6 Dispersion , an Important Radiation Mechanism for Ectomycorrhizal Fungi in Neotropical Lowland Forests ?

Mycorrhizas are symbiotic associations between plant roots and fungi. They are divided into different categories depending on morphological characteristics and the identity of the fungi and plants (Smith & Read, 2008). In tropical rainforests, most tree species are associated with arbuscular mycorrhizas (AM) (Alexander, 1989). Neotropical tree species belonging to confirmed ectomycorrhizal (EcM) host genera are diverse and generally scattered in a wide range of vegetation types (Table 1). Pakaraimaea dipterocarpacea, an endemic tree species from Guayana region (Maguire et al., 1977; Maguire & Ashton, 1980; Moyersoen, 2006), is one of the few known locally dominant EcM tree species in the Neotropical lowland forests. This tree is phylogenetically related to the most important EcM tropical tree family in SE Asia: the Dipterocarpaceae. The disjunct distribution of P. dipterocarpacea suggests that the capacity to associate with EcM evolved in the ancestors of Dipterocarpaceae before the splitting of South America from Gondwana, c. 135 million year ago (Ma) (Moyersoen, 2006). There are alternative proposals to explain P. dipterocarpacea EcM status such as independent acquisition of EcM association in South America or a more recent transoceanic long-distance dispersal of EcM Dipterocarpaceae to the Neotropics (Alexander, 2006). Biotrophic fungi such as EcM can be an important source of information for improved understanding of disjunct distributions (Pirozynski, 1983). The phylogeny of P. dipterocarpacea EcM fungi might reflect the disjunct distribution of this endemic tree species. A pioneering survey on P. dipterocarpacea indicated that broad host range fungal lineages distributed across the tropics are associated with this host plant (Moyersoen, 2006), but no phylogeographic studies on P. dipterocarpacea EcM fungi are available to date.


Introduction
Mycorrhizas are symbiotic associations between plant roots and fungi.They are divided into different categories depending on morphological characteristics and the identity of the fungi and plants (Smith & Read, 2008).In tropical rainforests, most tree species are associated with arbuscular mycorrhizas (AM) (Alexander, 1989).Neotropical tree species belonging to confirmed ectomycorrhizal (EcM) host genera are diverse and generally scattered in a wide range of vegetation types (Table 1).Pakaraimaea dipterocarpacea, an endemic tree species from Guayana region (Maguire et al., 1977;Maguire & Ashton, 1980;Moyersoen, 2006), is one of the few known locally dominant EcM tree species in the Neotropical lowland forests.This tree is phylogenetically related to the most important EcM tropical tree family in SE Asia: the Dipterocarpaceae.The disjunct distribution of P. dipterocarpacea suggests that the capacity to associate with EcM evolved in the ancestors of Dipterocarpaceae before the splitting of South America from Gondwana, c. 135 million year ago (Ma) (Moyersoen, 2006).There are alternative proposals to explain P. dipterocarpacea EcM status such as independent acquisition of EcM association in South America or a more recent transoceanic long-distance dispersal of EcM Dipterocarpaceae to the Neotropics (Alexander, 2006).Biotrophic fungi such as EcM can be an important source of information for improved understanding of disjunct distributions (Pirozynski, 1983).The phylogeny of P. dipterocarpacea EcM fungi might reflect the disjunct distribution of this endemic tree species.A pioneering survey on P. dipterocarpacea indicated that broad host range fungal lineages distributed across the tropics are associated with this host plant (Moyersoen, 2006), but no phylogeographic studies on P. dipterocarpacea EcM fungi are available to date.
Apart from human transport, there are three possible hypotheses to predict the phylogeography of EcM fungal community associated with P. dipterocarpacea, i.e. vicariance, dispersion and/or migration.Both migration and vicariance assume that there is a close and specialized relationship between the fungus and the plant partners in EcM symbiosis (Pirozynski, 1983;Halling et al., 2008).Any discovery of fungi with an old Gondwanan origin associated with P. dipterocarpacea would support the vicariance hypothesis.Radiations of EcM fungi across continents are explained by co-migrations of both the fungus and the host partners (Halling et al., 2008;Pirozynski, 1983) and the capacity of the fungi to switch hosts (Halling, 2001).The possible long term co-existence of Neotropical Dipterocarpaceae with other EcM host families in the same region (including the EcM Fabaceae Dicymbe) might have favored this radiation scenario (Moyersoen, 2006).Dispersion assumes that EcM fungi are able to cross environmental barriers, usually by spore dispersal.Long distance dispersion, presumably by trans-Tasman airflow, was proposed to explain the disjunct distribution of Pisolithus (Moyersoen et al., 2003) and Inocybe (Matheny et al., 2009) species between Australia and New Zealand.The biotrophic status of EcM fungi is an important constraint for the movement of these fungi and the relative importance of long distance dispersion versus migration and vicariance in global EcM fungi distribution is under debate (Halling et al., 2008).The objective of this study was to evaluate the different hypotheses about the possible origin of EcM fungi associated with P. dipterocarpacea.For comparisons between P. dipterocarpacea EcM fungi and other tropical tree species in the same region or elsewhere, EcM fungi diversity and community structure was evaluated in a plot, where clumps of P. dipterocarpacea are present together with individuals of the EcM tree Aldina sp., Fabaceae, Papilionoïdeae.Inocybe was selected for further phylogenetic analysis, because of its global importance as an EcM fungal lineage, the general knowledge about its biogeography and its hypothesized paleotropical origin (Kuyper, 1986;Matheny et al., 2009;Ryberg, 2009).
A great diversity of EcM fungi, comparable to other tropical or temperate EcM rich communities, was found in this study.The EcM community structure was similar to that found in another forest dominated by EcM Fabaceae Dicymbe sp. in the same region.This study was the first evidence that Neotropical host tree species belonging to the Dipterocarpaceae and the EcM Fabaceae share EcM fungi both within the same forest and across forest stands.The floristic similarities and phylogenetic relationships of the EcM fungi between P. dipterocarpacea and Dicymbe forests suggested that fungal dispersion is an important radiation mechanism for EcM fungi in the region.Close phylogenetic relationships between Inocybe species associated with P. dipterocarpacea and African strains confirmed EcM phylogeographic links between the two continents.This study should be extended to examine a new P. dipterocarpacea dominated forest.The possible importance of dispersion in EcM fungal radiation and the link with paleotropical fungi should be tested in additional fungal lineages.

