Morphological variability within the indigenous sheep population of Benin

Knowledge of both the genetic diversity and geographical distribution of animal genetic resources is a prerequisite for their sustainable utilization, improvement and conservation. The present study was undertaken to explore the current morphological variability within the sheep population in Benin as a prelude for their molecular characterization. From November 2018 to February 2020, 25 quantitative linear body measurements and 5 qualitative physical traits were recorded on 1240 adult ewes from the 10 phytogeographic zones that comprise the three vegetation zones of Benin. Fourteen morphological indices were calculated based on the linear body measurements. The collected data were first analyzed using multiple comparisons of least-square means (LSmeans), followed by generalized linear model (GLM) procedures, to explore the relationships among the measured morphometric traits and the 10 phytogeographic zones. Next, the presence of any genetic sub-populations was examined using multivariate analytical methods, including canonical discriminant analysis (CDA) and ascending hierarchical clustering (AHC). Univariate analyses indicated that all quantitative linear body measurements varied significantly (P<0.05) across the phytogeographic zones. The highest values (LSmean± standard error) of withers height (68.3±0.47 cm), sternum height (46.0±0.35 cm), and rump height (68.8±0.47 cm) were recorded in the Mekrou-Pendjari zone, the drier phytogeographic zone in the North, whereas the lowest values, 49.2±0.34, 25.9±0.26, and 52.0±0.35 cm, respectively, were recorded in the Pobe zone in the South. Multivariate analyses revealed the prevalence of four distinct sheep sub-populations in Benin. The sub-population from the South could be assimilated to the short-legged and that from the North to the West African long-legged sheep. The two other sub-populations were intermediate and closer to the crossbreeds or another short-legged sub-breed. The proportion of individuals correctly classified in their group of origin was approximately 74%. These results uncovered a spatial morphological variation in the Beninese sheep population along a South-North phytogeographic gradient.

Introduction sustainable use, management and conservation. To date, neither of these characterization tools have not been covered in depth to describe the diversity existing within the Beninese sheep population. Hence, in this study, to further document this existing diversity and to explore the actual spatial distribution within the indigenous sheep population of Benin, we primarily characterized their morphology based on a large panel of collected morphological/phenotypic traits. We hypothesized that the sheep population of Benin is highly diverse and unevenly distributed according to ecological conditions. The current study aimed to establish the relationships among sheep morphometric traits and the 10 phytogeographic zones of Benin using univariate analyses and then explore the presence of sheep sub-populations in the Beninese indigenous sheep population using multivariate analyses. The findings of this study will provide the basis for a comprehensive molecular study on the same samples, based on both simple sequence repeat and single nucleotide polymorphism marker genotyping. Morphological data could then be compared with molecular data and association analyses (i.e., genome-wide association studies) to appropriately address possible breeding strategies for the indigenous sheep population of Benin.

Study area
This study was conducted in the 10 phytogeographic zones (Fig 1) that comprise the three vegetation zones of Benin [22,23], namely the Guinea-Congolian (GCZ), the Guineo-Sudanian transition (GSZ) zone and the Sudanian zone (SZ). The characteristics of the 10 phytogeographic zones, such as climatic conditions, temperature, humidity index, soil characteristics, and predominant vegetation, are presented in Table 1.

Sampling procedure
A longitudinal survey was conducted from November 2018 to February 2020 in the 10 phytogeographic zones of Benin. In each zone, two to five communes were selected depending on the presence or abundance of sheep flocks. Thirty-two out of the 77 communes of Benin were included in the survey. At least four distinct villages were randomly chosen from each commune. In each village, 5 to 20 individual flocks were selected based on farmers' willingness to participate in the study. Approximately four or more unrelated animals were sampled per flock based on farmers' knowledge of the individual animals present in their sheep flocks. Thus, a total of 1240 ewes that were at least two years old and multiparous (at least two lambings) were randomly selected, described and phenotypically characterized. The age of animal estimated by the farmers was ascertained by examining their teeth according to the procedure described in [25,26]. The sampling distribution across the vegetation and phytogeographic

PLOS ONE
zones is presented in Table 1. All individuals sampled in a phytogeographic zone were considered as a sub-population.

