• We analysed the ectomycorrhizal (ECM) fungal diversity in a Mediterranean old-growth Quercus ilex forest stand from Corsica (France), where Arbutus unedo was the only other ECM host.
• On a 6400 m2 stand, we investigated whether oak age and host species shaped below-ground ECM diversity. Ectomycorrhizas were collected under Q. ilex individuals of various ages (1 yr seedlings; 3–10 yr saplings; old trees) and A. unedo. They were typed by ITS–RFLP analysis and identified by match to RFLP patterns of fruitbodies, or by sequencing.
• A diversity of 140 taxa was found among 558 ectomycorrhizas, with many rare taxa. Cenococcum geophilum dominated (35% of ECMs), as well as Russulaceae, Cortinariaceae and Thelephoraceae. Fungal species richness was comparable above and below ground, but the two levels exhibited < 20% overlap in fungal species composition.
• Quercus ilex age did not strongly shape ECM diversity. The two ECM hosts, A. unedo and Q. ilex, tended to share few ECM species (< 15% of the ECM diversity). Implications for oak forest dynamics are discussed.
Quercus ilex L. (Holm oak) is a characteristic evergreen oak species in the Mediterranean basin (Quézel, 1985; Scarascia-Mugnozza et al., 2000). Despite heavy anthropic pressure, old-growth forests still exist in the island of Corsica (Quézel & Médail, 2003). In such stands, overstorey oaks coexist with understorey chaparral shrubs such as Arbutus unedo L. (strawberry tree) and Erica arborea L. (tree heath). In general, oak trees do not exceed 200 yr of age (Panaïotis et al., 1997). In old-growth forests, mortality leads to tree falls that create numerous small-scale canopy gaps.
A wide variety of ectomycorrhizal (ECM) fungi are symbionts of many tree species in temperate climatic zones. More than 5000 species from the Ascomycetes and Basidiomycetes form ectomycorrhizas on secondary tree roots (Trappe, 1962; Smith & Read, 1997). The majority of ECM species have large host spectra (Molina et al., 1992). This allows a diffuse interaction, i.e. the sharing of common fungal associates by plant individuals of identical or different species. Ectomycorrhizas are critical for nutrition of both partners, and plant protection against soil parasites and toxic compounds. The mycorrhizal network can also reduce carbon costs of ectomycorrhiza formation for some plants, as the extraradical mycelium is already established and sustained by other plants (Högberg et al., 1999).
The fungi that form ectomycorrhizas with trees also form arbutoid mycorrhizas on the roots of ericaceous plants from the Arbutoidea suborder (e.g. Arctostaphylos and Arbutus spp.; Molina & Trappe, 1982). In addition to the fungal sheath and hyphal intercellular growth (Hartig net) that are typical of ECM, hyphae penetrate the cell wall and produce intracellular coils in living cells (Smith & Read, 1997). These fungi may mediate interactions between arbutoid plants and ECM trees. For instance, in Californian chaparral Arctostaphylos glandulosa Eastw. may allow the establishment of Pseudotsuga menziesii (Mirb.) Franco seedlings (Horton et al., 1999) by acting as a symbiont reservoir that may contribute to successional transition to forest stages. In the Mediterranean basin Q. ilex naturally establishes in A. unedo-dominated chaparral (Gamisans, 1999). However, sharing of fungal symbionts between Q. ilex and A. unedo has hitherto not been explored. To our knowledge, studies of A. unedo symbionts have mainly focused on mycorrhizal ultrastructure (Fusconi & Bonfante-Fasolo, 1984; Giovannetti & Lioi, 1990; Münzenberger et al., 1992).
Studies of ECM communities are based either on identification of mycorrhizas (the so-called below-ground view), or on monitoring of fruitbody production (above-ground view). Identification of mycorrhizas can be conducted according to root-tip morphotype or using molecular tools, such as restriction fragment length polymorphism (RFLP) or sequencing of the internal transcribed spacer (ITS) region, an efficient way to dissect ECM communities (Gardes & Bruns, 1996; Horton & Bruns, 2001; Tedersoo et al., 2003). Fruitbody surveys reveal the presence of ECM taxa in a fast and inexpensive way (Vogt et al., 1992; Richard et al., 2004). However these studies, mainly carried out on fleshy macromycetes, often underestimate the presence of numerous resupinate taxa (e.g. Thelephoraceae or Sebacinaceae), hypogeous fungi, and taxa lacking an apparent sexual stage (e.g. Cenococcum geophilum Fr.) (Horton & Bruns, 2001).