Study site
The sampling site was as described by Moyersoen (2006), located at 4º20'N, 61º48'W, altitude 500 m, near Icabarú Village, in Gran Sabana, Estado Bolivar, Venezuela, (Fig. 1).The precise location of the 20 X 20 m plot was selected based on a previous report of a stand of P. dipterocarpacea ssp.nitida described by Maguire & Steyermark (1981) in the same area (Fig. 2a).A second EcM tree, Aldina sp., also occurs in the plot (Moyersoen, 2006).

Fruit body and fine root sampling
All terricolous fruit bodies belonging to different morphotypes as well as resupinate fruit bodies growing under dead trunks and branches were collected inside the 20 X 20 m plot in 2007 and2008 (Fig. 2-b, c, d).A brief description from fresh material was made in the field for each collection and a piece of mycelium from the cap was taken and fixed in cetyltrimethylammonium bromide (CTAB) for further molecular study.Time lapse between collection and fixation was less than one day.Microscopic observations of 3% KOH rehydrated material was carried out using a light microscope (Leica DM 2500) .Fruit body collections were assigned to families or genus using general keys and assigned to morphospecies.Fungi were considered to be putatively EcM on the basis of identification (e.g.Rinaldi et al., 2008;Singer, 1986;Smith & Read, 2008).Fine roots were sampled in each sampling expedition.They were traced from identified P. dipterocarpacae trees in the first three surveys and sampled in nine 10 X 10 X 5 cm deep soil cores scattered in the plot in the last harvest.The soil cores were located either in places where P dipterocarpacea and Aldina sp trees occurred at < 2 m distance from each other (5 cores) or in random locations (4 cores).
Traced root samples were morphotyped using a field dissecting microscope (Novex AP-7) or a 10X magnifying glass before fixation in CTAB.Time lapse between tracing and root fixation was always up to one day.Morphology of traced P dipterocarpacea and Aldina sp.fresh secondary roots was characterized for subsequent sorting in the soil cores.Color was rust brown to yellowish and silvery creamish in P. dipterocarpacea and Aldina sp., respectively (Fig. 2-e, f).Frequent longitudinal peels of bark were often observed on P. dipterocarpacea roots.
A 5 X 5 X 5 cm 3 subsample including top organic soil was taken from each soil core and most fine roots attached to secondary roots as well as loose root tips were carefully washed in the field Lab.EcM and non-EcM root systems were separated using a 10X magnifying glass.EcM roots were subsequently separated on P. dipterocarpacea and Aldina root systems using color and morphological features.Loose EcM tips were grouped into a third "unclassified" category.All roots from the three categories were fixed in CTAB for further morpho-anatomical study in the laboratory in Belgium.Time lapse between root sampling and fixation was less than one day.
For each soil core, all fine roots belonging to P. dipterocarpacea, Aldina and "unclassified" categories were screened separately under a dissecting microscope to assign them to EcM morphological categories (Fig. 2-g, h).Each morphological category, including several tips, was cross checked using anatomical features on mantle peels following Agerer's (1991) method with some modifications.Different morphotypes from each P. dipterocarpacea, Aldina and "unclassified" root category were store d s e p a r a t e l y i n C T A B a n d k e p t a s a representative sample for further molecular analysis.

