Abstract

COMPARING LINEAR MIXED MODELS FOR PRELIMINARY YIELD TRIALS THAT FOLLOW AUGMENTED EXPERIMENTAL DESIGNS

Sudha Neupane Adhikari, Jixiang Wu, and Melanie Caffe-Treml

Agronomy, Horticulture, and Plant Science Department,

South Dakota State University, Brookings, SD 57007

Abstract

Ineffective control of spatial variation when analyzing field trials data may lead to biased conclusions, which in turn could impact selection efficiency in plant breeding programs. In this study, a group of 78 oats breeding lines were evaluated in preliminary yield trials at four locations in South Dakota in 2015. Four linear mixed models (with and without row and column effects) were compared regarding reduction in error variance, heritability, and model relative efficiency for three traits (grain yield, test weight, and heading date). Results showed that accounting for row and column effects in the model was effective in reducing error variance and thus improved heritability and model relative efficiency for grain yield and heading date. Inclusion of row and column effects in the statistical models reduced the error variance by 20% and 14% for grain yield and heading date, respectively. For test weight, there was 11% reduction in error variance when only row effect was included in the model suggesting the absence of column effect. Results suggests that for traits affected by spatial trends, the inclusion of row and column effects in statistical models should improve the selection efficiency.

Keywords

Augmented experimental design, Linear mixed model, Variance components, Error variance.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
May 1st, 3:00 AM

COMPARING LINEAR MIXED MODELS FOR PRELIMINARY YIELD TRIALS THAT FOLLOW AUGMENTED EXPERIMENTAL DESIGNS

COMPARING LINEAR MIXED MODELS FOR PRELIMINARY YIELD TRIALS THAT FOLLOW AUGMENTED EXPERIMENTAL DESIGNS

Sudha Neupane Adhikari, Jixiang Wu, and Melanie Caffe-Treml

Agronomy, Horticulture, and Plant Science Department,

South Dakota State University, Brookings, SD 57007

Abstract

Ineffective control of spatial variation when analyzing field trials data may lead to biased conclusions, which in turn could impact selection efficiency in plant breeding programs. In this study, a group of 78 oats breeding lines were evaluated in preliminary yield trials at four locations in South Dakota in 2015. Four linear mixed models (with and without row and column effects) were compared regarding reduction in error variance, heritability, and model relative efficiency for three traits (grain yield, test weight, and heading date). Results showed that accounting for row and column effects in the model was effective in reducing error variance and thus improved heritability and model relative efficiency for grain yield and heading date. Inclusion of row and column effects in the statistical models reduced the error variance by 20% and 14% for grain yield and heading date, respectively. For test weight, there was 11% reduction in error variance when only row effect was included in the model suggesting the absence of column effect. Results suggests that for traits affected by spatial trends, the inclusion of row and column effects in statistical models should improve the selection efficiency.