Mixotrophy in orchids: insights from a comparative study of green individuals and nonphotosynthetic individuals of Cephalanthera damasonium


  • Thomas Julou,

    1. Service de Systématique Moléculaire (IFR CNRS 101), Muséum National d’Histoire Naturelle, 43, rue Cuvier, 75005 Paris, France;
    2. These two authors contributed equally to this work;
    Search for more papers by this author
  • Bastian Burghardt,

    1. Lehrstuhl für Pflanzenökologie, Universität Bayreuth, 95440 Bayreuth, Germany;
    Search for more papers by this author
  • Gerhard Gebauer,

    1. Lehrstuhl für Pflanzenökologie, Universität Bayreuth, 95440 Bayreuth, Germany;
    Search for more papers by this author
  • Daniel Berveiller,

    1. Laboratoire d’Ecologie, Systématique et Evolution, Université Paris XI, Bât. 362, 91 405 Orsay cedex, France;
    Search for more papers by this author
  • Claire Damesin,

    1. Laboratoire d’Ecologie, Systématique et Evolution, Université Paris XI, Bât. 362, 91 405 Orsay cedex, France;
    Search for more papers by this author
  • Marc-André Selosse

    Corresponding author
    1. Service de Systématique Moléculaire (IFR CNRS 101), Muséum National d’Histoire Naturelle, 43, rue Cuvier, 75005 Paris, France;
    2. These two authors contributed equally to this work;
    3. Present address: Centre d’Ecologie Fonctionnelle et Evolutive, CNRS, 1919 route de Mende, 34 293 Montpellier Cedex 5, France
      Author for correspondence: Marc-André Selosse Tel: +33 (0)6 07 12 34 18 Fax: +33 (0)4 67 41 21 38 Email: ma.selosse@wanadoo.fr
    Search for more papers by this author

Author for correspondence: Marc-André Selosse Tel: +33 (0)6 07 12 34 18 Fax: +33 (0)4 67 41 21 38 Email: ma.selosse@wanadoo.fr


  • • Some green orchids obtain carbon (C) from their mycorrhizal fungi and photosynthesis. This mixotrophy may represent an evolutionary step towards mycoheterotrophic plants fully feeding on fungal C. Here, we report on nonphotosynthetic individuals (albinos) of the green Cephalanthera damasonium that likely represent another evolutionary step.
  • • Albino and green individuals from a French population were compared for morphology and fertility, photosynthetic abilities, fungal partners (using microscopy and molecular tools), and nutrient sources (as characterized by 15N and 13C abundances).
  • • Albinos did not differ significantly from green individuals in morphology and fertility, but tended to be smaller. They harboured similar fungi, with Thelephoraceae and Cortinariaceae as mycorrhizal partners and few rhizoctonias. Albinos were nonphotosynthetic, fully mycoheterotrophic. Green individuals carried out photosynthesis at compensation point and received almost 50% of their C from fungi. Orchid fungi also colonized surrounding tree roots, likely to be the ultimate C source.
  • • Transition to mycoheterotrophy may require several simultaneous adaptations; albinos, by lacking some of them, may have reduced ecological success. This may limit the appearance of cheaters in mycorrhizal networks.


To fulfil their carbon (C) requirements, organisms can reduce CO2, using light as an energy source (phototrophy), or assimilate organic matter (heterotrophy). Mixotrophy is an intermediate strategy that simultaneously combines both carbon sources and is common among algae feeding by uptake of dissolved organic C (Heifetz et al., 2000) or by phagotrophic predation (Zhang & Watanabe, 2001). Algal mixotrophy is considered to play a major role in aquatic ecosystems (Boraas et al., 1988), but to be less important in terrestrial ecosystems. Among land plants, mixotrophy has been long reported for green hemiparasitic plants, such as some Loranthaceae (Bannister & Strong, 2001), which derive a part of their C from their host. Recently, some terrestrial orchids were suggested to achieve mixotrophy by using C from their mycorrhizal fungi (Gebauer & Meyer, 2003; Bidartondo et al., 2004; Selosse et al., 2004).

Most plants form mycorrhizas with soil fungi that usually exchange mineral nutrients for plant carbon (Smith & Read, 1997). Since this symbiosis is rarely specific, some fungi colonize several plant hosts simultaneously. This creates mycelial links connecting plants and allowing C exchanges between plants, even of different species. Some achlorophyllous plants, the ‘mycoheterotrophic’ (MH) plants, specialize on this C source (Leake, 1994) and behave as C sinks in these mycorrhizal networks (McKendrick et al., 2000). Bidirectional interplant C exchanges were also demonstrated for green plants in field conditions (Simard et al., 1997; Lerat et al., 2002), but the extent and physiological relevance of such mixotrophy is often questioned for photosynthetic plants (Robinson & Fitter, 1999; Kytöviita et al., 2003). However, mixotrophy was recently described in some orchids from temperate-zone forests, the Neottieae tribe. 13C abundance in some Cephalanthera and Epipactis spp. are intermediate between abundances of fully MH plants and photosynthetic plants: up to 85% of the orchid C could thus be derived from mycorrhizal fungi (Gebauer & Meyer, 2003; Bidartondo et al., 2004). However CO2 exchanges have not so far been investigated, limiting discussion about the C budget of these mixotrophic species.

Additional interest comes from the phylogenetic relatedness of mixotrophic and MH orchids. MH species evolved repeatedly in Neottieae (Bateman et al., 2005), for example Cephalanthera austinae (Taylor & Bruns, 1997) or Neottia nidus-avis (Selosse et al., 2002a). Assuming that mixotrophic orchids represent an ancestral state, they allow reconstruction of evolutionary steps toward MH strategies. Each MH species shows a high specificity to a narrow fungal clade, which also forms mycorrhizas on surrounding plants (Bidartondo & Bruns, 2001; Taylor et al., 2002, 2003). For example, MH orchids are not associated with the usual orchid fungal partners, the polyphyletic ‘Rhizoctonias’ (Rasmussen, 2002), but with fungi forming ectomycorrhizas (ECM) on trees (Taylor et al., 2002), perhaps because they allow more constant C inflows. Following identification of ECM fungi mycorrhizal in mixotrophic Neottieae (Bidartondo et al., 2004; Selosse et al., 2004), we proposed that the shift to ECM occurred before emergence of fully MH strategies, and probably predisposed to MH emergence by allowing a mixotrophic nutrition. All orchids, however, are somewhat predisposed to MH strategy, since their reserveless seeds initially germinate by using fungal C (Rasmussen, 1995). Neottieae seem especially preadapted to mycoheterotrophy at adult stage by virtue of their association with ECM fungi. Evolution to MH strategy would thus result in three successive transitions: first a ‘mycorrhizal shift’ to ECM fungi allowing mixotrophic nutrition; second a transition to high specificity instead of the usually low mycorrhizal specificity; and third a transition to fully heterotrophic (MH) nutrition. The two latter transitions, and their succession order, still remain unclear.

