A field experiment was conducted in the Tarim River Basin of southern Xinjiang, China (40°53°03″N, 86°56°58″E; 900 m a.s.l.). The mean annual precipitation is 58.0 mm, and annual evaporation is approximately 2500.0 mm (Chen et al., 2018). The study area belongs to warm temperate continental desert climate. Groundwater depth is close to 3 m, and the total dissolved solid ranges from 2 to 7 g/L. Soil texture mainly consists of sandy loam, silt loam, and sand in the 0-100 cm soil depth, with the particle size distribution for clay, silt, and sand ranging from 0.00% to 4.36%, from 15.36% to 57.29%, and from 38.82% to 84.64%, respectively. In addition, the soil bulk density is 1.52 g/cm³. Soil electrical conductivity (EC) before and after the experiments is 3.35 and 0.97 dS/m, respectively (Table 1). And the soil has large amounts of NaCl and Na₂SO₄. Soil pH is 8.03. Irrigation water from the channel has a salinity of 0.7 g/L, where the ions, i.e., Ca²⁺, Na⁺, SO₄^2-, and HCO₃^- are the dominant ions, consisting of 37.10%, 38.20%, 25.30%, and 59.10% of the cations and anions, respectively. Total precipitation amounts during the cotton growth season were 32.4 and 13.8 mm in 2020 and 2021, respectively (Fig. 1). Average daily maximum temperatures were 39.1°C in 2020 and 41.5°C in 2021, and the corresponding average daily minimum temperatures were 7.7°C and 8.1°C, respectively (Fig. 1).

The field experiment was performed in the saline-alkaline soil with no irrigation in winter and spring seasons. There were five leaching amounts (0.0 (W0), 75.0 (W1), 150.0 (W2), 225.0 (W3), and 300.0 mm (W4)) and three leaching periods (seedling stage (P1), seedling and squaring stages (P2), and seedling, squaring, flowering, and boll setting stages (P3)). In addition, a control treatment (CK) with a leaching amount of 300.0 mm through flood irrigation in spring was set up. The details of leaching scheduling are given in Table. 2. Each treatment had three replicates. A HOBO U30 meteorological station (Onset Computer Corporation, Bourne, USA) was installed to record weather data. The regular irrigation amount over the two years was 85.00% crop water requirement for cotton. Experiment began in early June, and the irrigation interval was 9 d. Irrigation was postponed in the event of precipitation, and crops were not irrigated at the end of August (Table 2).

Irrigation amount was calculated by the data obtained through the weather station in the experimental field (Allen et al., 1998):

where ET_c is the crop water requirement for cotton (mm); K_c is the crop coefficient; and ET₀ is the reference crop evapotranspiration (mm). K_c of cotton was 0.75 during vegetative growth period (i.e., seeding and squaring period), 1.15 during flowering boll period, and 0.70 during boll opening period (Hou et al., 2022b). ET₀ was calculated as follows (Allen et al., 1998):

where Δ is the slope of the vapor pressure curve (kPa/°C); R_n is the net solar radiation (MJ/(m•d)); G is the soil heat flux (MJ/(m²•d)); γ is the psychrometric constant (kPa/°C); T_mean is the mean temperature (°C); U₂ is the wind speed at 2 m height (m/s); e_s is the saturation vapor pressure (kPa); and e_a is the actual vapor pressure (kPa).

Threshold of salinity for cotton was 7.7 dS/m (Minhas et al., 2020). Experimental cotton variety was Xinluzhong 67. Irrigation type was mulched drip irrigation. Dripper flow rate was 2.4 L/h and dripper operating pressure was 0.2 MPa. Planting pattern consisted of one 1.06-m-wide strip of plastic film, two drip belts, and four rows. In other words, the width of the two bare areas between films was 46 cm. Two drip irrigation belts were laid under the film at a distance of 30 cm from the edge of film, and four rows of cotton were planted at a distance of 10 cm from the edge of film and to the middle of two drip irrigation belts, respectively. Planting spacing of cotton was 10 cm (Fig. 2). Seeds were planted on 16 April, 2020 and 14 April, 2021, and harvested on 28 September, 2020 and 30 September, 2021, respectively. Phenological phase of cotton in 2020 and 2021 is shown in Table S1. The area of experimental plot was 4.3 m×15.0 m, and each experimental plot had three sheets of plastic mulch. A buffer zone was created between two plots to prevent the effect of lateral soil water movement. We implemented fertilization according to local practices, i.e., 300 kg N/hm², 90 kg P/hm², and 45 kg K/hm². Fertilizers used were urea, ammonium dihydrogen phosphate, and potassium oxide. Fertilizers were applied equally at eight times. Chemical fertilizers were dissolved in differential pressure fertilization tank one day before irrigation. There was one fertilizer tank for each plot. Other agronomic measures such as weeding and pesticide spraying were consistent with local practices.

