Kernel lot distribution assessment (KeLDA):: a study on the distribution of GMO in large soybean shipments

The reliability of analytical testing is strongly affected by sampling uncertainty. Sampling is always a source of error and the aim of "good" sampling practice is to minimize this error. Generally the distribution of genetically modified (GM) material within lots is assumed to be random in order to use binomial distribution to make inferences. This assumption was never verified in practice and no experimental data investigating the distribution of genetically modified organisms (GMOs) exist. The objectives of the KeLDA project were: (1) to assess the distribution of GM material in soybean lots (2) to estimate the amount of variability of distribution patterns among lots. The GM content of 15 soybean lots imported into the EU was estimated (using real-time PCR methodology) analyzing 100 increment samples systematically sampled from each lot at predetermined time intervals during the whole period of off-loading. The distribution of GM material was inferred by the one-dimensional (temporal) distribution of contaminated increments. All the lots display significant spatial structuring, indicating that randomness cannot be assumed a priori. The evidence that the distribution of GM material is heterogeneous highlights the need to develop sampling protocols based on statistical models free of distribution requirements.