Lactation Curves of Montbeliarde-Sired and Viking Red-Sired Crossbred Cows and Their Holstein Herdmates in 7 Commercial Dairies Using Random Regression and Best Prediction

Abstract

Lactation curves were estimated for Montbéliarde (MO) × Holstein (HO) and Viking Red (VR) × HO 2-breed crossbred cows and for MO × VR/HO and VR × MO/HO 3-breed crossbred cows and their HO herdmates from test-day observations in 7 high-performance dairies that participated in a designed research study. Cows calved from 2010 to 2017. Test-day observations from milk recording for the first 3 lactations were used to fit the lactation curves. Lactations of cows were required to have at least 250 days in milk (DIM) and to have at least 6 test days ≤ 265 DIM. Lactation curves from random regression (RR) functions were compared: 1) Ali-Schaeffer, 2) Wilmink, and 3) Third-order Legendre polynomial. Lactation curves from RR were also compared to lactation curves from Best Prediction (BP). The alternative RR functions and BP were compared for the lactation curve characteristics of 305-d production (kg), peak production (kg), peak d of production, and production from 4 to 103 DIM (kg), from 104 to 205 DIM (kg), and from 206 to 305 DIM (kg) for milk, fat, and protein production. Primiparous and multiparous cows were analyzed separately. Furthermore, two alternative measures of persistency of production were developed for each of the alternative RR functions and BP. The 2-breed and 3-breed crossbred cows and their respective HO herdmates were analyzed separately. For primiparous cows, Legendre polynomial RR best fit the actual test-day observations for milk, fat, and protein production among the alternative RR functions as well as BP for each breed group separately and across breed groups for both 2-breed and 3-breed crossbreds and their HO herdmates. For multiparous cows, BP best fit the actual test-day observations for milk and fat production for each breed group separately and across breed groups for both 2-breed and 3-breed crossbreds. However, Legendre polynomial RR best fit the actual test-day observations for protein production in all cases.