15N and 13C natural abundance of autotrophic and myco-heterotrophic orchids provides insight into nitrogen and carbon gain from fungal association


  • G. Gebauer,

    Corresponding author
    1. Lehrstuhl für Pflanzenökologie, Universität Bayreuth, 95440 Bayreuth, Germany;
    Search for more papers by this author
  • M. Meyer

    1. Lehrstuhl für Pflanzenökologie, Universität Bayreuth, 95440 Bayreuth, Germany;
    2. Present address: Lehrstuhl für Physikalische Chemie I, Universität Bayreuth, 95440 Bayreuth, Germany
    Search for more papers by this author

Author for correspondence: Gerhard Gebauer Tel: +49 921552060 Fax: +49 921552564 Email: gerhard.gebauer@uni-bayreuth.de


  • • Whereas mycorrhizal fungi are acknowledged to be the sources of nitrogen (N) and carbon (C) in achlorophyllous (myco-heterotrophic) orchids, the sources of these elements in autotrophic orchids are unknown. We have determined the stable isotope abundance of N and C to quantify their gain from different sources in these two functional groups and in non-orchids of distinctive mycorrhizal types.
  • • Leaves of each plant were collected from four forest and four grassland sites in Europe. The N and C isotope abundance, and total N concentrations of their tissues and of associated soils were determined.
  • • Myco-heterotrophic orchids were significantly more enriched in 15N (ɛMHO-R= 11.5‰) and 13C (ɛMHO-R= 8.4‰) than co-occurring non-orchids. δ15N and δ13C signatures of autotrophic orchids ranged from values typical of non-orchids to those more representative of myco-heterotrophic orchids.
  • • Utilization of fungi-derived N and C probably explains the relative 15N and 13C enrichment in the myco-heterotrophs. A linear two-source isotopic mixing model was used to estimate N and C gain of autotrophic orchids from their fungal associates. Of the putatively autotrophic species, Cephalanthera damasonium obtained the most N and C by the fungal route, but several other species also fell into the partially myco-heterotrophic category.


The orchid family with an estimated 20 000 species is one of the largest plant families and is believed to have undergone rapid diversification and speciation (Dressler, 1993). Many orchid species are rare, endangered or even facing extinction. Thus, orchids are of interest in the ongoing discussion on biodiversity and species conservation. The family is cosmopolitan with species adapted to a wide range of habitats from tropical rainforests to boreal tundra.

The Orchidaceae have an intimate association with fungi. The obligate dependence of orchids on fungi at least for a part of their life cycle has been known for 100 yrs: Noel Bernard identified Rhizoctonia as an orchid symbiont and demonstrated its important role in the germination of the extremely small and endospermless orchid seeds (Bernard, 1909). In later stages of the orchid life cycle its dependence on mycorrhizal fungi is highly variable. Some tropical epiphytic orchids are non-mycorrhizal when adult, while many terrestrial orchids remain mycorrhizal (Burgeff, 1932; Smith & Read, 1997). In the germinating seed stage, the orchids are dependent on the fungus for their complete nutrition. In adult orchids the mycorrhiza is assumed to be important for mineral nutrition (mainly of nitrogen and phosphorus), largely because the root systems of many terrestrial orchids are poorly developed (Smith & Read, 1997; Brundrett, 2002). The mineral nutrition via the fungus may be realized by the acquisition and transfer of inorganic and/or organic N and P compounds from the soil to the plant root cells through the fungal mycelium. However, it has also been demonstrated that orchid root cells ‘digest’ fungal mycelium, that is they incorporate compounds from the fungal tissue into their own biomass (Burgeff, 1932; Smith & Read, 1997). The quantitative importance of this digestion for the mineral nutrition of orchids is currently not clear. However, orchids will also gain C from the fungus if they digest fungal mycelium or if they receive organic compounds from the fungal mycelium. It is not yet clear whether this C gain through mycorrhizas is of any quantitative importance for chlorophyllous orchids. However, there is some indirect evidence that those chlorophyllous terrestrial orchids that spend several years underground continue to gain C via the fungus (Smith & Read, 1997). By extrapolation from other mycorrhizal types, it might be assumed that C-autotrophic orchids also transfer organic C compounds to the associated fungi. Attempts to demonstrate C transfer from orchids to fungi, however, have failed (Smith & Read, 1997).

The Orchidaceae also contain about 200 achlorophyllous species, which obtain their C completely from symbiotic fungi, and are therefore called myco-heterotrophic (Leake, 1994). They depend also on the fungus for their mineral nutrition. Some myco-heterotrophic orchids are connected to basidiomycetes, which form ectomycorrhizas with trees (Zelmer & Currah, 1995; Taylor & Bruns, 1997; McKendrick et al., 2000b; McKendrick et al., 2002). Furthermore, C transfer can occur from ectomycorrhizal trees to myco-heterotrophic orchids through linked fungal mycelia of an ectomycorrhizal basidiomycete (McKendrick et al., 2000a).

Our present knowledge of the nutritional interactions between orchids and their mycorrhizal fungi is mainly based on: (first) microscopy of mycorrhizal orchid roots, (second) tracer techniques with radioactive isotopes to investigate principles of nutrient fluxes in laboratory experiments and (third) classical and molecular biological approaches to identify fungal species associated with orchids. Surprisingly, variation in stable isotope natural abundance of N and C has been neglected as a tool to study nutrient fluxes between fungi and orchids under field conditions, despite the fact that N and C in fungi are isotopically distinguished from N and C of accompanying nonorchid vegetation. The isotopic distinction between fungi and a selection of forest plants was first described by Gebauer & Dietrich (1993) for N and by Gleixner et al. (1993) for C. This isotopic distinction has been confirmed repeatedly and extended to a broad spectrum of fungi and forest plants forming different types of mycorrhizas (Högberg et al., 1996; Michelsen et al., 1998; Gebauer & Taylor, 1999; Högberg et al., 1999a; Kohzu et al., 1999; Henn & Chapela, 2001; Hobbie et al., 2001), but not to include orchids.

The isotopic distinction between fungi and non-orchid plants provides the opportunity to test whether and to estimate how much N and C autotrophic orchids gain from fungi. The principle of this analysis is based on the assumption that non-orchids are not capable of digesting fungal mycelium and therefore do not receive N and C compounds that have the fungi-specific isotope signature, and that myco-heterotrophic orchids digest mycelium of their fungal hosts, and therefore are composed of N and C compounds that have the fungi-specific isotope signature. A linear two-source isotopic mixing model with the isotopic composition of nonorchids and of completely myco-heterotrophic orchids as endpoints may then be used to estimate the N and C gain of the autotrophic orchids.

