Irrigation Regime Optimization Plays a Critically Important Role in Plastic-Shed Vegetable Production to Mitigate Short-Term and Future N Leaching Pollution

Abstract

In northern China, plastic-shed vegetable production significantly contributes to nitrogen (N)-induced groundwater eutrophication due to excessive fertilization and irrigation. However, the impact of optimized farming practices on N leaching has seldom been systematically examined. We conducted a four-season field study to evaluate the impacts of optimal farming measures on tomato yield, water percolation, N concentration in leachate, and total N (TN) leaching. The treatments included conventional fertilization and flood irrigation (CON), fertilization decreased by 20% and flood irrigation (OPT1) or drip fertigation (OPT2), fertilization decreased by 30% and drip fertigation (OPT3), and no fertilization with flood irrigation (CK). Compared with the CON treatment, the optimal treatments significantly reduced annual TN leaching by 9.92–50.7% without affecting tomato yield (57.1–98.2 t ha⁻¹ for CON and 48.1–106 t ha⁻¹ for three optimal treatments). Drip irrigation contributed 73.8–79.0% to the mitigation of TN leaching. The N originating from soil and irrigation water exhibited a similar contribution to TN leaching (45.4–58.6%) to that of fertilizer N. The daily TN leaching at the basal fertilization stage was much greater than that at the top-dressing stage, due to over-fertilization. Optimizing fertilization, particularly basal fertilization, in combination with drip irrigation could substantially reduce N leaching in plastic-shed vegetable production. Other optimal practices, such as decision support systems (DSSs) and fertilizer amendments, could also be investigated to further mitigate the N leaching.

1. Introduction

Vegetables have been increasingly produced in plastic sheds over the past three decades to meet the increasing vegetable supply and consumption demands in China and worldwide [1]. The total plantation acreage of vegetables in China reached 22 × 10⁶ ha in 2021 [2,3] and approximately 22.5% of vegetables were produced in plastic sheds [3,4].

Large amounts of fertilizer and irrigation water are applied in plastic-shed vegetable production to ensure high yields and favorable incomes for farmers [5]. The average nitrogen (N) fertilizer rate in plastic-shed vegetable production ranges from 550 to 2000 kg N ha⁻¹ a⁻¹ in China [6,7], four times the N input rate in greenhouse vegetable production in Europe [8] and four to seven times the vegetable N uptake level [6,9]. The amount of conventional flood irrigation in plastic-shed vegetable production in China ranges from 60 to 120 mm per irrigation event and the total irrigation rate ranges from 1220 to 2244 mm a⁻¹ [1,10], much higher than that in greenhouse vegetable production in Europe [11] and cereal crop production in China [12]. Excessive fertilization and irrigation in plastic-shed vegetable production cause high N losses, such as N leaching into deeper soil or groundwater [13,14,15], and these losses are the main reasons for the increase in groundwater N concentration in northern China [16]. In Shandong Province, where plastic sheds for vegetable production were first established during the 1990s and have been intensively operated, the groundwater nitrate N (NO₃⁻-N) concentration increased to 6.6–471 mg L⁻¹ in 2020 and the concentration in more than 85.7% of the tested groundwater exceeded the drinking water standard (10 mg N L⁻¹) [17]. Since the 2010s, environmental and agricultural departments have optimized farming practices to mitigate this type of nonpoint source pollution, especially N leaching from plastic sheds [18,19]. However, the major factor causing significant N leaching from vegetable production in plastic sheds is not yet fully understood [20].

N leaching in plastic-shed vegetable production increases exponentially [6] or linearly [21] with the increasing fertilizer N rate, indicating that fertilizer N reduction is, therefore, one of the main practices for inhibiting N leaching [6]. Compared with the application of chemical fertilizer alone, the combined application of chemical and organic fertilizers is also considered to inhibit N leaching [21,22] as organic fertilizer decomposes and releases N slowly [23]. However, the impact of organic fertilization on N leaching remains controversial: it can either decrease because N is immobilized with increasing soil organic matter (SOM) and cation exchange capacity (CEC) [24,25] or it can increase due to the extremely high accumulation of organic N (ON) via the continuously high application rate of organic fertilizer [26,27]. High organic fertilization before transplanting is a common farming practice during the autumn–winter (AW) season, especially in plastic-shed vegetable production in northern China [4,28]; therefore, it is necessary to investigate the effect of optimizing organic fertilization on N leaching in vegetable production throughout the year.

Irrigation greatly affects N leaching [29]. A positive linear relationship was observed between N leaching and irrigation in plastic-shed vegetable production [9,30]. Compared with flood irrigation, water and fertilizer are directly and precisely supplied to crop roots via drip irrigation [31,32,33], thus reducing water percolation and the N concentration in leachate [5]. However, studies have also demonstrated that, compared with flood irrigation, drip irrigation causes an increase in the N concentration in leachate because of the lower water percolation [33,34]. Therefore, the influences of drip irrigation on water percolation and the N concentration in leachate in plastic-shed vegetable production must be clarified.

Fertilization and irrigation are the two most important factors that jointly influence N leaching in vegetable production [33]; however, most previous studies focused on either fertilization or irrigation [35] while few studies aimed to compare the contributions of fertilization and irrigation to N leaching or to analyze their interactive effects. We explored the impacts of fertilization and/or irrigation regime optimization via a two-year experiment of highly intensive plastic-shed vegetable production in China and investigated N leaching during different fertilization periods. We aimed to examine the efficacies of optimizing farming practices, such as the fertilization rate and type, irrigation regime and amount, on N leaching, and the underlying mechanisms, which could provide technical support for the optimization of plastic-shed vegetable production.

