Seedbed: Making sense of variable rate nitrogen

Monday, April 5, 2010

If we can get it right, it will increase crop yields, save money and reduce losses to the environment

by KEITH REID

Variable rate nitrogen is the "Holy Grail" of fertilizer application technology. If we can get it right, it will increase crop yields, save money and reduce losses to the environment through leaching and denitrification.

Three components are necessary for any variable application program – the ability to variably apply, the ability to measure variation, and the "smart program" to interpret the variation to predict optimum fertilizer applications. The first two components we have had for many years, but we haven't quite figured out the final piece of the puzzle.

Providing the optimum amount of nitrogen to the crop means accounting for both sides of the equation – how much the supply of N from the soil varies across the field, and how much the demand for N varies in areas with different yield potentials.

Many of the simple models that have been proposed either ignore or make assumptions about one side of this equation. The consequence is that models eitherdon't work well at all, or they only work within a very narrow geography and cannot be transferred to other areas. A further complication is that weather can affect both the N supply and the demand, so the accuracy of predictions can vary from year to year.

This does not mean that successfully applying variable rate N is impossible. You should, however, be aware of the strengths and limitations of each system. This will at least give you some tools to sort through the salesmen's claims before you invest big dollars.

In all likelihood, when we do sort out variable rate N, it will be a combination of methods that are entered into a sophisticated model to get us closer to optimum rates than a single rate for the entire field could.

Soil-based prediction. The supply of N from the soil is the net of the amount of mineral N left over from previous crops, plus the amount of N released from organic compounds in the soil, minus the amount of N lost through denitrification or leaching. Collecting a lot of soil samples and analyzing them for nitrate content can give us the first factor and some of the second, but is not particularly good on the third.

Adding in a factor for soil organic matter, either by analysis of the same samples or by predictions using soil colour or soil depth, should improve the prediction of total N supply, except where there is a large variation in the rate of mineralization from organic matter. These approaches have only rarely predicted N rates any better than a single rate for the entire field.

Many of the factors that affect N mineralization are implicit in the soil factor of the Corn N Calculator, but I don't know of anyone who has tried to combine the variation in yield potential (from yield maps) with the variation in soil types across the field to produce a N application map.

It might be best to illustrate the challenges of using soil factors with an example.  Consider a field with a gently rolling topography, with a soil texture ranging from clay loam on the side slopes and hollows to loamy sand on the knolls. Organic matter is highest in the hollows, and the entire field has been systematically tile drained.

With normal weather, the sandy knolls suffer from moisture stress and so don't have the yield potential to respond to above normal rates of N. The organic matter in the hollows provides a significant amount of N as it breaks down, so the part of the field that has the greatest response to added N is the side slopes.

In years with lots of rainfall in mid-summer, however, the knolls have the highest yield potential and the lowest capacity to provide N from the soil, so the responsive areas of the field switch. Excess rain can also cause denitrification in the hollows, reducing the amount of N in the soil, but these conditions would also hurt the crop enough that it is probably unable to respond to added fertilizer N.

Vegetation sensors. Some researchers are trying to get around the limitations of the soil measurements by using the plant itself as an indicator of nitrogen supply. Since the amount of chlorophyll in the plant is related to the amount of nitrogen available, the greenness of the plant can be used as an indicator of nitrogen supply. Various approaches have been tried, from remote sensing using satellite or air photo data, to hand-held sensors that clip onto the leaf to read the chlorophyll content, to active sensors that emit light at specific wavelengths and measure the reflection from the plants and use this to vary N applications "on-the-go." 

The theory behind all of these systems is sound, but there are a number of challenges in implementation. The first is that the greenness of a crop will vary with the growing conditions and the particular hybrid being grown, so it is crucial to have a reference strip with full N nutrition to which you can compare the rest of the field. Of course, if the field is variable, then this reference strip should cover the variability as well.

Second, the sensors will not be reliable on crops with small size plants because the sensor will be picking up the reflectance from the soil rather than the leaves, plus the small plants will not have had enough N demand for any shortages to show up in the leaves.

Finally, it ignores the many other factors besides nitrogen that can reduce the greenness of the crop (disease, insects, saturated soil, etc.), so the sensor may actually call for more N on the crop in areas that actually have little ability to respond to added N. BF

Keith Reid is soil fertility specialist with the Ontario Ministry of Agriculture Food and Rural Affairs based in Stratford. Email: keith.reid@ontario.ca