Recovery of Grain Yield and Protein with Fertilizer Application Post Nitrogen Stress in Winter Wheat (Triticum aestivum L.)

Abstract

Unfavorable weather conditions and lack of appropriate farm machinery often delay N application. This results in nitrogen (N) deficiency during the vegetative and early reproductive growth stages of winter wheat. The objective of this study was to evaluate the impact of N application timings (from tillering to flag leaf growth stages) on winter wheat grain yield and protein. The study was conducted across 12 site–years in Oklahoma, US. The treatments included a non-fertilized check, a pre-plant application of 100 kg N ha⁻¹, and ten in-season application timings at 100 kg of N ha⁻¹. The in-season treatment applications were initiated at the point when an N deficiency was visually identified by comparing the pre-plant treatment to the non-fertilized check. The treatments were applied in a progressive order every seven growing degree days (GDD > 0 °C) until a cumulative GDD of 63 was reached after visual deficiency (DAVD). All in-season treatments increased grain yield and protein as compared to the non-fertilized check, showing that N was a yield-limiting factor. The nitrogen applications made post Feekes 8 decreased grain yield when compared to pre-plant applications. Across this data set, that timing corresponded to a range of 21 to 63DAVD. The results suggested that forgoing N application until the growth stage Feekes 7, even when the visual N deficiency was highly apparent before that stage, had no negative impact on the yield, and it even increased the yield as compared to the pre-plant application in some cases. The plant developmental stage at which the N application takes place is more critical than the level of N deficiency. Our results show that N fertilizer applications should be made posteriorly to the crop dormancy to maximize both yield and protein, and that plants can recover from N deficiency when applications are made until the late-vegetative phase (Feekes 7). This document shows that winter wheat producers have a much wider N application window than traditionally believed.

1. Introduction

Wheat (Triticum aestivum L.) is one of the most planted cereal crops worldwide, representing 8% of the world crop production [1] and accounting for about 20% of the calories and protein of the daily nutritional requirements of humans [2]. Wheat yield needs to increase by approximately 50% to supply the growing population’s food demand by 2050 [3]. However, yields have stagnated in 27% of the wheat-producing areas in the world [4]. Nitrogen is the most limiting nutrient in wheat production and, overall, the second most limiting factor after water [5]. Nonetheless, N fertilization can promote the growth and development of stems, leaves, and roots, increasing photosynthesis and nutrient accumulation of winter wheat plants [6]. About 48% of the wheat yield improvements have been attributed to increased N fertilization and 28% due to improved genetics [7]. In Oklahoma, winter wheat is the most valuable crop, planted on approximately 1.78 million ha [8], and its production is predominantly rainfed [9]. In water-limited environments, improvements in N management practices could represent enhanced N use efficiency (NUE) (which is commonly below 50% [10]), decreasing the potential for environmental pollution due to over-fertilization.

Soil N application that is uncoordinated and/or excessive to meet crop demands can lead to potential environmental pollution [11]. The utilization of in-season optical sensors can increase the yield and NUE of winter wheat by optimizing the N rates when compared to conventional methods (i.e., the yield goal approach) [12,13]. However, for this management strategy to be better utilized, the in-season yield potential needs to be predicted. For an accurate yield potential estimation using optical sensors, winter wheat needs to be under N stress for the N application yield improvement to be calculated [12]. However, there is a lack of information concerning how long N applications can be delayed and yields maintained after a confirmed in-season N stress. Thus, determining the N application timing threshold is crucial for this optical sensing technology to be used by producers.

