Effects of Irrigation Techniques on Yield, Quality and Water Use Eciency of Greenhouse Grown Celery (Apium graveolens L.)

Background: Traditional irrigation methods in protected vegetable production such as furrow irrigation result in low water use eciency. New techniques, such as drip irrigation, micro-sprinkling irrigation have been developed for improving water use eciency. However, these techniques have not been tested in greenhouse celery production. In this study, three different irrigation techniques micro-sprinkler irrigation (MS), furrow irrigation under plastic lm mulching (PF) and micro-sprinkler irrigation under the plastic lm mulching (MSP) were investigated whether the three techniques can improve the yield, quality and water use eciency of greenhouse-grown celery, compared to furrow irrigation (FI). Results: The individual plant weight of celery was higher under MS, PF and MSP than under FI in both autumn season crop (AC) and spring season crop (SC), compared to FI. In AC and SC, the economic yield of celery increases under MSP by 54.18% and 49.55%, the economic yield of celery increases under PF by 30.37% and 34.10%. The irrigation amount of MSP was 151.69 and 179.91 m 3 667 m -2 in AC and SC, which was 23.13% and 27.27% lower than that of FI. The irrigation amount of PF was 151.69 and 196.78 m 3 667 m -2 in AC and SC, which was 23.13% and 20.45% lower than that of FI. PF and MSP reduced the irrigation amount of celery cultivation in greenhouse, and soil evaporation content. Conclusions: In short, MSP and PF promoted the growth and yield of celery in greenhouse with improved quality and water use eciency.


Background
In China, the area of facility cultivation is increasing, especially in horticultural production [1] . Su cient irrigation guarantees crop yields. However, some outdated techniques, such as diffuse irrigation and furrow irrigation are still used for facility cultivation in China, resulting in low water use e ciency. There is a need for the development of new irrigation techniques for water saving [2] . Nowadays, several irrigation techniques have been veri ed to decrease the irrigation amount without hampering yields. For example, micro-spray irrigation results in the increase of tomato yield 27% and water saving 21% by improving the root growth with improved air humidity and microclimate [3] . The micro-tube, drip and micro-sprinkler irrigation improve the mean head weight, mean head diameter and plant height of cabbage and reduce irrigation water for cabbage by 61.44%, 59.28% and 36.82% respectively, compared to surface furrow irrigation due to it had better retention of soil moisture without drought or excess water supply [4] .
Subsurface drip irrigation improved nutrient uptakes by onion roots, leading to the highest onion yield, compared to surface furrow irrigation [5] . Su cient irrigation maintains the crop yield, while it had a negative effect on the fruit quality [6] . For example, the content of soluble sugar and organic acid decreased with the increase of irrigation amount in muskmelon [7] and watermelon [8] .
Apart from using advanced irrigation techniques, mulching applications can effectively modify the hydrothermal conditions in micro-environment for plant growth. Plastic lm mulching may reduce ineffective evaporation by reducing surface air ow [9] , increase soil temperature and improve soil physical properties by reducing wind-candles and keeping the soil loose [10] . Moreover, it can inhibit weeds by reducing the light intensity beneath the plastic lm and decrease the occurrence of diseases and insect pests due to its re ective properties [11,12] . With such bene ts, plastic lm mulching promotes the growth of wheat and maize, resulting in improved crop yield and water use e ciency [13,14] . Sub-lm micro-spraying, where the micro-spraying is implemented under plastic lm mulching, increases water use e ciency in lettuce compared to traditional irrigation method like furrow irrigation [15] . Meanwhile, plastic lm mulching can further reduce the soil water evaporation directly in the vertical direction [16] .
Both black polyethylene mulch and wheat straw mulch improved the fruit yield, fruit size, dry matter, total leaf area, and nutrient concentrations in leaves of cucumber [17] .
Water demands of crops refers to the sum of crop transpiration, crop water content and ground evaporation between plants under appropriate water conditions. Therefore, water demands of crop depend on environmental factors such as temperature [18] , and it varies during the different growing stages. The intensity and duration of water stress had effects on crop growth. Water de cit slow the growth and development of plants and reduce the leaf area of pepper [19] . The pepper yield decreased by 4.5% when the irrigation amount was 75% of the total irrigation amount [22] . Therefore, it is critical to optimize the allocation of irrigation water resources, which directly re ects in the amount of crop water demand [20,21] .
Celery (Apium graveolens L.) is a chimonophilous, biennial crop of the Umbelliferae, rich in vitamins, dietary ber and minerals [23] . It is sensitive to changes in soil moisture [24] . The soil moisture has a great in uence on its growth, development, yield and quality [25] . The purpose of this study is to investigate whether irrigation techniques such as spray irrigation can improve the yield, quality and water use e ciency of greenhouse grown celery in spring and autumn season, by celery growth, water allocation and nutrient contents.

