Comparison between two multivariate analyses for the evaluation of genetic divergence for carcass and meat quality traits in alternative lines of chicken

The univariate analysis is becoming harder to use due the increasing number of characteristics of economic importance in agricultural industry. The multivariate approach provides an answer to this issue, allowing to analyze several traits when comparing different methodologies, genetics and products. This study aimed to use and compare the canonical variable analysis (CVA) and principal component analysis (PCA) to evaluate 7 genotypes of alternative lines of chicken (Caboclo, Carijo, Colorpak, Gigante Negro, Pesadao Vermelho, Naked Neck and Tricolor). The study evaluated 840 male chicks reared at 91 days from these genotypes, in a completely randomized design with 4 replicates per genotype. Different traits (23) were measured, and only 7 remained relevant after the multivariate approach: Carcass yield, Breast Yield, Back Yield, Cooking Loss, lightness, yellowness and water holding capacity. Both analyses remained with two variables explaining the variating. The Pearson correlation was used to measure the traits responsible for the most variance between genotypes. On the principal component cooking loss, carcass and breast yields, and the color parameters lightness and yellowness were the most relevant, while on canonical variables it was carcass yield, breast yield, lightness, yellowness and back yield. Both analysis resulted in similar conclusion, allowing to classify the genotypes in three major groups: 1 (Pesadao vermelho, Carijo, Colorpak, Nacked Neck), 2 (Gigante Negro and Caboclo) and 3 (Tricolor). PCA and CVA facilitate the interpretation of data with several traits of importance, showing the main traits responsible for genetic divergence.