ORIGINAL ARTICLE Genetic Resources (2022), 3 (5), 36–50 DOI: 10.46265/genresj.KPIL8781 https://www.genresj.org ISSN: 2708-3764 Phenotypic characterization of Gesha horses in southwestern Ethiopia Amine Mustefa *, Aweke Engdawork, Seble Sinke and Abebe Hailu Ethiopian Biodiversity Institute, Addis Ababa, Ethiopia Abstract: Fifteen qualitative and 21 morphometric variables on a total of 394 adult horses (282 stallions and 112 mares) from three selected districts were recorded to characterize the horse populations in southwestern Ethiopia. General linear model, frequency, and multivariate analysis procedures of Statistical Analysis Software (SAS 9.0) were used to analyze the data. Sex and location significantly affected the studied traits. Stallions were larger than mares, and the Gesha horse population was the tallest, longest, and largest among the studied populations. The majority of the studied horses possess plain body colour patterns with red-coloured medium hair size. A higher frequency of white-coloured horses was observed with increasing age. Stepwise discriminant function analysis revealed that pelvic width, cannon bone length, and height at croup were the top three morphometric variables to discriminate the populations while head length, head neck circumference, chest width, cannon bone circumference, and croup length had the lowest discriminatory power. The results of discriminant function analysis showed advanced classification (76.7%) of the studied horses into their respective populations/locations. Finally, canonical discriminant function analysis categorized the horse populations into three distinct categories. The Gesha horse population was different from Masha and Telo horse populations while having a relatively higher relationship with the Masha horse population. However, the distances calculated in this study show only the relative size differences between each population. Such differences might not necessarily be due to breed (genetic) differences. Therefore, diversity studies through further genetic characterization are recommended to design conservation and breeding programmes. Keywords: Ethiopia, Horse, Gesha, Phenotypic characterization Citation: Mustefa, A., Engdawork, A., Sinke, S., Hailu, A. (2022). Phenotypic characterization of Gesha horses in southwestern Ethiopia. Genetic Resources 3 (5), 36–50. doi: 10.46265/genresj.KPIL8781. © Copyright 2022 the Authors. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Introduction Horses are among the most important livestock species in the highlands of Ethiopia. In rural areas, horses are the main source of transportation, both for humans and agricultural goods. They are used in public events including social and cultural festivals, and are the most culturally respected and highly valued domestic animals in the country in general, and in southern and southwestern Ethiopia in particular (Kefena et al, 2012). The highlands of Keffa and Sheka zones in southwest Ethiopia are also among the most benefitted areas from the indigenous horses (Kefena et al, 2012). In these areas, horses were also used for traditional racing shows. ∗Corresponding author: Amine Mustefa (aminemustefa32@gmail.com) Ethiopia is reported to possess 2.1 million horses (Central Statistical Agency, 2020). However, in terms of standard characterization and documenta- tion, the equine sector has received little attention. Until now, only one country-wide general study by Kefena et al (2012) was performed to phenotypically charac- terize the country’s horse breeds, their geographical distribution and production environments. Accordingly, eight breeds (Abyssinian, Bale, Boran, Horro, Kafa, Kun- dido feral horse, Ogaden/Wilwal and Selale horse) were officially reported to exist in the country (Kefena et al, 2012; EBI, 2016). However, due to different reasons, the study by Kefena et al (2012) did not cover or characterize three horse breeds (Boran, Kundido feral horse and Ogaden/Wilwal horses) out of the total eight breeds. Additionally, the lack of qualitative morphological data in the study, and the small sample size taken (95–106 Received: 23.09.21 Accepted: 11.03.22 Published online: 06.04.22 https://www.genresj.org https://www.dx.doi.org/10.46265/genresj.KPIL8781 https://www.genresj.org https://www.dx.doi.org/10.46265/genresj.KPIL8781 mailto:aminemustefa32@gmail.com Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 37 horses per breed) can be noted as limitations of the study. Similarly, the selected sampling sites were too narrow to represent the horse populations of the area. For example, the horse populations of southwestern Ethiopia were represented by a sample from a single site (Masha district). A preliminary study by a team of live- stock experts from Keffa zone hinted at the presence of an unstudied unique horse population in Gesha district. According to the results of this preliminary study, Gesha horses are said to be typical riding horses of the Keffa zone highlands. However, in the country- wide study by Kefena et al (2012), this population was represented by horses from the neighbouring Masha district. Therefore, further characterization studies were required to better understand the horse populations and quantify the level of relationships among them, thus providing a clear country-wide picture. Hence, the current study was designed to characterize the horse populations in southwestern Ethiopia using both quantitative morphometric measurements and qualitative morphological characteristics. Materials and methods Locations This study was conducted in Keffa and Sheka zones of the Southern Nations Nationalities and Peoples Regional State (SNNPR), Ethiopia. Three locations were selected for the current study (Table 1, Figure 1). Gesha and Masha districts were sampled purposively: Gesha district (one of the ten districts in Keffa zone) is the location of the horses which were supposed to be unique and unaddressed before, while Masha district (one of the three districts in Sheka zone) is where the samples were taken for the previous country-wide study by Kefena et al (2012). Telo district was sampled randomly from Keffa zone to study the relationship of its horses with Gesha horses. The sampling frame was defined after collecting available background information (origin, distribution, population size, and unique features) of the unstudied horse population through focus group discussions with livestock keepers and experts. Additionally, information regarding the sampling sites of the country-wide study was also taken from the reports of Kefena et al (2012). Data collection Quantitative and qualitative data were recorded from a total of 394 adult horses (282 stallions and 112 mares) based on the data collection procedures outlined in FAO (2012) and the previous country-wide study by Kefena et al (2012). Studied horses were carefully handled by their owners and trained personnel. Data were collected when the animals were calm and standing in an upright position on flat ground and early in the morning of the day before feeding and watering. To minimize measurement error, data were not taken from aggressive horses that did not stand properly. Similarly, to minimize subjectivity error, measurements and data recording were performed by the same researchers throughout the study. A centimetre-unit textile measuring tape was used for the morphometric measurements. Twenty-one quantitative morphometric measure- ments (Table 2) and 15 qualitative