Milk quality assessment in automatic milking systems: accounting for the effects of variable intervals between milkings on milk composition

Abstract

Variation in milk yield and milk composition was studied in two commercial herds with automatic milking systems (AMS). All cows were milk recorded and milk sampled at all milkings throughout 48 h. All samples were analysed for fat, protein, lactose, urea and somatic cell count (SCC). The resultant data set for herd 1 included 606 milk samples from 138 cows, for herd 2 the data set included 551 milk samples from 125 cows. There was a strong relationship between milk yield per milking and time interval since last milking. Within cow, the standard deviations for milk contents (%) of protein, lactose and urea were 0.12, 0.08 and 0.69. Adjusting these measures for milk yield at that milking had a negligible effect on this variation. Within cow, the standard deviations for milk fat and log SCC were 0.68 and 0.17. This variation could be significantly reduced, and bias corrected, by accounting for milk yield at that milking, expressed as a deviation from mean milk yield (MILKDIF). The correction equations were: mean daily milk fat %=individual milk fat %+0.120(MILKDIF) rS.D.=0.62 mean daily log SCC=individual log SCC+0.035(MILKDIF) rS.D.=0.15 However, there was still substantial variation in these measures. It is concluded that in AMS, single milk samples provide measures of milk protein, lactose and urea whose variability was comparable to that measured in conventional milking systems and could not be improved by correcting for variation in milk yield at that milking. Single milk samples in AMS systems provide poor measures of milk fat content. Milk fat and SCC values can be usefully corrected for variation in milk yield at that milking. In practical terms, artificially increasing the time-interval prior to a milking at which samples will be taken would decrease the variability associated with the estimation of true daily milk fat content from a single sample.