In a similar manner, two-source isotopic mixing models have already been successfully applied in various ecological field investigations including quantifications of plant N gain from symbiotic N2 fixation (Shearer & Kohl, 1986; Schulze et al., 1991b), the N gain from insect capture by carnivorous plants (Schulze et al., 1991a; Schulze et al., 2001), the plant above-ground uptake of S and N compounds from atmospheric deposition (Gebauer et al., 1994; Ammann et al., 1999; Stewart et al., 2002), the absorption of ant-provided C and N by plants (Treseder et al., 1995) and the C gain of hemiparasites with C3 photosynthesis on hosts with C4 (Press et al., 1987) or CAM photosynthesis (Schulze et al., 1991c; Ziegler, 1996).

This investigation aimed to test (first) whether myco-heterotrophic orchids and autotrophic orchids are distinguished by their N and C isotope signature from each other and from co-occurring non-orchids, (second) whether terrestrial orchids and nonorchids from different European forest types (coniferous and broadleaf forest) and from grassland sites are distinguished by their isotopic pattern, and (third) whether the two-source isotopic mixing model is applicable to quantify the N and C gain of autotrophic orchids from their fungal association. Furthermore, total N concentrations in the leaves of orchids and nonorchids were investigated.

Materials and Methods

Sites of investigation

Plant and soil samples were taken from four forest and four grassland sites in Europe. The forest site in Thezan (T) is located in southern France (about 20 km WSW of Narbonne; 43°07′ N 02°45′ E) at an elevation of 170 m above sea level (a.s.l.). This forest is dominated by Pinus pinaster. In addition to the conifer, several deciduous and evergreen broadleaf trees occur: Castanea sativa, Quercus coccifera, Q. ilex and Q. pubescens. Furthermore, the stand has a dense understorey dominated by ericaceous shrubs. The soil type is a chromic luvisol with a pH of 5.7 developed over Permian sandstone. The site has a Mediterranean climate with summer drought. The mean annual temperature is 15°C and the mean annual precipitation is 800 mm. For further details of the site description see Persson et al. (2000). Samples were collected from this site in April 1993. The three other forest sites are located in the Veldensteiner Forst in NE Bavaria, Germany (about 30 km S of Bayreuth; between 49°38′ N and 49°40′ N and between 11°22′ E and 11°31′ E) at elevations between 460 and 600 m a.s.l. One of these sites is a dense broadleaf forest dominated by Fagus sylvatica with some Acer platanoides and a sparse and patchy cover of understorey vegetation (V1). The two other sites are open Pinus sylvestris forests with a small percentage of Picea abies and Fagus sylvatica trees mainly in the understorey and a dense and species-rich herbaceous ground vegetation (V2, V3). Soils at all three sites are lithic leptosols originating from Jurassic dolomite with a pH of 6 (Hemp, 1995). Samples from these sites were collected in June 1995 (V1, V2) and 1997 (V3), respectively. The four grassland sites of this investigation are located in the Frankenwald in NE Bavaria, Germany (about 50 km NNW of Bayreuth; between 50°17′ and 50°24′ N and between 11°21′ and 11°24′ E) at elevations between 450 and 600 m a.s.l. The grassland sites are distinguished by their species composition (Balzer, 2000). On three of the four sites the rare orchid Dactylorhiza sambucina grows either together with other orchid species (F1) or not accompanied by other orchids (F3, F4). On one grassland site (F2) two orchid species occur without Dactylorhiza sambucina. The soils are dystric cambisols originating from acidic paleozoic sediments with pH between 3.8 and 4.5. Samples were collected in May 1994 (F1, F4) and June 1995 (F2, F3), respectively. The climate of the forest and grassland sites in NE Bavaria is humid continental with mean annual temperature of 5–8°C and mean annual precipitation of 600–1000 mm (Balzer, 2000).

Sampling scheme and investigated species

Within each of the eight sites, four (T, V3) or five 1 m2 plots (all other sites) were selected using the following criteria: (a) at least one (if possible more) orchid species (of, if possible, different functional groups); (b) > 1 non-orchid (reference) plant species of (if possible) different functional groups and growth forms growing close to the orchids; for trees and shrubs, reference plants were always small (seedlings or small saplings). Only one leaf of each orchid plant was sampled to minimize damage to the orchids. For reference with large-sized leaves, one leaf per plant was sampled; for species with small-sized leaves, several leaves or needles of one individual were pooled. For evergreen species, only current-year leaves or needles were sampled. One soil sample from the humus layer (Of/Oh) or from the humus layer + mineral soil (Of/Oh/A0–5; only V3) of each forest site plot or from the uppermost 5 cm in the soil of each grassland site plot was taken, giving a total of 38 samples. As a consequence of this sampling scheme, the number of sampled individuals per species was in some cases smaller than the number of plots per site (Table 1).

Table 1.  List of species (orchids and non-orchids), functional groups, life forms (according to the system of Raunkiaer, 1910), leaf persistence types, N figures (according to Ellenberg et al., 1991) and number of samples (leaf material and soil samples) as collected on four forest and grassland sites in Europe
SpeciesFunctional groupLife formLeaf persistenceEllenberg N figureForest sitesNumber of samples (n) Grassland sites
  1. All samples were analyzed for δ15N and total N concentration. In some cases the number of samples for δ13C analysis was smaller. These cases are indicated by numbers in brackets. Nomenclature of the plant species follows Tutin et al. (1964–80) except for Alchemilla vulgaris agg. and Chrysanthemum leucanthemum L. (Schmeil, 2000). The functional groups include ECM ectomycorrhizal plants, ERI ericoid mycorrhizal plants, AM/NON arbuscular mycorrhizal or non-mycorrhizal plants, FIX potentially N2 fixing plants, AO autotrophic orchids, and MHO myco-heterotrophic orchids. The life forms include Phanero Phanerophyte, Nanophanero Nanophanerophyte, W. chamae Woody chamaephyte, Hemicrypto Hemicryptophyte, Geo Geophyte. Leaf persistence types include d deciduous, e evergreen. The Ellenberg N figures range from 1 = indicating extremely N-poor sites to 9 = indicating sites with extreme N oversupply, × indicates indifferent behaviour; species with missing N figures are not included in the list of Ellenberg et al. (1991). Site codes are given in the Materials and Methods section.

