Multi-host ectomycorrhizal fungi are predominant in a Guinean tropical rainforest and shared between canopy trees and seedlings


  • Abdala Gamby Diédhiou,

    Corresponding author
    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    2. Laboratoire Commun de Microbiologie, IRD/UCAD/ISRA, BP 1386 Dakar, Sénégal.
      E-mail; Tel. (+221) 33 849 33 32; Fax (+221) 33 849 33 02.
    Search for more papers by this author
    • Present address: Laboratoire Commun de Microbiologie, IRD/UCAD/ISRA, BP 1386 Bel-Air, Dakar, Senegal.

  • Marc-André Selosse,

    1. Centre d'Ecologie Fonctionnelle et Evolutive (CNRS, UMR 5175), Equipe Interactions Biotiques, 1919 Route de Mende, 34293 Montpellier Cedex 5, France.
    Search for more papers by this author
  • Antoine Galiana,

    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    Search for more papers by this author
  • Moussa Diabaté,

    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    2. Institut de Recherche Agronomique de Guinée, Division des Cultures Pérennes, Programme Recherche Forestière, BP 1523, Conakry, République de Guinée.
    Search for more papers by this author
  • Bernard Dreyfus,

    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    Search for more papers by this author
  • Amadou Moustapha Bâ,

    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    2. Laboratoire de Biologie et Physiologie Végétales, Faculté des Sciences Exactes et Naturelles, Université des Antilles et de la Guyane, BP 592, 97159 Pointe-à-Pitre, Guadeloupe, France.
    Search for more papers by this author
  • Sergio Miana De Faria,

    1. Embrapa Agrobiologia km 47, antiga estrada Rio-São Paulo, Seropédica 23851-970 Brazil.
    Search for more papers by this author
  • Gilles Béna

    1. Laboratoire des Symbioses Tropicales et Méditerranéennes, UMR113 – INRA/AGRO-M/CIRAD/IRD/UM2 – TA10/J, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France.
    Search for more papers by this author

E-mail; Tel. (+221) 33 849 33 32; Fax (+221) 33 849 33 02.


The diversity of ectomycorrhizal (ECM) fungi on adult trees and seedlings of five species, Anthonotha fragrans, Anthonotha macrophylla, Cryptosepalum tetraphyllum, Paramacrolobium coeruleum and Uapaca esculenta, was determined in a tropical rain forest of Guinea. Ectomycorrhizae were sampled within a surface area of 1600 m2, and fungal taxa were identified by sequencing the rDNA Internal Transcribed Spacer region. Thirty-nine ECM fungal taxa were determined, of which 19 multi-hosts, 9 single-hosts and 11 singletons. The multi-host fungi represented 92% (89% when including the singletons in the analysis) of the total abundance. Except for A. fragrans, the adults of the host species displayed significant differentiation for their fungal communities, but their seedlings harboured a similar fungal community. These findings suggest that there was a potential for the formation of common mycorrhizal networks in close vicinity. However, no significant difference was detected for the δ13C and δ15N values between seedlings and adults of each ECM plant, and no ECM species exhibited signatures of mixotrophy. Our results revealed (i) variation in ECM fungal diversity according to the seedling versus adult development stage of trees and (ii) low host specificity of ECM fungi, and indicated that multi-host fungi are more abundant than single-host fungi in this forest stand.


Ectomycorrhizal (ECM) symbiosis involves soil fungi and tree roots. It provides mineral nutrients, water and protection against pathogens to the plant which, as a reward, provides carbon to its fungal partner (Smith and Read, 2008). Each individual tree associates with several ECM fungal species, while fungal species display variable levels of specificity, ranging from highly specific species to generalists (= multi-host; Selosse et al., 2006; Smith et al., 2009). Multi-host ECM fungi are often abundant, at least in Holarctic and Mediterranean ECM communities (e.g. Kennedy et al., 2003; Dickie et al., 2004; Richard et al., 2005), but sometimes display host preferences as shown in mixed Japanese forests (Ishida et al., 2007), and Tasmanian sclerophyllous forest (Tedersoo et al., 2008). Some fungal taxa are more specific, e.g. suilloids, which are almost entirely restricted to Pinaceae (Bruns et al., 2002), and associates of Alnus (Tedersoo et al., 2009) in Holarctic regions.

Very few studies have addressed the question of specific versus multi-host abilities in tropical forests, where ECM associations remain little studied (Alexander et al., 1992; Alexander and Lee, 2005; Tedersoo et al., 2007a). In tropical Africa, there are currently few reports of ECM host preferences, particularly in mixed forests. In this region, ECM trees include Caesalpinioideae, Phyllanthaceae as well as some Dipterocarpaceae, Proteaceae and Sapotaceae (Högberg and Piearce, 1986; Newbery et al., 1988; Torti and Coley, 1999; Ducousso et al., 2004), but ECM fungal communities and their links withcomposition of the plant community remain little studied. Fruit body collections from mixed African forests have already suggested that numerous ECM fungi are multi-host (Sanon et al., 1997; Rivière et al., 2007). Thoen and Bâ (1989) suspected host effects on ECM fungi collected under Afzelia africana and Uapaca guineensis in southern Senegal. Nevertheless, fruit body survey cannot predict the ECM association patterns at the root level, because below-ground fungal diversity and abundance differ from those observed above-ground (Richard et al., 2005). Inoculation of seedlings demonstrated that some African fungal isolates are compatible with multiple tree species (Diédhiou et al., 2005). Hence, in situ studies of ECM diversity at root level are still required to assess the diversity and specificity (versus host sharing) of the ECM fungal community from tropical Africa.

Some ECM tropical tree species tend to aggregate in patches where they dominate together or alone (Newbery et al., 2004; Alexander and Lee, 2005), surrounded by trees forming arbuscular mycorrhizae (AM; McGuire, 2007). In the Korup National Park (Cameroon), ECM trees form up to 70% of local patches (Newbery et al., 1997); similarly, the ECM Gilbertiodendron dewevrei represents more than 90% of trees in some stands of the Congo basin, with abundant seedlings (Hart et al., 1989). Such mono- or oligo-dominant stands suggest efficient regeneration under adults. It has been claimed that adult trees could provide ECM inocula to seedlings, in the form of already established (and perhaps even already nourished) fungi. This likely favours the survival of seedlings in temperate ecosystems (Nara, 2006) and tropical ones (Onguene and Kuyper, 2002; McGuire, 2007). However, no data are available on fungal sharing between adults and seedlings in the latter ecosystems.

