In the present work, two uniformity trials were carried out during thefirst and the second winter seasons of 2011/2012 and 2012/2013 in theExperimental farm Demo, Faculty of Agriculture, Fayoum University. Themain objectives were to estimate the optimum plot size, plot shape andnumber of replicates for wheat yield traits using the variety Sakha 93 as plantmaterial. The cultivated area of each field trial was divided into 12 strips;each of which consisted of 100 rows, 0.2 m width and 3.0 m long. Twostatistical methods including soil variability index and maximum curvaturewere used to estimate the optimum plot size and shape using the yield data of1200 basic units (each of 0.6 m2). The data were subjected to two proceduresof statistical analysis to estimate the optimum plot size, when the cost ofconducting the experiment is not taken into consideration and to evaluate theeffect of changing the plot shape on the variability. The first statistical methodwas that of maximum curvature which is based on the exponentialrelationship between plot size and the coefficient of variability. The secondmethod was that developed by smith’s method (1938). Bartlett's test forhomogeneity of variances, as outlined by Steel and Torrie (1980), was usedto study the effect of changing plot shape. The obtained results could besummarized as follows: Increasing the plot size decreased the variance perbasic unit and the coefficient of variability. However, the reduction was not inproportion with the increase in plot size. The index of soil variability rangedfrom 0.6433 to 0.6018 as an average for the 1st and the 2nd seasons,respectively. The relationship between the coefficient of variability (C.V.)and plot size (X) were mathematically expressed by the following equationC.V. = 19.21 X-0.2595 for the 1st season and C.V. = 19.60 X – 0.2725 for the 2ndone. Accordingly, using the soil variability index, the optimum plot size was2 basic units (1/3500 fed.) for the two seasons, while it was 4 basic units(1/1750 fed.) in both seasons when the maximum curvature method wasapplied. The required number of replications for the optimum plot size usingSmith method detecting a 15% difference among treatment means varied 13and 14 in the 1st and the 2nd seasons, respectively. But, for detecting a 20%difference among treatment means, 7 replications in the 1st season and 8replications in the 2nd one were found necessary. Optimum plot size estimatedusing the maximum curvature method detecting a 15% difference amongtreatment means varied 7 and 8 in the 1st and the 2nd seasons, respectively. But, for detecting a 20% difference among treatment means, 4 replications inthe 1st and 2nd season's one were found necessary. Generally, the plot shapedid not affect on the precision of wheat yield trial in most cases in the twogrowing seasons.
Read full abstract