The deposition of gliadin and glutenin proteins in the kernel is ordered and asynchronous and can be impacted by plant stress and N management (Daniel and Triboi, 2002; Jamieson et al., 2001; Przednowek et al., 2002). Gliadin, which is deposited in the kernel first (Gupta et al., 1996; Panozzo et al., 2001), generally decreases in relative concentration as grain filling progresses (Triboi et al., 1990). The relative amount of gliadin contained in the kernel can be increased by decreasing N stress (Triboi et al., 2000). Others have reported that adopting practices that probably decrease N stress results in higher HMW-GS concentrations (Wieser and Seilmeier, 1998; Fuertes-Mendizábal et al., 2010).
Water or heat stress during grain filling can impact gluten composition and dough quality (Blumenthal et al., 1991; Ciaffi et al., 1996; Corbellini et al., 1997; Stone et al., 1997; Jiang et al., 2009). Park (2011) reported that for winter wheat grown in South Dakota, the relative glutenin (r = 0.45, P < 0.05) and gliadin (r = –0.49, P < 0.05) amounts were correlated with yield loss due to water stress (YLWS) as calculated using the 13C isotopic discrimination approach (Clay et al., 2001). By better understanding the impacts of N and water on grain yield and quality, it may be possible to optimize yearly irrigation and N management. The objective of this study was to quantify the combined and individual impacts of N and water stress on winter wheat grain yield, grain protein, dough quality, and water and N use efficiencies.
MATERIALS AND METHODS
The experiment was established at Dakota Lakes Research Farm (44°17´ N, 99°59´ W, elevation approximately 560 m) near Pierre, SD. The soil was a Lowry silt loam (a coarse-silty, mixed, superactive, mesic Typic Haplustoll). The sand, silt, and clay contents were 114, 686, and 200 g kg−1, respectively. The pH (1:1 water/soil ratio), electrical conductivity (saturated extract), cation exchange capacity, and organic matter content were 7.5, 1 dS m−1, 15 cmolc kg−1, and 30 g kg−1, respectively (websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx; verified 27 June 2011). Bulk densities for the 0- to 15- and 15- to 60-cm soil depths were 1.15 and 1.2 g cm−3, respectively. Growing degree days (GDD) were summed from planting to crop maturity and were calculated asThe calculated GDD for base 0°C were 1891 in 2007 and 2262 in 2008. The average temperature during flowering and grain fill, May to July, was 21.3 and 18.3°C in 2007 and 2008, respectively. Total precipitation during the same period was 17.3 and 22.3 cm in 2007 and 2008, respectively.
The hard red winter wheat ‘Overley’ and hard white winter wheat ‘Alice’ were planted at 4.44 × 106 seeds ha−1 (145 kg ha−1) on 21 Sept. 2006 and 8 Sept. 2007, respectively. The characteristics of Overley are available in Fritz et al. (2004) and the characteristics of Alice are available in Ibrahim et al. (2006).
Natural rainfall from 31 July 2006 to 1 Aug. 2007 and from 31 July 2007 to 1 Aug. 2008 for the 2007 and 2008 wheat crops was 45.7 and 45 cm, respectively. These totals were supplemented with additional water by placing a line source irrigation system in the center of the experiment. Rain gauges were used to measure the irrigation amounts 2.3 and 16 m from the line source. Plots 2.3 m from the line source were identified as the A (adequate) water treatment and plots 16 m from the line source were identified as the D (deficient) water treatment. Tensiometer readings from the 45- and 90-cm soil depths in the A water treatment were used to schedule irrigation applications.
In 2007 and 2008, irrigation was applied between the tillering and boot growth stages. Total irrigation applied to the A and D treatments in 2007 was 21 and 4.2 cm, respectively. The irrigation plus natural rainfall amounts in 2007 in the A and D treatments were 68.7 and 51.9 cm, respectively. In 2008, total irrigation water applied to the A and D treatments was 13.9 and 5.8 cm, respectively, and the total irrigation plus natural rainfall amounts in the A and D treatments in 2008 were 60.9 and 52.8 cm, respectively.
