Fig. 6 Germination probability in an incubator (at 35/20 °C, light/dark, 18/6 hours cycled) of seeds buried and then exhumed from the soil for up to three months. Seeds were buried in the exposed Spring compartment.

Discussion

  1. Use of semi-natural and simulated habitats for germination ecology experiments
In the present study we used a Semi-NH from the native range of M. balbisiana , and several Simulated-NHs in glass houses located in a temperate region, to examine wild banana seed germination ecology. We established that with such approximations of NHs, it is possible to link germination responses to ecological factors such as foliar-shading and burial-depth. Hence with this approach we overcame limitations when access to experimental NHs was not possible, this allowed for greater ecological interpretation than with LCs alone.
Temperature
We found that Musa seed germination is stimulated by exposure to the sun. It was the maximum part of the temperature fluctuation, in our results, that is most closely associated with germination. Above the threshold of 23°C, germination increased to a maximum at 35°C (in Simulated-NHs and 42°C in Semi-NHs) in exposed conditions. Our findings are broadly consistent with previous LC results, where optimal maximal and minimal temperature for germination were diurnal cycles of 35/18-20°C respectively, for both M. acuminata and M. balbisiana (Stotzky & Cox 1962; Kallow et al. 2021). Interactions between the elements of warming and cooling cycles play an important role in simulating germination.
Light
One might think that as germination responses were directly associated with soil exposure to sun and, in simulated-NHs, light intensity, germination is stimulated by light. Additionally, we also found that seeds germinated to a greater extent from 1 cm compared to 7 cm. However, light waves cannot usually penetrate the soil to greater than 4-5 mm depth (depending on soil moisture and particle size), and not to any amount that can illicit germination responses to light sensitive seeds (Woolley & Stoller 1978; Tester & Morris 1987), but in our experiment, seeds germinated from a depth of 7 cm. We therefore infer that whilst Musa seed germination may be correlated with factors associated with light (light intensity, shallow burial) theses are correlations rather than causal, and it is temperature that regulates germination.
Gap detection
Musa seed germination in response to sun exposure demonstrates adaptation to detect suitable niches for seedling establishment following disturbance in forest NHs. Conversely, inhibition of germination in shade is also an adaptation for seedling survival (Kos & Poschlod 2007; Poschlod et al. 2013). Musa germination responses were sensitive to sun/shade even when microclimates were very similar. For instance, mean temperatures were the same and range differed by only a few degrees in sun and shaded semi-NHs. Germination is therefore finely tuned to respond to microclimate such as that which would occur when a forest gap is formed (Pearson et al. 2002; Pearson et al. 2003). The effect of forest disturbance on temperature dynamics was studied by Harwick et al . (2015). The authors measured soil (10 cm depth) and air temperature (1.5m height) at three levels of forest disturbance in Borneo. Diurnal temperatures in oil palm plantations (formerly forested) were around 7°C greater at the hottest part of the cycle compared to old growth forests and soil temperatures were around 3 °C warmer at this point and around 1 °C cooler in the night; these were similar to our results in semi-NHs. Temperatures in logged forests were somewhat in between these two, but more similar to old growth forest. One could imagine that even in small forest gaps, microclimates could therefore also vary considerably (Pearson et al. 2002). These responses are in line with adaptations of other disturbance-adapted species that also require alternating temperature cycles rather than constant temperatures to germinate (Vázquez‐Yanes & Orozco‐Segovia 1982; Vázquez-Yanes & Orozco-Segovia 1993; Pearson et al. 2002; Seiwa et al.2009).
Seed burial-depth
When seeds were in the shade, burial-depth made no difference to germination, they did not germinate irrespective of burial-depth. In exposed sites, shallow buried seeds were more likely to germinate than deeply buried seeds. This was because temperature dynamics are likely buffered by burial depth. For some species, sensitivity to alternating temperatures is an adaptation to detect burial depth (Thompson, Grime & Mason 1977; Thompson & Grime 1983). Pearson et al. (2002) found, for large-seeded species, diurnal temperature sensitivity was more likely to related to forest gap size than seed burial-depth. This was also the case for large-seeded Musa , in that exposure was by far the most significant factor in simulated-NHs, and burial-depth was only secondary. For small seeds it is important to detect burial-depth as seedlings must reach the surface with small endosperm reserves; for larger seeds with greater nutrient reserves this is less of a limiting factor for survival.
Survival and dormancy loss in the soil
We found seeds can persist and remain viable buried in the soil for at least two years. In fact, there was no loss of viability after one year, there was however loss of actual seeds. Seed loss was more pronounced with shallow burial, suggesting it is the result of predation or perhaps splashing during watering, rather than decomposition. This is also supported by the fact that accessions with less viable seeds were more likely to be found i.e., seeds were not lost by decomposition of dead seeds.
For seeds adapted to disturbance, seed persistence in the soil seed bank is important. Musa clearly invest considerably in seed coat defenses (Graven et al. 1996), to survive the intense pressure present in the soil community (Dalling et al. 2011).
Not only do seeds persist in the soil, but dormancy is reduced during this process. In our results, when seeds were stratified for three months, or when they were in the soil for a year, germination increased. A stratification requirement is in keeping with results from our previous study (Kallow et al. 2021). Although, stratification was not required for freshly extracted M. balbisiana seeds in Semi-NHs, implying drying induces secondary dormancy in Musaseeds, as proposed by Chin (1996).
There was greater germination synchronization in cooler Simulated-NHs as seeds responded to threshold temperatures when sun was stronger during the summer. When temperatures were consistently warmer, synchronicity was reduced - this may again be a disturbance adaptation. We found this response more evident in M. balbisiana than in M. acuminata, suggesting it may also relate to seasonality as M. balbisiana has a large distribution that includes subtropical seasonal climates (Mertens et al. 2021).

