Rethinking the approach to viability monitoring in seed genebanks

Seed viability monitoring, usually through a germination test, is a key aspect of genebank management; a low viability result triggers the regeneration of an accession in order to ensure that the genetic diversity of the accession is conserved and available for distribution. However, regular viability monitoring of large collections is costly in terms of seeds, labour and other resources. Genebanks differ in how they conduct their viability monitoring and how they collect, manage and store the data that are generated. In this article, we propose alternatives to the current norm of conducting an initial germination test soon after arrival at the genebank and then testing after regular, set storage intervals, as recommended in the Food and Agriculture Organization's Genebank Standards for Plant Genetic Resources for Food and Agriculture. We use real data from the International Rice Genebank (held at the International Rice Research Institute) to illustrate some of the issues regarding the accuracy and reliability of germination test results, in particular when they are used to predict the longevity of a seed lot in storage and to set viability monitoring intervals. We suggest the use of seed storage experiments on samples of seeds to identify which seed lots from a particular crop season to test first. We also give advice on the use of sequential testing schemes potentially to reduce the number of seeds used for viability testing; the use of tolerance tables to identify unlikely results when samples are subdivided into replicates; and what data to include in a genebank management database to improve the management of seed collections.