2. Materials and Methods

2.1. Experimental Design

This study was performed in Shouguang County (36.84° N, 118.93° E), Shandong Province, China (Figure 1). The study area exhibits a warm temperate continental monsoon climate with a mean annual precipitation of 550 mm and a mean annual air temperature of 12.7 °C [36]. The experiment was performed between August 2017 and July 2019. There are two tomato cultivation seasons per year, namely, the AW season from August to March of the next year and the winter–spring (WS) season from March to June. Conventional greenhouse soil was employed for tomato and cucumber production for approximately 12 years before our experimental study.

A plastic shed (90 m (length) × 12 m (width)) covered with polyethylene was used to produce tomatoes. To ensure tomato growth in winter without heating, the greenhouse encompasses a north-facing back wall and the top is covered with a 1 cm layer of waterproof felt at night in winter. The plastic shed was reestablished in 2011. The upper 0–20 cm of soil exhibited a pH of 7.7 (determined via a 2.5:1 water:soil ratio), a bulk density of 1.50 g cm⁻³, a total N (TN) content of 1.31 g kg⁻¹, an inorganic N (IN) content of 50.8 mg kg⁻¹, a total phosphorus (P) content of 1.67 g kg⁻¹, an Olsen P content of 267 mg kg⁻¹, and a soil organic carbon (C) content of 10.5 g kg⁻¹.

Five treatments, with three replicates per treatment, for a total of fifteen plots, were randomly distributed at the experimental farm (Figure S1). The length of each plot was 11.5 m and the width was 3.5 m (40.3 m²) (Figure S2). Three furrows were established and five rows of tomato plants were planted in each plot. The distance between the furrows’ ridges was 0.66 m and the spacing between the rows in a furrow was 0.45 m. In total, 22 tomato plants were planted in each row (Figure S2). Each plot was surrounded by a 1.5 m high polyethylene film barrier that was buried 1.2 m deep within the soil to prevent surface runoff between adjacent plots.

Under the conventional farming practice (CON) treatment, during the AW season, 780 kg N ha⁻¹, i.e., 65% of the seasonal N input, and 480 kg P₂O₅ ha⁻¹ were applied as basal fertilizers (Figures S3 and S4). All of the N fertilizer and 73.3% of the calcium superphosphate were applied as organic fertilizer (91.7 t ha⁻¹, compost of rice husks and chicken manure). For top dressing, 420 kg N ha⁻¹, i.e., 35% of the seasonal N input, and 420 kg P₂O₅ ha⁻¹ were applied. The fertilizers (urea, mono-ammonium phosphate, and potassium nitrate) were applied by dissolving/mixing them in water through flood irrigation. During the WS season, the same rate of top-dressing fertilization as in the AW season was applied through flood irrigation but no organic fertilizer was applied. Compared with those under the CON treatment, 20% less fertilizer and the same amount of water were applied under the OPT1 treatment and flood irrigation was employed (similar to the CON treatment) during both the AW and WS seasons. Under the OPT2 treatment, the same fertilizer rate as that under the OPT1 treatment was employed, but drip irrigation was adopted, and 27.8–37.3% less water was provided under the OPT2 treatment than under the CON treatment. Under the OPT3 treatment, 30% less fertilizer was applied than under the CON treatment while drip irrigation was also employed (similar to the OPT2 treatment). A control treatment (CK) with flood irrigation was also established but no fertilization was applied.

During the tomato growing season, all the plots were irrigated after tomato planting. According to the needs of tomato plants, there were 9–13 irrigation events each season at 7–15 d intervals [37,38]. Three to five irrigation events occurred before the first top dressing. From the first top dressing to the end of the tomato growing season, there were 6 to 8 top-dressing events and irrigation was conducted simultaneously with fertilization. The interval between each irrigation event was also adjusted according to weather conditions (Figure S5), such as shortening the interval days between irrigations during continuous sunny days or high temperatures and lengthening the interval days between irrigations during continuous cloudy days or low temperatures. To prevent root rot and the occurrence and spread of diseases in tomato plants, no irrigation was applied on rainy or cloudy days [37]. In general, there were 13, 11, 13, and 11 irrigation events during the 1st, 2nd, 3rd, and 4th tomato growth seasons, respectively (Figure S3).

Irrigation was performed until the 0–30 cm soil layer was thoroughly moistened, which was 50–60 mm of water per irrigation event [4,10]. Under the flood irrigation treatments (CON, OPT1, and CK treatments), the duration of each irrigation event was approximately 10 min. In the drip-irrigated plots (the OPT2 and OPT3 treatments), irrigation lasted 3 to 4 h during each irrigation event. Five irrigation belts, each with a diameter of 2 cm, were placed parallel to the tomato plant rows with a 5 cm distance between the belts and the plants. The water outlets on each irrigation belt were 20 cm apart from each other. The irrigation areas of flood irrigation and drip irrigation were 45% and 14% of the total plot area, respectively. For flood-irrigated plots, the irrigation water amount was 683, 490, 473, and 689 mm during the 1st, 2nd, 3rd, and 4th tomato growth seasons, respectively. Accordingly, the drip-irrigated plots received 34.0%, 27.8%, 34.7%, and 37.3% less water, respectively, than the flood-irrigated plots (the CK, CON, and OPT1 treatments) (Figure S3).

During the AW season (Season 1) from 2017 to 2018, the dates of transplanting tomato seedlings and removing tomato plant residues were 31 August 2017 and 24 March 2018, respectively. For the WS season (Season 2) of 2018, the transplanting of tomato seedlings occurred on 19 April 2018 and the removing of tomato plant residues was conducted on 10 August 2018. Similarly, seedling transplantation occurred on 26 August 2018 during the AW season (Season 3) and on 19 March 2019 during the WS season (Season 4) and tomato plant residues were removed on 16 March and 2 July 2019, respectively. The number of days from transplantation to the first top dressing (defined as the basal fertilization stage) was 27, 19, 46, and 32 during the 1st, 2nd, 3rd, and 4th tomato growth seasons, respectively. The durations of the top-dressing stages (from the first top dressing to the removal of tomato plant residues) were 177, 92, 155, and 72, respectively. The plants were treated with pesticides to control pests and insects, similar to local farming practices. In Seasons 1 and 3, the round tomato cultivar Qidali was used; Qianxi, a cherry tomato variety, was planted in Season 2; and Maofen, another round tomato variety, was employed in Season 4. Each tomato plant was supported individually by a nylon rope wrapped around the stem and secured to a wire rope strung between the supporting posts at a height of 2.5 m above the ground (Figure 1).

