Desiccation-induced loss of seed viability is associated with a 10-fold increase in CO2 evolution in seeds of the rare tropical rainforest tree Idiospermum australiense


  • P. J. Franks,

    Corresponding author
    1. School of Tropical Biology, James Cook University, Cairns, Queensland, Australia; Present address: Organismic and Evolutionary Biology, Harvard University, 3119 Biological Laboratories, 16 Divinity Avenue, Cambridge MA 02138, USA
      Author for correspondence: Peter Franks Tel: +61 07 40421237 Fax: +61 07 40421284 Email:
    Search for more papers by this author
  • P. L. Drake

    1. School of Tropical Biology, James Cook University, Cairns, Queensland, Australia; Present address: Organismic and Evolutionary Biology, Harvard University, 3119 Biological Laboratories, 16 Divinity Avenue, Cambridge MA 02138, USA
    Search for more papers by this author

Author for correspondence: Peter Franks Tel: +61 07 40421237 Fax: +61 07 40421284 Email:


  • • Here the relationship was investigated between metabolic activity, state of hydration and seed viability in the desiccation-intolerant (recalcitrant) seeds of Idiospermum australiense, a rare and primitive angiosperm tree restricted to wet tropical forest.
  • • Seed CO2 evolution rate, R, was monitored in fully hydrated (control) seeds and seeds that were allowed to desiccate under ambient conditions over a period of c. 90 d.
  • • During desiccation R increased dramatically toward a peak at a seed relative water content of 39 ± 3% (relative to maximum water content, which corresponded to 0.45 ± 0.03 g water g-1 d. wt) followed by a decline toward zero with total desiccation. This peak constituted a 10-fold increase in mean R, relative to the control. Exposing seeds to O2-free air at this peak induced a further large, but transient, increase in CO2 evolution, indicating that the peak developed in the presence of oxidative phosphorylation, rather than due to the absence of it.
  • • The magnitude and mode of the observed increase in CO2 evolution in response to desiccation is unlike any reported so far and thus adds new information about metabolic changes that may occur as the water content of desiccation-intolerant seeds declines.


Humid tropical forests have many species that produce seeds that are intolerant of desiccation (Farnsworth, 2000). Although these forests usually experience very high rainfall, there are occasional dry periods, driven by the El Niño Southern Oscillation (Wright et al., 1999), during which seeds intolerant of desiccation may be lost from the recruitment cycle. Through this direct effect on seed survival, dry periods may alter seedling recruitment patterns in humid tropical forests. Over geological time, the evolution of seed desiccation tolerance is likely to have influenced the structure and composition of forests, yet little is known about the origins or physiological basis of this important adaptive trait. Although it is possible that seed desiccation tolerance has evolved independently several times, desiccation intolerance is thought to be the ancestral state (von Teichman & van Wyk, 1994; Pammenter & Berjak, 2000). In order to understand the evolutionary patterns of seed desiccation tolerance, as well as the longer-term ecological consequences of drought in humid tropical forests, much more needs to be known about the physiology of desiccation-intolerant seeds.

Desiccation-intolerant seeds are often referred to as ‘recalcitrant’ in agricultural terminology because they cannot survive storage for extended periods of time (Roberts, 1973). They are known to lack the ability to undergo crucial metabolic adjustments that ensure recovery from desiccation (Farrant et al., 1989; Vertucci & Farrant, 1995; Farrant et al., 1997; Berjak & Pammenter, 1997), yet much remains unknown about the metabolic changes that do occur while desiccation-intolerant seeds succumb to desiccation. In particular, and despite the fact that the majority of species with desiccation-intolerant seeds occur in humid tropical forests (Farnsworth, 2000; Pammenter & Berjak, 2000), almost nothing is known of the metabolic response to desiccation in seeds of species restricted to this habitat. Here we examine CO2 evolution rate, a basic indicator of metabolic activity, during the desiccation of seeds of the rare, wet tropical forest primitive angiosperm tree, Idiospermum australiense.