DNA protocols
Genomic DNA was extracted using QIAGEN Dneasy TM Plant Mini Kit (Qiagen S.A. Courtaboeuf, France) from fruit body morphospecies, traced ECM and morphotypes from the soil cores.To test the accuracy of morphotypes in the soil cores, DNA was extracted from several replicates (between two and 6) in EcM morphotypes including more than one sample.Fungal internal transcript spacer (ITS) and partial large subunit (LSU, 25-28S) nuclear rDNA were PCR-amplified with forward primer ITS1f in combination with LR6 (Gardes & Bruns, 1993;Vilgalys & Hester, 1990).If multiple or no PCR products were obtained, DNA was extracted from another EcM tip or the same extract was reamplified using the following primers in different combinations: ITS1f, ITS4b, ITS4, 5,8SR, LR21, LROR, LR6 (Gardes & Bruns, 1993;Vilgalys & Hester, 1990;White et al., 1990; R Vilgalys Lab http://www.biology.duke.edu/fungi/mycolab/primers.htm).The PCR protocols were as described in Moyersoen (2006) with some modifications in cycling parameters for different primers.To check host identity of traced roots and morphotypes from the soil cores, rbcL DNA was amplified from 7 EcM samples of P. dipterocarpacea and Aldina sp. using the primers rbcLN and rbcLR (Käss & Wink, 1997), following the same PCR protocols as in Moyersoen (2006).ITS-LSU and rbcL amplification products were electrophoresed in 1% agarose gels stained with ethidium bromide and visualized under UV light.1kb + DNA ladder (Invitrogen) was used as a marker.Controls with no DNA were included in every set of PCR amplifications.PCR products were purified using QIAquick protocol (Qiagen) or 96well filtration system (Multiscreen-PCR plate, Millipore Corporation, MA, USA).
Sequencing was performed by Genotranscriptomics Platform, GIGA, University of Liège , using the same primers as for PCR.Sequence editing was done using SEQUENCHER, version 4.0 (Gene Codes Corporation).ITS sequences of EcM specimens have been deposited at the National Center for Biotechnology Information (NCBI, GenBank: http://www.ncbi.nlm.nih.gov)under accession numbers JQ063044-JQ063063.Sequences of new fruit body species will be published separately.

Sequence analysis and phylogenetic analysis of Inocybe
ITS sequences were assigned to molecular species on the basis of arbitrary 3% similarity cutoff value (Nilsson et al., 2008).Sequence similarities were determined using the BLASTN sequence similarity tool (Altschul et al., 1997) in GenBank together with BLAST comparisons with recently published ITS sequences from a Dicymbe forest kindly supplied by M Smith (Smith et al., 2011).
A preliminary phylogenetic analysis was performed to place Inocybe sequences in Inocybaceae clades (Matheny et al., 2009).5' end nLSU sequences of two Inocybe species from P. dipterocarpacea forest were aligned using Clustal X (V.2.0.9) to Matheny et al. (2009) global Inocybaceae sequences data set kindly supplied by the first author.Maximum likelihood (ML) analysis was implemented in the program RAxML V. 7.0.6,using GTRCAT approximation.
Taxon sampling was made on the basis of this analysis.All strains in Inocybe subgenus Inocybe sequences and a strain from Nothocybe lineage were selected in Matheny et al. (2009) data base together with sequences from recently published tropical strains Y01 (UDB 004238), Y02 (UDB004239) (Tedersoo et al., 2010a) and I. tauensis (GU97711, GU977212, GU977213) (Kropp & Albee-Scott, 2010).In total, the data set included 77 RNA polymerase II (RPB1), 68 RPB2 and 114 LSU sequences.The RPB1, RPB2 and nLSU sequences were aligned using Clustal X.For consistency, the same criteria as Matheny et al. (2005Matheny et al. ( , 2009) ) were used for manual alignment using Bioedit V. Introns 1 of RPB1 and the intron of RPB2 were excluded from the analysis.Intron boundaries were inferred both by sequence comparison using published I. lilacina RPB1 intron 1 (AY351834), I. sindonia RPB1 intron 2 (AY351839) and by insertion between conserved amino acid and the canonical guanine-thymine and adenosine-guanine splice sites.Other positions, too ambiguous to align, were removed from the data set.RPB1 exon, RPB1 intron 2, RPB2 and nLSU were then concatenated in Bioedit.Data set partitioning was done following Matheny (2005).To test the phylogenetic relationships of Inocybe strains from P. dipterocarpacea forest with Inocybe data set, a ML Rapid Bootstrapping algorithm was implemented for 1000 replicates in RAxML, using GTRCAT approximation.

