Despite the large size of the contemporary nomadic Fulani population (roughly 13 million people), the genetic diversity and degree of differentiation of Fulanis compared to other sub-Saharan populations remain unknown. We sampled four Fulani nomad populations (n = 186) in three countries of sub-Saharan Africa (Chad, Cameroon, and Burkina Faso) and analyzed sequences of the first hypervariable segment of the mitochondrial DNA. Most of the haplotypes belong to haplogroups of West African origin, such as L1b, L3b, L3d, L2b, L2c, and L2d (79.6% in total), which are all well represented in each of the four geographically separated samples. The haplogroups of Western Eurasian origin, such as J1b, U5, H, and V, were also detected but in rather low frequencies (8.1% in total). As in African hunter-gatherers (Pygmies and Khoisan) and some populations from central Tunisia (Kesra and Zriba), three of the Fulani nomad samples do not reveal significant negative values of Fu’s selective neutrality test. The multidimensional scaling of F^sub ST^ genetic distances of related sub-Saharan populations and the analysis of molecular variance (AMOVA) show clear and close relationships between all pairs of the four Fulani nomad samples, irrespective of their geographic origin. The only group of nomadic Fulani that manifests some similarities with geographically related agricultural populations (from Guinea-Bissau and Nigeria) comes from Tcheboua in northern Cameroon.
The Fulani (also known as the Foulah, Peulh, Peul, Fulfulde, or Fulbe) are an ethnic group of sub-Saharan Africa who live in 17 states and number almost 30 million people. Although most Fulani now live settled lives, they spring from an originally nomadic population. Both the settled and nomadic communities are collaborating. The settled communities cultivate the cereals (mainly sorghum) and raise a small number of domestic animals; the activities of the pastoral groups are concerned almost exclusively with animal husbandry. It must be stressed that cattle keeping is really the central point of cultural identity of all Fulani nomads. Because of the specific needs of this practice, more pronounced division of labor, resulting in longer time separation of the husband and wife, is encountered in Fulani society. Males deal with the pasture, and females sell milk and milk products in town markets. Nomadic Fulani are known from almost all localities of the West African savanna and the Chad basin.
Archeological indicators date the origin of this nomadic population, on the basis of the rock art of the central Sahara, to about 5,000 years ago (Dupuy 1999); some indicators even suggest a Neolithic origin of the Fulani population (Ba and Dieterlen 1966). Some Fulani groups settled to form a number of important states: the kingdom of Tekrur on the lower Senegal River in the 11th century, the Massina Empire on the middle Niger in the 15th century, and the Sokoto Empire in the 19th century in northern and eastern central Nigeria.
The modern Fulani, who live in sub-Saharan Africa between the Sahara and the tropical rain forests, can be divided into the settled Fulani (15 million people) and the nomadic Fulani (up to 13 million people), sometimes called the M’Bororo (or Bororo) or the Wodaabe. The nomadic Fulani live in the African middle savanna belt, from eastern Senegal to the Central African Republic, and are the most numerous nomadic group in this area. Linguistically, both Fulani groups (the herders and the agriculturalists) belong to the Atlantic branch of the Niger-Congo language family (Ruhlen 1987).
Fulani nomads remain detached from the settled Fulani. Over the course of the year they practice seasonal migration; in the rainy season they move with their herds to the north, and in the dry season they return to the south. The northsouth line varies by tribe and family but is generally about 500 km long (Dupire 1962). In addition to these seasonal movements, there is large-scale migration, by which the nomadic Fulani have spread across the West African savanna and the Chad basin (Mohammadou 1975).
From an anthropological perspective the Fulani are rather heterogeneous; they show similarities to other sub-Saharan populations, but some characteristics-such as pale skin, a long, straight nose, and thin lips-link them to North African groups. In the 1930s it was assumed that the Fulani had migrated into the Lake Chad region and further into the West African savanna from East Africa, and even Egyptian and Near Eastern origins were proposed (Tauxier 1937); on the basis of the Rh system, however, it has been possible to link these populations to West African groups (Excoffier et al. 1987).
mtDNA Variability in Sub-Saharan Africa
The mtDNA diversity of African populations is relatively well known, but not all regions and ethnic groups have been sufficiently sampled yet; indeed, the mtDNA data of people from such inaccessible areas as eastern Chad or the Congo basin have not been studied at all.
