There’s a report in Science about a new short paper about Neandertal pigmentation genetics. The context is this. First, in 2007 an ingenuous paper was published which inferred that it may be that Neandertals had red hair, at least based on an N = 2 from two divergent locations. The new study looks at three Croation samples, and reports genotypes which are correlated with a swarthier phenotype in modern populations. But the results are neither here nor there: everyone interviewed in the paper assumes that like modern Europeans Neandertals were a polymorphic set of populations when it comes to pigmentation. There are lots of reasons for this agreement, despite issues one might take with this paper.
The report on the paper in Science has two sections which I want to zoom in on. First, “Nearly 60% of the formula’s predictions matched the subjects’ actual physical appearance, the authors say. The team considers that accuracy rate satisfactory, given the complexity of the genetics behind skin color and other physical traits.” Do you consider 60 percent satisfactory? What curve are you grading on? I’m willing to bet that the reporter didn’t consider 60 percent satisfactory, and neither do I. If you look in the paper you’ll see that their method predicts that a Yoruba in the HGDP sample has blue eyes and red hair. Several of the Papuans are predicted to have blue eyes.
But I also think that the Science piece did not do justice to what we know about pigmentation genomics, which is a lot. Case in point: “A particular form of the gene known as TPCN2, for example, bestows brown hair in modern humans; any other form means hair that’s another color.” Since the reporter is probably not an expert someone must have fed them this information. Or, the reporter distorted the researcher’s meaning. Because this is totally misleading. There are SNPs on this locus which correlate with brown vs. blonde hair, but my blonde father-in-law shares the same genotype as some of my siblings (who have black hair). The effect of this variant is strongly conditional upon other variants. (I looked at the same SNPs they did in the paper)
The scientists themselves sometimes have a problem communicating. For example:
…He argues that some pigmentation genes have such a powerful effect that they override the combined contributions of many weaker genes—a phenomenon that would render the new study’s simple gene addition inaccurate. The lighter skin color seen in Europeans, for instance, is due almost entirely to a single gene, he says. “We know that there are some genes that have a very strong effect” on physical appearance, he says.
This is artlessly rendered. If light skin in Europeans was due to a single gene of large effect then ~20 of African Americans would have nearly white skin (assuming dominant effect, it would be 4% if it was recessive). There are two candidate genes which come to mind as possibilities for what might make Europeans white in the mind of this researcher. The first is SLC24A5, a locus which is nearly disjoint in frequency between Europeans and Africans. This locus alone can explain 25-40% of the between population difference in pigmentation. But this locus does not make you white; otherwise I would be white! I’m a homozygote for the “European” variant of the SNP at this locus which differs across populations, as are both my parents. The total frequency of the “European” variant in South Asians is probably comfortably above 50%. A better candidate would be a SNP on SLC45A2 which is present in high proportions in European populations (~90 percent within Europe, ~10 percent in South Asia). But observe that this allele on SLC45A2 is not sufficient for light skin.
To better understand what’s going on here one just has to keep in mind how we now know pigmentation genetics works. It is most definitely not a single gene trait. But neither is pigmentation highly polygenic in the same manner than height or intelligence are. Rather, the genetic architecture of pigmentation is finitely comprehensible. For example, you could explain more than 90 percent of the between population variation between Africans and Europeans in complexion by the differences on just 5 genes. If you took every population in the world I would be willing to lay down a bet that over 90 percent of the between population variation in the pool could be explained by variants on 30 genes. 30 genes is a human-sized number that you can comprehend. A subtle nuance though is that there are different variants within particular genes. For example, both Europeans and East Asians have derived mutations in the OCA2 gene which seem to result in lighter skin, but they’re on separate mutations.
Another phenomenon to keep in mind is that the effect size of the variants seems to exhibit a power law distribution. One or two variants explain an outsized proportion of the difference, a few more a modest proportion, while many others occupy the long tail which explains less than 10 percent of the distribution. In other words, the state of just a few variants can give you a sense of the pigmentation of a population.
To bring it back to Neandertals, another issue is that this human lineage flourished for at least ~100,000 years. The best recent work on natural selection in humans implies that pigmentation can change on the order of ~10,000 years, in particular from dark to light. It would be totally unsurprising if Neandertals were polymorphic in the trait, and, they evolved toward a lighter condition for whatever reason several times, just as modern humans have. But just because we know this general truth does not mean that we have a good grasp of specific aspects of the phenomenon.