1.1. Nitrogen Dynamics in Winter Wheat

Both the genotype and the environment have a great influence on crop N uptake. According to Austin et al. [14], in a study analyzing 47 wheat genotypes, at anthesis, plants generally contained 83% of the total N taken up at physiological maturity, and, on average, 68% of the total N in the plant was present in the grain. Moreover, the authors also suggested a strong correlation between dry matter accumulation and N content at Feekes 10.5.3 and grain filling. Others [15,16] also reported this correlation between the biomass and N accumulation. Justes et al. [15] presented that when there is no N deficiency, the total N accumulation is determined by the plant’s growth rate and the soil N-nitrate availability. In addition, the authors noticed that under no stress, wheat could accumulate more N than the plants need to achieve a maximum growth rate. In a study performed under field conditions in Oklahoma, Girma et al. [16] documented that 61% of the total N accumulated by the plant was at growth stage Feekes 5. De Oliveira Silva [17] found that the N uptake by the hard winter wheat in Kansas at Feekes 10.5.3 would be equivalent to 82%.

Nitrogen deficiency can directly affect grain formation. By Feekes 5, plants begin stem elongation. The leaves are formed and all meaningful tillers contributing to the yield have ceased their growth. The growing point that generates new cells for the plant begins to develop an embryo head. At this stage, the number of spikelets is determined. A deficiency of N at this stage can be detrimental since N availability directly influences the number of seeds per head, as well as seed size.

During the final grain fill period (Feekes 10.5.4 to 11.3), protein accumulation is source-limited, whereas starch is sink-limited. Lollato and Edwards [18] documented that in low-yielding environments, the starch accumulation in the grains is reduced, resulting in a high protein concentration in the grains. This negative relationship between grain yield and protein concentration is well elucidated elsewhere in the literature [19,20]. Hence, under environmental conditions where the production potential is substantially higher, a starch accumulation was observed, which significantly reduced the protein concentration in the grain [18].

1.2. Nitrogen Application Timing

In general, most of the publications on N timing in winter wheat have found that the timing of application had an effect on the grain yield and grain protein concentration. The findings suggest that higher N losses during the growth period usually cause low N availability at the end of the plant life cycle, thus, reducing yield and protein content [21,22,23]. The highest N requirement found in the literature is during the period of the most rapid plant growth, which normally occurs after dormancy. Spring applications of N have been found to aid in the recovery from N stress during green-up [22,24]. An alternative for producers is to split the applications into multiple timings. This strategy can assist in the in-season evaluation of the crops and weather, thus, giving a better indication of the best timing for N fertilization, instead of a limitation on early-season decision [25].

One of the greatest limitations of the studies discussed previously is the lack of a wide breadth of N application timings in a single study. The majority of the work focuses on multiple rates applied to only a few timings. An additional limitation is the lack of documentation on if and when the crop displays signs of N stress. The objective of this study was to evaluate how long N application can be delayed during the winter wheat crop cycle without affecting grain yield and protein concentration when compared to pre-plant applications. For this, subsequent N applications were performed after the point at which N deficiencies were identified.

2. Materials and Methods

2.1. Site Description and Treatments

A field study was conducted over four growing seasons (2016–2017, 2017–2018, 2018–2019, and 2019–2020) and at five locations (Lake Carl Blackwell research farm (LCB) near Perry, OK; Cimarron Valley research station near Perkins, OK; Cross Country research farm near Stillwater, OK; the Raymond Sidwell research station near Lahoma, OK; and the Ballagh Family research farm near Newkirk, OK), comprising 12 site–years (

Table 1. Soil series classification and description, location, previous crop and planting date, and total rainfall (September to August) for all experimental sites utilized in the study.

