A two-year field experiment (2018-2019) was conducted at the agricultural test station in Jingzhou City, central China (29° 26′N, 111° 15′E, 28 m above sea level). This area features a typical subtropical monsoon climate, with an average yearly rainfall of about 1,050 mm. The region enjoys a mean annual sunshine duration surpassing 1,725 hours and an average yearly temperature of 16.6°C. Figure 1 displays monthly precipitation, mean air temperature, and sunshine duration during both cotton cultivation seasons, together with the related 30-year averages (1988-2017). According to FAO standards, the soil at the experimental site is categorized as calcareous alluvial, with a field capacity (F_c) averaging 23.8% and a pH of 6.9. Analysis of the topsoil (0–40 cm) showed that organic matter content, total N, total phosphorus, and total potassium were 17.58, 1.25, 0.48, and 22.23 g kg^-1, respectively. Moreover, available phosphorus, available potassium, nitrate N (NO₃^--N), and ammonium N (NH₄⁺ -N) were 12.21, 85.10, 4.87, and 9.28 mg kg^-1, respectively.

A split-plot design was utilized, where drainage regime served as the main plot and nitrogen (N) fertilizer rate acted as the sub-plot factor. Each plot covered an area of 24 m² (6 m × 4 m) and was repeated three times. Two parallel drainage ditches were built in every plot, each being 6 m long, 10 cm deep, and 15 cm wide. Polyethylene film was placed within backfilled trenches to a depth of 1 m along the boundary of each plot to form a hydraulic barrier. The drainage regimes consisted of free drainage (FD) and controlled drainage (CD). In the FD regime, field ditches were handled according to natural drainage patterns, in line with locally suggested farming practices. In the CD regime, an iron sluice gate was installed at one end of the drainage ditches, and the other end was blocked with a polyvinyl chloride board to retain surface runoff within the experimental plot. The sluice gate stayed shut until the water level in the ditch reached 5 cm—a level recognized as possibly inducing waterlogging stress in cotton plants (Qi and Zhu, 2023). The sluice gate was manually operated based on visually observed water depths during rainfall occurrences. A graduated steel ruler, 20 cm long, was set up in the middle of each ditch to check the depth of collected water. Four N application rates were assessed: a reference rate of 280 kg N ha^-1, together with three reduced levels equivalent to 90%, 80%, and 70% of the suggested nitrogen rate, namely 252, 224, and 196 kg N ha^-1, designated as N1, N2, N3, and N4, respectively. The rate of 280 kg N ha^-1 was taken as the recommended N level for local cotton growing according to soil test outcomes (Qi and Zhu, 2023).

Before sowing, calcium superphosphate (with 17% P₂O₅) and muriate of potash (with 60% K₂O) were applied at rates of 529 kg ha kg^-1 and 300 kg ha kg^-1 respectively. Urea (46% N) was used as the nitrogen source and applied in split doses: 30% as basal fertilizer, then 30% at the bud stage and 40% at the flowering stage as topdressing. The basal fertilizers of N, P₂O₅, and K₂O were applied by banding, while N topdressing was put into the planting holes. A commercial cotton variety (Gossypium hirsutum L.), Zhongmiansuo No.63, was used as the test material. Sowing occurred on May 10 and 12, and harvesting was carried out on November 19 and 20 in the 2018 and 2019 growing seasons respectively. Seeding furrows, each 3.5 cm deep and 5.0 cm wide, were made by an machine-drawn plough with a row spacing of 80 cm. By using a manual hill-drop sowing method, four to six seeds were placed per hill at intervals of 23.7 cm within the rows. Cotton seedlings were thinned to a density of 5.24 plants per square meter at the two-leaf stage. Each experimental plot had five rows, each 6.0 meters long and spaced 80 cm apart. Throughout the growing season, the crop depended only on natural rainfall without any additional irrigation. During both years of the study, diseases, weeds, and insect pests were well managed in all treatments.

All the measured data were individually processed using a randomized complete block design (RCBD) method with the PROC GLM procedure in SAS for variance analysis. The means were compared by Duncan’s multiple range test at a significance level of P < 0.05. Although most of the measured N, water, and physiological parameters exhibited variation between years, there was neither year × drainage regime nor year × N interactions (Table 1). Therefore, we merged the data from the two different years.

