Gene Expression

Skin color is not race

One of the peculiarities of American discussion about race is that skin color is assumed to be synonymous with racial distinctions. That is, skin color is not just a trait, but it is the trait which defines between population differences. There’s a reason for this, the skin is the largest organ and it is very salient. Populations with little phylogenetic relationship to each other, from India to the Pacific to Southeast Asia have been referred to as “black” by lighter-skinned populations. No population is referred to by their neighbors as those “straight hairs,” to my knowledge. But another point in the United States is that historically the black-white dichotomy has dominated our historical narrative to the exclusion of others. (with an asterisk for Native Americans) Though black Americans are an admixed population, which is around 20% European in ancestry with considerable variance (e.g., ~10% of the black American population is more than 50% European by ancestry), white Americans are relatively homogeneously European. Unlike the populations of Latin America admixture did not occur along a continuum, and due to hypodescent there was a categorical binning where black Americans with any visible African ancestry were categorized as non-white (and some with no visible African ancestry!).

But the past is not the present. Today Americans of Latin American origin are a larger proportion of the population than black Americans. These groups are often admixed in ancestry, and exhibit a continuum of appearance. The mean proportions of ancestry can also vary from region to region, as evidenced by the recent paper I pointed to earlier this week. While black Americans and white Americans are nearly disjoint in physical appearance, there are Latinos of all phenotypes. Most understand that the quasi-racial status of the Hispanic/Latino category is awkward, and at some point the old paradigm will need to shift.

Not only is the United States transitioning to beyond a black-white model but we also now understand the genetics of human pigmentation. The genes which are responsible for most of the variation between human populations in complexion have been identified, including their effect sizes. About half a dozen genes of large effect can explain ~90% of the between population variance. Some of the genes, such as SLC24A5, are so ancestrally informative that sans selection you can ascertain ancestry in mixed populations simply by inspecting its frequency.

With that said, I recently reread an “old” paper from the ancient days before human pigmentation was elucidated with any depth, that is, 2004. It might lack a firm theoretical framework because of ignorance of the genetic architecture of skin color, but there’s some interesting data on admixed populations and phenotypes worth revisiting. Implications of correlations between skin color and genetic ancestry for biomedical research.


There are 4 populations here. Latinos from Colorado were left out of the chart because the method used to measure skin color was different and so not comparable. Note the unsurprising variation across all groups, and the mean values. African Caribbeans are darker skinned, on average, than African Americans. This is what you would expect from the social-historical context. In the United States majority black regions (e.g., the cotton country of Mississippi and Alabama, the coastal regions of South Carolina) were exceptional. In contrast in the Caribbean black slaves were the overwhelming majority in most of the British colonies. As a point of fact they were slave-societies where a small transient, and sometimes absent, white minority extracted surplus through the forced labor of Africans who had to be imported at very high rates because of mortality. The Mexican sample is from Guerrero, which is a region where mixed-race individuals are more Amerindian than European. The Puerto Ricans are from New York. Some qualitative research I’ve read suggest that the Puerto Ricans who immigrated to New York City were darker-skinned and more African in background than those who remained on the island, because the more European elites had less of an incentive to leave. So it may be that the New York Puerto Rican sample is not totally representative of Puerto Rico itself. Note that the variance for the Latino samples is smaller than from the predominantly African ones. Why might this be? First, the difference in skin color between Africans and Europeans is greater than between Europeans and Amerindians. The potential range is therefore greater in a European X African scenario than a European X Amerindian one. Additionally, the variance will be lower as you change the proportions so that they are less balanced. Consider SLC24A5. A population which is 50% European and 50% African in ancestry will have a heterozygosity of ~0.5. A population which is 75% European and 25% African in ancestry will have a heterozygosity of ~0.38. One which is 90% European and 10% African will have a heterozygosity of ~0.18.* Note that the average melanin index for white Americans is ~30, while it is ~60 for West Africans.

In any case, what’s the admixture of these groups?


You can look at the paper to see the average admixture for the populations as a whole. Looking at this chart seems to overstate the proportion of indigenous ancestry in the African groups, and African ancestry in the Mexican one, probably because we don’t discern very easily the enormous numbers concentrated on the edges of the diagram. On the other hand, the Puerto Rican sample is clearly trihybrid, We know this from other research. As one would expect there also seems to be somewhat more African than Amerindian ancestry among the Puerto Ricans. Unlike the Amerindians, there was constant gene flow from Africa (and Europe) for generations. The European proportion is also preponderant among Puerto Ricans, which matches what you might expect based on their skin color (a substantial minority of Puerto Ricans have the mean complexions of white Americans).