The N rates were applied perpendicular to the line source. The five N treatments were 0, 0.25, 0.5, 1, and 1.5 times the current recommended N rate:where RNR is the recommended N application rate (kg ha−1), YG is the yield goal (Mg ha−1), STN is the NO3–N (kg ha−1) contained in the surface 60 cm of soil, and LC is the N credit from the previous legume crop (Gerwing and Gelderman, 2005). For STN, NO3–N was extracted with a dilute salt solution and then determined by electrode (Gelderman, 2008). the STN was 16.8 and 56 kg N ha−1 in 2007 and 2008, respectively. The yield goals were 5.7 and 6 Mg ha−1 for 2007 and 2008, respectively. Based on the previous field pea (Pisum sativum L.) yields, the LC was 22.4 and 33.6 kg N ha−1 in 2007 and 2008, respectively. All plots received starter fertilizer (9.6 kg N ha−1, 10.1 kg P ha−1 and 12 kg K ha−1) at planting. Based on yield goals, soil NO3–N levels, and legume credits, the N rates for 2007 were 0, 50 (0.25 RNR), 100 (0.5 RNR), 200 (RNR), and 300 (1.5 RNR) kg N ha−1. The N rates for 2008 were 0, 40 (0.25 RNR), 80 (0.50 RNR), 160 (RNR), and 240 (1.5 RNR) kg N ha−1. Fertilizer treatments (liquid urea–NH4NO3, 28–0–0) were applied using stream bar technology in the spring.
Water and Nitrogen Use Efficiency
Soil samples from the 0- to 15-cm (nine cores) and 15- to 60-cm (five cores) soil depths were collected in spring and during harvest. Spring samples were collected from each block, while harvest samples were collected from each plot. A subsample from each composite was analyzed for gravimetric soil moisture (10 g soil dried at 105°C for 48 h), with the volumetric water content calculated. The remaining sample was dried at 40°C, ground, sieved, and analyzed for NO3–N and NH4–N (Astoria-Pacific International, 1999a,b). For samples from the spring of 2007 and 2008, the amount of NO3–N plus NH4–N in the unfertilized controls was 76 and 69 kg N ha−1, respectively. The percentage of NH4 in the total soil inorganic N of the spring soil samples was approximately 36% in 2007 and 48% in 2008.
Plant samples were collected from 2 m of row at tillering, flag leaf, and crop maturity. These samples were dried in a forced-air oven at 60°C for at least 96 h. The dry plants were weighed and the biomass was determined. Grain yield was measured at crop maturity with a plot combine with a 1.52-m head. Grain yields were adjusted to 135 g kg−1 moisture content. Grain samples were ground and analyzed for total N, total C, δ15N, and δ13C on an isotope ratio mass spectrometer (Europa 20-20, Europa Scientific, Westchester, UK). The R (14N/15N and 13C/12C) ratios were used to calculate δ15N and δ13C (Clay et al., 2001, 2005). The value of δ13C was calculated aswhere Rsample is the 13C/12C ratio of the sample and Rstandard is the 13C/12C ratio of Pee Dee belemnite, a limestone standard for 13C. The δ13C values were used to calculate the 13C isotopic discrimination (Δ) value aswhere δ13Ca is the δ13C value of air (–8‰) and δ13Cs is the δ13C value of the grain sample.
Using the relationship between Δ and grain yield, the yield loss due to N stress (YLNS) and YLWS were calculated. The relationship between Δ and stress is based on water and N stress having opposite impacts on Δ. Water stress contributes to stomatal closure and lower Δ, while N stress contributes to reduced C fixation and increased Δ. Using the upper boundary line approach, which compares Δ and grain yield, YLNS and YLWS were calculated (Clay et al., 2001, 2005). This calculation assumed that water and N were the limiting factors.
The harvest index (HI), grain water use efficiency (GWUE), and GNUE were calculated asNitrogen mineralization (Nmin) from spring to harvest in the control plots was determined asThis calculation assumed that N loss during the growing season was minimal. This assumption was appropriate because soil water contents were maintained at levels that minimized NO3 losses to either denitrification or leaching.
