by Ken Janovicek, University of Guelph & Greg Stewart, OMAFRA

DON CONCENTRATION MEASUREMENT IN GRAIN CORN

MANY CORN PROCESSORS in Ontario have recently reduced acceptable maximum DON concentration from two to one parts per million (PPM).  Although management decisions such as hybrid choice are important for reducing the chance of high DON concentrations, it is also important that corn sample collection and analysis is conducted properly to ensure a low chance of false high DON readings.

A recent study found that four to six samples per truck is the optimum number to reduce inaccurate DON concentration estimates. Properly blending samples and using accredited labs are also important.

The project, conducted in the summer and fall of 2007, evaluated sampling protocols of 21 40-tonne truck loads of Ontario corn. Using grain probes or tailgate swiping 10 samples were taken from each truckload.

Each of the 10 samples was well blended and divided into smaller samples.  One of the smaller samples was analyzed to provide an estimate of DON concentration for the probe or sample.  DON concentration for the truck load as a whole could be obtained by averaging the DON concentrations from each of the 10 samples collected per truck.  

Another method evaluated was based on first mixing equal portions of grain from each of the 10 probes into a single composite sample.  The composite sample was thoroughly blended and then a single sample was drawn to provide an estimate of truck load DON concentration.

SAMPLE NUMBER NEEDED FOR ACCURATE ASSESSMENT
DON concentration for a single sample was within a range of 65 to 150 percent of the average of all 10 probes per truck nine times out of 10 (Table 1 shows the variability in DON concentration for each of the 10 probes for two truckloads). 

Similar variability was observed in each of the 21 truck loads in this study.  It is clear that use of single probe sampling protocols would provide an unreliable assessment of DON concentrations for the truck load as a whole.

Increasing sample number decreases the error associated with estimates of DON concentration.  Increasing samples to four per truck reduced the range in DON concentration estimates by half with a range between 80 and 125 percent of the actual concentration nine times out of 10.  Taking more than four samples will further reduce the error, but reductions in the size of the errors are small and it may not be worthwhile to collect more than six samples per truck.

This project clearly indicates that single samples can almost always identify truck loads of corn with actual DON levels of 0.5 parts per million (PPM) or less as having concentrations that are less than the 1 ppm limit.  Similarly, a single sample will almost always identify loads that are actually 1.8 ppm or higher as exceeding the 1 ppm limit.  However, inaccuracy can become apparent for loads with actual DON concentrations that are within 0.5 to 1.8 ppm. For a truck load with actual DON concentration of 0.8 ppm, there is about a one in five chance of a DON estimate that exceeds 1 ppm using a single sample.  If six samples are taken, the chance of wrongly obtaining a DON estimate that exceeds 1 ppm is reduced to less than one in 50.

SAMPLE COMPOSITION
Blending multiple samples taken from a truck can provide a reasonably accurate estimate of DON concentration.  Accuracy of DON concentration estimates obtained from blended samples can be improved by ensuring that equal amounts of corn are taken from the individual probes or samples and that the composite sample is thoroughly blended before a sample is taken for analysis.  

ASSESSMENT OF DIFFERENT LABS
Composite samples for each truck were also sent to four Ontario labs to determine accuracy across labs.  One of the labs did produce DON concentration estimates that significantly differed from the other three labs.  When this lab reanalyzed the samples, their results were more accurate.  

Reasons for incorrect analysis are not clear and could have been due to problems with sample preparation or with test kits.  Accuracy of lab analysis can be improved by routinely including samples of known DON concentration and perhaps review of accreditation of lab protocols which will help standardize results produced by the various labs.

DECREASING THE RISK OF INNACURATE ESTIMATES
Assessment of DON concentrations should be based on collection of four to six samples per truck which represent the entire load.  Collecting less than four samples can significantly increase the chance of incorrectly rejecting a truck load for DON concentration above 1 ppm when concentrations are actually below the limit.  

Similarly, collecting less than four samples per truck can significantly increase the chance of incorrectly accepting a truck with actual DON levels that exceed 1 ppm.  Collecting more than six samples will increase accuracy of estimates, but the improvement may not be worth the effort.  

Care should be taken to ensure that samples are collected in a way which represents the entire load, thoroughly blended and that the various labs conduct the analysis in a way that is verifiable and repeatable.