To the Bottom of Race

As Andrew observes, there has been a kerfuffle recently on science blogs between anthropologists and geneticists, precipitated it seems by the two parties working simultaneously on contiguous topics without much interaction with the other and building separate and seemingly contradictory consensuses.  A review in The American Scientist of two recent anthropological books declaring race a null concept biologically, much to the surprise of biologists, forcing the two views into conflict.

The anthropologists argue that genetic variation in anscetral human populations is continuous with respect to geography, a finding that the recent glut of genomic data has amply supported.  This clinal variation implies that there are no discrete groupings of like people: you are likely most similar to those who originate from the exact same location as you and less so to others roughly proportionate to how far their point of origin differs from your own.  What more, the genetic differences between geographically disparate groups are quite shallow (a product of recent radiation and regular inter-migration) and few alleles are exclusive to any one location, albeit present in different proportions (e.g., both Norwegians and Masai have the blue eyed alleles, but the Norwegian population simply has them in greater proportion to other eye color alleles).

Because there are no “natural” breaks in the genetic variation and the differences between all groups are quite meager, any attempt to divide humans by anscestry will necessarily be arbitrary.  There is no reason to suggest that the current division by continent of origin is any more valid than calling Frenchmen a race separate from Germans (in fact, it might even be less valid, as the massive genetic variation in Africa makes African to non-African comparisons similar to comparing apples to orange trees).  The best way to conceive of humanity is to imagine that we are one cohesive, interbreeding population with genetic variation purely a function of isolation-by-difference.

Jan Sparr, the author of the review, also cites biologist Richard Lewontin’s argument that most variation at any locus exists within populations not between them (85% to 15%, according to Lewontin).  However, in 2003, statistician and evolutionary biologist A. F. W. Edwards noted an elementary mistake in Lewontin’s math: he assumed that allele frequencies vary independently across loci.  In fact, there seems to be strong correlations between different alleles between groups, and, when one considers variation at multiple loci simultaneously, major differences between all groups (no matter how one hierarchically defines them) emerge.  This is how programs like ADMIXTURE (see below) can be used to track nest differences between and within human populations.  However, my sense of things is that anthropologists, Sparr apparently excepted, have acknowledged Edwards’ critique and no longer lean heavily on Lewontin’s findings.  Criticisms of Edwards and of genetic clustering identified by software can be easily rolled into the above argument about continuous variation, and thus that remains the primary argument.

Trolling through the genetics blogs, I have seen three basic variants of responses to the anthropologists’ argument:

1. That it is trivially true.  All parties seem to agree at this stage that there are substantive genetic differences between people of different geographic origins and anscestries.  Understanding and cataloguing the extent of this variation is what is actually interesting for researchers of the human populations, anthropologist and biologist alike.  That race as commonly understood is an arbitrary (and leaky) heuristic for conceptualizing ancestry is almost a moot point.

2. That human populations are not distributed evenly.  The anthropologists argue that genetic variation between peoples is a continuous function of geographic distance: genetic closeness descends continuously the further one gets from their point of origin without any obvious discontinuities.  However, this point is mostly relevant insofar as human populations are roughly evenly distributed, which is obviously not the case.  Here is a simple demonstration of this point.  Imagine looking at a narrow longitudinal cross-section of West Africa, let’s say from Tunisia to Nigeria.  If you were to sample one person every 100 miles going north, you would likely note a roughly continuous degree of genetic change relative to distance traveled with no discernable breaks, just as the anthropologists theorize.  However, if you were to sample all people within that cross-section, you would likely find that most fall into two discrete, easily differentiable groups: those from central Nigeria and those from the Mediterranian coast, where the huge concentration of people live, with the small number of Tuareg and Berber tribesmen falling somewhere between.  These differences in population densities relative to geography cause noticeable discrete clusters to fall out of any random genetic sample of human beings, even with continuous variation across geography.

3.  The focus on anscestral origins ignores major population changes of the last 10,000 years.  This point is related to the previous one: not only are certain populations more densely packed than others, some have expanded tremendously based on differential access to agriculture and technology, displacing or eliminating the native inhabitants and leaving ethnically homogenous swathes with discontinuous boundary lines.  This view agrees with the anthropologists in disgarding discrete primordial differences (original variation is clinal with respect to geography), but notes that certain groups gained access to agriculture before others, allowing them to expand their populations rapidly, displace their nomadic neighbors, and leave their genetically like progeny in their place, disrupting for all practical purposes the relationship between genetic variation and geography.  The most obvious example of this would be the Han Chinese, who expanded up and around the Yellow and Yangtze River basins, creating a largely uniform and densely populated mass of genetically similar people in their wake.  However, this model can be recreated repeatedly: for the Bantu people in the African savannah and Sahel, for Polynesian wayfarers in the Pacific, and even for the population of Europe by hypothetical Indo-Europeans.  These expansions create significant genetic differences between neighboring agriculturists, which enables the discrete differentiation of people.

I have not found a satisfying response to any of the above counter-claims from the anthropology blogs, some of whom have instead responded with equal parts dudgeon and spittle.  However, via Andrew, Kenan Malik has a very comprehensive assessment of the strengths and weaknesses of both sides and comes to a conclusion (race may be arbitrary but an understanding of ancestry of some sort is important) that seems basically correct in my mind.  Of all the posts surveyed, its the one I recommend the most.