Evaluation of Sensor-Based Nitrogen Rates and Sources in Wheat

Walsh, Olga S.; Shafian, Sanaz; Christiaens, Robin J.

doi:https://doi.org/10.1155/2018/5670479

International Journal of Agronomy

On this page

Abstract Introduction Methods and Materials Results Conclusions Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article | Open Access

Volume 2018 | Article ID 5670479 | https://doi.org/10.1155/2018/5670479

Evaluation of Sensor-Based Nitrogen Rates and Sources in Wheat

Olga S. Walsh,¹Sanaz Shafian,¹and Robin J. Christiaens²

Academic Editor: Yuanhu Xuan

Received25 Sept 2017

Accepted08 Nov 2017

Published01 Jan 2018

Abstract

Nitrogen (N) is one of the most essential nutrients needed to reach maximum grain yield in all environments. Nitrogen fertilizers represent an important production cost, in both monetary and environmental terms. The aim of this study was to assess the effect of preplant nitrogen (N) rate and topdress N source on spring wheat (Triticum aestivum L.) grain yield and quality. Study was conducted in North-Central and Western Montana from 2011 to 2013 (total of 6 site-years). Six different preplant nitrogen (N) rates (0, 220, 22, 44, 67, and 90 N rate, kg ha⁻¹) followed by two topdress N sources (urea, 46-0-0, and urea ammonium nitrate (UAN), 32-0-0) were applied to spring wheat (Triticum aestivum L.). The results showed that there were no significant differences in grain yield, protein content, or protein yield, associated with topdress N source.

1. Introduction

Wheat is one of the most important crops which provide nutrition (proteins, energy, and minerals) to most of the world population. United States Department of Agriculture (USDA) has in its World Agricultural Supply and Demand Estimates (WASDE) forecasted global wheat production to reach 743 million metric tons (MMT), and the demand for wheat in developing countries is projected to increase 60% by 2050 (https://www.proag.com/news/world-wheat-consumption-increases-spring-wheat-in-short-supply/). Wheat ranks third among the US field crops in planted acreage, production, and gross farm receipts, behind corn and soybeans (https://www.ers.usda.gov/topics/crops/wheat/). Excellent growing conditions for much of the US, especially throughout the Great Plains states such as Kansas, Montana, North Dakota, Nebraska, and Oklahoma, contribute to excellent US wheat production levels. In 2016-17, the US farmers produced a total of 24.6 million US tons of winter, spring, and durum wheat on 50.2 million acres of cropland (https://www.ers.usda.gov/topics/crops/wheat/).

Wheat grain yield and quality depend on multiple genes interacting with environmental factors, such as nitrogen (N) availability, water, and temperature [1, 2]. Nitrogen is generally the most limiting nutrient factor for wheat production by influencing on chlorophyll production, photosynthesis process, and grain yield and quality [3]. Increasing N supply to a wheat crop can increase photosynthesis rate and consequently increase canopy biomass and grain yield. However, excessive N fertilization in cultivated land has profound environmental impacts such as nitrate leaching, soil denitrification, ammonia volatilization, and nitrous oxide emissions, which contaminate water and air and aggravate the climate change [4, 5]. While N losses cannot be avoided completely, they can be substantially reduced with sustainable agricultural practice such as adjusting N rate, timing and using the most appropriate fertilizer source, and adopting precision nutrient management technologies [6].

