Experimental conditions, setup, and treatments
The muskmelon experiment was conducted in a 108 m long and 8 m wide greenhouse located in Yangling, Shaanxi Province, Northwest China (34°17′N latitude, 108°02′E longitude) from April 24 to July 12, 2014. Tomato experiment was conducted from Oct 18 2014 to May 20, 2015 in the same greenhouse. The climate is semi-arid with a long-term average annual precipitation of 550-650 mm, with the average annual sunshine of 2163.8 hours and 210 frost-free days. The soil texture of root growth area was clay loam with 25.4 % sand, 44.1 % silt, and 30.5% clay. pH was 7.82, dry bulk density 1.35 g cm-3, porosity 49.4 %, and gravimetric field capacity 28.2% (volumetric field capacity 38%). The physical and chemical characteristics of the irrigation water: pH was 7.90, EC was 2.71 ds m-1, total suspended solids were 15 mg L-1, CODMn was 1.2 mg L-1.
The experimental plots consisted of 5.5 m by 0.2 m high flat ridges that were separated by 0.15 m deep furrows. The experimental setup of muskmelon was designed to examine effects on plants under three levels of sub-surface drip irrigation in combination with drip-tubing placed at each of 3 depths in the soil, and 4 levels of artificial soil aeration. The drip irrigation tubing placement depths were 10, 25, and 40 cm below the surface of the ridge designated as Dcm (i.e. D10, D25, and D40). The tomato experiment was examining drip irrigation tube burial depth plus aeration frequency. The main drip irrigation pipe was connected to an air pump, and water and air were supplied to the soil through the subsurface drip irrigation tubes (Figure 7). Drip irrigation tubes (φ16 subsurface drip irrigation pipe, Qinchuan water-saving irrigation equipment engineering Co. Ltd, Yangling, China), 16 mm diameter with emitter spacing of 0.30 m, were buried to the appropriate depth spaced 0.50 m apart along the ridges. The subsurface drip irrigation tubes had a peculiar labyrinth channel structure, so it can promote the uniformity discharge rate of air/water as much as possible. The drip irrigation tubes were connected to a distribution system designed to supply both water and air. Each plot had 35 to 36 emitters.
Muskmelon cultivar Shantian No.1 seeds were sown in commercial seedling plugs (Northwest New Horizon Facilities Agriculture Development Co. Ltd, Yangling Shaanxi, China). Shantian No.1 is one of over 100 cultivars of the Chinese 'Hami' reticulated melon group (Cucumis melo L.). It is an early maturing, monoecious cultivar with separate male and female flowers on the same plant. Each plant produces 4-5 small (300 - 350 g), fragrant, sweet, crisp-fleshed, fruits that are very popular with consumers. The tested cultivar of tomato (Lycopersicon esculentum Mill.) was Fen-Yu-Yang-Gang (Northwest New Horizon Facilities Agriculture Development Co. Ltd, Yangling Shaanxi, China).
Before transplanting, the soil was rototilled, and 120 t/ha of decomposed organic manure (pig and sheep manure), 400 kg/ha of compound fertilizer (18% N, 15% P2O5, and 12% K2O), and 1,500 kg/ha of diammonium phosphate (18% N and 46% P2Oe) were broadcast uniformly in the soil as the basal fertilizer. After 20 days, 26 plugs of muskmelon (or tomato) were transplanted to the experimental plots spaced 0.40 m apart within 2 rows spaced 0.5 m apart on the ridges. To prevent the lateral spread of air and water into adjacent treatments, the plots were separated from each other by a 1.5 m wide empty space. Post-transplanting management practices (i.e. fertilization, agricultural chemicals spraying, fruit pruning etc.) for all plots were consistent with local production practice. Fruiting was not restricted, although some farmers prune the flowers.
