Abstract: Internal transcribed spacer regions of the nuclear ribosomal repeat have been sequenced from 47 isolates belonging to 38 recognized species of Suillus sensu lato. The sequences have been analyzed for phylogenetic and taxonomic implications using parsimony as well as distance methods: Based on these data, the genera Boletinus and Fuscoboletinus, that are often recognized within Suillus sensu lato, are not monophyletic. Isolates of Suillus granulatus derived from either North America or Europe and Asia are polyphyletic and seem to represent at least two different species. Our data also suggest that within Suillus sensu lato, mycorrhizal associations with Larix are primitive and host changes to Pinus and Pseudotsuga seem to have occurred only once. Changes among host species of the same genus appear to be more frequent. Finally, the most primitive clade within Suillus sensu lato seems to be formed by organisms with a strongly boletinoid hymenophore.
Suillus sensu lato includes organisms sometimes also classified as Boletinus or Fuscoboletinus. Delineation of Suillus sensu stricto from the related genera Boletinus and Fuscoboletinus has remained controversial (Pomerleau & Smith, 1962; Smith & Thiers, 1971; Pegler & Young, 1981; Singer, 1986; Sutara, 1987). Boletinus is a rather old name and the genus has traditionally been recognized by its radially elongated pores (boletinoid hymenophore) (Kalchbrenner, 1867). Other characters that have been used to differentiate Boletinus from Suillus are the presence of clamp connections, a sterile stipe surface without glandular dots and the absence of fasciculate encrusted pleurocystidia. Delineations based on these characters are, however, inconsistent (Singer, 1962). Finally, Singer (1986) restricted the genus to three species (B. cavipes, B. paluster and B. asiaticus) which are characterized by the presence of clamp connections in their fruitbodies. It is well known, however, that species of Suillus sensu stricto are also able to develop clamp connections in mycelium cultures (Hübsch, 1961; Pantidou & Groves, 1966). Fuscoboletinus on the other hand is a comparably recent segregate in which Pomerleau & Smith (1962) have grouped together all Suillus or Boletinus species that give a vinacious to purple brown spore print.
In the present study, we have determined internal transcribed spacer (ITS) sequences from 47 isolates belonging to 38 different species of Suillus sensu lato. We have used those data to perform phylogenetic analyses and to clarify the confused state of taxonomy.
Polymerase chain reaction.--Reaction mixtures were made up from equal volumes of DNA solution and a master mix containing all the other necessary components. Final concentrations were: 10 mM Tris/HCl pH 8.3, 50 mM KCl, 2.5 mM MgCl2, 0.1 mg/ml gelatin, 200 mM of each of the four deoxyribonucleotide triphosphates, 0.5 mM of each of two different primers, 25 U/ml Taq polymerase, and empirically determined dilutions of template DNA (see paragraph on DNA extractions). Varying sets of primers were used in the reactions. In most cases, use of the primer pairs ITS5/ITS4 or ITS5/ITS4-S (that amplify both ITS regions including the 5.8S rDNA) was found to be convenient. From older specimens, however, yields could significantly be increased by individual amplification of the two ITS regions with the primer pairs ITS5/ITS2 and ITS3/ITS4-S. Primer sequences and maps have been published before (White et al., 1990; Gardes and Bruns, 1993) except for ITS4-S the sequence of which is: 5'-CCTCCGCTTATTGATATGCTTAAG-3'. PCR reactions were overlaid with a drop of mineral oil (Sigma) and subjected to 35 amplification cycles on a Techne Thermal Cycler (model PHC-2). Conditions for the thermal cycling are described in Bruns and Gardes (1993). If the first amplifications yielded some but not enough DNA, aliquots of the amplified fragment were run on a 1.5% low melt agarose gel, bands were cut out, melted into TE buffer (10 mM Tris/HCl pH 8.0, 1 mM EDTA) at 65-70 C, diluted (10-2 in water) and subjected to another round of 35 amplification cycles.
