RSS
@razibkhan this looks like a v nice paper but do we really believe that Egyptian & Ethiopian genetics is reflective of what was there 60ka?
— Pontus Skoglund (@pontus_skoglund) May 28, 2015
There’s a new open access paper in AJHG, Tracing the Route of Modern Humans out of Africa by Using 225 Human Genome Sequences from Ethiopians and Egyptians, which is nice in that it uses state-of-the-art methods to analyze the genetics of a part of the world that warrants greater investigation. As the title of the paper implies the authors are focusing on a region which is likely the site of the exit of an ancient African population ~50-100,000 years ago which is responsible for over 90 percent of the ancestry of non-Africans. In short, they’re looking at the variation of modern populations across the region, and relating it to populations outside of the region, to infer historical relationships. This method has a long pedigree, at least by the standards of historical population genetics. About 15 years ago the Oxford geneticist Bryan Sykes wrote Seven Daughters of Eve, where he traced ancient European migrations to the most common mtDNA haplogroups in the continent. Using these results Sykes asserted that most of the ancestry of modern Europeans derives from Paleolithic hunter-gatherers; not Middle Eastern farmers. More precisely, modern Europeans exhibit overwhelming continuity with the Pleistocene populations. It turns out that this is wrong.
We know this because of ancient DNA, which is coming to various novel conclusions and overturning older understandings. One of them is that the genetic variation you see in a locale today has limited time depth into the past. That is why I state that it is likely that Cro-Magnons may contribute to less than 1 percent of the ancestry of modern Europeans. There are regions, such as the New World, where over the past 10,000 years genetic turnover on the whole has been modest, to negligible (most of the Holocene turnover in the New World before the arrival of Europeans is in northern North America). But this seems the exception rather than the rule. In South Asia, Africa, Europe, Siberia, East Asia and Southeast Asia, there is no dispute that the Holocene witnessed enormous changes in the genetic and demographic makeup of the dominant population. The flip side is that very ancient “archaic” lineages in some regions of the Eurasia have modern descendants. That is why I say we need to update our priors; the ancient branches of our family were mostly, but not entirely pruned, while many of the recent branches were mostly or even entirely pruned.
This brings me to the main question: how plausible it is that the genetic patterns on evidence in the paper in AJHG tell us about human evolutionary history with time depths of ~50,000 years. Color me skeptical. There are some specific issues that I’m confused by, in addition to the bigger framework. Greg Cochran has already put them into focus rather trenchantly. First, this section of the paper:
Using ADMIXTURE and principal-component analysis (PCA)18 (Figure 1A), we estimated the average proportion of non-African ancestry in the Egyptians to be 80% and dated the midpoint of the admixture event by using ALDER to around 750 years ago (Table S2), consistent with the Islamic expansion and dates reported previously.
A plain reading implies that 750 years ago non-African ancestry admixed into the population of Egypt so that it’s now 80% of the ancestry. Obviously this is insane. Egypt has a long history, and all the evidence that is not genetic indicates that ancient Egyptians were predominantly a population with Near Eastern and North African, not Sub-Saharan, affinities. The Roman era Fayum portraits suggest a people who resemble by and large modern Egyptians. Some do seem to have aspects of appearance which strike one as Sub-Saharan, but the presence of Nubians, as well as likely an ancient admixture event that occurred when Middle Eastern farmers arrived in the Nile Valley, can explain that. But when ascertaining the “Out of Africa” event you need to focus on the oldest element of ancestry. So you would have to look the people who contributed indigenous African ancestry well before the emergence of Egypt as a distinct civilization.
Here is the confusing part which inverts expectations. This last component is most likely to be within the “Non-African” segments of the Egyptian genome. I say this because the latest period of a mass population movement into Egypt from the Near East is ~8,000 year ago. 8,000 years is a long time, so recombination every generation would break apart the association between tracts of ancestry traceable to the newcomers, and that traceable to indigenous hunter-gatherers. Over time a new synthetic populatoin with its own distinctive population profile emerges. This is the case with South Asians, who are genetic compound of two very distinctive groups with extremely diverged histories. The latest evidence suggests that the admixture occurred on the order of ~4,000 years ago. That’s half the time depth of what likely occurred in ancient Egypt.
And about the African ancestry they did focus on, the 750 year time depth gives you a clue about where it came from: the rise of the Islamic empires and trans-Saharan trade enabled by camel triggered a massive influx of slaves from Africa into North Africa and the Near East (there was also an influx of slaves from the Caucasus and Central Asia, and for a time Europe, in large part because Islamic law banned the enslavement of believers). In the Maghreb these slaves were from West Africa. In the Persian Gulf the sources were diverse, but many were from East Africa. The natural source of Egyptian slaves is likely to be from the Sudan, what was ancient Nubia. Also, the Gumuz, who are used as a relatively unadmixed Ethiopian population (i.e., low Eurasian admixture fraction), are themselves of possible Sudanic origin and background!
I can agree that the Nubian/Sudanic ancestry exhibits a closer relationship to the population basal to non-Africans than West Africans. But, to me this paper does not make a strong case for a “northern” route through Egypt compared to the “southern” route, via the Bab-el-Mandeb. First, 50,000 years is a long time. My null assumption is that there has been enough population movement in Northeast Africa even before the Holocene to obscure the signal. Second, even without this consideration in mind, it strikes me that the African ancestry in Egyptians that they are focusing on is not a good geographic proxy in the first place, since it derives from Sudanic groups from further south. Finally, I do observe that this region of the world is relatively dry, making ancient DNA a possibility. So I have optimism that greater clarity will be achieved in the near future.
English
Arabic
Chinese (Simplified)
Dutch
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Persian
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Vietnamese
I wonder how good their masking method really is at distinguishing the African ancestry of Ethiopians and Egyptians. Beginning with the phasing, which may be more reliable with WGS, but is still subpar in comparison to using parent-offspring pairs to split up the segments definitively. Otherwise, you will always have the issue of alleles jumping between segments on different chromosomes, muddling the divergence between segments. I’ve seen it happen very clearly at 23andMe. With Egyptians having more non-African ancestry than Ethiopians, this mixing between segments would be more likely/common in Egyptian samples, with lower divergence from Eurasians as a result. To test this hypothesis, you would consequently expect the “SSA segments” in Amhara to also have a lower divergence from Eurasians than e.g. Wolayta, since the Amhara have more Eurasian admixture.
