Fig. 3 Estimated marginal means of minimal adequate GLMs on germination of Musa seeds buried in Simulated-NHs in Belgium (a) Binomial germination outcome; (b) mean germination time; (c) synchronization index.
Effects of microclimate variables on seed germination
The effect of each microclimate variable on final germination percentage was visualized (Fig. S3). These were then used to estimate breakpoints on GLMs produced from binomial germination outcomes (Fig. 4). Results of this show temperatures (at seed burial level) operate a threshold mechanism, above which germination increases linearly (on the logit scale), within the limits of temperatures achieved in simulated-NHs. Therefore, for germination to occur mean soil temperature needs to be above 19°C, maximum temperature above 23°C, minimum temperature has much less of an effect. At the soil surface, the effect of mean temperature is positively linear to an optimal 24°C after which germination is reduced. The effect of maximum temperature is positive above a threshold of 28°C. Germination increases with soil surface minimum temperature to 15°C, then the positive slope is less steep. Soil warming relates to light intensity asymptotically (as described above); germination increased above 1076 lux (mean). In relation to maximum light intensity, germination increased to a breakpoint at 54000 lux, after which the effect on germination is negative. Comparisons of AICs of these models (Table S1) show at seed burial level the maximum temperature is the best fit; at soil surface level mean temperature is the best fit, and mean light intensity is a better fit than the maximum value during cue periods.