This study was carried out at the farm station of Azad Islamic University, Karaj, Iran to study the effect of water deficit stress and sowing date on the quantitative and qualitative yield of Roselle (Hibiscus sabdariffa L.) during the two consecutive growing seasons of 2017 and 2018. This experiment included nine treatments which were included interactions among three irrigation regimes (well-watered, mild and drought stress) and three planting dates (4^th June, 18^th June and 2^nd July). The climate in this province is semi-dry with an annual rain fall of about 250 mm which mostly occurs in autumn and winter between November and April. Average minimum and maximum temperatures and precipitation in both years of the present study are shown in Figure 1. The soil type of the field was sandy-loam and low in organic matter (0.28%) with total N (0.031%) as measured by the Kjeldahl method (Beljkas et al., 2010), available P (35 mg kg^-1), using the Olsen procedure (Olsen and Sommers, (1982) and available K (225 mg kg^-1) after extraction with ammonium acetate, with pH and EC of 6.96 and 0.82 dS m^-1, respectively.

The nine treatments were employed in split plot with three replicates in a randomized complete block design (RCBD). Irrigation regimes were distributed among the main plots and included three irrigation levels ( 30, 45 and 60% depletion of available water, labelled control, mild and drought stress, respectively) (Keshavarz Mirzamohammadi et al., 2021). Planting date treatments were arranged in subplots. Roselle (Hibiscus sabdariffa L.) seeds were directly sown in prepared plots (divided into four rows at 60 cm apart and 5 cm between the plants on each row) on 4^th June, 18^th June and 2^nd July in both years. After germination, the seedlings were thinned to reach optimum density. The seedlings were irrigated when needed until they completely developed. Soil moisture content in each plot was monitored daily using a time domain Reflectometry (TDR). Irrigation water needed prior to irrigation (V_w in m³) was estimated based on the soil water content (θ _i ) (by TDR) and effective rooting depth (D, 0.3m here) according to the following equation:

Where, θ _F.C is the volume of soil moisture at field capacity and A is the main plot area (m²). Data on volumetric water content were collected daily prior to setting the experiment calibrates before seed sowing and during the growing season to calculate the time of irrigation. The amount of water irrigation applied was according to the soil deficit to bring it back to field capacity. All plots were fertilized uniformly, during soil preparation in the spring of each year.

The leaves and calyxes were harvested within four weeks after flowering. The fresh calyxes of each plot were washed with water, air-dried at room temperature, and weighed. The seeds were removed manually, air dried at room temperature and the whole calyces were used for the analysis of chemical properties. The extraction of dried calyxes with water at a ratio of 1:10 was conducted in a water bath (50 °C, 30 min) (Chumsri et al., 2008). The total phenolic and anthocyanin contents of calyx extract were assessed according to the Folin-Ciocalteu reagent method (Temraz and El-Tantawy, 2008) and Krizek et al. (1993), respectively. Vitamin C was measured spectrophotometrically (Hitachi, U-1800, Tokyo, Japan) by measuring Fe²⁺ complexes with 2, 2-dipyridyl (Sarker and Oba, 2018) and reading the absorbance of the sample solution at 525 nm. Antioxidant activity was measured according to Sarker and Oba (2018) by means of the diphenyl-picrylhydrazyl (DPPH) radical degradation method.

Harvesting was done at the physiological maturity stage on the first of October until the first of November (in both years) by harvesting the four middle rows. The traits examined included plant height, stem diameter, biological yield and calyx yield. Biological yield was determined after oven drying at 75 °C for 72 hours to constant weight. Then the dried biological part of plant was weighed by a digital 0.01-precision scale. Stem diameters and plant height were measured manually from the ground level with a caliper and a barcoded height stick ( 1 mm resolution), respectively. Therefore, they were considered to be precise and accurate. For this purpose, five individual plants from each plot were measured, and the averaged stem diameter and plant height were used as the plot-level. Seed oil content was measured by Inframatic 8620 Percor, England. Calyx water use efficiency (WUE_calyx) and calyx harvest index (HI_calyx) were calculated by dividing the calyx yields by the volume of applied water and biological yield, respectively.