Site–Year	Coordinate	Soil Series and Description	Previous Crop	Planting Date	Cultivar	Total Rainfall (mm)
Perkins2017	35°59′43.9″ N 97°02′35.9″ W	Teller; (fine-loamy, mixed, active, thermic Udic Agriustoll)	Wheat	13 October 2016	Bentley	997
Stillwater2017	36°08′12.9″ N 97°04′48.2″ W	Kirkland; (Fine, mixed, superactive, thermic Udertic Paleustolls)	Wheat	13 October 2016	Bentley	1003
LCB2017a	36°08′27.1″ N 97°17′05.1″ W	Port; (Fine-silty, mixed, superactive, thermic Cumulic Haplustolls)	Canola	12 October 2017	Doublestop CL Plus	896
LCB2017b	36°08′24.7″ N 97°17′00.2″ W	Pulaski; (Coarse-loamy, mixed, superactive, nonacid, thermic Udic Ustifluvents)	Wheat	12 October 2017	Doublestop CL Plus	896
Perkins2018	35°59′44.7″ N 97°02′38.9″ W	Konawa; (fine-loamy, mixed, active, thermic Ultic Haplustalf). Teller; (fine-loamy, mixed, active, thermic Udic Agriustoll)	Wheat	12 October 2017	Doublestop CL Plus	817
Lahoma2018	36°23′14.4″ N 98°06′30.9″ W	Grant; (Fine-silty, mixed, superactive, thermic Udic Argiustolls)	Wheat	12 October 2017	Bentley	1222
LCB2018a	36°08′24.7″ N 97°17′02.9″ W	Pulaski; (Coarse-loamy, mixed, superactive, nonacid, thermic Udic Ustifluvents)	Wheat	11 October 2017	Doublestop CL Plus	937
LCB2018b	36°08′22.4″ N 97°17′00.8″ W	Pulaski; (Coarse-loamy, mixed, superactive, nonacid, thermic Udic Ustifluvents	Fallow	11 October 2017	Doublestop CL Plus	937
Newkirk2019	36°47′45.1″ N 96°59′48.7″ W	Agra-Foraker (Fine, mixed, superactive, thermic Udertic Paleustolls)	Alfalfa	24 October 2018	Bentley	1390
LCB2019	36°08′23.6″ N 97°16′59.9″ W	Pulaski; (Coarse-loamy, mixed, superactive, nonacid, thermic Udic Ustifluvents)	Wheat	7 November 2018	Doublestop CL Plus	1368
Newkirk2020	36°47′43.5″ N 96°59′49.0″ W	Agra-Foraker (Fine, mixed, superactive, thermic Udertic Paleustolls)	Wheat	18 October 2019	SY Monument	892
LCB2020	36°08′25.0″ N 97°17′00.0″ W	Pulaski; (Coarse-loamy, mixed, superactive, nonacid, thermic Udic Ustifluvents)	Wheat	15 October 2019	Doublestop CL Plus	861

). The site–year and location, soil series and description, previous crops for each location (all sown at 84 kg ha⁻¹), planting date, cultivar, and total rainfall are described in Table 1. The total rainfall values were retrieved from the Mesonet for the station closest to the research area (www.mesonet.org, accessed on 15 February 2022).

A month before the initiation of the trials, 0–15 cm composite soil samples, composed of 15 soil cores, were collected from the entire trial areas, as outlined in Zhang and Arnall [26]. The soil pH was determined with a 1:1 soil to deionized water ratio by using an ion-selective (H⁺) glass electrode [27]. For the nitrate (NO₃─N), 5 g of soil was extracted with 25 mL of 1 M KCl, filtered through a 0.45 μm Whatman filter, and then analyzed on a Lachat QuickChem 8500 Series 2 Flow Injection Analyzer (Hach Co., Loveland, CO, USA). The plant-available phosphorus (P) and potassium (K) were extracted using the Mehlich 3 (M3) [28] solution and determined using an inductively coupled plasma by atomic emission spectroscopy (ICP-AES) (SPECTRO Analytical Instruments Kleve, Germany). The soil analysis results are presented in

Table 2. Soil pH and nutrient contents of all experimental sites.

Location/Year	pH	NO3─N	P	K
		mg kg−1
Perkins2017	5.6	15.5	18	96.5
Stillwater2017	6.1	15.5	5	54
LCB2017a	5.9	15.5	12	62
LCB2017b	5.7	10.5	14	49.5
Perkins2018	5.6	13	12.5	71.5
Lahoma2018	5.6	10.5	25	139
LCB2018a	5.8	2.5	10.5	45.5
LCB2018b	6.2	8.5	16.5	75.5
Newkirk2019	7.1	8	24	107
LCB2019	5.3	14.5	110	373
Newkirk2020	7.7	12	56	482
LCB2020	6.3	38	98	494

.