Furthermore, the African ancestry in Egyptians is recently mixed, not to mention the prehistoric migrations that would have taken place since 60 kya. Modern Copts lack West African admixture entirely, found at about ~5% in the regular Muslim Egyptian samples, clearly related to the 750 ya event. Incidentally, a similar difference is seen in Levantine Muslim/Christian communities. This is presumably what accounts for the “low” divergence of the Egyptian segments from West Africans (20 kya), in comparison to the ~40 kya Ethiopian-West African divergence (the latter fits nicely with the ~40 kya date of Y-DNA E-P2, and the timing of L3 expansions in West Africa).
You gotta love the professional courtesy from Skoglund: “this looks like a very nice paper but…”
Yeah, but, it totally reeks.
Interestingly, this is the second paper from the Wellcome Trust in only a few weeks making some very strange claims. That Ayub et al. Kalash paper was also from the Wellcome Trust.
Greg Cochran writes “everyone whose name is on the paper must be assumed to be equally ignorant and incurious – as well as the reviewers”. As a reviewer, I’d like to dispute that. Below is the full unedited text of my review:
“In this paper, the authors analyze new whole genome sequences from a set of Egyptian and Ethiopian populations.
The main observation is: after separating the genomes into “African” and “non-African” components, “African” haplotypes identified in Egypt are at higher frequency outside of Africa than are “African” haplotypes identified in Ethiopia. A related tentative observation–more recent coalescent times between “African” Egyptian haplotypes and non-African haplotypes–is confirmed via a separate method (MSMC), though the small difference between Egypt (masked) and Ethiopia (masked) in Figure 3 does not appear to be statistically strong.
The main conclusion made from this observation is that modern humans exited Africa to colonize the rest of the world via a route that went through northern Africa rather than eastern Africa.
Getting to this extremely strong conclusion from the initial observations involves a number of steps are not well supported. There are of course many reasons that Egyptian haplotypes could be at higher frequency outside of Africa relative to Ethiopian haplotypes that have nothing to do with the out-of-Africa expansion, most of which are mentioned and dismissed by the authors for reasons I find unconvincing. Most importantly:
1. There could have been gene flow from non-African populations into Egypt before the time frame that is detectable using their methods. The authors dismiss this possibility with “However, such confounding backflow would need to have taken place prior to the split between East Asians and Europeans (ca ~40kya) and, if this genetic component originated from the main OOA founding event, is likely to have been removed by the non-African masking procedure.”
I see no evidence that either of these claims is supported. That is: first, the authors present no evidence that confounding backflow need to have taken place prior to the split between east Asians and Europeans. Presumably this claim is based on the observation that some east Asian haplotypes match the Egyptian haplotypes. But any hypothetical migrant middle eastern population would also be carrying some haplotypes shared with east Asia as well, so my intuition is that gene flow at any time point would be confounding.
Second, the authors present no evidence that their masking procedure could mask out signals of older admixture, and I find it extremely unlikely that it could do so. As a hypothetical example, imagine gene flow from a Levantine population into Egypt 15,000 years ago (~500 generations ago). By today, the extent of admixture linkage disequilibrium will have decayed such that the average block of local ancestry is on the order of tens of kilobases. No local ancestry algorithm can even hope to detect something like this; these algorithms were designed for recent admixtures where local ancestry blocks are megabases long.
2. Asymmetry of errors in the local ancestry procedure. The authors are well aware that there has been extensive gene flow into both north and east Africa from non-African populations, and so mask out this ancestry. This masking procedure is of course imperfect, and the authors show that in simulations that error rates are indeed non-trivial (Supplementary Figure 4) [My error: this should read Supplementary *Table* 4].
The authors argue that “the two effects [i.e. error rates in Ethiopia and Egypt] are likely to cancel each other out.” But a perfect canceling out is unlikely, so the authors perform simulations to test whether an increased error rate in Ethiopia could lead to results similar to theirs, and conclude in the negative (Supplementary Figure 6). This Figure is puzzling-a simulated 50% mis-assignment rate seems to have basically little effect on the results. What if the authors simulated a 100% mis-assignment rate (i.e. label all “non-African” haplotypes in Ethiopia as “African”)? If this also has little effect on the results, then I suspect some aspect of the simulation is leading to misleading answers.
The details of this simulation procedure might be helpful, but none are given. For example, the authors say the simulation misclassified “CEU sites” as African. Does this mean individual SNPs? Or haplotypes?
3. Gene flow into Ethiopia from other places in Africa. Another plausible possibility is that the Ethiopian populations have been more affected by population movements within Africa itself. That is, the “African” component of ancestry in Ethiopia may be different from that in Egypt for reasons that are unrelated to the out-of-Africa event. It seems difficult to reject this possibility.
Other specific comments:
1. As the authors are likely aware, historians and anthropologists will be a bit dubious of the simplified demographic model for Egypt (for the purposes of this paper, Egypt can be modeled as a native African population representing the out-of-African population plus non-African gene flow 750 years ago). It is extremely unlikely that all gene flow into Egypt was due to the Arab conquest; for example, it was under Greek control ~2000 years ago and there is written historical attestation of extensive migration at that time. Additionally, there is historical evidence of various invasions and foreign dynasties during the pharaohs. Finally, it seems plausible that the Neolithic in Egypt involved the arrival of agriculturalist populations (as happened, for example, in Europe).
Models are of course simplifications, but it seems unlikely that the main conclusion of the paper is robust to these violations of the model.
2. I am unconvinced that the masking procedure has left the authors with a population that is free of more ancient non-African ancestry. A simple sanity check might be useful: If the authors compute a test of the population phylogeny [Yoruba, [Egypt,[Han, CEU]]] (using e.g. a D-statistic/four-population test), before masking this will not fit the data because of gene flow into Egypt. After masking I suspect it will still not fit the data, but this is testable.
