Introduction
For many plant species in fire-prone environments, flushes of seedling emergence are often observed post-fire from a soil-stored seed bank (Whelan, Reference Whelan1995; Auld et al., Reference Auld, Keith and Bradstock2000; Ooi, Reference Ooi2007). The period immediately post-fire is an advantageous time for germination and seedling establishment, with reduced competition and increased soil nutrients (Purdie, Reference Purdie1977; Whelan, Reference Whelan1995; Auld et al., Reference Auld, Keith and Bradstock2000). One of the key factors in the creation and maintenance of long-lived seed banks, seed dormancy, allows seeds to remain in the ungerminated state, even when conditions for germination are otherwise favourable (Baskin and Baskin, Reference Baskin and Baskin2014). This generates persistence that can last over multiple reproductive seasons (Fenner and Thompson, Reference Fenner and Thompson2005). To time germination with the post-fire environment, dormancy-breaking cues are often related to fire (Merritt et al., Reference Merritt, Turner, Clarke and Dixon2007; Ooi, Reference Ooi2007). The link between fire and dormancy break is the clearest for physically dormant species, where fire-associated heat cracks open the hard seed coat (Aronne and Mazzoleni, Reference Aronne and Mazzoleni1989; Auld and O'Connell, Reference Auld and O'Connel1991; Gama-Arachchige et al., Reference Gama-Arachchige, Baskin, Geneve and Baskin2013; Ooi et al., Reference Ooi, Denham, Santana and Auld2014), allowing moisture to enter the seed and germination to occur (Baskin and Baskin, Reference Baskin and Baskin2014). For physiologically dormant (PD) species, the relationship between fire and germination is more complicated due to a seasonal component to dormancy release (Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016; Collette and Ooi, Reference Collette and Ooi2017).
Despite an increasing understanding of dormancy and germination of PD species from fire-prone systems, there is still a large knowledge gap for many groups of species. Species from some fire-prone environments can have very deep long-held dormancy that makes identifying dormancy-breaking requirements experimentally difficult (Dixon et al., Reference Dixon, Roche and Pate1995; Ooi et al., Reference Ooi, Auld and Whelan2006; Merritt et al., Reference Merritt, Turner, Clarke and Dixon2007; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016). Furthermore, within plant communities, the level of dormancy can differ, even between closely related species. Quantifying the different dormancy mechanisms of multiple species allows us to compare the different life-history strategies between co-occurring species and to identify those groups of species potentially at risk from changing environmental conditions.
For physiological dormancy to break, hormonal changes need to occur within the seed which are triggered by environmental cues (Finch-Savage and Leubner-Metzger, Reference Finch-Savage and Leubner-Metzger2006; Baskin and Baskin, Reference Baskin and Baskin2014). The primary dormancy-breaking cue is stratification at seasonal temperatures and dry after-ripening (Baskin and Baskin, Reference Baskin and Baskin2014). In fire-prone ecosystems, further germination cues are also required after dormancy has been broken and these are often fire-related, including smoke, heat, changes in light and soil nutrient changes (Thomas et al., Reference Thomas, Morris and Auld2003; Fenner and Thompson, Reference Fenner and Thompson2005; Merritt et al., Reference Merritt, Turner, Clarke and Dixon2007; Ooi, Reference Ooi2007; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016). For example, Roche et al. (Reference Roche, Dixon and Pate1998) explored the interactions between season and fire cues for 37 species from Western Australia. They found that fire cues applied in autumn promoted a greater germination response than treatment in winter or spring. In the east of Australia, Collette and Ooi (Reference Collette and Ooi2017) found that Asterolasia buxifolia, a PD species, required winter seasonal temperatures, smoke cues and light for germination, with no germination at any other seasonal temperature or without smoke.