Crop canopy sensors have been developed to help growers to assess crop N status and make the best management decision based on actual crop condition and increase crop NUE [7]. Crop canopy sensors are designed to measure crop spectral reflectance at specific wavelength and generate vegetation indices (VIs) such as Normalized Difference Vegetation Index (NDVI) to assess specific crop characteristics of interest such as N status [8, 9]. These VIs can be used to generate field maps that show spatial variability of crop condition. These VIs have also been integrated into different algorithms to estimate crop N content, make fertilizer N recommendations, and increase economic return for producers [10]. Over 30 sensor-based algorithms have been developed since early 1990s. These algorithms are crop, region, and sensor specific and need to be calibrated for new crop in new region and new sensor used. For example, spring wheat (US, Canada, and Mexico) algorithm was developed as a collaborative effort of Oklahoma State University, Agriculture and Agri-Food Canada, and International Maize and Wheat Improvement Center (CYMMIT). This algorithm is a modified version of the spring wheat algorithm developed by the Oklahoma State University [11]. This algorithm is based on the relationship between sensor data collected early in the growing season and the yield attained across years and locations. The maximum wheat yield potential was set at 8000 kg ha⁻¹ and the NUE at 35%. [12]. This algorithm is based on one single N application, immediately after sensing event, which is recommended to occur at Feekes 5-6 while the timing of applications and the source of N can affect the amount of N recovered or the nitrogen use efficiency (NUE) in a given year.

In the Great Plains, wheat growers can choose to apply all N before planting or split N fertilizer between preplant starter and the midseason topdress. Several N fertilizer sources—dry and liquid— are available to producers. The growers often chose N source based on current price per unit of N as well as currently available application equipment. Contrasting results in comparing dry with liquid N products in wheat have been reported. Several studies demonstrated that there are no differences in NUE between dry and liquid fertilizers [13, 14]. Some studies showed that NUE might be lower with liquid N sources due to significant ammonia loss [15]. Other researchers found that liquid N products may be superior in terms of crop yield and quality, as well as being more environmentally friendly, due to greater plant availability and thus more efficient uptake [16]. For example, in a long-term winter wheat study in Oklahoma (total of 80 site-years), liquid N products resulted in an almost 20% advantage in NUE compared to dry granular N source. Knowing the relative NUE of preplant and topdress and what type of N products is more efficient, dry or liquid one, are important for calibrating the developed algorithms, estimating crop N needed with higher accuracy to reach the potential yield plateau.

The objective of this study was to determine whether dry or liquid N source would result in significantly superior results in spring wheat production. Specifically, we assess the effect of preplant N rate and topdress N source on spring wheat grain yield and quality.

2. Methods and Materials

2.1. Study Areas

This study was conducted for three consecutive growing seasons (2011–2013) at two experimental locations: one dryland site, at Western Triangle Agricultural Center (WTARC), Conrad, MT (48.309794 and −111.924684), and one irrigated site at Western Agricultural Research Center (WARC), Corvallis, MT (46.328179 and −114.089873). Initial soil test results (2011) for each location are detailed in Table 1. The experimental design was a randomized complete block design (RCBD) with four replications. The main plot treatments included six preplant N rates (0, 22, 44, 67, 90, and 220 kg N ha⁻¹) applied at planting by side-banding granular urea (46-0-0) approximately 2.5 cm below the seed. The subplot treatments were two topdress N sources (dry granular urea and liquid urea ammonium nitrate (UAN)) (28-0-0) (Table 2). Treatment 1 (0 kg N ha⁻¹) was established as an unfertilized check plot and treatment 2 (220 kg N ha⁻¹) served as a nonlimiting N-rich reference. In total, there were 40 plots at each experimental location. Hard red spring wheat (Triticum aestivum cv. Choteau) was planted with a small plot drill with Conserva Pak™ openers manufactured by Swift Machining (Washougal, WA) at a density of approximately 1.8 million plants per hectare.

2.2. Field Data Collection and Generating Topdress N Prescriptions

Within each study plot, Normalized Difference Vegetative Index (NDVI), was measured using a GreenSeeker hand-held optical sensor (Trimble Navigation Ltd., Sunnyvale, CA) at Feekes 5 of growing stage (early jointing, beginning of stem elongation, prior to first visible node). Then, the Spring Wheat (US, Canada, Mexico) algorithm (http://www.soiltesting.okstate.edu/SBNRC/SBNRC.php) was used to calculate topdress N rates for each treatment. The algorithm uses maximum NDVI (measured NDVI in treatment 2, nonlimiting N reference), measured NDVI for each treatment, seeding date, date of sensing (NDVI measurement date), and yield goal (average yield goal for the area) as input parameters to calculate topdress N rates. In some cases, treatment 2 did not result in the highest NDVI for the location, possibly, due to the negative impact of a high N rate applied at planting affecting the seed. Then, a treatment with the highest NDVI values was chosen and used as a reference. The topdress N fertilizer was applied as urea (as dry prill, broadcasted manually) or as UAN (as a foliar spray, using a battery-operated backpack sprayer with a fan nozzle).

