Effects of irrigation regime on the growth and yield of irrigated soybean in temperate humid climatic conditions

Agricultural Water Management 193 (2017) 30–45

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Research Paper

Effects of irrigation regime on the growth and yield of irrigated soybean in temperate humid climatic conditions F. Montoya ∗ , C. García, F. Pintos, A. Otero National Institute of Agricultural Research, INIA, Camino al Terrible s/n, Salto 50000, Uruguay

a r t i c l e

i n f o

Article history: Received 1 March 2017 Received in revised form 31 July 2017 Accepted 1 August 2017 Keywords: Supplemental irrigation Total dry matter Crop evapotranspiration Water use efficiency Profit margin

a b s t r a c t This research was conducted during two irrigation seasons (2014–2015 and 2015–2016) in Salto, Uruguay. This growing region is characterized by high annual precipitation and periods of soil water deficit of different intensities. This characterization casts much doubt to soybean growers regarding whether supplementary crop irrigation is useful for maximizing soybean yield, maintaining stable productivity and increasing profit margins. Three supplemental irrigation treatments in addition to a rainfed treatment were evaluated for their effects on soybean growth, development and yield with respect to the vegetative and reproductive stages. The results show that supplemental irrigation during the reproductive stage (R1–R8) has a positive effect on soybean growth and development, regardless of treatment. The total dry matter and leaf area index were between 8% and 40% higher in irrigation treatments compared with rainfed conditions. Actual evapotranspiration data, estimated with soil moisture sensors, showed that the crop coefficients (Kc) used in these experiments can be generalized for use in the region. During both cropping seasons, the rainfed treatment produced the lowest grain yield, with a 35% reduction in yield compared with that of the irrigated treatments. However, the water use efficiency values were inversely related to the amount of water applied. The profit margin showed that supplemental irrigation is useful in conditions during which the soybean price was greater than 350 U$D per ton, given the hypotheses considered. In the northwestern region of Uruguay, no irrigation would be the best option when the soybean price is less than U$D 350 or when rainfall is more stable during crop growth seasons. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Soybean (Glycine Max L.) is the sixth-most grown agricultural crop in the world (FAOSTAT, 2016). During 2006–2013, the USA, Brazil and Argentina were the main soybean-producing countries, whose soybean production equaled 70% and 80% of the total area harvest and total production in the world, respectively (FAOSTAT, 2016). Uruguay, as the eighth soybean producer worldwide (1.2% of the total), experienced a large increase in crop area and production (288% and 373%, respectively) during the 2006–2013 period, with crop yields close to those of the main producer countries (FAOSTAT, 2016). The high soybean demand from countries such as China and the high international price achieved (MGAP, 2014), coupled with the soil and climatic conditions in Uruguay, were the main causes of the growth in Uruguay.

∗ Corresponding author. E-mail address: [email protected] (F. Montoya). http://dx.doi.org/10.1016/j.agwat.2017.08.001 0378-3774/© 2017 Elsevier B.V. All rights reserved.

Irrigation significantly increases soybean yield and profit margins when the crop is growing in soils with permanent or periodic soil water deficit (Karam et al., 2005; Salassi et al., 1984). The soybean response to water deficit has been studied in many experimental trials (Dogan et al., 2007; Karam et al., 2005; Martín de Santa Olalla et al., 1994; Payero et al., 2005; Sincik et al., 2008), mainly under arid or semiarid conditions. Moderate soil water deficit for short periods of time during the vegetative stage generally do not reduce soybean yield (Lich et al., 2013; Oya et al., 2004). However, a more severe or long-term water deficit can lead to reductions in soybean yield (Lich et al., 2013). The reproductive stage shows the largest sensitivity to potential yield reduction during water deficit, while deficit during the flowering stage has a small negative effect on yield (Andrade, 1995; Foroud et al., 1993; Lich et al., 2013). However, water deficit during the pod-enlargement and seed-filling stages has a significant negative effect on the final yield and the components of yield (Andrade et al., 2002; Cox and Jolliff, 1986; Foroud et al., 1993). Many areas of the world with a moderate humid climate in addition to the high variation in rain distribution, especially during the spring and summer, can negatively affect soybean production, as

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45 Table 1 Soil water content (Hv%) of soil experimental trial. Soil Depth (m)

FC (%)

PWP (%)

0–0.10 0.10–0.20 0.20–0.30 0.30–0.40 0.40–0.50

55.60 52.40 58.90 56.90 60.90

38.15 44.00 45.60 43.20 49.60

FC: field capacity (0.01 MPa); PWP: permanent wilting point (1.5 MPa).

well as production of other agricultural systems that are based in pastures, compromising profitability and productive stability (Failde et al., 2013; Sincik et al., 2008). The use of the supplemental irrigation (Fereres and Soriano, 2007) applied during the most important stages, such as the reproductive stage, would allow growers to maximize yield, reducing the yield variability between years and increasing the profitability in temperate and humid climates, such as that of Uruguay. However, the typical rainfall season distribution of these areas, the productive costs (seeds, fertilizers, energy, etc.) and the soybean sale price reached during the harvest crop are the main reasons to decide whether growers ought to invest in irrigation system or not, under temperate humid climate conditions. Therefore, the purpose of the present work was to evaluate the soybean response to three supplemental irrigation strategies compared with the rainfed crop (fully irrigated and two deficit irrigations during the reproductive stage), focusing on the crop growth and development, grain yield, actual evapotranspiration and water use efficiency in a clay soil under a temperate climate in Uruguay. As an additional goal, the profit margins between supplemental irrigation and rainfed soybean crops were calculated. 2. Materials and methods 2.1. Location of the experiment The experiment was carried out during the 2014–2015 and 2015–2016 crop seasons in a 3500-m2 plot in Salto, Uruguay (31◦ 22 31 S, 57◦ 42 54 W). The soil was classified as either a Vertic Argiudolls (USDA-NRCS, 2006) or a Typical Brunosol Eutrico (MGAP, 1979), composed of an A horizon (0.35 m deep) and a B horizon (0.40 m deep), with clay texture (8.8% sand, 38.6% silt and 52.6% clay) in the A horizon. The soil hydraulic characteristics (field capacity and permanent wilting point) were determined using a Richard’s chamber for the soil water extraction with undisturbed samples from different depths up to 0.50 m (Table 1). It was carried out approximately one month before sowing in three representative sampling plots. The study area has a moderate humid climate (Linderman et al., 2013) characterized by a cyclical distribution of temperatures and evaporative demand, in which summer is the hottest season with the highest ET0 values, while winter has mild air temperature (5–8 ◦ C) with less evaporative demand. The meteorological data during both cropping seasons were obtained using a Davis station (Davis Instruments Corp. Inc., CA, USA) located near the experiment (Table 2). The accumulated precipitation during the cropping season was 730 mm during 2014–2015 and 1174 mm during 2015–2016. The reference evapotranspiration (ET0 ) was calculated according to FAO-Penman Monteith equation (Allen et al., 1998). 2.2. Crop management The soybean cultivar used in both years was an intermediate maturity group, (cultivar 6262 IPRO, Don Mario), which average

