Regression Analysis for Predicting AU Triumph Tall Fescue Forage Quality from Chemical Composition Data in North Alabama

Abstract

Lema, M., Knox, N. and Walker, L. 2002. Ṙegression analysis for predicting AU Triumph tall fescue forage quality from chemical composition data in North Alabama. J. Appl. Anim. Res., 22: 81–90. Statistical models may be better alternatives for forage quality estimation as chemical analysis, near infrared reflectance spectroscopy (NIRS) and in vitro digestion are in most instances time-consuming, expensive and hard to obtain and some times utilize hazardous chemicals. A study was conducted to calibrate AU Triumph tall fescue (Festuca arundinacea Shreb) forage quality (WDMD, NDF, ADF and CP contents) predictive equations from chemical analysis data. AU Triumph tall fescue pasture samples collected over a 2-year period were analyzed for crude protein (CP), acid-detergent fiber (ADF), neutral-detergent fiber (NDF), ether extract (EE), acid-detergent lignin (ADL), cellulose (CEL), ash and in vitro dry matter digestibility (IVDMD). Quadratic, cubic and logarithmic relations between independent variables and quality parameter were considered and the data subjected to stepwise regression and correlation analysis. The best fitting equation to predict IVDMD and CP contained only NDF (IVDMD-154.97–1.41 NDF, R2=0.92, RMSE=1.99; CP=41.39–0.51 NDF, R2=0.87, RMSE=0.90) while the best equation to predict NDF contained only EE (NDF=86.80–19.03 EE + 2.98 EE2, R2=0.92, RMSE=1.34). Ash and CEL were the best predictors of ADF (ADF=0.47 + 0.14 Ash + 0.96 CEL, R2=0.99, RMSE=0.04). Model validation using independent samples gave correlation coefficients of validation ranging from 0.86 to 0.97. The results show that tall fescue forage quality parameters (TVDMD, NDF, ADF and CP) can be estimated with high accuracy using simple equations requiring one or two chemical determinations. These equations provide convenient alternatives to chemical analysis for estimating tall fescue forage quality in North Alabama.