Based on the balance between soil air update rate and labor cost, artificial aeration treatments of tomato were none or aeration at 2-day, and 4-day intervals, and interval of aeration for muskmelon were none or aeration at daily, 2-day, and 4-day intervals beginning the first day after transplanting and designated as Ainterval in day (i.e. A1, A2, A4, and AN). For each treatment, 405 L of air was applied to each plot via the drip tubing using a manifold connected to an air compressor. The flow rate for each plot was about 10.2 L min-1. The air which injected via subsurface drip irrigation tubes have a characteristic of high O2 concentration and low CO2 concentration compared with air which originally stored in soil pore space. The injected air/oxygen consumed by soil microorganisms, soil animals, crop roots. Moreover, a fraction of oxygen diffuses to the atmosphere because of the chimney effect. This volume was calculated as:
[Please see the supplementary files section to access the equation.] [44]
where VA was the volume of air injected (liter), S the area of a 30-cm deep cross-section between rows (1,500 cm2 = 50 cm x 30 cm for 2 rows spaced 50 cm apart), L the plot length (550 cm), the soil dry bulk density of soil (1.35 g cm-3), and the soil particle density (2.65g cm-3). = 0.95 an application efficiency coefficient for subsurface drip irrigation.
Irrigation timing and amount demand of tomato experiment is mainly driven by farmers’ perceptions and climatic conditions. For the muskmelon experiment, following recommended production practices the plots were surface-irrigated at the time of transplanting. Soil water content was measured and controlled using a Field TDR 200 soil moisture meter (Spectrum, Aurora, IL, USA). A 60-cm deep probe was installed in the center of each plot. Soil water content was measured at 10-cm intervals down to a depth of 60 cm. The gravimetric water content (qg) averaged over the 0-60 cm at the time of transplanting was measured, and all the plots were surface irrigated with 385 litre (equivalent to 70 mm depth) applied to each plot. Subsequent sub-surface drip irrigation treatment levels, applied at 23 and 60 days after transplanting, were based on the measured gravimetric water content averaged over the 0-60 cm depth of the soil profile on these dates. The location of soil sampling was between two rows of muskmelon (or tomato), each soil sampling with three replicates. Muskmelon treatments, designated as I to% field capacity (i.e. I70, I80, and I90), were based on replenishing the water in the soil volume (Vs) in 60 cm of the soil profile (Vs = 5.5 m2 x 0.6 m) to 70, 80, and 90 % of the gravimetric field capacity. The irrigation treatment amount in litre of muskmelon were calculated as:
[See supplementary files.] (2)
Where VI was the irrigation amount (liter), rb was soil dry bulk density (1350 kg m-3), gravimetric water content at field capacity, the irrigation level(0.7, 0.8, or 0.9), the measured gravimetric soil moisture content for a given plot, and = 0.95 an application efficiency coefficient for subsurface drip irrigation[45]. Irrigation timing demand is mainly driven by farmers’ perceptions and climatic conditions. Details of irrigation timing and amounts during muskmelon growth periods are shown in Table 1. Pre-irrigation soil moisture of muskmelon for different treatments and soil depth at 22 and 59 DAT are shown in Figure 8.
Experimental design
Muskmelon experimental design
A full factorial [4(levels of aeration frequencies) x 3 (levels of sub-surface drip irrigation) x 3 (levels of burial depths)] of 36 experimental treatment combinations would require 108 plots for three replicates. Since there was not enough space to accommodate all these plots, a balanced one-third fractional factorial design [46, 47] was developed using 12 of the 36 treatment combinations. Two L9(34) orthogonal arrays were prepared and superposed. Removing duplicates after superposition yielded the L12(4×32) quasi-orthogonal design matrix (Table 2). The treatments were assigned to the experimental plots in a completely randomized layout.
The design is balanced since each level occurs the same number of times for each factor (row). However, every pair of levels occurs the same number of times for factor pairs (A D) and (A I) but not for pair (D I). For the factor pair (D I), there are 9 level pairs but pairs (D10 I70), (D25 I80), and (D40 I90) occurs twice while the remaining 6 level pairs occur once. While not fully orthogonal, the design was as statistically efficient as possible.