31

crop cycle duration is 142 days in the region, requiring from sowing around 9 days to emergence, 61 days to the beginning of bloom and 131 days to the beginning of ripening (Fassio et al., 2017). Sowing in the first experimental year occurred on November 8, 2014, whereas in the second year, it was carried out on November 24, 2015. During both seasons, the crop was planted after a ryegrass pasture, which was used as a cover crop during winter, producing 800 kg ha−1 of dry matter in the first cropping season and 3500 kg ha−1 in the second season, with the goal of avoiding erosion. Planting distance between rows was 0.40 m. A 300 kg ha−1 NPK (7-40-40) fertilizer were used in both cropping seasons applied at sowing date together with the seeds. Planting density was 41.2 and 28.0 plants m−2 in the first and second experimental years, respectively. The crop was harvested once the seedlings matured, which occurred on April 2, 2015, and April 20, 2016. The traditional cultivation techniques regarding pest and disease management of the area were used to maximize crop yield and quality. 2.3. Experimental design Four treatments were designed to supply different water satisfaction levels of the crop according to the soybean phenological stage. During the vegetative stage, 100% of the crop water requirement (CWR) was supplied with irrigation in three of the four treatments (Fig. 1a). However, these three treatments received a different CWR percentage (100%, 75% and 50%) during the reproductive phase (R1–R8) (Fig. 1b). The diary CWR was computed according to the FAO methodology (Allen et al., 1998). The reference treatment (T1; 100–100%) provided all crop water requirements throughout the crop growth cycle, allowing maximum crop production (Table 3). The two irrigation water deficit treatments, T2 (100–75%) and T3 (100–50%), received 75% (T2) and 50% (T3) of the CWR during the reproductive stage (Table 3). Rainfed treatment (T4) was used during the entire growing season (Table 3), allowing the crop to receive only the rainfall (Fig. 1). The experimental design was a completely randomized block with 3 replicates (Fig. 1). The treatment plot size was 12.8 m wide by 9.0 m long, providing enough surface to allow suitable crop development under a specific treatment. However, to ensure a crop water deficit and to achieve the T1, T2 and T3 treatments during the reproductive phase, it was necessary to delimit a small area that could prevent incoming rain in each plot. The rainfall-exclusion area (12.0 m2 ; 3.0 m by 4.0 m), called the elemental plot (Fig. 1b), was protected by a polyethylene cover film (similar to a tunnel-like structure) fixed to the ground with flexible tubing and wires (Fig. 2). This tunnel structure was not installed on the rainfed plots. Before any rain event, the polyethylene cover was extended over the crop, after which the cover was folded (Fig. 2). In addition, the elemental plot served as a sampling area. 2.4. Irrigation management and soil water measurement The irrigation schedule for the T1 treatment was accomplished diary using the simplified water balance method for the root zone of the crop (Allen et al., 1998; Pereira and Allen, 1999) (Eq. (1)). Dri = ETci − Pei − Ii + DPi + Dri-1

(1)

where, Dr: root zone depletion (mm); I: irrigation depth (mm); ETc: maximum crop evapotranspiration (mm) computed as ET0 *Kc; Pe: efective precipitation (mm); DP: deep percolation outside of root zone (mm); i: actual day; i-1: day before The crop coefficients (Kc) used for the reference treatment were based on those in FAO-56 (Allen et al., 1998) according to the soybean phenological stage as follows: 0.4 during crop establishment, 1.15 for the reproductive stage and 0.5 before ripening. The irrigation schedule for the T2 and T3 treatments was carried out

32

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

Table 2 Climatic conditions during the 2014–2015 and 2015–2016 experimental seasons. Year

Month

Average Tmax (◦ C)

Average Tmin (◦ C)

Average Rad (MJ m−2 day−1 )

Accumulated P (mm)

Accumulated ET0 (mm)

2014

November December January February March April November December January February March April

27.9 28.6 30.6 31.3 30.0 28.9 25.6 29.0 32.5 32.1 27.4 23.1

16.0 17.2 18.4 18.4 16.2 12.5 14.1 17.4 19.3 19.4 15.3 15.7

21.5 21.7 22.0 21.3 18.2 15.1 18.8 19.0 22.8 22.8 18.5 8.9

179.6 232.9 205.4 61.3 24.5 25.4 157.6 385.0 42.1 223.0 12.0 354.4

143.1 169.5 168.5 165.0 161.3 139.1 145.5 156.0 205.9 154.7 112.8 56.5

2015

2016

Tmax: maximum daily temperature; Tmin: minimum daily temperature; Rad: daily sun radiation; P: rainfall; ET0 : FAO-Penman Monteith reference evapotranspiration.

Fig. 1. Spatial treatment distribution according to the crop water requirement at vegetative stage (a) and reproductive stage (b). tp: Treatment plot (12.8 × 9.0 m); ep: Elemental plot (3.0 × 4.0 m).

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

33

Table 3 Irrigation water depth and effective precipitation during both experimental seasons. Irrigation treatment

T1 T2 T3 T4 a

% CWRa

100−100% 100−75% 100−50% Rainfed-Rainfed

Irrigation depth (mm)

Effective precipitation (mm)

2014–2015

2014–2015

2014–2015

2014–2015

382.7 282.7 201.6 –

309.5 236.5 116.6 –

384.9 384.9 384.9 442.8

524.8 524.8 470.6 470.6

Soybean Crop Water Requirement (% of ETc) according to two phenological stages, from sowing to R1 (vegetative stage) and from R1 to R6 (reproductive stage).

Fig. 2. Distribution tunnel in the experimental plots (a); Tunnel design over an elemental plot to avoid the rainfall (b).

using the same methodology as that for T1, except the Kc values during the reproductive stage and ripening were reduced by 25% and 50%, respectively. The effective precipitation (Pe) was used in the soil water balance equation, calculated according to the runoff curve number (NCRS, 2004) from the daily precipitation. Irrigation depth was computed in order to the depletion water root zone was between field capacity and readily available water (Allen et al., 1998). A complete drip irrigation system was used to supply the irrigation water. The system consisted of a lateral pipe of 16-diameter (13.6 mm) PE (0.25 MPa) with 0.50 m of emitter spacing; the pipe line was separated by 0.40 m in each crop row. The flow of the selfcompensating emitters was 1.60 l h−1 . Therefore, the system was designed to apply 8.0 mm h−1 of water with an average pressure in the lateral pipe of 100 kPa, allowing 100% wetting of the soil surface. The soil water content was measured with FDR sensors (10HS, Decagon Devices, Pullman, WA, USA). In the middle of each elemental plot, a set of five sensors was installed according to the sensor manual (Decagon, 2017) when crop reached the establishment stage (over 29 days after emergence), with intervals of 0.10 m in depth, up to 0.50 m. The FDR sensors data were recorded every 20 min. Capillary rise was not detected by the sensor data. However, some water percolation below the maximum effective rooting depth of the crop (0.45 m measured in field) and some surface runoff when major rainfall events occurred (>30 mm) was evident. The maximum root depth observed in the field agreed with Hoogenboom et al. (1987), who recorded more than 76% of the root system in the top 0.40 m of the soil.

2.5. Crop growth and development According to the soybean crop classification of Fehr et al. (1971), we monitored the crops weekly to determine crop phenology. The duration of each crop stage was calculated using the average temperature method to estimate the number of growing degree-days (McMaster and Wilhelm, 1997), considering 30 ◦ C and 5 ◦ C as the

upper and lower temperature thresholds, respectively (Raes et al., 2012). Sampling of soybean plants occurred every 15–20 days in each plot. During both cropping seasons (2014–2015 and 2015–2016), several samples were taken from crop establishment, beginning at 34 days after emergence (DAE) (December 18) and ending before harvest (136 DAE; March 30) during the first season (8 samples); for the second experimental season, plots were sampled 6 times, from 63 to 134 DAE (February 1 to April 12, respectively). On each sampling date, two plants were picked per plot and were analyzed in the laboratory by separating the organs (leaves, stems and pods) (Martín de Santa Olalla et al., 1994; Karam et al., 2005). After being oven-dried at 60 ◦ C for 48 h (until constant weight), the total dry matter (TDM) for each plant organ was recorded. On the same sampling day and with the same plants, the plant leaf area was measured using a LI-COR 3100C (LI-COR Inc., Lincoln, Nebraska, USA), and the LAI was calculated from the plant density. The TDM and LAI data were used to model the crop growth, fitted to a second-degree expo-polynomial model (Buttery, 1964; Martín de Santa Olalla et al., 1994), with days after emergence (DAE) as the variable of time (Eq. (2)). 2 X = e(a+b · DAE+c · DAE )

(2)

where X: Total dry matter (TDM; kg ha−1 ) or leaf area index (LAI; m m−2 ); DAE: days after emergence. The experimental plots were harvested when seed ripening occurred (140 DAE during both experimental seasons). Manual crop harvest was carried out in the center of each elemental plot (12 m2 ), with which 1.6 m2 was used to establish the plant density at harvest, aboveground biomass (kg ha−1 ), grain yield (kg ha−1 ), weight of 1000 seeds (g) (corrected to 13% seed moisture content) and harvest index (HI) 2.6. Crop evapotranspiration Actual evapotranspiration (ETa) was determined following two methodologies. On one hand ETa was estimated using the soil water content data at different depths provided by the FDR sensors (ET