Now we look at how skin color genes relate to admixture in the individuals.

Table 1 Relationship of melanin content and individual ancestry

  Spearman’s rho (95% c.i.)
Relationship of melanin index and ancestry
African Americans (W-AF) 0.440 (0.330 to 0.538) <0.001
African Caribbeans (W-AF) 0.375 (0.239 to 0.496) <0.001
Mexicans (I-AM) 0.212 (0.057 to 0.357) 0.008
Puerto Ricans (W-AF) 0.633 (0.457 to 0.761) <0.001
Relationship of skin reflectance and ancestry
Hispanics (I-AM) -0.259 (-0.141 to -0.369) <0.001

Ancestral proportion axes are indicated by the abbreviations W-AF (West African) and I-AM (Indigenous American).
Ninety-five percent confidence intervals for the correlation estimates are shown in parentheses.

As you can see the correlation is relatively weak between ancestry and complexion in the Mexican and Mexican American sample. It is stronger among Puerto Ricans. And modest among the African groups. In hindsight we know that there are a few genes of large effect segregating within the population, so sample variance will result in a relatively large range even within families, who presumably tend to share approximately the same proportion of ancestries.

I will now quote from the paper:

It is important to consider the implications of the wide variation in the strength of the correlation between constitutive pigmentation and individual ancestry in these admixed samples. This variability is presumably a reflection of differences in the degree of population structure present in each population or of the levels of pigmentation differences between the parental populations and the number of genes involved. For example, the strong correlation observed in Puerto Ricans seems to indicate that continuous gene flow, assortative mating or both factors are important in this population. On the contrary, the correlation between melanin content and ancestry in Hispanics, though significant, is weak. This is consistent with historical data indicating that this population appeared as a result of a relatively old admixture event and that independent assortment has greatly decreased the association between unlinked markers created by the admixture process…Alternatively, the differences in the extent of the correlation between constitutive pigmentation and ancestry may be due in part to admixture histories involving populations with widely different pigmentation levels. The Puerto Rican individuals have substantial contributions from three parental groups (Europeans, West Africans and Indigenous Americans). The African American and African Caribbean individuals have contributions mainly from West Africans and Europeans, and the Mexican and Hispanic individuals have primarily European and Indigenous American ancestry.

Remember the Uyghurs? They are an admixed population of East and West Eurasian origins, but as there has been little new gene flow and there has been around 2,000 years of random mating, within population variance in ancestral quanta is relatively small. Nevertheless there are Uyghurs who look mostly European, and Uyghurs who look mostly East Asian, with most a mix. The reason is that only a small subset of genes control the variation which we use to classify people as members of a particular population. It looks like something similar happened among the Mexican Mestizo populations.

Now to Puerto Rico. On the one hand the caste structure might have been stronger, resulting in assortative mating among phenotypes. But I think gene flow is a more likely answer. Puerto Rico is a small island, and has been subject to a lot of immigration from Spain in the 20th century. Correlation between skin color and European ancestry probably has to do with a great deal of recent admixture.

The point is that though there is a correlation between skin color and ancestry, that correlation is imperfect, and its informativeness is sensitive to various population genetic parameters. On an operational level this does not mean that people need to know what the allele frequencies of SLC24A5 in the HGDP data set are. Rather, Americans probably need to start being more cautious about using the phenotypic shorthands that they use to signify race. I’ll conclude with a quote from Eugene Volokh’s How the Asians Became White:

Asians are now white.

Don’t believe me? A recent MSNBC news headline announced a “Plunge in Minority University Enrollment” at the University of California, with UC Berkeley reporting that “minority admissions had declined 61 percent.” Actually, the total percentage of racial minority students at Berkeley, Asians included, fell from 57% to 49%. If you exclude the burgeoning group of people who decline to state their race, the minority percentage fell only three percentage points, from 61% to 58%.

The drop was exclusively among blacks, Hispanics, and American Indians. Asians, who make up less than 10% of the California population, apparently aren’t a “minority.”

