Process-based modelling of the nutritive value of forages: a review

Modelling sward nutritional value (NV) is of particular importance to understand the interactions between grasslands, livestock production, environment and climate-related impacts. Variables describing nutritive value vary significantly between ruminant production systems, but two types are commonly used: 1) variables related to cell wall content, digestibility and energy, and 2) variables related to protein content.This study reviews and compares alternative modelling approaches simulating forage NV. It is intended to: 1) provide model users essential information for a fit-purpose use of grassland models; 2) give model builders feedback helpful in improving the models; 3) promote the development of state-of-the-art model assessment. The focus is on approaches applicable for European grasslands, as implemented in the grassland models; BASGRA, CATIMO, IFSM, MCPy, ModVege, PaSim, QUAL, SPACSYS, and STICS. These models cover swards cut for silage or hay, grazed swards, monocultures and mixtures, permanent grasslands and leys kept for less than five years.Six out of nine models use neutral detergent fibre (NDF) to describe content of structural carbohydrates. Digestibility of harvested dry matter (DM) is estimated as a function of NDF content and NDF digestibility in five models. Two of the models use an alternative approach that consists in estimating organic matter digestibility or in-vitro true digestibility of DM directly. Three of the models include variables to describe net energy content of DM whereas one model use metabolizable energy. To estimate protein value of harvested DM seven approaches use crude protein concentration calculated as a function of plant nitrogen concentration and one models uses digestible intestinal protein content.