence, the Bluefish Caves site in Canada shows definite human presence dating back to 22,000 BP (https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0169486). Meadowcroft Rockshelter in Pennsylvania has been dated to at least 16,000-19,000 BP (https://www.sciencedirect.com/science/article/pii/S0305440313000733?via%3Dihub)(objections that it’s “only” 14,000 years old have been rejected by third-party radiocarbon dating). There is pretty obvious evidence of bone marrowing (by some type of hominid) dating back ~32,000 BP in Pendejo Cave, Mexico (https://web.archive.org/web/20060414084853/http://www.umass.edu/anthro/chrisman/page2.html), and earlier evidence of tool use from this cave dated back to 50kya. The Toca da Tira Peia rock art site in Brazil has been dated using luminescence to 22,000 BP (https://www.sciencedirect.com/science/article/pii/S0305440313000733?via%3Dihub)(with contested earlier dating as deep as 36kya (http://www.bradshawfoundation.com/south_america/serra_da_capivara/index.php)). The most famous pre-Clovis site is Monte Verde in Chile, with the now mostly accepted date of 18,500 BP. However the deepest layer of Monte Verde (MV-I, where burnt charcoal was found) was dated to 33,000 BP. Deepest of all, the Hueyatlaco butchering site (https://en.wikipedia.org/wiki/Hueyatlaco)has been dated to 250kya by both uranium-thorium and fission-track dating, although both younger and older proposed dates using other methods range from 80kya to 780kya. The Xalnene Tuff footprints (https://en.wikipedia.org/wiki/Xalnene_Tuff_footprints) have a similar dated range-- from the mid-Lower Paleolithic to the early Upper Paleolithic. Depending on which dating method is more accurate, Hueyatlaco and Xalnene Tuff could represent early humans, or maybe some archaic hominid. Either way, these sites demonstrate the mobility and ecological range of hominids -- a lesson that somehow many still refuse to accept.
While the existence of pre-Clovis American cultures are now widely accepted, we still lack definitive pre-Clovis DNA—but hints of deeper layers of American ancestry have already been published--nearly unnoticed. There is the Amazonian PopY, the Mesoamerican “UPopA”, and the recently discovered Antilles_Archaic. The latter two of these clades appear to harbor "deep" ancestry that has no close relationship with any of their supposed recent ancestors in Eurasia. In other ancient American samples, clades are inferred to "split" from neighboring populations (often in the absence of physical barriers) and live in isolation for 5k, 10k, or even 15k years--this is the only way qpGraph can explain the differential geographic patterns of drift. The large FST ratios between modern Amerindian populations have been explained by the post-Columbian collapse, whereby many Native groups lost up to 90% of their diversity, increasing homozygosity and skewing their inter-population relationships. But the ancient Amerindians we have on record also are much more diverse than expected—although qpAdm/QpGraph modeling has papered over this. Before examining the evidence in America however, let’s review some of the inconsistencies and shortcomings of the “orthodox” OOA theory as currently constituted:
The conventional Eurasian phylogeny (~60-80kya Out of Africa—or up to 150kya OOA if you believe in “Basal Eurasian”-- followed by a ~50kya “East-West Eurasian split”) has been shaken as incongruous new data comes in: “Western” Yana RHS is near Beringia, Belgian Goyet-Q2 has “eastern” affinities, Bulgarian Bacho Kiro has “Andamanese” mtdna, Tianyuan is EQUALLY related to E Asians and Amerindians, and various DNA and archaeological evidence exists for Australasians in Sahul BEFORE the proposed East/West OOA split. Why is fossil evidence of undisputed “modern” humans found in Upper Paleolithic Northeast Asia and Australia before Africa and the Near East? How do the Americas have such astounding linguistic diversity (even after post-Columbian devastation!), if the bottlenecked 15-10kya Beringians were really the First Americans? What is the “Pop Y” signal in Amazonians, if not a relic from a “deep” Old American population that must’ve been BASAL to (and not DERIVATIVE from) Oceanians? How is it possible Ust Ishim was in Siberia 45kya, Mungo Man in Australia 40kya (probably earlier, according to other evidence), modern human cave art in Borneo ~40kya, and Tianyuan in China 40kya, while the last Neanderthal holdout was Gibraltar (28kya)—and only then were they replaced from the northeast, not from right across the Straits? Why were Neandersovans wiped out so decisively by 40kya across most of Eurasia, while the West African Iwo Eleru and East African Wadi Halfa and Jebel Sahaba skulls show pronounced ‘archaic’/pre-modern features deep into the Mesolithic post-15kya?
Why can't we even find any modern human sites in pre-Mesolithic Africa? The only example I can see is the Hofmeyr Skull (36kya), which has long been said to morphologically resembler UP Eurasians. The Willerslev lab is reportedly working on getting DNA out of this. As of October 2019 (https://elmuseumscience.wordpress.com/2016/07/26/hofmeyr-skull-research-update/#comment-659), the first attempt was unsuccessful, but they're trying again.
So how could an Out of America hypothesis possibly "fit" with what we know about ancient and modern DNA? Something like this:
a) 300kya-200kya – Anatomically modern humans (Jebel Irhoud, Omo I, Florisbad) emerge and diversify across Africa. Development of y-haplogroup A00 and eventually, A
b) 200kya – 100kya – Out of Africa occurs (Apidima-1, Misliya-1, Skhul & Qafzeh, Al-Wusta etc) but while early humans are unable to challenge Neanderthals deep into Europe, they ARE successful penetrating deep into East Eurasia. One of these “eastern” OOA groups reaches the Americas no later than 100kya - 75kya (and possibly earlier), when there was still a land bridge in Beringia (https://www.youtube.com/watch?v=USIAcXfv39k). These Old American humans might have carried the CT macrohaplogroup, or that mutation could have developed soon after in the Americas.
c) Sealed off from Eurasia until 50kya by re-formation of the Beringian/Canadian ice sheets, Old American groups further diversify and spread through the Americas. Dziebel’s theory was that the entire “behavioral modernity” package developed in the Americas—language, symbolic/pictoral art, pair bonding etc. While these latter 2 elements are challenging to prove, the incredible linguistic diversity of the Americas does suggest that language as we know it originated with one or more of these Old American groups—and this could have formed their decisive advantage over other humans and archaic hominids.
d) 50-40 kya – Groups of “Old American” modern humans pour back west across Beringia, as soon as the land bridge was reopened, quickly spreading throughout Eurasia and Oceania and rapidly replacing Neandersovans along the way. The Old Americans, IOW, would be “basal” to all Upper Paleolithic Eurasians (“Crown Eurasians”), the latter being a subset of Old American diversity. In Southwest Eurasia and Africa at the terminus of their back-migration, Crown Eurasians admix with groups of “Basal Eurasians” and “Basal Africans”— their human cousins who never left Africa/SW Asia, in situ descendants of some of the early humans mentioned in a) & b). West Eurasians (and possibly Horners) can probably be modeled as varying mixtures of Crown + Basal Eurasian, while West, Central, and South Africans would have Crown, BE, BA, and additional archaic admixture. This would account for the “basal” phylogenetic position of sub-Saharan Africans. It would also explain why linguistic diversity generally decreases as one moves west from the Americas to Africa.
e) ~39kya – The Beringian land bridge closes again as the Laurentide ice sheet re-forms, pushing to its maximum southern extent ~21kya as the ice caps stretch from New England to Tennessee to Vancouver. This drives the migration of northerly Old Americans south, admixing with southerly Old American groups. Further admixture events occur— ie, possibly a ~14kya Clovis expansion associated with y-dna Q-M3, the 5.4kya Mesoamerican agricultural expansions both north and south (supported by aDNA), the Numic expansion, et cetera. These admixture events dampen some (but NOT ALL, as I will demonstrate!) of the signal of Old American inter-population diversity.
f) 15-10kya – As ice sheets retract and the land bridge reopens, some groups of Americans migrate back into Beringia/Siberia (ie, 10kya USR-1, 9.8kya Kolyma-1) while some Siberian groups move east. This American/Siberian exchange continues sporadically throughout history (Eskimos, Na Dene expansions, etc).
From the existing stock of published data, I hope to show that the preponderance of evidence is right under our noses—and the intent of this thread is to simply connect the dots.
The Big Picture
The Orthodox Out of Africa model has been partially bolstered by population geneticists’ widespread adoption of qpAdm/Graph over older tools like Treemix and FST (fixation index). QpGraph has fostered the development of more and more “ghost” populations to explain Eurasian and especially American diversity, and as this tool is “supervised” and subject to the assumptions of the user, I maintain that it has obscured the true state of affairs. While FST and Treemix may have their own flaws and limitations, they've been basically marginalized as addenda exhibits, while qpAdm/qpGraph has dominated the main analyses and topology-construction. FST has always indicated that Amerindians are the most distant group of humans from Africans—substantially more distant than Northern Europeans and non-Siberian East Asians are—which makes no sense if Amerindians are simply an ANE+East Asian composite. We are told these inflated Amerindian FST values are simply artifacts of demographic oddities—post-Columbian population collapse, long periods of isolation, etc. I aim to demonstrate that these explanations are not entirely convincing.
The 2015 paper "The Equidistance Index of Population Structure (https://www.biorxiv.org/content/biorxiv/early/2015/12/08/033852.full.pdf)" contained a basic FST matrix (in red) with selected populations:
https://i.imgur.com/abnZuzw.png
The authors noted:
"The Surui and Karitiana have an unusually high pairwise FST. In fact, the Karitiana are as diverged from the neighboring Surui in terms of FST as they are from the Mongola on the other side of the world (Table 1, Figure 3, and Figure S6). Moreover, FST actually decreases initially with distance from the Amazon, from 0.13 between the two Amazonian tribes, to 0.08-0.1 between Amazonians and Colombians, further decreasing to 0.07-0.09 between Amazonians and the more distant Maya. Remarkably, the highest FST among all HGDP Native American populations is between the two geographically closest populations, the Surui and Karitiana. These apparent anomalies can be explained by the inflation of FST in genetic isolates. FST between pairs of isolates can be nearly twice as high as between either one of the isolates and a more cosmopolitan population, as pairwise FST reflects the combined isolation of both populations. Since the Surui and Karitiana are both isolated, their pairwise FST is nearly double that between any one of them and a larger, less isolated population such as the Maya. In other words, the Maya’s contribution to the pairwise FST is dwarfed by that of the Amazonians."