Data collection procedure
A total of 25 quantitative linear body measurements (Fig 2 and Table 2) and 5 qualitative physical traits drawn from the FAO guidelines [25] and from a previous study [27], were used to describe the morphological characteristics of each animal. To minimize collecting biases, all measurements were taken by a young researcher and a trained field assistant. The live bodyweight of each animal was measured using a scale. The other 24 body measurements were taken using a flexible measuring tape and a measuring stick, early in the morning before the animals were fed to avoid biases on certain traits due to feed intake. In addition, the reproductive history of each sampled animal, including the number and type of parities (single, twins, triplet and quadruplets), was recorded from its owner. The geographical position of the herds in which the sheep individuals were sampled was recorded using a Garmin GPS (etrex vista TM).

Data analysis
Fourteen morphological indices (Table 3) were calculated based on the collected quantitative linear body measurements (or morphometric traits). All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Descriptive statistics for the quantitative linear body measurements and qualitative physical traits were obtained using the procedures PROC UNIVARIATE and PROC FREQ, respectively. The frequencies and Pearson chi-

PLOS ONE
Indigenous sheep of Benin square (χ2) tests were used for qualitative physical traits to explore the relationships among qualitative variables. The least-square means (LSmeans), their standard errors (SEs), and the coefficients of variation (CVs) of the morphometric traits were calculated for each phytogeographic zone. The comparison of LSmeans between phytogeographic zones was performed using Tukey's test multiple mean comparison tests. Subsequently, the general linear model procedure (PROC GLM) was used to analyze the relationship between phytogeographic zones and morphometric traits. A stepwise discriminant analysis was performed using PROC STEPDISC to identify the most useful morphometric traits and morphological indices for further discriminant analyses. The relative discriminatory ability of a quantitative variable was assessed using the partial Rsquare, F value, and level of significance (Pr>F). Then, the canonical discriminant analysis (CDA) function (PROC CANDISC) was used to perform univariate and multivariate one-way analyses, derive canonical functions and linear combinations of the quantitative variables that summarize variation between populations, and calculate the associated Mahalanobis distances. The ability of the computed canonical functions to classify each individual animal into its a priori phytogeographic zone was measured using the discriminant procedure (PROC DISCRIM). The degree of morphological similarity or dissimilarity among individuals from the different phytogeographic zones was determined based on the ascending hierarchical clustering (AHC) analysis procedure (PROC CLUSTER). The PROC TREE procedure was used to build a dendrogram displaying the interrelationships among individuals within and across phytogeographic zones. Finally, a multiple correspondence analysis (MCA) using the PROC CORRESP procedure was used to associate the qualitative physical traits with the phytogeographic zones.

Relationships among sheep morphometric traits and the 10 phytogeographic zones of Benin using univariate analyses
The result of the univariate analysis showed significant differences (P<0.05) among the 10 phytogeographic zones for all measured quantitative morphometric variables (S1 Table) and the
The proportion of birth type varied significantly (p<0.0001) among the 10 phytogeographic zones (S3 Table). Irrespective of the zone, single-born lambs were the most common. The highest percentages of twin-born lambs were recorded in the Oueme Valley, Pobe, and Zou zones, whereas the highest proportions of triplets and quadruplets were recorded in the Pobe zone. The percentage of multiple births appeared to increase with the parity number of ewes.