3. In Figure 2B, the authors present the fraction of haplotypes in, e.g. CEU that are found in each of a number of sets of population (e.g. Egypt only, Egypt and Ethiopia, etc.). They present the enrichment levels for these fractions relative to “the general African population”. I was unsure what this meant, does “the general African population” mean Yoruba? Perhaps some mathematical notation in the methods would be helpful. That is, if we let f_ij be the frequency of haplotype i in population j, can the authors write out exactly how their enrichment level is defined? Without this, evaluating the presented number involves several bits of guesswork about exactly what is being calculated.”
https://twitter.com/razibkhan/status/604426015465603072
"In this paper, the authors analyze new whole genome sequences from a set of Egyptian and Ethiopian populations.
The main observation is: after separating the genomes into "African" and "non-African" components, "African" haplotypes identified in Egypt are at higher frequency outside of Africa than are "African" haplotypes identified in Ethiopia. A related tentative observation--more recent coalescent times between "African" Egyptian haplotypes and non-African haplotypes--is confirmed via a separate method (MSMC), though the small difference between Egypt (masked) and Ethiopia (masked) in Figure 3 does not appear to be statistically strong.
The main conclusion made from this observation is that modern humans exited Africa to colonize the rest of the world via a route that went through northern Africa rather than eastern Africa.
Getting to this extremely strong conclusion from the initial observations involves a number of steps are not well supported. There are of course many reasons that Egyptian haplotypes could be at higher frequency outside of Africa relative to Ethiopian haplotypes that have nothing to do with the out-of-Africa expansion, most of which are mentioned and dismissed by the authors for reasons I find unconvincing. Most importantly:
1. There could have been gene flow from non-African populations into Egypt before the time frame that is detectable using their methods. The authors dismiss this possibility with "However, such confounding backflow would need to have taken place prior to the split between East Asians and Europeans (ca ~40kya) and, if this genetic component originated from the main OOA founding event, is likely to have been removed by the non-African masking procedure."
I see no evidence that either of these claims is supported. That is: first, the authors present no evidence that confounding backflow need to have taken place prior to the split between east Asians and Europeans. Presumably this claim is based on the observation that some east Asian haplotypes match the Egyptian haplotypes. But any hypothetical migrant middle eastern population would also be carrying some haplotypes shared with east Asia as well, so my intuition is that gene flow at any time point would be confounding.
Second, the authors present no evidence that their masking procedure could mask out signals of older admixture, and I find it extremely unlikely that it could do so. As a hypothetical example, imagine gene flow from a Levantine population into Egypt 15,000 years ago (~500 generations ago). By today, the extent of admixture linkage disequilibrium will have decayed such that the average block of local ancestry is on the order of tens of kilobases. No local ancestry algorithm can even hope to detect something like this; these algorithms were designed for recent admixtures where local ancestry blocks are megabases long.
2. Asymmetry of errors in the local ancestry procedure. The authors are well aware that there has been extensive gene flow into both north and east Africa from non-African populations, and so mask out this ancestry. This masking procedure is of course imperfect, and the authors show that in simulations that error rates are indeed non-trivial (Supplementary Figure 4) [My error: this should read Supplementary *Table* 4].
The authors argue that "the two effects [i.e. error rates in Ethiopia and Egypt] are likely to cancel each other out." But a perfect canceling out is unlikely, so the authors perform simulations to test whether an increased error rate in Ethiopia could lead to results similar to theirs, and conclude in the negative (Supplementary Figure 6). This Figure is puzzling-a simulated 50% mis-assignment rate seems to have basically little effect on the results. What if the authors simulated a 100% mis-assignment rate (i.e. label all "non-African" haplotypes in Ethiopia as "African")? If this also has little effect on the results, then I suspect some aspect of the simulation is leading to misleading answers.
The details of this simulation procedure might be helpful, but none are given. For example, the authors say the simulation misclassified "CEU sites" as African. Does this mean individual SNPs? Or haplotypes?
3. Gene flow into Ethiopia from other places in Africa. Another plausible possibility is that the Ethiopian populations have been more affected by population movements within Africa itself. That is, the "African" component of ancestry in Ethiopia may be different from that in Egypt for reasons that are unrelated to the out-of-Africa event. It seems difficult to reject this possibility.
Other specific comments:
1. As the authors are likely aware, historians and anthropologists will be a bit dubious of the simplified demographic model for Egypt (for the purposes of this paper, Egypt can be modeled as a native African population representing the out-of-African population plus non-African gene flow 750 years ago). It is extremely unlikely that all gene flow into Egypt was due to the Arab conquest; for example, it was under Greek control ~2000 years ago and there is written historical attestation of extensive migration at that time. Additionally, there is historical evidence of various invasions and foreign dynasties during the pharaohs. Finally, it seems plausible that the Neolithic in Egypt involved the arrival of agriculturalist populations (as happened, for example, in Europe).
Models are of course simplifications, but it seems unlikely that the main conclusion of the paper is robust to these violations of the model.
2. I am unconvinced that the masking procedure has left the authors with a population that is free of more ancient non-African ancestry. A simple sanity check might be useful: If the authors compute a test of the population phylogeny [Yoruba, [Egypt,[Han, CEU]]] (using e.g. a D-statistic/four-population test), before masking this will not fit the data because of gene flow into Egypt. After masking I suspect it will still not fit the data, but this is testable.
3. In Figure 2B, the authors present the fraction of haplotypes in, e.g. CEU that are found in each of a number of sets of population (e.g. Egypt only, Egypt and Ethiopia, etc.). They present the enrichment levels for these fractions relative to "the general African population". I was unsure what this meant, does "the general African population" mean Yoruba? Perhaps some mathematical notation in the methods would be helpful. That is, if we let f_ij be the frequency of haplotype i in population j, can the authors write out exactly how their enrichment level is defined? Without this, evaluating the presented number involves several bits of guesswork about exactly what is being calculated."Replies: @Razib Khan, @Douglas Knight
2nd & 5th author seem to concur with my critique to some extent.
https://twitter.com/razibkhan/status/604426015465603072
. I say this because the latest period of a mass population movement into Egypt from the Near East is ~8,000 year ago.
