https://www.biorxiv.org/content/bior...1/423079-1.pdf
"The analysis using the All set showed that Levantine (Natufians and PPNB) and North African (Taforalt and Morocco_EN) populations could be modeled as a mixture of Dzudzuana with extra Basal Eurasian ancestry.
The study of the Ibero-Maurusian remains from Taforalt was initially interpreted as suggesting that this population was formed by admixture between Natufians and a Sub-Saharan population15. However, the admixture graph model suggests the opposite scenario: that Natufians were formed by admixture from a Taforalt-related population and a Dzudzuana-related one"
(...)
"Taforalt could not be modeled as any 2-way mixture. The best model involving Natufians and an African population (Yoruba) could still be strongly rejected (p=2.7e-13). Taforalt could also not be modeled as a 3-way mixture.
However, Natufians could be convincingly modeled as a 2-way mixture of ~86% Dzudzuana and ~14% Taforalt (p=0.405) with small standard errors of 1.9%. Thus the affinity between Natufians and Taforalt described in ref.15 may have come about by admixture from a North African/Taforalt-related population into Natufians, rather than by admixture in the opposite direction. The results of our analysis using the All set, as well as the results of the analysis of ref.15 do suggest that Taforalt can be modeled as a mixture of a West Eurasian related population (represented by Dzudzuana in our case) and a Sub-Saharan African lineage. However, when one uses only a single African population as a source without using others as outgroups, this mixture can only be interpreted as evidence of ancestry from a lineage basal to members of the All set, rather than as evidence of ancestry specifically specifically 44 related to the chosen African population.
No Sub-Saharan African populations appear to be good sources for the ancestry of Taforalt as described previously. The admixture graph model suggests an alternative possibility: that it is West African populations like the Yoruba that may have ancestry from a North African Taforalt-like population. Under such a scenario, North Africa and the Levant were occupied by populations that experienced gene flow from each other, with more ancestry from a Basal lineage in North Africa, and more ancestry from a West Eurasian specific lineage (represented by Dzudzuana) in the Levant, thus explaining the presence of Dzudzuana related admixture in Taforalt and of Taforalt-related admixture in the Levant. Under this scenario, a North African-related population may have contributed some ancestry to Sub-Saharan populations to its south, perhaps during the Holocene Green Sahara period (~11-6kya)22 that postdates the sampled Taforalt individual which may have facilitated north→south gene flow across the Sahara. Based on the very low presence of Neandertal admixture in Yoruba, it has been estimated that >2.7±0.9% of the ancestry of Yoruba came from West Eurasia 9618 ± 1825 years ago 23 .
The admixture graph model predicts that 13% of the ancestry of Yoruba came from Taforalt, which in turn was 55% descended from Dzudzuana and which in turn was 72% descended from Villabruna, for a total of 0.13*0.55*0.72≈5% Villabruna-related ancestry that would have carried Neanderthal DNA. This is consistent with the >2.7±0.9% estimate of ref. 2"
(...)