For quality analysis, grain samples were machine and hand cleaned to ensure complete removal of foreign materials and broken kernels. Cleaned samples were analyzed for moisture and protein content using the ISIscan analysis package (InfraSoft International, State College, PA) for the Foss near-infrared analyzer (FOSS NIRSystems, Laurel, MD). Protein percentages are reported at 120 g kg−1 moisture content. Nanopure water was added and mixed with grain samples to make a final moisture content of 150 g kg−1. After 18 h of tempering (water and grain mixing), either 600 g (2007) or 500 g (2008) of tempered samples was milled following AACC International (2011a). Milled samples were weighed and sieved through a 0.25-mm sieve.
Farinograph analysis was conducted in a 50-g dough bowl using the constant dough method of AACC International (2011b). Flour moisture and protein were measured using the near-infrared reflectance method of AACC International (2011c,d). Water absorbance was determined based on the moisture content of the flour and the amount of water added to optimize the dough (discussed below). Water absorbance values are reported on a 14% moisture basis. The temperature of the farinograph was maintained at 30°C.
For each sample, the arrival time, peak time, water absorption, departure time, stability, mixing tolerance index (MTI), breakdown time, and 20-min drop values were determined (Fig. 1). The arrival time is the period of time (in min) from adding water until the top of the curve touches the 500 Brabender units (BU) line and indicates the flour hydration time.
The peak time is the time interval (to the nearest 0.5 min) from the first addition of water until the curve reaches its maximum height (Shuey, 1997). Peak time measures the length of time required to reach a maximum consistency (Bloksma and Bushuk, 1988; Sarker et al., 2008). If the height of the curve during peak time was not within the 500 ± 20 BU interval, the sample was mixed with a different flour/water ratio and rerun. A new ratio was determined based on the difference in water absorbance for the previous run. If the peak time occurred when the height of the curve was within the range (500 ± 20 BU), this time and the BU line were used for other calculations (Fig. 1).
The departure time is the period of time required for the curve to drop below the 500 BU line. This point indicates that gluten is breaking down and the dough has become overmixed. The stability is the period of time that the top of the curve remains above the 500 BU line. It is calculated by subtracting the arrival time from the departure time. The MTI is the difference in BU from the top of the curve at the peak time to the top of the curve 5 min after the peak time. The time to breakdown is the time from the start of mixing until the curve decreases by 30 BU from the peak point, whereas the 20-min drop is the difference in BU from the center of the curve at the peak time to the center of the curve 20 min after water addition.
Protein characterization was conducted on milled wheat samples from the 0 and recommended N rate treatments using high pressure liquid chromatography (Waters Corp., Milford, MA). Wet and dry gluten contents were determined with a hand-washing method according to AACC International (2011e,f). The relative concentrations of glutenin and gliadin were determined following the techniques of Sissons et al. (2005), whereas the HMW-GS and LMW-GS contents were determined following the techniques of Fu and Sapirstein (1996). Each analysis was done in duplicate.
Statistical Design and Analysis
The experiment contained two factors, N and water. The five N treatments were randomly assigned to plots within each block. A line source irrigation system running through the center of each plot was established, making N rate and water treatment perpendicular to each other. Since water treatments were not randomized and two sides of the lines source “direction” could have an impact on the analysis, a special technique was applied to analyze the data (Stroup, 1989). In this analysis, a first-order autoregressive model under PROC MIXED in SAS 9.1 (SAS Institute, 2008) was used to determine the treatment differences for the individual years. Block and its interaction with water and direction were random factors, whereas N and water were fixed factors. Water × N interactions were not significant for the measured variables.