At Perkins2017, Perkins2018, and Stillwater2017, 84 kg ha⁻¹ of diammonium phosphate (DAP) was applied in-furrow at planting because of the soil acidity at Perkins and the low soil test for P at Stillwater2017, which would limit the yield potential [29].

At all locations, the treatments were arranged in a randomized complete block design (RCBD) with three replications (n = 3); the sites Newkirk2019, LCB2019, Newkirk2020, and LCB2020 were replicated four times (n = 4). The plot size was 3.1 m × 6.1 m. All of the areas were under no-till (NT) management. Due to half of the field at LCB being in a systematic crop rotation, the project established two studies at that site during the growing seasons. The best management practices for small grains followed the recommendations of the Oklahoma State University for pest and weed management.

The treatments included a non-fertilized check, a pre-plant N application, and ten in-season N application timings (

Table 3. Treatment structure implemented at all locations. Nitrogen application timings of the in-season treatments based on the growing degree days after visual N deficiency (DAVD) between the non-fertilized check and the pre-plant N treatment.

Treatment	Fertilizer Timing	Nitrogen Rate (kg ha−1)
1	Pre-plant	100
2	Check	0
3	0 DAVD	100
4	7 DAVD	100
5	14 DAVD	100
6	21 DAVD	100
7	28 DAVD	100
8	35 DAVD	100
9	42 DAVD	100
10	49 DAVD	100
11	56 DAVD	100
12	63 DAVD	100

). All timings received a rate of 100 kg N ha⁻¹, applied as ammonium nitrate (AN, NH₄NO₃). The rate of 100 kg N ha⁻¹ was chosen as a yield-limiting rate to represent any impact on N fertilization efficiency [22]. The hypothesis for the approach was that if N was unlimited, the study would be unable to quantify the occasional difference in efficiencies gained or lost due to the application timing. Ammonium nitrate was applied in place of the commonly used urea (46-0-0) to reduce N volatilization [30,31]. The AN was applied to the plots via a surface broadcast.

2.2. Determination of the Visual N Deficiency for the Treatment Application

The initiation of the in-season application was based on the visual N deficiency between the pre-plant treatment and the non-fertilized check. The locations were visited weekly after sowing. At the point at which a visual difference in the crop biomass or the leaf greenness was identified, it was considered visually responsive. At this point, a GreenSeeker^® handheld sensor was used to collect the normalized difference vegetation index (NDVI) values from the pre-plant fertilized treatment and the non-fertilized check plot (

Table 4. Documented normalized difference vegetation index (NDVI) from the pre-plant (treatment 1) and the non-fertilized check (treatment 2) at visual deficiency in all site–years. Fertilizer applications and NDVI collection were initiated when the pre-plant treatment within at least one block was visually determined to be greener or had more biomass than the non-fertilized check.

Location	Mean Pre-Plant	Mean Check	Mean Difference	Prob > F
Perkins2017	0.56	0.44	0.12	0.05
Stillwater2017	0.61	0.48	0.13	0.09
LCB2017a	0.61	0.55	0.07	0.01
LCB2017b	0.39	0.33	0.05	0.16
Perkins2018	0.64	0.54	0.10	0.09
Lahoma2018	0.47	0.46	0.01	0.79
LCB2018a	0.35	0.31	0.05	0.14
LCB2018b	0.32	0.31	0.01	0.59
Newkirk2019	0.30	0.29	0.02	0.05
LCB2019	0.35	0.24	0.11	0.09
Newkirk2020	0.45	0.39	0.06	0.09
LCB2020	0.49	0.42	0.06	0.12