34

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

BSFDR ). In this case, soil moisture records were used to perform a daily soil water balance (BSFDR ), in which the variation of the soil moisture content (S) was equal to the difference between the total water content in the soil at time 00:00 of a particular day and the total water content in the soil of the previous day, from 0 to 0.45 m in depth. The data from days with a major rain event and the following day were discarded from the calculations to avoid errors due to drainage processes and surface runoff (Allen et al., 2011). Thus, the ETa was estimated using the following soil water balance equation (Eq. (3)): ETa (ETBSFDR ) = I + Pe + S

(3)

where ETa (ET BSFDR ): actual crop evapotranspiration estimated by FDR sensors (mm); I: irrigation depth (mm); Pe: effective precipitation (mm) and S: change in the stored soil moisture (mm). On the other hand, ETa was also calculated using the theoretical soil water balance (ET BST ) to compute the irrigation schedule (Eq. (1). To determinate ET BST , the water stress coefficient (Ks) was calculated following the Allen et al. (1998) methodology (Eqs. (4) and (5)). ETa (ETBST ) = Ks · ETC

(4)

TAW − Dr Ks = TAW − RAW

(5)

where ETa (BST ): actual crop evapotranspiration calculated by soil water balance (mm); Ks: water stress coefficient (dimensionless); ETc: maximum crop evapotranspiration (mm); TAW: total available soil water in the root zone (mm); Dr: root zone depletion (mm); RAW: readily available water in the root zone (mm), which fraction of TAW (p) used was 0.5. 2.7. Profit margin computations The economic analysis of the soybean crop was performed based on the profit margins (PM), which were calculated as the gross profit minus the cost associated with the irrigation water application (Eq. (6)): PM = Y · P s − CI · I T

for total dry matter and leaf area index according to the crop water treatments. The significance levels used were as follows: p ≥ 0.05 not significant; 0.01 ≤ p < 0.05 significant and p < 0.01 very significant. The Tukey test was used to estimate whether the differences between the mean treatments were significant. Standard statistics were used to determine the goodness of fit, such as the coefficient of determination (r2 ) and the accuracy of the regression model coefficient estimation (p) during the crop growth model analyses. To determine the validity of the statistical assumptions of each model, a normality test (Kolmogorov-Smirnov test), a homoscedasticity test (Spearman correlation) and an independence test (Durbin-Watson test) were performed. 3. Results 3.1. Crop development Crop development was similar in both years with respect to the total crop cycle and the accumulated thermal time. The twelve-day difference in sowing date between both experimental years did not influence the total crop cycle length. The crop development length for treatments T1 and T2 was 137 days from emergence, and the length took nearly 2700◦ -days to complete its cycle (Fig. 3). In contrast, the T3 and T4 treatments completed their cycle with around 133 days from emergence (Fig. 3a), requiring 2650 and 2600◦ -days, respectively (Fig. 3b). During the first experimental year, the duration of the crop phenological stages of emergence (VE), flowering (R1), beginning of pod formation (R3) and beginning of seed formation (R5) were similar across all treatments (Fig. 3). On the other hand, the physiological maturity of the seeds (R8) occurred 7 days in advance for plants of the rainfed treatment (T4) while T3 reached R8 2 days before with respect to T1 and T2 (Fig. 3a). During the second cropping season, the crop cycle duration from sowing to R3 was similar across all treatments (Fig. 3). However, the beginning of the R5 stage varied depending on the crop water treatment. T4 and T3 reached the R5 stage 8 and 4 days earlier, respectively, than the T1 and T2 treatments (100 and 55 fewer degree-days) (Fig. 3).

(6)

where PM: profit margin (U$D ha−1 ); Y: soybean yield (kg ha−1 ); PS : soybean price received by the grower (U$D kg−1 ); CI : irrigation cost (U$D m−3 ) and IT : total water applied to the crop (m3 ha−1 ). The soybean price received by the growers had been highly variable, decreasing during the 2006–2013 period, with maximum, minimum and average of prices of 547, 207 and 408 U$D per ton, respectively (MGAP, 2014). Due to the large difference in soybean prices, we studied the profit margin variation using the soybean prices between 200 and 550 U$D per ton with intervals of 50 U$D. The water cost was set to 0.182 U$D per cubic meter, considering that irrigation water was supplied with drip irrigation system (energy and operation cost). The crop management costs, such as tillage, planting, seed, crop protection, fertilizer and harvest, were not taken into account for the profit margin calculation. The costs of these inputs are assumed to be constants in both cropping seasons, as the purpose of this study was to evaluate the effect of only supplementary irrigation on the profitability. 2.8. Statistical analysis The analysis of variance (PolyANOVA) was used to evaluate the effect of treatments on the biomass, yield, weight of 1000 seeds, water use efficiency and plant density, taking into account the irrigation treatment, the effect of the season and the irrigation-season interaction. An analysis of variance (ANOVA) was also performed

3.2. Water use by the crop The accumulated rainfall received by the crop during the 2014–2015 and 2015–2016 growing seasons was 567 and 768 mm, respectively (Table 4). The onset of the irrigation treatments (Fig. 4a,b) began 69 DAE during the 2014–2015 growing season (January 23) and started 57 DAE during the second season (January 27), close to flowering (51 DAE). During the first season, it was not possible to start the treatments closer to flowering (57 DAE) because the constant and uninterrupted rainfall did not allow the installation of the cover tents on time (Fig. 4a). The soil water content estimated by the water balance model (Allen et al., 1998) at a root depth of 0.45 m was very close to the readings taken by the FDR sensors in the four treatments (Fig. 5). Treatment of 100% of the CWR kept the soil moisture content near field capacity, while in the other treatments the soil moisture decreased as the deficit conditions occurred. It can be observed from 74 DAE in 2015 (Fig. 5a–d) and from 43 DAE in 2016 (Fig. 5e–h) that the soil moisture in the T3 and T4 treatments is between 35% and 85% less than that in the T1 and T2 treatment in both experimental years, causing the shortened cycle of the T3 and T4 treatments. The ETa estimated by the FDR (ET BSFDR ) sensors was compared with the ETa calculated by the theoretical soil water balance equation (ET BST ) (Fig. 6); such a comparison was made only for the T1 treatment (100% of the CWR), which was not subject to water

Table 4 Water received by the crop (precipitation, effective precipitation and irrigation depth), irrigation ratio, maximum and actual crop evapotranspiration and evapotranspiration ratio for the different treatments during both experimental seasons. EY

2014–2015

T1 T2 T3 T4 T1 T2 T3 T4

P (mm)

Pe (mm)

Irrigation events

Veg.

Repr.

Total

Veg.

Repr.

Total

322.9 322.9 322.9 322.9 456.2 456.2 456.2 456.2

172.2 172.2 172.2 245 380 380 312 312

495.1 495.1 495.1 567.9 836.2 836.2 768.2 768.2

245.6 245.6 245.6 245.6 255.9 255.9 255.9 255.9

139.3 139.3 139.3 197.2 268.9 268.9 214.7 214.7

384.9 384.9 384.9 442.8 524.8 524.8 470.6 470.6

23 17 12 – 19 12 7 –

Irrigation depth (mm)

Irrigation ratio

ETc (mm)

ETa (mm)

ETa/ETc ratio

Veg.

Repr.

Total

Veg.

Repr.

Total

Veg.

Repr.

Total

Veg.

Repr.

Total

Veg.

Repr.

