Análise multivariada de características reprodutivas e produtivas de vacas da raça Holandesa
Carregando...
Data
2016-04-29
Autores
Castaño, Pablo Dominguez [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O desempenho medido por características de produção de leite e de eficiência reprodutiva são os principais critérios de seleção nos rebanhos leiteiros devido às consequências diretas na rentabilidade econômica do sistema. O objetivo deste trabalho foi identificar as variáveis que representem a maior parte da variação fenotípica e genética, além de estimar os parâmetros genéticos de características de produção e reprodução. Foram analisados registros de 5.217 matrizes da raça Holandesa utilizando análise de componentes principais (ACP) e análise de fatores (AF). As características avaliadas foram: produção de leite aos 305 dias de lactação (P305), produção de leite por dia de intervalo de partos (PLIEP) e pico de lactação (PICO), intervalo da data do parto ao primeiro cio (I1C), intervalo da data do parto a última cobertura (IUC), intervalo de partos (IP) e duração da gestação (DG). A estrutura de associação das características foi verificada utilizando os métodos ACP e AF para extrair novas variáveis a partir da matriz de correlação dos fenótipos por meio do software estatístico SAS®. Os resultados indicaram que ACP e AF extraíram três e quatro variáveis latentes retendo 81,5 e 88,9% da variação total dos dados, respectivamente. As características que tiveram maior participação nos componentes e fatores selecionados foram P305, PLIEP, PICO, IUC e IP. A característica DG apresentou independência frente ao conjunto de variáveis avaliadas. Parâmetros genéticos foram estimados utilizando seis modelos diferentes: modelo multicaracterística e cinco modelos de posto reduzido ajustando aos 2, 3, 4, 5 e 6 componentes principais genéticos. Para todos os modelos foram considerados os efeitos aleatórios genético, ambiente permanente e residual e os efeitos fixos de grupo contemporâneo e idade ao parto da vaca. Os componentes de (co)variâncias foram estimados pelo método da máxima verossimilhança restrita. Utilizou-se o critério de informação Bayesiano (BIC) para comparar os diferentes modelos. Os resultados indicaram que cinco componentes principais (CP5) são necessário para modelar a estrutura de (co)variância genética entre as sete características avaliadas. A correlação genética assim como a fenotípica evidenciou que características associadas à produção de leite estavam fortemente correlacionadas entre si, mas a correlação destas com as características reprodutivas foi baixa. Correlação genética entre características produtivas foi alta (>0,90), entre características reprodutivas variou de -0,19±0,15 (DG com IUC) a 0,99±0,00 (IP com IUC) e entre estes dois conjuntos de características variou de -0,14±0,14 (PLIEP com IUC) a 0,23±0,09 (PICO com DG). Estimativas de herdabilidade no modelo CP5 para as características P305, PLIEP PICO, I1C, IUC, DG e IP foram 0,18±0,02, 0,18±0,02, 0,16±0,02, 0,09±0,02, 0,04±0,01, 0,11±0,02 e 0,03±0,01 respectivamente. ACP reduziu o número de parâmetros a estimar, sem reduzir a qualidade do ajuste. Características IUP ou IP e P305, devem ser consideradas como critério de seleção em programas de melhoramento genético em bovinos leiteiros, devido a sua importância econômica e variabilidade genética.