Sequencing.--For use in sequencing, amplified DNA was cleaned from the PCR reaction mixture either by successive dilution with H2O and reconcentration using centrifugal filters (Millipore Ultrafree-MC filters) or with the BIO 101 Geneclean II kit. In the latter case, DNAs were first run on an agarose gel (1% normal agarose, 2% NuSieve agarose (FMC BioProducts), bands were cut out and subsequently cleaned according to the kit instructions. Sequencing was done by the cyclic reaction termination method using fluorescence labeled dideoxyribonucleotide triphosphates. The sequence reaction and the processing of the reaction products for electrophoresis were performed following the instructions for the sequencing kit (PRISM TM Ready Reaction DyeDeoxy TM Terminator Cycle Sequencing Kit, Perkin-Elmer Corporation). Electrophoresis and data collection were done on an ABI Model 373A DNA Sequencer (Perkin-Elmer Corporation) to which a Macintosh Quadra 650 computer was connected. DNA Sequencing Analysis (version 2.01) and SeqEd (version 1.03) were used for processing the raw data. A few sequences were determined by conventional dideoxy termination reaction with 35S labeled ATP as described by Bruns et al. (1990).
Phylogenetic analysis.--Sequences were aligned by visual estimation using a matrix created in PAUP 3.1.1 (Swofford, 1993) and a color font. Computer alignment was not found to be useful due to short hypervariable areas in the ITS-sequences. Obvious insertions/deletions (indels) were considered to be informative and were weighted equal to one or two transition(s) (except for a very large insertion present in the outgroups that was weighted as four transitions). Technically, this was done as follows: Alignment gaps were treated as missing data, but a new character state "I" was introduced at deletion sites. The weight attributed to the proper deletion was taken into account by the number of "I"s being introduced. As far as possible, only gaps with exactly the same length and position were coded for with "I"s introduced at precisely the same position. Phylogenetic analysis was done using both parsimony (Fitch, 1971) and distance methods (neighbor joining, Saitou & Nei, 1987). Searches for most parsimonious trees were done in PAUP 3.1.1, and neighbor joining was performed as follows: First a distance matrix was created in PAUP 3.1.1 that was subsequently adjusted manually to precisely fit the output format of DNADIST, a program of the PHYLIP 3.5 package that computes distance matrixes from sequence data. The advantage of using PAUP 3.1.1 over DNADIST for computing distances was that PAUP 3.1.1 recognized the introduced character state "I", when the data type was not defined to be "DNA". Finally, the edited distance matrix was fed into NEIGHBOR, another program of the PHYLIP 3.5 package, for neighbor joining. Bootstrap analyses were done under the parsimony criterion in PAUP 3.1.1, and values are based on 500 replicates. Gomphidius glutinosus and Chroogomphus vinicolor have been chosen as outgroups for the phylogenetic analyses, because both are closely related to Suillus (Bruns and Szaro, 1992). Rhizopogon subcaerulescens, Truncocolumella citrina and Gastrosuillus laricinus were also included in the study. Rhizopogon subcaerulescens and Truncocolumella citrina have been shown before to be closely related to Suillus (Bruns et al., 1989; Bruns et al., 1990). Gastrosuillus laricinus is a recent derivative of Suillus grevillei (Baura et al., 1992).
Alignment of the full set of sequences was in some areas ambiguous. Inclusion or exclusion of these areas did not, however, seem to influence the main conclusions drawn from the study, nor did differential weighting of transitions and transversions (either equal or transversions twice over transitions). Long branches essentially proved to be well supported independent of the treatment, whereas short branches were unstable. Results presented in Figures 1 and 2 are based on unweighted analyses from which areas of ambiguous alignment have been excluded. Under these conditions, the data matrix comprised 676 characters out of which 168 were informative. Figure 1 summarizes results from parsimony analysis; only branches with more than 50 % support by bootstrap analysis are shown. Figure 2 presents a neighbor joining tree derived from the same dataset. Although a number of branches on the neighbor joining tree are weakly supported (as can be seen by comparison to Figure 1), it currently represents our best phylogenetic estimate.
Finally, we have assessed intraspecific sequence variation and relationships between closely related taxa in a number of separate analyses, and the results are shown in Figure 3. Trees were constructed from the corresponding subset of taxa by neighbor joining. As only closely related species were included in each dataset, sequences could be aligned unambiguously, and no areas of ambiguous alignment needed to be excluded. Transitions and transversions were, again, weighted equally.