Above and below ground EcM richness
A total of 64 EcM fruitbody samples were collected in the plot (Table 2).From these samples, 41 specimens (including replicates) were selected for molecular analysis and 38 (93%) were successfully sequenced.These sequences belonged to 26 molecular species.Descriptions of new fruit body species will be published separately.
From 15 adult trees and the soil cores, 150 EcM samples were recovered (  From these samples, 64 EcM were subject to molecular analysis, resulting in 56 (83%) successful sequences and 22 species.Some preliminary categories were lumped together and 20 morphotypes were defined after cross molecular analysis.Only two morphotypes belonging to Cortinarius and Tomentella were accurate at genus level.The remaining morphotypes were accurate at the species level.Between one and 12 different morphotypes could be retrieved from a single 125 cm 3 soil core (Table 3).Species density per soil core was similar or above values reported in an highly diverse temperate Picea abies and Tilia cordata dominated EcM community where similar sampling strategy was used (Table 3).
Overall, 40 fungal species were recovered in the plot.Comparison with surveys in similar plot size in other tropical regions or in temperate regions showed that observed species richness was great and similar to or greater than values both in temperate and tropical areas (Table 4).
In total, 22 EcM fungal species were associated with P. dipterocarpacea; 11 (50 %) of these species belonging to 7 fungal groups were also putative or molecularly confirmed associates of Aldina, and three singletons were associated with Aldina only (Table 5).
Eighty percent (80%) of the best aligned sequences were from the tropical areas including both the Paleotropics and the Neotropics (

Inocybe EcM frequency and occurrence on host plants in the plot
Two Inocybe species were recovered in the plot.Inocybe were characteristic and could be distinguished from other fungal genera on the basis of morphology and anatomy.They were both characterized by a smooth and whitish mantle with frequent rhizomorphs and a pinkish color at the apex (Fig. 2a).Hyphae in the rhizomorphs were uniformly shaped and agglutinated (type B in Agerer 2006).EcM habit corresponded to medium-distance smooth exploration type in Agerer's (2001) EcM mycelial system.Typical Inocybe prominent clamp connections (Fig. 2b) were observed in the rhizomorphs.
Despite the low number of Inocybe species in the total EcM fungal diversity, Inocybe EcMs were frequent and were observed in 55 % of the soil cores.Two Inocybe species were recovered from traced roots of P. dipterocarpacea and the host identity was confirmed molecularly.They were also recovered from Aldina sp.roots in soil cores and Aldina rbcL could be sequenced for one Inocybe EcM species.

Phylogeny and biogeography of Inocybe species associated with P. dipterocarpacea
One Inocybe species from P. dipterocarpacea forest (BM03M2) matched (372/381, 98% similarity) Inocybe EcM1091 ITS1/5.8S/ITS2recovered from Aldina insignis and Dicymbe alstonii (Table 5).Phylogenetic analysis including LSU, together with fruit bodies' morphological comparisons, are necessary to confirm that EcM1091 and BM03M2 are conspecific.No significant match could be found for the second Inocybe species (Table 5).Best match with 25S LSU in GenBank was with I. epidendron (EU569840) for both species.Preliminary alignment of the two Inocybe species to Matheny et al. (2009) global Inocybaceae data set confirmed their position in Inocybe subgenus Inocybe (data not shown).These results were consistent with the characteristic Inocybe s. str.nodulose spores and pleurocystidia observed on Inocybe BM08C27 fruitbody species.LSU alignments with Matheny et al. (2009) global Inocybe subgenus Inocybe data set demonstrated that the two species were phylogenetically close to neotropical Inocybe species associated with Dicymbe and Aldina sp. in Guyana (Fig. 3).These species clustered within a tropical only, basal clade.A weakly supported node suggested a sister relationship with an African strain from Zambia.This clade was sister with a couple of related species from Zambia and Guyana.