characteristics (hair size, body colour pattern, colour of the body, head, muz- zle, tail and hoof, presence/absence of stripe at dorsal body, shoulder and leg, profile of the face, back and croup, length of the tail and mane) were collected. The following body measure indices were calculated from morphometric measurements (adapted from Bodó and Hecker (1992); Cabral et al (2004); Druml et al (2008); Bene et al (2013)). • Body index = (Body length/Thorax girth) x 100 • Quadratic index = (Height at withers/Body length) x 100 • Caliber index = (Thorax girth/Height at withers) x (Cannon circumference/Height at withers) x 1000 • Overbuilt index = (Height at croup/Height at withers) x 100 • Chest index = (Chest width/Thorax girth) x 100 • Conformation index = (Thorax girth2/Height at withers)/100 Data analysis Data entry and management were performed using Microsoft Excel© worksheet. Analysis of the quantitative traits was performed separately for stallions, mares and sex-aggregated by fitting location and age as fixed variables. UNIVARIATE procedure of Statistical Analysis Software (SAS) 9.0 was used to detect outliers and test the normality of morphometric data (SAS Institute, 2002). Data on qualitative traits were subjected to chi- square (χ2) tests of the frequency (FREQ) procedure of SAS 9.0 software. Quantitative morphometric and body measure indices data were analyzed using the general linear model (GLM) procedure of SAS 9.0 software, with adjusted Tukey-Kramer test to separate the least square means (LSM). Data analysis was performed using the following model: Yijk = µ + Si + Lj + Ak + eijk where Yijk is an observation, µ is the overall mean, Si is the fixed effect of ith sex (i = stallion, mare), Lj is the fixed effect of jth location (j = Telo, Gesha, Masha), Ak is the fixed effect of kth age (k = 4–11), and eijk is the random error attributed to the nth observation. The sex effect was removed from the class variables when the analysis was done separately for each sex. Morphometric traits that better discriminate the horse populations from different locations were identified using the forward selection method of the stepwise discriminant function analysis (STEPDISC) procedure of SAS 9.0. The discriminant function analysis (DIS- CRIM) procedure of SAS 9.0. was also used to assign observations to locations and evaluate probabilities of misclassifications. Linear combination of morphomet- ric variables that provide maximal separations between locations was performed using the canonical discrim- inant function analysis (CANDISC) procedure of SAS 9.0. The scored canonical variables were used to plot 38 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 1. Climatic and agroecological features of the studied areas. Data from Bezabih (2012), Assefa et al (2013), Gebrmichael (2019). Climate factors Telo Gesha Masha Altitude (m) 2,436–2,451 1,501–3,000 1,700–3,000 Temperature (◦C) 17–25 15.1–20 16.7 Rainfall (mm) 1,278 2,001–2,200 2,192 Agroecology Highland Midland and highland Midland and highland Figure 1. Map of the sampled locations and districts pairs of canonical variables to get visual interpretation of location differences. Pairwise squared Mahalanobis dis- tances between locations were computed as: D2 (i|j) = (xi − xj) ′ cov−1 (xi − xj). Where D2 (i|j) is the dis- tances between locations i and j, cov−1 is the inverse of the covariance matrix of measured variables, xi and xj are the means of variables in the ith and jth popula- tions. Results Morphometric measurements and body measure indices The effect of sex on the studied morphometric variables is presented in Table 3. Most measurements were higher for stallions than mares while ear length and barrel length measurements of the mares were higher than the stallions. On the other hand, body length and back length measurements were not significantly affected by sex. To have a clear picture of the differences among locations, the analysis was performed separately for both sexes. The effect of location on the morphometric measurements of the stallions is presented in Table 4. All stallions’ measurements were affected significantly by their location. Gesha stallions had significantly the highest values for most of the measurements except for cannon bone length where Telo stallions had higher values. Masha stallions had relatively higher measurement values than their counterparts from Telo district, and these populations shared more similarities. On the other hand, chest width, shoulder depth, body Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 39 Table 2. Description of the collected quantitative measurements. Adapted from FAO (2012); Kefena et al (2012). No. Morphometric measurements Explanation of the measurements 1 Head length Distance from the nape to the alveolar edge of the incisors I of the upper jaw 2 Head width Distance between the upper side of the eyes measured perpendicularly to the head length 3 Ear length Distance from the tip of the ear to the connection point with the head 4 Head neck circumference Circumference of the neck at the connection point to the head 5 Neck length Distance from the highest point of the withers to the nape with the neck in a relaxed position 6 Neck body circumference Circumference of the neck at the connection point with the body 7 Chest width Distance between two outer points of the humeral bones from the front 8 Shoulder depth Distance from the withers to the shoulder joint 9 Thorax depth Distance from the withers to the sternum 10 Thorax width Distance between two hypothetical vertical parallel lines drawn at the thorax sides and along the withers’ height line 11 Thorax girth Measured in the place of the saddle girth 12 Cannon bone length Distance from the lateral tuberculum of the os metacarpale IV to the fetlock joint 13 Cannon bone circumference Smallest circumference of the forelimb’s cannon bone 14 Height at wither Distance from the highest point of the processus spinalis of the vertebra thoracic to the floor 15 Height at back Distance from the deepest point of the back to the floor 16 Height at croup Distance from the croup (rump) to the floor 17 Body length Distance from the most cranial point of the shoulder joint to the most caudal point of the pin bone (scapulo-ischial length) 18 Back length Distance from the caudal point of the shoulder joint perpendicular to the wither to the most cranial point of the hip joint measured in the saddle place 19 Pelvic width Distance between the right and left coxal tubers of the ilium 20 Croup length Distance between the sacral tuber (the highest point of croup) and ischiatic tuber (most posterior point of ischium or point of buttock or seat bone) 21 Barrel length Distance from the most caudal point of the scapula to the most cranial and dorsal point of the point of the hip length and back length measurements of Telo stallions were higher than Masha stallions. The effect of location on the morphometric measure- ments of the mares is presented in Table 5. Most of the mares’ measurements were affected significantly by their location except ear length, neck length, chest width and barrel length. Gesha mares were the biggest and heaviest among the studied populations: their circumferences of head–neck, neck–body and thorax, and heights at with- ers, back and croup, and pelvic width were significantly larger than Telo or Masha. The effect of location on the morphometric measure- ments of the studied horse populations (sex-aggregated) is presented