Ancient Egyptians: black or white?

One of the most fascinating things about ancient Egypt is its continuity, and our granular and detailed knowledge of that continuity. We can thank in part the dry climate, as well as the Egyptian penchant for putting their hieroglyphs on walls and monuments (as well as graffiti!). And we can also thank the fact that both the ancient Greeks and Hebrews, Athens and Jerusalem so to speak, were deeply connected to and perceived themselves to be indebted to Egyptian civilization. Even before the translation of the Rosetta Stone and the deciphering of ancient Egyptian writing the Hebrews’ interactions with Egyptians, in particular in , mean that their memory would echo down through the millennia (the newly Christianized Irish interpolated Egyptian ancestry into their own genealogy).

The Greek relationship with Egypt was less fraught and at greater remove than the Hebrews. But the Classical period philosophers correctly perceived that Egyptian civilization was ancient, and preceded their own. Aegean-Egyptian connections were actually more longstanding than the Classical scholars knew, in , the correspondence in state archives which have been retrieved are rather clear that Minoan civilization was part of the orbit of Egypt early on. Though Egyptians never conquered the Aegean polities, mercantile and diplomatic connections were extremely old and persistent. The late Bronze Age eruption of barbarian Sea Peoples who attacked the whole civilized Near East may have been facilitated in part by the broad familiarity engendered by widespread trade networks.

The most recent book devoted to ancient Egypt I have read was Toby Wilkinson’s . Synthesizing extensive written material with archaeology, perhaps the most impactful argument in Wikinson’s narrative was the persistence of the temple based institutions from the Old Kingdom down to the Ptolemaic era. Religious institutions carried on even with the shocks of Nubian and Libyan conquest in the post-New Kingdom period, down to Late Antiquity. The temple at Philae in southern Egypt was an active center of the traditional religion, and therefore the culture which dates to the Old Kingdom in continuous form, down to the 6th century A.D. (when it was closed by Justinian in his kulturkampf against ancient heterodoxies).

For various ideological reasons though many people are very curious about the racial characteristics of the ancient Egyptians. There are two basic extreme positions, Afrocentrists and Eurocentrists. Though I have not done a deep dive of the literature of either group, I’ve read a few books from either camp over my lifetime. In fact I believe the last time I read the “primary literature” of Afrocentrist and Eurocentrism was when I was an early teen, and it was rather strange because both groups seem to be recapitulating racial disagreements and viewpoints relevant to the American context, and projecting them back to the ancient world.

In college I stumbled upon Mary Lefkowitz’s , a book length argument against the more sophisticated Afrocentrist views articulated in the wake of Martin Bernal’s . Lefkowitz was a classicist, so many of her objections were exceedingly scholarly. The reality is that the best refutation of an Afrocentrist view of of ancient Egypt, which reduces to the idea that ancient Egyptians would be recognizably black African today, are the Fayum portraits. It is notable to me how similar these portraits are to modern Copts. In fact the actor Rami Malek, of Coptic background, looks strikingly like someone who stepped out of the Fayum portraits.

I have read no book length refutation of the Eurocentrist, usually Nordicist, perspective. Mostly because this is a view associated with white supremacism, and that ideology is generally attacked on normative, not positive, grounds. But the visible evidence of the Fayum portraits is a strong refutation of the Nordic model. Of course, there is the reality that we now know that the Nordic phenotype, and the genetic components which congealed into that typical of Northern Europe today, was only coming into existence when the Old Kingdom of Egypt was already a mature civilization.

Both Afrocentrists and Eurocentrists will reject the evidence of the Fayum portraits became they came from the Roman era, and they would argue that the demographic nature of Egyptians changed quite a bit between that period and the end of the New Kingdom. And they are not incorrect that the period between the arrival of the Romans and the fall of the New Kingdom was characterized by a great deal of change. There were Libyan dynasties, Nubian dynasties, and periods of rule by Assyrians, Persians, and Macedonians. Large colonies of Greeks, Macedonians, and Hebrews-becoming-Jews were also resident in Egypt. Especially, but not limited to, the urban areas.

