Statistical considerations in seed purity testing for transgenic traits

Recent advances in agricultural biotechnology have produced many new crop varieties with valuable transgenic traits. These varieties are being, and will continue to be, marketed alongside conventional non-transgenic varieties. As a result, seed purity in commercial seed lots is of particular importance to both seed consumers and seed producers. A key step in the seed production process is the design of sampling and testing procedures used to evaluate seed lot purity. However, due to uncertainties in such methods, there is always a risk of incorrectly rejecting or accepting a seed lot. This paper discusses factors that should be considered when designing and implementing seed purity testing procedures to manage this misclassification risk – especially with regard to the presence or absence of transgenic traits. Many sources of uncertainty in both seed lot sampling and in the assay methods are described, and recommendations for reducing their impact are provided. This paper also explains the statistical concepts of misclassification risk as it affects seed producers and seed consumers. Sampling plans and formulas for determining the sample sizes necessary to control these misclassification errors when accepting or rejecting seed lots are also provided. Both simple, single-stage testing plans and the, often more efficient, double-stage testing plans are described. Testing seed pools rather than individual seeds is introduced as another way of adding efficiency to the testing process. Formulas are given for determining, from a seed sample, the confidence limits for the actual purity level of a commercial lot.