Other than Boletinus and Fuscoboletinus, Suillus sensu lato appears to be monophyletic with the exception of Gastrosuillus laricinus which is a recent derivative of Suillus grevillei (Baura et al., 1992) and probably other Gastrosuilli (unpublished results). As we are planning more detailed studies on the genus Gastrosuillus, we will discuss this problem at a later point. For now, we propose to collapse both the genera Boletinus and Fuscoboletinus into Suillus. As neither Boletinus nor Fuscoboletinus has been consistently recognized by authors in the past, most of the species have at some point already been included in Suillus. Those species that haven't and that are treated or mentioned in this study are transferred below.
Isolates of Suillus granulatus derived from either North America (S. granulatus TDB-878) or Europe and Asia (S. granulatus AD-11 and S. c. f. granulatus VC-1042) were widely separated into different clades by strongly supported branches. These findings are in agreement with earlier studies based on mating behavior and the physiology of mycorrhiza formation which indicated that isolates from America and Europe (Fries & Neumann, 1990) or America and Asia (Jacobson and Miller, 1992) might represent two different species. To our knowledge, no comparison has so far been made between isolates from Europe and Asia. Although isolates from Sweden and Nepal were more closely related to each other in this study than either of them was to North American isolates, they were still separated by a strongly supported branch which puts S. collinitus next to S. granulatus from Nepal and makes these two organisms paraphyletic to the Swedish S. granulatus and others. In addition, S. granulatus associates with a very different pine in Sweden than in Nepal (Pinus sylvestris L. section Pinus versus Pinus wallichiana Jackson section Strobus). Finally, S. c. f. granulatus from Nepal also seems to be morphologically distinct from other organisms being classified as S. granulatus in that it has a less mottled cap and sometimes stains slightly blue green to blue at the stipe base (Cotter, 1987). Overall, current evidence suggests that there might be three different taxa to be recognized within the Suillus granulatus complex, but more isolates need to be included in future studies to fully resolve the genetic heterogeneity within this complex. There are at least two more members of the S. granulatus complex that need to be examined: (i) another Asian S. granulatus from Korea which grows with Pinus densiflora Sieb. & Zucc. from the section Pinus (in contrast to the Nepalian isolate included in this study that grows with Pinus wallichiana from the section Strobus); (ii) S. granulatus found in the southwestern United States growing with Pinus ponderosa Laws. (section Pinus), while other North American isolates (including TDB-878) grow with Pinus strobus L. (section Strobus).
Besides the S. granulatus problem, minor conflicts with current species concepts have been observed in two more cases both of which involve specimens from Nepal: (i) Suillus c. f. placidus from Nepal has been described as fitting current species descriptions well but not completely, although observed differences were not explicitly stated (Cotter, 1987). We found S. c. f. placidus VC-1022 (Nepal) and S. placidus TDB-725 (North America) to be fairly closely related, but separated by a well supported branch which puts American S. placidus TDB-725 and American S. granulatus TDB-878 tightly together and makes the two S. placidus isolates from either North America or Nepal paraphyletic instead of monophyletic (see Fig. 3A). (ii) Finally, the isolate VC-1231 (Nepal) of Suillus laricinus fits into the clade of S. laricinus and relatives but is paraphyletic to S. laricinus sensu stricto (Fig. 3B). The species concept for S. laricinus has been considerably narrowed in the last decades by elevating the former variants S. bresadolae and S. serotinus to the level of independent species (Gerhold, 1985, and Smith & Thiers, 1971, respectively). Future work including more isolates should show whether this refined species concept is supported by molecular data. If that is the case, a new species will be needed to accommodate "S. laricinus" from Nepal which might also be supported by the unusual staining pattern observed on these specimens (Cotter, 1987).
Intraspecific sequence divergence was less pronounced in two more cases examined and did not conflict with current species concepts (Fig. 3C&D). Suillus luteus was the least divergent species looked at with only one base difference even between isolates from Europe and North America. This finding is in agreement with the idea of a recent introduction of S. luteus into North America. Suillus cavipes formed a divergent but monophyletic group, with the European isolate HDT-31407 being more distantly related to the two American isolates WJS-618 and TDB-646.