Identification of sheep sub-populations using multivariate analyses
The results of the stepwise discriminant analysis (Table 5) showed that 38 out of the 39 quantitative variables (i.e., 25 quantitative linear body traits and 14 morphological indices) included in the analysis significantly contribute to discrimination between the phytogeographic zones (P<0.0001). The traits rump width (RW) and sternum height (SH) showed higher partial R 2 and F values, illustrating their greater discriminant power than the other variables used to assess the morphological diversity in the Benin sheep population. Nevertheless, the use of the 32 significant (P<0.0001 for column Pr > F) quantitative variables (i.e., 22 quantitative linear body traits and 10 morphological index) in the CDA generated two significant (P<0.0001) canonical variables (CAN 1 and CAN 2) that explain 76% of the total variation, as revealed by the standardized coefficients for the discriminant function, the canonical correlation, the eigenvalue, and the share of total variance taken into account (Table 6). Canonical loadings that measure the simple linear correlations between each independent variable and canonical variables are reported in Table 6. The plot of the centroid values of the first two canonical discriminant functions (CAN1 and CAN2) showed many distinct and homogenous sheep sub-populations with overlapping events (Fig 3).
The Mahalanobis distances among the 10 phytogeographic zones are presented in Table 7. All pairwise distances were significant (P<0.0001). The two largest measured squared Mahalanobis distances were between the Mekrou-Pendjari and Pobe zones (69.02) and between the Oueme Valley and Bassila zones (61.65). The closest distance (2.85) was between the Chaîne Atacora and Borgou-Nord zones. The discriminant functions accurately classified a relatively high proportion (74.59%) of the individual sheep into their a priori group (Table 8).
Based on the squared Mahalanobis distances, AHC generated a dendrogram that indicated four distinct sub-groups or sub-populations of sheep (Fig 4). The first sub-population was composed of Borgou-Nord and Chaîne Atacora zones joined by the Borgou-Sud zone, the second was only composed of the Mekrou-Pendjari zone, the third was composed of the Pobe and Costal zones joined by the Bassila zone, and the fourth was composed of the Oueme Valley, Zou, and Plateau zones.

Multiple correspondence analysis of sheep qualitative traits
Multiple correspondence analysis (MCA) highlighted the association between the different qualitative physical traits and phytogeographic zones (Fig 5). The first two dimensions (Dim 1 and Dim 2) explained 68.88% and 18.21% of the total variation, respectively. On the righthand side of the plot, the Zou, Plateau and Oueme Valley zones were closely associated with sheep with a straight back, long hair, a predominant coat color of spotted white and brown patterns. Animals from the Pobe, Coastal, and Bassila zones were characterized by back profile slopes up towards the rump and either a plain/uniform white or a composite coat color with predominantly spotted black or white patterns. Moreover, the left-hand side shows that the Mekrou-Pendjari zone was plainly associated with sheep that had dropping ears, flush hairs, an ultra-convex head profile, and pie-black or pie-brown coat color. Conversely, the zones of Borgou-Sud, Borgou-Nord and Chaîne Atacora were associated with sheep that have a convex head profile, a dipped back profile and a coat color with predominantly spotted white and brown patterns (Fig 5).