Little is known about the below-ground community of ECM fungi in broadleaved forests. For instance, most descriptions of ECM communities in Q. ilex forests have been based on fruitbody surveys (Signorello, 1996; Laganàet al., 1999; Richard et al., 2004). The problems with the use of fruitbody sampling are obvious to anyone who has collected fungi for many years. First, fruiting may vary tremendously from year to year. Second, sampling must be intensive because fruit bodies of many species are short-lived. Thus, in addition to the analysis of fruitbody patterns, there is a need to explore the ECM community in the soil from either ectomycorrhizas or mycelia. A study conducted recently by De Román & De Miguel (2002) has revealed the presence of numerous species of Thelephoraceae in managed Q. ilex stands. However, further research using molecular tools is necessary to document the below-ground diversity in Q. ilex forests.
In a previous study (Richard et al., 2004), we analysed the temporal and spatial patterns of fruitbody production in an old Holm oak forest in Corsica during three consecutive fruiting seasons. Fleshy epigeous macromycetes were surveyed in a permanent plot (160 × 40 m) from September 1999 to March 2002. Here we sampled ectomycorrhizas from Q. ilex and arbutoid mycorrhizas from A. unedo shrubs at the same research site in March 2001. On Q. ilex we collected ectomycorrhizas from seedlings, young saplings and old trees. Our objectives were to: (i) document the below-ground ECM richness in an old-growth Mediterranean forest; (ii) investigate two factors potentially shaping this richness, i.e. host age and host species; and (iii) relate the structure of the ECM community, as determined by mycorrhizas, to that obtained from fruitbody surveys. To identify the fungal symbionts on roots, we compared RFLP types from mycorrhizas to those from fruitbodies of known species collected from the same site. Dominant fungal associates of A. unedo were also sequenced to investigate in more detail the composition of the below-ground community. We relied on this typically Mediterranean plant species to ascertain the relative importance of fungal groups not sampled during our fruitbody survey, such as resupinate or hypogeous fungi.
Materials and Methods
The research transect was the same as that studied in our previous paper (Richard et al., 2004). It was located in the Fango valley (42°20′ N; 8°49′ E) in Corsica, on slightly acidic soils. The vegetation at the research site is an old Q. ilex forest that consists mainly of old Q. ilex trees (≈ 460 stems ha−1), 2- to 10-yr-old saplings, and 1-yr-old seedlings (Panaïotis et al., 1997). A dense 7 m high oligospecific chaparral develops under oak canopies, made up of Phillyrea latifolia L., Erica arborea and A. unedo (Panaïotis et al., 1997). Two small individuals of Cistus monspelliensis L. were also present at this site. A detailed description of the research site is provided by Richard et al. (2004).
Sampling of mycorrhizas
To investigate the structure of ECM diversity, we sampled four plant categories: old A. unedo shrubs (A.un.) and three Q. ilex categories representing the age sequence: 1-yr-old seedlings (Q.il.1); 2- to 10-yr-old saplings (Q.il.2); and 170-yr-old senescent trees (Q.il.3) (Table 1). In each category, 30 individuals scattered over the whole study site (from 30 different 100 m2 plots) were sampled (Fig. 1). Plant age was determined either by counting shoot ring number (for Q.il.1 and Q.il.2) or using data from Panaïotis et al. (1997) (for Q.il.3). Host species was ascertained by tracking roots to the shoot (for Q.il.1 and Q.il.2) or using cambium colour and root architecture which differ between the two host species (for A.un. and Q.il.3).
Table 1. Sampling design and colonization by the dominant Cenococcum geophilum for the four host categories investigated
Arbutus unedo Old shrubs
Mean number ± SD, values followed by different letters differ significantly according to anova (P < 0.05).