Neottieae offers an additional unique, hitherto overlooked, opportunity of understanding these transitions. Apparently achlorophyllous (= ‘albino’) individuals are observed in otherwise green species, for example in Epipactis spp. (Salmia, 1989; Selosse et al., 2004) and Cephalanthera spp. (Renner, 1938, Fig. 1). Although we do not know the basis (mutation or phenotypic plasticity) and exact C metabolism for such phenotypes, they may represent an intermediate state of the transition to MH plants. In addition, subterranean (= ‘hypogeous’) Neottieae individuals of adult size also survive (Rasmussen, 1995; Selosse et al., 2004). In Epipactis microphylla, green, underground and albino individuals share identical fungal symbionts, and show some preference toward ascomycetes (mainly Tuber spp.) forming ECM with surrounding trees (Selosse et al., 2004). However, we do not yet know whether albinos from other Neottieae species also share fungal partners with green individuals, or whether they generally have a high fungal specificity. We also do not yet know why albinos are so rare, although this rarity may critically protect mycorrhizal networks against the too frequent emergence of C sinks in their evolution.

Figure 1.

Albino and green Cephalanthera damasonium individuals from the Boigneville (France) population (courtesy of P. Pernot).

Here, we investigate a widespread species of Neottieae, Cephalanthera damasonium, suspected to be strongly mixotrophic (Gebauer & Meyer, 2003) and associated with ECM fungi, at least on some sites from Germany (Bidartondo et al., 2004). We use a French population (Dusak & Pernot, 2002) harbouring the highest proportion of albinos, to our knowledge, to first describe the albino phenotype (mycorrhizas, N and C sources), and second preliminary characterize of the C budget in green C. damasonium, compared with albinos. Although the reduced size of this population limits samplings, and therefore statistical analyses, albinos are sufficiently frequent to allow comparisons with green individuals. We analyzed first traits related to morphology and fertility; second chlorophyll contents; third in situ photosynthetic abilities; fourth fungal partners; and fifth N and C stable isotope abundances. Hypogeous individuals and ECM from surrounding trees were included in the analysis of fungal partner diversity to further characterize the C sources of this C. damasonium population.

Materials and Methods

Main study site and model species

The investigated population of Cephalanthera damasonium (Mill.) Druce encompasses albinos (referred to as ‘BA’, Fig. 1), green individuals (‘BG’) and hypogeous individuals (‘BH’). In this species, each shoot is from a single, independent root system, and likely to be from a single seed (= single genet), due to monopodial growth of the rhizome (Rasmussen, 1995). The population grows on a calcareous soil at Boigneville (Essonne, France; 2°21′ E, 48°21′ N; elevation 105 m) in a Quercus robur L. and Corylus avellana L. forest, with various other herbaceous plants and shrubs such as Crataegus laevigata L., Cornus sanguinea L., Clematis vitalba L. and Hedera helix L. The highest trees reach 6 m and the canopy is closed.


In June 2003, at orchid fruiting, we measured for all plants the shoot height, fruit and leaf number, length and width of each leaf, as well as the distance to the nearest green individual for each albino. For comparison, the same measurements taken from 32 randomly chosen individuals from another C. damasonium population (with green individuals only, referred to as ‘CG’) growing in similar conditions at Chauriat (Puy de Dôme, France; 3°17′ E, 45°46′ N; elevation 500 m). How early growth happened was estimated by measuring height during early shoot growth (on 12 May, at time of first leaf expansion for green individuals). Morphological data were submitted to mean comparison by anova, or to χ2 tests for proportion of fertile individuals. A Factorial Correspondence Analysis (FCA; Lebart et al., 1995) was conducted for shoot height, and length and width of the four largest leaves. Homogeneity of the distribution of individuals along the first factor was tested with a χ2 test, followed by a more detailed analysis of the repartition where local χ2-values were compared with χ2 with one degree of freedom (Cibois, 1984).

Gas exchanges and chlorophyll analysis

Net assimilation and stomatal conductance were measured on mature leaves in June, under controlled levels of photosynthetic active radiations (PAR). We used an infrared gas analyzer (LI 6400, Li-Cor, Inc., Lincoln, NB, USA) on five or six randomly selected leaves (for each leaf, the value is obtained as the average of five repetitions). Measurements were performed at PAR = 15 µmol m−2 s−1 (a value representative of the day light received by orchids, data not shown) and PAR = 1000 µmol m−2 s−1. For comparison, the latter measurement was repeated on two leaves of four orchid species growing on the same site (Ophrys insectifera L., Platanthera chlorantha Cust. ex Reichb., Listera ovata (L.) R.Br., Orchis purpurea Huds.) in May, that is at phenological status comparable with C. damasonium in June, since they flower earlier. Measurements of chlorophyll fluorescence of PSII (photosystem II) were conducted with a portable pulse amplitude modulation fluorimeter (PAM 2000, Walz, Germany) on six individuals (between two and five leaves per individual were measured) for each type of individuals. The quantum efficiency of PSII electron transport (ΔF : inline image) was calculated as (inline image − F) : inline image (Genty et al., 1989) where inline image is the maximal fluorescence during light saturation and F is the level of steady-state fluorescence recorded at ambient PAR (15 µmol m−2 s−1).

To analyze pigments, one leaf of four BA and four BG individuals was randomly sampled. Leaf pieces (9 cm2) were weighted and extracted in acetone after grinding in liquid nitrogen. Optical densities were measured with a Varian DMS90 spectrophotometer, and chlorophyll a and b concentrations were estimated after Ziegler & Egle (1965). Mean total chlorophyll concentrations of BG and BA individuals were tested for differences using a Wilcoxon rank test. For two green and four albinos, absorbance spectra were obtained on an Aminco DW2 spectrophotometer.

