Heterogeneity of residuals variances of milk fatty acids in dairy cattle

Routine genetic evaluation for milk fatty acids is under development in the Walloon Region of Belgium. The objective of this study was to test the heterogeneity of residual variances and therefore indirectly the potential need to adjust for this heterogeneity if it exists. The residuals were computed as the difference between the observed and the estimated values using a multi-trait random regression test-day model, similar to the Walloon routine model, used for first lactation only milk yield, quantities and percentages of protein (PROT) and fat (FAT), content of saturated fatty acids in milk (g/100g of milk, SAT) and, content of mono-unsaturated fatty acids in milk (g/100g of milk, MONO). Residuals were considered homogeneous inside strata defined, among others, by weeks of lactation, by days in milk and by calendar months of test date. About 6,687,000 records were available for milk yield and for FAT and PROT parameters. For SAT and for MONO, about 184,000 records were available in this database. Means of residuals were stable and close to zero for all traits. Variances were more variable for MONO and SAT than for milk yield, for example. Daily and weekly variances tended to decrease at the end of the lactation (50%). When the variances were computed by month of test date, some variations were observed and some periods of year were more marked. In conclusion, the observed residual variances were less stable for MONO and SAT. We can conclude that introduction for heterogeneous residual variance is more important for the new traits (MONO, SAT) than it was for the old, traditional ones