Diversity and Genetic Relationships among Seven West African Goat Breeds

This study was carried out to determine the genetic relationships among seven west African goat breeds : Casamance Goat (Kolda, Senegal), Labe Goat (Fouta Djallon, Guinea), three Sahel Goat (Djoloff, Senegal ; Maradi, Niger; Gorgol, Mauritania) red Sokoto Goat (Maradi, Niger) and Guera goat (Atar, Mauritania).The polymorphism of six microsatellites and the αs1-casein locus was analysed. The six microsatellite loci were polymorphic with a mean number of alleles ranging from 2.71 to 4.0. At the αs1-casein locus, A and B were the most frequent alleles, which are known to be associated with a high level of protein synthesis. A neighbour-joining tree and a Principal Component Analysis were performed and the reliability of both methods was tested. Our study shows that the genetic relationships among the breeds analysed correspond to their geographical distribution and in addition, that the Labe Goat is strongly separated from the other breeds. Among the seven markers used, four have an effect on the distribution of breeds while three seem to be non-informative. (


INTRODUCTION
The domestic goat capra hircus is an important livestock species throughout the entire asian and african continents.Goats are an important resource for agricultural, economic, cultural and religious purposes.Their origin remains uncertain and controversial, and the history of their domestication is more complex than indicated by previous studies.Archaeological evidence indicates that goats were one of the first animals to be domesticated by man around ten thousands years ago, in the ganj-darch in the zagros mountain pastures (Zeder et al., 2000).However, a world wide survey of domestic goat mtdna diversity has identified a multiple maternal origin with a possible centre of origin in asia as well as in the fertile crescent region (Luikart et al., 2001).In their study, three major mtdna lineages were revealed while in a more recent study (Joshi et al., 2004) concerning indian domestic goats two additional lineages were observed.
Apart from advances on the evolutionary and demographic history of humans and domesticated animals, these studies are also important for the management and conservation of today's animal resources.Unfortunately, many livestock breeds are still weakly characterised to date especially locally adapted breeds.
According to Doutressoulle (1947), West Africa presents a unique geographic mixture of two basic types of goats, particularly adapted to the local environmental conditions : • Savanah or Sahel goats in the Sahel belt south of the Sahara, extending from lake Chad to Senegal.This type which is long legged and not resistant to trypanosomiasis is found in an area free of tsetse flies.
• Dwarf and trypanotolerant goats in humid zones, extending from South-Senegal through Nigeria.
Among these types, breed classification is still unclear.For instance, the red Sokoto Goat is classified as a Sahel type by Epstein (1971), a dwarf type by Doutressoulle (1947) and a crossbreed of both by Wilson (1992).The Guera Goat was first described by Kane (1995) simply as a breed from Mauritania.Moreover, some breeds are known by the same name in different places, but are phenotypically different from one place to another.Conversely, there are breeds that look alike but have different names in different places (ILRI, 1996).
Our study concerns the characterisation of seven west African goat breeds : Casamance Goat (Kolda, Senegal); Labe Goat (Fouta Djallon, Guinea); Sahel Goat from Djoloff (Senegal), from Maradi (Niger) and fom Gorgol (Maurtania); red Sokoto Goat (Maradi, Niger) and Guera Goat (Atar, Mauritania) with six microsatellites and with the α s1 -casein locus.This study contributes to the global process initiated by the FAO to document the world's Animal genetic resources.

Sampling
The sampling process is of great importance since it will determine the kind of inferences which can be made.In order to reflect the current genetic composition and to try to ensure that the animals chosen were unrelated, individuals were sampled in herds from different locations (breeding centres and surrounding areas of the main goat breeds of West Africa).Between 74 to 100 blood and milk samples were collected per breed in lactating goats from at least 30 different villages with an average distance of two to 10 km.Within these samples, 20 were chosen for microsatellite typing procedures (one per village).
Dwarf samples were provided from Kolda (Casamance, Southern Senegal) for the Casamance Goat and from Fouta Djallon (Guinea) for the Labe Goat.The three Sahel Goat were provided from Djoloff (Central Senegal); from Maradi (Niger) and from Gorgol (Mauritania).The red Sokoto Goat and the Guera Goat were provided respectively from Maradi (Niger) and Atar (Mauritania) (Figure 1).The Casamance Goat and the Labe Goat are dwarf and trypanotolerant while the other breeds are long legged and trypanosusceptible.
Blood and milk samples were frozen and were respectively sent to the Inter-States School of Veterinary Medecine (EISMV), Dakar, Senegal and to the Laboratory of Biochemical Genetics and Cytogenetics (LGBC), INRA, France, for analysis.DNA was extracted by LABOGENA (Laboratory of genetic analysis for animal species).Microsatellites and α s1 -casein polymorphism analysis were carried out respectively in EISMV and LGBC.

