Discriminant analysis of morphometric traits across growth stages in native goat breeds Authors: Esra Yavuz, Hasan Önder, Uğur Şen, Mustafa Şahi̇n, Ömer Faruk Yilmaz Published: 2026 ABSTRACT AbstractBackgroundMorphometric measurements can be used for the separation of animal breeds. The hypothesis of this manuscript is based on the assumption that the discriminatory power of morphometric features can change during the growth process. For this aim Angora, Hair, Kilis, and Honamlı goat breeds in three different ages that at birth, weaning (90 days of age), and mature (one year old) animals from were evaluated.MethodsFor dimension reduction Principal Component Analysis was used and Multivariate Discriminant Stepwise Analysis Method (MDASM) was used to determine the differences and grouping levels in terms of biometric measurements of nine animals per genotype such as height at wither (HW), height at sacrum (HS), body length (BL), chest girth (CG), chest width (CW), and chest circumference (CC) for discrimination of goat genotypes.ResultsOur results showed that there were differences among ages that at birth, weaning, and mature in spite of discrimination according to biometric measurements. The highest correct classification rate obtained for mature age (91.18%) which was 25.66% higher than age of birth.ConclusionsThese findings supported that the measurements taken at mature age should be used for robust discrimination. The chest girth, chest circumference, and chest width were found the most significant variables for classification.Keywords: Goat breeds, Classification, Age effect, Correct classification rate Introduction In animal production systems, accurately identifying and classifying breeds is crucial for both productivity and the preservation of genetic resources. Breed refers to subgroups within the same species that are similar in terms of genetic, morphological, and performance characteristics and that can be inherited across generations. Therefore, the concept of breed is a fundamental unit of classification, particularly in domestic animal breeding, for determining the productivity potential of animals, understanding their environmental adaptations, and implementing breeding programs [1]. Turkey boasts rich genetic diversity in small ruminants. Native breeds such as the Angora, Hair, Kilis, and Honamlı goats are populations that evolved in different geographical regions and are distinct from each other in terms of morphological and production traits. The Angora, Hair, Kilis, and Honamlı goats representing approximately 92% of Türkiye’s goat population [2]. Body measurements are considered a reflection of individuals’ genetic makeup and their interaction with environmental factors and, are relatively easy to collect. Morphometric data, particularly based on body measurements, allows for the numerical analysis of physical differences observed between breeds and is frequently used in classification studies [3–4]. Morphometric measurements are one method used to classify animal breeds. Body measurements are considered a reflection of individuals’ genetic makeup and their interaction with environmental factors. Therefore, body measurements are preferred as a numerical criterion for determining morphological differences between breeds. However, factors such as age, sex, and rearing conditions can influence body size, so these variables are also considered in the evaluation [5]. Body measurements in goats can vary significantly depending on the physiological stages (age groups) of the individuals. Developments in the skeletal and muscular systems throughout the growth and development process cause morphometric characteristics to change over time. While significant increases in body size are observed, particularly in the early stages of rapid growth, this increase slows with age, and measurements become more stable in adulthood [6]. This can make it difficult to directly compare morphometric data across individuals in different age groups and necessitates considering the age factor when determining breed differences. Therefore, the impact of age-related physiological changes should be considered in morphometric analyses, and assessments should be made on an age-group basis [7–8]. Multivariate statistical analysis are considered particularly useful for identifying distinguishing characteristics between breed and grouping individuals. Discriminant analysis, is widely used to determine the boundaries between different groups and to test classification accuracy. Discriminant analysis not only predicts which group (genotype) individuals belong to base on their measured morphometric characteristics, but also statistically demonstrates which variables best facilitate this distinction [9–10]. This study was used to basic body measurements such as height at wither (HW), height at sacrum (HS), body length (BL), chest girth (CG), chest width (CW), and chest circumference (CC) to determine breed discrimination in Angora (Tiftik), Hair, Kilis, and Honamlı goats of different