Is variable rate nitrogen based on yield maps a good idea?

By: Brad Carlson, Extension educator

Last year around this time I wrote a blog post that detailed the long track record in Minnesota of the most profitable crop producers spending significantly less on fertilizer than the least profitable growers. The recently released FINBIN data for 2024 show that this trend is continuing. In 2024, the 20% most profitable farms spent an average of $170 per acre on fertilizer while the 20% least profitable farms spent $224. This is a 31% difference, and very similar to the 33% I reported from 2023.

This difference is especially notable given the current challenging ag economy. The FINBIN data also showed the lowest median net farm income for Minnesota crop producers in the last 30 years (just under $22,000 in 2024).

My article last year detailed some of the possible explanations for the difference. One of these was applying too high of a fertilizer rate, and another was applying fertilizer more times than necessary (split-applying) . A recent trend of applying variable rate nitrogen (N) based on yield maps combines both of these issues.

Variable rate technology and nutrient management

The technology of variable rate (VR) fertilizer application has been around for over 30 years and has been primarily used for phosphorus (P) and potassium (K) based on soil test values. Many theories and technologies have been explored to use VR for nitrogen.

The simplest approach is to calculate how much N is in the grain and apply fertilizer N based on yield. Doing so, however, ignores much of what is known about corn response to N fertilizer. First is the calculation itself, which is frequently based on a simple formula of 1.2 X yield (bushels per acre)= how many pounds of N you should apply per acre. This formula was abandoned by the University of Minnesota way back in 1974. This is because there are other factors that affect yield beyond N applications and the formula rarely predicts accurate N recommendations.

Nitrogen is primarily found in protein in grain. Breeding efforts for corn have focused on yield and other characteristics such as drought tolerance, insect and disease resistance, and stand-ability. There has been little to no focus on grain protein. Analysis of actual grain samples shows a decline in the N content of grain from about 1.2 pounds per bushel back in the early 1980’s to an average of about 0.6 by 2010. This means that, if the overall response to N fertilizer is linear, the formula should be 0.6 X yield, not 1.2.

Pounds of nitrogen fertilizer per bushel of corn (1986 to 2011)

Source: John Lamb

 
The response, however, is not linear. Data obtained from N rate trials records the N rate (Maximum Return to Nitrogen, or MRTN) at which yield levels off for each trial. If you look at the MRTN rate versus the yield obtained at that rate for each trial and plot them on a graph (below), you can see that there is virtually no pattern, other than what looks like a shotgun blast. If you plug the yields into the 1.2 X yield formula (or 0.6 X yield for that matter), the results are numbers that, when plotted, form a line. As you can see, the points are far from being linear, indicating that there are significant factors beyond applied N fertilizer that are factoring into yield.

Source: Dan Kaiser

Furthermore, if one looks at sites with yields above 200 bushels per acre versus those under 200, there is also no significant difference in response to N. Extension nutrient management specialist Dan Kaiser conducted a trial looking at landscape positions (top of hill, side of hill, bottom of hill) and overall N response. Data obtained from this project shows that the MRTN rate was the same in all three positions, however, the area at the bottom of the hill had lower yield overall.

Source: Dan Kaiser

If one examines University-recommended N application rates, you see that higher recommendations do tend to correlate to places with lower soil organic matter. Specifically, sandy irrigated sites in Minnesota, or, if one looks at Illinois’ recommendations, in central Illinois. It should be noted that places with lower soil organic matter (OM) usually yield lower than places with higher OM (the exception being wet soils which are prone to N loss). This indicates that, in many places, higher N rates should correspond to lower yields, rather than higher yields, which is contrary to the theory of VRN based on yield.

Computer-based crop models

I was involved in several VRN trials over the past decade. The technology that showed the most promise was computer-based crop models. The field trials I was involved in showed that they:

1. Nearly always recommended less N than the farmer was inclined to apply, and
   
   
2. They resulted in the same yields as higher flat N rates.

One large field scale trial conducted by a farmer who received a Discovery Grant from Minnesota Corn in a continuous corn field showed an average yield of 251 bushels per acre with an average of 152 pounds of N applied, as advised by the N advisory tool of the Encirca program (Corteva, no longer available on the market). This is compared to a yield of 248 for the flat rate of 170 pounds of N per acre. More impressively, one of the strips (12 rows wide, 1,200 feet long) yielded 277 bushels per acre on 154 pounds of N per acre. This data obviously does not show the need for increasingly higher N rates based on overall yield.

Some wonder why the crop models mostly disappeared from the market, and it is simple. They were designed to make an accurate N rate recommendation, and when it turned out that most sites needed less N, and not more, it was also clear that yield was already maximized. This meant that the only opportunity for profit was to reduce input costs, which was a much smaller dollar amount than would have been achieved had the technology increased yields.

Conclusion

So, what is going on? Every farmer knows that their best yields come from their best soils. This fact is completely independent of N rate decisions made on a year-to-year basis. One cannot simply apply higher rates of N and expect higher yields. To put it another way, your best soils are supplying more N to the growing crop (via mineralization of soil OM) and therefore do not need more N fertilizer. In conclusion, there is a mountain of evidence to show that VRN based on yield expectations is not a valid method for soils with average to above average amounts of OM. To go back to my opening point, spending on application technology and rates that do not achieve a return on the investment is one reason why some farmers are spending too much per acre on fertilizer. Remember, just because something seems technologically advanced does not necessarily mean that the science behind it is sound.

---

For the latest nutrient management information, subscribe to the Nutrient Management Podcast wherever you listen and never miss an episode! And don't forget to subscribe to the Minnesota Crop News daily or weekly email newsletter, subscribe to our YouTube channel, like UMN Extension Nutrient Management on Facebook, follow us on X (formerly Twitter), and visit our website.

If you have questions or comments, please email us at nutmgmt@umn.edu.