Safflower (Carthamus tinctorius L.) crop adaptation to residual moisture stress: conserved water use and canopy temperature modulation are better adaptive mechanisms

Experiment-1: evaluation of safflower genotypes for transpiration under progressive soil moisture stress

For measuring transpiration using gravimetric method among the genotypes, pot experiments were conducted at ICAR-Indian Institute of Oilseeds and Research, Rajendranagar, Hyderabad (17.32012°N, 78.40931°E, and 487 m above MSL). Pots were filled with 12 kg of black soil (Order: Vertisol) and five seeds were sown in each pot during the last week of October. After the successful establishment of the crop, the plants were thinned and only two healthy plants were retained. At 45 days after sowing (DAS), when the plants were at the vegetative stage, pots were irrigated until saturation and left drained for overnight to remove the excess moisture. At this point, the soil in the pots was at 100 per cent field capacity. Before the onset of different water regimes, a replication set from the control as well as water-stressed pots were harvested and the initial biomass was recorded. The soil in pots was covered with a polypropylene cover to avoid evaporation. An approximate 400 g of filler, plastic beads were spread over to reduce soil temperature, leaving transpiration as the only way to lose water from the soil. The pots were weighed and recorded the initial weights. Later on, daily wise, the pots were weighed at 8:00 AM IST. The weight reduction in successive days was considered as an amount of water lost via transpiration, and it was replenished to maintain the initial weight throughout the experiment period for well-watered (WW) pots, whereas water-stressed (WS) pots were made to mimic progressive soil drying conditions by gradually reducing the amount of water needed to replenish.

Physiological traits measured under pot and field conditions

Data pertaining to SPAD chlorophyll metre, Infra-Red (IR) Camera, Infra Red Gas Analyzer (IRGA) were measured at both vegetative and flowering stage of respective genotypes, whereas RWC was measured only at flowering stage. SPAD-502 (Soil Plant Analytical Development; Konica, Ramsey, NJ, USA) metre was used for measuring the relative chlorophyll content of leaves. A third/fifth leaf from top of the plant (overall five samples) per plant were selected to record SPAD chlorophyll metre readings. The five readings from five individual leaves were pooled to get an average value, five plant data were recorded from each genotype for each replication. The gas-exchange measurements such as net photosynthesis, transpiration rate and stomatal conductance were recorded by using IRGA. Index leaf from tagged plants were used to record gas exchange parameters (IRGA; Model: LICOR 6400). The images were captured at vegetative and flowering stage, plant height corresponding to flowering stage was mentioned in Table S2. The thermal images of the plant canopies were captured using an infrared camera, the IR FLEXCAM (Infrared Solutions, Inc., Plymouth, MN, USA) with a sensor size of 160 × 120 pixels, sensitivity of 0.09 °C, accuracy of 2% and emissivity (ε) of 0.95. The images were taken from the north to prevent shading of the target region, which was approximately 30 cm × 20 cm at one of the middle rows of each plot (Kashiwagi et al., 2008). After removing the emissions from the soil (background), the image processing and estimation of canopy temperatures were done using the software SmartView 2.1.0.10 (Fluke Thermography, Everett, WA, USA) (Zaman-Allah, Jenkinson & Vadez, 2011a; Sivasakthi et al., 2017). The observations were with the camera attached to a shoulder at a. approximate height of 1.0 m. At the crop canopy level, a temperature and relative humidity recorder (Gemini Tiny Tag Ultra 2 TGU-4500 data logger, Chichester, UK) was used to measure the ambient temperature. The gas-exchange traits, SPAD chlorophyll meter readings and canopy temperature were captured during the early hours of 9:00 to 11:00 AM. CTD was calculated as the difference between the ambient and leaf temperature, i.e., CTD = Tambient – Tcanopy. Figure 1, provides insights into image capture and process after removal of background noisy data for genotypes A1 and GMU 2644 respectively. Relative leaf water content was measured based on the method described by Turner (1981).