From the phylogenetic point of view the mtDNA sequences from subSaharan Africa have been classified into L-type haplogroups (Chen et al. 1995, 2000; Watson et al. 1996, 1997; Rando et al. 1998; Bandelt et al. 2001; Pereira et al. 2001; Torroni et al. 2001; Brehm et al. 2002). About 30 sub-Saharan L-type haplogroups have been identified, and their ethnic or geographic origins and coalescence times have recently been summarized (Salas et al. 2002, 2004; Kivisild et al. 2004; Rosa et al. 2004). It seems that the main diversifications originated in East Africa but that the West African regions also contributed to the recent, wide mtDNA diversity.
The nomadic Fulani have not been studied with regard to mtDNA so far. The only samples (n = 61) from a Fulani-speaking population have been presented in research by Watson et al. (1996, 1997), who investigated a mixed sample of settled Fulani from Cameroon, Niger, and Nigeria. Close relationships to the neighboring settled populations of West Africa (the Tuareg, Kanuri, Hausa, Songhai, Yoruba, and Mandenka) have been confirmed by other studies [e.g., Pereira et al. (2001) and Salas et al. (2002)]. This Fulani group cannot be separated from the other populations, even by the indexes of molecular diversity (Pereira et al. 2001; Trovoada et al. 2004). Rosa et al. (2004) has recently studied another group of sedentary Fulani (n = 77); these are Fulani from GuineaBissau, unique in their slight divergence from six otherwise similar populations in Guinea-Bissau. Their haplogroup profile shows high proportions of haplogroups L1b, L2a, and L3b in particular, although a few Eurasian haplogroups were also found among them.
The aim of this study is to present the HVS-I mtDNA sequences of nomadic Fulani groups from four different locations that have not yet been described in this way. There are four questions that we seek to answer: (1) What is the mtDNA profile of the sample-does it include haplogroups of East or of West African origin? (2) Are the sampled Fulani genetically homogeneous, and do they reveal similar degrees of molecular diversity? (3) Are there genetic differences between the nomadic and sedentary Fulani populations? (4) What are the genetic relationships between the Fulani and their neighbors?
Materials and Methods
Sampling. Biological samples (buccal swabs) were collected from Fulani nomads at different places within their geographic range across the Chad basin and the West African savanna. The Chad basin samples come from the middle Logone River in Chad (Bongor area, n = 49) and from the territory of Cameroon south of the Benue (Tcheboua area, n = 40). A third sample set comes from the eastern part of Burkina Faso, south of the city of Fada Ngourma (Tindangou area, n = 47), and the fourth and final sample set was collected in the western part of Burkina Faso, south of the city of Bobo Diulasso (Banfora area, n = 50) (Figure 1). Camps were visited during the dry seasons (January-February 2002-2004), when the nomads rested at the southern extremity of their lines of migration. The only exception is the sample from Tcheboua (Cameroon), where the Fulani have recently settled.
All possible measures were taken to avoid sampling individuals with a known common maternal ancestor. The samples were secured under conditions of informed consent, and with the authorization of the ministries of education of the countries concerned.
Laboratory Methods. DNA was isolated from the buccal swabs using the method described by Cerný et al. (2004), and the HVS-I was amplified with the primers F-15971 (5′-TTA ACT CCA CCA TTA GCA CC-3′) and R-16410 (5′-GAG GAT GGT GGT CAA GGG AC-3′)- Products were purified using the QIA-quick PCR purification kit (Qiagen, Hilden, Germany) and then sequenced using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, California) with the F-15971 forward primer on an ABI Prism 3100 Avant Genetic Analyzer. Samples that contained length heteroplasmy (a homopolymeric tract of cytosines between positions 16184 and 16193) were also sequenced with the reverse primer R-16410.
Data Analysis. The sequences were aligned using DNA Alignment 184.108.40.206. (Fluxus Technology Ltd., Germany) and BioEdit 5.0.9 (Hall 1999). Each sequence was compared to the Cambridge Reference Sequence between nucleotide positions (np) 16030 and 16370 (Andrews et al. 1999). Individual haplotypes were ranked into haplogroups according to the published phylogenetic literature (Watson et al. 1997; Rando et al. 1998; Quintana-Murci et al. 1999; Alves-Silva et al. 2000; Bandelt et al. 2001; Pereira et al. 2001; Torroni et al. 2001; Salas et al. 2002, 2004; Rosa et al. 2004; Kivisild et al. 2004) and controlled for possible new clades with the help of the Network software (Fluxus Technology Ltd.).
