Avaliação de modelos para estimação de componentes de (co)variância em características de desempenho e reprodutivas em suínos

Abstract

RESUMO -Foram analisados dados de características de desempenho e reprodutivas de suínos de uma linha em desenvolvimento da raça Large White, para verificar a importância da inclusão dos efeitos materno e comum de leitegada nos modelos de avaliação genética animal, utilizando quatro diferentes modelos: modelo 1 -efeito genético aditivo direto; modelo 2 -efeito genético aditivo direto e materno; modelo 3 -efeito genético aditivo direto e comum de leitegada; e modelo 4 -efeito genético aditivo direto e materno e comum de leitegada. As características foram avaliadas pelo método da máxima verossimilhança restrita (REML), obtendo um valor de verossimilhança para cada característica. O teste da razão de verossimilhança foi aplicado para verificar qual o modelo mais adequado à avaliação genética. Foram obtidas as correlações amostral e de ordem entre os valores preditos para cada característica, em diferentes modelos, com o objetivo de verificar possíveis alterações no ordenamento das predições dos valores genéticos dos animais, quando o modelo mais adequado não fosse utilizado. O modelo 4 foi o indicado para as características de desempenho e idade ao primeiro parto e o modelo 1 foi o mais indicado para as demais características reprodutivas. Para algumas características em que o modelo 4 foi mais adequado, as correlações entre os valores genéticos preditos, por este modelo e os demais, foram próximas à unidade, indicando que, em condições de limitações computacionais, modelos menos parametrizados poderiam ser utilizados sem comprometer o ganho genético real.Palavras-chave: correlação amostral, correlação de ordem, efeito materno, efeito comum de leitegada, teste da razão de verossimilhança Evaluation of Models to Estimate (co)variance Components in Performance andReproductive Traits of Swine ABSTRACT -Large White data set with performance and reproductive traits for a development line were analyzed in order to verify the importance of including maternal and common litter effects in genetic evaluation models, using four different models: model 1 -direct genetic effect; model 2 -the direct and maternal genetic effects; model 3 -the direct and common litter effects; and the model 4 -the direct, maternal and common litter effects. The traits were analyzed by Restricted Maximum Likelihood method (REML). The Likelihood ratio test was applied in order to verify which model was more adapted to the genetic evaluation. Pearson and Spearman correlation among predicted values for each trait in different models were obtained, in order to verify ranking alterations when the most adapted model was not used. The model 4 was the most adapted to performance traits and age at first farrowing and, for the litter traits, the model 1.For some traits, in which the model 4 was the most adapted, the correlations between predicted values, by the model 4 and the remaining, were close to unit, indicating that with limited computational conditions, models less parameterized could be used without prejudice to the real genetic gain.

Genetic and environmental effects on age at first farrowing in sows in southeastern Brazil

“…A mean value lower than the one found in the present study was reported by Torres Filho et al (2004), who found a mean of 318.56 days in animals of the Large White breed. This difference may be due to the fact that this study was performed with commercial herds, for which each breeder determines in his own way the best time to have his sows mated.…”

“…The author considered at the time that this value was rather high. Based on this statement, on the findings of Torres Filho et al (2004) and on the results obtained in this study, we suggest that there has been genetic progress and/or environmental improvement regarding this trait in Brazil.…”

“…Some producers delay the entry of their sows in reproduction, determining an age or weight for starting their reproductive life, thus making it even more difficult to identify the sexually precocious females. However, the herd used by Torres Filho et al (2004) originates from a genetic improvement program, which makes it likely that they received better follow-up for this purpose. Roehe and Kennedy (1995), who worked with animals of the Yorkshire and Landrace breeds in Canada, obtained mean values of 365 and 360.5 days, respectively.…”

Genomic selection for slaughter age in pigs using the Cox frailty model

“…The heritability estimates found in this study using the L1 and L2 models based on 120 markers with the largest effects agree with those obtained by Torres Filho et al (2004), who reported values of between 0.13 and 0.20 for age until achieving 100 kg in Large White pigs. For the models S1 and S2, the heritability was calculated using Equation 11, with the error variance replaced by 1 / (1 -c), where c is the proportion of censored data.…”

Genetic trends in the performance and reproductive traits of pigs

“…According to Torres Filho et al (2004), the most suitable model for evaluating the performance traits and DAFF includes the direct genetic effect of the animal, the maternal effect and the litter common effect. For litter traits, the most suitable model includes only the direct genetic effect of the animal.…”