A new type of specialized morphophysiological dormancy and seed storage behaviour in Hydatellaceae, an early-divergent angiosperm family

Abstract

Background and Aims

Recent phylogenetic analysis has placed the aquatic family Hydatellaceae as an early-divergent angiosperm. Understanding seed dormancy, germination and desiccation tolerance of Hydatellaceae will facilitate ex situ conservation and advance hypotheses regarding angiosperm evolution.

Methods

Seed germination experiments were completed on three species of south-west Australian Hydatellaceae, Trithuria austinensis, T. bibracteata and T. submersa, to test the effects of temperature, light, germination stimulant and storage. Seeds were sectioned to examine embryo growth during germination in T. austinensis and T. submersa.

Key Results

Some embryo growth and cell division in T. austinensis and T. submersa occurred prior to the emergence of an undifferentiated embryo from the seed coat (‘germination’). Embryo differentiation occurred later, following further growth and a 3- to 4-fold increase in the number of cells. The time taken to achieve 50 % of maximum germination for seeds on water agar was 50, 35 and 37 d for T. austinensis, T bibracteata and T. submersa, respectively.

Conclusions

Seeds of Hydatellaceae have a new kind of specialized morphophysiological dormancy in which neither root nor shoot differentiates until after the embryo emerges from the seed coat. Seed biology is discussed in relation to early angiosperm evolution, together with ex situ conservation of this phylogenetically significant group.

INTRODUCTION

Hydatellaceae, a small family of aquatic plants, has recently undergone taxonomic realignment from the Poales to the waterlily clade, Nymphaeales, near the base of the angiosperm phylogenetic tree (Saarela et al., 2007). This repositioning has not only significantly improved our understanding of early angiosperm evolution but has also increased the conservation value of the group. Hydatellaceae consist of a single genus, Trithuria, containing 12 species (Sokoloff et al., 2008a). Ten of the species occur in Australia, with four endemic to the Southwest Australian Floristic Region (Hopper and Gioia, 2004). South-west Australian species are annuals, occurring in temporary wetlands, most often in clay-based vernal pools, but also in rock-pools on granite (locally known as gnammas) and on stream banks (Western Australian Herbarium, 1998–).

Research into the structural morphology, developmental processes and behavioural characteristics of Hydatellaceae and other early-divergent [ANA-grade (Amborella, Nymphaeales, Austrobaileyales)] species can help to clarify aspects of early angiosperm evolution (Friis and Crane, 2007; Rudall et al., 2007, 2008, 2009; Saarela et al., 2007; Friedman, 2008; Sokoloff et al., 2008a, b). For example, the seedling morphology of four Hydatellaceae species resembles that of both Nymphaeaceae (with united cotyledons) and some monocots with a single cotyledon (e.g. Schizocarpa and Heliconia), suggesting that Hydatellaceae could represent a possible evolutionary pathway to a monocot-like embryo (Sokoloff et al., 2008b).

Embryos of ANA-grade angiosperms are diverse, ranging from small linear (longer than broad; Trimeniaceae, Schisandra), rudimentary (as broad as long; Amborellaceae, Illiciaceae, Schisandraceae and Austrobaileyaceae) or broad (Cabombaceae and Nymphaeaceae) (Martin, 1946; Bailey and Swamy, 1948; Morat and MacKee, 1977; Endress, 1980; Saunders, 1998; Baskin and Baskin, 2007). Species with small linear, rudimentary or spatulate embryos that must grow inside the seed and take ≤30 d or >30 d to germinate (i.e. radicle emergence) have morphological dormancy (MD) or morphophysiological dormancy (MPD), respectively (Baskin and Baskin, 2004). In MD seeds with undifferentiated embryos, germination does not take place until both differentiation and growth have occurred (Baskin and Baskin, 2004). In two highly specialized kinds of MPD the embryos never differentiate into a radicle or cotyledons, even after germination; the embryo either grows to form a protocorm, which requires an association with a mycorrhizal fungus for further development (i.e. orchids), or the embryo forms a haustorium that connects to the host (e.g. Orobanche spp.; Baskin and Baskin, 1998). Embryos in Hydatellaceae are undifferentiated at seed maturity (dispersal) and, at least in T. submersa, increase in size prior to germination (Hamann, 1976; Cooke, 1983; Hamann, 1998; Rudall et al., 2009), implying that the seeds are morphologically dormant. However, qualitative data on embryo growth are lacking and seed dormancy type has not been determined. Seed germination studies on other temporary pool aquatic species growing sympatrically with Hydatellaceae in south-west Australia have shown that some species are non-dormant and germinate to high percentages in the light at temperatures between 5 and 10 °C (Tuckett et al., 2010). The seed germination response to temperature and light in Hydatellaceae is unknown and critical to inform conservation practitioners for ex situ conservation and propagation.

