• Leccinum;
  • host specificity;
  • glyceraldehyde 3-phosphate dehydrogenase (Gapdh);
  • internal transcribed spacer (ITS);
  • specialist;
  • generalist;
  • speciation


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • • 
    Species of the ectomycorrhizal genus Leccinum are generally considered to be host specialists. We determined the phylogenetic relationships between species of Leccinum from Europe and North America based on second internal transcribed spacer (ITS2) and glyceraldehyde 3-phosphate dehydrogenase (Gapdh).
  • • 
    We plotted host associations onto the phylogenies using maximum likelihood and parsimony approaches.
  • • 
    Resolution of the phylogeny was greater with Gapdh vs ITS2, plus the Gapdh and ITS phylogenies were highly incongruent. In Leccinum the coding region of Gapdh evolved clocklike, allowing the application of a molecular clock for the reconstruction of host specificity. Almost all species of Leccinum are highly host tree specific, except Leccinum aurantiacum, which associates with a broad range of host trees. Maximum likelihood reconstructions of the ancestral host associations show that this taxon evolved from a specialist.
  • • 
    Our results indicate episodes of rapid speciation coinciding with or immediately following host switches. We propose a model where host niche contraction through geographic isolation and host niche expansion through ecologically equivalent hosts drive cycles of speciation. The role of host race formation and incipient speciation is discussed.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Mycorrhizal fungi and, to a lesser extent, mycorrhizal plants, display different degrees of host specificity. What the evolutionary advantage of specialization is for either of the symbiotic partners is still unclear. Bruns et al. (2002) hypothesized the advantage for the fungus of specializing on a (phylogenetically) narrow range of hosts is found in a greater physiological compatibility of the fungus to its host. This increased physiological compatibility would then enable a specialist fungus to obtain more carbohydrates from the plant host than generalist competitors do. The advantage for the plant host to associate with specialist mycorrhizal fungi is less clear, because the costs of associating with a specialist are greater than that of associating with a generalist. From the plant's perspective, specialization may lead to decreased functional compatibility. Finlay (1989) reported that Suillus grevillei and Suillus cavipes– two associates with larch – were able to form ectomycorrhizas with pine, but hardly any nutrients were transferred to the plant host. Alternatively, associations with specialized fungi would reduce the chances of indirectly helping competing plant species (Molina et al., 1992), as generalist fungi can connect individuals of hosts from the same or different species and are able to translocate carbon between hosts (Simard et al., 1997). However, the significance of carbon translocation by generalist fungi is unclear. Robinson & Fitter (1999) suggested that in the case of AM the carbon stays in the generalist fungus, instead of being translocated from host to host, while in the case of ectomycorrhiza (EM) the evidence is still equivocal.

While the processes that select for or against specialization in EM symbiosis are still unknown, the fact that EM fungi display different levels of specialization is well known. Some species of EM fungi are associated with a phylogenetically broad range of hosts (generalists), such as Amanita muscaria (Trappe, 1962), while others are specialized to a phylogenetically narrow range of hosts, for example species of the genera Rhizopogon or Suillus, which are almost exclusively associated with Pinaceae (Massicotte et al., 1994; Molina & Trappe, 1994; Kretzer et al., 1996). The evolutionary history of specificity of EM fungi towards their plant host has received little attention, despite host specificity being, without doubt, a key element in understanding the present day distribution and diversity of EM fungi.

Leccinum S.F. Gray (Boletaceae, Boletales) is a genus of ectomycorrhizal fungi associated with a wide range of hosts (Table 1). The genus occurs mainly in the temperate and boreal regions of the northern hemisphere with some secondary expansion to the neotropics (Halling & Mueller, 2003). Reports of species of Leccinum occurring in Africa (Heinemann, 1964) probably refer to species that are better classified in the genus Tylopilus. Their relatively large size and distinctive appearance make their gross distribution well known. Most species of this genus are considered specialists (Singer, 1986). Consequently, host association is used as a distinctive character in keys. In addition to the fact that it is not always easy to determine which host plant the fruitbody is associated with (especially if more than one possible candidate host is present), the possibility that species are generalists and associated with more than one host species seems to be ruled out. (In this paper we will deal with the European species L. aurantiacum a generalist according to Den Bakker and Noordeloos (unpubl. data); Table 2.) Other authors consider this species to consist out of two specialists, Leccinum quercinum, when associated with Fagaceae and Leccinum populinum, when associated with Populus (Korhonen, 1995). A molecular phylogeny allows an evaluation of the status of these species and an investigation of the evolutionary history of host specificity from the fungal perspective.

Table 1.  Host specificity and general distribution on the northern hemisphere of the main clades found in phylogenetic research of Binder & Besl (2000) and Den Bakker et al. (2004)
Luteoscabrum Mainly Fagaceae and Betulaceae (subfamily Coryloideae)Temperate and subtropical regions
LeccinumLeccinum1. Salicaceae (Populus), Betulaceae (Betula) rarely FagaceaeTemperate and sub-boreal regions
2. Ericaceae (subfamily Arbutoideae)California and Costa Rica
ScabraBetulaceae (Betula)Mainly subboreal regions
Table 2.  Taxonomic changes of European taxa according to Den Bakker and Noordeloos (unpubl.)
New namesSynonymsHost range
  1. * Lannoy & Estades (1995), Korhonen (1995).

Leccinum leucopodiumLeccinum aurantiacum sensu PilatPopulus
Leccinum aurantiacumLeccinum quercinumPopulus, Betula, Quercus (Fagus, Castanea and Tillia reported*)
Leccinum populinum 
Leccinum versipelleLeccinum cerinumBetula, occasionally Arctostaphylos?
Leccinum callitrichum 
Leccinum roseotinctum 

Previous phylogenetic studies (Binder & Besl, 2000; Den Bakker et al., 2004) with conventional nuclear ribosomal markers (28S, ITS) have left the relationships within the sections Scabra/Leccinum clade relatively unresolved. In this study, sequences of the second internal transcribed spacer (ITS2) will be used in combination with about 1200 bp of the single copy nuclear gene glyceraldehyde 3-phosphate dehydrogenase (Gapdh, EC, in the hope of obtaining better resolution. The latter gene has proven to be phylogenetically informative for various groups of Ascomycetes (Berbee et al., 1999; Yun et al., 1999; Câmara et al., 2002), especially for studies at the species level. The gene has so far not been used for phylogenetic studies in Basidiomycetes.

