Representative image and trait analysis of cotton
Drought stress had significant effects on above-ground and root traits in cotton (Fig. 1). As indicated in supplementary tables 4 and 5, among the above-ground traits, leaf area, plant height, SPAD, leaf water potential, and relative water content displayed diminishing tendencies in response to drought stress. In contrast, canopy temperature exhibited an increasing pattern. In terms of root traits, root dry weight, root surface area, root volume, average lateral root emergence angle, average lateral root tip angle, and lateral root count demonstrated declining trends in response to drought stress. Meanwhile, average length - all roots, average length - lateral roots, width/depth ratio, specific root length, and specific root surface area displayed diminishing tendencies. The coefficients of variation for the measured traits ranged from 1.47 to 57.40 in 2021 and from 0.70 to 60.44 in 2022, with most traits exceeding 20% in both years (Supplementary Tables 4 and 5).
Spearman's correlation analysis, which revealed strong correlations between different traits, resulted in data redundancy and potentially undermined an accurate assessment of drought tolerance in cotton (Fig. 2). Consequently, we conducted comprehensive evaluations using PCA and cluster analysis.
PCA of each traits
Six principal components were extracted for both growing seasons (Tables 1 and 2). In 2021, the contribution rates of the top six comprehensive evaluation indicators, Cl Composite Index (Cl1-Cl6), were 45.70%, 14.09%, 6.26%, 5.16%, 4.45%, and 4.05%, respectively. The cumulative contribution rate of the six principal components was 79.71% (Table 1). In 2022, the contribution rates of Cl Composite Index (Cl1-Cl6) for the top six comprehensive evaluation indicators were 59.02%, 7.25%, 5.84%, 5.56%, 4.29%, and 3.76%.
The cumulative contribution rate of the six principal components was 85.72%. The two years of experimental data indicated that by transforming the original 28 single traits into six independent composite indicators, i.e., six principal factors, most of the information could be covered (Table 2). In 2021, root volume, surface area, average lateral root emergence angle, average diameter, total root length, and root dry weight had higher load coefficients. However, in 2022, average length - all lateral roots, average lateral root tip angle, root dry weight, average length - all roots, canopy temperature, and leaf area had higher load coefficients. These traits primarily reflected cotton's root traits and canopy parameters.
Comprehensive evaluation and screening of drought tolerance traits
Subordinate function values of 80 variety composite indicators were calculated according to formula (3) (Tables 3 and 4). In this principal component, higher Cl values indicate greater drought tolerance, while lower values indicate weaker drought tolerance. Based on the contribution rates of each composite indicator, indicator weights were calculated using formula (4).
After calculation, the weights of the six composite indicators in 2021 were 0.57, 0.18, 0.08, 0.06, 0.06, and 0.05, respectively (Table 1). In 2022, the weights of the six composite indicators were 0.69, 0.08, 0.07, 0.06, 0.05, and 0.04, respectively (Table 2). Eq. (5) was used to calculate the drought tolerance of different cotton varieties (Tables 3 and 4), and the drought tolerance of different cotton varieties was ranked according to the D-value (Tables 5 and 6). A smaller D-value indicates poorer drought tolerance, whereas greater drought tolerance corresponds to larger D-values (Tables 5 and 6).
We used PCA to evaluate 28 traits under well-watered and drought stress conditions (Fig. 3). In both well-watered and drought stress conditions, the first two principal components contributed 36.6% and 43.2%, respectively (Fig. 3). Specifically, under well-watered conditions, PC1 accounted for 28.7%, while PC2 explained 7.9% of the variance. PC1 was primarily characterized by SPAD, average diameter, canopy temperature, and water loss rate of the shoot, whereas PC2 predominantly featured average length - all lateral roots, projected area, average lateral root emergence angle, and average lateral root tip angle. Conversely, under drought stress, PC1 explained 34.2%, and PC2 explained 9.0%. PC1 was principally represented by water loss rate of the shoot, dry root/shoot ratio, root tissue density, and specific root surface area, while PC2 mainly showcased root width, average lateral root emergence angle, lateral root count, and maximum depth.
Comprehensive evaluation of drought tolerant varieties
In this investigation, we employed the Euclidean distance flattening method and systematic cluster analysis were used to classify 80 cotton varieties into five categories based on D-values, including drought-tolerant varieties, weak drought-tolerant varieties, intermediate varieties, and weak drought-sensitive varieties, drought-sensitive varieties. Our two-year field screening experiments revealed that Ji668, Guoxinmian02, Xuzhou 1818, and Han 6203 demonstrated higher D-values, contrasting with Ji 228, Guoxinmian 9, Zhongmiansuo 23, and Hanwu 216, which displayed lower D-values. Consequently, Ji668, Guoxinmian02, Xuzhou 1818, and Han 6203 can be categorized as drought-tolerance varieties, whereas Ji 228, Guoxinmian 9, Zhongmiansuo 23, and Hanwu 216 fall into the drought-sensitive varieties (Fig. 4).
Validation of biomass and yield drought tolerance index
Under drought stress, there was a significant reduction in both the above-ground biomass and cotton yield. Nonetheless, drought-tolerance varieties exhibited considerably higher drought tolerance coefficients in comparison to their drought-sensitive varieties. Specifically, in 2021, the drought tolerance index of above-ground biomass for drought-tolerance varieties Ji668, Guoxinmian 02, Xuzhou 1818, and Han 6203 were 0.31, 1.73, 0.57, and 1.43, while for drought-sensitive varieties Ji 228, Guoxinmian 9, Zhongmiansuo 23, and Hanwu 216 were 0.22, 0.39, 0.11, and 0.08 (Fig. 5A). The trends in 2022 closely mirrored those in 2021 (Fig. 5B).
In 2021, drought tolerance index of yield for drought-tolerance varieties Ji668, Guoxinmian02, Xuzhou 1818, and Han 6203 were 1.36, 1.39, 1.30, and 1.33, while for drought-sensitive varieties Ji 228, Guoxinmian 9, Zhongmiansuo 23, and Hanwu 216 were 0.07, 0.29, 0.12, and 0.15 (Fig. 6A). Similar trends were observed in 2022 compared to those in 2021 (Fig. 6B).