At maturity, wheat was harvested with a self-propelled Wintersteiger Classic Combine (Wintersteiger Inc., Salt Lake City, USA). At harvest, plot grain yield was recorded for each experimental plot using a Harvest Master GrainGage, by Wintersteiger. The harvested wheat grain was dried in the drying room for 14 days at the temperature of 35°C; then, the dried samples were weighed to determine the accurate by-plot grain yield, which was adjusted to 12% moisture. The by-plot subsamples were analyzed for total N content using near infrared reflectance spectroscopy (NIR) with a Perten DA 7250 NIR analyzer (Perten Instruments, Inc., Springfield, IL) at Agvise Laboratories (Northwood, ND).

2.3. Statistical Analysis

Grain N uptake for each plot was calculated by multiplying grain yield by total N concentration. The analysis of variance was conducted using the PROC GLM procedure in SAS v9.4 (SAS Institute, Inc., Cary, NC) and the mean separation was performed using the Orthogonal Contrasts method at a significance level of 0.05.

3. Results and Discussions

3.1. Relationship between Preplant N Rate and NDVI

Responses of NDVI to preplant N application rate for 2011, 2012 and 2013 at each study site are presented in Figure 1. Generally, there was a significant positive correlation between NDVI and preplant N application rate for all site-years.

The relationships between NDVI and preplant N application rate were linear or quadratic in nature (Figure 1). Positive quadratic relationships were more frequently observed between preplant N application rate and NDVI when wheat NDVI values were relatively high.

At Conrad site, the average NDVI values were 0.40, 0.47, and 0.58 in 2011, 2012, and 2013, respectively. Conrad was a dryland site, so differences in NDVI values for same preplant N application rate in different years could be related to varied precipitations and, consequently, differences in soil moisture content. In 2011 season, the amount and the distribution of rainfall from May to July, from planting time to the time the NDVI was measured, were higher than in 2012 and 2013. Moreover, the daily rainfall distribution was erratic; at times, heavy rain has caused significant runoff and leaching in 2011 compared to 2012 and 2013. This may explain lower average NDVI value at Conrad in 2011 compared to other two growing seasons. In 2013 Conrad study site received almost 20 cm of snowfall in April, before wheat planting, which increased the soil moisture content considerably. The higher soil moisture amounts available at wheat planting has likely resulted in higher NDVI values in 2013 compared to 2011 and 2012.

At Corvallis study site, the average NDVI values were 0.54, 0.49, and 0.33 in 2011, 2012, and 2013, respectively. At Corvallis, plots received 15.24 cm of irrigation water in three equally space periods between planting and soft dough stage, annually. Precipitation data can explain the differences in average NDVI values in different years. This field received 11.25 cm, 8.08 cm, and 7.76 cm of precipitations in 2011, 2012, and 2013, respectively, from planting to the date of the NDVI measurement. The higher rainfall in 2011 apparently may have resulted in higher NDVI compare to other years.

For Conrad in 2012 and 2013 and for Corvallis in 2011 and 2012, treatment 2, 220 kg N ha⁻¹, did not reach the highest NDVI value. This was possibly because the 220 kg N ha⁻¹ preplant application rate might have been damaging to seeds and/or seedlings and produced relatively poor stand. The decision was made to use a treatment with the highest NDVI value as an N-rich reference for those particular site-years.

