Dairy cattle

Prediction of dry matter intake and residual feed intake of Holsteins in Japan from milk traits including fatty acids

---

Abstract

Improving feed efficiency in dairy cows is important, but measuring actual feed intake on-site is costly and labor-intensive, making it difficult to obtain enough data for genetic evaluation. Therefore, it is considered useful to predict feed intake using milk production traits such as milk fatty acids that can be applied to herd test records. The objective of this study was to develop equations to predict dry matter intake (DMI) and residual feed intake (RFI) using milk production traits, and to investigate the differences in selected explanatory variables and Adjusted-R2 of the developed equations across lactation stages. The data used to develop the prediction equation were 49,491 records from 157 lactations of 96 cows calving between 2020 and 2023 at the Hokkaido Research Organization Dairy Research Center. Feed intake and milk yield were measured daily, while milk components and body weight were measured weekly. Weekly average values of DMI were calculated from 6 to 306 days in milk. RFI was defined as the difference between the DMI estimated using the Japanese Feeding Standard equation and the actual measured value. Multiple regression analysis was performed with DMI and RFI as dependent variables. Two models were developed: a full model with explanatory variables including milk yield, fat %, protein %, lactose %, milk urea nitrogen, body weight, and three milk fatty acids (De Novo (DnF), Mixed (MiF), Preformed (PrF)); and a reduced model excluding body weight and MiF from the full model. Records were divided weekly into 43 datasets, and optimal explanatory variables were selected for each dataset using stepwise regression. For DMI, Adjusted-R2 of the prediction equation was highest at the 5th week postpartum for the full model (0.82) and at the 6th week for the reduced model (0.81). For RFI, Adjusted-R2 was highest at the 2nd week postpartum for both the full and reduced models (0.56). During the early lactation period up to 15 weeks postpartum, the differences in Adjusted-R2 between the full and reduced models were small. Adjusted-R2 decreased as the lactation stage progressed in both models, ranging from 0.34 to 0.59 for DMI and 0.04 to 0.34 for RFI after 16 weeks postpartum. DMI maintained a higher Adjusted-R2 than RFI throughout lactation, with a difference ranging from 0.21 to 0.48. For DMI, DnF and PrF were always selected during the early lactation stage. For DMI and RFI, the selected variables differed depending on the lactation stage. The prediction equations for DMI and RFI using milk fatty acids showed good fitness in the early lactation stage. Milk fatty acids were particularly useful for predicting DMI in the early lactation stage. It is expected that applying prediction equations to herd-test records will enable genetic evaluation of feed efficiency.

Keywords: 2026

How to Cite:

Nishiura, A., Sasaki, O., Tanigawa, T., Saito, Y. & Tatebayashi, R., (2026) “Prediction of dry matter intake and residual feed intake of Holsteins in Japan from milk traits including fatty acids”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2279627. doi: https://doi.org/10.31274/wcgalp.23434

Rights: 1