A contribution to reduce sampling variability in the evaluation of deoxynivalenol contamination of organic wheat grain.

Fusarium head blight caused by different varieties of Fusarium species is one of the major serious worldwide diseases found in wheat production. It is therefore important to be able to quantify the deoxynivalenol concentration in wheat. Unfortunately, in mycotoxin quantification, due to the uneven distribution of mycotoxins within the initial lot, it is difficult, or even impossible, to obtain a truly representative analytical sample. In previous work we showed that the sampling step most responsible for variability was grain sampling. In this paper, it is more particularly the step scaling down from a laboratory sample of some kilograms to an analytical sample of a few grams that is investigated. The naturally contaminated wheat lot was obtained from an organic field located in the southeast of France (Rhône-Alpes) from the year 2008-2009 cropping season. The deoxynivalenol level was found to be 50.6 ± 2.3 ng g⁻¹. Deoxynivalenol was extracted with a acetonitrile-water mix and quantified by gas chromatography-electron capture detection (GC-ECD). Three different grain sampling techniques were tested to obtain analytical samples: a technique based on manually homogenisation and division, a second technique based on the use of a rotating shaker and a third on the use of compressed air. Both the rotating shaker and the compressed air techniques enabled a homogeneous laboratory sample to be obtained, from which representative analytical samples could be taken. Moreover, the techniques did away with many repetitions and grinding. This study, therefore, contributes to sampling variability reduction in the evaluation of deoxynivalenol contamination of organic wheat grain, and then, at a reasonable cost.