SPSS v.18.0 software was used for the analysis of variance (ANOVA) for multivariate comparisons. A general linear model had performed to explore the impact of each treatments on PH, SD, LAI, dry matter, yield, water productivity, soil moisture, and salinity by the SPSS software. Least significant difference (LSD) was applied to determine significant differences at P<0.05 level. Sigmaplot v.14.0 software was used to create figures and Microsoft Excel v.2016 software was applied for data analysis.

Variation of soil moisture in the 0-100 cm depth among various leaching amounts and periods in 2020 and 2021 are given in Figures 3 and 4, respectively. Figure 4 showed that mean soil moisture in the 0-40 cm depth increased by 38.26%-52.75% as leaching amount increased from seedling to boll stages. Similarly, as the times of leaching irrigation under the same leaching amount increased, mean soil moisture in the 0-40 cm depth also increased by 38.57%-50.76% and satisfied cotton growth requirement. In 2020, soil moisture for each treatment was the highest at seedling stage, intermediate at the other reproductive stages, and the least at boll opening stage (Fig. 3). In 2021, soil moisture was greater at seedling and boll opening stages and less at budding and boll flowering stage. Notably, average soil moisture values in the 0-100 cm soil depth at boll opening stage in 2020 under W3P1, W4P1, W3P2, W4P2, W3P3 and W4P3 treatments were 45.25%, 43.30%, 17.81%, 28.43%, 18.23% and 27.90% less than those at seedling stage, respectively. However, in 2021, it increased by 13.72%, 14.28%, 10.24%, 7.45% and 4.10% under W3P1, W4P1, W3P2, W3P3, and W4P3 treatments, respectively.

Figures 5 and 6 show soil EC changes in the 0-100 cm depth during the cotton reproduction period among different treatments in 2020 and 2021, respectively. Soil EC under W4P1 treatment decreased by 31.21% compared with that of CK, but there was no significant decreased among W0, W1 and W2 treatments. And soil EC was less in the 0-40 cm depth at the early stage under W3 and W4 treatments. There was some increase in soil EC in the 60-100 cm depth. P1 showed high soil EC at boll flowering stage in 2020 but a low soil EC in 2021. Under P2 treatment, soil EC throughout the growth period was less under W4 and W3 treatments than under W1 and W2 treatments in both 2020 and 2021. Under P3 treatment, soil EC existed throughout the cotton reproduction period probably due to low leaching amount, plentiful sunlight, and intense evaporation. Thus, P3 did not create a suitable water-salinity balance for plant development in the early stages of growth. Additionally, our research showed that 225.0 mm leaching amount at seedling and squaring stages significantly increased cotton yield by 8.00% and decreased soil EC by 11.79% compared with CK.

Figures 7 and 8 show the influences of different treatments on cotton plant height, stem diameter, and LAI. At seedling and squaring stages, plant height increased under CK treatment compared with the other treatments. Plant height increased under W3 and W4, P1 and P2 treatments at flowering, boll, and opening-boll stages (Figs. 7c1-c3). At all growth stages except for seeding stage, plant heights under W3 and W4 treatments were significantly higher than those under the other treatments at each growth stage in 2020 and 2021. Plant height under W4P2 treatment during boll opening stage was 34.18% higher than that of W4P1 and 41.19% higher than that of W4P3 in 2020; the corresponding values were 16.10% and 19.84% in 2021, respectively. In addition, plant height under W4P2 treatment during the boll opening stage was 34.46% greater than that of CK in 2021. As shown in Figure 7a and 7c, W4P2 treatment increased plant height.