). The in-season treatments were applied at 0, 7, 14, 21, 28, 35, 42, 49, 56, and 63 growing degree days > 0 (GDD > 0) after visual deficiency (DAVD). The daily temperature values were retrieved from the Mesonet Wheat Growth Day Counter for the station closest to the research area (www.mesonet.org, accessed on 15 February 2022). GDD > 0 is calculated as follows:

1 GDD > 0 = (Day Max Temperature + Day Min Temperature)/2 − 4.4 °C

(1)

Only if

(Day Max Temperature + Day Min Temperature)/2 − 4.4 °C > 0 °C

(2)

or

(Day Max Temperature + Day Min Temperature)/2 − 4.4 °C < 30 °C

(3)

2.3. Grain Yield and Protein

At grain maturity, the center 1.5 m of the plots were harvested with a Massey Ferguson 8-XP plot combine (Kincaid Equipment Manufacturing; Haven, KS, USA). Data on the grain yield and percent moisture content were recorded by the onboard Harvest Master Yield monitoring computer (Juniper Systems; Logan, UT) and grain samples were collected from each plot at harvest. To standardize all grain yields, the moisture content was adjusted to 12.5%. The grain protein was determined post-harvest using near infrared spectroscopy, Diode Array NIR Analysis Systems model DA 7200, Perten (Kungens Kurva, Sweden).

2.4. Data Analysis

Data were analyzed using the JMP 15 PRO^® (SAS Institute, Cary, NC, USA) for year-specific crop production factors, such as grain yield and protein. Data were differentiated using the ANOVA methods and Dunnett’s to separate the means at α = 0.05. The controls used for the test were as follows: the check (trt1), pre-plant (trt2), and 0DAVD (trt3). For the pre-plant and 0DAVD comparisons, the check treatment was removed from the data.

3. Results and Discussion

3.1. Timing of Response to Nitrogen

As we aimed to evaluate the N fertilizer application after visual N deficiencies were observed, the N treatments were applied at a range of dates across all site–years. Supplementary Table S1 provides the dates for all treatment applications. Even though the planting date and application of the pre-plant were very similar across all sites, the date of 0DAVD ranged from 11 November to 28 March. Even in the same season within the same field, LCB, the previous crop impacted 0DAVD by 40 and 45 days in the first and second crop years, respectively. The difference in the 0DAVD date across the site–years, along with the impact of location on the accumulation on GDD’s > 0, created a range of dates for 63DAVD from 12th February to 14th May, spanning a window of growth stages from Feekes 4 to Feekes 10.5. This range of application dates presented an opportunity to evaluate the N application over a wide range of physiological growth stages and, yet, created a challenge in that there is an inconsistency in the growth stages evaluated across the site–years.

3.2. Grain Yield and Protein Response to Nitrogen

To provide a general overview of all of the site–years, Figure 1, Figure 2 and Figure 3 graphically represent the grain yield and grain protein concentration for each treatment. In order to determine if an N response occurred, a multiple comparison utilizing a Dunnett’s test (pre-plant N application as a control) was performed by location (Supplementary Table S2). When evaluating the grain yield, a significant main effect of N was found when comparing the pre-plant N applied treatment to the non-fertilized check in 9 out of the 12 site–years. Perkins2018, LCB2018b, and LCB2019 were the locations at which there was no difference between the pre-plant N and the non-fertilized check. However, the grain yield of 0DAVD was statistically greater than the check at LCB2018b and LCB2019. The completion of this study resulted in 11 N-responsive experimental trials.

The in-season applications at LCB2018b and LCB2020 started earlier in the growing season than the visual deficiency between the check plots and the pre-plant. The actual deficiencies for LCB2018b and LCB2020 were noticed during the application of treatment five (14DAVD) and treatment four (7DAVD), respectively. Nitrogen was applied prior to the visual deficiency as the crop was already in an advanced stage and the range of application of the other treatments would override the agronomic interest.