Effects of water irrigation techniques on WUE
The WUE e and WUE b were affected statistically signi cant from irrigation techniques (p < 0.05) in both AC and SC. WUE e , and WUE b were all signi cantly higher than of the control (FI), indicating that the irrigation techniques improved WUE, and the MSP treatment had highest WUE. In AC (Fig. 2), the WUE e in MS, PF and MSP treatments were signi cantly higher than that of the control (FI), of which were 30.06%, 74.47% and 94.57%, respectively. The WUE b were signi cantly higher than that of the control (FI), of which were 28.74%, 72.79% and 92.99%, respectively. In SC (Fig. 3), the WUE e in MS, PF and MSP treatments were signi cantly higher than that of the control (FI), of which were 30.15%, 58.12% and 89.01%, respectively.
The WUE b were signi cantly higher than that of the control (FI), of which were 29.05%, 58.18% and 86.77%, respectively. Effects of water irrigation techniques on plant growth In AC, there was no signi cant difference in plant height between different treatments until the peak growth stage of leaf cluster (35 days after transplanting, from September 4, 2015 to November 11, 2015) (Fig. 4). The celery heights in PF and MSP started to be higher than those in FI and MS from August 30, 2015 to the harvest time (p < 0.05)). In SC, there was no signi cant difference in plant height between different treatments until the peak growth stage of leaf cluster (40 days after planting, from April 12, 2016 to May 31, 2016) (Fig. 4). The celery heights on April 12 and May 1, 2016 was higher in PF and MSP than that in FI (p < 0.05). The celery heights in PF and MSP started to be higher than those in FI and MS from April 12, 2016 to the harvest time (p < 0.05)).
In AC, the dry weights of the whole plant was lowest in FI, the order of whole plant weights was MSP > PF > MS > FI, with signi cant difference (p < 0.05) between treatments and control (FI) ( Table 3). The aboveground dry weights was lowest in FI, the order of aboveground dry weights was MSP > PF > MS > FI, with signi cant difference (p < 0.05) between treatments and control (FI) ( Table 3). The underground dry weight was lower in FI than, those in MSP,PF and MS (p < 0.05) ( Table 3). Under different irrigation methods, the root-shoot ratio (R/S) of PF and MSP was larger than that of FI (p < 0.05) ( Table 3). In SC, the dry weights of the whole plant was lowest in FI, the order of whole plant weights was PF > MSP > MS > FI, with signi cant difference (p < 0.05) between treatments and control (FI) ( Table 3). The aboveground dry weights was lowest in FI, the order of aboveground dry weights was MSP > PF > MS > FI, with signi cant difference (p < 0.05) between treatments and control (FI) ( Table 3). The underground dry weight was lower in FI than, those in MSP,PF and MS (p < 0.05) ( Table 3). Under different irrigation methods, the R/S of MS, PF and MSP was lower than that of FI (p < 0.05) ( Table 3). The premise of selecting irrigation method is to ensure the yield and quality of plants, and economic output is the main indicator to measure the yield and economic bene t of celery. Effects of water irrigation techniques on celery root system The RL, RSA, and RV were affected statistically signi cant from irrigation techniques (p < 0.05) in both AC and SC. In AC, the RL of MS, PF, MSP were 20.26%, 33.31% and 45.45% higher than that of control (FI), respectively, the RSA was 2.83%, 0.70% and 18.34% higher than that of control (FI), respectively, the RV was 11.14%, 32.27%, 34.77% higher than that of control (FI), respectively (Table 4). In SC, the RL of MS, PF, MSP were 5.90%, 34.18% and 49.03% higher than that of control (FI), respectively, the RSA was − 3.40%, 50.42% and 60.49% higher than that of control (FI), respectively, the RV was 14.58%, 17.59% and 97.45% higher than that of control (FI), respectively (Table 4). Effects of water irrigation techniques on soluble sugar, vitamin C and ber contents of celery The contents of