Our best tree estimate shown in Figure 2 suggests some evolutionary aspects that are interesting to note though they are not strongly supported by the data: First, the pattern of host specificity within Suillus sensu lato does not seem to have evolved through cospeciation with the ectomycorrhizal hosts. In conflict with the idea of cospeciation is primarily the finding that, while Larix is considered to be more recently derived than Pinus (Hart, 1987), larch association seems to be primitive in Suillus, and associations with pines, Douglas-fir, and hardwoods seem to be derived. Rather than cospeciation, multiple jumps to new host species and genera seem to have occurred during the evolution of Suillus sensu lato. While changes in host specificity to closely related hosts seem to be fairly frequent events (at least five changes between the pine subsections Pinus and Strobus are implied in Figure 2), all changes to new host genera seem to be unique events in the evolution of Suillus sensu lato. Changes in host specificity have also occurred in the outgroups, Rhizopogon, Truncocolumella and the Gomphidiaceae, but are not discussed in this paper. Finally, the most primitive clade within Suillus sensu lato seems to be formed by organisms with a very pronounced boletinoid hymenophore, that is with radially strongly elongated pores. It seems that the boletinoid hymenophore represents an intermediate evolutionary state between the gills found in related genera such as Gomphidius and Chroogomphus and the pores found in Suillus. This is basically an old idea; Smith and Thiers (1964) stated: "We find a tendency toward a lamellate type of hymenophore in Fuscoboletinus paluster, and it is only a step from this group to the Gomphidiaceae, whose species are truly lamellate." The lines and directions of evolution, however, have started to become clear only recently: Within the suilloid group, tubes seem to be derived from gills (Bruns & Szaro, 1992).
Suillus asiaticus (Singer) Kretzer, Li, Szaro & Bruns, comb. nov.
Basionym: Boletinus asiaticus Singer. Rev. Mycol. 3. 164. 1938.
Suillus grisellus (Peck) Kretzer, Li, Szaro & Bruns, comb. nov.
Basionym: Boletinus grisellus Peck. Mem. N. Y. State Museum 4: 169. 1900.
Synonym: Fuscoboletinus grisellus (Peck) Pomerleau & Smith. Brittonia 14:168. 1962.
Suillus paluster (Peck) Kretzer, Li, Szaro and Bruns, comb. nov.
Basionym: Boletus paluster Peck. Ann. Rept. New York State Cab. 23: 132. 1872.
Synonyms: Boletinus paluster (Peck) Peck. Bull. N.Y. State Museum 2: 78. 1889.
Boletinellus paluster (Peck) Murrill. Mycologia 1: 8. 1909.
Fuscoboletinus paluster (Peck) Pomerleau and Smith. Mycologia 56:708. 1964.
Suillus serotinus (Frost) Kretzer, Li, Szaro & Bruns, comb. nov.
Basionym: Boletus serotinus (Frost) Bull. Buffalo Soc. Nat. Sci. 2: 100. 1874.
Synonym: Fuscoboletinus serotinus (Frost) Smith & Thiers. Boletes of Michigan, p. 85. 1971.
Suillus weaverae (Smith & Schaffer) Kretzer, Li, Szaro & Bruns, comb. nov.
Basionym: Fuscoboletinus weaverae Smith & Schaffer, Mich. Bot. 4: 27. 1965.
Bruns, T. D., R. Fogel, T. J. White, and J. D. Palmer. 1989. Accelerated evolution of a false-truffle from a mushroom ancestor. Nature 339: 140--142.
---, ---, and J. W. Taylor. 1990. Amplification and sequencing of DNA from fungal herbarium specimens. Mycologia 82: 175--184.
---, and M. Gardes. 1993. Molecular tools for the identification of ectomycorrhizal fungi - taxon-specific oligonucleotide probes for suilloid fungi. Mol. Ecol. 2: 233--242.
---, and J. D. Palmer. 1989. Evolution of mushroom mitochondrial DNA: Suillus and related genera. J. Mol. Evol. 28: 349--362.
---, and T. Szaro. 1992. Rate and mode differences between nuclear and mitochondrial small-subunit rRNA genes in mushrooms. Mol. Biol. Evol. 9: 836--855.
Cotter, H. V. T. 1987. The systematics and ecology of boletes with special reference to the genus Suillus and its ectomycorrhizal relationships in Nepal. Ph.D. Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia.
Fitch, W. M. 1971. Toward defining the course of evolution: Minimal change for a specific tree topology. Syst. Zool. 20: 406--416.
Fries, N., and W. Neumann. 1990. Sexual incompatibility in Suillus luteus and S. granulatus. Mycol. Res. 94: 64--70.