Besides an indirect interaction (with adults paying the cost of the fungus more than the seedlings), facilitation can also occur thanks to carbon and nutrients transferred from adults to seedlings by way of mycorrhizal networks, interconnecting root systems of the same or different species (Simard et al., 1997). Labelling experiments have shown in both laboratory and field conditions that carbon and nutrients can be transferred between plants through mycorrhizal networks (Selosse et al., 2006). However, the role of the hyphal pathway during carbon movement, and the physiological or ecological relevance of inter-plant transfers remain controversial (e.g. Fitter, 2001; Wu et al., 2001). Natural abundance of stable isotopes has given increased support to such exchanges, at least in overstorey receiver species in forest ecosystems (Selosse and Roy, 2009). Such plants receiving carbohydrates from their mycorrhizal fungi exhibit 13C and 15N abundances intermediate between mycorrhizal fungi providing some carbon and autotrophic plants placed in similar conditions (Julou et al., 2005; Tedersoo et al., 2007b) – but this has not yet been used to test for the origin of seedlings' carbon.

In the Southern Guinea rainforests (West Africa), Caesalpinioideae and Phyllanthaceae trees are the most abundant native ECM species, growing in mixed patches, which raises the possibility of some fungal sharing. A high regeneration occurs under adult trees, strongly suggesting whether any ‘nurse effect’ occurs, that is: (i) do coexisting trees and seedlings harbour the same fungal communities? and/or (ii) do ECM fungi provide carbohydrates to seedlings? In this typical primary forest area, we first described the fungal diversity on ECM root tips, to assess ECM association patterns among the five dominant host species and between both development stages, i.e. adults versus seedlings. Second, in order to test whether some carbohydrates were transferred to seedlings, we compared the 13C and 15N contents of the latter with those of adult trees, fruit bodies of ECM fungi and surrounding AM tree species.


Ectomycorrhizal diversity

From 12 adult trees (Fig. 1) and 138 surrounding seedlings, 150 and 190 root samples were, respectively, collected (Table 1). From these samples, 362 ECM tips were obtained, and 293 were successfully amplified and sequenced (81%, Table 1); only 10 ECM tips exhibited two different Internal Transcribed Spacer (ITS) sequences at the same time. The 303 different ITS sequences obtained represented 39 ECM fungal taxa (i.e. diverging sequences) from seven fungal families: Boletaceae, Clavulinaceae, Russulaceae, Sclerodermataceae, Thelephoraceae, Tricholomataceae and three other undetermined families (Table 2).

Figure 1.

Map of the study site, with adults (big squares) and seedlings (circles) of the five ECM tree species, Anthonotha fragans (grey), A. macrophylla (yellow), Cryptosepalum tetraphyllum (red), Paramacrolobium coeruleum (blue), Uapaca esculenta (green). The field stand is divided into 16 plots (A to P). The square shapes with bars represent the adults from which ECM were sampled.

Table 1.  Number of root samples, ECM tips and ITS sequences obtained from adults (a) and seedlings (s) of the five host tree species.
 A. fragransA. macrophyllaC. tetraphyllumP. cœruleumU. esculenta
  • a. 

    Each sequence differing from the other is considered as a different taxon.

No. of sampled individuals124337233330324
No. of root samples30343037304630413032
No. of ECM tips sampled for DNA analysis34363139334932433035
No. of ECM tips successfully sequenced32302127224026382532
No. of ITS sequences obtained32312228224126412832
No. of ECM taxaa810810613715118
No. of ECM taxa (after rarefaction to 22)
Table 2.  List of ECM fungal taxa found on the study stand.
ECM taxaGenBank Accession No.Best blastn full-length ITS matchHost species (frequency)Development stage of host
Specimen / Accession noIdentity (%)
  • The closest blast match represents the most accurate taxonomic match between our ITS sequences and those in the NCBI database.

  • Host species are Anthonotha fragrans (Af), A. macrophylla (Am), Cryptosepalum tetraphyllum (Ct), Paramacrolobium cœruleum (Pc), and Uapaca esculenta (Ue).

  • a. 

    The fungi corresponding to these sequences were considered ectomycorrhizal since blast analysis revealed a lot of sequences amplified from ectomycorrhizas in GenBank (but an unclear taxonomic position); anyway, a status of endophytes or contaminant cannot be rigorously ruled out until a finer taxonomic position is obtained.