Conclusions

Studying germination ecology has intrinsic challenges, possibly the biggest being access to seeds and experimental set-ups in suitable conditions and timeframes. In the present study we demonstrate an approach for dealing with such difficulties in studying tropical seed germination ecology, which is a challenge when researchers are outside of a plant’s native region. Using Semi-NHs and Simulated-NHs we found: (1) foliage-shading inhibits germination of non-dormant seeds, and exposure to sun stimulates germination; this response is most closely associated with maximum temperature variation found under direct sunlight; this effect is marginally buffered by deep burial in the soil; (2) freshly extracted seeds are non-dormant, but stored seeds lose their dormancy during burial in the soil; (3) Musa seeds remain viable in the soil for at least a year without any loss in viability. Thus, wild banana species are well adapted to exploit canopy gaps following disturbance.

Acknowledgments

We thank Isla Kallow, Jasmin Kallow, Manuela Garcia Zuluaga and Kevin De Pauw for assisting with the setting up the glass house experiments. We gratefully acknowledge the help of John Mark Barios, Lyka Yanos and Paulo Jerome Lopez at NPGRL for setting up and monitoring the germination experiments in the Philippines and for Michelle Lyka V. Descalsota for assistance in viability assessing seed collections at NPGRL. We also thank the late Daniele Roques (CIRAD, Centre de Ressources Biologiques Plantes Tropicales (CRB-PT), Guadeloupe), Josephine Agogbua (International Institute of Tropical Agriculture (IITA), Nigeria) and Zhiying Li (Institute of Tropical Crop Genetic Resources, CATAS Tropical Crops, Danzhou, China) for providing seeds to use. This work was funded as a sub-grant from the University of Queensland from the Bill & Melinda Gates Foundation project ‘BBTV mitigation: Community management in Nigeria, and screening wild banana progenitors for resistance’ [OPP1130226]. The authors thank all donors who supported this work also through their contributions to the CGIAR Fund (http://www.cgiar.org/funders), and in particular to the CGIAR Research Program Roots, Tubers and Bananas (RTB-CRP). This study was supported by a bilateral grant between the Research Foundation - Flanders (FWO-Vlaanderen) and the Vietnamese National Foundation for Science and Technology Development (NAFOSTED) under grant number G0D9318N.

Author contributions

SK: conceptualization, methodology, software, formal analysis, investigation, data curation, writing-original draft, writing-review and editing, visualization; KQ: investigation; BP: conceptualization, methodology, resources, writing-review and editing, supervision, funding acquisition; SBJ: conceptualization, methodology, resources, writing-review and editing, supervision, funding acquisition; JD: writing-review and editing, supervision, funding acquisition; LG: resources; RS: writing-review and editing, supervision; FV: conceptualization, methodology, resources, writing-review and editing.

Conflict of interest

The authors declare that there is no conflict of interest associated with this article and research.

Data availability

All data available at Kallow, Simon (2021): Using semi-natural and simulated habitats for seed germination ecology. figshare. Dataset. https://doi.org/10.6084/m9.figshare.14884470.v1

Supplementary figures and tables

Fig. S1. Average and standard deviations of temperature and light intensity during cue periods of exposed and shaded environments, temperatures are °C, light intensity is lux.

Fig. S2. Light intensity and temperature measured at simulated natural environments in glass houses during germination tests; asymptotic non-linear regression shown (residual standard error 4.119 on 417,661 degrees of freedom).