Seedlings (Q.il.1) and saplings (Q.il.2) were carefully removed from the soil in order to keep the root system intact and avoid fine root disruption. ECM fungal diversity was evaluated in these two plant categories by exhaustive hand-picking of ECM after pulling up. Soil cores were collected for old A. unedo shrubs (A.un.) and Q. ilex senescent trees (Q.il.3): for each individual, four 10 × 10 cm soil samples were taken at the four cardinal points, from the humus organic horizon to a depth of 20 cm. Roots were sieved from the soil cores, carefully washed, and samples from a given plant were pooled. For each individual plant, five root samples were randomly selected, and 30 ECMs from 30 distinct aggregates. They were then hand-picked and stored at −20°C in 700 µl CTAB lysis buffer (2% cetylammoniumbromide, 100 mm Tris–HCl, 20 mm EDTA, 1.4 m NaCl). Picking was performed to maximize the number of morphotypes recovered on each plant, in order to obtain the most complete description of the community.
Cenococcum geophilum mycorrhizas were considered characteristic enough to be identified by morphology (Agerer, 1987–93). However, in order to ascertain this identification, two randomly selected C. geophilum mycorrhizas from five different individuals per plant category were screened by RFLP analysis, and three were sequenced. All other ECM morphotypes were submitted to molecular analysis according to the following sampling design: (i) exhaustive typing for Q.il.1 and Q.il.2 because of the low number of non-C. geophilum ECM recovered; (ii) five mycorrhizas per plant for A.un.; (iii) seven ECM per plant for Q.il.3.
Polymerase chain reaction
DNA was extracted as described by Gardes & Bruns (1993). Extracted DNA was resuspended in 30 µl TE buffer (1 mm Tris–HCl, 0.1 m EDTA pH 8.0) and diluted for PCR with sterile distilled water at a ratio of 1 : 300 (v/v). The internal transcribed spacer (ITS) region of the rDNA was amplified by PCR using the primer pair ITS1–F/ITS4 (White et al., 1990; Gardes & Bruns, 1993) in a PTC 200 DNA thermocycler (MJ Research, Inc., Waltham, MA, USA). A fraction of 12.5 µl of the diluted DNA extract was added to 12.5 µl of the PCR mix (final concentrations: 2 mm MgCl2, 0.2 mm dNTP, 1 mm of each primer, 1× buffer for Taq DNA polymerase, 1 U Taq DNA polymerase) and amplified using the same temperature profile as Gardes & Bruns (1996). Negative controls without DNA were used to detect DNA contaminations of the reagents in every PCR.
Aliquots of 8 µl of each amplified DNA were digested using the endonucleases Cfo1, Hinf1, Mbo1 and HaeIII. PCR products were size-fractionated on 3% agarose gels (38% agarose + 62% NU-sieve agarose (FMC BioProducts, Philadelphia, PA, USA), stained with ethidium bromide, and photographed under UV light. ΦX174 digested by Hinf1 was used to estimate fragment sizes. Gels were scanned using BioCapt 97.03 (Vilber Lourmat, France) and fragment sizes calculated using Bio1D++97.06 (Vilber Lourmat).
Molecular identification of mycorrhizas
To allow identification of mycorrhizas, small pieces of fresh tissue were taken from at least one fruitbody for each of the 166 ECM taxa fruiting on the study site between September 1999 and January 2002 (Richard et al., 2004). For 66% of the RFLP-typed species only one fruitbody was used; for 48 species (29%) two fruitbodies were used. For the eight remaining taxa, which belong to abundantly fruiting taxa poorly investigated from a taxonomic point of view, more than two fruitbodies were used as follows (n = number of fruitbodies tested): Cortinarius elatior Fr. (n = 4) and Cortinarius pseudosalor J. E. Lange (n = 3), Inocybe tigrina R. Heim (n = 4) and Inocybe flocculosa (Berk.) Sacc. (n = 3), Leccinum lepidum (Bouchet ex Essette) Quadr. (n = 4), Russula decipiens (Singer) Svrcek (n = 4), Russula fragilis (Pers. Fr.) Fr. (n = 3) and Russula persicina var. rubrata Romagn. (n = 4). DNA extraction, PCR amplification and RFLP digestion were carried out as for mycorrhizas. RFLP patterns from mycorrhizas and from fruitbodies were matched, and ECMs whose patterns did not correspond to any fruitbody were considered as unidentified taxa.
The most abundant taxa were sequenced to specify their taxonomic position: sequencing was restricted to A. unedo, a typically Mediterranean species. Based on RFLP results, A. unedo symbionts that were represented by at least two mycorrhizas in our sampling were sequenced as described by Selosse et al. (2002), and identified according to BLAST analysis at the NCBI page http://www.ncbi.nlm.nih.gov/blast/Blast.cgi, using default settings. Sequences were deposited in GenBank (Table 2).