In this paper a molecular phylogenetic study of a representative sample of species of sections Leccinum and Scabra is provided. These phylogenies will be used to assess the level of host specificity of the individual lineages/species found and to reconstruct the evolutionary history of host specificity.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Taxon sampling

Data of all collections used in this study are summarized in Table 3. A priori designation (through identification by the first author) of the collections to morphospecies and nomenclature are according to Den Bakker & Noordeloos (unpubl. data) for the European material and according to Smith and Thiers (1971), Thiers (1975) and Halling and Mueller (2003) for the American material. In some cases fresh material was placed in cetyltrimethylammonium bromide (CTAB) (100 mm Tris-Cl, 1.4 m NaCl, 20 mm ethylenediaminetetraacetic acid (EDTA), 2% CTAB, pH 8.0) in the field for further processing, otherwise dried herbarium material was used. We have sampled all known host associations within sections Leccinum and Scabra. Care was also taken to obtain (if possible) both American and European representatives of a known host association. Voucher specimens are deposited in L, GENT, H, O, P, NY and SFSU (herbarium abbreviations according to Holmgren et al., 1990).

Table 3.  Samples used in analyses, including voucher number, geographic origin, host and GenBank accession numbers. Numbers behind the geographical origin of some of the accessions refer to numbers used in Figs 2 and 3
Species designationVoucher collectionGeographical originHostGenBank accession number
Leccinum crocipodiumrw1659Sommauthe/Beaumont-en-Argonne, Ardennes, FranceCarpinusAF454589AY538783
Leccinum carpinihdb065Breukelen, Utrecht, The NetherlandsCarpinus/CorylusAF454588AY538785
Leccinum talamancaehalling8001San Gerardo, Dota, San José, Costa RicaQuercusAY544779AY538783
Fumosa clade
Leccinum duriusculumwtoo1Wassenaar, Zuid Holland, The NetherlandsPopulusAF454576AY538787
Leccinum nigellum4676PVibraye, FrancePopulus AY538815
Leccinum uliginosumhdb330Whitefish Falls, Ontario, CanadaPopulusAY538825AY538786
L. aurantiacum sensu lato
L. aurantiacumvan BrummelenForet de Belleme, Orne, France (2)BetulaAY538853 
L. aurantiacumvan BrummelenIge, Orne, France (3)PopulusAY538857AY538823
L. aurantiacumhdb003AWD, Noord Holland, The Netherlands, (3)PopulusAY538856 
L. aurantiacumHills2001219Windsor Great Park, Berkshire, EnglandBetulaAY538854AY538819
L. leucopodiumrw1656Sommauthe/Beaumont-en-Argonne, Ardennes, France?AF454469AY538795
L. leucopodiumhdb93Sogndal, Sogn og Fjordane, NorwayPopulus AY538817
L. sp. 1halling6580Twobridge Swamp, Franklin County, NY, USAPopulusAY538836 
L. sp. 2tdb304USA?AY538835 
L. sp. 3arora 00–53Along Dempster Highway, Yukon Territory, CanadaPopulusAY538841AY538824
L. sp. 4hdb317Manitoulin Island, Ontario, CanadaPopulus AY538821
L. brunneumhdt49122Cascade, Valley County, ID, USAPopulusAY538850 
L. insignehdb320Manitoulin Island, Ontario, CanadaPopulusAY538851AY538822
L. insignehdt50455Vicinity North Adams, MA, USABetulaAY538842 
L. aurantiacummk11850Vantaa, Nylandia, FinlandPopulusAY538861AY538797
L. aurantiacumhdb94Sogndal, Sogn og Fjordane, NorwayPopulusAY538860AY538817
L. aurantiacumhdb286Leusden, Gelderland, The Netherlands (2)QuercusAY538855 
L. aurantiacumvan BrummelenForet de Cessey, Doubs, France (1)QuercusAY538852AY538796
L. aurantiacumrw1683Oignies-enThiérarche, BelgiumQuercusAY538859 
L. aurantiacumhdb102Roden, Drenthe, The Netherlands (1)QuercusAY538858AY538816
L. versipelle sensu lato
L. atrostipitatum 1halling3131Togue Pond Road, Piscataquis County, ME, USABetula/PopulusAY538834 
L. atrostipitatum 2halling3081Baxter State Park, Piscataquis County, ME, USABetulaAY538833 
L. atrostipitatum27-8-84/3Nouveau Quebec, Quebec, CanadaBetulaAY538832AY538802
L. versipelle2270PAumont-Aubrac, Lozère, FranceBetulaAY538829AY538818
L. versipellehdb070Kall, Jämtland, Sweden (1)BetulaAY538827AY538801
L. versipellehdb285Leusderheide, Gelderland, The NetherlandsBetulaAY538831AY538799
L. versipelleOF64036Lærdal, Sogn og Fjordane, NorwayArctostaphylosAF454574AY538798
L. versipellemen95702Utsjoki, Inarilapland, Finland (1)BetulaAY538828AY538800
L. versipellemk11452Kilpisjarvi, Enontekio Lappi, Finland (2)BetulaAY538826AY538802
L. versipellehdb74Kall, Jämtland, Sweden (2)BetulaAF454575 
L. versipellehdb57Borgsjö, Jämtland, Sweden (3)BetulaAY538830 
Pinaceae associates
L. piceinumMEN2048Obertiliach, Lienz, AustriaPiceaAF454579AY538794
L. vulpinumhdb92Sogndal, Sogn og Fjordane, NorwayPinusAF454580AY538792
Ericaceae associates
L. arbuticolaarora 00–293Boonville, Mendocino County, CA, USAArbutusAY538837AY538789
L. manzanitaeLG464Santa Cruz Island, CA, USA,ArctostaphylosAY538838AY538789
L. manzanitaeEcv2404California, USAArctostaphylos AY538790
L. monticolahalling8288Cerro de la Muerte, Dota, San José, Costa RicaComarostaphylosAY538839AY538788
L. monticolahalling8325Costa RicaComarostaphylosAY538840AY538820
Section Scabra
Leccinum scabrumhdb048Hoogeveen, Drenthe, The NetherlandsBetulaAF454585AY538813
L. scabrumhdb301Midhurst, Ontario, CanadaBetulaAY538849AY538814
Leccinum holopushdb329Manitoulin Island, Ontario, CanadaBetulaAY538844AY538808
Leccinum holopushdb40Nieuwkoop, Zuid Holland, The NetherlandsBetulaAF454561AY538807
Leccinum brunneogriseolumhdb39Schiermonnikoog, Friesland, The NetherlandsBetulaAF454560AY538806
Leccinum cf snelliihalling6914Indian Creek, Swain County, NC, USABetulaAY538845AY538811
Leccinum cf snelliihalling4472Raquette Lake, Hamilton County, NY, USABetulaAY538846AY538812
Leccinum schistophilumMK11145Vantaa, Nylandia, FinlandBetulaAY538847AY538809
Leccinum schistophilumhdb121Orne, Foret Dominial du Perche, FranceBetulaAY538848AY538810
Leccinum variicolorhdb051Erica, Drenthe, The NetherlandsBetulaAF454572AY538804
Leccinum snelliihdb327Manitoulin Island, Ontario, CanadaBetulaAY538843AY538805

Host designation

In most cases the labels that accompanied the herbarium material noted the host(s). In two cases no host tree species information was provided. In one case two potential hosts were indicated. In one case the host could not be designated unambiguously in the field. Here, ectomycorrhizal root tips were collected under the fruit body. The host was then identified by means of DNA sequencing. The DNA was extracted from the root tips of the presumed host with the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany), following the protocol supplied by the manufacturer. The EM was identified using Gapdh primers (see below) and compared with the above-ground fruit body. The identity of the root tips was determined using the plastid trnL–trnF sequence. Amplification of this region used the primers TabE and TabF (Taberlet et al., 1991). For this PCR we used the same conditions as used for the amplification of Gapdh (see below). We did a blast (GenBank) search on the plastid sequence to compare it with known sequences.