In their natural environments, desiccation-intolerant seeds are potentially vulnerable to changed environmental moisture regimes. At the time of shedding, these seeds have high moisture contents and are metabolically active (Farrant et al., 1985; Salmen Espindola et al., 1994; Lin & Chen, 1995; Vertucci & Farrant, 1995; Farrant et al., 1997; Leprince et al., 1999; Kozeko & Troyan, 2000). In favourable moisture conditions, germination usually occurs soon after shedding (Farnsworth, 2000). However, under abnormally dry environmental conditions, even mild desiccation can reduce the likelihood of germination upon rehydration. Although the level of dehydration at which viability is lost varies considerably within and between species (Artlip et al., 1995; Farrant et al., 1997; Han et al., 1997; Normah et al., 1997; Pammenter & Berjak, 1999; Sun & Liang, 2001), all desiccation-intolerant seeds have a critically low moisture content from which they cannot recover. Acquisition of desiccation tolerance is known to be linked to the action of ABA and other plant hormones, as well as the synthesis of dehydrin proteins (Farrant et al., 1993; Bewley & Black, 1994; Finch-Savage et al., 1994; Baker et al., 1995; Farrant et al., 1996; Farnsworth, 2000). In desiccation-intolerant seeds, the ultimate cause of desiccation-induced loss of viability is irreversible damage to cellular membranes (Hendry et al., 1992; Finch-Savage et al., 1994; 1996; Leprince et al., 1999; Walters et al., 2001). This damage is thought to be due to increased generation of reactive oxygen species (ROS) and an inadequately protective antioxidant system (Senaratna et al., 1987; Hendry et al., 1992). However, the metabolic changes that contribute to this damage are poorly understood.

A characteristic of desiccation-intolerant seeds is their sustained metabolic activity. Indeed, it is thought that their persistently high metabolic activity and inability to down-regulate this activity in a balanced manner contributes significantly to desiccation damage (Vertucci & Farrant, 1995; Berjak & Pammenter, 1997; Leprince et al., 2000) However, no clear trend has been identified for the progression of these metabolic changes during desiccation, and no clear link has been established between certain metabolic characteristics and the degree of desiccation sensitivity. Furthermore, almost no studies have been conducted on desiccation-induced metabolic changes in desiccation-intolerant seeds of very wet tropical forest tree species.

Idiospermum australianse is a tropical rainforest subcanopy tree that occurs naturally in only a few known pockets of very-high-rainfall lowland tropical forest in north-eastern Australia (Briggs & Leigh, 1996). It is part of a group of extant flowering plant families that are said to have retained anatomical features considered to be ancient (Lamont, 1974; Wilson, 1979). Given these characteristics, we sought to examine the relationship between desiccation-induced changes in metabolic rate and loss of seed viability in this species, with a view to understanding more about the range of metabolic characteristics of recalcitrance in angiosperm seeds.

Materials and Methods

Plant material

Idiospermum australiense (Diels) ST Blake, a sub-canopy rainforest tree, is found in two isolated pockets of very moist lowland tropical rainforest in north-east Queensland, Australia (Cooper & Cooper, 1994). This geographical distribution virtually confines the species within the regional 3750 mm mean annual rainfall isohyet (Turton et al. 1999). Idiospermum australiense seeds are large, broadly spheroidal (about 8 cm in diameter), and are exceptional in that they typically comprise three or four (up to five) cotyledons. The seeds lack endosperm, and the large cotyledons comprise most of the seed mass. Due to the fact that this species is uncommon and produces relatively few seeds, our experiments were designed with this constraint in mind.

Seed storage

Fresh I. australiense seeds were collected from their natural population at what was observed to be close to a fully hydrated state, that is immediately after seed fall. This was important because it was desirable that they had no prior exposure to desiccation. Immediately after collection seeds were washed in distilled water and suspended above a water reservoir in sealed containers according to Farrant et al. (1989). Additionally, air was pumped into the basal water layer using an aquarium aerator. This provided the seeds with ambient oxygen concentration to carry out respiration and aided in the humidification of the chamber. A small hole in the lid of the container served as an outlet for gas. Alternative storage methods such as dry storage or chilling were not used because these conditions are known to induce tissue damage in desiccation-intolerant seeds (Mycock & Berjak, 1995). F. wt was monitored to ensure a stable state of hydration. Seeds were assumed to be fully hydrated when fresh weight stabilised. All seeds in the experiments outlined below (both humidified and desiccated) were stored in brown paper bags, which prevented germination, at a temperature of 23 ± 2°C.

