Germination Ecophysiology of Annona crassiflora Seeds

Abstract

Background and Aims

Little is known about environmental factors that break morphophysiological dormancy in seeds of the Annonaceae and the mechanisms involved. The aim of this study was to characterize the morphological and physiological components of dormancy of Annona crassiflora, a tree species native to the Cerrado of Brazil, in an ecophysiological context.

Methods

Morphological and biochemical characteristics of both embryo and endosperm were monitored during dormancy break and germination at field conditions. Seeds were buried in the field and exhumed monthly for 2 years. Germination, embryo length and endosperm digestion, with endo-β-mannanase activity as a marker, were measured in exhumed seeds, and scanning electron microscopy was used to detect cell division. The effect of constant low and high temperatures and exogenous gibberellins on dormancy break and germination was also tested under laboratory conditions.

Key Results

After burial in April, A. crassiflora seeds lost their physiological dormancy in the winter months with lowest monthly average minimum temperatures (May–August) prior to the first rainfall of the wet season. The loss of physiological dormancy enabled initiation of embryo growth within the seed during the first 2 months of the rainy season (September–October), resulting in a germination peak in November. Embryo growth occurred mainly through cell expansion but some dividing cells were also observed. Endosperm digestion started at the micropylar side around the embryo and diffused to the rest of the endosperm. Exogenous gibberellins induced both embryo growth and endo-β-mannanase activity in dormant seeds.

Conclusions

The physiological dormancy component is broken by low temperature and/or temperature fluctuations preceding the rainy season. Subsequent embryo growth and digestion of the endosperm are both likely to be controlled by gibberellins synthesized during the breaking of physiological dormancy. Radicle protrusion thus occurred at the beginning of the rainy season, thereby maximizing the opportunity for seedlings to emerge and establish.

INTRODUCTION

The Brazilian Cerrado resembles the African savannah vegetation, covering nearly 2 million km² of central Brazil, representing 22 % of the Brazilian territory (Ratter et al., 1997). In the State of Minas Gerais the Cerrado covers an area of 30·8 million ha, which is 53 % of that state's territory. The Cerrado is distributed from 5 °N to almost 34 °S and the altitude varies from sea level to 1800 m. The climate is typically hot with rainy summers and dry winters from April until September (Ratter et al., 1997). The annual precipitation in 90 % of the territory is 800–2000 mm; the average annual temperature varies from 18 °C to 28 °C (Dias, 1992). The Brazilian Cerrado has the greatest biodiversity of all the savannah regions in the world. It contains high numbers of local and regional endemic species and is considered one of the earth's hot spots for the preservation of biodiversity (Mittermeyer et al., 1999). However, the Cerrado region has lost much of its natural vegetation in the past and is still under constant threat due to the development of modern agriculture (mainly soybean production), charcoal production and fire (Ratter et al., 1997).

Annona crassiflora (‘araticum’ or ‘marolo’) is a native Cerrado tree species, belonging to the Annonaceae (Lorenzi, 1998; Silva, 2001). The fruit of A. crassiflora is appreciated for its smell and flavour. The fruit pulp is consumed in natura, and it also is used to prepare liqueur, sweets, ice creams and cakes (Ribeiro et al., 2000). Extracts of A. crassiflora seeds are also used in folk medicine to treat snake bites (Correia, 1926). Santos et al. (1996) purified a compound from A. crassiflora seeds, named araticulin, which showed in vitro cytotoxicity to human lung carcinoma and melanoma cells, whereas extracts obtained from A. crassiflora leaves showed antifungal properties (Silva et al., 2001). Because of the importance of the species, fruits of A. crassiflora are frequently being collected in the Cerrado (extractivism). In addition, agricultural occupation of the Cerrado is hindering the natural regeneration of species.

Thus, actions need to be taken to prevent this species from becoming extinct in the near future. Preservation and conservation of A. crassiflora requires an integration of sociological, economical and biological interests. However, this is a long-term goal and at present a reforestation programme using A. crassiflora seedlings is needed. However, A. crassiflora seeds display prolonged dormancy (Rizzini, 1973), due to embryo immaturity (Rizzini, 1973; Baskin and Baskin, 1998). Seed germination of A. crassiflora at field conditions takes 230–300 d (Rizzini, 1973). This makes seedling production and conservation of native populations a very expensive and time-consuming procedure. Recently, seed germination from species of the Cerrado has received more attention, because it was thought that reproduction of the species was preferably vegetative and that seedling establishment through seeds was an uncommon event (Baskin and Baskin, 1998).