3.2. Relationship between NDVI and Prescribed Topdress N Rates

The algorithm has prescribed topdress N rates ranging from 0 to 137 kg N ha⁻¹ depending on the yield goal for the location and the obtained NDVI values (Tables 3–5). As it mentioned in previous section, in several instances that treatment 2 did not reach to the highest NDVI value the decision was made to use a treatment with the highest NDVI value as an N-rich reference for that particular site-year.

In some cases, the algorithm prescribed higher topdress N rate for treatment with lower NDVI values in order to boost yield potential. For example, at Conrad, in 2012, measured NDVI showed higher value in treatment 6 compared to treatment 2 (Table 6). Prescribed 30 kg N ha⁻¹ topdress N for treatment 6 resulted in a total of 120 kg N ha⁻¹ applied, while 78 kg N ha⁻¹ topdress N prescribed for treatment 2 resulted in a total of 298 kg N ha⁻¹ applied. However, treatment 6 yielded over 1000 kg ha⁻¹ more compared to treatment 2. This suggests that prescribed topdress N rate for treatment 2 was excessive and did not help to optimize yield.

In some instances, the prescribed N rates did not allow to reach optimum yield (Conrad, 2011 (Table 6)). Although higher topdress N rate of 33 kg N ha⁻¹ was prescribed to treatment 7 compared to 11 kg N ha⁻¹ recommended for treatment 6, it was not adequate to optimize yield. In fact, significantly lower yield was obtained with treatment 7 (1480 kg ha⁻¹), which received only 22 kg N ha⁻¹ at seeding, compared to treatment 6 (2152 kg N ha⁻¹) to which 90 kg N ha⁻¹ was applied at seeding. Moreover, the highest grain yield for that site-year was 2690 kg N ha⁻¹ (treatment 2, 220 kg N ha⁻¹ applied at seeding), which suggests that topdress N rates should have been higher for both treatment 6 and treatment 7.

In some cases, the topdress N rates seemed appropriate. For instance, at Conrad, in 2012, higher preplant N rate for treatment 6 (90 kg N ha⁻¹) resulted in better plant stand, reflected by the greater NDVI value compared to treatment 3 (22 kg N ha⁻¹ preplant). This prompted higher N rate recommendation of 25 kg N ha⁻¹ for treatment 6 compared to only 15 kg N ha⁻¹ for treatment 3. A difference of 98 kg N ha⁻¹ total N applied has resulted in a surplus of 605 kg ha⁻¹ of wheat grain yield.

These results show that, at all site-years, N fertilizer rates recommended by the USA/Canada/Mexico Algorithm were not appropriate for grain yield optimization. For example, much higher topdress N rates were prescribed for Corvallis (the irrigated site) compared to the Conrad (dryland sites). This makes sense since the expected yield potential (YP) at the irrigated site was much greater. On the other hand, grain yields obtained at Conrad were just as high as at Corvallis, indicating that the YP was either overestimated at Corvallis or underestimated at Conrad. This puts forward a question of whether there is a need for two separate algorithms, one developed for dryland spring wheat and the other for irrigated spring wheat production systems.

3.3. Relationship between N Rate and Source and Wheat Yield and Quality

Based on different N application rates and sources, a wide range of yields ranging from 942 kg ha⁻¹ to 7667 kg ha⁻¹ were obtained (Tables 3–5). In 2011 and 2012 growing seasons, grain yield response to preplant N application rate was significant at all locations (Table 7). In Conrad, the highest mean grain yield of 7263 kg ha⁻¹ was obtained in 2012 when 90 kg N ha⁻¹ preplan followed by 30 kg N ha⁻¹ topdress was applied as urea. In Corvallis, the highest mean yield of 7667 kg N ha⁻¹ was obtained in 2012 when 90 kg ha⁻¹ preplant followed by 110 kg ha⁻¹ topdress was applied as UAN. In contrast, the lowest grain yield, except the control, was recorded when 22 kg N ha⁻¹ preplant followed by 33 kg N ha⁻¹ of topdress in Conrad in 2011. These results showed that, in most cases, the higher amount of N applied resulted in higher grain yield, which in consistent with results from other studies [17, 18].

