The effect of alternate wetting and drying (AWD) on spikelet fertility, grain traits and the biosynthesis of 2-acetyl-1-pyrroline (2-AP) in rice grown under high temperature Authors: Nnaemeka Emmanuel Okpala, Bin Du, Daniel Bimpong, Mawuli Korsi Amenyogbe, Kelvin Dodzi Aloryi, Yangxuan Liu, Peng Yan, Longzhou Chen, Wenyan Liu, Guoli Wang, Xiaohai Tian, Tianyue An Published: 2025 ABSTRACT BackgroundHigh temperature attributed to climate change is a major obstacle to rice production. This study examined how alternate wetting and drying (AWD) irrigation can mitigate the effects of high temperatures on rice production. Three indica rice cultivars-Ye Xiang You You Si (YXYYS), Jing Zhan 1 (JZ1), and Qing Xiang You Yi 19 (QXYY19)-with varying heat susceptibility levels were used to analyse the impact of AWD on spikelet fertility, grain traits, and the biosynthesis of 2-acetyl-1-pyrroline (2-AP) from heading to harvest in rice grown under high temperature.ResultsThe findings revealed that AWD significantly reduced spikelet fertility in all cultivars, but its effect on spikelet fertility depends on the heat susceptibility level of each cultivar. AWD alleviated the negative effect of high temperature on the biosynthesis of 2-AP and increased 2-AP content in leaves and grains of rice. This study also provided new insights on the effect of starch granules and brown rice grain width on chalkiness and demonstrated that starch granule arrangement and airspaces-not starch granule size-as well as brown rice grain width influenced chalkiness in rice. It also revealed that rice husk affected grain shape. Cooked rice elongation and rice whiteness were significantly increased by AWD, but it reduced the seed germination rate.ConclusionsThis study presents a roadmap for the improvement of spikelet fertility in rice grown under high temperature. It also presents a simplified and cost-effective means to further increase the biosynthesis of 2-AP, proffered solutions for the reduction of chalkiness, and proposed methods for the improvement of grain shape by targeting genes, quantitative trait loci (QTLs), and signalling pathways associated with those traits in rice. Background High temperature attributed to climate change is a major obstacle to rice production and possesses a great threat to food security. High temperature has been shown to negatively affect rice growth [1], rice pollen [2, 3], rice spikelet number [4], rice spikelet fertility [5] and rice grain quality [6]. Numerous studies have shown that the severity of high temperature depends on varietal differences in rice [5–7]. However, water scarcity presents a different kind of challenge to rice farmers and breeders. It has also been reported that rice production requires flooded paddy fields [8]. Lowland rice are usually grown under flooded conditions and consume about three times more water compared to other crops [9]. There are many factors that cause water scarcity. It has been reported that water scarcity can be caused by erratic rainfall, pollution, malfunctioning of irrigation systems, and increase in competition from other sectors and excessive ground water extraction [9]. Rice producers, farmers, and breeders are making efforts to devise irrigation methods that can reduce water consumption during rice production. One of the methods that rice farmers have employed to confront the issue of water scarcity during rice production is alternate wetting and drying (AWD) irrigation method. Lampayan et al. [9] stated that in AWD, the soil is allowed to dry out for one to several days, before it is flooded again. The number of days of non-flooded soil could range from 1 to 10 days [10]. The climatic conditions during drying determine the degree of AWD. For instance, water dries faster when temperature is high compared to low temperature conditions. Therefore, the impact of AWD could vary during early and late planting seasons of rice. Based on the drying duration and climatic conditions, AWD can be classified as moderate AWD or severe AWD. AWD has been shown to increase water productivity with respect to water input [11], which resulted in 17% to 38% increase in farmer’s income during rice production, with no other significant changes required in crop management [9]. However, Bouman and Tuong [11] reported that numerous field experiments conducted across Asia showed that 92% of AWD treatments during rice production resulted in decrease in rice yield, which ranged from negligible to as much as 70% reduction. Bao et al. [12] showed that moderate AWD increased rice yield and yield characters, while severe AWD resulted in significant decrease in rice yield. The large variability in rice yield as a result of AWD was due to differences in the duration of soil drying between irrigations and hydrological conditions [11]. Rice yield is important to rice breeders and producers because of the economic value of high yield varieties. One of the factors that affect yield in rice is the flag leaf area. It has also been reported that flag leaf is the most important organ for photosynthesis and provides at least 50% of photosynthetic products for grain [13, 14]. Flag leaf possesses the highest photosynthetic capacity and plays a major role in determining grain productivity [15]. Any impact on the photosynthesis of flag leaves affects grain productivity [16]. Spikelet fertility is an important yield component of rice. Xiang et al. [17] stated that successful double fertilisation is needed for spikelet fertility to occur. However, climatic conditions affect the spikelet fertility and grain quality of rice. Rice fragrance, cooked rice elongation and amylose content percentage are some of the traits used in determining rice grain quality [6, 18]. 