Constraints to symbiotic germination of terrestrial orchid seed in a mediterranean bushland


  • A. L. Batty,

    Corresponding author
    1. Botanic Gardens and Parks Authority, Kings Park and Botanic Garden, West Perth, Western Australia 6005, Australia;
    2. Soil Science & Plant Nutrition, University of Western Australia, Nedlands, Western Australia 6907, Australia
    Search for more papers by this author
  • K. W. Dixon,

    1. Botanic Gardens and Parks Authority, Kings Park and Botanic Garden, West Perth, Western Australia 6005, Australia;
    Search for more papers by this author
  • M. Brundrett,

    1. Botanic Gardens and Parks Authority, Kings Park and Botanic Garden, West Perth, Western Australia 6005, Australia;
    Search for more papers by this author
  • K. Sivasithamparam

    1. Soil Science & Plant Nutrition, University of Western Australia, Nedlands, Western Australia 6907, Australia
    Search for more papers by this author

Author for correspondence: A. L. Batty Fax: +618 94803641 Email:


  •  The dependence of seeds of terrestrial orchids on specific fungi for germination provides a means of locating these fungi in the wild and to investigate the role of appropriate fungi in the germination of orchid seed and development of seedlings under natural field conditions.
  •  Seed baits, comprising orchid (Caladenia arenicola) seed enclosed in fine nylon mesh, were placed at sample points along four transects through two orchid populations in bushland in Western Australia. Seed germination was scored and compared with adult orchid plant distribution and soil factors.
  •  A small fraction of available seed (< 1%) germinated to a stage of tuber formation where survival over the subsequent dry season would have been possible. Germination increased in the vicinity of adult C. arenicola plants, but other factors, such as soil potassium levels and presence of leaf litter, were also correlated with seed germination.
  •  The measurement of the spatial variability in germination events within an orchid habitat demonstrated the availability of new recruitment sites. This information is required to assess the natural recruitment capacity and the potential for orchid reintroduction in natural habitats.


Terrestrial orchids have associations with mycorrhizal fungi that are considered necessary for seed germination and growth of orchid plants (Warcup, 1973; Clements & Ellyard, 1979; Clements et al., 1986; Rasmussen, 1995). While it is possible to germinate orchid seed in vitro utilizing asymbiotic protocols (Arditti, 1992), symbiotic germination using appropriate fungi is generally more efficient (Clements et al., 1986; Rasmussen, 1995). In situ seed germination in natural habitats requires the presence of suitable fungi for seedling development and establishment (Zelmer et al., 1996; Zettler & Hofer, 1998). Under natural field conditions some orchids have been found to have a greater specificity of fungal partners than under laboratory conditions (Masuhara & Katsuya, 1994; Perkins & McGee, 1995). The term ‘ecological specificity’, applied to terrestrial orchids, refers to the role of plant–fungus interactions in defining orchid habitats under natural field conditions, whereas those associations occurring under laboratory conditions may be termed ‘potential specificity’ and should not be assumed to be relevant to field situations (Harley & Smith, 1983; Masuhara & Katsuya, 1994). For example, Perkins et al. (1995) found that Microtis parviflora has a narrow ecological specificity, as it only formed mycorrhizas with two Epulorhiza species in the field. Terrestrial orchids may have narrow or broad potential specificity (Warcup, 1981; Alexander & Hadley, 1983; Muir, 1989), but the specificity of their associations with endophytes in natural habitats is still poorly understood.

There are various reports of the detection of orchid protocorms in the field. Rasmussen & Whigham (1998a) summarize the cases where spontaneous seedlings have been observed in the field, with such observations, until recently, being the only published basis of our understanding of protocorm–fungal interactions in the wild.

Methods have been developed that overcome the experimental problems associated with the dust-like seed of orchids, and allow in situ studies of orchid seed germination (Rasmussen & Whigham, 1993, 1998b; Masuhara & Katsuya, 1994; Perkins & McGee, 1995; McKendrick et al., 2000). These techniques, which enable identification and enumeration of fungi in the field, are essential for an accurate understanding of the ecology of orchids and their fungal associates in natural habitats. Benefits of understanding the ecological specificity of fungi include improved management and translocation opportunities for terrestrial orchids particularly with the aim of rebuilding wild populations (IUCN/SSC Orchid Specialist Group, 1996). Seed baits were successfully used in an earlier study of soil seed bank dynamics of terrestrial orchids in south-western Western Australia (Batty et al., 2000).