Total

40.9 41.9 41.9 – 47.7 47.3 45.3 –

341.8 240.8 159.6 – 261.7 149.1 71.3 –

382.7 282.7 201.6 – 309.5 196.4 116.6 –

1.00 1.02 1.02 – 1.00 0.99 0.95 –

1.00 0.70 0.47 – 1.00 0.57 0.27 –

1.00 0.74 0.53 – 1.00 0.63 0.38 –

197.9 197.9 197.9 197.9 188.0 188.0 188.0 188.0

410.7 410.7 405.2 391.6 411.9 411.9 399.1 399.1

608.6 608.6 603.0 589.5 599.8 599.8 587.1 587.1

194.0 190.3 194.0 184.2 175.7 175.7 174.6 142.3

410.7 377.7 290.5 207.1 384.3 292.6 225.8 161.0

604.7 568.0 484.5 391.2 560.1 468.3 400.5 303.3

0.98 0.96 0.98 0.93 0.93 0.93 0.93 0.76

1.00 0.92 0.72 0.53 0.93 0.71 0.57 0.40

0.99 0.93 0.80 0.66 0.93 0.78 0.68 0.52

EY: experimental season; P: rainfall; Pe: effective precipitation; Veg.: vegetative stage; Repr.: reproductive stage; ETc: maximum crop evapotranspiration according to Allen et al. (1998); ETa: actual evapotranspiration according to soil water balance (Allen et al., 1998). Veg.: vegetative stage; Repr: reproductive stage.

Table 5 Statistical coefficients for the expo-polynomial model, describing the relationship between total dry matter or leaf area index and days after emergence, during both irrigation seasons. 2014–2015

Total dry matter

Leaf area index

T1 T2 T3 T4 T1 T2 T3 T4

r2

p

0.92 0.93 0.95 0.97 0.85 0.91 0.90 0.93

** ** ** ** ** ** ** **

2015–2016 p par a

b

c

−0.218 ns −1.646 ns 1.664 ns 1.515 ns −5.571* −6.087* −4.487* −5.720**

0.1240* 0.1449* 0.0888** 0.0909** 0.1560* 0.1700** 0.1360** 0.1610**

−0.00053* −0.00064* −0.00039** −0.0004** −0.00081* −0.0009** −0.00074** −0.00089**

KS

D

S

r2

p

yes yes yes yes yes yes yes yes

yes yes no no yes yes yes no

yes yes yes yes yes no yes yes

0.98 0.98 0.99 0.98 0.89 0.93 0.89 0.92

** ** ** ** * * * *

p par a

b

c

−2.327 ns −1.046 ns −0.863 ns −1.911 ns −9.143 ns −9.932* −8.089 ns −12.969*

0.154** 0.128** 0.125** 0.146** 0.224* 0.239* 0.205* 0.291*

−0.00065** −0.00053** −0.00053** −0.00063** −0.00114* −0.00123** −0.00108* −0.00147*

KS

D

S

yes yes yes yes yes yes yes yes

no no no no yes yes yes yes

yes yes yes yes yes yes yes yes

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

2015–2016

Treatment

T: treatment; r2 : regression coefficient; p: significance of the regression model; p par: significance of the coefficients model; KS: normality (Kolmogorov-Smirnov); D: independence (Durbin-Watson); S: homoscedasticity (Spearman correlation); ns: not significant. * p < 0.05. ** p < 0.01.

35

36

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

Fig. 3. Soybean crop phenological stages according to days after sowing (a) and thermal time (b). textEY: Experimental season; DAS: days after sowing; initials between the brackets are phenological stages according to Fehr et al. (1971).

deficit. In both irrigation seasons, the ET values estimated with both methodologies were similar (Fig. 6a,b). The ET estimated by BSFDR ranged, in both experimental years, from 1.0 to 8.9 mm d−1 , with a coefficient of determination (r2 ) of 0.83 with respect to BST (Fig. 6c). Likewise, the maximum ET values were achieved during the reproductive stages (R2–R6), while the lowest values were recorded during the ripening phase (R7–R8) (Fig. 6a,b). Assuming that the Kc used in the experiment was valid, the relationship between the actual evapotranspiration (ET BST ) and the maximum crop evapotranspiration (ETc) calculated by the theoretical soil water balance was carried out. The goal of providing the crop with 100% of the crop water requirements during the vegetative phase was achieved during the two experimental years, where the ETa/ETc ratios were very close to 1.00 (Table 4). In addition, the rest of the treatments achieved ETa/ETc ratios similar to that of T1, with the exception of T4 in the 2015–2016 season (0.76; Table 4). During the reproductive stage in the first cropping season, the ETa/ETc ratios for the irrigation treatments were 1.00 for T1, 0.92 for T2 and 0.72 for T3, while the ETa/ETc ratio obtained by the irrigation treatments during the second experimental year were 0.93,

0.71 and 0.57, respectively (Table 4). The ETa/ETc ratio for the T4 treatment in the first and the second experimental years was 0.53 and 0.40, respectively.

3.3. Crop growth Progression of the crop growth was evaluated by the total dry matter (TDM) related to the days after plant emergence (DAE), adjusting all the recorded TDM data to a single second-degree expopolynomial model. In all treatments, the r2 coefficients were very high in both years of the study, being 0.92–0.97 for the 2014–2015 season and 0.98–0.99 for the 2015–2016 season (Table 5). The coefficients b and c (origin intercept of the weight plant and response to water, respectively) were statistically significant for all treatments during both seasons. On the other hand, the coefficient a (maximum plant weight) was not significant for any treatment in both crop cycles. Analyzing the statistical validity of the model used for TDM, the normality test (Kolmogorov-Smirnov test) showed that the residuals were normally distributed at the 5% significance level (Table 5).

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

37

2014800

(a) 700

600

500

400

300

200

100

0 -7

0

7

14

21

28

35

42

49

56

63

70

77

84

91

98

10 5

11 2

11 9

12 6

13 3

14 0

133

140

DAE

P T1

P T2

P T3

P T4

Pe T1

Pe T2

Pe T3

Pe T4

Irrigaon T1

Irrigaon T2

2015-2016

800

(b)

700

Rain fall o r irrigatio n d ep th (mm)

600 500 400 300 200 100 0 -7

0

7

14

21

28

35

42

49

56

63

70

77

84

91

98

105

112

119

126

DAE

Fig. 4. Precipitation and irrigation depth accumulated in the different treatment plots during both experimental growing seasons (2014–2015, a; 2015–2016, b). Vertical bar is beginning of flowering (R1); P: precipitation; Pe: effective precipitation; Irrigation: irrigation depth; T: treatment.

The hypothesis of independence, using the Durbin-Watson statistic, showed that there was correlation between the residuals of the expo-polynomial model, with the exception of treatments T1 and T2 during the first experimental season (Table 5). With respect to the homoscedasticity test, it showed that the variance of the TDM model residuals was constant in all treatments (Table 5). In general, for both seasons, the treatments with greater crop water requirements (T1 and T2) showed a greater accumulation of TDM during the cycle, with the highest differences at 124 DAE

(2014–2015) and 119 DAE (2015–2016) (Fig. 7a; Table 6). No difference in TDM between treatments was observed during the last sampling (close to harvest) in either experimental season (Table 6). There was no difference in dry matter accumulation during the vegetative phase, whereas in the reproductive phase differences were observed between 111 and 124 DAE in the first season, while in the second season were between 105 and 119 DAE (Fig. 7a; Table 6). When comparing the accumulated growth curves between both experimental seasons, regardless of the treatment, the accumu-

38

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

Fig. 5. Soil water balance at the root zone during the 2014–2015 (a, T1 treatment; b, T2 treatment; c, T3 treatment; d, T4 treatment) and 2015–2016 (e, T1 treatment; f, T2 treatment; g, T3 treatment; h, T4 treatment;) growing seasons. Irrigation: irrigation depth; Pe: effective precipitation; Tbalance: theorical soil water balance at the root zone (mm); FDR: soil water balance with FDR sensors; FC: field capacity; PWP: permanent wilting point.

lated TDM was lower during the second season compared with the first (Fig. 7a). On the other hand, at the beginning of senescence, the accumulated TDM values were no different between treatments for both cropping seasons (Fig. 7a; Table 6).

