Tishkoff and her colleagues analyzed 1,327 markers (848 microsatellites, 476 insertions/deletions, and 3 SNPs) of 2,432 Africans (113 African populations), 98 African Americans, and 21 Yemenites and these were compared to 952 CEPH-HGDP, 432 Indians, and 10 Native Australians in order to understand the genetic variation and population structure in Africa as well as world-wide. Tishkoff and her colleagues have been working on understanding genetic variation exists in Africa and world-wide for over a decade. Probably at the time of publication, they used the largest data set for multilocus study of human population structure and genetic variation.
Like many other genetic studies, they found the highest genetic diversity among the Africans and phylogenetic trees of the world-wide populations show that non-Africans are within sub-branch of African populations. In addition to use of phylogenetic tree and STRUCTURE, Tishkoff and her colleagues retain typological thinking and ask whether linguistic classification or geography explain the population structure in Africa.
However, the results from their analyses are quite different from Rosenberg et al. (2002; 2005) and Li et al. (2008), who argue that human populations are well structured. Tisikoff et al. found evidence of gene flow across ethnic, racial, or continental boundaries. First, Africans share alleles with the Middle Eastern populations more than other populations suggesting significant gene flow between the Middle Eastern and African populations (note. European and Middle Eastern populations are often classified as Caucasians in human population genetic research). Second, the major human geographic groups overlap rather than cluster on their principle component analysis of world-wide data set. Third, Africans have heterogeneous ancestry possibly due to constant gene flow.
Probably the most interesting finding is the results of STRUCTURE analyses showing African populations subdivided as much as all non-African populations combined. Tishkoff’s K=2 and K=3 are similar to other studies (Rosenberg et al., 2002, 2005; Li et al., 2008), except for at K=3 Indians have significant admixture between East Asians and Caucasians. East Africans, Khoisans, and Pygmies, each form a distinct cluster, before non-Africans form other clusters. As K become larger, STRUCTURE show more complicated pattern of clustering as well as high level of admixture.
As Weiss and Long (2009) suggest, this study shows that the results and interpretations of STRUCTURE and world-wide population structure are strongly influenced by the population samples included. If more intermediate populations, such as the Middle Eastern and Indian populations, are included, cline rather than clustering explains the world-wide human population structure and genetic differences among human racial groups. Also, the major human geographic groups, or race, such as Africans, Caucasians, and Asians, are not homogenous groups and these groups are further subdivided in a very complex way.