Sunday, June 21, 2009

Reviewing: The Genetic Structure and History of Africans and African Americans

This is a carryover from the "miscellaneous notes" of the last blog posting (The Creation of a Fake Controversy: The Fula Origins). It pertains to extracts taken from: The Genetic Structure and History of Africans and African Americans, a publication by Tishkoff et al (2009). that just came out recently. Here, we will examine the work a bit more:

Sampling entailed...

We studied 121 African populations, four African American populations, and 60 non-African populations for patterns of variation at 1327 nuclear microsatellite and insertion/deletion markers. We identified 14 ancestral population clusters in Africa that correlate with self-described ethnicity and shared cultural and/or linguistic properties.

And mentions of interest here...

Ethnic groups, sample size, language classification, and subsistence classification are given in Table S1. For the Kenyan, Sudanese, Nigerian, and Cameroonian samples, white cells were isolated in the field from whole blood with a salting out procedure modified from (S1) and DNA was extracted in the lab with a Purgene™ DNA extraction kit (Gentra Systems Inc., Minneapolis, MN). Dogon samples were obtained from blood spots donated by participants in a cohort study of malaria incidence in Bandiagara, Mali. Ghanaian DNA was extracted onsite from whole blood, with the Purgene™ DNA extraction kit. The Juspeaking !Xun (a.k.a.Vasekela) and Khoe-speaking Khwe samples were collected from individuals in the area of Schmidtsdrift in the North-West Cape of South Africa. The Cape Mixed Ancestry (CMA) population, commonly referred to as Cape Coloured in South Africa, was collected in the Western Cape Province. The Dogon sample was collected in Bandiagara, Mali. Nigerian samples were collected in Abuja and Adamawa State districts. Cameroon samples were collected from the Eastern Province (Baka Pygmies and neighboring Bantu groups), Southern and Ocean Provinces (Bakola Pygmies and neighboring Bantu groups and coastal groups), Center Provinces (Medzan Pygmies and neighboring groups, mostly Bantu populations), Western Province (Bamileke and Mbororo Fulani groups), Extreme North Province (Mandara mountains and northern plains; Fulani and Afroasiatic/Nilo-Saharan speaking populations). All Cameroonians were sampled in their native village; the Hausa sample (a population who emigrated mainly from the Kano area two generations ago) was sampled in the city of Yaounde. Samples from Chad, CAR, Congo, DRC and Rwanda were obtained from individuals who recently immigrated to Cameroon. Tanzanian DNA samples were collected from individuals residing in the Arusha and Dodoma provinces of Tanzania. Samples from Kenyan populations of southern Ethiopian origin (Burji, Konso) samples were collected in the Rift valley, Nyanza, and Eastern provinces of Kenya. Sudanese samples were collected in the Khartoum and Kasala provinces of the Sudan. The Yemenite Temani and Ethiopian Beta Israel samples were purchased from the National Laboratory for the Genetics of Israeli Populations (S2) . The South African !Xun/Khoe, Xhosa, Vende, Cape Mixed Ancestry (CMA), Yemenite Temani, Beta Israel, and the Malian Dogon DNA samples were amplified by Whole Genome Amplification (WGA) with Illustra GenomiPhi HY™ kits provided by GE-Healthcare (Buckinghamshire, UK). It should be noted that the DNA for the Dogon population extracted from blood spots appeared to be of lower quality and microsatellite markers did not amplify as well as other samples obtained from whole blood (43% of markers had missing data).


Africa is the source of all modern humans, but characterization of genetic variation and of relationships among populations across the continent has been enigmatic. We studied 121 African populations, four African American populations, and 60 non-African populations for patterns of variation at 1327 nuclear microsatellite and insertion/deletion markers. We identified 14 ancestral population clusters in Africa that correlate with self-described ethnicity and shared cultural and/or linguistic properties. We observed high levels of mixed ancestry in most populations, reflecting historical migration events across the continent. Our data also provide evidence for shared ancestry among geographically diverse hunter-gatherer populations (Khoesan speakers and Pygmies). The ancestry of African Americans is predominantly from Niger-Kordofanian (~71%), European (~13%), and other African (~8%) populations, although admixture levels varied considerably among individuals. This study helps tease apart the complex evolutionary history of Africans and African Americans, aiding both anthropological and genetic epidemiologic studies.

