Precision of Some Statistical Procedures in Evaluating Seed Cotton Yield of Twenty Five Cotton Genotypes | ||||
| The Egyptian Statistical Journal | ||||
| Article 3, Volume 58, Issue 2, December 2014, Page 145-170 PDF (15.56 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/esju.2014.314446 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
| El-Kadi D.A.* ; A.M.M. Al-Naggar; Manal M.S.; Mona E.Shalaby | ||||
| Fac.Agric. c.u. Giza, Egypt | ||||
| Abstract | ||||
| The present investigation was carried out at the Agricultural Experiment and Research station of Faculty of Agriculture, Cairo University, Giza, Egypt during 2008 and 2009 seasons. Twenty five cotton families, lines and cultivars were used. The balanced lattice design (5X5) with six replications was used as a basic design. All recommended agricultural practices were used. The studied trait was seed cotton yield in grams per plot. The studied statistical procedures were traditional designs (randomized complete blocks design (RCBD), balanced lattice design (6-replications) and partially balanced lattice with 2,3,4 and 5 replications). Also, non-traditional analyses of restricted maximum likelihood (REML) method as ordinary, spatial and meta models were proceeded, for all replication combinations. All combinations from 6 replications were analyzed; first combination consisted of two replications (15 combinations), second of three replications (20 combinations), third of four replications (15 combinations) and fourth of five replications (6 combinations), using the respective lattice analysis for each combination. Relative precision was calculated for each replication combination in each season. The highest one in relative efficiency for each combination was identified. Non-traditional methods of statistical analyses were applied to the best -2, -3, -4 and -5 replication data sets. In both seasons, certain genotypes either in F3 or F6 shower significantly higher seed cotton yield than commercial cultivars under study. Based on results obtained, either quintic or balanced lattices could be recommended instead of RCBD. In general, as long as number of replications increase, the precision increases as in quintic and balanced lattices. The results were extended to detect the most précised REML models using four estimated parameters, i.e., residual variance (e12), x², deviance (DV) and akaike information criterion (AICD) were used to detect the best REML model. For all data sets, meta REML model was detected as a best REML model for increasing the precision of cotton field trials compared with ordinary and spatia: REML models. Except for 2-replications data set, either replications or replications and blocks alternative sub- models revealed their importance in increasing precision of experiments. Precision of REML models compared with the traditional designs were included in the present study. In both seasons, except for 2-replication data set, highest C.V. values were obtained for RCBD followed by balanced lattice, ordinary-, spatial- and meta- REML models, respectively. Concerning 6-replication data set, based on the averages of C.V. estimates for the two seasons of study, the lowest C.V. estimates were obtained for meta REML model (18.58%) in comparison with 51.14 and 35.31% for RCBD and lattice design, respectively. Furthermore, the same trend of results was detected for other data sets. Results showed the effect of adjustment of genotypes for unexplained variability in ranks for selection of the best genotypes. Finally we could conclude that the meta REML model is recommended for analyses of data for cotton field trials. | ||||
| Keywords | ||||
| Cotton Yield Trial; Balanced and Partially Balanced Lattice Design; Relative Precision; REML Model; Spatial Model; Meta Model; C.V. and RD | ||||
|
Statistics Article View: 118 PDF Download: 512 |
||||