Defining seed moisture content

The state of seed hydration, or seed relative water content RWC, is expressed here as a percentage of maximum total seed water content:

image(Eqn 1)

where f. wt = actual seed weight at any time, d. wt = oven dry weight (48 h at 105°C) and f. wtmax= weight at full hydration. Thus, RWC varies from 100% at full hydration to 0% at full desiccation. Mean f. wtmax (g) and d. wt (g) for I. australiense seeds were, respectively, 93.9 ± 6.7 and 43.4 ± 3.22 (n = 10). Seed RWC is usually expressed as grams water per gram seed d. wt (gwater/gd.wt). This is often a practical necessity because seeds in a humid atmosphere will often keep imbibing water and germinate without reaching a steady f. wtmax. However, the advantage of expressing RWC as a percentage of maximum water content rather than gwater : gd.wt is that the range is always from 100% to 0% whereas for measurements that express water content relative to d. wt, the range varies from an arbitrary percentage to 0%. Based on the values quoted above, RWC (% of maximum water content) for I. australiense seeds is easily converted to gwater/gd.wt by multiplying by 0.0116.

Measuring seed CO2 evolution rate

Rate of CO2 evolution, R, was quantified as nmol CO2 g-1 seed d. wt s-1. The experiments utilised an open-flow infrared gas analyser (IRGA) gas exchange system (model: Li-cor 6400, Li-cor Inc, Lincoln, NB, USA) for measurements of CO2 and H2O vapour concentrations. The seed sample chamber, placed in series with the gas sample line of the gas exchange system, comprised an airtight stainless steel container housing a stirring fan and a type-E thermocouple. After sealing a seed inside the sample chamber, R was measured at an ambient (chamber) CO2 concentration (ca) of 350 µmol CO2 mol−1 air and 70% relative humidity (rh). The final expression of R on a per gram dry weight basis was only possible after determining seed d. wt. Seed temperature was not measured directly, but was assumed to be in equilibrium with room and chamber air temperature, which were controlled at 23 ± 2°C throughout the course of the study.

Experiment 1: the effect of desiccation on seed CO2 evolution rate

Idiospermum australiense seeds were divided into two groups. Group 1 (treatment, n = 4) were allowed to desiccate slowly under ambient laboratory conditions (23 ± 2°C, 70 ± 5% RH, and ambient O2, assumed to be 21%). Group 2 (controls, n = 5) were stored as described above at 23 ± 2°C, 100% RH and ambient O2. CO2 evolution rates and f. wt of both groups of seeds were measured at regular intervals (1–2 d) over c. 90 d. When rate of CO2 evolution had declined to almost zero in seeds of Group 1, all seeds (Group 1 and Group 2) were oven-dried for 48 h at 105°C to determine d. wt, thus allowing back-calculation of RWC (%) and R (nmol CO2 g−1 d. wt s−1).

Experiment 2: the effect of O2-free air on CO2 evolution rate in fully hydrated and partially desiccated seeds

The results from experiment 1 (see below) showed a dramatic peak in CO2 evolution rate when RWC reached approximately 40%. Experiment 2 was designed to test whether this peak developed in the presence of oxidative phosphorylation. One group of seeds was kept at full hydration (Group 3, n = 3) and another group (Group 4, n = 4) were allowed to desiccate down to RWC = 40%. To determine when seeds from Group 4 had reached RWC = 40%, a value for d. wt had to be inferred so that Eqn 1 could be used to calculate RWC. Therefore, d. wt was estimated from the measured linear relationship between f. wtmax and d. wt of a sub-sample of seeds (n = 10), yielding the equation d. wt = 0.428 × f. wtmax.

