Repeatability and random regression models to estimate genetic parameters for oocyte and embryo production in the Gir breed

Context. Greater production of oocytes and embryos from Gir donors contributes to greater fertility and genetic progress. Aims. This study aimed to obtain genetic parameters for oocyte and embryo production in the Gir breed. Methods. Repeatability and random regression models were applied to data consisting of 17 526 Ovum Pick Up observations from 1641 Gir donors from five different herds. Single and multi-trait analyses were carried out with the application of both models for the traits: number of viable oocytes, number of total oocytes and number of embryos, using the BLUPF90 family programs. Legendre polynomials of second order were used in the random regression model. Key results. Considering the repeatability model, additive genetic variance ranged from 0.06 to 0.13 and permanent environment variance ranged from 0.05 to 0.08 for all evaluated traits. Residual variance ranged from 0.30 to 0.45. Heritability estimates were 0.10 for number of embryos, 0.24 for total oocytes, and 0.25 for viable oocytes. Repeatability estimates were moderate, ranging from 0.20 to 0.40, and genetic correlation estimates were always above 0.80. Phenotypic correlation was high only between viable and total oocytes (0.95), and moderate in the other cases. Random regression model results were consistent with those from the repeatability model. The heritability values remained similar throughout the donors' ages, with moderate values for viable and total oocytes, and low values for number of embryos. Genetic correlations among ages for each trait were moderate to high. Also, the genetic correlations between traits within each age were high, with values always above 0.7. Conclusions. Selection of Gir donors for total oocyte production at any time, between 1 and 16 years of age, might lead to an increase in the number of viable oocytes and embryos obtained, but it's preferable at younger ages to hasten genetic progress. Repeatability models could be the best method, as they require less computational effort when compared to the random regression models and the parameter estimates do not vary substantially throughout different ages of the donor.