Very deep dormancy is likely to be prevalent in fire-prone regions where species have very long generational turnover times, and subsequently, where there is little selective pressure for rapid loss of dormancy. For example, in many parts of temperate Australia, the length of time between fires determines generation turnover and can broadly range between 10 and 30 years (Enright and Thomas, Reference Enright and Thomas2008). However, there are numerous environmental factors that influence the level and type of dormancy maintained by a species. Dormancy depth is known to differ along gradients of temperature (Wagmann et al., Reference Wagmann, Hautekèete, Piquot, Meunier, Schmitt and Van Dijk2012; Debieu et al., Reference Debieu, Tang, Stich, Sikosek, Effgen, Josephs, Schmitt, Nordborg, Koornneef and de Meaux2013) and rainfall (Harel et al., Reference Harel, Holzapfel and Sternberg2011; Ramos et al., Reference Ramos, Diniz, Ooi, Borghetti and Valls2017) both within species and between closely related species. Seasonality of rainfall could also drive the development of dormancy but has not often been tested.
In temperate Australia, species with PD make up ~49% of shrub species (Collette, unpublished data). However, dormancy and germination cues for many of these species are poorly known (Auld, Reference Auld2001; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016). Here, we investigate how dormancy-breaking seasonal cues interact with fire-related germination cues and other environmental factors for 18 species from the Rutaceae family from across temperate Australia. This region of Australia consists of similar vegetation types and fire regimes, as well as closely related component species that are adapted to fire (Miller and Murphy, Reference Miller, Murphy and Keith2017), making it ideal for assessing how dormancy changes along rainfall and temperature gradients while controlling for other key factors. Including environmental variables in our analysis, will allow us to make more accurate predictions about how fire season may affect post-fire emergence across different climate regions. Furthermore, this work will contribute to developing germination protocols for these species, four of which are threatened, making the storage of their seeds in seed banks a more powerful ex situ conservation tool.
Methods
Study species and seed collections
For the germination experiments, species from the Rutaceae family were selected (Table 1), many of which are presumed to have physiological dormancy (Auld, Reference Auld2001; Ooi, Reference Ooi2007), a seasonal germination response and fire-cued germination (Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016; Collette and Ooi, Reference Collette and Ooi2017, Reference Collette and Ooi2020). The majority (16/18) of our study species were from temperate Australia. Five species were from Western Australia (WA) and South Australia (SA), both of which have Mediterranean-type (winter) rainfall, while 11 species were from the aseasonal regions of eastern Tasmania. The remaining two species were from the sub-tropical south east corner of Queensland, which has summer rainfall patterns. The regions are relatively large, spanning ranges of several hundred kilometres. Environmental variables can therefore vary for species within regions.
List of study species and the treatments that were applied during the experiments
State represents the states within Australia from where the seeds came from (TAS, Tasmania; SA, South Australia; QLD, Queensland; WA, Western Australia); rainfall region (seasonality) is the season when rainfall is dominant for each species’ region (based on the Köppen climate classifications); incubator temperatures (mean day/night) were those used representing each season during experiments (NB. Each incubator temperature is independent, not a move-along experiment); actual field temperatures for each season (winter, spring/autumn and summer) are long-term day/night climate averages from nearby Australian Bureau of Meteorology weather stations. Treatments denote those applied for germination experiments (NB. smoke × dark or smoke × heat denotes that species were subject to a fully factorial design at each incubation temperature). Seeds/rep are the number of seeds used in each of the three sample Petri dishes. Seed age, viability (viab) and source (seed bank within each state or field collection) are also reported.
The majority of species were obtained via seeds that had been stored in seed banks from across Australia, including the South Australian Seed Conservation Centre, Tasmanian Seed Conservation Centre and the Threatened Flora Seed Centre, WA (Table 1). Seeds from two species, Boronia keysii and B. rivularis were obtained via fresh collection. Seeds that were obtained from different seed banks had slight differences in their age (Table 1).
Germination experiments
Viability of seeds of each seed lot was assessed via X-ray (Baskin and Baskin, Reference Baskin and Baskin2014). For each species, 20 seeds were X-rayed with a Faxitron® MX-20 to assess seed fill, as a surrogate for viability. Seeds which were fully filled were considered viable (Erickson and Merritt, Reference Erickson, Merritt, Erickson, Barrett, Merritt and Kingsley2016). Seeds that were subjected to X-ray were not used in germination experiments.
