Different models for estimation of litter size variability phenotypes

Selection to increase litter size in pigs has simultaneously increased its variability. In Large White differences in litter size between sows can be larger than 20 piglets, which causes additional labour and affects welfare of animals.
Thus there is a need to dene phenotypes describing variability of litter size in order to reduce it. The main objective of this study was to obtain and compare new phenotypes of phenotypic variability of total number born (TNB) by
using the residual variance of TNB. The dataset contained 246,799 litter observations from 53,803 sows from Topigs
Norsvin multiplication farms. Three models were used to obtain estimates of residual variance for TNB in ASReml 4.1: basic (BM), basic with parity as fixed effect (BMP), and random regression (RRM). The within individual’s variance of the residuals was calculated and log-transformed, to obtain three new variability traits: LnVarBM, LnVarBMP and LnVarRRM. Then, variance components, heritability (h2), genetic coefficient of variation (GCV) and genetic correlations (r
g) between LnVar’s and between LnVar’s and mean TNB were estimated with multi-variate models. Results for additive genetic component (0.026-0.028), residuals (1.23-1.29), h2 (0.02) and GCV (0.081-0.083) indicated that LnVar’s as variability phenotype were robust and not affected by model applied for
their estimation. Furthermore, rg between additive genetic components of LnVar’s (~0.98) proved that genetically LnVar’s are the same trait. The positive r
g between LnVar’s and mean TNB (~0.60) conrmed that selection for
increased TNB increases its variability. Whereas GCV for LnVar’s indicated that TNB variability can be reduced by 1.1 piglet within four generations. These results indicate that phenotypic variability of litter size is under genetic
control and could be reduced by selection. The project was nanced by Polish National Center for Science (grant no. 2016/23/D/NZ9/00029).