Comparison of Methods of Predicting Breeding Values of Swine

Best linear unbiased predictions (BLUP) using information from all known relatives; selection index using phenotype, full-sib average and half-sib average; and phenotypie deviation from contemporary group average were compared as methods of predicting breeding values for days to 100 kg and backfat. Swine records (n = 203,869) from five Hampshire, one Duroc and six Yorkshire herds were obtained from the Nebraska SPF Swine Accrediting Agency. Heritabili ty values used to compute BLUP or index were either estimates based on within-breed offspring on parent regression or values recommended by the National Swine Improvement Federation (NSIF) guidelines. Within-breed estimates of heritabili ty ranged from .11 to .25 for days to 100 kg and from .10 to .22 for backfat. Heritabilities recommended by NSIF were .35 for days to 100 kg and .40 for backfat. Correlations between index and phenotypic deviation were larger than correlations between BLUP and phenotypic deviation or BLUP and index. Correlations between BLUP and index were slightly larger than correlations between BLUP and phenotypic deviation. Increasing the heritabili ty values used to compute BLUP or index increased the correlations among methods. Value of heri tabili ty had little effect on the correlation between a parent 's predicted breeding value and its progeny average. On the average and assuming the NSIF heritability, the correlation between BLUP of a sire and its progeny average was 33% larger for days to 100 kg and 44% larger for backfat than the correlation between the phenotypic deviation of a sire and its progeny average. The advantage of BLUP over phenotypic deviation for dams was less than for sires: 25% for days to 100 kg and 18% for backfat. Selection of pigs using BLUP instead of phenotypic deviation or index would substantially improve response to selection. (