Gene diversity, nucleotide diversity, and the average number of pairwise differences were calculated using Arlequin 2.000 (Schneider et al. 2000). Irregularities in the distribution of the average number of pairwise differences were tested using the raggedness index (Harpending 1994), which gives higher values for stable populations and lower values for expanding populations. Selective neutrality was analyzed using the methods of Tajima (1989) and Fu (1997).
For the purposes of the mtDNA study some previously published HVS-I data have been taken from the literature (Table 1). The criteria of selection were geographic proximity to or possible genetic relationship with Fulani nomads, available sequences of np 16030-16370, and population samples with n > 20. To evaluate the genetic distances between the populations, we calculated F^sub ST^ (the pairwise difference method) using Arlequin 2.000. Multidimensional scaling (MDS) analysis based on F^sub ST^ distances was carried out using Statistica software. The genetic structure of the populations and their different regional groupings were further evaluated by analysis of molecular variance (AMOVA) (Excoffier et al. 1992). Admixture proportions were calculated using the Admix 2.0 software (Dupanloup and Bertorelle 2001). Variations at np 16182-16185 and length polymorphic polyC regions were not taken into account in all analyzed samples.
mtDNA Haplogroups. The total sample from the four geographically distinct Fulani populations (n = 186) yielded 58 haplotypes, which were classified into 14 haplogroups (see Appendix 1); as expected, a significant majority belonged to sub-Saharan type L. The most numerous haplogroups are L3b (determined by the motif 16124, 16223, 16278, and 16362) and L3d (determined by the motif 16124 and 16223); these two haplogroups could not be distinguished from each other when only the HVS-I sequence was available. However, because both share the same geographic origin in West Africa, the HVS-II motif and RFLP polymorphisms were not further studied for a more specific molecular determination. Together, the L3b and L3d haplogroups contain 62 sequences with 12 haplotypes.
The second well-diversified haplogroup found in the Fulani nomads sample is L1b (50 sequences, 10 haplotypes), with the determining HVS-I motif 16126, 16187, 16189, 16223, 16264, 16270, 16278, and 16311, the origin of which also lies in West Africa (Salas et al. 2002, 2004).
All the other haplogroups identified are represented by conspicuously lower numbers; the most numerous among them, comparatively speaking, are L2b and L2c, which are also of West African origin. It is interesting that the otherwise common L2a haplogroup (determined by mutation at np 16294) is not present to any great degree in the overall sample. The distribution of the haplotypes in each of the sampled Fulani populations is quite even-all the aforementioned (more numerous) haplogroups were present in each population sample at comparable frequencies. In addition, it is worth noting that 15 sequences (8.1%) of North African/Eurasian origin (U5, V, J1b, and one sequence corresponding to the Cambridge Reference Sequence) occurred.
We have detected new mutations in the L1b haplogroup, which is otherwise well classified by HVS-I mutations at np 16126, 16187, 16189, 16223, 16264, 16270, 16278, and 16311 (Salas et al. 2002). The L1b network (Figure 2) shows that most of the Fulani mtDNA sequences fall within clade L1b1, which is determined by the presence of a mutation at np 16293 plus the aforementioned L1b motif. One haplotype with a mutation at np 16170, not yet reported elsewhere, was observed for one individual from Banfora (Burkina Faso), and two new haplotypes-one presenting a mutation at np 16093 (one Fulani from Bongor, Chad) and the other a mutation at np 16255 (one Fulani from Tindangou)-were also observed. Also, in the L1b* haplogroup there is a new haplotype with a mutation at np 16368.
