1462 longifolia (, ) is mixotrophic: a comparative study between green and nonphotosynthetic individuals

Jean-Claude Abadie, U¨ lle Pu¨ ttsepp, Gerhard Gebauer, Antonella Faccio, Paola Bonfante, and Marc-Andre´ Selosse

Abstract: We investigated an Estonian population of the orchid (L.) Fritsch. (Neottieae tribe), which harbours green and achlorophyllous individuals (= albinos), to understand albino survival and compare mycorrhizal associates, development, and nutrition of the two phenotypes. Albinos never changed phenotype over 14 years and had de- velopment similar to green individuals; their chlorophyll content was reduced by 99.4%, making them heterotrophic. Mo- lecular typing by polymerase chain reaction amplification of fungal intergenic transcribed spacer and microscopic analyses showed that (Basidiomycetes, usually forming ectomycorrhizae with trees) were mycorrhizal on both phe- notypes. Molecular typing also demonstrated that additional fungi were present on roots, including many endophytes (such as ) and various ectomycorrhizal taxa, whose role and pattern of colonization remained unclear. Mycorrhizal col- onization was increased in albinos by about twofold, but no obvious difference in fungal partners compared with green in- dividuals was demonstrated. Analysis of stable isotope composition (N and C) showed that albinos were dependent on their fungi for carbon (mycoheterotrophy), while green individuals recovered 33% of their carbon from fungi (mixotrophy). Surrounding trees, which formed ectomycorrhizae with at least one Thelephoraceae found in orchids, were likely the ulti- mate carbon source. These data are discussed in the framework of evolution of mycoheterotrophy in orchids, especially in Neottieae. Key words: Cephalanthera, mixotrophy, mycoheterotrophy, mycorrhizae, stable isotopes, thelephoraceae. Re´sume´ : Nous avons e´tudie´ une population estonienne de l’orchide´e Cephalanthera longifolia (L.) Fritsch. (tribu des Neottieae) comprenant des individus chlorophylliens et non chlorophylliens (= albinos), afin de comprendre la survie de ces derniers et de comparer les champignons mycorhiziens ainsi que le de´veloppement et la nutrition des deux phe´notypes. Les albinos n’ont pas varie´ de phe´notype en 14 ans et ont un de´veloppement similaire aux individus verts; leur teneur en chlorophylle est re´duite de 99,4 % — ils sont donc he´te´rotrophes. Le typage mole´culaire par amplification par re´action de polyme´risation en chaıˆne de l’espaceur interge´nique transcrit fongique et l’analyse au microscope sugge`rent que des Thele- phoraceae (Basidiomyce`tes habituellement ectomycorhiziens des arbres) mycorhizent les deux phe´notypes. Le typage mo- le´culaire de´tecte aussi d’autres champignons sur les racines, dont des endophytes (par exemple des He´lotiales) et d’autres taxons ectomycorhiziens, dont le roˆle et la nature de la colonization racinaire reste a` e´claircir. Les albinos sont deux fois plus colonise´s que les individus verts, mais aucune diffe´rence de partenaire fongique n’a pu eˆtre mise en e´vidence entre phe´notypes. L’analyse des isotopes stables de l’azote et du carbone conforte l’origine fongique de la biomasse des albinos (mycohe´te´rotrophie), tandis que les individus verts obtiennent 33 % de leur carbone de leurs champignons (mixotrophie). Les arbres voisins, qui sont ectomycorhize´s par au moins l’une des The´le´phores trouve´es sur les orchide´es, sont probable- ment la source ultime de carbone. Ces donne´es me`nent a` discuter l’e´volution de la mycohe´te´rotrophie chez les Orchide´es, et plus particulie`rement les Neottieae. Mots cle´s : Cephalanthera, isotopes stables, mixotrophie, mycohe´te´rotrophie, mycorrhizes, thelephoraceae.

Received 1 February 2006. Published on the NRC Research Press Web site at http://canjbot.nrc.ca on 20 October 2006. J.-C. Abadie1 and M.-A. Selosse.1,2 Centre d’E´ cologie Fonctionnelle et E´ volutive (Centre national de la recherche scientifque (CNRS), Unite´ mixte de recherche (UMR) 5175), E´ quipe co-e´volution, 1919 Route de Mende, 34293 Montpellier CEDEX 5, . U¨ .Pu¨ttsepp.1 Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 64, 51014 Tartu, Estonia. G. Gebauer. Laboratory of Isotope Biogeochemistry, Bayreuth Center of Ecology and Environmental Research (BayCEER), Universita¨t Bayreuth, 95440 Bayreuth, . A. Faccio and P. Bonfante. Dipartimento di Biologia Vegetale dell’Universita`, Istituto per la Protezione delle Piante – Consiglio Nazionale delle Ricerche, Viale Mattioli 25, 10125 Torino . 1These authors contributed equally to this work. 2Corresponding author (e-mail: [email protected]).

Can. J. Bot. 84: 1462–1477 (2006) doi:10.1139/B06-101 # 2006 NRC Canada Abadie et al. 1463

Introduction neottioids are associated with non-rhizoctonia, ECM fungi (Bidartondo et al. 2004; Selosse et al. 2004; Julou et al. Most orchids have an intimate relationship with their 2005; Girlanda et al. 2006), exactly as MH orchids. Sec- symbiotic fungi throughout their life cycle. These fungi col- ond, photosynthetic abilities are very limited in some neo- onize cortical root tissues and cross the cell wall to form hy- ttioids that perform photosynthesis at or below the phal coils (or pelotons) closely associated with the orchid compensation point (Julou et al. 2005; Girlanda et al. plasmalemma (Smith and Read 1997). At a later stage, pelo- 2006), and therefore need an additional C source. Third, tons are lysed and reinfection occurs, either immediately or fully achlorophyllous individuals, the so-called ‘‘albinos’’ during the next growing season (Rasmussen and Whigham are reported to grow and reproduce sexually, for example, 2002). As in other mycorrhizal symbioses, the fungus pro- in helleborine (L.) Crantz (Salmia 1988, 1989a, vides the with mineral nutrients (Smith and Read 1989b; Delforge 1998), E. microphylla Sw. (Selosse et al. 1997). In addition, orchid seeds are small, devoid of re- 2004), and Cephalanthera spp. (see references below). serves, and fed by the fungus during germination and early Although their low frequency in populations strongly im- subterranean development (Rasmussen 1995). This fungus- pairs statistical analyses, albinos are of high interest by (i) based heterotrophy, called mycoheterotrophy (MH), persists demonstrating an ability to obtain C heterotrophically, and at adult stages in several achlorophyllous species (Leake (ii) providing relevant comparisons in the study of green 2004) from various phylogenetically independent taxa conspecifics, for example, for stable isotope investigations (Molvray et al. 2000). (Julou et al. 2005), since they are full heterotrophs with a Traditional methods of identifying orchid mycorrhizal similar genetic background. fungi (collectively placed in the polyphyletic, asexual genus From an evolutionary point of view, Bidartondo et al. Rhizoctonia) were based on isolation and characterization in (2004) and Selosse et al. (2004) suggested that, in neo- pure culture (Rasmussen 2002). Recently, molecular biology ttioids, the symbiotic ‘‘jump’’ to ectomycorrhizal fungi al- tools allowed easier identification, without isolation bias, by lowed MX nutrition and predisposed to evolution of MH amplification and sequencing of ribosomal DNA stretches. nutrition, an evolution that occurred at least twice in neo- They identified the fungal associates of MH orchids that are ttioids (Bateman et al. 2004). In this framework, albinos often recalcitrant to isolation (Taylor et al. 2002; Taylor document evolutionary transition from MX to MH nutrition 2004). First, each MH species studied to date showed a (Selosse et al. 2006). However, full analysis (i.e., metabo- high specificity to a different narrow fungal clade (Taylor lism and mycorrhizal association) of MX neottioids and al- et al. 2002). Second, these fungal clades belonged to the bino individuals is so far limited to a very small number of Thelephoraceae, , and Russulaceae, but not to neottioid species and sites. any Rhizoctonia clade (even if some of them were phyloge- This work aims to test the occurrence of the MX strategy netically close to some Rhizoctonias, for example, within based on ECM fungi for another neottioid species, Cepha- the Sebacinales; Weiss et al. 2004). Third, these mycorrhizal lanthera longifolia (L.) Fritsch. The genus Cephalanthera,a fungi were also associated with tree roots, forming ectomy- basal clade among neottioids (Bateman et al. 2004), may corrhizae (ECM), whereas most Rhizoctonias are supposed document plesiomorphic traits and allow discussion of MX to be parasitic or saprotrophic. Surrounding ECM are and MH evolution in neottioids. So far, only C. damasonium thus likely the ultimate C source for MH orchids. Indeed, Druce and C. rubra (L.) Rich. are demonstrated to be MX similar fungal genets colonized the orchid and surrounding (Gebauer and Meyer 2003; Julou et al. 2005), and no data ECMs (Taylor and Bruns 1997; Selosse et al. 2002a), and are available for C. longifolia. Albinos are often described C labelling demonstrated C transfer from ECM trees in C. damasonium (Renner 1938; Mairold and Weber 1950) (McKendrick et al. 2000). but occur more rarely in C. longifolia (Renner 1943; Zba¨ren More recently, interest has shifted back to chlorophyllous 1968). Fortunately, a recently discovered Estonian orchid species that meanwhile had been somewhat less C. longifolia population (Pu¨ttsepp 1993) harbours several al- studied. They were shown to have variable specificity levels binos (Fig. 1). Based on this rare situation, our aim was (i)to (McCormick et al. 2004; Otero et al. 2004; Shefferson et al. test for MX nutrition in C. longifolia by investigating C nu- 2005), questioning whether high specificity was restricted to trition and root associates, and (ii) to compare albino and MH species. In addition, several photosynthetic orchids green phenotypes. For both phenotypes, we thus (i) assessed were shown to have unusually high 13C/12C and 15N/14N ra- growth parameters of genets over years, (ii) identified root tios as well as high leaf nitrogen (N) concentrations as com- fungi at the adult stage through molecular and microscopic pared with photosynthetic non-orchid species from the same investigations, and (iii) investigated the heterotrophy level sites (Gebauer and Meyer 2003; Julou et al. 2005). These using relative N and C stable isotope abundances. features are reminiscent of MH species that are N-rich and 13 12 15 14 have high C/ C and N/ N ratios, similar to their my- Materials and methods corrhizal fungi. Green individuals could thus use fungal C and N in addition to their photosynthesis and autotrophic N Study site nutrition, a strategy called mixotrophy (MX): up to 85% of The investigated population grows in a calcareous coastal their carbon could be derived from the fungus (Gebauer and plain at Pussa (Saaremaa island, West Estonia; 58814’44@N Meyer 2003; Selosse et al. 2006). and 22800’41@E), 300 m inland from the sea. The site is cur- MX species may be frequent among the Neottieae,a rently undergoing an ecological transition from grassland tribe of forest orchids encompassing green and MH species (alvar) to forest stage. The population of Cephalanthera (Bateman et al. 2004), for three reasons. First, several green longifolia appeared in the early 1980s and now covers an

