Indigenous sheep are widely distributed in the tropics and subtropics due to their unique adaptive features that enable them to fit in a wide variety of environments (Berihulay et al., 2019). In Botswana, the indigenous sheep constitute about 195 000 of the 300 000 national flock (Botswana Statistics, 2016) and is adapted to different geographical regions of the country. It contributes significantly to the livelihoods of resource poor farmers by providing meat and milk as a source of nutrition to the household, and income which is used for a wide range of economic activities. Indigenous Tswana Sheep retain certain adaptive features such as drought, heat and disease tolerance (Nsoso et al., 2004b). Very little has been done towards improvement and characterization of Tswana Sheep. The major threat facing indigenous Tswana sheep genetic resources is uncontrolled breeding with exotic breeds or breed replacement with exotic breeds carried out in an endeavor to improve the breed to meet current market demands for more mutton.

It is therefore very important to develop strategies for sustainable utilization of Tswana sheep genetic resources. The prerequisite to developing these strategies is the characterization of the genetic resource under its natural environment (Msanga et al., 2012; Monau et al., 2018). Phenotypic characterization of local genetic resources is essential for conservation, breed inventory and monitoring, policy formulation and design of breeding programmes (Baker and Gray, 2004). Phenotypic characterization of indigenous Tswana sheep was undertaken more than a decade ago and may not reflect the current situation due to changes in production systems and within population’s changes resulting from evolutionary forces (Sölkner et al., 1998). Morphological and productive aspects in a population evolve over time as a result of natural and artificial selection, mutation, migration, and random genetic drift (Song et al., 2006). Differences in environment and differences in climatic factors such as rainfall and temperatures also influence adaptive features of local populations which might result in phenotypic differences in the general population (Schierenbeck, 2017). There is therefore need to carry out routine inventories and monitoring of the indigenous Tswana sheep genetic resource. The aim of this was therefore to phenotypically characterize indigenous Tswana sheep under its native environment and to develop a prediction equation for body weight by using linear body measurements in the Kgatleng, Kweneng, Southern and South-East districts of Botswana.

Description of the study area

The study was conducted in Kgatleng (24°15'S 26°30’E with total land area of 7,960 km²) Kweneng (24°00'S 25°00'E with total land area of 35,890 km²), South East (25°00'S 25°45'E with total land area of 1,780 km²) and Southern (25°00'S 25°00'E with total land area of 28, 470 km²) districts of Botswana from November 2020 to January 2021 (Figure 1). The climate in the four districts is mainly semi-arid with high temperatures occurring from October to April and low temperatures occurring from around May to August. Rainfall is low, unreliable highly variable from one year to the next and not evenly distributed. The vegetation type in all districts is savannah with tall grasses, bushes and trees (Makhabu et al., 2002; Nsoso et al., 2004b).

Where: Y_ijk = Observed body weight or linear measurements    

µ = Overall mean

A_i = the fixed effect of i^th age groups (i = 0PPI, 1PPI, 2PPI and ≥ 3PPI) 

D_j = the fixed effect of j^th district (j= Kgatleng, Kweneng, South-East and Southern    districts)

S_k = the fixed effect of the k^th sex (k= male, female)

(AxD)_ij= the effect of the interaction of i of age group with j of district

 (AxS)_ik= the effect of the interaction of i of age group with k of sex   

 (DxS)_jk= the effect of the interaction of j of district with k of sex

e_ijk= random residual error

Model used for the least square mean analysis in males for scrotal circumference was:

Y_ijk =µ + A_i + D_j + (AxD)_ij + e_ij

Where: Y_ijk = Scrotal circumference    

µ = Overall mean

A_i = the fixed effect of i^th age groups (i  =  0PPI,  1PPI,  2PPI  and  ≥ 3PPI) 

D_j = the fixed effect of j^th district (j= Kgatleng, Kweneng, South-East and Southern districts)