Molecular Diversity. The indexes of genetic diversity for all four Fulani samples are given in Table 2. The gene diversities reach values between 0.893 and 0.953 (for the merged population, 0.936). This measure, considered equivalent to heterozygozity in haplotype studies (Schneider et al. 2000), together with the nucleotide diversity and the average number of pairwise differences attains values similar to those known from other sub-Saharan populations (Cerný et al. 2004). On the other hand, the relatively low levels of the raggedness index are interesting; in one sample from Burkina Faso (Tindangou) this index even attains statistical significance. Tajima’s D test of selective neutrality, based on the infinite-site model of the studied groups, was not significant in any of the samples. Fu’s F^sub S^-which is much more sensitive to population demographic expansions and is highly significant (p ≤ 0.001) in all agricultural populations [see Pereira et al. (2001, Table 3)]-is significant only for the Fulani sample from Bongor. Significant levels were obtained from merged Fulani groups in the Chad basin or when all four Fulani samples were merged (n = 97). However, the Fulani from Burkina Faso, when taken together, did not attain a statistically significant value. Neglecting the hypothesis of selection, the nonsignificant values of the D and F^sub S^ statistics provide an indication of reduced demographic expansion hitherto observed mainly in hunter-gather populations such as the Pygmies or Khoisan (Excoffier and Schneider 1999).
Comparison with Other Populations. The populations listed in Table 1 were compared using F^sub ST^ values. The most important finding was that none of the sampled Fulani groups were differentiated from the others. No F^sub ST^ distance between any pair of Fulani nomads is statistically significant (data not shown). Considering the actual geographic distances between the Fulani groups in the Chad basin and those in the West African savanna (about 2,000 km), this finding is extremely interesting. On the other hand, the sampled Fulani groups differ from all other neighboring settled populations, including the Fulbe (Fulani) reported by Watson et al. (1997); the only exception to this is the Fulani sample from Tcheboua (Cameroon), which is not differentiated from three populations of the Central Sahel (the Hausa, Fulbe, and Yoruba).
The F^sub ST^ distances of 29 analyzed African populations were plotted on a graph using MDS analysis (Figure 3); populations of the Western Sahel group, except Mandenka, were not included because of their outlier position. The clear separation of three Fulani groups is evident at first glance; the only population not differentiated from the rest is the Fulani sample from Cameroon. It is worth noting the special position of the Mandenka, the only population from the Western Sahel, which is relatively close to the Fulani samples.
Genetic Structure. Genetic structure was approached through AMOVA. When all the population samples were considered as a single group, 6.37% (p
Admixture Analysis. The question of the maternal origin of the Fulani was investigated using Admix 2.0. This software was developed to estimate the contribution of parental populations to the population under study. Data were entered according to group C from the AMOVA (i.e., the Mandenka were not considered; see Table 3); the mutation rate was 0.00005 per year. Admixture coefficients (m^sub Y^) were calculated from allele frequencies without taking into account the molecular divergence between the alleles. Because the exact period of Fulani migration cannot be estimated from archeological studies, we selected 4,000 years rather intuitively as the time to the possible admixture event. Bootstrap estimates of the coefficients and their standard deviations were calculated using 1,000 replicates. The results are summarized in Table 4 and show that none of the selected groups made a dominant contribution to the mitochondrial pool of the sampled Fulani. The only weakly prominent parental populations are from the Nile valley.
HVS-I analysis of four Fulani populations revealed the different proportions of the mtDNA gene pool. A major role is played by West African mtDNA haplogroups, such as L1b, L3d, L3b, L2b, L2c, and L2d, which together make up 79.6% of the whole. The far from negligible presence of some haplogroups from western Eurasia (8.1%), such as U5, U6, and J1, is not particularly surprising in a sub-Saharan context because these haplogroups currently appear in North Africa. This may suggest an ancient origin of the nomads in the more northerly mountain massifs of the Central Sahara (Dupuy 1999). According to our own anthropological examination (data not shown), the non-sub-Saharan haplogroups are not carried by “West Eurasian-like” individuals, as might be anticipated, but were rather detected in common “Fulani type” peoples.
The analyzed samples of the Fulani show levels of genetic diversity similar to those of their neighbors in the West African savanna and the Chad basin. However, like hunter-gatherers (Pereira et al. 2001) and some others (Cherni et al. 2005), the Fulani-particularly the groups from Burkina Paso-show nonsignificant values of Tajima’s D and Fu’s Fs statistics, meaning that the signal of demographic expansion was lost by subsequent population events (Bandelt and Forster 1997; Excoffier and Schneider 1999). The Fulani’s lower values of the raggedness index are also interesting in this regard. It can be hypothesized that the mating pattern is the main difference encountered between the nomadic and settled populations. Similar relationships have been observed in Africa between hunter-gatherers and farmers; Pygmies (as hunter-gatherers), for example, offer their girls to neighboring food producers but do not obtain any girls in return (Sebesta and Lebzelter 1933; Bailey 1991), and for this reason their mtDNA gene pool remains constant and is gradually reduced over time-Pygmies mainly have the haplogroup L1c (Destro-Bisol et al. 2004). An even more remarkable reduction of genetic diversity can be observed among the Khoisan, another African hunter-gatherer population; their mtDNA gene pool contains mainly the haplogroups L0d and L0k, which occur in virtually no other populations (Vigilant et al. 1991; Salas et al. 2002).