in Table 6. All the morphometric measure- ments of the studied horse populations were affected sig- nificantly by their location. Significantly, the Gesha horse population had the highest values for most of the mea- surements except for cannon bone length, which was higher in Telo horses. Pearson correlation coefficients of the morphometric measurements of the horses (both sexes) from different locations are presented in Table 7. The majority of the traits were positively correlated. Higher positive correla- tion was observed between height at withers and height at back while lower positive correlation was observed between ear length and head neck circumference. Nega- tive correlation was observed between thorax width and cannon bone length. The effect of location on body measure indices of the studied horse populations (separately for each sex) is presented in Table 8. All the body measure indices of the studied horse populations were significantly affected by sex. Similarly, most of the body measure indices were significantly affected by their location. Multivariate analysis Stepwise discriminant function analysis revealed the order of importance of the studied morphometric vari- ables in discriminating the horse populations (Table 9). The results were also confirmed by Wilk’s lambda test (Table 9) where all the selected variables had a highly significant (P < 0.0001) contribution in discriminat- ing the horse populations. Pelvic width, cannon bone length and height at croup were the first three important traits used in discriminating the studied horse popula- tions. However, some morphometric variables like head length, head neck circumference, chest width, cannon bone circumference and croup length had the lowest dis- 40 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 3. Least-square means ± standard errors of quantitative body measurements (cm) of the horse populations by sex. Traits Stallions Mares p-value N 282 112 Head length 53.2 ± 0.16 52.1 ± 0.23 < 0.0001 Head width 21.5 ± 0.06 21.2 ± 0.09 0.0015 Ear length 15.0 ± 0.09 15.5 ± 0.13 0.0033 Head neck circumference 58.9 ± 0.24 54.7 ± 0.36 < 0.0001 Neck length 59.4 ± 0.27 57.3 ± 0.40 < 0.0001 Neck body circumference 91.4 ± 0.37 84.3 ± 0.55 < 0.0001 Chest width 25.9 ± 0.13 24.3 ± 0.20 < 0.0001 Shoulder depth 53.4 ± 0.18 51.1 ± 0.27 < 0.0001 Thorax depth 61.8 ± 0.23 59.8 ± 0.33 < 0.0001 Thorax width 34.2 ± 0.16 32.9 ± 0.24 < 0.0001 Thorax girth 143.0 ± 0.44 138.6 ± 0.65 < 0.0001 Cannon bone length 24.0 ± 0.09 23.6 ± 0.14 0.0066 Cannon bone circumference 16.4 ± 0.06 15.6 ± 0.09 < 0.0001 Height at withers 131.8 ± 0.29 127.8 ± 0.43 < 0.0001 Height at back 129.0 ± 0.28 125.7 ± 0.41 < 0.0001 Height at croup 132.1 ± 0.28 129.1 ± 0.42 < 0.0001 Body length 125.0 ± 0.38 124.1 ± 0.56 0.1796 Back length 70.0 ± 0.26 70.2 ± 0.38 0.6114 Pelvic width 40.4 ± 0.16 39.5 ± 0.24 0.0016 Croup length 39.7 ± 0.18 38.6 ± 0.27 0.0020 Barrel length 67.0 ± 0.29 69.4 ± 0.43 < 0.0001 Table 4. Means and pairwise comparisons of morphometric measurements of the stallions from different locations. Means within a row bearing different superscripts are significantly different; a indicates the largest value. Traits Least Square Means (LSM ± SE) Mean ± SE CV p-value Telo Gesha Masha N 94 136 52 Head length 52.4 ± 0.27b 54.2 ± 0.23a 53.1 ± 0.36b 53.2 ± 0.16 4.5 < 0.0001 Head width 21.0 ± 0.11b 21.9 ± 0.09a 21.6 ± 0.15a 21.5 ± 0.07 4.6 < 0.0001 Ear length 14.5 ± 0.15b 15.1 ± 0.13a 15.4 ± 0.20a 14.9 ± 0.08 8.9 0.0002 Head neck circumference 58.1 ± 0.43b 60.8 ± 0.36a 57.9 ± 0.57b 59.3 ± 0.24 6.4 < 0.0001 Neck length 59.3 ± 0.48b 60.6 ± 0.40a 58.4 ± 0.63b 59.4 ± 0.27 7.1 0.0049 Neck body circumference 88.8 ± 0.63b 95.9 ± 0.53a 89.9 ± 0.83b 92.2 ± 0.41 6.1 < 0.0001 Chest width 26.0 ± 0.25ab 26.7 ± 0.21a 25.2 ± 0.33b 26.0 ± 0.14 8.5 0.0008 Shoulder depth 53.5 ± 0.32b 54.6 ± 0.27a 52.1 ± 0.42c 53.6 ± 0.18 5.3 < 0.0001 Thorax depth 59.6 ± 0.40c 63.7 ± 0.33a 61.9 ± 0.52b 61.9 ± 0.25 5.7 < 0.0001 Thorax width 32.8 ± 0.30c 35.5 ± 0.25a 34.2 ± 0.39b 34.1 ± 0.18 7.7 < 0.0001 Thorax girth 141.5 ± 0.78b 149.0 ± 0.65a 138.8 ± 1.02b 143.6 ± 0.53 4.8 < 0.0001 Cannon bone length 24.8 ± 0.16a 23.5 ± 0.14b 23.6 ± 0.21b 24.0 ± 0.09 6.0 < 0.0001 Cannon bone circumference 16.34 ± 0.11b 16.9 ± 0.10a 16.1 ± 0.15b 16.5 ± 0.07 6.1 < 0.0001 Height at withers 130.7 ± 0.51b 135.2 ± 0.43a 129.6 ± 0.67b 132.2 ± 0.31 3.4 < 0.0001 Height at back 127.8 ± 0.48b 132.4 ± 0.41a 127.0 ± 0.63b 129.5 ± 0.30 3.3 < 0.0001 Height at croup 131.3 ± 0.50b 135.8 ± 0.42a 129.6 ± 0.7b 132.8 ± 0.31 3.3 < 0.0001 Body length 125.3 ± 0.67b 127.5 ± 0.56a 122.4 ± 0.88c 125.3 ± 0.38 4.7 < 0.0001 Back length 70.7 ± 0.44a 71.1 ± 0.37a 68.2 ± 0.58b 70.1 ± 0.25 5.6 0.0003 Pelvic width 38.7 ± 0.29c 42.3 ± 0.24a 40.5 ± 0.38b 40.5 ± 0.19 6.3 < 0.0001 Croup length 38.7 ± 0.33b 41.1 ± 0.28a 39.3 ± 0.43b 39.7 ± 0.18 7.3 < 0.0001 Barrel length 66.1 ± 0.51b 67.7 ± 0.43a 67.1 ± 0.66ab 66.8 ± 0.28 6.7 0.0287 Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 41 Table 5. Means and pairwise comparisons of morphometric measurements of the mares from different locations. Means within a row bearing different superscripts are significantly different; a indicates the largest value. Traits Least Square Means (LSM ± SE) Mean ± SE CV p-value Telo Gesha Masha N 29 47 36 Head length 51.9 ± 0.49ab 53.0 ± 0.38a 51.3 ± 0.45b 52.1 ± 0.25 4.9 0.0128 Head width 20.6 ± 0.18b 21.5 ± 0.14a 21.5 ± 0.17a 21.2 ± 0.10 4.4 0.0001 Ear length 15.2 ± 0.27 15.4 ± 0.21 15.9 ± 0.25 15.5 ± 0.13 9.1 0.1536 Head neck circumference 53.2 ± 0.70b 56.7 ± 0.53a 54.3 ± 0.64b 54.9 ± 0.37 6.6 0.0002 Neck length 56.7 ± 0.75 58.4 ± 0.57 56.5 ± 0.69 57.2 ± 0.38 6.8 0.0597 Neck body circumference 81.3 ± 1.20b 89.4 ± 0.91a 81.7 ± 1.10b 84.7 ± 0.71 7.3 < 0.0001 Chest width 23.9 ± 0.32 24.8 ± 0.24 24.0 ± 0.29 24.3 ± 0.16 6.8 0.0535 Shoulder depth 51.0 ± 0.54ab 52.2 ± 0.41a 49.9 ± 0.49b 51.0 ± 0.30 5.4 0.0015 Thorax depth 58.7 ± 0.63b 61.1 ± 0.48a 59.7 ± 0.58ab 59.9 ± 0.34 5.4 0.0069 Thorax width 31.4 ± 0.44b 34.1 ± 0.34a 33.2 ± 0.41a 32.9 ± 0.27 7.0 < 0.0001 Thorax girth 135.9 ± 1.26b 144.4 ± 0.96a 134.6 ± 1.16b 138.6 ± 0.81 4.7 < 0.0001 Cannon bone length 24.1 ± 0.27a 22.9 ± 0.21b 23.6 ± 0.25ab 23.5 ± 0.14 6.0 0.0014 Cannon bone circumference 15.6 ± 0.15ab 15.9 ± 0.11a 15.4 ± 0.13b 15.6 ± 0.08 4.9 0.0067 Height at withers 127.3 ± 0.81b 130.3 ± 0.62a 125.4 ± 0.75b 127.9 ± 0.45 3.3 < 0.0001 Height at back 124.8 ± 0.84b 128.4 ± 0.64a 123.4 ± 0.77b 125.8 ± 0.46 3.5 < 0.0001 Height at croup 128.3 ± 0.81b 132.1 ± 0.61a 126.3 ± 0.74b 129.3 ± 0.47 3.2 < 0.0001 Body length 124.5 ± 1.13ab 126.6 ± 0.86a 121.0 ± 1.04b 124.0 ± 0.65 4.7 0.0004 Back length 70.6 ± 0.80ab 71.4 ± 0.61a 68.1 ± 0.74b 70.0 ± 0.44 6.0 0.0033 Pelvic width 38.6 ± 0.43b 41.4 ± 0.33a 38.4 ± 0.40b 39.6 ± 0.27 5.7 < 0.0001 Croup length 38.8 ± 0.52ab 39.6 ± 0.39a 37.7 ± 0.47b 38.7 ± 0.27 6.9 0.0120 Barrel length 68.8 ± 0.89 70.2 ± 0.68 69.4 ± 0.82 69.3 ± 0.47 6.7 0.4347 Table 6. Means and pairwise comparisons of morphometric measurements of the horses (both sexes) from different locations. Means within a row bearing different superscripts are significantly different; a indicates the largest value. Traits Least Square Means (LSM ± SE) Mean ± SE CV p-value Telo Gesha Masha N 123 183 88 Head length 52.0 ± 0.24b 53.6 ± 0.20a 52.2 ± 0.27b 52.9 ± 0.14 4.6 < 0.0001 Head width 20.8 ± 0.10b 21.7 ± 0.08a 21.6 ± 0.11a 21.4 ± 0.06 4.6 < 0.0001 Ear length 14.8 ± 0.14b 15.3 ± 0.11a 15.7 ± 0.15a 15.1 ± 0.07 8.9 < 0.0001 Head neck