Right off the bat, the first thing that probably jumps at any reader, is the heavy reliance on microsatellite repeats, as opposed to uniparental binary markers. Restricted fragment length nucleotide sequences or polymorphisms [relying on insertions/deletions] can add precision to identifying discrete monophyletic units, but even here, the results are not always as clear cut as the case generally is with "classic" uniparental binary markers of Y-DNA and mtDNA respectively, relying on unique event SNPs. For instance, certain RFLP haplotypes have been linked to two different haplogroups; case studies on RFLP haplotypes on this website have shed light to such occurrences: see Lucotte et al.'s haplotype IV, RFLPs: Lucotte et al., A case study — Pt. 1, and RFLPs: Lucotte et al., A case study — Pt. 2. In the first link of the three for instance, we were told:

A total of 21 different 49a,f haplotypes were found and are illustrated in Fig. 4 as a sub-classification of the Iraqi Y-chromosome haplogroups. The most represented haplotype of haplogroup E is haplotype 5 (A 2 C 0 D 0 F 1 I 1 ). This is followed by haplotype 11 (A 3 C 0 D 0 F 1 I 1 ) at a much lower frequency. Haplotypes 5 and 11 were observed both in Africa (Lucotte et al., 2001; Passarino et al., 1998; Persichetti et al., 1992; Santachiara-Benerecetti and Semino, 1996; Spurdle and Jenkins, 1992; Torroni et al., 1990) and Eurasia (Passarino et al., 2001; Semino et al., 2000b) but in Africa they belong to haplogroup E, whereas in Eurasia, particularly in *Northeastern Eurasia*, they belong mainly to the haplogroup R-M17. Interestingly, the proportion of haplotypes 5 and 11 in haplogroups E and R-M17 is reversed, with haplotype 5 prevalent in haplogroup E and haplotype 11 in haplogroup R-M17. By considering that the two haplotypes differ by a single band change and their different proportion in the two lineages, it is likely that haplotype 11 is a derivative of haplotype 5 in haplogroup E and just the opposite in haplogroup R-M17. - N. Al Zahery et al. 2003, Y-chromosome and mtDNA polymorphisms in Iraq, a crossroad of the early human dispersal and of post-Neolithic migrations.

In cases where it might be difficult to link a RFLP haplotype to a specific sub-clade or else superclade, classic uniparental (Y-DNA or mtDNA) SNP and/or UEP markers are called on, as we see in the aforementioned link 2 and link 3 on Lucotte et al. Questionable allocation(s) of ancestral origin based on "shared" microsatellite markers and/or insertion/deletion markers is therefore bound to occur, as one comes across in the study at hand; ancestry, which would otherwise actually be more a reflection of an ultimate shared common recent ancestry prior to the ethnogenesis of the implicated contemporary groups into recognizable ethnic or socio-geopolitical entities they are known today, as opposed to some recent unidirectional gene flow from one designated contemporary group to another.

Moving onto excerpts of interest...

As some have correctly picked up on, and as the authors themselves openly admit, there are issues with their Dogon sample that need to be ironed out...

