Introduction:  Safflower has a special place among oilseeds compatible with the climatic conditions of Iran as a plant tolerant of drought and salinity stress and having different growth types. On the other hand, due to the fact that Iran is one of the arid and semi-arid regions with an average of one-third of global rainfall, in such areas some important stages of plant growth, including safflower, are affected by reduced soil water potential. In order to adapt the plant to stressful climatic conditions, morphological and physiological changes and the accumulation of new metabolites, along with structural changes, increase efficiency under stressful conditions. Therefore, a better understanding of the relationship between environment and genotype and the study of physiological, biochemical, and morphological traits of plants in adverse environmental conditions, especially drought stress, will lead to a better understanding of plant adaptation and events involved in drought tolerance mechanisms. Canonical correlation analysis, like other multivariate statistical methods, is a way to estimate or reduce the number of data points. This method is a generalized multiple regression mode that is used to determine the relationship between two sets of variables. In this regard, this research was conducted to better understand the relationships among different traits under drought stress and use these relationships to select more desirable cultivars.
Materials and methods: This experiment was performed on a farm located at 5 km of Zahedan-Bam road in the Lakhshk area. This research was conducted as a split plot based on a randomized complete block design with four replications. An Experiment with three stress levels, including severe drought stress (irrigation after 250 mm of evaporation from the Class A evaporation pan surface), mild drought stress (irrigation after 175 mm of evaporation from the Class A evaporation pan surface), and normal (irrigation after 100 mm of evaporation from the Class A evaporation pan surface), was performed as the main factor, and 10 safflower cultivars were used as the sub-main factor. Drought stress started at the stem elongation stage. To measure the studied morphological and physiological traits in the experiment, sampling was performed at the grain formation stage. Before data analysis, a data normality test was performed, and after ensuring normal data distribution, analysis of variance and mean comparison by the Duncan method were performed. Due to the insignificance of the interaction of stress and cultivar in the analysis of variance of the data, canonical correlation analysis was performed for the mean of all stress levels using Statistica software.
Results and discussion: According to the analysis of variance, the studied traits were affected by irrigation treatment, and there was a significant difference between the cultivars in terms of all studied traits under stress. This indicates the existence of high genetic diversity in terms of the mentioned traits and the possibility of selection for these traits among the studied cultivars. Also, the interaction of cultivar and stress was not significant for all traits. On the other hand, in order to investigate the relationships between traits related to drought stress tolerance in this study, canonical correlation analysis between two groups of physiological traits, including total leaf water potential, leaf relative water content, leaf osmotic potential, chlorophyll fluorescence, proline concentration, and glycine betaine concentration, and morphological traits, including plant height, plant dry weight, plant fresh weight, and number of sub-branches, was determined. The Wilkes -Lambda statistic was significant at the 1% probability level, so there was a significant correlation between physiological traits and morphological traits. In explaining the canonical function related to physiological traits, all variables except proline concentration, leaf relative water content, and leaf osmotic potential had positive coefficients. Total leaf water potential (X1) with a coefficient of 0.299 played the greatest role in justifying the first canonical variable. For canonical functions related to morphological traits, plant height had a negative coefficient, and other traits had a positive coefficient. Plant fresh weight with a coefficient of 0.987 had the greatest effect in explaining this canonical variable. On the other hand, the relative water content of the leaves, with a coefficient of -0.376, played the least role in explaining the first canonical function of physiological traits. It seems that cultivars that have higher total leaf water potential under stress conditions have higher plant fresh weight and other morphological traits (except plant height).
Conclusion: According to the results of the mean comparison, the Parnian cultivar was the most tolerant, and the Padideh and Zarghan cultivars were the most sensitive cultivars under drought stress. Also, based on the results obtained from canonical correlation analysis, it was found that there is a significant correlation between the pair of canonical variables resulting from physiological traits and morphological traits. There was a high correlation between total leaf water potential and the relevant canonical variable, as well as between plant fresh weight and the number of sub-branches with the relevant canonical variable; therefore, genotypes that have a higher plant fresh weight and a larger number of sub-branches have more total leaf water potential. In general, according to these results, it can be stated that more favorable morphological traits of the plant are associated with drought tolerance.