DNA extraction, polymerase chain reaction (PCR) and sequencing of fungal material

The DNA of a small number of accessions was extracted by means of a modified CTAB procedure, as described by Den Bakker et al. (2004). DNA of all other accessions was obtained from either CTAB-preserved or herbarium material using the DNeasy Plant Mini Kit (Qiagen) following the protocol supplied by the manufacturer.

Internal transcribed spacer 2 was amplified using the primers ITS3 and ITS4 (White et al., 1990). The PCR reactions for amplification of ITS2 followed Den Bakker et al. (2004). Initially, a small section of Gapdh (c. 400 bp) was amplified by using the primers GPD0623F (all the primer sequences used for amplification of Gapdh are listed in Table 4, relative positions in Fig. 1) and GPD1035R (designed by Rasmus Kjøller, University of Copenhagen, Denmark). Because amplification failed for a number of species we designed an alternative forward primer, GPDlecF. With this primer, its reverse complement GPDlecR and two general primers GPDforward and GPDreverse (designed by slight modification of the primers published by Kreuzinger et al., 1996) we managed to amplify c. 1100 bp of the Gapdh gene in two pieces: a c. 600 bp piece (using primers GPDforward and GPDlecR) and a 500 bp piece (using primers GPDlecF and GPDreverse).

Table 4.  Primer sequences used for GPD, and corresponding primer position in Fig. 1
Primer name5′−3′Fig. 1
  1. Data from R. Kjøller (pers comm.)

GPD forward generalcgg ccg tat cgt cct ccg taa tgc1
GPD reverse generalgag ta(at) cc(gc) cat tcg tta tcg tac c2
Primer internal forward GPD Leccinumcga agg tct cat gag cac tat cca5
Primer internal reverse GPD Leccinumtgg ata gtg ctc atg aga cct tcg6
GPD0623F*ttg cca agg tcg tca acg3
GPD1035R*gtg taa gca acg ata ccc ttc ag4

Figure 1. Primer positions of Gapdh. Numbers refer to the primer sequences given in Table 4. Dotted areas indicate the position of introns.

Download figure to PowerPoint

To amplify the desired regions we used 2 µl of genomic DNA in a 25-µl reaction mixture. The mixture contained 1× PCR buffer (Qiagen), 2.5 nmol dNTPs, 4 pmol of both the forward and reverse primer, 2 ng bovine serum albumin (BSA), 3.75 mm MgCl2, and 1 unit Taq polymerase. Cycling parameters were: initial denaturing at 95°C for 2 min followed by 34 cycles of 30 s at 95°C, 30 s at 54°C and 30 s at 72°C, with a final extension of 2 min at 72°C. The PCR products were electrophoresed in a 1.25% agarose gel in 1× Tris-borate-ethylenediaminetetraacetic acid (TBE) (pH 8.3) buffer, stained with ethidium bromide to confirm a single product and cleaned following the Qiaquick PCR Cleanup protocol (Qiagen). In cases where multiple bands were encountered, PCR products of the right length were extracted from the agarose gel following the QIAquick Gel Extraction Kit (Qiagen).

The purified PCR products were directly sequenced using the amplification primers. Samples were sequenced on an ABI 377 automated sequencer (Applied Biosystems, Foster City, CA, USA) using standard dye-terminator chemistry following the manufacturer's protocols.

Phylogenetic analyses

The Gapdh sequences were aligned with clustal x (Thompson et al., 1997) and refined by eye. The ITS2 sequences were aligned with the online version of Partial Order Alignment (Lee et al., 2002, and subsequently refined by eye. Large sections of the ITS2 sequences of Leccinum talamancae, Leccinum crocipodium and Leccinum carpini could not be aligned with confidence to the ingroup taxa and were left out of the alignment.

Maximum parsimony (MP) and maximum likelihood (ML) analyses were conducted using paup*4.0b10 (Swofford, 2002). In all analyses gaps were treated as missing data. MP and ML phylogenies were obtained using the heuristic search option, 10 random sequence additions and tree bisection and reconnection (TBR) branch swapping. Maxtrees was set to 20 000 trees. In the MP analyses characters were treated as unordered and unweighted. For the ML analyses the program modeltest version 3.06 (Posada & Crandall, 1998) was used to find the model of sequence evolution least rejected given the data set. The model and its parameters were chosen based on the outcomes of a hierarchical likelihood ratio test (α = 0.01) as implemented in the software. Initially, Boletus edulis s.l., Boletus subglabripes and Tylopilus chromapes (considered by some authors as Leccinum chromapes) were used as outgroups. The use of these outgroups significantly lowered the resolution of the topology of the ingroup and ITS2 sequences and intron regions of Gapdh were hard to align without ambiguity. Analyses of Gapdh with these outgroups, however, showed the Costa Rican endemic L. talamancae to have a well-supported sister group relationship with the other accessions of Leccinum. Leccinum talamancae was therefore used as the outgroup for the MP, ML and Bayesian analyses presented in this paper.

Bayesian and bootstrap analyses

Bayesian analyses were performed using mrbayes v3.0b4 (Huelsenbeck & Ronquist, 2001) In order to perform a Bayesian analysis of the Gapdh data set the data were divided into eight partitions: The coding region was divided in three partitions representing the different coding positions. The noncoding region consisted of five introns, each treated as a separate partition. The program mrmodeltest (J.J.A. Nylander, available from the internet: was used to select (based on the implemented hierarchical likelihood ratio test (α = 0.01)) the least rejected model of sequence evolution for each individual partition. Likelihood and prior settings were changed in mrbayes to meet with the settings necessary to apply the models found for each partition. The analysis was initiated with a random starting tree and was run for 5 × 106 generations, keeping one tree every 1000 generations. The first 106 generations (burn-in) were discarded and the remaining 4000 trees (representing 4 × 106 generations) were used to calculate a 50% majority rule tree and to determine the posterior probabilities for the individual branches. The ITS2 data set was not partitioned. mrmodeltest was used to find the least rejected model of sequence evolution and likelihood and prior settings were changed according to the model found. The ITS2 analysis was conducted under the same settings as the Gapdh set. In order to check whether both analyses converged to the same optimum, we repeated the analyses several times with 1 × 106 generations.

