Bases para o melhoramento genético da resistência à mastite em rebanhos Guzerá

• Ano de defesa: 2018
• Autor(a) principal: Silva, Roberta Polyana Araújo da
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Correlação genética; Contagem de células somáticas; Herdabilidade; Inferência bayesiana; Modelo animal
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/63407

Somatic cell count (SCC) has been used as a tool to reduce the incidence of mastitis. Thus, breeding programs for dairy cattle have been using CCS on a logarithmic scale (ECS) as a selection criterion for udder health. In this work, two studies were performed to estimate genetic parameters for CCS. In the first one this trait was analyzed together with milk production traits, estimating their genetic associations, and also verified two CCS normalization strategies (CCS1 = Log10 (CCS); CCS2 = Log2 (CCS / 100) +3). In the second, after defining the best CCS standardization method, its components of variance and genetic parameters were estimated in two ways: 1) CCS in different lactations of the same animal under animal model of repeatability and 2) CCS on test-day, with animal model of random regression. The data file contained information on 6,513 cows from 95 herds. In both studies, Bayesian inference was used via Gibbs sampling. Single chains of 2,000,000 iterations were used with sampling discard of the first 5,000 chains and sampling period every 50 iterations. The deviation information criterion (DIC) was used to evaluate the best transformation for normalization of CCS data. The heritability estimates for the productive traits and for the CCS evaluated in the first study were low, indicating small probability of expressive genetic gains, from the direct selection for these traits. However, the estimated repeatability indicated an increase potential at this potential, if the effects of permanent environment were reduced. The genetic correlations between the productive traits indicated potential of correlated reperse from simple trait selection. According to the available data structure, the normalization of CCS by log 10 (Log10 (CCS)) was more efficient and produced better estimates than the normalization by the ECS method (Log2 (CCS / 100) +3). In the second study, the random regression model was more effective in capturing the genetic variability of CCS of the evaluated herds. This model allowed better detailing of the genetic parameters during lactation. Despite the trend of low CCS heritability values, it was observed that the best selection period for CCS in this dataset would be between 50 and 200 days of lactation.