# 2006 NRC Canada 1464 Can. J. Bot. Vol. 84, 2006

Fig. 1. Albino Cephalanthera longifolia from Pussa. (a) An albino in its shaded habitat (photograph by Fred Ju¨ssi). (b) Schematic map of all albinos in 1992–1998 and 2002–2004 (except No. 20, position unknown). Cep.lon., C. longifolia albinos; Pin.syl., Pinus sylvestris; Pic.abi., Picea abies. Green C. longifolia and Juniperus communis shrubs are not plotted because of their abundance.

– -1 + area of 15 m  39 m. Vegetation is a shrubland where ju- 12.1) and composition of NO3 0.00 mgÁkg ,NH4 5.71 nipers (Juniperus communis L.) with an average height of mgÁkg-1,K+ 0.03 cmol+Ákg-1,Ca2+ 30.0 cmol+Ákg-1, and 1 m form dense groups. Scots pines (Pinus sylvestris L.) Mg2+ 2.62 cmol+Ákg-1 at depth of and roots of of 1-6 m in height and a single 8 m Norway spruce (Picea C. longifolia (8-10 cm). abies (L.) Karst.) form a discontinuous canopy layer cover- ing <30% of the area. This results in shaded areas (average Genet monitoring, morphology, and pigment analysis Ellenberg light indicator calculated on the basis of occur- In a monitoring from 1992 (when the first albino was ring species: 5.7) and sunny open places (average Ellen- seen) to 1998 and in another from 2002 to 2005, all emerg- berg light indicator: 7.3). (The Ellenberg indicator values ing albino individuals were mapped, marked in the field, represent the preferences of individual species, based on and their state (vegetative, forming flowers, or dormant empirical field observations, ranging from 1, deep shade; underground) was scored. to 9, full sun light; Ellenberg et al. 1991.) Distance between individual shoots (>20 cm) allowed Cephalanthera longifolia plants are restricted to shaded genet tracking. The stock of C. longifolia exhibits zones with intermediate light conditions, between shrubs very short yearly increments (three to four internodes), pro- and trees. The dominating species in the herb layer are Se- viding vegetative mobility (annual horizontal distance be- sleria coerulea Arduin. (15%) and Tetragonolobus mariti- tween mother and daughter ramets) close to zero (Pu¨ttsepp mus Roth (5%), accompanied by Antennaria dioica and Kull 1997). We therefore used a conservative definition Gaertn, Campanula glomerata Hegetschw., Cirsium acaule of the genet in this work: only shoots growing less than (L.) Scop., Festuca ovina L., Filipendula vulgaris Hill, In- 5 cm away from each other were considered as belonging ula salicina C.B. Clarke, Molinia caerulea Moench, Ophio- to the same genet. The only risk was thus the pooling of glossum vulgatum L., Pimpinella saxifraga L., Plantago two different genets growing closely in the same year, or media L., Trifolium montanum L., etc. Total coverage of two genets growing at the same place in successive years. the herb layer is 65%. Various other orchid species occur For comparison of the two phenotypes, several parameters (e.g., Epipactis atrorubens (Hoffm.) Besser, Gymnadenia were measured each year for each albino genet and ran- conopsea (L.) R.Br., Ophrys insectifera L., Epipactis helle- domly selected green ones: number of shoots per genet, borine, Orchis mascula, and Listera ovata L.Br.). Moss number of flowers and number of leaves per shoot, length cover reaches 50%, with Hypnum cupressiforme, Thuidium and width of the longest leaf, and height of shoot (whenever delicatulum (Hedw.) B. S. G. as most frequent species. The two or more shoots occurred for a given genet, only the soil is a limestone rendosol with pHCaCl2 of 6.8-7.1 with highest value was considered). The latter measurements 3.65% of organic C and 0.301% of total N (thus C/N = were done at the end of June, in 1996 and 1997. Since nor-

# 2006 NRC Canada Abadie et al. 1465

Fig. 2. Morphological comparison between albino (white columns) and green Cephalanthera longifolia individuals (black columns) in 1996 (a) and 1997 (b). Mean values with SD are shown. The asterisk indicates a significant difference according to a Mann–Whitney test (P < 0.05). For green individuals, values were obtained from randomly selected individuals.

mality of variables was not verified, because of the low tigation was performed on five root fragments from number of albinos, data were tested with a nonparametric individual G4 and three from individual A3. These frag- Mann–Whitney test using Minitab (v12.2) software. Leaves ments were randomly chosen among those selected for mo- of four green and five albino individuals were collected in lecular typing of fungal partners, as previously described. A July 2005, quickly frozen at –20 8C, and handled as in Julou 0.3 cm subfragment, flanking the subfragment kept for mo- et al. (2005) for estimation of chlorophyll content. lecular analysis, was quickly fixed in 2.5% (v/v) glutaralde- hyde in a 10 mmolÁL–1 Na-phosphate buffer (pH 7.2). The Mycorrhizal sampling samples were then rinsed, postfixed, embedded, and cut as Root systems were harvested at flowering time in June semi-thin sections (for light microscopy) or thin sections 2002 (three green individuals (G1, G2, and G3) and two al- (0.05 mm for transmission electron microscopy, TEM) as in binos (A1 and A2)) and June 2004 (one green individual Selosse et al. (2004). (G4) and one albino (A3), Table 1). Six full-length roots were sampled from soil without disturbing the rhizome and Mycorrhizal typing other roots (a protocol designed to allow genet survival, see All root sections harbouring pelotons were submitted to genets Nos. 12 and 13, Table 1). The roots were cut in DNA extraction and polymerase chain reaction (PCR) am- 1 cm long fragments, and at each cutting a section of the plification of the fungal ITS, using primers universal for root was checked for the presence of fungal pelotons under fungi (ITS1F+ITS4) as in Selosse et al. (2002a). To ensure the light microscope. Whenever pelotons were seen, a thin that all host orchids were from the same species, plant ITS (ca. 0.2 mm) subfragment flanking the investigated section were sequenced as in Selosse et al. (2004) from one root was kept for molecular typing of fungal partners. The frag- per individual, and all proved to be identical (GenBank ac- ments were then kept for (i) estimation of the whole fungal cession number DQ182464). To obtain a more detailed view diversity by intergenic transcribed spacer (ITS) cloning, of the fungal partners (including nonmycorrhizal endo- and (ii) microscopic analysis (see below), as well as (iii) phytes), DNA was extracted in the same conditions from quantification of the mycorrhizal colonization. To quantify several 1 cm long root fragments. For individuals G1, G2, mycorrhizal colonization, 10, 1 cm fragments were ran- G3, A1, and A2, four root fragments were randomly chosen domly chosen for G1, G2, G3, A1, and A3 individuals, ex- and amplified as previously. PCR products were then pooled cluding root tips and young roots (<2 cm in length). They two by two, and the resulting mixes (two mixes per individ- were longitudinally sectioned and stained by trypan blue ual) were cloned as in Julou et al. (2005). A minimum of 70 following the method of Koske and Gemma (1989) to esti- clones was recovered for each mix. Clones were submitted mate the mean length percentage of mycorrhizal infection. to two restriction fragment length polymorphism (RFLP) analyses using HaeIII + HinfI and HhaI+NdeII: 2 mLof In June 2004, tree ECMs were carefully searched for in PCR product were added to 2 mL of each enzyme in the buf- the vicinity (<10 cm away from the roots) of individuals G4 fer provided by the manufacturer (Qiagen SA, Courtaboeuf, and A3 at the time of their harvest, to compare orchid my- France) and incubated at 37 8C for 2 h. Digestion products corrhizal fungi with surrounding tree mycorrhizal partners. were separated by migration on 3% agarose gels in 0.5Â In all, 192 ECMs were sampled. All the material prepared Tris–borate–EDTA buffer. ECM were submitted to DNA for molecular analysis was kept at –80 8C. extraction and ITS amplification as previously, and then compared in RFLP pattern to the orchid ITS (for compari- Microscopic analysis son, RFLP patterns were also generated from ITS amplified A first observation was performed under the light micro- root section harbouring pelotons). scope during checking and quantification of the mycorrhizal All sequences were obtained using ITS1F and ITS4 as in infection, by looking for clamp connections that are charac- Selosse et al. (2002a). Sequencing was performed for (i) all teristic of dikaryotic basidiomycetes. A second, closer inves- ITS amplified root sections harbouring pelotons, (ii)upto

# 2006 NRC Canada 1466 Can. J. Bot. Vol. 84, 2006

Table 1. Occurrence of Cephalanthera longifolia albino genets between 1992 and 2005.

