In the third in a series of three articles on precision agriculture, Les Henry gets down to the nitty gritty details of variable rate fertilizer
Variable rate fertilizer (VRF) is probably the input that is talked about the most.
The starting point to VRF is establishing management zones within a field. That can take many forms.
Pretty pictures of past crop performance is one way to divide your field into zones. But, maybe the areas that grew really good crops last year are already very fertile and need less nitrogen rather than more. When it comes to setting variable rates, nitrogen is the input most often varied, because it costs the most and is used the most.
Another way to establish zones is to spend thousands of dollars doing soil tests from dozens of individual sites established in a pre-determined grid. Then, you can return to the grid each year using GPS. I have never been a fan of grids. I think a much better way is to know the plan Mother Nature had in establishing the individual soils within a land type and to use that to establish individual soil boundaries based on soil properties. Then, you can test each zone to get information about current available nutrient status.
Grids may have merit in a very uniform land type, where the objective is to determine where old manure dumps from adjacent farmyards are located. A much simpler way to locate old farmyards would be old air photos. In Saskatchewan, old Municipal Assessment sheets could be used but they are not now accessible. I hope to change that.
Across the Prairies, probably the best source of old farmyard locations is the collection of 1:50,000 topographic maps — most of which are interpreted from air photos old enough to still locate old farmsteads. These topographic are easily available at the click of a mouse. (See http://geogratis.cgdi.gc.ca.)
The first thing we must realize with fertilizers is that they are like elephants — they have a long memory when applied at high enough rates. At many of our current rates, there will be residual effects.
Farmyard, feedlot or liquid manures also have very long memories. In data from Rothamsted, England (established 1843), manured plots were still producing double the yield of barley compared to un-manured plots 30 years after manure applications ceased.
The residuals from manures and fertilizers is the reason I never conducted a fertilizer experiment on a university or Agriculture and Agri-Food Canada research station.
Long term experiments are just now starting to pay off by showing the true long term residuals. I will expand on that in a future article.
Soil testing will be important in any VRF plan. It is important that we realize that soil test levels of nutrients provide an index of the availability of that nutrient. Soil testing is not like a dipstick in a crankcase. It is not as simple as crop need – soil test = fertilizer to apply.
We must also remember that soil testing is only validated by soil test correlation work. To prove that a soil test works requires dozens of field experiments over several years, on soils testing from very low to very high for the nutrient in question.
Much soil sampling is now done with set GPS-located sites that are returned to each year. That will remove a lot of the variation — as long as you’re sure the right locations are used. And care is required to ensure that the work of procuring the sample in the same location each time does not make changes to that site.
If GPS sampling is done within management zones, enough sites must be sampled to get a true average of that zone. I never have much faith in one soil sample.
Let us deal with it one nutrient at a time, starting with phosphorus (P).
P soil test correlation has been going on for 50 years in Western Canada. Using the most common current P soil test it works like this:
- Soil Test P < 10 pounds of P/acre or less: 75 per cent chance of response
- Soil Test P > 60 pounds of P/acre or more: 25 per cent chance of response
- Soil Test P in the middle range: response in any given year is crap shoot, about 50 per cent chance of a response.
That is why I have always said, “P fertilizer is an investment in the land, and in the long run it will pay off in spades.”
But in Weyburn loam and similar lands, soil test P could be very useful. First, sample the knolls and confirm very low P, then put on a little extra P every year (or put on a big batch once when price of P is right). Then, pick out the very high-testing arable sloughs and leave P out altogether. In the middle grounds, put on the usual old middle rate every year whether it is needed or not. In the long run it will pay.
The P soil test is valuable for documenting historical manure applications. At very high soil test P values, no more P fertilizer should be used until the test comes down. When the Saskatchewan Soil Testing Lab opened in 1966 there were few high soil test P values, and in almost every case the high P level could be related to past manure applications. I remember a field next to a dairy farm that was off the scale on every nutrient, but especially P.
N is a little more complex. In much of Western Canada rainfall variations from year to year are still an important determinant of how hard the trucker works in the fall. And, the soil water at seeding plus growing season rain (or the probability of it) has been used very effectively to refine N fertilizer recommendations.
And, the plot thickens. Think of N as the fuel in your half ton that has two tanks. In Tank 1 is the real stuff — gas that has an immediate bang every time you press the pedal. Tank 2 is bigger than Tank 1, but the gas supply from Tank 2 is metered out at an unknown rate with unknown timing. Better make sure Tank 1 doesn’t run too low to avoid walking.
The amount of Nitrate-N to two feet in a soil is like the gas in Tank 1 of our pickup. We’ve used the nitrate test like a dipstick in a crankcase in the past and it has worked not too bad. With the great increase in N fertilizer use in the past 20 years, there are now times when Tank 2 is supplying a lot of N (that is, mineralization of N from the soil organic matter). That is a result of continuous cropping, higher N application rates and zero tillage that helps to retain any N excess to current need.
Most folks that have been continuous cropping with zero till for decades find they can grow better crops with much less N fertilizer.
But, at the moment we have no good way of estimating the quantity and timing of flow of gas from Tank 2. So, I see limited gains from VFR for nitrogen fertilizer until some of those issues are solved. I am sure there are already professional agrologists out there who have made headway on the moisture-N interaction front. Measuring the available soil water as of freezeup is easy — and that information can be used all winter in planning. Water in the soil is like money in the bank. Of course the huge water excesses of recent years in southeast Saskatchewan and elsewhere makes that statement nonsense — but it usually does apply in much of Saskatchewan and Alberta.
With low organic matter eroded knolls, an extra dose of N could still be needed, depending on the history of N fertilizer. VRF could easily supply that extra N where needed. And, in soils with great accumulations of topsoil from past erosion, reducing N rates in those areas could pay well.
I think we have reached the stage on many soils where residual N from past fertilization is coming out as mineralized N to meet more of the needs of current crops. A soil test that will measure N that will be mineralized in the early part of the growing season would be a big boost to VRT for N.
Potassium (K) is a special nutrient. Routine K soil tests work well to establish seriously K-deficient and well supplied soils. The probability of response in the mid ranges is less well defined than for P.
Many of our soils are still well supplied. The areas of seriously K-deficient soils in northern sandy Grey and Peaty soils were defined many years ago. Almassippi soils in Manitoba and Carrot River soils in Saskatchewan are the two most K-deficient soils in Western Canada — or were. K fertilizers applied in the past 30 years will have changed some of that.
Individual agrologists are identifying soils marginal in K where crops are benefiting from application. We did find interactions of K with diseases of wheat, and more of that is likely being identified. I see no big advantage to VRF for K on a broad scale; some serious research would be required to prove any value.
The sulphur (S) soil test in garden patch agriculture (small plots) works great. But, in the field it is less useful. S does vary considerably within fields of very flat land. But the routine test may not always sort it out.
The problem is the natural soil gypsum that is in the subsoil in many soils. One core sample in with a load of gypsum in a collection of 30 samples will make the composite soil test read lots of S — when much of the area does not have enough S. If the soil test says S is needed, then it is needed. If the soil test says S is not needed, it still may be needed in parts of the field.
I expect with some more detailed soil testing it may be possible to vary S fertilizer over a field — some folks may already be doing it.
When I do any soil testing on my farm I always take 30 cores to make up a composite and can duplicate N, P and K easily but not S.
In summary, variable rate fertilization is not as simple as it might first seem and it will take an innovative approach and knowledge of individual land types to succeed. †