1. Introduction
With the rapid development of modern agriculture, the cultivation technology of solar greenhouses was used to improve vegetable yield, quality and water use efficiency, and had gradually attracted much attention from farmers, especially in northern China [
1]. Many greenhouses have a semi-enclosed structure, in which crops are greatly influenced by meteorological factors and the soil environment [
2,
3]. The root system is an important organ for the transfer of soil moisture and crop water; mastering the root distribution characteristics and water absorption law can well explain the response mechanism of crops to water and environmental stress.
Crop root distribution characteristics in greenhouses have been reported in some previous studies. For example, the root distribution characteristics of pepper and tomato grown in greenhouse were studied by Zapata-Sierra et al. (2021) [
4], who indicated that 90% of the root system was concentrated in the range of 0–9 cm (pepper) and 0–11 cm (tomato). Different irrigation amounts would lead to different water conditions around the roots [
5,
6], thus affecting the growth and distribution of crop roots. Insufficient irrigation would inhibit tomato root growth, while excessive irrigation may lead to a waste of water resources [
7,
8]. Wu et al. (1999) [
9] studied various crop root distribution characteristics and found that the relative root length density and relative soil depth can be described as probabilistic functions. Zuo et al. (2006) [
10] built a relative root length density distribution model for winter wheat by using the normalization method. Thereafter, Ning et al. (2019) [
11] compared the normalized model with the exponential distribution model, linear distribution model and piecewise distribution model, and indicated that the root length density calculated by the normalized model was in good agreement with the measurements, with an error of 4.0%. Some studies have also reported that the normalized method can better express the spatial distribution characteristics of roots, and calculate root length density based on these theories. However, these studies mainly focus on crops grown under field conditions, and few reports focus on vegetables grown in greenhouse conditions.
In recent years, some studies have tried to analyze the root distribution under different ventilation modes, and have concluded that ventilation is also one of the important factors affecting root growth in greenhouse conditions. It is known that crop roots need soil oxygen to maintain respiration and growth properly [
12] and that most oxygen (O
2) is obtained directly through diffusive gas exchange from the atmosphere to the inter-root soil. Ventilation can increase the concentration of oxygen in greenhouse. Ventilation can also reduce the physiological stress of the crop by changing the temperature and humidity in the greenhouse [
13]. At the same time, suitable temperature and humidity would reduce the production of crop pests and diseases [
14], hence promoting crop growth. In addition, the effect of different drip irrigation amounts on tomato root distribution was also studied [
15,
16], and the results indicated that mild water stress favored the growth of deep root. Ullah et al. (2021) [
17] reported that root length and root surface area can significantly increase by reducing the irrigation amount. Mild water stress could promote root growth and plant development, which is beneficial for improving water use efficiency [
18].
One study found that ventilation and irrigation affected crop root growth and distribution by altering the greenhouse environment and soil structure, and suitable ventilation mode promoted root growth [
19], while another study found that excessive irrigation led to poor root zone permeability, thus limiting root development [
20]. In addition, crop root growth indicators, such as root length, root diameter, root surface area and root volume, also changed with ventilation and irrigation management. Previous studies on root systems mainly focused on the distribution patterns of crop root length density in the profile [
21]. Root length density represents the total length of the root system per unit volume of soil and reflects the number of roots. Relative root length density represents the proportion of root length distributed at different relative depths and it is used to describe the relative distribution of root length density. Novak (1994) [
22] built an exponential distribution model of root length density by analyzing the horizontal and vertical distribution of the underground root system. Yang et al. (2009) [
23] established the relative root length density distribution model of wheat by polynomial fitting and verified it through measured and simulated water content with high accuracy. Zuo et al. (2004) [
24] established a nonlinear equation with four parameters to simulate the root distribution. Wu et al. (1999) [
9] expressed the plant root distribution in terms of a third-order polynomial equation. However, these models only described the distribution of root systems in the horizontal and vertical directions [
25,
26] and had not considered the test conditions. Therefore, test variables were added to the simulation of root distribution to obtain a more accurate model of root distribution.
From the above research results, we know that ventilation and irrigation amount affect root growth simultaneously in greenhouse conditions. Crop root growth and distribution also affect fruit yield and water use efficiency [
27]. However, few studies have focused on the interaction effect of ventilation and irrigation on root distribution, and root length density models under interactive conditions are also lacking. Therefore, two years of studies were conducted in a solar greenhouse to achieve the following objectives: (1) analyze the variations in soil water content, temperature and meteorological factors under different ventilation and irrigation amounts; (2) explore the root distribution and root length density of tomato with drip irrigation under different combinations; (3) establish the relative root length density distribution model and evaluate its performance; and (4) explore the effect of an optimal combination of ventilation and water conditions on tomato yield.