To test their theory on isolates: the Maya number in the millions. Yet their FST with Surui is still 0.09, larger than the Maya:Russian FST. And even if we exclude the isolated Amazonians, the Pima:Colombian FST is still equivalent to the Russian:Yakut FST. (Both Colombians [Piapoco + Curripaco] and Pima each number ~20,000 --possibly larger than their pre-contact population). And if "isolation" is responsible for skewing the FST ratios-- what about the San? According to the traditional phylogeny they've been isolated for 250k years, save for some minor Afroasiatic ancestry absorbed in the Iron Age. Yet the San:Bantu FST doesn't show the hallmarks of two groups that have been separated 200kya+.
The authors came up with an alternate statistic "Est" (the green matrix) to try and "rectify" these patterns, but this just seems to muddle things further. Due to these "issues", population geneticists have generally abandoned FST as a priority informative measure. But here we will stay on the case:
In "Insights into human genetic variation and population history from 929 diverse genomes (https://science.sciencemag.org/content/367/6484/eaay5012)" from Bergstrom et al, the authors used modern DNA from the HGDP broken into geographic population clusters. They ran Fst statistics using 6 different allele array sets, but all 6 arrays demonstrate the same basic patterns that all approaches to quantifying human diversity relying on fixation indices have shown:
Estimates of FST for a selection of population pairs, grouped by region, using the Li 2008 array sites (“li2008”), the Human Origin 2012 sites (“ho2012”), the MEGA array sites (“mega”), all variants discovered in the sequencing data (“all”), all variants excluding singletons (“all.c2”) and variants ascertained in archaic genomes (“archaics”). Each line connects the FST values for one pair of populations across the different set of sites
https://i.imgur.com/E7eHQxq.jpg?1
1) Americans are the most divergent group of humans relative to Africans, Oceanians second, East Asians the third-most divergent and West Eurasians (Europeans and Middle Easterners) the closest group to Africans.
2) Americans are more closely related to East Asians than to Europeans, BUT except near the extremes (ie, Yakuts on the one hand and Sardinians on the other hand) East Asians are more closely related to Europeans than to Americans
3) East Asians are [substantially] more closely related to Europeans than to Papuans or Australians
4) Africans are the most divergent of human groups and have famously vast intra-group diversity (ie, San vs San or Bantu vs Bantu) relative to Eurasians. Yet the inter-population distance between Africans is somehow lower than the distance between Africans and Eurasians (see the first FST table above: San:Bantu FST is ~4x smaller than San:Surui and 2.5x smaller than San:Russian). This makes no sense if South Africans split from all other humans 250kya-300kya.
5) Some American populations have higher inter-population Fst ratios between them than between Africans (Fst for Surui:Karitiana/Pima/Colombian/Maya > Bantu:San).
All five of these points are incompatible with the basic phylogeny presented in the orthodox OOA theory. However, the Out of America theory I have outlined provides a coherent explanation for every point above.
Regarding 1), West Eurasians’ affinity with Africans has been explained by “Basal Eurasian” ancestry, and this may be partly true. However, Basal Eurasian admixture (which may be as low as ~10% in North Euros, assuming Anatolia_N is ~25% BE) cannot sufficiently explain the magnitude of this attraction. And since East Asians lack Basal Eurasian ancestry, why are they also substantially closer to Africans than Oceanians, Siberians, or Amerindians are?
The answer to these questions is hinted at by a very significant recent paper by Cole et al (Ancient Admixture into Africa from the ancestors of non-Africans (https://www.biorxiv.org/content/10.1101/2020.06.01.127555v1)) which UPenn’s Ian Mathieson and Max Planck Institute collaborated on:
Here we present SMCSMC, a Bayesian method for inference of time-varying population sizes and directional migration rates under the coalescent-with-recombination model, to study ancient demographic events. We find evidence for substantial migration from the ancestors of present-day Eurasians into African groups between 40 and 70 thousand years ago, predating the divergence of Eastern and Western Eurasian lineages.
……
We find that the method infers high rates of migration from descendants of the OoA event (’non-Africans’) to Africans, but not in the opposite direction, in the period 30–70kya corresponding to the Late Middle Paleolithic…. Inferred IMFs are not significantly different between Han Chinese and European populations in non-Afroasiatic populations (p=0.14, two-tailed paired t-test; Figs. 1h and S6, Table S1), consistent with [back-]migration occurring before the European-East Asian split approximately 40kya.
Now, this back-migration admixture was inferred to occur 30-70kya—the 40kya figure is merely the authors setting a baseline for the latest possible “East-West” split in accordance with the Orthodox OOA. Of course in the Out of America scheme there really is no distinct, dateable “East-West” split: East and West Eurasians would represent different subsets of Old American diversity that only gradually coalesced into discrete geographic clusters over time, with later West Eurasians (maybe as early as WHG or even earlier) especially distinguished by Basal Eurasian admixture. And if the HGDP European test pops share Basal Eurasian ancestry w/Africans (the authors stress their method lacks the power to pick up discrete signals older than 70kya) this could skew the inferred back-migration timeframe deeper than necessary.
The nature of this back-migrating population (“equally related to Han and French”) is speculative, but my guess would be something with affinities to one or more Upper Paleolithic Europeans on the one hand, and the ambiguous yet apparently deep “Eastern” ancestry that seems to comprise ~30% of Iran_N’s genome on the other. As for y-dna, how about DE and maybe LT (with L and D gradually accumulating in Near East/Turan/S Asia while E and T play the largest role in Africa). Remember those six basal E1b1b (and possibly one E1c) samples found in Pakistan_Udegram? Or Narasimhan’s finding that Indus Valley farmers were basal to Iran_N?
Now, this Bayesian method attributed the following “Eurasian back-migration” admixture proportions to these African ethno-linguistic groups:
Afroasiatic: 50% / Nilo-Saharan: 41% / Niger-Kordofanian: 35% / Mbuti-Biaka: 27% / Khoe-San: 21%
(Afroasiatics also have substantial post-Holocene West Eurasian ancestry, not included above)
Consider the following table I put together to model the latter 3 African groups. The highlighted row copies the “[Crown] Eurasian back-migration” proportions directly from the Cole et al paper (I am assuming this population split from other Crown Eurasians ~40kya). These components below, I will stress, are speculative—there may be more or less Basal/Archaic populations, some Basal & Archaic lineages could have been already admixed or discrete, the split times may vary, etc. It is merely one potential instructive scenario. Note that while my topology is radically different, the aggregate split times are generally in agreement with Harvard’s latest work in the Shum Laka DNA paper (Central African & South African HGs inferred to split ~250kya, but modern Mbuti & Khoisan have admixture from less basal groups—~26% Bantu+17% Afroasiatic-related for the former and 10-20% Afroasiatic for the latter—hence their shallower projected splits below):
https://i.imgur.com/nEANSWa.jpg?1
Assume the Crown Eurasian “Back-Migration pop” reaches the Near East 30-40kya (Levantine Aurignacian?), absorbing more and more “Basal Eurasian” and then “Basal African East” ancestry as it crosses the Red Sea. This group later fans out in waves across the rest of Africa, acquiring additional localized variants of Basal/Archaic ancestry. Now, all 3 of the African populations modeled above harbor lots of “deep” ancestry relative to Eurasians—hence their overall “basal” phylogenetic position, and the relatively high heterozygosity and intragroup genetic diversity observed between African individuals even within the same tribe. However, note how Yoruba and Khoisan share 65% of their ancestry in common—viz, components 1 through 3. At a population level, both Khoisan and Yoruba are predominantly of the same Crown + BE + BA_East stock. Moreover, they acquired most of this common ancestral “base” from a small founding group that coalesced in East Africa as late as the Mesolithic. And this explains why the pairwise FST between Khoisan and Yoruba is much smaller than the FST between Khoisan and non-Africans.
Since Crown Eurasians represent only a subset of Old American diversity, Amerindians have the most distant relationship with Africans. East Asians have moderate attraction to Africans because they’re more closely related than Siberians and Oceanians to the Crown Eurasian back-migration population. West Eurasians as well, but they have an additional attraction to Africans due to shared Basal Eurasian ancestry.
Regarding Oceanians, when the authors used Papuan instead Han or Europeans as the source for the back-migration, the proportions of inferred Eurasian ancestry in African groups dropped by an avg of ~3% across the board:
When we estimate these proportions using a Papuan sample to represent non-African descendants we find slightly but significantly smaller values compared to estimates using either the Han Chinese or European populations (mean difference of 0.029±0.002, p=9.2⇥10
They theorized that the excess Denisovan ancestry in Papuans was responsible for this decrease, which is reasonable. I do wish they had used Australians (who have "normal" amounts of Denisovan admix) as a test pop as well—another factor for Oceanians’ distance from Africans might be that ancestral Oceanians were the first Old American group to back-migrate west across Beringia, separating from other Crown Eurasians almost immediately (~50kya) and thus being more distal from the back-migration population. Alternatively, modern humans arriving in Sahul might have picked up minor admixture from a substrate of ancient human groups (see ie, some evidence of humans in Australia dating back to 65kya, or even 120kya).
Turning back to the Bergstrom paper, we can see in Figure 3: "Counts and properties of geographically private variants":
The highest frequencies [of private variants] are reached by a few tens of variants present at >70% (and a few thousands at >50%) in each of Africa, the Americas and Oceania. In contrast, the highest frequency variants private to either Europe, East Asia, the Middle East or Central and South Asia reach just 10-30%.
...
Even comparing Central and South America, we find variants private to one region but absent from the other reaching >40% frequency. Within Africa, ~1000 variants private to the rainforest hunter-gatherer groups Mbuti and Biaka reach >30%, and the highly diverged San of southern Africa harbour ~100,000 private variants at >30% frequency,
Within the framework of Out of America, these patterns are expected. But within the Orthodox Out of Africa framework the Americas shouldn't have more private variants than Eurasia, since American diversity would be merely a subset of Crown Eurasian diversity. But as the private variants demonstrate, the opposite is the case.
