Predicting food intake in dairy heifers from early lactation records

Abstract

ABSTRACTNucleus breeding schemes for dairy cattle give opportunities for selection on characteristics other than milk production, such as food intake or efficiency, and for the application of reproductive technologies such as embryo transfer. The emphasis in such schemes involving embryo transfer will be on early lactation measurements of production and food intake, to minimize generation intervals. The aim of this study was to examine the value of early lactation measurements of food intake and other characteristics in predicting longer-term food intake. Intakes of a complete diet, offered ad libitum, were available for 101 heifers up to week 38 of lactation, from the Edinburgh School of Agriculture's Langhill herd. Partial correlations between weekly dry-matter (DM) or metabolizable energy (ME) intakes in early lactation and cumulated intakes to week 38 of lactation, after fitting year and month of calving as fixed effects, and proportion of Holstein blood as a covariate, ranged from 0·27 for week 1, to 0·70 for week 12. Cumulated ME intakes, up to week 38, were regressed on shorter measures of ME intake, together with fat plus protein yield in weeks 1 to 10 of lactation. Other independent variables, such as point estimates of, or changes in, live weight, condition score and backfat depth did not further increase the precision of prediction. The means and standard deviations for milk yield, DM intake and ME intake up to week 38 of lactation were 5877 (s.d. 1087) kg, 4070 (s.d. 400) kg and 51579 (s.d. 4614) MJ respectively. For a fixed duration of intake recording, measurements taken later in lactation gave the most precise prediction of 38-week ME intake (e.g. residual s.d.s from models including 4-week cumulated ME intakes in weeks 1 to 4, 3 to 6 and 5 to 8 of lactation were 2865, 2636 and 2501 MJ respectively, with R2 values of 0·62, 0·67 and 0·71). Shorter periods of intake recording started in week 5 of lactation gave slightly more precise prediction than longer periods of recording started in weeks 1 to 4 (e.g. residual s.d.s from models including cumulative ME intakes in weeks 1 to 10, 3 to 10 and 5 to 10 were 2391, 2298 and 2277 respectively, with R2 values of 0·69, 0·75 and 0·76). These results have implications for the cost: benefit of food intake recording in breeding schemes.