This is inaccurate. There was no such “mass population movement” into Egypt. You reference Wikipedia, which is modified almost daily to fit assorted agendas, but even the Wikipedia reference says no such thing about any mass population movements. It notes some migration into Egypt, which is to be expected, (note it also mentions the North African Neolithic already in place as well), but the cited footnotes in Wikipedia ALSo do not claim any mass migration. One of the footnotes in a cluster re “migration” for example is Bar-Josef on the Natufians, but Bar-Josef does not even mention the word Egypt in his article, showing some of the sleight of hand “edits” going on, with multiple footnotes piled in to look “authentic.” People can see through what’s going on.
Domesticates from the Middle East were incorporated primarily by peoples already in place in Egypt, though as a country with its SInai border right into Palestine and beyond, Egypt always had small scale movement. In addition the founding of the Dynasties was from the tropical south not the north whichis closer to the “Middle East.”
Archaeological data (Wendorf 2001, Wettstrom 1999) suggests that the peoples of the Sahara had already independently domesticated cattle in the early Holocene eastern Sahara, followed by the gradual adoption of grain cultivation, or gradual adoption of Near Eastern domesticates into an already established foraging and subsistence economy, rather than some mass influx of outsiders bringing benefits to the indigenes. QUOTE from 3 peer-reviewed scholars:
and
Even a popular publication like National Geographic is skeptical about any “mass movement” claim.
This is inaccurate. There was no such "mass population movement" into Egypt. You reference Wikipedia, which is modified almost daily to fit assorted agendas, but even the Wikipedia reference says no such thing about any mass population movements. It notes some migration into Egypt, which is to be expected, (note it also mentions the North African Neolithic already in place as well), but the cited footnotes in Wikipedia ALSo do not claim any mass migration. One of the footnotes in a cluster re "migration" for example is Bar-Josef on the Natufians, but Bar-Josef does not even mention the word Egypt in his article, showing some of the sleight of hand "edits" going on, with multiple footnotes piled in to look "authentic." People can see through what's going on.
Domesticates from the Middle East were incorporated primarily by peoples already in place in Egypt, though as a country with its SInai border right into Palestine and beyond, Egypt always had small scale movement. In addition the founding of the Dynasties was from the tropical south not the north whichis closer to the "Middle East."
Archaeological data (Wendorf 2001, Wettstrom 1999) suggests that the peoples of the Sahara had already independently domesticated cattle in the early Holocene eastern Sahara, followed by the gradual adoption of grain cultivation, or gradual adoption of Near Eastern domesticates into an already established foraging and subsistence economy, rather than some mass influx of outsiders bringing benefits to the indigenes. QUOTE from 3 peer-reviewed scholars:andEven a popular publication like National Geographic is skeptical about any "mass movement" claim.Replies: @Razib Khan
yes, egypt is TOTALLY different from the rest of eurasia. *roll eyes*
And speaking of Egypt, I also have a separate OT request: I'd like to see your take on the origins of the "Sea Peoples": http://en.wikipedia.org/wiki/Sea_Peoples.Replies: @Razib Khan, @ohwilleke
Furthermore, the African ancestry in Egyptians is recently mixed, not to mention the prehistoric migrations that would have taken place since 60 kya. Modern Copts lack West African admixture entirely, found at about ~5% in the regular Muslim Egyptian samples, clearly related to the 750 ya event. Incidentally, a similar difference is seen in Levantine Muslim/Christian communities. This is presumably what accounts for the "low" divergence of the Egyptian segments from West Africans (20 kya), in comparison to the ~40 kya Ethiopian-West African divergence (the latter fits nicely with the ~40 kya date of Y-DNA E-P2, and the timing of L3 expansions in West Africa).Replies: @Labayu, @jack shindo
What’s the your source for modern Copts completely lacking West African admixture versus ~5% in Muslim Egyptians? I don’t doubt it, in fact I assumed something like that would be the case, but I might want to cite it at some point.
I can’t give you any source, since it’s based on the results of individual Copts who tested using a personal genomics service.
Mr. Khan, I find this post and the associated comments very interesting and informative, and would like to see (even if short) a substantive rebuttal from you to what “Enrique Cardova” commented about mass migration into Egypt (or lack thereof).
And speaking of Egypt, I also have a separate OT request: I’d like to see your take on the origins of the “Sea Peoples”: http://en.wikipedia.org/wiki/Sea_Peoples.
the genetic divergence of egyptians from other near easterners is not great enough. that's the simple fact of it. suggests recent common ancestry. not deep local lineages. if it isn't the arab migration, which few assume (given coptic genetic background), then logical conclusion is back migration from the levant.
The one group about which we have a pretty solid bit of information is the Philistines who ended up in the Southern Levant and were mentioned in the Hebrew Bible. They were one of the Sea Peoples and were almost certainly migrating Mycenean Greeks, based upon a combination of historical and archaeological evidence.Replies: @LevantineJew
And speaking of Egypt, I also have a separate OT request: I'd like to see your take on the origins of the "Sea Peoples": http://en.wikipedia.org/wiki/Sea_Peoples.Replies: @Razib Khan, @ohwilleke
Mr. Khan, I find this post and the associated comments very interesting and informative, and would like to see (even if short) a substantive rebuttal from you to what “Enrique Cardova” commented about mass migration into Egypt (or lack thereof).
the genetic divergence of egyptians from other near easterners is not great enough. that’s the simple fact of it. suggests recent common ancestry. not deep local lineages. if it isn’t the arab migration, which few assume (given coptic genetic background), then logical conclusion is back migration from the levant.
To what extent do Copts show other sub-Saharan ancestry, perhaps Nilotic?
No way, no how, can anybody deduce from the populations that are living today in Ethiopia and Egypt what was there 70,000 years ago when this migration occurred. The people seriously asking this question need to be educated on how dynamic human population changes have been over that long a period of time. They simply don’t know.
"In this paper, the authors analyze new whole genome sequences from a set of Egyptian and Ethiopian populations.
The main observation is: after separating the genomes into "African" and "non-African" components, "African" haplotypes identified in Egypt are at higher frequency outside of Africa than are "African" haplotypes identified in Ethiopia. A related tentative observation--more recent coalescent times between "African" Egyptian haplotypes and non-African haplotypes--is confirmed via a separate method (MSMC), though the small difference between Egypt (masked) and Ethiopia (masked) in Figure 3 does not appear to be statistically strong.
