PSXII-17 - Fertility prediction model for Nilli-Ravi buffalo bulls through Fourier harmonic analysis of sperm

A model to predict Nili-ravi buffalo bull fertility was developed based on Fourier harmonic analysis of sperm. Seventeen bulls with 3032 AI records were categorizes based on fertility rate (FR) as low (36.5±0.2, n = 6: SD< ˗1 from mean FR), medium (39.9±0.2, n = 3; SD +1 to -1 from mean FR) and high fertility (41.4±0.1, n = 8; SD > +1 from mean FR). Cryopreserved semen samples from these bulls were investigated for Fourier harmonic analysis of sperm nuclear shape. Hoechst-33342 and YOYO-1 fluorescent stains were used to identify live and dead sperm. Digital images were analyzed to get sperm nuclear perimeter points at different phase angles to generate Fourier functions. Mean harmonic amplitude (HA) 0 was different (P < 0.05) for 1700 live vs. 1294 dead sperm from the 17 bulls, thus live sperm were used for remaining analyses. The mean, variance, skewness and kurtosis values of 100 live sperm nuclei/bull were compared for HA0-5 between high (n = 6) and low (n = 6) fertility groups, considering equal number of bulls in each category. The mean HA2 (0.739±0.01 vs 0.686±0.00) and 4 (0.105±0.001 vs 0.007±0.001) were higher in high vs low fertility group respectively (P < 0.05). Sperm nuclear perimeter among high fertility group sperm was more elongated. There was also an increased skewness of HA0 as fertility increased (P < 0.05). Discriminant analysis defined a fertility model by using mean HA4, skewness HA0 and variance HA2, that resulted in 91.7% bulls into their correct fertility group upon cross-validation (canonical correlation=0.928; P < 0.05). Higher values of mean HA4, skewness HA0 and variance HA2 increase the chance of bulls being placed in the high fertility group. In conclusion, sperm nuclear shape in Nili-ravi buffalo bull is related to in vivo fertility. A fertility model using reported discriminant measures could be used to objectively identify Nili-ravi buffalo bulls of varying fertility.