Dienekes and Maju have both commented on a new paper which looked at the likelihood of lactase persistence in Neolithic remains from Spain, but I thought I would comment on it as well. The paper is: Low prevalence of lactase persistence in Neolithic South-West Europe. The location is on the fringes of the modern Basque country, while the time frame is ~3000 BC. Table 3 shows the major result:
Lactase persistence is a dominant trait. That means any individual with at least one copy of the T allele is persistent. As Maju noted a peculiarity here is that the genotypes are not in Hardy-Weinberg Equilibrium. Specifically, there are an excess of homozygotes. Using the SJAPL location as a potentially random mating scenario you should expect ~7 T/C genotypes, not 2. Interestingly the persistent individual in the Longar location also a homozygote.
HWE makes a few assumptions. For example, no selection, migration, mutation, or assortative mating. Deviation from HWE is suggestive of one of these dynamics. The sample size here is small, but the deviation is not to be dismissed. Recall that lactase persistence has dominant inheritance patterns. If the trait was being positively selected for you would only need one copy. The enrichment of homozygotes is unexpected if selection in situ is occurring here. It can not be ruled out that one is observing the admixture of two distinct populations. One generation of random mating would generate HWE, but when populations hybridize in realistic scenarios this is not always a plausible assumption. Rather, assortative mating often persists over the generations, slowing down the diminishing of population substructure.
Stepping back from speculation in this case what can we say? First, the LCT locus has a large mutational target. The trait of lactase persistence has arisen multiple times via different mutational events across the Old World. But, there does seem to be one particular variant which is found from Spain to Northern India. There is some circumstantial evidence that the allele had its origin somewhere in Central Eurasia, but currently its modal frequency is in Northern Europe, Scandinavia and Germany. The region in the genome around this mutation is characterized by a very long haplotype. It is one of the most definitive loci as a candidate for natural selection in the human genome. There is now a fair amount of ancient DNA evidence that lactase persistence in Europe is a feature of the last ~5,000 years or so. Among the modern Basques the frequency of the allele is 66 percent.
For me the key issue is teasing apart the role of migration and selection in each specific case. It does not seem to be correct that the frequency of the -13910T LCT allele in Basques and Punjabis is reflective of the frequency of recent common ancestry. That implies that natural selection is at work at this locus. On the other hand, the haplotype which is present in both the Basque and Punjabis is likely to be descended from a common set of individuals, implying that there is a genealogical chain connecting these two very distinct and distant Eurasian populations. Therefore, we can potentially make some inferences about the power of migration in spreading distinctive alleles. Often we partition selection from genealogical information, because selection so often serves to distort the signal. But the genealogical patterns may lay at the heart of the distribution of different natural selective events at the LCT locus.
Overall, I would say that the results from ancient DNA are disordering and clouding simple elegant models. One hopes and presumes that as sample sizes increase in this domain we'll start to see more clarity as new paradigms crystallize.
Citation: European Journal of Human Genetics, 10.1038/ejhg.2011.254