Or listen to former California Chief Justice Rose Bird. Last year, she wrote a commentary saying that, without race preferences, the UC system would be “nothing more than a group of elitist, `lily white´ institutions.” A coorganizer of Jesse Jackson’s recent march in favor of race preferences called UC Berkeley’s law school, whose entering class last year was 20% minority, including 14% Asian, “lily-white.” Asians aren’t just white: They are lily-white.

I first noticed this effect 10 years ago, at a party where a friend of mine commented that the guests were all white. I responded by mentioning about a dozen Asians; oh, she said, that’s right, but you know what I mean. At a recent UCLA conference I attended, two speakers complained that everyone on the panel was white, without even realizing that one of the speakers was ethnically Chinese,and another was an Asian Indian with skin darker than that of many American blacks.

* Frequencies are disjoint, so do the HWE based on ancestry.


  1. #1 toto
    May 14, 2009

    So, in full opposition to the European method of using “European” as a euphemism for “White”, in the US, “White” is increasingly being used as a euphemism for “non-African-Black”?

  2. #2 bioIgnoramus
    May 14, 2009

    As I understand it, the out-of-Africans had dark skin; their descendants in the unsunny zones of, for example, Europe and NE Asia evolved lighter skins, probably for the vitamin D advantage. Did the Aborigines of Tasmania? Did the Red Indians of Patagonia, Southern Chile and Tierra del Fuego? If not, why not? Could it be an effect of population density, or of population size, or of….?

  3. #3 elambend
    May 14, 2009

    This is what is interesting about children with parents of mixed ancestry; it can become a crap-shoot as to what the children look like. This can be tragic in societies that prefer one skin tone over another, but interesting and neat for those who prefer variation.

  4. #4 pconroy
    May 14, 2009

    I find that for people who act, think and behave like me, I view as kin – despite phenotype differences.

  5. #5 Ikram
    May 14, 2009

    You may want to re-read Micheal Lind’s classic 1998 article The Beige and the black. It’s a little dated now, and too focused on intermarriage, but still relevant.

  6. #6 jim
    May 14, 2009

    Despite the rise in Asian and Hispanic immigration in recent decades, America still is primarily a bi-racial country. There are whites and blacks. And everybody who isn’t black eventually becomes white.

    This might change. Mexicans might become America’s third race. But once they lose their Spanish assimilation kicks in hard. And with Mexicans, unlike blacks, intermarriage groups them with whites. I’ve known many half-Mexican suburbanites and they have very weak Mexican identities. The continuous influx of Mexican immigrants, and their concentration in the Southwest, slows Mexican assimilation, but it’s steady and ongoing. Mexican immigration will slow in the next few decades and America will rapidly assimilate their kids and grandkids.

    As to Asians, Asian assimilate so rapidly that, yeah, Asians are white. Sure there are immigrant pockets. But 2nd generation suburban Asians become super white, super quick. Often literally, since Asian chicks are desirable to white men.

    When I say Asian, I mean East Asian, but most South Asians are basically white, too. Large scale South Asian immigration is the most recent new group, but it’s also the group with the most stringent IQ filter applied to it. The wealth and high IQs of American South Asians make them white, despite their dark complexion.

    If we ever import a few million South Asian morons, it’ll be interesting to see how America’s racial categories expand. I also don’t foresee white dudes chasing South Asian chicks; not the same way white guys chase East Asian girls. But I do see white women chasing the high-IQ, high-earning South Asian men. All American women love rich doctors.

  7. #7 ringo
    May 14, 2009

    I – Cuba had the opposite population pressure from Puerto Rico – the Cubans who came to the US are light skinned, the ones who stayed behind are much darker.

    II – Skin color is a verbal shorthand, but not sufficient a measure. I am regularly identified as “black” in the south east, and “mexican” in the south west. (Pale olive skin, curly black hair, brown eyes). Like gender, this is a constellation of traits, not a single organ.

    III – The Asians-are-white issue (and South-Asians-are-white issue, and Jews-are-white issue) has had some bizarre implications for affirmative action. My alma mater (which shall remain nameless, but it’s a vertical palindrome) has spaces set aside for “historically underrepresented minorities”, for which the bar is academically lowered, purportedly in an attempt to have the student population look something like the general population. “Whites,” on the other hand, compete with Asians and Jews and Indians for the remaining majority of slots. The result being that there is a massive skew in the population, which no longer bears any resemblance to the general population whatsoever.