The LAI at the seeding and squaring stages were comparable for the different treatments (Figure 2). However, the LAI at the other measured stages varied among the treatments. Compared to N1, the reduced N treatments significantly reduced LAI at the bud, flowering, boll setting and maturity stages (10.6%-35.5% smaller) under the two drainage regimes. Moreover, CD significantly increased LAI at the flowering, boll setting and maturity stages (9.5%-30.0% greater) at the reduced N rates when compared to FD. The CDN1 resulted in the greatest LAI at the flowering and boll setting stages, and the FDN4 resulted in the smallest LAI (Figure 2). SPAD values in the measured growth stages under different treatments showed similar variations compared with the LAI in the corresponding stages (Figure 3).

In all treatments, the maximum SOD, POD and CAT activities were found at the boll setting stage. Reduced N treatments significantly reduced activities of the SOD, POD and CAT at the flowering, boll setting and maturity stages (decreased by 7.8%-47.6%) under FD, while their activities at the boll setting and maturity stages were only significantly smaller in N4 (decreased by 11.1%-32.7%) under CD (Table 2). CD rather than FD resulted in 3.6%-31.4% higher SOD, POD and CAT activities at the three growth stages at each N rate, although the difference did not reach a significant level at N1. N4 resulted in the smallest SOD, POD and CAT activities under the two drainage regimes. The CDN1 resulted in the greatest SOD, POD and CAT activities at the three growth stages, and the FDN4 resulted in the smallest SOD, POD and CAT activities (Table 2).

Compared to normal nitrogen (N1) application, all the reduced N treatments (N2, N3, and N4) significantly reduced P_n (decreased by 10.2%-33.1%) at the flowering, boll setting and maturity stages under the two drainage regimes (Figure 4). Moreover, CD significantly increased P_n by 9.6%-23.7% at the flowering, boll setting and maturity stages at each N rate when compared to FD. The CDN1 resulted in the greatest P_n at the three growth stages, and the FDN4 resulted in the smallest P_n (Figure 4).

N2, N3, and N4 significantly reduced soluble protein content (decreased by 8.7%-24.5%) at the flowering, boll setting and maturity stages under FD when compared to N1 (Figure 5). However, N2 and N3 had a comparable soluble protein content at the flowering and maturity stage when compared to normal N rate (N1). Moreover, CD significantly increased soluble protein content by 10.1%-29.4% at the boll setting stage at each N rate when compared to FD. The CDN1 resulted in the greatest soluble protein content at the three growth stages, and the FDN4 resulted in the smallest soluble protein content (Figure 5).

All the reduced N treatments significantly increased MDA content by 9.2%-38.7% at the flowering, boll setting and maturity stages under the two drainage regimes (Figure 6). Moreover, CD significantly decreased MDA content by 12.1%-36.8% at the three growth stages at each N rate when compared to FD. The CDN1 resulted in the smallest MDA content at the three growth stages, and the FDN4 resulted in the greatest MDA content (Figure 6).

Reductions of N application rates significantly reduced seed cotton yield by 9.2%-18.6% under FD; while only N4 significantly reduced seed cotton yield (decreased by 13.0%) under CD. FDN4 resulted in the smallest seed cotton yield (Figure 7).