Nonparametric bootstrapping (Felsenstein, 1985) was performed to determine the levels of support for the internal nodes. We performed 1000 bootstrap replicates. The MP parameters were the same as in the heuristic search except the branch swapping option was set to search for 10 s for each replicate and the sequence addition procedure was set to simple.

Molecular clock analysis

To test if the Gapdh sequences in Leccinum evolve clockwise, we used a likelihood ratio test to test for rate constant evolution (Huelsenbeck & Rannala, 1997). This likelihood ratio test determines whether there are significant differences between the likelihood scores of trees where the branch lengths are unconstrained compared with a tree with the same topology where the branch lengths are constrained so that the terminal ends are contemporaneous. beast version 1.0.3 (Drummond & Rambaut, 2003) was used to calculate the posterior probabilities of the clades found when a molecular clock could be assumed.

Compatibility tests and topology tests

The compatibility of the different datasets was tested a priori with the partition homogeneity test (Farris et al., 1995) as implemented in paup*. A total of 10 000 replicates were performed and maxtrees was set to 100. In order to test if the topologies of the different analysis and the different datasets were significantly different we used the likelihood based Shimodaira–Hasegawa (SH) test as implemented in paup*, using the RELL option and 10 000 bootstrap replicates to calculate the test distribution. This test is more robust to violations of the model of sequence evolution than other likelihood-based topology tests (Buckley, 2002).

Reconstruction of the evolution of host associations

To trace the history of host associations we used a likelihood reconstruction method (the stochchar package, Maddison & Maddison, 2003b) as implemented in mesquite version 0.966 (Maddison& Maddison, 2003a). The one-parameter Markov k-state model (Lewis, 2001) was chosen to estimate the ancestral states, using the default settings. Differences in likelihood of two possible ancestral states were considered significant when they exceeded a cut-off point of two log units (Pagel, 1999). The different host associations were coded as one multistate character. A well-resolved and well-supported topology was chosen to trace the history of host associations. Additional to the likelihood reconstruction a parsimony-based reconstruction was performed as implemented in macclade 4.0.5 (Maddison & Maddison, 2002).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Host designation by molecular methods

One collection Leccinum sp. 4 from Ontario, Canada was found near a Pinus banksiana tree. A blast search of the trnL-F sequence from root tips collected under the fruit body and colonized by the mycorrhiza of that species, showed a close match to Balanophoraceae, a family that belongs to the Malphigiales. The genus Populus (Salicaceae) also belongs to this order. Populus trees were present in the area and we concluded these must have been the host trees and not a pine.

Gapdh phylogeny

For 26 accessions both the first c. 600 bp and the second c. 500 bp of the Gapdh gene were sequenced. For one accession only the first 600 bp was sequenced, for 14 other accessions only the second 500 bp. The position of the five introns was congruent with that of B. edulis, as shown by Kreuzinger et al. (1996). The data set comprised 41 accessions, 1160 characters and 213 potentially phylogenetically informative characters.

Using modeltest, the general time-reversible model was chosen for the ML analysis, with variable sites assumed to follow a gamma distribution (shape set to 0.5222), nucleotide frequencies set to A 0.2417, C 0.2661, G 0.2260, T 0.2662 and substitution rates set to 1 (AC), 2.7316 (AG), 1 (AT), 1 (CG), and 4.5152 (CT). The models used for the individual partitions in the Bayesian analysis can be found in Table 5. Trees obtained by MP (> 20 000 MP trees, 500 steps, CI = 0.796, RI = 0.895), maximum likelihood (three trees, –Ln L 4452.98) and Bayesian analyses of the Gapdh data did not differ significantly from each other. The Bayesian inference topology is depicted in Fig. 2 and shows that Leccinum can be subdivided into four very well supported groups. (1) L. carpini and L. crocipodium (clade H) show a well-supported sister group relation with the rest of Leccinum examined. The remaining accessions form three well to moderately supported clades: (2) a clade (the Scabra clade) formed by species that are all associated with Betula; (3) a clade comprising L. duriusculum, L. nigellum and L. uliginosum (the Fumosa clade), accessions that are all associated with Populus; and (4) a clade which will be referred to as the Leccinum clade and is composed of species which are associated with Populus, Betula, Arbutoideae, Pinaceae and Fagaceae. The relation between the Leccinum, Scabra and Fumosa clades remains unresolved.

Table 5.  Models of sequence evolution used for the individual partitions in the Bayesian analysis of Gapdh sequence data
Codon 1Felsenstein 81 model (Felsenstein, 1981), variable sites assumed to follow a gamma distribution.
Codon 2Felsenstein 81 model (Felsenstein, 1981), variable sites assumed to follow a gamma distribution.
Codon 3General time reversible model (Rodríguez et al., 1990), variable sites assumed to follow a gamma distribution.
Intron 1Kimura 2-parameter model (Kimura, 1980).
Intron 2Symetrical model (Zarkikh, 1994).
Intron 3Kimura 2-parameter model (Kimura, 1980).
Intron 4Hasegawa–Kishino–Yano model (Hasegawa et al., 1985).
Intron 5Kimura 2-parameter model (Kimura, 1980), variable sites assumed to follow a gamma distribution.

Figure 2. Tree based on the outcome of a Bayesian analysis of the Gapdh data. Thickened branches receive posterior probabilities of 95% or more. The values below the branches are bootstrap support values based on maximum parsimony analysis. Bootstrap support values < 50% are not indicated. Squares, Fagaceae; closed circles, Betula; open circles, Populus; tinted circles, Corylus/Carpinus; open triangles, Ericaceae; closed triangles, Pinaceae.

Download figure to PowerPoint

The Leccinum clade is very strongly supported (100% Bootstrap Support (BS), 100% Posterior Probability (PP)). Within this clade we can recognize five well to highly supported clades: (i) clade E formed by the two collections of L. monticola (associated with Comarostaphylis, Arbutoideae); (ii) clade D formed by L. vulpinum and L. piceinum (associated with Pinaceae); (iii) clade C formed by L. manzanitae and L. arbuticola (associated with Arbutoideae); (iv) clade B containing the North American L. atrostipitatum and the European L. versipelle accessions (except for one accession, all associated with Betula); and (v) a clade A comprising the European L. aurantiacum, L. insigne, L. leucopodium and some North American samples morphologically similar to L. aurantiacum (associated with a diversity of hosts). None of the relationships between these five clades receive any significant support. Clade A is composed of two well-supported subclades, one comprising L. aurantiacum (with a diversity of broad leaved hosts) and a moderately supported clade with four accessions under Populus and one accession of which the host plant associate was not recorded.

