a= (European Investment Bank 2018)
In LC germination experiments, such as those using incubators, most variables are kept constant (e.g. light intensity, sowing medium, water-availability, timing of diurnal temperature cycle), and one or two are manipulated with a few combinations (e.g. three or four temperatures of diurnal cycles). Researchers select the conditions of variables based on knowledge from NH microclimates, but in reality, these are notoriously difficult to exactly define (e.g. Dinsdale, Dale & Kent 2000). Indeed, important variables may inadvertently be omitted from experimental designs. Interpretation of LC experiments in an ecologically meaningful way is difficult because it will be based on many assumptions, both in selecting variables to test and extrapolating interpretation to NHs - especially if NHs have not been adequately studied.
By contrast, in NH experiments, many, often unknown, variables interplay, one or two of which may be controlled. If researchers were to control all combinations of NH variables in LCs they would soon run out of seeds, time and space. Interpretation of NH experiments depends on dynamically recorded variables that cannot be well controlled, some even being irrelevant to germination ecology. It goes without saying that NH experiments can only truly be performed in regions where the plant is native, whereas LC experiments can be carried out anywhere with suitable equipment.
Alternatives to NH and LC are semi-natural habitats (Semi-NHs) or simulated natural environments (Simulated-NHs). Examples of Semi-NHs include fields or farm edges, in or close to a species’ native region. The term simulated-natural environment (or habitat as used here) was used by Kaeberlein et al. (2002) to define an environment of natural seawater and sediment the authors placed in aquariums to culture previously ‘uncultivable’ marine microorganisms. Such Simulated-NHs are less controlled than LCs, but allow better interpretation of findings and may include important factors that are not well understood or known and so may inadvertently be omitted from LCs. Glass houses, such as those in botanic gardens, are examples of Simulated-NHs, as they mimic NHs and variables are not well under control.
In botanic gardens, living collections are often arranged according to geographic plant communities, each compartment or grouping representing a pseudo or Simulated-NH. These living collections are a valuable resource in studying plant ecology, particularly when NHs are challenging to access (Perez et al. 2019). Many botanic gardens also hold seed banks (469 gardens), and carry out seed or spore research (155 gardens) (BGCI 2021). As botanic gardens are biased towards temperate regions in the Northern hemisphere (Mounce, Smith & Brockington 2017), there is opportunity to enhance interpretation of seed germination studies using botanic gardens as Simulated-NHs when it is not possible to do so in native regions (Faraji & Karimi 2020).
Wild banana species (Musa L.) are native to tropical and subtropical Asia to the western Pacific (Govaerts & Häkkinen 2006). Their fruit contains many hard darks seeds, 3-7 mm in diameter (Chin 1996). The conditions for germination are not well understood and germination is notoriously inconsistent and often very low (Kallowet al. 2020; Panis, Kallow & Janssens 2020; Singh et al.2021). LC experiments show a requirement for alternating temperatures (Stotzky & Cox 1962; Kallow et al. 2021), but no NH experiments have been executed to interpret this. For instance, is this requirement a gap or depth detection mechanism affected by microclimates? And do species respond differently?
Understanding seed germination ecology of wild bananas is not only of ecological interest, it is also important for global food security. Seed banking crop wild relatives efficiently protects genetic material and makes it available for phenotyping and breeding (Dempewolf et al.2017), it is included in UN Sustainable Development Target 2.5 (UN General Assembly 2015). Optimized germination is a vital component of seed bank management and breeding - without it, viability is difficult to monitor and access to plants for research and breeding is constrained (FAO 2014; Batte et al. 2019; Amah et al. 2020).
In the present study we examined germination responses in a Semi-NH and Simulated-NHs of the two primary crop wild relatives of banana:Musa acuminata (subsp. siamea N.W. Simmonds, and subsp.burmanicca N.W. Simmonds), and M. balbisiana Colla (De Langhe et al. 2009). Specifically, we aim to answer the following questions that cannot be answered in LCs: (1) What environments stimulate or inhibit Musa germination? (2) Are Musa seeds dormant, and if so how is this broken in the environment? (3) CanMusa seeds remain viable in the soil?

Materials and methods

  1. Semi-natural habitat (nursery, Philippines)
  2. Plant material
We collected a bunch (an infructescence) of Musa balbisiana(accession GB61996) containing seeds from the field genebank at the National Plant Genetic Resources Laboratory (NPGRL), Institute of Plant Breeding, University of the Philippines, Los Baños. Seeds were extracted by opening fruit and washing seeds in flowing water to remove all pulp. Seeds were then left on a tray in the laboratory to surface dry for seven days prior to sowing.
Table 2. Accessions used for germination experiments in simulated natural environments, V= viability percentage from embryo rescue tests in 2019 and 2020.