Castanea sativaECMPhanerod×4 (0)       
Fagus sylvaticaECMPhanerod× 53     
Picea abiesECMPhaneroe×  5     
Pinus pinasterECMPhaneroe 4 (0)       
Quercus cocciferaECMPhaneroe 4 (0)       
Quercus ilexECMPhaneroe×4 (0)       
Quercus pubescensECMPhaneroe×4 (0)       
Calluna vulgarisERIW. chamaee14 (0)       
Erica arboreaERINanophaneroe 4 (0)       
Acer platanoidesAM/NONPhanerod× 5      
Alchemilla vulgarisAM/NONHemicryptod6    22  
Anthoxantum odoratumAM/NONHemicryptod×       5
Briza mediaAM/NONHemicryptod2  4     
Chrysanthemum leucanthemumAM/NONHemicryptod3    55 4
Convallaria majalisAM/NONGeod4 32     
Dactylis glomerataAM/NONHemicryptod6  1     
Daphne mezereumAM/NONNanophanerod5 3      
Euphorbia cyparissiasAM/NONHemicryptod3  5     
Galium odoratumAM/NONHemicryptod5 2      
Geranium sylvaticumAM/NONHemicryptod7     5  
Hieracium pilosellaAM/NONHemicryptod2       1
Juniperus communisAM/NONNanophaneroe×      5 
Lychnis viscariaAM/NONHemicryptod2  5     
Melica nutansAM/NONGeod3 5      
Meum athamanticumAM/NONHemicryptod3    5  5
Phyteuma spicatumAM/NONHemicryptod5     5  
Plantago lanceolataAM/NONHemicryptod×    555 
Plantago mediaAM/NONHemicryptod3     3  
Polygala chamaebuxusAM/NONW. chamaee2   4    
Polygala vulgarisAM/NONHemicryptod2      55
Sanguisorba minorAM/NONHemicryptod2      5 
Sesleria albicansAM/NONHemicryptod3   4    
Sorbus aucupariaAM/NONNanophanerod× 2      
Lathyrus montanusFIXHemicryptod2     5 5
Lotus corniculatusFIXHemicryptod3    5   
Trifolium montanumFIXHemicryptod2      5 
Trifolium pratenseFIXHemicryptod×    2   
Cephalanthera damasoniumAOGeod44 (0)5      
Cephalanthera rubraAOGeod4  3     
Dactylorhiza sambucinaAOGeod2    1 55
Epipactis atrorubensAOGeod2   1    
Epipactis helleborineAOGeod5   1    
Listera ovataAOGeod7  31 (0)23  
Ophrys insectiferaAOGeod3  24 (2)    
Orchis masculaAOGeod×    55  
Orchis ustulataAOGeod3    5   
Platanthera bifoliaAOGeod×  31    
Limodorum abortivumMHOGeod34 (0)       
Neottia nidus-avisMHOGeod5 5      
Plant samples    36 (0)353616 (13)42333030
Soil samples    4 (0)5545555

In total 68 samples from 12 orchid species were investigated (Table 1). The number of orchid samples collected from the respective sites ranged from five to 13 from one to five species. The orchid species were either autotrophic (AO) or myco-heterotrophic (MHO). MHO were found in two of the forest sites. One hundred and ninety samples from 37 species of non-orchids were collected, with 8–29 samples from two to seven species per site. The non-orchids were grouped into four functional types (Harley & Harley, 1987): ectomycorrhizal plants (ECM), ericoid mycorrhizal plants (ERI), plants forming arbuscular mycorrhizas or non-mycorrhizal plants (AM/NON) and plants potentially living in symbiosis with N2 fixing microbes (Legumes, FIX). There are indications that these four groups may be distinguished by their N isotope signature (Shearer & Kohl, 1986; Gebauer & Dietrich, 1993; Schulze et al., 1994; Michelsen et al., 1998). Major differences in the C isotope signature were not expected since all species in this study had C3 photosynthesis (Ziegler, 1995). Additional effects of water availability, light climate and soil CO2 respiration on the leaf C isotope signature were minimized by sampling leaves in close spatial proximity. The investigated species belong to a broad spectrum of plant life forms and include deciduous and evergreen species (Table 1). Most species were characteristic of nitrogen-poor sites (‘N figures’ of 1–3: Ellenberg et al., 1991).

Analytical methods

The leaf material was carefully cleaned in de-ionized water. Leaves and soil samples were dried at 105°C, ground in a ball mill (Retsch Schwingmühle MM2, Haan, Germany) and stored in a desiccator before further analysis. Relative N isotope abundances and total N concentrations of the plant and soil samples were measured with an elemental analyzer (Heraeus CHN-O Rapid, Hanau, Germany) equipped for Dumas combustion of the samples and linked to a Finnigan MAT Trapping Box (Bremen, Germany) (for automatic cryo-purification of the gaseous combustion products) and a Finnigan MAT delta D gas-isotope ratio mass spectrometer with a dual inlet system (Gebauer & Schulze, 1991). Relative C isotope abundances were later measured from identical material with the following on-line equipment: elemental analyzer (Carlo Erba 1108, Milano, Italy) for Dumas combustion and GC separation of the gaseous combustion products and linked to a Finnigan MAT delta S gas-isotope ratio mass spectrometer via a ConFlo II open–split interface (Finnigan MAT). In some cases all plant material was used for N isotope analysis and therefore data for C isotope abundances are missing (Table 1).

Relative isotope abundances are denoted as δ-values, which were calculated according to the following equation:

δ15N or δ13C = (Rsample/Rstandard − 1) · 1000 [‰](Eqn 1)

where Rsample and Rstandard are the ratios of heavy isotope to light isotope of the samples and the respective standard. Standard gases (N2 or CO2, respectively) were calibrated with respect to the international standards (N2 in air or PDB, respectively) by use of the reference substances N1 and N2 for the N isotopes and NBS 19 and ANU sucrose for the C isotopes. All reference substrates were provided by the International Atomic Energy Agency, Vienna. Total N concentrations of the samples were calculated from partial pressure measurements of the trapped N2 gas and are expressed as mmol eq N inline image(Gebauer & Schulze, 1991).