But now we have ancient DNA! :

Egypt, located on the isthmus of Africa, is an ideal region to study historical population dynamics due to its geographic location and documented interactions with ancient civilizations in Africa, Asia and Europe. Particularly, in the first millennium BCE Egypt endured foreign domination leading to growing numbers of foreigners living within its borders possibly contributing genetically to the local population. Here we present 90 mitochondrial genomes as well as genome-wide data sets from three individuals obtained from Egyptian mummies. The samples recovered from Middle Egypt span around 1,300 years of ancient Egyptian history from the New Kingdom to the Roman Period. Our analyses reveal that ancient Egyptians shared more ancestry with Near Easterners than present-day Egyptians, who received additional sub-Saharan admixture in more recent times. This analysis establishes ancient Egyptian mummies as a genetic source to study ancient human history and offers the perspective of deciphering Egypt’s past at a genome-wide level.

Because modern people care about the Afrocentrist question, the extent of Sub-Saharan African ancestry is highlighted in this paper. I do not think this is actually the most interesting aspect. But I’ll get to that. Since this post will be read by a fair number of people I’ll talk about the relationship of ancient and modern Egyptians to (Northern) Europeans and Sub-Saharan Africans.

The figure to the left is looking at 90 ancient Egyptian mitochondrial genomes (and some modern ones in the two rightmost columns). Since mtDNA is copious it was relatively easy to extract and analyze.  Haplogroup L, the red to orange shades in the bar plots, are associated without dispute with Sub-Saharan Africa. Haplogroup U6, M1 and a few others may be “back to Africa” variants of different periods (they are generally found in Afro-Asiatic groups).

What you can see is that somewhat more than half of Ethiopia’s mtDNA lineages are L, in keeping with the whole genome estimate of Sub-Saharan African ancestry in most Cushitic populations. In Egypt there is a difference over time; haplogroup L goes from low frequencies to much higher frequencies in modern periods. The ~20% fraction in the modern samples is in line with the population wide admixture one sees in modern Egyptians of Sub-Saharan admixture.

I actually recomputed the haplogroups to a finer granularity from the supplements. A quick inspection of mtDNA haplogroup frequencies shows that ancient Egyptians are not typical of modern Europeans. Not that much H, and lots of T, J and K. What that does remind me of are Early European Farmers. These people, who brought agriculture to Europe from Anatolia contributed a large fraction of the ancestry of modern Southern Europeans, and a lesser component to Northern Europeans.

But ultimately what’s great about this paper is that they have ancient autosomal DNA. That is, genome-wide results.

They got three samples of reasonably high quality. More precisely: “Two samples from the Pre-Ptolemaic Periods (New Kingdom to Late Period) had 5.3 and 0.5% nuclear contamination and yielded 132,084 and 508,360 SNPs, respectively, and one sample from the Ptolemaic Period had 7.3% contamination and yielded 201,967 SNPs.”

You can see the three samples on this bar plot. What is interesting is that they’re all pretty similar.

What you can see here is that to a great extent ancient Egyptians were descended from a population closely related to Natufians, or Natufians themselves. This easily explains the mtDNA affinity to Neolithic farmers: Natufians and Anatolian Neolithic populations were sister populations. The f3 statistic which looks at shared drift shows an affinity of ancient Egyptians with ancient farmer populations with Near Eastern provenance, but also with modern Sardinians. This is a common pattern, as ancient groups do not have later migration waves, with the Sardinians the modern population closest to this.

You see in the bar plot that northern Levantine populations are placed between Anatolian Neolithics and Natufians, as one might expect based on their geographical position and gene flow between these two regions. Additionally, the cyan color is associated with eastern farmers from the Zagros. I’ve already talked about gene flow from this area to the Levant recently. If you compare the Bronze Age Sidon samples I think you’ll see broad affinities with these Late Period Egyptians.

The PCA gives us results consonant with the model-based clustering. If you plot the genetic variation of ancient Egyptians they’re closest to Neolithic eastern Mediterranean populations. No great surprise.

Not the modern Egyptians. Why? It’s pretty clearly because modern Egyptians are shifted toward Sub-Saharan Africans. But there is also another component: modern Egyptians have more of the cyan eastern farmer component. What could this be?

An immediate thought comes to mind. We focus a great deal on Sub-Saharan African slavery. One reason is that it is visible. Black Africans are physically distinct from most Middle Eastern populations. But Egypt was long the center of another slave trade: “white slaves” from the Caucasus. Circassians. For hundreds of years Mamluks were recruited from the Caucasus as military slaves. They eventually became the ruling class of Egypt, until their decimation in the 19th century under Muhammad Ali (who himself was an Albanian Ottoman who never learned to speak Arabic well).