Below we can see the provenance of the private mutations: in Africans and Oceanians, their private mutations are shifted more towards "ancestral"/"archaic" as opposed to "derived" variants. But in the Americas, the highly "derived" and minimally "archaic" nature of the private variants shows they were very obviously related to the evolution of early modern humans. These private variants (and not merely "isolation") diversely distributed across Amerindian groups must contribute to the comparatively large FST ratios between Amerindian populations. And these variants take lots of time to accumulate: post-Colombian collapse and contractions in population sizes have nothing to do with the immense wealth of private gene variants in the Americas-- in fact the Columbian Exchange likely reduced the number of these variants.
https://i.imgur.com/bpCprnx.jpg
In the recent ancient East Asian DNA paper (https://anthrogenica.com/showthread.php?20302-Mammoth-Paper-from-Qiaomei-Fu-s-group-on-Genetic-history-of-East-Asia&p=666491&viewfull=1#post666491), the authors noted “genetic diversity was greater in these samples than in modern day East Asians”. Interestingly, in some ancient East Asian samples an affinity to Surui was detected, although this affinity was erased when restricting analysis to transversion sites. Moreover, the authors surprisingly failed to find Denisovan DNA in ancient or modern East Asians. While these results are contentious, is it possible the two findings are related? ie, that there is “deep” ancestry in East Asians related to Amazonians or other Old Americans, and this has been mischaracterized as “Denisovan”? Additionally, one of the Mesolithic East Asian samples had a slight but perceptible affinity to Villabruna.
These findings, along with the confusing affinities of various Upper Paleolithic Eurasians (mentioned earlier, ie, Bacho Kiro, Yana RHS, Goyet), and the fact that Europe has the lowest number of private variants in the world (shown above) help illuminate the true state of affairs: all of the Upper Paleolithic Eurasian diversity is radiating from the east, not from the south.
Let us now examine the ancient DNA from the Americas:
Ancient American Autosomal Diversity
The 2017 USR-1 paper revealed the closest analog to an ancestral Amerindian population in the “Ancient Beringians”.
“Ancient Beringian[s]… descended from a single founding population that initially split from East Asians around 36 ± 1.5 ka, with gene flow persisting until around 25 ± 1.1 ka. Gene flow from ancient north Eurasians into all Native Americans took place 25–20 ka, with Ancient Beringians branching off around 22–18.1 ka.”
USR-1 was deemed basal to Amerinds, though not a direct ancestor. The authors posited a North Amerind-South Amerind (NNA-SNA) split somewhere in Beringia or North America. This “clean split” theory is already dubious—12kya Anzick-1 is directly in the way of the proposed “inland” path for the NNA branch, yet falls in the “coastal” SNA clade.
As an aside, note how the split times between Amerinds & Eurasians, and even between USR1 and Amerinds, are continually pushed back beyond the elastic limits. Most climate modeling shows the ice-free land bridge across Beringia was only available for a brief time during this epoch—around 15kya to 10kya. Around 20kya there were still thick ice sheets from Beringia to Alberta. But now AB’s “split time” from Eurasians is pushed back (for the ANE ancestry to 20-25kya, and for the E Asian ancestry to 36kya), when these groups all would’ve been penned up together in Siberia. And pushing Ancient Beringians back to ~20kya forces the models to pin them along the narrow Beringian coastline in recognition of the ice sheet covering the Alaskan interior (and the interior is where USR-1 and Alaska_TrailCreek were actually found—but dated no earlier than ~ 10kya). The forced “coastal” route, skirting the ice, also makes the “clean split” of NNA/SNA ancestries difficult to conceive, spatially.
But let’s move on:
In June 2018’s Ancient human parallel lineages within North America contributed to a coastal expansion (https://science.sciencemag.org/content/360/6392/1024.full) from Scheib et al, the authors posited the existence of a two post-Beringian clades: one (ANC-B_) represented by ASO (Ancient South Ontario, dating to 500-3000 BP), and another represented by Clovis and South Americans (ANC-A). The distribution of these ancestries is downright bizarre: after ASO, the populations with the highest ANC-B ancestry are.... Chileans, from the Pacific coast of South America, with up to 75% ASO-related ancestry.
qpGraph also indicated West Eurasian admixture into ASO (“Canada_Lucier”), but the authors decided this was “modern contamination”:
The inferred 2% West Eurasian admixture into Canada_Lucier_4800BP-500BP is most likely explained by contamination in these samples by people of European ancestry.
We can surmise that all the usual techniques to remove contamination were adhered to across the dataset—and yet none of the other samples were asserted to have excess West Eurasian admixture. And in fact, in the Shuka Kaa (10kya western Canada) DNA paper from 2017, Treemix indicated an admixture edge from Shuka Kaa into West Eurasians (p. 11 of the Supp Info pdf).
Within Scheib et al, the authors included an ADMIXTURE plot. ADMIXTURE might be the least sophisticated measure of genomic diversity, however it's also unsupervised, and in cases where "deep" population history is less well known it can identify otherwise hidden patterns. For example, the "teal" component that showed up ADMIXTURE runs of West Eurasians was long the object of speculation. When the first Yamnaya were sequenced and revealed to be up to 50% "teal", speculation increased-- but all we knew is that Teal had some kind of affinity with the northern Near East/Caucasus. Of course it was not until the sequencing of CHG that "Teal" was revealed to be its own ancestral component that likely coalesced in the Mesolithic, with a complex Paleolithic history of which we're still unsure about. So for the Americas-- a vast geographic area with only several extant samples dating back to ~10kya-- ADMIXTURE can be very revealing.
At K=9, the colored ADMIXTURE components are (https://i.imgur.com/xJdTopb.jpg): Purple, Green, Orange, Beige, Magenta, Sky blue, Rust, Ocean blue, and Tan (or lime green? I’m mildly colorblind…)
The most “diverse” East Asians are those with Siberian-related ancestry: Oroquen, Hezhen, Mongola, etc. I have written before about how the American-Siberian exchange was more extensive and prolonged than generally accepted: for example, Nganassan-like ancestry first shows up in ancient Eurasia only in the Bronze Age (https://anthrogenica.com/showthread.php?20302-Mammoth-Paper-from-Qiaomei-Fu-s-group-on-Genetic-history-of-East-Asia&p=667534#post667534). This ancestry is obviously a recent arrival from the Americas that's only distantly related to other East Asians.
But south of Siberia, Han, Dai, Japanese etc are all modeled with 90-95% of the “Rust” component. In nMonte Han can be modeled pretty well as a 3-way mix of “Devils Gate” (“Mongolic/Tungusic”), Taiwan_Hanben (“Austric”), and NPL_Chokopani ("Tibetic") -type ancestries. Each of these components must have been the complex byproduct of alternating isolation and admixture events rooted in the Paleolithic peopling of East Asia. Yet compared to the Americas, the diversity between these mainland E Asian ancestral strains is apparently insignificant.
In contrast, there are 4 major ADMIXTURE components assigned to the Americas, compared to just 1 in E Asia and 2 in Oceania: “Beige” ancestry peaks in Californians and Mexicans, “Purple” in North Americans (not just Athabaskans w/recent Siberian ancestry, but also ASO/Ancient Ontario), while South Americans show up as either “Orange”, “Green” or some admixture therein. But the Americas actually harbor all 9 colored components in the Admixture run. Shuka Kaa has nearly 30% Sky Blue (“Onge”) ancestry, Anzick, ASO and Kennewick Man all score ~5% Ocean Blue (“Nganassan”), while Magenta (“Papuan”), Rust (“East Asian”) and “Tan” ancestries are found in trace quantities across geographically diverse American population clusters.
ADMIXTURE is demonstrating here what qpGraph, with its "subjective" topology, is obscuring: the Americas are more diverse than Eurasia.
The next big releases came within one week of each other in November 2018: Reconstructing the Deep Population History of Central and South America (https://www.cell.com/cell/fulltext/S0092-8674(18)31380-1), from Reich's Harvard group, and Early human dispersals within the Americas (https://science.sciencemag.org/content/362/6419/eaav2621.full), from Eske Willerslev's Copenhagen lab.
Reich's group rejected the ANC-A/ANC-B ancestral clades but came up with similar analogs: Ancestral-A and Ancestral-B. They rejected Ancient Ontario-related ancestry flowing into the Andes, explaining this affinity in other (opaque) ways-- basically multiple early split of the Ancestral-A ancestry, while Ancestral-B once again represents ASO. Yet in the Supp Data, we find that:
Although we were not able to reject the null hypothesis that ASO is equally related to Anzick1 and present-day populations (|Z| between 0.09 and 1.61), we observed that the Lagoa Santa population is most closely related to ASO (|Z| between 0.87 and 3.7), followed by the Spirit Cave genome (|Z| between 0.5 and 2.49) (Fig.2F). Together with the best-fitting model presented in (Fig.2F), these tests suggest that ASO carries admixture from a SNA population most closely related to Lagoa Santa and Spirit Cave that split after Anzick1. Thus, we consider that the apparent ASO-related ancestry in South Americans (7) could be explained by shared ancestry between Spirit Cave and/or Lagoa Santa and ASO. Spirit Cave and Lagoa Santa are both supposed to belong to the "Southern" clade that branches from Ancestral-A, like Anzick. Yet they're simultaneously most closely related to ASO from the Ancestral-B cade??? I recommend everyone to read the Supp Data to see the other puzzling stats and assertions from this study.
The Reich paper did include the first very old Central American/Mesoamerican samples: Mayahak Cab Pek (9300 BP) and Saki Tzul (7400 BP). As their neighbor joining tree indicates (https://i.imgur.com/vbEGIpm.jpg), these ancient Central Americans appear to be pretty divergent lineages. But there's surprisingly little analysis of these samples-- possibly because of low coverage, but also perhaps because they'd spoil their model.
The Willerslev group didn't have access to the ancient Mesoamerican genomes in their near-simultaneous paper's release. But merely by analyzing modern Mixe relative to to other ancients and moderns, they came to this surprising conclusion:
“further SFS-based modeling indicates that Mixe most likely carry gene flow from an unsampled outgroup, and form a clade with Lagoa Santa. Including nonzero outgroup admixture into Mixe when fitting an f-statistics-based admixture graph resulted in a significantly better fit (likelihood ratio test, P < 0.05) (Fig. 3, A and B ). Hereafter we refer to that outgroup as 'Un-sampled population A′ (UPopA), which is neither Ancestral Beringian, NNA or SNA, and which we infer split off from Native Americans ~24.7 ka, ranging from 30-22 ka (95% CI; this large range is a result of the analytical challenge of estimating divergence and admixture times in the absence of UPopA genome data). This age range overlaps with the inferred split of Native Americans from Siberians and East Asians 26.1-23.9 ka (1), and the divergence of USR1 from other NA (23.3-21.2 ka). This temporal overlap, which cannot be fully resolved into a relative sequence with current data, suggests there were multiple splits that took place in Beringia close in time.”