Calculations and statistical methods

All data sets (δ15N, δ13C and total N concentrations in leaves and soil samples from each of the eight sites) were first tested separately for normal distribution and homogeneity of variances. More than 75% of the data sets fulfilled these requirements. Accordingly, a one-way anova was used to evaluate differences in δ15N, δ13C or total N concentrations, respectively, between groups (ECM, ERI, AM/NON, FIX, AO and MHO plants and soil) separately for each of the eight sites. When effects of groups on the dependent variables were significant at the 0.05 level, the least-significant difference test (LSD0.05) was used to compare means. When mean values of groups or species are given, ± refers always to 1 SE.

The relative contribution of N or C derived from fungal material to the N or C content of the autotrophic orchids (% xdF with × as N or C, respectively) was calculated using a linear two-source isotopic mixing model based on individual δ-values of autotrophic orchids (δxAO), mean δ-values of suitable (see Results section) reference plants of each site (δxR) and on the mean relative enrichment of the myco-heterotrophic orchid species

MHO-R = δxMHO − δxR):
image(Eqn 2)

Means ± 1 SE were calculated for the percentage xdF values of each AO species sampled on the respective sites. These data were tested for significant difference from zero (i.e. no N or C gain from fungal material) using Student's t-test.

The two-source isotopic mixing model is based on the assumption that all of the N and C in the MHO is derived from the fungal partner and that reference plants have no access to this N and C source. Furthermore, it is assumed that the fungal partners of the MHO and of the AO are similar in their stable N and C isotope abundance. Neither of these assumptions could be tested for their validity in this study. This limitation has to be kept in mind for the interpretation of the data obtained from the model calculation.


Comparison of δ15N, δ13C and total N concentrations in functional groups

Significant effects of groups (ECM, ERI, AM/NON, FIX, AO and MHO plants and soil) on δ15N and total N concentrations were found at all sites (Table 2). Group effects on δ13C were significant for all of the investigated forest sites and for two of the four grassland sites.

Table 2.  Mean values of δ15N ± 1 SE, δ13C ± 1 SE and total N concentrations ± 1 SE in leaves of plants forming ectomycorrhiza (ECM), ericoid mycorrhiza (ERI), arbuscular mycorrhiza or no mycorrhiza (AM/NON), of plants potentially living in symbiosis with N2 fixing microbes (FIX), of autotrophic orchids (AO) and myco-heterotrophic orchids (MHO), and in soil samples collected on four forest and four grassland sites in Europe
Functional groupnδ15N [‰]δ13C [‰]Total N conc. [mmol N inline image]
  1. Different letters indicate significant differences between the functional groups for each of the respective parameters and sites (LSD0.05). Variation in the respective parameters due to functional groups and sites as tested by one-way anova is indicated by F-values and probabilities (P). Number of samples (n) for the respective functional groups as indicated. n in brackets refers to δ13C. n.a. no data available.

Forest sites
T (Pinus pinaster forest)
ECM20−5.2 ± 0.4an.a.  1.63 ± 0.19a
ERI8−4.8 ± 0.3an.a.  1.02 ± 0.05b
AO4 4.0 ± 0.6bn.a.  2.86 ± 0.03c
MHO4 5.7 ± 0.7bn.a.  1.86 ± 0.09a
soil (Of/Oh)4−6.1 ± 0.5an.a.  0.18 ± 0.03d
F, P 78.1, < 0.0001n.a. 10.9, < 0.0001
V1 (Fagus sylvatica forest)
ECM5−5.7 ± 0.3a−31.9 ± 0.3a  1.77 ± 0.03a
AM/NON20−5.9 ± 0.3a−31.1 ± 0.2a  2.01 ± 0.08a
AO5 3.3 ± 0.6b−24.1 ± 0.2b  2.87 ± 0.01b
MHO5 6.4 ± 1.2c−22.8 ± 0.2c  2.85 ± 0.12b
soil (Of/Oh)5−2.3 ± 0.9d−24.3 ± 0.7b  0.32 ± 0.04c
F, P 76.1, < 0.0001174.5, < 0.0001 56.5, < 0.0001
V2 (Pinus sylvestris forest)
ECM8−5.1 ± 0.4a−28.4 ± 0.3a  1.50 ± 0.13a
AM/NON17−5.3 ± 0.3a−28.5 ± 0.3a  1.66 ± 0.08a
AO11−2.7 ± 1.0b−27.9 ± 0.2a  2.13 ± 0.13b
soil (Of/Oh)5−4.1 ± 0.7ab−25.1 ± 0.5b  1.05 ± 0.16c
F, P  4.0, 0.0149 17.3, < 0.0001 10.6, < 0.0001
V3 (Pinus sylvestris forest)
AM/NON8−5.5 ± 0.5a−29.2 ± 0.5a  1.43 ± 0.15a
AO8 (5) 0.3 ± 2.1b−28.0 ± 1.0a  2.13 ± 0.17b
soil (Of/Oh/A0–5)4 1.7 ± 2.1b−16.0 ± 1.0b  0.39 ± 0.07c
F, P  5.3, 0.0166 76.3, < 0.0001 23.0, < 0.0001
Grassland sites
AM/NON22−0.5 ± 0.4a−26.4 ± 0.3a  1.93 ± 0.09a
FIX7−1.0 ± 0.4a−28.0 ± 0.3b  2.70 ± 0.12b
AO13 4.7 ± 0.6b−27.4 ± 0.4b  2.18 ± 0.11a
soil (0–5 cm)5 4.9 ± 0.2b−26.9 ± 0.1ab  0.44 ± 0.01c
F, P 39.7, < 0.0001  3.9, 0.0142 37.5, < 0.0001
AM/NON20−1.6 ± 0.3a−27.6 ± 0.1  1.71 ± 0.04a
FIX5 0.0 ± 0.1b−27.0 ± 0.2  2.51 ± 0.08b
AO8 1.9 ± 0.3c−26.9 ± 0.5  2.25 ± 0.20b
soil (0–5 cm)5 2.5 ± 0.3c−26.5 ± 0.1  0.42 ± 0.02c
F, P 28.9, < 0.0001  2.7, 0.0633 52.6, < 0.0001
AM/NON20−3.7 ± 0.3a−27.6 ± 0.2a  1.72 ± 0.07a
FIX5−1.0 ± 0.1b−26.2 ± 0.1b  2.41 ± 0.09b
AO5−2.2 ± 0.3b−27.2 ± 0.4a  2.32 ± 0.04b
soil (0–5 cm)5 0.8 ± 0.2c−26.3 ± 0.1ab  0.50 ± 0.07c
F, P 20.7, < 0.0001  7.3, 0.0007 60.8, < 0.0001
AM/NON20−1.7 ± 0.4a−26.6 ± 0.3  2.47 ± 0.06a
FIX5−0.5 ± 0.2a−25.8 ±  0.3  2.88 ± 0.09b
AO5−1.6 ± 0.3a−25.6 ±  0.4  2.35 ± 0.11a
soil (0–5 cm)5 4.1 ± 0.1b−25.9 ±  0.1  0.23 ± 0.01c
F, P 29.0, < 0.0001  1.8, 0. 1771137.4, < 0.0001