Molecular clock Gapdh

It has been shown for Drosophila that the protein coding sequences of Gapdh evolves clocklike at the nucleotide level (Ayala et al., 1996). To calculate if Gapdh in Leccinum also evolved clocklike we used a data set containing 25 taxa. A pairwise relative-rate test as implemented in the hyphy package 0.95beta (S.L. Kosakovsky Pond and S.V. Muse available from the authors at showed that the mutation rate of L. duriusculum significantly differed from most other taxa and therefore this taxon was removed form the clock analyses. When only the protein coding sequences were used, the hypothesis of a constant rate could not be rejected (–Ln constrained 2779.9693, –Ln unconstrained 2763.794, 2Δ = 32.35, df = 22, P = 0.07). When L. talamancae, L. crocipodium and L. carpini were excluded, a molecular clock could be assumed for the complete Gapdh sequences (–Ln constrained 3373.395, –Ln unconstrained 3359.998, 2Δ = 26.794, df = 19, P = 0.11). The topology of the tree based on the complete Gapdh gene sequences differed from the trees based only on the coding part of Gapdh by the fact that all the Arbutoideae-associated species are placed together with the Pinaceae associated species. The trees resulting from the ML analysis of the complete Gapdh sequences contradict the monophyly of this group, the Californian Arbutoideae-associated L. manzanitae is placed basal to all other species in the Leccinum clade, while the European Pinaceae-associated species form a separate clade with the Costa Rican Arbutoideae-associated species. The Bayesian analysis shows there is no significant support for this separate placement of L. manzanitae and therefore the topology cannot be considered incongruent with the one inferred from the complete sequences.

ITS2 phylogeny

The data set comprised 50 accessions of 536 characters of which 60 characters were potentially phylogenetically informative. Six accessions (L. versipelle Norway, L. cf. aurantiacum Canada, L. insigne Massachusetts, L. manzanitae California and both accessions of L. monticola from Costa Rica) shared a 40 bp deletion.

The MP analysis yielded more than 20 000 most parsimonious trees (154 steps, CI = 0.805, RI = 0.918). The ML analysis yielded 8 trees (–Ln = 1638.17), one of these trees is shown in Fig. 3. The MP, ML and Bayesian inference topologies did not differ significantly, though the Bayesian analysis showed somewhat less resolution. L. talamancae (the outgroup), L. crocipodium, and L. carpini are sister to the remaining Leccinum samples (69% BS). The other accessions fall into three main clades: (1) a weakly supported clade formed by the Populus-associated L. duriusculum and L. uliginosum (the Fumosa clade); (2) a highly supported clade containing most accessions of the Scabra clade (except L. variicolor and L. snellii) and a part of the Leccinum clade as found in the Gapdh analysis. Within the Scabra clade resolution shows three well-supported clades: (i) uniting L. holopus and L. brunneogriseolum; (ii) formed by accessions of L. schistophilum; and (iii) uniting L. cf. snellii and L. scabrum. The third major clade contains L. variicolor and the larger part of accessions of the Leccinum clade. However, this clade receives bootstrap support and posterior probability lower than 50%.


Figure 3. One of eight maximum likelihood trees based on ITS2 sequences. Thickened branches receive posterior probabilities of 95% or more. Values below clades indicate maximum parsimony bootstrap values. Values < 50% are not indicated. Squares, Fagaceae; closed circles, Betula; open circles, Populus; tinted circles, Corylus/Carpinus; open triangles, Ericaceae; closed triangles, Pinaceae.

Download figure to PowerPoint

Compatibility of ITS2 and Gapdh

The partition homogeneity test showed that the phylogenetic signal of the two data sets (Gapdh and ITS2) are highly incongruent (P < 0.001). The SH test showed that the topology of the trees obtained from the different data sets yielded significantly different (P < 0.001) likelihood scores when tested with either the gapdh dataset or the ITS2 dataset.

Reconstruction of the evolution of host associations

The ML trees with a molecular clock enforced of the Gapdh data were used to make a likelihood reconstruction of the ancestral character states (Figs 4 and 5). Leccinum versipelle was treated as an associate of Betula, although one accession was associated with Arctostaphylos uva-ursi. This is the only report of this species with this host and therefore we consider this an exception. Because one taxon (L. aurantiacum) was a generalist, we had to overcome the problem that mesquite cannot handle polymorphisms. Therefore, we compared reconstructions where the host association of L. aurantiacum was coded in different ways: (1) Betula; (2) Populus; (3) Fagaceae plus Coryloideae in the reconstruction based on the tree in Figs 4; (4) Fagaceae in the reconstruction based on the tree in Fig. 5. The reconstruction where Populus was coded as the mycorrhizal associate of L. aurantiacum received the highest likelihood score (see Table 5) in the reconstruction based on the coding sequences of Gapdh as well as in the reconstruction based on the complete Gapdh sequences. For the remainder of the discussion of the results we will mainly discuss the results of the reconstructions based on the complete Gapdh sequences, because this tree shows more resolution and most relationships are better supported. Betula received the highest likelihood score for being the mycorrhizal associate of the most recent common ancestor (MRCA) of taxa of the Scabra, Leccinum, and Fumosa clade, irrespective of the coding of the mycorrhizal association of L. aurantiacum and the tree used. The different coding of the mycorrhizal association of L. aurantiacum did affect the reconstructions of ancestral host associations of the two basal nodes (nodes 1 and 2 in Fig. 5) of the Leccinum clade and the host association of the MRCA of the species of node 3. When L. aurantiacum was coded to be a Fagaceae associate, Arbutoideae received the highest likelihood score for being the associate of the MRCA of the Leccinum clade (nodes 1 and 2), and Populus the associate of the MRCA of L. aurantiacum and L. leucopodium, L. insigne and Leccinum sp. 3 and 4 (node 3). Coding of the mycorrhizal association of L. aurantiacum as either Betula or Populus resulted in a likelihood score of the ancestral states of nodes one, two and three in favour of all being either Betula or Populus, respectively (Table 6).


Figure 4. Maximum likelihood tree with molecular clock enforced based on only the coding sequences of Gapdh. Thickened branches receive posterior probabilities of 95% or more in Bayesian analysis. Hatched branches receive posterior probabilities of between 90% and 95%. The axis below the tree gives the estimated number of substitutions per site. The likelihood reconstruction of ancestral host associations pictured here is the one where Populus was used as host for Leccinum aurantiacum s.s. Pie chart diagrams indicate proportional likelihood scores of nodes that could not be reconstructed unambiguously. Superimposed grey areas indicate episodes of rapid speciation.