Hydatellaceae seed studies may also contribute to resolving hypotheses as to whether seed desiccation tolerance (‘orthodox’ storage behaviour) is a derived or ancestral state for angiosperms (Pammenter and Berjak, 2000; Dickie and Pritchard, 2002; Tweddle et al., 2003; Berjak and Pammenter, 2008). The hypothesis that desiccation tolerance is an ancestral state for angiosperms is based upon observations that species with desiccation-sensitive seeds are usually those that have putatively ancestral ovule character states (von Teichman and van Wyk, 1994). This hypothesis has been further developed to suggest that desiccation-sensitive seeds are the ancestral state and that desiccation tolerance has been derived multiple times in early angiosperm evolution (Pammenter and Berjak, 2000). The alternative hypothesis that desiccation tolerance is the ancestral state for angiosperms is based on a study of 886 tree and shrub species, in which the simplest explanation for the phylogenetic occurrence of desiccation tolerance/sensitivity is that desiccation tolerance is the ancestral state (Tweddle et al., 2003). Relatively few extant spermatophytes have desiccation-sensitive seeds; those that do are scattered throughout the phylogenetic tree. As desiccation tolerance is a complex character state thought to involve many genes and molecular processes, it has been argued that this condition would be much easier to become ‘lost’ rather than ‘gained’ several times in the course of angiosperm evolution (Dickie and Pritchard, 2002; Tweddle et al., 2003).

The seasonally inundated wetlands where the south-west Australian Hydatellaceae occur may be particularly vulnerable to the potential threats faced by other aquatic habitats, including climate change, eutrophication, pollution, and algal and vascular weeds (Hay et al., 2000; Withers, 2000; Grillas et al., 2004; Horwitz et al., 2008). In addition to threats at an ecosystem level, increasing demand for Hydatellaceae specimens and seeds may put pressure on natural populations. Conservation and management strategies are therefore required to ensure that the genetic diversity of populations is maintained. A comprehensive understanding of the environmental conditions required for germination and the identification of seed storage behaviour are two of the key areas essential in establishing and sustaining ex situ conservation programmes. This study aims to test the following hypotheses, as they relate to Hydatellaceae: (a) seeds have MD at maturity (seed dispersal); (b) seeds will not have physiological dormancy (PD) and will germinate to high percentages at low temperatures in the light; and (c) if seed desiccation tolerance is taken as the ancestral state for angiosperms, Hydatellaceae seeds will tolerate drying and therefore be amenable to conventional (–18 °C) seed bank storage as a strategy for the ex situ conservation of these species.

MATERIALS AND METHODS

Seed collections

At the time of dispersal, mature seeds of Trithuria austinensis were collected in November/December 2007 at Tolkerlup Swamp (S34°19′20·9″ E116°43′23·2″) and those of T. bibracteata and T. submersa at the same time at Mersa Road Swamp (S34°07′27·8″ E116°12′124·4″) in the Warren region of Western Australia. Voucher specimens were deposited at PERTH. Both sites are temporary wetlands that are inundated during winter (July–August) and dry during the summer (December–February). Seeds were transported to the laboratory at Kings Park and Botanic Garden in ziplock plastic bags, where they were cleaned by hand under a microscope. Cleaned seeds were stored in sealed glass phials at room temperature (approx. 18 °C) until they were used. Seed viability was assessed using a cut test on a random sample of four replicates of 25 seeds. A seed was scored as viable if it had a turgid, white embryo and perisperm (ISTA, 1999).