EcM community diversity and structure in P. dipterocarpacea forest
This is the first above and below ground EcM fungal diversity survey in a Neotropical forest dominated by a Dipterocarpaceae.Only few pioneering surveys have been reported on EcM fungal communities in Neotropical monospecific Dipterocarpaceae genera Pseudomonotes and Pakaraimaea (Franco-Molano et al., 2005;Moyersoen, 2006;Vasco-Palacios et al., 2005).Moyersoen (2006) study mostly focused on P. dipterocarpacea EcM status and only a small proportion of EcM community was described.The present study was necessary to get a more precise EcM diversity estimate in Pakaraimaea forest.
Despite the relatively small plot size, a great number of EcM fungal species was recorded.Total above and below ground species richness and species density in soil cores was comparable to highly diverse EcM temperate forest communities.It has been suggested that EcM diversity in tropical forests is lower than in temperate forests (Tedersoo & Nara, 2010;Tedersoo et al., 2011).In contrast, with boreal forests, where a niche differentiation of EcM fungi is observed at the level of soil horizons (Rosling et al., 2003;Tedersoo et al., 2003), a study by Tedersoo et al. (2011) suggested that EcM communities are homogeneous across soil horizons in the African forests.Recent studies indicate that trends in EcM fungal biodiversity are not simply driven by a latitudinal gradient (Smith et al., 2011).Biodiversity depends on the forest type under study (Brearly, 2011;Smith et al., 2011) and is also influenced by host identity (Haug et al., 2005;Tedersoo et al., 2010a).Amongst the EcM surveys in Dipterocarpaceae vegetation (see Brearly, 2011 review and references therein), comparisons could be made with a SE Asian Mixed-dipterocarp forest (Peay et al., 2010), where diversity was measured in plot size identical to this study.EcM richness (Table 2) was at the top end of observed values (Table 4) in SE Asia and total above and below ground richness (Table 4) was within the range of estimated richness (9.3-60.4) in Mixed-dipterocarp forest.This comparison still needs to be taken with caution since root sampling strategy differed between the two studies.Contrary to the relatively low species density observed on Neotropical Pisonieae (Tedersoo et al., 2010a), density in P. dipterocarpacea was within the range or greater than in the temperate forests (Table 3).Both total fungal diversity and species density indicated that EcM fungi occur at a fine scale in P. dipterocarpacea forest such as in boreal forests.A greater EcM fungal diversity might be expected by sampling both organic and mineral soil layers in P. dipterocarpacea forest.This study extends the findings of a previous survey in a Neotropical Fabaceae forest where great levels of EcM fungi biodiversity were observed (Smith et al. 2011).EcM fungi in the P. dipterocarpacea forest belong to fungal orders generally observed in the tropical forests and elsewhere (e.g.Dickie & Moyersoen, 2008;Diédhiou et al., 2010;Lee et al., 2003;Peay et al., 2010;Riviere et al., 2007;Smith et al., 2011;Tedersoo et al., 2011).Species rich fungal orders included Agaricales, Thelephorales, Cantharellales, Boletales and Russulales.Particularly relevant was the number of Clavulina species.Many Clavulina species were also observed in Dicymbe forests and Smith et al. (2011) hypothesized that the Guayana region might be a centre of diversification for the Clavulina lineage.Cortinarius is usually considered to be rare in the tropical environments (Tedersoo et al., 2010b) and the great number of species identified in this study was striking.Interestingly, Peay et al. (2010)  white sands (Maguire & Steyermark, 1981).Whether preference for sandy soils is an important ecological character in these tropical Cortinarius species is worth investigating.
Tricholomataceae fruit body species in this study did not match any RPB2/LSU/ITS sequences either deposited in GenBank or from a recent Dicymbe forest EcM survey.This fruit body species is currently under further investigation and is potentially a new endemic species.
Cortinarius, Amanita and Russula are typical "late stage" fungi characteristic of undisturbed forest stands in the temperate forests (Deacon & Fleming, 1992;Nara et al., 2003).The great total diversity, together with the abundance of these late stage fungi and the presence of a possible novel endemic species, indicated that the plot is a well preserved P. dipterocapacea stand despite human activities in the region.This study also indicated that Neotropical Pakaraimaea and Dicymbe forests show similarities in EcM fungal community structure.