Gardes, M., and T. D. Bruns. 1993. ITS primers with enhanced specificity for basidiomycetes - application to the identification of mycorrhizae and rusts. Mol. Ecol. 2: 113--118
Gerhold, N. 1985. Taxonomische Klarstellungen in der Sektion Larigni Singer der Gattung Suillus Micheli ex S. F. Gray. Z. Mykol. 51: 161.
Hart, J. A., 1987. A cladistic analysis of conifers: preliminary results. J. Arnold Arbor. 68: 269--307.
Hübsch, P. 1961: Das Vorkommen von Schnallen bei Boletaceen. Z. Pilzk. 27: 110--113.
Jacobson, K. M., and O. K. Miller. 1992. Physiological variation between tree-associated populations of Suillus granulatus as determined by in vitro mycorrhizal synthesis experiments. Can. J. Bot. 70: 26--31.
Kalchbrenner, C. 1867. Notitz über eine neue Polyporeengattung. Bot. Zeitung 25: 181--182.
Pantidou, M. E., and J. W. Groves. 1966. Cultural studies of Boletaceae: some species of Suillus and Fuscoboletinus. Can. J. Bot. 44: 1371--1392.
Pegler, D. N., and T. W. K. Young. 1981. A natural arrangement of the Boletales with reference to spore morphology. Trans. Br. Mycol. Soc. 76: 103--146.
Pomerleau, R., and A. H. Smith. 1962. Fuscoboletinus, a new genus of the Boletales. Brittonia 14: 156--172.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: A new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4: 406--25.
Singer, R. 1962. The Agaricales in modern taxonomy. 2nd ed. J. Cramer, Weinheim, Germany. 915 pp.
---. 1986. The Agaricales in modern taxonomy. 4th ed. Koeltz Scientific Books, Koenigstein, Germany. 981 pp.
Smith, A. H., and H. D. Thiers. 1964. A contribution toward a monograph of North American species of Suillus. Publ. by the authors. Ann Arbor, Michigan. 116 pp.
Smith, A. H., and H. D. Thiers. 1971. The Boletes of Michigan. The University of Michigan Press, Ann Arbor, Michigan. 428 pp.
Snell, W. H., and E. A. Dick. 1941. Notes on Boletes. VI. Mycologia 33: 23--37.
Sutara, J. 1987. The limit between the genera Boletinus and Suillus. Ces. Mykol. 41: 139--152.
Swofford, D. L. 1993. PAUP: Phylogenetic analysis using parsimony, version 3.1. Computer program distributed by the Illinois Natural History Survey, Champaign, Illinois.
Taylor, J. W., and E. C. Swann. 1994. Dried samples: Soft tissues. 11 DNA from herbarium specimens. Pp. 166--181. In: Ancient DNA. Eds., B. Herrmann, and S. Hummell. Springer-Verlag, New York.
White, T. J., T. Bruns, S. Lee, and J. W. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315--322. In: PCR protocols; A guide to methods and applications. Eds., M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White. Academic Press, San Diego, California.
Fig. 1. Phylogenetic analysis of the ITS sequence data: Bootstrap analysis under parsimony criterion showed that most of the major lineages are recognized with reasonable confidence, while relationships among them are not. From the major lineages it is evident that neither Boletinus nor Fuscoboletinus
as defined by Singer (1986) and Pomerleau and Smith (1962), respectively,
is monophyletic. Furthermore, isolates of S. granulatus
derived from either North America (S. granulatus TDB-878)
or Europe (S. granulatus AD-11) and Asia (S. c. f. granulatus VC-1042) are polyphyletic and do not seem to represent the same species. Branch lengths are arbitrary.
Fig. 2. Neighbor joining tree representing a best tree estimate: Within the phylogeny of Suillus sensu
lato, mycorrhizal associations with larches seem to be primitive
and switches to different host genera such as pines, Douglas-fir,
or hardwoods seem to have occurred only once. On the other hand,
host switches between pines of the different subgenera Pinus
or Strobus seem to be rather frequent. The shown host specificities are based on observations of fruiting habits both by the authors and other authorities (Smith & Thiers, 1964; Smith & Thiers, 1971; Singer, 1986; Cotter, 1987). Horizontal branch lengths represent the standard mean distances (Swofford, 1993). a Fruitbodies have been found in mixed conifer forests under Pinus, Abies
and Picea .
Fig. 3. Assessing intraspecific sequence divergence: Bootstrap values are given for selected branches of interest only when above 60%. Horizontal branch lengths represent standard mean distances (Swofford, 1993).