Download figure to PowerPoint


Figure 5. Maximum likelihood tree with molecular clock enforced based on total Gapdh sequences. Thickened branches receive posterior probabilities of 95% or more in Bayesian analysis. Hatched branches receive posterior probabilities of between 90% and 95%. The axis below the tree gives the estimated number of substitutions per site. Numbers near nodes refer to the maximum likelihood reconstructions in Table 5. The likelihood reconstruction of ancestral host associations pictured here is the one were Populus was used as host for Leccinum aurantiacum s.s. Pie chart diagrams indicate proportional likelihood scores of nodes that could not be reconstructed unambiguously. Superimposed grey areas indicate episodes of rapid speciation.

Download figure to PowerPoint

Table 6.  Differences in results of likelihood reconstructions of ancestral host associations when the association of Leccinum aurantiacum is coded as being either one of the observed associated host species. The different nodes refer to the nodes with the same number in Fig. 5
Host association L. aurantiacumEstimated marginal probability (−log likelihood)Nodes 1 and 2Node 3
  1. *Significantly higher likelihood score for ancestral state of given host as compared to likelihood scores of other ancestral host states.


An additional parsimony reconstruction was performed based on the same trees as the ML reconstruction, with the exception that branches with a length close to zero were collapsed. This resulted in an unresolved relationship between the Scabra, Leccinum and Fumosa clades and the merging of nodes 1 and 2 (data not shown). The resulting polytomy were considered to be soft polytomies. An advantage of parsimonies reconstruction methods is that polymorphisms are allowed. Therefore, the associations could be coded according to genus or (sub)family (Fagaceae, Populus, Betula, Arbutoideae, Pinaceae, Coryloideae). In the parsimony reconstruction L. crocipodium was coded as being associated both with Fagaceae and Coryloideae, and L. aurantiacum as being associated with Fagaceae, Populus and Betula. The parsimony reconstruction showed the association of the MRCA of the Fumosa, Leccinum and Scabra clade could not be reconstructed unambiguously, as all hosts, except Pinaceae and Arbutoideae, were equally possible as the associate of this MRCA. The MRCA of the Leccinum clade was associated with Betula and/or Populus as was the MRCA of node 3.

The ML and parsimony reconstructions gave complementary information about ancestral mycorrhizal associations in Leccinum. Where parsimony showed an ambiguous reconstruction for the association of the MRCA of the Fumosa, Leccinum and Scabra clades, the ML reconstruction indicated that Populus and Betula were most likely the ancestral host. With both reconstruction methods Pinaceae or Arbutoideae can be ruled out as the ancestral host. Both reconstruction methods pointed toward Populus and/or Betula being the host of the MRCA of the Leccinum clade. This indicated that the contemporary Pinaceae and Arbutoideae associates evolved out of an ancestor that was associated with Populus and/or Betula. The second conclusion that can be drawn from these reconstructions is that the ability of L. aurantiacum to form mycorrhiza with Fagaceae is newly derived, and indicates a recent broadening of its host range.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Host specificity

Species of Leccinum are generally considered to be highly host specific (i.e. specialized on a phylogenetically restricted range of hosts). Our results show this to be generally true but with one major exception. Leccinum aurantiacum is associated with a broad range of hosts, found with Fagaceae (Quercus and Fagus), Betula and Populus. There are further records of associations with Tilia (Tiliaceae). Interestingly, the reconstruction of the ancestral host association provided clear evidence that this generalist evolved from an ancestor that was associated with a narrower host range, most likely Betula and/or Populus. It is not possible with the genes that we investigated to determine whether L. aurantiacum still behaves as a panmictic population or whether evidence exists of subsequent host race formation. Further investigations to address that question based on other molecular markers would be very useful. Schluter (2000) showed through compiling diverse phylogenetic studies that more often than expected generalists can evolve from specialists. His compilation and our observations on L. aurantiacum show that the generally held concept that ecological specialization must lead to more increased specialization may not always be valid.

Although within the Leccinum clade a generalist evolved from a more specialized ancestor when it concerns host specificity, a trend towards increased edaphic specialization is observed in the Scabra clade. This clade has a long history of association with Betula. Although all found on one host, in The Netherlands, in various locations, several species of this clade co-occur, showing edaphic niche differentiation: Leccinum scabrum on dry acidic soils, L. holopus in humid acidic areas and L. schistophilum on slightly calcareous, humid areas (Den Bakker, unpubl. obs.).

Incongruence of ITS2 and Gapdh

The ITS2 sequences and phylogeny showed two peculiarities. First, the presence of a shared 40 bp deletion in six accessions (L. versipelle Norway, Leccinum sp. 3 Canada, L. insigne Massachusetts, L. manzanitae, California and both accessions of L. monticola from Costa Rica). With the exception of L. monticola, closely related species or even sequences from different individuals of the same species (for example L. versipelle, clade 3 in Fig. 2) did not show this deletion. Most likely this represents an ancestral polymorphism, which is the best explanation for the exactly identical position of the deletion.

Another peculiarity of the ITS2 gene tree is the well supported (BS 85%, PP 98%) placement of the European L. aurantiacum and the North American Leccinum sp. 4 and L. brunneum (Leccinum clade 2), together with most species of the Scabra clade, except L. variicolor and L. snellii. In the Gapdh gene tree L. aurantiacum forms a monophyletic group with L. leucopodium, L. insigne and Leccinum sp. 4. Comparison of two loci in the ITS2 alignment (Table 7) shows the length of a single nucleotide ‘A’ repeat and sequence identity of these two loci are congruent with clades B, C, D and E in the Gapdh gene tree. In clade A in the Gapdh gene tree. however, we found several different sequences at the ITS2 loci. An explanation for this phenomenon would be that we are dealing here with paralogous copies of either gene. However, paralogous copies of Gapdh appear to be rare and are (to date) only found in photosynthetic plants (Figge et al., 1999). By contrast, paralogy in ITS is often encountered in plants and is associated with phenomena such as ancient introgression, hybridization and polyploidy (Álvarez & Wendel, 2003). The taxonomy of the North American species of the group of L. insigne and aurantiacum-like species is notoriously difficult and processes such as hybridization might account for these difficulties. More data are needed on this group.