circumference 55.8 ± 0.38b 58.7 ± 0.31a 56.0 ± 0.41b 58.1 ± 0.23 6.5 < 0.0001 Neck length 58.1 ± 0.42b 59.5 ± 0.34a 57.4 ± 0.46b 58.8 ± 0.23 7.0 0.0004 Neck body circumference 85.1 ± 0.58b 92.5 ± 0.48a 86.0 ± 0.64b 90.1 ± 0.39 6.4 < 0.0001 Chest width 25.0 ± 0.21b 25.7 ± 0.17a 24.6 ± 0.23b 25.5 ± 0.12 8.1 < 0.0001 Shoulder depth 52.3 ± 0.28b 53.4 ± 0.23a 51.0 ± 0.31c 52.8 ± 0.17 5.3 < 0.0001 Thorax depth 58.9 ± 0.35c 62.6 ± 0.29a 61.0 ± 0.39b 61.3 ± 0.21 5.7 < 0.0001 Thorax width 32.1 ± 0.25c 34.8 ± 0.21a 33.7 ± 0.28b 33.8 ± 0.15 7.5 < 0.0001 Thorax girth 138.9 ± 0.68b 146.6 ± 0.56a 136.9 ± 0.75b 142.2 ± 0.46 4.8 < 0.0001 Cannon bone length 24.5 ± 0.14a 23.2 ± 0.12b 23.6 ± 0.16b 23.9 ± 0.08 6.0 < 0.0001 Cannon bone circumference 15.9 ± 0.09b 16.4 ± 0.08a 15.7 ± 0.10b 16.2 ± 0.06 5.8 < 0.0001 Height at withers 128.8 ± 0.45b 132.8 ± 0.37a 127.7 ± 0.50b 131.0 ± 0.28 3.4 < 0.0001 Height at back 126.1 ± 0.43b 130.4 ± 0.36a 125.5 ± 0.48b 128.4 ± 0.26 3.4 < 0.0001 Height at croup 129.7 ± 0.44b 134.0 ± 0.36a 128.2 ± 0.48b 131.8 ± 0.27 3.3 < 0.0001 Body length 124.8 ± 0.59b 127.0 ± 0.49a 121.7 ± 0.65c 124.9 ± 0.33 4.7 < 0.0001 Back length 70.8 ± 0.40a 71.2 ± 0.33a 68.2 ± 0.44b 70.1 ± 0.22 5.7 < 0.0001 Pelvic width 38.4 ± 0.25c 41.8 ± 0.21a 39.6 ± 0.28b 40.3 ± 0.16 6.3 < 0.0001 Croup length 38.4 ± 0.29b 40.4 ± 0.23a 38.6 ± 0.32b 39.4 ± 0.16 7.2 < 0.0001 Barrel length 67.4 ± 0.45b 69.0 ± 0.37a 68.3 ± 0.50b 67.5 ± 0.25 6.6 0.0113 42 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 7. Pearson correlation coefficients between each morphometric measurement (above diagonal) and level of significance (below diagonal) of the horses (both sexes) from the three locations. HL, Head length; HW, Head width; EL, Ear length; HNC, Head neck circumference; NL, Neck length; NBC, Neck body circumference; CW, Chest width; SD, Shoulder depth; TD, Thorax depth; TW, Thorax width; TG, Thorax girth; CBL, Cannon bone length; CBC, Cannon bone circumference; HAW, Height at withers; HAB, Height at back; HAC, Height at croup; BOL, Body length; BAL, Back length; PW, Pelvic width; CL, Croup length; BRL, Barrel length. *, p < 0.05; **, p < 0.01; ***, p < 0.0001; NS, Not Significant. Traits HL HW EL HNC NL NBC CW SD TD TW TG CBL CBC HAW HAB HAC BOL BAL PW CL BRL HL 0.48 0.16 0.35 0.22 0.48 0.41 0.49 0.51 0.38 0.55 0.08 0.41 0.59 0.56 0.56 0.39 0.36 0.42 0.28 0.29 HW *** 0.24 0.37 0.30 0.44 0.39 0.41 0.46 0.44 0.48 0.06 0.36 0.42 0.42 0.41 0.36 0.30 0.42 0.35 0.31 EL ** *** 0.001 0.14 0.07 0.14 0.08 0.17 0.20 0.11 0.03 0.10 0.11 0.11 0.11 0.11 0.16 0.19 0.20 0.23 HNC *** *** NS 0.34 0.74 0.49 0.57 0.46 0.48 0.67 0.12 0.53 0.57 0.52 0.52 0.43 0.19 0.49 0.47 0.11 NL *** *** ** *** 0.49 0.35 0.42 0.39 0.39 0.53 0.13 0.44 0.53 0.53 0.54 0.47 0.32 0.38 0.40 0.23 NBC *** *** NS *** *** 0.55 0.66 0.62 0.56 0.78 0.06 0.61 0.70 0.67 0.67 0.49 0.31 0.60 0.52 0.16 CW *** *** ** *** *** *** 0.56 0.45 0.41 0.59 0.23 0.47 0.47 0.41 0.43 0.41 0.40 0.45 0.40 0.25 SD *** *** NS *** *** *** *** 0.57 0.47 0.73 0.20 0.55 0.69 0.65 0.66 0.56 0.39 0.49 0.51 0.29 TD *** *** ** *** *** *** *** *** 0.56 0.67 0.06 0.52 0.62 0.61 0.59 0.42 0.32 0.56 0.44 0.30 TW *** *** *** *** *** *** *** *** *** 0.66 -0.04 0.49 0.54 0.54 0.52 0.47 0.30 0.57 0.48 0.33 TG *** *** * *** *** *** *** *** *** *** 0.09 0.66 0.78 0.76 0.76 0.67 0.47 0.72 0.60 0.36 CBL NS NS NS * * NS *** *** NS NS NS 0.26 0.16 0.14 0.15 0.15 0.22 0.002 0.09 0.05 CBC *** *** NS *** *** *** *** *** *** *** *** *** 0.63 0.60 0.61 0.51 0.39 0.51 0.44 0.20 HAW *** *** * *** *** *** *** *** *** *** *** ** *** 0.96 0.94 0.60 0.45 0.60 0.55 0.28 HAB *** *** * *** *** *** *** *** *** *** *** ** *** *** 0.93 0.57 0.43 0.57 0.54 0.28 HAC *** *** * *** *** *** *** *** *** *** *** ** *** *** *** 0.63 0.47 0.59 0.56 0.31 BOL *** *** * *** *** *** *** *** *** *** *** ** *** *** *** *** 0.51 0.54 0.50 0.58 BAL *** *** ** ** *** *** *** *** *** *** *** ** *** *** *** *** *** 0.41 0.29 0.37 PW *** *** ** *** *** *** *** *** *** *** *** NS *** *** *** *** *** *** 0.59 0.35 CL *** *** *** *** *** *** *** *** *** *** *** NS *** *** *** *** *** *** *** 0.26 BRL *** *** *** * *** ** *** *** *** *** *** NS *** *** *** *** *** *** *** *** Table 8. Body measure indices of the studied horse populations Traits Least Square Means (LSM ± SE) Mean ± SE p-value Telo Gesha Masha Stallions Body index 88.71 ± 0.40a 85.60 ± 0.34b 88.33 ± 0.53a 87.36 ± 0.24 < 0.0001 Quadratic index 104.5 ± 0.47b 106.2 ± 0.40a 106.0 ± 0.62ab 105.7 ± 0.25 0.0095 Caliber index 135.4 ± 1.16ab 138.0 ± 0.98a 132.8 ± 1.53b 135.5 ± 0.65 0.0119 Overbuilt index 100.4 ± 0.16 100.5 ± 0.14 100.0 ± 0.21 100.4 ± 0.09 0.2523 Chest index 18.40 ± 0.15a 17.89 ± 0.13b 18.17 ± 0.20ab 18.1 ± 0.08 0.0178 Conformation index 1.53 ± 0.014b 1.65 ± 0.011a 1.49 ± 0.018b 1.56 ± 0.009 < 0.0001 Mares Body index 91.67 ± 0.75a 87.75 ± 0.58b 90.00 ± 0.69a 89.57 ± 0.40 0.0002 Quadratic index 102.3 ± 0.79 103.2 ± 0.60 103.9 ± 0.72 103.4 ± 0.42 0.3730 Caliber index 130.7 ± 1.57 135.3 ± 1.20 131.4 ± 1.44 132.1 ± 0.89 0.0580 Overbuilt index 100.8 ± 0.25 101.4 ± 0.19 100.7 ± 0.23 101.0 ± 0.13 0.0617 Chest index 17.61 ± 0.21ab 17.23 ± 0.16b 17.88 ± 0.19a 17.53 ± 0.10 0.0379 Conformation index 1.45 ± 0.02b 1.60 ± 0.02a 1.45 ± 0.02b 1.50 ± 0.01 < 0.0001 Both sexes Body index 90.01 ± 0.37a 86.71 ± 0.30b 89.07 ± 0.41a 87.99 ± 0.21 < 0.0001 Quadratic index 103.3 ± 0.42b 104.8 ± 0.34a 105.05 ± 0.46a 105.1 ± 0.22 0.0036 Caliber index 133.5 ± 0.97b 136.6 ± 0.80a 131.9 ± 1.07b 134.5 ± 0.53 0.0005 Overbuilt index 100.7 ± 0.14 100.9 ± 0.12 100.4 ± 0.16 100.6 ± 0.07 0.0579 Chest index 18.02 ± 0.13a 17.55 ± 0.10b 17.99 ± 0.14a 17.95 ± 0.07 0.0019 Conformation index 1.50 ± 0.01b 1.62 ± 0.01a 1.47 ± 0.01b 1.54 ± 0.01 < 0.0001 Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 43 criminatory power and were not used in discriminating the horse populations. The values and significant levels of different statistical tests used in the discriminant function analysis are shown in Table 10. All the statistical tests were significant showing the appropriateness of the model used in discriminating the horse populations. Outputs of the canonical discrimination analysis including eigenvalues and class means under the first two canonical structures are presented in Table 11. Sim- ilarly, Table 11 also presents raw canonical coefficients used in constructing the two canonical variables (Can 1 and Can 2). Accordingly, the first canonical structure (Can 1) explained the majority (65.7%) of the total variability among the three horse populations. It also produced a greater eigenvalue and multiple correlation (0.70) between the classes (locations) and the morpho- metric measurements than the second canonical struc- ture (Can 2). These results show the higher power of Can 1 compared with Can 2 in separating the horse pop- ulations from the studied locations. However, Can 2 also separated one-third of the population, which Can 1 is unable to separate. Accordingly, Can 1 separated Telo horses from the others while Can 2 separated Masha horses from the others. Discriminant function analysis classified each indi- vidual observation into a known population/location (Table 12). Accordingly, an average of 76.7% of the sam- pled animals were classified into their respective popu- lation/location. The highest classification of individual horses into their respective locations was observed in the Telo horse population (79.7%) with a small error rate (20.3%). On the other hand, a high error rate (26.1%) was detected in the