The genome of “Cheddar Man” is about to be published

If you are American you have probably heard about “Cheddar Man” in Bryan Sykes’ Seven Daughters of Eve. If you don’t know, Cheddar Man is a Mesolithic individual from prehistoric Britain, dating to 9,150 years before the present. Sykes’ DNA analysis concluded that he was mtDNA haplogroup U5, which is found in ~10% of modern Europeans, and which ancient DNA has found to be overwhelmingly dominant among European hunter-gatherers. But for years there has been controversy as to whether this result was contamination (after all, if it’s found in ~10% of modern Europeans it wouldn’t be surprising if the DNA was contaminated).

Today that is a moot point. On February 18th Channel 4 in the UK will premier a documentary that seems to indicate genomic analysis of Cheddar Man’s remains have been performed, and he turns out to be exactly what we would have expected. That is, he’s a “Western Hunter-Gatherer” (WHG) with affinities to the remains from Belgium, Spain, and Central Europe. These WHG populations were themselves relatively recent arrivals in Pleistocene Europe, with connections to some populations in the Near East, and with unexplored minor genetic admixture from an East Asian population. Their total contribution to the ancestry of modern Europeans varies, with lower fractions in the south of the continent, and the highest in the northeast.

Overall, the consensus seems to be that in Western Europe the genuine descent from indigenous hunter-gatherers passed down through admixture with Neolithic farmers, and then the Corded Ware and Bell Beaker groups, is around ~10%. This is the number that shows up in the press write-ups. But, there are some researchers who contend it is far less than 10%, and that that fraction is misattribution due to early admixture with relatives of these hunter-gatherers as steppe and farmer peoples were expanding.

Phylogenetics aside, one of the major headline aspects of the Cheddar Man is that reconstructions are now of a very dark-skinned and blue-eyed individual. Some of the more sensationalist press is declaring that the “first Britons were black!” As far as the depiction goes, this is literally true. The reconstruction is of a black-skinned individual in the sense we’d describe black-skinned.

But on one level it is entirely expected that this is what Cheddar Man would look like. The hunter-gatherers of Mesolithic Western Europe were genetically homogenous. They seem to derive from a small founder population. And, on the pigmentation loci which make modern Europeans very distinctive vis-a-vis other populations, SLC24A5, SLC45A2 and HERC2-OCA2, they were quite different from anything we’ve encountered before. First, these peoples seem to have had a frequency for the genetic variants strongly implicated in blue eyes in modern Europeans close to what you find in the Baltic region. The overwhelming majority carried the derived variant, perhaps even in regions such as Spain, which today are mostly brown-eyed because of the frequency of the ancestral variant. Second, these European hunter-gatherers tended to lack the genetic variants at SLC24A5 and SLC45A2 correlated with lighter skin, which today in European is found at frequencies of ~100% and 95% to 80% respectively.

The reason that one of the scientists being interviewed stated that there was a “76 percent probability that Cheddar Man had blue eyes” is that they used something like IrisPlex. They put in the genetic variants and popped out a probability. The problem is that the training set here is modern groups, which may have a very different genetic architecture than ancient populations. Recent work on Africans and East Asians indicate that the focus on European populations when it comes to pigmentation genetics has left huge lacunae in our understanding of common variants which affect variation in outcome.

East Asians, for example, lack both the derived variants of SLC24A5 and SLC45A2 common in Europeans but are often quite light-skinned. A deeper analysis of the pigmentation architecture of WHG might lead us to conclude that they were an olive or light brown-skinned people. This is my suspicion because modern Arctic peoples are neither pale white nor dark brown, but of various shades of olive.

As far as blue eyes go, it is reasonable that these individuals had that eye color because that trait seems somewhat less polygenic than skin color. There are darker complected people with light eyes, from the famous “Afghan girl” to the first black American Miss America, Vanessa Williams. The homozygote of the derived HERC-OCA2 variant seems relatively penetrant. From what I recall the literature indicates many people with blue eyes are not homozygotes on this locus for the derived haplotypes, but those who are homozygotes for the derived haplotypes invariably have blue eyes.

Addendum: It isn’t clear in the press pieces, but it looks like they got a high coverage genome sequence out of Cheddar Man. They refer to sequencing, and, they seem to have hit all the major pigmentation loci. This indicates reasonable coverage of the genome.

