I will be using the following equation to estimate some times in human prehistory:
![]()
Time t can be estimated if one has an estimate of the effective population size Ne, and of Fst between a pair of ancestral populations. We also need some estimate of the generation length to convert times into years.
I take the long-term human generation length to be 23 years (Hawks et al. 2000)
I will use the Fst values recently inferred by myself using ADMIXTURE on a large dataset of Old World variation; I note that this dataset did not include African hunter gatherers.
![]()
I use the estimates of Ne by Li et al. My procedure is as follows: I note which population is modal for each of the 12 ancestral components, and note its Ne. I take the average of the two populations for which there is an Fst value.
For example, the Atlantic_Med component is modal in the HGDP French_Basque population (Ne=6,137), and the Sub_Saharan in the HGDP Yoruba population (Ne=9,960). The average Ne=8,049. The Fst betwen these two components is 0.185. Hence, the time of divergence is:
23*log(1-0.185)/log(1-1/(2*8049)) = 75.7ky
Below are the pairwise divergence times of the maximally differentiated components: Atlantic_Med, Southeast_Asian, Sub_Saharan; within the HGDP, these are modal in French_Basque, Dai, and Yoruba
Atlantic_Med vs. Sub_Saharan = 75.7ky
Atlantic_Med vs. Southeast_Asian = 43.0ky
Sub_Saharan vs. Southeast_Asian = 86.4ky
These dates appear plausible. An important calibration point is the appearance of the first modern humans in Europe which has been archaeologically placed at precisely the time given above for the Atlantic_Med vs. Southeast_Asian divergence.
We can probably assume that the differentiation between Eurasians and Sub-Saharan Africans is best estimated by the Sub_Saharan vs. Southeast_Asian comparison, since Southeast_Asians would have extremely limited opportunity to experience subsequent gene flow with Africans in either direction. The time of 86.4ky is older than the 70ky estimated for mtDNA haplogroup L3 which almost certainly marks the major human population expansion. The difference can be probably attributed to the absorption of different populations in Africa and Asia (archaic humans or pre-L3 expansion modern humans).
Of particular interest to me is an evaluation of my womb of nations theory, with respect to the origin of West Eurasians. The Caucasus component exhibits Fst distances between 0.033 and 0.052 with the components I have labeled "the Six". Within the HGDP, the following populations are modal for them:
Caucasus: Adygei; Atlantic_Med: French_Basque; North_European: Orcadian; Southwest_Asian: Bedouin; Northwest_African: Mozabite; Gedrosia: Brahui
Divergence times are:
Caucasus vs. Atlantic_Med = 9.9ky
Caucasus vs. North_European = 11.1ky
Caucasus vs. Southwest_Asian = 16.8ky
Caucasus vs. Northwest_African = 16.2ky
Caucasus vs. Gedrosia = 13.6ky
There is reason to think that the last 3 ages are inflated: the Northwest_Asian and Gedrosia component appear divergent in a Sub_Saharan and South_Asian direction respectively, so they probably represent stabilized mixtures of a West Eurasian with a substrate element. Moreover, the effective population size used for the Bedouin is quite higher than for any other West Eurasian population, presumably because of the African admixture in that population; that high effective size is, hence, not reflective of the effective size of the Southwest_Asian ancestral component.
All in all, these results are consistent with a mainly Neolithic origin of West Eurasian populations, and incompatible with their pre-LGM differentiation. As I have argued in the womb of nations, the expanding Neolithic population probably absorbed -to a limited extent- aborigines from outside the core area of the Near East, however, the bulk of the population appears to share a fairly recent common ancestry.![]()
![]()
Time t can be estimated if one has an estimate of the effective population size Ne, and of Fst between a pair of ancestral populations. We also need some estimate of the generation length to convert times into years.
I take the long-term human generation length to be 23 years (Hawks et al. 2000)
I will use the Fst values recently inferred by myself using ADMIXTURE on a large dataset of Old World variation; I note that this dataset did not include African hunter gatherers.
I use the estimates of Ne by Li et al. My procedure is as follows: I note which population is modal for each of the 12 ancestral components, and note its Ne. I take the average of the two populations for which there is an Fst value.
For example, the Atlantic_Med component is modal in the HGDP French_Basque population (Ne=6,137), and the Sub_Saharan in the HGDP Yoruba population (Ne=9,960). The average Ne=8,049. The Fst betwen these two components is 0.185. Hence, the time of divergence is:
23*log(1-0.185)/log(1-1/(2*8049)) = 75.7ky
Below are the pairwise divergence times of the maximally differentiated components: Atlantic_Med, Southeast_Asian, Sub_Saharan; within the HGDP, these are modal in French_Basque, Dai, and Yoruba
Atlantic_Med vs. Sub_Saharan = 75.7ky
Atlantic_Med vs. Southeast_Asian = 43.0ky
Sub_Saharan vs. Southeast_Asian = 86.4ky
These dates appear plausible. An important calibration point is the appearance of the first modern humans in Europe which has been archaeologically placed at precisely the time given above for the Atlantic_Med vs. Southeast_Asian divergence.
We can probably assume that the differentiation between Eurasians and Sub-Saharan Africans is best estimated by the Sub_Saharan vs. Southeast_Asian comparison, since Southeast_Asians would have extremely limited opportunity to experience subsequent gene flow with Africans in either direction. The time of 86.4ky is older than the 70ky estimated for mtDNA haplogroup L3 which almost certainly marks the major human population expansion. The difference can be probably attributed to the absorption of different populations in Africa and Asia (archaic humans or pre-L3 expansion modern humans).
Of particular interest to me is an evaluation of my womb of nations theory, with respect to the origin of West Eurasians. The Caucasus component exhibits Fst distances between 0.033 and 0.052 with the components I have labeled "the Six". Within the HGDP, the following populations are modal for them:
Caucasus: Adygei; Atlantic_Med: French_Basque; North_European: Orcadian; Southwest_Asian: Bedouin; Northwest_African: Mozabite; Gedrosia: Brahui
Divergence times are:
Caucasus vs. Atlantic_Med = 9.9ky
Caucasus vs. North_European = 11.1ky
Caucasus vs. Southwest_Asian = 16.8ky
Caucasus vs. Northwest_African = 16.2ky
Caucasus vs. Gedrosia = 13.6ky
There is reason to think that the last 3 ages are inflated: the Northwest_Asian and Gedrosia component appear divergent in a Sub_Saharan and South_Asian direction respectively, so they probably represent stabilized mixtures of a West Eurasian with a substrate element. Moreover, the effective population size used for the Bedouin is quite higher than for any other West Eurasian population, presumably because of the African admixture in that population; that high effective size is, hence, not reflective of the effective size of the Southwest_Asian ancestral component.
All in all, these results are consistent with a mainly Neolithic origin of West Eurasian populations, and incompatible with their pre-LGM differentiation. As I have argued in the womb of nations, the expanding Neolithic population probably absorbed -to a limited extent- aborigines from outside the core area of the Near East, however, the bulk of the population appears to share a fairly recent common ancestry.
