Fig. 3

Principal component analysis showing the relationship of breeds (a) with traits associated with stature (b) and the corresponding phenotypic breed scores (c). Breeds (a) colored in grey are the breeds with more than 50% of missing variables and, thus, their position on the PCA must be interpreted carefully. The variables (b) are colored according to quality of the representation, which is measured by squared cosine between the vector originating from the element and its projection on the axis. The variables that contribute most to the separation of the trait into PC1 and PC2 are colored black. Breeds (a and c) are colored according to their genetic similarity. Breeds (a and c) in green are genetically Landrace-like breeds, in purple Large White-like breeds, in blue Iberian-like breeds, in red Duroc-like breeds, and in light blue the Gascon and Basque breeds