Feed intake of Holstein, Danish Red, and Jersey cows in automatic milking systems

The voluntary dry matter intake (DMI) of a cow is an important parameter in dairy farming; however, it cannot be measured directly for individual cows within a group, therefore a model is needed to estimate the individual cow DMI. In a previous study a DMI model-the '2004 Model'-was developed for Israeli Holstein cows milked in a milking parlor. The present study aimed to apply the 2004 Model to robotically milked dairy cows of three different breeds: Danish Holstein, Danish Red, and Danish jersey. Robotic milking by means of an Automatic Milking System (AMS) influences both milk yield and feeding behavior. The 2004 Model incorporates individual animal factors: milk yield (MY), daily changes in body weight (BW), the MY/BW relationship, and milk fat content. A modern milking robot automatically measures BW and milk yield of the cow in its milking stall. The 2004 Model differs from the US National Research Council (NRC) model in its incorporation of continuously changing parameters: 1) coefficients appropriate to each day in lactation; 2) daily values of and changes in BW; 3) values of MY normalized according to BW; and 4) data from only the previous 2 d rather than from the whole lactation. The model was developed on the basis of three data sets acquired from 2002 through 2005 at the Danish Cattle Research Centre (DCRC) in Foulum, Denmark, and which covers a total of 206 lactations (62 Holstein cows, 77 Danish Reds, and 67 Jerseys). Results suggest that: (1) on the individual cow level, the 2004 Model takes daily fluctuations into account; (2) on the group level, the average Mean Absolute Error (MAE) of the DMI as calculated by the 2004 Model was significantly (t-test, p = 0.05) lower than that given by the NRC model. The MAE values from the 2004 Model and the NRC models, respectively were: 2.0 vs. 2.4 kg, 2.2 vs. 2.5 kg, and 1.8 vs. 2.1 kg for the Danish Red, Holstein and Jersey cows, respectively. DMI estimation is important for ration formulation, therefore, the 2004 Model may be considered for on-line decision-making, such as daily adjustment of the amount of concentrate supplied by the AMS. (C) 2010 Elsevier B.V. All rights reserved.