The experimental design for germination of all study species included a control and smoke treatment at three incubator temperature cycles simulating summer, spring/autumn and winter for each respective region. However, where seed numbers allowed (for 6 of the 18 study species), an additional assessment under dark conditions was also tested (see Table 1). Note that for one species, Boronia keysii, a heat-shock treatment was included in the design instead of a dark treatment because previous work by Mackenzie et al. (Reference Mackenzie, Auld, Keith, Hui and Ooi2016) had suggested that species from the Valvatae Section within Boronia responded to heat shock. Three samples of 20 seeds were used for each treatment; however, this number varied slightly for some species due to limited seed numbers (Table 1).
Germination experiments were conducted using 90 mm plastic Petri dishes filled with a 0.7% agar and 99.3% reverse osmosis (RO) water agar solution. Using agar dishes allowed us to eliminate light from the dark treated seeds, as we did not need to check or provide them with moisture. Agar dishes were prepared in a laminar flow fume hood to prevent infection via airborne microbes and then left for 24 h to confirm that infection had not occurred before seeds were placed onto it.
The control treatment consisted of applying 1 ml of RO water on top of the agar. The smoke treatment was applied via 1 ml of smoke water, which was made by burning 1 kg of leaf litter collected from native sclerophyll vegetation typical of the study species’ habitats (i.e. Eucalyptus overstorey, with dominant Proteaceae, Fabaceae, Rutaceae and Ericaceae understorey) and passing the smoke produced through water for 2.5 h via a vacuum pump. This provided a highly concentrated solution which was then diluted to 2% with RO for the experiment, a concentration that we had shown to produce the highest germination results in previous studies (Collette and Ooi, Reference Collette and Ooi2017). The dark treatment was achieved by double-wrapping Petri dishes in aluminium foil. The heat-shock treatment was applied by placing the seeds in a dry oven at 60°C for 10 min, heating each sample separately. By heating the seeds to 60°C, we were trying to illicit a germination response but avoid seed mortality, which can occur at higher temperatures in PD species (Collette and Ooi, Reference Collette and Ooi2017). The chosen temperature has been found to produce a positive response in many species in this region (Auld and O'Connell, Reference Auld and O'Connel1991; Collette and Ooi, Reference Collette and Ooi2017).
Treatments were applied at three seasonal temperatures in a fully factorial design, with seasonal average maximum and minimum daily diurnal ranges simulated using temperature- and humidity-controlled incubators. Note that seasonal temperatures were tested independently, not in a move-along design. In total, six incubators were used (two Memmert IPP110plus, two Labec ICC36, one Labec ICC24 and one Thermoline TRIL-200-DL). The temperature within each incubator was monitored using Thermocron iButtons, which were checked periodically through the experimental period. The minimum and maximum seasonal temperatures that the incubators were set to, were calculated from the average temperature from winter, spring/autumn and summer for the location that the seeds were from, although due to incubator availability, the actual seasonal temperatures were only approximated for some species (see Table 1). Seasonal mean maximum and minimum temperatures were calculated from data obtained from nearby Australian Government Bureau of Meteorology weather stations, for summer (combining data for the months December–February), spring/autumn (March–May; September–November) and winter (June–August). The incubators were set to 12 h day/night (or max/min) temperatures with 12 h light/dark cycles.
After seeds had been placed into a Petri dish and the treatment applied, the dish was sealed with Parafilm, which prevented the dishes from drying out. Dishes without a dark treatment were checked weekly for germination for 14 weeks, which was scored when the radicle emerged. Dark treatment dishes were only checked at the end of the 14-week trial period, so no light contacted the seeds.
Environmental variables
We used environmental data layers available on the Atlas of Living Australia (ALA) (available at ala.org.au). All data layers were produced by the CSIRO and are modelled for the current climate. The layers that we used were ‘Temperature – annual mean’, ‘Temperature – coldest month min’, ‘Precipitation – annual’, ‘Precipitation – mean driest month’ and ‘Precipitation – annual seasonality’. Minimums for coldest month and driest month were chosen to identify those regions likely to be subject to frost and drought, respectively, as both provide selective drivers for dormancy. Maximum mean seasonal temperatures in our temperate study region are less variable and were assumed to be covered by the annual mean layer. The temperature variables are measured in degrees Celsius, the annual precipitation is measured in millimetres and the rainfall seasonality is an index derived from the ratio of warm to cool season rainfall totals. Positive values indicate that rainfall is greater in warm seasons, zero values indicate aseasonal rainfall and negative values indicate that rainfall is greater in cool seasons. To get these environmental variables for each of the species included in this study, we added each species to the ALA map and chose a central point within the population where collection had occurred. All data collected is available in Supplementary Table S1.
