The image to the left is the ‘average’ face of a Mexican woman as generated by the University of Glasgow Face Research Lab. Aside from the fact that the face is prettier than the typical human because of the well known tendency of averaging facial features removing unattractive asymmetry it is racially what you might expect, a synthesis of an Amerindian and European face, with an Amerindian skew. But a phenotypic average only tells you so much. Variation is one of the key ingredients in evolutionary processes, and by getting a sense of a population’s variation you can infer things about its past and possible future history. For example, if that variation is heritable, then it is amenable raw material for adaptation. In contrast, if the variation is due to environmental parameters then it is not going to be appropriate input for adaptation via natural selection. In a nation like Mexico we see the full range, from ‘typical’ Amerindian phenotype, to someone who looks to be fully European (with a small minority with visible African ancestry).
But if the phenotype is heritable, then underlying this variation is genotype. The extent that genotype controls the variation is contingent upon heritability. The heritability of behavioral phenotypes is often around ~0.5. But for physical traits such as height or pigmentation the heritability is much closer to 1, on the order of ~0.8 to ~0.9. That means 80 to 90 percent of the variation of the trait across the population is due to variation in the genes. When we code someone as “Amerindian” or “European” or “African” we are assessing phenotypes with a strong underlying genotypic component. A new study in PLOS GENETICS outlines just how this plays out in Latin America, a region of the world which has the virtue of being a living experiment in admixture between different geographic races over the past 500 years.
The current genetic makeup of Latin America has been shaped by a history of extensive admixture between Africans, Europeans and Native Americans, a process taking place within the context of extensive geographic and social stratification. We estimated individual ancestry proportions in a sample of 7,342 subjects ascertained in five countries (Brazil, Chile, Colombia, México and Perú). These individuals were also characterized for a range of physical appearance traits and for self-perception of ancestry. The geographic distribution of admixture proportions in this sample reveals extensive population structure, illustrating the continuing impact of demographic history on the genetic diversity of Latin America. Significant ancestry effects were detected for most phenotypes studied. However, ancestry generally explains only a modest proportion of total phenotypic variation. Genetically estimated and self-perceived ancestry correlate significantly, but certain physical attributes have a strong impact on self-perception and bias self-perception of ancestry relative to genetically estimated ancestry.
The phylogeographic aspect of this paper is not too interesting to me, as it confirms what we’ve known (e.g., more Amerindian ancestry in northern Brazil, Mexicans are somewhat more Amerindian than they are European, etc.). Rather, the biggest findings are those which relate physical appearance, self-identity, and genetic ancestry. In Europe someone who identifies as “white” is invariably ~99% European when assessed using a genetic method (the ~1% balance is often from Iberia). More precisely, white Europeans are ~99% West Eurasian, since a non-trivial amount of trans-Mediterranean gene flow has occurred, meaning there isn’t a clear boundary between Europe and nearby regions. Similarly, in Sub-Saharan Africa someone who identifies as “black” is likely to be nearly all Sub-Saharan African. This is often not the case in Latin America. That is, those who identify as “white” or “black” often have substantial admixture from other geographic racial groups.
One of the major drawbacks of this study is that it relies on 30 ancestrally informative markers (AIMs). Though this is acceptable in forensics, some of the ancestry inferences made on an individual basis are a touch less accurate than they would be on a dense marker SNP chip (e.g., the 650,000 SNPs used on the HGDP). The modest correlations here are probably a little lower than they would be if the ancestry was more accurately adduced. But in the broad sketch the conclusions are likely defensible. One result which may surprise then is the very modest correlation between physical traits and ancestry. Here’s the quote from the paper:
Regression of phenotypic variation on genetic ancestry (taking Native American as reference) demonstrates a significant effect for most of the traits examined (p-value <10−3 using a conservative Bonferroni multiple testing correction, Table 2). Among the non-facial phenotypes (accounting for sex, country, age, educational attainment and wealth) higher European ancestry is associated with: increased height, lighter pigmentation (of hair, skin and eyes) (Figure S6), greater hair curliness and male pattern baldness. Hair graying approaches statistical significance (p-value 10−2). Higher African ancestry is associated with: increased height, higher skin pigmentation and greater hair curliness. The proportion of phenotypic variance explained by ancestry is highest for skin pigmentation (19%) followed by hair shape (8%) and color of eyes and hair (4% and 5%, respectively) but at most 1% for the other phenotypes.
As I said it could be that the AIMs aren’t quite as accurate as they should be, and are underestimating the ancestral fractions on the individuals at the extremes (e.g., someone who is 100% European is estimated to be 95% European, because the marker set lacks precision). So you might bump up the proportion of variance explained a bit, but likely this still seems way too low to you intuitively. There are a few things going on here. First, skin color is controlled between populations by a relatively small set of genetic loci. This means that in admixed populations the sample variance, the random draw of genotypes across the loci, is going to vary a lot even in individuals with the same ancestry. Because of the relatively small number of large effect loci skin color is a trait which shows a lot of variation within families where ancestry is geographically diverse. And within families, or at least across full siblings, total ancestry is not going to vary that much. Second, for some of the “traits” in question that are being measured there is just a lot of variation within geographic races. It makes sense that ancestry would explain only a small fraction within this pooled data set. And yet people can recognize a set of features which are clearly European or Amerindian or African. I think the answer here is that you are picking up on correlation structure across the traits. A suite of subtle facial contours for example connote “European” in a Gestalt manner, even if quantitatively each contour trait has a lot of variation within a population and overlaps across them.
Where this all “cashes out” though is in the intersection of the sociocultural and biological. Within the paper itself they observe a few trends which would not be surprising. Skin color and hair form are very salient characteristics, and lead individuals to shift their estimates of their own ancestries. Those with lighter skin tend to overestimate their European ancestry fractions, while those with curlier hair overestimate their African ancestry. These are traits which have the characteristics that they are quite ancestrally informative to particular geographic races, and, very visible (unlike, say, Duffy status). Within these data there are also particular patterns which are intriguing and less obvious; those with low amounts of Amerindian ancestry underestimate the fraction, while those with higher levels overestimate it. The details of these patterns are obviously contextual in terms of time and place (e.g., in Henry Louis Gates Jr.’s genealogy specials many celebrities seem to yearn for exotic lineages, which would not be the case in past decades). What is more interesting is that fine grain patterns of variation in genetic ancestry and how they deviate from perceived ancestry can finally allow social scientists to get a better grip on patterns of discrimination (or lack thereof). It is not entirely uncommon in Latin America for full siblings to sometimes be socially perceived to be different races because of the random segregation of salient characteristics. In the aggregate these sorts of cases would allow one to estimate the effect of social perceptions, slights, or advantages. With the genetic dimension one could also ascertain the possibility of group differences, because many subtle characteristics are going to track genome-wide patterns, rather than a few phenotypes which society privileges when sorting people by geographic origin.