These observations are probably not due to inconsistencies in the our field sampling strategy, because results consistent with the diversity levels of common African agricultural populations were obtained when the same field sampling strategy was used among sedentary Chadic-speaking peoples in northern Cameroon (Cerny et al. 2004).
The samples of the studied Fulani groups differ from practically all the neighboring populations, the only exception being the recently settled (one generation) Fulani of Tcheboua, where three nonsignificant comparisons were observed. There is no differentiation from the Hausa, Fulbe, and Yoruba peoples reported from Niger and Nigeria by Watson et al. (1997). Some differences of the maternal gene pool between the nomadic and sedentary Fulani populations are also apparent from research conducted in physiological characteristics, for example, lactase persistence; the incidence of this trait among sedentary Fulani is far lower than among their nomadic counterparts (Holden and Mace 2002).
The genetic relationships of the sub-Saharan populations presented in this study, and particularly among the Wolof, Serer, Fulani, and Mandenka, are little different from those revealed by classic genetic polymorphisms (Cavalli-Sforza et al. 1994), where the Fulani of Senegal (described as Peuls, from the French) were differentiated from the Fulani of Nigeria. As far as we are aware, however, the samples of Cavalli-Sforza and colleagues come mainly from settled Fulani populations. It is worth noting that, according to protein polymorphisms, the Fulani of Senegal are closer to the Wolof and Serer and more distant from the Mandenka (Cavalli-Sforza et al. 1994, pp. 181-182), whereas the Mandenka are linked to the populations of northern Nigeria (such as the Hausa or Fulani). The similarity among the Fulani, Serer, and Wolof (all from Senegal) is so strong that these three groups even form a discrete cluster (Cavalli-Sforza et al. 1994, p. 169). This does not correspond to the results of this study, which show that the mtDNA gene pool of the sampled nomadic Fulani is conspicuously different from that of the Serer and Wolof. The samples obtained from herding groups in the southern part of the Chad basin (Chad and Cameroon) and the West African savanna (eastern and western Burkina Faso) display perhaps closer relationships to the samples of the sedentary Mandenka, as suggested by the MDS constellation; a similar finding has also been reported by Rosa et al. (2004).
Y-chromosome data of 22 African populations, including the Fulani from Burkina Faso and northern Cameroon, were analyzed by Cruciani et al. (2002). The main result of Cruciani’s study is that different populations from northern Cameroon (Fali, Ouldeme, Daba, and some mixed samples) reveal traces of backmigration from Asia to Africa because of a high proportion of haplotype 117. However, the Fulani sample from northern Cameroon considered by Cruciani and colleagues shows a rather low frequency of this haplotype, and the Fulani, which have a high frequency of haplotype 43, are situated as outliers. Cruciani et al. (2002) also showed that the Fulani from Burkina Faso have reduced diversity, because only two Y-chromosome haplotypes were observed in their sample.
In the introduction we mentioned several hypotheses for the origin of the (nomadic) Fulani. One well-known hypothesis is that the Fulani come from the Nile valley (e.g., Tauxier 1937). Analysis of F^sub ST^ distances, however, shows no close relationship between the sampled Fulani and the analyzed Nilotic populations. Admixture analysis, however, does not exclude the possible parental role of the Nilotic populations because the admixture coefficient for these populations is high. It is necessary to state that the conclusiveness of this finding is rather low. Further geographic sampling, particularly from Niger and other parts of the Sudanic belt of Africa, is needed to acquire a deeper insight into the genetic structure of the nomadic people of the African Sahel.
Acknowledgments We wish to express our gratitude to the volunteers for their confidence and their helpful participation in the study. We are indebted to Peter Forster for his mtDNA data proofreading of some of the samples. This project was supported by the Grant Agency of the Czech Republic (grant 404/03/0318) and the Mellon Fellowship program (CAORC).