First studies based on the results obtained from Carne‑Aragón S.C.L selection scheme on Rasa‑Aragonesa sheep are presented. Selection objective is to increase numeric productivity and the selection criterion was the ewes prolificacy.
Prolificacy presents a discontinuous and asymmetrical phenotypic distribution with a small number of classes which is not the ideal situation to apply mixed models methodology. Animal breeding values can be estimated ignoring discontinuous nature of this trait (trusting in mixed models method robustness) or using more complex models as non linear or threshold (with two scales: visible and subjacent) ones. Under the threshold model the heritability for prolificacy in the subjacent scale is higher than that in the visible scale. Therefore animals should be selected in the basis of their breeding values estimated in the subjacent scale. Sorensen (1995) proposed a Bayesian approach using Gibbs Sampling to estimate breeding values and genetic parameters in the subjacent scale under a threshold model.
In this case a threshold animal model with repeated measures including herd‑year, lambing season, lambing number, period between lambings and covering mode as fixed effects was applied. Data consisted on 21,605 ewes with 71,130 lambings.
Heritability and repeatability estimates were 0.049±0.004 and 0.106±0.002 in the visible scale and 0.101±0.008 and 0.164±0.004 in the subjacent one, respectively. These values corroborate literature estimates in the sense of higher heritability values obtained in the subjacent scale.