Table 7.  Clade and accession specific nucleotide patterns found on two different loci in ITS2
Clade in Fig. 2Position 211Position 335
Clade A
Leccinum leucopodiumGCAAAC(3)
Leccinum sp. 4A(6)TCATT
Leccinum insigne and Leccinum sp. 3A(6)AC(3)
Leccinum aurantiacumGCAATCATT
Clade BA(5)TC(3)
Clades C and DA(10/9)AC(3/4)
Clade EA(8)ACTC

Host switches and speciation

The reconstruction of the ancestral host associations shows two major host switching events (Figs 4 and 5). First, a switch by the MRCA of the Fumosa and Leccinum clade from Betula to Populus. Second, a switch by the MRCA of the Leccinum clade from Populus to Betula and to Arbutoideae. Remarkably, these host switches are associated with or followed by episodes of rapid speciation, as indicated by the unresolved polytomies and short branch lengths in the clock trees. The same phenomenon of extensive speciation (adaptive radiation) after host switches has been noted in Suillus (Kretzer et al., 1996), Hebeloma (Aanen et al., 2000) and also in Pisolithus, where all four species of lineage B are associated with eucalypts and acacias, and the three species of lineage AII are associated with pines (Martin et al., 2002). The fact that the second episode of rapid speciation in the Leccinum clade seems to coincide with an episode of rapid speciation in the Scabra clade makes us think that the cause of this rapid speciation must be found outside host specificity, since there is no host shift taking place in the Scabra clade. We therefore think that genetic isolation of allopatric populations during times of glaciation in the Quaternary may account for this pattern. A possible scenario to explain the pattern of host shifts in the Leccinum clade could be genetic isolation of allopatric populations, leading to a narrowing of the host range as a consequence of a decrease in the number of potential host tree species in areas influenced by drastic climatic changes. Narrowing of the host range could also be driven by ecological specialization. Evidence for this scenario is found that most host switches took place between host communities of ecologically equivalent species instead of phylogenetic groups within genera or families. The switch from Coryloideae plus Fagaceae to Populus and Betula could then be explained by a separation of ancestral populations of warmer and colder climates, since Coryloideae and Fagaceae represent thermophilous hosts and Populus and Betula are typical representatives of sub-boreal vegetation types. The importance of ecology as a factor promoting niche expansion is also consistent with the observation that the species associated with Pinaceae and Arbutoideae share a common ancestor and have evolved from Populus and Betula. In the current distribution area of L. manzanitae and L. monticola (associates of Arbutoideae), the coastal forests of California and the highlands of Costa Rica, respectively, Betula and Populus are virtually absent. Possibly a host-switch occurred by the extinction or decrease of the distribution area of Betula and Populus that originally overlapped that of Arctostaphylos in the Californian floral region. A subsequent switch (or niche expansion) to an association with Pinaceae is likely, since Pinus and Pseudotsuga can co-occur with Arbutus and Arctostaphylos and share the same mycorrhiza (Molina et al., 1997; Horton et al., 1999). A similar host niche expansion from eucalypts to acacias may have occurred in Pisolithus lineage B (Martin et al., 2002).

If host specificity (or at least host niche contraction) is a side-effect of geographic isolation and allopatric speciation, this strongly suggests episodes of relaxed specificity in periods in which several hosts can be exploited, otherwise the disappearance of the one specific host will mean the extinction of the associated specialist fungi. Relaxation of specificity could also occur in marginal areas, for example as in the case of niche expansion from eucalypts to Kunzea (Myrtaceae–Leptospermoideae) in geothermal areas in New Zealand (Moyersoen et al., 2003).