Detection of relative pairs: Relative pairs and duplicated samples in the dataset were inferred from the pattern of shared genotypes and population allele frequencies with RELPAIR 2.0.1 ( 6-8) . Because the inclusion of closely related individuals can impact population genetic inferences [e.g. (S9)], we took the conservative approach of excluding individuals inferred to be third degree or more closely related, including inferred relative pairs between regional ethnic populations (e.g. all Tanzanian populations). An exception was made in the case of the Dogon as it is difficult to reliably infer relative pairs in a small sample and the Dogon are highly distinctive and could not be readily merged with other populations to improve allele frequency estimates. Merging the Dogon with other non-Pygmy West African populations inferred four unrelated individuals in the sample, but this may be overly conservative given the distinctiveness of the Dogon sample from other West Africans. Also, the Dogon are the only representatives from Mali in our study and since the sample size is already small we did not want to further reduce the sample size in the analyses, especially if the relative pair estimates were questionable. Therefore, RELPAIR inferred relative pairs among the Dogon were not excluded. In total 737 individuals were removed. Networks of relatives, which in some cases were quite complex, were plotted with neato from the GraphViz software package (S10), which was used to select the minimum number of individuals to exclude to break up networks of relative pairs.

Indeed, the Dogon sample size, which to reiterate what the authors tell us consisted of candidates who could also possibly be closely related and the only sample that represented Mali, was only one of 9 individuals, a fairly modest size. Keeping this in mind, we are told:

The Dogon from Mali, who speak a Niger-Kordofanian language, cluster with the Saharan and eastern Africans in the NJ tree (Fig. 1), consistent with the results from STRUCTURE analysis, showing considerable Saharan (blue) and Afroasiatic (purple) ancestry, and consistent with oral history of a northern African origin (although it should be noted that the sample size for this population, 9 individuals, is very small and many markers did not amplify well) (Fig. S13; Table S9).

"Saharan" without regional specificity is next meaningless. Just about any of the groups on the Sahara, Sahel or adjoining "sub-Saharan" territories have "Saharan ancestry". Furthermore, as has been the talk of this site recently, relics of "fertile-Sahara" era or earlier east African ancestry is scattered along the length of the Sahara, and generally finds expression in Sahelian and "Sub-Saharan" populations in relatively low incidences, if not somewhat patchy distribution. Such shared Saharan ancestry is bound to have some sort of manifestation across bi-parental microsatellite markers as well. As we've just seen, the authors note that "many markers did not amplify well", and indeed, if there is any further indication of an air of dubiousness about the results of the amplifications, then perhaps it is notion of some 45% ancestry, or specifically 45% of the sample identified with an AAC (Associated Ancestral Cluster) , being suggestive of European ancestry in that small sample of Dogon, with 9 individuals [see table S8 of the publication]; and even if one were to remotely entertain the possibility as a hypothetical scenario for argument's sake, it would implicate these individual as essentially "outliers", if anything, based off on previous information published on samples taken from sections of Malian society, including the Dogon. Additionally, with regards to such an outlier situation, one might even ask if this is not a function of "third party" interaction, with that "third party" being coastal north African Imazighen and Arabized groups. The authors themselves base AAC assignment on where said markers are allegedly the most maximum in 'frequency'...

African and non-African Associated Ancestral Clusters (AACs; labeled based on the populations showing the highest levels of ancestry for each inferred ancestral cluster)are highly divergent.

One can suppose 'frequency' could be a telltale sign of geographical origin of certain biparental microsatellite clusters, if only tenuously by itself, but as a matter of precision, it effectively plays that role only when taken into account with other molecular genetics information, like the internal molecular variation of a designated clade, and in association, how much of this variation is paraphyletic, along with definite geographical patterns of distribution. For instance, one might surmise that Hg R1b originated in Europe, merely on the account of the haplogroup's unparralleled prevalence in that region, and place the maximum likelihood of finding ancestral R1b clades in Europe before finding them elsewhere. However, as a matter of merely "frequency", Hg R1b's incidence in Europe in no way allows inference that either Hg R1* or its descendant, the upstream clade R1b* originates there; in fact, R1b* clades bearing upstream SNPs that are nearer to the root of that sub-clade, but lacking the well-known downstream SNPs, have so far only been identified in non-European samples. Interestingly, these non-European samples happen to be African ones. At the same, ironically, R1b itself is relatively rare in Africa, even in the coastal north African areas which are closer to Europe. So frequency as a single factor for adjudging origin can be misleading. Neutral factors like random genetic drift, for example, can elevate a marker's frequency in a designated locale, and yet, this locale in of itself need not be the actual location of origin for said marker. For microsatellite repeats, in absence of binary or UEPs, the assignment of origin based on frequency alone becomes even more uncertain, than the case is for binary or SNP markers. Microsatellite repeats could even be shared both between different and within single monophyletic clades in some occasions, due to homoplasy.

