Determination of the best model for estimation of genetic parameters on the Fars native chicken traits using Bayesian and REML methods

Abstract

This study was conducted to determine the best model for genetic parameter estimation on the Fars native chicken traits using Bayesian and REML Methods. Studied traits were body weight at first day (BW1), body weight at eighth weeks (BW8), body weight at 12th weeks (BW12), age at sexual maturity (ASM), egg number production (EGP) and mean egg weight during 28th ,30th and 32nd week ages (EGW) involving three generations 17, 18 and 19 during the years 2010 to 2012. Genetic parameters were estimated with REML method using WOMBAT software and with Bayesian approach using MTGSAM software. Based on AIC and DIC criteria, the most appropriate model was determined. Estimations of direct additive heritabilities for BW1, BW8, BW12, ASM, EGP and EGW by the best models using REML method were 0.31, 0.32, 0.29, 0.45, 0.24 and 0.22 and by Bayesian method were 0.36, 0.33, 0.30, 0.48, 0.26 and 0.25, respectively. The genetic correlation coefficients ranged from -0.709 between EGP and ASM to 0.844 between BW8 and BW12 (by Bayesian method) and ranged from -0.724 between ASM and EGP to 0.894 between BW12 and BW8 (by REML method). Generally, based on the employed criteria, the 1st and 2nd models can be suggested for analysis of body weight traits (BW1, BW8 and BW12), whereas for other traits (ASM, EGP and EGW), 1st, 5th, 4th and 6th models seems to be suitable for estimation of genetic parameters of the Fars Native fowls traits using Bayesian and REML Methods. The Bayesian approach recommended for estimation of genetic parameters on the Fars native chicken traits because this method used the prior distribution in the calculation process.