3.3. Impact of Nitrogen Application Timing on Grain Yield

To analyze the impact of in-season N application timing on the yield and grain protein concentration, a multiple comparison from a Dunnett’s test was performed using the pre-plant treatment as a control (Supplementary Table S2). The multiple comparison analysis was also performed utilizing the 0DAVD as another reference for grain yield and protein (Supplementary Table S3). The first allows the evaluation of in-season N application, while the second addresses the purpose of the study for evaluating the ability of winter wheat to recover from N stress.

As shown in Supplementary Table S2, only 8 out of the 117 (6.8%) in-season N application comparisons with the pre-plant application yielded less when the N application was delayed until 30th March (approximately Feekes 8) or later. When compared to the 0DAVD (Supplementary Table S3), 10 out of the 117 (8.5%) comparisons showed a significant decrease in the yield. Nine of those ten were in-season applications, and one was a pre-plant application. Six of those nine were found in the Newkirk2019 and LCB2019 trials. All those applications were made on the 15th April (approximately Feekes 8) or later. While not always significant, there was a numeric decrease in the yield when compared to 0DAVD as noted in the majority of the N applications made in April or later (Figure 1, Figure 2 and Figure 3).

In the 2016/17 cropping year, 1 (3%) out of the 38 in-season–application comparisons to the pre-plant N application showed a significant decrease in yield, while 16 (42%) resulted in a significant yield increase. Due to the early visual deficiency at Perkins2017 and Stillwater2017, the range of in-season applications was completed by mid/late February. Therefore, because of the early N application start, we were not able to verify the point that the N stress could no longer be recovered from. However, results displayed that delaying application was possible up to 63 days after the development of an N deficiency without grain yield or protein concentration loss. Conversely, there was no timing effect during the 2017/18 and 2018/19 cropping seasons, in which grain yield was significantly greater than the pre-plant or 0DAVD (Supplementary Tables S2 and S3). A total of 1 out of the 39 in-season applications made to the pre-plant resulted in a significant increase in the yield for the 2018–2019 and 2019–2020 crop seasons and 4 yielded significantly less. These four applications were made in Newkirk2020 after the 6th April, which corresponded to the growth stage of Feekes 9.

Across the 11 N responsive site–years, the grain yields achieved from applications made after the visual deficiency were equal to or greater than that of the pre-plant application if the N was applied before Feekes 8. This result was achieved regardless of the yield level or rainfall distribution, which ranged from 897 to 1390 mm across all site–years (Table 1).

3.4. Impact of Nitrogen Application Timing on Grain Protein Concentration

The grain protein concentration decreased only once when compared to the pre-plant across all locations (Supplementary Table S2). This one timing, LCB2018b at 64DAVD, was made on the 2nd May. During this time, the crop was in the early stages of grain-fill. Out of the 117 comparisons for protein concentration performed against the pre-plant treatment across all locations, the in-season application of N significantly increased protein concentration 48 times (41%). It should be noted that at Perkins2018, where there was no significant yield response to the N application, nine positive grain protein responses to the N fertilization were observed. Perkins2017 was the only site to present no positive response for the protein concentration to N application timing. Perkins2017 had the earliest date for the 63DAVD, which may have influenced such a result. As seen in Figure 1, Figure 2 and Figure 3, the protein concentration increased with time. However, the increase in the protein concentration was most dramatic when the timing of the N application progressed past Feekes 8.

As with the yield, the number of significant positive responses to the N application was reduced when the in-season treatments were compared to 0DAVD (Supplementary Table S3). This indicates that delaying N application until the first visual deficiency will likely produce higher protein values than when all of the N is applied at the pre-plant.

3.5. Nitrogen Stress Recovery

While the visual differences between the pre-plant fertilized plot and the non-fertilized check were highly variable, the data showed that the N applications, made at or after the point at which the N deficiency was visually evident, resulted in yields equivalent to or greater than the pre-plant application in all four cropping seasons regardless of the location. As stated by Ravier et al. [32], N deficiencies can be tolerated in the vegetative stages of the crop with no decrease in grain yield or protein content. In some cases, N deficiencies can be beneficial in the early stages compared to high levels of N nutrition. Increased biomass and leaf area caused by N fertilization can increase the risk of foliar disease [33], crop lodging due to excessive stem elongation [34], and intensify water stress due to higher evapotranspiration [35].