A decrease in the LAI and SPAD value can mirror the leaf senescence status. The SPAD value offers an indirect assessment of the relative chlorophyll content, can be employed to indicate the plant’s potential ability to absorb light energy (Ren et al., 2023). The LAI can serve as an indicator of the photosynthetic potential of the canopy., thereby influencing biomass accumulation and the final crop yield (Gitelson et al., 2014). In the current study, CD generally led to greater LAI and SPAD values (Figures 2, 3) during the bud, flowering, boll setting, and maturity stages. This implies that controlled drainage can improve functioning of cotton leaves in the middle and late growth stages. Three potential mechanisms are responsible for this phenomenon. Firstly, CD augmented the soil moisture content (Supplementary Figure S1) by prolonging the retention of shallow water in croplands after irrigation or precipitation (Tolomio and Borin, 2018). It also diminished the total nitrogen loss via runoff, resulting in a higher soil nitrogen availability in the plough layer during the cotton growth season (Qi and Zhu, 2023), as evidenced by the increased soil NO₃^--N content (Supplementary Table S1). The reduced total nitrogen loss through runoff was related to the significantly smaller drainage volume under CD (King et al., 2022; Tanga et al., 2025). Alternately, in low humidity (high suction) soil, the existing form of NO₃^--N is wholly or partially solid nitrate. In contrast, in high humidity (low suction) soil, the existing form of NO₃^--N is nitrate ions dissolved in the soil solution. Solid nitrate is fixed in the soil, whereas the soil solution containing nitrate ions can migrate to soil with lower moisture and high suction, driven by the matrix suction in unsaturated soil (Wang et al., 2021). During this process, the solid nitrate in the low moisture soil dissolves, leading to an increase in soil NO₃^--N contents in high moisture soil (Wang et al., 2021; Qi and Zhu, 2023). The improved soil moisture and nitrogen contents in the plough layer are conducive to expanding leaf size (Qi and Hu, 2022). Secondly, CD could improve the morphological characteristics of roots in oilseed sunflower (Dou et al., 2021) due to the improved soil water and nitrogen availability (Wang et al., 2021; Qi et al., 2023). The enhanced root growth resulted in various positive physiological effects mediated by abscisic acid (ABA) signaling (Liu et al., 2005). As a result, the capacity of roots to absorb soil water and nutrients was obviously enhanced (Wang et al., 2016; Zhang et al., 2021).This brought a greater leaf water content, which help to maintain large size and function of leaf (Yang et al., 2022). Thirdly, CD up-regulated the activities of SOD, POD, and CAT (Table 2) and down-regulated the MDA content at the post growth stages (Figure 5), suggesting a better reactive oxygen species scavenging ability for plants treated with CD. Moreover, CD lowering the reduction of LAI and SPAD values caused by decreased nitrogen rates as the enhanced soil moisture (Supplementary Figure S1) and NO₃^--N contents (Supplementary Table S1). Therefore, it was not surprised that the CDN1 treatment resulted in the greatest LAI and SPAD values (Figures 2, 3). This outcome suggests that when fertility is not a limiting factor, the improved soil moisture regulation inherent in CD can be fully harnessed by the crop, leading to enhanced leaf growth. In contrast, the FDN4 treatment had the lowest NO₃^--N content (Supplementary Table S1) and smaller soil moisture content (Supplementary Figure S1) during the cotton growth season, corresponding to the lowest LAI and SPAD values in the middle and late growth stages.

The capacity to scavenge reactive oxygen species is closely associated with plant senescence (Choudhury et al., 2017). MDA interacts with proteins in the cell membrane structure and inactivates them; its content indicates the level of lipid peroxidation (Qiu et al., 2025). CD led to higher activities of SOD, POD, and CAT (Table 2), along with a lower MDA content (Figure 6), suggesting an enhanced reactive oxygen species scavenging ability under controlled drainage. This was consistent with the previously published findings of that improved water management was useful to the antioxidant defenses and osmoregulation (Hu et al., 2010). Such kind findings could also serve as the new physiological evidence to support the beneficial effects of CD on crop production, as reported in previous studies (Jouni et al., 2018; Dou et al., 2021). This was associated with a more oxygen-enriched rhizosphere, improved soil moisture, and enhanced nutrient availability under controlled drainage condition (Kaur et al., 2020). Moreover, the CDN2 and CDN3 exhibited relatively higher activities of SOD, POD, and CAT (Table 2), indicating better reactive oxygen species scavenging ability in plants treated with CD when N fertilizer was reduced by 10%-20%. This is parallel with the findings of Meng et al. (2023) that the sufficient water supply treatment mediated cotton growth at reduced nitrogen fertilization by enhancing photosynthesis and the activities of nitrogen metabolism enzymes. Additionally, optimal water and nitrogen management strategies up-regulate activities of antioxidant enzyme by enhancing the expression of related genes (Ozcubukcu et al., 2014). These highlight a coupling effect between water and nitrogen fertilizer, achieving both ‘regulating water with fertilizer’ and ‘promoting fertilizer with water’.