Stem diameter increased under W3P2 treatment at flowering, boll, and opening-boll stages. At seedling and squaring stages, stem diameter increased under CK (Figs. 7d1-d3). Variations in stem diameter during each growth stage in 2020 and 2021 were positively correlated with leaching amount. Except for seeding stage, W3 and W4 treatments significantly increased stem diameter. W3P2 and W4P2 treatments produced the greatest stem diameter over the entire growth period. We found that salt leaching irrigation of 225.0 mm in the growing period produced the greatest stem diameter (10.02 mm in 2020 and 12.21 mm in 2021). Meanwhile, CK significantly increased stem diameter in the early stages (Fig. 7d1-d3), which was attributed to the reduction of soil salinity by spring irrigation.

Two-year data showed that LAI of each treatment after salt leaching was greater than that of W0 (Fig. 8). LAI increased in the early flowering stage as leaching amount increased. The maximum LAI was observed mainly in late July and early August. However, a delayed growth period was also observed, particularly under CK treatment (Fig. 8b1-b3). The reason for this may be the cotton plants in the experiment were subjected to salinity stress due to limited effectiveness of small-scale salt leaching or non-salt leaching, so LAI of cotton was small. W3P2 treatment showed greater LAI than W3P1 and W3P3 treatments. This result might be due to the salt leaching at squaring stage, which avoided salt stress to the root system and ensured plant growth.

Both leaching amount and period had highly significant impacts on dry matter weight in 2020 and 2021 (P<0.01). Their interaction also had significant impacts on dry matter (P<0.01; Table 3). Dry matter was significantly greater under W4P2 treatment than under W4P1 and W4P3 treatments in 2020 and 2021 (Table 3). Furthermore, dry matter under W3P2 and W4P2 treatments were 19.36% and 25.11% higher than those of CK in 2021.

Boll number and boll weight per plant decreased as leaching amount decreased. The largest number of bolls per plant (6.25 in 2020 and 7.63 in 2021) was observed under W3P2 treatment, with 33.86% increase in 2021 compared with CK. The largest number of bolls per plant was observed under P2 treatment, with 20.52% and 21.98% increase in 2020 and 2021 compared with P3 (Table 3). The optimal boll weight (CK treatment) was 5.02 g. Boll weight under W1P3 treatment significantly increased compared with W1P2 treatment in 2020 and 2021, and boll weight under W2P2 treatment significantly increased compared with W2P1 and W2P3 treatments. Boll weight under W3P2 treatment significantly increased compared with W3P3 treatment in 2021 (Table 3). Nevertheless, boll weight per plant for the same treatment decreased by 4.98% in 2021 compared with CK.

Both leaching amount and period had highly significant impacts on seed cotton yield (P<0.01) in 2020 and 2021. Leaching amount and period had a significant impact on boll number per plant and boll weight per plant (P<0.05; Table 3). Seed cotton yield in this experiment showed that salt leaching amount and period were directly related. The optimum for seed cotton yield under W3P2 treatment was 8.67% greater than that of CK in 2021. This result suggested that leaching period influenced crop growth. Seed cotton yield increased as leaching amount increased. In addition, yields under W1P1 and W1P2 treatments were significantly greater than that of W1P3 treatment in 2021. In both 2020 and 2021, yield under W2P2 treatment significantly increased as compared with W2P3 treatment. Similar findings were found under W3 and W4 treatments (i.e., yield was significantly greater for these two treatments at P2 than at the other stages) (Table 3).

A significant interaction (P<0.01) between leaching amount and period on water consumption was observed in 2020 and 2021 (Table 4). Water consumption was positively correlated with leaching amount. Water consumption was generally lower in 2021 than in 2020 because of variation in weather parameters (mainly precipitation). Water consumption of leaching amount of 300.0 mm was significantly higher than those of the other treatments. In 2020, under W1 and W2 treatments, water consumption under P3 treatment was significantly greater than those under P1 and P2 treatments. For greater leaching amounts (W3 and W4), water consumption under P1 treatment was significantly greater than those under P2 and P3 treatments in 2020. In 2021, for low leaching amounts (W1 and W2), water consumption under P2 treatment was less than that under P1 and P3 treatments. For greater leaching amounts (W3 and W4 treatments), water consumption under P3 treatment was significantly greater than that under P1 treatment.