The main conclusion made from this observation is that modern humans exited Africa to colonize the rest of the world via a route that went through northern Africa rather than eastern Africa.
Getting to this extremely strong conclusion from the initial observations involves a number of steps are not well supported. There are of course many reasons that Egyptian haplotypes could be at higher frequency outside of Africa relative to Ethiopian haplotypes that have nothing to do with the out-of-Africa expansion, most of which are mentioned and dismissed by the authors for reasons I find unconvincing. Most importantly:
1. There could have been gene flow from non-African populations into Egypt before the time frame that is detectable using their methods. The authors dismiss this possibility with "However, such confounding backflow would need to have taken place prior to the split between East Asians and Europeans (ca ~40kya) and, if this genetic component originated from the main OOA founding event, is likely to have been removed by the non-African masking procedure."
I see no evidence that either of these claims is supported. That is: first, the authors present no evidence that confounding backflow need to have taken place prior to the split between east Asians and Europeans. Presumably this claim is based on the observation that some east Asian haplotypes match the Egyptian haplotypes. But any hypothetical migrant middle eastern population would also be carrying some haplotypes shared with east Asia as well, so my intuition is that gene flow at any time point would be confounding.
Second, the authors present no evidence that their masking procedure could mask out signals of older admixture, and I find it extremely unlikely that it could do so. As a hypothetical example, imagine gene flow from a Levantine population into Egypt 15,000 years ago (~500 generations ago). By today, the extent of admixture linkage disequilibrium will have decayed such that the average block of local ancestry is on the order of tens of kilobases. No local ancestry algorithm can even hope to detect something like this; these algorithms were designed for recent admixtures where local ancestry blocks are megabases long.
2. Asymmetry of errors in the local ancestry procedure. The authors are well aware that there has been extensive gene flow into both north and east Africa from non-African populations, and so mask out this ancestry. This masking procedure is of course imperfect, and the authors show that in simulations that error rates are indeed non-trivial (Supplementary Figure 4) [My error: this should read Supplementary *Table* 4].
The authors argue that "the two effects [i.e. error rates in Ethiopia and Egypt] are likely to cancel each other out." But a perfect canceling out is unlikely, so the authors perform simulations to test whether an increased error rate in Ethiopia could lead to results similar to theirs, and conclude in the negative (Supplementary Figure 6). This Figure is puzzling-a simulated 50% mis-assignment rate seems to have basically little effect on the results. What if the authors simulated a 100% mis-assignment rate (i.e. label all "non-African" haplotypes in Ethiopia as "African")? If this also has little effect on the results, then I suspect some aspect of the simulation is leading to misleading answers.
The details of this simulation procedure might be helpful, but none are given. For example, the authors say the simulation misclassified "CEU sites" as African. Does this mean individual SNPs? Or haplotypes?
3. Gene flow into Ethiopia from other places in Africa. Another plausible possibility is that the Ethiopian populations have been more affected by population movements within Africa itself. That is, the "African" component of ancestry in Ethiopia may be different from that in Egypt for reasons that are unrelated to the out-of-Africa event. It seems difficult to reject this possibility.
Other specific comments:
1. As the authors are likely aware, historians and anthropologists will be a bit dubious of the simplified demographic model for Egypt (for the purposes of this paper, Egypt can be modeled as a native African population representing the out-of-African population plus non-African gene flow 750 years ago). It is extremely unlikely that all gene flow into Egypt was due to the Arab conquest; for example, it was under Greek control ~2000 years ago and there is written historical attestation of extensive migration at that time. Additionally, there is historical evidence of various invasions and foreign dynasties during the pharaohs. Finally, it seems plausible that the Neolithic in Egypt involved the arrival of agriculturalist populations (as happened, for example, in Europe).
Models are of course simplifications, but it seems unlikely that the main conclusion of the paper is robust to these violations of the model.
2. I am unconvinced that the masking procedure has left the authors with a population that is free of more ancient non-African ancestry. A simple sanity check might be useful: If the authors compute a test of the population phylogeny [Yoruba, [Egypt,[Han, CEU]]] (using e.g. a D-statistic/four-population test), before masking this will not fit the data because of gene flow into Egypt. After masking I suspect it will still not fit the data, but this is testable.
3. In Figure 2B, the authors present the fraction of haplotypes in, e.g. CEU that are found in each of a number of sets of population (e.g. Egypt only, Egypt and Ethiopia, etc.). They present the enrichment levels for these fractions relative to "the general African population". I was unsure what this meant, does "the general African population" mean Yoruba? Perhaps some mathematical notation in the methods would be helpful. That is, if we let f_ij be the frequency of haplotype i in population j, can the authors write out exactly how their enrichment level is defined? Without this, evaluating the presented number involves several bits of guesswork about exactly what is being calculated."Replies: @Razib Khan, @Douglas Knight
Better to keep silent and be thought a fool.
Too few samples to really say, but their East African looks comparable to regular Egyptians, if not slightly lower.
how many? would be neat to confirm with tree mix or admix tools copt vs. non-copt (which there are some public data sets) egyptians.
Furthermore, the African ancestry in Egyptians is recently mixed, not to mention the prehistoric migrations that would have taken place since 60 kya. Modern Copts lack West African admixture entirely, found at about ~5% in the regular Muslim Egyptian samples, clearly related to the 750 ya event. Incidentally, a similar difference is seen in Levantine Muslim/Christian communities. This is presumably what accounts for the "low" divergence of the Egyptian segments from West Africans (20 kya), in comparison to the ~40 kya Ethiopian-West African divergence (the latter fits nicely with the ~40 kya date of Y-DNA E-P2, and the timing of L3 expansions in West Africa).Replies: @Labayu, @jack shindo
Egyptians are Africans just as Italians are Europeans. Haplogroup variance often shows itself based on genetic drift, new inccoming populations (Mamluks from Albania? recent) and various groups (Greeks?) and their degree of assimilation into the social milieu
It may suggest that the Copts, at one time, were a majority but with the spread of Islam amongst the less fortunate using religion as a war cry, enabled various groups from various parts of the African continent to consolidate against the more sedentary Coptic population who were unable to fend of the warrior revenging soldiers of Allah!