2-acetyl-1-pyrroline has been reported as the major compound responsible for fragrance in rice [19]. Rice consumers are known to have a strong affinity for fragrant cultivars [19]. Fragrant rice types such as Basmati and Jasmine are not just popular in India, Pakistan and South East Asia where they are produced, but they are also popular in Europe, North America, and Australia [20]. However, Okpala et al. [19] noted that fragrant rice are also becoming increasingly more popular in West Africa and China. AWD has been successfully employed to confront the issue of water scarcity during rice production. It will be hugely beneficial to rice farmers and breeders, if AWD can also be used to mitigate the negative impact of high temperature on rice yield and grain quality. Therefore, the aim of this study was to investigate the impact of AWD-from the onset of heading to harvest-on spikelet fertility, grain traits and the biosynthesis of 2-AP in rice grown under high temperature. Investigations were conducted with three heat susceptible rice cultivars: Ye Xiang You You Si (YXYYS), Jing Zhan 1 (JZ1) and Qing Xiang You Yi 19 (QXYY19). Results Climatic conditions from the onset of heading to harvest Heading of JZ1 and YXYYS grown under DI (control) and AWD (treatment) began on 3rd August 2022, while harvest took place on 8th September 2022. The average daily minimum and daily maximum temperatures from the onset of heading to harvest were 26℃ and 35℃, respectively (Table 1). We recorded 9 h (h) of average daily sunshine during the entire period from heading to harvest. However, further analysis of the climatic data revealed 4 days with no sunshine, 2 days with less than an hour of sunshine, and 21 days with over 11 h of sunlight. The average daily relative humidity and wind speed were 65% RH and 2.5 m/s, respectively. Heading began on 11th August 2022, in QXXY19, and harvest took place on 12th September, 2022. The average daily minimum and daily maximum temperatures from the onset of heading to harvest were 25℃ and 34℃, respectively. The average daily sunshine, relative humidity, and wind speed were 8 h, 63% RH, and 2.4 m/s, respectively.Table 1Climatic conditions from the onset of heading to harvest of JZ1 and YXYYS grown under DI and AWDDateDaily Minimum TemperatureDaily Maximum TemperatureDaily Average TemperatureDaily Sunshine HoursDaily Average Relative HumidityDaily 2 M Wind Speed03/08/202227.636.532.111.974.81.404/08/202228.435.732.411.8731.905/08/202228.235.832.211.9702.206/08/202228.435.131.51270.33.407/08/202228.737.432.211.966.8408/08/202228.837.633.211.563.33.709/08/202228.436.932.612662.910/08/202227.436.832.811.759.52.811/08/202227.9363211.6653.112/08/202228.537.432.811.9653.313/08/202229.136.832.511.863.53.514/08/202228.536.632.611.763.33.615/08/202228.83833.411.763.33.316/08/202229.737.732.57.8722.317/08/202228.437.832.49.173.51.718/08/202228.63933.511.667.82.719/08/202228.439.233.611.663220/08/202229.238.63411.560.51.721/08/20223037.133.411.562.33.522/08/202229.538.233.110.157.33.723/08/202228.935.9326.9643.724/08/202224.934.729.411.458225/08/202225.735.430.511.452.31.926/08/202223.733.529.36.672.31.927/08/202222.331.9260.584.54.428/08/202220.525.221.9029/08/202220.124.522077.52.830/08/202221.223.222.1072.83.431/08/202219.226.722.70.2711.501/09/202221.126.323.4070.32.102/09/202221.731.325.33.365.81.503/09/202219.931.225.39.857.52.404/09/202221.632.826.310.452.5205/09/202221.635.627.311.556.3106/09/202220.837.327.811.558.31.407/09/202222.736.628.76.953.31.408/09/202225.535.229.24.451.51.3 Table 1Climatic conditions from the onset of heading to harvest of JZ1 and YXYYS grown under DI and AWDDateDaily Minimum TemperatureDaily Maximum TemperatureDaily Average TemperatureDaily Sunshine HoursDaily Average Relative HumidityDaily 2 M Wind Speed03/08/202227.636.532.111.974.81.404/08/202228.435.732.411.8731.905/08/202228.235.832.211.9702.206/08/202228.435.131.51270.33.407/08/202228.737.432.211.966.8408/08/202228.837.633.211.563.33.709/08/202228.436.932.612662.910/08/202227.436.832.811.759.52.811/08/202227.9363211.6653.112/08/202228.537.432.811.9653.313/08/202229.136.832.511.863.53.514/08/202228.536.632.611.763.33.615/08/202228.83833.411.763.33.316/08/202229.737.732.57.8722.317/08/202228.437.832.49.173.51.718/08/202228.63933.511.667.82.719/08/202228.439.233.611.663220/08/202229.238.63411.560.51.721/08/20223037.133.411.562.33.522/08/202229.538.233.110.157.33.723/08/202228.935.9326.9643.724/08/202224.934.729.411.458225/08/202225.735.430.511.452.31.926/08/202223.733.529.36.672.31.927/08/202222.331.9260.584.54.428/08/202220.525.221.9029/08/202220.124.522077.52.830/08/202221.223.222.1072.83.431/08/202219.226.722.70.2711.501/09/202221.126.323.4070.32.102/09/202221.731.325.33.365.81.503/09/202219.931.225.39.857.52.404/09/202221.632.826.310.452.5205/09/202221.635.627.311.556.3106/09/202220.837.327.811.558.31.407/09/202222.736.628.76.953.31.408/09/202225.535.229.24.451.51.3 Climatic conditions from the onset of heading to harvest of JZ1 and YXYYS grown under DI and AWD Flag leaf area and spikelet fertility AWD from the onset of heading to harvest significantly reduced flag leaf area in all three rice cultivars compared to DI (Fig. 1a). The flag leaf area of JZ1 grown under DI and AWD was 43 cm2 and 38 cm2, respectively; while the flag leaf area of YXYYS grown under DI and AWD were 40 cm2 and 22 cm2, respectively. Further analysis of the results showed that AWD decreased flag leaf area in JZ1 and YYXYS by 12% and 45%, respectively. The flag leaf area of QXXY19 under DI was 43 cm2. However, we observed that the flag leaves of QXXY19 grown under AWD were completely withered. AWD significantly decreased spikelet fertility in all cultivars (Fig. 1b). The spikelet fertility of JZ1 grown under DI and AWD was 64% and 33%, respectively, while the spikelet fertility of YYXYS under DI and AWD was 65% and 8%, respectively. QXXY19 grown under DI had spikelet fertility of 9%, but we observed complete spikelet sterility in QXXY19 under AWD. Consequently, QXXY19 was not included in further analysis. These results indicate that while AWD conserves water, it can detrimentally affect heat-sensitive rice cultivars. Therefore, careful variety selection is essential for successful cultivation under AWD conditions.Fig. 1The impact of AWD on flag leaf area and spikelet fertility; (a) the effect of AWD on flag leaf area: (b) the effect of AWD on spikelet fertility. Differences in lower case letters indicate significant difference (Pvalue < 0.05) among