Alternativas para avaliação genética na raça gir leiteiro

• Autor(a) principal: Reis Filho, João Cruz
• Data de Publicação: 2009
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: LOCUS Repositório Institucional da UFV
• Texto Completo: http://locus.ufv.br/handle/123456789/1309
• topic: Modelo de regressão aleatória; Modelo autoregressivo; Modelo de repetibilidade; Gir leiteiro; Endogamia; Random regression models; Autoregressive model; Repeatability model; Gyr cattle; Inbreeding; CNPQ::CIENCIAS AGRARIAS::ZOOTECNIA::GENETICA E MELHORAMENTO DOS ANIMAIS DOMESTICOS

Pedigree and milk production files from the National Brazilian Dairy Gyr Breeding Program (PNMGL), led by Embrapa Dairy Cattle in partnership with the Brazilian Association of Dairy Gyr Breeders (ABCGIL) were used in research alternatives for the genetic evaluation of milk production in the breed and to provide subsidies for the development of PNMGL. Comparative studies of genetic evaluations for cumulative 305 days milk yields (PL305), using random regression and repeatability models - the latter considered the traditional method - and studies in test day records of milk yields (TD), by means of autoregressive and repeatability models, which were also compared with the traditional method. Also, the possibility of the incorporation of lactations with non-conventional drying off causes (death or separation of the calf, illness, death or sale of the cow and removal of the cow from the official milk control) and the genetic relationship of proven bulls with the population in order to provide subsidies for the selection of young bulls for the progeny test was studied. In the edition of the data file, only the first three lactations were retained, also eliminating lactations whose first calving was outside the range of 24 and 66, 36 and 84, 48 and 102 months for the first, second and third parity orders, respectively. Controls whose production were below or above 3.5 standard deviations from the average in the test day were eliminated. Lactations with three or fewer controls were also excluded. Likewise, there was a minimum of three lactations for each class of fixed effect. The fixed effects included in the models varied according to the analysis, and may contain the herd-calving year, birth month or month and year of the test, parity or class of days in milk nested in parity order, number of milkings and management conditions. In the case of the analysis with TD, the tests were grouped into 31 classes of days in milk (by parity order), with the first class comprising the yields produced between the 6th and 10th day, the second of the 11th to 20th day and so on successively at intervals of 10 days where the last class only had yields produced between 301 and 305 lactation days. On average, each cow presented eight tests per lactation, regardless of parity. Except for the file with the non conventional drying off causes, which used 140,369 TD records of 17,733 lactations of 10,900 cows, the other files analyzed used the PL305 of 15,065 lactations or their respective 125,191 TD records from 9,242 cows. The cows calving occurred between 1983 and 2009 from Gyr cows subjected to two or three milkings, at three different levels of management: extensive (grass production), semiextensive (grass production with concentrate supplement) and intensive (confined animals and/or with galactogenic substance use). A recursive algorithm was used to reduce the pedigree file of 50,371 to 17,483 animals that were considered informative, starting from the output file of 10,900 cows and looking for all of their ancestors who were connected to at least two animals in the database. Genetic evaluations were made using MTDFREML system programs (repeatability models), REMLF90 (random regression model) and MATLAB routines (autoregressive model). Using the first three lactations of PL305, the random regression model can be used as an alternative to enhance the accuracy of genetic evaluations, providing higher estimates of heritability (0.26 to 0.28), predicted similar genetic merits (correlations above 0.97) and high coincidence between the selected animals (above 92%) compared with the traditional model of repeatability. Using the TD milk yields in the first three lactations, the autoregressive model provided smaller residual variances, a higher additive genetic variance and higher heritability estimates (0.26 to 0.32), and predicted genetic merit with greater accuracy being a recommended alternative for the genetic evaluation of the Dairy Gyr, compared to the repeatability model, which used both the TD as well as the PL305, whose heritability estimates were 0.18 and 0.24, respectively. In this sense, the use of an autoregressive model, by examining the daily milk yields, allows better use of the data and the inclusion of lactations not traditionally used in genetic evaluation, which can increase the accuracy of predicted genetic merits (in general, averaging 5%) without, however, generating major changes in the classification of the evaluated animals (in the full file, sample and rank correlations above 95%). Finally, the estimated average inbreeding coefficient for the active population of Dairy Gyr was 1.04%, indicating a trend of the more inbred animals not staying in their herds. Using the averagerelatedness coefficient of bulls with the population could prevent the increase of inbreeding in Gyr breed, to help in selection of candidates for the progeny test from lines less represented in PNMGL.