Molecular identification of C. damasonium root fungi

Two BG and two BA root systems were harvested in June 2003 (flowering stage) at more than 5 m from each other. During sampling of each BG individuals, a close (c. 30 cm) hypogeous individual (BH) was also found, likely representing a young plant. In addition, 71 ECM belonging to 24 different morphotypes were collected on tree roots growing less than 30 cm from a BG1 individual. All roots were washed, cut into 1.5-cm-long pieces (all colonized by fungi, as shown by their brownish colour) and frozen at −80°C. To ensure identification of BH, plant ITS were sequenced as in Selosse et al. (2004) and proved to be identical for all orchids (AY833027).

DNA extraction and PCR amplification of fungal ITS were performed, as in Selosse et al. (2002a) on 15 pieces (one per root at least + randomly chosen pieces) per individual, as well as on two ECM of each morphotype. Whenever unique ITS fragments were amplified, they were directly sequenced. To analyse fungal diversity in the remaining root pieces, a cloning was performed for each individual by pooling the PCR products of four randomly chosen pieces (i.e. the equivalent of 6 cm of roots). Cloning was performed as in Selosse et al. (2004) and at least 110 clones per individual were sequenced. Sequencing was conducted as in Selosse et al. (2002a); when many fragments of similar size occurred in a cloning, a RFLP was first performed as in Selosse et al. (2002b) and only 20 clones per RFLP type were sequenced. To further ensure absence of usual orchid symbionts with highly derived rDNA sequences, that is tulasnelloids and sebacinoids basidiomycetes, additional PCR amplifications were carried out using specific primers (ITS4tul and ITS3S, respectively: Selosse et al., 2004). Sequences (or consensus sequences for similar clones) were deposited in GenBank, and searches for similar sequences were conducted with BLAST (Altschul et al., 1997) as in Selosse et al. (2004).

In order to test for a sampling effect, a replicate cloning was performed for BA1 using four different root pieces. Total number, n, of fungal sequences within a root system can be estimated as follows: if k sequences have been found in the first cloning, and p sequences in the second cloning, with q new sequences compared with the first one, then the probability of finding a new species in a cloning is k : n or q : p, and thus n = kp : q. To compare the diversity of their fungal partners, BA, BG and BH individuals were described on the basis of their associated fungi, using a FCA followed by a Hierarchical Ascendant Classification (HAC) considering the presence or absence of each fungal sequence (as frequencies in each cloning may reflect only experimental biases). In addition, logistic regression, a method allowing quantitative comparison of binary data, was used to analyze the presence or absence of fungal species, trophic types or fungal classes (Asco-/Basidiomycetes) and to plant phenotype. Significances of these variables were tested with a χ2 test.

To identify the fungi forming mycorrhizal pelotons within roots, pieces not used for molecular analysis were submitted first to transmission electron microscopy analysis, as in Selosse et al. (2004), or second to isolation of pelotons by microdissection (Rasmussen, 1995). Pelotons were isolated from 10 root sections per plant. For each section, 15–20 pelotons were recovered and pooled (i.e. 150 samples), and then submitted for molecular analysis as above.

Isotope analysis

One leaf was collected from each of 10 albinos and 10 green C. damasonium individuals in July 2003, and leaves of several surrounding plants were sampled at the same height for comparison. All terrestrial fungal fruitbodies found on the site were collected between September and November 2003, and fungi belonging to genera present in C. damasonium roots were submitted to ITS analysis as previously described to assess their identity. All samples were predried for 3 days at 55°C and transported to the isotope laboratory. Subsequently, the material was handled as in Bidartondo et al. (2004) to measure total N concentrations and relative isotopic abundances. These abundances are denoted as δ-values: δ15N or δ13C = (Rsample : Rstandard− 1) × 1000 [‰] where Rsample and Rstandard are the ratios of heavy isotope to light isotope of the samples and the respective standard (Bidartondo et al., 2004). For species with more than six repetitions, δ13C and δ15N were submitted to mean comparison by anova (and posthoc LSD test) with α type I error fixed at 1%.

Statistics and data analysis

All statistical tests were conducted using the ‘R’ software package (Ihaka & Gentleman, 1996) with the α type I error fixed at 5% (unless otherwise stated), while the FCA and the HAC were performed with Bi (©logINSERM, Paris). FCA allows analysis of complex data sets by using linear combinations of variables and thus reducing the redundancy of information; it is based on χ2 distances and allows comparison of qualitative variables. HAC is a classification method that produces a dendrogram by iterative grouping of the two closest elements (considering χ2 distances), where elements are individuals or sets of individuals previously grouped together. Whenever the statistical tests required the normality of variables, this was previously tested using a Kolmogorov-Smirnov test. Unless otherwise stated, the values plotted on graphs are mean ± standard error (se).


Pigment and gas exchange analyses

In 2003, the Boigneville population harboured 15 albino and 16 green C. damasonium individuals. Chlorophyll concentration was 100 times higher in green than in albino individuals (Table 1): by comparison, a lilac (Syringa vulgaris) leaf had a content of 2.5 µg mg−1 f. wt, that is similar to green individuals. Absorbance spectra (at 650–660 nm, Fig. 2a) and fluorescence tests (not shown) supported the presence of chlorophyll traces in albinos. However, the albino spectrum is dominated by absorption of carotenoids (at 400–600 nm, Fig. 2a), whose concentrations were four times higher in albino than in green individuals (not shown).

Table 1.  Comparison of green, albino and hypogeous Cephalanthera damasonium individuals for photosynthetic abilities, early growth and fungal diversity, as estimated by ITS sequencing on isolated pelotons or after cloning
 Green individualsAlbino individualsHypogeous individuals
  • Mean ± SD; values followed by different letters differ significantly between green and albino individuals according to a Student test, P < 0.05.

  • *

    measured at PAR = 15 µmol m−2 s−1 (at 20°C and 40% air humidity), representative of the site conditions.

  • **

    see Table 2 for detailed report.

