The experiment was carried out in lysimeters under a large-scale rain shelter at the Xinxiang Comprehensive Experimental Station of the Chinese Academy of Agricultural Sciences located in Qiliying Town, Xinxiang, Henan, China (35°18′N, 113°54′E, 75 m a.s.l.) with temperate monsoon weather in the 2019 and 2020 maize growing seasons. All lysimeters are non-weighing with well-equipped irrigation and drainage systems. The dimensions of each lysimeter were 2.0 m wide × 3.33 m length × 2.0 m in depth. The top side of the steel outer frame of the lysimeter is 10 cm higher than the soil surface in the lysimeter to prevent runoff during rain or irrigation events. A total of 24 lysimeters were arranged in two rows under a rain shelter. There was a 2 m space between the rows and 20 cm between lysimeters in the same row. The physical and chemical properties of the top 40 cm of the soil layer are shown in Table 1 . A mobile rain shelter was installed above the two rows of lysimeters and closed before a rainfall and opened after the rainfall. This was done to avoid the severe effects of natural rainfall on the experiment of signed water deficit at different stages in maize growing seasons. An automatic weather station (YM-HJ03, Handan Chuangmeng Electronic Technology Co., Ltd., Hebei, China) was set on the edge of the lysimeter area. The average daily temperature and accumulated precipitation during the whole growing season of waxy maize in the two experiment seasons are shown in Figure 1 .

The experimental waxy maize variety was “Shenkenuo 1,” bred by the Shanghai Academy of Agricultural Sciences. The variety is a multi-resistant waxy maize variety with high taste quality and great planting promotion value (Wang et al., 2014). A total of 40 plants of waxy maize were maintained in each lysimeter, with row spacing of 60 cm and plant spacing of 30 cm. The experiment arranged only one factor (water deficit stage) with four treatment levels, i.e., water deficit occurred at V6–VT, VT–R2, and R2–R3 stages, and no water deficit in the whole waxy maize growing season (as CK), respectively. Following the soil water content arrangement shown in Xiao et al. (2011), no irrigation was carried out when the soil water contents were higher than 50% of field capacity during the process of water deficit treatment at the V6–VT, VT–R2, and R2–R3 stages of waxy maize, and irrigation was performed when the soil water content was reduced to or less than 50% of field capacity, or at the end of a growing stage with water deficit treatment. During the periods without arranging water deficit treatment, the soil water contents were maintained at more than 65% of field capacity, so the soil water contents in CK treatment were maintained at higher than 65% of field capacity in whole growing seasons. All four treatments were replicated three times. The planned soil water lower limits for all four treatments are shown in Table 2 . The beginning date of each growth stage is shown in Table 3 .

The waxy maize grains were collected at late R3 stages for determining fresh grain quality using the method described by Huang et al. (2022) in the 2019 and 2020 seasons. The soluble sugar content (determined with the anthrone colorimetric method), starch content (with the anthrone-sulfuric acid method), total protein content (with the total nitrogen content method, total protein content = total nitrogen content × 6.25), and lysine content (with the ninhydrin chromogenic method) of waxy maize grains were measured for each sample. The contents of amylopectin, amylose, gliadin, gluten, albumin, and globulin were measured by the Sanshu Bio-Tech company in China. The amylose content of starch was determined using a colorimetric amylose content assay (Knutson and Grove, 1994). The amylose content was analyzed using the GPC-RI-MALLs system developed by Park et al. (2007). Glutenins and gliadins were extracted and quantitated subsequently from two biological replicates by reverse-phase ultra-performance liquid chromatography (RP-UPLC) according to a method described by Han et al. (2015), and the sample size was modified in minor ways according to the protein concentration of waxy maize grain. The albumin and globulin content was analyzed using an automatic microplate reader (Multiskan GO, Thermo, USA).

