Department of Animal- Wildlife- and Grassland Sciences, University of the Free State, P.O. Box 399, Bloemfontein 9300, South Africa; Agricultural Modeling and Training Systems, 418 Davis Rd Cortland, NY 13045, United States
Scholtz, G.D.J., Department of Animal- Wildlife- and Grassland Sciences, University of the Free State, P.O. Box 399, Bloemfontein 9300, South Africa; van der Merwe, H.J., Department of Animal- Wildlife- and Grassland Sciences, University of the Free State, P.O. Box 399, Bloemfontein 9300, South Africa; Tylutki, T.P., Agricultural Modeling and Training Systems, 418 Davis Rd Cortland, NY 13045, United States
A study was conducted to evaluate current proposed models for assessing Medicago sativa L. hay quality, using near infrared reflectance spectroscopy (NIRS) analyses and Cornell Nett Carbohydrate and Protein System (CNCPS) milk production prediction as a criterion of accuracy. Application of the theoretically-based summative total digestible nutrients (TDNlig) model of Weiss et al. (1992), using lignin to determine truly digestible NDF, explained almost all of the variation in milk yield (MY) (r 2 = 0.98). However, this model involves high analysis costs to develop and maintain NIRS calibrations and several of its components were poorly predicted by NIRS and therefore, not suited for quality assessment in practice. Current available models (forage quality index (FQI), relative forage quality (RFQ); relative feed value (RFV)) for assessing Medicago sativa L. hay quality revealed lower accuracies (r 2 = 0.83, r 2 = 0.76, r 2 = 0.61, respectively), especially when protein was included in the model (total forage quality index (TFI); r 2 < 0.49). The developed empirical equation named lucerne milk value (LMV), including ADF, ash and lignin (Y = b0 - b1ADF - b2ash - b3lignin) (r 2 = 0.96), proved to be the most practical, simplistic, economical and accurate quality evaluation model for commercial application. © South African Society for Animal Science.