Although the application of urea resulted in slightly higher yields compared to UAN, with 1 site-year being virtually equal in yield for both N sources, the differences were no statistically significant. In addition, the results showed there were no significant differences in grain protein content values associated with topdress fertilizer N source (urea versus UAN), although slightly higher (but not statistically significant) grain protein content values were noted with urea topdress application compared to UAN (Table 7).

The relationship between total amount of N applied and spring wheat grain yield for each site-year is shown in Figures 2–4. In all cases, there was a positive correlation between total amount of applied N fertilizer and grain yield. In 2011 and 2012 for both fields, this correlation was strong with ranging from 0.82 to 0.95. In 2013, decreased for both locations. These results may indicate that the model should be calibrated for each location frequently. In 2011, at Conrad, wheat yield increased to 242 kg N ha⁻¹ and at Corvallis to 114 kg N ha⁻¹ (Figure 2). In 2012, at Conrad, yield was maximized at 120 kg N ha⁻¹ and at Corvallis at 200 kg N ha⁻¹ (Figure 3). In 2013, at Conrad, grain yield was maximized at 121 kg N ha⁻¹ and at Corvallis at 134 kg N ha⁻¹ (Figure 4).

4. Conclusions

Crop canopy sensors are convenient tools for assessment of crop N status and provide assistance to growers when making the best management decision based on actual crop condition. In this study, NDVI from a GreenSeeker sensor was integrated with spring wheat (US, Canada, and Mexico) algorithm to estimate crop N content and make fertilizer N recommendations.

Overall, no significant differences in wheat grain yield or grain protein content associated with topdress N fertilizer source were observed at any of the site-years. This suggests that topdress N fertilizer rates do not need to be adjusted based of fertilizer sources used, that is, the same N rates should be prescribed whether urea or UAN is applied. At the time of this writing, the cost of N sources per unit of N are volatile, with urea currently costing $0.45 more per kg of N. Thus, growers are strongly advised to pay attention to fertilizer prices per unit of N when making decisions on what N source to use in any particular growing season, especially when there is no clear consistent advantage from using one source versus the other.

Results indicated that the three assessed algorithms developed in other regions did not provide the appropriate topdress N rate recommendations for spring wheat. Our case study emphasizes the importance of (1) calibrating the crop sensors for the local crop varieties and growing conditions and (2) developing N fertilization algorithms based on locally established YP prediction studies.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Acknowledgments

The project was funded in part by the Montana Fertilizer Tax Advisory Board.