EcM fungal communities comparison between P. dipterocarpacea and other tropical tree host species
Until recently, biodiversity information on Neotropical lowland EcM fungi mostly relied on traditional fruit body surveys (e.g.Franco-Molano et al., 2005;Henkel et al., 2002Henkel et al., , 2005Henkel et al., , 2011;;Matheny et al., 2003;Moyersoen, 1993;Singer et al., 1983;Vasco-Palacios et al., 2005) with inherent difficulties to confirm the host plant identity.Relatively few EcM surveys have included molecular host identifications (Haug et al., 2004;Moyersoen, 2006;Smith et al., 2011;Tedersoo et al., 2010a).This study in P. dipterocarpacea forest provides the first report of multi-host EcM fungal species associated both with a Dipterocarpaceae and EcM Fabaceae in the Neotropics.
With the exception of Pisonieae and Gnetum sp.where patterns of host specificity were reported (Bechem & Alexander, 2011;Haug et al., 2005;Tedersoo et al., 2010a), recent tropical EcM biodiversity surveys have shown that co-occurring (sympatric) tree species including different tree families can be colonized by identical dominant EcM fungi (Diédhiou et al., 2010;Smith et al., 2011;Tedersoo et al., 2011).These sympatric tree families included the Dipterocarpaceae and Caesalpiniaceae in the Paleotropics (Tedersoo et al., 2007(Tedersoo et al., , 2011)).In addition to host overlap in the same forest, Tedersoo et al. (2011) also observed that some EcM fungal species are broadly distributed in tropical Africa.In contrast, host identity influences EcM fungal communities in the temperate forests (Ishida et al., 2007;Tedersoo et al., 2008).This study in P. dipterocarpacea forest indicated that several fungi from different fungal groups can colonise both P. dipterocarpacea and a sympatric Papilionoideae.Since EcM fractional colonisation and root length were not measured, sampling design did not allow to evaluate whether or not shared fungal species were true generalists (i.e. if they equally colonized the root system of both tree hosts).
The capacity of some EcM fungi to associate with Pakaraimaea and EcM Fabaceae was further demonstrated by the comparisons between Pakaraimaea forest fungal community and Dicymbe dominated forests.Around 15% of total species in P. dipterocarpacea forest (including undescribed Amanita sp.BM08C22, Boletellus ananas, Austroboletus rostrupii, Clavulina MCA4022 and two undescribed Tomentella species) were conspecific (≥ 99% ITS similarity) with specimens collected in Dicymbe forest 240 km apart from the study site.
Cantharellus guianensis might be an additional example of broadly distributed species but this hypothesis should be tested with complementary morphological and molecular study.Interestingly, C. guianensis was also described in P. tropenbosii dominated forest (Franco-Molano et al., 2005).Phylogenetics and morphological comparisons are necessary to confirm that the remaining 6 presumed shared species between P. dipterocarpacea and Dicymbe forests are conspecific.This study demonstrated that a proportion of EcM fungi associated with P. dipterocarpacea can also associate with sympatric and allopatric tree species belonging to EcM Fabaceae in the same region.The low proportion of Inocybe species in P. dipterocarpacea forest EcM fungal community was consistent with the general trend in the Neotropics.At least three different Inocybe lineages were observed in Dicymbe forests (Matheny et al., 2003;Matheny et al., 2009).P. dipterocaracea survey needs to be extended to test whether this tree species also associates with several Inocybe lineages.