42 thoughts on “The genome of “Cheddar Man” is about to be published

  1. From what I recall the literature indicates many people with blue eyes are not homozygotes on this locus for the derived haplotypes, but those who are homozygotes for the derived haplotypes invariably have blue eyes.

    People who meet my children are almost always surprised that a couple of them have green eyes. My wife (English-German-Swedish) has blue-green eyes and I have light brown (“caramel”) eyes. So the uninformed would start speculating that I must be part European somehow. Nope. 99.9% East Asian shared genes. The remaining 0.1% is Native American shared. But my genes did come out as having much higher probabilities of having lighter hair and eye color compared to average East Asian in the database. People don’t seem to realize that, though there is less phenotypical diversity, there is still some among East Asians. Certainly so genetically.

    with unexplored minor genetic admixture from an East Asian population

    Any speculation on this?

  2. The Afghan girl looks awfully white to me.

    But that quibble aside, just how do the Cro Magnons fit into this picture?

  3. Cro-Magnon, and Kents Cavern (much closer to Cheddar Gorge) are Palaeolithic, around 40-50,000 bp, so there’s not much to fit, especially as there was glaciation in between, stripping the land of human presence.

    For all the silly liberal air-punching, this is ten year old news. The best thing about the result is that it further constrains the appearance of light skin in north western Europe to less than 10,000 bp. I believe the previous estimates had it possibly out as far as 15,000 bp, making the arrival of farming people from the middle east within the range, but this makes the story more solid. So we have dark-skinned hunter gatherers, surviving through winter on meat, then farming arrives and there is no longer that dietary source of vitamin D.

  4. Where will/has the genome be published?


    But that quibble aside, just how do the Cro Magnons fit into this picture?

    depending on how u define they left very few or zero descendants in northern europe.

  5. @jim, not necessarily quite the case. To go on a tangent, as the European hunter gatherer results have rolled in, there’s a cline in SLC24A5 and SLC45A2 in Europe from “WHG proper” (own term) like Loschbour, La Brana, Bichon, through to populations like the Iron Gates HG, Latvia HG, SHG, EHG who have higher (more modern) derived frequencies of SLC24A5 and SLC45A2, at a summit in EHG / SHG who have almost completely derived SLC24A5 and generally higher derived levels of SLC45A2 compared to the early Anatolians and EEF.

    See Iain Mathieson’s latest paper’s supplement: on page 51

    This seems like it would be hard to explain through subsistence differences – why should the Baltic HG or Iron Gates HG who are clearly fisher gatherers need vit D absorption more?

    My suspicion (like Razib’s?) is that the WHG proper probably had the same selective pressures for lighter skin that we find in all northern latitude populations and which were certainly present in the mesolithic EHG and also the Devil’s Gate hunter gatherers in East Asia, and must have been present to some degree in the founding population of the Americas. It would surprise me if they were the exception.

    We may not just understand these variants well because the “WHG proper” probably contributed less than 10% to the ancestry of even Western Europeans (once we account for HG ancestry from other sources like Latvia HG / Iron Gates HG, it will be less than 10%), and the variants will probably only be detected through contribution to intra-European fine scale differences over large numbers of sample sizes (something like a pan-European 500,000 sample Biobank with pigmentation information). So even something close to fixture in “WHG proper” would show low-modest frequency differences in mostly West Europeans compared to other populations (assume a variant at 0.6 in “WHG proper”, Western Europeans have 0.07 “WHG proper” ancestry, expected frequency of this variant only 0.04 in Western Europeans, assuming no further positive selection and absence of negative selection).

    IRF4, a variant which is associated with lighter skin mainly in Ireland (to the extent that the Irish have a high frequency difference even with the English!) seems suspiciously frequent in the WHG proper so far, for instance, for a population under no selection for lighter skin. Variants in OCA2, which is derived in WHG, also may have some pleiotropic action on the skin as well.

  6. i think they probably had different variants.

    strong selection for pigmentation loci in modern europeans indicates to me that these groups were not as pale as modern pops. but i doubt they were as dark skinned as africans or oceanians or south asians, as these depictions make them be.

