Structural Equations Modeling of Hidden Genetic Constructs of Production, Reproduction and Longevity in Holstein Cows in Iran

     In structural equation modeling, it is possible to create hidden variables that cannot be directly measured by the researcher. The aim of this study was to model the structural equations of hidden genetic constructs of production, reproduction and longevity of Holstein dairy cows in Iran using breeding value of the traits that make up these constructs. Accordingly, the breeding value of the traits of the age at first calving, calving interval and the open days made up the hidden genetic construct of reproductive production; the adjusted milk, fat, and protein production traits formed the hidden genetic construct of the product and the lifetime traits in the herd and the productive lifespan constituted the hidden genetic construct of longevity. Estimation of single-trait and multi-trait animal model of genetic variance components, breeding values of each of the above traits was performed using Bayesian method conducted in GIBBS3F90 software. For this purpose, 500,000 rounds were performed and 50,000 initial samples – as burnt-in were removed from analysis. Then, using the variance-based method implemented in SMARTPLS software, 4 conceptual models of structural equation modeling (SEM) were created using the breeding value of the mentioned traits. Using the single animal model, the heritability of milk, fat and protein productions, age at first calving, calving interval, open days, lifetime in herd and production lifespan were 0.36, 0.55, 0.56, and 0.0250, 0.24, 0.07, 0.07 and 0.011, and using the multi-animal model, these values were 0.21, 0.52, 0.42, 0.48, 0.38, and 08, 0.0, 0.74 and 0.51, respectively. The estimated genetic correlation between the traits varied from -0.87 (milk production-fat production) to 0.98 (lifetime in herd and productive lifespan). The report root means square residuals (RMSR) on different SEM models fitted to breeding values of the traits that made up the various structures showed that the best fit was obtained when reproduction and production constructs had causative effect on longevity construct. This type of modeling approach has been used for the first time in the animal sciences which is rooted in social science researches. Therefore, adapting SEM modeling assumptions to the reality of animal sciences data can be a new field for breeding specialists, especially managers of production farms, because researchers in breeding sciences can ask many questions and hypotheses in the form of different constructs and create appropriate and establish a better management for the breeding farm due to the significant causal relationship between the extracted constructs.