Nitrogen Recommendation Recalibrations for Wheat in North Dakota
A multiple year research study updating N response and recommendations for wheat in various agro-ecological areas of North Dakota.
IPNI-2007-USA-ND15
21 Dec 2007
Justification
Fertilizer prices are at an all-time high. Crop prices are high, but input costs are so high that wheat producers continue to ask how to cut nitrogen and phosphate rates. Our current recommendations (2.5 X Yield Potential less credits) were developed more than 30 years ago when there was no ability to vary rates within a field and N prices were less than 10 cents/lb of N. Our review of old and more modern N calibration data do not support our current recommendation in this new price and technological environment. Growers need methods to develop N rates that are economical and relevant to their wheat-growing abilities. The objective of this two-year study is to add necessary modern sites, complete with modern soil analysis, to our older data in order to develop better, more efficient N recommendations for growers in every corner of North Dakota.
In North Dakota, the N recommendation formula for wheat of 2.5 X Yield Goal has not substantially changed since the mid-1970’s (Wagner et al., 1977). Although the concept of relying exclusively on Yield Goal as a measure of need, and soil test nitrate determination as the major adjustment to N availability, is best adapted to dry, hot environments, other potential modifiers to rate have essentially been ignored (Dahnke and Johnson, 1990). The most glaring omission is some relationship of N mineralization from organic matter, and how it may vary due to soil or climatic region.
Yield Potential as a determining factor appears reasonable at one level. If one examines the N content of grain with yield, the resulting graph is relatively linear. However, if one graphs the response of grain to added N, the line is usually a curve (Fig. 1).
Figure 1. Wheat yield with N rate. Studies conducted and compiled from NDSU records during the period from 1970-2006.
The difference between what we might expect with a Yield Potential approach, and the actual data is the influence of climate and organic matter mineralization. Note that in the data points in Figure 1, some of the higher yields occur at lower N rates. Also, one can find similar yields over a 100 lb/acre range of N rates.
If one looks at the data with economic numbers, the recommended rate of N that supplies the most economic return varies with cost of N (Figure 2). At lower costs of N, the optimal N rate exceeds 200 lb/acre, but at higher costs of N, the optimal rate is over 150 lb N/acre less.
Figure 2. Return to N from data in Figure 1, compiled from 1970-2006, with N costs varying from 20 cents/lb N to 50 cents/lb N. Left graph is with $3/bu wheat, right graph is with $6 wheat.
However, when wheat price is at $6/bu and above, there is no best N rate. The optimal N rate becomes whatever the grower can afford. The data set used to produce Figures 1 and 2 is not large enough in my opinion to result in a basic change in our recommendations, although I have used it to illustrate that N rates should probably be reduced in high N cost growing seasons. One reason that I am reluctant to use this data set to change recommendations is the lack of information in this data regarding soil organic matter content. Some states have used a nitrogen-availability index, usually based on soil organic matter content, as a tool to develop N recommendations; however, this has been resisted in North Dakota in the past.
We believe that N rate responses are different in different parts of the state due to their unique agri-climatology characteristics, soils, organic matter and N mineralization potential. A wheat grower in the west may look at soil moisture and decide that his prospects for yield greater than 20 bu/acre are bleak and fertilize weakly according to our current recommendation formula. Good rains after seeding may make that N rate decision a poor one. But even more importantly, over a range of studies in our state and others, optimal N rate appears to be similar whether yields are high or low (Goos, 1983; Sawyer and Nafziger, 2005). In Goos’s studies on wheat, he recorded the responses as differences in formulas; for a lower yielding environment in the southwest, the formula recommendation was 4.2 X Yield Goal, while in a better environment it was as low as 0.6 X Yield Goal. What it means looking at it another way, the lower yielding wheat was 13 bushels, or an optimal N rate of about 55 lb N/acre, while in the other location, a 37 bushel crop had an optimal N rate of 23 lb/acre. If the recommended N rate for both locations had been similar, perhaps 50 lb N/acre, the yield results would have been similar. Good growing conditions appear to generate more N from the soil, and poor growing conditions require that a greater percentage of N needs to be supplied by fertilizer. Figure 3 shows widely different N responses comparing data from the Langdon area with sites west of Bismarck/Minot.
Figure 3. Langdon responses to N at $6/bu wheat, compared with western ND responses at both $3/bu wheat and $6/bu wheat. Note the presence of an optimal N rate at Langdon even at high wheat price. This may be due to higher mineralization rate in the higher organic matter soil in a cooler and moister climate than western sites.
A possible method for predicting N response within a field or between fields in addition to our traditional nitrate soil test is the Illinois Soil Nitrogen Test, or ISNT (Khan et al., 2001). A new analysis of a component of soil organic matter, amino-sugars, has been developed and tends to determine whether a response to N is possible with a certain soil in a given year. The tool has been tested by a number of researchers with varying levels of success (Mulvaney et al., 2005). However, at the present time, it is a tool that holds some promise to help determine the mineralization potential of soils that up to now has been impossible to calculate in our recommendations (Figure 4).
Figure 4. Relationship of ISNT with % yield increase over check from our small current dataset.