2.2. N Leaching Measurement and Calculation

As the tomato roots are mainly distributed in a 0–1 m soil layer [5,39], the soil leachate originating from each plot was collected via a microporous ceramic suction cup at a 100 cm soil depth [33]. Two tensiometers were installed at depths of 90 and 110 cm to measure the soil water potential. Tensiometer data were recorded every day between 08:00 and 09:00. Percolating water was collected by lowering the pressure in the suction cup to −70 kPa via a hand vacuum pump on Days 1 and 3 after irrigation. Leachate was collected in a 50 mL centrifuge tube and stored at −20 °C for analysis. The nitrate N (NN) and ammonia N (AN) concentrations in the collected leachate were determined via a TRAACS2000 continuous flow analyzer (Bran and Luebbe, Norderstedt, Germany). The leachate and irrigation water aliquots were subjected to alkaline persulfate oxidation, after which the TN concentration was determined via dual-wavelength (220 and 275 nm) ultraviolet spectrophotometry [1]. The ON concentration was defined as the difference between the TN and inorganic N (NN + AN) concentrations.

Seasonal water percolation (QT, mm) at a depth of 100 cm was calculated via Equation (1):

$$Q_T={\int }_0^T{10q}_t\mathrm{d}t,$$

(1)

where T denotes the entire tomato growing period (d) and q_t denotes the soil water flux (cm d⁻¹) at the 100 cm depth, with t denoting the sampling day. Moreover, q_t was calculated via Darcy’s law:

$$q_t=K\left(H_{90}\left(t\right)-H_{110}\left(t\right)\right)/∆\mathrm{Z},$$

(2)

where H₉₀(t) and H₁₁₀(t) are the soil water potentials (cm) at depths of 90 and 110 cm, respectively, as monitored by the installed tensiometers; ΔZ is the depth difference between the two tensiometers, i.e., 20 cm; and K is the unsaturated hydraulic conductivity (cm d⁻¹) calculated via the Van Genuchten Equation (3) [40]:

$$\mathrm{K}={\mathrm{K}}_{\mathrm{s}}{\displaystyle \frac{{\left\{1 -{\left(\mathsf{\alpha }\left|\mathsf{\phi }\right|\right)}^{\mathrm{n}-1}{\left[1+{\left(\mathsf{\alpha }\left|\mathsf{\phi }\right|\right)}^{\mathrm{n}}\right]}^{-\mathrm{m}}\right\}}^2}{{\left[1+{\left(\mathsf{\alpha }\left|\mathsf{\phi }\right|\right)}^{\mathrm{n}}\right]}^{\mathrm{m}/2}}}$$

(3)

where K_s is the hydraulic conductivity at saturation (5.95 cm d⁻¹) calculated via the constant-head permeameter method [41]; α, n, and m are parameters derived from the soil moisture characteristic curve (α = 11.09, n = 1.13, and m = 0.11 in this study); and φ is the soil matric potential (cm H₂O) measured via a tensiometer [40].

The seasonal N loss through leaching (NL, kg N ha⁻¹) was calculated via Equation (4):

$$\mathrm{NL}={\int }_0^T{C_{t}q}_t/10\mathrm{d}t,$$

(4)

where C_t denotes the TN, IN, or ON concentration (mg N L⁻¹) and q_t is the soil water flux (cm d⁻¹) at the 100 cm depth on the t-th day.

The partial factor productivity (PFP) from applied N was calculated via Equation (5):

$$\mathit{PFP} =(Y/FN) \times 100\%,$$

(5)

where Y is tomato yield (kg ha⁻¹) and F_N is the total fertilizer N input (kg N ha⁻¹).

The water use efficiency was calculated via Equation (6):

$$\mathit{WUE} =(Y/W) \times 100\%,$$

(6)

where W is irrigation water quantity (m³ ha⁻¹).

The fertilizer TN leaching factor was calculated via Equation (7):

$$\mathrm{TN} \mathrm{leaching} \mathrm{factor} =\left(\right({\mathit{NL}}_F{- \mathit{NL}}_0)/FN) \times 100\%,$$

(7)

where NL_F and NL₀ are the cumulative TN leaching losses (kg N ha⁻¹) from the N-fertilized and unfertilized plots, respectively.

The apparent TN leaching factor, i.e., without deducting the amount of TN leaching from the CK treatment, was calculated via Equation (8):

$$\mathrm{Apparent} \mathrm{TN} \mathrm{leaching} \mathrm{factor} =({\mathit{NL}}_F/FN) \times 100\%,$$

(8)

The percolation ratio was calculated via Equation (9):

$$\mathrm{Percolation} \mathrm{ratio} =(Q_T/I) \times 100\%,$$

(9)

where I is the amount of irrigation water added (mm).