 (AxD)_ij= the fixed effect of the interaction of i^th age group with j^th of district

e_ijk= random residual error

Correlations and regression

Pearson’s correlation coefficients of indigenous Tswana sheep were estimated between body weight and other linear body measurements (LBMs) within each sex using the procedure correlation (PROC CORR) of Statistical Analysis System (SAS release 9.1 2003)  to describe the strength and direction of relationships between the response variable (live body weight) and explanatory variables (LBMs). Body weight and other LBMs (BL, HG, WH, RH, SW, EL, RL, CBC, CBL, NL, RW, HW, HL, TL, TC and SC) were included for males whereas SC was excluded when calculating correlations coefficients for female sheep. Based on the correlations of body weight with other LBMs, a stepwise regression procedure (PROC REG) of Statistical Analysis System (SAS release 9.1 2003) was then used to regress body weight for each sex in order to determine the best-fit regression equation for the prediction of body weight using LBMs. The best-fit models were selected based on the coefficient of determination (R²) and the simplicity of measurements of the LBMs under field conditions. The following models were used for the analysis of multiple linear regressions.

For males:

Υ_j =β₀ + β₁X₁ + β₂X₂ + β₃X₃ + β₄X₄ + β₅X₅ + β₆X₆ + β₇X₇+ β₈X₈ + β₉X₉+ β₁₀ X₁₀ + β₁₁X₁₁ + β₁₂X₁₂ + β₁₃X₁₃ + β₁₄X₁₄ + β₁₅X₁₅ + β₁₆X₁₆ + ej

Where:

Y_j =the response variable (body weight)

β₀ = the intercept

X₁, X₂, X₃, X₄, X₅, X₆, X₇, X₈, X₉, X₁₀, X₁₁, X₁₂, X₁₃, X₁₄, X₁₅ and X₁₆ are the explanatory variables BL, HG, WH, RH, SW, EL, RL, CBC, CBL, NL, RW, HW, HL, TL, TC and SC, respectively.   

β₁, β₂…β₁₆ are the regression coefficients of the variables X₁, X₂…X_16.

e_j = the residual random error.

For female:

Υ_j =β₀ + β₁X₁ + β₂X₂ + β₃X₃ + β₄X₄ + β₅X₅ + β₆X₆ + β₇X₇+ β₈X₈ + β₉X₉+ β₁₀ X₁₀ + β₁₁X₁₁ + β₁₂X₁₂ + β₁₃X₁₃ + β₁₄X₁₄ + β₁₅X₁₅ + ej

Where:

Y_j =the response variable (body weight)

β₀ = the intercept

X₁, X₂, X₃, X₄, X₅, X₆, X₇, X₈, X₉, X₁₀, X₁₁, X₁₂, X₁₃, X₁₄ and X₁₅ are the explanatory variables BL, HG, WH, RH, SW, EL, RL, CBC, CBL, NL, RW, HW, HL, TL, and TC, respectively.  

β₁, β₂…β₁₆ are the regression coefficients of the variables X₁, X₂…X_15.

e_j = the residual random error.

Table 1 shows some qualitative characters of both male and female indigenous Tswana sheep raised in Kgatleng, Kweneng, South-East and Southern districts of Botswana. There was variation on coloration patterns amongst the sheep populations with predominantly plain coat colour patterns (81.25%) across districts, followed by patchy and spots of different colours (Figure 2). The results are consistent with the findings of Asaminew et al. (2016), Edea et al. (2010) and Tibbo and Ginbar (2004) for Bonga, Horro and Woliata sheep types of Ethiopia. The higher proportion of animals with white dominated coat colour and plain white coat colour could be a reflection of natural selection for animals manifesting white colour to withstand the hot environment of Botswana. This observation is different in other part of the African continent such as Ethiopia where Hailemariam et al. (2018) reported the dominance of black coat colour for Gamogofa sheep which helps in absorption of solar radiation to maintain an optimum body temperature in the cold Gamogofa zone.  