The effects of water deficits imposed at different stages on the waxy maize growth index, grain yield, yield components, and grain quality were analyzed by analysis of variance using the General Linear Model procedure (GLM) in SPSS 19.0 (IBM Inc., Chicago, IL, USA). Duncan’s new multiple range difference method was used to test the significance of the difference at the P<0.05 level. Figures were drawn with Origin 2017 (OriginLab, USA). Principal component analysis was used to determine the comprehensive impact of the water deficit.

Plant height and leaf area index (LAI) of waxy maize varied significantly with growing stages and growing seasons (P<0.01; Table 4 ). The results of the two growing seasons showed that plant height and LAI decreased most significantly under the water deficit imposed at the V6–VT and VT–R2 stages ( Table 4 ), and the change trends in the two growing seasons were basically the same ( Figure 2 ). At the end of the water deficit period of V6–VT and VT–R2, all plant heights and LAI values of D_V6–VT and D_VT–R2 were significantly lower than those of the CK treatment. Compared with CK, the plant height under D_V6–VT and D_VT–R2 treatments at the R3 stage decreased by 10.7% and 9.0% (P<0.05), and the LAI decreased by 22.4% and 19.5% (P<0.05), respectively. The D_R2–R3 treatment did not exhibit significant effects on plant height and LAI, but because of temperature, light, and other reasons factors, the plant height of waxy maize in 2019 was higher than in 2020. It was clearly indicated that a water deficit imposed at the V6–VT and VT–R2 stages may severely limit the plant growth and leaf development of waxy maize.

MDA content and antioxidant enzyme activities in waxy maize leaves under different water-deficit treatments are shown in Figure 3 . Compared with CK, the MDA content in D_V6–VT and D_VT–R2 increased by 40.8% and 46.0%, respectively (P<0.05, Figure 3A ), while the MDA content in D_R2–R3 was significantly decreased by 30.1% (P<0.05). These results indicated that the recovery of plants in the D_V6–VT and D_VT–R2 treatments was weak after re-watering. It may be the main reason that changes in CAT were insignificant under D_V6–VT treatment, and the activities of SOD and POD were significantly decreased by 26.8% and 25.8%, respectively, compared with CK (P<0.05). The oxidative damage to waxy maize plants still appeared obviously at the milk stage, even after a long-term release of the water deficit imposed at the jointing stage. D_VT–R2 significantly increased CAT by 33.6% (P<0.05) but decreased significantly SOD and POD by 19.1% and 18.4% (P<0.05), which indicated that the oxidative damage caused by the water deficit at the VT–R2 stage can only be alleviated to a certain extent. D_R2–R3 exhibited less oxidative damage and obvious increases in SOD, CAT, and POD by 24.5%, 83.8%, and 24.5%, respectively (P<0.05), which indicated that the damage caused by water deficit at the R2–R3 stage may have been almost completely alleviated at the milk stage.

The soluble protein, soluble sugar, and free proline in waxy maize leaves at the R3 stage are shown in Figure 4 . Results showed that the recovery of waxy maize plants after re-watering was poor under D_V6–VT and D_VT–R2 treatments. Compared with CK, the soluble sugars and soluble proteins in the D_V6–VT treatment increased by 95.4% and 38.1% (P<0.05) and by 42.0% and 26.4% (P<0.05) in the D_VT–R2 treatment, respectively. These increases were beneficial to maintaining cell water potential under water deficits, reducing leaf water loss, and improving leaf water use efficiency. The D_V6–VT treatment had the highest levels of soluble sugar, soluble protein, and proline. Meanwhile, the soluble sugar content in the D_R2–R3 treatment increased by 30.9% (P<0.05), but the soluble protein decreased by 19.1% compared with the CK treatment. It was further indicated that a water deficit in the V6–VT stage had the greatest impact on the relevant index in waxy maize leaves and the smallest effects from a water deficit in the R2–R3 stage.

