|WU, XIAO-LIN - Council On Dairy Cattle Breeding|
|PARKER GADDIS, KRISTEN - Council On Dairy Cattle Breeding|
|BURCHARD, JAVIER - Council On Dairy Cattle Breeding|
|NORMAN, HOWARD - Council On Dairy Cattle Breeding|
|NICOLAZZI, EZEQUIEL - Council On Dairy Cattle Breeding|
|CONNOR, ERIN - University Of Delaware|
|DURR, JOAO - Council On Dairy Cattle Breeding|
Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/18/2021
Publication Date: 11/1/2021
Citation: Wu, X., Parker Gaddis, K.L., Burchard, J., Norman, H.D., Nicolazzi, E., Cole, J.B., Connor, E.E., Durr, J. 2021. An alternative interpretation of residual feed intake by phenotypic recursive relationships in dairy cattle. Journal of Dairy Science Communications. 2(6):371-375. https://doi.org/10.3168/jdsc.2021-0080.
Interpretive Summary: Feed efficiency is an important trait for genetic improvement programs because feed costs are a large portion of expenses associated with dairy production. Residual feed intake (RFI) is commonly used to measure feed efficiency in beef and dairy cattle. However, RFI is usually calculated with a two-stage approach that requires a number of assumptions. These assumptions can result in unexpected results that do not make sense in mathematical terms. In this paper, we describe a one-step approach using a statistical model that describes RFI as a recursive system made up of dry matter intake and energy sinks. We propose a Bayesian recursive structural equation model as a one-step solution to predict RFI and estimate relevant genetic parameters simultaneously, and demonstrate its use with field data collected from the USDA Beltsville Agricultural Research Center Dairy.
Technical Abstract: Feed efficiency is an important trait for genetic improvement programs because feed costs comprise a large portion of costs associated with dairy production. In the past decades, there has been an increasing interest in the use of residual feed intake (RFI) as a measure of net feed efficiency in meat and lactation animals. Studies on RFI often use a two-stage approach. First, an energy sink model treats feed intake as a linear regression function of essential energy sinks, such as metabolic body weight, energy-corrected milk, and body weight change. It may also account for some additional systematic covariates or factors. In the second stage, the residuals are taken to be the phenotypes of RFI. In this model, the computed RFI phenotypes are fitted as the response variable in a quantitative genetic model to estimate the genetic values and relevant genetic parameters for net feed efficiency. Combining these two modeling stages leads to a one-step approach that eliminates the need to compute the residuals as the RFI phenotypes. Statistically speaking, the energy sink model, defined by a standard linear regression, is not self-contained concerning the model assumptions about the regressor variables. Statistical features of the RFI linear regression model also have not been addressed adequately. Re-arranging the energy sink model suggests an alternative interpretation of RFI as a phenotypic recursive system between dry matter intake and energy sinks. In this technical note, we propose a Bayesian recursive structural equation model as a one-step solution to predict RFI and estimate relevant genetic parameters simultaneously. Accounting heterozygous relationships between dry matter intake and energy sinks is discussed. Extending the model to handle errors-in-variables is also relevant, but it remains a topic of interest for future studies.