Genet No. 1992 1993 1994 1995 1996 1997 1998 2002 2004 2005 1FVFFFFFV0VV 2 V ——— — ——— — 3FV——————— 4V0VFVVF—— 5FDVV———— 6 V+V V+V V+V+V V+F 0 0 V — 7V0VV———— 8VD-—————— 9V————— 10 V –– — — — — 11 V – — — — — 12 F 0 V Fa V 13 F Fb V— 14 Fc —— 15 VV 16 V 17 V 18 V 19 V 20 V Total Albinos 1 4 6 5 8 6 3 4 5 8 All phenotypes 20 TU TU TU TU TU 98 TU TU 235 Note: Shoots are flowering (F), flowering and fruiting (FF), vegetative, that is, with a nonflowering shoot (V) or damaged in early development (D). Dormant individuals (0) lack shoots in year n, but form shoots in year n+1. Individuals that did not form shoots up to 1998 have an unknown state (—), that is, are dormant over a long period or dead; TU, total unknown. aGenet A3 in this study, sampled in 2004. bGenet A1 in this study, sampled in 2002. cGenet A2 in this study, sampled in 2002.

five ITS cloned from each RFLP type obtained after cloning We distinguished species growing in the same shaded parts (to correct sequence PCR-generated errors), and (iii) all ITS of the site as C. longifolia and those growing in more sunny from ECMs matching any orchid ITS in RFLP analysis. To parts. All terrestrial fungal fruitbodies found on the site were look for Tulasnellaceae, which have highly derived rDNA collected at the same time. Samples were predried for 3 d at sequences, PCR was carried out on all extracted DNAs us- 55 8C and subsequently handled as in Bidartondo et al. ing ITS1 and ITS4tul (a primer specific for Tulasnelloids) (2004) to measure total N concentrations and relative isotopic as in Selosse et al. 2004, and always failed to reveal any tul- abundances, that are denoted as d values: d15Nord13C= lasnelloid ITS. Sequences were edited and aligned using Se- (Rsample/Rstandard –1)Â 1000 [%], where Rsample and Rstandard quencher 4.5 for MacOsX from Genes Codes (Ann Arbor, are the ratios of heavy isotope to light isotope of the samples Michigan) and deposited in GenBank. Searches for similar se- and the respective standard (see Gebauer and Meyer 2003). quences allowing taxonomic identification were conducted us- ing the BLASTN algorithm available through the National Results Center for Biotechnology Information (NCBI; http://www. ncbi.nlm.nih.gov/BLAST/index.html). Putative ecology was Dynamics of albino genets in the population inferred from that of the closest relatives, when taxonomic A first albino C. longifolia genet emerged in 1992 identification was reliable. Phenetic comparisons of the vari- (Table 1) and appeared again in 1993–2005, together with ous orchid individuals based on shared ITS sequences were others (Fig. 1b). Some albinos flowered and genet No. 1 generated by UPGMA analysis using PAUP*4.0b2a (Swofford produced three capsules in 1995. Altogether, 20 distinct al- 2001). bino genets were found by 2005. Some had a one-year eclipse, suggesting that they stayed in an underground, dor- mant state (state 0, Table 1), a state also recorded for green Isotope analysis individuals (not shown). Albino genets never produced We collected nine C. longifolia white leaves from five al- green shoots over 13 years, and albino shoots never derived binos (two leaves per individual) and nine green leaves from from a previously green genet. A last survey in 2005 re- nine different green individuals in June 2004. Leaves of sur- vealed 8 albinos and 235 green genets, so that albinos re- rounding plant species (eight leaves, each from different in- mained in low proportion (ca. 3% of genets with dividuals) were sampled at the same height for comparison. aboveground shoots) in this expanding population.

# 2006 NRC Canada bdee al. et Abadie Table 2. Mycorrhizal symbionts from green (G1–G4) and albino (A1–A3) genets of Cephalanthera longifolia.

Plant individual and year of sampling G1, 2002 G2, 2002 G3, 2002 G4, 2004 A1, 2002 A2, 2002 A3, 2004 Mycorrhizal colonization (% of root length) 40±12ab 37±18a 38±15a NI 75±16b 76±9b NI Clamp connectionsa ++ +++++ Peloton-forming fungib Typing success 6 of 9 8 of 11 5 of 7 8 of 9 6 of 10 5 of 9 3 of 5 Thelephoroids Shared (= occurring in ‡2 individuals) DQ150118 (3) DQ150116 DQ150123 (4) DQ150126 (2)c DQ150128 (2) DQ150116 DQ150118 DQ150128 DQ150123 (3) DQ150126 (3) DQ150128 (3) Nonshared DQ150117 DQ150125 DQ150113 DQ150121 DQ150127 (2) DQ150122 DQ150129 (3) DQ150119 DQ150124 DQ150130 DQ150120 (2) Other fungi Sebacinaceae sp.1 Wilcoxina sp.1 (DQ150132)d (DQ150131)e Note: Means (± SE) followed by different letters are significantly different according to an ANOVA (P < 0.001); NI, not investigated. aObservation of clamp-connection, that is, typical basidiomycetous features (Fig. 3), on some peloton hyphae at least. bGenBank accession numbers, with number of investigated sections producing the same ITS sequence if n > 1 (in brackets). cAlso found on tree ECM surrounding G4. dAccession No. AF440647 (Sebacina endomycorrhiza of nidus-avis) is 96.8% identical over 1214 bp (the 5’ part of the 28S rDNA of Sebacinaceae sp. 1 was also amplified for comparison); Sebacinaceae sp. 1 thus belong to the ECM clade of Sebacinales (clade A in Weiss et al. 2004). eAccession No. DQ069002 (uncultured Wilcoxina) is 99.7% identical over 574 bp. # 06NCCanada NRC 2006 1467 1468 Can. J. Bot. Vol. 84, 2006

