PREDICTION OF TOTAL MUSCLE WEIGHT FROM MORPHOMETRIC LIVE TRAITS IN RABBITS USING PRINCIPAL COMPONENT ANALYSIS

Abstract

A total number of two hundred and eighteen New Zealand White rabbits were used to predict total muscle weight using nine live body measurements taken on the head, chest, loin, round and total body length and their three principle component varimax rotated scores. Total muscle weight (TMW) was more variable than any morphological live trait. The highest phenotypic correlations with total muscle weight were recorded with the live measurements of the most meat-producing regions in the body (r = 0.77, 0.75, 0.76, 0.60 and 0.58 for chest width, loin width, chest girth and abdomen girth, respectively). Results showed that out of the principal components (PCs) calculated, the first three explained 74.14% of the total variance. The PC1, PC2 and PC3 explained, respectively, 29.33, 23.23 and 21.58% of the generalized variance. Based on correlations of: (i) 0.913, 0.919, 0.582 and 0.594, PC1 is primarily a measure of width of chest and loin and girth and depth of chest, (ii) 0.705, 0.778 and 0.864, PC2 is primarily a measure of girth of chest, abdomen and round, (iii) 0.776, 0.791 and 0.696, PC3 is a measure of width of head and length of head and body. The stepwise involvement of PC1, PC2 and PC3 in regressions predicting TMW were as follows: TMW= 728.3 + 100.8 PC1 (R2 = 51%), TMW= 728.3 + 100.8 PC1 + 59.3 PC2 (R2= 69%) and TMW= 728.3 + 100.8 PC1 + 59.3 PC2 + 48.2 PC3 (R2 = 80%). Conclusively, Use of principal component analysis was advantageous for avoiding any possible multicollinearity to occur and, thus, excluding any wrong decisions to be taken when morphological body measurements were put together in a multiple regression.