Table 2. Identification by ITS sequence of the most abundant RFLP types on Arbutus unedo
For each plant category, and for the two plant species, the richness of ECM communities was estimated using various species diversity estimators based on abundance and frequency of taxa. Abundance was defined as the cumulative number of mycorrhizas of a given taxon divided by the total number of mycorrhizas, for a given plant category. Frequency was the number of plant individuals on which a given taxon was found divided by the total number of plant individuals, for a given plant category. Species diversity was estimated using: (i) richness, i.e. the total number of taxa, S; (ii) Simpson's diversity index, D; (iii) the Shannon–Wiener information index, H′; and (iv) Fisher's alpha (Fisher et al., 1943). The rarefaction method (Krebs, 1999) was used to compare one-to-one taxonomic richness of samples of various sizes. This method corrects for differences in sampling size by virtually reducing the size of all samples to that of the smallest one. Calculations were performed with the software biodiversity pro 2 (http://www.sams.ac.uk/activities/downloads/software/bdpro.zip) and rarefaction curves obtained were compared graphically for the minimal sample size of the various data sets. Rarefaction analysis was used for comparison between the various plant categories (with minimal sample size n = 121 ECM tips) and for comparison between above- and below-ground diversity (with minimal sample size n = 521 ECM tips or fruitbodies). A visual comparison of the distributions of relative species abundance and frequency between above- and below-ground views of the ECM community was performed using rank–abundance curves.
Differences in abundance of C. geophilum (the only species that was abundant and frequent enough to perform parametric tests), according to plant category, were tested by one-way anova with plant category as single factor using minitab 12.2 software (Minitab Inc., Paris, France).
Relatedness in the composition of ECM taxa among plant species and categories was compared using the Jaccard similarity index (Mueller-Dombois & Ellenberg, 1974), J =c/(a + b − c) × 100, where a is the number of taxa found on the first plant category, b is the number of taxa found on the other, and c is the number of taxa shared by the two plant categories. A percentage similarity (PS) based on RFLP-type abundance was calculated in order to take distribution of taxa into account (Pielou, 1984): PS = c′/(a′ + b′) × 100, where a′ is the number of mycorrhizas formed by taxa found on the first plant category, b′ is the number of mycorrhizas formed by taxa found on the other, and c′ is the number of mycorrhizas formed by taxa colonizing the two plant categories.
General description of the below-ground ECM community
In all, 2294 mycorrhizas were sampled from 120 plants (Table 1), among which a Cenococcum morphotype largely dominated. DNA extraction and RFLP typing were carried out on 558 ECMs, including 40 ascribed to this Cenococcum morphotype (Table 1). The 40 Cenococcum ECMs showed the same RFLP pattern, and three of them showed a unique C. geophilum ITS sequence (Table 2). The Cenococcum morphotype was therefore considered as homogeneous, accounting for 35.3% of the total ECM number and present on 97.5% of all investigated plant individuals (Table 1).
In all, including C. geophilum, ITS–RFLP analysis was conducted on 393 ectomycorrhizas (data not shown). The typing produced 140 different RFLP patterns (Table S1, available online as supplementary material), and resulted in high values of species richness estimators (Table 3). Identification of the mycorrhizal symbionts to species or species group was performed using ITS–RFLP matches with fruitbodies and direct sequencing of the ITS for the most common symbionts of A. unedo (Tables 2 and S1). No intraspecific polymorphism was observed whenever more than one fruitbody per species was examined, except for R. fragilis and R. persicina var. rubrata which produced two different patterns (data not shown). Out of 140 RFLP types, 60 (42.9%) were identified at the genus level (Table S1).
Table 3. Ectomycorrhizal fungal diversity on the different host plant categories
Taxa are RFLP types for below-ground diversity and species based on a morphological concept for above-ground diversity. Values in parentheses indicate number of taxa for which RFLP patterns from mycorrhizas and fruitbodies from the same study site successfully matched.
Rarefaction analysis conducted on Q.il.1, Q.il2, Q.il3 and A. un. (minimal sample size, n = 121 root tips).
