Although previous studies have considered the role of admixture in shaping the genetic diversity among southern Africans, in particular the Coloured, no study has assessed (i) the significance of these contributions, (ii) how this admixture has shaped or contributed to distinct population subgroups among southern Africans, or (iii) the possibility that southern Africans may be harboring ancient vestiges of a ‘lost’ or understudied source of genetic diversity. The extent of admixture within people today defined broadly as Khoesan complicates these analyses, further compounded by subject heterogeneity. We attempt to assess sources of admixture and heterogeneity and ultimately identify and characterize a Khoesan-representative population that displays little to no significant non-Khoesan ancestral contribution. Such a population we identify as the Ju/'hoan.
This study suggests that the Ju/'hoan form a unique ancestral population for the human lineage, distinct (i.e., most dissimilar) from all contemporary populations for which data is currently available, including other forager populations. Gender-specific analysis confirms genetic isolation of the Ju/'hoan from non-Khoesan populations, while autosomal analysis shows no significant non-Ju/'hoan ancestral contribution. While the rest of the world was driven into agriculture at the end of the Last Glacial Maximum 
, the Ju/'hoan appear to have maintained their hunter-gatherer based subsistence. Significant agricultural-driven genomic signatures were absent from the study subjects, while previously described functionally significant ancestral forager-based alleles were identified. One of the most interesting findings to emerge from our analysis of foraging versus agricultural genome profiles was a potential for an increased chemical dependency for tobacco. We observed heavy tobacco usage by all study participants, both male and female. Historical accounts include the successful use of tobacco as a means of trade or coercion of indigenous Khoesan by European settlers 
. Anthropological observational studies suggest an unusual devotion of Ju/'hoan to master the difficult task of tobacco cultivation over food-based cultivation when minimal farming is adopted 
. Our data therefore suggests that the Ju/'hoan have not had adequate time to adapt to selective pressure associated with the use of tobacco. The significance of genes associated with inflammatory, autoimmune or immune diseases, being significantly enriched between forager and agriculturalist requires further investigation. Coined ‘the harmless people’ 
, it may not be surprising that we found a greater representation of loci associated with mood-based disorders. Physical characteristics within the Ju/'hoan with possible links to enriched pathways include (i) maintaining both thermal and fluid homeostasis within desert climates, (ii) the need for rapid wound repair, and (iii) a possible state of semi-erection in males. The latter, a locally accepted trait, has been documented in Bushmen rock art 
and reported as a defining characteristic 
Unlike the Ju/'hoan, the !Xun exhibit significant male-derived non-Khoesan African ancestral contribution to their gene pool. While autosomal marker analysis suggests roughly 20.5% non-Khoesan admixture, Y-chromosomal analysis suggests a possible East African Nilotic contribution, although extended autosomal substructure analysis suggests a proto-Bantu and Sandawe contributions while excluding for a Nilotic contribution. Evidence for Bantu migration into the northern Kalahari region of Namibia appears as early as the 7th century 
. Bantu-Khoesan interaction is evident by the introduction of iron-based arrow tips and cooking utensils, as well as the use of cultivated tobacco by the Khoesan, and conversely the inclusion of clicks within the non-click languages of early Bantu immigrants, for example isiXhosa (the language of the amaXhosa). The possibility of a pre-Bantu, likely east African migration into the region requires further investigation. The Ju/'hoan-Yoruba differentiating AIMs defined two unique !Xun subgroups suggesting independent genomic prehistories. The Ju/'hoan-ancestral !Xun share on average 54.8% (range 43.6–67.5%) of their genomic heritage with contemporary Ju/'hoan, and include the Angolan !Xun from this study (). In contrast, we identify a new non-Ju/'hoan (range 0.9–5.3%) ancestral contribution to 50% of the !Xun, averaging 71.1% (range 65.9–76.9%) (). We suggest that the !Xun identifier as used today incorporates different Khoesan prehistories, one independent from contemporary Ju/'hoan. Interestingly, the non-Khoesan African contribution to the !Xun appears to be uniform with ancestral signatures shared by contemporary Bantu and Sandawe. Our data therefore suggests that these two independent !Xun lineages carry the same non-Khoesan African contributions. The amaXhosa Bantu carry an almost equal ancient ancestral Khoesan contribution, while AIMs analysis suggests that this contribution is largely non-Ju/'hoan. It is highly feasible to assume that the southward migration of the amaXhosa along the eastern coast would constitute differing Khoesan contribution from the more westerly located inland Ju/'hoan. This observation is further supported by the lack of L0k mtDNA representation within the amaXhosa. Further analysis would be required to determine the relationship between the Khoesan contribution to the amaXhosa and the ‘unknown’ !Xun lineage identified in this study.
