A study on milk yield persistency using the best prediction and random regression methodologies in Iranian Holstein dairy cows

The data consisted of 435,390 test day milk yield records of primiparous cows in 659 herds calving from 2001 to 2011. Evaluation of persistency using best prediction methodology showed that the phenotypic correlation between this persistency measure and total milk yield was 0.450, while the best reference day, the heritability of persistency and 305 d milk yield  estimated by this method, were day 130, 0.11 and 0.305, respectively. Heritabilities of milk yield persistency for Pers1 predicted breeding value from 106-205 days in milk, subtracted from predicted breeding value from 6-105 days in milk) and Pers2 (predicted breeding value from 206-305 days in milk subtracted from predicted breeding value from 6-105 days in milk) calculated by Random regression methodology were 0.09 to 0.185, respectively. The results showed that the best prediction method is powerful and accurate in measuring persistency. However, due to the flexibility of random regression methodology, some measures of persistency using this method can have higher heritability and genetic correlation with total milk yield compared to the best prediction methodology. It can therefore be concluded that calculation of persistency using random regression methodology is preferred to the best prediction method. Key words: additive genetic effects, lactation curve, persistency, total milk yield