Analyses of variance (ANOVA) were performed using the SAS software (SAS Institute Inc. ver. 9.2). Two years’ data were analyzed by combined years because the Bartlett test was not significant for all traits measured. Differences among mean values for the roselle plant responses to the level of irrigation regimes and planting date were analyzed with the least significant difference (LSD) test at a significance of α ≤ 0.05. Principal Component Analysis (PCA) based on biplot (SAS 9.1) and coefficient of correlation were applied to consider the visualization of similarities or differences and interrelationships by acute and obtuse angles among all parameters. Clustering analysis (S-PLUS ver. 6.1 software, Insightful Corporation, USA) aimed at classifying objects based on the minimum variance linking method and similarity of input data for the existing parameters using Ward’s hierarchical approach and Euclidean distance to organize data into groups.

The data related to plant height (Tables 2 and 3), the well-watered irrigation (181.2 cm) and first sowing date (4^th June) (168.7 cm) showed maximum value for plant height. The well-watered plants measured significantly higher stem diameters (2.85 cm) than the mild and severe drought stress by 14.03 and 28.7%, respectively. Our results are in agreement with those found by Karami et al. (2016) for soybean cultivars [Glycine max (L.) Merr.], which indicated that drought stress reduced plant growth, plant height and stem diameter due to disturbance in photosynthesis and low carbohydrate production. In addition, prolonged growth periods allowed the crops to use growth resources like light, water and nutrients which finally increased the growth of the crops. Similar findings with respect to drought stress and sowing date have been reported by Apeyuan et al. (2017) and Rah Khosravani et al. (2017) for roselle (Hibiscus sabdariffa L.) and maize (Zea mays L.) hybrids, respectively.

A decrease of 14.69 ton.ha^-1 to 11.40 ton.ha^-1 in biological yield of roselle was recorded due to decreasing the irrigation supply from control to severe drought stress. However, there were no significant difference between well-watered and mild drought stress treatments (Table 2). This may be related as stress conditions reduce the biological yield production because of limitations in the water and essential fertilizer uptake by the plants (Seghatoleslami et al., 2013). Reductions in total dry matter of rice (Oryza sativa L.) under drought conditions seemed to be attributed to the reduction in water availability, which reduced cell division, lowered LAI and plant height and finally resulted in lower dry matter and biological yield (Aghdasi et al., 2018a). The authors stated that an increased allocation of photosynthates to the plant root relative to the shoots was another reason for the reduction in biological yield. Regarding sowing dates, a significant maximum biological yield (14.24 ton.ha^-1) was achieved when the plants were sown on 4^th June with the minimum biological yield (13.18 ton.ha^-1) in the case of the crop sown on 2^nd July. However, there were no significant difference between 18^th June and 2^nd July. These results are in accordance with those of Keshavarz and Khodabin (2019) and Aghdasi et al. (2018b). Higher biological yield with early sowing was mainly due to a greater number of effective branches per m², more plant height and greater stem diameter (some data not show). Another reason for this result is probably due to the fact that the first sowing dates resulted in longer growing periods, and therefore, greater total rainfall received during the growing periods. An adequate irrigation supply would eventually yield the higher growth rate. These results are in line with Rah Khosravani et al. (2017). There was a significant positive correlation between calyx yield, total phenolic content, vitamin C and seed oil yield with biological yield. However, biological yield had a negative correlation with total anthocyanin content and antioxidant activity (Table 4). In a similar study on roselle, Fallahi et al. (2017b) concluded that biological yield had a positive effect on the number of capsules per plant and plant height which may be attributed to genetic effects rather than environmental ones and these traits may be good selections for improving seed and calyx yield in roselle.

