Genetic parameters of calving ease in dairy cattle using threshold and linear models

Genetic parameters of direct and maternal effects for calving ease (CE) in Portuguese dairy cattle were estimated applying a threshold model (TM) and three linear models (LM) using 320,953 CE records. In LM, CE scores were directly used as dependent variables (LM1) or were transformed into values on an underlying continuous liability scale (average and maximum values for LM2 and LM3, respectively). The estimate of heritability for direct effects was lower on linear models (0.04–0.05) than in TM (0.09). Furthermore, direct heritability was higher than maternal heritability, ranging between 0.04–0.09 and 0.01–0.02, respectively. The genetic correlation between direct and maternal effects was negative in all models, ranging between –0.75 and –0.82. We investigated the association between bulls estimated breeding values (EBV) attained by TM and three LM, for 2223 bulls. Results show that all correlations were higher than 0.95 and slightly greater correlations were observed for direct effects than for maternal effects. Particular emphasis was given to individual differences between EBV ranking lists. The average difference between rank positions of TM and the three LM ranged from 64 to 122. Therefore, the selection of breeding animals is influenced by the model’s choice. Even with clear theoretical advantages in analysing categorical data, perhaps due to its more recent implementation, TM has not yet achieved the popularity of LM. The results of this work show that TM ranking list is more in accordance with LM2 than with LM1 or LM3. Therefore, with linear models, CE should be considered as average liability value instead of usual numerical score or maximum liability