In conclusion, species within the genus Leccinum are generally host specific as widely assumed. However, L. aurantiacum associates with a broad range of ectomycorrhizal broad-leaved trees. This shift from a Populus-associated specialist to a generalist probably took place recently in the evolutionary history of the genus and shows that, in contrast to the theory that evolution of a symbiont leads to increased specialization, the opposite can occur. This has taxonomic and evolutionary implications. Taxonomically the ability to grow on a new host cannot be taken a priori as evidence that a new Leccinum species has evolved. Phylogenetic studies can serve as a starting point for further research on the evolutionary biology of host specificity in mycorrhizal fungi. Cycles of niche contraction (switches from generalists to specialists) and niche expansion (from specialists to generalists) are essential to explain speciation and the evolution of host specificity in mycorrhizal fungi.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Tom Bruns for his hospitality and help in developing the Gapdh primers. Martin Bidartondo and Else C. Vellinga for making the stay in the Bruns laboratory a productive one. We thank Rasmus Kjøller for the development of the Gapdh primers, which served as a starting point for our work on this gene. Alan Hills and the curators of the herbaria of SFSU, P, H are thanked for sending material. We are particularly grateful to Roy Halling for making available the material from Costa Rica and Nancy Ironside for making it possible to conduct fieldwork in Canada. We also thank Barbara Gravendeel for creating the opportunity to visit California. Finally we thank Natasha Schidlo for her help both in the laboratory and in field. The first author was funded by a study bursary of the Rijksherbarium Kits van Waveren fund.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Aanen DK, Kuyper TW, Boekhout T, Hoekstra RF. 2000. Phylogenetic relationships in the genus Hebeloma based on ITS1 and 2 sequences, with special emphasis on the Hebeloma crustuliniforme complex. Mycologia 92: 269281.
  • Álvarez I, Wendel JF. 2003. Ribosomal ITS sequences and plant phylogenetic inference. Molecular Phylogenetics and Evolution 29: 417434.
  • Ayala FJ, Barrio E, Kwiatowski J. 1996. Molecular clock or erratic evolution? A tale of two genes. Proceedings of the National Academy of Sciences, USA 93: 1172911734.
  • Berbee ML, Pirseyedi M, Hubbard S. 1999. Cochliobolus phylogenetics and the origin of known, highly virulent pathogens, inferred from ITS and glyceralde-3-phosphate dehydrogenase gene sequences. Mycologia 91: 964977.
  • Binder M, Besl H. 2000. 28S rDNA sequence data and chemotaxonomical analyses on the generic concept of Leccinum (Boletales). In: Associazone MicologicaBresadola, ed. Micologia 2000. Brescia, Italy: Grafica Sette, 7586.
  • Bruns TD, Bidartondo MI, Taylor DL. 2002. Host specificity in ectomycorrhizal communities: what do the exceptions tell us? Integrative and Comparative Biology 42: 352359.
  • Buckley TR. 2002. Model misspecification and probabilistic tests of topology: Evidence from empirical data sets. Systematic Biology 51: 509523.
  • Câmara MPS, O'Neill NR, Van Berkum P. 2002. Phylogeny of Stemphylium spp. based on ITS and glyceraldehyde-3-phosphate dehydrogenase gene sequences. Mycologia 94: 660672.
  • Den Bakker HC, Gravendeel B, Kuyper TW. 2004. An ITS phylogeny of Leccinum and an analysis of the evolution of minisatellite-like sequences within ITS1. Mycologia 96: 102118.
  • Drummond AJ, Rambaut A. 2003. beast v1.0.3.
  • Farris JS, Kallersjo M, Kluge AG, Bult C. 1995. Testing significance of incongruence. Cladistics 10: 315319.
  • Felsenstein J. 1981. Evolutionary trees from DNA sequences: a maximum likelihood approach. Journal of Molecular Evolution 17: 368376.
  • Felsenstein J. 1985. Confidence limits on phylogenies – an approach using the bootstrap. Evolution 39: 783791.
  • Figge RM, Schubert M, Brinkmann H, Cerff R. 1999. Glyceraldehyde-3-phosphate dehydrogenase gene diversity in eubacteria and eukaryotes: Evidence for intra- and inter-kingdom gene transfer. Molecular Biology and Evolution 16: 429440.
  • Finlay RD. 1989. Functional aspects of phosphorus uptake and carbon translocation in incompatible ectomycorrhizal associations between Pinus sylvestris and Suillus grevillei and Boletinus cavipes. New Phytologist 112: 185192.
  • Halling RE, Mueller GM. 2003. Leccinum (Boletaceae) in Costa Rica. Mycologia 95: 488499.
  • Hasegawa M, Kishino H, Yano Y. 1985. Dating the human-ape splitting by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution 21: 160174.
  • Heinemann P. 1964. Boletinae du Katanga. Bulletin du Jardin Botanique de l’État à Bruxelles 34: 425478.
  • Holmgren PK, Holmgren NH, Barnett LC. 1990. Index herbariorum. Part I: The herbaria of the world, 8th edn. New York, USA: New York Botanical Garden, 693.
  • Horton TR, Bruns TD, Parker VT. 1999. Ectomycorrhizal fungi associated with Arctostaphylos contribute to Pseudotsuga menziesii establishment. Canadian Journal of Botany 77: 93102.
  • Huelsenbeck JP, Rannala B. 1997. Phylogenetic methods come of age: Testing hypotheses in an evolutionary context. Science 276: 227232.
  • Huelsenbeck JP, Ronquist F. 2001. mrbayes: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754755.
  • Kimura M. 1980. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16: 111120.
  • Korhonen M. 1995. New boletoid fungi in the genus Leccinum from Fennoscandia. Karstenia 35: 5366.
  • Kretzer A, Li YN, Szaro T, Bruns TD. 1996. Internal transcribed spacer sequences from 38 recognized species of Suillus sensu lato: phylogenetic and taxonomic implications. Mycologia 88: 776785.
  • Kreuzinger N, Podeu R, Gruber F, Göbl F, Kubicek CP. 1996. Identification of some ectomycorrhizal basidiomycetes by PCR amplification of their gpd (glyceraldehyde 3-phosphate dehydrogenase) genes. Applied and Environmental Microbiology 62: 34323438.
  • Lee C, Grasso C, Sharlow MF. 2002. Multiple sequence alignment using partial order graphs. Bioinformatics 18: 452464.
  • Lewis PO. 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50: 913925.
  • Maddison W, Maddison D. 2002. macclade, version 4.0.5. Sunderland, MA, USA: Sinauer.
  • Maddison W, Maddison D. 2003b. stochchar: a package of mesquite modules for stochastic models of character evolution, Version 0.996.
  • Maddison W, Maddison D. 2003a. mesquite, a modular system for evolutionary analysis, version 0.996.
  • Martin F, Díez J, Dell B, Delaruelle C. 2002. Phylogeography of the ectomycorrhizal Pisolithus species as inferred from nuclear ribosomal DNA ITS sequences. New Phytologist 153: 345357.
  • Massicotte HB, Molina R, Luoma DL, Smith JE. 1994. Biology of the ectomycorrhizal genus Rhizopogon. II. Patterns of host-fungus specificity following spore inoculation of diverse hosts grown in mono- and dual-cultures. New Phytologist 126: 677690.
  • Molina R, Massicotte H, Trappe JM. 1992. Specificity phenomena in mycorrhizal symbiosis: community-ecological consequences and practical implications. In: AllenMF, ed. Mycorrhizal functioning: an integrated plant–fungal process. London, UK: Chapman & Hall, 357423.
  • Molina R, Trappe JM. 1994. Biology of the ectomycorrhizal genus Rhizopogon. I. Host associations, host-specificity and pure culture syntheses. New Phytologist 126: 653675.
  • Molina R, Smith JE, McKay D, Melville LH. 1997. Biology of the ectomycorrhizal genus Rhizopogon. III. Influence of co-cultured conifer species on mycorrhizal specificity with the arbutoid hosts Arctostaphylos uva-ursi and Arbutus menziesii. New Phytologist 137: 519528.
  • Moyersoen B, Beever RE, Martin F. 2003. Genetic diversity of Pisolithus in New Zealand indicates multiple long-distance dispersal from Australia. New Phytologist 160: 569579.
  • Pagel M. 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on phylogenies. Systematic Biology 48: 612622.
  • Posada D, Crandall KA. 1998. modeltest: testing the model of DNA substitution. Bioinformatics 14: 817818.
  • Robinson D, Fitter A. 1999. The magnitude and control of carbon transfer between plants linked by a common mycorrhizal network. Journal of Experimental Botany 50: 913.
  • Rodríguez F, Oliver JL, Marin A, Medina JR. 1990. The general stochastic model of nucleotide substitution. Journal of Theoretical Biology 142: 485501.
  • Schluter D. 2000. The ecology of adaptive radiation. Oxford, UK: Oxford University Press.
  • Simard SW, Perry DA, Jones MD, Myrold DD, Durall DM, Molina R. 1997. Net transfer of carbon between ectomycorrhizal tree species in the field. Nature 388: 579582.
  • Singer R. 1986. The Agaricales in modern taxonomy, 4th edn. Koenigstein, Germany: Koeltz Scientific Books.
  • Smith AH, Thiers HD. 1971. The Boletes of Michigan. Ann Arbor, MI, USA: The University of Michigan Press.
  • Swofford DL. 2002. paup* – phylogenetic analysis using parsimony (* and other methods), version 4.0. Sunderland, MA, USA: Sinauer Associates.
  • Taberlet P, Gielly L, Patou G, Bouvet J. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 11051109.
  • Thiers HD. 1975. California mushrooms: a field guide to the Boletes. New York, NY, USA: Hafner Press.
  • Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. 1997. The Clustal–windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24: 48764882.
  • Trappe JM. 1962. Fungus associates of ectotrophic mycorrhiza. Botanical Review 28: 538606.
  • White TJ, Bruns T, Lee SS, Taylor J. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: InnisMA, GelfandDH, SninskyJJ, WhiteTJ, eds. PCR protocols: a guide to methods and applications. New York, NY, USA: Academic Press, 315322.
  • Yun SH, Berbee ML, Yoder OC, Turgeon BG. 1999. Evolution of the fungal self-fertile reproductive life style from self-sterile ancestors. Proceedings of the National Academy of Sciences, USA 96: 55925597.
  • Zarkikh A. 1994. Estimation of evolutionary distances between nucleotide-sequences. Journal of Molecular Evolution 39: 315329.