the mean values The impact of AWD on flag leaf area and spikelet fertility; (a) the effect of AWD on flag leaf area: (b) the effect of AWD on spikelet fertility. Differences in lower case letters indicate significant difference (Pvalue < 0.05) among the mean values The biosynthesis of 2-AP and cooked rice elongation AWD significantly increased the biosynthesis of 2-AP in both JZ1 and YXYYS. Leaves of JZ1 grown under DI and AWD had 2-AP content of 56 μg/kg and 113 μg/kg, respectively, while the grains of JZ1 grown under DI and AWD had 2-AP content of 19 μg/kg and 30 μg/kg, respectively (Fig. 2a). The leaves of YXYYS grown under DI and AWD had 2-AP content of 98 μg/kg and 169 μg/kg, respectively, while the grains of YXYYS grown under DI and AWD had 2-AP content of 10 μg/kg and 16 μg/kg, respectively (Fig. 2b). Further analysis found that AWD increased the biosynthesis of 2-AP in the leaves and grains of JZ1 by 102% and 56%, respectively (Fig. 2c). Also, AWD increased the biosynthesis of 2-AP in the leaves and grains of YXYYS by 73% and 62%, respectively. The increased 2-AP levels demonstrate that AWD is a viable irrigation strategy for enhancing rice fragrance.Fig. 2The impact of AWD on the biosynthesis of 2-acetyl-1-pyrroline (2-AP) and cooked rice elongation: (a) the effect of AWD on the biosynthesis of 2-AP in the leaves and grains of JZ1 grown under heat stress; (b) the effect of AWD on the biosynthesis of 2-AP in the leaves and grains of YXYYS grown under heat stress; (c) Increase in 2-AP content in JZ1 and YXYYS as a result of AWD; (d) the impact of AWD on cooked rice elongation of JZ1 and YXYYS. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values The impact of AWD on the biosynthesis of 2-acetyl-1-pyrroline (2-AP) and cooked rice elongation: (a) the effect of AWD on the biosynthesis of 2-AP in the leaves and grains of JZ1 grown under heat stress; (b) the effect of AWD on the biosynthesis of 2-AP in the leaves and grains of YXYYS grown under heat stress; (c) Increase in 2-AP content in JZ1 and YXYYS as a result of AWD; (d) the impact of AWD on cooked rice elongation of JZ1 and YXYYS. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values The impact of AWD on cooked rice elongation on both cultivars was further investigated. The results found that the cooked rice elongation percentage of JZ1 grown under DI and AWD was 44% and 55%, respectively, while the cooked rice elongation percentage of YXYYS grown under DI and AWD was 51% and 53%, respectively (Fig. 2d). These findings suggest that AWD is not only water-saving irrigation method but but also a strategy to enhance key quality traits in rice, potentially improving both yield and profitability. Starch granule morphology Starch granules of JZ1 grains under DI were found to be flat and tightly packed with small airspaces (Fig. 3a), as compared to the starch granules of the grains grown under AWD, which appear somewhat oval, irregular, and loosely packed, with bigger airspaces (Fig. 3b). The starch granule morphology of YXYYS grown under DI and AWD, respectively, displayed flat and tightly packed starch granules with small airspaces (Fig. 3c, d). Further analysis showed that AWD significantly increased starch granule size and diameter in both cultivars (Fig. 4). The starch granule size of JZ1 under DI and AWD was 6.9 μm and 7.7 μm, respectively, while the starch granule diameter of JZ1 under DI and AWD was 6.7 μm and 7.5 μm, respectively. The starch granule size of YXYYS under DI and AWD was 7.9 μm and 8.2 μm, respectively, while the starch granule diameter of YXYYS under DI and AWD was 7.6 μm and 8.0 μm, respectively. The results suggest that AWD promotes granule expansion and structural looseness, which may differentially regulate starch biosynthesis pathways.Fig. 3The effect of AWD on starch granule morphology: (a) starch granule morphology of the grains of JZI grown under DI; (b) starch granule morphology of the grains of JZI grown under AWD; (c) starch granule morphology of the grains of YXYYS grown under DI; (d) starch granule morphology of the grains of YXYYS grown under AWDFig. 4The impact of AWD on the starch granule size and starch granule diameter: (a) effect of AWD on starch granule size; (b) effect of AWD on starch granule diameter. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values The effect of AWD on starch granule morphology: (a) starch granule morphology of the grains of JZI grown under DI; (b) starch granule morphology of the grains of JZI grown under AWD; (c) starch granule morphology of the grains of YXYYS grown under DI; (d) starch granule morphology of the grains of YXYYS grown under AWD The impact of AWD on the starch granule size and starch granule diameter: (a) effect of AWD on starch granule size; (b) effect of AWD on starch granule diameter. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values Impact of AWD on rice grain traits The grain length of JZ1 grown under DI and AWD was 7.4 mm and 6.4 mm, respectively, while the grain length of YXYYS under DI and AWD was 7 mm and 6.6 mm, respectively (Table 2). AWD reduced brown rice grain length in JZ1 and YXYYS by 15% and 6%, respectively. Statistical analysis showed that AWD resulted in a significant decrease in the grain length of both JZ1 and YXYYS. Also, the grain width of JZ1 grown under DI and AWD was 1.67 mm and 1.85 mm, respectively, while the grain width of YXYYS grown under DI and AWD was 1.72 mm and 1.66 mm, respectively (Table 2). Statistical analysis found that AWD resulted in a significant increase in grain width of JZ1. Conversely, it resulted in a significant decrease in the grain width of YXYYS. Further analysis showed that in JZ1, AWD increased brown rice width by 10%, but in YXYYS, we observed that AWD decreased grain width by 4%. The length-to-width ratio of JZ1 grown under DI and AWD was 4.42 mm and 3.46 mm, respectively, while the length-to-width ratio of YXYYS grown under DI and AWD was 4.08 mm and 3.96 mm, respectively (Table 2).Table 2Rice grains traits of JZ1 and YXYYS grown under DI and AWD. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean valuesRice cultivarsBrown rice length (mm)Brown rice width (mm)Brown rice length: width (mm)Brown rice weight (g)Paddy rice weight (g)Husk %Chalkiness (%)Milled rice whiteness (%)JZ1 (DI)7.39 ± 0.09a1.67 ± 0.023bc4.42 ± 0.10a0.32 ± 0.017a0.39 ± 0.02a18.23 ± 6.13a3.97 ± 2.84a37.63 ± 0.929bJZ1 (AWD)6.42 ± 0.21c1.85 ± 0.04a3.46 ± 0.18c0.32 ± 0.01a0.34 ± 0.01b7.31 ± 3.41d19.63 ± 12.70a39.30 ± 1.852bYXYYS (DI)7.02 ± 0.05b1.72 ± 0.04b4.08 ± 0.07b0.32 ± 0.01a0.36 ± 0.01ab11.38 ± 0.23c19.43 ± 7.94a42.37 ± 0.40aYXYYS (AWD)6.60 ± 0.13c1.66 ± 0.01c3.96 ± 0.05b0.28 ± 0.01b0.33 ± 0.03b12.82 ± 7.33b17.50 ± 8.18b44.33 ± 0.67a Table 2Rice grains traits of JZ1 and YXYYS grown under DI and AWD. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean valuesRice cultivarsBrown rice length (mm)Brown rice width (mm)Brown rice length: width (mm)Brown rice weight (g)Paddy rice weight (g)Husk %Chalkiness (%)Milled rice whiteness (%)JZ1 (DI)7.39 ± 0.09a1.67 ± 0.023bc4.42 ± 0.10a0.32 ± 0.017a0.39 ± 0.02a18.23 ± 6.13a3.97 ± 2.84a37.63 ± 0.929bJZ1 (AWD)6.42 ± 0.21c1.85 ± 0.04a3.46 ± 0.18c0.32 ± 0.01a0.34 ± 0.01b7.31 ± 3.41d19.63 ± 12.70a39.30 ± 1.852bYXYYS (DI)7.02 ± 0.05b1.72 ± 0.04b4.08 ± 0.07b0.32 ± 0.01a0.36 ± 0.01ab11.38 ± 0.23c19.43 ± 7.94a42.37 ± 0.40aYXYYS (AWD)6.60 ± 0.13c1.66 ± 0.01c3.96 ± 0.05b0.28 ± 0.01b0.33 ± 0.03b12.82 ± 7.33b17.50 ± 8.18b44.33 ± 0.67a Rice grains traits of JZ1 and YXYYS grown under DI and AWD. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values The brown rice weight of JZ1 grown under DI and AWD was 0.32 g and 0.32 g, respectively, while the brown rice weight of YXYYS grown under DI and AWD was 0.32 g and 0.28 g, respectively (Table 2). The result suggests that AWD reduced the brown rice grain weight of YXYYS by 12%, but there was no difference in the brown rice grain weight of JZ1 grown under DI and AWD. Statistical analyses showed no significant difference in the brown rice weight of JZ1 under DI and AWD, but there was a significant difference in the brown rice grain weight of YXYYS grown under the two irrigation methods. The lack of change in JZ1 may indicate its resistance to AWD, enabling it to maintain yield. In contrast, the decrease observed in YXYYS could reflect its vulnerability to stress under AWD conditions. The paddy rice weight of JZ1 grown under DI and AWD was 0.39 g and 0.34 g, respectively, while the paddy rice weight of YXYYS grown under DI and AWD was 0.36 g and 0.33 g, respectively (Table 2). Statistical analyses showed a significant difference in the paddy rice weight of JZ1 grown under DI and AWD, but no significant difference was found in the paddy rice weight of YXYYS grown under the two irrigation methods. However, the results also showed that 18% of the paddy rice grains of JZ1 grown under DI were made up of husk, while 7.3% of the paddy rice grains grown under AWD were made up of husk. Conversely, the results also showed that 11.3% of the paddy rice grains of YXYYS grown under DI were made up of husk, while 12.8% of the paddy rice grown under AWD were made up of husk. This results suggest that AWD may lead to less filled grains, potentially lowering milling yield and profitability. In JZ1 grown under DI and AWD, our analysis found chalkiness of 4% and 20%, respectively, while in YXYYS under DI and AWD, we found chalkiness of 19% and 18%, respectively (Table 2). Statistical analyses showed a significant difference in the chalkiness of JZ1 grown under DI and AWD, but no significant difference was found in the chalkiness of YXYYS grown under the two irrigation methods. The milled rice whiteness of JZ1 grown under DI and AWD was 38% and 39%, respectively, while the milled rice whiteness of YXYYS grown under DI and AWD was 42% and 44%, respectively. Statistical analyses showed a significant difference in the whiteness of rice grown under DI and AWD in both cultivars. The observed increase in chalkiness for both cultivars indicates that AWD can enhance specific aesthetic grain qualities, despite its potential to compromise other grain characteristics. Germination rate and hypocotyls length We further investigated the germination rate of the grains after 18 months of storage. In JZ1 grown under DI, an average of 2.3, 15.6, and 17.6 seeds germinated after 24 h, 48 h, and 72 h, respectively, while in JZ1 grown under AWD, 2, 13.6, and 16 seeds germinated after 24 h, 48 h, and 72 h, respectively (Fig. 5a). In YXYYS grown under DI, an average of 2.6, 14.6, and 18.3 seeds germinated after 24 h, 48 h and 72 h, respectively, while in YXYYS grown under AWD 2.6, 11.6 and 16 seeds germinated after 24 h, 48 h, and 72 h, respectively (Fig. 5b). Statistical analysis showed a significant difference in the number of germinated seeds in JZ1 grown under DI and AWD after 48 h, but no significant difference was observed after 24 h and 72 h. Similarly, in YXYYS grown under DI and AWD, we found no significant difference after 24 h and 72 h, but a significant difference was observed in the number of germinated seeds after 48 h. The absence of significant differences at 24 and 72 h suggests that the initial stages of germination are largely resilient to AWD; however, prolonged exposure may induce latent stress responses that ultimately compromise germination success.Fig. 5The effect of AWD on seed germination and hypocotyl growth: (a) the number of germinated seeds of JZ1 grown under DI and AWD; (b) the number of germinated seeds of YXYYS grown under DI and AWD; (c) seed germination rate of JZ1 grown under DI and AWD; (d) see d germination rate of YXYYS grown under DI and AWD; (e) the hypocotyl length of JZ1 and YXYYS grown under DI and AWD. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values The effect of AWD on seed germination and hypocotyl growth: (a) the number of germinated seeds of JZ1 grown under DI and AWD; (b) the number of germinated seeds of YXYYS grown under DI and AWD; (c) seed germination rate of JZ1 grown under DI and AWD; (d) see d germination rate of YXYYS grown under DI and AWD; (e) the hypocotyl length of JZ1 and YXYYS grown under DI and AWD. Differences in lower case letters indicate significant difference (P value < 0.05) among the mean values Furthermore, JZ1 grown under DI had germination rates of 12%, 78%, and 88% after 24 h, 48 h, and 72 h, respectively, while JZ1 grown under AWD had germination rates of 10%, 68%, and 80% after 24 h, 48 h, and 72 h, respectively (Fig. 5c). YXYYS grown under DI had germination rates of 13%, 73%, and 92% after 24 h, 48 h, and 72 h, respectively, while YXYYS grown under AWD had germination rates of 13%, 58%, and 80% after 24 h, 48 h, and 72 h, respectively (Fig. 5d). After 72 h, hypocotyl lengths differed significantly between the treatments: JZ1 seedlings averaged 2.7 cm (DI) and 2.2 cm (AWD), while YXYYS averaged 2.5 cm (DI) and 2.2 cm (AWD) (Fig. 5e). Discussion Generally, high temperature has been shown to have a negative impact on rice spikelet fertility [5], biosynthesis of 2-AP [21], starch properties, and grain quality [6]. Heat-susceptible rice cultivars were used for this study because the impact of high temperature on them is extremely harmful. Studies have shown a correlation between flag leaf and yield in rice [22, 23]. A decrease in spikelet fertility has been associated with a reduction in rice yield [24]. In the present, a study positive correlation was found between flag leaf area and spikelet fertility of rice. In all three cultivars, rice plants grown under DI (control) -from the onset of heading to harvest-had larger flag leaf area compared to rice plants grown under AWD (treatment). Rice plants grown under DI had significantly higher spikelet fertility as compared to rice plants grown under AWD. Plants tend to shed or reduce their leaves as a coping mechanism during water stress. The results of this study are consistent with Cabuslay et al. [25], which found that water stress led to the reduction in the leaf area of rice. Reduction in flag leaf area reported in rice grown under AWD could be their means of minimising the impact of water shortage. AWD initiates a cascade of physiological and molecular responses that enable plants to adapt to varying water availability. This process leads to partial stomatal closure, which conserves water by reducing transpiration; however, it also limits carbon dioxide (CO₂) uptake, which impairs photosynthesis and the functionality of the flag leaf [26]. The resulting decline in photosynthetic efficiency may explain the observed reductions in both spikelet fertility and flag leaf area. Drought conditions significantly impact abscisic acid (ABA) signaling pathways, which are crucial for the plant's response to water stress. Under AWD, elevated levels of ABA activate stress-responsive transcription factors such as DREB and NAC. These factors regulate genes involved in osmoprotection, antioxidant defense, and cell wall remodeling [27, 28]. Such molecular responses may also be linked to increased biosynthesis of 2-acetyl-1-pyrroline (2-AP), as proline metabolism—associated with 2-AP production—becomes upregulated under AWD conditions [12]. Moreover, AWD may influence the expression of aquaporin genes, leading to alterations in water transport dynamics and cellular turgor. These changes can significantly impact grain filling and the morphology of starch granules [29]. The interplay between these physiological and molecular mechanisms under AWD likely determines cultivar-specific outcomes, as seen in the differential responses of JZ1 and YXYYS. In QXXY19-which is a highly heat susceptible cultivar-grown under AWD, the flag leaves were completely withered, while the spikelet fertility was 0% (Fig. 1). Since some of the other leaves of QXXY19 grown under AWD were also withered, the spikelet sterility might not be solely due to the lack of flag leaves, but also as a result of the impact of AWD on other leaves. Another possible explanation for the spikelet sterility reported in QXXY19 could be due to the inability of the cultivar to cope simultaneously with high temperature and water shortage as a result of AWD. JZ1-the least heat susceptible of all the cultivars-and YXYYS-moderately heat susceptible cultivar-grown under DI had almost the same spikelet fertility rate of 64% and 65%, respectively. However, under AWD, the spikelet fertility rate of JZ1 and YXYYS reduced to 33% and 8%, respectively (Fig. 1). These results demonstrate that heat susceptibility levels of the cultivars influenced the impact of AWD on their spikelet fertility rate. Spikelet sterility is a major setback associated with AWD. However, spikelet fertility in rice during high temperature was significantly increased by artificially assisted pollen shedding combined with intensified fertilisation by promoting pollen germination through chemical application [30]. In order to increase spikelet fertility during AWD, rice farmers and breeders can explore artificial pollination techniques. Breeding techniques can also be used to develop cultivars with a high spikelet fertility rate. This study used gas chromatograph-mass spectrometer (GCMS) analyses to investigate the impact of AWD on the biosynthesis of 2-AP. Previous studies have found that high temperature-with a daily minimum of 26℃ and daily a maximum of 32℃-from the onset of heading to harvest significantly reduced the biosynthesis of 2-AP in leaves and grains of rice [6]. In this study, the average daily minimum temperature and daily maximum temperature from the onset of heading to harvest were 26℃ and 35℃, respectively (Table 1). We found that AWD ameliorated the negative impact of high temperature on the biosynthesis of 2-AP and significantly increased the biosynthesis of 2-AP in the leaves and grains of JZ1 and YXYYS (Fig. 2a, b, c). The only compound that rice breeders can select to adjust fragrance in rice is 2-AP [19]. Fragrance is one of the most highly valued grain quality traits in rice [19, 31]. However, the sensory approach to improving fragrance is often time-consuming, expensive, and unreliable [32]. This study has shown that even under high temperature, AWD can simplify and reduce the cost of improving fragrance in rice. If rice farmers can save water by using AWD and still harvest rice grains with high fragrance, it will reduce the