Although some information is available regarding the behaviour of A. crassiflora seeds during seed dispersal and germination, there is a lack of more detailed studies about the level of embryo differentiation at the time of seed dispersal, the type of dormancy that seeds of A. crassiflora display, the environmental requirements needed to break dormancy and promote seed germination. Studies at these levels may provide a clue as to which technological, physiological and molecular strategies should be adopted to break and understand the dormancy mechanism in A. crassiflora seeds to further accelerate seed germination.

MATERIALS AND METHODS

Seed material

Fruits of Annona crassiflora Mart. were collected from natural populations in the Cerrado region in the State of Minas Gerais Brazil in 2003. At the time of collection, the fruits were mature as judged from colour and size, and most of them had already dispersed. For seed processing, the fruit's skin (exocarp) was removed manually after which the pulp (mesocarp) was separated from the seeds by using a concrete mixer machine. The seeds with fruit pulp stayed in this concrete mixer for approx. 2 h, until complete separation of the seed from the pulp was achieved. The rotation speed and time used for separation of the seeds from the fruit pulp did not cause mechanical damage to the seeds. After processing the seeds were dried at room temperature (20–25 °C) until a moisture content of 10 % (fresh weight basis) was reached after maximally 3 d. The dried seeds were packed in sealed impermeable plastic bags and stored at 10 °C for 1–2 d until the beginning of the experiments.

Germination

Seeds were surface-sterilized in a 1 % sodium hypochlorite solution for 10 min. Eight portions of 500 seeds each were placed separately in a polyester bag with holes of 0·5 cm diameter and buried in a nursery at a depth of 5 cm in April and in September of 2003. The holes in the sacks allowed gaseous and water exchange necessary for the germination process. The substrate that was added to the bags was composed of ravine earth, organic soil and carbonized rice peel in a proportion of 5:2:1. Seeds were exhumed on the fifth day of every month and the number of germinated seeds (radicle protrusion) was determined. The non-germinated seeds were immediately put back in the bags with the same substrate and buried again. Maximum and minimum ambient and soil temperatures were measured and recorded continuously with an electric thermograph during the experiment and the daily mean, maximum and minimum temperatures were calculated. Maximum and minimum day temperatures occurred between 1200 h and 1500 h and 0400 h and 0600 h, respectively. Local precipitation data were obtained from the Engineering Department at Federal University of Lavras-MG, Brazil (21 °14′S; 45 °00′W; at 918·87 m a.s.l.). For germination in GA₄₊₇ solutions, seeds were imbibed in 10 µm, 100 µm, 500 µm and 1000 µm GA₄₊₇ (Berelex, ICI, UK) for 6 d at room temperature to allow complete imbibition of the seeds. Previous experiments have shown that A. crassiflora seeds are fully imbibed after 6 d. As a control, seeds were allowed to imbibe in sterilized water at room temperature. After 6 d of imbibition, treated and control seeds were rinsed in running water and placed at 30 °C under constant light. The substrate used was sterilized sand that was kept moist during the experiment. Four replicates of 25 seeds each were used. Germination was scored daily and seeds were considered to be germinated when the radicle was 2 mm.

Seedling emergence

In an adjacent part of the nursery seeds were planted at a depth of 5 cm in the same substrate as used for the burial experiment. These seeds were not placed in polyester bags. Seedlings were considered emerged when the seedling appeared above the soil surface.

Embryo growth

To measure embryo length and the occurrence of cell division during the germination experiments at field conditions, seeds were placed to germinate following the same methodology as described above. Seeds were exhumed monthly and three replicates of five intact embryos were isolated by using a scalpel, immersed in modified fixative solution of Karnovisk (gluteraldehyde 25 %, formaldehyde 10 % and sodium cacodylate buffer 0·2 m, pH 7·2) and stored in a cold chamber at 10 °C until use. Embryos were washed in 0·05 m cacodylate buffer (three times 10 min), fixed in 1 % osmium tetroxide for 1 h and subsequently dehydrated in an ascending series of acetone (30, 50, 70, 90 and 100 %). The specimens obtained were mounted in aluminium supports with the help of a carbon ribbon with a double face placed on a film of aluminium paper and covered with gold in a sputter apparatus (BAL-TEC SCD 050). Observations were made with a scanning electron microscope (LEO EVO 40XVP). Images were generated and registered at 20 kV and a distance of 9 mm. Embryo length (hypocotyl and radicle) was measured on the images and recalculated to actual dimensions.

