How to get more accurate DON toxin readings

Elevator operators are advised to grind the entire sample before testing to get more consistent results

Reading Time: 3 minutes

An industry committee has recommended changes to elevator testing procedures for deoxynivalenol (DON) after probing into the inconsistent results farmers found so frustrating during the fusarium-plagued 2018 corn harvest.

While many have blamed the sampling probe used to collect samples as trucks pull into the elevator, subsequent testing has shown the inconsistent results arose from how those samples were ground and tested, said the University of Guelph’s Art Schaafsma, who worked with staff from Ontario Ministry of Agriculture, Food and Rural Affairs, elevator operators and ethanol plant workers to better understand the source of variation.

Related Articles

Why it matters: Inconsistent DON tests in 2018 were frustrating for buyers and sellers. Not being able to identify where the problems were occurring aggravated the issue.

Numerous trucks were denied the ability to dump at elevators in 2018 as the DON levels in the corn they were carrying were too high.

“Many folks would turn around, take a break, get re-tested and be able to unload even though it was the same load. It was no good for either the buyers or the sellers. Everyone was frustrated,” says Schaafsma.

He now recommends that the entire two kg sample be ground before testing for DON.

A wet and warm growing season provided excellent conditions for the growth of mould in corn in 2018, resulting in high volumes of corn being rejected by buyers or sold at a great discount by farmers.

As trucks pull into elevators, four probe samples of corn are taken to test for moisture, test weight, DON and dockage.

“There’s potential for the problem to be within taking the probe sample, the subsampling, the grinding, the kit and the test procedure itself.”

Schaafsma and the team took numerous samples from the rear of 100 trucks as they unloaded into elevator pits to ensure they were receiving precise, non-varied samples.

As they compared the sample results from both the rear of the truck and the probe, they found the variation in test results was consistent with that of the probe samples, proving the type of sample taken isn’t where the issues lie.

“We found the probe sample wasn’t really the problem. The subsampling in the grading house was a big problem and we were able to find out some reasons behind that,” says Schaafsma.

The remainder of corn not used from the probe samples, about 1.5 kilograms, was then divided into four sub samples and the remaining corn from the five-gallon pail was mixed and ground.

The 1.5 kilograms was divided into four subsamples, these were tested and then compared to the corn from behind the truck.

The results from the DON tests of these samples were then plotted on a graph (see below).

High variation of results with first splitting the sample and then grinding.
photo: Graph courtesy of Art Schaafsma

Lower variation of results with grinding the whole sample first and then testing.
photo: Graph courtesy of Art Schaafsma

“You can see that it’s not very well correlated, there is a lot of variation showing up within that small sample that we took with the probe, which we divided into four. That’s already a clue that something is wrong.”

The four sub samples from the probe were then recombined, mixed well, and ground. That removed much of the variation.

Further research and tests showed the variation in DON was among individual kernels, some having really high volumes and some having none.

“It is luck of the draw as to which DON kernels you are going to get when you subsample. Every time you sample, you are going to have a different number of highly concentrated kernels. That is going to throw your test results off.”

Schaafsma says that if whole two-kilogram samples are mixed and ground before testing, the test results are much more uniform.

Tests were being completed by taking the four samples from trucks to equal two kilograms, with the grain then split, divided into 200 gram lots, then grounded and tested.

“Every time you do that, every 200 grams of whole grain you take showed a different answer each time. (What) we are advising is, rather than split it, first grind the whole sample, take some flour and do your test.”

One of the issues now arising is “how do we grind this stuff up?”

Currently, Schaafsma and the team are working on testing prototypes. They hope to bring them out to elevators this fall.

“We are testing in-house to ensure they all grind the same and understand the best way to clean them out. We are trying to get the best protocol in place.”

About the author


Jennifer Glenney

Jennifer lives on a farm in Cayuga, Ontario and has a lot of experience in the many aspects of agriculture.



Stories from our other publications