  7. But my genes did come out as having much higher probabilities of having lighter hair and eye color compared to average East Asian in the database.

    i’m a pretty typical looking south asian in complexion (not dark, not light). but my three kids all look far whiter than we had expected, and most other people would expect. so there’s something going on in my genome which leads to this (two oldest have hazel eyes, with green accent, though my youngest,who has skin nearly as white as his mother happens to have brown eyes and his went non-blue far earlier than the other two).

    since my oldest son has light brown hair but a nice tan complexion people always joke he looks like a perpetual surfer (his features also seem very “northern european” to people from northern europe for what it’s worth).

  8. I understand high-latitude people in Europe, Asia, and North America have light-er skin, showing selective pressure. I was thinking of the difference between that normal phenomenon, and the unusually-white appearance of people in the extreme high latitudes on the Atlantic coast of Europe since Cheddar Man’s time.

  9. I was thinking of the difference between that normal phenomenon, and the unusually-white appearance of people in the extreme high latitudes on the Atlantic coast of Europe since Cheddar Man’s time.

    the genotypes didn’t come together really until about 5 to 6 thousand years ago. and the sweeps kept going (are still going on?) down to the present. no one knows why.

    sexual selection is a popular rationale. but mostly because they can’t think of much else. i think it’s something else, but with low confidence

  10. Thanks for that. So while I’m content with speculating on extreme depigmentation in Western Eurasia, you’re looking for a theory with the explanatory power to account for similar extreme depigmentation in Eastern Eurasia too, a tougher call because we haven’t got the warm waters of the Gulf of Mexico to blame for the East as we do for the West.

  11. jim most of east eurasia that is agricultural is at far lower latitudes. the latitudes which are like europe are HG territory (or nomad). i think that might make a difference. but i don’t know.

  12. I am curious about this statement:

    “From what I recall the literature indicates many people with blue eyes are not homozygotes on this locus for the derived haplotypes, but those who are homozygotes for the derived haplotypes invariably have blue eyes”

    According to 23andMe, homozygotes have the following eye color distribution. Light eyes are near universal, but blue is actually only about a 50% chance (and this is in a European population, as I understand their reporting):

    52% Blue
    21% Greenish blue
    17% Green
    8% Light hazel
    2% Dark Hazel
    <1% Light brown
    <1% Dark brown

    This accords with my own experience, being a white homozygote with green eyes, and I have met many others (including nearly all my family members). Interestingly I had blue eyes until I was about 4 or 5 and then they slowly turned green. It occurred so late that I can actually vaguely remember my eyes becoming less blue and being kind of disappointed!

    I haven't seen data on it, but I would guess that true blue eyes are even less common among non-white homozygotes than in Europeans. I suspect the genotype may be more likely to manifest itself as light brown or hazel in the absence of other lightening genes. Fully blue eyes, like those of Vanessa Williams, are extremely rare among black Americans (yet probably 1 or 2% are homozygotes), and all those with blue eyes I have ever seen or met invariably had relatively light skin as well.

  13. gadget, you’re right. must have subbed ‘blue’ for ‘light’.

    are extremely rare among black Americans (yet probably 1 or 2% are homozygotes)

    this sounds right.

  14. Has anyone every found someone who is fully European, but by chance is homozygous ancestral on both SLC24A5 and SLC45A2? Would be statistically very rare, but in a population of hundreds of millions maybe a handful might exist out there. It would be very interesting to see what such a person’s skin tone turned out to be. Perhaps it would say something about Cheddar Man.

  15. 45a2 is totally findable. there is some research that shows ppl who are hets on 45a2 have olive skin.

    24a5 would be REAL hard. the frequency is so low in europeans. but there has to be someone.

    fwiw, my kids are homoz derived on 24a5 (i’m homoz. derived on that) and hets on 45a2 (and herc2/oca2). they have olive skin. well, the two oldest do. the youngest is about 1 year old and still totally white skinned. so perhaps he’s different….

  16. I always find these pigmentation presentations/assumptions interesting. Although I do have to wonder how accurate they are.