References

C. L. Da Silva, G. Benin, E. Bornhofen, E. Beche, M. H. Todeschini, and A. S. Milioli, “Nitrogen use efficiency is associated with chlorophyll content in Brazilian spring wheat,” Australian Journal of Crop Science, vol. 8, no. 6, pp. 957–964, 2014.
View at: Google Scholar
X. Xu, Z. Wu, Y. Dong, Z. Zhou, and Z. Xiong, “Effects of nitrogen and biochar amendment on soil methane concentration profiles and diffusion in a rice-wheat annual rotation system,” Scientific reports, vol. 6, Article ID 38688, 2016.
View at: Publisher Site | Google Scholar
D. K. Biswas and B.-L. Ma, “Effect of nitrogen rate and fertilizer nitrogen source on physiology, yield, grain quality, and nitrogen use efficiency in corn,” Canadian Journal of Plant Science, vol. 96, no. 3, pp. 392–403, 2016.
View at: Publisher Site | Google Scholar
A. R. Ravishankara, J. S. Daniel, and R. W. Portmann, “Nitrous oxide (N₂O): the dominant ozone-depleting substance emitted in the 21st century,” Science, vol. 326, no. 5949, pp. 123–125, 2009.
View at: Publisher Site | Google Scholar
D. S. Reay, E. A. Davidson, K. A. Smith et al., “Global agriculture and nitrous oxide emissions,” Nature Climate Change, vol. 2, no. 6, pp. 410–416, 2012.
View at: Publisher Site | Google Scholar
N. Gupta, A. K. Gupta, V. S. Gaur, and A. Kumar, “Relationship of nitrogen use efficiency with the activities of enzymes involved in nitrogen uptake and assimilation of finger millet genotypes grown under different nitrogen inputs,” The Scientific World Journal, vol. 2012, Article ID 625731, 10 pages, 2012.
View at: Publisher Site | Google Scholar
L. J. Thompson, R. B. Ferguson, N. Kitchen et al., “Model and sensor-based recommendation approaches for in-season nitrogen management in corn,” Agronomy Journal, vol. 107, no. 6, pp. 2020–2030, 2015.
View at: Publisher Site | Google Scholar
J. Crain, I. Ortiz-Monasterio, and B. Raun, “Evaluation of a reduced cost active NDVI sensor for crop nutrient management,” Journal of Sensors, vol. 2012, Article ID 582028, 10 pages, 2012.
View at: Publisher Site | Google Scholar
M. Huang, W. Zhang, L. Jiang, and Y. Zou, “Impact of temperature changes on early-rice productivity in a subtropical environment of China,” Field Crops Research, vol. 146, pp. 10–15, 2013.
View at: Publisher Site | Google Scholar
J. T. Bushong, J. L. Mullock, E. C. Miller, W. R. Raun, and D. Brian Arnall, “Evaluation of mid-season sensor based nitrogen fertilizer recommendations for winter wheat using different estimates of yield potential,” Precision Agriculture, vol. 17, no. 4, pp. 470–487, 2016.
View at: Publisher Site | Google Scholar
W. R. Raun, J. B. Solie, M. L. Stone et al., “Optical sensor-based algorithm for crop nitrogen fertilization,” Communications in Soil Science and Plant Analysis, vol. 36, no. 19-20, pp. 2759–2781, 2005.
View at: Publisher Site | Google Scholar
J. I. Ortiz-Monasterio and W. Raun, “Reduced nitrogen and improved farm income for irrigated spring wheat in the Yaqui Valley, Mexico, using sensor based nitrogen management,” Journal of Agricultural Science, vol. 145, no. 3, pp. 215–222, 2007.
View at: Publisher Site | Google Scholar
G. Silva, All Fertilizers are not Created Equal, 2016, http://msue.anr.msu.edu/news/all_fertilizers_are_not_created_equal.
The Mosaic Company, Fluid and Dry Fertilizers, Fluids and Solids are Equal Agronomically, 2013, http://www.cropnutrition.com/efu-fluid-dry-fertilizers#overview.
C. J. Watson, R. J. Stevens, R. J. Laughlin, and P. Poland, “Volatilization of ammonia from solid and liquid urea surface-applied to perennial ryegrass,” The Journal of Agricultural Science, vol. 119, no. 2, pp. 223–226, 1992.
View at: Publisher Site | Google Scholar
E. Lombi, M. J. McLaughlin, C. Johnston, R. D. Armstrong, and R. E. Holloway, “Mobility and lability of phosphorus from granular and fluid monoammonium phosphate differs in a calcareous soil,” Soil Science Society of America Journal, vol. 68, no. 2, pp. 682–689, 2004.
View at: Publisher Site | Google Scholar
Z. Abebe and H. Feyisa, “Effects of nitrogen rates and time of application on yield of maize: rainfall variability influenced time of N application,” International Journal of Agronomy, 2017.
View at: Google Scholar
O. S. Walsh, A. Pandey, and R. Christiaens, “Sensor-based technologies for nitrogen management in spring wheat,” in Proceedings of the Western Nutrient Management Conference, vol. 11, Reno, NV, USA, 2015, http://www.ipni.net/ipniweb/conference/wnmc.nsf/e0f085ed5f091b1b852579000057902e/4be3031d1d87927a85257e37004fa7a8/$FILE/WNMC2015%20Walsh%20pg163.pdf.
View at: Google Scholar

Copyright

Copyright © 2018 Olga S. Walsh et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

2075

Downloads

1605

Citations