Germination of fresh seeds

Experiments on fresh seeds commenced within 28 d of collection. Seeds were sterilized in 2 % (w/v) calcium hypochlorite [Ca(OCl)₂] for 10 min and then rinsed three times in sterile deionized water. A complete three-way factorial design was devised to test seeds for treatment effects and presence of dormancy. Each treatment consisted of three replicates of 25 seeds, except for T. bibractata where each replicate consisted of 15 seeds, because seed material was more limited. Factors investigated included light, temperature and germination stimulant. Seeds were sown in 90 mm diameter sterile plastic Petri dishes containing 25 mL of water agar (0·7 %, w/v), 25 mL of water agar with 0·29 mm gibberellic acid (GA₃; Sigma, Australia), or 25 mL of water agar with 0·67 µm karrikinolide (KAR₁; provided by G. Flematti, University of Western Australia; Flematti et al., 2005). Petri dishes were sealed with plastic film (Glad^® Wrap, Australia) and, for dark treatments, immediately wrapped in aluminium foil and placed in light-proof cardboard boxes for the duration of the experiment. Seeds exposed to light were left unwrapped and not placed in boxes. Dishes were then placed in incubators set at 5, 10, 15 or 20 °C with a daily 12 h photoperiod of 30 µmol m⁻² s⁻¹, 400–700 nm, cool-white fluorescent light. As seeds of T. bibracteata were limited, they were not incubated at 5 °C or treated with KAR₁. Scoring of light treatments occurred weekly until no further germination was observed for 3 weeks. Scoring of dark-treated seeds only occurred on the final day of scoring of the light-treated replicates (after 140 d).

Seed desiccation and storage

To test whether the seeds tolerate desiccation and sub-zero storage and whether there was a change in the dormancy breaking and/or germination requirements, the same three-way factorial experiment (as above) was carried out on dried and stored seeds (GA₃ and KAR₁ data not presented for dried and stored treatments). Seeds were dried in open containers in a controlled environment room [15 °C, 15 % relative humidity (RH)] for 14 d. ‘Dried’ seeds were then either tested for germination immediately or were hermetically sealed in a single laminated aluminium foil bag and placed at –18 °C for 28 d (‘stored’ seeds). Germination tests were carried out on dried and stored seeds (foil bags allowed to equilibrate with room temperature before opening), as described above with three replicates of 25 seeds per treatment, except for T. bibracteata, where we used three replicates of 15 seeds.

Embryo growth

To investigate whether the embryo grows prior to germination, i.e. the presence of MD or MPD, seeds of T. austinensis and T. submersa were sectioned at regular intervals prior to and during germination. Dried seeds were placed on water agar as described above and incubated in the light at 10 °C. Samples of T. austinensis seeds were taken for sectioning at 0, 5, 10, 15, 20 and 25 d, and those of T. submersa seeds were taken at 0, 10, 20 and 30 d. Seeds were placed in concentrated hydrochloric acid (HCl) for 2 h before fixing in formalin–acetic–alcohol (40 % formaldehyde, glacial acetic acid and 70 % ethanol, 10 : 5 : 85, v/v/v). Seeds with a ruptured seed coat were not treated with HCl. Fixed seeds were sent to the Royal Botanic Gardens, Kew, where five seeds from each sample time were sectioned, stained and viewed under light microscopy following Rudall et al. (2009). Material was embedded in Histo-Technovit 7100 resin and sectioned using a Leica RM 2155 rotary microtome fitted with a tungsten–carbide knife. Sections were stained in toluidine blue and mounted in DPX resin (a mixture of distyrene, plasticizer and xylene). Optical sections were photographed using a Leitz Diaplan photomicroscope fitted with a Zeiss Axiocam digital camera, in some cases using differential interference contrast microscopy (Rudall et al., 2009).

