Accuracy of alternative methods of estimating weekly yields of milk fat and protein

Abstract

Thirty-four procedures were used to estimate the weekly yields of milk, milk fat and milk protein of 19 grazing cows on five occasions between the third and eighth months of lactation. The standard deviation of daily milk yield, fat and protein percentage of individual cows averaged 1 .3 litres, 0.58% fat and 0 - 10% protein, respectively. The four factors involved were: sampling frequency, sampling distribution, basis of compositing subsamples for compositional analyses, and the use of total weekly yields or milk yields on day(s) of sampling, for estimating yields of fat and protein. There were no substantial interactions between the 34 procedures and stage of lactation. The average difference between the estimate of weekly milk yield from any one day's yield and the true weekly milk yield (accuracy) was 7.2 litres/week. The equivalent values for fat and protein yields were 0.59 and 0.22 kg/week. When sampling frequency and measurement of milk yield was increased from one to six days per week, the accuracy of the weekly estimate of milk yield (MY) improved by 5.7 litres/week and that of fat yield (FY) by 0.49 kg/week. If total weekly milk yield was used to estimate fat yield, instead of using milk yield on days of testing only, the improvement in accuracy ranged from 0. 15 to 0.03 kg/week for one and six samplings, respectively. Two equally spaced sampling days gave a marginally better (0.74 litres/week) estimate of milk yield than two consecutive days but a marginally poorer (0.10 kg/week) estimate of fat yield. Yield weighting of subsamples for fat or protein analyses did not improve the estimate of fat yield. As the day to day variation in level of protein in the milk was low, frequency and method of sampling for protein analysis was not important but the accuracy of estimating milk yield had a large effect on estimated weekly protein yield.