ages. We also investigated whether this discrimination changed across three different ages at birth, weaning (90 days of age), and mature (one year old) animals. This study aimed to contribute to the literature on both the identification of native goat breeds and the impact of age-related morphological variation on breed differentiation. Materials and methods Animal material The study was conducted at the Ondokuz Mayis University Agricultural Faculty’s Research and Application Farm. All the animal material of Angora, Hair, Kilis, and Honamlı goat kids (nine animals per genotype) were born in the Research and Application Farm between February 10 and March 15, 2025 as the second parity of does. In the first two weeks, all the kids were kept in the pen with their dams and allowed to suckle freely to ensure they received sufficient colostrum and milk. For two weeks, the goats were fed an average of 250 g concentrate (2800 kcal/kg dry matter metabolizable energy, 15% crude protein, and approximately 90% dry matter) and alfalfa of 1 kg per day. Until weaned, the goats were allowed to graze on pasture during the day and care for their kids overnight in the pen in two weeks after birth. In addition to grazing on pasture, the goats were fed an average of 150 g of concentrate per day during lactation. When all kids were one month old, they were treated against internal and external parasites. The kids were also introduced to ad libitum concentrate diets for creep feeding and dry alfalfa in addition to their dam’s milk until weaning in the pen to ensure adequate rumen development. At 90 days, all kids were weaned. Measurements Basic body measurements such as height at wither (HW), height at sacrum (HS), body length (BL), chest girth (CG), chest width (CW), and chest circumference (CC) to determine breed discrimination in Angora, Hair, Kilis, and Honamlı goat in three different ages that at birth, weaning (90 days of age), and the mature age (one year old) measurements were measured. These ages coded as B: birth, W: weaning, and M: mature. In body measurements, length, height and depth measurements were taken with a measuring stick, width measurements were taken with a measuring compass, and circumference measurements were taken with a tape measure [11]. Statistical analysis To determine the effective variables for B, W, and M the Principal Component Analysis (PCA) was used as a dimension reduction tool. For any group of variables to summarize the highest possible variability in a reduced number of variables, principal components, PCA can be applied with great robustness for many types of data for dimension reduction [12– 13]. Correct interpretation of PCA analysis results is crucial, and it should not be overlooked that the total variance must be explained by at least 85% when determining the number of components to be used [14]. The PCA analysis was executed using R software with FactoMineR and factoextra packages [15]. The Multivariate Discriminant Stepwise Analysis Method (MDASM) was used to determine the differences and grouping levels in terms of biometric measurements and discrimination of goat genotypes [10]. MDASM analysis was executed using SPSS [16] with license of Ondokuz Mayis University. The total correct classification rates was calculates dividing the correctly classified number of animals into total number of animals. Multivariate discriminant stepwise analysis method (MDASM) Multivariate Discriminant Stepwise Analysis Method (MDASM) is a statistical classification and variable selection approach used to identify variables that best discriminate between groups in multivariate datasets. This method is based on classical discriminant analysis and works on the principle of adding or removing variables from the model stepwise. Thus, the most effective variables that maximize between-group variation and minimize within-group variation are selected. In MDASM, variable selection is generally based on Wilks’ Lambda statistic, F statistic, or similar discrimination measures. Each variable added to the model at each step should make a statistically significant contribution to the discriminatory power between groups. Variables that do not make a significant contribution are excluded from the model. This approach reduces the problem of multicollinearity between variables, especially in studies involving numerous biometric measurements, and allows for the creation of more parsimonious models. In biometric and zootechnical studies, MDASM is used as an effective method for revealing morphological differences between different genotypes, age groups, or populations. The method increases the accuracy of classifying individuals into their respective groups while also allowing for the biological interpretation of distinguishing characteristics. Therefore, MDASM is considered a reliable and complementary statistical analysis tool in the evaluation of multivariate morphometric data [10]. Results The PCA results showed that the total variance explanation of four components (the threshold