Discussion
In this study, we aimed to further document the existing diversity and spatial distribution within the sheep population raised in Benin based on a large panel of qualitative and quantitative traits. Univariate analyses revealed significant differences among phytogeographic zones for all measured morphological traits and derived indices, suggesting the possible influence of these zones on the evolutionary adaptation of the sheep population in terms of these  morphological traits. This result is in line with the finding of a previous study [16], who reported a significant impact of the breeding area on morphological traits in the sheep population from Ivory Coast. The mean values of thoracic development (TD), greater than 1.2,   indicated that all measured animals had good thoracic development, regardless of the phytogeographic zone. However, following the classification based on the body index (IBR) [27], the sheep from the phytogeographic zones of Borgou-Nord, Chaîne Atacora, Plateau, and to some extent Borgou-Sud, can be considered as of the brevigline breed (mean values of IBR < 0.85), whereas those of Bassila, Mekrou-Pendjari, Pobe, Oueme Valley, Coastal and Zou zones are of the medigline breed (mean values of IBR > 0.85). The overall calculated cephalic index (IC) of 51.20% indicated that sheep from Benin are dolichocephalic, regardless of the phytogeographic zone. This result is considerably lower than that reported in a previous study [35] for sheep breeds in Nigeria. Based on the main qualitative traits generally used for breed description (i.e., coat color, facial profile, ear orientation), the composite coat color with a dominance of spotted white associated with both straight facial profile, long hairs, and erected ears, most frequently recorded in the Southern zones (i.e., BZ, PlZ, PoZ, CZ, ZZ and VOZ) are physical characteristics of Djallonké/WAD sheep, as described in previous studies [10,16,36]. This result suggests that the sheep in the aforementioned zones are closer to the Djallonké sheep. In contrast, specific physical traits of the long-legged Sahelian sheep breed, such as convex facial profile, short hair, and dropped ears orientation, were more common in the sub-populations surveyed in the Northern zones (i.e., BSZ, BNZ, CAZ, and MPZ). Some individuals from the Sahelian sheep breed presented a bicolored coat (the front being brown or black and the rear white), mainly in Mekrou-Pendjari (MPZ). These characteristics are specific to Fulani sheep, also known as Oudah. Notably, in these Northern zones, some of the long-legged Sahelian sheep breed's physical characteristics were also observed in crossbreeding products between Sahelian and West African Dwarf sheep. This distribution of sheep populations along the South-North gradient was confirmed by the results of the multiple correspondence analysis for qualitative physical traits. This finding could be explained either by the increasing introgression of Sahelian longlegged sheep from the Sahel through transhumance and trade or by the selection pressure for specific traits. The mobility of herders with diverse sheep breeds in West Africa could favor genetic introgression and be a dynamic factor of animal genetic diversity. According to a recent report on sheep transhumance between Niger and Benin [37], transhumant sheep herders move from Niger towards localities of the Alibori department in North Benin and stay for about six months to ensure feeding and watering of their animals. Therefore, many exchanges are made between transhumant and local breeders favoring the mixture of Sahelian or long-legged type sheep (i.e., Oudah, Bali-Bali, and Balami) with WAD sheep. The distribution gradient could also result from an adaptive response to changing local environments since morphological adaptations (body size and shape, coat and skin color, and hair type) are physical changes in the animal that enhance its fitness in a given environment [38,39]. The multivariate discriminant analyses confirmed the significant morphological variability among sheep from most of the 10 phytogeographic zones of Benin. The proximity between sheep from CZ, PoZ, and BZ and their farness from those of VOZ, MPZ, BNZ, BSZ, CAZ, ZZ, and PlZ (Fig 3) could be explained by the geographic proximity and exchange of animals, as well as by similarities or dissimilarities in their ecologies [7]. Nonetheless, the low proportion of individuals from CAZ (55.4%) and BNZ (64.5%) correctly classified in their origin groups, as well as the low values of Mahalanobis distance between these two zones, revealed some overlap among sheep breeds of these phytogeographic zones with those from other zones except for BZ, PoZ, CZ, and VOZ. These overlaps could be the result of crossbreeding, especially in the BNZ and CAZ phytogeographic zones that host transhumant sheep flocks. Additionally, the subdivision of the sheep population into four sub-populations in the hierarchical cluster analysis (Fig 4) seems to reflect differences in the type of vegetation, climate, and humidity among phytogeographic zones. This result is likely to confirm the effect of environmental factors on the morphology of sheep [40][41][42][43] and transhumance and management practices. Many convergences with other reports on the measured morphological traits were found in the current study. For example, the mean values of the two quantitative body linear traits, WH and HG, obtained for the sheep from the Southern zones (i.e., PlZ, ZZ and VOZ), (53.2±3.40 cm and 60.9±1.67 cm, respectively), were highly similar to previously reported values (52.3 ± 1.07 cm and 65.0 ± 1.60 cm, respectively) for the Djallonké/WAD sheep in Nigeria [44], and to the WH mean value of 56.5± 0.22 cm reported for the Djallonké/WAD sheep in Burkina Faso [31]. Therefore, the sheep sub-population found in these zones appeared to be an ecotype of the Djallonké sheep (WAD). Likewise, the mean value of these traits (HW and HG) obtained for the sheep from PoZ, CZ and BZ (52.9±5.07 cm and 73.6±3.72 cm, respectively) were highly similar to those (54.6 ± 8.23 cm and 74.7 ± 8.28 cm, respectively) reported for the Djallonké/ WAD sheep in Togo [36]. Thus, the sheep from PoZ, CZ, and BZ might represent another ecotype of Djallonké/WAD sheep with a size relatively larger than the first sheep sub-group of PlZ, ZZ, and VOZ. As for the sheep population from MPZ, which is dominated by individuals with physical characteristics of the Sahelian long-legged sheep, the mean value of WH obtained in the current study (i.e., 68.4±0.47 cm) was similar to that (69.1 ±0.12 cm) reported for the Sahelian sheep in Burkina Faso [31]. The mean values of WH and HG for the sheep from BNZ, CAZ, and BSZ (61.2±0.92 cm and 68.9±0.90 cm, respectively) were intermediate between those for the sheep from MPZ and each of the two other groups of Southern zones (PlZ, ZZ, and VOZ; PoZ, CZ, and BZ) (Figs 3 and 4), suggesting that these zones may be considered as very favorable zones for crossbreeding. This result also suggests the co-existence of several sheep morphotypes in these zones.