To observe cell division, five embryos from germinating and germinated seeds were longitudinally sectioned with a razor blade and mounted on a cup-shaped holder with tissue freezing medium. After mounting, the samples were plunge-frozen and stored in liquid nitrogen for subsequent cryo-planing and observations. Cryo-planing, which attempts to produce flat surfaces for observations in Cryo-SEM, was performed using a cryo-ultra microtome with a diamond knife, according to Nijsse et al. (1999). For observations, the specimens were sputter-coated with platinum and placed in the cryostat of the scanning electron microscope (JEOL 6300 field emission SEM). Observations were made at –180 °C using a 2·5–5 kV accelerating voltage. Digital images were taken and printed. To assess embryo growth in 500 µM GA₄₊₇, 30 embryos (axis plus cotyledons) were isolated and their lengths were measured daily with a digital caliper. Photographs of seeds were taken with a digital camera (Canon Power Shot S40) mounted on a Leica MZ75 binocular.

Endo-β-mannanase activity

Extracts from ten micropylar or lateral endosperms were prepared from seeds that were placed to germinate at field conditions. Alternatively, micropylar and lateral endosperms from seeds treated with 500 µm GA₄₊₇ solutions were also used for quantification of endo-β-mannanase activity. Endo-β-mannanase was extracted in McIlvaine buffer (0·05 m citric acid/0·1 m Na₂HPO₄, pH 5·0) with 0·5 m NaCl and assayed in an activity gel (0·5 mm thick) containing 0·5 % (w/v) locust bean gum (Sigma) in McIlvaine buffer (pH 5·0) and 0·8 % type III-A agarose (Sigma) on gelbond film (Pharmacia). Then, 2 µL from the extracts were applied to holes that were punched in the gel with a 2-mm paper punch. Gels were incubated for 21 h at 25 °C, and then washed in McIlvaine buffer (pH 5·0) for 30 min, stained with 0·5 % (w/v) Congo Red (Sigma) for 30 min, washed with 96 % ethanol for 10 min, and destained in 1 m NaCl for 5 h. All staining and destaining steps were performed on a rotating platform. Commercial endo-β-mannanase from Aspergillus niger (Megazyme, North Rocks, Sydney, Australia) was used to generate a standard curve. Calculations were performed according to Downie et al. (1994). Thirty seeds, collected monthly during germination at field conditions from April to December, were used for localization of endo-β-mannanase activity in the endosperm of A. crassiflora. In addition, ten seeds allowed to imbibe in 500 µm GA₄₊₇ were also used for localization of endo-β-mannanase activity. The integument was removed manually, the seeds were rinsed in deionized water and cut in half with a razor blade. Seed parts were blotted dry on filter paper and laid on top of an activity gel, cut side down. Activity gels were incubated for 20 min at room temperature. Seed parts were then removed from the gel with tweezers. After incubation, the gels were stained and destained, as described above.

RESULTS

Anatomy and morphology

The seed of A. crassiflora is 2 cm long with a hard seed coat (Rizzini, 1973; Fig. 1A). The embryo is located at the micropylar end and is approx. 2 mm long at the time of seed dispersal, when radicle, axis and cotyledons are distinguishable (Fig. 1B, C; de Melo et al., 2006). There are nine or ten cell layers located at the micropylar end of the endosperm in front of the embryonic radicle (Fig. 1E). The endosperm has a surface that is irregular in shape (Fig. 1C, F). Annona crassiflora seeds possess a ruminate endosperm (de Melo et al., 2006). The rumination extends through the endosperm, beginning at the epidermal cells and extending to the middle (Fig. 1C, F). Normally the rumination ends with a bifurcation. In parallel with embryo growth within the seed, prior to radicle protrusion, the seed coat showed ruptures starting from the micropylar region and continuing until the micropylar region of the seed was exposed completely (Fig. 1B). The presence of ruptures coincided with the beginning of embryo growth within the seeds and, therefore can be used as a morphological marker for germination.

Fig. 1.

(A) Intact Annona crassiflora seed at time of dispersal. (B) Seed with ruptured seed coat. (C) Seed longitudinally cut showing the micropylar endosperm (MEn), the lateral endosperm (LEn) and the embryo (Em). (D) Embryo at the time of seed dispersal. Note the presence of axis and cotyledons. (E) Scanning electron micrograph of the micropylar endosperm (MEn) and embryo (Em) of imbibed seeds. Observe that the micropylar endosperm contains eight or nine cell layers in front of the embryo. The dotted line indicates the border between embryo and endosperm. (F) Lateral endosperm. Arrows indicate ruminations in the endosperm. Scale bars: A–C = 5 mm; D = 200 µm; E = 10 µm; F = 100 µm.

