Path Analysis of Wheat Grain Yield with Overcoming Multi-Collinearity of Traits

Path analysis is one of the oldest methods of knowing the direct and indirect effects of traits on yield, but the widespread application of this approach is limited by the problem of multi-collinearity, that has been addressed in the present study. A total of 298 bread wheat genotypes were evaluated during the two cropping years in the lattice design and the path analysis of grain yield was studied by simple and ridge methods. The highest direct effect on grain yield in the first year were related to spike weight, number of seeds and 1000-seed weight, respectively, and in the second year were related to number of seeds, 1000-seed weight and spike weight, respectively. Spike weight and number of seeds in the first year had an indirect effect on yield through each other as well as through spike length and plant height. In the second year, the number of seeds and 1000-seed weight through the weight of spikes showed the most indirect effect. Day to booting and day to flowering traits in both years and spike weight in the second year had a high multi-collinearity that the use of constant k =0.10 in ridge causal analysis overcame this problem. Finally, in contrast to simple path analysis, ridge path analysis showed that the spike weight through the number of seeds had an indirect effect on grain yield in the second year. According to the results, it is recommended to consider the issue of multi-collinearity between traits in future path analysis studies.