soluble sugar and vitamin C in the celery were highest in MSP, followed by MS and PF, and was lowest in FI in both AC and SC, while the contents were increased to different extents in two growing seasons. The soluble sugar contents were 44.64%, 106.96% and 141.74% higher in MS, PF and Discussion MSP treatments respectively than that in the control (FI) in AC (Table 5), and 4.19%, 33.26% and 59.69% higher in SC (Table 5). The vitamin C contents were 3.85%, 26.93%, and 50.01% higher in MS, PF and MSP treatments respectively than that in the control (FI) in AC (Table 5), and 24.97%, 58.94%, and 79.42% higher in SC ( Table 5).
The total ber contents of celery in different treatments was not signi cantly different in both seasons. However, the hemicellulose contents of the inner petiole were lower in MSP than that in FI (p < 0.05) in AC ( Table 5). The hemicellulose contents in the outer petiole were 8.23%, 11.34%, and 11.49% lower in MS, PF and MSP treatments respectively than that of the control (FI) in AC (  (Table 5).
This study mainly investigated the effects of plastic lm mulching and micro-sprinkling irrigation techniques on the WUE, growth, quality, total yield and water distribution of celery in greenhouse of AC and SC. It was found that plastic lm mulching and micro-sprinkling irrigation could improve WUE (Fig. 2,  3) and promote the growth (Fig. 4) of celery. In addition, these techniques improved the quality (Table 5) and promote root growth of celery (Table 4).
Micro-sprinkler and plastic lm mulching accelerated celery growth (Table 3), which showed transverse thickening and longitudinal growth. Previous study showed that lm mulching signi cantly increased onion yield, compared to bare soil. Irrigation water is also reduced by the presence of mulch covers [26] . In both AC and SC, the plant height of the two treatments covered with plastic lm was higher than that of the other two treatments (p < 0.05) in harvest. This increased height could be ascribed to increased soil temperature and soil moisture content with full lm mulching. With full lm mulching, soil temperature and soil moisture content increased more than those under partial lm mulching [9] . The plastic lm mulching can improve soil properties and increase crop yield [27,28] . Similarly, the irrigation techniques improved the yield of celery and WUE. the MSP and PF had high total yield and WUE due to it has better root system (Table 4) both in AC and SC, the roots absorbed more nutrients and substances. In addition, plastic lm mulching can reduces soil moisture evaporation, improving WUE. Previous research has indicated that the combined plastic lm mulching-micro-sprinkler irrigation was the most suitable irrigation approach for increasing lettuce yield and it had best WUE [15] .
The quality of celery was better when covered with plastic lm than that uncovered by plastic lm, MSP and PF had higher plant height and individual weight, this was probably due to it had higher transpiration content, which had improvement of the water and mineral nutrients transported in celery (Table 5) and tomato [29] . Plastic lm mulching can improve soil quality [30] and prevent water evaporation [31] , which were bene cial to accumulation of plant biomass (Table 3). This study shows that PF and MSP have obviously improved the celery's soluble sugar, vitamin C, reduce the content of cellulose, hemicellulose contents (Table 5), related to the promotion on the growth of the root and the absorption of nutrients. These promotions under the plastic lm mulching and micro-sprinkling have been ascribed to the increase of soil temperature and decrease of the soil bulk density [32] . Root water absorption capacity is the key factor determining the balance of transpiration and plant water status, while root growth and distribution are closely related to soil moisture and nutrient status [33] . In this study, MSP and PF had better root system, it had promote the growth and improve quality of celery.

