Introduction
Vascular epiphytes account for c. 27,000 species worldwide (Reference ZotzZotz, 2013b). Traditionally, physiological ecologists have studied larger individuals of this phylogenetically and ecologically diverse group of plants in an endeavour to understand species distributions in time and space (Zotz and Hietz, Reference Zotz and Hietz2001), but more recently there has been increased interest in the earliest ontogenetic processes, namely germination and establishment (for field studies see, for example, Mondragón and Calvo-Irabien, Reference Mondragón and Calvo-Irabien2006; Cascante-Marín et al., Reference Cascante-Marín, Wolf, Oostermeijer and Den Nijs2008; Goode and Allen, Reference Goode and Allen2009; and for laboratory studies, Fernandez et al., Reference Fernandez, Beltramo and Caldiz1989; Manzano and Briones, Reference Manzano and Briones2010; Tsutsumi et al., Reference Tsutsumi, Miyoshi, Yukawa and Kato2011).
Germination is arguably the most vulnerable stage of the plant life cycle and thus a key element of plant life history strategy (Harper, Reference Harper1977). A large proportion of published germination studies with plants in general, and epiphytes in particular, has been performed under controlled conditions in the laboratory. While allowing for strong inference by controlling relevant ambient factors, such studies may also lead to erroneous conclusions because of oversimplified conditions or arbitrary selection of treatment differences. One way to avoid such pitfalls is to inform experimental designs by field data (e.g. on in situ temperature regimes, Tsutsumi et al., Reference Tsutsumi, Miyoshi, Yukawa and Kato2011), another is a general, critical revision of currently used methods (see also Baskin and Baskin, Reference Baskin and Baskin2001).
For epiphytes, the majority of germination studies have been performed with members of a single family, the Bromeliaceae, and consequently our study also focuses on this group. We touch a number of issues, from the collection of samples and seed storage, to the appropriate design of experiments dealing with the influence of water supply and temperature on germination in epiphytic bromeliads.
Many authors give detailed and unambiguous information on the sampling procedure of the seeds used in experiments (e.g. Fernandez et al., Reference Fernandez, Beltramo and Caldiz1989; Montes-Recinas et al., Reference Montes-Recinas, Marquez-Guzman and Orozco-Segovia2012), in other studies there is room for interpretation (e.g. ‘collected from natural populations’, Bader et al., Reference Bader, Menke and Zotz2009; or ‘mature fruit’, Pereira et al., Reference Pereira, Andrade, Pereira, Forzza and Rodrigues2009). Since bromeliad capsules may look ‘mature’ weeks or even months before natural dehiscence, differences in germination success could easily stem partly, or entirely, from varying levels of maturity and not from difference among populations or treatment effects. For the proper interpretation of past and future studies, information on the potential of afterripening would be essential. To this end, we investigated whether collection time (from immediately after anthesis to close to dehiscence) affects the outcome of germination experiments in three species.
For soil-rooted terrestrial plants, there has been long-standing interest in the response of the germination process to different water potentials (Ψ), both in agriculture and in basic plant science, because germination is an important bottleneck for plant recruitment in natural systems (Evans and Etherington, Reference Evans and Etherington1990). A few studies with epiphytic bromeliads have also investigated this relationship (Pereira et al., Reference Pereira, Andrade, Pereira, Forzza and Rodrigues2009), but it can be debated whether such experiments allow relevant conclusions for the situation in nature. In contrast to soils, which provide a relatively constant environment for germinating seeds over longer periods, conditions in the epiphytic habitat are extremely volatile, in particular in the case of bark epiphytes: their seeds are either wetted by rain or fog (Ψ ≈ 0 MPa) or, with rapid transition, exposed to very dry conditions even in moist rainforest conditions. For example, a relative air humidity (rh) of 99% at 20°C already corresponds to a Ψ of − 1.4 MPa, while rh of 90% represents a ten times lower Ψ (Nobel, Reference Nobel2005). For this reason, some researchers (e.g. Bader et al., Reference Bader, Menke and Zotz2009) have begun to study the influence of water on the germination process in vascular epiphytes, not by varying Ψ, but by alternating periods of drought and wetness. Although this approach seems much closer to the real world with its irregular rainfall, the experimental details have not been analysed systematically.
