EVALUATION OF STATISTICAL METHODS FOR DETERMINING THE RELATIVE CONTRIBUTION OF YIELD FACTORS IN WHEAT | ||||
| Egyptian Journal of Agricultural Research | ||||
| Article 30, Volume 76, Issue 4, December 1998, Page 1733-1750 PDF (5.3 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/ejar.1998.365564 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| SAED M. NASR | ||||
| Central Laboratory for Design and Statistical Analysis, Agricultural Research Center Giza, Egypt | ||||
| Abstract | ||||
| Three statistical procedures of relating yield components to yield namely; multiple linear regression (in the form of full model, stepwise and pathe analysis), factor and cluster analysis were applied to 10 yield factors to evaluate these statistical techniques. A set of six widely grown wheat varities, namely; (Giza 155, Giza 157, Giza 163, Sakha 8, Sakha 61 and Sakha 69) were used. They were planted at Sids during two successive seasons of 1995/96 and 1996/97 in a randomized complete block design with four replications. The most important results obtained from this investigation can be summarized as follows: 1.Simple correlation coefficients revealed that No. of spikes/m2, spike weight, No. of plants/m2 and plant weight had the greatest influence on grain yield. Path analysis indicated that No_ of spikes/m2, plant weight and No. of plants/m2 were the three factors that exerted the greatest influence both directly and indirectly upon grain yield. When studying correlations or path-analysis it is important to recognize the nature of population under consideration. The magnitude of correlation coefficient can often be influenced by the choice of individuals upon which the observations are made. For this reason path analysis may not be applicable to in some cases. 2.Multiple linear regression and stepwise regression had the same adjusted R2 (93%) denoting similar accuracy_ The test of mu[ticol-linearity and linearity in the two procedures between variables indicated that, these models showes highly multiple coefficient of determination (R2 = 0.94) for prediction only and were not valid to interpret the partial correlation coefficient. 3.Factor analysis and cluster analysis would seem to be more suitable and efficient than the other techniques. They provide more information about cluster of intercorrelated variables. The two procedures indicated greatest relationships between (No. of spikes/m2 & plant weight),(1000-kernels weight & grain yield/spike) and (spike length & No. of spikelets/spike). These relationships will enable wheat breeders to realize high grain yield. | ||||
|
Statistics Article View: 57 PDF Download: 118 |
||||