The leached TN under the CK treatment mainly originated from the soil N and irrigation water N stocks. The contribution rates of soil N and water N (CSW) and fertilizer N (CFN) to the leached TN were calculated via Equations (10) and (11), respectively:

$$\mathit{CSW} (\%)={\mathit{NL}}_0/{\mathit{NL}}_{\mathrm{F}} \times 100\%$$

(10)

$$\mathit{CFN} (\%)={({\mathit{NL}}_{\mathrm{F}}-\mathit{NL}}_0)/{\mathit{NL}}_{\mathrm{F}} \times 100\%$$

(11)

The mitigation of annual TN leaching under the OPT1 treatment was regarded as the effect of applying 20% less fertilizer than that under the CON treatment. The contributions of decreasing fertilization and switching from flood irrigation to drip irrigation to the mitigation of annual TN leaching under the OPT2 and OPT3 treatments were calculated via Equations (12)–(15):

$${\mathit{CRF}}_{\mathrm{OPT}2} (\%)=({\mathit{NL}}_{\mathrm{CON}}-{\mathit{NL}}_{\mathrm{OPT}1})/({\mathit{NL}}_{\mathrm{CON}}-{\mathit{NL}}_{\mathrm{OPT}2})\times 100\%$$

(12)

$${\mathit{CDI}}_{\mathrm{OPT}2} (\%)=(1-{\mathit{CRF}}_{\mathrm{OPT}2}) \times 100\%$$

(13)

$${\mathit{CDI}}_{\mathrm{OPT}3} (\%)=\left(\right({\mathit{NL}}_{\mathrm{OPT}1}{-\mathit{NL}}_{\mathrm{OPT}2})/({\mathit{NL}}_{\mathrm{CON}}-{\mathit{NL}}_{\mathrm{OPT}3}\left)\right) \times 100\%$$

(14)

$${\mathit{CRF}}_{\mathrm{OPT}3 }(\%)=(1-{\mathit{CDI}}_{\mathrm{OPT}3}) \times 100\%$$

(15)

In Equations (12)–(15), CRF_OPT2 and CDI_OPT2 are the contributions of decreasing fertilization and switching from flood irrigation to drip irrigation, respectively, to the mitigation of annual TN leaching under the OPT2 treatment; CRF_OPT3 and CDI_OPT3 are the contributions of decreasing fertilization and switching from flood irrigation to drip irrigation, respectively, to the mitigation of annual TN leaching under the OPT3 treatment; and NL_CON, NL_OPT1, NL_OPT2, and NL_OPT3 are the annual TN leaching losses under the CON, OPT1, OPT2, and OPT3 treatments, respectively.

2.3. Statistical Analysis

Statistical analysis was performed via SPSS 20.0 for Windows (IBM, Armonk, NY, USA). The seasonal and annual N leaching losses, water percolation amount, percolation ratio, N concentration in leachate, and TN leaching factor for the various treatments and different fertilizer types were investigated via analysis of variance. Repeated measures analysis of variance was performed via the GLIMMIX procedure in SAS 9.2 (SAS Institute, Cary, NC, USA) to analyze the effects of fertilization and irrigation on the water flux and the N concentration in leachate using the treatment as the fixed effect and the block, plot, and block × sampling time as random factors and to investigate the effect of the fertilizer type on the TN leaching factor using the fertilizer type as the fixed effect and the block, plot, and block × treatment as random factors.

Regression analysis was performed to determine the best-fitting models for the relationships between the amount of water percolation and the irrigation water quantity and between the mean TN concentration in leachate and the fertilizer N application rate via flood and drip irrigation. The adjusted R² (adj. R²) values of the regression equations were used to identify the best models. Multiple linear regression analysis was performed to assess the contributions of the irrigation water quantity, fertilizer N rate, water percolation amount, and the mean TN concentration in leachate to TN leaching. F-tests were employed to assess the regression at the p < 0.05 level.

A structural equation model (SEM) was applied to identify the relationships among the irrigation interval, irrigation water quantity, inorganic and organic fertilizer inputs, and TN leaching loss. The least significant difference method was used to compare the means of different datasets. The data are presented as the mean ± standard error. SEM analysis was performed via AMOS 22.0 (Amos Development Corporation, Chicago, IL, USA).

3. Results

3.1. Tomato Yield and Water Percolation

Regarding 2-year (four seasons) tomato production, there were no significant tomato yield differences between the optimized treatments (the fertilizer input was decreased by 20% or 30% and the amount of irrigation water was reduced by 27.8–37.3%, i.e., the OPT1, OPT2, and OPT3 treatments) and the CON treatment (Figure S6). The tomato yields under the CON, OPT1, OPT2, and OPT3 treatments during the first season ranged from 95.9 to 106 t ha⁻¹, which were significantly greater (38.9–99.2%) than those during the second, third, and fourth seasons. This difference was due to the short tomato growth period and different tomato varieties during the second and fourth seasons, as well as the occurrence of severe disease during the third season.

The PFP among the three optimized treatments exhibited no significant differences, but the PFP of each optimized treatment was significantly higher than that of CON (Figure S7). The PFP during the third season was lower than those during the other three seasons, mainly due to lower tomato yields compared to the first season and a higher fertilizer N input compared to the second and fourth seasons. Optimized fertilization did not change WUE when comparing with OPT1 and CON (Figure S8) but drip irrigation (OPT2 and OPT3) significantly increased the WUE by 13.1–53.4% compared with flood irrigation (CON and OPT1).

Fertilization did not significantly affect water percolation, except that the CON treatment exhibited significantly more water percolation in Seasons 1 and 2 than that under the CK treatment (

Table 1. Seasonal water percolation rates and percolation ratios during two years of plastic-shed tomato production.

