There is a new paper discussing in some good depth with the peculiarities that the irregular mutation patterns of mtDNA, particularly in the macro-haplogroup R, show and its implications and complications for the idea of a molecular clock that can estimate the age of the various haplogroups, so dear of some and so much hated by others.
While I do not necessarily agree with what the authors conclude in this paper, I do applaud their critical approach in general and I do recommend the (mostly free access) bibliography for those interested in digging deeper in the matter.
Denis Pierron et al., Mutation Rate Switch inside Eurasian Mitochondrial Haplogroups: Impact of Selection and Consequences for Dating Settlement in Europe. PLoS ONE. Open access.
Probably figure 3 illustrates quite well the problem:
As you can see the actual number of mutations found in each of the sublineages of R varies a lot! Some sublineages have accumulate as many as 16 mutations, while others barely have four. Also the excess or defect of mutations follows some obvious patterns along haplogroups.
The authors suggest that there are two issues: in the case of J1 (only), they find that there must be a selective constrain of some sort that blocks further neutral evolution. But this does not apply to the rest of JT nor to the big problem child: R0 (notably HV and specially H under it).
They conclude that there must be some other circumstance such as the lack of mutations for some lengthy period at each lineage.
I must say here that I found this argument faulty because the problem is not, I understand, absolute lack of novel mutations but lack of effective mutations (i.e. those that survive and forge new lineages). I understand that this was surely caused because the corresponding haplogroup was already solidly established and therefore novel mutations had no room to fructify in most cases, being reabsorbed by the dominant ones in a totally normal drift process (where the most common lineages almost invariably succeed).
We can say this is the cannibal mum model... though nobody had to actually eat anyone in reality, just "daughter" lineages with novel mutations were systematically drifted out in most cases.
Instead where populations were very low, all lineages, novel or ancestral had similar chances of survival, so the effective mutation rate was increased instead.
I reached to this conclusion because I noticed that it is actually the haplogroups with large star-like structures, notably M and H, which suffer from this symptom most intensely. As star-like phylogenies are clear indicators of sudden expansions, I concluded that it was the success of mum what aborted that of the daughters, delaying and even nearly stopping the process of accumulation of new mutations.
That is why, when doing molecular clock exercises myself, I count mutations from the root and not present day haplotypes. This last makes sense only when the number of mutations is so huge and common in all generations that every newborn has some novel mutations inside. This is true for nuclear and Y chromosome DNA but not mtDNA, which has such a small genetic chain that each mutation probably only happened every many dozen generations.
It is easy to understand, I believe, that, with so rare mutation events, the novel mutation lineage (not the carrier!) had in most cases very very low chances of survival, unless the population was so tiny that it was one among a handful and not one among hundreds or even thousands.
Back to the paper
I am not sure at the moment on what Pan-Homo divergence estimate they have used (this is one of my greatest criticisms to the usual molecular clock guesstimates and does not seem to be clarified in the paper) but, regardless, I am favorably surprised by the age estimates they have been able to calculate.
Naturally (my method is too different) I am not really in agreement but at least they have come with age estimates with some plausibility. They are all in table 5 but here there are some examples:
- R2'JT - 53 Ka
- J2 - 28 Ka
- T - 27 Ka
- R0 - 41 Ka
- V - 17 Ka
- H - 28 Ka
- H1 18 Ka
- H3 17 Ka
- U - 44 Ka
- U4 - 25 Ka
- U5 - 20 Ka
- U6 - 25 Ka
- B - 43 Ka
I still think that these dates are too recent in most cases and the reason is probably that they are still counting the age estimates, in spite of all corrections, from present backwards and not from the root to the relevant node.
Of course my method requires some other point(s) of calibration (instead of present), something like an archaeological event (for example equating the colonization of Europe c. 40 Ka with the H star-like node) and that is a point of controversy on its own...
More stuff to read
As I said at the beginning one of the virtues of this paper is that it has an extensive free access bibliography on the issue of why mtDNA molecular clock is problematic. I have selected the following (not all of which I have read yet):
- A. Torroni et al., A Signal, from Human mtDNA, of Postglacial Recolonization in Europe. AJHG 2001. (link)
- Neil Howell et al., African Haplogroup L mtDNA Sequences Show Violations of Clock-like Evolution. MBE 2004. (link)
- Neil Howell et al., Relative Rates of Evolution in the Coding and Control Regions of African mtDNAs. MBE 2007. (link)
- H-J Bandelt, Clock debate: when times are a-changin': Time dependency of molecular rate estimates: tempest in a teacup. Heredity 2007. (link)
- Brenna M. Henn et al., Characterizing the Time Dependency of Human Mitochondrial DNA Mutation Rate Estimates. MBE 2008. (link)
- N. Howell et al., Molecular clock debate: Time dependency of molecular rate estimates for mtDNA: this is not the time for wishful thinking. Heredity 2008. (link)