Experiment 2: Evaluation of safflower genotypes for CTD and yield traits in field under residual moisture conditions

The same set of 12 genotypes were further evaluated under field conditions for two years by growing them under residual moisture conditions using the recommended packages of practices given by ICAR-IIOR (Alivelu et al., 2014). Residual moisture conditions impart moderate to severe stress scenarios during the later stages of crop growth. Data pertaining to soil moisture levels, generated by gravimetric method revealed a reduction of 30.43 per cent from 10 DAS (vegetative) to 105 DAS (seed filling) at 15 to 45 cm depth, whereas it was 38 per cent at 45 to 60 cm. During both the seasons, the crops were sown by following a spacing of 45 × 20 cm, where only one sprinkler irrigation was given at sowing time, accounting for around 30 mm of water received. The experiments were conducted at ICAR-IIOR Research farm at ICRISAT, Patancheru (17.51°N, 78.27°E and 545 m above MSL). Information pertaining to prevailing climatic conditions in the experimental plot during experimentation were given in Fig. S1. The genotypes were raised in black soil (order: Vertisol) following randomized block design in an average plot size of 20 sq m accounting for an approximate 150 healthy plants. Observations of CTD, SPAD chlorophyll meter and gas exchange parameters were recorded similarly to the pot experiment at vegetative and reproductive stages in three replications. Morphological traits such as total biomass, leaf area index and yield traits were recorded at the time of harvest. Plant height was measured during flowering stage from tip to base with the help of 1 m scale from five tagged plants in each replication avoiding border rows. As for total dry matter the plants samples were harvested from base and samples were sundried for two days later oven dried at 72 °C till constant weight was attained. Leaf area index was measured from individual plots in replication with an AccuPAR LP-80 Ceptometer. Seeds from all the capitulum were pooled and calculated the total yield per plant. Harvest index (HI) = Economic yield/Biological yield was calculated at the end of experiments and the data recorded are given in Table S3.

Statistical analysis

The experiments were carried out at two factors, genotype and progressive soil drying. In both the experiments, the data generated from both the years at respective vegetative, flowering and maturity stages were pooled before statistical analysis. To obtain replication mean, data sets generated for various traits were averaged. The mean data thus generated for various traits were compared at 1 per cent significance level using the least significance difference test. Further the mean data sets were analysed using appropriate statistical tools viz., FTSW-NTR threshold graphs using non-linear regression analysis, ANOVA was conducted for traits recorded from pot experiments i.e., TE (WW), TE (WS) and TDM (WS) to understand the existing variance between and within the genotypes under WW and WS conditions, comparison of TE (WW) and TE (WS) means using Tukey’s honestly significant difference test, trait associations among various traits measured under pot and field experiments, Pearson correlation analysis to understand the linear relations among the variables and PCA to reduce the size of the data sets generated and create a smaller set of new variables (principal components).

FTSW-NTR plots were generated for each genotype using individual replicated data on daily basis. GraphPad Prism 2.0 version 6 (GraphPad Software Inc., San Diego, CA, USA) was employed for construction of graphs and non-linear regression analysis to be in line with the exponential model given by Muchow & Sinclair (1991). R square values which quantify the goodness were generated using GraphPad Prism 2.0 version 6, which make use of sum of the squares of the distances of the points from the best-fit curve determined by nonlinear regression. The relationship among various morpho-physiological and yield traits were studied using linear regression analysis. Pearson correlation table was constructed to understand strong correlation and consistency among the traits using “corrplot” packages.

PCA for phenotypic traits recorded under residual moisture conditions was done using “FactomineR” (Husson, Josse & Pages, 2010), “factoextra” and “ggfortify” packages (R Development Core Team, 2008) under R software version 4.1.2 (R studio). Eighteen traits were selected for PCA analysis, involving calculating the covariance matrix, eigenvectors and eigenvalues of the covariance matrix. Principal components whose eigen values more than one were selected and PCA scores generated. The number of principal components to be retained in the analysis is often determined based on the amount of variance explained, with a threshold of 80% or greater commonly used. The biplot was generated by using the ‘FactoMineR’ (factor analysis and data mining) with function ‘biplot’.