Progression of the leaf area index (LAI) curve was also evaluated using the second-degree expo-polynomial model (Table 6), which more properly described the progression in time of the LAI in both seasons. The coefficients of determination ranged from 0.85 to 0.93

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

39

-

(a)

10 9

(b)

10 9 8

7

7

6

6

ET (mm)

8

5

4

5

4

3

3

2

2

1

1

0

0 30

40

50

60

70

80

90

10 0

11 0

12 0

13 0

14 0

30

40

50

60

70

80

DAE ET BSFDR

ET BST

ET BST (m m d-1 )

90

10 0

11 0

12 0

13 0

DAE

10 9 8 7 6 5 4 3 2 1 0

ET BSFDR

ET BST

ET BST = 0.8357ET BSFDR + 0.9637 R² = 0.83

(c) 0

1

2

3

4

5

6

7

8

9

10

ET BSFDR (mm d-1 ) 2014-2015 Fig. 6. ETa calculated by the theoretical water balance (ET BST ) and estimated by the water balance using FDR sensors (ET BSFDR ) during 2014–2015 (a) and 2015–2016 (b). The relationship between both seasons is also shown (c). ET BST: actual crop evapotranspiration calculated by theoretical water balance; ET BSFDR: actual crop evapotranspiration estimated by the water balance using FDR sensors.

Table 6 Statistical analysis of the total dry matter and leaf area index throughout soybean cycle during both experimental seasons. Variable

DAE

34

55

68

83

96

111

124

136

63

77

91

105

119

134

Total Dry Matter (g m−2 )

T1 T2 T3 T4 p-valor

50.9 55.7 53.5 55.4 ns

187.3 212.9 169.7 160.5 ns

368.0 340.5 333.7 327.0 ns

499.6 687.7 694.9 623.5 ns

801.1 959.9 665.8 655.0 ns

1153.0 a 1089.5 a 832.1 b 710.6 b

1323.8 a 1087.8 a 745.6 b 771.7 b

313.9 333.5 311.1 292.9 ns

546.9 487.5 451.4 422.6 ns

999.0 a 855.3 b 738.7 bc 681.3 c

**

148.5 122.4 142.0 117.7 ns

775.0 a 677.3 b 668.3 b 632.5 b

**

768.9 763.2 723.5 614.1 ns

*

*

778.1 724.6 639.6 529.4 ns

T1 T2 T3 T4 p-valor

0.7 0.7 0.7 0.7 ns

2.1 2.5 2.2 1.8 ns

3.7 3.4 3.6 2.9 ns

5.7 6.0 5.4 4.3 ns

6.4 7.0 5.0 4.5 ns

6.4 a 6.0 a 4.5 b 3.9 b

4.9 a 4.1 b 3.3 c 1.6 d

*

**

0.0 0.0 0.0 0.0 ns

1.8 1.6 1.8 1.1 ns

4 3.1 3.8 2.2 ns

5.5 4.8 4.8 3.6 ns

5.9 4.7 4.5 4.4 ns

4.9 3.4 3.6 2.8 ns

Leaf Area Index (m2 m−2 )

2014–2015

2015–2016

0.3 0.1 0.0 0.0 ns

DAE: days after emergence; ns: not significant; Means in the columns followed by different letters are significantly different according to the Tukey test. * p < 0.05. ** p < 0.01.

during the first season and from 0.89 to 0.93 during the second season. In addition, the hypothesis of normality was fulfilled in all treatments, and the assumptions of homoscedasticity and independence were met in all the plots in both seasons, with the exception of two treatments, T2 and T4, respectively, during the first season (Table 6). The model chosen was able to simulate the maximum observed LAI value in all treatments, with values ranging from 6.4 to 7.0 at 96 DAE during the first season and values ranging from 5.9 to 4.7 at 105 DAE during the second (Table 6; Fig. 7b). The maximum LAI reached was recorded between the R4 and R5 stages. The effect of the irrigation treatments on the LAI began to manifest at 111 DAE during the first season. However, during the second season, there were no differences in the LAI among treatments due to the greater variability of LAI in the treatment replicates (Table 6).

With respect to the LAI curve modeled, the rainfed treatment (T4) curve in both experimental years showed less growth than the other treatments, not only in the maximum LAI but also in the rise and fall phases of the LAI curve (Fig. 7b). 3.4. Water deficit and crop yield The soybean harvest started three to five days after the commercial crop maturity. During both seasons, the maximum biomass and grain yield values were achieved in the T2 treatment (9027 kg ha−1 and 4173 kg ha−1 , respectively) (Table 7). On the other hand, the minimum values occurred in the rainfed treatment (T4), with 35% less biomass and grain yield during the two growing seasons. The water replenishment of 100% and 50% of the crop requirements (T1 and T3 treatments, respectively) for the reproductive stage

40

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

Fig. 7. Total dry matter (a) and leaf area index (b) progression during both experimental cropping seasons. TDM: total dry matter; LAI: leaf area index; DAE: days after emergence.

Table 7 Effect of crop water treatment and irrigation season on the biomass, yield, harvest index, weight of 1000 seeds, water use efficiency and plant density.

Crop water treatment

Irrigation season

Main effects

T1 T2 T3 T4 SEM 2014–2015 2015–2016 SEM Irrigation Season Irrigation × Season

Aboveground biomass (kg ha−1 )

Grain Yield (kg ha−1 )

Harvest Index

Weight of 1000 seeds (g)

WUE (kg m−3 )

Plant density (plants m−2 )

8716a 9027a 8414a 6344b 191.5 8594a 7657b 135.41

3925ab 4173a 3730b 2689c 77.05 3686 3572 54.48 **

150 152 154 151 2.08 154a 149b 1.47 ns

0.46a 0.56bc 0.59c 0.52b 0.0132 0.51a 0.55b 0.0094

**

0.45 0.46 0.44 0.43 0.0112 0.43a 0.47b 0.0079 ns

32 35 35 36 1.44 41a 28b 1.01 ns

**

ns ns

**

*

**

**

ns

ns

ns

ns

ns

**

WUE: water use efficiency; SEM: standard error of the mean; ns: not significant; Means in the columns followed by different letters are significantly different according to the Tukey test. * p < 0.05. ** p < 0.01.

resulted in biomass and yield values similar to those of the T2 treatment, but the values were between 3% and 15% lower, respectively (Table 7). The combined analysis of irrigation and growing season demonstrate the significant effect of the irrigation on the total biomass and

grain yield, but a year effect was only observed in terms of the total biomass (Table 7). The interaction between irrigation and season for both variables was not significant (Table 7). The total biomass produced in the rainfed treatment was lower than that of the other three treatments in both seasons. No difference in the total biomass

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

produced was found among the three irrigation treatments, at different soil water deficit levels (Table 7). The harvest index (HI) was similar between treatments for both growing seasons, with values ranging from 0.43 to 0.46 (Table 7). However, the HI tended to be higher in the T2 treatment for the two experimental seasons, and the HI was significantly different between both growing seasons (Table 7). Likewise, this occurred with the weight of seeds, in which the crop water treatments had no effect, although the seed weight showed significant differences between both experimental years (Table 7). The functions of water production calculated for both cropping seasons showed that the soybean crop responds positively to the amount of water supplied; the production functions were fitted to a second-degree polynomial model, with a high coefficient of determination (>0.95; Fig. 8a). Regardless of the irrigation treatment, the maximum crop yield was achieved with a total water depth (irrigation and effective rainfall) greater than 580 mm. Despite of that, the maximum crop water requirement treatment (T1) did not increase grain yield compared with T2 (Fig. 8a; Table 7). The production functions according to the ETa/ETc ratio (Fig. 8b) showed coefficients of determination similar to those of the water production function, in which an ETa/ETc ratio greater than 0.75 led to the maximum soybean yield. However, with an ETa/ETc ratio less than 0.65, a lower soybean yield was achieved. The soybean water use efficiency (WUE, kg m−3 ) reached values between 0.47 and 0.65 kg m−3 , with high coefficient of determinations for both experimental years (Fig. 8c; Table 7). Treatments with greater soil water deficit (T2, T3 and T4) showed the highest WUE values, with the T3 treatment reaching the highest WUE (0.59 kg m−3 ; Table 7). In contrast, the T1 treatment presented the smallest WUE (0.46 kg m−3 ; Table 7) during both cropping seasons.

