Prediction of 30S-day milk yield based on test- day values in Murrah buffaloes

Abstract

The first lactation monthly test-day nlilk yield of the 485 Murrah buffaloes sired by 58 bulls spread over 22 years (1983-2004), maintained at National Dairy Research Institute, Kamal. \vere used to predict 305-day nlilk yield. The test-day milk yields recorded from 6th to 305th day at monthly sequential intervals were used to predict 305-day milk yield by regression method. The regressions of 305-day milk yield on lTIonthly test-day milk yield were highly significant but did not exhibit a consistently increasing or declining trend. The regression coefficients ranged bet\veen 148.31 kg (TDY-9th) and 208.50 kg (TDY-6th). The accuracy of prediction (R2) of 305-day milk yield based on I110nthly test-day milk yields varied between 42 (TDY-lst) and 670/0 (TDY-6th). Prediction equations derived through step-wise regression for predicting the 305-day milk yield from monthly test-day luilk yields showed that the average error and residual mean sum of squares continuously declined in prediction of 305-day milk yield with the increase in number of variables up to 6, however there was no use of increasing the variables beyond 5. It was concluded that the third, sixth and ninth test-day milk yield should be pooled for prediction of the 305-day Inilk yield with accuracy of 91% in Murrah buffaloes.