As noted in the paper earlier work looking at patterns in ancestry tracts and LD decay had made it obvious that much of the admixture of Sub-Saharan ancestry in Egypt, as in much of the Middle East, is relatively recent. In particular, it dates to the Islamic period, when trade and conquest took on new dimensions in Africa and north into Central Asia. One way ethnic minorities like Assyrians and Lebanese Christians differ from their Muslim neighbors is that they have much lower fractions of Sub-Saharan African ancestry, and no East Asian component. The latter might surprise, but remember that Central Asian Turkic slaves have been prominent in Muslim armies since at least the 9th century.

But some of the Sub-Saharan ancestry in Egyptians is old. The ancient Egyptian samples have it. To have none of it would seem strange, considering the history of contact between Nubia and Egypt, dating back to the Old Kingdom. Second, there is evidence of low levels of Sub-Saharan African gene flow into Southern Europeans. How did that happen? The highest fractions are in Spain, and can there be attributed to the Moorish period. But that explanation does not hold in much of Italy, where there are a few percent of haplogroup L. This probably is due to south-to-north gene flow across the Mediterranean during the Classical period. Some of the peoples on the south shore of the Mediterranean almost certainly already had some Sub-Saharan African admixture.

Not getting into the details of it, there are ways to explicitly model gene flow into a target population from donors defined by a phylogeny. In this case the authors tested various models of gene flow from Sub-Saharan Africans and Eurasians (non-Africans) to generate allele frequency patterns we see in modern Egyptians and ancient Egyptians.

What they consistently found is that modern Egyptians are about twice as much Sub-Saharan African as ancient Egyptians. The proportions for modern Egyptians ranged from ~10 to ~20 percent Sub-Saharan African against a Eurasian background, with a bias toward the higher values (depending on which populations you put into the phylogeny for non-Africans), and ~0 to ~10 percent for the ancient Egyptians, again with a bias toward the higher values. The pattern is consistent in these tests.

An issue here is that we’re going off three samples. That being said, the authors observe that despite differences in contamination/quality and time period they’re very concordant with each other. If I had to bet I think Old Kingdom samples would have somewhat less Sub-Saharan and eastern farmer ancestry. But the basic pattern persisted down to the Roman period, and was only shifted by admixture due to slavery.

And not to belabor the point, but a paper from a few years ago which had some Copt samples looks familiar in its broad outlines. You see that the Copts have very little Sub-Saharan African ancestry, though it does seem to be evident (the marker set is in the hundreds of thousands of SNPs). Additionally, they are quite distinct from the Qatari Arab sample.

Unfortunately the data for this paper just published is not on the European Nucleotide Archive. I really want to dig a little deeper into it.

What are the takeaways here? Egypt has been the sink for a lot of migration and gene flow over the past several thousand years, and probably earlier. Not surprising considering that it was relatively wealthy in the aggregate. The Natufian population that the Late Period Egyptians resemble the most did not have Sub-Saharan African ancestry according to earlier research. These Late Period Egyptians do have some. This is reasonable in light of the long interaction with Nubia which is historically attested. Similarly, there was clearly gene flow from Southwest Asia. This is again historically attested, especially in the Nile Delta (though foreign garrisons of mercenaries are recorded in Upper Egypt as well).

The Roman period probably did introduce some gene flow from Southeast Europe and Southwest Asia. But these populations are not that distinct from Egyptians.

Similarly, the Islamic period also brought in different peoples from Arabia and the Caucasus. But the most salient dynamic during the Islamic period was a massive trans-Saharan slave trade (though the Caucasus impact may have been comparable, and I think these results support the proposition that it was).

It seems entirely likely that the Copts are descended from a mix of Roman era Egyptians. Not only do they resemble the people in the Fayum portraits, but the circumstantial genetic data is that they have fewer “exotic” components which increased in frequency during the Islamic era. This would be exactly parallel to ethno-religious minorities in the Levant and Iraq.

One curious element to me is the suggestion gene flow before ~5,000 BCE between Sub-Saharan Africa and the lower Nile valley was low. If it hadn’t been low, it seems unlikely that the fraction of Sub-Saharan ancestry (or shift in that direction in relation to other Eurasians) in Copts would be so small.

So what explains the lack of earlier gene flow? I think the answer is going to be the fact that the human demographic landscape is characterized by lots of local population extinctions. As ancient DNA sampling coverage gets better and better meta-population dynamics are coming into focus, and we see gene flow, and die offs, in several areas. It is fashionable to say that human population variation is characterized by clines. But much of this clinal aspect is an outcome of the period after massive admixture over the last ~10,000 years.

