Germination and viability assays
Germination rate under standardized conditions was used to quantify seed dormancy (as in Postma and Ågren 2015). Only lines that had produced at least 150 seeds were included in germination tests. For seeds produced in the greenhouse, we monitored changes in seed dormancy over time by determining the proportion of viable seeds germinating 1, 3, 12, 30 and 54 weeks after harvest. Due to limited seed availability, dormancy of seeds produced at the two field sites was possible to test only once. For these seeds, we chose to test germinability 12 weeks after harvest, because this was the time point at which variation among recombinant inbred lines derived from a cross between an Italian and a SwedishA. thaliana population was highest in a previous study of dormancy of seeds produced at the two field sites (Postma and Ågren 2015). In all tests, about 30-100 seeds of each line were sown on 2 filter papers watered with 2 mL of demineralized water in each of three petri dishes. After sowing, petri dishes were placed in a growth room and regularly watered for one week (20 °C 16 h light with photosynthetically active radiation of 150 μmol m-2s-1 and 16 °C 8 h dark). One week after sowing, we recorded for each line the number of germinated seeds and the total number of seeds sown.
To quantify germination proportion as number seeds germinating per viable seed sown, we tested seed viability of lines with a germination proportion below 95% in a given maternal environment by exposing seeds to conditions expected to break any seed dormancy. About 50 seeds per line were sown on agar in petri dishes under sterile conditions. The petri dishes with sown seeds were kept in darkness at 6 °C for two weeks, and then moved to randomized positions in a growth room (22 °C 16 h light with photosynthetically active radiation of 150 μmol m-2 s-1 and 16 °C 8 h dark). Viability was scored as proportion of seeds that had germinated after two weeks in the growth room (see Postma & Ågren, 2015). If the viability of a given line was below 95%, its viability was tested a second time and the mean viability of the two tests was used in the analysis. We estimated number of seeds germinating per viable seed using the expression, number germinated seeds/(number of seeds sown × proportion viable seeds), and included only lines that had a mean viability greater than 95% in the statistical analysis of variation in seed dormancy. In total, we included in the analysis 214 lines representing 45 populations grown in the greenhouse, 172 lines representing 43 populations grown at the Italian site, and 108 lines representing 28 populations grown at the Swedish site.