This UPopA autosomal signature is inferred to have formed when Amerindians’ ancestors were still in Siberia! So UPopA tagged alongside on the trip to Beringia, *without admixing with proto-Ancient Beringians*, then around 15kya AB splits into NNA-SNA branches, and UPopA once again tags alongside these two groups, settles directly between the NNA and SNA branches near the Panama isthmus—*again without admixing with SNA/NNA until more modern times* (since the UPopA signal was not detected in other ancient Amerinds, save for those that received post-5.4kya Mesoamerican-related admixture)
How absurd is this scenario? And if it somehow occurred, wouldn’t we then expect to see ancient or modern East Asians who share additional alleles with Mixe? But we don’t. The logical conclusion is that “UPopA” is an older American autosomal signature which must have been largely paved over by later admixture events.
The next surprising finding from this paper comes from western British Columbia:
While the coastal British Columbia ancient individuals clustered together with present-day Athabascan and Tsimshian speakers from the region, ASO and Kennewick were placed in an intermediate position between NNA and SNA. Although the Big Bar individual was placed close to NNA populations not carrying recent Siberian admixture (Fig. 1, C and F) (13), D-statistics of the form D(Aymara, NA; BigBar, Yoruba) and D(USR1, NA; BigBar, Yoruba) suggest Big Bar represents a previously undetected outgroup to non-[Ancestral Beringian] Native Americans, one that diverged prior to the NNA/SNA split (13)
.........
In this region’s long history, we find evidence that groups on the coast (e.g., 939) and their contemporaries in the interior (Big Bar), were as genetically distinct as are present-day groups (18) (Fig. 6A).
Now, Big Bar from British Columbia is dated to 5.4kya. It seems highly unlikely this individual spent 10,000 years as an unadmixed clade straight out of Beringia. The “contemporary” “NNA” clade formed by the “939/302/443” samples is just several hundred miles northwest, while the older Anzick-1 from the SNA clade was found only ~500 miles southeast. The nearest sample spatially is Kennewick Man (9kya) who is assigned as being a SNA/NNA admixture—or at least that’s how Kennewick appears on PCA. If Big Bar is admixed (which seems likely), whatever is causing its divergent ancestral profile would by definition have to be considerably “deeper” than the hypothesized NNA-SNA split.
The Willerslev paper also detected the presence of "PopY"-- an ancient Amazonian substrate with excess affinities to Andamanese relative to other Amerinds-- in the 7.4kya Brazil_Lapo_De_Santo. This ancient sample is inferred to harbor 3% PopY ancestry, compared to modern Amazonians (2%), whose PopY ancestry has been diluted by later Mesoamerican-related admixture.
In the most recent Amerind paper, A Genetic History of the pre-Contact Caribbean (https://www.biorxiv.org/content/10.1101/2020.06.01.126730v1), the authors discovered yet another very “deep” ancestral signature, this time in the hitherto-unsampled Caribbean (map of samples (https://i.imgur.com/0fhY7qV.jpg)). The Greater_Antilles_Archaic clade consists of 13 individuals from NW Cuba (3200 – 800 BP) and 2 individuals from the Dominican Republic (note: “Archaic” refers to the pottery style; these are not archaic hominids). The authors searched for affinities with other ancients and moderns and this was their conclusion:
“Using present-day populations, a similar result of no specific affinities was found . In qpGraph, we fit GreaterAntilles_Archaic in several positions along a skeleton admixture graph from Posth et al.23, again reflecting poor power to differentiate between a Central or South American origin (Supplementary Information section 11). Likewise, in a maximum likelihood phylogenetic tree based on allele frequency covariances, GreaterAntilles_Archaic is inferred to split before all South and some Central American populations. Together, these results point to the origin of GreaterAntilles_Archaic in a deeply divergent Native American population that is not particularly closely related to any sampled present-day groups.”
Here is a copy of their qpGraph attempt (https://i.imgur.com/1QjpWXW.jpg?1)
Now take a look at the FST(x100) table:
https://i.imgur.com/LAUACok.jpg?1
Now, In Eurasia the FST between Mesolithic populations is higher than in later populations-- agriculture seemed to always result in admixture (Neolithic farmers to Europe and European herders to Central/SW Asia, for example). These Caribbean populations are "young" , relatively speaking. BUt just for comparison, the FST between the Archaic samples and the other Caribbeans (.1082 to .1211) dwarfs the FST between WHG:EHG (.078), LBK:EHG (.084), WHG:Iran_C (.099), and EHG:Iran_C (.078), to name just a few ancient Eurasians who don't share recent geneflow. For that matter, even the Venezuela_Ceramic samples are quite divergent-- much more than expected under the terms of the Orthodox OOA.
“While short ROH measures population sizes of the past few dozen generations, the consistent signal across all Caribbean sites (including large and small islands) indicates that these estimates share some signal of larger meta-populations, as expected due to a high rate of mobility.
…..
“We also evaluated heterozygosity levels and found lower genetic diversity in the individuals from GreaterAntilles_Archaic and Haiti_Ceramic than in any of the other Ceramic-related (sub-)clades from the Greater Antilles, Curaçao, and Venezuela (Extended Data Fig. 2). These results agree with the median sums of ROH suggesting lower genetic diversity and a small population pool in the Archaic-related than in the Ceramic-related individuals. However, the Ceramic-associated Caribbean groups here studied were found to have an overall similar genetic diversity to that of some contemporary groups from continental South America, such as from the Peruvian Middle and Late Horizon periods (Nakatsuka et al. 2020).”
Note that Antilles_Archaic and Haiti_Ceramic have similarly low levels of heterozygosity/intragroup diversity—but Haiti_Ceramic didn’t exhibit the massive fst values vs other Caribbeans, or have nearly as much difficulty fitting into qpGraph as Antilles_Archaic.
From their Treemix run, a (~30%?) migration path is inferred from Haiti_Cermaic into Antilles_Archaic:
https://i.imgur.com/bMlJfQU.jpg
“Orange arrows represent admixture events, although the indicated direction of admixture might be inaccurate based on observations from analyses such as qpAdm admixture modeling - e.g., we believe it likely that there is GreaterAntilles Archaic admixture into Haiti_Ceramic rather than the reverse direction of flow (Supplementary Information section 8).”
The reason the authors “believe” admixture occurred from Antilles_Archaic into Haiti_Ceramic, instead of vice versa is obvious: if Antilles_Archaic indeed has ~30% Haiti_Ceramic related ancestry, then the component causing its divergence must be even deeper—likely deeper than the split between Ancient Beringians and Eurasians. They can ‘force’ supervised qpGraph modeling to conform to their prejudices. But Treemix is unsupervised, and it indicates admixture from Haiti_Ceramic into Antilles_Archaic and not vice versa. The fact that Early_Archaic (n=2, 3150BP) samples are more divergent from surrounding Caribbean populations than the Late_Archaic (n=11, 800 BP) samples, only lends more weight to the Treemix inference of [sustained] geneflow from neighboring populations into the Archaic group (though perhaps increased RoH over time could be another factor). In the expanded fstx100 table in the addenda, the “Early_Archaic” samples have pairwise values vs other ancient Caribbeans ranging from 11.15 to 13.04, compared to reduced values of 10.38 to 12.16 for the “Late” samples. This is expected if Antilles_Archaic absorbed admixture between the Early and Late periods, and it further undermines any theories of long-term isolation.
Finally, the lone high-coverage Antilles_Archaic individual found outside of Cuba was from Andres in the Southeast Dominican Republic. This individual was the only sample in the entire paper that had a strong “PopY” ancestral signal:
When Onge was used as the Australasian proxy, several of our ancient groups showed weak positive statistics (Z > 2), but only the Archaic individual I10126 from the site of Andrés in the Dominican Republic (Z = 3.4) surpassed our threshold of Z > 2.8 (Extended Data Table 5).
Cuba is only ~100 miles from both the Florida Keys and the Yucatan Peninsula, and even closer to Hispaniola and the Bahamas. It’s highly unlikely GreaterAntilles_Archaic (which only dates to 3150-800 BP!) represents some unadmixed Mesolithic population that’s been isolated for many millenia. While Dominican_Andres (n=2) forms a clade with Cuba_Carajo, and Haiti_Ceramic ancestry is present in both samples, only the Dominican sample possesses PopY ancestry. These patterns point to long-term genetic substructure-- not isolation-- across the Antilles_Archaic clade. That is what makes its “divergence” even more significant: whatever is causing it must be a very, very “deep” American population signal—[B]and this population has seemingly no analogs, ancient or modern, in Eurasia, Oceania, or Africa.
In reality, there is no ANC-A and ANC-B, or NNA/SNA, or any other "15-10kya Beringian splits" conjured up by qpGraph for the Americas. The Americas obviously have a very "deep" population history, which must reach back much further than the Mesolithic.
Between 200,000 and 100,000 years ago, a subset of this original [East Eurasian] hominid population migrated to the New World (via the Bering Strait Land Bridge), where speciation into modern humans occurred. While, under out-of-America II, the unique social behaviors shared between modern humans and New World primates (pair bonding, paternal investment, cooperative breeding and speech) are interpreted as homoplasies, the fact that such key aspects of human social and cognitive behavior are shared with Platyrrhines suggests that the immediate ancestors of modern humans were exposed to the same New World environment as the New World monkeys.
