Morphometric characterization and decision tree-based prediction of phenotypic traits in Pantaneiro sheep

The Pantaneiro sheep is a locally adapted genetic resource shaped by the seasonal flooding, thermal stress, and heterogeneous forage availability of the Brazilian Pantanal. Characterizing its morphological structure using objective and field-applicable traits is essential for standardized phenotyping and conservation-oriented breeding strategies. This study evaluated 211 Pantaneiro ewes raised under semi-extensive management to identify morphometric patterns and identify simple predictors of qualitative phenotypic traits. Twenty-three quantitative and ten qualitative traits were measured and analyzed using multivariate techniques, including factor analysis and hierarchical clustering, to describe morphological structure. Decision tree models were then developed to predict key phenotypic traits from morphometric variables. Four main morphological dimensions explained 64.6% of total variance, reflecting body capacity, limb robustness, and cranial morphology. Among the ten decision tree models generated, three achieved predictive accuracies above 80% head profile (86.3%), muzzle type (84.8%), and wool color (82.5%), indicating that metatarsus and tarsus circumferences are reliable non-invasive predictors of cranial and tegumentary traits. The integration of multivariate morphometry and machine learning proved effective for identifying simple predictors applicable to field management. Overall, these findings reinforce the adaptive potential of Pantaneiro ewe and their importance for sustainable and resilient livestock production in tropical floodplain ecosystems.