    My own HIrisplex results don’t fit reality so I always am a bit skeptical. They are close on the hair but eye color shows likely Blue with a pretty high AUC. But I have brown eyes, light brown, but brown. My brother with gray eyes shows a lower AUC under blue. Strange. My ancestry is NW Euro and Baltic.
    Predicted phenotype
    p-value AUC Loss
    blue eye 0.972 0
    intermediate eye 0.025 0
    brown eye 0.002 0
    blond hair 0.786 0.007
    brown hair 0.068 0.002
    red hair 0.138 0.027
    black hair 0.007 0.002
    light hair 0.992 0
    dark hair 0.008 0

  17. @Razib Khan

    How is this 10% ancestry left behind by WHG constructed? Is this estimated on the amount of mtDNA U5? In that case one needs to consider this. The gene flow of WHG into farmers was asymetrical, as is made clear in the latest Mathieson, presenting “evidence for sex-biased admixture between hunter-gatherers and farmers in Europe, showing that the Middle Neolithic “resurgence” of hunter-gatherer ancestry in central Europe and Iberia was driven more by male than by female hunter-gatherers.”

    There is a reason I2 became farmers Y-DNA.

    If then later a second male-biased expansion happened (R1b/R1a) the mtDNA and Y-DNA ratio’s aren’t really reflecting real admixture numbers.

  18. depending on how u define they left very few or zero descendants in northern europe.
    This is new to me, I always assumed Cro-Magnons were WHG.

  19. I’m not totally convinced by the supposition that WHG likely had their own skin lightening alleles for two reasons.

    1. Some populations in high latitudes did not lighten – namely, the Tasmanians. One could perhaps argue that was because their small population size didn’t allow for advantageous lighter-skinned variants to be selected for, but even in southern mainland Australia you would have expected some lightening of skin (since it’s about where southern Europe is in terms of latitude) yet AFAIK they were just as dark brown as other aboriginal Australians.

    2. From what I have read, the WHG population only expanded from the Near East (or a refugia very close to it in the Balkans) following the LGM. This means that WHG may not have been in cooler climates long enough to have significant selection for lighter skin.

    As an aside are the genetics of hair texture well known enough now for these reconstructions to be accurate? I know pigmentation broadly is, and genetic understanding of facial morphology is improving, but I don’t typically see tracking for genes related to hair curl (aside from EDAR) in these archeo-genetics papers.

  20. @Karl, I can’t answer your first point, but on the second point:

    A) If EEF evolved light skin in an Middle Eastern urheimat, then the selective pressure is there in the ME (or at least, the part of the ME that EEF came from, likely Anatolia). So it’s a strange one for this pressure to work for EEF but not for WHG. Even assuming WHG came from the ME.

    I suppose we could come up with a complicated story about how WHG wouldn’t need light pigmentation, while evolving in ME, because they are HG, while even the earliest EEF who were in the paleolithic transition and had scarcely been farming, the Boncuklu samples, had fully derived SLC24A5 *because* of farming… as did the CHG hunter gatherers from the Caucasus. But it would be a complicated story to say the least.

    B) Per the Mathieson paper above, since we have Iron Gates HG and Latvia HG who are fairly to identical to WHG, modeled as 85% WHG and 70% WHG respectively remainder EHG (or something like 90% and 80% if we model EHG as WHG+ANE), and they have derived SLC24A5 respectively of 0.5 and 0.6, then time wouldn’t be a strong argument for me for lack of frequency rise in SLC24A5. SLC24A5 has to be under selection in these populations, cannot be the product of EHG+WHG frequencies (and note frequencies of SLC45A2 and HERC2 are not consistent with being intermediate EHG / WHG in these populations).

    (Forgot in the last post, but SLC24A5 derived variant was present in one “WHG proper” sample in Western Europe – Canes_Meso –, very late Mesolithic 5000 YBP).

  21. Matt,

    You raise some interesting points. But to my mind, that the (now typical) West Eurasian light skin alleles are elevated in these “intermediate” HG populations suggests that perhaps WHG had dark skin. I say this because in contemporary populations it appears SLC24A5 has been enriched in areas West Eurasians mixed with dark-skinned groups (South Asia and East Africa) but in areas like Central Asia where East and West Eurasians mixed it tends to track very closely with West Eurasian ancestry. I’ve always found this a bit curious. The only conclusion I can draw from this is that whatever SLC24A5 is under heavy selection for (might not be light skin, it could just be a side effect) East Asians not only found a parallel way to do it, but putting the two genes together in the same population does not have an additive effect.