Masha horse population. The priors (33.3%) show the chance of every individual observation to be classified into the given three popula- tions/locations. Pairwise squared Mahalanobis distances between locations are shown in Table 13. All distances were significant. Gesha and Masha horse populations are closely related, while their distance from the Telo horse population is large. A plot of the first two canonical structures discrimi- nating the studied horse populations is presented in Fig- ure 2. Accordingly, Can 1 separates the Telo horse pop- ulation from the others, while Can 2 discriminates the Masha horse population from the others. Overall, the analysis categorized the horse populations into three dis- tinct categories. Therefore, the Gesha horse population is different from the Masha and Telo horse populations. Furthermore, the Gesha horse population has more rela- tionship with the Masha than the Telo horse population. Qualitative characteristics Chi-square and Cramér’s V statistical values and level of significance for the effect of the class variables on the qualitative characteristics of the studied horse populations are presented in Table 14. All the traits were significantly affected by the location of the horse populations except body colour pattern and shoulder stripe. On the other hand, only five traits were significantly affected by the horses’ sex and age. Face and back profile of the studied horse populations were found to be highly associated with location while the level of relationship of shoulder stripe with location was insignificant. A higher level of relationship between the horses’ sex and age with their head colour was also observed. The majority of the studied horse populations possess a plain body colour pattern with red, medium hair size, and long tail and mane with a mainly black muzzle, tail and hoof (Tables 15 and 16, Figure 3). All horses had sloppy croup with the absence of leg stripe. Short hair size, convex face and straight back profiles were observed more frequently on stallions than mares. The majority of the Gesha horses had red body and head (Figure 3, C and D) while white-striped red head was also frequently observed. White body and head colour were observed more frequently on Telo horses. Around half of the horse population from Masha district had black and white hoof, which was rarely observed in the other horse populations. The effect of age on the colour-related qualitative characteristics of the studied horse populations is presented in Figure 4. Little effect of age on the colour- related qualitative characteristics was observed. As the age of the studied horses increased, the proportion of horses with white body colour showed a significant increase (p < 0.0001), while the proportion of the other colours decreased. Similarly, the proportion of horses with white head colour showed a significant increase (p< 0.05) with age, while the proportion of horses with grey head colour decreased. The proportion of the others (red and red with white stripe) remained constant. Finally, older horses also showed a higher proportion of white tail colour (p < 0.01) while the proportion of horses with black tail decreased. The proportion of the others (red and grey) remained the same. The majority of the Gesha horses had a dorsal stripe and slightly convex face profile, which can be considered their unique characteristics (Table 9). A curved back profile was predominantly observed in Telo horses, which distinguished them from the others. A slight effect of sex on the qualitative characteristics was observed: shorter hair, a slightly convex face and a straight back profile were observed mainly in stallions. Discussion Morphometric measurements The studied morphometric measurements produced reliable information to characterize and differentiate the three horse populations phenotypically. Besides studying the main effect (location), the effects of age and sex were also analyzed to see if they could cause a significant difference. The effect of age was not significant, 44 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 9. Summary of the stepwise discriminant function analysis. Traits are listed in ascending order used in discriminating the horse populations from different locations. Step Variables entered Partial R-square F value Pr > F Wilks’ Lambda Pr < Lambda 1 Pelvic width 0.2214 55.60 < 0.0001 0.7785 < 0.0001 2 Cannon bone length 0.1561 36.06 < 0.0001 0.6570 < 0.0001 3 Height at croup 0.1362 30.68 < 0.0001 0.5675 < 0.0001 4 Head width 0.0888 18.91 < 0.0001 0.5171 < 0.0001 5 Body length 0.0574 11.79 < 0.0001 0.4874 < 0.0001 6 Ear length 0.0500 10.16 < 0.0001 0.4630 < 0.0001 7 Thorax depth 0.0381 7.62 0.0006 0.4454 < 0.0001 8 Shoulder depth 0.0531 10.76 < 0.0001 0.4218 < 0.0001 9 Neck body circumference 0.0393 7.84 0.0005 0.4052 < 0.0001 10 Back length 0.0308 6.07 0.0025 0.3927 < 0.0001 11 Barrel length 0.0336 6.63 0.0015 0.3795 < 0.0001 12 Thorax width 0.0272 5.32 0.0053 0.3692 < 0.0001 13 Thorax girth 0.0306 5.99 0.0028 0.3578 < 0.0001 14 Height at withers 0.0191 3.67 0.0264 0.3510 < 0.0001 15 Height at back 0.0227 4.39 0.0131 0.3430 < 0.0001 16 Neck length 0.0200 3.83 0.0225 0.3362 < 0.0001 - Head length 0.0029 0.55 0.5754 - - - Head neck circumference 0.0002 0.05 0.9555 - - - Chest width 0.0028 0.52 0.5947 - - - Cannon bone circumference 0.0019 0.36 0.6946 - - - Croup length 0.0011 0.20 0.8210 - - Table 10. Values and significant levels of different statistical tests. DF, degrees of freedom. Statistic Value F value Num DF Den DF Pr > F Wilk’s lambda 0.3362 17.03 32 752 < 0.0001 Pillai’s trace 0.8298 16.71 32 752 < 0.0001 Hotelling-Lawley trace 1.4280 17.35 32 668.29 < 0.0001 Roy’s Largest Root 0.9718 22.90 16 377 < 0.0001 Figure 2. Plot of the first two canonical structures discriminating the three horse populations. Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 45 Table 11. Canonical correlations, eigenvalues, and class means. Can 1 Can 2 Multivariate Statistics Canonical Correlation 0.7020 0.5805 Eigenvalue 0.9718 0.5083 Proportion 0.6566 0.3434 Class (location) means Telo -1.4394 0.1662 Gesha 0.7827 0.5109 Masha 0.3841 -1.2949 Raw canonical coefficients Head width 0.3332 -0.2810 Ear length 0.1426 -0.1680 Neck length -0.0526 -0.0230 Neck body circumference 0.0552 0.0332 Shoulder depth -0.1501 0.0289 Thorax depth 0.0693 -0.1063 Thorax width 0.0581 -0.1371 Thorax girth -0.0134 0.0875 Cannon bone length -0.3522 -0.1375 Height at withers -0.1633 -0.0871 Height at back 0.1627 -0.0541 Height at croup 0.0777 0.2509 Body length -0.0567 0.0267 Back length -0.0606 0.0513 Pelvic width 0.1924 -0.0206 Barrel length 0.0350 -0.0723 which might be due to the nature of the sampling, which included adult horses only. On the other hand, sex significantly affected the studied traits. Stallions had higher values than mares on most morphometric measurements, in line with Rensch’s rule (Rensch, 1950). According to Rensch (1950), males of a given species are usually larger than females. Such differences between stallions and mares may be ascribed to levels of testosterone secreted by stallions, which leads to larger muscle mass and skeletal development (Baneh and Hafezian, 2009). Similar results were also reported by Kefena et al (2012), Ghezelsoflou et al (2018) and Sadek et al (2006) on Ethiopian, Iranian Turkoman and Arabian horses, respectively. According to Kefena et al (2012), Selale horses (the tallest and typical riding horses in Ethiopia) had