On the forest sites mean δ15N values varied systematically between groups by more than 12‰ (Table 2). MHO, with δ15N values ranging from 5.7 ± 0.7 to 6.4 ± 1.2‰, was the functional group most enriched in 15N. In AO, δ15N values ranged from 4.0 ± 0.6 to −2.7 ± 1.0‰. Leaves of nonorchids on the forest sites had δ15N values ranging only from −4.8 ± 0.3 to −5.9 ± 0.3‰ and ECM, ERI and AM/NON plants were not significantly different. However, they had significantly more negative δ15N than both AO and MHO on all the forest sites. Thus, all δ15N values of non-orchids from each of the forest sites were pooled to one δ15NR value per site for the percentage NdF calculation. From the differences in δ15N between MHO and non-orchids, the mean relative enrichment of myco-heterotrophs (ɛMHO-R) was calculated as 11.5‰. Soil samples from the forest sites had δ15N values ranging from 1.7 ± 2.1 to −6.1 ± 0.5‰ and thus were in some cases depleted in 15N in a similar to non-orchid plants and in others enriched. The positive δ15N value of soil samples from the site V3 (1.7 ± 2.1) probably reflects the presence of mineral soil, which is more enriched in 15N than humus (Gebauer & Schulze, 1991).

Soil samples from the four grassland sites had δ15N values ranging from 0.8 ± 0.2 to 4.9 ± 0.2‰ (Table 2) and thus, tended to be more enriched in 15N than soil samples from the forest sites. For leaves of AM/NON plants from the grassland sites, which are assumed to prefer soil mineral N as an N source, δ15N values ranging from –0.5 ± 0.4 to −3.7 ± 0.3‰ were found, and thus were also more enriched in 15N than leaves of non-orchids from the forest sites. Leaves of FIX plants had mean δ15N values ranging from −1.0 ± 0.4 – 0.0 ± 0.1‰, which is close to the natural N isotope abundance of atmospheric N2 (0‰). FIX plants can potentially use both atmospheric N2 and soil mineral N. There was a significant difference in δ15N between FIX and AM/NON plants on only two of the four grassland sites. Nevertheless, δ15N values of FIX plants were not considered for the δ15NR calculation, which employed data of only all AM/NON samples per site. δ15N values in the leaves of the AO group ranged from −2.2 ± 0.3 − 4.7 ± 0.6‰, a broader range than for the other functional groups. On three of the four grassland sites δ15N values of AO leaves were significantly less negative than in AM/NON leaves.

δ13C values of leaf samples on the forest sites showed a site-dependent pattern (Table 2). In the Fagus sylvatica forest (V1) δ13C values of MHO (−22.8 ± 0.2‰) and AO (−24.1 ± 0.2‰) were significantly different from each other and from ECM (−31.9 ± 0.3‰) and AM/NON leaves (−31.1 ± 0.2‰). From these data an ɛMHO-R of 8.4‰ was calculated. In the two Pinus sylvestris forests (V2, V3) mean δ13C values of AO leaves (−27.9 ± 0.2 or −28.0 ± 1.0‰, respectively) were again less negative than mean δ13C values of the non-orchids (−28.4 ± 0.3 to −29.2 ± 0.5‰). However, differences between AO and non-orchids in δ13C were much lower than in the Fagus sylvatica forest and statistically not significant on the functional group level. δ13C values within each plant group were always more negative for leaves from the Fagus sylvatica forest than for leaves from the two Pinus sylvestris forests. In neither the Fagus sylvatica forest nor in the Pinus sylvestris forests were δ13C of ECM and AM/NON leaves significantly different. Because of this, the ECM and AM/NON leaves were pooled to one δ13CR value per site for the percentage CdF calculation. Soil samples from the forest sites had δ13C values ranging from −25.1 ± 0.5 to −16.0 ± 1.0‰ and thus, tended to be less depleted in 13C than most of the leaf samples. The high δ13C values in the soil samples from the site V3 (−16.0 ± 1.0‰) again probably reflect the presence of mineral soil, here derived from dolomite, which contains inorganic carbonate with a δ13C of about −10–0‰ (Ehleringer & Rundel, 1988).

δ13C values on the grassland sites ranged from −28.0 ± 0.3 to −25.6 ± 0.4‰ (Table 2). δ13C between sites and functional groups for the grassland sites varied less than for the forest sites. No consistently significant differences between functional groups were found.

Mean total N concentrations of leaves of AO from the forest sites ranged from 2.13 ± 0.17 − 2.87 ± 0.01 mmol eq N inline image (Table 2). MHO had only slightly lower N concentrations in their rudimentary leaves ranging from 1.86 ± 0.09 − 2.85 ± 0.12 mmol eq N inline image. Mean total N concentrations in orchid leaves were thus always higher than mean N concentrations of non-orchid leaves from the respective sites (range from 1.02 ± 0.05 − 2.01 ± 0.08 mmol N inline image), irrespective of whether they belonged to ECM, ERI or AM/NON. This distinction in total N concentration between leaves of orchids and non-orchids from the forest sites was significant with only one exception. Irrespective of the functional plant group, total N concentrations in leaves were always significantly higher than total N concentrations in soil samples (range from 1.05 ± 0.16 −0.18 ± 0.03 mmol eq N inline image) from the respective forest sites.

Leaves of FIX plants had total N concentrations ranging from 2.41 ± 0.09 − 2.88 ± 0.09 mmol eq N inline image, higher than all other functional groups on the four grassland sites (Table 2). As with the forest sites, leaves of AO in grassland tended to have higher total N concentrations (2.18 ± 0.11 − 2.35 ± 0.11 mmol N inline image) than AM/NON plants (1.71 ± 0.04 − 2.47 ± 0.06 mmol N inline image), although differences were only significant on two sites. Grassland soil samples had consistently lower total N concentrations (0.23 ± 0.11 − 0.50 ± 0.07 mmol eq N inline image) than leaves of all plant groups.

