Prediction of bodyweight from body measurements in rabbits using principal component analysis

Abstract

The study aimed atdescribing objectively the interdependence among the biometric traits and topredict body weight from their orthogonal body shape characters using principalcomponent factor analysis. Body weight and ten biometric variables namely bodylength, height at wither, thigh circumference, shoulder to tail drop, earlength, heart girth, tail length, length of front and back leg and nose toshoulder were measured on eight week old 104 F1 progeny Hyla rabbits of twogenetic groups (NZWXNZW purebred and NZWXCAL crossbred rabbits). General linearmodel and principal component analysis procedure of statistical analysis system(S.A.S 9.0) was used to compute the variations between the two breeds. Pairwise correlations between bodyweight and biometric traits were positive andhighly associated (r = 0.60-0.90, 0.62–0.94; P<0.01) for Hyla purebred andcrossbred rabbits. In the factor solution of the principal component analysis,with varimax rotation of the transformation, two factors were identified for thefirst genetic group and three factors for the second genetic group (ratio ofvariance = 83.97 and 89.88% for NZWXCAL crossbred and NZWXNZW purebred rabbitsrespectively). The first factor in each case accounted for the greatestpercentage of the total variation, and was termed general size. The subsequentfactors (indices of body shape) presented patterns of variation independent ofgeneral size. The principal component based regression models, which arepreferable for selecting animals for optimal balance, accounted for 88% of thevariation in the body weight for both Hyla purebred and crossbred rabbitsrespectively.