Germination and embryo growth in field conditions

Radicle protrusion began to occur 150 d after seeds were sown (Fig. 2). The experiment was started in April and the first radicle protrusion was observed in September. Seedling emergence started 30–60 d after radicle protrusion, which corresponds to the months of October–November. Germination and seedling emergence continued until December when a maximum percentage of germination of 60 % was attained (Fig. 2). Additional germination was only observed again in September 2004. Germination started after the increase of the minimum and maximum temperature, which coincided with September of 2003 and 2004 (Fig. 2).

Fig. 2.

(A) Germination at field conditions during the years 2003 and 2004. Data points are averages of eight replications of 500 seeds each; error bars indicate standard deviation. (B) Average monthly temperature, and average monthly maximum and minimum temperatures and average monthly precipitation (bars) at the experimental site during the germination experiment.

At the time of seed dispersal the embryo possessed cotyledons and axis (Fig. 1D). However, the embryo is very small (approx. 1·8 mm) when compared with the length of the seed (approx. 17·0 mm). In the field the embryo did not show any increase in length from April until August (150 d) (Fig. 3A). After August the embryo grew within the seed, prior to germination, from 1·8 mm to 6·0 mm of length (Fig. 3A, B). Thus, the embryo increased 3·3 times in length prior to germination. Radicle protrusion took place only after this increase in embryo length within the seed (Fig. 3A, B). Cotyledons also showed an increase in length to a similar extent as the embryonic axis (Fig. 3A, B). The increase in embryo length coincided with the beginning of cell division prior to radicle protrusion, as observed by scanning electron microscopy, which started in September (Fig. 4). Embryos from seeds that stayed dormant in the soil (around 40 %) did not grow. Thus, only seeds that were germinating showed an increase in embryo length.

Fig. 3.

(A) Length of axis and cotyledons during germination at field conditions. (B) Images of embryo growth during germination at field conditions in 2003, following dispersal in April. Note that the embryo grows inside the seed prior to radicle protrusion in December. First seed = seed at time of dispersal; second seed = August; third seed = September; fourth seed = October; fifth seed = November.

Fig. 4.

Scanning electron micrograph of cells in the embryonic axis in September 2003, when the embryo started to grow, but prior to radicle protrusion. Arrows indicate dividing cells. Scale bar = 3 µm.

Endo-β-mannanase activity

Endo-β-mannanase was only detected from September onward, when germination started. The activity was observed in both the micropylar and lateral endosperm (Figs 5 and 6). However, in September (prior to radicle protrusion) the activity of endo-β-mannanase was higher in the micropylar than in the lateral endosperm (Fig. 5). In October, when germination had already started, the activity was practically the same in both regions of the endosperm. Germinated seeds (November) showed higher activity of endo-β-mannanase in the lateral endosperm (Figs 5 and 6). Thus, the initial occurrence of endo-β-mannanase activity in the micropylar endosperm in September coincided with the increase in embryo length, although radicle protrusion was not yet observed. Endo-β-mannanase activity in the lateral endosperm was also detected prior to radicle protrusion and increased until and after radicle protrusion.

Fig. 5.

Endo-β-mannanase activity in micropylar and lateral endosperm during germination at field conditions from April 2003 (seed dispersal) to November 2003 (after radicle protrusion). Error bars are too small to be shown. Coefficients of variation were 10 % and 6 % for micropylar and lateral endosperm, respectively.

Fig. 6.

Tissue printing of longitudinally cut seeds before and after radicle protrusion, showing endo-β-mannanase activity as clearings in micropylar and lateral endosperm. (A) Absence of endo-β-mannanase activity. (B) Activity in the micropylar region only. (C) Activity in the micropylar and lateral endosperm before radicle protrusion. (D) Activity in the micropylar and lateral endosperm after radicle protrusion. Drawn representations of embryo growth and endo-β-mannanase activity are included for clarity.

Tissue printing was used to locate endo-β-mannanase activity in the endosperm during germination at field conditions. Both the micropylar and lateral endosperm showed activity, but activity in the micropylar endosperm (surrounding the embryo) started to appear earlier (Fig. 6). The activity first appeared in the embryo cavity in September and subsequently spread throughout the whole endosperm. The appearance of activity in the micropylar endosperm coincided with the increase in embryo length within the seed prior to radicle protrusion (Fig. 6). The activity in the lateral endosperm coincided with ruptures in the seed coat and subsequent radicle protrusion.

