Phenotypic characterization has been shown to be an accessible and easy-to-use tool in conservation and breeding programs . This could be explained in part by the high heritability of measures of size across ontogeny . Marked differences were observed in the morphological traits of the sheep breeds in this study. The present values of BW and linear body measurements of Yankasa, Uda and Balami sheep are higher than the range of values reported for Ganjam sheep in India by Arora et al. . Most researchers consider an organism’s phenotype as a multivariate set of variables and the covariation of traits an important analytical consideration . Discriminant analysis of morphometric traits is a mathematical approach that has been widely used in determining the relationships between different breeds of livestock [6, 23–25]. The most discriminating variables obtained in this study are similar to previous reports by Dossa et al.  and Vargas et al. . The present results indicate that there is significant morphological differentiation among Nigerian sheep populations. This morphological diversity pattern could be as a result of inherent genetic potential of each breed, alongside geographical isolation, ecological variation and community isolation [5, 27]. This is noticeable in the remarkable morphological differences between WAD (more adaptable to the wet, dense forest and derived savannah zones of southern Nigeria) and Yankasa, Uda and Balami sheep (more suited to the dry climatic conditions of Northern Nigeria).
The larger values reported for the conformation traits of males are in consonance with earlier reports on sheep , goats  and cattle . However, Bacchi et al.  found no sexual dimorphism in the morphometric characters considered in Lama guanicoe guanicoe in Argentina. The higher values recorded for intensively managed animals might be as a result of better nutrition and management practices as they tend to gain more attention than those reared semi-intensively. Riva et al.  reported that husbandry system was a source of variation in the body measurements of Bergamasca sheep; theirs was a case between transhumance and sedentary systems, where the former was taller, longer and showed a wider hock while the latter was wider at the chest and rump.
The three canonical functions (linear combinations of the continuous variables that summarize variation between the four sheep breeds) obtained could be used as criterion for establishing phenotypic standards for Nigerian sheep. The closeness between Uda and Balami sheep compared to their WAD and Yankasa counterparts might be as a result of near biometric convergence, which may function as a guide to genetic and evolutionary relationships between the two breeds. The longer distance between WAD and other breeds, especially Balami and Uda revealed that phenotypic differences are maintained in part by the reduction of gene flow among populations separated by large distances as well as physical-ecological barriers. The WAD sheep predominate in the trypano-endemic humid zones of Southern Nigeria. The intermediate morphology of Yankasa sheep might be maintained by natural selection on body size of individuals inhabiting intermediate or a range of different environments (this breed of sheep, which occupies a central geographical position, is more widely distributed in the country than the other three sheep breeds). Selective advantage might therefore favor Yankasa sheep from the biogeographical context, although more heterotic gains might be attained from crosses involving WAD and Uda or Balami. This is because populations are dynamic units which adapt physiologically and genetically to their environments and sensitive to, and within limits, responsive to any change in their environmental conditions. An appreciable percentage of animals were classified into their distinct breeds. However, some level of intermingling was observed between Uda and Balami, which could partly be attributed to indiscriminate crossbreeding due to geographical proximity.
Molecular genetic diversity between populations
The high number of alleles observed in Yankasa must have contributed to its adaptability to more agro-ecological zones in Nigeria compared to other sheep breeds that may confer selective advantages . Genetic variation is necessary to allow organisms to adapt to ever-changing environments with some of this variation stemming from introduction of new alleles by the random and natural process of mutation, since the frequency of occurrence of an allele changes regularly as a result of mutation, genetic drift, and selection . The number of alleles identified in this study is slightly higher than those reported by Adebambo et al.  with observed number of alleles in Yankasa, Balami, Uda and WAD being 11.5, 3.9, 5.85 and 5.05, respectively, based on relatively small sample sizes. Higher numbers of alleles may also be reflective of larger sample sizes used in this study. However, the values fall within the range reported for other sheep breeds. Arora et al. reported values for observed number of alleles in the range of 7 and 25 with a mean value of 13.96. The small average number of alleles per locus in Uda and WAD may be the result of a small number of founder animals since the degree of genetic variability among these animals will be small even in the presence of crossbreeding and mutation rates .
Different indices used in this study demonstrate the existence of genetic diversity among Nigerian sheep breeds. Variations in allelic richness were observed across all loci. Mean values among the various loci ranged from 3.66 in OarHH47 to 14.73 in OarJMP29, with mean values of 8.63. Dalvit et al.  found similar values of allelic richness with an average of 8.8, when they analyzed 10 European sheep breeds. Measures of genetic diversity based on allelic richness are considered important in conservation genetics because marker-assisted methods for maximizing the number of alleles conserved have been shown to be effective . Allelic richness may be a useful indicator of a decrease in population size or of past bottlenecks . It is also relevant in a long-term perspective, as selection limits are determined by the initial allelic composition rather than by heterozygosity . As noted by El Mousadik & Petit  the often reported numbers of alleles per locus for the whole population or averaged over subpopulations are not comparable because of the much larger sample size of the whole population.
