Up in the air?

21 Feb 2014

Most people know what a CV is but not so many know what a CV% represents. It is the coefficient of variation and, crudely put, is a measure of the variation in the data from individual treatments in an experiment as a proportion of the mean of the individual treatments. Most research organisations use this as a guide to the reliability of the results of individual experiments.

In NIAB TAG we become concerned when the CV exceeds 5% in cereal field experiments and typically would reject results when the CV% rises significantly above this guideline.

The reason I mention all this is that I have just re-read the results from Defra Project NT26 because of the current debate over the role of additives/coating for urea in reducing the loss of nitrogen due to ammonia emissions. NT26 suggests that there’s a need to increase the dose of nitrogen supplied in urea by 20% to match the crop nitrogen uptake (not yields) achieved by ammonium nitrate. The difference is attributed to losses due to ammonia emissions which can be reduced by the addition of a chemical that inhibits the soil urease enzymes which are responsible for converting urea into ammonia.

The bland data from the ten field trials in NT26 (nine of which are in winter wheat and one in winter barley) identifying the optimum dose of nitrogen does indeed suggest that, on average, 20% more nitrogen in the form of urea is required to match the nitrogen uptake of ammonium nitrate at its optimum dose. However, the authors state that there is no significant difference in the nitrogen required to optimise winter cereal yields or the protein content at optimum yields whether it is applied as ammonium nitrate, urea, urea + the urease inhibitor Agrotain or UAN. The question is why such a difference is not statistically significant? In my opinion, the reason is that the errors in the majority of trials are unacceptably high.

I spent a little time looking at the results of the four trials that resulted in the most accurate response curves and CVs of around 5% or less. These confirm that urea is slightly less efficient in crop uptake but the difference is a lot less than 20%; more like 6-7%. This is a tiny database. The much larger database comprised of trials ADAS carried out in the mid-1980s and the NIAB TAG trials suggests that the difference is even lower.

So anyone can draw almost any conclusion that they wish from NT26. The evidence from previous ADAS trials, the four trials in NT26 with CVs of around 5% or less and the NIAB TAG data suggests that there is little difference between ammonium nitrate and urea in achieving optimum yields in winter cereals but the protein levels can be a little down where urea is used. Published data on urea + Agrotain is restricted to NT26 and so is thin on the ground but suggest that, in terms of crop uptake, it is equivalent to ammonium nitrate.

There is one specific concern about the evaluation of Agrotain in NT26. One reading of the field trials implies that around 8% more applied nitrogen in the form of urea + Agrotain is required to get similar optimum yields to ammonium nitrate and urea. However, the efficiency of use of nitrogen is equivalent to ammonium nitrate because at these optimum yields, the protein contents are higher with urea + Agrotain. This may reflect how the total dose of nitrogen was split in the trials. After the first 40 kg N/ha applied in March, the remainder of the total dose was split between an application at GS30/31 and at GS32 of the wheat. Hence, particularly with the overall higher doses, relatively large doses were being applied as late as GS32. This kind of split may not be entirely suitable for Agrotain because there is the implication in these trials that it slowed the release of nitrogen during the growth stages that most influence yields but more was available later for protein formation.

So nitrogen fertiliser requirements still continue to confound everyone and, as with other projects, NT26 seems to raise as many questions as it answers. Who’d be a soil scientist?