Buckle up. In this thread I will present comprehensive evidence for a modified version of this theory, except instead of some archaic “Eastern” hominid, my contention is that it was homo sapiens who trekked from Africa through Asia and into America by around 100kya-- and later back-migrated into Eurasia and Africa. The Apidima-1, Misliya-1, Skhul & Qafzeh, and Al-Wusta remains all demonstrate human presence outside of Africa in the 215-100kya time-frame. Further east, Chinese researchers have discovered what they describe as 47 fully modern looking human teeth in South China dated 80-120kya (https://www.sciencemag.org/news/2015/10/trove-teeth-cave-represents-oldest-modern-humans-china), showing that humans made it at least this far. The Tongzi teeth (also from South China) were dated 172-240kya (https://www.sciencedirect.com/science/article/abs/pii/S0047248418303464#fig3), although their provenance is more dubious. There seems to be a definite homo sapiens presence in North India at least 80kya (https://www.thehindu.com/sci-tech/science/early-humans-lived-in-northern-india-80000-years-ago/article30924041.ece) and some type of advanced hominid in Australia as early as 120kya (http://ancientnews.net/2019/03/11/archaeology-places-humans-in-australia-120000-years-ago/). And there is one very suggestive hint found in ancient DNA: the Altai Neanderthal has modern human admixture that introgressed approximately 100kya (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4933530), while the more westerly Croatian Vindija Neanderthal does NOT. Hitherto many (including myself) have just regarded most of these “early” OOA populations as genetic dead-ends…
It’s true, we don’t have definitive evidence of AMH in the Americas anywhere near deep in history as I’m suggesting (100kya or so), but I would posit this is partly due to lack of initiative. Only a century ago the “consensus” was that the Americas were populated no earlier than 5kya. Only in the 1960s or so was the date pushed back to 10-15kya or so with the acceptance of the Clovis Culture. But in recent DNA papers to make sense of American diversity, this date has been pushed back to 20kya or or 25kya—even though the Bering Land Bridge was closed between 40-15kya (https://www.youtube.com/watch?v=USIAcXfv39k). And how much “deep” archaeology has really been undertaken in the Americas compared to Africa or Europe? For that matter, even East Asia has been mostly terra incognito-- the first Denisovan was discovered there only in 2010, vs. the mid-19th century for European Neanderthals.
In terms of affirmative evidence, the Bluefish Caves site in Canada shows definite human presence dating back to 22,000 BP (https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0169486). Meadowcroft Rockshelter in Pennsylvania has been dated to at least 16,000-19,000 BP (https://www.sciencedirect.com/science/article/pii/S0305440313000733?via%3Dihub)(objections that it’s “only” 14,000 years old have been rejected by third-party radiocarbon dating). There is pretty obvious evidence of bone marrowing (by some type of hominid) dating back ~32,000 BP in Pendejo Cave, Mexico (https://web.archive.org/web/20060414084853/http://www.umass.edu/anthro/chrisman/page2.html), and earlier evidence of tool use from this cave dated back to 50kya. The Toca da Tira Peia rock art site in Brazil has been dated using luminescence to 22,000 BP (https://www.sciencedirect.com/science/article/pii/S0305440313000733?via%3Dihub)(with contested earlier dating as deep as 36kya (http://www.bradshawfoundation.com/south_america/serra_da_capivara/index.php)). The most famous pre-Clovis site is Monte Verde in Chile, with the now mostly accepted date of 18,500 BP. However the deepest layer of Monte Verde (MV-I, where burnt charcoal was found) was dated to 33,000 BP. Deepest of all, the Hueyatlaco butchering site (https://en.wikipedia.org/wiki/Hueyatlaco)has been dated to 250kya by both uranium-thorium and fission-track dating, although both younger and older proposed dates using other methods range from 80kya to 780kya. The Xalnene Tuff footprints (https://en.wikipedia.org/wiki/Xalnene_Tuff_footprints) have a similar dated range-- from the mid-Lower Paleolithic to the early Upper Paleolithic. Depending on which dating method is more accurate, Hueyatlaco and Xalnene Tuff could represent early humans, or maybe some archaic hominid. Either way, these sites demonstrate the mobility and ecological range of hominids -- a lesson that somehow many still refuse to accept.
While the existence of pre-Clovis American cultures are now widely accepted, we still lack definitive pre-Clovis DNA—but hints of deeper layers of American ancestry have already been published--nearly unnoticed. There is the Amazonian PopY, the Mesoamerican “UPopA”, and the recently discovered Antilles_Archaic. The latter two of these clades appear to harbor "deep" ancestry that has no close relationship with any of their supposed recent ancestors in Eurasia. In other ancient American samples, clades are inferred to "split" from neighboring populations (often in the absence of physical barriers) and live in isolation for 5k, 10k, or even 15k years--this is the only way qpGraph can explain the differential geographic patterns of drift. The large FST ratios between modern Amerindian populations have been explained by the post-Columbian collapse, whereby many Native groups lost up to 90% of their diversity, increasing homozygosity and skewing their inter-population relationships. But the ancient Amerindians we have on record also are much more diverse than expected—although qpAdm/QpGraph modeling has papered over this. Before examining the evidence in America however, let’s review some of the inconsistencies and shortcomings of the “orthodox” OOA theory as currently constituted:
The conventional Eurasian phylogeny (~60-80kya Out of Africa—or up to 150kya OOA if you believe in “Basal Eurasian”-- followed by a ~50kya “East-West Eurasian split”) has been shaken as incongruous new data comes in: “Western” Yana RHS is near Beringia, Belgian Goyet-Q2 has “eastern” affinities, Bulgarian Bacho Kiro has “Andamanese” mtdna, Tianyuan is EQUALLY related to E Asians and Amerindians, and various DNA and archaeological evidence exists for Australasians in Sahul BEFORE the proposed East/West OOA split. Why is fossil evidence of undisputed “modern” humans found in Upper Paleolithic Northeast Asia and Australia before Africa and the Near East? How do the Americas have such astounding linguistic diversity (even after post-Columbian devastation!), if the bottlenecked 15-10kya Beringians were really the First Americans? What is the “Pop Y” signal in Amazonians, if not a relic from a “deep” Old American population that must’ve been BASAL to (and not DERIVATIVE from) Oceanians? How is it possible Ust Ishim was in Siberia 45kya, Mungo Man in Australia 40kya (probably earlier, according to other evidence), modern human cave art in Borneo ~40kya, and Tianyuan in China 40kya, while the last Neanderthal holdout was Gibraltar (28kya)—and only then were they replaced from the northeast, not from right across the Straits? Why were Neandersovans wiped out so decisively by 40kya across most of Eurasia, while the West African Iwo Eleru and East African Wadi Halfa and Jebel Sahaba skulls show pronounced ‘archaic’/pre-modern features deep into the Mesolithic post-15kya?
Why can't we even find any modern human sites in pre-Mesolithic Africa? The only example I can see is the Hofmeyr Skull (36kya), which has long been said to morphologically resembler UP Eurasians. The Willerslev lab is reportedly working on getting DNA out of this. As of October 2019 (https://elmuseumscience.wordpress.com/2016/07/26/hofmeyr-skull-research-update/#comment-659), the first attempt was unsuccessful, but they're trying again.
So how could an Out of America hypothesis possibly "fit" with what we know about ancient and modern DNA? Something like this:
a) 300kya-200kya – Anatomically modern humans (Jebel Irhoud, Omo I, Florisbad) emerge and diversify across Africa. Development of y-haplogroup A00 and eventually, A
b) 200kya – 100kya – Out of Africa occurs (Apidima-1, Misliya-1, Skhul & Qafzeh, Al-Wusta etc) but while early humans are unable to challenge Neanderthals deep into Europe, they ARE successful penetrating deep into East Eurasia. One of these “eastern” OOA groups reaches the Americas no later than 100kya - 75kya (and possibly earlier), when there was still a land bridge in Beringia (https://www.youtube.com/watch?v=USIAcXfv39k). These Old American humans might have carried the CT macrohaplogroup, or that mutation could have developed soon after in the Americas.
c) Sealed off from Eurasia until 50kya by re-formation of the Beringian/Canadian ice sheets, Old American groups further diversify and spread through the Americas. Dziebel’s theory was that the entire “behavioral modernity” package developed in the Americas—language, symbolic/pictoral art, pair bonding etc. While these latter 2 elements are challenging to prove, the incredible linguistic diversity of the Americas does suggest that language as we know it originated with one or more of these Old American groups—and this could have formed their decisive advantage over other humans and archaic hominids.
d) 50-40 kya – Groups of “Old American” modern humans pour back west across Beringia, as soon as the land bridge was reopened, quickly spreading throughout Eurasia and Oceania and rapidly replacing Neandersovans along the way. The Old Americans, IOW, would be “basal” to all Upper Paleolithic Eurasians (“Crown Eurasians”), the latter being a subset of Old American diversity. In Southwest Eurasia and Africa at the terminus of their back-migration, Crown Eurasians admix with groups of “Basal Eurasians” and “Basal Africans”— their human cousins who never left Africa/SW Asia, in situ descendants of some of the early humans mentioned in a) & b). West Eurasians (and possibly Horners) can probably be modeled as varying mixtures of Crown + Basal Eurasian, while West, Central, and South Africans would have Crown, BE, BA, and additional archaic admixture. This would account for the “basal” phylogenetic position of sub-Saharan Africans. It would also explain why linguistic diversity generally decreases as one moves west from the Americas to Africa.
e) ~39kya – The Beringian land bridge closes again as the Laurentide ice sheet re-forms, pushing to its maximum southern extent ~21kya as the ice caps stretch from New England to Tennessee to Vancouver. This drives the migration of northerly Old Americans south, admixing with southerly Old American groups. Further admixture events occur— ie, possibly a ~14kya Clovis expansion associated with y-dna Q-M3, the 5.4kya Mesoamerican agricultural expansions both north and south (supported by aDNA), the Numic expansion, et cetera. These admixture events dampen some (but NOT ALL, as I will demonstrate!) of the signal of Old American inter-population diversity.
f) 15-10kya – As ice sheets retract and the land bridge reopens, some groups of Americans migrate back into Beringia/Siberia (ie, 10kya USR-1, 9.8kya Kolyma-1) while some Siberian groups move east. This American/Siberian exchange continues sporadically throughout history (Eskimos, Na Dene expansions, etc).
From the existing stock of published data, I hope to show that the preponderance of evidence is right under our noses—and the intent of this thread is to simply connect the dots.
The Big Picture
The Orthodox Out of Africa model has been partially bolstered by population geneticists’ widespread adoption of qpAdm/Graph over older tools like Treemix and FST (fixation index). QpGraph has fostered the development of more and more “ghost” populations to explain Eurasian and especially American diversity, and as this tool is “supervised” and subject to the assumptions of the user, I maintain that it has obscured the true state of affairs. While FST and Treemix may have their own flaws and limitations, they've been basically marginalized as addenda exhibits, while qpAdm/qpGraph has dominated the main analyses and topology-construction. FST has always indicated that Amerindians are the most distant group of humans from Africans—substantially more distant than Northern Europeans and non-Siberian East Asians are—which makes no sense if Amerindians are simply an ANE+East Asian composite. We are told these inflated Amerindian FST values are simply artifacts of demographic oddities—post-Columbian population collapse, long periods of isolation, etc. I aim to demonstrate that these explanations are not entirely convincing.