Leaching amount and period had highly significant interaction impacts on water productivity (P<0.01; Table 4). Water productivity generally decreased with increasing leaching amount. In 2020, water productivity under W3P2 treatment was 1.17 kg/m³ (the greatest water productivity among all treatments in 2020). In 2021, water productivity under W1P2 treatment was the greatest (1.29 kg/m³) among all treatments. This result was due to the fact that the yield under W1P2 treatment was about 36.59% less than that of W3P2, which used 150.0 mm more leaching water (about three times the average annual precipitation in the area). The optimal water productivity in 2021 increased by 72.00% compared with that of CK. Furthermore, water productivity under W3P2 treatment in 2021 increased by 25.33% compared with that of CK, which showed leaching at seedling and squaring stage could improve water productivity.

There were highly significant interaction effects of leaching amount and period on irrigation water productivity (Table 4; P<0.01). Under P2 treatment, WPI values under W1, W2, and W4 treatments were 36.84%, 29.82%, and 22.80% smaller than that under W3 treatment in 2020, while the corresponding values were 13.64%, 15.79%, and 15.79% in 2021, respectively. The optimal irrigation water productivity under W3P2 treatment increased by 23.38% compared with that of CK. Moreover, under W4 treatment, irrigation water productivity under P2 treatment was 3.90% greater than that for CK, while those under P1 and P3 treatments were 24.68% and 18.18% smaller than that of CK, respectively.

The relationships among dry matter, yield, water productivity, soil EC, and leaching amount are shown in Figure 9. Dry matter accumulation increased as leaching amount increased, and seed cotton yield also tended to increase as leaching amount increased (Fig. 9a and b). However, a greater leaching amount (W4) slightly reduced the available air in the soil. WP had a quadratic relationship with leaching amount, but the maximum water productivity differed between the two years. This result might be the fact that the rainfall in the two years were not similar, i.e., the effective rainfall in 2020 was 2.3 times higher than that in 2021 during the cotton reproductive period. Therefore, the maximum water productivity was shifted backward (Fig. 9c). In Figure 9d, soil EC in the 0-40 cm depth tended to reduce as leaching amount increased (P<0.05). Soil EC in 2021 was significantly less than that in 2020, which revealed that salt leaching in addition to regular irrigation was effective in reducing soil salinity. In general, leaching during the reproductive period could effectively decrease soil salinity and increase cotton production and water productivity.

Salt leaching regulated soil water-salinity dynamics and increased crop production in saline soils (Minhas et al., 2020). Many factors affect water-salinity dynamics, such as meteorology and hydrology (Hosseini and Bailey, 2022; Li et al., 2022), water management (Akramkhanov et al., 2011), and agronomic measures (Meng et al., 2021). In this study, soil moisture inconsistency was realized for the studied two years (Figs. 3 and 4). The reasons for this phenomenon were as followed: there was the rainfall event before sampling (Fig. 1) and soil moisture was lost mainly through soil evaporation (Tugwell-Wootton et al., 2020), when plants were small. Similar results had been confirmed by Zheng et al. (2021). In addition, we found that the soil became saline due to irrigation intervals would be expected to induce water uptake from shallow soil layers, increase unproductive evaporative losses from soil surface, and increase the salt load of soils (Wang et al., 1993; Minhas et al., 2020). These observations indicate that salt leaching was effective in reducing salinity in the 0-40 cm soil depth (Chen et al., 2010; Kang et al., 2012; Barnard et al., 2021). According to Chen et al. (2020), the main root of cotton is in the 0-40 cm soil depth. And salt stress changes the osmotic pressure of soil, which makes it difficult for the root system to take up nutrients and water (Abdelraheem et al., 2019). In Figure 9d, soil EC in the 0-40 cm depth tended to reduce as leaching amount increased, which was consistent with the result of Yang et al. (2019). However, there was some salinity reversion in the deep soil depth (Ren et al., 2021), probably due to the high clay content of loamy soil and the short time interval (generally 2 d after leaching) between salt leaching treatment and soil sample extraction. The short time interval meant that not all salts from the upper soil layer was available to percolate into the lower layer, so salt accumulated in the upper layer. Salinity in 2021 was less than that in 2020 (Fig. 9d), possibly due to the reduction in salinity after leaching and planting multi-year crops (Tan et al., 2008). In addition, the low level of leaching resulted in little salt or water penetrating the deep soil depth, and the deep soil depth was not leached and salt did not enter the shallow groundwater (Kang et al., 2010). Due to plentiful sunlight and intense evaporation in the southern Xinjiang, leached salts dissolved or entrained in the water became evaporation residue between two irrigation events and remained in the upper soil depth. These results agreed with that of Wang et al. (2011). Additionally, our research showed that 225.0 mm leaching amount at seedling and squaring stages significantly increased cotton yield by 8.00% and decreased soil EC by 11.79%, compared with irrigation in spring. This result might be attributed to the fact that cotton growth before squaring stage was mainly vegetative and biomass increased when soil was low in salinity and high in moisture, leading to the greater yield. Furthermore, a single cotton boll had formed when salt leaching occurred at boll setting stage. There was no obvious effect on cotton yield. This result indicated that the cotton boll setting stage was less sensitive to salinity than seedling and squaring stages, which was consistent with the results obtained by Tian et al. (2019). Thus, in the early stage of cotton, a suitable water-salinity balance for plant development was not created.

