Models and Comparison Methods for Yield Analysis of Wheat Backbone Parents and Its Derivative Progeny

Tāo Zhāng,Yang De,Ji WanQuan.,Xinlun Liu
2010-01-01
Abstract:In order to explore the models and analysis methods of wheat backbone parents and its derivative progeny on character comparision,Base on R language program,The 2 models of linear fixed model and linear mixed model on grouping block within repeat,and the 2 analysis methods of Tukey's multiple comparison test and the Scott-Knott grouping method were chosen for studying on the data of plot yield from yangling dwarf wheat variety regional trials in 2008.The results showed that,the Standard error mean on the estimated treatment effects of linear fixed model and linear mixed model value were 140.35 and 96.75,respectively,the Standard error mean on the estimated treatment effects of linear mixed model on grouping block within repeat value was lowered 31.06%than linear fixed model mean value.The Results using linear mixed model,for Tukey's multiple comparison tests (Significance level 0.05),would allow 252 treatment were divided into 35 groups with overlapping letters.However,the Scott-Knott grouping method would allow all treatment were divided into four groups without overlapping letters.As can be seen,the error would be smaller if we used mixed models to estimate treatment effect when a large number of treatments and unbalanced missing-data in in- complete block experiments.Through the Scott-Knott test,It can be got the results without overlapping letters.The results also have simple expression and easy to explain in large sample size,the linear mixed model and the Scott-Knott grouping method was suggested to apply in the same trials in the future,and would obtain the better results.
What problem does this paper attempt to address?