Many questions abound the objectivity of result inferred for the small Malian sample. In a related matter, the Beja samples were implicated in some ~ 32% and ~ 34% AACs [the Banuamir and Hadandawa samples respectively] placed under "European" [see: Table S8 of the journal]. Well, to the extent that ancestry suggestive of "possible" European origin has been implicated in a Beja sample before, it is reasonable to assume that this may render a section of the Beja to display shared clusters with Europeans in a segment of nuclear DNA that is outside uniparental binary markers; however, it is worth noting that the uniparental gene pool, the paternally-inherited one in particular, featured fairly low incidence of ancestry suggestive of ultimate European ancestry. Citing Hassan et al. (2008), the Beja male gene pool comprised of only ~ 5% ancestry that could *potentially* be linked to European origin [Hassan et al. (2008) provide little information on specific downstream clades of Hg R1b]; this is fairly consistent with R1b's general distribution in northern Africa, where it is generally low, in contrast to the fairly considerable representation of European maternal ancestry in coastal northwest Africa. So, the 32 or so % of microsatellite AAC linked to a possible European origin is out of character, when uniparental ancestry information is taken into account. On the other hand, substantial AAC sharing between coastal northwest African samples and the European ones is consistent with uniparental DNA examinations, which to reiterate, mainly finds expression in the maternally-inherited gene pool.

Mitochondrial DNA analysis indicates that Fulani have lineages of predominantly West African origin and that they cluster together and close to the Mandenka population from Senegal [S93].

By contrast, Y chromosome analyses of Fulani
sampled in the Sudan indicates shared ancestry with Nilo-Saharan and Afro- Asiatic speaking populations [S89].

The line about the mtDNA is essentially a reversion back to what has been cited in the main post above, as observed by Cerny et al. (2006). The line about the Y chromosome analysis however, is quite misleading; one only needs to refer to Hassan et al. (2008) [see: Y-Chromosome Variation Among Sudanese] to see how: The Sudanese Fula sample here, as noted in the main post above, predominantly featured Hg R1*-M173 markers in the uniparental paternal gene pool. This marker is in fact generally quite rare in "Afro-Asiatic" speaking populations; rather, its highest incidences has thus far been reported in Niger-Congo, Nilo-Saharan and "Afro-Asiatic" (Chadic) speaking groups in northern Cameroon, with smaller occurrences in the rest of central Africa [see Luis et al. 2004]. Nilo-Saharan groups in Sudan itself, it should be noted, did NOT share this ancestry with the Sudanese-based Fulbe at all. This reverts back to what had already put forward in the main post, about the implication being that these Sudanese Fula derived from a section of the northern Cameroonian Fula, wherein the Hg R1*-M173 is featured in a considerable but not the predominant frequency.

The only marker that the Sudanese Fula shared with Nilo-Saharan and Afrasan ("Afro-Asiatic") speaking Sudanese groups was Hg E3b1 (M78). Common sense however, intimates that this aspect of the paternal gene pool was largely picked up from neighboring populations, only after the Fula situated themselves in Sudan. It makes sense, since after all, E3b1 is one of the prominently featured markers found in the Sudanese sample.

