Dear @Bidhan K. Mohapatra I appreciate you for asking a highly important scientific question. The contribution of all the narrated components of yield has been estimated individually in front line demonstrations. However, estimation in isolation cannot be extrapolated when we want to know proportional contribution of individual components (taken together) on yield. Literature exactly on this aspect is not available (if I have not missed it). It may be perhaps because as the number of variables increases, interaction among variables increases exponentially. Moreover, increasing levels of variables further complicate the design of experiment. For example, with five variables (which you have mentioned) there will be 10+10+5+1=26 interactions, and even merely with two levels of each variable with 3 replications, there will be 96 plots. I think perhaps for these reasons, researchers hardly conduct experiments with such a number of variables. To my opinion, until such an experiment is conducted, it will be very difficult to precisely answer your question.