Recent prediction tools for estimation of first lactation 305 days milk yield using monthly test days for genetic evaluation in crossbred cattle

Abstract

First lactation 305 days milk yield (FL305DMY) of progenies is an important criteria for selection of sires. In field progeny testing programme, prediction of FL305DMY based on test day records will reduce the cost of recording. The present study was aimed at evaluating the efficiency of different methods of prediction for FL305DMY based on monthly test day milk yield records in Karan Fries cattle and evaluation of sires based on the performance of its progenies. Methods of prediction were Wood’s incomplete gamma function and Artificial neural network (ANN). Efficiency of prediction was compared using error of prediction. Root mean square error was least for ANN method (5.72%). The attempt was made to evaluate Karan Fries sires based on actual and predicted FL305DMY so obtained utilising BLUP- AM method of sire evaluation. The breeding values of 48 Karan Fries sires having 3 or more daughters were estimated. Comparison of ranking of sires based on different methods of prediction was done using Spearman’s Rank correlation and RMSE values. The sire evaluation using ANN method based predicted values showed marginally higher rank correlation (0.91) and least RMSE (2.39%) with actual yield based on monthly test day records. Thus it was concluded that recoding at monthly intervals and use of ANN and conventional fitting Wood’s function can be used for evaluation of sires. Though both the methods are equally efficient, ANN showed marginal superiority.

Keywords

• Artificial neural network
• BLUP-AM
• Monthly test days
• Wood’s incomplete gamma function