Caladenia arenicola is a widespread and common terrestrial orchid from the south-west mediterranean zone of Western Australia (WA). The species grows in open woodlands dominated by species of Eucalyptus, Banksia and Allocasuarina with a diverse sclerophyllous understorey. A single hairy leaf (15–25 cm long and 5–12 mm wide) is produced at the onset of the wet season, remaining conspicuous for approx. 7 months (Hoffman & Brown, 1992). C. arenicola is typical of many orchids in the south-west of WA being summer (dry season) dormant with growth and flowering events coinciding with the arrival of cooler winter (wet season) conditions typical of a mediterranean-type climate (Gentilli, 1989).

The aims of this study were to investigate the timing and frequency of in situ seedling development and tuberization of C. arenicola, by the use of buried, orchid seed baits to locate naturally occurring fungi capable of stimulating seed germination under field conditions. Soil factors related to successful germination in the field were also investigated. This study is part of a strategic programme to develop methodologies and acquire knowledge required for the successful conservation of Australian terrestrial orchid species. In particular, the long-term success of translocation and propagation efforts, particularly with threatened orchid taxa, requires knowledge of the distribution, abundance and host specificity of orchid fungi in natural habitats.

Materials and Methods

Seed collection

Seed was collected from wild plants growing in seminatural woodland dominated by species of Banksia, Allocasuarina and Eucalyptus in a secure conservation reserve at Kings Park in Perth, Western Australia. This woodland had a diverse understorey dominated by sclerophyllous shrubs. In this mediterranean environment, seed set for C. arenicola (Orchidaceae) Hopper and A. P. Brown occurs in early spring, with seed maturation taking up to 6 wk and seed dehiscence coinciding with the drying of soils before the onset of summer drought (Batty et al., 2000). Plants were hand pollinated to ensure sufficient seed for experimentation. Seeds were obtained by collecting capsules from individuals of uniform appearance occurring within 1 km of the study area, just before seed release. Contents of C. arenicola capsules were combined and thoroughly mixed. Seeds were cleaned by passing through a fine sieve, dried over silica gel for 24 h at 22°C and stored in air-tight containers at 4°C before experimentation.

Seed burial transects

Seed baits consisted of 400–600 orchid seed and 0.5 g of dry white silica sand sandwiched between two 40 mm × 40 mm sections of nylon cloth (pore size 90 µm), held in place between two halves of standard plastic 35 mm photographic slide mounts (Fig. 1a) using the method of Rasmussen & Whigham (1993), with minor modifications. Locational tags were attached to the slide mounts to assist recovery of baits (Fig. 1b). Field observations by the authors show that protocorms occurring under natural field conditions typically form at the interface between the litter layer and the soil A horizon. Thus, seed baits were positioned under the litter layer (approximate depth 4 cm), at 45 degrees to the vertical, along transect lines. Each transect consisted of two rows of 101 sample points at 50 cm intervals (Fig. 1c). Four transects were centred around two areas where C. arenicola is relatively common (Fig. 1d) producing a total of 808 sample points.

Figure 1.

Diagrammatic representation of (a) seed bait design; (b) assembled seed bait and locating tag; (c) overview of individual transects; and (d) relationship of each transect to adult orchid plants (indicated by coloured squares). Transects one and two intersect through population one while transects three and four intersect through population two.

Seed baits were placed into field sites at the beginning of the wet season (autumn) and removed, with the exception of a single row from one transect, before the onset of the dry season (mid-spring), 20 wk later. The remaining row of seed baits was recovered after a further eight weeks, to determine the level of tuberization. These data were combined as no substantial differences in tuberization were noted.