Total chlorophyll (µg mg−1 f. wt)3.15 ± 0.50 (n = 4) a0.029 ± 0.004 (n = 4) b
Net assimilation (µmol CO2 m−2 s−1)*−0.10 ± 0.35 (n = 5) a−0.65 ± 0.39 (n = 6) a
Stomatal conductance (mol H2O m−2 s−1)*0.017 ± 0.003 (n = 5) a0.036 ± 0.019 (n = 6) b
Quantum efficiency of PSII (ΔF/inline image)*0.728 ± 0.052 (n = 6) a0.053 ± 0.041 (n = 6) b
Early growth (cm on 12 May 2004)14.5 ± 1.9 (n = 9) a9.2 ± 0.3 (n = 5) b
Fungi identified in isolated pelotons:Hymenogaster sp. #1Cortinariaceae sp. #1Hymenogaster sp. #1
Thelephoraceae sp. #2Thelephoraceae sp. #2 
Thelephoraceae sp. #3Thelephoraceae sp. #3 
Thelephoraceae sp. #1  
Fungi identified by cloning**:
  ECM or mycorrhizal484
  ECM Thelephoraceae252
  other ECM fungi222
  Ceratobasidium sp.1
 Fungi specific to the phenotype3 out of 1312 out of 215 out of 12
 Basidiomycetes specific to the phenotype1 out of 46 out of 82 out of 4
Figure 2.

Photosynthetic abilities of green and albino Cephalanthera damasonium. (a) Absorbance spectra of C. damasonium. Black line, green individuals (mean of two individuals); grey line, albinos (mean of four individuals; for clarity, scale for this spectrum was increased by 10-fold); dotted zones indicate se. (b) In situ net CO2 exchange (µmol m−2 s−1) as a function of level of photosynthetic active radiations (PAR) for an albino (open circle) and a green individual (closed circle).

In situ measurements (Fig. 2b) showed a classic assimilation response to light for green leaves and a net respiration, with gas exchange unresponsive to light, for albinos. Fluorescence measurements (Table 1) confirmed that a functional PSII apparatus existed in green individuals, but was lacking in albinos. At PAR representative of the site, the net leaf gas exchange was higher (although not significantly) for the green than for the albinos (Table 1), but green individuals remained near the compensation point. Stomatal conductance in albinos is twice as high as in green individuals (Table 1), implying a higher transpiration. The maximal measured net assimilation at saturation light (PAR = 1000 µmol m−2 s−1) was 5 µmol m−2 s−1 for green individuals, that is somewhat lower than measured for four orchid species from the same site (between 6 and 8 µmol m−2 s−1, data not shown). Green individuals had thus effective, but limited, photosynthetic abilities, while albinos had not.

Morphological analysis

Albino (BA) plants had, on average, slightly but not significantly, lower size and fertility than green individuals (BG –Fig. 3). Only the largest leaf was significantly shorter for albinos (85% shorter, P < 0.05), but this may be of little relevance for nonphotosynthetic plants. The Chauriat green individuals (CG) used for comparison showed some significant differences from BA, and even more from BG individuals (Fig. 3), suggesting that morphological polymorphism exists within this species, and that BA plants development is not strongly affected.

Figure 3.

Morphological comparisons of Chauriat green individuals (grey bars) with Boigneville albino (open bars) and green (filled bars) individuals (mean ± se, populations with different letters significantly differ according to a Student test or a χ2 test, P < 0.05). Lmax and lmax: length and width of the largest leaf.

As we ignore which parameters fairly characterize plant development, a FCA was conducted. The first (F1) and second (F2) factors represented much of the total variability (20.6% and 10.4%, respectively, Fig. 4); in the first factorial plane, parameters had a parabolic distribution (Gutman effect) indicating that the F1 axis bears most of the information. BA, BG and CG individuals were evenly and widely distributed when plotted on the first factorial plane (Fig. 4). The number of individuals in each of sectors A to C (Fig. 4) was compared with the number expected assuming a random repartition, by comparing the local χ2 in each sector with one degree of freedom χ2 (not shown): no significant trend was observed for BA individuals, whereas BG and CG individuals were significantly over-represented in sector A and BG individuals were significantly under-represented in sector D. We suspected that the heterogeneous distribution along F1 occurred because different plant sizes coexisted among BA, BG and CG individuals. We therefore distinguished individuals shorter or taller than 12 cm: on this basis, BAs were more often small (seven out of 15) than BG (two out of 15). Local χ2 tests (not shown) support the prior hypothesis that large Boigneville individuals dominate in sector A and small individuals dominate in sector D (where BG are under-represented, Fig. 4). This repartition explains an important part of the heterogeneity (Fig. 4) and also applies to CG individuals. Altogether, this shows that first the Boigneville population (BA + BG) does not strongly differ from the Chauriat population (CG); second individuals of different size coexist within BA and BG individuals; and third albinos tend to be less developed.

Figure 4.

First factorial plane (F1 x F2) of the FCA of BG (black symbols), BA (white symbols) and CG (grey symbols) morphological development. Symbols indicate individuals smaller (circles) or larger (squares) than 12 cm in height. The four delineated sectors A, B, C and D are the quartiles used for testing repartition of the various phenotypes.

Albinos were observed between 10 cm and 12 m away from the nearest BG. Their development, estimated by shoot size or value on F1 axis previously defined in FCA, was independent of the distance to the nearest BG (no significant correlation was shown using a Pearson correlation test, P > 0.05, not shown). BG appeared earlier and developed faster (Table 1) than BA individuals. In addition, BA individuals dry sooner than green individuals: in 2003 and 2004, all albino shoots were dry by 25 July, whereas all BGs were still green with fleshy fruits.

Fungal diversity

Fungal ITS were amplified from 87 of the 90 root pieces, suggesting a homogeneous colonization (as seen under light microscope, not shown), and often consisted in multiple PCR fragments. For the 19 products that yielded a single fragment, sequencing produced seven divergent fungal sequences (Table 2). After cloning of the other products, 24 additional sequences were found together with all the seven previous sequences (Table 2). All sequenced fungi belonged to asco- and basidiomycetes, and some taxa were over-represented, namely Thelephoraceae, Helotiales, and other Pezizomycetes (including Pezizales, Tuber, Phialophora, Leptodontidium and Geopyxis spp.) and Cortinariaceae (including Hymenogaster) (Table 2). Primers designed for amplification of tulasnelloids and sebacinoids ITS (see Materials and Methods) did not detect additional fungi.

Table 2.  Fungi found in Cephalanthera damasonium roots and on ECM (underlined) collected at Boigneville (France) near BG1
GenBank accession NumberTentative IdentificationPutative ecology+*Closest matches found by BLAST analysis at the NCBI homepage (with BLAST expected value)Cloning frequency in % (and nb of direct sequencing)°:
  • +

    Trophic type of the closest relatives (we assume that ecology is similar for the sequenced species): ECM, ectomycorrhizal; M, mycorrhizal but not necessarily ECM; S, saprobes; PE, plant endophyte or parasite.

