There has been an explosion in the availability of tools and advisory services intended to make nitrogen recommendations for corn over the past several years. While soil scientists know in general how these products are making their recommendations, the specifics of how these products work, and therefore, how well they work, is usually a mystery due to the proprietary nature of any commercial product.
This has left us on the university-side to not pass judgement on these products, pending careful research and evaluation. Now that several years have passed since the introduction of many of these technologies, some of that data is starting to emerge.
There are three general categories of N advisory tools: sensing, modeling, and sample-based. The sensing tools have been around the longest and include devices that measure light and color in the plant or use photography or imagery to make a color analysis.
These tools have been shown to be of limited value in corn due to the high likelihood that the crop will have already lost yield at the point that measurable color differences based on N inputs can be detected, and are therefore best suited for use when making rescue treatments.
The other two methods have been the focus of more recent evaluation and are showing more promise. Several field trials have been conducted comparing the input from these methods to the customary “BMP” flat-rate application. To date, there has not been enough information gathered under a wide enough range of sites or conditions to either support or reject using them, but we have learned and observed a lot.
To understand how to evaluate variable rate N, one must first understand how university N rate guidelines are generated. Nitrogen rate trials occur in various forms and locations every year. The economic optimum N rate is calculated for each of these sites and is entered into a database, becoming a point on a graph.
A mathematical model called a “quadratic plateau” is fit to the data on this graph. This looks like a line that increases yield as N rate increases to a certain point and then levels off. This line is “fit” to the data so that most (but not all) of the points are under the line.
This ensures that on most sites and in most circumstances the suggested N rate is adequate. You can use the corn nitrogen rate calculator, which includes the most recent data from U of M research plots, to calculate economic return to N and find profitable N rates.
One can think of this as one large variable rate N study conducted across the state, where different rates of N were necessary to maximize profit in different locations and in different years. By its very nature, it is more likely (66-80% of the time) that a lower rate of N than suggested would maximize profit and yield.
AgFax Weed Solutions
A higher than suggested N rate is less likely to result in a profitable yield increase (20-33% of the time). The purpose of an N advisory tool is to predict where these instances occur by measuring (or predicting) the probable causes. This also means that a properly calibrated tool is much more likely to make a recommendation below the BMP rate than above.
Taking this all one step further, one quickly realizes that there is no possibility for a yield increase when using a lower N rate (but the assumption is that the rate used is adequate).
Only in instances where more N is necessary will a yield increase be realized. This means that the cost of the N advisory tool must be recovered by a reduction in fertilizer expenses under a much larger range of situations than where a yield increase will result in a profit.
Field trials in Minnesota
Most of the studies conducted over the past 6 years confirm what is outlined above: the N rate advisory tools will make recommendations below flat rate guidelines in a high percentage of sites and situations.
So the dilemma is this: The data used to construct the N rate calculator told us that we could get away with less than the suggested rates in many instances. So did the technology work or did it simply take advantage of this reality?
One major clue is how these products performed in 2019. The way crop models work is to keep a running balance of N available versus N demand by the crop both in a past tense and projected to the end of the growing season.
The 2019 growing season gave us every reason to believe more than the guideline rate of N was necessary. A wet spring ensured there was little to no carryover N in the profile. The season stayed wet, with ample soil moisture to cause N loss (both leaching and denitrification).
Additionally, the relatively cool conditions were likely to result in less than normal mineralized N from soil organic matter.
Unfortunately, we didn’t have many variable rate N trials last year, but one site (a Minnesota Corn Innovation Grant) that was intensely evaluated showed us that a crop model (in this case it was Encirca) predicted this N deficit, while two methods involving soil sampling and tissue testing did not.
The increased N (an average of 70 lbs./Ac. more than suggested) resulted in a yield increase of 45 bu./Ac. It is worth noting that the crop model in this study had gone low on N rates the two years prior with no yield penalty (as did the other two methods).
In one corn-following-corn plot in 2017, an average N rate of 154 lbs./Ac. produced a yield of 277 bu./Ac. It is unfortunate that we were not able to evaluate other crop models, as the study was not intended to be a product endorsement.
One takeaway from this study is that N predictive technology that involves sampling soil or tissue does not account for future conditions. It is obviously not possible to know what the weather will be like through the end of the growing season with precision, but at least the crop models are working to predict it based on past conditions and current trends.
A substantially wet or cool period in the middle of summer could render earlier soil or tissue data dubious by the end of the growing season. The results of this study are similar to a project at the U of M Southern Research and Outreach Center in Waseca in 2017 that studied the cover crops and N rates.
In this study, chlorophyll readings in July were the same, but a lower N rate ran short in September, causing a yield reduction. In other words, a snapshot in time showing adequate plant nutrition does not guarantee sufficiency through the end of the growing season.
One final word of caution is for producers to be sure to consider their own economics when analyzing these products. They each have a cost per acre that needs to be weighed against fertilizer cost savings and potential yield increases, as well as with potential risks associated with performing timely field operations (like getting back into a field to sidedress).