To recap, there's something still unmistakable about the Sudanese Fula sample: they retain western African ancestry, as signified by Hg E1-M33, which was absent in all autochthonous Sudanese groups; the only other groups where this marker was implicated in Hassan et al.'s (2008) work, were, well, the well-established west African groups like the Hausa and Wolof. When pressed on to explain away this anomalous feature of the Sudanese Fula's paternal gene pool, dissidents of the west African origin noticeably become mute on the matter.

These results raise the possibility of differential patterns of male and female gene flow into this population.

Whatever may be "differential" about male and female gene flow patterns in Fula sample, no doubt influenced by matters like life style recalling on nomadic vs sedentary, both the maternal and paternal gene pools communicate the same thing: west African origin!

Our analysis, using genome-wide nuclear markers and STRUCTURE, indicates that the Fulani have distinctive ancestry [fuchsia] at K = 14 in the global analysis [Figs. 3,4] and at K = 9 -14 in the Africa analysis [Fig. S13]. The Fulani cluster with the Chadic and Central Sudanic speaking populations at K <13> They also cluster near the Chadic and Central Sudanic speaking populations in the NJ tree based on population genetic distances [Fig. 1].

Fula samples taken from either central Africa or Sudan are expected to cluster with groups in central Africa-Chad, as the Fula would have arrived in Sudan via a central African and/or Chadic corridor, having experienced a bottleneck [see main blog post]. The central African Fula in turn would have arrived ultimately from Atlantic-coastal regions of western Africa, where their ethnogenesis as the Fula-"proper" likely occurred.

The pattern of the Sudanese Fula's Y-DNA composition is for instance, undoubtedly different from their western African brethren, but continuity still exists via recurring typical western African markers that unite all Fula groups, regardless of where they spread to; e.g. primarily Hgs E1 and E3a, and a lesser extent, Hg E2. Either of these clades are telltale signs of so-and-so Fula groups' western African ancestry.

In the global STRUCTURE analysis, the Fulani show low to moderate levels of European/Middle Eastern ancestry [blue], consistent with mtDNA and Y chromosome [S89] analyses, as well as the presence at low frequency of the -13910 mutation associated with lactose tolerance in Europeans in this population [S94].

Again quite misleading. Yes, understandably, "low" levels of shared ancestry attributable to potential "European" ancestry, might be correlated with "low level" incidences (~1.7 %) of R1b [pertaining the Rosa et al.'s (2007) Guinean sample] or relics of this ancestry attained from coastal northwestern populations [Imazighen] seen on the mtDNA side [Cerny et al. 2006], as cited in the main post. However, in the case of Fula samples, from northern Cameroon through to Sudan, a good degree of presumed shared "European/Middle Eastern ancestry [blue]" may very well actually be linked to common "distant" ancestry, as the paraphyletic R*-M207 and R1*-M173 markers suggest; these latter two are essentially rare to absent in "Middle Eastern" and "European" populations, which are mainly characterized by more downstream markers, especially European populations, who have no upstream markers for R1. Now, because R1b and R1a markers are still 'molecularly' linked to the upstream R*, any group bearing the latter or else any intermediary nodes between the upstream R* and the downstream R1a and R1b will "appear" to share ancestry with Europeans and/or possibly "Middle Easterners", which could misleadingly be misread as ancestry from "European/Middle East". Given the features of northern Cameroonian and Sudanese Fula sample uniparental male gene pool, it is a safe bet that this is the underlying issue at work, especially when the red flag of "moderate" levels is evoked; the Fula uniparental gene pool specifics, whether maternal or paternal, simply do not bear out "moderate" ancestry from "Middle East" or "Europe".

Recalling a previous blog post here
, R1*-M173 in Africa, we have:

Interestingly, upon revisiting Wood et al. (2005), it should be pointed out that paraphyletic clade of R*-M207 was detected amongst some "Afro-Asiatic" African groups, along with the paraphyletic clade R1*-M173 [it is worth noting that Wood et al. implicate the Egyptian sample here as something other than that of Semitic speakers (Arabic)], while some Niger-Congo groups — though in small frequencies [pooled] — tested positive for the paraphyletic R1b*, lacking the established downstream R1b markers. Henceforth, R*-M207, lacking downstream mutations have been identified in African groups via this study; and yes, the basic nodes of all presently known Hg R's downstream clades had been accounted for, which means that R*, as predicted above, is NOT relegated to the Indian sub-continent. All in all, this suggests that African Hg R pool is actually more diverse than many seem to think.