Most of sheep across the districts had a characteristic short fat tail with a straight tip pointing downwards (Figure 2). The fat tail is an adaptive attribute that serves as an energy reserve to enable indigenous Tswana sheep to adapt and survive feed fluctuation periods throughout the year (Ermias et al., 2002). This characteristic has also been reported in some South African sheep breeds including the South African Namaqua Afrikaner, the Zulu and the Pedi sheep (Soma et al., 2012). To the contrary, Getachew et al. (2010) reported short fat tails curved upwards in Menz sheep of Ethiopia. Contrary to the current study, Giwaz et al. (2007) and Edea et al. (2010) who reported a long fat tail characteristic in Adilo and Bonga sheep of Ethiopia. Differences between tail types is associated with genetic variations of sheep types (Ermias et al., 2002). The predominant ear form or orientation observed in indigenous Tswana sheep (overall 98.65%) was horizontal ear orientation. Almost all sheep across districts (97.2% for Kgatleng, 100% for Kweneng, 100% for South-East and 98.3% for Southern district) had no wattles. The findings are similar to Gamogofa sheep of Ethiopia (Hailemariam et al., 2018) and a lower proportion of Tswana sheep with wattles in Kgatleng and South-East district is consistent with Melesse et al. (2013) and Tibbo and Ginbar (2004) for Bonga sheep of Ethiopia. A comparatively high proportion of indigenous Tswana sheep across districts did not have horns and a high proportion of sheep with horns had their horns curving backwards as opposed to straight (Table 1).

Table 2 shows the flock structure of indigenous Tswana sheep in Kgatleng, Kweneng, South-East and Southern districts of Botswana. Generally, the proportion of female sheep increased with age, hence females out-numbered males (rams and castrates) across all age groups except for the 6-11 month age group (Table 2). This is because selection of breeding rams is mostly done after 12 months of age after the rams had reached sexual maturity. Rams that have not been selected for breeding are mostly castrated or culled while almost all females are retained for breeding purposes. A few males are selected for breeding purposes based on favourable desired traits while culls are sold for meat production or to other farmers who further do some selection to suit their farming needs (Nsoso et al., 2004b). Again, the traditional meat market also requires animals of an older age group (>12 months) to attract a favourable price (Nsoso and Madimabe, 1999). Similar findings have been previously reported by Katongole et al. (1996) on goats and Nsoso et al. (2004b) on goats and sheep.  

Body weight and linear measurements

Age effect

In this study, body weight and most quantitative linear body measurements were significantly (P<0.05) affected by the age group except cannon bone circumference (Table 3). All body measurements   increased as the animal advanced in age from the 0PPI age group to the ≥3PPI age group. This is expected and has been echoed by several scholars who observed that age had a significant effect on body weight and other linear body measurements and that the size and shape of sheep increased as the animal advanced with age, until optimum growth or maturity (Fasae et al., 2006; Yoseph, 2007; Tesfaye, 2008; Asaminew et al., 2016).

District effect

The least square means and standard errors for the effect of district on live body weight and other linear body measurements of indigenous Tswana sheep are presented in Table 3. The district effect was significant (P<0.05) for body weight and most linear body measurements except cannon bone circumference. The significant district effect  on body weight and linear body measurements is consistent with Asefa et al. (2017), Alemayehu (2011) and Kunene et al. (2007). Southern district sheep were the heaviest (38.93±0.55 kg) and Kweneng sheep were the lightest (34.14±0.53 kg). Generally, Tswana sheep were comparable in body weight to Zulu sheep (39.76 to 40.26 kg) (Kunene et al., 2007) and heavier than several indigenous sheep of Ethiopia (Asaminew et al., 2016; Michael et al., 2016; Mohammed et al., 2017). Tswana sheep are however lighter than Balami and Uda sheep types from South, Middle belt and North West districts of Nigeria (Agaviezor et al., 2012).