Conclusions
Micro-sprinkler and plastic lm covering can increase root growth. Meanwhile, it promoted the growth of celery in both aboveground and underground parts, improved the quality of celery in both AC and SC, increased the total yield and WUE of celery. All in all, in this study, it is considered that MSP and PF treatment is the best treatment under the conditions of this experiment, it promotes celery growth, improves quality, and is most effective in increasing yield and water use e ciency.

Field sites and experimental design
The experiment was conducted in the greenhouse at the Experimental Station, China Agricultural University, Beijing (39˚54N', 116˚23E'), China, during June 2015 to October 2016. The experiment design was complete random block design with three replicates. Four irrigation treatments were furrow irrigation (FI), micro-sprinkler irrigation (MS), furrow irrigation under plastic lm mulching (PF) and micro-sprinkler irrigation under the plastic lm mulching (MSP) (Fig. 1). Moisture transducer (Shengshixintong Technology, Beijing, China) was used to monitor soil moisture. The irrigation started when soil moisture lowered to 60% of eld capacity, and the total irrigation in four treatments are shown in Table 1. The irrigation water given each time was assumed to increase the soil moisture to 80% of eld capacity. Thus, the irrigation amount for each time was 15.21 m 3 667 m − 2 in AC and 16.87 m 3 667 m − 2 in SC, calculated with the following formula [15] :  (1) where r is the soil bulk density (g cm − 3 ), p is the soil wetting ratio with a value of 100%, h is the planned wetting layer (0.3 m), θ f is the eld capacity (%), q 1 and q 2 are 80%, 60%, respectively, s is the area of 667 m 2 , η is the water use coe cient, taken as a value of 1 in this experiment.
The celery (variety 'Ventura') was germinated in greenhouse use 128-plug tray (one plug with one seed) with vegetable seedling substrate (charcoal, vermiculite, slag and perlite are mixed by volume ratio 20: 20:  (Fig. 1). Plastic lm inserted vertically 50 cm into the soil pro le between adjacent plots to prevent water runoff to adjacent plots. 3000 kg 667 m − 2 decomposed chicken manure (the total content of N, P and K ≥ 4%, total organic matter content ≥ 30%) and 150 kg compound fertilizer 667 m − 2 (the contents of N, P 2 O 5 , K 2 O were 15%, 10% and 15% respectively) were applied before planting. The celery was harvested on November 11, 2015 and on May 31, 2016. The detailed soil physical and chemical characteristics are shown in Table 2. In the treatments with micro-spraying irrigation technique, main rigid pipes were with a diameter of 50 mm and micro-sprinkler pipes were with a diameter of 28.6 mm, a hole with 0.1 mm diameter (Figure ). In the MSP treatment, the same pipes were used as in the MS treatment and placed under the plastic lm. The plastic lm used in PF and MSP treatments was an ordinary black agricultural plastic lm with a thickness of 0.01 mm (Figure 1). Five soil cores on raised bed were taken in each plot using the soil-drilling in an S-shape. Soil was thoroughly mixed, and about 1 kg of soil was sieved through a 1-mm mesh and afterwards air-dried [34,35] .