Statistical analyses

Seed germination data were analysed using logistic regression in GenStat 10th edition (VSN International Ltd, UK). For each experiment, a full model including all main factors and interactions was fitted. Graphs show exact binomial standard errors (95 % confidence interval). Images of sectioned seeds were analysed using ImageJ for Windows 1·41. Images from each sample time were calibrated and measured for total area, length and width for both seeds and embryos. In addition, the number of cells was counted along the midline section of each embryo for each section. As cells were quantified along the midline section, some cells of the embryo were missed due to the 3-D nature of the embryo. However, as the 3-D shapes of the embryos between species are generally similar, some comparisons can be made between species.

RESULTS

Germination of fresh seeds

Cut tests indicated high seed viability for all three species: 95, 82 and 91 % for T. austinensis, T. bibracteata and T. submersa, respectively. Initially, germination of fresh T. austinensis seeds was low. Germination was lower in the dark; maximum germination occurred at 10 °C in light, reaching 28 % in water and 37 % in the presence of GA₃ (Fig. 1A, D). It took 50 d for 50 % of the total number of germinants to germinate at 10 °C, and maximum germination occurred on day 119.

Fig. 1.

Effect of gibberellic acid agar (GA₃) and karrikinolide agar (KAR₁) on seed germination (mean ± 95 % CI) for T. austinensis (A, D), T. bibracteata (B, E) and T. submersa (C, F). Seeds were incubated under a 12 h light / dark regime (A–C) or full darkness (D–F). The y-axis shows percentage germination after 119, 85 and 88 d for T. austinensis, T. bibracteata and T. submersa, respectively.

Seed germination of T. bibracteata occurred at 15 °C in the light. Treating seeds with GA₃ increased germination from 29 % (on water) to 36 % (Fig. 1B). There was no germination when seeds were incubated in the dark (Fig. 1E). Time to 50 % germination at 15 °C was 35 d, and maximum germination was reached on day 85.

Germination of fresh T. submersa seeds on water was highest at 5 °C in the light (75 %), and when seeds were treated with KAR₁ this increased to 90 % (Fig. 1C). When data were pooled across incubation temperatures, GA₃ and KAR₁ increased germination at sub-optimal temperatures (10, 15 and 20 °C). Similar to T. bibracteata, germination in the dark was negligible (Fig. 1F). Seeds of T. submersa took 37 d for 50 % of the total number of germinants to germinate. Maximum germination was reached between 77 and 88 d, at 10 °C.

Seed desiccation and storage

Seeds of T. austinensis, T. bibracteata and T. submersa germinated to equal or greater percentages once fresh seeds were dried at 15 °C and 15 % RH for 14 d and stored at –18 °C for 28 d (Fig. 2).

Fig. 2.

Seed germination (mean ± 95 % CI) of fresh (28 d old), dried and stored seeds of T. austinensis, T. bibracteata and T. submersa. Seeds were incubated at the indicated temperatures under a 12 h light/dark regime for a maximum of 140 d.

The maximum germination percentage of T. austinensis seeds was significantly greater than that of fresh seeds following drying (70 %; 15 °C, 15 % RH for 14 d) Germination further increased following drying and storage (79 %; as above followed by 28 d at –18 °C; Fig. 2A). Seeds of T. bibracteata germinated to the highest percentage after drying and storage, and incubation at 10 °C in the light. However, maximum germination was only 48 % and, in contrast to fresh seeds, GA₃ did not increase germination further (Fig. 2B, Table 1). Maximum germination of T. submersa seeds increased from 75 to 90 % following drying and storage (Fig. 2C) and occurred under the same incubation treatments as with fresh seeds.

Table 1.

