MULTIVARIATE APPROACH IN THE EVALUATION OF THE PRODUCTION AND COMPOSITION OF BUFFALO MILK IN NORDESTINE SEMIARID

ABSTRACT The objective was to evaluate the production, chemical composition and hygienic -sanitary parameters of the milk of Murrah buffaloes raised in the northeastern semi-arid region. Data from 13,752 observations of lactations collected from the years of 2013 to 2016 from a herd in Rio Grande do Norte were used. The milk quality through chemical composition (contents of fat, protein, lactose, and total solids), somatic cell counts (CCS), and milk production was evaluated. Data were submitted to Pearson correlation, factorial analysis, multivariate analysis of variance per year, and per period (dry, rainy and transition) and canonical discriminant analysis by year and period. A high and positive correlation was observed for total solids and fat (0.91) and positive correlation for protein and total solids, yield, and lactose. Through factor analysis, four factors were selected that explained 88% of the total variation. The first factor was considered "milk quality factor, " the second factor as "milk production factor," and the third factor as " sanity factor." There was significant effect based on year but not significant effect based on period for the studied variables. In the canonical discriminant analysis per year, the discriminant power variables were (CCS), milk production and total solids. For periods were protein, production, lactose, fat, and total solids. Multivariate analysis was efficient in evaluating the production and chemical composition of buffalo milk and can be safely used in future studies. RESUMO Objetivou-se avaliar a produção, composição química e parâmetro de qualidade higiênico sanitária do leite de búfalas da raça Murrah criadas no semiárido nordestino. Foram utilizados dados de 13.752 observações de lactações coletados dos anos de 2013 a 2016 em plantel no Rio Grande do Norte. Foram avaliadas a qualidade do leite através da composição química (teores de gordura, proteína, lactose, e sólidos totais), contagem de células somáticas (CCS) e produção do leite. Os dados foram submetidos a correlação de Pearson, análise fatorial, análise de variância multivariada por ano e por período (seco, chuvoso e transição) e análise discriminante canônica por ano e por período. Correlação alta e positiva foi observada para sólidos totais e gordura (0,91), e correlação positiva para proteína e sólidos totais, produção e lactose. Por meio da análise fatorial, foram selecionados 3 fatores que explicaram 88% da variação total. O primeiro fator foi