DNA extraction, PCR amplification and electrophoresis of milk samples
DNA was extracted from 10 ml of frozen blood as described by Jean-Pierre (l987).Eleven microsatellites chosen from the most polymorphic ones published by Luikart et al. (1999) were tested.In our samples, only six were polymorphic: ILSTS011, INRABERN172, MAF65, OarFCB20, OarFCB48 and SRCRSP9.References and annealing temperature are described in Luikart et al. (1999).PCR was carried out in 25 µl reactions containing 1-30 ng of template DNA, 100 µM of each dNTP, 0.1 µM of each primer, MgCl 2 (25 mM), 1 U of Taq polymerase, 1×PCR Buffer (500 mM Tris-Hcl pH 7.2 at 25°C, 100 mM MgSO4, 1 mM DTT).Cycling involved initial denaturation at 95°C for 5 min, 35 cycles of 15 s at 95°C, 15 s at 55°C or 50°C and 30 s at 72°C and a final extension step at 72°C for 10 min using a Perkin Elmer Gene Amp PCR System 2400.The PCR products were typed by electrophoresis on polyacrylamide gels (10%) and allele identification was performed by silver staining (Budowle et al., 1991).
The polymorphism of milk proteins was determined by isoelectric focusing according to Mahé and Grosclaude (1993).

Statistical analysis
Allele frequencies, estimates of heterozygosities (Ht = 1-∑(⎯Pi 2 )), expected heterozygosities, F-statistics (Fis, Fit, Fst) and gene flow were calculated using the GENETIX package version 4.03 (Belkhir et al., 1998).The goat breeds studied were considered to be closely related and the main factor describing their genetic variability is random drift.Under this assumption, the Reynolds distance (Reynolds, 1983) is the most appropriate genetic distance to measure the degree of diversification (Eding et al., 1999, Laval et al., 2002).The Reynolds distances estimation, the neighbourjoining tree construction and bootstrap analysis were performed using the PHYLIP package (Felsenstein, 1993).
Prior to multivariate analysis, we tested the congruence of loci following the two step procedure developed by Moazami-Goudarzi and Laloë (2002), because a consensus representation of breed relationships is not meaningful if single markers are not congruent.First, distance matrixes among all the breeds were generated for each locus and correlations estimated using the Mantel procedure (Mantel, 1967).Next, marker congruence was investigated by performing a Principal Component Analysis (PCA) on the matrix of correlations and a Kruskal-Wallis test (NPAR1WAY procedure; SAS Institute, 2000) on rank scores of standardized distances among breeds.The correlation circle produced by the analysis gives a visual representation of marker congruity, while a significant Kruskall-Wallis test indicates that a compromise structure exists because distances are unequal among breeds.If a compromise structure exists, then a PCA using all the markers will be meaningful.The contribution of each marker in the structure of the principal components have been computed to see if these components are due to the action of a few markers only or to the whole set of markers.All computations relative to PCA were performed with ADE-4 package (Chessel et al., 2004) of R software (Ihaka and Gentleman, 1996).