This study highlights a highly diverse sheep population in Benin, as in other African countries (e.g., Burkina Faso, Ivory Coast, Togo, and Nigeria), within which the distribution of individuals is affected by natural and also anthropogenic factors. Thus, the sheep subgroups observed in the different phytogeographic zones of Benin also exist in other African countries in similar or different environments [16,31,36,44]. The most important natural factors at the origin of the recorded sheep diversity across the 10 investigated phytogeographic zones might be climate-related factors (temperature, humidity, and/or vegetation cover), which affect the availability of feed resources and induce natural selection pressures. Anthropogenic factors mainly concern animal management practices in different zones, cultural preferences, and livestock marketing systems. Thus, the phenotypic traits (small size, stocky appearance, small ears, and long hair) of the Djallonké/WAD sheep, which are predominantly found in Southern Benin, are likely a response to natural selection over several generations under the influence of the constraints of the environment in which the animals are raised. Furthermore, the larger phenotypic traits of the Djallonké ecotype in the PoZ, CZ, and BZ could be explained, in addition to the influence of the environment, by changes in sheep farmers' breeding practices in these phytogeographic zones, especially the practice of crossbreeding short-legged with long-legged animals from the North. This is undoubtedly influenced by the annual flow of Sahelian animals to these regions during the Aid El-Kebir cultural ceremonies when sheep sacrifice takes place in Muslim households. In addition, the breeders of these areas would try to adapt to new consumer demands, as expressed by their preference for animals that possess larger physical features than the Djallonké during ceremonies and festivals. Likewise, the Sahelian sheep, which are predominant in the MPZ in northern Benin, are larger and slender, with varied but predominantly light coats, short hair, a long tail, dropped, and larger ears. Several hypotheses about the adaptive value of these traits have been put forth. For instance, [45] argue that these specific traits might allow them to reflect solar radiation better, and thus, to be less prone to heat stress. In addition, according to these authors their long legs might predispose them to travel long distances when searching for pastures. Moreover, their large height might allow them to feed easily in tree and shrubs savannah pastures, which are predominant in these regions [22,23]. But confirmation of these hypotheses requires further study and remains inconclusive. The BNZ and CAZ, with their intermediate climatic gradient between the humid south and the dry north, promote, on the one hand, the extension of the distribution area of the Sahelian types, and on the other hand, the cross-border sheep transhumance practices that are at the origin of the admixture of sub-populations observed in these two zones. Referring to transhumance, it is worth noting that during the migratory period, and to meet their own subsistence needs, transhumant sheepherders often sell or exchange a few heads of animals in their herds for food and salt [37]. In contrast, the attraction of certain sheep farmers for large animals in areas hosting transhumantherds sometimes encourages them to herd their animals to the same grazing areas in the hope of mating their animals with those kept by the transhumant herders.
Although morphological variation is largely under genetic control [30], it is subject to the influence of the environment and management practices [46,47]. Thus, the preservation of local populations that adapt to their environment is essential. This calls for the development of new management strategies for sheep farming in Benin as well as in other African countries aiming to improve farm profitability by improving animal performance while preserving the diversity within the local sheep populations. In this way, sheep farming would overcome current and future challenges in production systems in Africa, including climate change and market demand.

Conclusion
This study aimed to explore the morphological variability of indigenous sheep reared in different phytogeographic zones of Benin. The results showed significant variations in phenotypic traits, both qualitative and quantitative, among phytogeographic zones. Four sheep sub-populations were identified. Animals in the phytogeographic zones of Southern Benin could be identified as short-legged (Djallonké/WAD) sheep, whereas those from the zones located in the northern regions of the country were much closer to the long-legged Sahelian sheep breed. The intermediate sub-populations included an ecotype of Djallonké/WAD sheep and various crossbreeds. These results could be due to several factors, such as adaptation of animals or natural selection, changes in farmers' breeding practices, and gene flow. Further research is ongoing to better understand the genetic, environmental, and socio-economic determinants of these recorded morphological variations. Thereafter, a breeding program could be developed and implemented for better management of the diversity existing between and within recorded sheep sub-populations and for the sustainable production of this livestock species in Benin.