Data analysis
Final germination proportions were calculated with the percentage of viable seeds from the X-ray analysis factored in. Statistical analyses were all conducted using the R environment for statistical computing (R Core Team, 2018). For all species that produced a germination response across at least two treatments, we first fit a binomial generalised linear mixed model including the two factors temperature and smoke and their interaction, plus seed age as a covariate and species as a random factor to identify the broad drivers of germination. We then used binomial generalised linear modelling (GLMs) with a logit link function to analyse each of these species separately, with either two (temperature × smoke) or three (temperature × smoke × dark (NB. heat for B. keysii)) factors, depending on species (Table 1). Models were therefore in the form: glm(proportion germination ~ temperature * smoke, weighted by the number of viable seeds), where the temperature was the incubator-simulated winter, spring/autumn and summer, and smoke was applied or not applied. For species where dark was also tested, we used models in the form glm(proportion germination ~ temperature * smoke * dark, weighted by the number of viable seeds). For B. keysii only, models were in the form glm(proportion germination ~ temperature * smoke * heat). For 13 species, the models did not converge due to complete separation in at least one of the treatments, in which there was no germination. To enable analysis, one ‘germinant’ was added to one sample from that treatment so the models converged. Model assumptions were assessed by checking for overdispersion using the residual degrees of freedom against residual deviance. Model selection procedures using Akaike's Information Criterion corrected for small sample sizes (AICc) were used via the ‘MuMIn’ package (Barton, Reference Barton2020), and the final model selected based on the lowest AICc score. Where there was competing support for two models, both were considered (Table 2). Overall model significance was tested using analysis of deviance. If the temperature was a significant factor in the final model, we used treatment contrasts in the summary function to generate significance values between temperatures. If smoke was a factor in the final model, we tested if smoke had a significant effect on germination within each season, using the same approach.
Model selection statistics for candidate models for each species that germinated
AICc is the measure of model fit. Where there was competing support for two models, with a change in AICc value (Δ) < 2, both are listed. P values generated from analysis of deviance with χ2 and degrees of freedom (d.f.) also provided. Significance (P < 0.05) is denoted in bold. NB. ‘temp’ or ‘temperature’ denotes incubation temperature representing different seasons.
AICc, Akaike's Information Criterion corrected for small sample sizes.
a The Δ listed is in comparison to the null model.
To initially test correlations between germination and environmental variables, we used Pearson's correlation coefficient. To visualise the correlation matrix, we used the ‘ggcorrplot’ package (Kassambara, Reference Kassambara2019). For the proportion germination, the seasonal temperature and treatment combination that yielded maximum germination for every species was used, meaning that every species had a single ‘proportion germination’ ranging from 0 to 1, regardless of whether it germinated under multiple treatment combinations. Once key environmental variables were identified, we then used linear or quadratic regression to model each of them against maximum germination, and retained the best fitting model. The four environmental variables tested were rainfall driest month (mm), annual mean temperature (°C), mean temperature (°C) of coldest month and the rainfall seasonality index (where values <0 indicate winter dominated rainfall, values of 0 indicate aseasonal rainfall and values >0 indicate summer dominant rainfall).
Results
Overall, there were high levels of dormancy and low levels of germination. Out of the 18 species tested, only eight germinated greater than 10% (Fig. 1), while seven germinated between 1 and 10% and three did not germinate at all (Table 1).
(a) Results for species subject to smoke treatment experiments only, and that germinated at a proportion >0.1 at each of the seasonal temperature treatments applied. Treatments were control (RO water) and smoke (smoke water). Different letters indicate significant differences between seasonal temperatures at less than P = 0.05 level, with control treatments excluded, * indicates significant differences for smoke treatments within seasonal temperatures. (b) Germination results for Boronia keysii, which was the only species where a heat treatment was applied. Germination was primarily restricted to summer temperatures and, as above, different letters indicate statistical differences within the summer treatment.