Results of logistic regression analysis (significance of terms) of the germination data for T. austinensis (n = 75), T. bibracteata (n = 45) and T. submersa (n = 75)

	Light (L)	Temperature (Te)	Treatment (Tr)	L × Te	L × Tr	Te × Tr	L × Te × Tr
Trithuria austinensis	<0·001	<0·001	<0·001	<0·001	<0·001	<0·001	<0·001
Trithuria bibracteata	<0·001	<0·001	0·027	<0·001	<0·001	<0·001	<0·001
Trithuria submersa	<0·001	<0·001	NS	<0·001	<0·001	<0·001	<0·001

Seeds were placed at a range of temperatures (5, 10, 15 and 20 °C), with or without addition of GA₃ or KAR₁ (T. bibracteata was not incubated at 5 °C or treated with KAR₁) and incubated in darkness or under a 12 h light/dark regime.

Embryo growth

Prior to incubation, embryos were undifferentiated, positioned at one end of the seed and surrounded by 1–2 cell layers of endosperm (Figs 3A, B and 4A). The embryo : seed (E : S) ratio (area) of T. austinensis of dry seeds (i.e. before placing them on agar) was 0·021 (Table 2). This ratio increased to 0·054 following 25 d incubation at 10 °C, more than doubling in embryo size. The embryo also underwent cell division during incubation, with the number of cells (in longitudinal sections) increasing from 23 (±s.e. 1·5) prior to incubation to 67 (±s.e. 9·0) following 25 d at 10 °C (Table 2, Fig. 3). Embryos of T. submersa seeds also increased in size during incubation at 10 °C. The E : S ratio (area) increased from 0·019 to 0·080 after 25 d, a 4-fold increase (Table 2, Fig. 4). During incubation, the mean number of cells in the embryo increased from 18 (±s.e. 0·9) to 79 (±s.e. 6·8).

Fig. 3.

Longitudinal sections of dried and stored seeds of T. austinensis. Undifferentiated embryos grew inside the seed until the operculum was ruptured, after which most growth occurred outside the seed coat. (A and B) Resting seed showing limited growth. (C) Embryo growth following 15 d of incubation at 15 °C. (D–F) Seeds after 20 d incubation. Abbreviations: e, endosperm; em, embryo; o, operculum; p, perisperm; ttl, thickened tegument layer. Scale bars: (A, E) = 100 µm; (B–D, F) = 50 µm.

Fig. 4.

Longitudinal sections of post-dispersal seeds of T. submersa. Most embryo growth occurred outside the seed coat, with a 4-fold increase in cell number before differentiation into a radicle and primary photosynthetic leaf. Images of seeds are rotated to the horizontal for ease of comparison of embryo growth and differentiation; seedling growth in nature is vertical. (A) Resting seed. (B–F) Seeds showing germination following incubation for 20–30 d. Abbreviations: e, endosperm; em, embryo; fp, first photosynthetic leaf; p, perisperm; r, primary root. Scale bars = 200 µm.

Table 2.

Seed morphological characters of resting (before incubation) and germinated seeds of T. austinensis and T. submersa

	Trithuria austinensis		Trithuria submersa
	Resting seed	Germinating seed	Resting seed	Germinating seed
Seed length (μm)	498 ± 24·7	485 ± 7·5	690 ± 32·9	763 ± 27·1
Seed width (μm)	335 ± 17·4	320 ± 13·9	284 ± 15·5	286 ± 22·8
Embryo length (μm)	43 ± 10·1	66 ± 4·4	42 ± 1·7	108 ± 6·6
Embryo width (μm)	83 ± 3·9	86 ± 10·1	90 ± 5·9	156 ± 10·1
E : S (length)	0·086	0·136	0·131	0·204
Seed area (1000 µm2)	126 ± 7·9	116 ± 6·5	141 ± 11·4	154 ± 13·3
Embryo area (1000 µm2)	3 ± 0·4	6 ± 1·1	3 ± 0·2	12 ± 1·8
E : S (area)	0·021	0·054	0·019	0·080
No. of cells in embryo section	23 ± 1·5	67 ± 9·0	18 ± 0·9	79 ± 6·8

Data are presented on seed and embryo sizes of resting seeds and seeds just prior to differentiation of the embryo for T. austinensis and T. submersa (mean ± s.e.).