While the !Xun and amaXhosa show evidence for historical admixture, inter-continental migrations to the region has led to the emergence of more recent admixture. Considering a highly variable non-Khoesan contribution to the #Khomani, the Coloued and Baster populations represent a complex admixture pattern that transverses both the earliest and the most recently diverged human lineages. Defining and tracing such significant ancestral contributions provides a unique model not only to track human expansion and prehistories, but also define gene regions undergoing selection 
and recombination 
. The datasets presented in this study provide a unique resource for further genomic analyses. In the Ju/'hoan we speculate that the fraction of ROH has been lowered as a result of early divergence with other populations, while increased as a result of a smaller effective population size (Ne
). Unlike cosmopolitan societies, the maintenance of population size is an essential survival mechanism for foragers. As a result of varied contribution of ancestrally distinct chromosomal segments, contemporary southern African populations would display admixture-based recombination, decreasing total ROH. The complex ‘Khoesan-African-Asian-European’ ancestral admixture fractions of the Baster and Coloured would be further impacted by gender-specific meiotic recombination rates 
. The observation of gender biased ancestral contributions include a paternally-driven ‘African non-Khoesan’ contribution to the !Xun, maternally-driven ‘Khoesan’ contribution to the amaXhosa, and maternally-driven ‘Khoesan’ and paternally driven ‘non-African’ (likely European) contribution to the Baster and Coloured.
Although previous studies have looked at the ancestral contributions to the Coloured 
, no studies have to date addressed complex admixture within the Basters. Emerging from a common historical background to the Coloured, the Baster population have since the late 1800 s distinguished themselves as independent from the Coloured, migrating to the now Baster nation of Rehoboth in Namibia 
. In contrast to the Coloured we show the Baster population to carry the largest Khoesan-derived maternal contribution (91.7% compared to 64.3% in the Coloured) and the largest paternal European-derived contribution (93.3% compared to 71.8%), while autosomal marker analysis confirmed increased ‘Khoesan’ and ‘European’ contributions and decreased ‘Asian’ and ‘African non-Khoesan’ contributions. Geographic distribution of the ‘African non-Khoesan’ admixture fraction showed an increased contribution and significance from west to east (Baster, NC-, D6- to EC-Coloured, Figure S7
), with significance of the Bantu-derived fraction (1.6%, 5.8%, 15.4% and 16.6%, respectively) based on nine ancestral fractions () and mirroring Bantu population distributions (Statistics South Africa Census 2011 and Community Survey 2007, (http://www.statssa.gov.za
)). The most significant ‘Asian’ contribution was found within persons who were residents of District Six. Previously a residential region of Cape Town, District Six was geographically located at the heart of the Dutch-East Indian slave trade 
. In this study we define an almost equal ‘broadly Indian’ and ‘Sino-Tibetan’ contribution to the D6-Coloured. Besides fixation for the dry earwax allele in the Han Chinese and Koreans, an elevated frequency (71%) has been reported for the Indian Dravidian inhabitants of Tamil Nadu (correlating to the DR-S-LP3 population from this study) 
. Lack of this allele in our subjects alludes to a non-Dravidian Indian contribution which was further supported by non-contributing independent GC4 Dravidian subgroup substructure.