Conceivably, the outcome of such studies may depend strongly on the relative lengths of wet and dry periods, or on a possible interaction between the moisture treatment and other factors, such as the temperature used. Consequently, we analysed how variation in the duration of dry/wet cycles affected the results of a germination experiment in four bromeliad species. In addition, for four other species we compared the outcome of an experiment with intermittent drought (Bader et al., Reference Bader, Menke and Zotz2009) with the germination response to varying Ψ of the solutions in which the seeds were immersed.
The use of constant versus fluctuating temperatures in experiments is a long-standing issue in germination biology ((Baskin and Baskin, Reference Baskin and Baskin2001). Clearly, there are plant groups where fluctuating temperatures are required for germination, or are at least strongly stimulating, e.g. in many species of wetlands and flood plains (e.g. Mollard and Insausti, Reference Mollard and Insausti2011; Carta et al., Reference Carta, Bedini, Müller and Probert2013). In most studies with epiphytes, constant temperatures have been used, but this approach has apparently rarely been based on a critical evaluation of the effect of constant versus fluctuating temperatures on germination in this group. A few researchers did apply fluctuating temperatures, but only two studies compared the outcome of such a treatment with germination under the constant, mean temperature. The results were inconsistent. While germination of the two bromeliads studied by Pinheiro and Borghetti (Reference Pinheiro and Borghetti2003) was slightly lower and slower under fluctuating conditions, germination in four other bromeliads was significantly enhanced by varying temperatures (Pereira et al., Reference Pereira, Andrade, Pereira, Forzza and Rodrigues2009). Such inconsistent results are hard to interpret in an ecological context, because the maximum and minimum temperatures are rarely based on relevant measurements of thermal fluctuations in the field (but see, for example, Pinheiro and Borghetti, Reference Pinheiro and Borghetti2003). We compared germination under both conditions for a range of species. In addition, we studied the interactive effect of varying temperature and water supply on germination with another three species.
To conclude, laboratory experiments are simplifications and cannot capture the complexity of the natural situation, while field work (e.g. Cascante-Marín et al., Reference Cascante-Marín, Wolf, Oostermeijer and Den Nijs2008) will hardly allow the unambiguous identification and quantification of the effect of individual factors, e.g. temperature, on germination. This contribution will hopefully assist in ‘bridging the gap’, by improving the quality of the results from controlled, ex situ experiments to help understand the complexities in situ for studies with vascular epiphytes.
Materials and methods
All experiments were conducted with material from epiphytic Bromeliaceae collected from natural populations in Panama (Table 1), which were brought to the Plant Functional Ecology Laboratory at the university in Oldenburg, Germany, where the experiments were carried out. Seeds were sown in disposable Petri dishes (100 × 15 mm) with filter paper (Machery-Nagel, 651 mm, Ø 83 mm), which were kept in climate cabinets (Economic Delux, Snijders Scientific, Tilburg, The Netherlands) with a light/dark period of 12/12 h (photon flux density c. 60 μmol m− 2s− 1). If not mentioned otherwise, temperature was set to 25°C. Before starting an experiment all seed comas were clipped off and the seeds sterilized, following Pickens et al. (Reference Pickens, Affolter, Wetzstein and Wolf2003). Coma removal can slow germination compared to intact seeds, as shown for Catopsis sessiliflora (Wester and Zotz, Reference Wester and Zotz2011), but the final proportion of germinated seeds seems to be unaffected. In the experiments without special water treatments seeds were watered every other day to ensure continuously moist conditions. Germination (defined as breakage of the testa by the protruding, swollen hypocotyl) was recorded daily with a dissecting microscope (Zeiss 57 50 57, Jena, Germany). Before starting experiments with intermittent water supply it was necessary to determine the amount of water evaporating from Petri dishes with and without a lid. To this end, we added different amounts of water to the filter paper in open Petri dishes and weighed the dishes at 30-min intervals for 2 h. The same was done with closed Petri dishes, but at 6-h intervals over 24 h. This procedure was repeated in all chambers at all the temperatures used in the subsequent experiments. The results of these trials allowed us to vary the absolute amount of water applied to each Petri dish at the beginning of each wet period, so that the length of the wet and dry periods were as long as planned irrespective of temperature and chamber.
