An approximate Bayesian analysis of somatic cell score curves in Holsteins

Somatic cell score (SCS) decreases to a minimum at around 60 days postpartu m and then increases until the end of lactation. This pattern was described with three non-linear mixed effects models. Parameters of the models were linear in fixed and random (cow) effects. Data were 1705 SCS test-day recor ds (155 primiparous Holsteins) taken from 14 to 300 days in milk. Point est imates were the mode of the joint posterior distribution of fixed and rando m effects (given dispersion parameters) and approximate marginal maximum li kelihood estimates of covariance components. Intramammary infection during lactation, season and period of calving affected SCS patterns. Estimates we re similar with or without a relationship matrix. There was between-cow var iation of trajectories, suggesting that SCS curve parameters can be modifie d by selection in some advantageous manner.