18 January 2002

Perfect predictor, please

Many stabs at predicting

optimum nitrogen input for

wheat have been made.

But no system is yet

perfect, independent

testing suggests.

Andrew Blake reports

WHATEVER your approach to assessing N needs for wheat, using a scientific prediction system is almost sure to prove better than relying on experience alone; but no system is yet infallible.

That is the view of John Ward, recently retired director of the Scott Abbott Arable Crops Station, near Peterborough. Three years work testing various methods highlights the difficulties of producing a practical tool which gives good results whatever the conditions.

"Our experiments show there is still an urgent need to find a truly reliable prediction system to improve margins and protect the environment," says Mr Ward.

The N response and prediction trials on the shallow limestone brash soil at Sacrewell Farm began with the 1998 harvest crop, a year when a very wet April and wet June followed more normal winter rainfall.

Systems tested

Six systems (see box) were tested on a first wheat after peas receiving a range of treatments, from 0 to 250kg/ha total N, to determine the optimum.

In that season, 76.6kg/ha produced top yield, with 72.5kg/ha being the economic optimum.

"Only one prediction system, Kemiras N-Min, suggested a rate close to the optimum." It indicated 57kg/ha.

Yields from Hydros PrecisioN plan and IACR-Rothamsteds Sundial computer system were much the same but more costly to achieve. Advising 120 and 140kg/ha, respectively, they also risked extra leaching losses, he notes.

The French Jubil system, favouring heavy rates of early N, was highly successful the previous year. But in 1998 its recommended 150kg/ha gave 0.55t/ha less than the same total applied more evenly through the season.

"That underlines the importance of correct N timing," says Mr Ward. "With no guarantee that subsequent weather will support the high tiller counts that Jubil induces, the system seems a bit of a gamble especially on land prone to drought."

Guidance from the ministrys RB209 booklet, before its recent revamp, had always proved reliable. But in that season its recommendation of 190kg/ha for shallow soil at index 1 was clearly too high, although its target yield was almost spot on at 7.56t/ha, he notes.

A year later, when yields were generally higher, the economic optimum N rate, again for first wheat after peas, was more than doubled at 183.4kg/ha.

Advice from Sundial, N-Min, RB209 and PrecisioN Plan all came close to that. "But it is fair to say that Sundial gave us a range of applications from which we chose the highest – 175kg/ha. The lowest might have given a different result."

Recommendations from N_Able and the Solomon porous pot system, which replaced Jubil in the trials, were both clearly too low at 111 and 140kg/ha, respectively.

The Solomon system is not particularly "farmer-friendly", adds Mr Ward. "It involves placing the pots at two depths in the crop and may have limitations on shallow limestone soil in a dry season when the cups fail to collect enough leachate."

The harvest of 2000 followed a generally wet season when all systems underestimated wheats response to nitrogen after peas.

All produced similar, but slightly below optimum output.

With nitrogen at the time costing £85/t and wheat worth £60, the most economic dressing was 226kg/ha.

"The most successful system, Hydro PrecisioN, used only 157kg/ha, 50 of which was not applied until after the N tester had shown the need for it at GS37."

Sundial, without the backing of Solomon, suggested 74kg/ha at that timing and produced almost the same yield.

"Kemira N-Min and Sundial with an adjustment from Solomon indicated a need for only 25kg/ha at GS37, and it seems likely that in such a wet year the late N was necessary to maintain growth."

Many difficulties

2001 illustrated just one of the difficulties facing prediction system, says Mr Ward. The exceptionally wet winter leached nitrogen, but also restricted root growth.

"At Sacrewell much of the extra N applied could not be utilised, an RB209 prediction of 250kg/ha over-estimating requirement by 41kg/ha. So even when the soil N supply is known, dressings, in my view, must be interpreted in the light of crop potential." &#42

Prediction systems tested at Sacrewell Farm

&#8226 Hydro PrecisioN.

&#8226 Jubil (1998 only).

&#8226 Kemira N-Min.

&#8226 N_Able.

&#8226 RB209.

&#8226 Solomon (1999 & 2000 only).

&#8226 Sundial.

N need forecasting

&#8226 Several approaches.

&#8226 None infallible.

&#8226 More reliability required.

&#8226 Crop potential still a factor.

&#8226 Several approaches.

&#8226 None infallible.

&#8226 More reliability required.

&#8226 Crop potential still a factor.

&#8226 Hydro PrecisioN.

&#8226 Jubil (1998 only).

&#8226 Kemira N-Min.

&#8226 N_Able.

&#8226 RB209.

&#8226 Solomon (1999 & 2000 only).

&#8226 Sundial.