  • *

    Sequence also found on ECM surrounding BH1 (that is, near BH1): note that the same sequences were amplified from BH1 or BG1.

  • The BLAST expected value (in parentheses) represents the number of sequence matches expected by random chance (the smaller the value, the better the match between our sample sequences and those in the NCBI database).

  • °

    In parentheses, number of occurrences of the sequence by direct amplification from pieces of the same root system.

  • I

    ·Two independent clonings, using different root pieces, were performed on the root system of individual BA1.

  • °°

    Although some Sebacina spp. belong to the Rhizoctonia, this ITS sequence, that is 100% identical to an ITS found on Neottia nidus-avis roots (Selosse et al., 2002a), is related to ECM-forming sebacinoids, but not to sebacinoids belonging to rhizoctonias from green orchids (Weißet al., 2004).

AY833028Helotiales #1M?AY578281 Uncultured Phialophora (0.0)
AF081442 Mycorrhizal sp. (0.0)
 63  7 
AY833029Helotiales#2ECM*AJ430405 Axenic ECM isolate (0.0)
AY345353 ECM isolate (Helotiales) (0.0)
AY833030Helotiales#3ECM*AF269068 Ericoid mycorrhizal sp. (0.0)
AF502889 Leaf litter ascomycete (2.0e-83)
AY833031Helotiales #4ECM?AJ430227 ECM isolate (Pyrenopeziza) (5.0e-68)
AJ430226 Grass root isolate (Pyrenopeziza) (5.0e-68)
AY833032Helotiales #5S? ECM?AY578277Phialophora from Cypripedium (0.0)
AF335454Bisporella citrina (1.0e-127)
AY833033Pezizales #1ECMAJ534699 Pezizales sp. (1.0e-75)
AY351626 ECM (Pezizales) 2.0e-74
AY833034Pezizomycetes #1?AY634148Nalanthamala vermoesenii (2.0e-92)
AY554216Nalanthamala vermoesenii (2.0e-92)
AY833035Pezizomycetes #2?AY634148 Mycorrhizal ascomycete (0.0)
AJ293878Phomopsis quercella (0.0)
   1  5
AY833036Pezizomycetes #3?AF284133 Salal root associate (1.0e-115)
AF383954Lophiostoma vagabundum (1.0e-115)
AY833037Geopyxis sp. #1ECMAY465441Geopyxis sp. (1.0e-111)
Z96992Geopyxis sp. (1.0e-111)
2313 776
AY833038Phialophora sp.ECM*U31845Phialophora sp. (1.0e-137)
AF083206Phialophora lignicola (1.0e-134)
7  228 
AY833039Leptodontidium sp. #1PEAF486133Leptodontidium orchidicola (0.0)
AF168783 Dark septate endophyte (0.0)
3 4 1  
AY833040Tuber excavatum#1ECMAY286191 Uncultured ECM (Tuberaceae) (0.0)
AF073509Tuber excavatum (0.0)
AY833041Tuber uncinatum#1ECMAF516792Tuber aestivum (0.0)
AF51679Tuber aestivum (0.0)
AY833042Exophiala sp. #1S?AF050274Exophiala salmonis (0.0)
AF050272Exophiala pisciphila (0.0)
AY833043Sordariales #1S?AF177155Cercophora appalachianensis (1.0e-158)
AY587911Cercophora areolata (1.0e-155)
AY833044Nectriaceae #1PE?AY295332Cylindrocarpon sp. (0.0)
AJ279482Cylindrocarpon sp. (0.0)
AY833045Nectriaceae #2PEAY295332Cylindrocarpon sp. (0.0)
AJ007353Nectria radicicola (0.0)
AY833046Verticillium sp. #1PEAF504848 Uncultured soil fungus (0.0)
AY805596Verticillium sp. (0.0)
AY833047Ceratobasidium sp. #1PEAY634128 ECM (Ceratobasidiaceae) (0.0)
AJ242901Rhizoctonia sp. (0.0)
AF440655°°Sebacinaceae #1°°ECMAF440655Sebacina endomycorrhiza in Neottia (0.0)   4   
AY833048Cortinariaceae sp. #1ECMAF495459 ECM of Kobresia (5.0e-89)
AY310821 Uncultured ECM fungus (1.0e-86)
AY833049Hymenogaster sp. #1ECM*AY351629 ECM (Hymenogastraceae) (0.0)
AY634136 ECM (Hymenogastraceae) (0.0)
  3 16
AY833050Hymenogaster sp. #2ECMAF325637Hymenogaster populetorum (0.0)
AY351629 ECM (Hymenogastraceae) (0.0)
AY833051Thelephoraceae sp. #1ECMAJ510270 ECM (Thelephoraceae) (0.0)
AF272913Tomentella ellisii (0.0)
AY833052Thelephoraceae sp. #2ECMAJ581550Tomentella sp. (0.0)
AJ581535Tomentella sp. (0.0)
AY833053Thelephoraceae sp. #3ECMAJ581550Tomentella sp. (0.0)
AJ581552Tomentella sp. (0.0)
AY833054Thelephoraceae sp. #4ECMAJ510270 ECM (Thelephoraceae) (0.0)
AJ581550Tomentella sp. (0.0)
AY833055Thelephoraceae sp. #5ECMAJ510270 ECM (Thelephoraceae) (0.0)
AJ581550Tomentella sp. (0.0)
AY833056Thelephoraceae sp. #6ECMAJ510270 ECM (Thelephoraceae) (0.0)
AF272913Tomentella ellisii (0.0)
AY833057Thelephoraceae sp. #7ECMAJ421255Tomentella galzinii (0.0)
AJ421251Tomentella galzini (0.0)
 Number of clones successfully sequenced:1028411295118109103
Total number of fungal sequences:99149693

Assuming that they share the same ecology as their closest GenBank relatives, at least 18 out of 31 sequences found are from ECM fungi (Table 1); others are from plant endophytes or parasites, and possibly saprobes. Indeed, by investigating 71 ECMs sampled near BG1 and BH1 individuals, five fungi occurring in these two orchids were proved to form ECM on surrounding trees: three ascomycetes, Hymenogaster sp. #1 and Thelephoraceae sp. #3 (Table 2).

