Site Description
The field experiment was conducted during 2016 and 2017 at Tianye group agricultural demonstration park (44°26′ N, 85°95′ E), Shihezi City, Xinjiang, China. The experimental site was located in an arid temperate continental climate zone with large diurnal temperature variations. Mean annual temperature was 11.2 (2016) and 14.0°C (2017), and annual precipitation was 395 (2016) and 203 mm (2017). The previous crop was cotton (Gossypium spp.). The physical and chemical properties of the 0-20 cm plough layer soil are shown in Table 1.
Experimental Design
The experimental treatments were the factorial combinations of three irrigation rates (5.25 ML ha-1 (W1), 6.0 ML ha-1 (W2, the actual irrigation amount of local alfalfa high yield field) and 6.75 ML ha-1 (W3)) and four P fertilizer rates (as P2O5 equivalent; 0 kg ha-1 (P0), 50 kg ha-1 (P1), 100 kg ha-1 (P2) and 150 kg ha-1 (P3)), with three replicates in a randomized complete block design. The P was monoammonium phosphate (P2O5 52 %). Each block was 40 m2 (5 × 8 m).
There were 8 irrigation times in each growing season, the specific irrigation time was 8–10 days before harvest and 3–5 days after harvest, and the P fertilizer was evenly divided into four times and applied to the soil with irrigation under drip irrigation, beginning at the branching stage of spring growth following winter dormancy and subsequently 3–5 d after the each cuts. The specific cutting time was May 25, June 26, August 1 and September 25, 2016. The cutting time was May22, June26, July 30 and September 23, 2017.
The alfalfa cultivar WL354HQ (Fall dormancy class, FDC 3.9) was sown on April 19, 2015. The crop was sown with artificial drilling (seed drill) using a seed rate of 18.0 kg·ha-1 with a row-spacing of 20 cm, and the sowing depth was 2.0 cm. Drip irrigation belts were buried at a depth of 8–10 cm with a distance between drip irrigation belts of 60 cm. Inlaid drip irrigation belts (produced by Green Source Co., Ltd. in Beijing) were used, and the distance between the drip heads was 20 cm. Monthly rainfall and average temperature during the growing seasons in 2016 and 2017 are presented in Fig. 1.
Soil sample collection
Soil samples of 0–20 cm were taken from soil drills in each plot by the "S" sampling method in October of each year. Five soil samples from the same soil layer were mixed to make composite soil samples. After removing impurities such as alfalfa roots and stones, the soils were brought back to the laboratory and dried to constant weight in an oven at 65℃. The fine soil was sifted through a 100 mesh sieve for reserve18.
Sampling and Measurements
The hay yield of each cut of alfalfa was measured by cutting three 1 m × 1 m quadrats in each plot at the early flowering stage (10 % blooming) and cut four times a year. The specific harvesting dates were May 25, June 26, August 1, and September 25, 2016 and May 22, June 26, July 30 and September 23, 2017. The alfalfa plants in the sample plot (cut height 5 cm) were cut with scissors and weighed, and the yield of fresh alfalfa forage was recorded. A sample of 300 g per plot fresh alfalfa was taken back to the laboratory. The samples were first oven-dried at 105°C for 30 min and then at 65°C to a constant mass.
Crude protein (CP) was determined by the semimicro Kjeldahl method. The neutral detergent fibre (NDF) and acid detergent fibre (ADF) were determined according to procedures of Van Soest19. The relative feeding value (RFV) was calculated by NDF and ADF using the following equation20:
RFV = (88.9 - 0.779 × ADF) × (120 / NDF) / 1.29 (1)
In the process of measuring alfalfa hay yield, three fresh alfalfa samples were dried and crushed. Forage P concentration was determined using the molybdenum-antimony spectrophotometric method21.
The P concentration of the alfalfa was multiplied by the respective yields to calculate the shoot P uptake based on the hay yield. The P uptake based on the hay yield was added together to determine the total P uptake and was converted to kg P uptake ha-1. Phosphorus recovery efficiency (PRE) was calculated as the following equation21:
PRE = (Up – U0) / Fp (2)
where Up and U0 are the P taken up by alfalfa from soils with (Up) and without (U0) added P and Fp is the amount of P applied, and the result expressed as a percentage.
The WUE of alfalfa was calculated using the following equation23:
WUE = HY / ET (3)
where HY is the alfalfa hay yield, and ET (evapotranspiration) is the crop water consumptio16. Then:
ET = P + U + I – F – R – ΔW (4)
where P is precipitation, U is the groundwater recharge, I is the amount of irrigation, F is the deep drainage, R is the runoff, and ΔW is the change in soil moisture from the beginning to the end of the trial24, soil moisture content was determined by drying method. According to the conditions during the experiments (no slope, deep water table), the contributions of groundwater recharge, runoff and deep drainage were negligible.
Total phosphorus (TP) was determined by the sulfuric acid-perchloric acid decoction molybdenum antimony colorimetric method, and AP was determined by the NaHCO3 extraction molybdenum antimony colorimetric method25.
Economic benefits
Economic benefit analysis refers to the assessment and evaluation of the size or level of economic benefits, and the analysis and Research on the causes of its formation26.
EB=YB-TC (5)
YB=Y-P (6)
TC=LRC+SC+WC+PC+WEC+LC+HC (7)
where EB is the Economic benefit, YB is the yield benefit, TC is the total cost, Y is the yield of alfalfa, P is the price of alfalfa, LRC is the land rent cost, SC is the seed cost, PC is the phosphorus cost, WE is the water and electricity cost, LC is the labor cost, and HC is the harvesting cost.
Statistical Analysis
The effects of water and P on the hay yield, CP, RFV, WUE, PRE and P concentration of alfalfa were examined using two-way (W, P, W×P) ANOVA for each of the 8 harvest dates separately. The means were compared using Duncan tests at P < 0.05. The statistical analyses were determined with 7.05 (Data Processing System, China).
The subordinate function evaluation method was used to comprehensively evaluate the optimal treatment using the following formulas:
where X is the measured value of each index of the sample; UX(+) is the positive correlation low function value of each index; and UX(-) is the negative correlation low function value of each index18.