3.5. Profit margin Regardless of the soybean crop water treatment used in this report, the profit margins obtained were positive (>0; Fig. 9) for all soybean sale prices considered. However, according to the yield obtained in this experimental trial, the rainfed treatment reached the highest profit margins when the soybean sale price was around U$D 300 t−1 or less. The best profit margins were achieved by the T2 and T3 treatments when the soybean sale price was higher than U$D 300 t−1 . In general, the least profit margin was achieved by T1 treatment for all soybean sale prices considered, which differs with respect to the rest of the treatments by approximately U$D 260 t−1 , with the exception of the rainfed treatment when the soybean sale price was greater than U$D 450 t−1 , during which both profit margins were closer.

4. Discussion 4.1. Crop development During both experimental years, the soybean crop cycle for the T3 and T4 treatments shortened due to the soil water deficit created during the R1 to R8 reproductive stages. During the first cropping season, the beginning of the R5 stage did occur sooner than that of the soil water deficit treatments or the rainfed treatment (T4; Fig. 3) because the rainfall was continuous up to 80 DAE (close to the beginning of the R3 stage), where the accumulated effective precipitation (Pe) from sowing to the beginning of the R3 stage was 350.5 mm (Fig. 4a). Despite the decreased rainfall from the R3 stage to the end of the growing season (70.6 mm), the R5 stage for the T4 treatment was reached concurrently with the other treatments (99 DAE; Fig. 3a).

41

During the second growing season, there were two periods of low rainfall. The first one was between 43 DAE and 73 DAE (R1–R3), and the second was between 92 DAE and 124 DAE (R5–R7), with 27.6 mm and 16.2 mm, respectively. These two rainfall periods occurred during the phenological stages most sensitive to yield reduction by soil water deficit, which can result in reduced number of pods per plant, reduced number of seeds per pod and lower seed weight (Andrade et al., 2002; Oya et al., 2004). Soil water deficit can shorten the crop growth length, as Farré and Faci (2006) achieved with corn and McMaster and Wilhelm (1997) observed in barley and wheat. Thus, the prolonged effect of the soil water deficit caused the anticipation of the R5 stage in T3 and T4 with respect to the T1 and T2 treatments. In addition, the shortening of the cycle was maintained until the maturity (R8; Fig. 3). 4.2. Crop water use During both experimental years, the precipitation (P) and the effective precipitation (Pe) were different among treatments. During the first growing season, the plots of the T1, T2 and T3 treatments received less rainfall than those of T4 (Table 4) due to the reduction of the rainfall received over the soil by the plastic tents. In contrast, during the second growing season T1 and T2 received more rain because the crop cycle length was somewhat longer than those of T3 and T4 (Table 4; Figs. 3 and 4). In addition, during the second growing season, the cover tents were not used over the treatments since they suffered severe damage after two intense rainfall events (>35 mm h−1 ), but this lack of tents did affect the crop. We decided to not cover the elemental plots and to balance the soil water, taking advantage of the little rainfall and adjusting the soil water balance to the crop water requirement. It is noteworthy that the total rainfall that occurred during the cropping season was high enough to allow maximum production of soybean (Doorenbos and Kassam, 1979; Sincik et al., 2008). However, the uneven distribution and intensity of rainfall throughout the growing season leads to situations of water shortages at key phenological stages that affect the maximum performance of the soybean crop. The relationship between the theoretical soil water balance with respect to the FDR sensors was similar for all irrigation treatments (Fig. 5a,b,c,e,f,g), with the exception of the driest treatment (T4), whose relationship was weak (Fig. 5d,h). This lack of agreement can be explained by the FDR sensor equation, which could be not adjusted for the clay soils used in the experiment (Regalado et al., 2010), where the main lack of fit was in the low soil moisture content. Nevertheless, the daily variation of soil moisture between the water balance model and the FDR readings was very similar. The progression of soil water content during the 2014–2015 season shows that polyethylene tunnel prevented the entry of rainfall to the elemental plots, in which T4 showed an increase of soil water content in contrast to the rest of the treatments (Fig. 5a–d). The crop ETa estimates were calculated on select days distributed throughout the crop growth stages with different evaporative demands; during these specific days (always between watering events), the condition of the water movement in the soil was appropriate for analyzing the soil water balance with sufficient guarantee (Camargo et al., 2015b). The ETa results obtained with both methodologies are consistent with those of Karam et al. (2005), who measured maximum daily soybeans ETa values that ranged from 8.0 to 9.2 mm day−1 , whereas the ETa in the final phase of the crop dropped to between 80 and 85%. Therefore, the Kc values used in the experiments (Allen et al., 1998) can be considered valid for the present study. However, differences were found between the ETa and ETc (Fig. 6), indicating that a Kc calibration under local conditions would result in greater accuracy in the estimation of the crop water requirements.

42

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

Fig. 8. Production function with respect to the water received by the crop (a) and the ETa/ETc ratio (b). Water use efficiency function with respect to the water received by the crop (c). I: irrigation; Pe: effective precipitation; ETa: actual crop evapotranspiration; ETc: maximum crop evapotranspiration; WUE: water use efficiency.

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

43

800 600 400 200 0 200

250

300

350

400

450

500

Sale soybean price (U$D t-1 )

Fig. 9. Profit margin achieved at the different crop water treatments according to the variation in soybean sale price.

The contribution of rain during both experimental years was significant with respect to irrigation water (Table 4), allowing the proper crop development during the vegetative phase. Thus, the ETa/ETc ratio reached for all treatments during the vegetative stage was very close to 1.00, with the exception of T4 (rainfed conditions) during the second cropping season, as there was insufficient rainfall (36.7 mm) during January (the period during which the crop was still in the vegetative stage). In contrast, the rainfall that occurred between sowing until the end of January and its proper distribution during the first experimental year (Fig. 4a) allowed the rainfed treatment (T4) to reach a high ETa/ETc ratio (0.93; Table 4). The ETa/ETc ratios achieved for the T2 and T3 treatments during the first experimental year (higher values than the water deficit goals of 0.75 and 0.50, respectively) were due to the onset of irrigation treatments carried out 12 days after the beginning of flowering. With respect to the second experimental year, the ETa/ETc ratios achieved during the reproductive stage in both treatments were much closer to the targets due to the low rainfall from January until the end of the cycle (Fig. 4b). In general, the total ETa values obtained during both experimental seasons (Table 4) are close to those reported by Sincik et al. (2008) in a subhumid climate and by Payero et al. (2005) in a semiarid environment. 4.3. Crop growth The recorded TDM data for the of all treatments were adjusted to a single second-degree expo-polynomial model (Buttery, 1964; Martín de Santa Olalla et al., 1994) because this model showed a better fit with respect to other single-degree models, such as the Gompertz or the logistic model. The estimated coefficients for the expo-polynomial model with TDM data met the criteria of significance and quality. However, there were some specific cases that did not meet the test of independence. Nevertheless, the model is robust and therefore is not a problem for interpreting the results. Buttery (1964) and Martín de Santa Olalla et al. (1994) found that the expo-polynomial model better model conformed to their observed TDM data, which we also observed with our data. Likewise, the accuracy of the estimation of the model coefficients was very similar to that in other studies in soybean (Martín de Santa Olalla et al., 1994) or in other crops, such as potato (Camargo et al., 2015a). In general, the average accumulated TDM was higher during the 2014–2015 season than 2015–2016 because the plant den-