Germination response
In our global model, both smoke (χ2 = 170.03, d.f. = 1, P < 0.001) and temperature regime (χ2 = 47.30, d.f. = 2, P < 0.001) had a strong influence on germination. There was no significant effect of seed age (χ2 = 0.31, d.f. = 1, P = 0.579). For individual species analysis, smoke appeared to have a strong influence on germination and was included as a factor in the best fitting model for 13 of the 15 species that germinated, however, was a significant factor for only six species (Table 2). Lack of significance appeared to be mostly related to very low germination response for seven species (Supplementary Table S2). Within-season assessment with treatment contrasts showed that the application of smoke water had a significant positive effect on five species, all of which had germinated over 10% (Fig. 1a).
Temperature regime was selected as a factor in the best fitting model for six species and was significant as a main factor in five of those, while dark only appeared for two species but not significant in either (Table 2). Only one interaction, temperature × smoke for B. keysii, was included in any of the final models. Boronia keysii and B. rivularis had a strong germination response to summer temperatures, B. gunnii only germinated at winter temperatures and B. parviflora and Philotheca spicata germinated at two seasonal temperatures (Fig. 1a). There was a slight but a significant increase in germination in winter for B. ramosa (Fig. 1a). Heat shock was only applied to Boronia keysii, and it significantly increased germination. The maximum germination for this species occurred in summer seasonal temperatures with both smoke and heat-shock treatments applied (Fig. 1b).
Relationships between maximum germination and climatic variables
Maximum germination proportion of all species was highly correlated with annual mean temperature and mean temperature of the coldest month, and negatively correlated with rainfall of the driest month and rainfall seasonality. Germination was not correlated with annual rainfall total (see Supplementary Fig. S1). Furthermore, these correlations were supported by regression analyses, where the maximum germination proportion of each species had a negative linear relationship with rainfall of the driest month (F 1,55 = 18.39, P < 0.001; r 2 = 0.24) and a positive quadratic relationship with mean temperature (F 2,54 = 96.71, P < 0.001; r 2 = 0.77) and mean temperature of the coldest month (F 2,54 = 93.87, P < 0.001; r 2 = 0.77). A quadratic shallow U-shaped relationship was also the best fit for the rainfall seasonality index (F 2,54 = 36.82, P < 0.001; r 2 = 0.56; Fig. 2).
Regression models for four key environmental variables (see Supplementary Fig. S1). Values for each environmental variable were obtained from each species’ region and modelled against maximum germination. Rainfall driest month (mm) was fit with a linear regression model and is the long-term average rainfall of the driest month of the year. Mean temperature of coldest month (°C) refers to the long-term average temperature of the coldest month of the year. Rainfall seasonality index refers to rainfall seasonality. Values <0 indicate winter dominated rainfall, values of 0 indicate aseasonal rainfall and values >0 indicate summer dominant rainfall. Annual mean temperature (°C) refers to the mean temperature throughout the calendar year. Both temperature variables and the rainfall seasonality variable were fit with quadratic regression models.
Discussion
The high levels of dormancy found in our study, with approximately half of the species tested germinating below 10%, appears to be typical of PD species in temperate Australian fire-prone ecosystems (Ooi et al., Reference Ooi, Auld and Whelan2006; Merritt et al., Reference Merritt, Turner, Clarke and Dixon2007; Downes et al., Reference Downes, Light, Pošta, Kohout and van Staden2014; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016) and their reputation as ‘difficult to germinate’ for practitioners aiming to utilise seeds for restoration (Dixon et al., Reference Dixon, Roche and Pate1995; Daws et al., Reference Daws, Downes, Koch and Willyams2014). Deep dormancy allows species to build up multiple cohorts of seeds into a soil seed bank (Baskin et al., Reference Baskin, Baskin, Yoshinaga and Thompson2005; Fenner and Thompson, Reference Fenner and Thompson2005), an advantageous trait for species from these fire-prone environments, as it allows the seed bank to build for long periods between fires. In our study region, generational turnover is often associated with fire, which can broadly be in the range of 10–30 years (Enright and Thomas, Reference Enright and Thomas2008).
