Genetics of milk protein composition and milk colour in irish dairy cattle

The overall aim of this thesis was to determine the feasibility of breeding for improved milk quality and in particular protein fractions, free amino acids (FAA) and milk colour. To breed for a characteristic such as milk quality it must be; (i) economically or socially important (ii) exhibit genetic variation (i.e be heritable), and (iii) be measurable or genetically correlated with a measurable trait. Gold standard data was determined from 715 milk samples. Spectral data used consisted of ~ 95,000 milk samples from seven research herds and ~ 40,000 milk samples (morning and evening milk samples combined) from 69 commercial herds. The greatest correlation coefficients of external validation obtained for protein fractions, FAA and milk colour were 0.74 (total casein), 0.75 (glycine) and 0.72 (yellowness), respectively. Milk protein fractions and FAA change across calendar months of the year, stage of lactation and parity. A peak in the concentration of all casein fractions was evident in the months of August, September and October. The concentration of glutamic acid was greatest during the months of February, March, April and June when adjusted for milk yield. Changes in individual milk protein fractions and FAA across calendar months of the year and across stages of lactation could provide useful input parameters for decision support tools in the management of product portfolios by processors over time. Heritability of the predicted protein fractions and FAA ranged from 0.04 (beta casein) to 0.61 (total lactoglobulin) and from 0.05 (aspartic acid) to 0.58 (serine), respectively. The coefficient of genetic variation of gold standard protein fractions and FAA ranged from 3.01 (alpha lactalbumin) to 22.98 (total lactoglobulin) and from 1.01 (glutamic acid) to 25.65 (serine), respectively. Milk colour traits were low to moderately heritable ranging from 0.29 (lightness) to 0.35 (yellowness), respectively. The coefficient of genetic variation of milk colour ranged from 0.37 (lightness) to 1.72 (greeness), respectively. Results from this thesis clearly show that some protein fractions, some FAA and milk colour are predictable from MIRS and these predictions exhibit genetic variation and thus breeding for improved milk quality is feasible. The outcome of this thesis is primarily that the prediction of these traits by MIRS could benefit the dairy breeding industry worldwide through genetic selection of animals with higher quality milk and allowing for the more accurate selection of milk for human consumption, infant milk formula, and cheese production. The generated predictions could also be useful for herd and processor management strategies