Single step genomic prediction accuracies for lactation and reproduction traits in Yorkshire sows

Abstract Text:

Most economically important traits associated with lactation and reproduction in pigs are either less heritable, sex-limited, expressed later in life, or difficult to measure on a routine basis. Genomic predictions using single step BLUP (SSBLUP) methodologies, which utilizes information on phenotypes, pedigree and markers from genotyped and non-genotyped animals simultaneously, is an alternative to phenotype and pedigree based (BLUP) methods. The objective of this study was to estimate and compare prediction accuracies for lactation and reproduction traits using SSBLUP and BLUP methods. Data from 2481 litters from 1431 Yorkshire sows farrowed between August 2011 and August 2015 were used in this study.  Of these 1161 sows were genotyped using Illumina’s Porcine SNP60 Bead Chip. The sows were weighed and scanned for back fat and loin depth before farrowing and at weaning. Piglets were weighed at birth, weaning and death. Feed consumption of each sow during lactation was measured using the Gestal feed recording system. Reproductive traits studied were total number born, number born dead, number alive at 24 hours and number weaned. Lactation traits analyzed were average daily feed intake of sow during lactation, sow body weight, back fat and loin depth at farrowing, body weight, back fat and loin depth loss during lactation, and litter weight gain from birth to weaning. The training data included sows born before April 1, 2013, and validation data included sows (n=242) born on or after April 1^st 2013. BLUP breeding values for animals in the validation data were computed using all information from the test data plus pedigree information for animals in the validation data. Both SSBLUP and BLUP evaluations were computed using MiX99 software. Accuracies for animals in the validation data were estimated as the correlation between their estimated breeding values and phenotypes, corrected for fixed effects, divided by the square root of heritability of the trait. For all traits studied, prediction accuracies using SSBLUP were higher (0.23 to 0.84) than those from BLUP (0.18 to 0.72). On average the SSBLUP accuracies were 39% and 33% higher, respectively for reproductive and lactation traits. The results indicate that the SSBLUP methodology produces more accurate estimated breeding values for lactation and reproduction traits in pigs.

Keywords: single step genomic prediction, pig, reproduction, lactation