By contrast, in various autosomal DNA studies, the northern Nilotes have appeared to be almost completely of African ancestry. For example, in the African Genome Variation Project’s analysis, the Dinka sample showed no extraneous influence at K=2 (cf.
Gurdasani et al. (2015)).
This apparent lack of non-African affinities is why the Dinka, Nuer and other northern Nilotes have often been used by researchers as proxies for the African component. In 2017,
Skoglund et al. compared the same AGVP Dinka sample to that of an ancient South Cushitic pastoralist (Luxmanda), the first such specimen to be genetically analyzed. In their admixture analysis, the Dinka individuals now all of a sudden showed almost 30% non-African ancestry at the K=2 level. Judging by the existing uniparental marker data, it’s pretty clear why that is: there was non-African ancestry buried within the northern Nilote gene pool, and that ancestry was specifically derived from earlier Cushitic peoples such as Luxmanda.This, in a nutshell, explains why the Dinka, Nuer and other northern Nilotes often fit well as proxies for the African component in admixture testing models. However, the problem with using such admixed groups in these analyses is that doing so leads to highly distorted estimates of ancestry proportions. For example, let’s say a researcher is testing admixture models on qpAdm and finds that his modern Cushitic or Ethiosemitic-speaking population is best modeled as 50% Neolithic Levantine + 50% Dinka. What that program is really telling him is that his sample is best modeled as 50% Neolithic Levantine + X% ancient Nilotic + X% ancient Cushitic.
This again stems from the fact that the Dinka are not a purely African population, but rather a Nilote-descended community with significant Cushitic admixture (whence was derived their non-African ancestry). When factoring in the ~30% of non-African ancestry that Skoglund et al. (2017) detected in their Dinka reference sample, the estimated whole genome ancestry apportionment then actually becomes 50% Neolithic Levantine + 35% African + 15% unclassified Eurasian. This necessary adjustment therefore brings the non-African total to around 65%. The same corrective adjustment would have to be made if the African reference sample were a southern Nilotic group, such as the Maasai Nilotes (e.g. when the ~30% of West Eurasian admixture in
Ali et al. (2020)‘s East African proxy sample (Maasai) is taken into account, the estimated non-African ancestry for their northeastern Somali sample rises to 70%; this more accurate total is close to Hodgson et al.’s 66% average for their general ethnic Somali sample).
That’s also before correcting for linkage disequilibrium bias, which, as Hodgson et al. (2014) observed and
Choudhury et al. (2020) also demonstrated, would increase the Cushitic speakers’ Eurasian total even further. Put simply, there is considerable extra non-African ancestry in the genome of the Afro-Asiatic-speaking sample which is not being counted. This results in an inaccurate overall estimation of ancestry proportions. It also conflicts with scientific data gathered through other means (viz. craniometric analysis, anthropometric analysis, linguistics).
The maternal haplogroups distributed today among the northern Cushitic and Ethiosemitic-speaking populations in the Horn of Africa do not support the claim that there was significant admixture between their Afro-Asiatic-speaking male ancestors and Nilotic females. This idea first gained prominence after the publication of
Watson et al. (1996) and
Watson et al. (1997), two early mtDNA studies which reported that the analyzed Somali (sampled in Kenya) and Tuareg Berber (sampled in Niger) individuals primarily belonged to the L3 mtDNA haplogroup.
Comas et al. (1999) similarly indicated that only 5%-27% of Ethiopians bore non-African mtDNA lineages. However, it turns out that these results were inaccurate since the L3 macroclade’s phylogeny had not yet been fully resolved. Specifically, many of the samples that were at the time presumed to represent L3 carriers actually belonged to derivatives of the N haplogroup.
Martin Richards, who served as a co-author on Watson et al. (1997), explains the situation thusly:
source:
@Garaad Awal
What do you think of this?