Since the submission of this paper, two publications have emerged that have addressed genomic variation within the southern African region we studied. The first assessed ~500 K custom designed variants including study subjects described as Ju/'hoan and !Xun (!Xuun) and grouped together as Kx'a speakers 
. Significant findings consistent with our analyses include ~20% non-Khoesan contribution to the !Xun (after fixing non-Khoesan contribution to the Ju/'hoan at 6%), while confirming minimal admixture contribution within the Ju/'hoan. Additionally this study dates the !Xun African non-Khoesan-mixture time to around 450 years ago and implies an ancient genetic link between southern and Eastern Africa. Our observation for a predominance of the East African Nilotic (non-Bantu) E1b1b Y-chromosomal haplogroup within the !Xun may provide further confirmation for a southern-eastern link, although our autosomal analysis suggests that this link is more likely related to the Sandawe and not the Nilotic peoples. No ancestral link was observed between east Africans and the Ju/'hoan from our study. The second paper looked at ~2.3 million variants including study subjects described as Ju/'hoan, !Xun, Coloured (Colesburg), Coloured (Wellington) and undefined South African Bantu-speakers 
. Consistent with our findings and the first paper, this study depicts the Ju/'hoan as a relatively homogenous population, while depicting a non-Khoesan contribution to the !Xun. In contrast to both studies, we suggest additional !Xun substructure and present the notion of two distinct !Xun prehistories. Our assumption is that contemporary !Xun represent a unique ancestral Khoesan lineage with an ancient non-Khoesan African (predominantly Bantu) contribution, with one subgroup having shared an ancient genetic link with the Ju/'hoan while the other remained genetically isolated from the Ju/'hoan. Notably the second study reports a predominance of Angolan !Xun study representation, represented in our study by the Ju/'hoan-ancestral genetic link. Further between study confirmation includes the representation of a South Asian (Indian) contribution to the Coloured, in particular the Wellington-Coloured (approximately 60 miles from Cape Town and District Six) compared with the Colesburg-Coloured (approximately 500 miles from Cape Town and District Six), with minimal East Asian ancestral contribution. Unlike the Wellington-Coloured, however, no subject in this study presented with non-African ancestry (Bantu and/or Khoesan). A single individual from District Six lacked any observable Khoesan contribution. No distinction was made for the Southern Bantu included in the latter study, so no correlation could be made with regards to the amaXhosa. The availability of new southern African datasets will allow for a more comprehensive analysis of population substructure within the region.
This study demonstrates both ancient and recent admixture within southern Africans. Cautionary concerns include: (i) bias in current content arrays towards non-African populations will greatly impact inferences about diversity among southern Africans, while lack of rare allele representation would diminish an ability to separate southern African subpopulations, (ii) lack of an available common ancestral genome that truly represents the earliest modern humans results in biases in methods used to attain divergence times among populations, (iii) inferences regarding population structure and recent admixture events are currently based on analyses of data reflecting contemporary genetic variation between populations, which is still largely lacking for the region of Southern Africa, and (iv) this is confounded by lack of data for populations that may actually be extinct. Taking these cautionary observations into consideration, we present an analysis of a set of individuals that, to the best of our knowledge, most accurately defines a homogenous ancestral Khoesan contribution, the Ju/'hoan. Additional cultural differences that may have restricted interbreeding between our Ju/'hoan and local agro-pastoral groups include; economic distinction (those without and those with possessions), language (Khoesan versus Bantu), social practices (egalitarian versus patriarchal society), kinship (bilineal/patrilineal versus matrilineal), marital locality (matrilocal versus patrilocal), and marital practices (monogamy versus polygamy, and no bridal payment versus a bridal payment). While a recent study acknowledges western influences as a result of the establishment of a Ju/'hoan ‘reserve’ near Tsumkwe 
, for this reason we actively avoided recruitment within the immediate vicinity of Tsumkwe. In contrast to the Ju/'hoan, we describe not only a ‘African non-Khoesan’ (almost equal proto-Bantu and Sandawe) contribution to the !Xun, but define two distinct !Xun lineages, with, and largely without, a shared Ju/'hoan ancestry. Additionally we describe a new population with complex ancient and recently diverged genomic contribution, the Basters of Namibia. Sharing a history with the South African Coloured, population-defining genetic signatures include increased significance of Khoesan and European contribution with gender-specific bias to a maternal and paternal contribution, respectively. In contrast, while we confirm increased ‘African non-Khoesan’ (largely Bantu and to a lesser extent Sandawe) and ‘Asian’ (Indian and Indonesian) contribution to the Coloured, we demonstrate significant regional-based ancestral differences which would have important implications for gene mapping studies that rely on self-reported ancestry among Coloured and non-Coloured populations. As inter- and intra-continental migration increases globally, so will the impact of admixture on disease gene mapping studies. The dataset presented provides an opportunity to investigate the impact of arguably some of the most diverse genomic contributions within single population identifiers.