1. The effect of model structure on the model performance to fit milk production data in Isfahan Holstein cows.
- Author
-
Gholizadeh, Sajad, Zamani, Pouya, and Ghafouri-Kesbi, Farhad
- Subjects
- *
MILK yield , *HERITABILITY , *SOMATIC cells , *INFORMATION retrieval - Abstract
The structure of the fixed and random parts of the genetic evaluation model plays a significant role in fitting data and the estimation of genetic parameters for economic traits in livestock. The present study was conducted to investigate the effect of different fixed and random effects combinations in an animal model framework on the general properties of the model and estimates of the genetic parameters for milk production traits. Traits studied were 305-day milk production (305-MY, 15920 records), fat percentage (FP, 27954 records), protein percentage (PP, 26183 records), average daily milk production (ADM, 30954 records) and milk somatic cell score (SCS, 25408 records) in Isfahan Holstein cows. In general, 54 scenarios were studied which differed in fixed and random parts of the model. Variance components were estimated using the animal model fitting restricted maximum likelihood (REML) procedure. The best model for each trait was selected based on the Bayesian Information Criterion (BIC). Results showed that for all traits studied, models in which the effect of contemporary groups Herd-Year-Season (HYS) or Herd-Year-Month (HYM) were fitted as the random or fixed effect together with age at the first calving and inbreeding as a classified fixed effect or covariate lead to a significantly better data fit instead of fitting herd, year, season and month of calving separately. For each trait, a wide range of heritability was obtained by fitting 54 models. Based on the best models, the estimates of heritability for 305-MY, ADM, FP, PP and SCS were 0.33, 0.28, 0.21, 0.16 and 0.61, respectively. It was concluded that a single model should not be used for analyzing all milk production traits and that for each trait a series of models which differ in random and fixed parts should be tested to find the most suitable model which describes the data best. Fitting the effects of herd and year, season and month of calving as contemporary groups HYS or HYM was recommended for genetic evaluation of milk production traits as resulted in better data fit. Depending on the trait, inbreeding and age at first calving can be fitted as a classified fixed effect or as a covariate. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF