The Prediction of Brody, Logistik and Von Bertalanffy Models By Using The Bayesian Approach for Modeling The Growth Curves in Holstein Calves

Abstract

In cattle breeding, growth curves are used for determining the most appropriate slaughter age, obtaining information about the general health status of animals, estimating the age of sexual maturity and the age of use in breeding, and selection studies. The aim of this study is to estimate the growth curves of Holstein calves using the Bayesian Approach of Brody, Logistik, and Von Bertalanffy models. The live weight data was collected from 34 Holstein calves raised at the cattle research farm of Niğde Ömer Halisdemir University, Ayhan Şahenk Agricultural Research and Application Center in 2019. Furthermore, for estimating the frequency modeling of the Holstein breed the predicted parameter values and standard deviation of parameters were used as the prior information. The Bayesian approach was used for making the statistical analysis. Monte Carlo Method Markov Chains (MCMC) algorithms were used to estimate the posterior distributions and it was 900,000 in total while excluding the 8000 burn-up periods. Random distribution graphs and autocorrelation graphs were used to control the iterations for the detection of posterior distributions. In this study, no problems arising from iteration were found. Moreover, the distribution information of the Brody, Logistic, and Von Bertalanffy model was calculated for the results. The Brody, Logistik and Von Bertalanffy model parameters distributions results can be used for modeling studies of the Holstein cattle breed. In addition, the compatibility of Brody, Logistik and Von Bertalanffy models was investigated by using data set, mean information of the posterior distributions estimated at the end of the study. The information for Brody, Logistik, and Von Bertalanffy model parameters was calculated, and the results of the posterior distributions showed the Deviation Information Criteria (DIC) values. For the comparison between the three models DIC values were calculated as 55.19, 33.17 and 38.02, respectively, and it was decided that the most compatible model was the Bayesian Logistics Model. The Bayesian Logistic Model, which is decided to be the most compatible, is a study-specific result.