The 2015 paper "The Equidistance Index of Population Structure (https://www.biorxiv.org/content/biorxiv/early/2015/12/08/033852.full.pdf)" contained a basic FST matrix (in red) with selected populations:
https://i.imgur.com/abnZuzw.png
The authors noted:
"The Surui and Karitiana have an unusually high pairwise FST. In fact, the Karitiana are as diverged from the neighboring Surui in terms of FST as they are from the Mongola on the other side of the world (Table 1, Figure 3, and Figure S6). Moreover, FST actually decreases initially with distance from the Amazon, from 0.13 between the two Amazonian tribes, to 0.08-0.1 between Amazonians and Colombians, further decreasing to 0.07-0.09 between Amazonians and the more distant Maya. Remarkably, the highest FST among all HGDP Native American populations is between the two geographically closest populations, the Surui and Karitiana. These apparent anomalies can be explained by the inflation of FST in genetic isolates. FST between pairs of isolates can be nearly twice as high as between either one of the isolates and a more cosmopolitan population, as pairwise FST reflects the combined isolation of both populations. Since the Surui and Karitiana are both isolated, their pairwise FST is nearly double that between any one of them and a larger, less isolated population such as the Maya. In other words, the Maya’s contribution to the pairwise FST is dwarfed by that of the Amazonians."
To test their theory on isolates: the Maya number in the millions. Yet their FST with Surui is still 0.09, larger than the Maya:Russian FST. And even if we exclude the isolated Amazonians, the Pima:Colombian FST is still equivalent to the Russian:Yakut FST. (Both Colombians [Piapoco + Curripaco] and Pima each number ~20,000 --possibly larger than their pre-contact population). And if "isolation" is responsible for skewing the FST ratios-- what about the San? According to the traditional phylogeny they've been isolated for 250k years, save for some minor Afroasiatic ancestry absorbed in the Iron Age. Yet the San:Bantu FST doesn't show the hallmarks of two groups that have been separated 200kya+.
The authors came up with an alternate statistic "Est" (the green matrix) to try and "rectify" these patterns, but this just seems to muddle things further. Due to these "issues", population geneticists have generally abandoned FST as a priority informative measure. But here we will stay on the case:
In "Insights into human genetic variation and population history from 929 diverse genomes (https://science.sciencemag.org/content/367/6484/eaay5012)" from Bergstrom et al, the authors used modern DNA from the HGDP broken into geographic population clusters. They ran Fst statistics using 6 different allele array sets, but all 6 arrays demonstrate the same basic patterns that all approaches to quantifying human diversity relying on fixation indices have shown:
Estimates of FST for a selection of population pairs, grouped by region, using the Li 2008 array sites (“li2008”), the Human Origin 2012 sites (“ho2012”), the MEGA array sites (“mega”), all variants discovered in the sequencing data (“all”), all variants excluding singletons (“all.c2”) and variants ascertained in archaic genomes (“archaics”). Each line connects the FST values for one pair of populations across the different set of sites
https://i.imgur.com/E7eHQxq.jpg?1
1) Americans are the most divergent group of humans relative to Africans, Oceanians second, East Asians the third-most divergent and West Eurasians (Europeans and Middle Easterners) the closest group to Africans.
2) Americans are more closely related to East Asians than to Europeans, BUT except near the extremes (ie, Yakuts on the one hand and Sardinians on the other hand) East Asians are more closely related to Europeans than to Americans
3) East Asians are [substantially] more closely related to Europeans than to Papuans or Australians
4) Africans are the most divergent of human groups and have famously vast intra-group diversity (ie, San vs San or Bantu vs Bantu) relative to Eurasians. Yet the inter-population distance between Africans is somehow lower than the distance between Africans and Eurasians (see the first FST table above: San:Bantu FST is ~4x smaller than San:Surui and 2.5x smaller than San:Russian). This makes no sense if South Africans split from all other humans 250kya-300kya.
5) Some American populations have higher inter-population Fst ratios between them than between Africans (Fst for Surui:Karitiana/Pima/Colombian/Maya > Bantu:San).
All five of these points are incompatible with the basic phylogeny presented in the orthodox OOA theory. However, the Out of America theory I have outlined provides a coherent explanation for every point above.
Regarding 1), West Eurasians’ affinity with Africans has been explained by “Basal Eurasian” ancestry, and this may be partly true. However, Basal Eurasian admixture (which may be as low as ~10% in North Euros, assuming Anatolia_N is ~25% BE) cannot sufficiently explain the magnitude of this attraction. And since East Asians lack Basal Eurasian ancestry, why are they also substantially closer to Africans than Oceanians, Siberians, or Amerindians are?
The answer to these questions is hinted at by a very significant recent paper by Cole et al (Ancient Admixture into Africa from the ancestors of non-Africans (https://www.biorxiv.org/content/10.1101/2020.06.01.127555v1)) which UPenn’s Ian Mathieson and Max Planck Institute collaborated on:
Here we present SMCSMC, a Bayesian method for inference of time-varying population sizes and directional migration rates under the coalescent-with-recombination model, to study ancient demographic events. We find evidence for substantial migration from the ancestors of present-day Eurasians into African groups between 40 and 70 thousand years ago, predating the divergence of Eastern and Western Eurasian lineages.
……
We find that the method infers high rates of migration from descendants of the OoA event (’non-Africans’) to Africans, but not in the opposite direction, in the period 30–70kya corresponding to the Late Middle Paleolithic…. Inferred IMFs are not significantly different between Han Chinese and European populations in non-Afroasiatic populations (p=0.14, two-tailed paired t-test; Figs. 1h and S6, Table S1), consistent with [back-]migration occurring before the European-East Asian split approximately 40kya.
Now, this back-migration admixture was inferred to occur 30-70kya—the 40kya figure is merely the authors setting a baseline for the latest possible “East-West” split in accordance with the Orthodox OOA. Of course in the Out of America scheme there really is no distinct, dateable “East-West” split: East and West Eurasians would represent different subsets of Old American diversity that only gradually coalesced into discrete geographic clusters over time, with later West Eurasians (maybe as early as WHG or even earlier) especially distinguished by Basal Eurasian admixture. And if the HGDP European test pops share Basal Eurasian ancestry w/Africans (the authors stress their method lacks the power to pick up discrete signals older than 70kya) this could skew the inferred back-migration timeframe deeper than necessary.
The nature of this back-migrating population (“equally related to Han and French”) is speculative, but my guess would be something with affinities to one or more Upper Paleolithic Europeans on the one hand, and the ambiguous yet apparently deep “Eastern” ancestry that seems to comprise ~30% of Iran_N’s genome on the other. As for y-dna, how about DE and maybe LT (with L and D gradually accumulating in Near East/Turan/S Asia while E and T play the largest role in Africa). Remember those six basal E1b1b (and possibly one E1c) samples found in Pakistan_Udegram? Or Narasimhan’s finding that Indus Valley farmers were basal to Iran_N?
Now, this Bayesian method attributed the following “Eurasian back-migration” admixture proportions to these African ethno-linguistic groups:
Afroasiatic: 50% / Nilo-Saharan: 41% / Niger-Kordofanian: 35% / Mbuti-Biaka: 27% / Khoe-San: 21%
(Afroasiatics also have substantial post-Holocene West Eurasian ancestry, not included above)
Consider the following table I put together to model the latter 3 African groups. The highlighted row copies the “[Crown] Eurasian back-migration” proportions directly from the Cole et al paper (I am assuming this population split from other Crown Eurasians ~40kya). These components below, I will stress, are speculative—there may be more or less Basal/Archaic populations, some Basal & Archaic lineages could have been already admixed or discrete, the split times may vary, etc. It is merely one potential instructive scenario. Note that while my topology is radically different, the aggregate split times are generally in agreement with Harvard’s latest work in the Shum Laka DNA paper (Central African & South African HGs inferred to split ~250kya, but modern Mbuti & Khoisan have admixture from less basal groups—~26% Bantu+17% Afroasiatic-related for the former and 10-20% Afroasiatic for the latter—hence their shallower projected splits below):
https://i.imgur.com/nEANSWa.jpg?1
Assume the Crown Eurasian “Back-Migration pop” reaches the Near East 30-40kya (Levantine Aurignacian?), absorbing more and more “Basal Eurasian” and then “Basal African East” ancestry as it crosses the Red Sea. This group later fans out in waves across the rest of Africa, acquiring additional localized variants of Basal/Archaic ancestry. Now, all 3 of the African populations modeled above harbor lots of “deep” ancestry relative to Eurasians—hence their overall “basal” phylogenetic position, and the relatively high heterozygosity and intragroup genetic diversity observed between African individuals even within the same tribe. However, note how Yoruba and Khoisan share 65% of their ancestry in common—viz, components 1 through 3. At a population level, both Khoisan and Yoruba are predominantly of the same Crown + BE + BA_East stock. Moreover, they acquired most of this common ancestral “base” from a small founding group that coalesced in East Africa as late as the Mesolithic. And this explains why the pairwise FST between Khoisan and Yoruba is much smaller than the FST between Khoisan and non-Africans.
Since Crown Eurasians represent only a subset of Old American diversity, Amerindians have the most distant relationship with Africans. East Asians have moderate attraction to Africans because they’re more closely related than Siberians and Oceanians to the Crown Eurasian back-migration population. West Eurasians as well, but they have an additional attraction to Africans due to shared Basal Eurasian ancestry.
Regarding Oceanians, when the authors used Papuan instead Han or Europeans as the source for the back-migration, the proportions of inferred Eurasian ancestry in African groups dropped by an avg of ~3% across the board:
When we estimate these proportions using a Papuan sample to represent non-African descendants we find slightly but significantly smaller values compared to estimates using either the Han Chinese or European populations (mean difference of 0.029±0.002, p=9.2⇥10
They theorized that the excess Denisovan ancestry in Papuans was responsible for this decrease, which is reasonable. I do wish they had used Australians (who have "normal" amounts of Denisovan admix) as a test pop as well—another factor for Oceanians’ distance from Africans might be that ancestral Oceanians were the first Old American group to back-migrate west across Beringia, separating from other Crown Eurasians almost immediately (~50kya) and thus being more distal from the back-migration population. Alternatively, modern humans arriving in Sahul might have picked up minor admixture from a substrate of ancient human groups (see ie, some evidence of humans in Australia dating back to 65kya, or even 120kya).