Comparison of albinos and green individuals for tigations, this demonstrated that at least Thelephoraceae morphology and chlorophyll were mycorrhizal within C. longifolia roots. Chlorophyll content was considerably lower for albinos To further assess the fungal diversity in roots, a cloning (1.9 ± 5Â10–2 mgÁmg–1 fresh mass, n = 5) than for green in- of ITS PCR from four root pieces per individual was under- dividuals (2.97 ± 0.56 mgÁmg–1 fresh mass, n = 4), suggest- taken (Table 3). In all, 59 ITS sequences were retrieved ing that albinos are not able to carry out photosynthesis from the ca. 700 clones analyzed, that is, between 10 and (values significantly different according to Student’s t test, 22 sequences per individual, encompassing 9 sequences al- P < 0.05). Number of shoots per genet showed opposite ready detected in the previous analysis (8 thelephoroids and trends over the 2 years of survey (Fig. 2). For both years, Wilcoxina sp.1). Basidiomycetous ITS were mostly from the number of leaves tended to be higher for albinos, while ECM taxa, for example, Thelephoraceae (dominating in number of flowers per shoot and total height were lower for number and frequency – not shown), Cortinariaceae, and albinos. However, except for the shoot length in 1996, none Amphinema spp. Three ITS belonged to Rhizoctonia clades of these differences were significant according to a Mann– (Ceratobasidium sp. 1 and sp. 2, Table 3, and Thanatepho- Whitney test (P < 0.05). No significant difference in length rus sp. 1, Appendix A). Ascomycetous ITS encompassed or width of the largest leaf was demonstrated (not shown). many Helotiales (putatively biotrophic, symptomless endo- In all, no obvious trend appeared in morphological compari- phytes, but possibly some ECM), some Sordariales (possibly son, but albinos tended to be smaller and to have decreased saprotrophic), and Nectriaceae (endophytes or parasites), to- internode size. Albino shoots tended to have shorter life gether with Pezizales (probably ECM) (Table 3; Appendix spans than green ones, developing and flowering later but A). A total of 24 ITS putatively corresponded with ECM drying earlier (in the first week of July, when green individ- species, and 23 corresponded with endophytic species. Un- uals still ripened their fruits). For these data again, no statis- expectedly, two Glomus-like (arbuscular mycorrhizal fungi) tical support was found (not shown). ITS were retrieved from G3 (Appendix A). Some ITS (24.5%, Table 3) were amplified from at least Root colonization two individuals. Albino individuals did not show any ob- Fungal hyphae were constantly present inside root frag- vious similarity between them, as compared with green in- ments. However, the level of root colonization was signifi- dividuals, both on the basis of ITS shared in peloton cantly lower for green than for albino plants (Table 2). Some analysis (Fig. 5) and in cloning (similar topology, not pelotons from all root systems showed clamp-connections, a shown). Similarly, w2 analysis could not reject the null hy- structure formed during cytokinesis in dikaryotic basidiomy- pothesis that no significant difference occurred between cetes only (Fig. 3a; Table 2). The presence of basidiomycetes green and albino individuals, considering fungal classes (Ba- was further confirmed by an extensive TEM survey of eight sidiomycetes, Ascomycetes, or Glomales; w2 = 1.83, df = 2, fixed samples, which consistently showed endophytes P < 0.05), fungal orders (w2 = 10.44, df = 13, P < 0.05) or (Fig. 4). The morphology of the mycorrhizal fragments was fungal ecology (w2 = 1.66, df = 4, P < 0.05). rather homogeneous, irrespective of the samples or plant phe- Last, we investigated 192 tree ECMs surrounding individ- notypes. As seen on semi-thin sections from both green and uals G4 and A3. From the 78 morphotypes found (not albino individuals, pelotons were restricted to the cortical pa- shown), RFLP followed by sequence analysis revealed only renchyma and underwent a final lysis (Figs. 4a–4d). Large two similarities, namely a thelephoroid (DQ150126, found nuclei of host cells were usually surrounded by fungal hy- near G4 where it also occurred) and Sebacinaceae sp.2 phae. Starch was abundant in noninfected cells as well as (DQ182435, found near G4 and already found on G3). around the clumped hyphae. Pelotons colonized living root cells and were surrounded by an extension of the host N concentrations and isotope signatures plasma membrane (Fig. 3b). Two samples (one from G4 Leaf N concentrations revealed strong interspecific varia- and the other from A3) showed only collapsed hyphae at tions (Table 4). While most plants (including the N-fixing the lytic stage, while all the remaining samples revealed legume Tetragonolobus maritimus) had N concentrations be- –1 septate hyphae, with dolipores connecting neighbouring hy- low 2 mmolÁgdm , the neottioids Epipactis helleborine and phal cells (Fig. 3c). The presence of clamp connections Listera ovata had significantly higher concentrations, similar and dolipores unambiguously demonstrates that basidiomy- to green C. longifolia individuals. Albinos had even signifi- cetes are mycorrhizal on C. longifolia in both green and cantly higher leaf N concentration, in the range of ECM fun- albino roots. No ascomycetous hypha was seen during gal fruitbodies collected on this site (Helvella acetabulum TEM analysis. (L: Fr.) Que´let and Inocybe sp., Table 4). Leaf d15N and d13C (Fig. 6) clearly distinguished albinos Identification of root fungi from all other plants (difference significant according to a A first typing of fungal ITS on 60 root pieces harbouring Mann–Whitney test, P < 0.001). When compared with non- pelotons succeeded in 41 samples (68%; Table 2). All ITS orchids growing on the same shaded parts of the site, albi- were from thelephoroid species, with the exception of two nos were enriched in 15N by 11.8% and in 13C by 6.6%. root pieces from G1 and G2, respectively, exhibiting ITS Green C. longifolia and the neottioid E. helleborine were from a Sebacinaceae and a Wilcoxina (Ascomycetes). All distinguished in d15N from most other plants (the difference these ITS are from putatively ECM species. Five of the 16 is significant according to a Mann–Whitney test, P < 0.001) thelephoroid ITS were shared by more than one individual, and their d13C was less negative (but not significantly, P = and some sequences were found several times on the same 0180) than for other plants growing in similar light condi- root system (Table 2). Congruently with microscopic inves- tions (Fig. 6), a feature expected for mixotrophic plant. All

# 2006 NRC Canada Abadie et al. 1469

Fig. 3. Microscopic analysis of fungal pelotons from green individuals of Cephalanthera longifolia.(a) Light microscopy of the intracellu- lar hyphal pelotons (F) with clamp connection (circled and inset) in squashed root cells. (b) TEM of fungal hyphae (F) inside a cortical host cell developing close to the plant nucleus (Pn). The fungal hyphae are surrounded by mitochondria (Pm) and vacuoles (Pv). (c) TEM of dolipores (D) with perforated parenthesomes surrounding the pore of the septum (S), which confirms the basidiomycetous origin of the hyphae (Fv, fungal vacuole; Fw, fungal cell wall; Fpm, fungal plasma membrane; arrows, host plasma membrane). Scale bars = 10 mmin Fig. 3a and 0.36 mm in Figs 3b and 3c.

other orchids were also enriched in 15N in comparison with photosynthesis is unlikely to contribute significantly to the C the non-orchids and non-legumes, but to a lesser degree than budget of albinos. Their association with ECM fungi the two previous neottioids. In addition, their d13C were (Table 2, see below) is also reminiscent of MH orchids similar to non-orchids (differences are nonsignificant ac- (Leake 2004). The denser root colonization, as compared cording to a Mann–Whitney test, P < 0.001). The legume with green individuals (Table 2), is reminiscent of Epipactis (T. maritimus) was distinguished by d15N from the non- albinos (Salmia 1989a; Selosse et al. 2004) and may con- orchids because of N2 fixation. Non-orchid plants from the tribute to a better C supply. Albinos had high leaf N concen- shade tended to be somewhat more depleted in 13C than the trations, similar to ECM fungal fruitbodies collected on this non-orchids that were in light. The investigated ECM fungi site (Helvella acetabulum and an Inocybe sp., Table 4), a were distinguished from all non-orchids in d15N and feature expected for MH plants. Indeed, albino organic mat- d13C. However, they were only enriched in 13C when com- ter should have a C/N ratio similar to the fungi they fed on. pared with most of the orchids, except for C. longifolia Since part of the recovered biomass is respired into CO2, and E. helleborine. The two latter orchid species have they may even have a lower C/N ratio: unfortunately, this is higher d15N values than the ECM fungi. Unfortunately, no not testable here, as no fruitbodies of their root fungal part- fruitbody corresponding to C. longifolia root fungi was ners were found. The 15N enrichment of albinos in compari- found, so that no data on isotope signatures of C sources son with the non-orchids and non-N-fixing plants on the potentially used by albinos were obtained. Saprotrophic shaded part of the site (11.8%, Fig. 6) is similar to previous fungi (all growing on soil with woody remains) were dis- findings for MH orchids versus non-orchids (Gebauer and tinguished from ECM fungi because of a greater depletion Meyer 2003; Trudell et al. 2003; Bidartondo et al. 2004; in 15N and a higher enrichment in 13C. Julou et al. 2005). Cephalanthera longifolia albinos are thus completely N heterotrophic. The 13C enrichment of the Discussion albinos in this study (6.6%) is slightly lower than that found in these other studies for MH (e.g., 8.4%, Gebauer and Albino phenotype in Cephalanthera longifolia Meyer 2003) and albino orchids (9.1% in Julou et al. 2005). Data on C. longifolia albinos from the Pussa population This could relate to some particular 13C enrichment of their strongly support their MH status. Their total chlorophyll mycorrhizal, C-providing fungi. The alternative assumption content is only 0.6% that of green individuals, similar to that they were green in the previous year is unlikely, because heterotrophic C. damasonium albinos (Julou et al. 2005). of phenotype stability (Table 1). Indeed, d13C of the two Although measurement of gas exchanges would be required, ECM species found at this site are low (Fig. 6). Strong limi-

# 2006 NRC Canada 1470 Can. J. Bot. Vol. 84, 2006

Fig. 4. Root anatomy and cytology in green and albino individuals of Cephalanthera longifolia. Light microscopy sections of root samples from green (a and c) and albino (b and d) individuals reveal intracellular hyphal coils (Fh) and fungal clumps (Fc) resulting from hyphal lysis in the cortical host cells. Hyphae are not seen in the epidermal cells (E), or in cells (Pc) containing calcium oxalate crystals (Pn, plant nu- cleus). Scale bars = 40 mm in Fig. 4a, 50 mm in Fig. 4b, and 14 mm in Figs. 4c and 4d.

