Linear Models for Predicting Body Weight of Crossbred Chickens

International Journal of Research and Innovation in Applied Science (IJRIAS) |Volume VIII, Issue IV, April 2023|ISSN 2454-6194

Linear Models for Predicting Body Weight of Crossbred Chickens

U.C. Isaac1*, A. I. Adeolu2
1Department of Animal Science and Technology, Faculty of Agriculture, Nnamdi Azikiwe University, P.M.B. 5025, Awka, Anambra State, Nigeria.
2Department of Agriculture (Animal Science Programme), Alex Ekwueme Federal University, Ndufu-Alike Ikwo, Abakaliki, Ebonyi State, Nigeria.
*Correspondence Author
DOI: https://doi.org/10.51584/IJRIAS.2023.8407
Received: 20 February 2022; Accepted: 19 March 2022; Published: 29 April 2023

Abstract: Prediction of body weight with linear body measurements of 531 day-old crossbred chickens was determined by stepwise regression analysis for mixed sexes at 4 and 8 weeks and separate sexes at 12, 16 and 20 weeks. Shank length was the best single predictor of body weight in Isa Brown x naked neck (IBxNa) at 4 weeks (coefficient of determination, R2 = 65%) and in feathered x Isa Brown (FxIB) at 8 weeks (R2 = 96%). In males, body weight was best predicted by drumstick length (DL) in IBxNa (R2 = 90%) at 12 weeks and in Isa Brown x frizzle feathered (IBxF) at 16 (R2 = 85%) and 20 (R2 = 81%) weeks. The best single predictors of body weight in females were body length (BL) (R2 = 91%), body girth (BG) (R2 = 90%) and body width (BW) (R2 = 90%) in naked neck x Isa Brown (NaxIB) at 12, 16 and 20 weeks, respectively. The best partial predictors of body weight were BG and wing length at 4 weeks (R2 =97%) and BG, BL and keel length (KL) at 8 weeks (R2 =97%) in IBxF; BW and DL (R2 =76%) in normal feathered x Isa Brown (NxIB) males at 16 weeks and BW and DL (R2 =97%) in NxIB females at 20 weeks. The higher R2 values obtained in the models for females made prediction of their body weight more accurate than that of the males. In general, the R2 of mixed sexes ranged from 50-97% and 62-97% at 4 and 8 weeks and for males and females, it ranged from 50-90% and 57-91%; 51-85% and 53-90%; 49-81% and 51-90% at 12, 16 and 20 weeks, respectively. Body weight was best predicted at 8 weeks, and irrespective of genotype, sex and age, the best predictors in single or partial state were BG, BL, KL, BW and DL.

Key words: Linear models, body weight, predictors, sex, crossbred chickens.

I. Introduction

Stepwise regression is a method of multiple linear regression used to analyse the relationship between a set of independent variables or predictors and a dependent variable in linear models. The independent variables produce partial regression coefficients that specify the amount of change in the dependent variable as a result of a unit change in each of the independent variables. Stepwise regression, however, differs from entry procedure of multiple linear regression by employing forward and backward steps to exclude some variables, leaving only those that fit the model (Agostinelli 2002). By so doing, it reduces the interdependency or multicollinearity among independent variables which has been shown to be associated with unstable estimates of regression coefficients (Keskin et al. 2007; Yakubu et al. 2009) that reduces the accuracy of prediction (Chatterjee et al. 2000). The reliability of prediction is reported to depend on positive or negative linear relationship between the dependent variable and the predictors (Semakula et al., 2011; Ojedapo, 2012; Sanda et al. 2014) as well as the magnitude of the coefficient of determination. The coefficient of determination explains the total variation in a dependent variable that is accounted for by the explanatory variables or predictors in a linear regression model (Agomy et al. 2015).

Knowledge of animal weight is important in determining breed characterization (FAO, 2012). It is also useful in management decision (Dingwell, 2006) and in determination of the market price of animals (Semakula et al. 2011). The linear body measurements are important estimators of body weight and have been used extensively in prediction analyses (Gunawan and Jakaria, 2010; Birteed and Ozoje, 2012). Predicting body weight using linear body measurements is more reliable than the conventional method of weighing by scale (Lukuyu et al. 2016). This is because weight measurement by scale is often biased by gut fill. Moreover, measurements of linear body traits involve the use of simple measuring tape that is cost effective (Heinrichs et al. 2007). Prediction of body weight with linear body measurements has been reported in cattle (Lukuyu et al. 2016), sheep (Birteeb and Ozoje 2012), goats (Sam et al. 2016) and various species of poultry. Most reports on chicken mainly focused on either pure breeds, crossbreds or mixed sexes (Gambo et al. 2012; Ajayi et al. 2008; Dzungwe et al. 2018; Adenaike et al. 2015) and not on combinations of genotypes, sexes and ages in chicken.

Full Text  PDF