How to make fertilizer recommendations

Tuesday, April 6, 2021

Learn about the key considerations you should keep in mind.

By Patrick Lynch

How do you decide which fertilizer you use? Do you use the same as last year? Do you ask your dealer for suggestions on what you should apply? Or do you use a soil test and get help from a certified crop advisor or other competent agronomy person?

Do you use a program like SWAT or SoilOptix or a similar program for soil sampling? Do you use grid sampling and variable rate application? Does the person who provides you with recommendations use a build, maintain or drawdown strategy? If you don't know, you should look into it.

A big part of making fertilizer recommendations is to test, apply fertilizer, record what you added, measure yields and then retest. It's important to do this with the same lab.

I have connected with John Heard, a soil fertility extension specialist at Manitoba Agriculture, to gather his insights on fertilizer recommendations.

Whatever you do and whether you do your own recommendations or have someone else do them, there are some things to keep in mind. Also, when you make recommendations based on soil analysis or even yield maps, you have to make compromises on rates and even application.

For instance, a liquid starter fertilizer may be more convenient, and you may want to broadcast the rest, but according to Heard, you have to be cautious about broadcasting fertilizer. Heard says, "On the calcareous soils of the Prairies – banded nitrogen (N) is 20 per cent more efficient, phosphorus (P) is two to six times more efficient, and potassium (K) is two times more efficient than broadcasting – so broadcasting, especially low rates, is not very effective.

"Contrary to Eastern Canada and much of the United States, there is not much broadcast fertilizer in the Prairies. A survey of growers suggests that 3 to 10 per cent of canola/wheat acres get broadcast P, but 85 to 90 per cent of fields get P during seeding operation (with 55 to 60 per cent putting P in the seedrow). Corn is different; 40 per cent of farms broadcast some P, 32 per cent apply with seed," says Heard.

Soil analysis is not exact. A good lab on a good day will have about 25 per cent variation in analysis results. Thus, if your P test result is 12, it could actually be between 10 and 14. That variance is significant if you are watching your soil test levels and do not want them to drop.

Similarly, with potassium, a test result of 100 could actually be between 88 to 112. On the other hand, if your test for P is 30, the range is around 27 to 34. At these high levels it doesn't matter. Both levels suggest a very small probability of economic return to extra P except as a starter. Even then, you only need a smaller amount of P in the starter.

Most starter research trials have treatments of zero, then 30 lb. and maybe 60 lb. of P. The research will show an advantage to 30 lb. of P versus none. But in some cases, depending on your soil test, 15 to 20 lb. of P may be all you need.

This brings us to economics. Your biggest economic return comes from your first few pounds of nutrients. Each subsequent pound gives a response but that response becomes less and less with each pound until you get to a rate where there is no response. This is much different than herbicides, fungicides and insecticides where you have to use a certain amount to get any response.

So, why does this matter? Well, if you are applying a starter fertilizer on corn and let's say you are limited to 135 lb. of starter, you can decrease the P rate by five to 10 lb. without significantly affecting yield.

And if you are broadcasting fertilizer and part of the field is testing somewhat lower than the rest, you can use less on the lower-testing areas without giving up significant yield. I do this when I have a field where 10 to 20 per cent is testing lower than the rest of the field. Similarly, if 10 to 15 per cent of a field is high and likely unresponsive, for convenience and time I will recommend fertilizer on these higher-testing areas.

The art is to know what compromises to make. According to Heard, Prairie farmers make these compromises routinely. A survey of growers indicates 70 per cent of wheat farmers put the same fertilizer rate on all their fields, 20 per cent vary the rate between fields and about 10 per cent use some measure of variable rate applications.

And part of making fertilizer recommendations is determining what your expectations are. Do you want a program that is simple to use or are you willing to treat small areas separately? In the spring and in the interest of time, often you treat different testing areas in the same way.