  22. @Razib Kahn

    “no. it’s genome-wide autosomal.”

    Have a link for that? I was under the impression that the amount was roughly 20%, accumulated from LBK on.

  23. karl, I’d just like to pick a nit with the phrase “cooler climates”. I would prefer “darker latitudes”, because it’s precisely where the climates are warmest at high latitudes that I would expect to see extreme depigmentation.

    On Australia, I would suppose that people at latitude x in contact with much paler people at higher latitudes, would be paler than people at the same latitude x, who are only in contact with people at lower latitudes, if that makes sense. Southern Australia and Southern Europe might be at the same latitude, but one is at the highest latitude possible, with only lower latitudes nearby; while the other is at the same latitude, but is neighbor to occupied lands of much higher latitudes.

    Useful latitude-comparing tool:

    To summarise: the conundrum of why incredibly pale people appeared in north western europe less than ten thousand years ago is complicated 🙂

  24. @karl, I’m still not 100% sure on that point exactly, even in modern people – need to gather more information on frequencies and population modelling with adna to see if this is a robust trend; I don’t think I’ve seen anything like a test on X populations for SLC24A5 derived variant, accompanied by an ancestry model for same populations using adna, and finally cross checking for expected vs predicted frequencies, though I’d think we can now do this OK for East Africa.

    Usually I think it’s something more like “We know populations in East Africa have 0.6 SLC24A5, and have 0.4 West Eurasian ancestry via ADMIXTURE, therefore…” (or “We know X South Asian population have Y ANI and Z ASI, but they have >Y frequency of derived SLC24A5 variant, therefore…”). Which is not wrong if that’s the best we can do, but I’d quite like to be able to see the very controlled version above, using real adna references and a very large panel of populations.

    Though have to say that, while I would assume SLC24A5 was present in ancestral in a WHG population rather than introduced through introgression EHG->WHG, the patterns of selection do seem a bit of a mystery still. (Is it really the case that “WHG proper” have very low frequency (effectively 0) of SLC24A5 derived variant due to different selective pressures? What is derived variant for?)

  25. razib, I went looking for maps of temperature to show warmth extending into the dark latitudes in west eurasia, but what I started finding to my surprise was maps showing darkness extending into the warm latitudes in east eurasia!

    I don’t know what’s going on there in China. Has it always been like that, or is it some sort of modern pollution? I had never heard of extreme depigmentation in east eurasia before you mentioned it, but I thought you might be interested in the maps.

  26. Hardly ‘dark’. More likely cloud cover as % of time – China has always been humid (so poor preservation of ancient human remains cf Europe).

    What ‘colour’ did you think Koreans were, then?

  27. I have argued for years that there is no substantive correlation between the ancestral version of SLC24A5 and dark skin, or that of the derived one and light skin. There is only a link to farmers, so perhaps the ability to change skin color seasonally (tanning) is the important aspect, here. I hope the authors of this study have more than that…

  28. Actually, Vanessa Williams is mulatto elite from a long line of other mulattoes. That’s where the blue eyes come from (as well as the other European features). Real blacks don’t produce blue eyes.

  29. To expand on my earlier post, here are some more data points:

    1. Double carriers of the derived SLC24A5 can be towards the darker side of the spectrum in S Asia. People in Afghanistan to Northern Pakistan and many Northern Indians are generally lighter. People around the Mediterranean are yet lighter (often “olive” in color). People in Central Europe are generally lighter, yet. People in Northern Europe (e.g., Iceland, Ireland, Scandinavia, Northern Holland and Germany, the Baltics, etc.) tend to have yet lighter skin color, if they have any pigmentation to speak of, at all. So there appear to be something like at the minimum five additional mutations that must affect skin color, and that are much more important than SLC24A5 and likely work completely independent of SLC24A5, see also (2) below.

    2. Most North-East Asians have the ancestral version of SLC24A5 – yet, many of them have very light skin color, some with nearly no pigmentation to speak of. Again, it appears that SLC24A5 is not an important factor in general skin coloration, in many populations.