A suitable plant height for high-yield cotton varied from 90 to 102 cm, depending on plant density and environmental conditions (Rosolem et al., 2013), with LAI peaking at 100 d after sowing (Grantz et al., 1993). Similar values were obtained in this study by leaching salts during growth period. However, LAI of CK reached the maximum in late August. Cotton plants in the experiment were subjected to salinity stress due to limited effectiveness of small-scale salt leaching or non-salt leaching, so LAI of cotton was small. Xiao et al. (2021) investigated salt leaching in southern Xinjiang during growth period and showed that cotton stem diameter was the greatest after three times of salt leaching irrigation (240.0 mm). In contrast, we found that 225.0 mm salt leaching irrigation during growing period produced the greatest stem diameter (10.02 mm in 2020 and 12.21 mm in 2021). The difference can be attributed to the fact that brackish water was used to irrigate cotton (Xiao et al., 2021). Therefore, soil moisture contained a large quantity of salts at the early growth stage, accumulating and resulting in an increase in salinity. Meanwhile, stem diameter significantly increased in the early stages under CK treatment (Fig. 7d) due to the reduction of soil salinity in spring irrigation (Zong et al., 2022). In addition, we also found that dry matter accumulation increased for different salt leaching treatments (Fig. 9a) due to increased leaching water availability from treatments (Ma et al., 2021). As leaching amount increased, salts in the root zone of cotton was leached to a deeper layer, and cotton was thus less salt-stressed, and a suitable soil water-salinity environment for cotton growth was created (Che et al., 2022).

Greater boll weight and increased boll number per plant are important in achieving high yields in cotton (Pettigrew, 2004; Grundy et al., 2020). Liu et al. (2020) found that both boll number and boll weight per plant decreased as leaching amount decreased in Xinjiang, which was consistent with our results. We observed the largest number of bolls per plant (6.25 in 2020 and 7.63 in 2021) under W3P2 treatment, with an increase by 33.86% in 2021 compared with CK. Nevertheless, boll weight per plant for the same treatment decreased by 4.98% in 2021 compared with CK. This result could be due to the decrease in the transfer of photosynthetic products to the buds and bells (Fig. 8). As a result, bell weight decreased (Luo et al., 2008). Irrigation schemes and other agricultural practices significantly affect cotton yield (Hou et al., 2021; Zulfiqar et al., 2021). Yield in this experiment showed that salt leaching amount and period were directly related (Zhang et al., 2021). Yield peaked under W3P2 treatment because leaching before boll setting ensured that crop water requirements were satisfied and salt leaching was effective. This result suggested that leaching period influenced crop growth. Seed cotton yield increased as leaching amount increased because cotton under water and salinity stresses in salt-alkaline soils required more water for salt leaching than that in non-alkaline soils to maintain growth (Rengasamy, 2006). The maximum salt leaching thus occurred under W3 treatment rather than under W2 treatment. Similar result was also found by Lokhande and Reddy (2015).

Cotton water consumption varied because of the changes in climate parameters (expecially precipitation), which was also confirmed by Grismer (2002) and Howell et al. (2004). An appropriate salt leaching amount increased water productivity and irrigation water productivity (Wang et al., 2015; Hou et al., 2021; Xiao et al., 2021). Our results showed that leaching at seedling and squaring stages could improve water productivity and irrigation water productivity. The maximum water productivity was found under W3P2 treatment. However, the inconsistent leaching treatments corresponding to the maximum water productivity (Fig. 9c) might be caused by the fact that the two years were not hydrologically similar. Furthermore, irrigation water productivity decreased with increasing leaching amount under P3 treatment in 2021.