Basidiomycota #1aAM113461Uncultured soil fungus / DQ42087793Am (1), Ct (2)Both stages
Basidiomycota #2aAM113462Fungal endophyte / FJ45005082Ue (2)Adults
Basidiomycota #3aAM113463uncultured Basidiomycota / AY96951897Af (1), Am (1), Ct (1), Pc (1), Ue (1)Both stages
Boletaceae #1AM113453Boletus bicolour / GQ16687780Af (3), Am (1), Pc (2), Ue (2)Both stages
Boletaceae #2AM113454Bothia castanella / DQ86711484Af (5), Pc (1)Both stages
Boletaceae #3AM113455Phylloporus rhodoxanthus / DQ53398079Af (2)Adults
Clavulinaceae #1AM113459Clavulina castaneipes / EU66920982Af (4), Ct (1)Seedlings
Clavulinaceae #2AM113460Clavulinaceae sp. / AJ53470881Af (1)Seedlings
Russulaceae #1AM113427Russula compacta / EU59817283Am (6), Ct (1), Pc (2), Ue (1)Both stages
Russulaceae #2AM113428Uncultured Russulaceae / DQ77797884Af (1), Pc (1)Adults
Russulaceae #3AM113429Uncultured Russulaceae / DQ77797882Ct (3)Adults
Russulaceae #4AM113430Lactarius pelliculatus / AY60697895Am (1), Pc (1)Seedlings
Russulaceae #5AM113431Lactarius pelliculatus / AY60697895Am (1), Ct (1), Pc (2)Both stages
Russulaceae #6AM113432Lactarius pelliculatus / AY60697896Ue (1)Adults
Russulaceae #7AM113433Uncultured Russula / AY66742689Pc (2)Adults
Russulaceae #8AM113434Lactarius pelliculatus / AY60697895Ct (3)Both stages
Russulaceae #9AM113435Lactarius pelliculatus / AY60697895Ue (1)Adults
Russulaceae #10AM113436Lactarius pelliculatus / AY60697895Ue (1)Adults
Russulaceae #11AM113437Lactarius pelliculatus / AY60697895Pc (2)Both stages
Russulaceae #12AM113438Lactarius pelliculatus / AY60697895Ue (1)Adults
Russulaceae #13AM113439Lactarius pelliculatus / AY60697895Pc (2), Ue (2)Both stages
Russulaceae #14AM113440Lactarius pelliculatus / AY60697895Ue (2)Seedlings
Russulaceae #15AM113441Uncultured fungus / FM99965985Af (1), Am (1), Ct (2), Pc (1), Ue (5)Seedlings
Russulaceae #16AM113442Lactarius pelliculatus / AY60697896Af (17), Am (29), Ct (34), Pc (34), Ue (29)Both stages
Sclerodermataceae #1AM113464Scleroderma sp. / AB09990093Am (1), Ct (2), Ue (2)Seedlings
Sclerodermataceae #2AM113465Scleroderma sp. / AB09990094Af (2), Am (1), Ct (3), Pc (1), Ue (5)Both stages
Thelephoraceae #1AM113443Uncultured Tomentella / EF21882691Af (3)Adults
Thelephoraceae #2AM113444Uncultured Thelephoraceae / DQ27342089Am (1), Pc (1)Seedlings
Thelephoraceae #3AM113445Uncultured ectomycorrhizal fungus / FM99321192Ct (5), Pc (1)Both stages
Thelephoraceae #4AM113446Uncultured Tomentella / GQ24090892Ct (1)Seedlings
Thelephoraceae #5AM113447Uncultured Thelephoraceae / GQ24090390Ct (1)Adults
Thelephoraceae #6AM113448Uncultured fungus / GQ20537288Af (11), Am (1), Ct (2), Pc (5), Ue (1)Both stages
Thelephoraceae #7AM113449Uncultured ectomycorrhizal fungus / FM99326288Af (1)Seedlings
Thelephoraceae #8AM113450Uncultured fungus / GQ20537288Af (9), Am (1)Adults
Thelephoraceae #9AM113451Uncultured fungus / FJ82058189Pc (3)Seedlings
Thelephoraceae #10AM113452Uncultured fungus / GQ20537188Af (2), Am (3), Ct (1), Pc (4), Ue (3)Both stages
Tricholomataceae #1AM113456Uncultured fungus / FJ82056097Ue (1)Adults
Tricholomataceae #2AM113457Tricholoma atroviolaceum / AY75016691Pc (1)Seedlings
Tricholomataceae #3AM113458Uncultured mycorrhizal fungus / AB45438294Am (1)Adults

Ectomycorrhizal abundance and distribution among host species

Among the 39 ECM taxa, 20 (including 11 singletons) were found on a single host species and 19 were multi-host: six were shared by the five tree species, two by four, two by three and nine by two tree species (Table 2). Each tree species associated with 10–15 multi-host ECM fungal taxa, 2–3 single-host ECM fungal taxa, and 1–5 singletons. To compare the abundance of single-host ECM taxa and multi-host ECM taxa, the singletons were excluded from the analysis. Therefore, from the considered 292 ITS sequences, the multi-host ECM taxa represented 92% of the total ECM abundance versus 8% for the single-host ECM taxa (Fig. 2). The five tree species displayed relatively similar abundances of multi-host ECM taxa (18–22% of total abundance of multi-host ECM). The most abundant ECM fungal taxon, Russulaceae #16 occurred on the five host species and represented 47% of the overall abundance.

Figure 2.

Multi-host ECM fungal taxa (above the zero line of the histogram), single-host ECM fungal taxa and singletons (below the zero line of the histogram) classified by rank of abundance. Taxa were recovered from seedlings only (white), adults only (grey), or both development stages (black). For clarity, the most abundant ECM taxon, Russulaceae #16 (47% of the total abundance) was omitted.

A comparison of the frequency of each multi-host fungal taxon on the five host species revealed that some multi-host fungal taxa tend to display a preferential occurrence (Table 2). For instance, Thelephoraceae #8 was found on only Anthonotha fragrans and Anthonotha macrophylla, and was more frequent (nine occurrences) on A. fragrans than on A. macrophylla (only one occurrence). Boletaceae #2 was also found on only two host plant species with five occurrences on A. fragrans and one occurrence on Paramacrolobium coeruleum.

At the fungal family level, the factorial correspondence analysis (FCA) showed that Boletaceae, Clavulinaceae and Thelephoraceae tended to associate preferentially with A. fragrans: these fungal families were associated with positive values of the axis F1 (with high contribution to this axis; 26%, 26% and 25% respectively), while Russulaceae was associated with negative values of this axis (with a 18% contribution) and tended to associate more with the other tree species (Fig. 3).

Figure 3.

First factorial plane projection of ECM taxonomical groups (closed circles), adults and seedlings of the five host tree species (open square). The axes F1 and F2 represent 54.4% and 24% of the total variability respectively. Afa, A. fragans adult; Afs, A. fragans seedlings; Ama, A. macrophylla adults; Ams, A. macrophylla seedlings; Cta, C. tetraphyllum adults; Cts, C. tetraphyllum seedlings; Pca, P. coeruleum adults; Pcs, P. coeruleum seedlings; Uea, U. esculenta adults; Ues, U. esculenta seedlings.