The woman on the portrait look Helenistic/Levantine to me.
[1] http://en.m.wikipedia.org/wiki/Fayum_mummy_portraits#Subjects_and_social_context_of_the_paintings
I had the same instinct as the OP at first read.
The region has had multiple rounds of mass migration:
1. One ca. 30 kya brining mtDNA M1 and U6 to Africa.
2. One ca. 8 kya with the first wave Neolithic that increased population density in Egypt by about 100x.
3. Substantial trade and migration in the Bronze and Iron Ages facilitated by Egypt’s Imperial reach (much of which was limited to the Nile Delta, however). Of particular importance, was the Hyskos Dynasty which is a historically attested period of rule by outsiders that could have been accompanied by mass migration.
4. The Islamic Empire.
5. Migration in connection with British colonial rule.
This said, in Ethiopia, near the source of the Southern Route, the Neolithic arrived much later and the migration attributable to first wave Neolithic and Ethio-Semitic influence (and later) can probably be peeled away statistically, leaving only Migration (1) above as a major, post-Out of Africa migration.
It is also likely that the number of waves of migration and the extent of their impact is smaller in Southern Arabia, than in the Levant or Egypt. But, lack of archaeological and historical data until fairly late in the historic era still leaves question marks. The fact that we don’t have clear data on the demic impact of herding in Southern Arabia doesn’t mean that it didn’t happen and we have good ancient DNA data from the Eurasian steppe that herding societies can give rise to mass population genetic replacement in even a single wave. We also know that Yemen was a center of maritime Indian Ocean trade dating back to the Copper Age.
One way to do a reality check would be to look at the TMRCA of Southern Arabian populations based on contemporary genetics. If the TMRCA << 40,000 years ago, it is a fair guess that there has been near complete population replacement there.
We do not have Out of Africa era data archaeology from Arabia, most of which isn't human remains, but can be correlated with contemporaneous African sources. And that data, whichever direction it points, is probably far more reliable.
Ethiopia also received major migration from Sub-Saharan Africa to Ethiopia in the time period after Out of Africa. It probably received some around the time of Bantu expansion.
But, it also probably received some migration earlier, particular of Sahel farmers from West Africa that helped bring about the integration of Ethiopian wild type domesticates and West African Sahel wild type domesticates into a common Neolithic Sahel crop package. This was sometime in the vicinity of 6,000 to 4,000 years ago, but the data is pretty thin.
Thus, Ethiopia may have had more regionally distinct population genetics ca. 125kya to 50 kya, at the time of Out of Africa, than its African component does today.
And speaking of Egypt, I also have a separate OT request: I'd like to see your take on the origins of the "Sea Peoples": http://en.wikipedia.org/wiki/Sea_Peoples.Replies: @Razib Khan, @ohwilleke
As the terminology you use suggests, the “Sea Peoples” were likely more than one group.
The one group about which we have a pretty solid bit of information is the Philistines who ended up in the Southern Levant and were mentioned in the Hebrew Bible. They were one of the Sea Peoples and were almost certainly migrating Mycenean Greeks, based upon a combination of historical and archaeological evidence.
The Philistines (Plishtim in Hebrew, PLST in Ancient Egyptian) were descendants of Sea People, just like tribe of Dan and both were related to Aegean or Greek civilization. Yet they were enemies, while Dan joined the Israelite tribe confederation, the Philistines were the main enemy of the Israelites. The Philistines driven out Dan from their tribal lands, near modern Gush-Dan in the central coastal area of Israel, to the Upper Galilee in the Northern Israel, near modern Tel-Dan [5].
Dan's primary trade characteristic was seafaring, unusual for the Israelite tribes [6].
In the Song of Deborah the tribe is said to have stayed on their ships with their belongings.
I think the main dividing line here is not ethnic, but political. Philistines were Sea People mercenaries serving Egyptians, while Dan and other Israelite tribes escaped from Egypt. Some think that tribe of Dan are related to Denyen Sea People, part of them went from Egypt straight to Greece and founded Sparta, where they called Dananoi. Thousands years later Jewish High Priest asked for help from Spartans against Rome, citing common ancestry, which Spartans acknowledged, but didn't sent military help [7].
[5] http://en.wikipedia.org/wiki/Dan_(ancient_city)
[6] Mediterranean archaeology, Volume 16. University of Sydney. Dept. of Archaeology. 2003. p. 117
[7] http://judaism.stackexchange.com/questions/12310/how-are-spartans-the-son-of-abrahamReplies: @ohwilleke
The region has had multiple rounds of mass migration:
1. One ca. 30 kya brining mtDNA M1 and U6 to Africa.
2. One ca. 8 kya with the first wave Neolithic that increased population density in Egypt by about 100x.
3. Substantial trade and migration in the Bronze and Iron Ages facilitated by Egypt's Imperial reach (much of which was limited to the Nile Delta, however). Of particular importance, was the Hyskos Dynasty which is a historically attested period of rule by outsiders that could have been accompanied by mass migration.
4. The Islamic Empire.
5. Migration in connection with British colonial rule.
This said, in Ethiopia, near the source of the Southern Route, the Neolithic arrived much later and the migration attributable to first wave Neolithic and Ethio-Semitic influence (and later) can probably be peeled away statistically, leaving only Migration (1) above as a major, post-Out of Africa migration.
It is also likely that the number of waves of migration and the extent of their impact is smaller in Southern Arabia, than in the Levant or Egypt. But, lack of archaeological and historical data until fairly late in the historic era still leaves question marks. The fact that we don't have clear data on the demic impact of herding in Southern Arabia doesn't mean that it didn't happen and we have good ancient DNA data from the Eurasian steppe that herding societies can give rise to mass population genetic replacement in even a single wave. We also know that Yemen was a center of maritime Indian Ocean trade dating back to the Copper Age.
One way to do a reality check would be to look at the TMRCA of Southern Arabian populations based on contemporary genetics. If the TMRCA << 40,000 years ago, it is a fair guess that there has been near complete population replacement there.
