Application of discriminant analysis, classification and regression tree (CART) methods to some morphometric data for...

Genetic diversity is essential for the adaptation of organisms to changing environments and the dynamic maintenance of their populations, and indigenous animal breeds serve as valuable reservoirs of both genetic and cultural heritage. Kouri cattle, native to the Lake Chad Basin, are notable for their large bulbous horns and adaptability to aquatic environments. Morphometric measurements (height at withers, pelvis width, body length, horn length, ear length etc.), combined with discriminant and classification and regression tree analyses, provide a cost-effective and efficient approach for classifying these breeds and identifying key traits that distinguish them. Discriminant analysis results showed that breeds were correctly classified at 82.5% for males and 81.4% for females. Kouri cattle from the Mamdi location had the highest accuracy rates for both males and females (100% and 93.8%), while females from Kouloudia exhibited a lower accuracy rate (50%). The analysis revealed that some individuals from the Kaya location showed partial overlap with the Mamdi and Kouloudia groups, reflecting morphometric similarities between the breeds. In CART-based classification, male cattle were correctly assigned to their breeds at a rate of 82.5%; Mamdi achieved 100% accuracy, Kouloudia 80%, and Kaya 72.2%. For female cattle, the overall classification accuracy was 82.2%, with Mamdi at 89.6%, Kouloudia at 55.9%, and Kaya at 93.6%. The analysis showed that most individuals from the Mamdi location formed distinct terminal nodes, while some Kouloudia and Kaya individuals overlapped, reflecting their morphometric similarities. Overall, CART effectively distinguished the Kouri cattle breeds reared at the three different locations using morphometric characteristics. This study investigates the practical utility of selected morphometric traits in differentiating Kouri cattle breeds and contributes to their morphological characterization.