Study species used in the five experiments (E1–E5). Collecting sites are lowland sites with wet (San Lorenzo), moist (BCI, Barro Colorado Island) and dry tropical vegetation (Azuero) and a lower montane site (Fortuna). Plant names follow The Plant List (2013)
Experiment 1: Seed maturity and germination response
One capsule per plant (Table 1, four individuals for each of three species) was collected at five different dates (November 2010–March 2011). The first collection was shortly after flowering (in October), the last immediately before natural dehiscence (in April). Seeds enclosed in the capsules were allowed to dry at room temperature (c. 22°C) and were kept in paper bags until the experiments were started a few weeks after the last collection date. We have shown recently that seeds of Tillandsioideae remain viable for at least 1 year under such conditions (Reference ZotzZotz, 2013a). For each species two replicates of every collecting date per individual with 20 seeds each were sown on filter paper in a Petri dish wetted with 2 ml of distilled water. The Petri dishes containing the seeds were sealed with Parafilm® in order to reduce evaporation and the germination success was monitored starting 3 d after sowing. Germination was followed for 19 d.
Experiment 2: Germination response to oscillating and constant temperatures
Germination with a fluctuating temperature regime (15–25°C) was compared with germination at the mean constant temperature of 20°C. In the case of Guzmania monostachia, Tillandsia fasciculata and Vriesea sanguinolenta, three replicates of 20 seeds were used; in the other nine species (Table 1) one batch of 20 seeds each were used. Germination was followed for 22 d.
Experiment 3: Germination response to wet–dry cycles of varying frequency
Seeds of four species (Table 1) were subjected to four different water treatments. Wet and dry periods made up consistently 50% of total time, but frequency of changes varied with cycles of 12/12 h, 24/24 h, 36/36 h and 48/48 h (wet/dry). A control determined germination under continuous moisture. Each treatment was repeated in five Petri dishes containing 25 seeds each, except for Tillandsia flexuosa for which three replicates of 18 seeds were used, related to a scarcity of seeds of this species. The experiment was terminated when seeds in each of the treatments had experienced a total of 216 h ( = 9 d) of wet conditions, which is enough for all controls to reach 100% germination.
Experiment 4: Germination response to varying water potentials
Germination in seven solutions with varying osmotic potential was studied over 35 d with four species (Table 1). Solutions with osmotic potentials of − 0.10, − 0.25, − 0.50, − 1.0, − 1.50 and − 2.0 MPa were produced by varying the concentrations of mannitol; a control used distilled water (0 MPa). Treatments and control were replicated three times with 20 seeds each per species. Using the same seed batches, we also studied germination with different moisture regimes following the protocol described in Bader et al. (Reference Bader, Menke and Zotz2009):
-
D0: continuous moisture in closed Petri dishes, adding 1 ml distilled water every other day to compensate for evaporation;
-
D1: mild repeated drought with 2 h dryness per day;
-
D2: moderate repeated drought with 6 h dryness per day;
-
D3: severe repeated drought with 22 h dryness in 2 d.
Each treatment was replicated three times with 20 seeds each. Germination was followed for 35 d.
Experiment 5: Simultaneous variation in the water and temperature regime and germination
Seeds of four different species (five Petri dishes with 20 seeds each, Table 1) were subjected to three different water treatments (D0, D1 and D2 at four different temperatures (15°C, 20°C, 25°C and 32.5°C). Consistent durations of dry periods were achieved by applying different amounts of water at the beginning of each wet period and removal of the lid about 2 h before the planned beginning of the subsequent dry period. Germination was followed for 35 d.
Data analysis
For statistical analysis we used R 3.0.3. (R Development Core Team, 2014). There is a range of response variables that can be used to analyse germination experiments (Scott et al., Reference Scott, Jones and Williams1984). In most cases we used the germination index (GI), which is defined as:
$$\begin{eqnarray} \end{eqnarray}$$
where T i is the number of days after sowing, N i is the number of seeds germinated on day i, and S is the total number of seeds planted. This index reflects both the speed of germination and the final germination success. Usually, we standardized GI by setting the maximum observed value in each experiment and species to unity, which facilitates comparison across experiments. Since final germination percentages under favourable conditions were close to 100% in individual seed batches in all experiments, we interpreted lower germination success as indicative of a treatment effect and not as an indicator of non-viable material, although we did not specifically test for viability after termination of an experiment. The responses of germination to varying wet/dry periods were analysed with one-way analyses of variance (ANOVAs) for each species separately, using ‘lm’. Within-group differences were explored with Tukey's honest significance difference (HSD) tests. The effects of the combined drought and temperature treatments were studied with a three-way ANOVA. The response to constant and fluctuating temperature was analysed with a t-test.
