Variability, correlation, path analysis and stepwise regression for yield components of different wheat genotypes

Abstract

In order to evaluate the variability and relationship between different wheat yield components, a randomized complete block design experiment with ten genotypes of wheat had been carried out during three growing seasons (2010-2012). The number of spikelet per spike and grain weight per spike had low genotypic and phenotypic variability, while plant height had the highest one. High heritability was observed for plant height (h(2) =93.1%), spike length (h(2) =92.3%) and spike density h(2) =92.9%). The low heritability was found for grain weight per spike (h(2) =35.6%). Grain weight per spike was in significant positive genotypic and phenotypic correlation with all the traits (plant height, spike height, number of spikelet per spike, number of grain per spike and spike weight) except spike density. The spike weight had the highest phenotypic (r(p) =0.988), while number of spikelet per spike had the highest genotypic correlation with grain weight per spike (r(g) =0.981). Path coefficient analysis revealed that all the traits had highly significant direct effect on grain weight per spike, except spike length. The stepwise regression revealed that 87.1% of the grain weight per spike variation was explained by model which excludes spike length. Spike weight and plant height had the highest shared and unique contribution to grain weight per spike.