You need a realistic fertilizer budget. With computer technology, you can easily change rates to arrive at a dollar value you can live with. If the recommendation is done properly, you can scale the rates down. Say you have budgeted $75/ac but the recommendation calls for $100/ac. You want to know that you can decrease the rate with minimal impact.

So, how much can you let your soil tests drop without sacrificing yields?

There are different thoughts/values on how many pounds of P it takes to drop or raise P one point. The range is 25 to 50 lb. of P2O5/ac to change the P level by one point. According to Heard, it takes about 20 lb. of P2O5/ac in neutral sand loams and up to 40 lb. P2O5/ac in calcareous clay loams to change one part per million. Similarly, there is a range in how much K you need to drop K levels by one point. This number is dependent on the cation exchange capacity (CEC) of your soil.

Soils with lower CEC (lighter) are different than soils with higher CEC (heavier). Also, there's a big difference in how much P and K are removed by crops. For a three-year crop rotation of wheat (75 bu), soybeans (45 bu), and canola (50 bu), you remove about 135 lb./ac P2O5 and 95 lb./ac K2O. But since these values vary, I look at soil test levels to see if levels are falling or increasing. (I use above-field-average values since parts of the field will be using more than the average.)

If soils are testing in the low ranges, I want levels to increase. If soils test at medium levels (a good working range), I do not want to see levels fall. I feel that if these levels are falling in these ranges, you are shorting the crop. If levels are high, I don't mind if levels drop.

Put another way, Heard says, "This is the same simple long-term sustainable approach we are advocating for P. Build depleted soils, maintain when in good working range and opportunity to drawdown with starter only in high-test soils." A word of caution: since labs are different, if you change labs you will need to account for this change.

When it comes to sulphur (S), Heard says, "High S soil levels are very common and often a result of salinity – natural soil salts coming to surface – but extremely variable by landscape and so you cannot trust the soil test well. So, for needy crops like canola, insurance applications are usually warranted and applied."

There is another factor that we rarely take into account – the depth of topsoil. All the calibration and testing are done in the acre slice in the top six inches. Our recommendations are based on the top six inches weighing two million lb. Thus, some results report in parts per million while others report in pounds per acre.

In either case, if you are tracking your soil test levels over time, it is important to stick with the same lab for the most accurate tracking.

A side note is that I think if you have two soils with the same soil test number, but one has four inches of topsoil and the other has eight inches, there will be a difference in nutrient- supplying power.

Where do recommended rates come from? There are a number of places. Most jurisdictions in North America have tables that are published for the public to view. These are recommendations based on what the people who put these tables together think.

So, with different jurisdictions there are different philosophies and thoughts. Some of the people putting these tables together get very defensive if you challenge their tables. Some tables are based on removal rates of nutrients, but there is disagreement on what these numbers should be. Some jurisdictions change these numbers based on what they have measured in crop removal over time.

Some of these tables are put together using a build, maintain or drawdown philosophy. Some of these tables factor yield into P and K recommended rates. Some tables take CEC into account when making K recommendations. When you talk to the people making your recommendations, ask them which approach they use.

And there are different extraction techniques. The standard in Manitoba has been Olsen (sodium bicarbonate) for P extraction. More of the soil analyses I see have Mehlich as the main extractant. Our American neighbours to the south recently officially embraced Mehlich as the choice of extractant for P.

If you were tracking P using sodium bicarbonate extractant and have switched to Mehlich, the numbers will be very different.

According to Heard, "On the Prairies (in fact all northern Great Plains), there are a lot of free carbonates or free lime in soils that will challenge the acidic Mehlich 3 – so proceed with caution! Or simply choose a lab that does the Olsen sodium bicarb."

There are other little nuances among labs that mean you cannot compare between labs. The scoop size for soil analysis and the shake time (time that a sample is shaken) before analysis differs among labs.

The bottom line is you should know where your fertilizer recommendations come from and you should track and record results over time. The last word goes to John Heard: It's the tracking over time that the soil test is most valuable for. BF