Experiment I: threshold levels of genotypes for NTR at FTSW

Significant differences were noticed from pot experiments at genotype and treatment level for FTSW threshold values. Threshold values are the soil moisture points beyond which normal transpiration of the plant gets affected. Genotypes exhibited wide variation for FTSW – NTR threshold values. At a FTSW threshold of 0.79 (R² = 0.99), genotype A-1 exhibited an early decline in NTR, followed by Bhima (0.74, R² = 0.98), CO-1 (0.71, R² = 0.91), GMU-2347 (0.71, R² = 0.92), and EC-523368-2 (0.70, R² = 0.99). Stress scoring that was done before the termination of experiments revealed a slow wilting followed by delayed senescence in genotypes CO-1, EC-523368-2, and NARI 6, which was attributed to their high soil moisture threshold. Furthermore, the adaptation governed by these genotypes was shown in terms of increased transpiration efficiency (TE) under stress, whereas genotypes (GMU-2644 (0.384, R² = 0.93), followed by GMU-2648 (0.46, R² = 0.90) and GMU 3438 (0.48, R² = 0.95)) that recorded the lowest threshold value were often prone to moisture stress during later stages of growth (Fig. 2). There also existed a positive correlation between the FTSW threshold values and the TE of the genotypes under stress. FTSW as independent variable plotted against TE a dependent variable using linear regression analysis represented in following Eq. (1) (Fig. 3).

(1) $$\mathrm{y}=0.222\mathrm{x}+0.1327(\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{d}{\mathrm{R}}^2\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}0.85)$$

Transpiration efficiency (TE) among the genotypes

TE varied significantly among genotypes under WW and WS conditions. Genotype CO-1 (4.88 g kg⁻¹) recorded a high TE under WW, whereas GMU-2644 (2.56 g kg⁻¹) recorded the least, 1.8 folds less than the highest value. A substantial reduction in mean TE was observed under WS conditions; genotypes GMU-2347 recorded the highest TE (2.72 g kg⁻¹), which was on par with A-1 (2.66 g kg⁻¹), EC-523368-2 (2.64 g kg⁻¹) and CO-1 (2.496 g kg⁻¹) with 0.23 CD. Similarly, the least TE was noticed on GMU-2644 (1.29 g kg⁻¹), followed by GMU-2648 (1.65 g kg⁻¹), GMU-3438 (1.71 g kg⁻¹) and GMU-3266 (1.91 g kg⁻¹) (Table 1). There was a significant positive relationship found in the amount of dry matter generated (above ground mass) and TE under WS conditions given in Eq. (2) (Fig. 3). ANOVA to understand the variance between and within the genotypes was given in Table S4.

(2) $$\mathrm{Y}=0.2851\mathrm{x}-0.7115(\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{d}{\mathrm{R}}^2\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}0.698)$$

Trait association

Among the various traits recorded, results from linear regression analysis revealed that seed yield under residual soil moisture was significantly associated with harvest index (R² = 0.39, p < 0.01) and weight of primary capitulum (R² = 0.19, p < 0.01). Similarly, TE under WS was significantly associated with, total dry matter under stress (R² = 0.69, p < 0.01), SPAD-chlorophyll meter reading (R² = 0.26, p < 0.01), canopy temperature depression (vegetative stage) (R² = 0.27, p < 0.01), canopy temperature depression (flowering) (R² = 0.67, p < 0.01), leaf area index (R² = 0.39, p < 0.01), assimilation rate (R² = 0.15, p < 0.05) and relative water content (R² = 0.67, p < 0.01) (Table 2).

Pearson correlation analysis

From both the experiments, the traits that were showing significant association with seed yield, were further subjected to correlation analysis and the results are given in Fig. 4. There was a positive correlation between transpiration efficiency under water stressed condition measured from pot experiment with, TDM under water stress (R = 0.83), CTD at vegetative stage (R = 0.54), SCMR values at flowering stage (R = 0.50), leaf area index (R = 0.65) and relative water content (R = 0.64), whereas a strong negative correlation was noticed from CTD at flowering stage (R = 0.69) under field conditions. Among various traits measured seed yield under field experiments exhibited a weak positive correlation TE under water stress (R = 0.15), number (R = 0.43) as well as weight of primary capitulum (R = 0.43) and strong positive correlation with harvest index (R = 0.61).