of 0.85 explanation rate was reached with four component) were 95.66%, 98.20%, and 97.80% for different ages B, W, and M, respectively. The contribution of the biometric variables height at wither (HW), height at sacrum (HS), body length (BL), chest girth (CG), chest width (CW), and chest circumference (CC) to the first four components were given in Fig. 1. Also, in Fig. 1, in the figure, the red line indicates the significance threshold for variable contributions. Contributions exceeding this threshold were interpreted as meaningful, reflecting variables with a greater influence on the principal components. The contribution of variables for first four components (B: Birth, W: Weaning, M: Mature, HW: height at wither, HS: height at sacrum, BL: body length, CG: chest girth, CW: chest width, CC: chest circumference) Variance of genotypic for contributing to the first two components were present in (Fig. 2). Variance of genotypes (Angora, Hair, Kilis, and Honamlı goat) for contributing to components (B: Birth, W: Weaning, M: Mature) The territorial maps of goat genotypes according to the discriminant function were given in (Fig. 3.) Discriminant function territorial map of goat genotypes (B: Birth, W: Weaning, M: Mature) Total correct classification rates were 65.52%, 79.31%, and 91.18% for B, W, and M, respectively. Unstandardized Canonical Discriminant Function coefficients were given in Table 1 and correct classification rates were given in (Table 2). Table 1Unstandardized canonical discriminant function coefficientsBirthWeaningMatureFunction 1Function 2Function 1Function 2Function 1Function 2HW-0.404-0.335-0.134-0.6200.279-0.143HS0.5780.0510.3340.6860.237-0.009BL0.3950.1450.144-0.1290.258-0.012CG-0.2590.067-0.4680.099-0.2120.931CW-1.226-0.460-0.5460.0830.0750.068CC-0.0080.2700.1240.155-0.107-0.228Constant-6.564-3.290-9.617-9.595-37.6091.589Abbreviations: HW height at wither, HS height at sacrum, BL body length, CG chest girth, CW chest width, CC chest circumference. Unstandardized canonical discriminant function coefficients Abbreviations: HW height at wither, HS height at sacrum, BL body length, CG chest girth, CW chest width, CC chest circumference. Table 2Correct classification rates of genotypes for agesEstimated (%)BirthHonamlıHairKilisAngoraHonamlı100.00Hair16.6750.0033.33Kilis45.4545.459.10Angora20.0080.00Weaning Honamlı71.4314.2914.29 Hair83.3316.67 Kilis9.1081.829.10 Angora20.0080.00Mature Honamlı100.00 Hair66.6733.33 Kilis100.00 Angora100.00 Correct classification rates of genotypes for ages The total variance explanation of four components were 95.66%, 98.20%, and 97.80% for different ages B, W, and M, respectively according to the PCA results. Based on the PCA results, these rates were so high and reassuring [17]. For the age of birth all variables except chest circumference had significant contribution to the PCA. Chest circumference was below the significance threshold and therefore did not contribute significantly to the variance. For the weaning age and the mature age chest girth, chest circumference, and chest width had significant and nearly similar contributions to the PCA. The height at sacrum, height at wither, and body length were found insignificant to contribution to the PCA for the weaning age and the mature age (Fig. 1). Variance of genotypes (Angora, Hair, Kilis, and Honamlı goat) for contributing to the first two components showed that the highest variances were obtained at the age of weaning. The lowest variances were obtained at the mature age except for Angora goat which showed the highest variance at all ages (Fig. 2). The territorial maps’ size driven using unstandardized Canonical Discriminant Function coefficients were wider for mature age (-10:10) than the ages of birth and weaning (-4:4), which indicated that at the mature age the genotypes discriminated better. The results showed that the correct classification rates were increased from 65.52% to 91.18% with increasing age. The Honamlı genotype was exactly discriminated with 100% correct classification rate at age of birth and mature age, but misclassified at weaning age by being mixed with Hair (14.29%) and Kilis (14.29%) genotypes. The Hair genotype had misclassification in all age groups. For Hair goat correct classification rate was only 50% at birth and misclassified by being mixed with Honamlı (16.67%) and Kilis (33.33%) genotypes. The Hair genotype misclassified by being mixed with Kilis genotype with misclassification rates 16.67% and 33.33% at weaning and mature ages, respectively. The Kilis genotype had misclassification in birth and weaning age groups. For Kilis genotype correct classification rate was only 45.45% at birth and misclassified by being mixed with Hair (45.45%) and Angora (9.10%) genotypes. The correct classification rate was increased to 81.82% at weaning and misclassified by being mixed with Hair (9.10%) and Angora (9.10%) genotypes for Kilis genotype. The Kilis genotype was exactly discriminated with 100% correct classification rate at mature