    3. A study found that in Asia, another mutation (rs2470102) also helps explain light skin color. But even the combination of dual alleles in both spots (“H1”) does not explain skin color in almost 40% of the (NE Indian) population: “People who had a combination of similar (homozygous) mutant alleles of both the new and the known SNP had the fairest skin; they are said to belong to the H1 haplotype. The frequency of the H1 haplotype was far higher (96 per cent) in people with lighter skin than in darker skin (37 per cent).”

    Lastly, I haven’t seen anything on SLC45A2 that would change the above interpretation: neither of these if present in the ancestral variety imply dark skin, quite the opposite, we know that additional mutations are required for very light skin, and can generate very light skin even with the ancestral versions of those two alleles. And from everything we know about the UV /vitamin D (and perhaps folic acid precursor) balance and world-wide skin color distribution (extremely high vitamin D diet excepted), pre-Neolithic peoples in Europe very likely had at least moderate to very light skin color.

  30. Cheddar man belonged to Haplogroup U5

    The highest concentration found today is amongst Finns, Estonians and Sami people.

    I was very interest to read that east asian’s lack both the derived variants of SLC24A5 and SLC45A2 found in Europeans but are still light-skinned.

    Surely the twins who made the recreation were given an idea of his appearance by the people involved in the study. I hope they didn’t just assume is appearance.

    I wonder if the genome sequence found contradicts the statue that was created? I’ll i’ve find is that everything just “suggests”, doesn’t prove.

  31. Sorry for this additional post, but I think it touches upon two subjects often ignored:

    1. Protection against UV does not necessarily present itself in the (to us) visible spectrum. Someone might not get a tan or might not show much “sun burn” after exposure (like many of my NE Asian friends), and still might have a very light skin color (perhaps exactly because of their ancestral SLC24A5 variant!).

    2. In early photography, you can find many pictures of peoples at extreme northern or southern latitudes that look darker than what you expected, given local insolation. However, don’t forget that early photographic emulsions were only sensitive to UV, so a person with a summer-reddened face appears dark, and a person with (for us invisible) UV protection/absorption appears genuinely dark on such images.

  32. Re: SLC45A2, I know the usual story is that this explains a good percentage of the variance between European and Middle Eastern populations, yet I do wonder about that.

    The variance seems roughly clinal: Southern Levant (Lebanese, Syrian, etc.)->Northern Middle East(Armenia, Turkey, roughly comparable to Korea/Japan/North China)->Southern Europe->Northern Europe.

    But derived variants at SLC45A2 at rs16891982 seem fixed in Southern Europe, at least in the North Italian panel on HGDP (at 0.8 in the Spanish panel), and the paper on Bronze Age Lebanon last year reported a frequency of 0.64 for present day Lebanese – (contrasted to 0 in BA_Sidon). It seems like if that were a big driver as well, you’d see a Mendelian pattern of people in ME with either a North European like pigmentation or a darker pigmentation, which doesn’t appear to be the case.
    (Instead people look more like they have levels of within population variation similar to Europeans, and less than within Africans or Indians and looks fairly typically normally distributed rather than bimodal?)

    Something like the Tishkoff/Crawford paper for African populations in 2017 with a focus on the ME might help explain some of this?

  33. What ‘colour’ did you think Koreans were, then?

    If you care to reframe your comment in a way that says what you think, rather than a question, I’m willing to respond to that, but I’m not willing to enter into a Socratic dialogue with you in the role of the teacher.

  34. “I had never heard of extreme depigmentation in east eurasia before you mentioned it.”

    Typing ‘Korean men’ or ‘Korean girls’ into Google images will get you any number of photos of very pale skinned people, easily as pale skinned as many north Europeans. Or you could just look at photos on news sites about the Olympic Winter Games.

    People who have been following for any length of time will already know that East Asians also became pale skinned, but via different genetic pathways from Europeans.

    You can also Google for ‘skin reflectance’ and find papers which show that, on average, East Asian skin reflects just as much light as European skin, but with minor spectral differences. Elli
    Angelopoulou at the University of Pennsylvania was an early researcher in this field, and her papers are still worth reading.

    But it was actually intended as a rhetorical question, not a Socratic one.

Comments are closed.