	2017–2018 Autumn–Winter Season	2018 Winter–Spring Season	2018–2019 Autumn–Winter Season	2019 Winter–Spring Season
	(1st Season)	(2nd Season)	(3rd Season)	(4th Season)
Water percolation (mm)
CK	117 ± 12.9 b	82.4 ± 3.81 b	90.1 ± 4.65 a	108 ± 3.59 a
CON	130 ± 2.11 a	90.4 ± 1.49 a	92.9 ± 0.73 a	120 ± 7.91 a
OPT1	126 ± 0.46 ab	85.3 ± 2.42 ab	95.1 ± 5.35 a	116 ± 4.48 a
OPT2	58.8 ± 0.29 c	56.0 ± 1.97 c	50.0 ± 0.32 b	60.1 ± 0.24 b
OPT3	59.8 ± 1.76 c	56.9 ± 1.28 c	49.9 ± 1.20 b	65.2 ± 0.76 b
Water percolation ratio (%)
CK	17.1 ± 1.89 a	16.8 ± 0.78 ab	19.1 ± 0.98 a	15.7 ± 0.52 ab
CON	19.1 ± 0.31 a	18.5 ± 0.30 a	19.6 ± 0.15 a	17.4 ± 1.15 a
OPT1	18.4 ± 0.07 a	17.4 ± 0.50 ab	20.1 ± 1.13 a	16.8 ± 0.65 a
OPT2	13.1 ± 0.06 b	15.8 ± 0.56 b	16.2 ± 0.10 b	13.9 ± 0.06 b
OPT3	13.3 ± 0.39 b	16.1 ± 0.36 b	16.2 ± 0.39 b	15.1 ± 0.18 b

Note: The data are presented as the mean ± standard error (n = 3). The different letters indicate statistical significance (at p < 0.05) of the difference between the seasonal totals of the experimental treatments. CON = conventional fertilization and flood irrigation, OPT1 = fertilization decreased by 20% and flood irrigation, OPT2 = fertilization decreased by 20% and drip irrigation, OPT3 = fertilization decreased by 30% and drip irrigation, CK = no fertilization and flood irrigation.

). The total amount of water percolation was 34.4–52.3% lower (a mean of 44%, p < 0.05) under drip irrigation (OPT2 and OPT3) than under flood irrigation (CK, CON, and OPT1; seasonal water percolation ranging from 82.4 to 130 mm). Compared with that under flood irrigation, drip irrigation significantly reduced the water percolation ratio by 9.1–27.6% (Table 1).

3.2. N Leaching

The N (TN, NN, AN, and ON) concentrations in the leachate during the AW season, i.e., Seasons 1 and 3, were significantly higher than those during the WS season, i.e., Seasons 2 and 4 (Figure 2 and Figures S9 and S10). The NN concentration accounted for 57.7–89.9% (77.4% on average) of the total TN concentration in the leachate during the four tomato seasons. Over the 2 years of tomato production, the mean ammonium AN concentration in the leachate was <7 mg N L⁻¹, accounting for <6% of the inorganic N (IN) concentration in the leachate. Under the fertilization treatments, the ON concentration in the leachate accounted for 19.5–26.1% of the TN concentration during the AW season and 12.5–18.5% during the WS season.

During Seasons 1 and 4, the seasonal mean TN concentrations in the leachate did not significantly differ among the OPT1, OPT2, and OPT3 treatments but were 2.8–10.5% lower than that under the CON treatment (292–386 mg N L⁻¹) (Figure 2a). During Seasons 2 and 3, the seasonal mean TN concentrations in the leachate under the OPT1 and OPT3 treatments were 2.5–13.6% lower (p < 0.05) than that under the CON treatment. However, compared with those under the CON and OPT1 treatments, drip irrigation (OPT2) significantly increased the average seasonal TN concentration by 9.2–18.8% and 11.6–30.9% during Seasons 2 and 3, respectively. The seasonal NN, AN, and ON concentrations in the leachate under all the treatments exhibited trends similar to those of the TN concentration (Figure 2b, Figure 2c and Figure 2d, respectively).

During year-round tomato production, N leaching mainly (63.6–74.7% of the annual TN leached) occurred during the AW season (

Table 2. Seasonal and annual N leaching losses and total N (TN) leaching factors at the 100-cm soil depth during the two-year period of plastic-shed tomato production.