Composition of the ECM community among host species and development stages

Anthonotha fragrans displayed different fungal communities relative to other species, except P. coeruleum for its seedlings. Conversely, whereas the adult trees of the four remaining host species displayed significantly different fungal communities from each other, except the pair P. coeruleumA. macrophylla, their seedlings had similar fungal communities (Table 3). This was also well supported in FCA, where seedlings were more closely patched than adults (Fig. 3).

Table 3.  Genotypic differentiation tests (upper half of the matrix) among fungal communities isolated from adults and seedlings of the five host species (P-values reported were estimated from the Fisher exact test using the Genepop program, and bold values indicate significant or marginally significant differentiation), and similarity index (lower half) between host species (similarity values calculated as the coefficient of Sørensen from occurrence of ECM fungal taxa on adults and seedlings).
Host speciesA. fragransA. macrophyllaC. tetraphyllumP. cœruleumU. esculenta
A. fragransAdult  < 0.001 < 0.001 < 0.001 < 0.001 
Seedlings0.33  0.019 0.005 0.059 < 0.001
A. macrophyllaAdults0.380.22  0.008 0.28 0.03 
Seedlings0.220.600.33  0.68 0.70 0.52
C. tetraphyllumAdults0.  < 0.001 0.002 
Seedlings0.290.520.570.610.32  0.12 0.32
P. cœruleumAdults0.400.240.400.350.150.40  < 0.001 
Seedlings0.350.560.260.640.290.430.27  0.14
U. esculentaAdults0.  

Moreover, similarity values were higher for all seedlings pairs (Sørensen coefficient ≥ 0.44) than for adult tree pairs or for seedlings and adult tree pairs, regardless of the host species, except for Cryptosepalum tetraphyllum seedlings and A. macrophylla adults (0.57; Table 3). Moreover, adults of Uapaca esculenta, P. coeruleum and A. macrophylla had higher similarity to non-conspecific seedlings than to conspecific seedlings (Table 3). For instance, U. esculenta adults displayed 21% similarity to their own seedlings versus 30 ± 3% (mean ± SE) to non-conspecific seedlings. The opposite trend was observedfor C. tetraphyllum (C. tetraphyllum adults were 32% similar to their seedlings versus 17 ± 4% to non-conspecific seedlings) while no significant difference was observed for A. fragrans (adults were 33% similar to their conspecific seedlings versus 28 ± 3% to non-conspecific seedlings). When pooling seedlings and adults together, species of Caesalpinioideae did not share significantly more ECM fungal taxa among them than with the Phyllanthaceae U. esculenta: the average similarity was 55 ± 5% for Caesalpinioideae species pairs versus 49 ± 3% for Caesalpinioideae and Phyllanthaceae species pairs, so that host phylogenetic position poorly structured the ECM community.

When considering the 19 multi-host fungal taxa exclusively and the five tree species combined, the Shannon index showed that seedlings had a larger diversity than adults (1.89 versus 1.72, respectively, for both accumulation curves rarefied at n = 109). Rarefaction analyses (Fig. S1) suggested that diversity did not greatly vary between cohorts (Table 1), and also supported a higher diversity for seedlings, except for U. esculenta. Although significant qualitative differences were observed between seedling and mature stages of the different trees species for fungal diversity (not shown), 12 out of 19 multi-host fungal taxa occurred on both adults and seedlings. Conversely, when considering the single-host fungal taxa solely or including singletons, adults had a higher level of diversity than seedlings (1.62 versus 1.32, respectively, when considering single-host taxa solely and rarefying at n = 8, and 2.22 versus 1.97, respectively, when admitting singletons and rarefying at n = 12). Altogether, in accordance with the previous results (Table 3), adults tended to harbour more single-host ECM fungi than their seedlings.

Ectomycorrhizal composition among the four plots

When combining all host species and stages, the similarity of the ECM fungal assemblages among the four plots B, G, L and O was relatively low (Table 4). The inter-plot distances did not correlate with similarity in fungalcomposition: e.g. plots B was more similar to O than to G, although G was closer to B. Similar results were obtained when considering adults and seedlings separately (not shown). Some ECM taxa tended to associate with particular plots, as shown by the χ2 independence test (P = 0.999). For example, Russulaceae #1, Thelephoraceae #3 and Thelephoraceae #6 (the most abundant taxa in plots O, L and G respectively) were recorded in only one or two plot(s), suggesting aggregated distribution for some taxa. Conversely, Russulaceae #16 did not prefer any particular plot (15–47% of the total abundance in each plot), and neither did Sclerodermataceae #2 (always < 4% in total abundance), showing that the detection on several plots was not necessarily related to the abundance.

Table 4.  ECM taxa similarity among plots B, G, L and O, pooling data from adults and seedlings.
  1. Similarity values calculated as the coefficient of Sørensen are reported in the upper halves of the matrices, and distances (m) among stands are reported in the lower half of the first matrix.


Moreover, ECM taxa did not present the same apparent extent inside the different plots. For example, Russulaceae #16 was among the most widespread ECM taxa in the four plots, while Thelephoraceae #10, which covered one of the largest apparent areas in plot B, was not found in plot G. The longest distance between the two outermost sampled ectomycorrhizas was recorded for Thelephoraceae #6 in plot G, while it displayed a relatively small apparent area size in plot O, and was not found in plots B and L. On the other hand, there was some overlap in spatial distribution of ECM taxa in plots (Fig. S2).

