Functional Classification of Feed Items in Pampa Grassland, Based on Their Near-Infrared Spectrum

International audience

English. Native pastures that present high botanical richness produce a variety of feed items for grazing herbivores. Diet selection takes place at the bite scale and, at a daily scale, the sum of bites composing the intake determines diet characteristics. The objective of this paper is to bring an original contribution to the understanding of diet selection in diverse foraging environments such as native Pampa grasslands. The issues are 1) to improve grazing management in Pampa ecosystems and 2) to evaluate the ability of native pastures to feed a herd in order to reduce supplementation and to improve the economic sustainability of extensive systems. We hypothesize that the information contained in the infrared spectra of plant samples corresponding to potential bites can be useful for establishing functional bite categories, which, in turn, can be used to understand and manage pastoral ecosystems. We analyzed the near-infrared spectrum (NIRS) of 1 515 bite samples gathered in 2012 using the continuous bite monitoring method applied to heifers continuously grazing on Pampa grassland at five grazing intensities (randomized block design with two replicates, applied since 1987). On the basis of the main wavelengths, a statistical classification of bite samples was produced and five classes were described: 1) small bites of wetland grasses and short grass; 2) medium bites of forbs and short grass; 3) small bites of short Poaceae and mixed species; 4) large bites on tussocks of Poaceae; and 5) extralarge bites on stiff tussock of Poaceae and Apiaceae. In order to evaluate the functional value of the classification, we tested its ability to take seasonal changes in pasture composition into account and to describe the variations in animal intake and performance in contrasted grazing conditions. We concluded that the NIRS is a good support for establishing functional classes of feed items that can be used to understand the foraging behavior and diet composition of grazing herbivores. Our approach was tested in Pampa grasslands grazed by cattle but should be applicable to a wide range of ecosystems, provided that chemical composition is more relevant for selection than feed item accessibility.