Effect of gibberellins

Different concentrations of exogenous GA₄₊₇ were effective in breaking dormancy in A. crassiflora seeds (Fig. 7A). An optimal concentration of 500 µm allowed germination of 43 % of the seeds whereas 1000 µm was inhibitory. The concentration of 500 µm was used for quantification of endo-β-mannanase activity and measurement of embryo growth. Exogenous GA₄₊₇ induced endo-β-mannanase activity in both the lateral and micropylar endosperm. Activity in the micropylar endosperm was slightly higher than in the lateral endosperm (Fig. 7B). Tissue printing showed initially more intense clearing around the embryo cavity in the micropylar endosperm region than in the lateral endosperm (Fig. 7B). In parallel with the induction of endo-β-mannanase activity, exogenous GA₄₊₇ also induced embryo growth of A. crassiflora seeds (Fig. 7C).

Fig. 7.

(A) Germination in various concentration of GA₄₊₇ at 30 °C. Data points are the average of four replicates of 25 seeds. Standard deviations were smaller than the symbols. (B) Endo-β-mannanase activity in micropylar and lateral endosperm from seeds incubated in 500 µm GA₄₊₇. Inset: tissue prints of seeds allowed to imbibe in water (1) and in 500 µm GA₄₊₇ for 15 d (2), displaying endo-β-mannanase activity in micropylar (top side) and lateral endosperm. Standard deviations were smaller than the symbols. (C) Total embryo length in seeds imbibed in 500 µm GA₄₊₇. Bars indicate s.d.

DISCUSSION

Seed dormancy of A. crassiflora has been classified as non-deep simple morpho-physiological (Baskin and Baskin, 1998). This implies that physiological dormancy has to be broken prior to, or in parallel with, the release of the morphological dormancy. The present results show that exogenous GAs can break physiological dormancy, as in many other species with physiological dormancy. This allows subsequent removal of morphological dormancy and the start of embryo growth. However, the question remains how this is accomplished under field conditions. In the field, germination first occurred with the first rains in October 2003, and peaked in November, when precipitation was substantial (Fig. 2). First significant embryo growth was observed in September (Fig. 2). Thus, field conditions prior to September may have been decisive for the breaking of (physiological) dormancy. This is corroborated by the observation that burial of seeds in September 2003 did not result in germination in the subsequent months (data not shown). In many species, physiological dormancy can be broken by a (prolonged) cold period (Probert, 2000). In 2003, seeds in the soil were exposed to 5 months with average minimum temperatures below 15 °C. Alternating sequences of high (30 °C) and low temperature (5–10 °C) periods (data not shown), as well as a constant optimal temperature of 30 °C, were tested (Fig. 7A) under laboratory conditions but this never resulted in germination. Thus, a simple combination of cold and warm stratification was not effective in the breaking of dormancy. A qualification of A. crassiflora seeds as having intermediate, rather than non-deep simple morpho-physiological dormancy therefore seems appropriate (Baskin and Baskin, 1998). The term ‘cold stratification’ is defined as temperatures between 0 °C and 10 °C being effective in the breaking of dormancy (Baskin and Baskin, 1998). In this sense, dormancy break in A. crassiflora is not the result of cold stratification as during the experimental period the lowest soil temperature at a depth of 5 cm was 11·6 °C on 28 June 2004. However, it has been shown for a number of temperate annual species that dormancy break may occur at temperatures up to 15 °C (Bouwmeester and Karssen, 1993). Hence, the definition of ‘cold stratification’ as occurring between 0 °C and 10 °C may be too strict and should rather be related to representative winter temperatures (Baskin and Baskin, 1998). It is also possible that dormancy breaking is related to the amplitude of minimum and maximum temperatures (night versus day temperatures), which attains its maximum in June/July (ΔT approx. 14 °C), mainly as a result of decreasing minimum temperatures (Fig. 2). Alternating temperatures may be very effective in the breaking of dormancy when constant temperatures are not, e.g. in Rumex species (Roberts and Totterdell, 1981; Probert, 2000). It has been established clearly that cold stratification results in the synthesis of GAs and an increase in sensitivity to the hormone in physiologically dormant seeds of Arabidopsis thaliana (Derkx et al., 1994; Yamauchi et al., 2004). It is likely that a similar mechanism is operational in A. crassiflora seeds since it has been shown here that GAs are effective in the removal of physiological dormancy in this species.