Upon incubation, the embryo began to undergo cell division (Fig. 3A–C) and emerged through the thickened tegmen layer and the operculum. Seeds were defined as germinated at this point. The majority of embryo growth occurred outside of the seed coat (Figs 3D, E and 4B). Following seed coat rupture, the embryo continued to undergo cell division, with cell numbers increasing 3- to 4-fold (Figs 3F and 4C). No differentiation was evident in any of the sections prior to day 30 in T. submersa seeds, or prior to day 25 in T. austinensis seeds (Figs 3F and 4C). However, in both species rapid differentiation of cells into the radicle occurred following emergence from the seed coat (Fig. 4D, E) followed closely by differentiation of the first primary leaf (Fig. 4E, F).

Temperatures at which maximum germination occurred were 5, 10 and 15 °C for T. submersa, T. austinensis and T. bibracteata, respectively, and all species required light for germination or germinated better in light than darkness. The time taken to achieve 50 % of maximum germination for seeds on water agar was variable between species, but all seeds took ≥35 d. Seeds of all species were desiccation tolerant, and drying of seeds increased maximum germination percentage. Some embryo growth and cell division in T. austinensis and T. submersa occurred prior to the emergence of an undifferentiated embryo from the seed coat (‘germination’). Embryo differentiation occurred later, following further growth and a 3- to 4-fold increase in the number of cells.

DISCUSSION

A new type of specialized morphophysiological dormancy in Trithuria

During incubation and prior to germination, embryos of T. austinensis and T. submersa underwent cell division, with the number of cells (as visible in a section under light microscopy) increasing from approx. 23 to 67 in T. austinensis and from approx. 18 to 79 cells in T. submersa, representing 2·9- and 4·4-fold increases in embryo size, respectively. Since (a) embryo growth within the seed is a requirement for germination, (b) seeds took >30 d to germinate and (c) germination increased following the drying treatment, we conclude that seeds of Hydatellaceae can be described as having MPD as defined by Baskin and Baskin (2004). Differentiation of the embryo into a radicle and shoot occurs after the embryo breaks through the seed coat, i.e. after the seed germinates (Figs 3 and 4). This combination of events in seed germination has not been reported previously and thus represents a new kind of specialized MPD.

Although there was some embryo growth inside the seeds of Trithuria prior to germination (embryo emergence), differentiation of the root and shoot occurred outside the seed coat. This pattern of germination differs from that of seeds of all other species known to have undifferentiated embryos that eventually give rise to normal embryos with a radicle and cotyledon(s). For example, undifferentiated embryos in seeds of the eudicots Anemone nemorosa and A. ranunculoides (Mondoni et al., 2008) and Corydalis solida and C. ledebouriana (Liden and Staaf, 1995) grow and differentiate inside the seed before emergence of a differentiated radicle. The only other taxa with specialized forms of MD or MPD are some mycoheterotrophic plants such as orchids and some holoparasitic plants such as Orobanche. In these taxa, the embryos have no (or very little) growth prior to germination and never differentiate into a shoot and root (Baskin and Baskin, 1998). In Hydatellaceae, most differentiation of the embryo into a radicle and a primary photosynthetic leaf occurs after the seed coat has ruptured and the embryo is largely outside the seed. Differentiation of the meristem and other ‘organs’ in holoparasites occurs exogenously during germination, rather than before (i.e. within the seed), as does the development of the radicle and primary leaf in Trithuria. However, formation of a seedling in Trithuria is based on its own reserves, whereas both other known forms of specialized MD occur in species that are heterotrophic and require a host or fungus to develop into a seedling (Baskin and Baskin, 1998).

Is the undifferentiated embryo of Trithuria an ancestral state in angiosperms?

