Fig. 4 Segmented GLMs of binomial germination outcomes back-transformed to the response probability scale as response to microclimate variables in simulated-NHs, dashed lines are breakpoints and pink shading is standard error (0.95), residuals are red circles.
Survival and dormancy loss in the soil
Overall, 40% of seeds that were buried in the soil for two years and 68% for one year were successfully exhumed from the soil. The probability of finding seeds was modelled against year, burial-depth, and accession (Fig. 5a). Seed burial-depth was the most important factor, deep-buried seeds were more likely to be successfully exhumed (z=16.722, p<0.001), followed by year (z=13.379, <0.001). Surprisingly, seeds with low viability when sown had high probability of being exhumed (bur60 z=7.242, p<0.001, sia61 z=7.025, p<0.001).
Viability of exhumed seeds were tested by TTC and incubation. Viability tests from these, original ER and maximum in Simulated-NHs, was modelled in a GLM. In contrast to the results of the above, accession was excluded from the MAM, year remained. Contrasts were made per year against original ER. Viability from the incubator test was not lost during one year in the soil. After two years in the soil, viability was reduced in both incubator and TTC tests. The TTC test consistently underestimated viability.