tations are the unavailability of co-occurring MH species and of the albino phenotype over years makes sense of inter- fruitbodies of C. longifolia root fungi. However, with the phenotype comparisons in C. longifolia. lowest plant d13C at this site, albinos are likely fully MH. In the growing Pussa population, albinos remain infre- Our 13-year survey, based on a very conservative defini- quent (ca. 3% in 1998 and 2005). Assuming genetic deter- tion of the genet (see Material and methods), demonstrates minism, this means that (i) albinos do not reproduce but are the stability of the albino phenotype in this population produced at a constant mutation rate (cryptopolymorphism), (Table 1). Neither an intermediate phenotype nor pheno- or that (ii) they reproduce with a fitness nearly identical to type transition was observed. This differs from Epipactis green individuals, or (iii) a combination of the two. Con- helleborine albinos, where there is an intermediate pheno- gruently, we did demonstrate significant impairment of albi- type or reversal (Salmia 1989a, 1989b; M.-A. Selosse, un- nos in aboveground development (Fig. 2) or survival (one published data). The albino phenotype could thus result half of the genets from Table 1 formed shoots in >2 years), either from a mutation-abolishing photosynthesis or from but statistical limitations may be the reason for this. How- micro-environmental conditions preventing greening, as de- ever, a limitation on capsule ripening (and thus on fitness) scribed in vitro for Anacamptis morio [(L.) R.M.Bateman, of albinos may be the short life span of their shoots, which Pridgeon & M.W.Chase] grown on high sugar concentra- dry early at Pussa, as described for albinos of other neo- tions (Beyrle and Smith 1993). Our data, as well as pre- ttioids (Ricek and Gutermann 1971; Salmia 1989b; Delforge vious findings on albino individuals (Salmia 1988, 1989a, 1998; Julou et al. 2005). Assessment of albino fitness is 1989b; Selosse et al. 2004; Julou et al. 2005) provide no thus needed, through investigations of seed production and definitive conclusions on determinism of albinism, but we viability (germination), but the rarity of reproducing plants favour genetic determinism as the most likely reason for at Pussa and the time to reach adulthood in C. longifolia phenotype stability at Pussa. Most importantly, the stability (Rasmussen 1995) will certainly complicate such work.

# 2006 NRC Canada Abadie et al. 1471

Table 3. Root fungi found in cloning on green (G1–G3) and albino (A1–A2) genets of Cephalanthera longifolia collected in 2002 (two cloning replicates per individual).

Plant individual Fungal type G1 G2 G3 A1 A2 Thelephoroid fungia Thelephoroids shared between individuals DQ150118 DQ150118 DQ150118 DQ150123b DQ150123b Thelephoroids not shared DQ150122 DQ150115 DQ150116 DQ150113 DQ150112 DQ150117 DQ150119 DQ150114 DQ150121 Other shared fungic Basidiomycetes Cortinariaceae sp.1 (DQ182417) ECM n =2 n =2 Ceratobasidium sp.1 ((DQ182418) Rhizoctonia n =1 n =1 Ceratobasidium sp.2 (DQ182419) Rhizoctonia n =1 n =1 n =1 n =1 Cortinariaceae sp.2 (DQ182420) ECM n =1 n =1 n =2 n =1 Amphinema sp.1 (DQ182421) ECM n =2 n =2 Ascomycetes Wilcoxina sp.1 (DQ150131)d ECM n =2 n =2 Unknown ascomycete sp.1 (DQ182422) Endophyte n =2 n =1 Leptodontidium sp.1 (DQ182423) Endophyte n =1 n =1 Helotiale sp.1 (DQ182424) Endophyte n =1 n =2 Nectriaceae sp.1 (DQ182425) Endophyte n =1 n =1 sp.1 (DQ182426) Endophyte n =2 n =2 Helotiale sp.2 (DQ182427) Endophyte n =2 n =1 Nectriaceae sp.2 (DQ182428) Endophyte n =1 n =1 Other nonshared fungie 5 1 13 5 11 Total no. of fungi 15 11 19 13 22 No. of ECM 5 7 5 7 7 No. of endophytes 7 2 8 3 10 No. of ‘‘Rhizoctonias’’ 2 1 2 1 1 No. of Glomales 2 No. of saprotrophic fungi 1 1 Note: We consider only the presence or absence information, as quantitative differences in clone frequency may only reflect PCR and (or) cloning biases. aUnderlined GenBank accession numbers indicate fungi already found in peloton analysis (Table 2); Bold GenBank accession numbers indicate fungi already found in peloton analysis on the same individual. bAlso found on tree ECM surrounding G3. cITS sequence found in one (n = 1) or two (n = 2) clonings per individual; GenBank accession numbers are in parentheses. dAlready found by peloton analysis on G2 (see Table 2). eSee Appendix A for GenBank accession numbers, and putative identifications and ecology.

Mycorrhizal and root endophytic fungi in as recorded for C. damasonium (Julou et al. 2005). Rigor- Cephalanthera longifolia ous assessment of mycorrhizal diversity at the species level One of our major conclusions is that mycorrhizal pelotons requires analysis of other populations and other develop- in C. longifolia of the Pussa population are formed by basi- mental stages (e.g., during germination). diomycetes, mostly belonging to Thelephoraceae. Micro- Cloning investigations revealed various fungal species scopic analysis reveals basidiomycetous features (Fig. 3), that are not necessarily mycorrhizal (a strong bias of this ap- and molecular analysis of colonized root portions reveals proach). Many ITS sequences putatively belong to endo- thelephoroid ITS (Table 2). The two other ITS found in pe- phytes (i.e., growing in living tissues, but not necessarily loton analysis are from clampless fungi, namely a Wilcoxina peloton forming) or possibly weak parasites (Table 3; Ap- (an ascomycete without dolipores) and a sebacinoid (with a pendix A). Among the latter, Leptodontidium is a common continuous, uninterrupted parenthesome, Selosse et al. orchid endophyte (Rasmussen 1995), and Phialophora spp. 2002b) that were not found in TEM analysis. Clamp connec- were already reported from Cypripedium species (Vujanovic tions and similar dolipores exist in Cortinariaceae, Amphi- et al. 2000; Shefferson et al. 2005 – our Phialophora sp.2 is nema spp., and Ceratobasidium spp. (Andersen 1996) found 98.8% identical over 424 bp to a sequence from Cypripe- by cloning (Table 3), but they were not found by molecular dium parviflorum Salisb., AY578279). The finding of Helot- analysis of isolated pelotons. Thelephoraceae and Cortinaria- iales considered as aquatic asexual taxa (Tricladium and ceae indeed associate with the related C. damasonium and Tetracladium, Appendix A) is unlikely to result from con- C. rubra (Bidartondo et al. 2004; Julou et al.2005). A certain tamination of the lab distilled water (as shown by control mycorrhizal specificity thus seems to exist in C. longifolia, cloning on the lab water, not shown), but adds to the grow-

# 2006 NRC Canada 1472 Can. J. Bot. Vol. 84, 2006

Fig. 5. Tree comparing the five investigated Cephalanthera long- Table 4. Mean total nitrogen concentrations (± SD) of fungal ifolia individuals on the basis of their peloton-forming fungi (data fruitbodies and plant leaves belonging to different functional from Table 2). groups compared with those of the leaves of the two Cepha- lanthera longifolia phenotypes.

Nitrogen concentration Species (or functional group) (mmolÁg–1 dry mass) n C. longifolia Albinos 3.76±0.24a 9 Green 2.66±0.21b 9 Other orchids Neottioidsa 2.51±0.26b 16 Non-neottioidsb 1.62±0.19c 24 N-fixing plantsc 1.71±0.31c 8 Non-orchid, non-N-fixing plantsd 1.43±0.11c 72 Fungi Saprotrophice 1.43±1.39c 12 ECMf 3.35±0.52a 10 ing evidence that these fungi are endophytes in many terres- Note: Means (± SE) followed by different letters are significantly differ- ent according to a Kruskal–Wallis test (P < 0.05). trial plants (e.g., Russell and Bulman 2005; Sati and Belwal a 2005). This unexpected ecology awaits further investigation. Epipactis helleborine and Listera ovata (n = 8 each). bOphrys insectifera, Orchis mascula, and Gymnadenia conopsea (n =8 Other endophytes belong to Basidiomycetes, such as un- each). known basidiomycete sp.1 (DQ182429, Appendix A), whose cTetragonolobus maritimus (n = 8). closest GenBank accession (by BLAST analysis) is a Cala- dPlantago lanceolata, Convallaria majalis, Rubus saxatilis, Pinus denia formosa G.W. Carr endophyte (AY463171). Individu- sylvestris, Filipendula vulgaris, Potentilla erecta, Picea abies, Frangula als of some of these root endophytes may grow in soil alnus, and Juniperus communis (n = 8 each). eAuriscalpium vulgare (n = 3), Mycena sp. (n = 2), and Strobilurus around plants (Jumpponen and Trappe 1998), so that they tenacellus (n = 7). could possibly contribute to nutrient uptake. Intriguingly, fHelvella acetabulum (n = 8) and Inocybe sp. (n = 2). two glomalean sequences were found in individual G3 (Ap- pendix A), a feature already reported from Cypripedium spp. fection. There is growing evidence that mycorrhizal fungi (Shefferson et al. 2005) that could result from unspecific respond to root exudates (Horan and Chilvers 1990; colonization by fungi from surrounding endomycorrhizal Akiyama et al. 2005), and neottioids may use or mimic sig- plants. A last particularity is the presence of Ceratobasidia- nals used by ECM plants in early recognition, entailing ceae related to the rhizoctonias, the usual orchid mycorrhizal some nonspecific, but limited, colonization. partners, that is, two Ceratobasidium and one Thanatepho- Last, many fungi were shared between years and orchid rus (Table 3; Appendix A). Although their dolipores resem- phenotypes (Tables 2 and 3). Since albinos did not obvi- ble those observed here (Fig. 3C; Andersen 1996), they were ously differ from green individuals by their fungal partners not reported to form pelotons (Table 2). A Thanatephorus (Fig. 5), this suggests that green individuals may be able to was occasionally found in mycorrhizal and nonmycorrhizal recover fungal carbon and live mixotrophically. parts of the MH orchid Hexalectris spicata Barnhart (Taylor et al. 2003), while Ceratobasidium spp. were found in neo- ttioid roots by Bidartondo et al. (2004) and Julou et al. Mixotrophic strategies in Cephalanthera longifolia (2005). No rhizoctonia related to Tulasnella were found As expected, non-orchid plants from the shade tend to be here. This raises the possibility that some Ceratobasidiaceae somewhat more depleted in 13C than non-orchids in sunny simply grow endophytically, or even parasitically, without conditions. Leaf d13C is negatively correlated with the leaf being mycorrhizal, as described in other plants (Hietala and intercellular CO2 concentration: this parameter increases Sen 1996; Sen et al.1999). with decreasing light supply under otherwise constant envi- Strikingly, a large number of ECM species were found in ronmental conditions (Farquhar et al. 1982; Gebauer and cloning analysis (40%), while microscopic investigations Schulze 1991), leading to greater 13C discrimination during failed to reveal corresponding hyphae (e.g., no ascomyce- CO2 assimilation. Green C. longifolia should thus be exclu- tous or sebacinoid septal features were seen). At least two sively compared with co-occurring plants from the shade. of these ECM species were also ectomycorrhizal on sur- Our data allow calculations using a linear two-source mixing rounding tree roots (Table 3). ECM fungal species detected model (Gebauer and Meyer 2003) calibrated with two end by the molecular approach but undetectable by TEM analy- points: albinos as MH standards and non-orchids, non-N-fixing sis have already been reported for two other neottioids, plants from the shade as fully autotrophic standards (Table 5). C. damasonium (Julou et al. 2005) and Epipactis micro- Green C. longifolia leaf biomass contains 33% of fungi- phylla (Selosse et al. 2004). Since these fungi are uncom- derived C. The value is well above that obtained for mon contaminants of soil samples, some biological reasons C. rubra by Gebauer and Meyer (2003), but at the lower may account for their presence. They could be attracted by end of those found for C. damasonium (between 33 and chemical clues similar to those attracting the ECM species 85% of fungal C; Gebauer and Meyer 2003; Bidartondo et forming pelotons, but fail to result in further mycorrhizal in- al. 2004; Julou et al. 2005). The only other MX orchid spe-

