Adaptation and evaluation of the GrazeIn model of grass dry matter intake and milk yield prediction for grazing dairy cows

Abstract

The prediction of grass dry matter intake (GDMI) and milk yield (MY) are important to aid sward and grazing management decision making. Previous evaluations of the GrazeIn model identified weaknesses in the prediction of GDMI and MY for grazing dairy cows. To increase the accuracy of GDMI and MY prediction, GrazeIn was adapted, and then re-evaluated, using a data set of 3960 individual cow measurements. The adaptation process was completed in four additive steps with different components of the model reparameterised or altered. These components were: (1) intake capacity (IC) that was increased by 5% to reduce a general GDMI underprediction. This resulted in a correction of the GDMI mean and a lower relative prediction error (RPE) for the total data set, and at all stages of lactation, compared with the original model; (2) body fat reserve (BFR) deposition from 84 days in milk to next calving that was included in the model. This partitioned some energy to BFR deposition after body condition score nadir had been reached. This reduced total energy available for milk production, reducing the overprediction of MY and reducing RPE for MY in mid and late lactation, compared with the previous step. There was no effect on predicted GDMI; (3) The potential milk curve was reparameterised by optimising the rate of decrease in the theoretical hormone related to secretory cell differentiation and the basal rate of secretory cell death to achieve the lowest possible mean prediction error (MPE) for MY. This resulted in a reduction in the RPE for MY and an increase in the RPE for GDMI in all stages of lactation compared with the previous step; and (4) finally, IC was optimised, for GDMI, to achieve the lowest possible MPE. This resulted in an IC correction coefficient of 1.11. This increased the RPE for MY but decreased the RPE for GDMI compared with the previous step. Compared with the original model, modifying this combination of four model components improved the prediction accuracy of MY, particularly in late lactation with a decrease in RPE from 27.8% in the original model to 22.1% in the adapted model. However, testing of the adapted model using an independent data set would be beneficial and necessary to make definitive conclusions on improved predictions.

The assessment of supplementation requirements of grazing ruminants using nutrition models

“…The RPE in herbage DMI prediction varies from 15% to 16% ( Delagarde et al, 2011b ; O’Neill et al, 2013 ). O’Neill et al (2014) adapted the model by increasing the intake capacity, resulting in lower RPE, particularly in mid and late lactations. The model is not sensitive enough with heavy pasture constraints such as low allowances and pasture in the reproductive phase.…”