Isotope signature, C/N ratio and N concentration of the plant leaves

Whenever seedlings or adults of the same species were available from the investigated plots B and L (n = 6 species), the differences in δ13C and δ15N values were not significant, except for A. macrophyllaδ13C values in seedlings (higher by 1.05‰ in plot L, P = 0.001). Altogether, when pooling data of adults on each site, δ13C values of ECM and AM legumes did not differ (P > 0.05), and were higher than those of non-legumes in plot L only (P = 0.001); δ15N never significantly differed among these three categories. No significant differences in δ13C and δ15N were recorded between seedlings and adult individuals for all ECM tree species on both plots (Fig. 4); these differences were not significant for the AM legumes, except for Bussea occidentalisδ13C values on plot B that were significantly higher for seedlings (+0.9‰, Fig. 4A). Among non-legumes, the only significant differences between adults and seedlings were for T. longifoliaδ13C values in plot B (seedlings values are 3.4‰ lower, Fig. 4A), and Newbouldia laevis in plot L (δ13C was 1.4‰ higher and δ15N was 2.2‰ higher for seedlings, Fig. 4D, 4E). The fungal fruit bodies differed in δ13C compared with all other tree samples (δ13C = –25.84 ± 1.80‰, mean ± SE; P = 0.001; not shown). Their δ15N values were largely dispersed (δ15N = 2.17 ± 2.31‰; not shown), and did not differ (P > 0.05) from all other tree samples.

Figure 4.

Values of δ13C and δ15N (‰) and C/N ratio of plants collected in plot B (A, B, C) and plot L (D, E, F), with indication of significant differences between seedlings and adults for each species according to pairwise anova (*P < 0.05; **P < 0.01; ***P < 0.001). Ectomycorrhizal trees: Ama and Ams correspond to Anthonotha macrophylla adults and seedlings respectively; and in the same way: Cta and Cts for Cryptosepalum tetraphyllum; Pca and Pcs, Paramacrolobium coeruleum; Uea and Ues, Uapaca esculenta. Non-ECM legumes: Boa and Bos, Bussea occidentalis; Ega, Entala gigas adults. Non-legumes: Baa and Bas, Bosquiea angolensis; Nla and Nls, Newbouldia laevis; Pba and Pbs, Pentadesma butyracea; Tla and Tls, Trichoscypha longifolia.

The C/N ratio values measured did not differ significantly for species investigated on the two plots, except for B. occidentalis (higher C/N ratio in plot L; P = 0.01; Fig. 4). When pooling data of adults on each plot, C/N ratio did not differ between AM legumes and non-legumes, or between ECM and AM species (P > 0.05), as well as when considering total N content (P > 0.05; not shown). C/N ratios in leaves of adults and seedlings never differed for all species, with the exception of lower C/N ratios in seedlings for the AM legume B. occidentalis in plot L (Fig. 4F) and for the non-legumes B. angolensis and T. longifolia in plot B (Fig. 4C).


Pattern of an ECM fungal community from tropical Africa

Up to now, with a few exceptions (Tedersoo et al., 2007a), most studies on ECM communities carried out in African tropical rainforests have focused on epigeous fruit bodies (e.g. Rivière et al., 2007). We report here one of the first samplings of ECM roots from five host species in a primary rainforest. It showed 39 ECM fungal taxa from 362 ECM tips on 1600 m2, a species number to area ratio a bit lower than for ECM communities in Holarctic regions (e.g. Richard et al., 2005; Walker et al., 2005; Ishida et al., 2007). Whether this reflects a trend to less diversity in tropical ECM community, as observed in Asia (K. Nara, pers. comm.) requires further analysis of other sites. In addition, our low sample size and small sampling area probably strongly underestimate the diversity.

The dominance of Russulaceae and Thelephoraceae in this community is reminiscent of temperate ECM forests (Horton and Bruns, 2001), in spite of the noticeable absence of Cenococcum geophilum, Cortinariaceae and Sebacinales, which are widespread or dominant in Holarctic communities (e.g. Richard et al., 2005; Walker et al., 2005; Diédhiou et al., 2009). Similar communities, including Tricholomataceae, Boletaceae and Sclerodermataceae, have already been reported from the African tropics, in Madagascar (Ducousso et al., 2004), Senegal (Diédhiou et al., 2004), Western Guinea (Rivière et al., 2007), and the Seychelles (Tedersoo et al., 2007a). Ectomycorrhizal fungal communities from tropical Guyana (Moyersoen, 2006) and Asia (Ingleby et al., 1998; Rivière et al., 2007) tend to harbour different dominant families, perhaps as a result of distance. Our study revealed a high spatial heterogeneity among the four plots analysed, due in part to the patchy distributions of some ECM fungal taxa, as reported for more northern communities (Richard et al., 2005), and also provided information on association patterns, in relationship with host age and species.

Multi-host versus single-host fungal taxa

Ectomycorrhizal host-specificity is little studied in tropical tree species. Outplanting experiments and habitat preferences for various fungi suggest that tropical ECM fungi are more specific than temperate ones (Smits, 1994), but very few data support this in African tropical forests (Onguene, 2000; Onguene and Kuyper, 2002). Specificity is thought to be low in tropical ECM Dipterocarpaceae from Asia (Alexander and Lee, 2005; K. Nara, pers. comm.). In our study, half of fungal taxa displayed a broad host-range. In temperate forests, early reports suggested dominance of multi-host fungi in forest ecosystems (up to 90% of the ECM fungal community: Horton and Bruns, 1998; Cullings et al., 2000), while later reports revealing more rare fungal species indicated only 15–35% multi-host species (Kennedy et al., 2003; Richard et al., 2005; Ishida et al., 2007; Tedersoo et al., 2008). However, multi-host ECM species tend to be the most abundant: here, they represent 89% (92% when excluding singletons from the analysis) of the overall ECM colonization. Congruently, in a Mediterranean forest, while representing only 15% of the fungal taxa, multi-host ECM fungi colonized c. 70% of the roots (Richard et al., 2005). Moreover, we cannot exclude that the apparent specificity of some fungal taxa only reflects their rarity in the restricted sampling area (Fig. 2): all single-host taxa displayed no more than 1% of the total abundance. Apparently single-host taxa and singletons may result from insufficient sampling. However, several multi-host taxa were not abundant, ranging from 2% to less than 1%, so that the ability to be a multi-host is not fully related to abundance. Thus, although possibly underestimating the number of multi-host ECM taxa, our study shows the potential for mycorrhizal networks links between tree species: however, only intra-specific molecular markers, such as microsatellites (Lian et al., 2006), and denser sampling could rigorously reveal the sharing of individual ECM mycelia between tree species.