A full 41×32 factorial of the 36 experimental treatment combinations and would require 108 plots for 3 replicates (Table 2). As with all fractional factorial designs, some effects are not estimable and confounding of lower and higher order effects is unavoidable. In this case, the three-factor interaction effect was not estimable. The main effects are independently estimable and it was possible to determine the existence of two-factor interactions. Even with these limitations, the design provided good stability and permitted estimation of the effects of primary interest comparable to the output of the much more expensive complete factorial experiment requiring 108 experimental plots.
Tomato experimental design
The experiment was limited to 6 treatments arranged as a randomised complete block design with 2 factors (drip irrigation tube burial depth and aeration frequency). The experiment used 2 drip irrigation tube burial depths (D): D15 and D40 represent drip irrigation tube burial depths of 15 and 40 cm, respectively. The tomato experiment used 3 aeration frequencies (A): AN, A2, and A4 represent no aeration, aeration once every 2 d, and aeration once every 4 d, respectively. The two experiments resulted in a total of 6 treatments, namely, D15AN, D15A2, D15A4, D40AN, D40A2 and D40A4, each of which was repeated three times.
Plant measurements
At 75 days after transplanting (DAT), all marketable fruits of muskmelon were harvested from each plot. And 214 days after transplanting (DAT), all marketable fruits of tomato (3 trusses) were harvested from each plot. The fruits were weighed and the fresh fruit weight per plant calculated. At 25, 55, and 75 DAT leaf area index (LAI) was measured by using AccuPARLP-80 canopy analyzer (Decagon Devices, Pullman, Washington 99163, USA) based on solar radiation transmittance. After harvesting, the remaining above-ground portion of all plants in each experimental plot were collected. The roots were removed and washed through a sieve. Stem and leaves and root were oven dried at 60°C for 48 h according to Shao et al., 2008; Ahmadi et al., 2014; Ramírez et al., 2014 [48-50]. Root dry weight was then measured on 3 muskmelon (tomato) plants from each plot. A cylinder of soil 0.6 m in diameter and 0.5 m in depth was isolated by digging a circular trench centered around the plant. The cylinder was removed, placed on the 100-mesh steel sieve, soaked, and gently washed to separate the roots from the soil. The plant roots were dried in the oven at 105 oC for 15 min to deactivate enzymes and allowed to dry completely at 75 oC before weighing.
Data analyses and statistics
The experimental data were organized in Microsoft Excel 2016. The statistical analyses were performed with the SPSS 22 software package (IBM, Armonk, New York). All figures were constructed using the graphing software OriginPro 9.0 (Origin Lab Corporation, One Roundhouse Plaza, Suite 303, Northampton, MA 01060, USA).
Muskmelon Data
Data are expressed as the mean. Three replicates were used for each experimental determination. Data were analyzed using a residual test method before statistical analysis, and the data met the assumption of homogeneity of variances and followed normal distribution. Mean differences between treatments were assessed by analysis of variance (ANOVA). Post-hoc pairwise comparisons of the treatment means were performed using Duncan's multiple range test. Differences were considered significant at the level of 0.05. The main purpose of the ANOVA was to assess the main effects and the 2-way interactions between drip irrigation tube burial depth, irrigation level and artificial aeration frequency on tomato yield, LAI, post-harvest dry matter partitioning, and irrigation use efficiency. The significances were defined at the level of 0.05 and 0.01.
Tomato Data
The experimental design was taken as a 2 x 3 factorial with 3 replicates. Data were analyzed using a residual test method before statistical analysis, and the data met the assumption of homogeneity of variances and followed normal distribution. Multiple comparisons using Duncan’s new multiple-range test were completed whenever the ANOVA indicated significant differences at the level of 0.05.