Of the eight species where germination was above 10%, seasonal incubation temperatures and smoke had the greatest effect on germination. We would not expect smoke effects on dormant seeds (Thompson and Ooi, Reference Thompson and Ooi2010, Reference Thompson and Ooi2013); however, for those that did germinate, smoke consistently provided a positive effect. For both Philotheca species tested, P. spicata and P. tuberculosum, germination within the cues tested here, was absolutely dependent on the smoke cue, while for Boronia keysii, B. rivularis and B. ramosa, germination almost tripled with smoke. Although limited, our study has tested across multiple temperature regimes and, combined with other studies that have tested multiple factors (Thomas et al., Reference Thomas, Morris and Auld2003; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016; Collette and Ooi, Reference Collette and Ooi2017, Reference Collette and Ooi2020), provides some evidence to support the hypothesis that an obligate smoke response is a fire-adapted germination cue. While smoke has been found to be a general promoter of germination across species from fire-prone and fire-free environments, the obligation of a smoke requirement indicates germination strongly bound to fire occurrence. This hypothesis is similar to that of a heat-shock fire adaptation (Keeley et al., Reference Keeley, Pausas, Rundel, Bond and Bradstock2011; Moreira and Pausas, Reference Moreira and Pausas2012), where soil temperatures that can only occur during fire heating are required overcome dormancy (Ooi et al., Reference Ooi, Denham, Santana and Auld2014). It is possible that testing different smoke concentrations would have stimulated germination in more species, or to higher levels (Çatav et al., Reference Çatav, Küçükakyüz, Akbaş and Tavşanoǧlu2014; Moreira and Pausas, Reference Moreira and Pausas2018; Alahakoon et al., Reference Alahakoon, Perera, Merritt, Turner and Gama-Arachchige2020), meaning that this obligate requirement could occur across a greater range of species.
Although there were too few species with a strong response to test formally or robustly, the maximum germination appeared to be correlated with rainfall season of the source populations. The two summer rainfall region species, B. keysii and B. rivularis, both had maximum germination levels at summer temperatures, while three of the four winter rainfall region species had their maximum germination at cooler incubation temperatures, albeit with only small increases. The two aseasonal rainfall species (B. parviflora and Zieria arborescens) germinated equally well across two or three seasonal temperatures, respectively. A pattern often assumed is that species have germination cued to the temperature range of the rainfall season they occur in, however this assumption is based on little comparative data across sites with different rainfall seasonality. Further investigation of a broader range of species from different rainfall regions would help elucidate this knowledge gap. More broadly, there was a U-shaped relationship between germination and the rainfall seasonality index, meaning that species were more likely to be dormant if they were from aseasonal rainfall areas, and a negative relationship with the average rainfall of the driest month. One possible explanation for this is that species in areas with distinct dry seasons have emergence driven by the timing of rainfall and may have less selective pressure for deep dormancy development.
Germination was also highly positively correlated with annual mean temperature and mean temperature of the coldest month. It is possible that this is due to deeper dormancy levels in colder environments to avoid frost. The cold climate areas of our study are located in Tasmania and the Blue Mountains region near Sydney, where the mean temperatures of the coldest month are between 0 and 2°. Frost conditions are lethal for seedlings of many species, even those that are from alpine environments (Marcante et al., Reference Marcante, Sierra-Almeida, Spindelböck, Erschbamer and Neuner2012). Avoiding germinating in frost conditions through the development of deeper dormancy mechanisms is one strategy to avoid frost (Spindelböck et al., Reference Spindelböck, Cook, Daws, Heegaard, Måren and Vandvik2013). Additionally, the germination of PD species has been shown to alter due to differences in climate between latitudinal gradients, where seeds from warmer environments are less dormant than seeds from cold environments within the same species, possibly through phenotypic plasticity (Chamorro et al., Reference Chamorro, Luna and Moreno2018).