RESULTS AND DISCUSSION
Nitrogen and Water Effect on Biomass and Grain Yield
In 2007, N fertilizer increased the biomass at the tillering and flag leaf stages (Table 1). The HI ranged from 0.54 to 0.41 and decreased with increasing N. Grain yield ranged from 3.31 to 4.71 Mg ha−1 and was lower for the 0 N than 1× N rate (Table 2; Fig. 2C). The GWUE was also lower for the 0 N than 1× N rate, whereas GNUE had the opposite result, decreasing with increasing N. Associated with the increase in GNUE in the A treatment was an increase in total grain protein, from 646 to 758 kg protein ha−1 in the D and A treatments, respectively. The A and D treatments had similar optimum N rates (Δyield/ΔN = 0 for the second-order equation relating yield and N) of 192 and 198 kg N ha−1 (Fig. 2C). These results indicate that irrigation water enhanced GNUE, whereas N enhanced GWUE.
|Treatment||Tillering||Flag leaf||Harvest||HI||Tillering||Flag leaf||Harvest||HI|
|Mg ha−1||Mg ha−1|
|Recommended N rate multiple|
|Treatment||Grain yield||YLNS||YLWS||Soil water at harvest||GWUE||GNUE|
|Mg ha−1||kg ha−1||kg ha−1||cm||kg cm−1||kg grain kg−1 N|
|Recommended N rate multiple|
|Recommended N rate multiple|
Grain yields ranged from 4.43 to 5.87 Mg ha−1 in 2008. Generally, the lowest yields and highest YLWS and YLNS were associated with the 0 N rates (Fig. 2F to 2H). The GWUE increased with increasing N and was 123 and 146 kg cm−1 in the 0 N and 1× N rate treatments, respectively. Associated with the increase in GWUE were decreases in YLWS and YLNS. These findings indicate that, in 2008, N had a beneficial impact on water use efficiency.
In 2008, the A water treatment had a lower YLWS and grain protein concentrations than the D water treatment (Tables 2 and 3; Fig. 2F). The decrease in protein was attributed to a water-enhanced increase in C fixation without a corresponding increase in protein. The response functions between yield and N rate suggest that, in 2008, wheat was N limited and had an optimum N rate (Δyield/ΔN = 0) that exceeded the maximum N rate used in the experiment. Differences in the water impacts on GNUE and optimum N rates between years were attributed to higher N mineralization in 2007 (192 kg N ha−1) than 2008 (99 kg N ha−1).
|Treatment||Protein||Arrival||Peak||Departure||Stability†||20-min drop||Time to breakdown||MTI||Water absorption|
|Recommended N rate multiple|
|Recommended N rate multiple|
These findings suggest that the implementation of a strategically applied N fertilizer program is more complicated than defining a yield goal and measuring N credits. In addition to the impacts of water on GNUE, it can also increase the yield response per unit of N applied. In both years, water increased the amount of grain produced per kilogram of N applied. For example, at the 1× recommended N rate in 2007, the grain production per kilogram of N was 20.8 ± 1.2 and 23.6 ± 1.1 kg in the D and A treatments, respectively, while the grain yield per kilogram of N applied in 2008 was 33.3 ± 2.1 and 36.0 ± 1.7 kg in the D and A treatments, respectively. In landscapes with different amounts of available water, similar responses would be expected (Mamani-Pati et al., 2010).
In 2007, the impact of N and water on protein was mirrored in many of the dough mixing characteristics (Table 3; Fig. 2E). Arrival time, peak time, departure time, stability, time to breakdown, and water absorption increased with increasing N and were correlated negatively with YLNS (Table 4), whereas the 20-min drop (inversely related to the dough strength) and MTI values decreased with increasing N and had positive correlations with YLNS. The YLWS was not correlated to any of these parameters except MTI, which provides a measure of dough softening during mixing. These findings suggest that N or water stress can result in weaker dough that softens more quickly during mixing. The impact of N on dough stability was consistent with the impact of N on protein.
|Pearson correlation coefficient†|
|Quality characteristic||YLNS||YLWS||Grain protein||Grain yield||YLNS||YLWS||Grain protein||Grain yield|
|Time to breakdown||–0.38||–0.14||0.74||0.39||–0.16||0.48||0.71||–0.3|
|Mixing tolerance index||0.39||0.24||–0.71||–0.45||0.15||–0.31||–0.56||0.17|
The longer arrival, peak, departure, stability, and breakdown times and higher water absorption in the 1× than the 0 N treatment were attributed to the influence of N on protein. Peak time is important because it provides valuable information about the length of time dough can be mixed before it starts to weaken. Others have reported that grain protein, peak time, and stability increase with increasing N (Terman et al., 1969; Boehm et al., 2004; Ma et al., 2009; Al-Eid, 2006).