4. Materials and Methods
4.1. Experimental Site and Design
The experiment was conducted from March to July in 2020 and 2021 in a solar greenhouse at the Xinxiang Integrated Experimental Base of the Chinese Academy of Agricultural Sciences (35°9′ N, 113°5′ E, 78.7 m above sea level). The greenhouse walls are made of brick and concrete construction, with an area of 510 m2. The direction of the greenhouse is east–west, and it sinks 0.5 m. A steel frame construction is used to support the roof of the greenhouse and is covered with a 0.2 mm thick polyethylene drip-free film to maintain the air temperature inside. Meanwhile, 5 cm thick insulation quilts are used to maintain warmth at the seedling stage. There are three vents, one on the roof (60 m × 30 cm) and another on the bottom of the south side (60 m × 1.5 m). The soil in the greenhouse at a depth of 0–100 cm is a silt loam, including 16.3% clay, 77.1% silt and 6.6% sand. Mean field water capacity and wilting water content at a soil depth of 0–100 cm are 0.31 and 0.11 cm3 cm−3, respectively, with an average bulk density of 1.59 g cm−3.
Tomato seedlings “
Solanum lycopersicum L.
c.v. Jinpeng M6” were transplanted on 4 March 2020 and 7 March 2021. The size of the experimental plot was 8.0 m long and 2.2 m wide, and a wide (65 cm) and narrow (45 cm) row planting mode was used, with an interval of 30 cm. The planting density was 5.7 trees m
−2. Drip irrigation was used, with a drip head flow rate of 1.1 L h
−1. Each plot was replicated three times, and six treatments were designed in blocks with the ventilation mode as the main treatment and the irrigation amount as the vice treatment. Here, three ventilation treatments were set:
TR (open the roof vents only),
TRS (open both the roof and south vents),
TS (open the south vents only), and two irrigation treatments were set according to the cumulative water evaporation (
Ep) from a standard 20 cm evaporation pan (20 cm diameter and 11 cm deep):
K0.9 (0.9
Ep) and
K0.5 (0.5
Ep). The irrigation events were performed based on the average surface evaporation of the three ventilation treatments. According to the results of our previous research [
49], soil water content of 0–60 cm accounts for 80–90% and 60–65% of field water capacity, respectively, under an irrigation amount of 0.9
Ep and 0.5
Ep. The evaporation pan was placed 30 cm above the crop canopy and adjusted according to tomato plants’ development. The evaporation water amount was measured every day at 8:00 using a measuring cylinder with an accuracy of 0.1 mm, and 20 mm of distilled water was refilled after the measurement. When the accumulated water evaporation reaches 20 ± 2 mm, irrigation events was conducted [
50]. The irrigation amount (
Ir) was calculated according to Equation (3).
where
Ir is the irrigation amount, mm;
Ep is the accumulated evaporation, mm; and
φ is the water surface evaporation coefficient.
A water meter with an accuracy of 0.001 m3 was installed at the head of each plot to precisely control the irrigation amount. A supplementary irrigation amount of 20 mm was performed by drip irrigation after transplanting to maintain the seedlings alive. In this study, 112 kg hm−2 urea (containing 46% N), 150 kg hm−2 potassium sulfate (containing 50% K2O), and 120 kg hm−2 superphosphate (containing 14% P2O5) were used as base fertilizers and ploughed to a depth of 16 cm with a rotary spade. Thereafter, differential pressure fertilizer tanks were used for topdressing urea at 18.8 kg hm−2 and potassium sulfate at 25 kg hm−2. The fertilizer was applied four times when the first, second, third and fourth truss fruits began to expand. The agronomic practices (e.g., topping, spraying, fruit thinning) were the same as those used locally. The irrigation amounts were 247.5 and 245.7 mm (K0.9) and 137.5 and 136.5 mm (K0.5) in 2020 and 2021, respectively.
4.2. Measurement
4.2.1. Environmental Factors
Air speed in the greenhouse was monitored by using an anemometer (Wind Sonic, Gill, UK) at the vents with an accuracy of ±0.02 m s−1. Data were collected every 5 s, and the 15 min average was recorded in a CR1000 data logger (Campbell Scientific Inc., Logan, UT, USA). The air temperature and relative humidity were measured by using an automatic climate station (CS215, Campbell Scientific, Inc, Monterrey, CA, USA) with accuracies of 0.02 °C and 0.18 °C, and 15 min averages were calculated and stored.
4.2.2. Soil Water Content
A ZL6 cloud data collector (METER Group, USA) with an accuracy of 1 × 10
−6% was used to determine the water content of the soil layer at 0, 10, 20, 30, 40 and 60 cm in the middle of two drip heads of the same drip tape [
51], with data collected automatically every 15 min.