As for the matter concerning lactose tolerance, it has been pointed out here before, how this has nothing to do with European ancestry. Recap, see:
More on R1*-M173 bearers

In summation, the piece essentially rehashes the well-understood facts of primary west African ancestry of the Fula, both in terms of genealogy and language, along with genealogical "admixture" elements picked up over the course of migratory events, but suffers from matters, as laid out in detail in the main blog post, pertaining to idealistic urges for mystification where none is necessary or warranted. To provide an example, consider the following line:

Our analysis, using genomewide nuclear markers and STRUCTURE, indicates that the Fulani have distinctive ancestry (fuchsia) at K = 14 in the global analysis (Figs. 3,4) and at K = 9 -14 in the Africa analysis (Fig. S13). The Fulani cluster with the Chadic and Central Sudanic speaking populations at K <13>Fulani show low to moderate levels of European/Middle Eastern ancestry (blue), consistent with mtDNA (S93) and Y chromosome (S89) analyses, as well as the presence at low frequency of the -13910 mutation associated with lactose tolerance in Europeans in this population (S94). Additionally, we observe moderate to high levels of Niger- Kordofanian ancestry in the Fulani populations (Figs. 3, 4, S13;Tables S8, S9). These results do not enable us to determine the definitive origin of the Fulani, although they indicate shared ancestry with Saharan and Central Sudanic populations and suggest that the Fulani have admixed with local populations, and possibly adopted a Niger-Kordofanian language, during their spread across central and western Africa. The origin of European (possibly via the Iberian peninsula) and/or Middle Eastern ancestry in the Fulani requires further exploration with additional genetic markers.

The words "admixed with local populations" is particularly interesting, as it is naturally inconsistent with mtDNA report [and Y-DNA] where western Africa is concerned, recalling Cerny et al. (2006) for example, who point out that their results show a primarily or predominantly west African ancestry of "charasteristic" markers of the Fula, but wherein the gene pool of nomadic Fula samples show very little genetic contribution from neighboring sedentary populations. This, to put it simply, means that the only way for the Fula gene pool to be primarily west African, is if they were autochthonously west African.

Another interesting choice of words, is the mention of "adopt" in Tishkoff et al.'s piece above, in relation to the Fulani language. "adopt" is usually suggestive of one taking on something that wasn't originally his/her's. Supposing one remotely entertained said insinuation of the authors for just a minute, their very idea that a small band of foreigners adopted the "Niger-Kordofanian" language along their migratory path, means that the locals along that path would have had to have already been Fula speakers; that is the only way a wandering group of immigrants/foreigners could have become speakers of a Niger-Congo language a language family that supposedly did not characterize their own which by chance, happens to be Fula. In other words, the Fula as an ethnic group was already established on the continent prior to the intrusion of this small band of immigrants; it's just common sense. Henceforth, this band of immigrants cannot be judged as proto-Fula antecedents by any stretch of the imagination; rather, they should be viewed merely as a foreign element that integrated into an already well established ethnic-group, i.e. the Fula, as opposed to assuming the role of a "prototype" for the group into which they allegedly integrated. To somehow use this [unsupported] scenario as something that mystifies Fula origins, is clearly not rational or logical deductive reasoning; at best, it is quite comical. The matter of Fula origins was recently posted here, in detail; recap: The Creation of a Fake Controversy: The Fula Origins

It is interesting to note that while the authors carry themselves as very confused individuals about Fulani ancestry, their own tables (see S8) show very little AAC sharing with Europeans; their Nigerian Fulani sample supposedly shared ~5 % or so AACs placed under "European", while the Cameroonian Fula samples shared about 2.5% and 2.6% respectively. Nothing "moderate" about these.