We do not have Out of Africa era data archaeology from Arabia, most of which isn't human remains, but can be correlated with contemporaneous African sources. And that data, whichever direction it points, is probably far more reliable.Replies: @ohwilleke
One more point on Ethiopian population history.
Ethiopia also received major migration from Sub-Saharan Africa to Ethiopia in the time period after Out of Africa. It probably received some around the time of Bantu expansion.
But, it also probably received some migration earlier, particular of Sahel farmers from West Africa that helped bring about the integration of Ethiopian wild type domesticates and West African Sahel wild type domesticates into a common Neolithic Sahel crop package. This was sometime in the vicinity of 6,000 to 4,000 years ago, but the data is pretty thin.
Thus, Ethiopia may have had more regionally distinct population genetics ca. 125kya to 50 kya, at the time of Out of Africa, than its African component does today.
The one group about which we have a pretty solid bit of information is the Philistines who ended up in the Southern Levant and were mentioned in the Hebrew Bible. They were one of the Sea Peoples and were almost certainly migrating Mycenean Greeks, based upon a combination of historical and archaeological evidence.Replies: @LevantineJew
Here is my speculation:
The Philistines (Plishtim in Hebrew, PLST in Ancient Egyptian) were descendants of Sea People, just like tribe of Dan and both were related to Aegean or Greek civilization. Yet they were enemies, while Dan joined the Israelite tribe confederation, the Philistines were the main enemy of the Israelites. The Philistines driven out Dan from their tribal lands, near modern Gush-Dan in the central coastal area of Israel, to the Upper Galilee in the Northern Israel, near modern Tel-Dan [5].
Dan’s primary trade characteristic was seafaring, unusual for the Israelite tribes [6].
In the Song of Deborah the tribe is said to have stayed on their ships with their belongings.
I think the main dividing line here is not ethnic, but political. Philistines were Sea People mercenaries serving Egyptians, while Dan and other Israelite tribes escaped from Egypt. Some think that tribe of Dan are related to Denyen Sea People, part of them went from Egypt straight to Greece and founded Sparta, where they called Dananoi. Thousands years later Jewish High Priest asked for help from Spartans against Rome, citing common ancestry, which Spartans acknowledged, but didn’t sent military help [7].
[5] http://en.wikipedia.org/wiki/Dan_(ancient_city)
[6] Mediterranean archaeology, Volume 16. University of Sydney. Dept. of Archaeology. 2003. p. 117
[7] http://judaism.stackexchange.com/questions/12310/how-are-spartans-the-son-of-abraham
It is obvious that population of the past in a nation/state do not reflect the authenticity of the modern citizenry who make up its present demographics. Compare NewWorld population before 1492 and 2000. Rape, massacre, assimilation, intermarriage and genocide figure prominently into the complex structure of what present day geneticists have to contend with. The same applies to population demographics of yore! Turkey/Byzantium from 1200 to 1900 will show a population displacement and integration showing a more European influence over a Central Asian one. Many of the occupants of the haram were captured and even volunteers, as they sought to gain the favour of their hated rulers, which allowed for personal power in the long run. the white European female saw the haram as a goal to be reached as her own personal power base as lacking and the haram was the ladder of opportunity. Compare the hagiographic representations of the Turkish horde from the view of their defeated neighbours to the reality of societal integration of the red headed Suleiman descendants and their offspring. Same applies to Africa and the Americas
The Philistines (Plishtim in Hebrew, PLST in Ancient Egyptian) were descendants of Sea People, just like tribe of Dan and both were related to Aegean or Greek civilization. Yet they were enemies, while Dan joined the Israelite tribe confederation, the Philistines were the main enemy of the Israelites. The Philistines driven out Dan from their tribal lands, near modern Gush-Dan in the central coastal area of Israel, to the Upper Galilee in the Northern Israel, near modern Tel-Dan [5].
Dan's primary trade characteristic was seafaring, unusual for the Israelite tribes [6].
In the Song of Deborah the tribe is said to have stayed on their ships with their belongings.
I think the main dividing line here is not ethnic, but political. Philistines were Sea People mercenaries serving Egyptians, while Dan and other Israelite tribes escaped from Egypt. Some think that tribe of Dan are related to Denyen Sea People, part of them went from Egypt straight to Greece and founded Sparta, where they called Dananoi. Thousands years later Jewish High Priest asked for help from Spartans against Rome, citing common ancestry, which Spartans acknowledged, but didn't sent military help [7].
[5] http://en.wikipedia.org/wiki/Dan_(ancient_city)
[6] Mediterranean archaeology, Volume 16. University of Sydney. Dept. of Archaeology. 2003. p. 117
[7] http://judaism.stackexchange.com/questions/12310/how-are-spartans-the-son-of-abrahamReplies: @ohwilleke
The Hebrew Bible is part of a larger corpus of Bronze and Iron Age “legendary history”, which is not an accurate account of history in the sense of later contemporaneous historical chronicles and Roman histories, but is also not entirely divorced from actual history (compare, for example, the Greek legends related to the City of Troy which were later corroborated, in part, archaeologically).
In the absence of corroboration from other sources and archaeology and population genetics it is very hard to know how much of it approximates historical reality, and how much of it is fiction.
Much of Genesis and Exodus shows very strong signs of being pure myth, particularly those portions with clear Mesopotamian antecedents like the story of Noah’s flood, some of the creation account, and the story of Baby Moses in the Nile. The absence of strong Egyptian corroboration of the Exodus story, in a culture that has the oldest and best historical records of that era, also casts doubts on its historicity (as of course, do its many fantastic elements which are at a minimum hyperbole).
I would be inclined to believe that there really was a historical tribe of Dan, and the events similar to those described in the Book of Joshua probably took place. But, the more detailed one gets and the further back one goes towards the ethnogenesis of the Jewish people and the Jewish religion, the less comfortable I would be that the Hebrew Bible account is historical.