4.2.3. Soil Temperature
An eight-channel soil temperature logger (JL-04, Ningbo, China) with an accuracy of 0.1 °C was used to monitor soil temperatures at 0, 10, 20, 30, 40 and 60 cm, with data collected automatically at 30 min intervals.
4.2.4. Root Distribution
At the end of picking, root systems were removed from the soil depth between 0 and 60 cm at different orientations by using a layered segmental soil auger with an auger diameter of 7 cm. The soil enclosing the roots was removed at intervals of 10 cm within and between rows (
Figure 7). Tomato roots were placed in mesh bags, rinsed and scanned into JPG files using a 4800 (H) × 9600 (V) dpi (MRS-9600TFU2L, WANSHEN, China) scanner and analyzed for morphological characteristics, root length and surface area image were analyzed by software (Win RHIZO Pro2004 b, Canada). Root length density (RLD, cm cm
−3) was calculated according to Equation (4):
where RLD is the root length density, cm cm
−3; RL is the root length in different soil layers, cm and V is the root auger volume (384.85 cm
3).
4.2.5. Yield
Twenty plants were selected in the middle of each plot to measure yield, and this was repeated 3 times. When the tomato fruits were picked, an electronic balance with an accuracy of 0.005 kg was used to weigh the tomato and record the number to calculate tomato yield.
4.3. Root Length Density Distribution Model
To facilitate the modeling of tomato root length density under different treatments, the normalization method [
9] was used to convert the penetration depth of tomato roots at different soil layers to a standardized root depth in the range of 0 to 1. The root length density was expressed as a generalization of the relative root length density by using the following equation:
where Z
r is the standardized root depth, between 0 and 1, dimensionless; Z
i is the depth of the rooted soil layer, cm; Z
max is the maximum rooting depth, cm, obtained for the root-free soil layer, and the maximum rooting depth in this experiment is 80 cm; NRLD (Z
r) is the relative root length density value, dimensionless; and RLD (Z
r) is the root length density value at Z
r, cm cm
−3.
A third-order polynomial was used to fit the mean values of NRLD (Z
r) at each lateral position at the relative sampling depth (Z
r) based on previous studies [
9,
52]. The equation is as follows:
where
R0,
R1 and
R2 are polynomial parameters and
R3 represents the theoretical value of the NRLD at the surface.
4.4. Model Evaluation
The regression fitting method was used to model the relative root length density distribution for the 2021 root data, and the model was validated using the 2020 measured data. Microsoft Excel 2010 and SPSS 26.0 were used for data processing and graphing. Model simulations were carried out using 1stOpt software, and the model was evaluated using the coefficient of determination (
R2), root mean square error (RMSE) and normalized root mean squared error (NRMSE).
where N is the number of samples; Y
i is the measured values;
is the simulated values;
is the measured mean value; and
is the simulated mean value.
R2 is close to 1, the better the correlation is; RMSE can indicate the average difference between the simulated and observed values, and the closer it is to 0, the smaller the deviation is. NRMSE indicates how good the model simulation performance is; when NRMSE < 10%, the model simulation performance is excellent; when 10% ≤ NRMSE < 20%, the model simulation performance is good; when 20% ≤ NRMSE < 30%, the model simulation performance is average; and when NRMSE ≥ 30%, the simulation performance is considered poor.
5. Conclusions
In this study, we investigated the effects of different ventilation and irrigation amounts on air environment, soil water and temperature conditions and the root distribution of greenhouse tomatoes under drip irrigation. The main conclusions were as follows.
The ventilation and irrigation amount had a significant effect on soil water, temperature and meteorological factors. Ventilation mainly had a significant effect on meteorological factors. TRS was more beneficial to greenhouse gas exchange than TR and TS. Ta and RH were affected by the ventilation treatments in the following manner: TR > TS > TRS. The ventilation and irrigation amount had an interaction effect on soil water and temperature.
The root distribution differences were substantially influenced by both the irrigation amount and ventilation. However, consistent trends of decreasing root length density with deepening of the soil layer were observed below the 10 cm level. Thereafter, a relative root length density distribution model was established according to the following principles: firstly, the relationship between NRLD and relative depth of soil profile was a third-order polynomial; secondly, the coefficient of cubic term (R0) had a bivariate quadratic polynomial relationship with the irrigation amount and air speed. The RMSE between the simulated and measured values of NRLD at K0.9 and K0.5 were 0.20, 0.23 and 0.27 in 2020 and 0.31, 0.23 and 0.28 in 2021, respectively, indicating that performance of the model was perfect.
The irrigation amount and ventilation had significant effects on RLD and yield. The combined treatment of TRS and K0.9 was the most beneficial to increase tomato yield.