age. To correct classification rate of the Angora genotype was 80% at ages of birth and weaning by being mixed with Kilis genotype (20%) and 100% at the mature age (Fig. 3; Table 2). Discussion The high cumulative variance explained by the first four principal components in the present study aligns with previous morphometric research, demonstrating that PCA effectively summarizes body conformation in goats. For example, Aduba et al. [14], reported that four principal components explained approximately 74% of total variation in different goat breeds, highlighting the utility of PCA in capturing morphological diversity across age groups and populations. Al-Atiyat et al. [18] used Multivariate Discriminant Analysis on three sheep breed (Naemi, Najdi and Hari) and found the correct classification rate of 77% using body and carcass measurements. Macena et al. [19] discriminated goat carcasses according to gender × slaughter weight groups. They found 87.5% correct classification rate. Maksimovic et al. [4] used discriminant analysis on three goat genotypes (Serbian White, Saanen and Balkan) and found 98.98% total correct classification rate using morphometric characteristics and indicated that there was a clear separation between Serbian White, Saanen, and Balkan goats. In Maksimovic et al. [4], work the heart girth, head length, chest depth, head width, pelvic width, and body length were the significant variables but our results showed that chest girth, chest width, and chest circumference were significant variables according to PCA results. They didn’t have any report about the dimension reduction. Ofori et al. [20], used morphometric characterization to discriminate four different dwarf goat populations with 67.82% correct classification rate which is too low than our results. Ofori et al. [20], also mentioned that chest girth and heart girth were the most significant variables. Okoro et al. [21] also worked to discriminate the three state for Dwarf goat populations using morpho‑biometric measurements and they reached 62.4% correct classification rate. The high misclassification rate of Ofori et al. [20] and Okoro et al. [21] can be seen as reasonable because they classified populations within same genotype. Rotimi et al. [22], used discriminant analysis to classify six indigenous cattle genotypes and found 65% correct classification rate. Tenagne et al. [23], discriminate three goat genotypes (Sahel, Sokoto Red and West African Dwarf) using the explanatory variables body weight, height at wither, body length, paunch girth, heart girth, and ear length with 99% correct classification rate which is high as our results. Zaitoun et al. [24], discriminated five goat genotype using morpho-structural characteristics and found 60.05% correct classification rate. Honamlı goats are easily separated at birth and in mature age, but not at weaning. This difference can be explained the growth and development structure that at the birth and the mature age Honamlı goats have higher values than others, but at the weaning the Honamlı goats mixed with Hair and Kilis because of Hair and Kilis goats’ rapid growth. Also, the difficulty in separating the Hair goats can be due to genetic similarity that Kilis and Honamlı goats are originated from Hair goats [25]. Conclusion Our results showed that there were differences among ages that at birth, weaning (90 days of age), and mature age (one year old) in spite of discrimination of Angora, Hair, Kilis, and Honamlı goats according to biometric measurements. The highest correct classification rate obtained for mature age (91.18%) which was 25.66% higher than age of birth. These findings supported that the measurements taken at mature age should be used for robust discrimination. The chest girth, chest circumference, and chest width were found the most significant variables for classification. The results also revealed that discrimination accuracy differed among goat genotypes, indicating that morphometric distinctiveness is breed-dependent. Authors' contributions HO and EY prepared the manuscript. MS, HO and EY contributed to its review and revision. HO, US and OFY were responsible for the sample collection. HO conducted the analyses. All authors read and approved the final version of the manuscript. Funding The authors acknowledges that this study was supported by Ondokuz Mayis University Scientific Research Office with project number of BAP01-2024-5282 and BAP01-2024-4873. This study was also supported by Şırnak University Scientific Research Office with project number of 2024.FNAP.02.04.01. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The authors confirm that the ethical policies of the journal, as noted on the journal’s author guidelines page, have been adhered to. This study was approved by the Local Ethics Committee for Animal Experiments of Ondokuz Mayıs University (approval date: February 29, 2024, protocol code: 2024/3). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Citations Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.