Statistical Analysis of the Relationships Between Yield and Yield Components in Some Durum Wheat (Triticum durum desf.) Genotypes Growing under Semi-Arid Conditions

Abstract

The aim of this study was to know the relation between the grain yield and its components in order to identify the most important attributes in grain yield prediction which would serve as a criterion for the selection of genotypes growing under semi-arid conditions. Fifteen genotypes were studied composed of 11 advanced lines, 03 local landraces and 01 introduced genotype used as controls. Analysis of variance, simple linear correlation multiple linear regression, stepwise multiple linear regression, path analysis, principal component analysis and hierarchical clustering analysis were used to evaluate six traits including grain yield, plant height, days to heading, number of grains spike^-1, number of spikes m^-2 and thousand kernels weight. Analysis of variance showed that genotype effect was significant for the majority of traits studied (p=0.001), advanced line G10 was shown to be most performing (4.723 t ha^-1). Simple linear regression revealed that the number of grains spike^-1, number of spikes m^-2 and thousand kernels weight contributed significantly in grain yield changes (R²=43, 17%). Path analysis showed that the number of grains spike^-1, number of spikes m^-2 and thousand kernels weight had a direct and significant effect on grain yield. Principal component analysis showed that thousand kernels weight and negative days to heading were most important factors traits in grain yield. According to these results number of grains spike^-1 and thousand kernels weight were crucial for the majority of static analysis.