“Ancestral North Indians”, Europeans and pigment

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

Something that has been nagging me about the recent paper by Reich et al. which models Indian populations as a hybridization event between two ancestral groups, “Ancestral South Indians” (ASI) and “Ancestral North Indians” (ANI). As a reminder, the ANI seem to have been rather like Europeans in their allele frequencies, or at least far closer to Europeans than they were to the ASI (it seems that they compared ANI with Western Europeans). This is interesting. They found in the populations surveyed that the low bound for ANI was 40%, the high ~80% (in the supplements they included some Pathans and Sindhis from the HGDP, and that’s where that number comes from). The ~40% low bound for ANI rather surprised me. The populations which they sampled included South Indian tribal groups. In other words, these were the groups arguably least affected by what we term Hinduism and Indian culture (their status as “tribals” as opposed to lower caste or outcaste was generally a function of the fact that they rejected integration and assimilation into mainstream Indian culture and isolated themselves both geographically and in terms of their customs). Just seems weird that these groups would be so ANI.

For a few weeks now Greg Cochran has been asking if I saw something in the paper above about when the admixture between ANI and ASI occurred, or at least if there was a hint about when the authors think it occurred. I said no, there are only hints. I was wrong, I skimmed over the supplement too quickly, they assume 200 generations ago as a parameter in a model they use for simulations. Bingo. Just click the image to the left, and look at the lower right. 200 generations = 5,000 years ago, assuming 25 years for generation time. Let’s assume that a South Indian tribal group is a small deme of ASI surrounded by a very large (infinite) deme of ANI for 200 generations. If I assume a constant outmarriage rate of 0.25% per generation (1 out of 400) then at the present time you’d have the tribal group being ~40% ANI.

OK, what about my idea which I presented to John Hawks that Indians “don’t really look” like a hybridization between Northern Europeans and the ASI, ASI assumed to be similar to the Andaman Islanders (who I do not believe were necessarily “Negritos,” insofar as I suspect their small stature is due to contact with Europeans and Indians, as those who have avoided such contact are seen to be of normal or even above average size for South Asians). Specifically the frequency of light eyes and hair is just way too low among groups which are on the 70-80% ANI range such as Punjabis and Kashmiris, though these groups do tend have more Caucasoid features and lighter (olive) skin. On the other hand, here is something which jumped out at me about the Reich et al. paper: they added two Pakistani populations who fit well in the ANI-ASI cline which most of the Indian groups mapped onto (some groups with “Eastern” origin in both Pakistan and India were discarded from the analysis), and their ANI frequency proportions seemed familiar to me. There are three ANI estimates for both groups:

Sindhi – 78%, 70.7%, 73.7% (78%)
Pathan – 81%, 74.2%, 76.9% (81%)

In the parenthesis is the frequency for the derived (European-like) variant of SLC24A5. The data sets were the same, from the HGDP, though the ancestry estimates used only 10 and 15 of the approximately 50 of each group respectively. There’s a suspicious correspondence here. The lowest frequency of the derived variant of SLC24A5 I’ve seen for a South Asian population is ~30% for Sri Lankan Tamils, with ~50% for Sri Lankan Sinhalese. Remember that a reasonable low bound for ANI for South Asian groups is on the order of 40%.

But what about my contention that other European-like pigmentation alleles don’t fit because the phenotype isn’t what you’d expect. You can look at a blue vs. brown eye variant of OCA2 in the HGDP. Another eye color variant, HERC2. And here is a variant of TYR which causes light skin. The interesting point would be to look at the Indian samples, but I don’t have really good proxies for that (in one paper which surveyed Indian Americans various language groups ranged from 70-100% in derived SLC24A5 frequency, but it is very difficult to imagine that these correspond well to many groups in the Reich paper. Specifically, it’s biased toward higher status/caste groups). I might have spoken too soon, though it still seems to me that something is off. Perhaps Europeans changed after ANI left. Or perhaps ANI changed when it arrived in India. One recent data point which I find curious is that a paper just came out which suggests that populations of the Andronovo culture in Trans-Siberia, which is assumed to be the precursor to the Indo-Iranians, seem to resemble modern day Russians in pigment phenotype. At least judging from the genes extracted and sequenced.
More later when my thoughts become more settled.

Labels: ,

19 Comments

  1. Disclosure: I’m a co-author of the Reich et al. paper.  
     
    Looking at pigmentation genes is interesting, but many of these  
    are known to have been under selection. So if there has  
    been selection in India since ANI/ASI mixing they are not  
    good choices to figure out mixing proportions.

  2. This is a somewhat off-topic question, but how far over on the “light-complexion” end of the Bell Curve do Indians have to be for their fellow Indians to regard them as “non-Indian looking?” For example, I have read some internet comments (Sepia Mutiny, etc.) from Indians who regard GOP gubernatorial contender Nikki Haley as “non-Indian looking.”

