Human population size over time (Theunert et al. 2012)

The authors of the current paper estimate a 13,601 ancestral and 22,915 current size for Yoruba, with an expansion beginning at 20.15ky. For Europeans, a more complex model provides a better fit, with an ancestral size of 10,065 going through a bottleneck to 3,300 at 39.5ky and recovering at 32.5ky to a current size of 18,300.

I note that the 2.5x10^-8 mutation rate was used to derive these estimates. The model used in this paper infers population size changes on the basis of both allele frequency and linkage disequilibrium, so it is not clear what the use of the slower directly measured mutation rate which led to the re-dating of the human-chimp divergence will have.

Since the direct rate is about half of the widely used 2.5x10^-8, it might be the case that the European bottleneck will approach the ~70ky turning point which is associated with the Toba eruption and the drying up of the Sahara-Arabia belt. That sounds like a most opportune time for a population living in Eurasia to go through a 3-fold reduction in size! Moreover, the Yoruba expansion may be re-dated to around ~40ky, which would correspond to the Middle to Lower Stone Age transition in Africa. Again, an excellent time for a population whose size was pretty constant to experience growth!

Indeed, according to a recent paper on climate history, the middle regions of Africa were more shielded from climate change than North Africa/Levant, so less extreme variations in size in tropical Africa vs. a drastic bottleneck in North Africa/Levant where the ancestors of Europeans were at the time, makes excellent sense to me.

In another recent paper that used a 2.35x10^-8 mutation rate (Lukic et al. 2012), the founding of the Eurasian population was estimated at 52ky. But, if the direct rate is used, this estimate will probably coincide with the crucial ~100ky mark that coincides with the first appearance of anatomically modern human in the Levant at Skhul/Qafzeh, associated with a Mousterian lithic technology, as well as the appearance of the Nubian Complex in South Arabia. So, there is no need to speculate about the early Out-of-Africa-that-failed with respect to these processes and postulate unwarranted coastal migrations.

A new calibration of the mtDNA phylogeny is probably also needed. Time-dependent evolution in mtDNA when the new human-chimp calibration is used will result in a new saturation curve, where mtDNA evolution will occur at recent times with something close to the genealogical rate, and at long (pre-AMH) intervals at the long-term evolutionary rate conditioned on the Pan-Homo divergence. Someone ought to "do the dates" again, but I don't think it's inconceivable that the L3 clade corresponding to the Out-of-Africa event will be updated from 70ky to something like100ky as well, with the origins of the Eurasian M and N lineages approaching 70ky themselves. So, the puzzle of L3 may in fact be resolved: it may be of ~100ky African origin, with later Eurasian clades N and M appearing at the same time as the ~70ky bottleneck.

It seems to me that the directly measured rate of 1-1.3x10^-8 is a better fit than the widely used one of 2.5x10^-8. Many of the papers currently appearing still use the old rate, so their conclusions need to be updated. 


It seems that the major Out-of-Africa event happened pre-100ky, the Eurasian bottleneck occurred at ~70ky and coincided with well-known adverse geo-climatic developments; some of the survivors went north Out-of-Arabia out this time, started interbreeding with Neandertals, and finally the MP/UP transition occurred post-50ka, probably first in the Levant, and followed very quickly in Russia, Central Europe, and the Mediterranean.

Mol Biol Evol (2012) doi: 10.1093/molbev/mss175

Inferring the History of Population Size Change from Genome-Wide SNP Data

Christoph Theunert, Kun Tang, Michael Lachmann, Sile Hu and Mark Stoneking

Dense, genome-wide single-nucleotide polymorphism (SNP) data can be used to reconstruct the demographic history of human populations. However, demographic inferences from such data are complicated by recombination and ascertainment bias. We introduce two new statistics, allele frequency-identity by descent (AF-IBD) and allele frequency-identity by state (AF-IBS), that make use of linkage disequilibrium information and show defined relationships to the time of coalescence. These statistics, when conditioned on the derived allele frequency, are able to infer complex population size changes. Moreover, the AF-IBS statistic, which is based on genome-wide SNP data, is robust to varying ascertainment conditions. We constructed an efficient approximate Bayesian computation (ABC) pipeline based on AF-IBD and AF-IBS that can accurately estimate demographic parameters, even for fairly complex models. Finally, we applied this ABC approach to genome-wide SNP data and inferred the demographic histories of two human populations, Yoruba and French. Our results suggest a rather stable ancestral population size with a mild recent expansion for Yoruba, whereas the French seemingly experienced a long-lasting severe bottleneck followed by a drastic population growth. This approach should prove useful for new insights into populations, especially those with complex demographic histories.

Link