Of the 140 RFLP types, 70 were represented by a single mycorrhiza (Fig. 2a), so that the community harbored a large number of rare taxa. Cenococcum geophilum and R. decipiens were frequent species, found on 117 and 17 plants, respectively (Fig. 2b). They also contributed to the overall number of mycorrhizas (839 and 28 for C. geophilum and R. decipiens, respectively; Table 3 and see Fig. 6a). Apart from these two species, only three taxa were represented by 10 or more mycorrhizas: Russula acrifolia Romagn. (16 mycorrhizas from four plants); a thelephoroid species (13 mycorrhizas from seven plants); and a sebacinoid species (11 mycorrhizas from three plants; data not shown).
Distribution of the ECM community among the two host species
In all, 46 RFLP types were found on A. unedo and 112 on Q. ilex (Table 3). In a rarefaction analysis performed to account for differences in sample size (Table 3), the diversity for A. unedo was intermediate between that of senescent oaks and those of seedlings and saplings. Similarly, Shannon entropy, Simpson diversity index and Fisher's alpha were higher for A. unedo than for young Q. ilex (Q.il.1 and Q.il.2), but lower than for old Q. ilex trees (Q.il.3, Table 3), suggesting that Q.il.3 harbored more rare species than the three others. Only 18 of the 140 RFLP types occurred both on A. unedo and Q. ilex roots (Fig. 3). These two-host taxa represented 12.9% of the taxonomic diversity, but colonized 69.4% of all mycorrhizas sampled (Fig. 3) because of the abundance of C. geophilum on both hosts. Relative taxa frequencies had similar distributions on the two hosts, with 50.9 and 60.9% of the RFLP types found only once on Q. ilex and A. unedo roots, respectively (data not shown). At the other extreme, only three RFLP types were found on at least four plant individuals of the same species (data not shown). Because of the low number of mycorrhizas representing each RFLP type (probably caused by the size of our sample), differences in abundance between hosts could not be tested statistically.
On Q. ilex roots, 48 RFLP types (42.9%) were identified to genus, species or family level, based on RFLP matches with fruitbodies (Table 4) or sequence analysis of taxa that were shared with A. unedo (Table 2). The genus Russula was the most represented (Table 4), accounting for 22–27.1% of the identified ECM and between 12 and 16.7% of the corresponding taxonomic diversity, depending on plant age. At the species level, apart from C. geophilum, R. decipiens (a species linked to the genus Quercus) was the most abundant (10.5% of the total number of Q. ilex mycorrhizas) and most frequent species (present on 28.3% of sampled Q. ilex, data not shown).
Table 4. Abundance of main fungal groups (excluding Cenococcum geophilum) on three Quercus ilex plant categories based on matching with fruitbody RFLP patterns
Values are either percentage of total number of typed mycorrhizas (number of mycorrhizas in parentheses), or percentage of total number of RFLP types (number of RFLP types in parentheses).
Tricholoma, Laccaria, Hygrophorus, Lactarius, Inocybe, Hebeloma and Clavulinaceae.
On A. unedo roots the molecular analysis allowed identification of 28 taxa, accounting for 81.4% of the typed mycorrhizas and 60.9% of the species richness (data not shown). Thelephoraceae was the most represented, accounting for 25% of taxonomic diversity and 35.7% of the identified ECMs (Fig. 4). To a lesser extent, the genus Inocybe (21.4% of the ECMs) and the Sebacinaceae family (18.6% of the ECMs) were well represented (Fig. 4). The genera Russula, Cortinarius and Laccaria, as well as the Clavulinaceae family, were also present (Table 2). At the species level, the ECM community on A. unedo was strongly dominated by C. geophilum, present on all 30 sampled plants (Table 1). In addition to C. geophilum, only three species, two Thelephoraceae species and I. tigrina, were present on at least three plant individuals (data not shown).
ECM community and Quercus ilex age
The three Q. ilex categories were dominated by C. geophilum ECM (Table 1). In addition to C. geophilum, variable numbers of RFLP types were found (Table 3), and rarefaction analysis suggested that the ECM richness was higher under senescent trees than for Q.il.1 and Q.il.2 (Table 3). For each age, R. decipiens was among the most frequently recorded, and the only species found on more than three plant individuals (data not shown). Most RFLP types (90 out of 112) were found on one Q. ilex age only, while 19.6% were shared by at least two categories (Fig. 5). Only C. geophilum, R. decipiens, and two unidentified species were found on the three plant categories (Fig. 5). Cenococcum geophilum was significantly more abundant on Q.il.1 and Q.il.2 than on Q.il.3 (Table 1). Based on taxonomic diversity, Q. ilex of various ages shared few fungal partners (between 6.9 and 14.8%; Table 5). However, based on RFLP type abundances, similarities ranged from 63.9 to 81.7% (or at least 17.9% excluding C. geophilum), with successive age categories being more similar than the distal ones, Q.il.1 and Q.il.3 (Table 5).
