INCREASING PRECISION OF EXPERIMENTAL DESIGNS USING DIFFERENT ANALYSES MODELS IN GROUNDNUT
Abstract: Really, the traditional designs and classical analysis of variance (ANOVA) may be not adequately to evaluate or screen large number of genotypes and reduce the heterogeneity effect of the new soil. To minimize standard error of differences (SED) between means for catching significance, data were analyzed using various designs and analyses models. This study was purposed to compare traditional randomized complete blocks design (RCBD) and alpha lattice design (ALD) by assessing relative efficiency, using two models of analyses, ANOVA as traditional model and restricted maximum likelihood (REML) as non-traditional one. The field trials were conducted during seasons 2017 to 2019 at Marashda Research Station (new soil), Kena, Egypt. A collection of twenty-five groundnut genotypes were grown in an alpha lattice design with three replications. Results elucidated that the studied genotypes differed significantly for all yield-traits. Relative efficiency of using designs of ALD over RCBD increased experimental accuracy by 72.77- 46.23 % and 21.92-14.78% for pod weight plant-1 and pod yield fed-1 in both seasons, respectively. Therefore, the results confirmed that the RCBD should be replaced by ALD experiments in testing large number of treatments. Relative precision of using REML analysis over ANOVA increased accuracy by 31.32 - 15.88 % and 8.54-1.76 % pod weight plant-1 and pod yield fed-1 in both seasons, respectively. Generally, REML analysis was more precise and effective in reducing the SED compared to ANOVA for individual and combined analysis. The rank of genotypes mean across the different analyses and seasons were not constant. Then, REML analysis increased the efficiency of genotype selection for further evaluation, providing unbiased estimates. The genotypes that score rank over the studied checks (Giza 6, Ismailia-2, Suhag 104, Suhag 107 and Suhag 110) that are considered the best elite group, included Line-25A and Line-110 genotypes with highest pod yield, followed by Line-32A, Line-2B, Line-19A, Intro.-335 and Intro.-504. Therefore, this collection which is the best promising elite group should be effective and satisfactory for successful breeding purposes under Marashda new soil.
URL
Publication year 2021
Pages 1075-1096
Organization Name
Author(s) from ARC
Publication Type Journal
↑