Shai Carmi has his article out on Ashkenazi Jewish ancestry. A few comments:
First, looks like a good job, on the whole. Perhaps Carmi had special training…
Second, about dates: they assume a mutation rate of 1.44 x 10-8 per generation, and a generation length of 25 years. I think both of those are a little off. All the directly-measured whole-genome rates are between 0.96×10−8 and 1.20×10−8: I wouldn’t go higher than 1.20 x 10-8 per generation, with what I know now. In the course of looking at paternal age effects, I also checked out the known data on average generation lengths. In no known population is it as short as 25 years: never less than 28 for females, and almost always longer than 28 for males, usually in the 30s. 30 is a much more reasonable generation length than 25.
The date estimates are inversely related to the assumed mutation rate and directly proportional to the generation length.
In the article, they estimate an Ashkenazi bottleneck with an effective size of 250-420, 25-32 generations ago. With a mutation rate of 1.2 x 10-8 per generation, that changes to an effective size of 300-500 (not much different) 30-38.4 generations ago.
By their numbers, we’re talking 625-800 years ago. Adjusted, 900-1150. Since we know for sure that at least some Ashkenazi Jews were in the Rhineland before the year 1000, I think that the revised estimate indicates that the original settlement was the bottleneck, which makes sense.
Third, about comparison populations: they used Flemings as a sample European population, and estimated the Ashkenazi Jews are 46-50% European. Why Flemings? All the previous analyses have suggested that their European component is from southern Europe. The mtDNA analysis is pretty specific: Italy and France. I think you’d get a better estimate using Italians, and it would probably yield a slightly higher estimate of European admixture.