Modelling approaches for the estimation of genetic parameters for calving ease and stillbirth in German Holstein dairy cattle

• Allocation of calving traits to the dam generated a longitudinal data structure across parities as a basis for innovative modelling approaches. • Maternal genomic multiple-trait models and maternal random regression models on a continuous parity scale displayed similar genetic parameter pattern. • Additive genetic variances and maternal heritabilities declined with increasing lactation number. • Largest genetic correlations between same traits from different parities were found between parities 1 and 2. • Direct-maternal genetic associations were close to zero or of antagonistic nature. The aim of the present study was to apply alternative modelling approaches for genetic evaluations of stillbirth (SB) and calving ease (CE) from the dam perspective, enabling consideration of a longitudinal data and genetic covariance structure across lactations. We considered a comprehensive dataset including 435,489 calf records for CE and 477,800 calf records for SB from the birth years 2001 to 2017, and genotype data including 41,304 SNPs from 24,133 animals. The calves with phenotypes were offspring from 184,012 Holstein Friesian (HF) cows for SB, and from 177,162 HF cows for CE. The calves and cows were kept in 45 large-scale German dairy contract herds. The applied three genetic-statistical models based on single-step methodology considering both pedigree and genomic relationship matrices. In the “classical” model 1, we allocated SB and CE observations to the calf by considering direct and maternal genetic effects with their respective covariances. Model 2 was a multiple-trait model (MTM) by allocating the observations to the dam and considering same traits in different parities as different traits. Accordingly, in the random regression model (RRM), SB and CE were defined as a trait of a dam and analyzed on a continuous parity scale by considering random regression coefficients for additive-genetic effects of intercept and slope. From both models MTM and RRM, we observed a gradual decrease of additive genetic variances and maternal heritabilities with increasing parity. Genetic correlations between same traits from different parities were larger than 0.80 for adjacent parities, but declined with increasing parity distance. Correlations between maternal genomic breeding values (GEBV) from the two different models MTM and RRM for the same trait and parity were throughout larger than 0.80, and in the range from 0.68 to 0.88 with the maternal GEBV from model 1. Genetic and breeding value correlations close to zero were found between the direct and maternal genetic effects. Correlations between maternal GEBV from the RRM and maternal GEBV from official genetic evaluations were throughout larger than 0.82, and the large rank correlations indicate only minor changes in top lists for sires.