I agree about the current paper. (The entire Mallick et al paper can be found here:
http://www.plosgenetics.org/article....1371/journal.pgen.1003912&representation=PDF)
It is really talking about the emergence, or perhaps more accurately, the coalescence of this mutation, and not when it expanded. Even then, their confidence intervals for that coalescence are huge.
"We estimated the coalescence time of the rs1426654 mutation at 28,100 years (95% CI - 4,900 to 58,400 years) using BEAST.Using the same mutation rate, the coalescent age estimated by rho statistics was 21,702 years 6-10,282 years.
For an understanding of the actual selective sweeps involved, I find this paper more informative.
http://mbe.oxfordjournals.org/content/early/2012/08/25/molbev.mss207.full.pdf+html
The Timing of Skin Pigmentation Lightening in Europeans, Beleza et al
This is one of their conclusions, based on the KITLG gene.
I find this interesting in light of the fact that the Mal'ta boy is dated to 24, 000 years ago, and so maybe from a period before there was complete divergence between West Eurasian, East Eurasian, and perhaps indeed Southeast Eurasian groups, which would explain his admixture results.
"the initial stages of European skin lightening occurred in a proto Eurasian population, about 30,000 years ago, after the out of Africa migration ~60,000 70,000 years ago and slightly more recently than the earliest archaeological evidences for the dispersal of anatomically modern humans in Europe, around 40,000 years ago Recent estimates based on genome wide patterns of variation have suggested that the European and East Asian divergence might have occurred as late as ~25,000 years ago.
This is where they discuss the timing of the selective sweep.
"Our estimates additionally show that the onset of selective sweeps at SLC24A5, SLC45A2,and TYRP1, the three genes in which the geographic distribution of the polymorphisms is primarily restricted to European populations, were much more recent than at KITLG
, and remarkably compressed within the last 11,000-19,000 years (Table 3)."
Based on these dates, they place the time of the sweeps into Europe during the Magdalenian, and posit that it occurred both because of reduced sunlight during the LGM, leading to high risks of Vitamin D deficiency, and the fact that they see a large increase in population during that period, and that would have made these mutations more available in the population.
I'm not sure I agree with that conclusion from the data. Snow and ice conditions would not necessarily decrease sunlight, or at least that's how I understand it, and these H/G's would still have been consuming a high fish diet, which would presumably have somewhat mediated their situation in terms of access to Vitamin D. I also still lean toward the view that the Neolithic technologies produced the large increase in population, instead of resulting from it...
Also, the dates they provide in Table 3 seem to me to support the Neolithic era as the most likely time for the sweeps to have begun, in particular because of the added Vitamin D deficiency stress caused by a majority cereals diet. The data in Table 3 shows that they
estimated that the selective sweep at
SLC24A5 occurred around 11.3 KYA (95% CI, 1–55.8 KYA) and 18.7 KYA (5.8–38.3 KYA) under additive and dominant models, respectively
[42]. With those kinds of dates and confidence intervals, it seems more than possible to me that the sweep took place during the Neolithic, funnelling out from the northern Near East and into Europe.
At any rate, the new Mallick et al paper which Dienekes posted does not necessarily see a contradiction between its own results and these slightly older papers. As they say:
"Our Bayesian coalescent age estimate of the rs1426654-A allele at ~28 KYA (95% HPD, 5–58 KYA), as well as the rho-based estimate at 21.7 (±10.3) KYA, are older in their point estimates than both of the above selective sweep date estimates, although these age estimates have broad and overlapping error margins. This finding is not surprising because sweeps can also operate on standing variation. "
Therefore, they conclude that:
"It appears that the most plausible scenario is that light skin evolved as an adaptation to local environmental conditions as humans started moving to northerly latitudes, with the initial phase of skin lightening occurring in proto Eurasian populations, while genetic variation in
SLC24A5 formed the later phase which led to lighter skin in Europeans and South Asians, but not East Asians. This was followed by a European-specific selective sweep, which favored the rapid spread of this mutation in these populations. Our coalescence age estimates of 28 KYA (95% HPD 5–58 KYA) show wide margins, also evident in the earlier sweep date estimates for the gene
[42]. This can be due to the fact that the power of our analysis was limited by the need to reduce our sequence range to a subset of sites from a region with sufficiently high LD around the rs1426654-A allele and very low level of sequence variation. Therefore, we speculate that narrowing down the coalescence age estimates and specifying the geographic source of the rs1426654-A allele will depend rather on the success of ancient DNA studies than on more extensive sequencing."
