Aging and longevity in decades-old genebanked seeds from U.S. endangered plant species: Assessments using survival and RNA integrity assays.

PREMISE: Seed longevity is critical for successful genebanking, but it is hard to detect or predict. We examined survival of genebanked seeds from species native to the United States to estimate longevity. We tested whether RNA integrity (RIN) can be used to detect aging and predict mortality.
METHODS: Dry seeds from >100 species were stored for 28 ± 7 yr at -18°C. A recently harvested sample (cohort) from the same population provides a zero-time reference. Germination and RIN were assessed and differences between cohorts were used to distinguish short-lived seeds from long-lived seeds.
RESULTS: No differences in germination or RIN were detected between cohorts in about one-fourth of the species. Viability and/or RIN was lower in the stored cohort than in the recently harvested cohort in most species, and the size of the difference was used to infer aging rates. Differences in germination and RIN were correlated among the 100 samples tested; moderate correlation coefficients indicate that additional factors are involved in seed aging and its detection.
CONCLUSIONS: Overall, longevity in the genebank appears to be similar for seeds from wild and domesticated species. We identified species that appeared to produce quite long-lived and short-lived seeds. Seeds from wild species tend to germinate slowly and asynchronously, and this confounds comparisons across storage times; deterioration is detected mostly after severe mortality. By contrast, RIN values decline before viability loss is detected and appear to be unaffected by wild seed traits. RIN tests during early storage can help predict seed longevity.