Darkness did not have any effect on the four species that we tested; however, this result is limited by the fact that only one of the species tested with dark germinated to levels that allowed robust analysis. Dark is commonly tested as a cue in PD species (Baskin and Baskin, Reference Baskin and Baskin2014) and has previously been shown to affect germination of Australian PD species (Bell, Reference Bell1999; Gilmour et al., Reference Gilmour, Crowden and Koutoulis2000; Merritt et al., Reference Merritt, Kristiansen, Flematti, Turner, Ghisalberti, Trengove and Dixon2006), including from within the Rutaceae family (Collette and Ooi, Reference Collette and Ooi2017, Reference Collette and Ooi2020). However, the general lack of germination studied here leads to a broader question regarding PD species in fire-prone regions such as Australia. Fire return intervals on the scale of decades provide very little selective pressure to have seeds that lose dormancy rapidly, such as over a single season or year. While it is common that seeds of PD species require stratification at warm or cold temperatures for a period prior to other seasonal temperatures in order to germinate (Baskin and Baskin, Reference Baskin and Baskin2014), simple single stratification treatments rarely work on Australian fire-prone species (Ooi, Reference Ooi2007; Merritt et al., Reference Merritt, Turner, Clarke and Dixon2007). Dry after-ripened or aged seeds may be required to gain a better understanding of germination response and temperature preferences (Ooi et al., Reference Ooi, Auld and Whelan2006) despite the fact that this will require forward planning by researchers.
The germination of B. keysii, the only species to be tested with heat shock, was significantly affected by seasonal temperatures, smoke and heat-shock treatments. The highest germination occurred at summer temperatures with smoke and heat shock applied. This supports the findings from Mackenzie et al. (Reference Mackenzie, Auld, Keith, Hui and Ooi2016) that showed that this section of Boronia (Valvatae) was likely to be affected by heat-shock cues. This mechanism, where fire-related heat shock increases germination of species with permeable seeds, has been identified for several PD species (e.g. Auld and Ooi, Reference Auld and Ooi2009; Mackenzie et al., Reference Mackenzie, Auld, Keith, Hui and Ooi2016) and deserves further investigation.
A limitation of this study was imposed by the use of seeds of different ages between species. Age can affect the dormancy of seeds via after-ripening (Holdsworth et al., Reference Holdsworth, Bentsink and Soppe2008; Baskin and Baskin, Reference Baskin and Baskin2014) and dormancy cycling (Baskin and Baskin, Reference Baskin and Baskin2014; Collette and Ooi, Reference Collette and Ooi2020); however, we found no effect of seed age on germination response. These effects may have been somewhat mitigated in our study, because cryostorage of our seed banked species would potentially have minimised these effects (Hay et al., Reference Hay, Merritt, Soanes and Dixon2010). Nevertheless, we suggest further study using a range of dormancy-breaking treatments would provide greater insights into comparative germination responses across these regions.
Ex situ seedbanking is an increasingly important method of conserving plant species around the world (Cochrane et al., Reference Cochrane, Crawford and Monks2007). It is a cost-effective way of storing plant species, because seeds take up little space, and in many cases can be stored for long periods (Walters et al., Reference Walters, Wheeler and Stanwood2004; Li and Pritchard, Reference Li and Pritchard2009). The difficulty with using seed banking for the ex situ conservation of deeply dormant seeds is that germination of those seeds is difficult, limiting their conservation potential. The germination protocols we have used in this study highlight the need to systematically assess seed lots under ecologically relevant conditions in order to gain information for these difficult to germinate species. The eight species that responded in our study showed that this approach can be useful, especially for species where significant data would not be expected from germination trials. Development of protocols such as these greatly increases the value of such species stored in seed banks.
Supplementary material
To view supplementary material for this article, please visit: https://doi.org/10.1017/S0960258520000422.
Acknowledgements
We thank the South Australian Seed Conservation Centre, Tasmanian Seed Conservation Centre and the Threatened Flora Seed Centre, WA, who supplied the seed used in this experiment. Sophie Yang and Josee Hart aided with the germination experiments. This work was funded by the Oatley Flora and Fauna Conservation Society Research Grant and the SE Queensland Fire and Biodiversity Consortium. Mark Ooi's research is supported by the NSW Government's Department of Planning, Industry & Environment via the Bushfire Risk Management Research Hub and ARC Linkage Project LP180100741. Justin Collette's research is supported by an Australian Government Research Training Program (RTP) Scholarship.