Results
Experiment 1: Seed maturity and germination response
With the exception of the first collection date, germination success was invariably close to 100% (Fig. 1), and we found no significant effect of collection date on final germination percentage for any of the three species (one-way ANOVA, P>0.05). Seeds from capsules collected shortly after anthesis in November 2010 did not germinate in the case of G. monostachia and V. sanguinolenta. Remarkably, those of T. fasciculata showed about 50% germination, although capsules were still greenish when collected – about 5 months before natural dehiscence and seed dispersal.
Cumulative germination of seeds collected at different stages of maturity in three bromeliad species. Data points are means of four individuals. Numbers (1–5) indicate the different collection dates: 1, immediately after anthesis (4 November 2010); 2, c. 2 months after anthesis (8 January 2011); 3, c. 3 months after anthesis (2 February 2011); 4, c. 4 months after anthesis (23 February 2011); 5, c. 5 months after anthesis (17 March 2011). For the species Vriesea sanguinolenta, data for the second collecting time are missing because all seed batches were attacked by a fungus.
Experiment 2: Germination response to oscillating and constant temperatures
The 12 tested bromeliad species showed no significant difference in germination response under constant and fluctuating temperature (Table 2, t-test, P= 0.42 for final germination percentage and P= 0.17 for GI). In three of the species, these trials were replicated three times, and again no treatment-related differences were found [G. monostachia (t-test, P= 0.08), T. fasciculata (t-test, P= 0.51) and V. sanguinolenta (t-test, P= 0.63)].
Germination success under constant and fluctuating temperature regimes in 12 epiphytic bromeliads (Experiment 2). Results are final germination in % after 22 d or the germination index. Sample size was 1 or 3 times 20 seeds, in the latter case the data are averages
Experiment 3: Germination response to wet–dry cycles of varying frequency
Germination was invariably fastest in the controls and slowest in the treatment with the most frequent dry periods (12 h/12 h), with intermediate responses of the other drought-treated samples (see supplementary Figure S1). A different picture emerged when relating the GI to the periods of moisture, i.e. hydrotime (Fig. 2). Although the 12 h/12 h treatment continued to impose a significant effect on germination (ANOVAs, HSD tests P< 0.05; with one exception, Guzmania monostachia), the effects of the other treatments were basically indistiguishable statistically, both between each other and compared to the control (ANOVAs, HSD tests P>0.05).
Standardized germination indices as a function of different wet–dry periods in four bromeliad species (left panels) and a control (constantly wet, right panels). Data are means ± SD; sample size was 5 × 25 seeds, except for Tillandsia flexuosa with 3 × 18 seeds. Significant differences (ANOVA, HSD, P< 0.05) are indicated by different letters. Detailed germination kinetics are shown in supplementary Figure S1.
Experiment 4: Germination response to varying water potentials
The GI followed a logistic relationship with varying water potential (Fig. 3). Remarkably, two of the four species still germinated, if at low percentage, at water potentials < − 1.5 MPa. Aligning these response curves with the GIs of the four drought treatments (D0-D3) yielded an inconsistent quantitative pattern. For example, the germination response under the driest wet/dry treatment in Vriesea gladioliflora was comparable to that with a water potential of − 1.7 MPa, while the effect of the same treatment on Vriesea viridiflora was equivalent to that of an osmoticum of only − 0.8 MPa.
Germination responses to different water potentials in four epiphytic bromeliads. Open symbols are the mean germination indices at different water potentials (0 to -2.0 MPa), lines show logistic regressions fitted to these data, and arrows indicate the mean germination indices of the four different treatments with intermittent drought (D0–D3). Detailed germination kinetics are shown in supplementary Figure S2.