The effect of O2-free air on CO2 evolution rate in the seeds from Group 3 (fully hydrated) and Group 4 (RWC c. 40%) was measured as follows: The seed was placed into the measurement chamber and R was allowed to reach steady state at a chamber CO2 concentration of 350 µmol CO2 mol−1 air, 70% RH and ambient O2 concentration. After measurement of a steady CO2 evolution rate over several minutes, humidified O2-free air (0.1% CO2, 99.9% N2) was then supplied to the Licor 6400 console air inlet port at a rate of 0.5–1 l min-1 with all other environmental variables kept constant. R was monitored until it reached a new steady state under these O2-free conditions, after which chamber air was returned to ambient O2 concentration and R allowed to reached steady state again. Over approximately the same time interval a control group (Group 5, n = 3) was kept fully hydrated at ambient O2 concentration and their CO2 evolution rate monitored.

Experiment 3: seed germination trial

Trials were conducted on I. australiense seeds in three states of hydration: first fresh, fully hydrated (RWC = 100%, n = 40); second moderately desiccated (RWC = 40%, n = 15); and third severely desiccated (RWC = 10%, n = 15). Seeds forming the two partially desiccated treatments were allowed to desiccate under ambient laboratory conditions at 23 ± 2°C, 70 ± 5% RH. RWC for the moderately and severely desiccated treatments was determined as described in the methods for Experiment 2, using Eqn 1 and an inferred d. wt.

After the desired level of desiccation had been reached, seeds were immediately placed half buried in 10 cm pots, containing a standard potting mix (40 : 30 : 30 organic matter: coarse sand: sandy loam) in a glasshouse (day/night temperature 30/25°C), and watered three times daily to encourage germination. The group of nondesiccated seeds, serving as the control, was germinated in the same manner. Germination was identified as the first appearance of the young shoot.


As RWC decreased over the 90-d period (Fig. 1) the CO2 evolution rate of all I. australiense seeds first increased to a maximum and then declined toward zero (Fig. 2). The maximum rate of CO2 evolution occurred at RWC = 39 ± 3% (0.45 ± 0.03 gwater/gd. wt). The mean maximum R was 1.38 ± 0.25 se. nmol CO2 g−1 d. wt s−1 in desiccated seeds, representing a 10-fold increase over controls, where mean R in control seeds at the time corresponding to the peaks in desiccated seeds was 0.130 ± 0.004 nmol CO2 g−1 d. wt s−1 (Fig. 3). This difference between mean R in control and mean maximum R in desiccated seeds was statistically significant (one way anova, F = 24.5, P = 0.002). Plotting R against the time course of desiccation (Fig. 3) enables a comparison between control and treated (desiccated) seeds for the entire duration of the desiccation treatment. This shows that, compared with the widely varying CO2 evolution rates of the desiccating seeds, the mean respiration rate of the control seeds (Group 2) remained relatively constant over the experimental period.

Figure 1.

Time course of desiccation for Idiospeermum australiense seeds in Group 1, which were allowed to desiccate under ambient laboratory conditions (23 ± 2°C, 70 ± 5% RH). (a) Seed f. wt and (b) relative water content compared with days of desiccation. Symbols for each individual seed trace correspond to those in Figs 2 and 3.

Figure 2.

Seed CO2 evolution rate, R, compared with seed relative water content (RWC) for Idiospeermum australiense seeds from Group 1. Desiccation was allowed to proceed over c. 90 d under ambient laboratory conditions (23 ± 2°C, 70 ± 5% RH). Each panel traces the change in R for an individual seed as RWC declined. Data and symbols correspond to the seeds in Fig. 1. Measurements of R were made at a CO2 concentration of 350 µmol mol−1, 21% O2 concentration, 70% RH and 23 ± 2°C to simulate the ambient laboratory conditions.

Figure 3.

Time course of seed CO2 evolution rate as seeds of Idiospeermum australiense from Group 1 desiccated. Each panel traces the change in R for an individual Idiospeermum australiense seed as desiccation proceeded over c. 90 d under ambient laboratory conditions (23 ± 2°C, 70 ± 5% RH). For comparison, the time course of R (mean ± s.e., denoted by the symbol Δ) for control Group 2, kept fully hydrated, is shown in each panel. Data and symbols correspond to the seeds in Figs 1 and 2. Conditions for measurement of R as for Fig. 2.