Electron microscopy showed that all pelotons, the typical mycorrhizal hyphae in orchids, exhibited intercellular dolipores characteristic of basidiomycetes (not shown). These perforate dolipores differed from imperforate dolipores found in rhizoctonias. Furthermore, ITS sequencing was successful for 32 out of the 150 investigated pools of mycorrhizal pelotons and produced sequences already found in clonings and beloning to basidiomycetes (Table 1). Basidiomycetes are therefore the dominant mycorrhizal fungi in C. damasonium roots.

Comparison of fungal diversity between phenotypes

The low success of pelotons typing does not allow comparison of BA, BG, and BH individuals (Table 1), so that we used the data from cloning, considering only the presence/absence information since quantitative differences may reflect only cloning biases. Many fungi were shared between orchid individuals, since 14 out of the 31 fungal sequences were found on more than one individual. Only two sequences were restricted to the two BA individuals and none to the two BHs (Table 2). Furthermore, the FCA and Hierarchical Ascendant Classification (HAC, see Materials and Methods) of BA, BG and BH individuals did not show clear differentiation between phenotypes (Fig. 5). In the three first factorial planes (65.5% of the total variability), the position of individuals was not independent of their fungal partners (χ2 test; P < 0.05), but was first independent of their phenotype, as supported by the absence of clustering of identical phenotypes in HAC, and second independent of distance between roots systems, as contiguous individuals (BG1 + BH1 and BG2 + BH2) did not cluster together. Besides, the groups obtained by HAC did not preferentially associate with fungi of a given ecology, as shown by their repartition on the three first factorial planes (Fig. 5). Furthermore, logistic regression shows that the number of associated fungal species is not significantly different for the three orchid phenotypes (P > 0.10). But restricting analysis to the basidiomycetes carries out new results: the number of associated fungal species depends on orchid phenotype (P < 0.05). Nevertheless, associated fungal species, trophic types, and classes (Ascomycete/Basidiomycete) do not significantly differ between orchid phenotypes (P > 0.14, P > 0.31, P > 0.22, respectively), considering the entire dataset or restricting it to Basidiomycetes. Hence, our analysis failed to detect any difference in specificity for fungal partners between orchid phenotypes, but showed that albinos have significantly more diverse associated basidiomycetes than green and hypogeous individuals.

Figure 5.

Classification of Boigneville individuals on the basis of their fungal partners, plotted on the F2 × F3 factorial plane (as the interpretation is similar for the three first factorial planes, only this more readable plane is plotted here). Crosses: fungi of different putative ecology (+, ectomycorrhizal; x, plant endophytic; grey +, saprophytic; grey x, unknown; some fungi are overlapping on this graph). Circles: orchid phenotypes (black, BG; white, BA; grey, BH; BA1 was plotted by two points grouped by a dotted circle, according to two independent cloning replicates). The dendrogram summarizes the HAC analysis, with vertical lengths of the branches proportional to χ2 distances.

To test for the accuracy of fungal detection, a replicate cloning was performed for BA1 using four different roots. Six sequences were common to both clonings (representing > 85% of the clones in each case; Table 2), whereas three different sequences occurred at low frequency in each cloning. The total number of sequences in this root system can be estimated to be n = 27 (see Materials and Methods), indicating that one-third (nine of 27) of all sequences from a root system was found in each cloning. In HAC, positions of BA1 by the two clonings closely clustered together (Fig. 5), so that other clonings would increase description of the fungal diversity without adding to phenotype comparison.

C and N isotopes of plants and fungi

All nonorchid plants from the site had δ13C typical of C3 photosynthesis of forest ground vegetation (Fig. 6). However, C. damasonium had significantly higher values, both in leaves and roots, with BA leaves having the highest 13C abundance (Fig. 6). C. damasoniumδ13C would be typical for C3 vegetation from sunny, dry ecosystems, suggesting that they instead exploit a different C source. Few fungal fruitbodies were found at Boigneville, including only two ECM species: Inocybe tenebrosa and an unidentified Thelephora sp. identical in ITS to Thelephora sp. #1 (Table 2) found in roots of three orchids. Several saprophytic Bisporella citrina fruitbodies were collected, a species belonging to Helotiales, but that differed in ITS (AY833058) from all helotiales from orchid roots (Table 2). All fruitbodies had δ13C values similar to orchid roots (Fig. 6). Albinos and, to a lesser extent, green individuals thus use a C source distinct from C3 photosynthesis, probably of fungal origin.

Figure 6.

Values of δ13C and δ15N for Boigneville plants and fungi (mean ± se, values followed by different letters differ significantly, P < 0.01; values for other taxa were not tested, due to the low number of replicates). Phaeom., Phaeomarasmius; Polyg., Polygonatum.

Total N concentrations were significantly higher in C. damasonium than in surrounding plants, suggesting a better N nutrition, and were significantly higher for albinos than for green individuals (Table 3). Fungal values were intermediate between C. damasonium and other plants (Table 3). C. damasoniumδ15N values significantly differed from surrounding plants (Fig. 6), supporting a different N source for this orchid. Among fruitbodies, most had δ15N values similar to green plants, with the exception of Thelephora sp. #1, which had values closer to C. damasonium. Based on δ13C and δ15N values, at least Thelephora sp. #1 is a plausible C and N source for this orchid.

Table 3.  Fungal and plant total nitrogen concentration at Boigneville (France)
SampleTotal N concentration (mmol g−1 d. wt)
  • Mean ± SD; values followed by different letters differ significantly according to an anova, P < 0.05.

  • *

    see Fig. 6 for species names.

Albino Cephalanthera damasonium3.98 ± 0.07 (n = 10) a
Green C. damasonium3.26 ± 0.08 (n = 10) b
Fungi2.67 ± 0.13 (n = 20) c
Nonorchid green plants*1.57 ± 0.06 (n = 59) d


The investigated C. damasonium population encompasses two phenotypes: while green individuals have functional photosynthetic apparatus, albinos have 100 times less chlorophyll and are nonphotosynthetic (Figs 1 and 2, Table 1). Morphological differences between phenotypes are limited, although albinos tend to be more often less well developed (Figs 3 and 4). Inter-population comparison with Chauriat green individuals (Fig. 3) also supports the suggestion that albinos do not undergo drastic phenotypic alterations. Morphological polymorphism exists in the investigated populations (Fig. 4), likely due to co-occurrence of several developmental states, but is independent of phenotypes. In spite of differences between individuals, possibly because of spatial structure of the fungal community, no significant difference in fungal partners was detected among phenotypes (Fig. 5; Table 2). We face, of course, statistical limitations inherent to analysis of rare and protected plants such as albinos, implying first a low plant number in the population, second a limited root sampling, and third unavailability of a population replicate. However, our data suggest that achlorophylly in C. damasonium is not linked to major symbiotic or morphological change.

