Different populations were simulated according to the genealogical and productive structure of Alava meat horse breeders' association in order to determine the reliability of a variance component estimation in the real population. The simulation model consisted of an additive model, which included herd‑year as fixed effect and direct and maternal genetic additive effects. A total of 27 populations were simulated for every combination of the following genetic parameters: heritability of direct effect (h_d² = 0.1,0.25 and 0.40), heritability of maternal effect (h_m² = 0.1 ,0.25 y 0.40), and genetic correlation between both effects (r_dm = 0.2, ‑0.2 and ‑0.40). Each popular ion was simulated 100 times.
Afterwards, variance components were estimated from the simulated data using a REML method. In this analysis, two models were considered: the first one dealing with direct and maternal effects, and the second one including only direct effects. In the first model, estimates of direct and maternal effects were accurate, but estimates of genetic correlation showed lack of accuracy. In the second one, the obtained estimates resulted in a highly positive bias.
It can be concluded that the data structure could allow an acceptable variance component estimation, provided that maternal effects are included in the model.