Comparison of δ15N and δ13C on the level of individual species

Leaves of AO had more variable δ15N and to a lesser extent δ13C values than the other functional groups, especially in the forest sites. This variability was due to larger differences in isotope signature among AO species than among other functional groups.

On the forest sites, species of the genera Cephalanthera (C. damasonium, C. rubra) and Epipactis (E. atrorubens, E. helleborine) within the AO group had much higher δ15N and δ13C values than non-orchids, while Listera ovata and Platanthera bifolia did not. Ophrys insectifera fell between these two groups. Cephalanthera damasonium had δ15N and δ13C values closest to the MHO species Limodorum abortivum and Neottia nidus-avis (Figs 1 and 2).

Figure 1.

Mean values of δ15N ± 1 SE in leaves of all individual plant species and in soil samples collected from four forest sites in Europe. (a) Pinus pinaster forest (Thezan, S France) (b) Fagus sylvatica forest V1 (c) Pinus sylvestris V2 (d) Pinus sylvestris forest V3 (all Veldensteiner Forst, NE Bavaria, Germany). Symbols represent classification according to functional groups: ECM (open circles), ERI (open triangles), AM/NON (open squares), AO (filled circles), MHO (filled squares), soil (open rhombs). For the respective numbers of replicates (n) see Table 1. Error bars are missing, if smaller than the symbols or if n < 3.

Figure 2.

Mean values of δ13C ± 1 SE in leaves of all individual plant species and in soil samples collected from three forest sites in the Veldensteiner Forst (NE Bavaria, Germany). (a) Fagus sylvatica forest V1 (b) Pinus sylvestris forest V2 (c) Pinus sylvestris forest V3. Symbols represent classification according to functional groups: ECM (open circles), AM/NON (open squares), AO (filled circles), MHO (filled squares), soil (open rhombs). For the respective numbers of replicates (n) see Table 1. Error bars are missing, if smaller than the symbols or if n < 3.

On the grassland sites species effects within the AO group were less pronounced and mainly limited to the δ15N values (Figs 3 and 4). Two species in the genus Orchis (O. ustulata, O. mascula) were largely responsible for significant differences in δ15N between AO and AM/NON on the grassland sites, while Dactylorhiza sambucina had δ15N values similar to AM/NON plants.

Figure 3.

Mean values of δ15N ± 1 SE in leaves of all individual plant species and in soil samples collected from four grassland sites in the Frankenwald area (NE Bavaria, Germany). (a) site F1 (b) site F2 (c) site F3 (d) site F4. Symbols represent classification according to functional groups: AM/NON (open squares), FIX (open triangles), AO (filled circles), soil (open rhombs). For the respective numbers of replicates (n) see Table 1. Error bars are missing, if smaller than the symbols or if n < 3.

Figure 4.

Mean values of δ13C ± 1 SE in leaves of all individual plant species and in soil samples collected on four grassland sites in the Frankenwald area (NE Bavaria, Germany). (a) site F1 (b) site F2 (c) site F3 (d) site F4. Symbols represent classification according to functional groups: AM/NON (open squares), FIX (open triangles), AO (filled circles), soil (open rhombs). For the respective numbers of replicates (n) see Table 1. Error bars are missing, if smaller than the symbols or if n < 3.

N and C gain of autotrophic orchids from fungal association

% NdF and % CdF data were calculated separately for all AO species collected on the four forest and the four grassland sites in an attempt to quantify the species specific and site dependent effects illustrated by the δ15N and δ13C values. Statistically significant N gain from the fungal association ranging from 37 ± 14 to 80 ± 5% of the leaf total N was estimated for Ophrys insectifera collected at the forest site V3 and for the two Cephalanthera species (Table 3). High N gain from fungi was also indicated for Ophrys insectifera collected at the forest site V2 and for the two Epipactis species, but their significance could not be tested due to insufficient replicates. N gain from fungi seems to be negligible for Platanthera bifolia and Listera ovata. On the grassland sites significant N gain from fungi ranging from 12 ± 3 to 63 ± 4% of leaf total N was found (Table 3). The two Orchis species gained more N from fungi than Listera ovata and Dactylorhiza sambucina. Again, some of the data could not be tested for their significance due to a too low number of replicates.

Table 3.  Percentages of nitrogen (mean percentage NdF ± 1 SE) and carbon (mean percentage CdF ± 1 SE) derived from fungal association in the leaves of 10 autotrophic orchid species as calculated from δ15N or δ13C values, respectively, and based on a linear two-source isotopic mixing model
SpeciesSite% NdF% CdF
  1. The orchids were collected from four forest and four grassland sites in Europe. Site codes are given in the Materials and Methods section. Stars indicate significance levels for deviations from zero (*P < 0.05, **P < 0.01, ***P < 0.001). For the respective numbers of replicates (n) see Table 1. SE is missing, if n < 3. n.a. no data available.

Cephalanthera dam.T 79 ± 5 ***n.a.
Cephalanthera dam.V1 80 ± 5 *** 85 ± 3 ***
Cephalanthera rubraV2 51 ± 1 ***  7 ± 1 **
Ophrys insectiferaV2 54  4
Platanthera bifoliaV2 −2 ± 11 11 ± 7
Listera ovataV2  0 ± 4  6 ± 4
Ophrys insectiferaV3 37 ± 14 *  9
Epipactis helleborineV3100 43
Epipactis atrorubensV3 80 32
Platanthera bifoliaV3  4−24
Listera ovataV3 15n.a.
Listera ovataF1 26 16
Orchis ustulataF1 63 ± 4 ***−10 ± 5
Orchis masculaF1 37 ± 4 ***−16 ± 7
Dactylorhiza samb.F1 22  3
Listera ovataF2 26 ± 6 ** 27 ± 6 **
Orchis masculaF2 33 ± 3 *** −6 ± 3
Dactylorhiza samb.F3 12 ± 3 **  1 ± 4
Dactylorhiza samb.F4  1 ± 2 10 ± 5

Statistically significant C gain from the fungal association was only found for the two Cephalanthera species growing on forest sites and for Listera ovata on the grassland site F2 (Table 3). Cephalanthera damasonium growing in the Fagus sylvatica forest was estimated to obtain 85% of its C from fungi. Epipactis species probably gain also considerable proportions of their leaf C from fungi, but those and some other data sets could not be tested for significance due to insufficient replicates. Some percentage CdF calculations produced negative results, but none of these were significant.


