About Us | Help Videos | Contact Us | Subscriptions
 

Abstract

 

This article in JAS

  1. Vol. 93 No. 3, p. 944-948
     
    Received: May 15, 2014
    Accepted: Dec 22, 2014
    Published: February 5, 2015


    2 Corresponding author(s): rrekaya@uga.edu
 View
 Download
 Share

doi:10.2527/jas.2014-8061

Genetic properties of residual feed intakes for maintenance and growth and the implications of error measurement1

  1. R. Rekaya 2* and
  2. S. E. Aggrey
  1. * Department of Animal and Dairy Science, University of Georgia, Athens 30602
     NutriGenomics Laboratory, Department of Poultry Science, University of Georgia, Athens 30602

Abstract

A procedure for estimating residual feed intake (RFI) based on information used in feeding studies is presented. Koch’s classical model consists of using fixed regressions of feed intake on metabolic BW and growth, and RFI is obtained as the deviation between the observed feed intake and the expected intake for an individual with a given weight and growth rate. Estimated RFI following such a procedure intrinsically suffers from the inability to separate true RFI from the sampling error. As the latter is never equal to 0, estimated RFI is always biased, and the magnitude of such bias depends on the ratio between the true RFI variance and the residual variance. Additionally, the classical approach suffers from its inability to dissect RFI into its biological components, being the metabolic efficiency (maintaining BW) and growth efficiency. To remedy these problems we proposed a procedure that directly models the individual animal variation in feed efficiency used for body maintenance and growth. The proposed model is an extension of Koch’s procedure by assuming animal-specific regression coefficients rather than population-level parameters. To evaluate the performance of both models, a data simulation was performed using the structure of an existing chicken data set consisting of 2,289 records. Data was simulated using 4 ratios between the true RFI and sampling error variances (1:1, 2:1, 4:1, and 10:1) and 5 correlation values between the 2 animal-specific random regression coefficients (–0.95, –0.5, 0, 0.5, and 0.95). The results clearly showed the superiority of the proposed model compared to Koch’s procedure under all 20 simulation scenarios. In fact, when the ratio was 1:1 and the true genetic correlation was equal to –0.95, the correlation between the true and estimated RFI for animals in the top 20% was 0.60 and 0.51 for the proposed and Koch’s models, respectively. This is an 18% superiority for the proposed model. For the bottom 20% of animals in the ranking, the superiority was 17%. Even when the ratio of variances was 10:1, the superiority of the proposed model was around 6%.

  Please view the pdf by using the Full Text (PDF) link under 'View' to the left.

Copyright © 2015. American Society of Animal Science