Experiment 5: Simultaneous variation in the water and temperature regime and germination
Temperature, water regime and species affected germination (Table 3, Fig. 4). Germination success of all four studied lowland species varied consistently with temperature – it increased from 15°C to 25°C with no further increase, or even a decrease, at 32.5°C. Much less consistent was the response to intermittent drought. A negative effect of longer drought on germination was strongest at low temperatures ( − 63% to − 100%), intermediate at 20 and 25°C ( − 24% to − 81%), while at 32.5°C a reduction was only found in V. sanguinolenta ( − 80%, Fig. 4).
Results of a three-way ANOVA on the effects of water, temperature and species on the standardized germination index for the species Guzmania monostachia, Tillandsia fasciculata, Tillandsia flexuosa and Vriesea sanguinolenta
Germination indices in response to different water and temperature treatments for four epiphytic bromeliads. Water treatments: continuous moisture, 4-h dry period and 8-h dry period. Temperatures: 15°C, 20°C, 25°C and 32.5°C. Data are means ± SE. Detailed germination kinetics are shown in supplentary Figure S3.
Discussion
Only a decade ago Baskin and Baskin (Reference Baskin and Baskin2001) concluded in a general review that ‘not much is known about the germination ecophysiology of epiphyte seeds’. This situation has improved considerably by now, with a number of recent publications on the subject (e.g. Cota-Sanchez and Abreu, Reference Cota-Sanchez and Abreu2007; Toledo-Aceves and Wolf, Reference Toledo-Aceves and Wolf2008; Bader et al., Reference Bader, Menke and Zotz2009; Goode and Allen, Reference Goode and Allen2009; Manzano and Briones, Reference Manzano and Briones2010; Valencia-Díaz et al., Reference Valencia-Díaz, Flores-Palacios, Rodríguez-López, Ventura-Zapata and Jiménez-Aparicio2010; Montes-Recinas et al., Reference Montes-Recinas, Marquez-Guzman and Orozco-Segovia2012; Sosa-Luría et al., Reference Sosa-Luría, Chávez-Servia, Mondragón-Chaparro, Estrada-Gómez and Ramírez-Vallejo2012). Overall, however, our understanding of this aspect of epiphyte ecology is still rather sketchy and highly biased – the majority of studies have focused on bromeliads. This contribution addresses methodological issues in an effort to ensure ecologically meaningful germination studies with bromeliads and epiphytes in general.
A first consideration in any germination study is to be sure about the level of maturity of the material used (Baskin and Baskin, Reference Baskin and Baskin2001). Unless seeds are collected from naturally opened fruit, ‘maturity’ is not unambiguously defined. In our experience, fruit may look ‘mature’ long before natural opening. Information on afterripening in epiphytes is scarce (Fernandez et al., Reference Fernandez, Beltramo and Caldiz1989, Schwallier et al., Reference Schwallier, Bhoopalan and Blackman2011). In Tillandsia recurvata, immature seeds germinated at a relatively high percentage, but lost viability much faster than mature seeds (Fernandez et al., Reference Fernandez, Beltramo and Caldiz1989). The advantage of collecting capsules before dehiscence are obvious: (1) during collecting trips material from species of different phenologies and conspecifics of varying developmental state can be collected together; and (2) as discussed by Ruiz et al. (Reference Ruiz, Laguna, Iglesias, Damon, Marin, Azpiroz and Moreno2008) seeds inside unopened capsules are in sterile conditions, which makes sterilization procedures unnecessary. The results of our germination trials with immature seeds (Fig. 1) suggest that it is safe to collect material before natural dehiscence and obtain meaningful results, although it is certainly preferable to collect mature seeds whenever possible (Baskin and Baskin, Reference Baskin and Baskin2001). The resulting question why these epiphytes postpone apparently possible dispersal for months may have a simple answer. Dispersal in the late wet season or early dry season would obviously offer no advantage for establishment, while seeds are clearly better protected in capsules until released immediately before the wet season. If this logic is true, afterripening should be less pronounced or absent in species with natural dehiscence during the wet season.
