Abstract
Numerous in vitro ruminal fermentation techniques have been developed to estimate forage digestibility. However, these methods are costly, time consuming, and require access to fistulated animals. This study aims to develop predictive models for in vitro ruminal organic matter digestibility (IVOMD) from pasture chemical components (CP, NDF, ADF, ADL, cellulose and hemicellulose) by examining the relationships between chemical composition and IVOMD of tropical pastures harvested at ground level (G-L) or by hand-plucking (H-P) during the dry, intermediate and wet season in Jamaica. Relationships were determined using simple linear and multivariate forward step-wise regression procedures. Selection of the best approximating multivariate models was done by finite Akaike Information Criterion. Chemical components were better predictors of IVOMD in the dry season than in intermediate and wet seasons. Dry season ADF was the best single predictor of IVOMD (R2 = 0.76). Crude protein accounted for 63.8% of variability of IVOMD during the dry season; 30.8% and 11.5% higher than in the intermediate and wet seasons, respectively. Neutral detergent fiber (R2 = 0.60 – 0.63) was consistently a moderate predictor of IVOMD across seasons. Hemicellulose poorly predicted IVOMD. Crude protein was the best single parameter for predicting IVOMD of G-L harvested herbage, but, with low accuracy (R2 = 0.31). Multivariate models improved the prediction accuracy of IVOMD. The best approximating models for dry, intermediate and wet season IVOMD prediction were; Y = 922 - 0.61ADF + 0.99CP - 1.24ADL - 0.32NDF, Y = 1306 - 0.81NDF - 0.46ADF + 0.83CP - 1.39ADL and Y = 1211 - 1.02NDF + 0.90CP - 1.10ADL, respectively. The optimum predictive models for G-L and H-P harvested herbage were; Y = 1225 + 1.05CP - 0.75NDF - 0.73ADF and Y = 1093 - 0.70NDF + 0.78CP - 0.40ADF, respectively. It was concluded that the accuracy of predicting IVOMD from chemical components was greatest during dry season.