sity during the first season was 47% higher for 2015–2016. In this sense, Steduto et al. (2012) showed that the plant density is directly related to the accumulation rate of the TDM, LAI and soil canopy cover in soybean and in many other crops. However, this higher TDM progression did not result in a higher accumulated TDM in the ripening crop (Fig. 7a). This aspect was noted by Andrade (1995) in a study carried out in Argentina, in which the stability of the soybean yield compared with the variation in plant density was high. Regarding the plant density in this trial, there was no difference in the density of plants between treatments (p > 0.05) during both growing seasons, with 39.8 and 42.8 plants m−2 in the first experimental season and 25.0 and 30.1 plants m−2 in the second year (Table 7). The LAI growth was also modeled with the expo-polynomial model. This model was also used in other reports of soybean (Buttery, 1964; Martín de Santa Olalla et al., 1994) to describe the progression of the LAI as well as in sunflower (Botella et al., 1997) and potato (Camargo et al., 2015a), whose coefficients and significance models were similar to those of our reported data. The same period of time at which the maximum LAI was reached by the soybean crop in our experimental trial was noted by Karam et al. (2005) and Sincik et al. (2008), who noted the period between the R3 and R5 crop stages. The LAI curve model of T4 showed less growth than the other treatments due to the soil water deficit that occurred between 52 DAE and 76 DAE, where the lack of soil moisture content in T4 resulted in a slower LAI increment with respect to the other treatments (Fig. 7b). Similarly, regarding the TDM, the LAI curve modeled during the initial growth phase in all treatments was less during the second growing season than the first one due to the plant density (Fig. 7b). Nevertheless, the maximum LAI of treatments was reached at a similar date, as noted above. 4.4. Water deficit and crop yield The grain yield was affected by the amount of water applied to the crop and the degree of water stress generated during the reproductive stage (R1–R8) for the two cropping seasons. In this sense, only with rain was maximum yield unattained (2689 kg ha−1 ; rainfed treatment). However, maximum yield (4173 kg ha−1 ) was achieved with 75% of the crop water requirement during the reproductive stage. The T1 and T3 treatments obtained similar yields, without showing significant differences with T2 (Table 7). Therefore, the supplemental irrigation during the R1 to R8 stages allowed for maximal soybean grain yield in this temperate climate. Other

44

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45

authors have found the same crop response to water in trials carried out under different climatic conditions (Andrade, 1995; Andrade et al., 2002; Candogan et al., 2013; Karam et al., 2005; Martín de Santa Olalla et al., 1994; Sincik et al., 2008), and they reported no difference in yield with water deficit treatments of 75% and 50% ETc in soybean, with a maximum yield with 100% ETc. In addition, the total rainfall accumulated in these trials was much lower, where irrigation treatments were carried out during the entire crop cycle. The significant differences between both experimental years for HI and seed weight can be explained by the effect of plant density and rainfall distribution during the reproductive stage, where the first experimental year reached a higher biomass than the second one, with similar seed yield. Sincik et al. (2008) reported HI values that were 20% lower than the data presented in Table 7 but with significant differences among irrigation treatments. In addition, Spaeth et al. (1984) reported soybean results in concordance with our data, in which the HI stability was a soybean cultivar characteristic that varied with its relationship with the soil water content and photoperiod. According to the ETa/ETc ratio reached in both experimental years (Table 4), the T2 treatment achieved a similar ratio (0.93) to that of the T1 treatment (0.99) during the first cropping season. Therefore, the behavior of T2 was similar to that of the T1 treatment, as observed in the yield results (Fig. 8b). During the second cropping season, T2 presented an ETa/ETc ratio of 0.78, which was very close to the ETa/ETc ratio of the T3 treatment during the first season (0.80; Fig. 8b) – both of them with similar yield and without showing differences in yield between years (Table 7). However, during the second experimental year, the ETa/ETc ratio of the T3 treatment was similar to the ETa/ETc ratio of the rainfed treatment (T4) during the first season, which were 0.68 and 0.66, respectively (Fig. 8b), but the yield of both showed differences of approximately 35%. This yield difference can be explained by the rainfall distribution from sowing to R3 (80 DAE) during the first season being more homogeneous than that during the second experimental year (Figs. 3, 4 and 7). Although the yield and the total water received in the T4 treatment were similar in both seasons (Fig. 8a), the ETa/ETc ratio was different by 20% between seasons (Fig. 8b) due to the rainfall distribution during the growing season. During the growing season, the soil water deficit in the T4 treatment beginning at R3 during the first cropping season affected the seed-filling period, while the T4 treatment, during the second cropping season, had two long periods without homogeneous rain (Fig. 3). Payero et al. (2005), in a deficit irrigation trial carried out under semiarid conditions for three years, reported that ETa/ETc ratios greater than 0.75 led to achieving maximum soybean yield. These results are consistent with the data observed in this trial. In general, the WUE values reached by the water irrigation treatment were similar to those of other soybean trials (Candogan et al., 2013; Karam et al., 2005; Sincik et al., 2008) as well as those in other crops (Camargo et al., 2015b; Onder et al., 2005; Ortiz et al., 2010; Payero et al., 2006). These results show that soybean maintains a constant water use efficiency under different climatic conditions. 4.5. Profit margin According to the soybean yield reached with each crop water requirement applied during both growing seasons, the profit margins have shown that the use of supplemental irrigation on a soybean crop is suitable for increasing the profit margin, taking into account two aspects. The first one is that the use of irrigation amounts less than the maximum crop water requirement can achieve a better profit margin (Fig. 8), independent of the considered soybean price. The second aspect is that when soybean commercial prices are more than U$D 350 t−1 , higher profit margins begin with the T2 and T3 treatments than the T4 and T1 treatments

(Fig. 8). In addition, soybean prices lower than U$D 350 t−1 suggest that the rainfed soybean crop may be the most desirable option for growers, assuming costs do not increase. This study’s profit margin is based on the results of two growing season; thus, the profit margin is not robust enough for growers given the variability of precipitation during the crop growth period. Therefore, it is advisable to use crop simulation models to verify whether supplemental irrigation is necessary in regions like this one used for the trial, where also may exist a high interest in soybean production. In this sense, it could produce greater robustness and reliability in the profit margins obtainable by the growers who irrigate. 5. Conclusions After two experimental seasons it was observed that the use of supplementary irrigation in soybean crop provided an increase in total dry matter as well as grain yield, enabling crop yield stability during the key phenological stages (R1–R6). The best irrigation strategy, which maximizes yield and profit margin, is to provide 75% of the crop water requirement between the R1 to R8 stages, since rainfall during the vegetative stage allows for proper soybean crop growth and development. Rainfed conditions reduced the grain yield and final biomass up to 35%. The expo-polynomial model used to represent the total dry matter and leaf area index progression, adjusted properly with observed field data from both seasons, allowed determination of the beneficial effects of irrigation on the growth of the soybean crop. The ETa/ETc ratio achieved with the theoretical soil water balance showed that it is possible to achieve maximum yields of soybean in Uruguay with a global ETa/ETc ratio greater than 0.70. Finally, the profit margins showed that for soybean retail prices less than U$D 350 t−1 , supplemental irrigation is not suitable to practice, taking into account the assumptions of the average yield obtained, total irrigation water amount received by the crop and cost of water. If the soybean price increases, the practice of supplement irrigation, with 75% or 50% of crop water requirement strategies during the reproductive stage, is suitable for maximizing the profit margin. Acknowledgements This paper was developed within the framework of the “INIASA 28 0 00” project that was funded by the National Institute of Agricultural Research of Uruguay. The authors also wish to thank the support given by Experimental Station of Agronomy in Salto (EEFAS). References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration-Guidelines for Computing Crop Water Requirements-FAO Irrigation and Drainage Paper 56, vol. 300. FAO, Rome, pp. 6541. Allen, R.G., Pereira, L.S., Howell, T. a., Jensen, M.E., 2011. Evapotranspiration information reporting: I. Factors governing measurement accuracy. Agric. Water Manag. 98, 899–920, http://dx.doi.org/10.1016/j.agwat.2010.12.015. Andrade, F.H., Aguirrezábal, L.A.N., Rizzalli, R.H., 2002. Crecimiento y rendimiento comparados. In: Andrade, F.H., Sadras, V.O. (Eds.), Bases Para El Manejo Del Maíz, El Girasol Y La Soja. INTA Balcarce, Facultad de Ciencias Agrarias, Buenos Aires, Argentina, p. 450. Andrade, F.H., 1995. Analysis of growth and yield of maize sunflower and soybean grown at Balcarce, Argentina. Field Crop. Res. 41, 1–12. Botella, O., de Juan, J.A., Martin de Santa Olalla, F., 1997. Growth, development and yield of five sunflower hybrids. Eur. J. Agron. 6, 47–59. Buttery, B.R., 1964. Analysis of the growth of soybeans as affected by plant population and fertilizer. Can. J. Plant Sci. 49, 675–684. Camargo, D.C., Montoya, F., Córcoles, J.I., Ortega, J.F., 2015a. Modeling the impacts of irrigation treatments on potato growth and development. Agric. Water Manag. 150, 119–128. Camargo, D.C., Montoya, F., Ortega, J.F., Córcoles, J.I., 2015b. Potato yield and water use efficiency responses to irrigation in semiarid conditions. Agron. J. 107, 2120–2131.