Turning back to the Bergstrom paper, we can see in Figure 3: "Counts and properties of geographically private variants":
The highest frequencies [of private variants] are reached by a few tens of variants present at >70% (and a few thousands at >50%) in each of Africa, the Americas and Oceania. In contrast, the highest frequency variants private to either Europe, East Asia, the Middle East or Central and South Asia reach just 10-30%.
...
Even comparing Central and South America, we find variants private to one region but absent from the other reaching >40% frequency. Within Africa, ~1000 variants private to the rainforest hunter-gatherer groups Mbuti and Biaka reach >30%, and the highly diverged San of southern Africa harbour ~100,000 private variants at >30% frequency,
Within the framework of Out of America, these patterns are expected. But within the Orthodox Out of Africa framework the Americas shouldn't have more private variants than Eurasia, since American diversity would be merely a subset of Crown Eurasian diversity. But as the private variants demonstrate, the opposite is the case.
Below we can see the provenance of the private mutations: in Africans and Oceanians, their private mutations are shifted more towards "ancestral"/"archaic" as opposed to "derived" variants. But in the Americas, the highly "derived" and minimally "archaic" nature of the private variants shows they were very obviously related to the evolution of early modern humans. These private variants (and not merely "isolation") diversely distributed across Amerindian groups must contribute to the comparatively large FST ratios between Amerindian populations. And these variants take lots of time to accumulate: post-Colombian collapse and contractions in population sizes have nothing to do with the immense wealth of private gene variants in the Americas-- in fact the Columbian Exchange likely reduced the number of these variants.
https://i.imgur.com/bpCprnx.jpg
In the recent ancient East Asian DNA paper (https://anthrogenica.com/showthread.php?20302-Mammoth-Paper-from-Qiaomei-Fu-s-group-on-Genetic-history-of-East-Asia&p=666491&viewfull=1#post666491), the authors noted “genetic diversity was greater in these samples than in modern day East Asians”. Interestingly, in some ancient East Asian samples an affinity to Surui was detected, although this affinity was erased when restricting analysis to transversion sites. Moreover, the authors surprisingly failed to find Denisovan DNA in ancient or modern East Asians. While these results are contentious, is it possible the two findings are related? ie, that there is “deep” ancestry in East Asians related to Amazonians or other Old Americans, and this has been mischaracterized as “Denisovan”? Additionally, one of the Mesolithic East Asian samples had a slight but perceptible affinity to Villabruna.
These findings, along with the confusing affinities of various Upper Paleolithic Eurasians (mentioned earlier, ie, Bacho Kiro, Yana RHS, Goyet), and the fact that Europe has the lowest number of private variants in the world (shown above) help illuminate the true state of affairs: all of the Upper Paleolithic Eurasian diversity is radiating from the east, not from the south.
Let us now examine the ancient DNA from the Americas:
Ancient American Autosomal Diversity
The 2017 USR-1 paper revealed the closest analog to an ancestral Amerindian population in the “Ancient Beringians”.
“Ancient Beringian[s]… descended from a single founding population that initially split from East Asians around 36 ± 1.5 ka, with gene flow persisting until around 25 ± 1.1 ka. Gene flow from ancient north Eurasians into all Native Americans took place 25–20 ka, with Ancient Beringians branching off around 22–18.1 ka.”
USR-1 was deemed basal to Amerinds, though not a direct ancestor. The authors posited a North Amerind-South Amerind (NNA-SNA) split somewhere in Beringia or North America. This “clean split” theory is already dubious—12kya Anzick-1 is directly in the way of the proposed “inland” path for the NNA branch, yet falls in the “coastal” SNA clade.
As an aside, note how the split times between Amerinds & Eurasians, and even between USR1 and Amerinds, are continually pushed back beyond the elastic limits. Most climate modeling shows the ice-free land bridge across Beringia was only available for a brief time during this epoch—around 15kya to 10kya. Around 20kya there were still thick ice sheets from Beringia to Alberta. But now AB’s “split time” from Eurasians is pushed back (for the ANE ancestry to 20-25kya, and for the E Asian ancestry to 36kya), when these groups all would’ve been penned up together in Siberia. And pushing Ancient Beringians back to ~20kya forces the models to pin them along the narrow Beringian coastline in recognition of the ice sheet covering the Alaskan interior (and the interior is where USR-1 and Alaska_TrailCreek were actually found—but dated no earlier than ~ 10kya). The forced “coastal” route, skirting the ice, also makes the “clean split” of NNA/SNA ancestries difficult to conceive, spatially.
But let’s move on:
In June 2018’s Ancient human parallel lineages within North America contributed to a coastal expansion (https://science.sciencemag.org/content/360/6392/1024.full) from Scheib et al, the authors posited the existence of a two post-Beringian clades: one (ANC-B_) represented by ASO (Ancient South Ontario, dating to 500-3000 BP), and another represented by Clovis and South Americans (ANC-A). The distribution of these ancestries is downright bizarre: after ASO, the populations with the highest ANC-B ancestry are.... Chileans, from the Pacific coast of South America, with up to 75% ASO-related ancestry.
qpGraph also indicated West Eurasian admixture into ASO (“Canada_Lucier”), but the authors decided this was “modern contamination”:
The inferred 2% West Eurasian admixture into Canada_Lucier_4800BP-500BP is most likely explained by contamination in these samples by people of European ancestry.
We can surmise that all the usual techniques to remove contamination were adhered to across the dataset—and yet none of the other samples were asserted to have excess West Eurasian admixture. And in fact, in the Shuka Kaa (10kya western Canada) DNA paper from 2017, Treemix indicated an admixture edge from Shuka Kaa into West Eurasians (p. 11 of the Supp Info pdf).
Within Scheib et al, the authors included an ADMIXTURE plot. ADMIXTURE might be the least sophisticated measure of genomic diversity, however it's also unsupervised, and in cases where "deep" population history is less well known it can identify otherwise hidden patterns. For example, the "teal" component that showed up ADMIXTURE runs of West Eurasians was long the object of speculation. When the first Yamnaya were sequenced and revealed to be up to 50% "teal", speculation increased-- but all we knew is that Teal had some kind of affinity with the northern Near East/Caucasus. Of course it was not until the sequencing of CHG that "Teal" was revealed to be its own ancestral component that likely coalesced in the Mesolithic, with a complex Paleolithic history of which we're still unsure about. So for the Americas-- a vast geographic area with only several extant samples dating back to ~10kya-- ADMIXTURE can be very revealing.
At K=9, the colored ADMIXTURE components are (https://i.imgur.com/xJdTopb.jpg): Purple, Green, Orange, Beige, Magenta, Sky blue, Rust, Ocean blue, and Tan (or lime green? I’m mildly colorblind…)
The most “diverse” East Asians are those with Siberian-related ancestry: Oroquen, Hezhen, Mongola, etc. I have written before about how the American-Siberian exchange was more extensive and prolonged than generally accepted: for example, Nganassan-like ancestry first shows up in ancient Eurasia only in the Bronze Age (https://anthrogenica.com/showthread.php?20302-Mammoth-Paper-from-Qiaomei-Fu-s-group-on-Genetic-history-of-East-Asia&p=667534#post667534). This ancestry is obviously a recent arrival from the Americas that's only distantly related to other East Asians.
But south of Siberia, Han, Dai, Japanese etc are all modeled with 90-95% of the “Rust” component. In nMonte Han can be modeled pretty well as a 3-way mix of “Devils Gate” (“Mongolic/Tungusic”), Taiwan_Hanben (“Austric”), and NPL_Chokopani ("Tibetic") -type ancestries. Each of these components must have been the complex byproduct of alternating isolation and admixture events rooted in the Paleolithic peopling of East Asia. Yet compared to the Americas, the diversity between these mainland E Asian ancestral strains is apparently insignificant.
In contrast, there are 4 major ADMIXTURE components assigned to the Americas, compared to just 1 in E Asia and 2 in Oceania: “Beige” ancestry peaks in Californians and Mexicans, “Purple” in North Americans (not just Athabaskans w/recent Siberian ancestry, but also ASO/Ancient Ontario), while South Americans show up as either “Orange”, “Green” or some admixture therein. But the Americas actually harbor all 9 colored components in the Admixture run. Shuka Kaa has nearly 30% Sky Blue (“Onge”) ancestry, Anzick, ASO and Kennewick Man all score ~5% Ocean Blue (“Nganassan”), while Magenta (“Papuan”), Rust (“East Asian”) and “Tan” ancestries are found in trace quantities across geographically diverse American population clusters.
ADMIXTURE is demonstrating here what qpGraph, with its "subjective" topology, is obscuring: the Americas are more diverse than Eurasia.
The next big releases came within one week of each other in November 2018: Reconstructing the Deep Population History of Central and South America (https://www.cell.com/cell/fulltext/S0092-8674(18)31380-1), from Reich's Harvard group, and Early human dispersals within the Americas (https://science.sciencemag.org/content/362/6419/eaav2621.full), from Eske Willerslev's Copenhagen lab.
Reich's group rejected the ANC-A/ANC-B ancestral clades but came up with similar analogs: Ancestral-A and Ancestral-B. They rejected Ancient Ontario-related ancestry flowing into the Andes, explaining this affinity in other (opaque) ways-- basically multiple early split of the Ancestral-A ancestry, while Ancestral-B once again represents ASO. Yet in the Supp Data, we find that:
Although we were not able to reject the null hypothesis that ASO is equally related to Anzick1 and present-day populations (|Z| between 0.09 and 1.61), we observed that the Lagoa Santa population is most closely related to ASO (|Z| between 0.87 and 3.7), followed by the Spirit Cave genome (|Z| between 0.5 and 2.49) (Fig.2F). Together with the best-fitting model presented in (Fig.2F), these tests suggest that ASO carries admixture from a SNA population most closely related to Lagoa Santa and Spirit Cave that split after Anzick1. Thus, we consider that the apparent ASO-related ancestry in South Americans (7) could be explained by shared ancestry between Spirit Cave and/or Lagoa Santa and ASO. Spirit Cave and Lagoa Santa are both supposed to belong to the "Southern" clade that branches from Ancestral-A, like Anzick. Yet they're simultaneously most closely related to ASO from the Ancestral-B cade??? I recommend everyone to read the Supp Data to see the other puzzling stats and assertions from this study.
