Practical tests of a prototype scheme for pre-harvest prediction of Hagberg falling number in wheat

Abstract

If Hagberg falling number (HFN) of wheat grain falls below 250 or 220 s, producers lose bread making or export premia. In the UK, low HFN is caused by four different origins of the enzyme alpha-amylase, which digests starch. HGCA project 0056/1/93 investigated the mechanisms of each alpha-amylase pathway and designed a prototype scheme for prediction of combine harvest HFN from pre-harvest HFN and germination measurements. The current project was undertaken to fine-tune the prediction scheme and assess the logistics of commercial operation compared to its use at research sites. A range of current commercial cultivars was grown at ADAS Bridgets, Aberdeen University and Harper Adams University College to allow derivation of specific HFN prediction equations and to further test the scheme. Samples from commercial crops were submitted by three crop consultants to NIAB Labtest to assess the logistics of commercial laboratory operation. Crop consultant samples were also used to assess the appropriate sampling method from the field. Methods of determining the stage of 35% grain moisture, required as a marker for earliest possible sampling, were investigated. This included a comparison of microwave and oven drying of ears, grains or milled samples and assessment of accumulated potential evapotranspiration as a marker of grain moisture. The optimum temperature of the pre- harvest germination test for assessment of sprouting risk was studied by analysis of data from the previous HGCA project as well as 1998 germination test data from research sites. Two methods of HFN prediction, either the prediction class system (developed in 0056/1/93 using pre-harvest HFN and germination data) or probability distribution functions (derived from a general HFN prediction equation without germination data) were assessed.

Due to heavy rainfall at Aberdeen, there were too few samples for derivation of significant individual HFN prediction equations for current cultivars, so a general equation derived from the previous project was used. Analysis of sample size showed that the coefficient of variation for pre-harvest HFN was reduced by sampling from five random locations within a crop, but that no further benefit was gained with more samples. Assessment of grain moisture content by microwave drying was problematical due to condensation and charring. However, oven drying of whole ears gave a good estimate of grain moisture content. A strong logistic relationship was found between accumulated potential evapotranspiration (PE) from ear emergence and grain moisture content, indicating that pre-harvest HFN sampling should not occur before 190 mm PE. Analysis of germination test data revealed different responses to temperature for different cultivars, sites and years with the possibility of optima for germination in three days below 10'C or around 20'C. Thus it was concluded that the best compromise temperature for germination testing was 15'C, although the information should be treated with care since temperature optima for sprouting in the field could be very different. Logistically, operation of the scheme proceeded smoothly, although crop consultants were concerned about the large size and time-intensive nature of pre-harvest sampling. Completion of analysis and reporting of data was possible in 4-5 days. Hand threshing and selection of grains for the germination test was the most labour- intensive process, with the three day test setting a minimum time for delivery of results at four days after sampling. Analysis of the HFN predictions made with the scheme showed the prediction class system, using HFN and pre-harvest germination data, to be about 65% accurate. Use of the probability distribution functions, without reference to germination data, gave an accuracy of about 77%, with data from this system available two days after sampling. Consequently, it seems predictions based solely on pre-harvest HFN data rather than with the time-consuming germination test would be most useful, since results could be returned more quickly and rapid re-testing of samples after significant rainfall events would be possible.