The Jewish diaspora, and near total disappearance of Judaism in the Levant for many centuries, coinciding with a particularly ill documented part of Jewish history, make it very hard to know what the population genetics of self-identified Jews were prior to the diaspora, and likewise make it hard to know how much of the population genetics of the pre-diaspora Jews is preserved in one or more of the modern Levantine populations whose ancestors were, in part, Jews who converted religious and in ethnic identity after the fall of the Second Temple ca. 70 CE.
It is similarly hard to know just what proportion of people in the territory ruled for some time as kingdoms of Israel and Judea in the Iron Age were religiously and/or ethnically Jewish, and what proportion, in contrast, remained religiously pagan and ethnically non-Jewish. The more historical books of the Hebrew Bible and contemporaneous accounts from other sources strong suggest that the portion of the Levant controlled by the kingdoms of Israel and Judea was never 100% Jewish, and quite possibly wasn’t even majority Jewish.
My intuition is that the various sects of Judaism in the couple of centuries before the diaspora probably had some strong ethnic correlation (much as religious denominations and schools in all of the major faiths extant today do). Rabbinic Judaism may have captured predominantly only the ethnic and population genetic feature of one of those sects, while others withered and were lost to conversion.
Given all of the uncertainties, I am agnostic regarding the precise origins of the Jewish people, the tribe of Dan included. There may be something to your speculation, or it may be just plain wrong. I don’t have enough evidence to know the answer.
According to Josephus, Essenes followed ancient religious traditions of their forefathers, therefore at least original Essenes were ethnically different from the main mass of Israelites. Some of Essene traditions are similar to Sun worshiping, which probably indicate either Assyrian/Mesopotamian origin or cult of Ra in Egypt.
Pharisees had this notion of Oral Torah, which was passed generation by generation from Moses. So they might be descended from different ethnicity.
It is common throughout the History that the nobility is of one ethnicity, while the commoners are from another.
Idumeans / Edomites were forcefully converted.
Galileans weren't considered "pure" Jews therefore Messiah can't come from Galilee.
According to Josephus all three Jewish sect counted only about 5000 people and this at the time when when 10% of Roman Empire were Jews.
So my speculation is that at that time Jewish religion was something like folk religion today in Japan or China and the Pharisees took it to the organized religion level, which eventually become Rabbinic Judaism.
This is how nations are usually formed several ethnic tribes united, then some common language, religion and/or ethics codex either voluntarily adopted or forced on them by nobility.
My original comment wasn't about Jews, but about my speculation that Philistines/Plishtim/PLST and Dan/Denyen/Dananoi might be a sister groups, but on different political sides: i.e. mercenaries for Egypt and those who become enemies of Egypt correspondingly.
BTW: there are extra-Biblical accounts, that Dananoi came to Greece out of Egypt. The Sparta's acknowledgement of being Sons of Abraham might be a fabrication.
Some archaeological finds in Tel-Dan, which connect it with Sardinia.
Likewise there is no doubt that Philistines were connected to Aegean civilization, according to the pottery found and the myths of Goliath and Samson (who was from tribe of Dan) having most-likely Greek origins.
I agree with everything you wrote, but you essentially said nothing controversial, except this:
I agree, we may speculate that Sadducees where Cohanim and belonged to haplogroup J2. Another branch of Cohanim where Kenites / Rehabites who belonged to haplogroup J1 (CMH) – as they were nomads out of Arabia (ancient Bedouins?) and received priestly status since they were relatives of Moses’ wife.
According to Josephus, Essenes followed ancient religious traditions of their forefathers, therefore at least original Essenes were ethnically different from the main mass of Israelites. Some of Essene traditions are similar to Sun worshiping, which probably indicate either Assyrian/Mesopotamian origin or cult of Ra in Egypt.
Pharisees had this notion of Oral Torah, which was passed generation by generation from Moses. So they might be descended from different ethnicity.
It is common throughout the History that the nobility is of one ethnicity, while the commoners are from another.
Idumeans / Edomites were forcefully converted.
Galileans weren’t considered “pure” Jews therefore Messiah can’t come from Galilee.
According to Josephus all three Jewish sect counted only about 5000 people and this at the time when when 10% of Roman Empire were Jews.
So my speculation is that at that time Jewish religion was something like folk religion today in Japan or China and the Pharisees took it to the organized religion level, which eventually become Rabbinic Judaism.
This is how nations are usually formed several ethnic tribes united, then some common language, religion and/or ethics codex either voluntarily adopted or forced on them by nobility.
My original comment wasn’t about Jews, but about my speculation that Philistines/Plishtim/PLST and Dan/Denyen/Dananoi might be a sister groups, but on different political sides: i.e. mercenaries for Egypt and those who become enemies of Egypt correspondingly.
BTW: there are extra-Biblical accounts, that Dananoi came to Greece out of Egypt. The Sparta’s acknowledgement of being Sons of Abraham might be a fabrication.
Some archaeological finds in Tel-Dan, which connect it with Sardinia.
Likewise there is no doubt that Philistines were connected to Aegean civilization, according to the pottery found and the myths of Goliath and Samson (who was from tribe of Dan) having most-likely Greek origins.
There’s a new paper on the autosomal genetics of Sudan, including plenty of Copts sampled in Khartoum.
Some of the more revealing figures: 1, 2.
The Copts seem very unique here, and distinct from the general Egyptian sample. Other than the obvious possibility of Arab ancestry in Muslims (highlighted in the paper), Coptic distinctiveness could reflect their isolation, West-Central African ancestry in the Muslims, and/or a possible distinction between Upper and Lower Egyptians (many Copts live in Upper Egypt).
It would be nice to see what you could do with this data, Razib. Here’s the sample info from the authors, since it’s apparently missing in the paper.
Some of the more revealing figures: 1, 2.
The Copts seem very unique here, and distinct from the general Egyptian sample. Other than the obvious possibility of Arab ancestry in Muslims (highlighted in the paper), Coptic distinctiveness could reflect their isolation, West-Central African ancestry in the Muslims, and/or a possible distinction between Upper and Lower Egyptians (many Copts live in Upper Egypt).
It would be nice to see what you could do with this data, Razib. Here's the sample info from the authors, since it's apparently missing in the paper.Replies: @Labayu
It would be interesting to see how much that presumed ancestral North African component in Copts corresponds to the ancestral North African component identified in this study.