Recovered seed baits were gently washed under running tap water and maintained moist between paper towel, for up to 1 d, until examination. Seed germination stages were determined by scoring six random fields of view from open seed baits (Fig. 2a) for each slide at 60× under a dissecting microscope. Germination stages were scored using the rating method of Batty et al. (2000), which is based on germination stages described by Ramsay et al. (1986). Ungerminated seeds (Fig. 2b) were assigned to three groups: no pro-embryo present with decayed testa; intact testa containing a pro-embryo; and swollen pro-embryo with cracked testa. Successful germination (Fig. 2c) was defined as the production of trichomes representing at least stage two development.

Figure 2.

General overview of recovered seed baits. (a) Opened slide frame ready for scoring; (b) ungerminated, imbibed and dead seed; (c) cluster of protocorms. Scale: (a) scale in cm; (b) bar, 0.2 mm; (c) bar, 1 mm.

Mycorrhizal infection

Protocorms resulting from the baiting trial were cleared using KOH and stained in trypan blue in lactoglycerol and observed under the light microscope to confirm the presence of internal hyphae (Phillips & Hayman, 1970; Brundrett et al., 1996). As a further test of the efficacy of endophytic fungi, hyphal coils (pelotons) extracted from wild adult orchid plants and protocorms recovered from seed baits were used to initiate mycorrhiza from C. arenicola (Clements & Ellyard, 1979; Dixon, 1989). Endophytes that were free of contamination, were tested by symbiotic germination on oat agar with surface sterilized C. arenicola seeds using standard orchid seed sowing procedures (Warcup, 1971, 1973; Clements & Ellyard, 1979). Seed germination was assessed using the criteria described above. Each treatment was replicated six times.

Soil factors

Soil samples were taken along the four transects, at 5 m intervals and at points where seed germination was observed. Approximately 300 g of soil was removed at each sample point, homogenized and a 100-g subsample used for the soil analysis. CSBP soil laboratories (CSBP and Farmers Ltd.) employed standard chemical soil analysis to determine levels of phosphorus, nitrogen in the form of nitrate and ammonium, potassium, organic carbon, iron and pH (1.5 water) (Rayment & Higginson, 1992). Leaf litter was sampled at each transect location by using a 70-mm diameter corer to remove material down to the A soil horizon, which was oven dried at 70°C for 24 h and weighed.


Seed germination and soil data from transect locations were statistically analysed by ANOVA. Fisher’s LSD with a 95% confidence interval was then applied to make comparisons between means. Correlation analyses were also performed to test relationships between soil properties and seed germination and between different categories of germination. Analysis was performed using the statistical program, Statview® (Statview, SAS Institute Inc.).


Transect germination results

A broad spectrum in seed integrity and germination stages was present in the 34578 seeds scored from the 808 recovered samples. Germination stages ranged from protocorms with trichomes to seedlings with droppers (sensuPate & Dixon 1982) (Fig. 3a,b) and tubers forming (Fig. 3c), and finally the production of dormant tubers towards the end of the study period (see double arrows in Fig. 3d). Etiolated leaves were observed in some recovered protocorms (Fig. 3b,c). Examination of cleared and stained protocorms and seedlings showed fungal material, present as pelotons, to be confined to specific locations (Fig. 3, indicated by single arrows) and absent from seeds that failed to germinate.

Figure 3.

Details of developmental stages recovered from the field during early spring (a–c) and following the onset of dry conditions (d). (a) Germination stages 2 (far left) through to 4 (far right); (b) stage 6 showing initiation of dropper; (c) tuber now present; and (d) formation of dormant tubers. Single arrows indicate position of fungal infection and double arrow presence of tuber. Scale: (a) bar, 1 mm; (b) scale in mm; (c) bar, 2 mm; and (d) bar< 3 mm.

The number of nonviable seeds sampled from each transect ranged from 20 to 37% and was similar to the number of cracked seeds (23–29%). The number of swollen seeds observed (37–51%) was significantly greater (ANOVA P < 0.001) than those scored as nonviable or cracked. Protocorms scored as stage four and above were classed as successful germinants, as only these protocorms were considered likely to produce a dormant tuber by the end of the growing season. Those that had not produced a leaf during the sample period were grouped in a separate category deemed unlikely to produce a tuber capable of sustaining a plant during summer dormancy (Fig. 4). Samples examined at the onset of the dry season showed that nontuberized seedlings were desiccating. Germination was only observed in 14.2% of sample points and was not evenly distributed along transects. Successful germination was restricted to certain patches of soil that were not necessarily colocated with high densities of adult plants (Figs 1d, 4).