# 2006 NRC Canada Abadie et al. 1473

Fig. 6. Mean d15N versus d13C values (± SD) for plants (squares) and fungi (triangles) from the Pussa site, including Cephalanthera long- ifolia (*), green (CLG) and albino (CLA) individuals (n = 9 for each phenotype). Plants collected in shaded situations (filled symbols) under Juniperus shrubs and Pinus, in addition to C. longifolia, are Plantago lanceolata (PL), Convallaria majalis (CM), Rubus saxatilis (RS), Pinus sylvestris (PS), and Filipendula vulgaris (FV). Plants collected in light situations (open symbols) are Epipactis helleborine (EH), Ophrys insectifera (OI), Orchis mascula (OM), Listera ovata (LO), Gymnadenia conopsea (GC), Tetragonolobus maritimus (TM, N- fixing), Potentilla erecta (PR), Picea abies (PA), Frangula alnus (FA), and Juniperus communis (JC) (n = 8 for each plant). ECM fungi (~) are Helvella acetabulum (H, n = 8) and Inocybe sp., (I, n = 2); saprotrophic fungi (~) are Auriscalpium vulgare (A, n = 3), Mycena sp. (M, n = 2), and Strobilurus tenacellus (S, n = 7).

Table 5. Percentages of nitrogen (% NdF) and carbon (% CdF) derived from fungi in the leaves of green orchids from on MX neottioids (Gebauer 2005), while low tree density Pussa, as calculated from d15Nord13C values, respectively, could reduce C availability in ECM fungi. and based on a linear two-source isotopic mixing model. Evidence for the MX strategy in C. longifolia was also found in an independent study on orchid physiology in Orchid species % NdF %CdF France (M.-A. Selosse, unpublished data) and may explain Albino Cephalanthera 100 (reference) 100 (reference) the tendency of this plant (i) to have long underground longifolia growth before first flowering, and (ii) to stay for some years Green C. longifoliaa 86 33 in a dormant state (i.e., underground, without forming Epipactis helleborineb 100 28 shoots; Rasmussen 1995). This dormant state is known for Orchis masculab 57 Negligible green genets from this and other Estonian C. longifolia Ophrys insectiferab 48 Negligible (Pu¨ttsepp and Kull 1997). Gymnadenia conopseab 42 Negligible b Listera ovata 30 Negligible Evolution of mixotrophic and mycoheterotrophic aAs compared with co-occurring plants from shaded places. strategies bAs compared with co-occurring plants from sunny places. Our study points to a particular link between Cephalan- thera spp. and thelephoroid ECM fungi. cies at Pussa is the neottioid Epipactis helleborine, which , a species phylogenetically contains 28% of fungal C, while Gebauer and Meyer (2003) close to C. longifolia (Fig. 7; Bateman et al. 2004), also as- obtained a value of 43% for this species at another forest sociates with thelephoroids (Bidartondo et al. 2004; Julou et site. The apparent autotrophy of Listera ovata at Pussa con- al. 2005), while the MH C. austiniae is highly specific to trasts with the range of values (6%–27%) found at several thelephoroids (Taylor and Bruns 1997). Thus, a preference sites by Gebauer and Meyer (2003). These relatively low C for thelephoroids may be plesiomorphic in the Cephalan- gains of neottioids in comparison with other studies (see thera genus, and should be tested on other Cephalanthera also Bidartondo et al. 2004) may be explained by the com- species. paratively higher light supply at this site, even in its shaded MH plants often exhibit mycorrhizal specificity (Taylor et areas (average Ellenberg light indicator: 5.7). Indeed, tree al. 2002; Leake 2004). However, recent work on symbiont density is still low in this recently colonized site (Fig. 1b). diversity in chlorophyllous orchids challenged the concept Light may enhance the contribution of photosynthesis to that MH orchids are more specific than chlorophyllous ones plant biomass, as suggested by comparisons between studies (Otero et al. 2002, 2004; McCormick et al. 2004; Shefferson

# 2006 NRC Canada 1474 Can. J. Bot. Vol. 84, 2006

Fig. 7. A phylogenetic tree of the Neottieae tribe, redrawn from fungi may enable adaptation to shady forest environments Bateman et al. (2004) and restricted to species for which mycohe- by allowing use of tree C. Moreover, ECM partners and the terotrophic (MH), mixotrophic (MX), or autotrophic strategies are MX strategy at the adult stage are the most parsimonious described. Asterisks indicate bootstrap supports greater than 75% plesiomorphic states in the sister clade (Epipactis + Neottia in Bateman et al. (2004). Thin branches indicate autotrophic taxa; + , Fig. 7). As a result, these features likely thick black lines and bold characters represent MH taxa; grey lines arose in the Neottieae common ancestor (arrowed on indicate MX taxa as shown by isotopic or gas exchange investiga- Fig. 7). An unclear situation occurs in E. palustris, which tions (Gebauer and Meyer 2003; Bidartondo et al. 2004; Julou et grows in open places (Rasmussen 1995), germinates with al. 2005; Girlanda et al. 2006; this study). a, denotes the existence true saprotrophic Rhizoctonias (Rasmussen 1995), and was of albinos; h, denotes the existence of hypogeous, dormant indivi- found by Bidartondo et al. (2004) to be autotrophic and as- duals (Salmia 1988, 1989a, 1989b; Selosse et al. 2004; Julou et al. sociated with rhizoctonias, but not ECM fungi. More de- 2005; this study and M.-A. Selosse, unpublished data). Question tailed Epipactis phylogenies will clarify whether a marks indicate taxa and ancestral states uncertain for trophic type. reversion to the usual orchid ecology occurred (i) in the The arrow points to the Neottieae ancestor where ectomycorrhizal Epipactis common ancestor, followed by a second rever- (ECM) partners and the MX strategy were acquired in the most sion to the MX strategy with ECM fungi in some Epipac- parsimonious scenario. tis spp., or (ii) only in the clade leading to E. palustris and related species. In any case, association with ECM partners and the MX strategy at adulthood remains the most parsi- monious evolutionary scenario for the neottioid ancestor in light of the available phylogeny (Fig. 7). MH species in neottioids are thus nested in MX clades (Fig. 7), suggest- ing that the MH strategy was selected in MX species. In this context, albinos at low frequency can be interpreted as a transition to the MH strategy, successful in terms of individual physiology but unsuccessful in evolutionary terms, since they do not invade the population. To summarize, our work demonstrates the stability of the albino phenotype and justifies the comparison between green individuals and albinos. Our data support previous conclusions regarding neottioid populations harbouring albi- nos: (i) the orchid is associated with ECM fungi, some of which cannot be found in roots by TEM investigation; (ii) some of these ECM fungi colonize surrounding tree roots at the same site; (iii) green individuals harbour the same my- corrhizal partners as their albino conspecifics; (iv) they are MX, with a significant heterotrophy level (33% here) that can be calculated by comparison with albinos. By demon- strating a new case of MX neottioid, we further support a ‘‘MX-first’’ model in the evolution of MH species in this tribe. However, orchids are unusual plants because of their MH germination, a feature that may predispose them to et al. 2005). Nevertheless, physiological data (e.g., d13C ra- evolve MX and MH strategies. Indeed, albino variants are tios or analysis of CO2 exchanges) are lacking for these spe- unknown in other plant taxa. Whether MX plants, with high cies, and some may be MX. From our data, the relation heterotrophy level, exist in non-orchid photosynthetic taxa between specificity and heterotrophy level is unclear, when remains an open question. comparing the Pussa population to the C. damasonium pop- ulation we previously investigated in France (Julou et al. Acknowledgements 2005). While the first is quite specific to thelephoroids and The authors thank Kaarel Sepp for his kind help in fol- 33% heterotrophic, the latter seems less specific (several lowing the development of the achlorophyllous plants on Cortinariaceae were found in addition to thelephoroids) and his home coast; Krista Lo˜hmus, Mai Olesk, To˜nu Talvi, half heterotrophic. We suspect, from our data and the survey and Arnoud de Vries for soil analyses; Tiiu Kull for help by Gebauer (2005), that the heterotrophy level may correlate in postage of samples; David Marsh for English correc- less with specificity than with environmental light condi- tions; Fred Ju¨ssi for Fig. 1; Prune Pellet, Thomas Julou, tions for MX species. and Annie Tillier for help in molecular investigations; Our data and the findings of others (Bidartondo et al. Bastian Burghardt for help with sample preparation in iso- 2004; Julou et al. 2005) suggest that the shift to ECM part- tope abundance analysis; and two anonymous reviewers for ners and the MX strategy may be ancestral in the genus improving the manuscript. This work was funded by grants Cephalanthera (Fig. 7). In addition, Ceratobasidiaceae from the Centre national de la recherche scientifique and found in C. longifolia show that rhizoctonia availability is the Socie´te´ franc¸aise d’orchidophilie to M.-A. Selosse and not a limiting factor. Instead, they support the hypothesis grants from The National Council of Research (Biodiver- by Taylor and Bruns (1997) that association with ECM sity Project) to P. Bonfante.