Treatment of fully hydrated seeds (Group 3) with O2-free air resulted in a steady state rise in the CO2 evolution rate (Fig. 4). When O2 concentration was returned to ambient, the CO2 evolution rate of these seeds gradually returned almost exactly to the value measured before the perturbation. Over the same period, seeds that were maintained at an ambient O2 concentration showed stable CO2 evolution rates (Fig. 4d).

Figure 4.

The effect of O2-free air on the CO2 evolution rate (R) of fresh, fully hydrated Idiospeermum australiense seeds (Group 3). Panels (a), (b) and (c) are traces for three individual seeds. R for control seeds (Group 5), which were maintained at ambient O2 and fully hydrated over a similar time period, are shown in panel (d). Mean R for controls = 0.41 ± 0.01 nmol CO2 g−1 d. wt s−1. Measurements of R were made at a CO2 concentration of 350 µmol mol−1, ambient O2 or O2-free atmosphere as labelled, relative humidity 70 ± 5% and 23 ± 2°C.

For the partially desiccated seeds in Group 4, O2-free air was applied at RWC = 41.47 ± 3.90%, which was consistent with the point of maximum CO2 evolution rate observed in Experiment 1, as illustrated in Fig. 2. Following exposure to O2-free air the CO2 evolution rate of these seeds showed an initial rise followed by a slow decline back to a rate similar or only slightly below that before treatment with O2-free air (Fig. 5). Thus treatment of these partially desiccated seeds with O2-free air had only a transient effect on the elevated CO2 evolution rate. A return to ambient O2 concentration had negligible effect on CO2 evolution rate (Fig. 5), except in one case where it rose to slightly above the initial (pre-O2-free) rate (Fig. 5d).

Figure 5.

The effect of O2-free air on rate of CO2 evolution for partially desiccated (RWC = 41.47 ± 3.90%) Idiospeermum australiense seeds. Panels (a), (b), (c) and (d) are traces for four individual seeds. Measurements of R were made at a CO2 concentration of 350 µmol mol−1, ambient O2 or O2-free atmosphere as labelled, relative humidity 70 ± 5% and 23 ± 2°C.

In the seed germination trial, seeds not subject to desiccation exhibited 95% germination. This fell to 46% for seeds that were allowed to reach RWC = 47–60%, while those allowed to desiccate down to RWC = 4–12% did not germinate at all.


The rate of CO2 evolution (R) of desiccating I. australiense seeds followed a defined pattern for all desiccated seeds, with a distinct peak at RWC = 39 ± 3% (0.45 ± 0.03 gwater/gd.wt). This peak constituted a 10-fold increase in mean R relative to controls. Such a massive increase in CO2 evolution rate as a result of desiccation has not been previously reported. Following this peak, further desiccation led to almost complete cessation of CO2 evolution at RWC c. 10% (Fig. 2). The relatively constant CO2 evolution rate of the control seeds when R was plotted against time confirms that the dramatic rise in CO2 evolution rate was caused by desiccation rather than maturation or other age-related processes.

Changes in respiration rate during desiccation in desiccation-intolerant seeds have been attributed to ‘uncontrolled metabolism’ or a metabolic imbalance (Vertucci & Farrant, 1995; Leprince et al., 1999; Pammenter & Berjak, 1999; Walters et al., 2001). Metabolic imbalance appears to begin early in the desiccation process, with the toxic products of this imbalance eventually leading to irreparable damage to cellular membranes (Salmen Espindola et al., 1994; Leprince et al., 1999; Buitink et al., 2000). Rate of CO2 evolution and/or O2 uptake can be a useful indicator of the course of these metabolic changes during desiccation. However, there are only a limited number of studies that have examined seed gas exchange in this context. Of these, only the study by Leprince et al. (1999) with sweet chestnut (Castanea sativa) shows evidence of a rise in CO2 evolution rate during desiccation, although this was only moderate compared with our observations with I. australiense.