Multi-stage versus single-stage fungal taxa

As far as the host developmental stage is concerned, three equally frequent kinds of ECM taxa were seen: 11 taxa were isolated solely from seedlings, 14 were isolated only from adults and 14 were shared by the two developmental stages, including the most common ones (Fig. 2). In terms of abundance, these multi-stage fungi represented 79% of the overall ECM colonization. In well-established forests, the external fungal mycelium supported by adult trees likely contributes to ECM community structure on seedlings (Simard et al., 1997; Walker et al., 2005), and ECM fungal diversity can be expected to be similar among trees of different ages. This was evidenced in Mediterranean holm oak (Quercus ilex) forests (Richard et al., 2005) and in temperate coniferous forests (Pinus sylvestris, Jonsson et al., 1999; Tsuga heterophylla, Kranabetter and Friesen, 2002). Accordingly, Kennedy and colleagues (2003) showed that Lithocarpus densiflora seedlings shared 17 of 56 ECM fungi (30%) with overstorey Pseudotsuga menziesii trees, but still the diversities of ECM fungi on seedlings and adults significantly differed. Ectomycorrhizal communities can thus change according to the developmental stage (seedling or adult) of host species in our forest. Similarly, Lee and Alexander (1996) observed in Malaysian tropical rainforests that the ECM community on dipterocarp seedlings changed within 7 months following germination. Thus, age and history of hosts may influence the ECM community, a feature also reported for AM fungal species in the tropics by Husband and colleagues (2002a,b). Whether the occurrence of ECM species preferring (or preferred by) seedlings is more pronounced in tropical regions or not remains unclear; similarly, the role of this change in buffering changes of environmental conditions over development (Simard et al., 1997; Husband et al., 2002a,b) requires further study.

Comparisons of the ECM community among host species

When seedlings and adults were considered together, there was no significant difference in terms of ECM composition among host tree species, except for A. fragrans. Although some host preferences were detected at the level of fungal species or families, the magnitude of host effects is lower than in those observed in previously investigated mixed forests (Ishida et al., 2007; Morris et al., 2008; Tedersoo et al., 2008; Smith et al., 2009). Furthermore, we anticipate that the host preference patterns may be related to spatial non-independence of the samples, low number and aggregated distribution of the adult trees and seedlings at the study site. However, the comparison between ECM associates of the various host species showed contrasting results, depending on the tree age. The seedlings of A. macrophylla, C. tetraphyllum, P. coeruleum and U. esculenta did not significantly differ in ECM fungal associates (Table 3, Fig. 3). This, together with the presence of multistage fungi and the intermingling of adult roots (particularly dense, our personal observations), supports the potential for mycorrhizal network formation between seedlings and adults.

The adult trees hosted a more specific fungal community than seedlings (Table 3), linked to a reduction of fungal diversity at the expense of the multi-host species. Thus, seedlings are generalist, but tend to specialize more in single-host ECM fungi when they become older. In a Japanese conifer-broadleaf ECM forest, Abies homolepis is colonized by more specific ECM fungi many years after its establishment (Ishida et al., 2007); in mixed western hemlock and Douglas-fir stands the two tree species share more ECM fungi in young forests (40 years old) than in older ones (400 years old; Horton et al., 2005). Three non-exclusive reasons can account for the lower specificity of seedlings. First, their smaller roots systems may limit their access to all fungal partners, especially in such patchy ECM fungal communities where possibly specific ECM taxa are rare. Second, a plant may be able to select among different fungal partners throughout its life cycle: it can be expected that older trees tend to recruit more specific fungal species in order to avoid interaction with surrounding species (Selosse et al., 2006), while seedlings tend to use and thus maximize this interaction. Third, changes over a life span could be driven by fungal competition: as recruitment of ECM fungi on root systems proceeds, the increasing competition may lead to progressive elimination of some species, favouring more specific fungal species, maybe because they are more adapted to a given host. Future research on symbiotic interactions between plant and fungi over their life cycle in realistic, ECM conditions (e.g. Kennedy and Bruns, 2005) will be required to understand such developmental changes, and their functional implications.

Possible advantage of lower specificity in seedlings

Thanks to their lower specificity, seedlings may receive inocula from adults: curiously, except for C. tetraphyllum and A. fragrans, adults tend to share more ECM fungal taxa with the non-conspecific seedlings than with their own seedlings. Adults likely function as ‘nurse trees’ for conspecific and non-conspecific seedlings, and therefore promote diversity and coexistence of species in this forest stand. The selective pressures behind this apparently ‘altruistic’ phenomenon are unclear, as it may be detrimental for adults and their conspecific seedlings. However, it may contribute to the formation of the stands where ECM trees dominate, as is often observed for ECM tropical forests (Hart et al., 1989; Newbery et al., 1997; 2004; Alexander and Lee, 2005; McGuire, 2007): perhaps, a global positive feedback of for nourishing ECM fungi allows AM species to be outcompeted.

Besides a simple inoculum effect, the connection to a ‘prepaid’ mycorrhizal network supported by carbon from adults may allow understorey seedlings to reduce symbiosis costs and thus compete better against ECM and AM adults (Molina et al., 1992; Cullings et al., 2000). Indeed, there is good evidence from temperate forests that overstorey trees contribute more to ECM fungal nutrition (Högberg et al., 1999). Interestingly, the survival and establishment rates of seedlings in tropical forests depend on the rate at which they become mycorrhizal (Alexander et al., 1992; Newbery et al., 2000). Seedlings in contact with adult ECM trees often have better survival and mycorrhizal colonization than seedlings isolated from adults, in tropical (Onguene and Kuyper, 2002) and temperate forests (Dickie et al., 2002). In Guyana's forest, access to mycorrhizal networks also increases survival and growth of ECM seedlings (McGuire, 2007).

Strikingly, only A. fragrans seedlings had a significantly different community as compared with other species (Table 3). Although it may persist in older stands, A. fragrans is the only species under study that can easily establish alone, earlier in the colonisation succession. Thus, its seedlings are adapted to grow without the assistance of any mycorrhizal networks. Additionally, due to its large seed reserves, A. fragrans might be less dependent upon ECM fungi at early stages, as reported for other large-seeded species (Allsopp and Stock, 1992; Zangaro et al., 2000).

