The application of recursive multivariate model for genetic evaluation of early growth traits in Raeini Chasmere goat: A comparison with standard multivariate model

Data collected on 2638 kids originated from 1305 dams and 197 sires during 1979 to 2012 in Raeini Cashmere goat breeding station located in Baft, Kerman province, south-eastern part of Iran, on body weights of 2638 Raeini Cashmere kids at birth (BW), at weaning (WW) and at six-month age (6 MW) were used to adopt a fully recursive multivariate model considering causal effects from BW on WW and on 6 MW and from WW on 6 MW. The adopted recursive multivariate model (RMM) was compared with a standard multivariate mixed model (SMM), in which causal relationships were ignored, in terms of deviance information criterion (DIC), predictive ability and the estimates of genetic and phenotypic parameters. Preliminary analyses revealed that the importance of direct additive genetic, maternal permanent environmental and temporary environmental effects for genetic evaluation of all the studied traits. Maternal additive genetic effects were only influencing on BW. Model comparisons via DIC revealed the plausibility of RMM over SMM; DIC value obtained by fitting RMM (-15,218.76) was lower than that of obtained under SMM (21,654.92). For BW, lower mean square of error (MSE) and higher Pearson's correlation coefficient between the observed and predicted values (r(y,y^)) were obtained under SMM but in the case of WW and 6 MW lower MSE and higher r(y,y^) were obtained under RMM. The causal effects of BW on WW and on 6 MW and that of WW on 6 MW were 1.94 kg, 2.48 kg and 1.03 kg, respectively. The estimates of direct heritability, maternal heritability (for BW) and ratio of permanent and / or temporary environmental variances to phenotypic variance were not significantly affected by model fitted (SMM and RMM). Direct additive genetic correlations of BW-WW and BW-6 MW were not statistically significant under SMM (95% HPD interval included zero) but the corresponding estimates under RMM were significant (99% HPD interval did not include zero) values of 0.28 and 0.84, respectively; implying that genes associated with BW also increase WW and 6 MW in Raeini Cashmere goat. Direct genetic correlation estimate for WW-6 MW were 0.88 (under SMM) and 0.29 (under RMM). Generally, RMMs perform better than SMMs and may provide a better explanation to biological relationships among the studied traits.