Inocybe ecology and phylogeography
Despite the few species recorded, the frequent occurrence of Inocybe EcM suggested that this fungal group is successful in P. dipterocarpacea plot.An interesting feature of morphotypes observed on two P. dipterocarpacea Inocybe species was the presence of rhizomorphs with agglutinated hyphae and ramified distal end in close contact with soil particles.A small proportion of Inocybe EcM morphotypes have been formally described (Agerer, 2006) and only one of them (I.avellana in association with Shorea, Dipterocarpaceae) (Ingleby, 1999) was tropical.One of the patterns in described morphotypes was the presence of emanating hyphae only as soil exploration structures (Agerer, 2006).Ryberg et al. (2010) demonstrated that soil preference reflects the evolutionary history of Inocybaceae.Preference for poor soil prevail in some phylogenetic groups.Agerer (2001)  In contrast with the hypothesis that Inocybe species associated with P. dipterocarpacea represent an isolated immigration event (Matheny et al., 2009), both BLASTN and phylogenetic analysis demonstrated that P. dipterocarpacea shared identical or phylogenetically close Inocybe species with EcM Fabaceae.The capacity of two Inocybe species to associate with both P. dipterocarpacea and co-occurring Aldina species B is consistent with the host generalist status of Inocybaceae within angiosperms in the temperate forests (Ryberg et al., 2010;Kuyper, 1986).This host sharing ability was demonstrated for several Inocybe species on Dicymbe and co-occurring Aldina in the Guayana region (Smith et al., 2011) The broad distribution of phylogenetically close Inocybe species might be the result of an efficient dispersion ability, the capacity to associate with different tree hosts and a long evolutionary history in the Guayana Region.A possible explanation for the disjunct distribution of Inocybe related species in the region is that extant forests with a dominance or codominance of one or several Neotropical EcM tree families including Dipterocarpaceae (Franco-Molano et al., 2005;Moyersoen, 2006;Vasco Palacios et al., 2005) and EcM Fabaceae (Henkel et al., 2002;Singer & Araujo, 1979, 1986) are remnants of more extensive EcM forests in the past.This hypothesis is difficult to evaluate since knowledge on Guayanan Region vegetation history is scanty (O Huber, pers.com).Inocybe is often mentioned in the literature as a pioneer and an effective coloniser by spore dispersal (Smith and Read, 2008).Although they are often scattered in AM dominated forests (Béreau et al., 1997;Lodge, 1987;Singer & Araujo, 1979;St John, 1980;Tedersoo et al., 2010a), species belonging to confirmed EcM tree hosts are diverse and widely distributed in the Guayana region (Table 1).These nondominant Neotropical EcM plants might have contributed to the dispersion of EcM fungal groups such as Inocybe subgenus Inocybe by co-migration (Halling, 2001) or by providing "EcM tree islands" for fungus colonization (Peay et al., 2007).Inocybe species belonging to different clades in Inocybe subgenus Inocybe (data not shown) have been found on Guapira and Neea sp. from Ecuador, Belize and US Virgin Islands (Matheny et al., 2009;Tedersoo et al., 2010a) as well as on Coccoloba species in Ecuador (Tedersoo et al., 2010a).Fruit bodies belonging to 8 EcM lineages including Inocybaceae were reported by Moyersoen (1993) in Bana vegetation, Southern Venezuela, where Pisonieae (Neea and Guapira) are particularly conspicuous together with A. kunhardtiana and C. excelsa.Additional surveys are necessary to test the importance of non-dominant EcM tree hosts on EcM fungi dispersion and diversity in Guayana Region.Phylogeographic analyses similar to this Inocybe study should also be undertaken on more diverse fungal lineages such as Clavulina that are associated with different confirmed tree hosts from different families across the Guayana region (Henkel et al., 2005(Henkel et al., , 2011;;Smith et al. 2011).
P. dipterocarpacea Inocybe clustering within a basal clade found only in the tropics demonstrated their tropical origin.The results are consistent with Matheny et al. (2009) in the species composition of the Inocybe subgenus Inocybe basal clade (tree node 84).These basal species were not included in the phylogenetic study by Kropp and Albee-Scott (2010), who selected a subset of Neotropical and Paleotropical Inocybe strains.According to molecular clock phylogeny, the estimated age of tree node 84 was 65 Myr (41-89 Myr, 95% CI) (Matheny et al. 2009).Inocybe species associated with P. dipterocarpacea are the recorded oldest origin on a Dipterocarpaceae.The present study highlights the need to get a more exhaustive Dipterocarpaceae Inocybe ITS/LSU/RPB1/RPB2 sequence data set across the tropics.Comparisons of EcM flora between P. dipterocarpacea and P. tropenbosii as well as Dipterocarpaceae from Africa or the related EcM family Sarcolanaceae (Ducousso et al., 2004) from Madagascar would be particularly appropriate.Basal clade topology fits with a vicariance hypothesis, but Matheny et al. (2009) excluded this hypothesis on the basis of an estimated divergence date.An alternative boreotropical migration route (Pennington & Dick, 2004) was proposed (Matheny et al., 2009) but this hypothesis is not reflected in the topology of Inocybe subgenus Inocybe phylogeny.Long distance dispersal between Africa and South America would be another alternative hypothesis.There is no clear cut biogeographic interpretation for the basal clade including Paleotropical and Neotropical Inocybe subgenus Inocybe strains.Observation of two of these basal species on P. dipterocarpacea, a tree host with possible Gondwanic ancestors, is intriguing.