F. Montoya et al. / Agricultural Water Management 193 (2017) 30–45 Candogan, B.N., Sincik, M., Buyukcangaz, H., Demirtas, C., Goksoy, A.T., Yazgan, S., 2013. Yield: quality and crop water stress index relationships for deficit-irrigated soybean [Glycine max (L.) Merr.] in sub-humid climatic conditions. Agric. Water Manag. 118, 113–121. Cox, W.J., Jolliff, G.D., 1986. Growth and yield of sunflower and soybean under soil water deficits. Agron. J. 78, 226–230. Decagon, 2017. 10HS Large Soil Moisture Sensor [WWW Document]. URL http:// www.decagon.com/en/soils/volumetric-water-content-sensors/10hs-largevolume-vwc/. (Accessed 14 July 2017). Dogan, E., Kirnak, H., Copur, O., 2007. Deficit irrigations during soybean reproductive stages and CROPGRO-soybean simulations under semi-arid climatic conditions. Field Crop. Res. 103, 154–159. Doorenbos, J., Kassam, A.H., 1979. Yield response to water. In: Doorenbos, J., Kassam, A.H. (Eds.), FAO Irrigation and Drainage Paper No. 33. FOA, Roma, Italy, pp. 1–57. FAOSTAT, 2016. Food and Agriculture Organization of the United Nations. Stadistic Division [WWW Document]. URL http://faostat3. fao.org/faostat-gateway/go/ to/down. (Accessed 6.14.16). Failde, A., Peixoto, C., Estol, E., Preve, A., 2013. Estudio sobre riego agropecuario en Uruguay. FAO-Red Mercosur Investig. Económicas. Farré, I., Faci, J.M., 2006. Comparative response of maize (Zea mays L.) and sorghum (Sorghum bicolor L. Moench) to deficit irrigation in a Mediterranean environment. Agric. Water Manag. 83, 135–143. ˜ Fassio, A., Ibanez, W., Rodriguez, M., Ceretta, S., Pérez, O., Rabaza, C., Vergara, G., Cesán, A., Restaino, E., 2017. Predicción de estados fenológicos de soja y otros cultivos de verano [WWW Document]. URL http://www.inia.uy/gras/Alertasy-herramientas/Utilidades#. (Accessed 15 July 2017). Fehr, W.R., Cavines, C.E., Burmood, D.T., Pennington, J.S., 1971. Stage of development descriptions for soybean [Glycine max (L.) Merrill]. Crop Sci. 11, 929–931. Fereres, E., Soriano, M.A., 2007. Deficit irrigation for reducing agricultural water use. J. Exp. Bot. 58, 147–159. Foroud, N., Mundel, H.H., Saindon, G., Entz, T., 1993. Effect of level and timing of moisture stress on soybean yield components. Irrig. Sci. 13, 149–155. Hoogenboom, G., Huck, M.G., Peterson, C.M., 1987. Root growth rate of soybean as affected by drought stress. Agron. J. 79, 607–614. Karam, F., Maasad, R., Sfeir, T., Mounzer, O., Rouphael, Y., 2005. Evapotranspiration and seed yield of field grown soybean under deficit irrigation conditions. Agric. Water Manag. 75, 226–244. Lich, M.A., Wright, D., Lenssen, A.W., 2013. Soybean Response to Drought, Agricultur. Iowa State University Extension and Outreach, Ames, IA (USA). Linderman, T., Plata, V., Sancho, D., Oyhantcabal, W., 2013. Clima de cambios. Nuevos desafíos de adaptación en Uruguay. FAO, MGAP. MGAP, 1979. Carta de reconocimiento de suelos del Uruguay, Tomo III. Ministerio de Agricultura y Pesca. Dirección de Suelos y Fertilizantes, Montevideo-Uruguay, 452 pag. MGAP, 2014. ANUARIO ESTADÍSTICO AGROPECUARIO 2014 [WWW Document]. URL http://www.mgap.gub.uy/estadisticas-y-documentos/agricultura. (Accessed 25 May 2016).

45

Martín de Santa Olalla, F., de Juan, J.A., Fabeiro, C., 1994. Growth and yield analysis of soybean (Glycine max (L.) Merr.) under different irrigation schedules in Castilla-La Mancha, Spain. Eur. J. Agron. 3, 187–196. McMaster, G.S., Wilhelm, W.W., 1997. Growing degree-days: one equation, two interpretations. Agric. For. Meteorol. 87, 291–300. NCRS, 2014. Part 630 Hydrology. National Engineering Handbook, Estimation of Direct Runoff from Storm Rainfall, Natural Resources Conservation Service. United States Department of Agriculture. Onder, S., Caliskan, M.E., Onder, D., Caliskan, S., 2005. Different irrigation methods and water stress effects on potato yield and yield components. Agric. Water Manag. 73, 73–86, http://dx.doi.org/10.1016/j.agwat.2004.09.023. Ortiz, J.N., de Juan, J.A., Tarjuelo, J.M., 2010. Analysis of water application uniformity from a centre pivot irrigator and its effect on sugar beet (Beta vulgaris L.) yield. Biosyst. Eng. 105, 367–379. Oya, T., Nepomuceno, A.L., Neumaier, N., Farias, J.R.B., Tobita, S., Ito, O., 2004. Drought tolerance characteristics of Brazilian soybean cultivars: evaluation and characterization of drought tolerance of various Brazilian soybean cultivars in the field. Plant Prod. Sci. 7, 129–137. Payero, J.O., Melvin, S.R., Ismark, S., 2005. Response of soybean to deficit irrigation in the semi-arid enviroment of west-central Nebraska. Trans. ASAE 48, 2189–2203. Payero, J.O., Melvin, S.R., Irmak, S., Tarkalson, D., 2006. Yield response of corn to deficit irrigation in a semiarid climate. Agric. Water Manag. 84, 101–112. Pereira, L.S., Allen, R.G., 1999. Crop water requirements. In: van Lier, H., Pereira, L.S., Steiner, F.R. (Eds.), CIGR Handbook of Agricultural Engineering, Vol. I: Land and Water Engineering. ASAE, St. Joseph, Michigan, pp. 213–262. Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2012. AquaCrop Reference Manual, AquaCrop Version 4.0 [WWW Document]. Regalado, C.M., Ritter, A., García, O., 2010. Dielectric response of commercial capacitance, impedance and TDR electromagnetic sensors in standard liquid media. In: The Third International Symposium on Soil Water Measurement Using Capacitance, Impedance and TDR, Murcia, Spain. Salassi, M.E., Musick, J.A., Heatherly, L.G., Hamill, J.G., 1984. An economic analysis of soybean yield response to irrigation of Mississippi River Delta soils. Miss. Agric. 60, 928–935. Sincik, M., Candogan, B.N., Demirtas, C., Büyükcangaz, H., Yazgan, S., Göksoy, A.T., 2008. Deficit irrigation of soya bean [Glycine max (L.) Merr.] in a sub-humid climate. J. Agron. Crop Sci. 194, 200–205. Spaeth, S.C., Randall, H.C., Sinclair, T.R., Vendeland, J.S., 1984. Stability of soybean harvest index. Agron. J. 76, 482–486. Steduto, P., Hsiao, T.C., Fereres, E., Raes, D., 2012. Crop yield response to wateStedutor. FAO Irrig. Drain Pap., 66. USDA-NRCS, 2006. Keys to Soil Taxonomy, 10th ed. USDA-Natural Resources Conservation Service, Washington, DC.