  3. nick, yeah. what would i would be curious about is if derived slc24a5 lines up perfectly with your estimates of ANI. if so, that’s awful suspicious.

  4. Razib, 
     
    But another issue is that Andronovo Culture individuals are part of the ancestry of the invading Indo-Iranians, the other part was BMAC

  5. That should be BMAC

  6. http://en.wikipedia.org/wiki/Bactria?Margiana_Archaeological_Complex

  7. “Perhaps Europeans changed after ANI left. Or perhaps ANI changed when it arrived in India.” 
     
    Or perhaps they weren’t Europeans at all? Perhaps they were Middle Easterners – the same Middle Easterners who brought agriculture to Europe?

  8. The scenario of a small South Indian deme constantly surrounded by an ocean of North Indians does not really seem to fit the conventional wisdom about Aryan migration (relatively small tribes taking control of a larger indigenous population, even in the North).  
     
    Perhaps the North Indians were actually the stock of the Indus Valley folks, presumably the first people who grew enough to have such a significant impact on India as a whole? That would imply that both the Harappans and the Aryans were actually of “Foreign” (European or Middle Eastern) descent. If confirmed, I predict mass fits of hysteria in the Hindutva rank and file…

  9. jan,  
     
    I am thinking that the Indus Valley civilization was populated mostly by people of Middle Eastern descent – perhaps relatives of the Elamites. By this model, the pre-agricultural South Asians did better than the pre-agricultural Europeans, who are even less well represented in the modern European genome.

  10. would be interesting for preform the same analysis as here on northern europeans.

  11. Razib, 
     
    Based on what I’ve seen on 23AndMe of Northern Europeans, Scandinavians show affinities with South/Central Asians, whereas Southern Europeans show affinities with Middle Easterners.

  12. Based on what I’ve seen on 23AndMe of Northern Europeans, Scandinavians show affinities with South/Central Asians, 
     
    autosomal? not just Y?

  13. Right, automsomally

  14. I share genomes with about 250 people, and my top 10 overall autosomal matches on 559259 SNPs are: 
    1. Irish-English-Dutch 
    2. Swedish 
    3. Irish 
    4. Swedish-Scots-German 
    5. Irish-German 
    6. English-German 
    7. Irish 
    8. English 
    9. Cornish-English-Irish 
    10. English-Spanish 
     
    But on certain traits my matches are not as expected, say for instance the 502 SNP’s tested under pigmentation, my top matches are: 
    1. Estonian 
    2. Finnish 
    3. Hungarian 
    4. English-French-Italian 
    5. Finnish 
    6. English-Scottish-Irish 
    7. Russian (Google co-founder) 
    8. English-German-French 
    9. Cornish-English 
    10. English 
     
    So my overall matches are mostly Irish, English, Swedish and German as might be expected, while my pigmentation skews towards North Eastern Europe.

  15. This is a somewhat off-topic question, but how far over on the “light-complexion” end of the Bell Curve do Indians have to be for their fellow Indians to regard them as “non-Indian looking?” For example, I have read some internet comments (Sepia Mutiny, etc.) from Indians who regard GOP gubernatorial contender Nikki Haley as “non-Indian looking.” 
     
    1) People from Northeastern states have had situations when they were not considered as ‘Indian’, due to thir looks. 
    2)We see a significant number of ‘Fair’ south indians and very dark Sikhs and other north Indians.

  16. People from Northeastern states have had situations when they were not considered as ‘Indian’, due to thir looks. 
     
    even i have a tendency to “look asian” in some photos, though i’m obviously brown. so i think this is particularly retarded. i’m sure that some people in my extended family would fall under the ‘not brown’ category because of their asian appearance (i’ve got relatives nicknamed “jackie chan” for their resemblance). 
     
    as to people looking “not indian,” i think the key is can they pass as white. to white people. but honestly, i think white looking indians get more slack than asian looking indians. why else would so many people in bollywood be white indians?

  17. So my overall matches are mostly Irish, English, Swedish and German as might be expected, while my pigmentation skews towards North Eastern Europe. 
     
    really strange for a ‘black irish’ like you :-)

  18. why else would so many people in bollywood be white indians? 
     
    Historically Bollywood had many heros/ male leads from Punjab, where people tend to be fair. The early female leads were Parsi/Anglo Indian/Muslim, who all tend to be fair. 
     
    That there is a preference for ‘Fair’ skin cannot be denied, given the huge profits skin creams are making.

  19. one thing that i think is important. SM readers are mostly american. they’ve probably been the “brown person” their whole life, blah, blah. might have some resentment issues with brownz who can “pass” if they want. i knew a half-brown kid who americanized his first name and easily passed as white. i got a sense his brother who was a touch too tan to pass easily resented him for that (though most of the racism directed at him was because people assumed he was puerto rican).

a