Under water deficit, plants usually reduced water loss by closing partially stomates, while the photosynthetic recovery of waxy maize was different after release from the water deficit imposed at different growing stages ( Figure 5 ). The changes in SPAD and photosynthetic characteristics in the two growing seasons were basically the same ( Figure 5 ). After water deficit at the V6–VT stage, SPAD and gas exchange parameters of leaves decreased, but leaf water use efficiency (LWUE) increased. After a water deficit at the VT–R2 stage, P_n , G_s , C_i , and T_r of leaves decreased significantly. Compared with CK, the D_V6–VT treatment reduced the SPAD of waxy maize leaves ( Figure 5A ), and the D_V6–VT and D_VT–R2 treatments significantly decreased P_n , G_s , and T_r ( Figures 5B, C, E ) and C_i of waxy maize leaves at the R3 stage. Both D_V6–VT and D_VT–R2 treatments increased LWUE significantly ( Figure 5F ). But no significant differences in P_n , C_i , G_s , T_r , and LWUE between D_R2–R3 and CK were investigated. With the postponement of the water deficit stage, P_n , G_s , and T_r showed an increasing trend, while C_i showed a trend of decreasing at the beginning and then increasing later ( Figure 5D ), while LWUE showed a firmly decreasing trend ( Figure 5F ).

The results of the ANOVA in two growing seasons showed that there were significant differences in grains per ear and 100-grain weight under different water deficits imposed at different growth stages of waxy maize (P<0.01; Table 4 ). The 100-grain weight of D_V6–VT was significantly lower than those of other treatments, decreasing by 19.4% compared with CK (P<0.05; Figure 6A ). Meanwhile, the grain yield per ear of the D_VT–R2 treatment was the lowest and was 23.1% lower than that of CK (P<0.05; Figure 6B ).

Ear length, ear diameter, bald tip length, grain rows per ear, and grains per row significantly varied with water deficit treatments (P<0.01; Table 4 ; Figures 7A–E ). The results of multiple comparisons in two growing seasons showed that the ear length of D_V6–VT, compared with CK, decreased the most obviously, followed by D_VT–R2. Meanwhile, the ear diameter under the D_VT–R2 treatment decreased the most significantly, and grain rows per ear and grains per row of D_VT–R2 decreased significantly due to the decrease in ear length and ear diameter ( Table 4 ). Compared with CK, the ear length of D_V6–VT and D_VT–R2 decreased by 12.7% and 8.5%, respectively (P<0.05; Figure 7A ), while the ear diameter, grain rows per ear, and grains per row of D_VT–R2 decreased by 3.2%, 12.6%, and 16.8%, respectively (P<0.05; Figures 7B, D, E ). It was the decrease in grain rows per ear and grains per row that resulted in the decrease in grains per ear. Due to the stage differences in imposed water deficit stages, the maturity dates of ears were obviously different, which resulted in different grain moisture contents at harvest. The grain moisture content of D_V6–VT treatment was significantly lower than that of CK, while no remarkable differences were investigated between the grain moisture contents of D_VT–R2 or D_R2–R3 treatments and that of CK at the end of R3 stage ( Figure 7F ).

Fresh ear yield with husks (HFY) and fresh ear yield (FY) were affected very significantly by the water deficit stage and growing seasons (P<0.01; Table 4 ). The ear yield of waxy maize varied under different water deficit treatments. Water deficit at the V6–VT stage showed the greatest effect on the ear yield, followed by those at the VT–R2 stage, and the least effect under water deficit at the R2–R3 stage ( Figure 8 ). Compared with those under CK, mainly due to the varying grain number per ear and 100-grain weight under water deficit at different stages, the HFY and FY decreased by 22.0% and 20.3% under D_V6–VT (P<0.05), by 14.3% and 14.2% under D_VT–R2 (P<0.05), and both less than 10.0% under D_R2–R3, respectively.