Hydatellaceae and Nymphaea lotus (Nymphaeaceae) (Rudall et al., 2009) are the only ANA-grade angiosperms reported to have undifferentiated embryos (Martin, 1946; Bailey and Swamy, 1948; Morat and MacKee, 1977; Endress, 1980; Saunders, 1998). Hydatellaceae share several other developmental characteristics with the waterlilies, such as a four-nucleate embryo sac, operculate seeds with perisperm and haustorial endosperm (Saarela et al., 2007; Rudall et al., 2009). At least one of these features, the four-nucleate embryo sac, is putatively ancestral in angiosperms (Friedman and Williams, 2003, 2004; Williams and Friedman, 2004). Thus, the phylogenetic placement of Hydatellaceae raises the question of whether the undifferentiated embryo state could also be ancestral in angiosperms, or whether it is merely consistent with the reduced habit and short life cycle characteristic of this family (Hamann, 1976; Dahlgren et al., 1985; Rudall et al., 2007).

We propose that the undifferentiated embryo is an ancestral state in angiosperms rather than an ecologically driven specialization based on several observations. First, it is unlikely that exogenous embryo growth is evolutionarily derived within crown clade Hydatellaceae, as it is a high risk strategy for germination. The lack of a protective seed coat during much of the embryo growth and differentiation leaves the embryo exposed and very vulnerable. Presumably this makes Trithuria more susceptible to desiccation than in typical seeds (with a full protective seed coat) that undergo embryo growth and/or differentiation prior to germination. Thus, Hydatellaceae would be unlikely to survive in a terrestrial ecosystem where desiccation is probable. Secondly, no other sympatric species of the temporary wetland environment of these Hydatellaceae have yet been identified to have an undifferentiated embryo (R. E. Tuckett, unpubl. res.). Furthermore, Trithuria species apparently have not developed other germination strategies that are common in many south-western Australian species, such as fire responsiveness and a subsequent strong response to KAR₁ (Chiwocha et al., 2009; Dixon et al., 2009). Unlike many dryland terrestrial species, if Hydatellaceae have remained in an aquatic habitat they are likely to have retained the ancestral state due to the lack of ecological drivers whereby pre-emergence differentiation of the embryo may be advantageous in aquatic habitats. Phylogenetic niche conservatism of Hydatellaceae is consistent with the greatest diversity and the global distribution of the family on old, climatically buffered and largely infertile landscapes (Hopper, 2009). Traits such as the presence of a chalazal endosperm haustorium and perisperm as the major nourishing tissue for the embryo could represent the ancestral state in Hydatellaceae and other early-divergent angiosperms (Hamann, 1998; Floyd and Friedman, 2001; Rudall et al., 2007, 2009). However the hypothesis that the undifferentiated embryo represents the ancestral state for the angiosperms requires further testing using phylogenetic optimization and more comparative data on early-divergent and stem-group angiosperms and their sister groups, both extant and extinct.

Evolution of desiccation tolerance

The observation that Hydatellaceae seeds are desiccation tolerant could also mean that this trait is ancestral for all angiosperms, a question that has not been resolved, not least because of the inability to assess desiccation tolerance in fossilized seeds (Pammenter and Berjak, 2000; Berjak and Pammenter, 2008). If, as seems likely, Hydatellaceae plants have remained in similar environments since their evolutionary origin, this trait may have been retained since the crown clade Hydatellaceae began to diversify, rather than being subsequently derived. However, further rigorous testing of this hypothesis is required, because some species of the sister group Nymphaeaceae (i.e. Nuphar lutea and Nymphaea alba) show desiccation sensitivity (Smits et al., 1989; Hay et al., 2000). It would be informative to test whether seeds of the other Trithuria spp. (some of which complete their entire life cycle submerged) are desiccation tolerant, to determine whether this is a universal characteristic of the family, particularly for the several species found in northern Australia and the single species in India. If seeds of all other Hydatellaceae species are also desiccation tolerant, and have inhabited similar environments since they evolved, this would suggest that desiccation tolerance may well be an ancestral state for the family. Admittedly, the ephemeral habitat of the taxa studied is highly specialized (Main, 1997) and desiccation tolerance of the seeds is an inevitable adaptation to the vagaries of this variable habitat. The question also arises as to what extent adaptation to an ephemeral rock pool habitat is associated with a particular syndrome of seed character states. Seeds of other temporary wetland aquatics of south-western Australia also have been shown to be desiccation tolerant and to share some of the germination characteristics displayed in Hydatellaceae (Tuckett et al., 2010).

