Evaluation of some Egyptian cotton genotypes in the Delta using both compressed and collected latin square design

The present investigation dealt with the variances of four Egyptian cotton (Gossypium barbadense L.) genotypes, with respect to yield, its components and fiber properties in the Delta region during 2009 and 2010 seasons. One of the genotypes, cultivar ; G.86 is normally grown in the Delta (control). The others were hybrids (Bulk families), viz. (G.89 x G.86), (G.75 x Sea) and (10229 x G.86). The genotypes were evaluated in the Delta (Sharkia, Dakahlia, Gharbia and Mounofia). Simple latin square design (4 x 4) was used in each experiment. The results exhibited that 10229 x G.86 was the best genotype in the three locations except Dakahlia region since it kept the first rank with respect to yields (seed and lint) and significantly surpassed G.86 in the two seasons. A compressed analysis was used to estimate location and genotype variances. In this research the data of each location (two seasons) considered column and each cell of the design includes eight readings. Statistical analysis of compressed was similar to analysis of simple latin square for more than one observation per experimental unit. A collected analysis was used to estimate locations and genotypes variances due to test one column for fiber properties in each location. Statistical analysis of the collected was similar to analysis of simple latin square with a single determination per plot. The objective of the compressed and the collected was developing simple analysis to use a combined analysis. The compressed surpassed combined since it does not need to calculate homogeneity test of variances (Bartlett test) before the start of analysis. It calculated among locations variances direct through columns, while combined calculated the same value indirect through partitioning locations. It calculated genotypes variance direct, while combined calculated the same value indirect through partitioning genotypes within locations. It gave more information with respect to performance of genotypes under different locations. The collected surpassed combined since it could estimate among locations and genotypes variances using one column of each location. Results of both traditional combined and compressed analysis exhibited that (G.75 x Sea) and (10229 x G.86) significantly surpassed G.86 in yields (seed and lint). 10229 x G.86 was the best genotype with respect to yield components except for boll weight since it kept the first rank and significantly exceeded G.86. Results of collected exhibited that fiber properties were not affected by locations except for strength.