Isotope signature of the non-orchids

The δ15N values of leaves of non-orchids and of soil samples from all eight sites were in the typical range for temperate vegetation and soils, with forest vegetation and soils more depleted in 15N than grassland vegetation and soils (Shearer & Kohl, 1986; Gebauer & Schulze, 1991; Martinelli et al., 1999; Gebauer et al., 2000a). Because all major pathways of N loss (denitrification, ammonia volatilization and nitrate leaching) cause 15N enrichment of the remaining nitrogen, increased N losses on grassland sites are considered to be mainly responsible for this difference (Shearer & Kohl, 1986; Högberg, 1997; Jung et al., 1997; Velthof et al., 2000; Vervaet et al., 2002; Tilsner et al., 2003). δ15N of leaves of non-orchids belonging to the group of ECM, ERI or AM/NON plants showed little variation within sites compared with intersite variation, not significantly different at a site, and were typical of data from temperate systems (Gebauer & Dietrich, 1993). Under more nutrient-limited conditions of boreal regions, ECM, ERI and AM/NON plants have distinct δ15N values (Schulze et al., 1994; Michelsen et al., 1998). The isotopic homogeneity within non-orchid leaves at all sites of this investigation suggests that all non-orchids within one site contained N of similar origin. This does not necessarily mean that all of these plants utilized one specific N compound (e.g. ammonium) as an N source, but that they utilized a mixture of N compounds (e.g. ammonium + nitrate + organic N from the soil + various N compounds from atmospheric deposition, Gebauer, 2000; Gebauer et al., 2000b; Harrison et al., 2000) that were isotopically more similar than the mixture of N compounds available at the other sites. The remaining variability between leaves of individual non-orchid species growing on the same sites represents species-specific characteristics in N source preference or N metabolism and allocation. The isotopic homogeneity of the leaf material on each of the respective sites was furthermore enhanced by the sampling scheme, which attempted to minimize undesired spatial and temporal heterogeneity in N utilization (see Materials and Methods and Gebauer & Dietrich, 1993). The FIX plant group was the only functional group to be significantly distinguished by leaf δ15N values from other non-orchids, at least on two of the four grassland sites. Since FIX plant leaves and leaves of other non-orchids from the grassland sites always had δ15N values close to zero, which is the δ15N value of atmospheric N2, we did not attempt to calculate the nitrogen gain from N2-fixation. Furthermore, leaves of these two groups differed in δ15N by less than 5‰ (Högberg, 1997 and discussion in the last section of this paper). Nevertheless, the significantly and consistently higher leaf total N concentration of the FIX plants indicates nitrogen gain from symbiotic N2-fixation. We conclude from these findings that the leaf δ15N values found for all non-orchids and non-FIX plants on each of the investigated sites represents the site-specific spectrum of leaf δ15N values that occurs when different plant species associated with different types of mycorrhizas (ECM, ERI and AM/NON) are competing at the same time and in close spatial proximity for soil-borne N under temperate European growing conditions.

The δ13C values of C-autotrophic terrestrial plant leaves are determined by the metabolic pathway of photosynthesis, by leaf conductance and by the C isotope ratio of atmospheric CO2 (Farquhar et al., 1982). δ13C values of leaves of all nonorchids of this investigation are typical for C3 plants (Ziegler, 1995). The greater 13C depletion of non-orchid leaves on forest sites relative to grassland sites is probably related to two factors (Gebauer & Schulze, 1991): (first) leaf intercellular CO2 concentrations increase with decreasing light availability (Ehleringer et al., 1986), allowing greater 13C discrimination during CO2 assimilation, and (second) δ13C of CO2 available for assimilation decreases with an increasing proportion of CO2 from soil respiration (Schleser & Jayasekera, 1985), leading to more negative δ13C of CO2 on the forest floor than in the grassland due to lower wind turbulence in the forest. Analysis of intrasite variation of leaf δ13C was more informative than that of intersite variation. We conclude that the leaf δ13C values found for all non-orchids investigated on each of the respective sites represents the site-specific spectrum of leaf δ13C values that occurs when different plant species of different life forms, but following the same type of photosynthesis (C3), are simultaneously and in close spatial proximity utilizing CO2 for photosynthesis.

Isotope signature of the myco-heterotrophic and autotrophic orchids

As indicated by the results, the leaves of MHO from the forest sites are considerably more enriched in 15N (ɛMHO-R= 11.5‰) and 13C (ɛMHO-R= 8.4‰) than leaves of C-autotrophic non-orchids living in close proximity on the same sites, irrespective of whether these non-orchids belong to different plant life forms or are associated with different types of mycorrhiza. δ15N and δ13C values of AO from both forest and grassland sites resulted in a range of signatures from those typical for non-orchids on the same sites to signatures typical for MHO.

For the MHO, the utilization of C and N sources different from the sources utilized by the non-orchids must be responsible for this isotopic distinction. MHO are completely dependent on host fungi for their C nutrition (Leake, 1994; Smith & Read, 1997). Preferential acquisition of N by MHO via their host fungi is also likely, although additional uptake of ammonium or nitrate from the soil through the poorly developed roots of the MHO can not completely be excluded. Thus, there are good reasons to assume that fungally metabolized N and C compounds serve as predominant N and sole C source for the MHO, irrespective of whether these compounds derive from soil uptake or from other plants and of whether they are transported by the fungi to the MHO or are gained through mycelium digestion (see Introduction). Fungi are known to be enriched in 15N and 13C when compared with co-occurring vegetation typically by 5–19‰ in sporophores for 15N (Gebauer & Dietrich, 1993; Högberg, 1997; Michelsen et al., 1998; Gebauer & Taylor, 1999; Kohzu et al., 1999; Hobbie et al., 2001) and by c. 4‰ for 13C (Gleixner et al., 1993; Högberg et al., 1999a; Kohzu et al., 1999; Hobbie et al., 2001). The isotopic differences between the MHO and the non-orchids (ɛMHO-R) found in this investigation are similar (N) or higher (C) than these isotopic differences between fungi and vegetation. Isotopic similarities between heterotrophic plants and their hosts have also been found for the δ13C of holoparasites living on higher plants and the δ13C of their host plants (Ziegler, 1996). It was hypothesised that the N gain via mycorrhizal symbiosis is the reason for the relative depletion of 15N in ECM plants (Högberg et al., 1999b; Hobbie et al., 2001). This paper demonstrates exactly the opposite isotope effect for the N gain of MHO via their mycorrhiza, that is a 15N-enrichment of the orchids due to incorporation of 15N-enriched fungi-specific N compounds. Other findings, such as the similar 15N-enrichment in ECM fungi and in saprotrophic fungi on identical substrate (Gebauer & Taylor, 1999; Bauer et al., 2000), also conflict with the above hypothesis.