An increase from 1.37 ton.ha^-1 to 1.52 ton.ha^-1 in calyx yield was recorded for roselle with an increase drought severity from well-watered to mild drought stress but severe drought stress significantly decreased calyx yield to 1.16 ton.ha^-1 (Table 2). It was demonstrated that well-watered irrigation is needed to guarantee high crop yield but in the case of medicinal plants a mild drought stress could be beneficial to the quality and quantity of these plants (keshavarz et al., 2018). Increases in roselle calyx may be attributed to the mild water stress imposed. This was in agreement with the reports of Fallahi et al. (2017a), who claimed that the calyx yield of roselle increased under mild drought stress compared to full irrigation. El-Dissoky et al. (2020) studied the effect of irrigation frequency on the calyx yield of the roselle plant and found that mild drought stress improved the calyx production but severe drought condition reduced the calyx yield. Stress conditions have been shown to decrease the number of flowers and calyx (depending on the time of the stress severity), because the flowering phase involves several processes that are vulnerable to stress conditions (Jasim et al., 2020). From Table 3, it is clear that first sowing dates (4^th June) registered maximum calyx yield (1.48 ton.ha^-1) while the lowest calyx yield (1.24 ton.ha^-1) was recorded for the third sowing date (2^nd July). Given the favorable growth conditions such as temperature and the sun light on 4^th June, plants produced more assimilates and yielded greater dry matter. Due to a shortened vegetative phase, flowering occurred when summer temperatures were high and the flowers aborted. Previous research has reported that optimal planting time leads to better developed plants with higher LAI than those sown later (Rah Khosravani et al., 2017). Accordingly, flowering abortion would be expected to increase during drought stress as it decreases the flux of photosynthate supply from source leaves to the vegetative tissues. Also, drought stress may change the concentration of ABA in the plants, and thereby induce flower abortion in drought-stressed crops (Salih, 2019).

The highest TPC (2.77 mg.100 g^-1 DW) was recorded for plants treated with mild drought stress, while the lowest TPC (1.94 mg.100 g^-1 DW) was recorded for severe drought stress treatment. Although, mild drought stress increased TPC, there was no significant difference between well-watered and mild drought stress treatments (Table 2). However, a greater increase in drought severity, to 55% FC, lowered the TPC of roselle. TPC, as a group of non-enzymatic antioxidant compounds with carbon skeleton, by osmotic regulation of plant cells serves to respond to drought conditions, stabilize proteins, prevent lipid peroxidation, and act as an osmolyte in the permeability of cell membranes (Keshavarz Mirzamohammadi et al., 2021). Under severe drought conditions, TPC decreases, which can be because of an increase in the production of oxygen radicals, peroxidation of chlorophyll pigments, degradation of genes related to the chlorophyll biosynthetic path, reduction in the photosynthetic level and increasing the allocation of carbon to the root.

The effect of irrigation treatment was significant for the TAC of roselle (Table 1). On average, TAC increased by 19.68 and 24.69% compared to the well-watered treatment when the plants were treated with mild and severe drought stress, respectively (Table 2). However, mild and severe stress were statistically at the same level. As documented in many papers (An et al., 2020; Cirillo et al., 2021), the total anthocyanin content of roselle was increased by drought stress to reduce excess light availability and provide useful protection to leaves without significantly reducing photosynthesis efficiency. This shows a mechanism to modulate light absorption and reduce photo-oxidative injury. It has been reported that anthocyanin has the potential to reduce photo-oxidative damage in leaves, both by scavenging reactive oxygen species and protecting chloroplasts from excess high-energy quanta (He et al., 2020). The influence of drought stress on TAC is in agreement with the findings of Fallahi et al. (2017a) and Hinojosa-Gómez et al. (2020) on roselle, which stated that the improvement in the plants’ qualitative traits under simultaneous drought stress, such as anthocyanins, has been related to improved secondary metabolism.

Vitamin C content was affected significantly by irrigation levels (Table 1) and the amount of these compounds under mild drought stress (395.7 mg 100 g^-1) was higher than well-watered irrigation (13% on average). In contrast, severe drought stress sharply decreased the vitamin C content by 21 and 32% compared to the well-watered and mild stress, respectively. Changes in vitamin C under drought stress have been shown to depend on stress severity, drought duration and cultivar (Dolatabadian et al., 2010). Mataa et al. (2020) showed that the concentrations of vitamin C, TPC, and TAC showed similar trends and were reduced in severe drought stress. Therefore, it is possible that in this study, the concentration of other antioxidants, such as superoxide dismutase and catalase (not measured in this study), can be increased in roselle under drought stress.

