Correlation and prediction equations for eight-week bodyweight in Sussex and Orpington chickens

Abstract

Correlation and regression models are useful tools for fast or early evaluation of the market body weight of farm animals. However, variation occurs in the reliability of the different curve estimation or regression models. The aims of this study were to determine the correlation between weekly body weights and to predict 8-week body weight from early (1^st and 4^th week) body weight measurements in Sussex and Orpington chickens. The 10 regression models used were: Linear; Logarithmic; Inverse; Quadratic; Cubic; Compound; Power; Sigmoidal; Growth; and Exponential equation. The results showed that Orpington breed had higher body weight than Sussex chicken for most weeks. Male chicks were generally higher in weekly body weight from 4 to 8 weeks of age. The coefficient of correlation of body weight at week 8 with body weight at weeks 1, 2, 4, 5 and 6 ranged from moderate to high (r=0.305 to 0.867). Regression models computed from 1st week body weight of chicks explained only 4.9% to 24.0% of variation in body weight at 8 weeks in Sussex chicks and 27.4% to 37.5% of variation in body weight at 8 weeks in Orpington chicks. Some curve estimation (linear, logarithmic, inverse, compound, growth and exponential) have significant (p<0.05) predicting values for 8-weeks body weight of male and female chicks, however they were associated with low r² (10.0% to 16.5%). The regression models obtained from 4^th week body weight of chicks were better (r² =37.0% to 56.8%) for predicting the 8-week body weight of male and female chicks belonging to the two genotypes. It was concluded that regression equations to predict 8-week body weight from early body weight are more reliable using 4^th week body weight than 1^st week body weight of chicks.

Keywords: Body weight, Chicken, Correlation, Regression models, Reliability