Through accumulation to augment the water-holding capacity of cells and safeguard the structure of biological membranes, the content of soluble protein is frequently employed as an indicator for detecting the abiotic stress-resistance capabilities of plants. The photosynthetic capacity can be denoted by the levels of P_n (Li et al., 2020). In the measured growth stages, CD led to an increase in P_n (Figure 4) and soluble protein content (Figure 5). This implies that controlled drainage facilitates the improvement of metabolic activities and the enhancement of photosynthetic capability, laying a solid foundation for shoot biomass accumulation (Qi et al., 2024). A high P_n was closely related with the enhanced soluble protein content, LAI, and SPAD values due to improved soil moisture content (Wang et al., 2016). Besides, an enhanced LAI was consistently accompanied by a higher leaf water content (Li et al., 2010). The elevated water status can suppress the production of ABA, resulting in a high stomatal conductance in leaves (Bahrun et al., 2002), and consequently, high P_n levels (Kang and Zhang, 2004). Alternatively, an obvious positive correlation existed between the activity of nitrogen-related metabolism enzymes and the root physiological characteristics in plants (Fu et al., 2024). CD optimized the rhizosphere soil environment (Youssef et al., 2023), which enhanced the root vitality (Qi et al., 2023). This phenomenon is corroborated by the relatively higher nitrogen accumulation in plants treated with CD (Qi and Zhu, 2023). Moreover, the CDN2 and CDN3 treatments had relatively high P_n and soluble protein content (Figures 4, 5), suggesting that CD can stabilize the plant’s photosynthetic capability with a 10%-20% reduction in normal nitrogen fertilizer input.

In this research, all the decreased N treatments led to a significant reduction in seed cotton yield under FD, whereas only the N4 significantly decreased seed cotton yield under CD (Figure 7). This suggests a positive interaction between CD and 10%-20% reduced nitrogen fertilization (N2 and N3) regarding cotton yield. One possible reason is that CD enhanced the soil moisture status in the plough layer (Supplementary Figure S1). Under conditions of ample water supply, reduced nitrogen fertilizer application was beneficial for increasing the nitrification rate and decreasing the denitrification level (Bateman and Baggs, 2005). As a result, nitrogen losses through emission, leaching, or runoff from crop fields were reduced (Ju et al., 2009; Qi and Zhu, 2023). Alternatively, at the N2 and N3 levels, CD exhibited elevated LAI, SPAD value, P_n, SOD, POD, and CAT activities, as well as high soluble protein content (Figures 2-5; Table 2), while demonstrating low MDA content (Figure 6). These contributed to the relatively high cotton yield. Besides, CD had a comparable number of bolls, boll weight, and lint percentage at N2 and N3 (Qi and Zhu, 2023). Consistently, enhanced water management practices (such as water-saving irrigation) can partly offset the adverse effects of reduced nitrogen fertilizer rates on plant growth, thus stabilizing crop yields (Xu et al., 2018; Hu et al., 2023). This is in line with previous findings and indicates that controlled drainage adjusts the soil water environment and/or nutrient availability to improve crop yield (Jouni et al., 2018; Dou et al., 2021; Tanga et al., 2025). Nevertheless, it has been shown that drainage patterns have no impacts on the growth and yields of maize and sugar beet (Awale et al., 2015) and may even cause a reduction in maize yield (Youssef et al., 2023). These contradictions might be associated with differences in drainage patterns, weather conditions, soil fertility, crop types, etc (Kaur et al., 2020). Indeed, the reasons are still unknown and require further exploration.

In the future, the underlying mechanisms by which CD contributes to a relatively high cotton yield with a 80%-90% of normal nitrogen fertilization should be investigated from the perspective of physio-ecological characteristics (including dry weight, volume, length, surface area, oxidation activity, and the content of indole-3-acetic acid) in root and soil microbial community structure. Moreover, effects of title drainage (a more popular drainage method and it is implemented by artificially raising the outlet elevation of a subsurface drainage system) and nitrogen application rates on crop growth and development merits a further study. Furthermore, as climate patterns undergo changes, specifically with the rise in growing-season temperatures and the unpredictable distribution of precipitation, the efficacy of CD in managing soil water and nutrient storage to ensure optimal crop utilization becomes increasingly significant.