In this investigation, antioxidant activity (DPPH) was significantly increased with the increase in the severity of drought stress in the order: control < mild < severe drought stress (Table 2). In general, with delay in planting date, the percentage of antioxidant activity in the plant increased and 4^th June exhibited the lowest (71.57%) antioxidant activity for septal extract, whilst the highest antioxidant activities were observed on 18^th June and 2^nd July (78.62 and 78.20, respectively) (Table 3). Free radicals are generated under stress conditions and the antioxidant capacity of plants includes enzymatic and non-enzymatic compounds, free radical species are oxidized by releasing electrons to free radicals (Keshavarz, 2020). A significant decrease was observed for total anthocyanin content (y = -0.76**) and antioxidant activity (y = -0.79**) with the increase in plant height (Table 4). This is due to the difference in red (R): far red (FR) ratio perceived by the plants grown at different irrigation regimes, planting time and neighbor detection (Khattak et al., 2016). The plants develop a strategy to grow taller in order to capture more photosynthetic radiation at the expense of physiological changes (Smith, 1982).

On average, the seed oil content (% of seed dry matter) of the plants under mild and severe drought treatment was significantly low (12.21 and 11.27%, respectively) compared to 16.88% in the seeds of the well-watered plants (Table 2). Delay in planting time significantly decreased the seed oil content in fruits compared to on-time sowing dates, where the highest values for seed oil content were recorded on 4^th June (14.14%) with an average of 6.15 and 8.41% higher than 18^th June and 2^nd July (Table 3). As previously explained, water stress conditions affected plant growth and development, leading to an effect on seed oil content. The optimal water supply throughout the growing period could enhance oil content in seed oil plants, while drought stress reduces it (Soheili-Movahhed et al., 2019). A reduction in seed oil content might be attributed to the polyunsaturated fatty acid oxidation by free radicals under drought conditions (Keshavarz, 2020). The reduction in seed oil content when drought stress occurred during the growing season is supported by various reports (Keshavarz et al., 2020). The quantity of seed oil content is mainly determined by water balance, soil fertilizers, sun light and other environmental factors throughout the growth period. The high temperature and consequently changes in water balance in the seed filling stage could likely be the reason for lower oil content on 2^nd July compared to 4^th June. Asadpour et al. (2020) and Yazdanpanah et al. (2018) reported that high seed oil content in roselle (Hibiscus sabdariffa L.) was related to lower temperature and water availability at the seed filling phase.

The highest water use efficiency for calyx (2.02 kg.kg^-1) was obtained with severe drought treatment and the lowest (1.79 kg.kg^-1) was with the well-watered treatment (Table 2). The results showed that water use efficiency increased when the sowing date was delayed to after June 4^th. Based on our results (Table 3) the highest use efficiency was observed on 2^nd July and it was higher by 15.42 and 7.94% than the first and second planting dates, respectively. However, there were no significant difference between the first and second planting dates. Based on harvested calyx yield and the amount of irrigation, severe drought stress was determined to be the best irrigation regime in terms of the WUE_calyx. As previously mentioned, the biological yield of roselle was decreased by drought stress. In our study, since the average amount of irrigation water was reduced among the drought stress or delayed planting treatments, higher calyx yield compared to biological yield led to the greater WUE_calyx.

A limited supply of carbohydrates for dry matter production leads to a reduction in biological yield. Previous research has reported that higher values for HI_calyx in roselle (Hibiscus sabdariffa L.) under drought conditions might be due to higher air temperature and heat stress, which could negatively affect biological yield and plant growth, resulting in earlier flowering due to shortened vegetative stages and increased economical yield/biological yield ratio (Javadzadeh, 2018). In fact, drought conditions reduced biological yield more than calyx yield. As shown in Table 2, calyx yield is higher under mild drought conditions. It was thought that the increase in calyx under mild drought conditions resulted from the increase in assimilate partitioning to flower due to a shortened vegetative phase. On the hypothesis that short distance translocation of is limited to first/second node (Liu et al., 2020), the improvement in HI_calyx means the increase in sink demand to source. So, we can conclude that the increase in calyx yields improved HI_calyx by increasing sink demand, therefore, the use efficiency of assimilates was increased under drought conditions.