Abstract: This study was aimed to investigate the effects of different data structures and animal models on the estimation of genetic parameters of economic traits of Alpine Merino sheep(14 months),and select the best animal model.The best model was used to estimate the genetic parameters of weight,greasy fleece weight,clean fleece yield,clean fleece weight,average fiber diameter,coefficient of variation of fiber diameter and staple length,which could provide the theoretical basis for the breeding of Alpine Merino sheep.The data of 20 720 sheep were divided into data set 1 and data set 2 according to the genealogical integrity and data amount using the correlation function of data sorting in R language.The significance of four non-genetic factors year of identification,birth type (single or twin),group and sex in two data sets were tested by ANOVA with R language.Extremely significant effect (P < 0.01) was put into the animal model as fixed effect.Four models were obtained by combining two data sets and two single-trait animal models.Model 1 and model 2 used data set 1 and data set 2,respectively.The random effects were individual additive genetic effect and residual effect.Model 3 and model 4 used data set 1 and data set 2,respectively,and the random effects were individual additive genetic effect,individual permanent environmental effect and residual effect.Variance component estimation was implemented by ASReml4 software.The Akzo information criterion (AIC) and Bayesian information criterion (BIC) were used to evaluate each model,and Likelihood ratio test (LRT) was used to compare each model.Finally,the optimal model was selected to estimate the heritability.The results showed that,①Fixed effect significance test showed that all traits in data set 1 and data set 2 of identification year and gender were extremely significant (P < 0.01),birth type was extremely significant for weight and greasy fleece weight in data set 1 (P < 0.01),and gender was only extremely significant for weight (P < 0.01).②The direct heritabilities were 0.1614-0.2392,0.1958-0.3254,0.4395-0.5539,0.2003-0.2393,0.4024-0.5897,0.3174-0.6077,0.2960-0.3669 for weight(WT),greasy fleece weight(GFW),clean fleece yield(CFY),clean fleece weight(CFW),average fiber diameter(FD),coefficient of variation of fiber diameter(CVAFD) and staple length(SL),respectively.③Likelihood ratio test showed that there was no significant difference between model 1 and model 3 for all traits (P > 0.05).Model 1 and model 4 showed significant differences in body weight and staple length (P < 0.01).Model 2 and model 3 showed significant difference in clean fleece weight (P < 0.01),but no significant difference in other traits (P > 0.05).There was no significant difference between model 2 and model 4 for all traits (P > 0.05).In conclusion,the optimal model of clean fleece weight was model 1,and the optimal model of WT,GFW,CFY,CFW,FD,CVAFD and SL was model 2.All traits were not significantly affected by individual permanent environment (P > 0.05).Based on the optimal model,WT,GFW,CFY,CFW,FD,CVAFD and SL heritability of alpine merino sheep were estimated to be 0.2392,0.3254,0.4394,0.2893,0.4222,0.3175 and 0.3670,respectively.

Key words: Alpine Merino sheep; economic traits; animal model; model evaluation; heritability