Looking back to the future

24 Aug 2018

I have been motivated to emerge from blogging retirement by the latest AHDB Project Report on applied nitrogen (N) doses for feed winter wheat. As regular readers of my blogs know, I have a ‘thing’ about the subject. This is because field trials data do not support some of the underlining principles that have been used to develop the recommendations in RB 209. I eventually wrote papers on the subject in 2010 and 2012 (referred to in the AHDB report) and came to the conclusion that on mineral long-term arable soils, where little or no organic amendments are used, using a fixed dose of N produces economic margins at least equal to those using RB 209 as a guide. The field trials data indicate that only when Soil Mineral Nitrogen (SMN) levels exceed around 100 kg N/ha, equivalent to N index 4 and above, is there a consistent need to contemplate reducing doses of applied N.

The papers I wrote were met by a howl of outrage by many soil scientists. It seemed that I had committed heresy. However, their field trials data said the same as NIAB TAG’s. Some soil scientists insisted that the scientific literature backed the approaches taken by RB 209, particularly the influence of measured SMN on recommendations. I had already read the papers they quoted and knew that these actually concluded the opposite by detailing the lack of influence of measured SMN on the optimum economic dose of N for feed winter wheat.

Let me acknowledge that I was not the first to suggest a fixed dose of N for feed winter wheat. During the 1980s, the then newly available measurement of SMN and also the first attempts to estimate Soil Nitrogen Supply (SNS = SMN plus nitrogen in the crop in early spring plus an allowance for N net mineralisation of the soil) were being tested by ADAS for their possible role in N recommendations. A range of tactics employed in Northern Europe was tested. However, in 1987 ADAS reported that using a fixed rate of N gave as good or better prediction of the optimum economic dose than using measured SMN or SNS or an N index system as the basis of a recommendation.

Let us look at a couple of quotes from the latest AHDB report:

“Little relation between soil mineral N or grain yield and N optimum was observed.” The lack of relationship between measured SMN and N optimum is nothing new and the relationship between grain yield and N optimum has always been rather tenuous.

“….. in long-term arable situations with high yields where N requirements were expected to be similar, it was difficult to improve on recommendations beyond RB209 (or a single average N rate).” The last caveat is extremely important and reflects the following table from the report which shows that adopting a single fixed rate of N in all the trials was as cost effective as using RB 209. This very important conclusion is not mentioned in the abstract of the project report.

Comparison of different approaches to deciding N fertiliser rates

Using average results from previous trials on mineral long-term arable soils with little or no organic amendments, the single standard rates in the table are predictable and this approach has been the basis of NIAB TAG recommendations for the past few years. Please remember that field yields tend to be lower than plot yields because of lower yielding headlands etc.

It is worth highlighting that the basis of the RB 209 recommendations in the report was the Field Assessment Method. This is based on previous cropping. The approach based on measured SNS is now only being advocated where very high levels may be expected. The lack of a relationship between measured SMN and optimum doses in this report underscores the advice in RB 209 on which method to adopt.

I have often mused why measured SMN or SNS has so little influence on applied N requirement for feed winter wheat. There are some possible explanations. First of all, the efficiency of use of every extra Kg/ha of SMN or SNS above a threshold (i.e. the marginal efficiency) of around 40-50 kg N/ha (N index 0) is far less than the 100% assumed in RB 209. In this newly reported AHDB project this marginal efficiency of use of SNS was less than 50% when no applied N was used. Secondly, higher levels of SNS are often a reflection of good moisture retentive soils and/or a ‘better’ rotation and/or a more healthy soil, resulting in higher yields and consequently a higher N demand by the crop. This could largely offset the need to reduce N doses with increasing measured SNS levels until they become very high. Analysis of an extensive trials database suggests that there may be more than a grain of truth in this theory but the statistical analysis is not convincing. A third, and less techy possible explanation, is that measuring SMN, estimating SNS and also identifying an economic optimum dose are not precise exercises. There may well be other possible explanations. It could be that it is a combination of several factors that explain the lack of influence of measured SNS on optimum economic doses of N for feed winter wheat.

Another quote from the report is “Grain protein cannot therefore be used as an entirely reliable indicator for N management.” This is slightly at odds with the advice in RB 209 which suggests that “Farm nitrogen strategies for wheat can be assessed periodically using information on grain protein concentration.” Perhaps this statement in RB 209 is too dogmatic when it is clear that the AHDB original project report on this approach included caveats on using protein as a guide to optimised N management.

RB 209 recommendations remain based on some underlining principles with which I disagree. However, its recommendations have now been adjusted to reflect the average results of field trials. Does this mean that we have all the answers on N nutrition of feed winter wheat? We certainly do not. The project again demonstrates that there is a huge unexplained variation in the actual optimum economic doses between fields, farms and years. The report does emphasise that there can be an error around the optimum economic dose of plus or minus 50 Kg of applied N/ha without affecting too much the financial margin over fertiliser costs. However, there remain an uncomfortably high proportion of trials where the error in predicting the optimum is outside this range. As AHDB (HGCA) project report 73 (1993) concludes “current recommendation systems are similarly poor because they fail to identify fields with aberrant responses to N”. The 1987 ADAS paper I quoted earlier said that “measurements of SMN have shown promise where small optima are suspected [i.e. very high SMN levels]”. To be perfectly honest we are now back to exactly the same situation we were 25-30 years ago despite a huge investment in research on the subject in the intervening years. It is truly looking back to help define the future research needs of this very important subject.

Now, back to my retirement. There has been sufficient rain for me to sow cover crops on my allotment. I am hoping that the spring oats that I liberated from a farmer’s grain store last year are still viable.