Principal component and stepwise discriminant analysis of the morphometric traits of three dual-purpose breeds of chicken

Abstract

Using the principal component procedure of SAS, 10 variables; body weight (BW), body length (BL), breast girth (BG), shank length (SL), thigh length (TL), comb length (CL), comb height (CH), wing length (WL), wattle length (WAL) and wattle thickness (WT) obtained from three dual-purpose chicken breeds (Shika, Kuroiler, and Sasso), variables were separated. Similarly, stepwise discriminant analsis procedure of the SAS software was employed to evaluate variables that contribute to the overall differences in breeds. Results showed three principal components (PC1, PC2 and PC3) were extracted for all the breeds and pooled data. 45.60% of total variance was accounted for by PC1, 28.17% by PC2 and 16.22% by PC3. The principal components partitioning of total variance were 50.80, 15.10 and 9.70%, 50.82, 19.90 and 14.90%, and 48.63, 14.00 and 12.67% for Shika, Kuroiler, and Sasso breeds respectively, with different factor loadings. Communalities ranged from 0.43-0.97, 0.45-0.83, 0.45-0.88 and 0.45-0.90, indicating that a good amount of variance was accounted for. Stepwise discriminant analysis indicated that seven morphometric traits, TL, SL, WAL, WL, CL, CH and BL, contributed significantly (P<0.001) to the separation of the birds into breeds. TL and SL, however, indicated higher discriminating power compared to others. The principal component analysis allowed for better understanding of the complex correlations among traits and reduced the number of traits along with high communalities, using only PC1, PC2 and PC3the first three PCs, without loss of information. Summary of stepwise discriminant analysis shows that lengthwise measures of long bones of the body of chickens, such as thigh and shank, are viable metrics for phenotypic differentiation of birds in the studied population.

Keywords: Principal Component, Stepwise discriminant Analysis, Morphometric traits, chickens