Figure 4.

Seed states (dead (white columns); swollen (light yellow); cracked testa (light green)) and germination stages (stages 2 & 3 (dark yellow) and stage 4 & 5 (dark green)) along four 50 m transects showing outcomes at 50 cm intervals. Percentages calculated from total number of seeds scored at each sample point.

Germination relative to adult C. arenicola plants

Three out of the four transects had significantly closer average distances between successfully germinated seeds and the closest adult C. arenicola plants, when compared to sites where seeds failed to germinate (Fig. 5). Analysis of average data for all transects showed that the distance of germinated seed to the closest adult C. arenicola plants was significantly less than the mean distance from an adult plant to seeds which failed to germinate (P < 0.001). The frequency of successful germination was greatest in seed frames where few seeds germinated (Fig. 6). However, in some cases 12 or more seedlings survived in the seed baits.

Figure 5.

Average distance of locations supporting germinated seed compared to the average distance of seeds which failed to germinate relative to the presence of adult Caladenia arenicola plants. Bars with different letters are significantly different, ANOVA P < 0.001.

Figure 6.

Frequency distribution of seed germination (stages 3–5) in relation to the number of protocorms contained in the seed bait. Stage 3, light grey columns; stage 4, dark grey columns; stage 5, black columns.

Germination relative to soil factors

Soil properties were compared with seed germination at particular transect locations. Seed germination was highly positively correlated with soil potassium and surface litter, and negatively with organic carbon (Table 1). Levels of potassium and leaf litter were significantly (P < 0.001) higher between sites where germination occurred than where it did not (Table 2). Significant differences in some soils properties were also observed between transects (Table 2).

Table 1.  Correlation between soil factors and orchid seed germination along the transects
FactorGerminationPhosphorusAmmoniumPotassiumOrganic CIronpH (water)Surface mulch
  1. ***, 1% confidence = 0.254; **, 5% = 0.195; *, 10% = 0.164.

Germination 1.000       
Phosphorus 0.037 1.000      
Ammonium 0.022 0.061 1.000     
Potassium−0.558***−0.145 0.047 1.000    
Organic C−0.195**−0.137 0.327*** 0.364*** 1.000   
Iron 0.065−0.095 0.244** 0.057 0.025 1.000  
pH (water) 0.124 0.166*−0.330***−0.091−0.119−0.480*** 1.000 
Surface mulch 0.240**−0.131 0.282***−0.113 0.134 0.759***−0.475***1.000
Table 2.  ANOVA of soil properties using presence/absence of germination, and transect, as factors
  1. *, Significant values; **, highly significant values.

Phosphorus 0.308 0.580 0.243 0.866
Ammonium 4.434E-4 0.983 3.703 0.014
Potassium62.505<0.001* 5.499 0.002
Organic C 2.844 0.095 4.968 0.003
Iron 0.084 0.77334.903<0.001*
pH (water) 6.070 0.01636.426<0.001*
Surface mulch18.264<0.001*90.491<0.001*

Isolation and resynthesis

C. arenicola endophytes from adult plants or protocorms were compared in ex situ germination trials with C. arenicola seeds. Fungi from adult plants and protocorms were nearly identical in cultural characteristics and symbiotic effectiveness and were considered to be similar fungi. Further research using molecular techniques as used by McKendrick et al. (2000) is required to determine how similar the fungal isolates are.


The use of an orchid seed baiting technique was found to be an effective means for investigating in situ orchid seed germination and seedling development when a large number of sample points were investigated. Rasmussen & Whigham (1993) and McKendrick et al. (2000) used similar methods to investigate seed germination and also found that in situ orchid seed germination was highly variable. Van der Kinderen (1995) used a similar in situ method to investigate seed germination of the terrestrial orchids Dactylorhiza maculata and Epipactis helleborine at different soil depths. The present study appears to be the first in which the production of dormant tubers directly from protocorms has been observed under natural conditions for herbaceous terrestrial species without recourse to an autotrophic phase.