Southern district sheep had significantly (P<0.05) higher heart girth than sheep in Kgatleng, Kweneng and South-East districts of Botswana (Table 3). The difference in heart girth between sheep in different regions has also been reported by Asaminew et al. (2016) who found higher heart girth in Soddo Zuria and Damote Gale sheep than Damote Sore sheep of Ethiopia. Generally Tswana sheep had similar heart girth with Uda sheep of Nigeria (Agaviezor et al., 2012) and Hulet eju sheep in Ethiopia (Michael et al., 2016). The heart girth of Tswana sheep in the Southern district was higher than the heart girth of sheep in Wogide, Borena and Legambo districts of Ethiopia (Mohammed et al., 2017) and Sinan and Hulet eju sheep of Ethiopia (Michael et al., 2016). Southern district Tswana sheep had the longest body length than sheep from other districts while Kweneng district sheep were the shortest. Body length of Tswana sheep in Kgatleng, Kweneng and Southern sheep were similar to those of Gozamen, Sinan and Hulet-eju sheep of Ethiopia (Michael et al., 2016) and longer than that of Borena and Legambo sheep of Ethiopia (Mohammed et al., 2017). Generally, Tswana sheep across the districts were shorter than Soddo Zuria and Damote Gale of Southern Ethiopia (Asaminew et al., 2016).

South-East and Southern district Tswana sheep had similar wither height and were significantly (P<0.05) taller than Kgatleng and Kweneng sheep which also had similar wither height. South-East and Southern district Tswana sheep had similar wither height to Borena sheep of Ethiopia (Mohammed et al., 2017) and Agarfa sheep of Ethiopia (Asefa et al., 2017). Generally, Tswana sheep are taller than Metta, Gorogutu and Deder sheep of Ethiopia (Shibabaw, 2012) and Damote Sore sheep of Ethiopia (Asaminew et al., 2016) and shorter than Gozamen, Sinan and Hulet eju sheep of Ethiopia (Michael et al., 2016). The differences in body weight and other linear body measurements could be as a result of differences breed structure emanating from in the influences of evolutionary forces of the world. Other differences in body weight and other linear body measurements could be due to nutritional and management practices between districts.  In females, the pregnancy status of the animal could be another reason for variations in body weight and other linear body measurements especially body length (Kunene et al., 2007). The wide variations in body weight and other linear body measurements of sheep between districts indicate healthy diversity which could be exploited in genetic improvement programmes of indigenous Tswana sheep (Berhanu and Haile, 2009).

Kgatleng and Kweneng rams had similar scrotal circumference that were significantly higher than scrotal circumference of sheep in South-East and Southern districts. The significant influence of district on scrotal circumference of Tswana sheep found in the current study is contrary to Mohammed et al. (2017) for Wogide, Borena and Legambo rams of Ethiopia and Michael et al. (2016) for Gozamen, Sinan and Hulet eju rams of Ethiopia who reported a non-significant influence of district on ram scrotal circumference. Scrotal circumference of Tswana sheep in Southern and South-East districts of Botswana is similar to that of Wogide and Legambo rams (Mohammed et al., 2017) and Zulu rams (Kunene et al., 2007). Generally, the scrotal circumference of Tswana rams was higher than that of Borena rams of Ethiopia (Mohammed et al., 2017) and Soddo Zuria, Damote Gale and Damote Sore rams of Ethiopia (Asaminew et al., 2016). Differences in scrotal circumference between breeds and districts might be due to the fact that testicular size varies with breed, age of an animal and season/time of the year (Söderquist and Hulten, 2006). Kunene et al. (2007) reported larger scrotal circumference in autumn and in summer compared to winter and spring in Zulu rams, probably because of fluctuations in fodder quality and quantity with changes in times of seasons of the year. Furthermore, Dana et al. (2000) reported reduced scrotal circumference by up to 10% in Ethiopian highland sheep fed low quality diet than those fed good quality diet. Low quality diet causes loss of fat from scrotal tissue of rams resulting in reduced testicular size (Coulter and Kozub, 1984).