  8. #8 Jason Malloy
    May 14, 2009

    But once they lose their Spanish assimilation kicks in hard.

    Nope, this is contradicted by the evidence. See the book Generations of Exclusion:

    “The study contains some encouraging findings, but many more that are troubling. Linguistically, Mexican Americans assimilate into mainstream America quite well–by the second generation, nearly all Mexican Americans achieve English proficiency. In many domains, however, the Mexican American story doesn’t fit with traditional models of assimilation. The majority of fourth generation Mexican Americans continue to live in Hispanic neighborhoods, marry other Hispanics, and think of themselves as Mexican. And while Mexican Americans make financial strides from the first to the second generation, economic progress halts at the second generation, and poverty rates remain high for later generations. Similarly, educational attainment peaks among second generation children of immigrants, but declines for the third and fourth generations.

  9. #9 TGGP
    May 14, 2009

    You’ve said what others think you are, ringo, so how about revealing the correct answer?

  10. #10 Sandgroper
    May 15, 2009

    Bio – there is a gradation in skin colour in Australia, from darker in the north to lighter in the south. Plus factor in that they were hunter gatherers who largely went naked, not farmers who wore clothes.

    I guess the same could apply to Tierra del Fuegans – despite the climate, they went naked.

    It suggests that white Australians living in Melbourne who suffer from Vitamin D deficiency due to inadequate UVB exposure in winter should also start going completely naked, but that conjures up an image too horrifying to even contemplate.

  11. #11 Tod
    May 15, 2009

    there is a gradation in skin colour in Australia, from darker in the north to lighter in the south“.

    Yes there is some, but not enough to explain why the extremes of skin colour evolved in Europe and Africa.

    Brace studied skin color variation in Amerindians, who have inhabited their continents for 12,000-15,000 years, and in Australian Aborigines, who have inhabited theirs for some 50,000 years. Assuming that latitudinal skin-color variation in both groups tracks natural selection by solar UV, their calculations show that this selection would have taken over 100,000 years to create the skin-color difference between black Africans and northern Chinese and ~ 200,000 years to create the one between black Africans and northern Europeans (Brace, C.L., Henneberg, M., & Relethford, J.H. (1999). Skin color as an index of timing in human evolution. American Journal of Physical Anthropology, 108 (supp. 28), 95-96“.

    white Australians living in Melbourne who suffer from Vitamin D deficiency due to inadequate UVB exposure in winter

    Vitamin D synthesis isn’t ‘inadequate’ in the N. European winter, it’s totally absent. Whites in Australia in t-shirts are getting more UVB exposure in a year than any Europeans, even naturists. If there is a problem with ‘D’ it’s more likely to do with Adiposity in Relation to Vitamin D Status
    ” Precisely measured total body fat is inversely associated with 25-OH-D levels ”

  12. #12 pconroy
    May 15, 2009


    I also don’t foresee white dudes chasing South Asian chicks; not the same way white guys chase East Asian girls

    Not true in my case at least, I find South Asian women far more desirable than East Asians – they are curvier and more feminine in general, for starters.

  13. #13 toto
    May 15, 2009

    But “race” is also a social construct, which is largely mediated by skin colour, even if you remove all the rest.

    In South Asia, parents with nearly identical skin tones can have children ranging from dark brown to pale “wheatish”. Large variation within families is routine.

    And still, the darker ones do get picked upon, even by their own family – especially if they’re girls. Remember, this is the country where Kareena Kapoor is a sex symbol – apparently for no other reason than being ice-white.

    Thus, India is a stark counter-example to the idea that widespread interracial marriage will free us from racial prejudice. Skin colour on its own is quite enough to keep the system going, even after the abolition of genetic dichotomies.

  14. #14 Sandgroper
    May 16, 2009

    I haven’t read Brace et al, but if Maciej Henneberg’s contribution to that was anything like his claim that the floresiensis skull had modern dental work, I think I might take it with a grain of salt. Did they factor in diet and clothing to those calculations? If so, how? If not…

    And I’m aware of the adiposity effect, but that doesn’t explain why Vitamin D deficiency in white Australians living in southern Australia has increased since the start of government campaigning about skin cancer and the use of protective clothing and sunscreens. Unless you think that using sunscreen makes people fat, or that it just happened to coincide with Australians becoming more obese.