Differences among populations are commonly quantified by the use of one of several statistics, including Wright’s inbreeding coefficient (F
) and Nei’s coefficient of gene variation (G
) . The rather high level of genetic variability in Yankasa is due to the fact that this is the most numerous sheep breed in Nigeria . These values (0.658 to 0.902) are within the range observed in other sheep breeds in other parts of the world. Arora et al.  observed 0.594 to 0.922 for sheep breeds from Southern Peninsular and Eastern regions of India. Values for Italian sheep breeds are 0.761 to 0.805  and European sheep breeds ranged from 0.538 to 0.807 . The observed heterozygosity values are generally lower than the expected heterozygosity in all the breeds and loci considered. The highest value of observed heterozygosity in Yankasa could also be attributed to its large number in Nigeria . The results of G
analysis in this study reveal that the proportion of gene variation among the breeds is still low.
The Shannon index of Nigerian sheep breed in this study revealed low species richness and evenness since all the indices were below 3.5, the mark set for high species evenness and richness . This might not be unconnected with the level of heterozygote deficiency observed among this population, possibly due to the management system . Data collected within the EU-ECONOGENE project on sheep and goat diversity in marginal areas indicate the presence of significant inbreeding in most of the breeds [7, 43]. This is likely due to poor breeding management of frequently small herds, which leads to partial isolation and fragmentation both at the local and breed levels .
Differences in the values of global F
over all loci considered in this study supports the suitability of some microsatellite markers over the others in the study of genetic diversity in Nigerian sheep breeds. Selection based on estimated breeding values rather than phenotypic merits can lead to the extensive use of a small number of elite individuals in purebred livestock populations, and could potentially increase inbreeding . It has therefore been suggested that breeds with wide range of genetic diversity are needed in the future for generating transgressive variation for quantitative trait loci mapping and developing new genotypes for particular management systems and market demands . On the average, within-breed heterozygote deficit (F
) was observed to be 33.5% whereas the total population (F
) exhibited 39.5% deficit with significant values (p < 0.05). Some investigations have reported relatively low variability in local non-selected breeds, as is the case of certain Nigerian  and Chinese sheep .
The relative higher genetic identity is probably due to the continuous crossing between populations at least in the recent past . The AMOVA results revealed that the greatest variation (60.716%) is within the individual, 30.545% among individuals within populations and 8.739 among populations which are consistent with F
results. Finally, the trend in genetic relationship between these Nigerian sheep breeds is the same for delta mu square, kinship coefficient and proportion of shared alleles. Indeed, our recent molecular characterization of Nigerian sheep using mtDNA sequences of the D-loop found 96 haplotypes, but only 5 haplotypes are common to all the breeds , showing relatively divergent haplotypes within breeds and geographical locations. This suggests that gene flow has occurred on a regional scale at some time in the recent past and that the breeds have not been subdivided by long term biogeographic barriers.
Genetic structure of the populations
The closest genetic distance between Yankasa and Balami at 0.184 and the farthest distance between Balami and WAD (0.665) may be reflective of their geographical locations in Nigeria since Yankasa and Balami are more of northern than southern breeds. This is further confirmed by the greater genetic distance between WAD and Balami observed in this study. These results are supported by Adebambo et al.  among these breeds. The pattern of differentiation revealed by the matrix of Nei’s genetic distances and the tree topology reflected the evolutionary history, geographical distributions and the gene flow among breeds. Genetic structure of a breed at any time is the result of a balance between genetic drift (founder effect and selection) and gene flow . WAD and Balami shared the highest number of alleles while the least is between Yankasa and Balami. However, the pattern of genetic distances deviated from the morphological distances obtained for the four sheep breeds. This may be related to varying sensitivity of the two distance estimates, although both methods were able to separate the southern WAD goats from their northern counterparts. A similar finding was reported in Ankole cattle  where the results of morphological analyses were not in concordance with the molecular genetic relationship results. The authors attributed this to the fact that microsatellites loci are selectively neutral whereas morphological traits are under selection.
Results from the STRUCTURE analysis revealed that varying the number of presumed ancestral populations (K) produces clusters that are consistent with the observed morphological categorization. The first level of clustering (K = 2) reflects the presence of two clusters in the four breeds examined and further evaluation revealed a third strong cluster. This result suggests that the four breeds originated from three ancestral populations which diverged as a result of several years of adaptation and domestication. Further evaluation of the clusters revealed the presence of sub clusters and admixtures which are indicative of substantial gene flow between these breeds. WAD is the only breed with a minimal case of admixture. This may be due to geographical delineation of the breed and the breeding practices of the owners of these animals predominantly in the southern part of the country. Yankasa possess alleles which are shared by the other three breeds and this is consistent with the results discussed above.
The mitochondrial data showed a different relationship in the neighbor-joining tree for the four Nigerian sheep breeds . In that tree, the first divergence was for Yankasa breed, followed by WAD and later by Uda and Balami. This can be explained by differences in the breeding patterns, by the use of dams and rams in different management schemes. The higher level of concordance of morphological and mitochondrial DNA data could mean a more ancestral relationship among the breeds revealed by mitochondrial DNA that is maternally inherited. Differences found in the microsatellite data may indicate recent crossbreeding due to geographical closeness among the sympatric breeds, especially involving males from one breed crossing with females of the other breeds.