Total protein, grain soluble sugar, and starch contents in waxy maize grains varied significantly with the different water deficit stage treatments and growing seasons (P<0.01; Table 4 ). Water deficit increased the contents of total protein, starch, and lysine in fresh waxy maize grains, but reduced the content of soluble sugar in fresh waxy maize grains. However, the effects of water deficits at different growing stages on grain quality were obviously different, and the changing trends in the two growing seasons were consistent ( Figure 9 ). The soluble sugar content of the D_V6–VT treatment was the lowest due to a 31.6% decrease over that of CK ( Figure 9B ; P<0.05), followed by the D_VT–R2 treatment with a decrease of 14.1% (P<0.05), while the decrease was 4.0% under the D_R2–R3 treatment. The total protein content and starch content were the greatest under the D_V6–VT treatment ( Figures 9A, C ) with an increase of 8.5% and 66.5%, respectively (P<0.05) than those under CK, respectively (P<0.05). The following were those under the D_VT–R2 treatment with an increase of 7.1% and 20.1% (P<0.05), and under the D_R2–R3 treatment with an increase of 2.0% and 1.5% under the D_R2–R3 treatment. Compared with the CK treatment, the D_V6–VT, D_VT–R2, and D_R2–R3 treatments increased the lysine content in grains by 16.7%, 23.8%, and 16.7%, respectively.

Figure 9 indicated that the water deficit at the V6–VT stage had the most significant effects on the total protein and starch content of waxy maize grains, so the changes in amylose, amylopectin, and component protein content of waxy maize grains under the D_V6–VT treatment were selected as representative to show the effects of water deficit on grain quality index ( Table 5 ). Compared with CK, the amylose and amylopectin content under the D_V6–VT treatment increased by 15.9% and 92.5% (P<0.05), the contents of glutelin and globulin decreased by 9.1% and 56.9%, respectively (P<0.05), while the contents of alcohol-soluble protein and albumin increased by 24.0% and 59.9%, respectively (P<0.05).

Principal component analysis (PCA) was performed on all measured indexes of waxy maize in the two growing seasons. Three principal components (PC1, PC2, and PC3) were extracted in 2019 and 2020 (λ >1), and the eigenvalues (λ) of principal component 1 (PC1) in 2019 and 2020 were 18.88 and 13.37, and explained 69.9% and 66.9% of the total variation, respectively. λ of PC2 in 2019 and 2020 were 5.19 and 4.22, which contributed 19.2% and 21.1% of the total variation, respectively. λ of PC3 in 2019 and 2020 were 2.93 and 2.41, which explained 10.9% and 12.0% of the total variation, respectively ( Table 6 ). The largest loading variable in 2019 was CAT, followed by bald tip length, SPAD, lysine, and ear diameter. The largest loading variable in 2020 was C_i , followed by ear length, ear diameter, grain number per ear, and fresh ear yield with husks. These results indicated that leaf physiology traits were most sensitive to water deficit, followed by ear traits and fresh ear yield in waxy maize ( Figure 10 ).

When plants are subjected to water stress, a large amount of reactive oxygen species will accumulate in the tissues, which will break the mechanism of reactive oxygen species production and scavenging, triggering the production of MDA by cell membrane peroxidation (Liu et al., 2015). Meanwhile, plant cells produce many kinds of antioxidant enzymes such as SOD, CAT, and POD to scavenge reactive radicals (Simova et al., 2008). Studies have shown that the accumulation of MDA varies with growing stages and that the longer a water deficit lasted, the more MDA was accumulated (Liu, 2013b). In our study, the MDA content in waxy maize leaves under the D_V6–VT and D_VT–R2 treatments was significantly higher than that under the CK treatment at the R3 stage, and the MDA value under D_V6–VT was greater than that under D_VT–R2. Our results are similar to the results of Li et al. (2017), which indicated that the magnitude of MDA accumulation at the jointing stage was greater than that at the filling stage. On the one hand, SOD, CAT, and POD activities in the cells increased to scavenge excessive reactive oxygen species (Liu et al., 2015). In this study, both SOD and POD activities were significantly lower in the D_V6–VT treatment compared with those in CK, mainly due to the inhibition of SOD and POD activities by excessive MDA (Liu, 2013b). But SOD, CAT, and POD activities in the D_R2–R3 treatment increased significantly, which indicated that the plant functions recovered better after re-watering from the R2 to the R3 water deficit. Bu et al. (2009) also demonstrated an increase in protective enzyme activity after re-watering. Meanwhile, the increases in proline and total carbohydrate contents under water deficit are beneficial to protect maize plant tissues from oxidative damage (Bu et al., 2009; Anjum et al., 2016). Previous studies have shown that proline and MDA are in a reciprocal relationship, with proline accumulation helpful for reducing MDA damage to the plant, while soluble sugars and soluble proteins increase beneficial for maintaining cellular osmotic pressure (Liu et al., 2015; Li et al., 2017). In our study, the soluble sugar, soluble protein, and proline contents of waxy maize leaves under D_V6–VT were significantly higher than those under CK at the R3 stage, while these contents under both D_VT–R2 and D_R2–R3 were similar to those under the CK treatment. Even though the soluble sugar content increased after re-watering (Bu et al., 2009), the damage caused by the water deficit imposed at the V6–VT stage remained unrecovered. After re-watering, the recovery of antioxidant enzyme activity was very poor, photosynthesis was adversely affected, and an irreversible effect resulted in waxy maize plants grown under the D_V6–VT treatment in our study, mainly due to the long water deficit period (approximately 20 d, see Table 3 ) in the V6–VT stage.