Results of this study are in agreement with earlier observations that seeds of the Western Australian terrestrial orchid C. arenicola germinate during winter (wet season) and seedlings continue to develop until the onset of dry conditions at the end of the growing season some 5–6 months later (Batty et al., 2000). As is typical for orchids and other herbaceous perennials in this mediterranean environment (Pate & Dixon, 1982), tubers formed by October enter dormancy in order to survive the onset of the dry summer period.

The high degree of variability in either the timing of seed germination or the rate of seed development resulted in a broad range of germination stages in the slide frames. Only those seeds which germinate early in the growing season, or which develop rapidly, attain a tuberization state and thus have a chance of recruitment into the adult plant population. This phenological variability could have resulted because: mycelia of appropriate fungi were not initially present at the seed microsite and required time to encounter orchid seeds (foraging activity of the fungus); the fungus was present at many sites, but only had sufficient resource for sustaining germination at a few locations; or other less effective fungi were present at sites where germination did not proceed fully.

The Orchidaceae produce vast numbers of seeds with very low maternal contributions (Arditti & Ghani, 2000). In earlier studies we have established that on average only 4% of the flowers of C. arenicola set seeds without artificial pollination, with each capsule containing approx. 30 000 ± 2000 seeds (Batty, 2001). Thus, the average seed production capacity of C. arenicola would be approx. 1200 seeds per plant per year in a natural population. The seed bank of terrestrial orchids in this environment lasts < 1 yr (Batty et al., 2000), so any seeds that fail to germinate in the first growing season will perish. Of the approx. 34 500 seeds examined in this study < 1% reached a stage where they would be capable of producing a plant able to survive summer dormancy. The average amount of seed entering the seed bank following seed dispersal for other taxa is estimated to be around 50% of the seed rain (Baskin & Baskin, 1998), but we would expect a smaller proportion of orchid seeds to reach a suitable position under the litter and avoid predation (probably < 10%), further reducing the probability of a seed resulting in a plant. Roche et al. (1998) and Rokich (1999) showed that 37% to 51% of terrestrial plant taxa, other than orchids, from habitats comparable to those investigated in this study survived as seedlings beyond the first year following germination. However, there are no comparable data for orchid seedling survival, although we have observed high incidences of grazing damage on orchid seedlings compared to seedlings of other herbaceous plants.

Data presented above can be used to estimate the overall success of C. arenicola recruitment from seed in this habitat equating to approximately 0.4 seedlings per parent plant per year. This estimate is based on what we believe are generous assumptions regarding plant longevity, seed dispersal and seedling survival rates. This estimate only applies to conducive years (periods of high rainfall), as a subsequent study of the same habitat failed to detect any successful seed germination for a subsequent year which had a highly variable growing season (P. Hollick pers. comm.). It should be noted that C. arenicola is a relatively common orchid and we must expect recruitment rates to be much less for other less common orchids with similar life expectancies.

Symbiotic seedlings of the study species, were unable to produce green leaves due to confinement within slide frames, but nevertheless developed tubers in the absence of light, presumably as myco-heterotrophic seedlings. We cannot be certain that the capacity for C. arenicola seedlings to reach the stage of tuberization would not have been enhanced if leaves had been exposed to light or allowed to develop outside the constraints of the nylon cloth necessary for seed recovery. This species normally produces a green leaf within 3 months of sowing in the field, suggesting a preference for a combination of myco-heterotrophy and autotrophy (A. Batty et al., unpublished). There are other reports of seedlings surviving underground for the first years of development (Rasmussen, 1995). For example, Spiranthes lacera may develop to flowering in two years with a mycotrophic phase in the first year after germinating and a partly autotrophic phase in the second season (Zelmer & Currah, 1997).

Inoculum potential is defined as the energy for growth of an organism at the surface of the host, and is a consequence of the number of active propagules of that organism and their nutritional status (Garrett, 1956). Known propagules of mycorrhizal fungi are thought to include spores, dead root fragments, sclerotia, colonized organic material and networks of hyphae in soil (Skinner & Bowen, 1974; Ba et al., 1991; Jasper et al., 1991). Aggregated (nonrandom) distribution patterns have been observed whenever soil was sampled for other types of mycorrhizal fungi (Dahlberg & Stenlid, 1990; Friese & Koske, 1991; Griffiths & Caldwell, 1992; Brundrett & Abbott, 1995), and also seem to be the normal situation for orchid fungi. The inoculum of soil-borne plant pathogens typically also has an aggregated distribution pattern (Campbell & Noe, 1985; Otten & Gilligan, 1998). We have shown, in this study, that the fungi associated with the terrestrial orchid C. arenicola occurs in patches within the habitat. This indicates that the fungi have low saprophytic activity outside the influence of the plant host.

