The prediction of the intake potential of grass silage in the supplemented diets of lactating dairy cows

This study was undertaken to develop models which could be used in conjunction with the near infrared reflectance spectroscopy (NIRS) analysis of grass silage to accurately predict the intake potential of grass silage when offered to lactating dairy cows as part of a mixed diet. Empirical models were developed with data collected from two large-scale studies carried out at the Institute. The models comprised of (1) a linear equation for converting the NIRS-based predicted intake of a given silage for beef cattle to dairy cows and (2) a model which corrected the intake potential of the grass silage for supplementary concentrates. Furthermore, a milk yield adjustment factor of 0.14 kg DM/kg milk was utilised to standardise milk yields. Both linear and exponential models were developed to describe the decrease in silage intake as concentrate intake increased, with y-axis intercepts corresponding to unsupplemented silage intakes (NIRS-based predictions for beef cattle adjusted for dairy cows) and common x-axis intercept of 168.0 (SE=20.50) and 203.8 (SE=5.64) g/kg W0.75, respectively, corresponding to concentrate intake when offered as a sole feed. A common r parameter (model curvature) of 1.0047 (SE=0.00397) was assumed for the exponential model. When the models were validated against the data from an independent study, the predictions from the two models were not significantly different, giving R2 values of 0.70. The intercept and slope from the linear model were 5.39 and 1.01, respectively, and the intercept and slope from the exponential model were 6.10 and 0.98, respectively. Both intercepts and slopes were not significantly different from 0 and 1, respectively. Ninety-three percent of predictions were within 10% of observed intakes in the validation data.