The soil and plant sampling and analysis
Plant samples were collected using the ve-spot method . Plant height (from ground to the highest point of plant) was measured by ruler (accuracy is 1 mm). The soluble sugar contents were analyzed by anthrone colorimetry method [37] . 0.5 g of fresh celery leaves were cut into pieces and put into a test tube, to which 5 mL distilled water was added and mixed. After 30 min in a boiling water bath, the supernatant was collected, this step was repeated twice. Extra distilled water was added to ll the solution to 10 mL, then analyzed with a spectrophotometer UV-2102C (Honglang Company, ZhengZhou, China) at 630 nm wavelengths.
For determination of vitamin C [38] , 10 g fresh celery leaves were grinded 60 mL of mixture of metaphosphoric and acetic acid (3% HPO 3 + 8% CH 3 COOH) was added instantly to avoid vitamin C breakdown in air. The dilutions were shaken in shaker reciprocating 190-200 rpm min -1 for one hour using. After that, 2 mL of starch indicator was added to it and the solution was immediately titrated with standard solution of iodine I 2 until the solution turns the color from initial reddish-brown color to greyishblue.
Fiber content was measured by ANKOM A200i semi-automatic ber analyzer [39] : 0.5 g of petiole dry sample was put into the lter bag. The lter bag was washed with the con gured neutral detergent solution in the ANKOM A200i semi-automatic ber analyzer (ANKOM Technology, New York, America) for 75 min. The lter bag was taken out and rinsed with acetone for 5 min. After being dried naturally, it was put into the oven at 100 o C for 4 hours. The sample was taken out from the previous step to a dryer and cool to room temperature, then recorded the weight of the sample and add acidic washing solution for 60 min, take out the lter bag and rinse with acetone for 5 min, dry naturally. The lter bag was put into the oven at 100 o C for 4 hours, then cooled to room temperature in a dryer and weigh it. The lter bag was immersed in 72% sulfuric acid solution for 3 hours, rinse the lter bag to pH neutral with water, then rinse with acetone for 3 min, dry naturally. The lter bag was put in a 100 o C oven for 4 hours, dry and weighed it after cooling to room temperature.
The root system was measured by scanning method [40] at the end of harvest (November 10, 2015 for AC and on May 30, 2016 for SC). The whole root system of a celery was collected and rinsed with water, in three replications, then scanned the root system in the EPSOM EXPRESSION 4990 (Seiko Epson Corporation, ChangYe, Germany). The data of root length (RL), root surface area (RSA), root volume (RV) were analyzed by WinRHIZO (Regent Instrument Inc, Ville de Québec, QC Canada).

Calculation of WUE
Economic and biological of WUE (WUE e and WUE b , respectively) were calculated with the following formulas: where EY is the economic yield (kg), including the fresh weight of the stems and leaves in celery, BY is the biological yield (kg), including the fresh weight of the roots, stems, and leaves in celery, and I is the irrigation content (m 3 ).

Data analysis
The data were analyzed by Excel and SPSS 13.0, and data from each sampling event were analyzed separately. The data were subjected to analysis of variance (ANOVA) and Duncan's least signi cant difference (LSD) test to determine the signi cance, and the signi cance was set as p < 0.05. The homogeneity of variances was checked by Bartlett tests, and the normality of the variables was tested by the Shapiro-Wilk test. Availability of data and materials All data were generated or analyzed during this study are included in this published article and its additional les.

Competing interests
The authors declare that they have no con ict of interest.

Funding
This study was supported by Beijing Leafy Vegetables Innovation Team of Modern Agro-industry Technology Research System (BAIC07-2020) and The Construction of Beijing Science and Technology Innovation and Service Capacity in Top Subjects (CEFF-PXM2019_014207_000032). The funding bodies were not involved in the design of the study, collection, analysis, and interpretation of data, and in writing the manuscript.
Authors' contributions WSL, WQ and JXT conceived and designed research. WSL, LC and JXT conducted experiments and analyzed data. WSL and XY con gured the structure of the manuscript. WSL wrote the manuscript. WSL, LC, XY and WYX revised the paper. All authors have read and approved the manuscript.   Effects of water irrigation techniques on WUE in SC: WUEe is WUE of economic yield and WUEb is WUE of biological yield. Values are means (n = 3), different lower case letters (a, b, c, d) indicate the signi cant differences at p < 0.05 according to Duncan's least signi cant difference (LSD) test.

Figure 4
Page 18/19 Effects of different irrigation techniques on plant height: AC (a), SC (b). Values are means ± standard deviation (n = 3), different lower case letters (a, b, c) indicate the signi cant differences at p < 0.05 according to Duncan's least signi cant difference (LSD) test.

Figure 5
Effects of different irrigation techniques on total yield of celery: AC (a), SC (b), EY is the economic yield (kg), BY is the biological yield (kg). Values are means ± standard deviation (n = 3), different lower case