The use of milk records in dairy sire evaluation, with special reference to simplified recording practice
Heifer records from New York State, U.S.A., were used to calculate a simple recording system for milk recording. Correction factors for environmental factors and genetic parameters were calculated by the use of the least-sguares method. Genetic parameters were derived from the analysis of paternal half-sibs. Analysis of total lactation yield showed the significance of herd-year-season, age, year-age interaction, season-age interaction and season-group of bull interaction for at least one of milk, fat and fat %. Records corrected for environmental effects were used to estimate genetic parameters. Heritability estimates were 0.33, 0.20, and 0.58 for milk, fat and fat % respectively. Genetic correlations of milk with fat and fat % were 0.69 and -0.62 respectively and that of fat and fat % was 0.15. The corresponding values of phenotypic correlations were 0.85, -.36 and .16 respectively. A gamma function was used to describe lactation curve and for correcting for stage of lactation effects on single day yield. The shape of the lactation curve has been shown to be affected by age at calving, month of calving and length of lactation and by their interactions, a values for individual heifers were calculated using the b and c value of heifers within a subclass of age-season-length of lactation. Analysis of a values shows the significant effect of herd-year-season, age, length of lactation, season-age interaction, length-age interaction and length-season interaction. a values corrected for these environmental factors and interactions were vised to calculate genetic parameters. Heritability estimates were lower than those of the total lactation based on ten recordings. The highest values were when the single recordings were made in the autumn months. Genetic correlations between total lactation yield and the corresponding a values were high for milk and fat yield but low for fat %. All genetic correlations were accompanied by high standard errors. Breeding values of bulls were calculated according to total lactation yield and to the a values, using the regressed least-squares constants. Correlation between the two estimates was highest for autumn months. Progeny testing accuracy and the theoretical genetic gain per generation using the simpler recording system were calculated. The theoretical correlated response in total lactation yield was also calculated when selection was on the basis of a values. Lower testing accuracy was found under the simpler recording system, because of the lower heritability estimate. This is true even when the same number of effective daughters are in use. Selection based on the single recording method for milk or fat was capable of giving a theoretical correlated response in total lactation milk or fat yield equal to that obtained from direct selection according to total lactation milk or fat yield. This could be achieved by using larger progeny size in some calendar months or by increasing the number of bulls tested per year. Selection based on the single recording method for fat % was not capable of giving a theoretical correlated response in total lactation fat % equal to that obtained from direct selection according to total lactation fat %. This is due to the low genetic correlation between single recording day fat % and total lactation fat %. It is concluded that the simpler recording system could be used for estimating the breeding value of dairy bulls under certain circumstances; an example could be its use in developing countries where a large proportion of the cattle population is at present unrecorded.