	2017–2018 Autumn–Winter Season	2018 Winter–Spring Season	2018–2019 Autumn–Winter Season	2019 Winter–Spring Season	Annual Average
	(1st Season)	(2nd Season)	(3rd Season)	(4th Season)
TN leaching loss (kg N ha−1)
CK	151 ± 17.3 d	86.3 ± 3.98 c	156 ± 7.74 c	106 ± 5.12 b	250 ± 12.0 c
CON	382 ± 4.41 a	142 ± 3.09 a	351 ± 3.23 a	154 ± 11.2 a	514 ± 4.64 a
OPT1	356 ± 1.32 b	120 ± 3.27 b	313 ± 25.5 a	137 ± 4.08 a	463 ± 13.0 b
OPT2	166 ± 1.25 c	94.9 ± 3.98 c	209 ± 3.29 b	72.9 ± 3.68 c	271 ± 6.10 c
OPT3	170 ± 5.79 c	84.8 ± 1.90 c	178 ± 5.13 bc	74.6 ± 0.52 c	251 ± 6.67 c
NN leaching loss (kg N ha−1)
CK	93.7 ± 9.85 d	67.9 ± 3.08 c	122 ± 5.87 c	83.8 ± 6.32 b	183 ± 7.01 d
CON	297 ± 3.46 a	114 ± 2.50 a	253 ± 1.90 a	128 ± 9.22 a	396 ± 4.14 a
OPT1	281 ± 0.05 b	96.4 ± 2.78 b	229 ± 19.0 a	114 ± 3.09 a	360 ± 9.37 b
OPT2	129 ± 0.88 c	74.9 ± 3.49 c	153 ± 2.41 b	61.5 ± 3.09 c	210 ± 4.93 c
OPT3	131 ± 3.64 c	67.8 ± 1.53 c	129 ± 3.60 bc	62.6 ± 0.38 c	195 ± 4.58 cd
AN leaching loss (kg N ha−1)
CK	5.35 ± 0.52 b	3.93 ± 0.19 a	5.28 ± 0.27 b	3.97 ± 0.11 b	9.26 ± 0.51 b
CON	7.63 ± 0.02 a	4.27 ± 0.04 a	6.39 ± 0.01 a	4.92 ± 0.32 a	11.6 ± 0.15 a
OPT1	7.44 ± 0.74 a	4.02 ± 0.18 a	5.94 ± 0.32 a	5.31 ± 0.36 a	11.4 ± 0.44 a
OPT2	3.01 ± 0.04 d	2.72 ± 0.13 b	3.08 ± 0.02 c	2.25 ± 0.07 c	5.53 ± 0.13 c
OPT3	3.31 ± 0.09 c	2.68 ± 0.08 b	2.96 ± 0.11 c	2.71 ± 0.12 c	5.83 ± 0.20 c
ON leaching loss (kg N ha−1)
CK	52.3 ± 7.05 c	14.4 ± 0.79 d	29.3 ± 1.61 d	18.6 ± 1.39 a	57.3 ± 5.09 c
CON	77.5 ± 0.94 a	23.9 ± 0.55 a	91.0 ± 1.32 a	20.3 ± 1.64 a	106 ± 0.36 a
OPT1	67.5 ± 0.53 b	19.9 ± 0.31 b	78.2 ± 6.23 b	17.4 ± 0.63 a	91.5 ± 3.22 b
OPT2	33.1 ± 0.34 e	17.3 ± 0.36 c	52.2 ± 0.86 c	9.1 ± 0.51 b	55.8 ± 1.04 c
OPT3	36.1 ± 2.06 d	14.4 ± 0.29 d	45.5 ± 1.42 c	9.3 ± 0.02 b	52.6 ± 1.89 c
Apparent TN leaching factor (%)
CON	31.9 ± 0.37 b	33.8 ± 0.73 a	29.2 ± 0.27 a	36.6 ± 2.66 a	31.7 ± 0.29 b
OPT1	37.1 ± 0.14 a	35.8 ± 0.97 a	32.6 ± 2.66 a	40.7 ± 1.21 a	35.7 ± 1.01 a
OPT2	17.2 ± 0.13 d	28.3 ± 1.18 b	21.7 ± 0.34 b	21.7 ± 1.09 b	20.9 ± 0.47 c
OPT3	20.3 ± 0.69 c	28.8 ± 0.64 b	21.2 ± 0.61 b	25.4 ± 0.18 b	22.4 ± 0.59 c
Fertilizer TN leaching factor (%)
CON	19.2 ± 0.37 b	13.2 ± 0.73 a	16.2 ± 0.27 a	11.3 ± 2.66 a	16.3 ± 0.29 a
OPT1	21.3 ± 0.14 a	10.1 ± 0.97 b	16.3 ± 2.66 a	9.1 ± 1.21 a	16.4 ± 1.01 a
OPT2	1.47 ± 0.13 c	2.58 ± 1.18 c	5.46 ± 0.34 b	−9.9 ± 1.09 b	1.62 ± 0.47 b
OPT3	2.25 ± 0.69 c	−0.49 ± 0.64 d	2.57 ± 0.61 b	−10.8 ± 0.18 b	0.32 ± 0.59 b

Note: The data are presented as the mean ± standard error (n = 3). The different letters indicate the statistical significance (at p < 0.05) of the difference between the seasonal totals of the experimental treatments. CON = conventional fertilization and flood irrigation, OPT1 = fertilization decreased by 20% and flood irrigation, OPT2 = fertilization decreased by 20% and drip irrigation, OPT3 = fertilization decreased by 30% and drip irrigation, CK = no fertilization and flood irrigation.

). TN leaching under the CON treatment (382, 142, 351, and 154 kg N ha⁻¹ during Seasons 1, 2, 3, and 4, respectively) was significantly greater than that under the other treatments, namely, under the OPT1 treatment, TN leaching decreased by 6.8–15.1%. In contrast, under the OPT2 and OPT3 treatments, TN leaching decreased by 33.0–56.7% during the four tomato seasons. There were no significant TN leaching differences between the OPT2 and OPT3 treatments, although there was an additional 10% fertilizer N reduction under the OPT3 treatment compared with that under the OPT2 treatment.

The apparent annual TN leaching factors under the CON and OPT1 treatments were 55.3% greater (p < 0.05) than those under the two drip irrigation treatments (OPT2 and OPT3) (Table 2). The fertilizer TN leaching factors under the CON treatment were similar to those under the OPT1 treatment and much greater than those under the OPT2 and OPT3 treatments. As the background TN leaching level under the CK treatment was very high, the OPT2 and OPT3 treatments exhibited much lower fertilizer TN leaching factors and even negative values (Table 2). During the 2-year experiment, the contribution rates of soil N and water N to TN leaching (CSW) were 48.7% and 54.3% under the CON and OPT1 treatments, respectively, and the remainder could be attributed to fertilizer N (CFN) (Figure 3). Thus, it was impossible to calculate the relative contributions under the OPT2 and OPT3 treatments, whereas different irrigation practices, i.e., drip irrigation, were implemented under the CK treatment.

The amount of TN leaching under all the treatments at the basal fertilization stage ranged from 18.3 to 161 kg N ha⁻¹ (a monitoring period of 19–46 days, 13.2–30.8% of the entire growth period), accounting for 19.9–46.2% of the overall TN leaching during the entire tomato season (Figure 4a). During the AW season, the amount of daily TN leaching at the basal fertilization stage was approximately 220% greater than that at the top-dressing stage (Figure 4b) while during the WS season, the amount of daily average TN leaching at the basal fertilization stage of the flood irrigation treatments (CK, CON, and OPT1) was approximately 46.3% greater than that at the top-dressing stage. However, there was no significant difference between the basal fertilization stage and the top-dressing stage for the drip irrigation treatments (OPT2 and OPT3).