cost of rice production and increase the desirability and the economic value of rice grains. Cooked rice elongation is an important trait used in the determination of rice grain quality [18, 33]. Rice with extensive elongation after cooking is desired by rice consumers and generates more revenue for rice farmers. High temperature has been reported to reduce cooked rice elongation percentage [6]. However, AWD increased the cooked rice elongation percentage in both JZ1 and YXYYS grown under high temperature (Fig. 2d). The mechanism underlying cooked rice elongation is not well understood. Tian et al. [18] reported that cooked rice elongation is likely controlled in the waxy gene region. It has been reported that during grain filling, high temperature had a negative effect on the waxy gene [34]. Targeting the waxy gene region of rice might be a reliable way to further improve the cooked rice elongation of rice grown under AWD. Starch granules are thought to be a contributing factor to starch functional properties [35]. Environmental factors such as high temperature and drought affect starch granule size [36]. Starch granule morphology has been associated with various degrees of chalkiness in rice [37, 38]. AWD had more impact on the starch granule morphology of JZ1 as compared to YXYYS (Fig. 3). The results of SEM revealed that the starch granules of JZ1 grown under AWD are irregular and loosely packed with larger airspaces, while the starch granules of JZ1 grown under DI are flat and tightly packed with small airspaces (Fig. 3a, b). However, in YXYYS grown under DI and AWD, the differences in starch granule morphology are not very pronounced (Fig. 3c, d). Wang et al. [39] showed that loosely packed starch granules in rice grains resulted in more chalkiness. The grains of JZ1 grown under AWD had significantly more chalkiness (20%) as compared to the grains under DI (4%). In YXYYS with similar starch granule morphology in both DI and AWD, no significant difference was found in the chalkiness of the grains grown under DI and AWD (Table 2). However, in both JZ1 and YXYYS, AWD increased both the starch granule size and starch granule diameter. These results suggest that starch granule arrangement and airspace influenced chalkiness, while starch granule size had no effect on chalkiness. Another factor that has been shown to affect rice chalkiness is brown rice grain width. Gao et al. [37] reported a positive correlation between grain width and chalkiness in rice grains. AWD increased the brown rice grain width of JZ1 but decreased the brown rice grain width of YXYYS. However, there was a positive correlation between brown rice grain width and chalkiness in both JZ1 and YXYYS. Even though the difference in the chalkiness of YXYYS grown under DI and AWD was not statistically significant, YXYYS grown under DI had higher grain width and more chalkinessas compared to YXYYS grown under AWD with lower grain width (Table 2). While this study has offered new insights on the effect of starch granule arrangement and brown rice grain width on chalkiness, there is a need to carry out further studies to elucidate the impact of AWD on starch granule morphology and brown rice grain width. Furthermore, AWD did not ameliorate the negative impact of high temperature on grain chalkiness. In fact, in JZ1, AWD resulted in the production of grains with significantly high chalkiness. Chalkiness is a grain trait that is not desirable to rice consumers and reduces the economic value of rice grains. Reports have shown that chalkiness has a negative effect on cooked rice [40, 41]. Rice breeders can use marker-assisted selection techniques to tackle the problem of chalkiness in rice grown under AWD. This can be achieved by targeting the genes and QTLs associated with chalkiness in rice. Previous studies have identified the genes and QTLs underlying chalkiness in rice [37, 39, 42]. Since there is a positive correlation between grain width and chalkiness, efforts should be made to reduce grain width in rice. Some studies have explained the mechanisms and pathways underlying rice grain width [43–46]. These mechanisms and pathways can be explored to control the grain width of rice grown under AWD. The varying effects of AWD on spikelet fertility and grain traits among different cultivars indicate genotype-specific molecular sensitivities. For instance, the complete sterility observed in QXXY19 may result from impaired pollen viability and fertilization processes under AWD. A study has demonstrated that heat stress disrupts auxin homeostasis and inhibits pollen tube elongation. Water stress can further exacerbate these issues by limiting nutrient transport and hormonal signaling [47]. To alleviate these effects, breeding strategies should prioritize enhancing the expression of genes associated with heat and drought tolerance [48]. Such strategies can help modulate reactive oxygen species homeostasis and involve heat shock transcription factors that protect reproductive tissues [49]. Additionally, marker-assisted selection targeting QTLs linked to spikelet fertility under stress, such as qSF1 and qHTSF4, could enhance resilience. Moreover, the exogenous application of compounds like salicylic acid or brassinosteroids has been shown to improve stress tolerance by modulating antioxidant enzyme activity and stabilizing membrane integrity [50]. These treatments could be explored to enhance spikelet fertility and grain quality under AWD conditions. Additionally, conducting transcriptomic studies of cultivars subjected to AWD could reveal novel regulatory networks and candidate genes that may improve grain traits, particularly those related to starch biosynthesis and husk development. Rice consumers are paying more attention to brown rice because of its high nutritional value and health benefits as compared to white rice [51]. Brown rice is made up of about 7% bran, about 3% embryo, and about 90% endosperm. The milling of brown rice to produce milled rice, or white rice, leads to the loss of about 70% of B vitamins, about 90% of calcium, about 85% of fat, nearly15% of protein, and