Prediction of body weight at eight weeks of age through body measurements at different ages of muscovy ducklings

Abstract EN

The objective of this study was to compute appropriate statistical models that describe the body weight at eight weeks of age by the input variables body measurements at different ages.

One-hundred and sixty one-day old Muscovy ducklings (72 males and 88 females) were used to determine the importance of phenotypic traits such as keel length (KL), shank length (SL) and breast circumference (BC) in addition to body weight (BW) and to fit different statistical models to predict body weight through body measurements in Muscovy ducks.

In order to compare the relative efficiency of various models and to select the most suitable model, the coefficient of determination (R2) and the standard error of the estimate (SEE) were used.

A larger value of R2 and smaller value on SEE indicate best fit of the model.

The following results were obtained: Muscovy males had significantly heavier BW and longer SL and KL and had higher BC than females at all studied ages.

SL, KL and BC growth trends were in general concomitant to body growth trends.

The differences in BW at 8 weeks of age appeared to depend more on differences in growth in length of shank bone than on the differences in growth in keel or breast.

Muscovy males' BW8 significantly (P? 0.001) affected by either lengths of shank or keel at 8 weeks of age indicating that about 68.4% of variability in BW8 were explained by both SL and KL at the same age.

Whereas, either females' BC or KL at 8 weeks of age were significantly increased BW8.

Model 2 (BC8 and KL8 used as predictors for BW8) had higher R2 and lower SEE than model 1 (BC8 used as the sole predictor for BW8).

The model for predicting Muscovy ducklings’ BW8 that depended firstly on KL8 followed by BW6 and SL8 had higher R2 (P? 0.05) and lower SEE than other models.

Each of BW6, SL6, SL4 and BW4 could be used as predictors for Muscovy ducklings' BW8, regardless of sex with higher R2 values and lower SEE estimates (P? 0.001).

Similarly, BW8 could be predicted through BC6, KL6, SL8 and KL4.