Nonetheless, the pre-plant application was statistically equivalent in yield when compared to the application made at the point of visual deficiency (trt3), with the exception of LCB2017a. In this site–year, the pre-plant N application yielded 0.53 Mg ha⁻¹, which is less than the 0DAVD, 3.84 and 4.37 Mg ha⁻¹, respectively. Recovery of the yield equivalent to the pre-plant occurred seven times at 63DAVD, twice at 56DAVD, and at one location at 42DAVD and 35DAVD. However, the ability of the crop to recover to the equivalent levels of the 0DAVD after a prolonged deficiency was similar to that of the pre-plant comparison. Eight locations documented a full recovery until 63DAVD and one location at 56DAVD, 35DAVD, and 28DAVD. As stated by Jeffroy and Bouchard [36], the impact of N deficiencies on grain yield is directly related to the period and intensity of the deficiency. Although this work lacks the ability to quantify the intensity of the deficiency, it did not show that the duration of the deficiency impacted grain yield when N applications were made prior to Feekes 8.

While applying N after a visual deficiency resulted in grain yields that were equal to or better than the pre-plant until the point of Feekes 8, grain protein concentrations increased as application dates moved beyond Feekes 8, regardless of the date that the deficiency developed. These results are not surprising as protein accumulation begins approximately 10 days after Feekes 10.5.3 [37]. In addition, the final grain protein concentration can be greatly influenced by the grain yield and late-season N applications. In agreement with other studies, we documented that in low-yielding environments, the starch accumulation in the grains was reduced, resulting in a higher protein concentration in the grain [19,38]. However, in environments where the productive potential was substantially higher, starch accumulation occured, thus, significantly reducing the grain protein content. The extreme delay on N resulted in a reduced final grain yield and a high N availability during Feekes 10.5.3.

4. Conclusions

The application of N and its timing had a significant effect on the yield and protein concentration of winter wheat. Contrary to conventional thoughts, data from this study suggest that the agronomic optimum timing was not related to the timing of N deficiency. Although a negative effect on the grain yield was observed, it was only in the scenarios where N was applied after late March, which for this work corresponded to the Feekes 8 growth stage. This study documented no negative impact on the grain yield or protein concentration when N application was made after a deficiency had occurred. Notably, the yield and protein values of the in-season N application for all site–years were equal to or greater than those of the pre-plant treatments.

A significant finding of this work was that reaching the maximum achievable yield levels was not related to the point in time at which the crop goes under N stress. Rather, an optimum grain yield recovery was mostly associated with the N applications made when the crop was most vigorously growing during Feekes 6 to 8 growth stages. The evaluation of N timing on protein concentration displayed a trend for increasing protein values with an increasing delay of application up to May, approximately Feekes 10. These results were consistent across a diverse range of soils, cultivars, yield levels, and precipitation amounts and distributions.

These results are significant for the winter wheat growers of the southern Great Plains as this research documents not only the ability, but also the necessity to move away from pre-plant and fall N applications for winter wheat grain production. The window for N application is likely much greater than most wheat producers would have ever considered. This work showed that not only could N be delayed and the yield not be sacrificed, but when delayed, the yield would be maintained and the protein concentration increased. Thus, the study provides significant evidence that N management strategies that rely upon optical sensing technologies can be utilized to limit potential pollution and economic risk, as well as minimize yield loss.

Further studies should take into consideration the timings of the applications in other environments and conditions, as this study was limited to the rainfed conditions of a limited region in Oklahoma. Expanding the study to other environments, including a range of planting dates and cultivars, would provide a more robust understanding of the influence of N application timing on grain yield and protein and the way producers should manage it.