Experiment II: genotypic differences in morpho-physiological and yield traits

From field experiments generated data, a significant difference was noticed among the genotypes for various morpho-physiological traits recorded under residual moisture conditions. NARI-6 (91.1 ± 2.1 cm) recorded the highest plant height on par with CO-1 (87.4 ± 4.0 cm), which is 1.6 folds higher than EC-523368-2 with least plant height (56.4 ± 2.137 cm). The leaf area index (LAI), trait that is proportional to transpiration and was measured at 50% flowering of genotypes. ISF-764 (1.13 ± 0.05), Bhima (1.1 ± 0.04) and GMU-2347 (1.1 ± 0.02) recorded the highest LAI under field conditions and were on par, followed by EC-523368-2 (1.02 ± 0.02) and A-1 (1.02 ± 0.02), whereas GMU-3438 (0.69 ± 0.03) and GMU-2648 (0.737 ± 0.02) recorded the least. In addition, genotypes that recorded the highest TE in pot experiments also recorded higher SCMR values in the field. The genotypes that differed significantly for SCMR values (p < 0.01) were PBNS-12 (55.45 ± 2.56), EC-523368-2 (52.37 ± 1.51), GMU-3266 (37.92 ± 1.31), and GMU-2644 (45.12 ± 1.04). Supporting the observations made from the pot experiment, genotypes A-1 (92.3 ± 2.13%), EC-523368-2 (87.3 ± 2.52%), Bhima (86.5 ± 2.99%) and GMU-2347 (84.7 ± 1.46%) had a relatively higher RWC, whereas genotype GMU-2648 (71.3 ± 2.05%) recorded a relatively lesser RWC, which was on par with GMU-3438 (72.4 ± 2.92%) and GMU-3266 (73.7 ± 2.55%) (Tables S2 and S3).

Canopy temperature and gas exchange parameters

Exception to previous results, EC-523368-2 measured high transpiration rate (3.44 ± 0.09 m mol H₂O m⁻² s⁻¹), high stomatal conductance (109 ± 3.147 m mol H₂O m⁻² s⁻¹) on par with ISF-764 (108 ± 4.98 m mol H₂O m⁻² s⁻¹), whereas net photosynthetic rate was higher in A-1 (29.5 ± 0.681µ CO₂ m⁻² s⁻²) on par with NARI 6 (27.9 ± 0.644 µ CO₂ m⁻² s⁻²) and at least by GMU-3266 (16.9 ± 0.58 (µ mol CO₂ m⁻² s⁻¹)). Bhima recorded the highest leaf temperature (34.98 ± 0.66 °C), whereas the least was recorded from A1 and NARI 6 (29.76 ± 0.41 °C). Net photosynthetic rate recorded under field conditions is positively correlated with TE (WS) from pot conditions (R² = 0.15, p < 0.05), whereas leaf temperature is related to seed yield under residual moisture (R² = 0.19, p < 0.01) (Table S3).

Seed yield and it’s components

The genotypes performed extremely well under well water conditions (Manikanta et al., 2023), whereas under residual soil moisture (RSM) conditions a reduction in yield ranging from 22 per cent to 42 per cent was noticed emphasizing the intensity of residual soil moisture stress the crop had undergone. Though genotypes NARI-6 and EC-523368-2 recorded high TE (correlated via SCMR values) and delayed senescence in field, both the genotypes recorded the least seed yield per plant. NARI 6 (8.50 ± 0.246 g) and EC-523368-2 (8.82 ± 0.204 g), whereas genotypes ISF-764 (21.694 ± 0.50 g) and A-1 (18.64 ± 0.7 g) recorded the highest seed yield under residual moisture stress (p < 0.01). Days to 50 per cent flowering showed no significant variation among the genotypes (p > 0.05). The Harvest index (HI) significantly varied (p < 0.01) among the genotypes, with the highest in ISF-764 (0.369 ± 0.0005), which was on par with A-1 (0.366 ± 0.00031) (Table S2).

Analysis of principal components

Under residual soil moisture conditions, PCA resulted in four independent components whose SD values were more than 1, accounting for a cumulative variance of 88 percent and with SD values greater than 1.13. The maximum variation in PC1 was accounted for by FTSW-NTR threshold values (0.42), TE under WS (0.37), and RWC under field (0.36), while the least variation was accounted for days to 50% flowering (−0.15). In PC2, TE under WS (0.33), days to 50% flowering (0.32), and harvest index (−0.38) contributed the most variation, with capitulum weight under field (−0.43), total dry matter under field (−0.38), and seed yield under field (−0.38) contributing the least. Findings from biplot revealed a closer association, i.e., a smaller angle between the vectors between seed yield and yield attributing traits in safflower that included plant height and total dry matter, whereas physiological traits such as RWC, CTD at vegetative stage, FTSW-NTR thresholds, and morphological traits such as leaf area index were in close relationship with TE under water stress in pot conditions (Table 3 and Fig. 5).