Annona crassiflora seeds therefore restrict their germination to the period of the year that is favourable to successful seedling establishment. Based on the present results, this species can be classified as belonging to the group of species in the intermediate-dry syndrome (Garwood, 1983). In this group, the occurrence of dormancy avoids germination during short wet spells in the dry season. However, other members of the Annonacea, e.g. the recalcitrant tropical wetland species Annona glabra, have adapted to different habitats, in which dormancy is broken by other environmental conditions (Mata and Moreno-Casasola, 2005). The observed peak in germination at the beginning of the rainy period only, indicates that the species has developed a strategy to maximize the survival rate of its seedlings. Approximately 40 % of the seed population remained dormant in the soil without displaying any morphological changes in embryo and biochemical changes in the endosperm. This behaviour indicates that the species can contribute to the formation of soil seed banks, allowing germination of only part of the seed population in the year of dispersal. About 20 % of the seeds germinated in the following year (2004), with a similar timing as in 2003, whereas the remaining seeds were predated by insects or deteriorated naturally in the soil (data not shown). This survival mechanism allows the plants to survive and perpetuate in the Cerrado.

The morphological component of dormancy in A. crassiflora seeds is caused by embryo immaturity (Rizzini, 1973). According to Rizzini (1973), the embryo is an undifferentiated tissue and very small at the time of dispersal, comprising a mass of cells about 2 mm long. However, de Melo et al. (2006) have shown that the embryo possesses clearly distinguishable axis and cotyledons. Thus, the embryo is fully differentiated at the time of dispersal. Once physiological dormancy has been broken the embryo starts to grow. In 500 µm GA₄₊₇ the embryo grows in approx. 20 d to a length of 6·3 mm, which marks the point of radicle protrusion. In field conditions this process seems considerably slower. From the data presented in Fig. 3A, a period of 2 months between the beginning of growth and radicle protrusion is estimated. The embryo increased its length >3-fold within the seed prior to radicle protrusion and was mainly caused by cell expansion since only few dividing cells were observed (Fig. 4). Seeds from the temperate member of the Annonacea, Asimina triloba, started to germinate at day 12 after 12-month cold stratification at 5 °C (Finneseth et al., 1998). Embryo growth prior to radicle protrusion occurred at a similar rate as in A. crassiflora but not during cold or warm stratification. Similar characteristics of embryo growth prior to germination have been found in coffee seeds (da Silva et al., 2004).

The endosperm of seeds from the Annonaceae family is thickened by galactomannans that were deposited in the cell walls (Buckeridge et al., 2000). The lateral endosperm of A. crassiflora displays thickened cell walls but the micropylar endosperm contains relatively thin cell walls (de Melo et al., 2006). Thus, the micropylar endosperm is modified to be less resistant to radicle protrusion and only embryo growth may be sufficient to allow radicle protrusion, as has been shown in seeds of fenugreek, Chinese senna, carob and date (Gong et al., 2005) and in Coffea arabica cv. Rubi (da Silva et al., 2004). Endo-β-mannanase activity has been associated with the hydrolysis of galactomannan-containing cell walls in the endosperm of many species, including Datura ferox (Sánchez and de Miguel, 1997), Solanum lycopersicum (Groot et al., 1988) and Coffea arabica (da Silva et al., 2004), thereby releasing carbohydrates as energy supply for embryo growth and reducing the mechanical restraint to growth of the radicle. The beginning of embryo growth in A. crassiflora seeds coincided with the appearance of endo-β-mannanase activity around the embryonic cavity in the micropylar region of the endosperm that subsequently diffused also to the lateral endosperm (Fig. 6). Further growth of the embryo under field conditions was tightly correlated with endo-β-mannanase activity. However, in 500 µm GA₄₊₇, the evolution of enzyme activity was considerably faster, as was embryo growth. It has been shown in tomato that endo-β-mannanase activity depends on the presence of GAs (Groot et al., 1988). The activity of endo-β-mannanase observed initially in the micropylar endosperm region where the small embryo is located may have a function in the degradation of the endosperm to supply energy-rich compounds to the embryo, but also to create space in the embryo cavity to accommodate embryo growth prior to radicle protrusion.

In summary, the physiological component of morpho-physiologically dormant A. crassiflora seeds is broken by low temperature and/or temperature fluctuations preceding the rainy season. Subsequent embryo growth and degradation of the endosperm are likely to be both controlled by GAs synthesized during the breaking of physiological dormancy. Radicle protrusion thus occurs at the beginning of the rainy season, thereby maximizing the opportunity for seedlings to emerge and establish.