values of 131.2 ± 0.4, 125.6 ± 0.4, and 131.7 ± 0.5cm for heights at withers, back and croup, respectively. The current study revealed that Gesha horses are the tallest horses in Ethiopia with a value of 132.8 ± 0.37, 130.4 ± 0.36, and 134.0 ± 0.36cm for heights at withers, back and croup, respectively (Table 6). However, these values were much lower than the reports of Zechner et al (2001) for Lipizzan horses studied in different locations in Europe, and Ghezelsoflou et al (2018) for Iranian Turkoman horses in Iran. The tall and big body of the Gesha horse population in Ethiopia indicates that they Figure 3. A, Telo stallion; B, Masha stallion; C, Gesha stallion; D, Gesha mare. Photo: Amine Mustefa, EBI can be categorized as typical saddle horses. This is in line with the study by Kristjansson et al (2016) in Iceland, which showed a higher riding ability as the horses’ height increased. Traditionally, Gesha horses, which are known for their aggressiveness, are also known and recognized as typical riding horses. The barrel and neck lengths, and cannon bone length and circumference for all the populations from the current study are comparable with the reports of Kefena et al (2012) on all Ethiopian horse populations. The 46 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 12. Number and percentage of observations classified into locations. From location Telo Gesha Masha Total Telo 98 (79.7%) 14 (11.4%) 11 (8.9%) 123 (100%) Gesha 19 (10.4%) 140 (76.5%) 24 (13.1%) 183 (100%) Masha 7 (7.9%) 16 (18.2%) 65 (73.9%) 88 (100%) Total 124 (31.5%) 170 (43.1%) 100 (25.4%) 394 (100%) Error rate 0.203 0.235 0.261 0.233 Priors 0.333 0.333 0.333 Table 13. Squared Mahalanobis distance between locations; output of the multivariate analysis calculated using the quantitative measurements. *** shows the significance of the distance calculations at p < 0.0001. From location Telo Gesha Masha Telo 0 Gesha 5.06*** 0 Masha 5.46*** 3.42*** 0 body length of Gesha horses (127.0 ± 0.49cm) is Figure 4. Effect of age on colour characteristics of horse populations. A) Body colour; B) Head colour; C) Tail colour. lower than the reports of Kefena et al (2012) for all Ethiopian horse populations. On the other hand, the head and back lengths of Gesha horses (53.6 ± 0.20 and 71.2 ± 0.33cm, respectively) is higher than all Ethiopian horse populations (Kefena et al, 2012). Such wide disagreement might be due to differences in points of measurement. The thorax girth of Gesha horses (146.6 ± 0.56cm) is comparable with Selale (146.6 ± 0.8cm), Bale (145.3 ± 0.7cm), and Horro horses (145.5 ± 0.6cm) while it was higher than Abyssinian horses (140.4 ± 0.5cm) and lower than Keffa horses (152.6 ± 0.7cm) (Kefena et al, 2012). Body measure indices The body index shows the length of the animal. A long animal is best suited for speed, a short animal for strength (Torres and Jardim, 1981). Long animals have a body index value greater than 90, while a value less than 85 indicates that the animal is short (Torres and Jardim, 1981). According to Table 8, the Telo and Masha mares were categorized as long horses. However, in reality, Gesha stallions are known for their speed. The caliber index, which shows the overall size of the horse, increases with age and size (Kaps et al, 2005). Kaps et al (2005) observed its increase from 119.1 to 135 in Lipizzan horses from 6 to 36 months of age. The current findings show the comparably big size of Gesha stallions. The overbuilt index of a horse indicates the proportion of its height at withers and at croup. A horse with downhill conformation (height at croup higher than height at withers) is indicated as the best riding horse by Padilha et al (2017), since stronger muscles in the hind limbs and taller hind limbs indicate greater power for jumping and the ability to give a solo performance. In line with the current findings, Mcmanus et al (2005) in Campeiro horses, Rezende et al (2014) in Brazilian sport horses and Mariz et al (2015) in Quarter horses reported a slightly downhill conformation. However, uphill conformation was reported as an important characteristic by Lucena et al (2015) in Marchador horses and Kristjansson et al (2016) in Icelandic horses. According to Torres and Jardim (1981), a riding horse must have a conformation index value of 2.1125. A value above this threshold shows the suitability of a horse for work. The conformation index values found in the current study were between 1.47 and 1.65 (Table 8), Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 47 Table 14. Statistical values for chi-square and Cramér’s V, and level of significance (probabilities) for the effects of location, sex and age on the qualitative characteristics of the studied horse populations: aggregate sex. χ2, chi-square; prob., probabilities; *, < 0.05; **, < 0.01; ***, < 0.0001; NS, Not significant. Qualitative traits Location Sex Age χ2 value Cramér’s V Prob. χ2value Cramér’s V Prob. χ2value Cramér’s V Prob. Body colour 43.1 0.234 *** 6.4 0.127 NS 95.2 0.201 *** Head colour 34.8 0.210 * 19.4 0.222 * 90.5 0.432 * Muzzle colour 37.1 0.217 *** 5.9 0.122 NS 42.2 0.164 * Tail colour 23.7 0.173 ** 9.6 0.156 * 58.1 0.192 ** Hoof colour 55.8 0.266 *** 1.9 0.069 NS 38.2 0.220 ** Hair size 21.3 0.233 *** 12.4 0.178 ** 9.7 0.157 NS Body colour pattern 8.7 0.105 NS 0.07 0.014 NS 8.3 0.103 NS Dorsal stripe 16.5 0.205 ** 0.2 0.021 NS 10.5 0.163 NS Shoulder stripe 1.8 0.068 NS 1.6 0.064 NS 4.8 0.111 NS Face profile 52.9 0.367 *** 4.1 0.102 * 4.3 0.105 NS Back profile 52.8 0.366 *** 4.0 0.101 * 2.6 0.081 NS Tail length 28.4 0.190 *** 4.2 0.103 NS 17.4 0.149 NS Mane length 52.8 0.259 *** 2.5 0.080 NS 10.0 0.112 NS Table 15. Percentages of colour-related qualitative traits of the horses (both sexes) from different locations. Colour-related qualitative traits Location Sex Telo Gesha Masha Stallions Mares Body colour Red 30.1 50.8 35.2 42.2 37.5 Brown 20.3 13.1 21.6 14.5 24.1 Gray 16.3 15.8 20.5 18.1 14.3 White 20.3 13.1 18.2 17.4 14.3 Tan 0.8 5.5 4.6 3.5 4.5 Black 9.8 1.1 0.0 3.2 4.4 Red and white 2.4 0.6 0.0 1.1 0.9 Head colour White 30.9 19.7 28.4 26.9 20.5 Gray 18.7 12.0 13.6 15.3 12.5 Red 21.1 26.2 18.2 24.5 18.8 Red with white stripe 5.7 21.9 12.5 14.9 14.3 Black 14.6 9.8 9.1 10.3 13.4 Black with white stripe 0.8 1.6 4.5 1.8 2.7 Brown 5.7 4.9 9.1 3.5 12.5 Brown with white stripe 1.6 0.6 1.1 0.7 1.8 Tan 0.8 1.1 0.0 1.1 0.0 Tan with white stripe 0.0 2.2 3.4 1.1 3.6 Muzzle colour Black 51.2 36.6 37.5 39.4 46.4 White 26.8 19.7 12.5 22.7 14.3 Red 10.6 25.7 18.2 19.1 10.6 Gray 11.4 9.3 21.6 13.5 10.7 White and Black 0.0 8.7 10.2 5.3 8.9 Tail colour Black 52.0 53.0 36.4 48.6 50.0 Gray 26.0 19.7 26.1 25.2 17.9 White 13.8 8.7 12.5 11.7 9.8 Red 4.9 14.8 12.5 11.0 11.6 Brown 3.3 3.8 12.5 3.5 10.7 Hoof colour Black 91.9 74.9 52.3 74.1 77.7 Black and White 4.9 21.8 47.7 22.7 21.4 White 3.2 3.3 0.0 3.2 0.9 48 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Table 16. Percentages of qualitative traits of the horses (both sexes) from different locations. Qualitative traits Location Sex Telo Gesha Masha Stallions Mares Hair size Short 42.3 43.2 15.9 42.2 23.2 Medium 57.7 56.8 84.1 57.8 76.8 Body colour pattern Plain 95.9 99.4 100 98.6 98.2 Pied 1.6 0.6 0.0 0.7 0.9 Shaded 2.4 0.0 0.0 0.7 0.9 Dorsal stripe Absent 67.5 44.3 57.9 53.9 56.3 Present 32.5 55.7 42.1 46.1 43.7 Shoulder stripe Absent 99.2 99.4 97.7 98.6 100 Present 0.2 0.6 2.3 1.4 0.0 Face profile Straight 86.2 45.4 65.9 59.6 70.5 Slightly convex 13.8 54.6 34.1 40.4 29.5 