# 2006 NRC Canada Abadie et al. 1475

References New Phytol. 166: 639–653. doi:10.1111/j.1469-8137.2005. 01364.x. PMID:15819926. Akiyama, K., Matsuzaki, K., and Hayashi, H. 2005. Plant ses- Jumpponen, A., and Trappe, J.M. 1998. Dark septate endophytes: a quiterpenes induce hyphal branching in arbuscular mycorrhi- review of facultative biotrophic root-colonizing fungi. New Phy- zal fungi. Nature, 435: 824–827. doi:10.1038/nature03608. tol. 140: 295–310. doi:10.1046/j.1469-8137.1998.00265.x. PMID:15944706. Koske, R.E., and Gemma, J.N. 1989. A modified procedure for stain- Andersen, T.F. 1996. A comparative taxonomic study of Rhizocto- ing roots to detect VA mycorrhizas. Mycol. Res. 92: 486–505. nia sensu lato employing morphological, ultrastructural and mo- Leake, J.R. 2004. Myco-heterotroph/epiparasitic plant interactions lecular methods. Mycol. Res. 100: 1117–1128. with ectomycorrhizal and arbuscular mycorrhizal fungi. Curr. Bateman, R.M., Hollingsworth, P.M., Squirrell, J., and Hollings- Opin. Plant Biol. 7: 422–428. doi:10.1016/j.pbi.2004.04.004. worth, M.L. 2004. Tribe Neottieae: phylogenetics. In Genera PMID:15231265. Orchidacearum (4): (part 1). Edited by Mairold, F., and Weber, F. 1950. Notiz u¨ber Cephalanthera albi- A.M. Pridgeon, P.L. Cribb, M.W. Chase, and F.N. Rasmussen. nos. Protoplasma, 39: 275–277. doi:10.1007/BF01248860. Oxford University Press, Oxford, UK. pp. 487–495. Beyrle, H.F., and Smith, S.E. 1993. Excessive carbon prevents McCormick, M.K., Whigham, D.F., and O’Neill, J. 2004. My- greening of leaves in mycorrhizal seedling of the terrestrial corrhizal diversity in photosynthetic terrestrial orchids. New orchid Orchis morio. Lindleyana, 8: 97–99. Phytol. 163: 425–438. doi:10.1111/j.1469-8137.2004.01114.x. Bidartondo, M.I., Burghardt, B., Gebauer, G., Bruns, T.D., and McKendrick, S.L., Leake, J.R., and Read, D.J. 2000. Symbiotic Read, D.J. 2004. Changing partners in the dark: isotopic and mo- germination and development of myco-heterotrophic plants in lecular evidence of ectomycorrhizal liaisons between forest orch- nature: transfer of carbon from ectomycorrhizal Salix repens ids and trees. Proc. R. Soc. Lond. B. Biol. Sci. 271: 1799–1806. and Betula pendula to the orchid Corallorhiza trifida through Delforge, P. 1998. Des Epipactis helleborine (L.) Crantz de´pourvus shared hyphal connections. New Phytol. 145: 539–548. doi:10. de chlorophylle dans les environs de Bruxelles. Natural. Belges, 1046/j.1469-8137.2000.00592.x. 79: 124–130. Molvray, M., Kores, P.J., and Chase, M.W. 2000. Polyphyly of Ellenberg, H., Weber, H.E., Du¨ll, R., Wirth, V., Werner, W., and mycoheterotrophic orchids and functionnal influences on floral Paulißen, D. 1991. Zeigerwerte von Pflanzen in Mitteleuropa. and molecular characters. In Monocots: systematics and evolu- Verlag Erich Goltze, Go¨ttingen, Germany. tion. Edited by K.L. Wilson and D.A. Morrison. CSIRO Publish- Farquhar, G.O., O’Leary, M.H., and Berry, J.A. 1982. On the rela- ing, Sydney,Australia. pp. 441–448. tionship between carbon isotope discrimination and the intercel- Otero, J.T., Ackerman, J.D., and Bayman, P. 2002. Diversity and lular carbon dioxide concentration in leaves. Aust. J. Plant host specificity of endophytic Rhizoctonia-like fungi from tropi- Physiol. 9: 121–137. cal orchids. Am. J. Bot. 89: 1852–1858. Gebauer, G. 2005. Partnertausch im dunklen Wald - Stabile Isotope Otero, J.T., Ackerman, J.D., and Bayman, P. 2004. Differences in geben neue Einblicke in das Erna¨hrungsverhalten von Orchid- mycorrhizal preferences between two tropical orchids. Mol. een. In Auf Spurensuche in der Natur: Stabile Isotope in der Ecol. 13: 2393–2404. doi:10.1111/j.1365-294X.2004.02223.x. o¨kologischen Forschung. Edited by Bayer. Akademie d.Wis- PMID:15245412. ¨ senschaften, Rundgespra¨che der Kommission fu¨rO¨ kologie, Pu¨ttsepp, U. 1993. Kahvatu orhidee. Eesti Loodus, 11–12: 357. Band 30. Verlag Dr. Friedrich Pfeil, Mu¨nchen, Germany. pp. Pu¨ttsepp, U¨ ., and Kull, T. 1997. Cephalanthera longifolia and 55–67. in Estonia. Botanica Lithuanica, Gebauer, G., and Meyer, M. 2003. 15N and 13C natural abundance 25(Suppl. 1): 133–135. of autotrophic and myco-heterotrophic orchids provides insight Rasmussen, H.N. 1995. Terrestrial orchids, from seed to myco- into nitrogen and carbon gain from fungal association. New Phy- trophic plant. Cambridge University Press, New York. tol. 160: 209–223. doi:10.1046/j.1469-8137.2003.00872.x. Rasmussen, H.N. 2002. Recent developement in the study of Gebauer, G., and Schulze, E.D. 1991. Carbon and nitrogen isotope orchid mycorrhiza. Plant Soil, 244: 149–163. doi:10.1023/ ratios in different compartments of a healthy and a declining Pi- A:1020246715436. cea abies forest in the Fichtelgebirge, NE Bavaria. Oecologia, Rasmussen, H.N., and Whigham, D.F. 2002. Phenology of roots 87: 198–207. doi:10.1007/BF00325257. and mycorrhiza in five orchid species differing in phototrophic Girlanda, M., Selosse, M.A., Cafasso, D., Brilli, F., Delfine, S., strategy. New Phytol. 154: 797–807. doi:10.1046/j.1469-8137. Fabbian, R., Ghignone, S., Pinelli, P., Segreto, R., Loreto, F., 2002.00422.x. Cozzolino, S., and Perotto, S. 2006. Inefficient photosynthesis Renner, O. 1938. U¨ ber blasse, saprophytische Cephalanthera alba in the Mediterranean orchid Limodorum abortivum (L.) Swartz und Epipactis latifolia. Flora (Jena), 132: 225–233. is mirrored by specific association to ectomycorrhizal Russula- Renner, O. 1943. Notiz u¨ber blasse Cephalanthera und Tozzia. ceae. Mol. Ecol. 15: 491–504. doi:10.1111/j.1365-294X.2005. Flora (Jena), 136: 309–312. 02770.x. PMID:16448415. Ricek, E.W., and Gutermann, W. 1971. Beobachtungen an Epipac- Hietala, A.M., and Sen, R. 1996. Rhizoctonia associated with forest tis purpurata Sm. Und ihrer chlorophyllfreien Form. Mitt. Nat- trees. In Rhizoctonia species: , molecular biology, urwiss. Ver. Steiermark, 100: 294–298. ecology, pathology and disease control. Edited by B. Sneh, S. Russell, J., and Bulman, S. 2005. The liverwort Marchantia Jabaji-Hare, S. Neate, and G. Dijst. Kluwer Academic Publish- foliacea forms a specialized symbiosis with AM fungi in ers, Dordrecht. pp. 351–358. the genus Glomus. New Phytol. 165: 567–579. doi:10. Horan, D.P., and Chilvers, G.A. 1990. Chemotropism — the key to 1111/j.1469-8137.2004.01251.x. PMID:15720667. ectomycorrhizal formation? New Phytol. 116: 297–301. doi:10. Salmia, A. 1988. Endomycorrhizal fungus in chlorophyll-free and 1111/j.1469-8137.1990.tb04717.x. green forms of the terrestrial orchid Epipactis helleborine. Kar- Julou, T., Burghardt, B., Gebauer, G., Berveiller, D., Damesin, stenia, 28: 3–18. C., and Selosse, M.A. 2005. Mixotrophy in orchids: insights Salmia, A. 1989a. Features of endomycorrhizal infection of chloro- from a comparative study of green individuals and non- phyll-free and green forms of Epipactis helleborine (Orchida- photosynthetic individuals of Cephalanthera damasonium. ceae). Ann. Bot. Fenn. 26: 15–26.