Photosynthesis is the main physiological process driving plant growth and is highly sensitive to water deficits (Chaves et al., 2009). Water deficits reduce P_n , T_r , and G_s , which in turn reduce maize plant biomass and grain yield (Ali and Ashraf, 2011; De Carvalho et al., 2011; Li et al., 2018). In our study, P_n and G_s decreased significantly under both the D_V6–VT and D_VT–R2 treatments with nearly the same variation pattern, which was very similar to those results from Cai et al. (2017). C_i was also significantly reduced in the D_VT–R2 treatment, mainly due to stomatal limitation, but was not significantly affected in the D_V6–VT treatment. SPAD value decreased obviously under the D_V6–VT treatment, because reactive oxygen species generated by water deficit severely degraded severely chlorophyll pigments (Anjum et al., 2016) and significantly decreased chlorophyll content (Ye et al., 2020b). Water deficit also may reduce photosynthetic enzyme (PEPCase, RuBPase) activity (Ye et al., 2020a), and it caused a remarkable reduction in photosynthetic rate under the D_V6–VT treatment in this study. Reduced G_s also limited transpiration and resulted in a significant decrease in T_r under the D_V6–VT and D_VT–R2 treatments. But LWUE increased when plants were exposed to a water deficit, due mainly to the obvious reduction in transpiration and water consumption (Yao et al., 2012).

The reduction of photosynthetic rate under water deficit may lead to a reduction of plant photoassimilate deposition and ultimately the loss of grain yield (Kimaro et al., 2015; Ye et al., 2020a; Ulfat et al., 2021). Previous studies have shown that a water deficit during grain filling reduced the grain fresh weight, water content, and grains per ear of waxy maize (Guo et al., 2022). In this study, fresh ear yield with husk and fresh ear yield were significantly reduced, and the grain number per ear and 100-grain weight also showed some reducing trends under the D_V6–VT and D_VT–R2 treatments. The most obvious decrease in grain number per ear was investigated in the D_VT–R2 treatment, while the most remarkable decrease in 100-grain weight occurred in the D_V6–VT treatment. Previous results already indicated that a water deficit imposed at the tasseling, flowering, and filling stages could reduce the grain number per ear and grain weight of waxy maize, resulting in yield loss (Sun, 2014). Similar to the results of Xiao et al. (2011), the yield loss under treatment with a water deficit that occurred at the jointing stage was less than that under a water deficit imposed at the filling stage. The yield loss trend under water deficit may be described as that the yield loss reduces gradually as the period of water shortage becomes later and later. The yield reduction was mainly caused by the decrease in grain number per ear and 100-grain weight (Wen, 2020). Yang et al. (2019) also showed that a water deficit imposed at the tasseling stage significantly reduced the grain number per ear and 100-grain weight of fresh waxy maize, mainly due to the decreases in pollen dispersal and grain filling rates. Water deficit not only affects the formation and vitality of pollen but also affects the differentiation of maize spikelets (Li et al., 2022), resulting in a decrease in the number of rows per ear, grain number per ear, and final yield reduction. This study showed that high accumulation of MDA in leaves resulted in quicker leaf senescence, and then decreases in LAI, photosynthesis rate, and dry matter accumulation under the D_V6–VT treatment. The greatest reductions in 100-grain weight and grain moisture content were investigated under D_V6–VT in this study, indicating that the D_V6–VT treatment caused the process of starch accumulation in advance and led to poor filling quality. The P_n values in 2020 were generally lower than those in 2019, mainly due to less radiation and a lower temperature in 2020 ( Figure 5B ), which resulted in a lower 100-grain weight and less dry matter accumulation in 2020 (Gao et al., 2017). The worse radiation environment in the 2020 growing season also led to a decrease in pollen viability and a subsequent decline in grains per ear, which may be the main reason for the remarkable differences in fresh ear yields between the two growing seasons (Zhou et al., 2013).