In nature, fluctuating rather than constant temperatures are the rule. Although some authorities therefore advocate the invariable use of alternating temperature regimes in germination studies (Baskin and Baskin, Reference Baskin and Baskin2001), it is far from clear that arbitrarily set temperature variation yields results that allow better, i.e. less ambiguous, interpretation in an ecological context. Undoubtedly, experimental designs informed by data of field conditions are preferable, but in the absence of such data we see no reason to promote fluctuating temperatures in studies with epiphytic bromeliads. This statement is based on the inconsistent results in published studies and our own results. For instance, some studies have shown that bromeliad species have elevated germination percentages at alternating temperatures (Pereira et al., Reference Pereira, Andrade, Pereira, Forzza and Rodrigues2009), while others showed the opposite pattern (Pinheiro and Borghetti, Reference Pinheiro and Borghetti2003). The bromeliad Pitcairnia albiflos even failed to germinate in an alternating temperature regime (Pereira et al., Reference Pereira, Andrade, Pereira, Forzza and Rodrigues2010). Our results with 12 species of epiphytic bromeliads (Table 2) are in line with these previous observations. There was no consistent difference between germination behaviour in constant and fluctuating temperature conditions.
Two recent papers have used wet/dry cycles to study the effect of intermittent water supply on germination in epiphytes (Bader et al., Reference Bader, Menke and Zotz2009, Wagner et al., Reference Wagner, Bogusch and Zotz2013). The ecological realism of this new approach is certainly much higher than the use of different osmotica, but considering the frequent use of the latter in germination studies (Baskin and Baskin, Reference Baskin and Baskin2001) a comparison of the effects of both approaches is desirable. Although there was a consistent, and unsurprising, reduction in germination response with increasing drought (Fig. 3), either produced by shorter periods of wetness or lower water potential, a quantitative comparison revealed substantial interspecific variation in that relationship: different osmotica are clearly not a simple proxy for the impact of intermittent droughts of differing severity.
There are many other conceivable variations of wet/dry cycles in nature. In another experiment, we explored how the length of wet and dry periods affected gemination, while keeping the total duration of wet and dry periods constant, with 216 h each in c. 18 d. Compared to the controls, the delay in germination was inversely related to the length of the wet/dry cycles (see supplementary Figure S1). However, calculating the GI for periods of hydration only (i.e. for hydrotime; Black et al., Reference Black, Bewley and Halmer2006), revealed very similar kinetics in most treatments (Fig. 2). Apparently, the germination process is only temporarily arrested during dry periods, and continues immediately after remoistening. A substantial reduction in the GI by the most rapid fluctuations of wet and dry periods (12 h/12 h) was only observed in one species, V. sanguinolenta. This is remarkable, because this tank bromeliad tends to occupy more exposed, and hence drier, microsites than two of the other studied species (G. monostachia and T. fasciculata; Zotz, Reference Zotz1997), which were hardly affected.
The response to intermittent water supply depends on temperature as well (Fig. 4). Although the expected reduction in GI with increasing duration of intermittent drought was found at most temperatures, GI did not show this trend in three of four species at the highest temperature. Moreover, the proportional reduction varied substantially among lower temperatures. Hence, results of such experiments may be quantitatively, and even qualitatively, misleading when experimental conditions are not based on the typical situation in the field. Unfortunately, we are largely ignorant of the temperatures that germinating seeds of epiphytic plants experience in nature. Records from climate stations may be used as a first approximation, but conditions, particularly on exposed branches, are likely to deviate considerably from such standardized measurements.
To conclude, we present the results of a series of germination experiments with epiphytic bromeliads. We show that it is safe to collect capsules for experiments before natural dehiscence. The application of fluctuating temperatures is not imperative. The effects of different water potentials and intermittent drought on germination are not quantitatively comparable among species, and we advocate the use of the latter because of the greater ecological realism. However, the impact of intermittent drought on germination depends on temperature. Hence, data on in situ temperatures during germination are needed to design experiments in such a way as to ascertain unambiguous interpretation of the results in an ecological context.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0960258514000312.
Acknowledgements
We thank the Republic of Panama for making its natural resources available for science. Many thanks to Wiebke Bogusch and Lars Vormstein (both at Oldenburg) for help with the experiments.
Financial support
This work is part of the doctoral thesis of the first author, who is supported by a fellowship from FUN DAYACUCHO (Fundación Gran Mariscal de Ayacucho, La Urbina, Venezuela) and DAAD (Deutscher Akademischer Austauschdienst, Bonn, Germany).
Conflicts of interest
None.
