Estimation of Variance Components for Litter Size in the First and Later Parities in Improved Jezersko-Solcava Sheep

Aim of this study was to estimate variance components for litter size in Improved Jezersko-Solcava sheep. Analysis involved 66,082 records from 12,969 animals, for the number of lambs born in all parities (BA), the fi rst parity (B1), and later parities (B2+). Fixed part of model contained the eff ects of season and age at lambing within parity. Random part of model contained the eff ects of herd, permanent eff ect (for repeatability models), and additive genetic eff ect. Variance components were estimated using the restricted maximum likelihood method. Th e average number of lambs born was 1.36 in the fi rst parity, while the average in later parities was 1.59 leading also to about 20% higher variance. Several models were tested in order to accommodate markedly diff erent variability in litter size between the fi rst and later parities: single trait model (for BA, B1, and B2+), two-trait model (for B1 and B2+), and single trait model with heterogeneous residual variance (for BA). Comparison of variance components between models showed largest diff erences for the residual variance, resulting in parsimonious fi t for a single trait model for BA with heterogeneous residual variance. Correlations among breeding values from diff erent models were high and showed remarkable performance of the standard single trait repeatability model for BA.