Interação genótipo X ambiente em codornas de corte criadas em diferentes níveis de treonina

Abstract

We aimed to evaluate the inclusion of fixed effects and residual variance classes in random regression models for estimating genetic parameters and to assess the presence of genotype-environment interaction for body weight at 21 and 35 days of age (BW21 and BW35) in meat-type quails of the ICA I and ICA II genetic groups raised at different threonine levels (1.14, 1.24, 1.34, 1.44, and 1.54%). Two articles were developed from the experiment. In the first article, the inclusion of fixed effects such as sex, incubation, interaction between sex and incubation, and birth weight was evaluated through multivariate analysis of variance (MANOVA) using the SAS program. Five models considering residual variance classes according to threonine level were compared: homogeneous with only one class (1.14-1.54%) and heterogeneous containing two classes (1.14-1.34% and 1.44-1.54%); three classes (1.14-1.24%, 1.34%, and 1.44-1.54%); four classes (1.14%, 1.24%, 1.34%, and 1.44-1.54%); and five classes (1.14%, 1.24%, 1.34%, 1.44%, and 1.54%). Models were compared using the likelihood ratio test (LRT), and model fit was assessed by Bayesian Information Criterion (BIC). Sex, incubation, and linear birth weight effects were significant and included in the models. There were no significant differences between the models by LRT. For BIC values, the best fit for both genetic groups for both traits (BW21 and BW35) was obtained with the homogeneous model. The fixed effects in random regression models should be evaluated, and models with homogeneous residual variance may be used for variance component estimation. In the second article, the presence of sensitive genes and genotype-environment interaction for BW21 and BW35 across threonine levels was assessed. Random regression coefficients were estimated considering direct additive genetic effects, homogeneity of residual variance, and adjustment by second-order ordinary polynomial. Reaction normswere created with the regression coefficients, and variance components were estimated. There was an increase and a decrease in the dispersion of estimated genetic values across threonine levels. Heritability estimates (h²) increased with increasing threonine levels, except for BW35 in ICA I. For BW21, h² at the lowest and highest levels was 0.35 and 0.54 for ICA I and 0.64 and 0.77 for ICA II. For BW35, h² at the lowest and highest levels was 0.64 and 0.49 for ICA I and 0.49 and 0.68 for ICA II. There was no reclassification of genetic values, indicating that value prediction can be made at any of the tested levels.