Age by sex interaction

The least square means and standard errors for the effect of sex, age group and their interaction on body weight and other LBMs of indigenous Tswana sheep are presented in Table 3. The interaction of sex and age group was significant (P<0.05) for body weight and most LBMs (BL, HL, HW, SW, CBL, HG, WH, RL, RH, RW, TL and TC). However, the interaction effect was not significant (P>0.05) for cannon bone circumference and ear length, implying that these parameters were not affected by the sex X age group interaction in the current study. In consonance with Tassew (2012) the sex X age group interaction was significant (P<0.05) for most LBMs except ear length in Habru and Gubalafto sheep. Contrary to the current findings, Michael et al. (2016) and Alemayehu  (2011)  reported  a   significant   sex  by  age group interaction (P<0.01) only for body weight in Dawro and Konta Special Woreda zones of Ethiopia. Kunene et al. (2007) also reported a non-significant sex by age interaction for LBMs between Zulu rams and ewes lambs of South Africa at their milk stage. 

The sex by age interaction found in the current study revealed that the differences in live body weight between males and females increased with the age of the animals from 4.47 to 5.20 kg and 7.90 kg at one, two and three pairs of permanent teeth, respectively (Table 3). Body weights obtained at 0PPI, 1PPI and ≥3PPI age groups in the current study were slightly higher than those previously reported by Nsoso et al. (2004b). The 2PPI age group in the current study had lower body weights than those reported by Nsoso et al. (2004b). The discrepancies might be mainly due to differences in management and evolving production systems and changing breeding goals of Tswana sheep farmers.

Generally, males were heavier and had higher LBMs than females across all age groups except ear length at 2PPI and ≥3PPI. This is consistent with Getachew et al. (2010) for Menz and Afar sheep and Tibbo et al. (2004) for Menz and Horro sheep in Ethiopia. The superiority of males over females in body weight and other LBMs might be attributed to differences in hormonal profiles between the sexes with males having hormones that promote rapid weight gain and muscularity than females, consequently resulting in superior body weight and higher LBMs in males than females (Gebreyowhens and Tasfay, 2016). In females, estrogen inhibits growth of long bones of the body resulting in slower growth rate and the reaching of puberty at a relatively smaller body size (Sowande and Sobola, 2007).

Correlation between body weight and other LBMs

Animal live body weight, linear body measurements, their interrelationships and correlations are very important in determining genetic potential for co-current improvement of traits in genetic improvement programs. The phenotypic correlation coefficients between live body weight and linear body measurements are presented in Table 4. In males, significant (P<0.05), positive and strong correlations were found between body weight and heart girth (r=0.97), followed distantly by rump height (r=0.79) and wither height (r=0.76). These linear body measurements were highly affected by the changes in body weight, hence important in predicting body weight of indigenous Tswana sheep males. Body length (r=0.67), shoulder width (r=0.67), rump length (r=0.66), cannon bone circumference (r=0.69), cannon bone length (r=0.72), neck length (r=0.69), rump width (r=0.72), head width (r=0.60), head length (r=70), tail length (r=55) and tail circumference (r=0.50) showed significant (P<0.05) moderate and positive correlations with body weight. Ear length showed a significant (P < 0.05) low and positive correlation while scrotal circumference did not show any significant correlation with body weight.

Likewise, in females, heart girth (r=0.90) showed the strongest significant (P<0.05) and positive correlation with body weight. Most linear body measurements, body length (r=0.51), wither height (r=0.60), rump height (r=0.58), shoulder width (r=0.48), rump length (r=0.50), neck length (r=0.56), rump width (r=0.51) and head length (r=0.48) had significant (P<0.05), moderate and positive correlations with body weight. Ear length (r=0.29), cannon bone length (r=0.45), head width (r=0.39), tail length (r=0.23) and tail circumference (r=0.13) showed significant low correlation with body weight whereas cannon bone circumference had no significant (P>0.05) correlation with body weight. The highest association of heart girth with body weight than other linear body measurements is consistent with Afolayan et al. (2006) and Asaminew et al. (2016).