roughly 75% of phosphorus [52]. Grain length, width, and weight are also important traits that determine the grain value of rice. Brown rice grain length and width can be indicators of the length and width of milled rice. AWD significantly decreased the grain length of brown rice. Ren et al. [43] stated that grain size and quality are strongly influenced by rice husk (or hull) and that rice grain size is regulated by cell proliferation and cell expansion in the husk. Several signalling pathways-such as the ubiquitin–proteasome pathway, G-protein signalling, MAPK signalling, phytohormone sensing and signalling, and transcriptional regulators-regulate and control grain size by affecting husk growth in rice [43]. AWD had a significant effect on rice husk. Interestingly, in JZ1 where AWD significantly reduced rice husk, we found a significant increase in grain width. The grain weight of JZ1 grown under DI and AWD was the same. The significant increase in the grain width of JZ1 grown under AWD possibly compensated for the decrease in grain length; hence, the same grain weight was observed in both DI and AWD. The findings of the study showed that AWD influenced rice husk, which ended up influencing grain shape. It has been reported that overexpressing WTG1 resulted in elongated grains [43], while a truncated version of OsMADS1 enhances grain elongation by promoting longitudinal cell division in the husk [46, 53]. To develop rice cultivars with improved grain size, it is important to carry out more studies on the molecular mechanisms underlying the development of rice husk. Conclusion This study demonstrates that the impact of AWD on rice under high temperature is complex and highly cultivar-dependent. The study found that starch granule arrangement and brown rice grain width—rather than granule size—influenced chalkiness in rice. Furthermore, AWD influenced rice husk development, which in turn affected grain shape and size. Although AWD significantly enhanced the biosynthesis of 2-AP and cooked rice elongation, it also substantially reduced spikelet fertility and germination rates, particularly in heat-susceptible cultivars. The findings suggest that AWD is not recommended for high-temperature environments unless it is implemented alongside careful cultivar selection and complementary agronomic strategies to maintain yield stability. Future work should focus on identifying the genetic and molecular bases of AWD-responsive traits to develop rice varieties that can benefit from water-saving irrigation without compromising fertility or grain quality. Methods Plant materials and growth conditions Three indica rice cultivars-Ye Xiang You You Si (YXYYS), Jing Zhan 1 (JZ1), and Qing Xiang You Yi 19 (QXYY19)-with different heat susceptibility levels were used for the study. QXYY19 is highly heat susceptible, while YXYYS is moderately heat susceptible, and JZ1 is the least heat susceptible of all cultivars. Rice grains were obtained from the College of Agriculture, Yangtze University, Jingzhou, Hubei Province, China. Seeds were soaked in distilled water for 24 h and sown on 15th May 2022 in randomized blocks. After 20 days, seedlings were transplanted into cylindrical plastic pots (28/30 cm in height/diameter), each filled with 14 kg of loamy clay soil sourced from the College of Agriculture experimental field. The soil was homogenized and sterilized prior to use to minimize microbial variability. Each pot received a basal application of 11 g of compound fertilizer (N:P₂O₅:K₂O = 22:8:12) from the Anhui Fengle Agrochemical Co., Ltd. (Hefei, Anhui Province, China). Fertilizer was thoroughly mixed into the soil before transplanting. Six uniform seedlings were transplanted per pot. To ensure biological replication, the experiment was conducted using three independent biological replicates per treatment per cultivar. Pots were placed outdoors under a high roof to prevent rainfall and ensure that water was controlled during rice growth, development, and maturity. From transplanting to heading, all pots were irrigated daily with 3 L of water to maintain saturated conditions. However, at the onset of heading-which began on the 3rd August 2022, in JZ1 and YXYYS, and on the 11th August 2022, in QXYY19-only 3 pots were given daily irrigation (DI) of 3 L of water and served as control, while the other 3 pots were given severe AWD [54] by watering (3 L of water) them above the soil level once every 3 days. Temperature, humidity, wind speed, and hours of sunshine were taken daily (Table 1). The findings from this study was subsequently confirmed in a separate repetition under the same experimental conditions during the 2024 growing season. Sample collection At maturity, flag leaf and grain samples were collected from each biological replicate. For each pot, samples were pooled from three randomly selected plants to minimize intra-pot variability. Samples for the determination of 2-acetyl-1-pyrroline were immediately frozen in liquid nitrogen before they were transferred to −80℃ pending further analyses. Other grain samples were dried and stored at room temperature pending further analysis. All downstream analyses were performed in triplicate for each biological replicate to ensure statistical robustness. Measurement of flag leaves and spikelet fertility Flag leaves were measured immediately after collection. The length and maximum width were measured for the calculation of flag leaf area. The flag leaf area was determined according to Yoshida et al. [55] which stipulates that:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Flag\;leaf\;area\left({cm}^2\right)=K\;\times\;Leaf\;length\;\times\;Maximum\;Leaf\;Width$$\end{document}where K is the adjustment factor. The value of K is 0.67 for rice leaves collected at the seedling stage and harvest. Spikelet fertility was determined according to Prasadet al. [56]. Each spikelet was pressed with the thumb and index finger to determine if fertilisation had occurred. Partially filled spikelets and fully filled spikelets were both categorised as fertilised spikelets. All spikelets that were completely empty were categorised