# 2006 NRC Canada 1476 Can. J. Bot. Vol. 84, 2006

Salmia, A. 1989b. General morphology and anatomy of chloro- Smith, S.E., and Read, D.J. 1997. Mycorrhizal symbiosis. 2nd ed. phyll-free and green forms of Epipactis helleborine (Orchida- Academic Press, San Diego, Calif. ceae). Ann. Bot. Fenn. 26: 95–105. Swofford, D.L. 2001. PAUP*. Phylogenetic analysis using parsi- Sati, S.C., and Belwal, M. 2005. Aquatic hyphomycetes as endo- mony (*and other methods). Sinauer Associates, Sunderland, phytes of riparian plant roots. Mycologia, 97: 45–49. Mass. PMID:16389955. Taylor, D.L. 2004. Myco-heterotroph-fungus marriages — is fide- Selosse, M.A., Bauer, R., and Moyersoen, B. 2002a. Basal hyme- lity over-rated? New Phytol. 163: 217–221. doi:10.1111/j.1469- nomycetes belonging to the Sebacinaceae are ectomycorrhizal 8137.2004.01124.x. on temperate deciduous trees in silva: microscopic and molecu- Taylor, D.L., and Bruns, T.D. 1997. Independent, specialized inva- lar evidence. New Phytol. 155: 183–195. doi:10.1046/j.1469- sions of ectomycorrhizal mutualism by two non photosynthetic 8137.2002.00442.x. orchid. Proc. Natl. Acad. Sci. U.S.A. 94: 4510–4515. doi:10. Selosse, M.A., Weiss, M., Jany, J.L., and Tillier, A. 2002b. 1073/pnas.94.9.4510. PMID:9114020. Communities and populations of sebacinoid basidiomycetes as- Taylor, D.L., Bruns, T.D., Leake, J.R., and Read, D.J. 2002. My- sociated with the achlorophyllous orchid Neottia nidus-avis corrhizal specificity and function in myco-heterotrophic plants. (L.) L.C.M. and neighbouring tree ectomycorrhizae. Mol. In Mycorrhizal ecology. Edited by S.I. Van der Heijden MGA. Ecol. 11: 1831–1844. doi:10.1046/j.1365-294X.2002.01553.x. Springer-Verlag, Berlin. pp. 375–413. PMID:12207732. Taylor, D.L., Bruns, T.D., Szaro, T.M., and Hodges, S.A. 2003. Di- Selosse, M.A., Faccio, A., Scappaticci, G., and Bonfante, P. 2004. vergence in mycorrhizal specialization within Hexalectris spi- Chlorophyllous and achlorophyllous specimens of Epipactis mi- cata (Orchidaceae), a nonphotosynthetic desert orchid. Am. J. crophylla (Neottieae, Orchidaceae) are associated with ectomy- Bot. 90: 1168–1179. corrhizal septomycetes, including truffles. Microb. Ecol. 47: Trudell, S.A., Rygiewicz, P.T., and Edmonds, R.L.S. 2003. Nitro- 416–426. PMID:15107957. gen and carbon stable isotope abundances support the myco-het- Selosse, M.-A., Richard, F., He, X., and Simard, S. 2006. Mycor- erotrophic nature and host-specificity of certain achlorophyllous rhizal networks: les liaisons dangereuses. Trends Ecol. Evol. In plants. New Phytol. 160: 391–401. doi:10.1046/j.1469-8137. press. doi:10.1016/j.tree.2006.07.003. PMID:16843567. 2003.00876.x. Sen, R., Hietala, A.M., and Zelmer, C.D. 1999. Common Vujanovic, V., St Arnaud, M., and Barabe´, D. 2000. Phialophora anastomosis and internal transcribed spacer RFLP group- victorinii sp. nov., endophyte of Cypripedium parviflorum. My- ings in binucleate Rhizoctonia isolates representing root cologia, 92: 571–576. endophytes of Pinus sylvestris, Ceratorhiza spp. from Weiss, M., Selosse, M.A., Rexer, K.H., Urban, A., and Ober- orchid mycorrhizas and a phytopathogenic anastomosis winckler, F. 2004. Sebacinales: a hitherto overlooked cosm of group. New Phytol. 144: 331–341. doi:10.1046/j.1469-8137. heterobasidiomycetes with a broad mycorrhizal potential. My- 1999.00522.x. col. Res. 108: 1003–1010. doi:10.1017/S0953756204000772. Shefferson, R.P., Weiss, M., Kull, T., and Taylor, D.L. 2005. PMID:15506013. High specificity generally characterizes mycorrhizal associa- Zba¨ren, E. 1968. Botanische Seltenheit: Schwertbla¨ttriges Waldvo¨- tion in rare lady’s slipper orchids, genus Cypripedium. glein ohne Chlorophyll. Kosmos, 64: 307. Mol. Ecol. 14: 613–626. doi:10.1111/j.1365-294X.2005. 02424.x. PMID:15660950.

# 2006 NRC Canada Abadie et al. 1477

Appendix A

Table A1. Root fungi found by cloning on a single Cephalanthera longifolia indi- vidual (i.e., not shared with the other investigated root systems).

Putative taxonomic identity GB accession No. Putative ecology In individual G1 Unknown basidiomycete sp.1 DQ182429 Endophytic Helotiale sp.3 DQ182430 Endophytic Geoglossum sp.1 DQ182431 Endophytic? Unknown septomycete sp.1 DQ182432 ? Helotiale sp.4 DQ182433 Endophytic In individual G2 Exophiala sp.1 DQ182434 Endophytic In individual G3 Sebacinaceae sp.2 DQ182435 Ectomycorrhizal Glomus sp.1 DQ182436 Endomycorrhizal Glomus sp.2 DQ182437 Endomycorrhizal Leptodontidium sp.2 DQ182438 Endophytic Leptodontidium sp3 DQ182439 Endophytic Unknown Septomycete sp.2 DQ182440 ? Helotiale sp.5 DQ182441 Endophytic Sordariale sp.1 DQ182442 Saprotrophic Cortinariaceae sp.3 DQ182443 Ectomycorrhizal Exophiala sp.2 DQ182444 Endophytic Clavariaceae sp.1 DQ182445 Ectomycorrhizal Tetracladium sp.2 DQ182446 Endophytic Cortinariaceae sp.4 DQ182447 Ectomycorrhizal In individual A1 Nectriaceae sp.3 DQ182448 Endophytic Unknown ascomycete sp.2 DQ182449 ? Cortinariaceae sp.5 DQ182450 Ectomycorrhizal Pleosporale sp.1 DQ182451 Endophytic Unknown ascomycete sp.3 DQ182452 ? In individual A2 Unknown agaricale sp.1 DQ182453 ? Amphinema sp.2 DQ182454 Ectomycorrhizal Unknown septomycete sp.2 DQ182455 ? Orbiliale sp.1 DQ182456 Endophytic Pezizale sp.1 DQ182457 Ectomycorrhizal Helotiale sp.6 DQ182458 Endophytic Pezizale sp.2 DQ182459 Endophytic Thanatephorus sp.1 DQ182460 Rhizoctonia Phialophora sp.1 (Helotiales) DQ182461 Endophytic Sordariale sp.2 DQ182462 Saprotrophic Phialophora sp.2 (Helotiales) DQ182463 Endophytic

# 2006 NRC Canada