The main causes of discard in dairy herds are the milk production and the reproductive efficiency traits due to direct consequences on economic profitability of the system. These two groups of characteristics are interconnected, occurring correlated response when selected some of them. The aim of this study was to evaluate the structure of phenotypic and genetic associations in addition to estimate genetic parameters of production and reproduction traits. Records of 5,217 Holstein cows were analyzed using principal component analysis (PCA) and factor analysis (FA). The traits studied were: 305-day accumulative milk yield (Y305), milk yield per day of calving interval (MYCI) and peak yield (PY), interval between calving and first estrus (CFE) interval between calving and last service (CLS), calving interval (CI), and gestation length (GL).The association structure of the traits was verified using the ACP and AF methods to extract new variables from the phenotypes correlation matrix using statistical software SAS®. The results indicate that ACP and AF managed to extract three and four latent variables explained 81 and 88.9% of the total variance of the dataset, respectively. The principal component analysis was more efficient than the analysis of factors, get summarize the information in fewer latent variables. Genetic parameters were estimated using six different models: standard multi-trait model and five reduced rank models fitting the first 2, 3, 4, 5 and 6 genetic principal components. The models include the random animal genetic effect, permanent environmental effects and residual effects and the fixed effects of the contemporary groups and age of cow. Variance components were estimated by restricted maximum likelihood method. Schwarz Bayesian information criteria were used to compare models. The results indicate that at most five principal components are required to model the genetic covariance structure among the seven traits. The genetic and phenotypic correlations showed to the production traits were more strongly correlated among themselves and more weakly correlated with the reproductive traits. Genetic correlation between yield traits was high (>0,90±0,02), between reproduction traits ranged of -0.19 ± 0.15 (GL with CLS) to 0.99 ± 0.00 (CI with CLS) and between these two sets of traits ranged of -0.14 ± 0.14 (MYCI with CLS) to 0.23 ± 0.09 (PY with GL).The estimates of heritability in CP5 model for the traits Y305, MYCI, PY, CFE, CLS, GL and CI were 0.18 ± 0.02, 0.18 ± 0.02, 0.16 ± 0.02, 0.09 ± 0.02, 0.04 ± 0.01, 0.11 ± 0.02 and 0.03 ± 0.01 respectively. ACP decreased the number of parameters to be estimated without reducing the goodness-of-fit. Traits CLS or CI and P305 should be considered as selection criteria in breeding programs in dairy cattle, due to its economic importance and genetic variability.
The main causes of discard in dairy herds are the milk production and the reproductive efficiency traits due to direct consequences on economic profitability of the system. These two groups of characteristics are interconnected, occurring correlated response when selected some of them. The aim of this study was to evaluate the structure of phenotypic and genetic associations in addition to estimate genetic parameters of production and reproduction traits. Records of 5,217 Holstein cows were analyzed using principal component analysis (PCA) and factor analysis (FA). The traits studied were: 305-day accumulative milk yield (Y305), milk yield per day of calving interval (MYCI) and peak yield (PY), interval between calving and first estrus (CFE) interval between calving and last service (CLS), calving interval (CI), and gestation length (GL).The association structure of the traits was verified using the ACP and AF methods to extract new variables from the phenotypes correlation matrix using statistical software SAS®. The results indicate that ACP and AF managed to extract three and four latent variables explained 81 and 88.9% of the total variance of the dataset, respectively. The principal component analysis was more efficient than the analysis of factors, get summarize the information in fewer latent variables. Genetic parameters were estimated using six different models: standard multi-trait model and five reduced rank models fitting the first 2, 3, 4, 5 and 6 genetic principal components. The models include the random animal genetic effect, permanent environmental effects and residual effects and the fixed effects of the contemporary groups and age of cow. Variance components were estimated by restricted maximum likelihood method. Schwarz Bayesian information criteria were used to compare models. The results indicate that at most five principal components are required to model the genetic covariance structure among the seven traits. The genetic and phenotypic correlations showed to the production traits were more strongly correlated among themselves and more weakly correlated with the reproductive traits. Genetic correlation between yield traits was high (>0,90±0,02), between reproduction traits ranged of -0.19 ± 0.15 (GL with CLS) to 0.99 ± 0.00 (CI with CLS) and between these two sets of traits ranged of -0.14 ± 0.14 (MYCI with CLS) to 0.23 ± 0.09 (PY with GL).The estimates of heritability in CP5 model for the traits Y305, MYCI, PY, CFE, CLS, GL and CI were 0.18 ± 0.02, 0.18 ± 0.02, 0.16 ± 0.02, 0.09 ± 0.02, 0.04 ± 0.01, 0.11 ± 0.02 and 0.03 ± 0.01 respectively. ACP decreased the number of parameters to be estimated without reducing the goodness-of-fit. Traits CLS or CI and P305 should be considered as selection criteria in breeding programs in dairy cattle, due to its economic importance and genetic variability.
Descrição
Palavras-chave
Gado Holandês, Modelos de posto reduzido, Parâmetros genéticos, Correlação, Fertilidade, Produção de leite, Correlation, Fertility, Genetic parameters, Holstein cattle, Milk yield, Reduced rank models