Estimation of Body Weight from Heart Girth in Sardi and Timahdite Sheep Using Different Models

The objective of this study was to determine the relationship between body weight (BW) and heart girth (HG) in Sardi and Timahdite sheep in order to develop a prediction equation of BW from HG. The data used for this study included 476 records on BW and HG (227 in Sardi and 249 in Timahdite) collected on males and females of different ages in 33 private farms. The BW and the HG averaged 34.8 ± 21.2 kg and 74.0 ± 16.3 cm, respectively in Sardi and 39.2 ± 22.7 kg and 78.4 ± 16.4 cm, respectively in Timahdite. Correlation coefficients between BW and HG were 0.958 in Sardi and 0.944 in Timahdite indicating a strong relationship between the two variables. Six predictive models for BW were fitted to the data; simple linear regression, polynomial quadratic and cubic regressions and three non-linear regressions (Gompertz, allometric and Mitscherlich). These models were used for the pooled data (regardless of breed and sex), separately for all the animals of a breed regardless of sex (breed-specific) and separately for males and females irrespective of breed (sex-specific). To determine the best fitted regression model, coefficient of determination (R2 or Pseudo-R2), residual mean square (MSE) and Akaike information criterion (AIC) were used. The six models fitted the dataset well since their R2 or Pseudo- R2 varied from 0.892 to 0.969. Nevertheless, based on the previous selection criteria, it seemed that the polynomial cubic model was the best and the allometric model should be discarded. Extreme observations of the three best models were checked using studentized residuals and an absolute value greater than two standard deviations implies considerable deviation. Once the outliers discarded, the best models were run on the clean dataset and compared. Thus, for the pooled data, Sardi breed and females, the Mitscherlich model was appropriate, whereas for Timahdite breed and males, cubic and Gompertz models, respectively were the best. Therefore, a tape measure was developed for each animal category in order to assist livestock farmers in managing their sheep better.