From language, culture to genealogical heritage, all indicators resolutely point to a west African ethnogenesis of the Fula. This has been supplemented with archaeological evidence, as that from Tassili, on rock art. Here, are renditions of a cattle-rearing group, bearing features reminiscent of contemporary Fula and/or western Sahel groups in general. One scholar often cited on the internet, regardless of how objective or otherwise the sites themselves are, for identifying definite links between the Tassili n'Ajjer rock painting and contemporary Fula, is Amadou Ba. For instance, one site notes:

At the Tin Tazarift site, for instance, historian Amadou Hampate Ba recognized a scene of the lotori ceremony, a celebration of the ox's aquatic origin. In a finger motif, Ba detected an allusion to the myth of the hand of the first Fulani herdsman, Kikala. At Tin Felki, Ba recognized a hexagonal carnelian jewel as related to the Agades cross, a fertility charm still used by Fulani women. With the disappearance of many traditions and other aspects of African culture, works of traditional African art are becoming more and more scarce. - This particular piece is from

The following are from Andrew Brown Smith [African Herders (2005)], who seemingly attempts to portray himself as a "voice of moderation":

Of note here, is the observation that Smith as any other person guided by reasoning could not pass over the significance of linguistics, i.e. recalling the Niger-Congo familial ties of Fula language.

While briefly citing examples of the sort of reception Ba's definite correlations got in certain quarters, mainly amongst "western" observers, even Smith could not resist pointing out that the "coincidences" surrounding the reminiscence of Fula socio-cultural traits in definite features of the rock arts are simply too considerable to not draw a link between the two, i.e. between the figures in the rendition and contemporary Fula. Notwithstanding this openly-admitted acknowledgment, Smith of course, goes onto characterize Ba's demonstrations as "perhaps what they wanted to see to support their thesis", but offers no substantive counter-thesis that puts to question the "specifics" of what Ba "sees", other than a mere opinionated presumption of Ba's "possible" motive behind making the definite links that he made.

Click on the above images to get higher resolution versions.

In the lower images for example, one notices the head gear, which is not unlike that featured on a contemporary Dogon dancer in the image on the right hand side.

Other notes: Fulani cattle are also deemed to be of west African origin, as a sub-phyla of the west African breed of Zebu cattle called the "West African Zebu", and they fall into the following two groups -

The Fulani have been classified further into two groups: the lyre-horned subgroup consisting of Senegalese Fulani (or the Gobra), the Sudanese Fulani, and the White Fulani (or Bunaji); and long-horned subgroup represented by the Red Fulani (or Rahaji). Diali (or Djeli) is a strain of Fulani found on the flood plains of Niger river in Niger and south-west Nigeria (Rege 1999; Rege and Tawah, 1999). - courtesy

Dictionary description...

Sudanese Fulani cattle
West African, lyre-horned, milking cattle, usually light gray.- Medical-dictionary

Continuing with examination of the study at hand, the authors note:

Individuals from Saharan and Eastern Africa show heterogeneous ancestry, reflecting descent from populations ancestral to non-Africans and/or gene flow from non-Africans into Africa.

Well, "exotic" gene flow as function of said heterogeneity, while reasonably a contributing element to some extent, is really quite trivial in the big scheme of things. African populations would still be quite heterogeneous *regardless* of gene flow from non-Africans
. In relation to this, a subset of Africans who formed the basis for non-African groups, would assume an intermediary position in a case study involving a "comprehensive" collection of African and some non-African samples, because quite simply, that subset of Africans only represents a portion of the overall African diversity. Non-Africans in turn, only have a portion of the diversity of the subset of Africans from whom they emanate.

*Watch for occasional updates or modifications of this post in future.

*Last edited on 10/26/2010.