Zero Rates in Site-Specific On-Farm Trials

J. Lowenberg-DeBoer

SSMC Director

Introduction

Many American farmers think that their fertilizer rate decisions could be improved. Some of those involved in precision agriculture understand that the state or regional recommendations used by the extension services and soil testing laboratories were meant for broad application across a wide range of conditions, even if those recommendations were not perfect for any specific area. They think that the fertilizer rate decision might be fine-tuned given their soil, microclimate and management strategy.

Some of those farmers think that the fertilizer rate choice is important enough to do on-farm trials. Precision farming technology has made those on-farm trials easier than ever before. But those same producers often balk at including the zero or low rates that are required to test crop response. The goal of this article is to consider why U.S. producers are so reluctant to use zero or low fertilizer rates on small areas in on-farm trials, when farmers in many other countries readily accept the need to have a zero or low rate control against which response can be measured.

It should be noted from the beginning that rate trials are probably not the best place to start on-farm experimentation. They typically require at least 3 or 4 treatments. Analysis of the data often requires some statistical skill. The best place for most farmers to start on-farm experimentation is with an either/or question, such as: Does herbicide A or B provide the best weed control? Should an insecticide be used on 1^st year corn? Does hybrid X or Y yield best on my soil? Either/or questions lend themselves to simple experimental designs like the split plant design. In many cases, no statistical analysis is required. In Precision Farming Profitability, Brouder and Nielsen provide some rules of thumb that can be used to identify the best choice without much calculation. In many cases the either/or choice is between to options that are already used on the farm. It is often not clear that there will be a yield loss from either option.

Why are low rate treatments needed?

Crop response to inputs can vary widely from year to year. Rainfall patterns, temperature and many other factors enter into the response. For instance, in the Eastern Corn Belt in a dry year, nitrogen response in corn may peak at 100 lbs/acre or lower because there is not enough water to support higher yield levels. A nitrogen rate trial, which uses rates of 100, 150, 200 and 250 bu./acre, would probably not see any response in that dry year. A trial with no visible response contributes little to next year’s decisions. It suggests only that in at least some years the response stops at a rate lower than the lowest rate used. The on-farm rate trial design that maximizes the value of the information to gained probably includes a zero rate. That trial with a zero rate treatment is the one that is most likely to show some response.

Other countries

I have been involved in on-farm rate trials in several countries in Africa and Latin America. Usually, the question of whether to include zero rates was raised by researchers and accepted by farmers without much discussion. My experience in on-farm trials in the U.S. has been very different. Zero rate strips are viewed very negatively by producers. In some cases just the mention of the possibility of zero rate strips is enough to cause cooperating farmers to loose interest. Acceptance of zero rates is perhaps easiest to understand for the African smallholder who uses very little fertilizer. He or she probably does not use any fertilizer on most crops anyway. The situation of commercial agriculture in Latin America is more similar to the U.S. case. What is different about the Latin American situation? Some ideas:

Farm size – The farms on which I have been involved in trials in Argentina are very large by U.S. standards, 10,000 to 30,000 acres. For them a few zero rate strips are a very small part of their overall acreage. But it must be noted that for both the U.S. and the Argentine cases is the trial likely to be a minor part of the farm area. For example, a half-mile strip of 16, 30 inch rows, an anhydrous applicator would be 0.24% of the area of a 1000 acre U.S. farm and 0.024% of a 10,000 acre Argentine estancia.
Other sources of rate information – U.S. farmers are accustomed to having reliable sources of rate information from public sources. Some producers complain about how well that information fits their specific location, but the price is right. It is usually free. In both Africa and Latin America there has usually been much less information available.
Culture of fine-tuning – Crop production systems in both Africa and Latin America tend to be more extensive than in the U.S. In commercial agriculture there is often less attention to making every acre count.

I think the most likely explanation of the reluctance to use low rate treatments is the easy availability of low cost information. For many U.S. producers the value of the lost yield on those low rate strips outweighs the value of the information that might be gained. For me this raises the question of how valuable it is for U.S. producers to do any rate trials.

Statistical fix

In some cases experimental designs that are costly for producers in terms of time or expense can be avoided by using more sophisticated analysis software. In the June SSMC newsletter I raise the possibility of such a “statistical fix” for the preference of some producers for comparisons in large blocks instead of in narrow strips.

There is some possibility for reducing the area of the low rate treatment. For instance, in a strip trial with four replications, it may be possible to gain some response information by having a single low or zero rate strip. This reduces the reliability of the response information at low rates, but does not eliminate it entirely. If no low rate strip is included no statistical fix can tease out that information. That season’s response information is gone forever.

Conclusions

Producers thinking about doing fertilizer or other rate trials should consider how valuable the resulting information would be to them. If it is not valuable enough to include at least one low or zero rate strip or plot, then maybe the trial is not valuable enough to spend any time on. Rate trials without a low or zero rate strip may end up being a waste of time and resources.