Table 5. Comparison of ectomycorrhizal fungal communities between plant species and plant categories based on presence/absence of taxa, using a Jaccard similarity index (J) or quantitative comparisons (abundance of taxa, considering the dominant Cenococcum geophilum or not), using a percentage of similarity (PS)
In all, 260 ECM taxa were revealed on the studied transect by combining the present ECM typing and the 166 species from fruitbody identifications by Richard et al. (2004) (Table 3). Rarefaction analysis (data not shown) suggested that, for the minimal samples size n = 521, below-ground richness was higher (127.3 taxa) than above-ground diversity (92.4 taxa). Only 17.7% (46 species) of these taxa were found in both analyses, while 36.2% (94 RFLP types) were found only below-ground (Table 3). In addition, nonfruiting species do exist on this site, as exemplified by C. geophilum. Reciprocally, 46.1% of the total taxonomic diversity (120 species) that fruited was not found on roots (Table 3).
Dominance–diversity curves showed very similar distributions of above- and below-ground diversities. The ECM community was strongly dominated by rare species (Fig. 2a): more than 60% of the taxa produced less than one fruitbody per 1000 m2, while 50% of the taxa were represented by only one mycorrhiza. Similarly, both distributions showed a few very abundant species (Fig. 2a): three dominant species (Laccaria laccata, I. tigrina and Lactarius chrysorrheus) produced 32.9% of all fruitbodies (Richard et al., 2004), while C. geophilum accounted for more than a third of all mycorrhizas (Table 1). An ordination based on either relative abundance (Fig. 6a) or frequency (Fig. 6b) of RFLP types was used to visualize correspondences between above- and below-ground views. Of the 22 most abundant ECM RFLP types (including C. geophilum), only eight produced epigeous conspicuous fruitbodies, including the three most productive ones, L. laccata, I. tigrina and L. chrysorrheus (Fig. 6a). Using mycorrhizas and fruitbody frequencies, a similar relative discrepancy was observed, with only nine taxa found above ground (including the four most frequent) belonging to the 22 most represented ECM RFLP types (Fig. 6b). A single fruiting species, R. decipiens, was present among the seven most frequent RFLP types found in this study (Fig. 6b). This taxon was not abundant (ranking 29th) and relatively infrequent (ranking 28th) above ground (data not shown). Symmetrically, some very infrequent species producing few fruitbodies were more represented below ground, such as Cortinarius subsgp. Phlegmacium-3 and Inocybe subsgp. Inocybium-3. They produced, respectively, one and three fruitbodies over 3 yr (Richard et al., 2004), but their ECM occurred on three different plant individuals (Fig. 6b).
The present study confirms the remarkable species richness of the fungal community measured by fruitbody surveys of epigeous macromycetes at the same site (Richard et al., 2004). High values of species diversity estimators were obtained, as illustrated by Fisher's alpha (Table 3), an estimator linking the number of taxa to the number of individuals sampled, which is not unduly affected by sample size (Tokeshi, 1993). Assuming that each RFLP type corresponds more or less to one species (see below), a total of 140 species were detected based on analysis of mycorrhizas (one RFLP type per 2.5 tips investigated, excluding C. geophilum), compared with 166 species that have been found using fruitbody surveys (Richard et al., 2004). Based on these two approaches together, there were at least 260 ECM fungal species at the site between 1999 and 2002. This is more than in most previously described late successional stands covering similar areas (Jonsson et al., 1999; Bidartondo et al., 2000, 2001), although similar values were found in old temperate coniferous forests, either by ECM typing (Dahlberg et al., 1997; Luoma et al., 1997) or fruitbody surveys (Villeneuve et al., 1989; O'Dell et al., 1999; Smith et al., 2002).