Back profile Straight 44.7 76.5 87.5 72.0 61.6 Curved 55.3 23.5 12.5 28.0 38.4 Tail length Short 2.4 0.0 0.0 1.1 0.0 Medium 40.7 22.4 14.8 28.7 20.5 Long 56.9 77.6 85.2 70.2 79.5 Mane length Short 4.9 0.0 0.0 2.1 0.0 Medium 48.0 16.9 39.8 31.9 31.3 Long 47.1 83.1 60.2 66.0 68.7 with Gesha stallions having the highest conformation index values among the studied populations. Multivariate analysis Stepwise discriminant function analysis selected and ranked the morphometric variables according to their importance in discriminating the studied horse popula- tions. The inclusion of height at croup and body length within the top five discriminatory variables is compara- ble with the reports of Kefena et al (2012), who clas- sified them among the top four variables to discrimi- nate Ethiopian horse populations. The results of discrim- inant function analysis showed an advanced classifica- tion (76.7%) of the studied horses into their respective populations/locations. This high value shows the dis- similarity among the studied populations. Canonical dis- criminant function analysis revealed the higher power of Can 1 than Can 2 to separate the horse populations. This shows the separation of Gesha and Masha horses from Telo horses while differences also occur between Gesha and Masha populations. However, the distances showed only the relative size differences between each popula- tion. Such differences might not necessarily be due to breed (genetic) differences (Zechner et al, 2001). There- fore, a diversity study through further genetic charac- terization is recommended to design conservation and breeding programmes. Qualitative characteristics Besides their aggressiveness and top-riding ability, the examined qualitative characteristics clearly differenti- ated the Gesha horse population from the other stud- ied populations. The majority of Gesha horses possess red body colour, red and white-striped red head colour, striped dorsal body, slightly convex face and long mane while some similarities were observed with the adja- cent Masha horses. A slight effect of sex and age on the qualitative characteristics was observed. Shorter hair, a slightly convex face and a straight-back profile were observed predominantly in stallions than mares. The current study revealed the level of relationship between age and body colour. As age advanced, the proportion of horses with white (body, head and tail) colour increased while the proportion of horses with grey and brown colours decrease, which might be due to the progressive depigmentation of the coat’s hairs (Locke et al, 2002). At birth, grey horses may have any colour but over time, white hairs begin to appear and become gradually more dominant as white hairs become intermixed with hairs of other colours. At a later age, most horses of this type ultimately become completely white, though some retain intermixed light and dark hairs (Locke et al, 2002). This is due to the presence of a greying allele of the KIT gene, which inhibits the hair follicles from producing melanin. The coat takes on a ’dappled’ pattern that increasingly becomes white. However, grey horses with a totally white coat can be distinguished from white horses by their underlying black skin, particularly around the eyes, muzzle, and genital area (Locke et al, 2002). Conclusion The studied phenotypic traits (morphometric measure- ments and qualitative characteristics) had produced reli- able information in characterizing and differentiating Gesha, Masha and Telo horse populations. Gesha horses Genetic Resources (2022), 3 (5), 36–50 Characterization of Ethiopian horses 49 were the tallest, longest and largest among the studied horse populations. Besides their size, the most important characteristics of Gesha horses are their aggressiveness, top-riding ability, red-dominated body colour, white- striped red head colour and slightly convex face. These results were also supported by the multivariate analysis, which differentiated the Gesha horse population from the Masha and Telo horse populations, and showed a relatively higher relationship with Masha horses. Further genetic characterization is recommended to confirm the above results and design conservation and breeding pro- grammes. Acknowledgments The authors are highly indebted to the Ethiopian Biodiversity Institute (EBI) for covering all the budget needs of the work. Our special appreciation also goes to the animal owners for providing their animals for this work for free. We also take this opportunity to thank the animal science experts and development agents in the districts for their endless help during data collection. A special word also goes to our friend and work partner Mr Tadesse Hunduma for mapping the study area. Author contributions All authors contributed to the study conception and design. Material preparation and data collection were performed by Amine Mustefa, Aweke Engdawork, and Seble Sinke. Amine Mustefa performed the data analysis and wrote the first draft of the manuscript. All authors commented on previous versions of the manuscript, and read and approved the final manuscript. Conflict of interest statement The authors declare that they have no conflict of interest. References Assefa, A., Demissew, S., and Woldu, Z. (2013). Floristic composition, structure and regeneration status of Masha forest, south-west Ethiopia. African Journal of Ecology 52, 151–162. doi: https://doi.org/10.1111/ aje.12098 Baneh, H. and Hafezian, S. H. (2009). Effect of envi- ronmental factor on growth traits in Ghezel sheep. African Journal of Biotechnology 8, 2903–2907. doi: https://www.ajol.info/index.php/ajb/article/view/ 60943 Bene, S., Giczi, A., Nagy, Z., Benedek, Z., Szabó, F., and Polgár, J. P. (2013). Live Weight and Body Measure- ment of Hungarian Thoroughbred Broodmares. Jour- nal of Central European Agriculture 14(3), 99–109. doi: https://doi.org/10.5513/JCEA01/14.3.1287 Bezabih, T. (2012). Study on the intensity and adoption of improved wheat varieties and associated agronomic practices in Kaffa zone, the case of Gesha woreda. MSc thesis, Indira Gandhi National Open University, India. Bodó, I. and Hecker, W. (1992). Handbook of Horse Breeders (Budapest, Hungary: Mezőgazda Publisher). Cabral, G. C., De Almeida, F. Q., Quirino, C. R., De Azevedo, P. C. N., Pinto, L. F. B., and Santos, E. M. (2004). Morphometric evalua- tion of Mangalarga Marchador horse: Confor- mation index and body proportions. Revista Brasileira de Zootecnia 33, 1798–1805. url: https: //www.researchgate.net/publication/262481742 Morphometric evaluation of Mangalarga Marchador horse Conformation index and body proportions. Central Statistical Agency (2020). Ethiopian agricultural sample survey 2019/2020 (2013 E. C.). In Report on Livestock and Livestock Characteristics, volume II, Central Statistical Agency, Addis Ababa, Ethiopia, 20- 20. Druml, T., Baumung, R., and Sölkner, J. (2008). Mor- phological analysis and effect of selection for confor- mation in the Noriker draught horse population. Live- stock Science 115, 118–128. doi: https://doi.org/10. 