Character process model for semen volume in AI rams: evaluation of correlation structures for long and short-term environmental effects

The objective of this study was to build a character process model taking into account serial correlations for the analysis of repeated measurements of semen volume in AI rams. For each ram, measurements were repeated within and across years. Therefore, we considered a model including three environmental effects: the long-term environmental effect, which is a random year* subject effect, the short-term environmental effect, which is a random within year subject* collection effect, and the classical measurement error. We used a four-step approach to build the model. The first step explored graphically the serial correlations. The second step compared four models with different correlation structures for the short-term environmental effect. We selected fixed effects in the third step. In the fourth step, we compared four correlation structures for the long-term environmental effect. The model, which fitted best the data, used a spatial power correlation structure for the short-term environmental effect and a first order autoregressive process for the long-term environmental effect. The heritability estimate was 0.27 (0.04), the within year repeatability decreased from 0.56 to 0.44 and the repeatability across years decreased from 0.43 to 0.37.