Regarding C metabolism, our data demonstrate that first albinos are MH plants, and second green individuals are mixotrophic. C. damasonium is significantly more enriched in 13C than nonorchids growing in close proximity (Fig. 6). The latter are heavily 13C-depleted, as expected for understorey C3 plants because low light increases 13C discrimination (Gebauer & Schulze, 1991; Högberg et al., 1999) and contribution of CO2 from soil respiration reduces 13C availability (Schleser & Jayasekera, 1985). Albinos show significant 13C enrichment compared with nonorchid, green plants (+9.1‰ on average), similarly to MHP species in other studies (Gebauer & Meyer, 2003: +8.4‰; Trudell et al., 2003: c. +7‰). Albinos are unlikely to have been photosynthetic in the previous years, because this would not explain while they are more 13C-enriched than green individuals (Fig. 6) and no phenotype switch was seen over 3 yr (not shown). Albino δ13C are in the range of fungal values (Fig. 6) that are less depleted in 13C than autotrophic plants, as to be expected (Högberg et al., 1999; Wallander et al., 2004). Albinos are thus true MH plants, feeding on fungal C. Green C. damasonium have intermediate values between autotrophic C3 nonorchids and MH plants, as already shown for this species on other sites (Gebauer & Meyer, 2003; Bidartondo et al., 2004). Beyond this, our data demonstrate that green individuals are photosynthetic (Fig. 2b) and allow calculation of their mixotrophy level by using for the first time a conspecific, and thus more realistic, MH plant as a reference (the albinos): green individuals derive 48.7% of their C from a fungal source, according to a two source linear mixing model (i.e. an albino-like MH source and a C3 photosynthesis; Gebauer & Meyer, 2003). This is in the range reported for other mixotrophs: hemiparasitic, xylem-tapping plants, such as mistletoes, derive up to 63% of their C from their host (Schulze et al., 1991; Marshall et al., 1994; Bannister & Strong, 2001), while the green alga Chlamydomonas reinhardtii grown on acetate derive 50% of its C from this source (Heifetz et al., 2000). Gas exchanges further support mixotrophy: beyond isotopic ratios, characterizing anabolic C, instantaneous CO2 exchanges show that photosynthesis does not compensate for catabolic C loss at PAR representative of the site (Table 1). However, an exact C budget remains to be determined, deserving in situ measures over the whole growing season to know whether fungal C is exploited continuously, or preferentially at some developmental stages.

N composition also supports the suggestion that C. damasonium feed on fungi since δ15N abundance is similar to MH plants in other studies (Gebauer & Meyer, 2003; Trudell et al., 2003) and differs from nonorchid plants (Fig. 6). Fungal δ15N values are unexpectedly low at Boigneville (Fig. 6), except for Thelephora sp. #1, which better fits fungal δ15N from most other investigated sites (Gebauer & Taylor, 1999; Hobbie et al., 2001; Trudell et al., 2003; but see Wallander et al., 2004). Interestingly, this species, which forms pelotons in C. damasonium roots (Table 1), has δ15N values similar to C. damasonium, and could be one of the N and C sources for the orchid. The hypogeous-fruiting Hymenogaster species (Cortinariaceae) that occur as unique basidiomycete in BH2 (Table 2) and in other C. damasonium plants from French populations (our unpublished data) are also likely to provide C to this orchid. The other ECM species, Inocybe tenebrosa, and saprophytic fungi have the same δ15N, differing from C. damasonium (Fig. 6). Indeed, δ15N values are not intrinsically different for ECM and saprobic fungi, but depend on the N source exploited, for example humus or mineral solutes (Gebauer & Taylor, 1999). Inocybe tenebrosa and the saprophytes are thus probably not N and C sources for C. damasonium. The fact that albinos have higher N concentration than fungi (Table 3) is also congruent with a fungal source for albino biomass: respiration (and thus CO2 loss) increases the N : C ratio in albinos compared with their source of organic matter. This statement obviously explains the observation (Table 3) that albinos, being fully mycoheterotrophic, have a higher N concentration than green mixotrophs, which also use their own photosynthetic C for respiration.

Our analysis of total fungal diversity allows further discussion of C-providing fungi (Tables 1 and 2). Identified fungi fall into three groups: first, ECM Thelephoraceae and Cortinariaceae are mycorrhizal on C. damasonium, likely furnishing C. These taxa were also found on specimens from Germany (Bidartondo et al., 2004), and their mycorrhizal state is additionally supported here by microscopy investigations. A second group consists of ascomycetes that belong to mycorrhizal taxa, including ECM species (such as Pezizales), root biotrophs and mycorrhizal species (such as Helotiales; Vrålstad, 2004) or ‘dark septate root endophytes’ (such as Phialophora and Exophiala spp.; Jumpponen & Trappe, 1998). These ascomycetes could be intercellular endophytes or rhizoplan colonizers, and their role for C. damasonium deserves further attention. This mirrors the case of Epipactis microphylla, where ECM basidiomycetes were found by ITS sequencing, but not by inspecting pelotons (Selosse et al., 2004). A third group encompasses ascomycetes that could be expected in any plant organ, either as endophytes (e.g. Leptodontidium orchidicola, a common orchid endophyte –Rasmussen, 1995), or as parasites (e.g. Nectriaceae or Verticillium spp.), or even as decaying fungi (e.g. Sordariales and maybe Exophiala sp.): indeed, related taxa were reported from nonmycorrhizal Neottieae roots (Bidartondo et al., 2004). Interestingly, isolation experiments already demonstrated the presence of some of the reported ascomycetes in green, putatively autotrophic orchid roots, for example Phialophora spp. (Currah et al., 1988), Nectriaceae, Verticillium and Phomopsis spp. (Richardson & Currah, 1995). This suggests that the amplified ascomycetes sequences do correspond to intraradical fungi and may not be linked to MH or mixotrophic strategies in C. damasonium. However, we cannot formally exclude a role in nutrition for fungi other than ECM basidiomycetes because first the low success of pelotons typing (21.3%) does not rule out the possibility that they form rare pelotons, and second C. damasonium individuals without pelotons exist at some period of the year (Bidartondo et al., 2004; our own observations). To summarize, our samples show a diversity of root fungi expected for mixotrophic orchids (Bidartondo et al., 2004; Selosse et al., 2004), with at least ECM basidiomycetes from a limited number of clades as mycorrhizal fungi (mycorrhizal preference) and almost no rhizoctonias, the usual mycorrhizal partners on nonmixotrophic orchids (Rasmussen, 2002). The sebacinoid sequence found (Table 2) belongs to ECM sebacinoids that are only distantly related to sebacinoid rhizoctonias (Selosse et al., 2002a; Weißet al., 2004), so that the only true rhizoctonia found is Ceratobasidium sp. #1: a similar sequence was reported from the mixotrophic Neottieae Epipactis helleborine (Bidartondo et al., 2004), suggesting that rhizoctonias are available in these forests sites and that they can colonize mixotrophic orchids.