Germination requirements of Trithuria

Germination of T. submersa was highest at 5–10 °C, which is considerably lower than the average optimal temperature (24 °C) for aquatic species (Baskin and Baskin, 1998). However, optimal temperatures for germination of some sympatric species occurring in the same environment as Hydatellaceae are also in the range of 5–10 °C (Tuckett et al., 2010). Final germination percentages were similar at 5 and 10 °C, but the germination rate was faster at 10 °C. Temperature profiles from a field site where Hydatellaceae are present suggest that, in situ, seeds are likely to experience sufficient time periods for germination when temperatures average around 10 °C, but not at 5 °C (R. E. Tuckett, unpubl. res.). World climate change models predict that in southern Western Australia temperature will increase and rainfall decrease (Gitay et al., 2002; Hilbert et al., 2007). Final germination percentages decreased at temperatures >10 °C, so we hypothesize that an increase in average winter temperature could lead to a reduction in the percentage of seeds germinating each year if the population is unable to adapt rapidly to a changing climate or to migrate to more suitable habitats.

Seed germination responses to temperature and light were considerably different between species. Seeds of T. submersa have higher germination at low temperatures (5–10 °C), whereas seeds of T. austinensis germinate to high percentages at all tested temperatures, although 10–20 °C was more favourable than 5 °C. Even though the geographical distributions of these species largely overlap (Sokoloff et al., 2008a) and both are restricted to temporary wetlands, they are rarely found growing in the same pool. It is possible that there are small differences in environmental variables between different sites that could account for this variation. These may be revealed with more detailed information on the phenology of species and of pool regimes. We have observed that T. submersa is often found in, but not restricted to, sites that are typically shallower than those of T. austinensis. Trithuria austinensis often grows on the edges of larger temporary pools where annual fluctuation of water depth is greater than that in the typical smaller pools of T. submersa sites. The greater depth at T. austinensis sites could result in lower light availability, and hence the observation that some seeds can germinate in the dark.

Conservation strategies for Trithuria

Hydatellaceae species are of significant conservation value and occur in a vulnerable environment that is potentially threatened by climate change, eutrophication, pollution, algal and vascular weeds, soil disturbance, in-filling and drainage, agriculture, tourism, feral animals and conversion to permanent water bodies (Hay et al., 2000; Withers, 2000; Grillas et al., 2004; Horwitz et al., 2008). If Hydatellaceae have persisted in south-western Australia since their evolutionary origin, then they have sustained several changes in climate and may have considerable evolutionary resilience (James, 1992, 2000; Hopper, 2009). However, the adaptability of present-day Hydatellaceae species to increased average temperatures and a reduced hydroperiod is currently unknown. Dispersal capabilities, pollination and genetic differences within populations are not fully understood. Considering the evolutionary significance of this group, ex situ conservation is necessary to provide an insurance policy for natural populations. The species of Hydatellaceae tested here are amenable to seed banking conditions and are orthodox in storage behaviour (Liu et al., 2008), indicating that ex situ storage is a viable option for their conservation.

We have shown that compared with other early-branching angiosperms and south-western Australian flora, Hydatellaceae have a unique combination of characteristics in the process of early plant establishment. Identifying the germination requirements, orthodox seed storage behaviour and new type of specialized MPD in this family is important to implement ex situ conservation strategies and contribute to current evolutionary hypotheses of the early-divergent angiosperms.