Generally, higher correlations coefficients between body weight and other linear body measurements were found in males than females. Selection for body weight will thus result in highest co-current improvements of linear body measurements in males than females. Traits that will benefit more from selecting for higher body weight in both male and female Tswana sheep include heart girth, rump height and wither height. Asefa et al. (2017) and Asaminew et al. (2016) similarly reported heart girth as one of the traits highly correlated to body weight and will thus be significantly improved if selection is based on body weight.

Prediction of live body weight from LBMs

Body weight is an important growth indicator and trait of economic importance that influence management interventions (drug doses) and determines the final market value of an animal (Otoikhian et al., 2008). The accuracy of functions used to predict live body weight from linear body measurements is of paramount importance in livestock production enterprises (Mohammed et al., 2017). For this study, in order to develop the prediction equation, only four linear body measurements (HG, WH, SW and RL) were selected in the prediction equation for rams (Table 5) and four linear body measurements (HG, WH, BL and RH) were selected in the prediction equation for ewes (Table 6). After comparing all coefficient values for all the relationships between body weight and other LBMs in males and females, heart girth showed the highest association with body weight in both males (0.97) and females (0.90), and thus was selected to predict body weight in each sex. This might be true because heart girth is made up of muscles, some fat and bone structure which are the main constituents of live body weight of an animal (Okpeku et al., 2011). Heart girth is deemed an easy variable to measure and is amongst the least affected by the posture of the animal (Mohammed et al., 2017). Therefore, low errors are incurred by individuals taking heart girth measurements under field conditions compared to other LBMs. 

The prediction of body weight in males can be based on the prediction equation y= -64.15+1.28x and for females y= -53.47+1.14x where, x and y are heart girth and body weight respectively. The fitted prediction models were selected based on higher of coefficient of determination (R²) and smaller coefficient of variation (CV %). The R² is a proportion of the total variability explained by the predicted model. Using heart girth in the model gave R² values of 93% in males and 80% in females meaning that heart girth accounted for large proportions of changes in body weight in males and females, respectively. Although there is a slight increase in adjusted R² values whenever a new variable is added to the model (Tables 5 and 6), using heart girth alone to predict body weight might be sufficient and preferable to combinations with other LBMs due to simplicity. This could prove very useful particularly under field conditions where animal restraint might be difficult during measurements. Several authors have reported that heart girth can be a sole predictor of live body weight based on its high correlation coefficients with body weight (Asefa et al., 2017; Asaminew et al., 2016).

The most dominant coat color patterns on indigenous Tswana sheep were plain. The white dominant and plain white was the most dominant coat colors in Tswana sheep.  Most Tswana sheep were characterized by short fat tails with a straight tip pointing downwards at the end. The predominant ear orientation in Tswana sheep was horizontal. Most Tswana sheep did not have horns and wattles. The study revealed variability in LBMs between sheep in different districts of Botswana. Tswana sheep in the Southern district of Botswana displayed superiority in body weight, body length and heart girth over sheep in other districts. Generally, higher correlation coefficients between body weight and LBMs were observed between in males than females. Heart girth accounted for most of the variability in body weight than other LBMs in both males and females and was thus used as the sole predictor of body weight. The prediction equation for body weight in Tswana sheep males was y= -64.15+1.28x and for females was y= -53.47+1.14x where, x and y are heart girth and body weight, respectively.

The authors have not declared any conflict of interests.

The authors appreciate Botswana University of Agriculture and Natural Resources for funding this study. They are also grateful to Chandapiwa Moses, Malebogo Radikara and Josephine Motseothata who provided assistance in animal handling during data collection. They appreciate the meticulous work by the government extension staff in the districts in organizing and liaising with farmers for data collection, and are forever indebted to the smallholder farmers who allowed their animals to be used for body measurements.