The Reich paper did include the first very old Central American/Mesoamerican samples: Mayahak Cab Pek (9300 BP) and Saki Tzul (7400 BP). As their neighbor joining tree indicates (https://i.imgur.com/vbEGIpm.jpg), these ancient Central Americans appear to be pretty divergent lineages. But there's surprisingly little analysis of these samples-- possibly because of low coverage, but also perhaps because they'd spoil their model.
The Willerslev group didn't have access to the ancient Mesoamerican genomes in their near-simultaneous paper's release. But merely by analyzing modern Mixe relative to to other ancients and moderns, they came to this surprising conclusion:
“further SFS-based modeling indicates that Mixe most likely carry gene flow from an unsampled outgroup, and form a clade with Lagoa Santa. Including nonzero outgroup admixture into Mixe when fitting an f-statistics-based admixture graph resulted in a significantly better fit (likelihood ratio test, P < 0.05) (Fig. 3, A and B ). Hereafter we refer to that outgroup as 'Un-sampled population A′ (UPopA), which is neither Ancestral Beringian, NNA or SNA, and which we infer split off from Native Americans ~24.7 ka, ranging from 30-22 ka (95% CI; this large range is a result of the analytical challenge of estimating divergence and admixture times in the absence of UPopA genome data). This age range overlaps with the inferred split of Native Americans from Siberians and East Asians 26.1-23.9 ka (1), and the divergence of USR1 from other NA (23.3-21.2 ka). This temporal overlap, which cannot be fully resolved into a relative sequence with current data, suggests there were multiple splits that took place in Beringia close in time.”
This UPopA autosomal signature is inferred to have formed when Amerindians’ ancestors were still in Siberia! So UPopA tagged alongside on the trip to Beringia, *without admixing with proto-Ancient Beringians*, then around 15kya AB splits into NNA-SNA branches, and UPopA once again tags alongside these two groups, settles directly between the NNA and SNA branches near the Panama isthmus—*again without admixing with SNA/NNA until more modern times* (since the UPopA signal was not detected in other ancient Amerinds, save for those that received post-5.4kya Mesoamerican-related admixture)
How absurd is this scenario? And if it somehow occurred, wouldn’t we then expect to see ancient or modern East Asians who share additional alleles with Mixe? But we don’t. The logical conclusion is that “UPopA” is an older American autosomal signature which must have been largely paved over by later admixture events.
The next surprising finding from this paper comes from western British Columbia:
While the coastal British Columbia ancient individuals clustered together with present-day Athabascan and Tsimshian speakers from the region, ASO and Kennewick were placed in an intermediate position between NNA and SNA. Although the Big Bar individual was placed close to NNA populations not carrying recent Siberian admixture (Fig. 1, C and F) (13), D-statistics of the form D(Aymara, NA; BigBar, Yoruba) and D(USR1, NA; BigBar, Yoruba) suggest Big Bar represents a previously undetected outgroup to non-[Ancestral Beringian] Native Americans, one that diverged prior to the NNA/SNA split (13)
.........
In this region’s long history, we find evidence that groups on the coast (e.g., 939) and their contemporaries in the interior (Big Bar), were as genetically distinct as are present-day groups (18) (Fig. 6A).
Now, Big Bar from British Columbia is dated to 5.4kya. It seems highly unlikely this individual spent 10,000 years as an unadmixed clade straight out of Beringia. The “contemporary” “NNA” clade formed by the “939/302/443” samples is just several hundred miles northwest, while the older Anzick-1 from the SNA clade was found only ~500 miles southeast. The nearest sample spatially is Kennewick Man (9kya) who is assigned as being a SNA/NNA admixture—or at least that’s how Kennewick appears on PCA. If Big Bar is admixed (which seems likely), whatever is causing its divergent ancestral profile would by definition have to be considerably “deeper” than the hypothesized NNA-SNA split.
The Willerslev paper also detected the presence of "PopY"-- an ancient Amazonian substrate with excess affinities to Andamanese relative to other Amerinds-- in the 7.4kya Brazil_Lapo_De_Santo. This ancient sample is inferred to harbor 3% PopY ancestry, compared to modern Amazonians (2%), whose PopY ancestry has been diluted by later Mesoamerican-related admixture.
In the most recent Amerind paper, A Genetic History of the pre-Contact Caribbean (https://www.biorxiv.org/content/10.1101/2020.06.01.126730v1), the authors discovered yet another very “deep” ancestral signature, this time in the hitherto-unsampled Caribbean (map of samples (https://i.imgur.com/0fhY7qV.jpg)). The Greater_Antilles_Archaic clade consists of 13 individuals from NW Cuba (3200 – 800 BP) and 2 individuals from the Dominican Republic (note: “Archaic” refers to the pottery style; these are not archaic hominids). The authors searched for affinities with other ancients and moderns and this was their conclusion:
“Using present-day populations, a similar result of no specific affinities was found . In qpGraph, we fit GreaterAntilles_Archaic in several positions along a skeleton admixture graph from Posth et al.23, again reflecting poor power to differentiate between a Central or South American origin (Supplementary Information section 11). Likewise, in a maximum likelihood phylogenetic tree based on allele frequency covariances, GreaterAntilles_Archaic is inferred to split before all South and some Central American populations. Together, these results point to the origin of GreaterAntilles_Archaic in a deeply divergent Native American population that is not particularly closely related to any sampled present-day groups.”
Here is a copy of their qpGraph attempt (https://i.imgur.com/1QjpWXW.jpg?1)
Now take a look at the FST(x100) table:
https://i.imgur.com/LAUACok.jpg?1
Now, In Eurasia the FST between Mesolithic populations is higher than in later populations-- agriculture seemed to always result in admixture (Neolithic farmers to Europe and European herders to Central/SW Asia, for example). These Caribbean populations are "young" , relatively speaking. BUt just for comparison, the FST between the Archaic samples and the other Caribbeans (.1082 to .1211) dwarfs the FST between WHG:EHG (.078), LBK:EHG (.084), WHG:Iran_C (.099), and EHG:Iran_C (.078), to name just a few ancient Eurasians who don't share recent geneflow. For that matter, even the Venezuela_Ceramic samples are quite divergent-- much more than expected under the terms of the Orthodox OOA.
“While short ROH measures population sizes of the past few dozen generations, the consistent signal across all Caribbean sites (including large and small islands) indicates that these estimates share some signal of larger meta-populations, as expected due to a high rate of mobility.
…..
“We also evaluated heterozygosity levels and found lower genetic diversity in the individuals from GreaterAntilles_Archaic and Haiti_Ceramic than in any of the other Ceramic-related (sub-)clades from the Greater Antilles, Curaçao, and Venezuela (Extended Data Fig. 2). These results agree with the median sums of ROH suggesting lower genetic diversity and a small population pool in the Archaic-related than in the Ceramic-related individuals. However, the Ceramic-associated Caribbean groups here studied were found to have an overall similar genetic diversity to that of some contemporary groups from continental South America, such as from the Peruvian Middle and Late Horizon periods (Nakatsuka et al. 2020).”
Note that Antilles_Archaic and Haiti_Ceramic have similarly low levels of heterozygosity/intragroup diversity—but Haiti_Ceramic didn’t exhibit the massive fst values vs other Caribbeans, or have nearly as much difficulty fitting into qpGraph as Antilles_Archaic.
From their Treemix run, a (~30%?) migration path is inferred from Haiti_Cermaic into Antilles_Archaic:
https://i.imgur.com/bMlJfQU.jpg
“Orange arrows represent admixture events, although the indicated direction of admixture might be inaccurate based on observations from analyses such as qpAdm admixture modeling - e.g., we believe it likely that there is GreaterAntilles Archaic admixture into Haiti_Ceramic rather than the reverse direction of flow (Supplementary Information section 8).”
The reason the authors “believe” admixture occurred from Antilles_Archaic into Haiti_Ceramic, instead of vice versa is obvious: if Antilles_Archaic indeed has ~30% Haiti_Ceramic related ancestry, then the component causing its divergence must be even deeper—likely deeper than the split between Ancient Beringians and Eurasians. They can ‘force’ supervised qpGraph modeling to conform to their prejudices. But Treemix is unsupervised, and it indicates admixture from Haiti_Ceramic into Antilles_Archaic and not vice versa. The fact that Early_Archaic (n=2, 3150BP) samples are more divergent from surrounding Caribbean populations than the Late_Archaic (n=11, 800 BP) samples, only lends more weight to the Treemix inference of [sustained] geneflow from neighboring populations into the Archaic group (though perhaps increased RoH over time could be another factor). In the expanded fstx100 table in the addenda, the “Early_Archaic” samples have pairwise values vs other ancient Caribbeans ranging from 11.15 to 13.04, compared to reduced values of 10.38 to 12.16 for the “Late” samples. This is expected if Antilles_Archaic absorbed admixture between the Early and Late periods, and it further undermines any theories of long-term isolation.
Finally, the lone high-coverage Antilles_Archaic individual found outside of Cuba was from Andres in the Southeast Dominican Republic. This individual was the only sample in the entire paper that had a strong “PopY” ancestral signal:
When Onge was used as the Australasian proxy, several of our ancient groups showed weak positive statistics (Z > 2), but only the Archaic individual I10126 from the site of Andrés in the Dominican Republic (Z = 3.4) surpassed our threshold of Z > 2.8 (Extended Data Table 5).
Cuba is only ~100 miles from both the Florida Keys and the Yucatan Peninsula, and even closer to Hispaniola and the Bahamas. It’s highly unlikely GreaterAntilles_Archaic (which only dates to 3150-800 BP!) represents some unadmixed Mesolithic population that’s been isolated for many millenia. While Dominican_Andres (n=2) forms a clade with Cuba_Carajo, and Haiti_Ceramic ancestry is present in both samples, only the Dominican sample possesses PopY ancestry. These patterns point to long-term genetic substructure-- not isolation-- across the Antilles_Archaic clade. That is what makes its “divergence” even more significant: whatever is causing it must be a very, very “deep” American population signal—[B]and this population has seemingly no analogs, ancient or modern, in Eurasia, Oceania, or Africa.
In reality, there is no ANC-A and ANC-B, or NNA/SNA, or any other "15-10kya Beringian splits" conjured up by qpGraph for the Americas. The Americas obviously have a very "deep" population history, which must reach back much further than the Mesolithic.