as infertile. Spikelet fertility percentage was calculated as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} &Spikelet\;fertility(\%)\\&=\frac{Partially\;filled\;spikelet\;+\;completely\;filled\;spikelet}{Total\;number\;of\;spikelet}\times100\end{aligned}$$\end{document} GC–MS analysis for the determination of 2-acetyl-1-pyrroline GC–MS analysis was carried out according to Okpala et al. [6]. GC–MS analysis was conducted with GCMS-QP2010, with the following specifications: SH-Rxi-5Sil MS chromatographic column was 30 m in length, 0.25 mm in diameter, 0.25 μl film thickness, and the heating temperature at its peak was 220 ℃. High purity Helium (99.999%) was used as carrier gas; the constant pressure splitless injection method was used; the injection volume was 2 μl; the mass spectrometry used was an electron bombardment ion source; the ion source temperature was 200 ℃; ionisation energy was 70 eV; interface temperature was 250 ℃; quadrupole temperature was 150 ℃; full scan mode with a scanning mass range of m/z 35–160. All experiments were conducted in 3 replicates, and each replicate was prepared with 1 g of pulverised sample. The pulverised sample was transferred to a 20 mL bottle, followed by the addition of 10 mL of dichloromethane (CH2CL2). Samples were then transferred to an ultrasonic cleanser (KQ-800ES), set at 0 ℃ for 4 h. The samples were transferred to 10 mL conical flask before the addition of 4 g of anhydrous sodium sulfite (Na2SO3). After subsiding for 30 s, 1 mL of the supernatant was transferred to a vial with a disposable micropipette. This was followed by the addition of 2 μl of 2,3,6-trimethylpyridine (1000 × dilution with CH2Cl2) as an internal standard. Vials were closed and transferred to GC–MS machine for analysis. GC Solution 2.3 software was used for data analyses. Cooked rice elongation Cooked rice elongation was here represented as the difference between the length of grains measured before and after cooking. Three replicates were used for each sample, and 20 grains were used per replicate. The analysis was conducted according to Okpala et al. [6]. Rice measurements were carried out with a CanoScan 5600 F scanner manufactured by Canon, while a BIO-RAD T100™ Thermal cycler was used to cook rice. To determine the cooked rice elongation percentage for each treatment, the following formula was used:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$CRE\mathit(\mathit\%\mathit)\mathit=\frac{\mathit A\mathit C\mathit M\mathit L\mathit-\mathit B\mathit C\mathit M\mathit L}{\mathit B\mathit C\mathit M\mathit L}\mathit\times\mathit{100}$$\end{document}Where CRE is cooked rice elongation, ACML is the mean length after cooking, and BCML is the mean length before cooking. Scanning electron microscopy for the determination of starch granules The samples were prepared for SEM analysis based on Kasem et al. [57] with minor modifications. Grains were manually dehusked and fixed for 24 h in 10% formalin. Fixed grains were rinsed three times (10 min each time) with distilled water to remove the formalin. Rinsed grains were then dehydrated by a standard series of ethanol washes: 25%, 50%, 75%, and 95%, each for 30 min and then three washes (40 min each time) in 100% ethanol. Samples were air-dried for 24 h. The dried samples were cut transversely in half. Specimens were gold coated for 60 s at 30 mA. Scanning electron microscopy was conducted using a TescanVega3 SBU scanning electron microscope. Samples were examined at a magnification of 5.00 k at 20.0 kV. The SEM micrographs were imported into ImageJ software and measured as previously described [58]. The set scale was set based on the scale bar in the SEM image. Individual starch granules were then manually traced using the freehand selection tool, and the area, perimeter, and fit ellipse of each granule were measured. Three biological replicates were set for each sample. Starch granule size was determined using the following formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Starch\;granule\;size=\frac{Major\;axis+Minour}2$$\end{document}while, the equivalent of starch granule diameter was calculated as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Starch\;granule\;diameter=2\times\frac{\sqrt{Granule\;area}}{\mathrm\pi}$$\end{document} Grain traits analysis Rice was manually dehusked to avoid damage to the grains and scanned with CanoScan 5600 F to measure grain length and width. Experiments were carried out in 3 replicates for each treatment. For each replicate, 20 grains were scanned, and the average length and width of each replicate were determined. Both brown rice and paddy rice weight were determined with FA2004B (Shanghai Yoke Instrument Co. Ltd.) electronic balance. 3 replicates were used for each treatment, and each replicate has 20 grains. Brown rice was milled with JDMZ100 milling machine manufactured by Beijing Dongfu Jiuheng Instrument Technology Ltd. Milled rice was used for the determination of chalkiness and whiteness of the grains. Here chalkiness was represented as the percentage of chalky area in the rice grains. Rice chalkiness and whiteness were determined with a CanoScan 5600 F scanner manufactured by Canon. Determination of germination rate and hypocotyl length To determine the number of germinated seeds, rice seeds were placed on moist Whatman No.2 filter paper in petri dishes (90 mm × 15 mm). The temperature of the chamber was 27 ℃ with relative humidity of 100%. The germination rates were calculated daily. The length of the hypocotyl was measured at the end of the third day. Statistical analysis Statistical differences among treatment groups were evaluated using analysis of variance (ANOVA) in Genstat 22.1. Mean separation was conducted using the Least Significant Difference (LSD) test at a 5% significance level. The Duncan Multiple Range Test (DMRT) was used as a post-hoc analysis to identify specific group pairs that differed significantly at the same significance level. Data were presented in tables and figures, with different lowercase letters indicating significant differences among group means, as determined by the DMRT.