No evidence for nutrient transfer from adult ECM trees to seedlings

The potential for mycorrhizal networks opened the possibility of some nutrient exchanges. As expected, δ13C values of fungal fruit bodies were significantly higher than those of plants (c.+4‰), as a result of an isotopic fractionation between donor trees and ECM fungi (Högberg et al., 1999; Boström et al., 2008). Thus, seedlings receiving carbon from fungi can be expected to have higher δ13C values than adult leaves, and closer to ECM fungi, because transfer to receiver plants usually does not entail any shift in δ13C (Julou et al., 2005; Tedersoo et al., 2007b). However, the absence of significant difference in δ13C between seedlings and adults suggests that both share indistinguishable sources of carbon (Fig. 4), with no carbon transfer to ECM seedlings. This result must be interpreted with caution, as there are considerable inter-specific variations in δ13C signatures among ECM adults (see Fig. 4), and there is a possibility that an undetectable transfer (thus of debatable role) occurs. Isotopic methods are powerful, because they integrate carbon metabolism over long periods, but they have low resolution for low carbon flux – however, here, even a few per cent of the whole carbon budget would be relevant, due to heavy competition for light.

The mean δ15N values did not differ between ECM legume, AM legume and non-AM legume species, an unexpected feature already reported for tropical rainforests (Högberg and Alexander, 1995). Among legume species, the high δ15N values obtained in A. macrophylla, C. tetraphyllum and P. coeruleum were consistent with their non-N2-fixing status considering that they were never found to be nodulated in a study performed in the same forest area, as all Caesalpinioideae ranged in the Amherstieae tribe in general (Diabatéet al., 2005). Similarly, the high δ15N value obtained in B. occidentalis belonging to the Caesalpinieae tribe confirmed its inability to fix N2 as previously observed. By contrast, δ15N values obtained in Entada gigas were almost nil, thus confirming its nodulation and N2-fixing ability previously reported (Diabatéet al., 2005) for about 95% of Mimosoideae species. No significant difference in δ15N was recorded between seedlings and adults of ECM plants, again suggesting that ECM seedlings did not receive carbon via their fungi. Interestingly, the non-legume AM N. laevis had significantly higher δ15N values for seedlings than adults. Differences in total N content between N. laevis seedlings and adults could explain different δ15N, but no such difference was observed here. Thus, together with a higher δ13C, we cannot rule out a C transfer to N. laevis seedlings by AM symbionts, although other explanations may account for these features.


Our results, with at least a half of multi-host ECM fungi which are the most abundant ECM types, showed a potential for mycorrhizal network formation, perhaps underestimated due to the sampling design. Moreover, seedlings tended to be more generalist and adults more specific. Although no carbon transfer to seedlings was detected, connecting to pre-existing mycorrhizal networks may allow seedlings to spare carbon. Therefore, the existence of mycorrhizal networks and individual fungal sharing, as well as their impact on growth and fitness of seedlings, remain to be rigorously assessed in this ecosystem.

Experimental procedures

Site description

South of Guinea is one of the last regions of West Africa retaining a primary tropical rainforest, at elevations ranging from 500 to 1750 m. The research was conducted in the typical evergreen Ziama rainforest near Sérédou (8°51N, 9°31W, altitude 600 m), characterized by a mean annual rainfall of 3000 mm and a short dry season from January to March Temperatures are in general above 24°C with relative humidity up to 80%. Soils are ferralitic, acid and very poor in available phosphorus. The canopy is dominated by C. tetraphyllum (ECM Caesalpinioideae, Fabales), as well as the AM trees Erythrophleum ivorense (Fabaceae) and Heritiera utilis (Malvaceae, Malvales), with rare Piptadeniastrum africanum (AM Fabaceae), A. fragrans (ECM Caesalpinioideae) and Uapaca spp. (ECM Phyllanthaceae, Malpighiales). Canopy trees are at least 30 m tall, with some emergent individuals reaching 50–60 m in height (Diabatéet al., 2005). We restricted our study to an area dominated by five ECM trees, i.e. four Caesalpinioideae [A. fragrans (Af), A. macrophylla (Am), C. tetraphyllum (Ct), P. coeruleum (Pc)] and the Phyllanthaceae U. esculenta (Ue), with abundant ECM seedlings (Fig. 1).

Sampling of ectomycorrhizae

Ectomycorrhizae were sampled in July 2003 from the five ECM host species, in a 1600 m2 area subdivided into 16 plots of 100 m2 each (A to P, Fig. 1), containing 20 adult trees and 332 seedlings (defined as less than 50 cm in height and 1.5 cm in stem diameter) that were individually mapped. For the adult trees, whenever possible root tips were sampled around each tree by tracing roots from the base of trunk. Ten to 30 root samples were randomly harvested per species, from 0 to 35 cm deep (Table 1). The seedling root tips (on a 4–5 cm tap root each) were simultaneously collected at no more than 50 cm from each adult root sampling point in order to avoid spatial heterogeneity (Table 1). All root samples were kept in cool conditions (4°C) until arrival at the laboratory 1–3 days after field collection. They were then gently separated from surrounding soil and rinsed in a sieve (0.5 mm diameter) with tap water. Ectomycorrhizal tips were observed under a binocular microscope, carefully separated from roots, and kept for subsequent molecular analyses (Table 1). Ectomycorrhizal tips were placed on a cotton layer in 10 ml tubes half-filled with silica gel, and stored at room temperature.

Molecular identification of fungi and plants from ECM

Total DNA was isolated from each ECM tip using a DNAeasy Plant mini-kit following the manufacturer's recommendations (Qiagen, Courtaboeuf, France). Identification of ECM fungi was performed for all ECM tips by amplifying and sequencing the nuclear rDNA ITS with the fungus-specific primer pair ITS1f–ITS4. Whenever several amplification products were obtained from the same ECM tip, each was extracted from agarose gel and re-amplified separately. Fungal DNA amplification was performed as described in Diédhiou and colleagues (2004).

