Analysis of Factors Affecting Daily Milk Yields: An Initial Case Study in an Automatic Thrice-Milking Farm

Abstract

The methodologies and parameters for estimating daily milk yields in the United States were mainly developed from the 1960s through the 1990s. A recent initiative by the Council on Dairy Cattle Breeding, USDA Animal Genomics and Improvement Laboratory, and the National Dairy Herd Improvement Association aims to update these methods and parameters for estimating daily yields by collecting and analyzing milking data from dairy farms. This study, serving as an initial case study, examined the factors influencing daily milk yield estimation at a dairy farm in New York State and compared the performance of the existing method with a recently proposed one. In total, 63,562 milking data were extracted from approximately 2,200 cows milked thrice daily in this farm. Data cleaning eliminated incomplete or missing records, retaining 47,670 entries from 1,869 cows for subsequent analyses. The average partial yields in kilograms (milking interval time in hours) of the three milkings were 14.6, 16.5, and 13.8 (7.88, 8.79, and 7.25), respectively. Analysis of variance based on an extended version of the Wiggans (1986) model revealed significant effects of milking interval time and months in milk on proportional daily milk yields. The lactation effects on proportional daily yields were significant for the first two milkings but not for the third milking. Nevertheless, the relative importance of milking interval time and lactations was very low. The polynomial-interaction-regression model analysis showed significant effects from partial yields and significant interactions between partial yields and milking interval times on daily yields. The latter model gave more accurate estimates than the Wiggans (1986) model. Regarding the relative predictability of the three milkings, the 2nd milkings, having the longest average milking interval time, gave more accurate estimates than the 1st and 3rd milkings. The calculated multiplicative correction factors in this farm increased slightly for the 1st milkings and remained roughly comparable (or slightly decreased) for the 2nd and 3rd milkings compared to the Wiggans (1986) assessment. These results revealed only minor changes in daily yield correction factors over the past four decades.