3.3. Contributions of Reducing Fertilization and Switching Flood Irrigation to Drip Irrigation to N Leaching Mitigation

Significant positive linear relationships were found between the amount of water percolation and the irrigation water quantity (p < 0.001; Figure 5a) and between the seasonal TN concentration and the fertilizer N rate (p < 0.001; Figure 5b) for both the flood and drip irrigation regimes. Water percolation per unit irrigation water input was lower (Figure 5a) under drip irrigation than under flood irrigation but the seasonal TN concentration per unit N fertilizer input exhibited the opposite trend, i.e., a higher level under drip irrigation than under flood irrigation, especially at higher N rates (Figure 5b).

The irrigation water quantity (IWQ) and the fertilizer N (FN) rate explained 77.6% of the total variance in the TN leaching loss (NLL in the equation) (Figure 5c). The corresponding equation was NLL (kg N ha⁻¹) = 0.290 IWQ (mm) + 0.203 FN (kg N ha⁻¹) − 79.3 (adjusted R² = 0.776, p < 0.001, n = 60), indicating that irrigation and fertilization were the two important factors influencing TN leaching. Irrigation affected TN leaching by influencing mainly the water percolation quantity and fertilization, in turn, affected TN leaching by influencing the TN concentration in leachate. Moreover, water percolation significantly negatively influenced the TN concentration (Figure 6a,b). TN leaching was greatly affected by the irrigation water quantity (standardized coefficient β = 0.386 during the AW season and β = 0.371 during the WS season), followed by the fertilizer N rate and irrigation interval (Figure 6c and Figure 6d, respectively).

A decrease in the fertilization rate of 20% was the only reason for the mitigation of TN leaching, according to a comparison of the OPT1 and CON treatments (Figure 7). A comparison of the OPT2 and OPT3 treatments with the CON treatment revealed that a decrease in fertilization of 20% or 30% explained only 21.0% and 26.2%, respectively, of the reduction in TN leaching and most of the reduction in TN leaching (79.0% and 73.8%, respectively) could be explained by the switch from flood irrigation to drip irrigation (Figure 7).

4. Discussion

4.1. Which More Effectively Mitigates N Leaching in Plastic-Shed Vegetable Production: Fertilization Reduction or Switching the Irrigation Regime?

The reduction in excessive N fertilization in plastic-shed vegetable production significantly mitigates N leaching [21,42,43]. However, our findings indicated that a reduction in the fertilizer N input alone imposed a limited mitigating effect on N leaching. Compared with that under the CON treatment (514 ± 4.64 kg N ha⁻¹), the amount of annual TN leaching decreased by 9.9% under the OPT1 treatment (fertilizer N was reduced by 20%) (Table 2), and similar yield and higher PFP were maintained, which agrees with the findings of Zhao et al. [39]. The mitigation effect was not notable as the fertilizer N input under the OPT1 treatment in our study was still high (1296 kg N ha⁻¹ a⁻¹), much higher than the critical fertilizer N rate of 1000 kg N ha⁻¹ a⁻¹ determined in the meta-analysis by Qasim et al. [6]. Under both the OPT2 and OPT3 treatments, in which the same irrigation regime was adopted, an additional 10% reduction in the fertilizer N input under the OPT3 treatment also did not significantly reduce the amount of annual TN leaching compared with that under the OPT2 treatment (251 vs. 271 kg N ha⁻¹). This demonstrates that irrigation optimization, instead of fertilizer optimization alone, should be considered in plastic-shed vegetable production [33,44,45].

In our study, drip irrigation (OPT2 and OPT3) significantly decreased water percolation by 34.3–53.3% (Table 1) compared with that under flood irrigation (CON, OPT1, and CK). Although the leachate TN concentration during Seasons 2 and 3 increased under the OPT2 treatment compared with that under the CON and OPT1 treatments (Figure 2a), the significant reduction in water percolation effectively alleviated this increase and ultimately significantly reduced the overall TN leaching level. Compared with those under drip irrigation, flood irrigation resulted in greater water inputs (60–70 mm per event and a total of 473–689 mm per season), which far exceeded the tomato demand and soil water-holding capacity [5]; thus, a large amount of water together with N percolated [4,13,46].

Conforming with the above analysis, we found that the contributions of fertilizer N reduction and the switch from flood irrigation to drip irrigation to the mitigation of TN leaching were 21.0–26.2% and 73.8–79.0%, respectively (Figure 7). The dominant importance of irrigation regime optimization in the mitigation of TN leaching was also supported by the SEM analysis results (Figure 6). In tomato production, the wetted area of drip irrigation (OPT2 and OPT3) was 68.9% smaller than that of flood irrigation (CK, CON, and OPT1), which resulted in greater water retention in the soil around crop roots [44] and maintenance of the tomato yield. In addition, drip irrigation directly and in a timely manner transports water and nutrients to crop roots and increases water and nutrient efficiency both spatially and temporally [4,44] (Figures S7 and S8); therefore, less N percolates out of the root area with water leakage. Compared with flood irrigation, drip irrigation maintained a lower and more stable soil moisture (Figure S11), which significantly restricted the amount of water percolation and contributed the most to the mitigation of N leaching.

