RPM points me to a post at Salamander Candy which discusses the usefulness of neutral markers in conservation genetics. Obviously this complements my recent posts about introgression, and in fact, my last entry was a comment on a conservation genetic paper. Here is the important point from Salamander Candy:
The problem is, a growing body of evidence suggests that patterns of variation and divergence in adaptive traits are not well reflected by neutral markers...In the hypothetical species mentioned above, a small amount of gene flow between east and west would be enough to swap small numbers of alleles. This would hardly affect the divergent neutral genotypes at all, but newly introduced advantageous alleles would increase in frequency even if they were originally rare. For example, maybe all the northern ones have adaptations for cold temperatures and the southern ones are adapted to warmth. This pattern would not be reflected in the neutral markers.
Bingo! This is exactly what I've been alluding to,
neutral variation does not necessarily reflect adaptively significant variation
. Obviously this matters for conservation geneticists for deep philosophical reasons having to do with the raison detre of their field: the perpetuation of biodiversity. Now, one issue with conservation genetics is that the primary reason they are focused on introgression is that they are attempting to assay populational variation accurately, and so the loci you select as your proxies matters a great deal. The cases I have been hypothesizing in relation to human evolution are situations where neutral variation, or disjoint neutral allelic distributions, may mask important selective sweeps constrained only to a few important, but significant, loci. In other words, though neutral regions of the Neandertal genome suggest little affinity with modern H. sapiens sapiens, that does not necessarily exclude the possibility that on a few significant loci there maybe more more similarity between Neandertals and some populations of H. sapiens sapiens (e.g., skin color through MC1R for Europeans). But what about the inverse scenario, that is, where neutral variation suggests a uniform population but there are loci which suggest hidden population structure driven by selective forces? Obviously this is one issue which is of interest to conservation geneticists, as local ecotypes may have emerged recently due to powerful selection on only a few traits. In these situations assays of neutral variation might underestimate the extent of diversity, and this has had relevance for public policy and the application of legislation designed to conserve biodiversity and protect "species." But this issue of local ecotypes is not one that is restricted to animals only. It effects humans as well! This issue was brought to my attention by genetical anthropologist Henry Harpending several years ago. Harpending points out that both Basques and Bushmen do not seem to suggest genetic discontinuity from their neighbors. But the Bushmen do look very different from their Bantu neighbors, while the Basque are famous for their worldwide modal frequency of the Rh- allele. What's going on here? Again, selection maybe at work. Consider the case of Bushmen. Their mtDNA does not suggest that they are very different from their Bantu neighbors, but their physical features are very distinct. How can it be that a neutral marker implies ancestral commonality while genes which code for appearance lead to sharply different physiognomies? Consider this thought experiment: imagine that two tribes, one of Bushmen, another of recently arrived Bantu, decide to "swap females." I am assuming maximal patriarchy here, even if it is not realistic, for illustrative purposes. The offspring of Bantu males would now all be genetically half-Bushmen while the offspring of Bushmen males would be genetically half Bantu. In fact, if you looked at mtDNA the children in the tribe of Bantu adult males would be Bushmen and vice versa! Now, because of sampling variance in segregation & independent assortment of genes the offspring in both tribes will exhibit a range in their gross appearance (assuming there is some overlap between the Bushmen and Bantu, if not, we'll have to push back the emergence of variance in the F2 generation). Though the expectation, the average, child in both tribes would look about the same, half-Bantu and half-Bushmen, both of the populations' F1 hybrid generations would exhibit a wide range in physique, favoring either Bantu or Bushmen. Now, imagine that fathers favor a particular appearance, and specifically, they favor their offspring who resemble their own physical type. In China the Hui Muslim ethnicity is known for speaking the local Chinese dialect and practicing Islam. Their origin likely lay during the influx of Central Asian officials and soldiers who arrived with the Mongol conquests, but in any case, over the centuries they have admixed with the local substrate. Nevertheless, I have read in the literature that when anthropologists visit Hui villages they are sometimes presented with an individual who looks far more Central Asian than East Asian, and the villagers will declare that this is a "true Hui," as this individual represents their idealized self-image. In other words, despite the reality that the Hui are now predominantly Chinese genetically through admixture they still emphasize a physical difference with their Han neighbors which has been diminished by generations of intermarriage (predominantly of Han women marrying into the Hui). If one assumes a situation of maximal patriarchy then the founding Bantu and Bushmen males could perpetuate social myths and legends of their idealized ancestors, who exhibited the "pure" Bushmen or Bantu look. In such a manner social selection could result in the sifting of the range of appearances for a particular physical type, and sexual selection could subsequently "fix" this type. And yet no matter the reality of phenotypic differences between the two populations, ancestrally informative neutral markers should, overall, show that both are genetically rather similar across the full genome! Now, let us move to the Basques. First, you have to know a little about the problems with Rh incompatibility. The basics are this: a minority of European females are Rh-, which is a recessive phenotype, and these women have serious problems carrying Rh+ fetuses to term, increasing sharply after their first birth. Basques have the highest frequency of Rh- in the world. Because of their problems with pregancies during the pre-modern era other Spaniards looked askance at their sickliness, and there was even some talk of them being cursed by Satan. It seems likely that this Rh- frequency is a product of some sort of selective event, as a trait with such clear fitness implications can't just arise due to drift or mutation (see the equation for mutation-selection balance of a recessively expressed allele). But in any case, the Rh- status of many Basque females poses a problem for intermarriage with Rh+ peoples in the premodern era. Since so many Basque females are Rh-, the fitness of an Rh+ man would be lower vis-a-vis an Rh- negative male who married into the Basque community. Additionally, the fitness of a Rh+ homozygote would be lower than that of a Rh+ heterozygote (one copy of the functional allele). Nevertheless, we know intermarriage occurred, we see it in the genes. And yet the Rh- frequency of the Basques remains inordinately high. Could be the local selective force which originally allowed the rise in Rh- frequency? Perhaps. But here is another dynamic: consider the offspring of a "mixed marriage." If the outsider man was an Rh+ heterozygote, clearly fertilizations where the offspring was Rh- (because it received the loss of function copy) would be more fit from the get go, and much more likely to be carried to term. In the case of an Rh+ homozygote all of his fertilizations would result in Rh+ offspring, but, the first child would likely survive, and there is a non-trivial chance of subsequent survivals. But, the offspring themselves would marry within the Basque community, and again, the gauntlet of selection would favor fertilizations which resulted in Rh- offspring! But, this does not effect other loci, so in this way neutral alleles can leak between populations while selective weeds out Rh+. Even modest amounts of gene flow between populations tend to equilibrate neutral alleles, so you can have a Basque population which is ancestral "non-Basque," and yet still retain a stamp of the Rh- trait which has characterized their people for ages. What does this mean in the big picture for humanity? Well, it means that we should be cautious about extrapolating from neutral variation. The evidence of powerful selective forces on the human genome within the last 10,000 years suggest that an analogy to "ecotypes" might be warranted for our species. Though we are relatively closely related our phenotypic variation should be a clue as the power of selection.