224 USE OF STATISTICAL MODELS BASED ON BAYESIAN INFERENCE TO ESTIMATE FIELD FERTILITY OF NELORE BULLS BY USING DATA OBTAINED IN AN IN VITRO FERTILIZATION PROGRAM

Abstract

The aim of the present study was to analyze the use of in vitro fertilization to predict the field fertility of Nelore bulls by using statistical models based on Bayesian inference. A total of 407 oocytes were obtained from slaughterhouse ovaries and submitted to in vitro maturation in TCM-199 and fertilization in Tyrode’s + BSA + lactate + pyruvate. Sperm cells from 3 different bulls, namely, V, T, and G, were used. All bulls had a known field gestational rate. After 18 h of IVF, a sample of zygotes (n = 40/bull) was stripped and stained with acetic orcein (1%) to estimate the number of pronuclei. The remaining zygotes (n = 287) were cultured in SOF for 7 to 8 days. The results are based on 3 replicates with 32 to 33 oocytes/bull per replicate. All 3 bulls analyzed presented similar percentages of oocytes with at least 2 pronuclei (52.5, 40, and 40% for bulls V, T, and G, respectively). However, when cleavage rate was analyzed, bull T presented a significantly (P = 0.0022) lower rate (57.4%) compared with bulls V (77.9%) and G (78.1%). In contrast, blastocyst formation rates were similar between bulls (27.9, 22.8, and 30.5% for bulls V, T, and G, respectively). When the cleavage and blastocyst rates were analyzed in comparison with the field pregnancy rates previously obtained for bulls V, T, and G by using statistic models based on Bayesian inference, the predictions of conception rate based on cleavage were 59.8 ± 2.5%, 54.2 ± 4.7%, and 59.8 ± 2.5% for bulls V, T, and G, respectively. When blastocyst formation rates were used, the predicted conception rate was 58.5 ± 2.2%, 52.3 ± 4.5%, and 61.5 ± 2.8% for bulls V, T, and G, respectively. The predicted conception rates obtained by using the cleavage and blastocyst data together were 54.7 ± 3.9%, 54.1 ± 4.8%, and 63.0 ± 3.1% for bulls V, T, and G, respectively, which was very similar to the real field conception rates for the same bulls (54.4, 54.1, and 63.3%, respectively). The statistical model obtained through the use of Bayesian inference with the combined data for cleavage and blastocyst rates presented a good correlation with the real field fertility of bulls used in this experiment. This result indicates that this statistical method could be a good and efficient alternative to predict the fertility of Nelore bulls. However, a larger number of animals will need to be tested by the same methodology to confirm these results. FAPESP.