Relationship between reproductive and productive traits in Holstein cattle using multivariate analysis

• Autor(a) principal: Dominguez-Castano, Pablo [UNESP]
• Data de Publicação: 2020
• Outros Autores: Vargas de Oliveira, Matheus Henrique [UNESP], El Faro, Lenira, Vasconcelos Silva, Josineudson Augusto de [UNESP]
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1111/rda.13679; http://hdl.handle.net/11449/196778
• topic: correlation; dairy cattle; multivariate statistics; production; reproduction

Multivariate procedures are used for the extraction of variables from the correlation matrix of phenotypes in order to identify those traits that explain the largest proportion of phenotypic variation and to evaluate the relationship structure between these traits. The reproductive traits (days from calving to first estrus [CFE], days from calving to last service [CLS], calving interval [CI] and gestation length [GL]) and milk production traits (milk yield at 305 days of lactation [MY305], peak yield [PY] and milk yield per day of calving interval [MYCI]) of 5,217 Holstein females (primiparous and multiparous) were measured. Principal component analysis (PCA) and factor analysis of the correlation matrix were used to estimate the correlation between traits. Analysis grouped the seven traits into three principal components and four latent factors that retained approximately 81.5% and 88.9% of the total variation of the data, respectively. The production variables exhibited positive phenotypic correlation coefficients of high magnitude (>.67). The phenotypic correlation estimates between the productive and reproductive traits were low, ranging from .13 to .22. A strong association (.99) was observed between CLS and CI. Our results indicate that multivariate analysis was effective in generating correlations between the traits studied, grouping the seven traits in a smaller number of variables that retained approximately 81% of the total variation of the data.