You can't really trust these estimates. The age of both the mtDNA and Y-DNA root has been constantly pushed back in time over the years. In other words the age estimates have always been underestimated, and I am pretty sure they still are. Just compare the estimates with the actual ancient DNA data. How can HV be only 19,000 years old when the 30,000-year-old sample from the Paglicci Cave was already HV ? Besides chances are that HV did not originated in southern Italy, but in the Middle East, so that would mean HV is at least a few millennia older than 30,000 years.
There are many other examples. I cited the H1b sample from Mesolithic Portugal (c. 8,500 years old), which doesn't agree with the age of 6,200 years given by Behar. Likewise, there is an H6 from Paleolithic Cantabria (12,000 years old), but Behar gives it an age of 11,000 years old. That is close enough, but only if H6 actually originated in Iberia. That's very unlikely considering that H6 is most common in Eastern Europe.
I may have confused matters by giving a wrong date for the HV, and for not providing the SD figures for the rest of the clades. I apologize for that. Here are the correct dates:
HV: 21,905.8, +/- 2,832.7
H: 12,846, +/- 773.4
H13: 12,475, +/- 867.7
H2: 11, 905.3, +/- 1364.4
H4: 10,617.3, +/- 1,471.3
H1: 9,888.6, +/- 880.6
H5: 9877.6, +/- 1,401.7
H2a2 (CRS): 9126.3, +/- 1,873.7
H3: 8919, +/- 1062.6
You mentioned H1b. These are his figures for that. I suppose if you stretch the SD to the very limits of the older range, it just might fit the Iberian sample, but that is indeed *stretching it*.
H1b: 6,237, +/- 1,805.3
These are the figures for H6, which comes out to about 12,000 years ago if we use the middle range of the SD:
10,945.6, +/- 1873.7
I have a date of 24,000 B.P. for Paglicci 25, and with the correct Behar date of about 22,000 to 23,000 for HV taking into account the SD it looks like a pretty good fit to me. Of course, the typing of the Paglicci sample also says HV or RO, and RO is certainly old enough.
All the dates are in the supplementary table in Behar et al,
http://www.sciencedirect.com/science/article/pii/S0002929712001462
Just generally, I am always a little skeptical of the mtDNA classifications done by academics in the past, and sadly even in the present occasionally, based solely on HVRI data, and when we're talking about ancient DNA, fragmentary sequences at that. That's why I tend to give more credence perhaps to the trees and the dating based on the sequencing of the entire mtDNA, although they certainly do change these dating schemes, and so, I am definitely keeping an open mind about all of this..
FWIW, I checked the Behar dates against those in this paper on mtDNA N, I, X and W. They used completely sequenced mtDNA. When taking into account the SD figures, the Behar figures are a good match for the ones in this paper. The authors provide some great distribution maps too.
The Arabian Cradle: Mitochondrial Relicts of the First Steps along the Southern Route out of Africa, Fernandes et al
http://www.sciencedirect.com/science/article/pii/S0002929711005453