The size of colonies of nonclonal adult orchid plants may be indicative of the health of populations of orchid fungi. Fungal density is likely to determine the chance of a fungus reaching the infection zone of a root before a competing fungus (Esnault et al., 1994). It has been suggested that the presence of other plant species at the site may also influence fungal density (Perkins et al., 1995). In heterogenous soil environments, a higher frequency of successful establishment of myco-heterotrophic seedlings near adult plants would be expected, given the proximity of both the seed source and the fungi associated with the roots of the adult plants. In the present study, there was significantly higher recruitment of seedlings closer to adult plants, but the patches where seedlings grew best did not coincide precisely with areas of higher plant density. Soil fungus activity likely moves from year to year in response to changing soil conditions and the exhaustion of substrates in certain areas. Habitat properties, such as leaf litter accumulation, can provide valuable clues about potential recruitment sites, but the high degree of spatial variability in orchid fungus activity means that most seedlings will fail to establish even in areas that appear highly favourable.

This study supports previous findings (Rasmussen & Whigham, 1998a) that there is considerable spatial variability in seedling recruitment in natural habitats. Our seed study detected a number of potential recruitment sites in an orchid habitat that had yet to be occupied by orchids. It is most likely that orchids, like other seed-bearing plants are also subject to the key driving factors affecting seedling recruitment at a site: arrival of a viable seed, survival, dormancy release, moisture and climatic factors favourable to seedling survival (Baskin & Baskin, 1998). Successful recruitment depends on many factors, as was discussed above.

Knowledge of the ecological specificity of orchids and their fungi is likely to be the key to devising successful orchid translocation programs. Orchids cultured in vitro with fungi for which only potential (laboratory tested) specificity has been demonstrated, may not survive under natural field conditions where ecological specificity may be operating. However, those propagated with fungi selected using ecological criteria (field testing) are more likely to establish and become self-sustaining populations capable of recruiting new individuals into the population. Masuhara & Katsuya (1994) suggest that isolation from fungal coils formed in cortical cells of adults from natural environments is more likely to result in obtaining fungal isolates with ecological specificity. In the present study, there were no clearly observable differences in the fungi associating with adult and seedling plants of C. arenicola.

Perkins et al. (1995) suggest that factors other than host specificity are important in determining which fungus or fungi become mycorrhizal with M. parviflora at any particular site. For C. arenicola, several soil factors examined appear to be related to seed germination (levels of potassium and leaf litter). Potassium often relates to the presence of clay which in turn may affect water holding capacity of the soil. The presence of leaf litter is likely to maintain higher soil moisture levels and also indicates the presence of organic matter on which soil fungi depend (Garrett, 1956). In addition, it is likely that potassium and organic carbon were related to soil structure and texture properties, and thus are indirectly rather than directly related to seed germination.

There are many factors or combinations of factors which may affect the distribution of orchid fungi, which in turn may affect recruitment potential and ultimately the distribution of orchid plants in the field. From this study we conclude that a number of potential recruitment sites exist, that are capable of supporting orchid seed germination and development of seedlings through to the production of a tuber. This information may be critical for synthesizing a self-perpetuating orchid population, ultimately the aim of any translocation, given the dependence of terrestrial orchids on myco-heterotrophy in the early stages of plant development. However, more detailed study is required into the characterization of fungi and the definition of fungal ecotypes and their role in seed germination and plant development stages.


The Western Power Endangered Plant Rescue Program in partnership with Kings Park and Botanic Garden sponsored this study. Volunteers from the Master Gardeners of Kings Park and Botanic Garden provided valuable assistance in the construction and processing of seed baits, with special thanks going to Ethel Lucas and Philip Shaw. Thanks also to Paul Moscardini for his assistance in the study.