The isotope signatures of some AO species, intermediate between non-orchids and MHO, suggest that they can acquire N and C from two different sources, namely N available to the non-orchids and non-FIX plants or C available to the non-orchids, respectively, and N and C available to the MHO. This conclusion is supported by the finding that δ15N and δ13C values closest to those of the MHO were found for AO species of Cephalanthera and Epipactis, both genera that contain myco-heterotrophic species (Leake, 1994; Smith & Read, 1997). Thus, within these two genera an evolutionary succession from autotrophic to myco-heterotrophic nutrition is likely. Furthermore, E. helleborine, which normally produces green leaves, is known to occasionally be achlorophyllous (Leake, 1994 and references therein). C. rubra, also a species normally producing green leaves, is known to be capable of spending several years below-ground before producing above-ground shoots, leaves and flowers (Smith & Read, 1997). These plants must therefore have access to additional nutrient sources. The isotope signatures of the AO species suggest that N gain from the source utilized by the MHO is not limited to Cephalanthera and Epipactis. Species of Ophrys and Orchis apparently also gain N from their fungal partner, while species of Platanthera, Listera and Dactylorhiza preferentially gained their leaf N from the source available to the non-orchids and non-FIX plants. The higher total N concentrations in the leaves of the MHO, as well as of the AO, also indicate – in analogy to the FIX plants – a better access to the nutrient N when compared with the non-orchids on the respective sites. Based on these findings we feel it is necessary to investigate a broader spectrum of orchid species and genera in the future. Furthermore, isotopic behavior found here for the MHO species may be more widespread among myco-heterotrophic plants: Delwiche et al. (1979) reported that two members of the Monotropaceae, which parasitise ECM fungi associated with other hosts, were c. 10‰ more enriched in 15N than other plants on the same site.

Applicability of isotope data to provide quantitative insight into the origin of orchid N and C gain

In ecosystem studies, stable isotopes at a natural abundance level contain both source-sink (tracer) and process information (Peterson & Fry, 1987). To extract solely source-sink information from stable isotope data and to apply a linear two-source isotopic mixing model as used in this investigation, it is essential (first) to minimize interferences between source-sink and process information, (second) to reduce the number of sources for each isotope pair under investigation to two, and (third) to work under conditions where the two sources under investigation are sufficiently distinguished by their isotope signature. Högberg (1997), for example, recommends a minimum of 5‰ difference in δ15N between N2-fixing plants and non-N2-fixing reference plants, when data on δ15N are used to estimate symbiotic N2-fixation. In ecosystem studies it is not easy to fulfil completely all of these criteria.

Handley & Scrimgeour (1997) questioned whether nitrogen stable isotopes at a natural abundance level can be used to obtain source-sink information. This criticism is only valid if one or more of the prerequisites listed above are ignored. Our sampling scheme was designed to fulfil the first and the second of the above-mentioned criteria. By selecting similar-aged C3 plant leaves occurring in spatial proximity under similar light, nutrient and water availability conditions at each site, we attempted to minimize interferences between isotopic source-sink and process information. The simultaneous consideration of a spectrum of reference plants of different life forms and associated mycorrhiza provided means and ranges for the δ-values of the N and C sources utilized by these plants on the respective sites. The alternative source used for the isotopic mixing model was the N and C utilized by the MHO. While it is quite obvious that MHO exclusively utilize a C source different from the atmospheric CO2 usually utilized by autotrophic plants, their use of soil-borne N sources in addition to fungal N cannot completely be excluded. A further critical assumption for the application of the two-source mixing model is that the isotopic behavior of the fungal partners of the AO in both forest and grassland was similar to that of the fungal partners of the MHO, which were found only on forest sites. Furthermore, this investigation proved that the third of the above-mentioned prerequisites for the application of a two-source isotopic mixing model was also fulfilled. The isotopic difference between non-orchids (for δ13C) or nonorchids and non-FIX plants (for δ15N), respectively, and MHO was considerably higher than the threshold of 5‰ recommended by Högberg (1997).

The fungi-derived N and C gain calculated on the basis of a linear two-source isotopic mixing model was especially high for the AO species Cephalanthera damasonium. The data indicate that this species, when growing in a dense beech forest, gained considerably more N and C from heterotrophic nutrition than from its own assimilation. For all other investigated AO species fungi-derived C gain was of lower importance than N gain. The high fungi-derived C gain by C. damasonium may be related to the living conditions under low light availability, which might limit C gain from photosynthesis. C. damasonium is known to be the most shade-adapted species of the orchids investigated in this study (Ellenberg et al., 1991). Further investigations have to test whether fungi-derived C gain by C. damasonium is related to light availability. A comparison of the data obtained for the N gain from fungal association for the AO species from the forest and the grassland sites, indicates that AO species living in forests tend to gain more N from their fungal association than AO species living on grassland. We postulate from our findings that the AO species from the forest sites, which were shown to be successful in N and C gaining from fungal association (especially C. damasonium, but potentially also C. rubra, E. helleborine, E. atrorubens and O. insectifera), are connected both to typical orchid mycorrhizal fungi and to basidiomycetes forming ectomycorrhizas with trees, as already shown for some MHO species (Zelmer & Currah, 1995; Taylor & Bruns, 1997; McKendrick et al., 2000b; McKendrick et al., 2002).


We would like to thank Margit Gebauer and Sandra Balzer for help with the field work on some of the investigated sites, Petra Dietrich and Tobias Mummert for skillful technical assistance in mass spectrometry and Andrew Angeli for help with language improvement of the manuscript. Alastair Fitter contributed valuable editorial remarks that improved the manuscript considerably. Kind permission for the orchid sampling by the respective regional authorities is gratefully acknowledged. The investigation contributes to the EU project FORCAST (EVK2-CT 1999–00035).