And yet it may not be that the period before the Holocene was not clinal. Rather, it may be that large depopulations of areas of human occupation fragmented clinal ranges, and resulted in new range expansions from “core” zones.

About ~8,000 years ago there was a major desertification period in the Sahara desert. Many trans-Saharan populations may have gone extinct during this time due to rapid climate change. Eventually repopulation may have occurred from outside of the Sahara, so that post-Natufian Levantines and Sub-Saharan Africans from what today call the Sahel pushed up and down the Nile drainage basic respectively, meeting in the zone of Nubia on the boundary of history and prehistory.

Unlike many other areas of the world we have a long attested record of Egyptian history. As we get more mummy samples it seems likely that we’re get a crisper, clearer, picture. And the time transects will not be narrative blind; we already know the general arc of Egyptian history. If, for example, we see a new ancestral component around ~1500 B.C., in Egypt it’s not mysterious what this might be: the Hyksos.

This is just the prologue to a fascinating book that will be written over the next decade.

Related: Blog post analyzing one Copt’s results suggests that Sub-Saharan admixture is more like Dinka than Yoruba (in contrast, Muslim Egyptians have a mix of both, the latter probably coming during the Islamic slave trade, while the former is probably ancient admixture).

Citation: Schuenemann, V. J. et al. Ancient Egyptian mummy genomes suggest an increase of Sub-Saharan African ancestry in post-Roman periods. Nat. Commun. 8, 15694 doi: 10.1038/ncomms15694 (2017).

11 thoughts on “Ancient Egyptians: black or white?

  1. Something you did not mention is skin color. Looking through the supplementary tables, one of the three samples was derived for SLC24A5, but not derived for SLC45A2, while another was derived for SLC24A5, but couldn’t be tested for the other SNP. The third didn’t have any coverage for these genes. Although a small sample size, this seems to suggest ancient Egyptians were not dark brown, but may have been about as swarthy as the older bronze age Sidonians discussed last week.

    Also, this might be self-evident from earlier studies, but the strange distribution of the Y haplogroup E3b1 – present in Northeastern Africa and southern Europe – is making a lot more sense. It seems it was the original “western farmer” Y-DNA haplogroup, but was replaced in the Near East by “eastern farmer” J, and although WHG admixture into EEF was comparably minor, I2a somehow took over soon after the expansion into Europe. If you hold that the Natufians/Neolithic Levantine populations were speakers of Proto-Afro Asiatic, this would mean that the first EEF could have spoken Afro-Asiatic languages as well, although there would have to have been a language shift to something else (WHG? Anatolian?) relatively early on.

    Also, while of course there is historically attested population movements between the Caucasus and Egypt, do we really need to rely on those to explain the elevated “cyan” component? It is common throughout the Near East, including among relatively isolated populations like the Druze and the Bedouins. It seems to me that the cosmopolitan nature of Islamic civilization would explain a great deal of this by itself – more people were migrating into Egypt from throughout the Near East, which elevated the cyan component over time.

    Finally, all the samples studied in this paper are from Lower Egypt. I would like to see if Upper Egypt similarly shows very little ancient East African admixture, or if there was more of a gradient as you traveled southward at that time.

  2. Something you did not mention is skin color. Looking through the supplementary tables, one of the three samples was derived for SLC24A5, but not derived for SLC45A2, while another was derived for SLC24A5, but couldn’t be tested for the other SNP. The third didn’t have any coverage for these genes. Although a small sample size, this seems to suggest ancient Egyptians were not dark brown, but may have been about as swarthy as the older bronze age Sidonians discussed last week.

    yeah. so i looked at the slc45a2 in the 48 copts from sudan. that snp is in there and it’s 25% derived. so reason i didn’t mention is if the ancient freqs were same we’d expect 50% ancestral homoz.

  3. How comfortable can we be that the mummies are representative of the general population and not just the upper class?

    Unlike a lot of pre-historic societies, we know for a fact that ancient Egypt was very stratified by class and that the upper class took endogamy to extremes including some instances of sibling or other marriages that would be considered incestuous today even in SW Asia? Also, elite marriages appear to have had a more international character (with aristocrats marrying foreign aristocrats to make alliances) to a greater extent than the rank and file.