Our results suggest that ITS–RFLP data are robust for characterizing community diversity, for two reasons. First, from 158 morphologically defined species that were used in the DNA analysis, 144 (91% of total) yielded a single species-specific RFLP type (Table S1 and data not shown). Second, intraspecific variation was a minor problem. Of the 58 species represented by at least two fruitbodies, 56 (96%) yielded a single RFLP type for all fruitbodies with the exceptions of R. fragilis and R. persicina var. rubrata (Table S1 and data not shown). Together our results highlight the fact that ITS–RFLP data are a valuable tool for grouping ECM species, and for identification of the mycorrhizal symbionts with the fruitbody RFLP-matching approach. These results are similar to those reported by Horton (2002), who also investigated the use of ITS–RFLP patterns to assess diversity of ECM fungi collected across a 7 km coniferous forest. In addition, Kårén et al. (1997) already reported that intraspecific variation was not a problem on a local scale. In the two polymorphic species of Russula, the RFLP variation found in the ITS is the result of variation in two of the four endonucleases (data not shown). Currently, we do not know if the variation observed is a reflection of cryptic species.
The below-ground method revealed the same distribution pattern as the above-ground survey with respect to the relative proportion of abundant vs rare taxa. The below-ground community was characterized by a few common types and a large number of rare types (Fig. 2). This pattern was also observed using fruitbodies (Fig. 2; Richard et al., 2004). Below ground, 50% of the RFLP types collected were represented by one mycorrhiza. The two dominant species were C. geophilum (Table 1) and R. decipiens (Fig. 6a). Cenococcum alone contributed to 35% of the ectomycorrhizas. Of the 120 total plants, 117 (98%) were colonized by this fungus (Table 1). Russula decipiens was found on 17 oaks (data not shown). However, several questions remain concerning the below-ground diversity because of the large number of rare types observed at our site. Which proportion of the local community is really sampled? Would comparable patterns be obtained at another time? Are all abundant species included? Our ability to detect community similarity (e.g. Arbutus vs Quercus) based on species abundance is also limited by the inherent distribution of the diversity.
Fungal species richness was comparable above and below ground, but the two levels exhibited little overlap (< 20%; Table 3) in fungal species composition. This result confirms and extends earlier observations on the complementarity of the two levels in obtaining a comprehensive view of community composition (Gardes & Bruns, 1996; Jonsson et al., 2000; Peter et al., 2001). For instance, without the below-ground approach we would have missed C. geophilum, an ascomycetes species that does not produce fruitbodies at all. It was particularly abundant on oak seedlings and saplings (Table 1). Its high dominance and frequency at our site may arise in part from its ability to sustain xeric conditions by formation of sclerotia (Lilleskov et al., 2004). The role of this fungus in ecosystem functioning is also intriguing – could it provide drought protection to plant roots, as suggested by Jany et al. (2003)? Or, alternatively, could C. geophilum be purely opportunistic, with little relevance to tree physiology? The observation that C. geophilum often dominates in ECM communities, for example in Spanish Q. ilex forests (De Román & De Miguel, 2002); in the Californian chaparral (Borchers & Perry, 1990); or in temperate Fagus sylvatica forests (Blaise & Garbaye, 1983), leads us to question the existence of ecotypes or cryptic biological species (Shinohara et al., 1999).
Combining the species composition viewed above and below ground, the following patterns were observed. Apart from Cenococcum, the community appeared to be dominated by members of the genus Russula and, to a lesser extent, by the genus Inocybe as well as members of the Thelephoraceae and Sebacinaceae (Fig. 4; Richard et al., 2004). In Spanish managed Q. ilex forests, thelephoroid morphotypes accounted for a quarter of the root tips investigated by De Román & De Miguel (2002). Russulaceae and Thelephoraceae also dominated the community in two other Californian Mediterranean ecosystems, the chaparral (Horton et al., 1999) and the bishop pine forest (Gardes & Bruns, 1996), whereas Sebacinaceae were among the most frequently encountered taxa in Eucalyptus sclerophyllous forests in Australia (Glen et al., 2002). An intriguing feature is the absence of hypogeous fungi (at least among dominant taxa on A. unedo, Table 2 and S1), which is perhaps caused by environmental conditions. For instance, the lack of species of Tuber may be explained by acidic soil conditions.