Grain quality was closely related to leaf physiology and yield. Studies have shown that an increase in wheat grain protein content is closely associated with a decrease in yield under water deficit (Ozturk and Aydin, 2004; Flagella et al., 2010). It is also indicated that the starch content of grain was also closely related to yield in barley (Worch et al., 2011). In this experiment with waxy maize, the D_V6–VT and D_VT–R2 treatments increased the total protein contents while decreasing the soluble sugar contents. In addition, the starch and lysine contents were increased by the water deficit imposed at all growing stages. Several studies have shown that water deficit at the grain formation stage may reduce the starch content and increase the total protein content of waxy maize grain (Sun, 2014; Wang et al., 2021b). The increase in protein content was mainly due to the concentration effect of reduced biomass under water deficits (Rotundo and Westgate, 2009). Ma et al. (2006) also showed that the protein and lysine contents of maize grains increased under moderate water deficit conditions. The changes in protein and lysine contents in this study showed very similar trends to those in Ma et al. (2006). D_V6–VT resulted in increases in glutenin and albumin contents, but decreases in alcoholic-soluble protein and globulin contents in grains. For the grain quality of waxy maize, the effects of water deficit on grain starch content in this study were opposite to those shown in some previous studies (Sun, 2014; Wang et al., 2021b), maybe mainly because the previous results were investigated from matured grains, whereas fresh grains were used in this study. Debon et al. (1998) and Silva et al. (2001) found that the increase in the starch concentration of fresh fruit of potato was mainly due to the concentration effect caused by the decrease in water content in fresh fruit. Guo et al. (2022) and Shi et al. (2018) also showed an increase in starch content in fresh waxy maize grains under drought stress and indicated that acceleration of grain filling and decreases in grain water content under drought ( Figure 7F ) might be the main reasons for the starch content increase. Meanwhile, water deficit caused a significantly increase in abscisic acid (ABA) content in maize plant tissues, and then led to the activity increases of four key enzymes related closely to converting soluble sugar into starch (Zhang et al., 2012). Yi et al. (2014) indicated that a water deficit imposed at V6–VT advanced the starch accumulation process in sorghum, and Dai et al. (2008) also indicated that a water deficit increased starch accumulation in wheat at the early filling stage. In our study, both amylose and amylopectin increased under water deficit at the V6–VT period, and the percentage of amylopectin increase was much greater than that of amylose. Water deficit reduced the expression of starch branching enzymes SEBI and SBEIIb genes, which resulted in decreases in amylose and amylopectin content, and then the final total starch content (Wu et al., 2022). The results of PCA ( Figure 10 ) showed that the antioxidant enzymes and gas exchange parameters in leaves were items affected early by water deficit, and ear growth, grain number per ear, 100-grain weight, the water content of fresh grains were affected lately, and fresh ear yield and grain quality as the finally affected items.