1016/j.livsci.2007.06.015 EBI (2016). Ethiopian National Strategy and Plan of Action for conservation and utilization of Animal Genetic Resources. FAO (2012). Phenotypic Characterization of Animal Genetic Resources. url: www.fao.org/ docrep/015/i2686e/i2686e00.pdf. Gebrmichael, A. (2019). Cattle milk production, processing and marketing situations of smallholder farmers in Telo district, Keffa zone, Ethiopia. African Journal of Agricultural Research 14(18), 806–812. url: https://academicjournals.org/journal/ AJAR/article-full-text-pdf/3270D8460881. Ghezelsoflou, H., Hamidi, P., and Gharahveysi, S. (2018). Study of factors affecting the body conformation traits of Iranian Turkoman horses. Journal of Equine Science 29(4), 91–96. doi: https: //doi.org/10.1294/jes.29.91 Kaps, M., Curik, I., and Baban, M. (2005). Genetic Analysis of the Caliber Index in Lipizzan Horses Using Random Regression. In 56th Annual Meeting of the European Association for Animal Production (EAAP). Kefena, E., Dessie, T., Han, J. L., Kurtu, M. Y., Rosen- bom, S., and Beja-Pereira, A. (2012). Morphological diversities and eco-zones of Ethiopian horse popula- tions. Animal Genetic Resources 50, 1–12. doi: https: //doi.org/10.1017/S2078633612000021 Kristjansson, T., Bjornsdottir, S., Albertsdóttir, E., Sig- urdsson, A., Pourcelot, P., Crevier-Denoix, N., and Arnason, T. (2016). Association of conformation and riding ability in Icelandic horses. Livestock Science 189, 91–101. doi: https://doi.org/10.1016/j. livsci.2016.05.010 Locke, M. M., Penedo, M. C. T., Bricker, S. J., Millon, L. V., and Murray, J. D. (2002). Linkage of the grey coat color locus to microsatellites on horse chromosome 25. Animal Genetics 33, 329–337. doi: https://doi.org/10.1046/j.1365-2052.2002.00885.x https://doi.org/10.1111/aje.12098 https://doi.org/10.1111/aje.12098 https://www.ajol.info/index.php/ajb/article/view/60943 https://www.ajol.info/index.php/ajb/article/view/60943 https://doi.org/10.5513/JCEA01/14.3.1287 https://www.researchgate.net/publication/262481742_Morphometric_evaluation_of_Mangalarga_Marchador_horse_Conformation_index_and_body_proportions https://www.researchgate.net/publication/262481742_Morphometric_evaluation_of_Mangalarga_Marchador_horse_Conformation_index_and_body_proportions https://www.researchgate.net/publication/262481742_Morphometric_evaluation_of_Mangalarga_Marchador_horse_Conformation_index_and_body_proportions https://www.researchgate.net/publication/262481742_Morphometric_evaluation_of_Mangalarga_Marchador_horse_Conformation_index_and_body_proportions https://doi.org/10.1016/j.livsci.2007.06.015 https://doi.org/10.1016/j.livsci.2007.06.015 www.fao.org/docrep/015/i2686e/i2686e00.pdf www.fao.org/docrep/015/i2686e/i2686e00.pdf https://academicjournals.org/journal/AJAR/article-full-text-pdf/3270D8460881 https://academicjournals.org/journal/AJAR/article-full-text-pdf/3270D8460881 https://doi.org/10.1294/jes.29.91 https://doi.org/10.1294/jes.29.91 https://doi.org/10.1017/S2078633612000021 https://doi.org/10.1017/S2078633612000021 https://doi.org/10.1016/j.livsci.2016.05.010 https://doi.org/10.1016/j.livsci.2016.05.010 https://doi.org/10.1046/j.1365-2052.2002.00885.x 50 Mustefa et al Genetic Resources (2022), 3 (5), 36–50 Lucena, J. E. C., Vianna, S. A. B., Neto, F., Filho, R. L. M. S., and Diniz, W. J. S. (2015). Estudo compar- ativo das proporções morfométricas entre garanhões e castrados da raça Campolina 36, 353–366. doi: http: //dx.doi.org/10.5433/1679-0359.2015v36n1p353 Mariz, T. M. A., Santos, W. K., Mota, L. F. M., Martins, R. B., Lima, C. B., Escodro, P. B., Júnior, D. M. L., Oliveira, L. P., Sousa, M. F., and Ribeiro, J. S. (2015). Avaliação de medidas morfoestruturais em equinos da raça Quarto de Milha utilizando análises de imagens. Acta Veterinaria Brasilica 9, 362–368. url: https://www.bvs-vet.org.br/vetindex/periodicos/acta- veterinaria-brasilica/9-(2015)-4/avaliacao-de- medidas-morfoestruturais-em-equinos-da-raca-quarto- de-mil/. Mcmanus, C., Falcão, R. A., Spritze, A., Costa, D., Louvandini, H., Dias, L. T., Teixeira, R. A., Rezende, M. J. M., and Garcia, J. A. S. (2005). Caracterização morfológica de equinos da raça Campeiro. Revista Brasileira de Zootecnia 34, 1553–1562. url: https:// repositorio.unb.br/handle/10482/6858. Padilha, F. G. F., Andrade, A. M., Fonseca, A. B. M., Godoi, F. N., Almeida, F. Q., and Ferreira, A. M. R. (2017). Morphometric measurements and animal-performance indices in a study of racial forms of Brazilian Sport Horses undergoing train- ing for eventing. Revista Brasileira de Zootecnia 46(1), 25–32. doi: https://doi.org/10.1590/ S1806-92902017000100005 Rensch, B. (1950). Die Abhangigkeit der relative sexual differenz von der korpergrosse. Bonner Zoologische Beitrage 1, 58–69. url: https://www. biodiversitylibrary.org/partpdf/119381. Rezende, M. P. G., Souza, J. C., Mota, M. F., Jardim, R. J. D., Ramires, G. G., Silva, R. M., and Souza, C. F. (2014). Morfometria corporal de equinos utilizados em trabalho, esporte e lazer em três munićıpios do Mato Grosso do Sul. Veterinária e Zootecnia 21, 569– 583. url: https://www.researchgate.net/publication/ 284178649 MORFOMETRIA CORPORAL DE EQUINOS UTILIZADOS EM TRABALHO ESPORTE E LAZER EM TRES MUNICIPIOS DO MATO GROSSO DO SUL MORPHOMETRY BODY OF EQUINES USED IN WORK SPORT AND LEISURE IN THREE CITIES OF MATO GROSS. Sadek, M. H., Al-Aboud, A. Z., and Ashmawy, A. A. (2006). Factor analysis of body measurements in Arabian horses. Journal of Animal Breeding and Genetics 123, 369–377. doi: https://doi.org/10.1111/ j.1439-0388.2006.00618.x SAS Institute (2002). Statistical Analysis Software (SAS). url: https://www.sas.com/en us/home.html. Torres, A. P. and Jardim, W. R. (1981). Criação do cavalo e de outros eqüinos. Nobel, São Paulo. Zechner, P., Zohman, F., Solkner, J., Bodo, I., Habe, F., Marti, E., and Brem, G. (2001). Morphological description of the Lipizzan horse population. Livestock Production Science 69, 163–177. doi: https://doi.org/ 10.1016/S0301-6226(00)00254-2 http://dx.doi.org/10.5433/1679-0359.2015v36n1p353 http://dx.doi.org/10.5433/1679-0359.2015v36n1p353 https://www.bvs-vet.org.br/vetindex/periodicos/acta-veterinaria-brasilica/9-(2015)-4/avaliacao-de-medidas-morfoestruturais-em-equinos-da-raca-quarto-de-mil/ https://www.bvs-vet.org.br/vetindex/periodicos/acta-veterinaria-brasilica/9-(2015)-4/avaliacao-de-medidas-morfoestruturais-em-equinos-da-raca-quarto-de-mil/ https://www.bvs-vet.org.br/vetindex/periodicos/acta-veterinaria-brasilica/9-(2015)-4/avaliacao-de-medidas-morfoestruturais-em-equinos-da-raca-quarto-de-mil/ https://www.bvs-vet.org.br/vetindex/periodicos/acta-veterinaria-brasilica/9-(2015)-4/avaliacao-de-medidas-morfoestruturais-em-equinos-da-raca-quarto-de-mil/ https://repositorio.unb.br/handle/10482/6858 https://repositorio.unb.br/handle/10482/6858 https://doi.org/10.1590/S1806-92902017000100005 https://doi.org/10.1590/S1806-92902017000100005 https://www.biodiversitylibrary.org/partpdf/119381 https://www.biodiversitylibrary.org/partpdf/119381 https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://www.researchgate.net/publication/284178649_MORFOMETRIA_CORPORAL_DE_EQUINOS_UTILIZADOS_EM_TRABALHO_ESPORTE_E_LAZER_EM_TRES_MUNICIPIOS_DO_MATO_GROSSO_DO_SUL_MORPHOMETRY_BODY_OF_EQUINES_USED_IN_WORK_SPORT_AND_LEISURE_IN_THREE_CITIES_OF_MATO_GROSS https://doi.org/10.1111/j.1439-0388.2006.00618.x https://doi.org/10.1111/j.1439-0388.2006.00618.x https://www.sas.com/en_us/home.html https://doi.org/10.1016/S0301-6226(00)00254-2 https://doi.org/10.1016/S0301-6226(00)00254-2 Introduction Materials and methods Locations Data collection Data analysis Results Morphometric measurements and body measure indices Multivariate analysis Qualitative characteristics Discussion Morphometric measurements Body measure indices Multivariate analysis Qualitative characteristics Conclusion Acknowledgments Author contributions Conflict of interest statement