Prediction of body weight from linear body measurement traits of Nguni goats in the Limpopo province

Abstract

The study aims to determine the relationship between body weight and linear body measurement traits and to establish a model to predict body weight from linear body measurement traits using MARS and CART. A total of 426 Nguni goats (Male = 69 and female = 357) aged between one to four years were used for body weight (BW) and some linear body measurements such as heart girth (HG), body length (BL) and withers height (WH). The average body weight of male and female goats was 32.50 and 33.02 kg, respectively. MARS and CART were used to predict BW from some morphometric traits. Correlation results indicated that in female goats, BW had a significantly positive correlation (P < 0.01) with HG (r = 0.84). While in male goats, BW had a highly significant positive correlation (P < 0.01) with BL (r = 0.79). The CART model indicated that HG was the best predictor variable for BW, while MARS found BL > 77 cm as the best predictor of BW in the study. CART outperformed MARS model with high R2 = 0.76, Adj R2 = 0.76 and r = 0.87. The study concludes that HG can be used as a predictor of BW.