Abstract
Seeds are highly resilient to the external environment, which allows plants to persist in unpredictable and unfavorable conditions. Some plant species have adopted a bet-hedging strategy to germinate a variable fraction of seeds in any given condition, and this could be explained by population-based threshold models. Here, in the model plant Arabidopsis (Arabidopsis thaliana), we induced secondary dormancy (SD) to address the transcriptional heterogeneity among seeds that leads to binary germination/nongermination outcomes. We developed a single-seed RNA-seq strategy that allowed us to observe a reduction in seed transcriptional heterogeneity as seeds enter stress conditions, followed by an increase during recovery. We identified groups of genes whose expression showed a specific pattern through a time course and used these groups to position the individual seeds along the transcriptional gradient of germination competence. In agreement, transcriptomes of dormancy-deficient seeds (mutant of DELAY OF GERMINATION 1) showed a shift toward higher values of the germination competence index. Interestingly, a significant fraction of genes with variable expression encoded translation-related factors. In summary, interrogating hundreds of single-seed transcriptomes during SD-inducing treatment revealed variability among the transcriptomes that could result from the distribution of population-based sensitivity thresholds. Our results also showed that single-seed RNA-seq is the method of choice for analyzing seed bet-hedging-related phenomena.