Markers polymorphism
A total of 34 distinct alleles were detected from the 137 goats studied with a mean number of alleles ranging from 2.71 to 4.0 (Table 1).Among the seven loci analysed, five private alleles were detected.These alleles were present with a relatively high frequency (15%) for three of them and low frequencies (2.5%) for the two others.
Population differentiation was examined by fixation indices Fis, Fit and Fst for each locus and across all loci.On average, we observed no significant deficit of heterozygotes for each of the breeds analysed (Fis = -0.52 (p<0.001)) or for the whole population (Fit = -0.19(p<0.001)).We can thus suggest that the consanguinity of the analysed samples is probably weak.The average of genetic differentiation among breeds, measured as the Fst value, was 21% (p<0.001) with all loci contributing significantly to this differentiation.Thus, seventy nine percent of the genetic diversity in the total population could be attributed to differences among individuals.This value is similar to the values reported for Swiss goat breeds, Fst = 0.170 (Saitbekova et al., 1999), and Sub-Saharian goats Fst = 0.146 (Chenyambuga et al., 2004), Chinese pig breeds Fst = 1.63 and is higher than the values reported for horses Fst = 0.078 (Cañon et al., 2000), river buffalo breeds Fst = 0.038  (Barker et al., 1997) or cattle Fst = 0.107 (Kantanen et al., 2000).The differences in allele frequencies may be due to migration, random genetic drift or mutations.In our study, the genetic migration would have played an important role between breeds of close geographical vicinity.The gene flow ranges from 0.46 to 6.21 between pairs of breeds.The greatest gene flow (N e m = 6.21) is between the Sahel Goat from Niger and red Sokoto Goat and the lowest is between the Labe Goat and other breeds (on average N e m = 0.65).Thus, the Labe Goat has maintained an important genetic isolation from all other breeds.
α s1 -Casein polymorphism The α s1 -casein locus is characterized by a high qualitative and quantitative genetic variability.In particular, milk coagulation properties are strongly influenced by this locus (Grosclaude et al., 1987(Grosclaude et al., , 1994)).The nucleotide sequence of the whole goat α s1 -casein-encoding gene (CSN1S1) has been determined, and at least, 15 alleles have been characterized (Ramuno et al., 2004).They are distributed among seven different classes of variants (A, B, C, E, F, G and O) and associated with four different synthesis efficiencies ranging from 3.5 g/L (α s1 -Cn A, B, C) to 0 g/L (α s1 -Cn0) per allele (Grosclaude et al., 1994).In this study, among the 620 animals analysed, it is clear that α s1 -cnA and α s1 -CnB were the major alleles, while α s1 -CnC, α s1 -CnE and α s1 -CnF were rare alleles present only in some breeds (Table 2).For example, the α s1 -CnB frequency varied from 0.62 in the Guera Goat to 0.93 in the Sahel Goat from Mauritania.The relatively high frequency of the α s1 -CnE (11%) in the Guera Goat is interesting.In addition, at the DNA level the null allele is detected with a high frequency 25%, while not observed or only in one animal in the Casamance Goat and in the Sahel Goat from Mauritania (data not shown).As these alleles have been reported to be highly present in European goat breeds (Grosclaude et al., 1994), the specific distribution pattern of α s1 -casein in the Guera Goat (also known as Spanish Goat or Western Sahara Goat) could be explained by crossbreeding with Spanish breeds introduced in Mauritania throughout Western Sahara.It is important to specify that due to selection programs the frequencies of α s1 -CnA, α s1 -CnB and α s1 -CnC have considerably increased during the last decade.For example, the frequency of these groups of alleles has increased from 20% (1985) to 83% (2000) for the Alpine breed.
It is surprising to note that the most frequent variant in these African goats (α s1 -B) is also the phylogenetically oldest variants reported (Grosclaude et al., 1994, Bevilacqua et al., 2002).Even if goats have multiple maternal origins, possibly arising through multiple independent domestications, Africa may likely be one of the centres of domestication (Luikart et al., 2001).

Relation between breeds
Genetic distances were significantly unequal among breeds (Kruskall-Wallis test, x 2 = 41.55,20 d.f, p-value = 0.003), indicating the existence of a multivariate compromise structure.The correlation circle (Figure 2) suggested that the markers are congruent.Since all the markers are clustered in the same half-circle, PCA was performed.
Due to the fact that the first two principal components (PC) explained 73% of the total variation, the four others were not considered.The first PC accounts for 45% of the total variance and it clearly distinguished the Labe Goat    species, Jouy-en-Josas, France) for DNA extraction and to LGBC for its collaboration and contribution to this paper.

Figure 1 .
Figure 1.Geographical location of milk and blood samples collected in the 7 west African goat breeds of this study.

Figure 3 .
Figure 3. Scatterplot of principal component analysis of allelic frequencies of 7 west African goat breeds of this study.

Figure 4 .
Figure 4. Phenogram showing genetic similarities among seven West African goat breeds from microsatellite and α s1 -casein.This tree is constructed by the neighbor-joining method from the Reynolds distance.Numbers at the node represent the percentage of group's occurrence in 500 bootstrap replicates.

Table 1 .
Mean number of alleles per locus and average heterozygosity values of 7 West African goat breeds of this study

Table 2 .
Allelic frequencies of α s1 -casein in samples of the 7 west African goat breeds of this study α s1 -cn