Conclusion
This study demonstrated the importance of fungal dispersion and host sharing in understanding of the EcM community associated with P. dipterocarpacea.Host sharing between sympatric tree species belonging to different lineages within Fabaceae had already been described in a Dicymbe dominated forest (Smith et al. 2011).This study highlighted that EcM fungal species can associate with allopatric tree hosts belonging to two typical EcM tree families in the Guayana region, Dipterocarpaceae and EcM Fabaceae.There is a need to extend Inocybe phylogeographic study to other fungal groups and to include non-dominant tree species including Pisoniae and Coccolobeae in the surveys in Guayana Region.These studies are necessary to evaluate the importance of host overlap in the area and the possible importance of non-dominant host tree species in EcM fungal dispersion and diversity.
The hypothesized origin of Inocybaceae is Paleotropical (Matheny et al., 2009).This study confirmed that forests dominated by typical EcM tree families host EcM rich fungal communities in the Guayana Region.EcM fungal species density was great and further studies are needed to evaluate the effect of soil substrate on EcM fungal communities.
The presence of a possible tropical endemic Tricholomataceae species in the P. dipterocarpaceae forest suggests that EcM flora could include old relictual species, therefore confirming the great biodiversity value of EcM flora in the Guayana Region (Smith et al., 2011).
Fig. 1.Location of the field site
observed a greater number of Cortinarius in Mixed-dipterocarp forests on sandy soils.P. dipterocarpacea typically grows on (a) (b)
Matheny et al. (2009) suggested that Inocybe flora in the Neotropics resulted from multiple recent immigration events in South America.The lower Inocybe diversity in the Neotropics than in the Paleotropics was attributed (apart from sampling effort) to a combination of factors including competition or extinction and paucity of EcM tree hosts.
The old origin of Inocybe strains associated both with P. dipterocarpacea and EcM Fabaceae and their tropical root suggest a long evolutionary history of EcM symbiosis in the Neotropics.Present knowledge of Inocybaceae molecular clock phylogeny is insufficient to say with certainty which EcM radiation scenario could explain the disjunct distribution of the basal clade of Inocybe subgenus Inocybe.A more complete ITS/LSU/RPB1/RPB2 sequence data set including Inocybe strains associated with Dipterocarpaceae across the tropics is needed to improve the understanding of Dipterocarpaceae associated Inocybe phylogeography.

Table 2
).A total of 113 EcM samples were selected for molecular analysis and 97 EcM tips (86%) were successfully amplified and sequenced.These sequences belonged to 27 EcM fungal species.Only 12 of these species matched DNA from fruitbody surveys.*Including ITS1, 5.8S, ITS2, LSU DNA regions depending on species.**Species were defined using 97% sequence similarity cutoff across the ITS1, 5.8S, ITS2 region as well as phylogenetic analysis of LSU on selected fungal groups.***This figure does not include one unsequenced Russula and Scleroderma species.

Table 2 .
Observed fruitbody and EcM species in 400m 2 plot dominated by Pakaraimaea dipterocarpacea morphotypes accuracy was tested before further measurement of species density.A total of 96 EcM samples from soil cores were classified in 28 preliminary morphological categories.

Table 5
(Smith et al. 2011)cies matched (≥ 97% similarity) the published ITS sequences.These species belonged to Amanita, Inocybe, Boletaceae, Clavulina, Sebacinales and Tomentella.All but one of these species were reported from a Dicymbe dominated forest situated ca 240 kms apart in the Guayana region (Guyana, Upper Potaro River Basin).Confirmed tree hosts included Dicymbe and Aldina species(Smith et al. 2011).

Table 3
. Comparison of observed EcM species density between Pakaraimaea dipterocarpacea forest and recent tropical and temperate surveys *mixed fruitbody and EcM survey.

Table 4 .
Observed EcM fungi richness in similar sized tropical and temperate forest plots Host for each fungal OTU was identified from EcM after rbcL sequencing°, root tracing* or root morphology δ .Fungal OTUs with no identified host were collected only from the fruiting bodies.Pd : P.

Table 5 .
Sequence matches of ectomycorrhizas and ectomycorrhizal fruit bodies collected in Pakaraimaea dipterocarpacea forest Peay et al. (2011)ssible ecological importance of rhizomorphs for soil exploration and plant nutrition.Peay et al. (2011)also suggested that differences in EcM exploration type influence EcM fungal dispersion ability.Rhizomorphs might contribute to the success of Inocybe in poor sandy soils associated with P. dipterocarpacea forest.Whether this morphological character is a conserved morphological characteristic in Inocybe subgenus Inocybe basal clade is worth investigating.
(Matheny et al., 2003)Inocybe species in P. dipterocarpacea forest were phylogenetically close to species in Dicymbe dominated forest.Singer et al. (1983)described I. amazoniensis from Campinarana vegetation near Manaus, Northern Brazil.They did not identify I. amazoniensis host plant and putative EcM hosts in Campinarana include EcM Fabacaeae (Aldina), Pisonieae and Coccolobeae.Confirmation of I. amazoniensis phylogenetic position which is morphologically close to I. epidendron(Matheny et al., 2003)and Inocybe sp.BM08C27 (unpublished data) would further support the evidence that several closely related Inocybe species associate with diverse tree hosts including Dipterocarpaceae and EcM Fabaceae across the Guayana region.