The respiration process, comprising the three key elements glycolysis, the Krebs cycle and the electron transport chain, is regulated by a series of enzyme-inhibition negative feedback loops. Although desiccation eventually disrupts and disables all of these elements, available evidence suggests that the onset of desiccation may affect them differentially at different stages of desiccation. For instance, in a detailed study involving the measurement of both O2 and CO2 fluxes over the course of desiccation, Leprince et al. (1999) showed that, although O2 uptake rates changed during the course of cotyledon desiccation, respiratory quotient (CO2 output/O2 input) remained constant until loss of membrane integrity. In a subsequent study, Leprince et al. (2000) showed that metabolic imbalance preceded the loss of membrane integrity during dehydration of germinating radicles. Our data do not reveal when loss of membrane integrity occurred, but the results predict that by the time CO2 evolution rate had reached its maximum during the course of desiccation, viability in approximately one in two seeds was lost. Further work is necessary to determine whether the pronounced increase in CO2 evolution rate is associated with a protective metabolic response, or merely represents a breakdown in balanced metabolic control, with toxic side-effects.

The experiments with O2-free air (Figs 4 and 5) reveal important information about the interplay between glycolysis and oxidative phosphorylation during the desiccation of these seeds. The Pasteur effect, initially described by Louis Pasteur as the oxygen inhibition of glycolysis during fermentation, results from the tight regulation of glycolysis and fermentation in the cytosol, in concert with oxidative phosphorylation in mitochondria. When O2 levels are high, increased ATP and citrate production from mitochondrial respiration inhibit phosphofructokinase, an allosteric enzyme that catalyses a regulatory reaction in the glycolytic sequence (Passonneau & Lowry, 1962; Krebs, 1972). Because oxidative phosphorylation results in considerably more ATP per glucose consumed than fermentation, its inactivation (e.g. by anoxia or mitochondrial damage) will require glycolysis to proceed at a much higher rate to maintain ATP supply, thus increasing CO2 evolution rate. This effect is demonstrated in Fig. 4, where fully hydrated seeds were exposed to an O2-free atmosphere. Due to the potentially toxic side-effects of this induced acceleration of glycolysis, organisms tolerant of hypoxia or anoxia respond by down-regulating metabolic activity. Increased cytoplasmic viscosity as a result of desiccation has been shown to inhibit O2 diffusion (Leprince & Hoekstra, 1998). If the rise in CO2 evolution rate observed in seeds of I. australiense (Figs 2 and 3) was due primarily to desiccation-induced hypoxia and removal of the Pasteur effect, then exposing desiccated seeds to an anoxic atmosphere should have had little effect on CO2 evolution rate. Instead, seeds responded with a large, albeit transient, increase in CO2 evolution rate, before settling with a steady state CO2 evolution rate only slightly below the original elevated rate (Fig. 5). This suggests that up until the time of application of O2-free air, a Pasteur effect was operating to suppress the already elevated CO2 evolution rate, indicating the maintenance of mitochondrial integrity, at least up until that point. This is supportive of the proposal by Leprince et al. (2000) that metabolic imbalances precede, rather than accompany loss of membrane integrity during the drying of desiccation-intolerant seeds. However, the reason why the Pasteur-like response in Fig. 5 was not sustained is unclear. Further work is required to determine whether it was due to substrate limitation, enzyme inactivation or reduced energetic demand.

Further work is necessary to determine the cause of the extraordinarily large increase in CO2 evolution rate during desiccation of I. australiense seeds, and to determine its role in the loss of seed viability. It is unclear as to why the peak in CO2 evolution rate was so sharp and occurred consistently at RWC = 39 ± 3%, but it could be a reliable indicator of the onset of irreversible cellular damage in this species. Given that the increase in CO2 evolution rate could be only partially simulated in fully hydrated seeds by depriving them of oxygen (Fig. 4), it cannot be fully attributed to removal of the Pasteur effect as a result of possible inactivation of oxidative phosphorylation. Insight into the role of oxidative phosphorylation in this unusual mode of response could be gained by simultaneous measurement of both CO2 evolution and O2 uptake, as recently employed by Leprince et al. (1999). Such studies will lead to a better understanding of the metabolic and other physiological characteristics that result in seed recalcitrance, and improve our knowledge of the adaptations that have led to desiccation tolerance.


We thank Professor David Day for helpful comments. We are also grateful to Daryl Graham and Marina Gurtzis for kind assistance with seed collection.