To determine if total protein or protein composition was the controlling factor impacting dough quality, protein characterization of the 0 N and 1× N rate treatments was conducted. The gliadin/glutenin ratios were 2.5 ± 0.10, 1.91 ± 0.22, 2.26 ± 0.069, and 1.68 ± 0.32 for the 0N D, 1×N D, 0N A and 1×N A, treatments, respectively. These findings suggest that the gliadin/glutenin ratio was decreased by reduced plant stress (N and water). In addition, there was (i) a negative relationship between YLNS and the relative glutenin percentage (r = –0.68, P < 0.001), (ii) a positive relationship between YLNS and the gliadin/glutenin ratio (r = 0.71, P < 0.001), and (iii) a negative relationship between the gliadin/glutenin ratio and dough stability (r = –0.75, P < 0.001). Nitrogen addition also increased the relative percentage of HMW-GS from 17.8 ± 0.62 in the 0 N treatment to 21.8 ± 1.22 in the 1× N treatment.
In 2008, slightly different results were observed, with N and water both impacting the dough mixing characteristics (Tables 3 and 4). Similar to 2007, N increased protein, arrival time, peak time, departure time, stability, and water absorption, and decreased the 20-min drop (Table 3), whereas protein, arrival time, peak time, and departure time (10% level) were lower in the A than D treatments. In addition, YLWS was positively correlated to arrival time, departure time, peak time, stability, and time to breakdown and negatively correlated to 20-min drop, MTI, and yield. Similar to 2007, the gliadin/glutenin ratios in 2008 were decreased by N and irrigation and were 2.57 ± 0.13, 1.93 ± 0.09, 2.22 ± 0.07, and 1.78 ± 0.09 in the 0 N D, 1× D, 0 N A, and 1× A, treatments, respectively. Unlike 2007, however, both irrigation and fertilizer increased the relative percentage of HMW-GS. For example, the percentage of HMW-GS was lower in the 1× D treatment (21.6, P = 0.03) than the 1× A treatment (23.4). In addition, the relative glutenin percentages were correlated to YLNS (r = –0.52, P < 0.05) and YLWS (r = –0.51, P < 0.05).
Analysis across years showed that the chemical tests provided different information about wheat quality. For example, protein had a strong relationship (r = 0.78, P < 0.001) with water absorption (which is related to bread moistness and flour yield), while the HMW-GS/LMW-GS ratio had a strong relationship with dough stability. These findings suggest that to predict management impacts on wheat quality, total protein and the HMW-GS/LMW-GS ratio should be measured.
In 2007, the A water treatment increased yields and GNUE from 0.20 to 0.25 kg grain N kg−1 N fertilizer, reduced YLNS from 1141 to 480 kg grain yield ha−1, and in the 1× N rate treatment, increased the grain produced per kilogram of N applied from 20.8 ± 1.2 to 23.6 ± 1.1 kg. These benefits were achieved without a loss of wheat quality. These findings were attributed to a relatively high N mineralization rate and beneficial relationships between N and water. Slightly different results were observed in 2008, when the A water treatment decreased the YLWS from 1099 to 689 kg grain ha−1 and increased the grain produced per kilogram of N applied from 33.3 ± 2.1 to 36.0 ± 1.7 kg. Associated with these results were decreases in protein concentration and the arrival, peak, and departure times. Differences between 2007 and 2008 were attributed to lower spring temperatures in 2008 that reduced N mineralization and available N. A major consequence of reduced mineralization was less soil available NO3 for plant growth. These findings suggest that (i) a conceptual model where N and water are simultaneously taken up by the plant, as proposed by Kim et al. (2008) in corn (Zea mays L.), is operational in wheat, and that (ii) climate, soil, and management interact to influence wheat production, protein composition, and dough quality and that by understanding these relationships it may be possible to improve food security.