The ITS region was sequenced using the PCR primers; forward and reverse DNA sequences were aligned to produce a consensus DNA sequence. All sequences were aligned with the Clustal X program (Thompson et al., 1997). Alignments were manually optimized with the Genedoc program (Nicholas et al., 1997). Whenever two sequences differed by one base only, amplification and sequencing were repeated from the samples to confirm the dissimilarity. Individual ECM tips generating two different ITS sequences were considered as harbouring two distinct ECM fungi. In order to determine their taxonomical affiliation, sequences were compared with the GenBank database using the blastn algorithm (Altschul et al., 1990). Ectomycorrhizal taxon names were designated following the blast score. The identity values were often ≤ 95% (Table 2), therefore only families were considered here.

Confirmation of roots identity for samples from adult trees was necessary, due to root intermingling. This was conducted by single strand conformational polymorphism (SSCP) analysis of the trnL intron region of the chloroplast DNA, comparing DNA from ECM and leaves of the five host species. Introns were amplified with primers trnL-c and trnL-d (Taberlet et al., 1991). Each purified PCR product (1–3 µl) was mixed with 7–9 µl of a denaturing buffer (95% formamide, 4% EDTA, 0.05% bromophenol blue and 0.05 xylene cyanol). The mixture (10 µl) was heated for 3 min at 98°C, subsequently cooled on ice, loaded onto an 8% non-denaturing polyacrylamide gel and stained with ethidium bromide after migration. The SSCP analysis always confirmed the identification obtained by root tracing (data not shown).

Isotopic analyses

Plant leaves were collected from two distant plots (B and L, Fig. 1) in March 2006, at 10–30 cm above soil (to avoid isotope distortion due to CO2 from soil respiration) and in similar light conditions (because low light supply results in slower photosynthesis, and thus higher discrimination against 13C; Julou et al., 2005). We collected leaves from (i) ECM legumes (A. macrophylla, C. tetraphyllum and P. coeruleum) on plot L and (ii) the same ECM legumes and U. esculenta on plot B. As controls, AM legume trees (B. occidentalis as non-N2-fixing species on both plots and Entada gigas as N2-fixing tree species on plot L) were sampled. We also collected leaves from non-legume AM trees, namely Pentadesma butyracea (Clusiaceae) and N. laevis (Bignoniaceae) on plot L as well as Bosquiea angolensis (Moraceae) and Trichoscypha longifolia (Anacardiaceae) on plot B. Sampling was also extended to places surrounding the plots, to allow the collection of leaves on independent adults. In addition, fruit bodies of the eight most abundant fruiting taxa of which seven ECM fungi (Amanita griseoflocosa, Amanita sp., Boletus sp., Cantharellus sp., Russula sp.1, Russula sp.2, Scleroderma sp.) and one non-ECM fungus (Hygrocybe sp.) were collected in 2005–2006 to estimate their isotopic contents. For each sampled plant species, when available, one leaf collected on six different adults or seedlings were dried at 50°C for 48 h. Samples were handled as in Tedersoo and colleagues (2007b), and isotope abundances were expressed in δ13C and δ15N values in parts per thousand relative to international standards for measurements of V-PDB and atmospheric N2:


where R is the molar ratio, i.e. 13C/12C or 15N/14N.

Statistical analyses

The relative abundance of each ECM taxon (as identified by ITS sequence) was estimated as the percentage of its occurrence in each or all tree species. The genotypic differentiation test was performed to test putative differences among fungal communities between each host species (or between seedlings and adults), using the Genepop program (Raymond and Rousset, 1995). The P-values were estimated from a log-likelihood-based exact test. The principle is the same as that of the Fisher exact test (Goudet et al., 1996). Each P-value was calculated as the sum of the probabilities of all tables (with the same marginal values as the observed one) with a probability lower than or equal to the observed table. To compare species accumulation and richness estimates among the various plant cohorts, rarefaction curves with 95% confidence intervals were calculated using the EstimateS software version 8.0.0 (Colwell, 2006). This corrects for differences in sampling size and allows virtual reduction of the size of all samples to that of the smallest one. The Shannon index (Shannon and Weaver, 1949) was also calculated to estimate the fungal diversity sampled from adults and seedlings, using the EstimateS software.

Factorial correspondence analysis was performed to visualize the relationships between ECM fungal communities and their hosts using the XLSAT™ software package (version 2009, Addinsoft, Paris, France). This involved the construction of a contingency table (host trees versus fungal taxa), in which fungal taxa were represented in terms of their occurrences (frequency) on both life stages of the five host species. The calculation of variance is based on the χ2 test, which measures the extent to which a sample distribution deviates from a theoretical distribution and is an integral part of the correspondence analysis. Factorial correspondence analysis was presented as a plane projection of the two most informative axes accounting for the ECM community composition (Bertout et al., 1999).

The comparison of ECM composition among the plots was restricted to four plots (B, G, L and O) that included the maximum number of individuals and species of ECM hosts with the most homogeneous distribution (Fig. 1; each selected plot was at least 10 m away from the others). Ectomycorrhizal roots sampled from these four plots were individually mapped. The similarity of the ECM composition among the plots was compared using the Sørensen coefficient (Krebs, 1999), calculated for all ECM sampled from each plot irrespective of the development stage. To check whether ECM fungi exhibit some trend for plot preference, we used a χ2 test of independence for the number of occurrences of each ECM taxon in each selected plot.

For isotope analysis, total N concentrations, δ13C and δ15N values were tested for normality and homogeneity of variances using a Wilk-Shapiro W-test and a Levene test respectively. One-way analysis of variance (anova) was separately performed for each variable and site combined with the multiple range test to distinguish homogeneous mean groups at α = 0.05.


We thank Jean Garbaye and Thomas W. Kuyper for their insightful comments on and discussion of earlier versions of this manuscript, Alexandre Geoffroy for help in isotope analysis, and Fabrice Elegbede for help in statistical analyses. We thank the anonymous referees for their valuable comments on this paper, and David Marsh for improving the English.