Identification of the predictors of preference for tifton and alfalfa hays on feeding of horses

• Grant number: 11/02597-9
• Support Opportunities: Regular Research Grants
• Start date: June 01, 2011
• End date: May 31, 2013
• Field of knowledge: Agronomical Sciences - Animal Husbandry - Pastures and Forage Crops
• Principal Investigator: Ciniro Costa
• Grantee: Ciniro Costa
• Host Institution: Faculdade de Medicina Veterinária e Zootecnia (FMVZ). Universidade Estadual Paulista (UNESP). Campus de Botucatu. Botucatu , SP, Brazil

Abstract

Horses preference to the conventional predictors, such as energy and protein, will lead to additional information on quality of hays which may be used by ranchers, as well as it will be helpful to guide them at haying process and hay sales. Thus, the objective of this study is: a) to identify predictors of preference in Tifton and Alfalfa hays for horses, b) to evaluate an quick and precise analytical procedure for hay producers and ranchers utilizing NIRS (Near infrared reflectance spectroscopy), c) to classify Tifton and Alfalfa hays according to the predictors of preference described above. Nine Quarter horses will be fed either 40 kinds of Tifton or Alfalfa hay. The intake, chewing activity and preference values of Tifton and Alfalfa hays will be determined for each kind of hay by means of three horses and replicated three times per animal in a short period of evaluation (10 min with each feedstuff). Each kind of hay, Tifton and Alfalfa, will be fed in separated experiments associated to four standardized hays to allow horses to demonstrate their preferences in many combinations, being about 3000 individual tests of preference. Also, chemical (dry matter, neutral detergent fiber, acid detergent fiber, crude protein, gross energy, water soluble carbohydrates and in vitro digestibility of dry matter), physical (color, leaf:steam ratio and steam diameter) and microbiological analysis (yeast and mold) will be performed in experimental and standardized hays. Data will be analyzed as simple linear or multiple regressions depending on number of dependent variables included in the model. It will be investigated the potential of NIRS to predict equations of preference as well. (AU)