4.2. Long-Term and High-Fertilization-Accumulated N in Soil Contributed Equally as Fertilizer N to N Leaching

In this study, both the apparent TN leaching factor (leached TN/fertilizer N input) and the fertilizer TN leaching factor ((leached TN from fertilization treatment − CK)/fertilizer N input) under the flood irrigation treatments (CON and OPT1) were much greater than those under the drip irrigation treatments (OPT2 and OPT3) while the TN leaching under the CK treatment was even greater than that under the OPT2/OPT3 treatment during Seasons 2 and 4. The annual TN leaching rates under the CK treatment were 237 and 262 kg N ha⁻¹, which were much higher than the meta-analysis-based mean value of 91 ± 20 kg N ha⁻¹ reported by Qasim et al. [6]. In conventional tomato production, continuous and high fertilizer N and water inputs cause the accumulation of large amounts of N within the soil [47,48]. The experimental soil in our study has been conventionally used for vegetable production for more than 10 years, indicating that the accumulated N under the CK treatment was relatively high and that the accumulated N was likely to leach into deeper soil [9,47]. After four seasons of tomato cultivation in our study, the IN stock in the 0–100 cm soil profile under the CON treatment increased from 598 ± 106 kg N ha⁻¹ to 723 ± 31 kg N ha⁻¹ (Table S1) due to excessive fertilization and flood irrigation. Under drip irrigation, N did not accumulate in the deeper soil due to the lower water input (Table S1), which reduced the extent of N leaching, which is comparable to the findings of other studies [1,44].

Because of the legacy effect of N accumulation resulting from long-term excessive N fertilization and flood irrigation, the effect of reducing fertilization on decreasing the TN leaching factor was not observed in our short-term study (four seasons of tomato production) (OPT1 vs. CON) and the joint contribution of the soil and water N stocks to N leaching was similar to that of the application of fertilizer N (Figure 3), which conforms with the findings of Li et al. [9]. The high N originating from irrigation water (667 kg N ha⁻¹ a⁻¹) (Table S2), which was pumped from a deep well (120 m) below the plastic shed, also served as the fertilizer N input during long-term vegetable production. These analyses highlighted that optimal water irrigation should be urgently introduced in intensive plastic-shed vegetable production zones within China to mitigate the potential bombing effect of accumulated soil N [49] exacerbated by the flood irrigation regime in the region.

4.3. Optimal Fertilization Timing to Reduce N Leaching in Plastic-Shed Vegetable Production

The appropriate timing of fertilization is also an important measure for reducing N loss [50]. TN leaching during the AW season was significantly greater than that during the WS season (Table 2), mainly because of the higher N fertilizer rate and the longer period of tomato growth, as well as the additional organic fertilizer applied at the basal fertilization stage during the AW season (546–780 kg N ha⁻¹ season⁻¹) (Figure S4). The lower rate of mineralization and slower release of N from organic fertilizers help alleviate N leaching [51]. However, given the limited capacity of soil for nutrient adsorption and fixation [51] and the low nutrient demand of relatively small tomato plants, excessive organic fertilization still resulted in significant N leaching [33,39]. This was especially true for the leaching of organic nitrogen, which accounted for 19.0–26.0% of the overall TN leached during the AW season, but this phenomenon has been largely neglected in most studies [52].

At the basal fertilization stage during the AW season (August–September), the high temperature inside the plastic shed promoted the turnover of soil organic N, which further increased the risk of N leaching [48,52] (Figure 4b). This finding corresponded to the observation that the main N leaching occurred at the early stage of tomato growth [39,53]. Thus, integrating the approaches of switching from flood irrigation to drip irrigation for N leaching and optimizing fertilizer application, especially the rate of organic fertilizer application at the basal fertilization stage, could effectively mitigate N leaching while ensuring high vegetable yields in plastic-shed vegetable production [5,14].

4.4. Implications for Nonpoint Source Pollution Control in Northern China

In conventional plastic-shed vegetable production, farmers mostly apply large amounts of fertilizer and irrigation water, especially in Shandong Province, where most plastic sheds are located [28]. Excessive fertilizer and irrigation water result in a significant loss of N from the vegetable system via leaching. Considering the large acreage of plastic-shed vegetable production in northern China (4.7 × 10⁶ ha) [1], the overall TN leaching from these plastic sheds has reached 2.4 × 10⁹ kg N a⁻¹, which poses a considerable risk to the groundwater quality [16]; moreover, the rapid and sustained increase in the NO₃⁻-N concentration in groundwater has been a growing concern among the public [9]. The optimized operations in our study may reduce TN leaching by 0.24–1.2 × 10⁹ kg N a⁻¹, equivalent to 2.4–12.2% of the annual fertilizer N usage in the EU in 2020 (FAOSTAT, 2021). However, the space for fertilization and irrigation optimization should be studied further, particularly when combined with other technologies, such as decision support systems (DSSs) and tensiometers, in which nutrients and water are supplied to crops promptly [54,55,56]. Additionally, the use of fertilizer inhibitors [57,58,59] and near-zero drainage technologies, such as employing substrates for tomato cultivation, should be explored [60,61,62].

5. Conclusions

Conventional plastic-shed vegetable production with excessive fertilization and irrigation induces high N leaching, thus posing a notable risk to the environment. Our study exhibited that the integration of fertilizer optimization with drip irrigation could significantly reduce N leaching while maintaining a high tomato production yield. And switching from flood irrigation to drip irrigation played a dominant role (>70%) in reducing TN leaching in conventional plastic-shed tomato production. The reduction in TN leaching caused by optimized fertilization was mainly due to the decrease in TN concentration in the leachate. However, the long-term excessive input of fertilizer and water had accumulated much N (the legacy effect); thus, the reduction in N leaching due to optimized fertilization was not as effective as we initially assumed. Drip irrigation reduced water percolation by 44% due to water savings, thus greatly reducing N leaching during the tomato production season and, more importantly, substantially decreasing and inhibiting the movement of N to deeper soil, thereby reducing potential N leaching pollution in the future. Our study also highlighted that high organic fertilization at the basal fertilization stage during the AW season promoted N leaching, especially organic N leaching. Therefore, in addition to optimizing the fertilizer rate and irrigation scheme, farmers should also improve the timing of fertilization. Other farming practices, including but not limited to the application of DSSs, employing a tensiometer for irrigation, the use of fertilizer inhibitors, growing tomatoes using substrate, and other applicable options, should also be studied and introduced, considering the greater space for the mitigation of N leaching in current plastic-shed vegetable production.