Since, as mentioned previously, we can not reject their biological relevance, all cloned sequences were included in our FCA comparing the various orchids phenotypes (Fig. 5), as well as in our logistic regression: our analysis failed to detect any obvious difference in fungal diversity between phenotypes. This also applies when comparison is restricted to putative mycorrhizal basidiomycetes: both involved families (Thelephoraceae and Cortinariaceae) are present in all phenotypes, as shown by cloning (Table 2) and peloton analysis (Table 1; with the exception of Thelephoraceae in hypogeous individuals). Albinos show a trend to larger symbiont number and higher specificity (Table 1), but analysis by logistic regression failed to show any specificity of a given phenotype, even if only Basidiomycetes are taken into account.

Because ECM fungi dominate in C. damasonium roots, surrounding trees are the most likely ultimate C source. We provide here additional evidence of a fungal link to trees, since at least five fungi simultaneously colonized BH1 and BG1 individuals and formed ECM on trees surrounding these individuals (Table 2). A similar link to trees was already demonstrated for albino and green Epipactis microphylla (Selosse et al., 2004). One could argue that green individuals could also be a C source, since they share fungal taxa with albino and hypogeous individuals (eight and six spp., respectively; four spp. are common to all phenotypes; Table 2). But this seems unlikely to us, because first green individual are near compensation point for photosynthesis, second some albinos grow at distance from green individuals (up to 12 m, i.e. more than usual genet size of ECM species; Selosse, 2001), and third albino development is not correlated with distance to green individuals. In addition, orchid digestion of fungal pelotons may also preclude substantial C uptake by the fungi in green individuals. This link to trees has strong implications for decisions about stand management, since trees are therefore crucial for C. damasonium survival, whereas their development can harm other sun-requiring orchid species growing at the same place.

Albinos represent a hitherto poorly analyzed step in evolutionary emergence of MH plants among mixotrophic species. Their genetic determinism remains to be investigated, for example by germination of seeds: albinos are either mutants, as suggested by the stability of their phenotype over the years, or phenocopies, in which photosynthesis genes can mutate without counter-selection. However, investigated albinos are often less developed and less fertile (Figs 3 and 4), although fruit number (Fig. 3) and seed content (not shown) does not significantly differ from green individuals. Albino Epipactis helleborine investigated by Salmia (1989) showed a similar trend to a lower vegetative success. This may reduce albino fitness, so that loss of photosynthesis could be a limiting step in evolution to MH strategy: indeed, most of the time, albino orchids are rare and do not invade the populations where they occur (Renner, 1938; Mairold & Weber, 1950), perhaps being evolutionary dead ends. We propose two nonexclusive explanations for this. First, investigated albinos may be less specific than MH species (Taylor et al., 2002, e.g. the related MH Cephalanthera austiniae is exclusively associated with Thelephoraceae, Taylor & Bruns, 1997). Specificity may be linked, for unknown reasons, to improved exploitation of fungal C, so that the investigated albinos are somewhat disfavoured. However, albinos in Epipactis microphylla seem quite specific to Tuber spp. (Selosse et al., 2004), but do not invade their population. Roles of specificity in MH strategy thus remain unclear, but could act here to limit albinos access to fungal C. Second, C. damasonium albino shoots have a shorter lifespan than green ones: they appear later, probably because their C is available only after plants have developed their leaves and transferred C to their mycorrhizal fungi (Lerat et al., 2002); they dry earlier, maybe because of higher transpiration (Table 1). In this context, it is noteworthy that stomata that are mainly required for CO2 exchanges, are lost in most MH plants (Leake et al., 2004). In C. damasonium, retention of features related to the photosynthetic state, such as flowering phenology and/or stomata, may limit albino success. In other words, a good fitness for MH mutants in mixotrophic species would require several changes at the same time – for example several mutations – that rarely occur simultaneously. Whereas a shift to mixotrophy can be progressive by adding ECM fungi to rhizoctonias (as suggested by the persistence of rare rhizoctonias in C. damasonium, Table 2; Bidartondo et al., 2004), loss of photosynthesis may be a discontinuous step, rarely successful in mixotrophic species.

Further fitness comparisons between albinos and green individuals, for example comparative germination, are obviously needed. If confirmed, a reduced albino fitness could limit appearance of pure C sinks in mycorrhizal networks, especially among herbaceous plants in forests where low light strongly selects for alternative C source. This could contribute to evolutionary stability of mycorrhizal networks, by limiting the frequence of cheater appearance. In addition, Neottieae provide tractable models to study in situ mixotrophic C budget and C flow through mycorrhizal networks. Although several other green orchid taxa may be mixotrophic, the relatively frequent occurrence of albinos in Neottieae will certainly help in addressing fascinating questions on the ecophysiology and evolution of mycorrhizal networks.


We thank F. Dusak and P. Pernot for access to the C. damasonium population, and P. Bonfante and A. Faccio (CNR Torino, Italy) for help in transmission electron microscopy. We also thank P.-A. Moreau, C. Delaruelle, J.-C. Thomas and T. Tully for help in experimental analyses and A. E. Douglas and R. Bateman for valuable comments on the manuscript. We are grateful to J.-Y. Bansard for help in statistical analysis of our data. The research was funded by the Société Française d’Orchidophilie, M.-A. Selosse's own funds and the IFR CNRS 101.