Tree diversity has been suggested to favour ECM diversity on a local scale (Nantel & Neumann, 1992; Kernaghan et al., 2003). We tested the hypothesis that the hosts contribute to ECM fungal diversity. Only 12.9% of the taxa were shared (Fig. 3), less than what was found in mixed forest stands by Horton & Bruns (1998), Cullings et al. (2000) and Kennedy et al. (2003), where multihost fungi dominated, accounting for 30 to 90% of the ECM fungal community in all three studies. Unfortunately our sampling is insufficient to provide statistically significant data, because of the high species richness of the community. Most species were too infrequent to draw conclusions about their distribution, a problem that is often limiting in studies of ECM communities (Horton & Bruns, 2001; Taylor, 2002). Nevertheless, even if restricted to a limited number of fungal taxa, sharing of symbionts may have ecological consequences as Q. ilex seedlings successfully establish and survive in A. unedo-dominated chaparral (Gamisans, 1999). This pattern suggests that A. unedo shrubs may provide conducive conditions for Q. ilex seedlings in early stages of forest succession, perhaps by providing a compatible fungal network.
Despite the important width of the age sequence, the ECM community was quite similar at the various developmental stages of Q. ilex investigated. We observed: (i) similar rank–abundance curves (reflecting high taxonomic diversity and a dominance of rare taxa, Table 3); (ii) among the identified taxa, similar dominance of genera such as Russula, Cortinarius and Amanita (Table 4); and (iii) a high abundance of C. geophilum (Table 1). Our findings support the conclusion that established seedlings recruit ECM symbionts in an opportunistic way among mycobionts colonizing the old surrounding trees. Similar observations were made in multi-aged stands dominated by conifers such as Pinus sylvestris (Jonsson et al., 1999) or Tsuga heterophylla (Kranabetter, 1999; Kranabetter & Friesen, 2002).
An intriguing question is whether or not the sharing of ECM partners between seedlings and old trees is under natural selection. Seedlings may take benefit from established ECM fungi that already have large extraradical soil-exploring mycelia built at the oldest trees’ expense (Högberg et al., 1999). In addition, shared symbionts may even transfer carbon from high-canopy trees to understorey seedlings (Simard et al., 1997; Lerat et al., 2002), counterbalancing low light influx. For the related species Quercus rubra, seedling nutrition and mycorrhization (infection level and diversity) are improved in the vicinity of adult conspecifics (Dickie et al., 2002). Symbiont sharing between seedlings and older Q. ilex could thus result in favouring of kin, as most Q. ilex acorns remain around the mother tree due to barochory (Darley-Hill & Carter Johnson, 1981).
This first report on the below-ground ECM diversity in a Mediterranean hardwood old-growth forest revealed a striking diversity of ECM fungi. The ascomycete C. geophilum and members of the Russulaceae, Cortinariaceae (genus Inocybe), Thelephoraceae and Sebacinaceae are the most abundant taxa on roots. Exhaustive inventory of the diversity is a real challenge because of the large number of rare types. Our results suggest that both above- and below-ground levels have to be explored to obtain a comprehensive overview of the composition of the ECM fungal community. The composition and diversity of the ECM community does not depend on host age. Our results also question the ecological importance of symbiont sharing between Q. ilex and A. unedo in old-growth forest dominated by Q. ilex.
This research represents a portion of the PhD dissertation submitted to the University of Toulouse III by F. Richard. We acknowledge A. Tillier and C. Bonillo (Muséum National d’Histoire Naturelle, MNHN) for help in sequencing. We thank P. Kennedy, A. Izzo and G. Chust for statistical advice and helpful suggestions with the rarefaction methods. Financial support was provided by the French Ministry of Ecology (DIREN Corse), the Territorial Collectivity of Corsica (CTC) and the National Forest Service (ONF) to F.R.; the Centre National de la Recherche Scientifique and the Ministère de la Recherche to M.G.; and the MNHN (Service de Systématique Moléculaire) to M.-A.S. We also thank to the Regional Natural Park of Corsica (PNRC) for its participation in this project and logistical assistance.
The following material is available as supplementary material at http://www.blackwellpublishing.com/products/journals/suppmat/NPH/NPH1382/NPH1382sm.htm. These supplementary data include (i) for each ITS-RFLP type, DNA fragment sizes in base pairs and (ii) for the 60 taxonomically identified taxa, genus- or species-level identification obtained by comparing ITS-RFLP types to those from voucher specimens of sporocarps or by BLAST analysis of the ITS sequences.
Table S1 ITS-RFLP types of ectomycorrhizas collected on the 120 sampled plants.