On the other hand, the self-fulfilling one-drop rule in America may have resulted in a large distinct black population in the U.S and their virtual disappearance in Mexico. But on the third hand, Brazil also has a latin-type continuous “color bar” and a large distinct black population in the northeast.

“Outside West Africa, sporadic reports of HIV-2 infection have been previously made in Portugal, Mozambique, Angola, southwestern India, and Brazil; these events are all related to the former ties Portugal had with West Africa.[21] Portugal itself appears to have low but stable rates of HIV‑2 prevalence in the population.[14] The other countries once shared common historical-political ties, with some economic trade relationships existing even today. As noted, HIV‑2 has been detected in some large cities in southwestern India, [22,23] perhaps because of exchange with the former Portuguese colonies of Africa. Goa, a former Portuguese colony, situated south of Bombay on the western coast, has reported 4.9% HIV-2 and 9.8% HIV-1 infection rates in patients with sexually transmitted infections (STIs).[22] To date, significant HIV-2 infection has not been reported in other parts of Asia. [My emphasis]”

Biology and Variation in HIV-2 and HIV-1 Kanki, PJ., et al.

See also, High proportion of HIV-2 and HIV-1/2 double-reactive sera in two Indian states, Maharashtra and Goa: first appearance of an HIV-2 epidemic along with an HIV-1 epidemic outside of Africa.
Rübsamen-Waigmann H., et al.

This is not relevant to the data in this particular post. I’m just parking this suggestion here as something to think about.

http://dienekes.blogspot.com/2011/04/comparing-five-methods-of-admixture.html

I wouldn’t bet the farm on ROLLOFF being correct, and certainly it’s a major flaw in the paper that introduced it that its discrepancy with HAPMIX (by the same group) is never explained.