    With panmixia or even a reasonable approximation thereof, autosomal results from even a few individuals are quite representative to the entire population after only a few generations after initial admixture. But, we could be seeing trends limits only to one elite strata of the society.

    On the other hand, the progression in the sub-Saharan percentage to the present and the perhaps less stratified nature of the aristocrats during the Roman era could argue against a caste-like interpretation of the data.

    I guess part of it comes down to how elite you had to be to be mummified, something I’d love to hear what you know about from your reading.

  4. Interesting how little contact there was with sub-Saharans absent the slave trade. For example, you might have thought that Copts and Ethiopians would mix a lot over the centuries, both before and after the Islamic conquest, through their shares Monophysite Christian faith, but contact between the groups seems to have been minimal and Christian Ethiopians remain quite distinct from Copts today both physiologically and culturally, despite the shared religion.

  5. re: representativeness. they discussed this issue in the text. don’t know if i find it persuasive. but note that all the samples look very similar similar despite one of them being roman era. additionally, they discuss the class ubiquity of mummification.

    (also, i can double check, but i’m pretty sure if one of the individuals was inbred they’d be mentioned it)

    finally, the similarities to modern copts argues is an important point to consider. if copts descend only from the elite that’s an argument, but that seems a bit too extreme re: elite overproduction.

    ultimately more samples will fix this issue. i am moderately confident they’ll validate these results.

  6. “Something you did not mention is skin color. Looking through the supplementary tables, one of the three samples was derived for SLC24A5, but not derived for SLC45A2, while another was derived for SLC24A5, but couldn’t be tested for the other SNP. The third didn’t have any coverage for these genes. Although a small sample size, this seems to suggest ancient Egyptians were not dark brown, but may have been about as swarthy as the older bronze age Sidonians discussed last week.

    yeah. so i looked at the slc45a2 in the 48 copts from sudan. that snp is in there and it’s 25% derived. so reason i didn’t mention is if the ancient freqs were same we’d expect 50% ancestral homoz.”

    RE: Skin color,

    According to these results, any idea as to where the Ancient Egyptians would be on the Fitzpatrick or Von Luschan Scales?

  7. “I actually recomputed the haplogroups to a finer granularity from the supplements.”

    Can you share this data, thanks.

  8. We still need DNA studies done on ethnic AE’s from both Upper and Lower Egypt from time periods preceding the New Kingdom (i.e. Middle Kingdom, Old Kingdom, and Predynastic Period).

    Sure this new study is very interesting and a strong wound to the Afrocentric ideal as right now it’s looking like AE was a civilization of Near Eastern origin that happened to be located in North Africa just like Carthage. However it’s still only mummies from one site in Central Egypt from the New Kingdom to Roman Egypt. These mummies don’t tell the whole story of Ancient Egypt.

    Razib do you know if DNA studies will be conducted on Ancient Egyptians from older periods anytime soon?

  9. Not sure if you seen this but here’s how these mummy samples show up on a DNA autosomal calculator:

    776-569 cal BC
    0.00% Ancestral_S_Eurasian
    0.00% East_Asian
    25.22% Iran_Neolithic
    61.91% Natufian
    8.00% WHG
    4.86% Sub_Saharan

    769-560 cal BC
    2.31% Ancestral_S_Eurasian
    1.94% East_Asian
    24.59% Iran_Neolithic
    57.60% Natufian
    6.64% WHG
    6.93% Sub_Saharan

    97-2 cal BC
    0.00% Ancestral_S_Eurasian
    0.01% East_Asian
    35.25% Iran_Neolithic
    54.16% Natufian
    2.36% WHG
    8.22% Sub_Saharan

    Now compare to a modern Copt:

    1.64% Ancestral_S_Eurasian
    1.75% East_Asian
    24.77% Iran_Neolithic
    55.94% Natufian
    4.56% WHG
    11.34% Sub_Saharan

    Clearly it looks like modern Copts have strong genetic similarities to these mummy samples. Indicating (at least based off these samples) that Copts are genetically the most related to Ancient Egyptians probably out of all groups of people on Earth.

    Not only that buy if your observe these samples by time period you’ll notice the SSA further increases as well as general decrease in Natufian ancestry as well.

  10. @ Razib : “I actually recomputed the haplogroups to a finer granularity from the supplements.”

    Yeah can you share this..pretty please with a cherry on top, thanks.

Comments are closed.