Secotioid agarics and boletes are morphological intermediates between normal epigeous mushrooms and hypogeous false-truffles. This has lead to speculation that they are ancestral to either false-truffles or to mushrooms (3, 14, 19). Gastrosuillus laricinus (Singer & Both) Thiers is a secotioid bolete. It was originally described as a species of Gastroboletus (14) and was later transferred to Gastrosuillus by Thiers when he erected that genus (20). Singer and Both discussed at length the possibility that G. laricinus was either ancestral or derived but concluded by saying: "The arguments for and against gastromycetation as opposed to arguments for and against an ancestral position of the genus Gastroboletus are not forceful enough to exclude the opposite hypothesis.... Further observations should be based on experimental evidence." In this study we provide what we believe is conclusive evidence based on sequence analysis of the internal transcribed spacer (ITS) that G. laricinus is indeed recently derived from local populations of S. grevillei (Kl.) Singer.Key Words: C-T transition bias, internal transcribed spacer, molecular evolution, secotioid fungi
The species we included in this study were chosen for the following reasons: Suillus grevillei was identified by Singer and Both as closely related to G. laricinus because they share several morphological and ecological features such as the presence of an annulus, identical spore size and shape, pileus color, chemical reactions, and mycorrhizal host (14). Suillus spraguei (Berk. & Curt.) Kuntze (= S. pictus (Peck) Smith & Thiers) is morphologically similar to S. grevillei (15), and evidence from mitochondrial DNA mapping suggests a close relationship between the two species (4). Rhizopogon subcaerulescens A. H. Smith is a false-truffle known to be closely related to Suillus (2, 3, 5).
We chose to sequence the ITS region because it contains two variable non- coding spacers located between highly conserved portions of the 18S, 5.8S and 28S rRNA genes. This location makes it easily accessible for sequence analysis with universal primers (21). Studies in other fungi and fungal-like protists have shown that the ITS region is often variable both within and between species (8, 10, 11).
DNA was extracted from the following herbarium specimens as previously described (2): S. grevillei , EB2296; S. grevillei , EB2040a; G. laricinus, EB2030; and G. laricinus , EB2031. All four were sympatric collections made by E. E. Both at the type locality, Krull Park, New York, the only known location for G. laricinus . The collections are deposited at Buffalo Museum. DNA was extracted from cultures of S. grevillei , TDB570; S. spraguei , TDB638; and Rhizopogon subcaerulescens, F 2882. The extraction methods and deposition of voucher collections have been described (2, 4). Sequences were determined for both strands by a slightly modified asymmetric primer ratio method (2). ITS5 and ITS4 primers were used for both symmetric and asymmetric amplifications (2, 21). ITS2 and ITS3 were used as additional internal sequencing primers (21).
Complete sequences for the ITS region are shown in FIG. 1. These reveal that the sequences of S. grevillei isolates differ from each other by 0-3% and differ from that of S. spraguei by approximately 12%. The sequence of R. subcaerulescens was too divergent to be aligned with any of the other taxa in two portions of the ITS region, but within more conserved, alignable portions, it differed by approximately 8% from all of the other taxa (FIG. I ). Sympatric collections of S. grevillei and G. laricinus had identical ITS sequences (FIG. 1). Parsimony analysis shows no close relationship between G. laricinus and Rhizopogon . Instead the closest relatives of G. laricinus are clearly sympatric and nonsympatric individuals of S. grevillei (FIG. 2). Bootstrap analysis (7) demonstrates that the branch uniting Rhizopogon with S. spraguei to the exclusion of S. grevillei and G. laricinus is supported by 99% of the replicate trees (N = 1000. Although alignments slightly different from that shown (FIG. I ) are possible, these results are robust to minor changes in alignment.
If we exclude the Rhizopogon sequence from the comparison, ITS divergence in the remaining species is primarily in the form of single point mutations with a bias towards C-T transition mutations. A similar C-T bias appears to be characteristic of several fungal nuclear genes (5) and has been reported from the ITS spacers of Fusarium (10). Between S. grevillei and S. spraguei the transition bias is 2.2:1, while within S. grevillei it is 7:1 (FIG. 3). The apparent decrease in relative frequency of transitions with greater divergence has been reported previously and attributed to the fact that as transversions accumulate they erase the record of prior transitions (6).
To examine the possible effect of a strong transition bias on the parsimony analysis we used a step matrix to weight transversions over transitions (16) by 7:1. The result was that this weighting further lengthened the branches leading to both S. spraguei and R. subcaerulescens but did not alter the tree topology nor the bootstrap confidence in the central branch.
These results strongly suggest that G. laricinus is recently derived from S. grevillei and that it is neither an ancestor of the boletes nor an evolutionary intermediate between the boletes and Rhizopogon . The position of G. laricinus within the S. grevillei clade demonstrated that it is related to Rhizopogon only though Suillus . One could argue that since this tree is unrooted, G. laricinus could still be ancestral to the entire assemblage, but the lack of any difference in ITS sequences between sympatric collections of G. laricinus and S. grevillei versus the highly divergent sequences of S. spraguei and R. subcaerulescens makes this rooting grossly incompatible with the expectation of progressive divergence with time. Furthermore, sequence data from three different rRNA genes and mitochondrial mapping data suggest that Suillus and Rhizopogon are themselves a derived group that is highly divergent from rest of the Boletaceae (2, 3, 4, 5).
What then is the possible role of G. laricinus in fungal phylogeny? In 1933 Goldschmidt (9) proposed the concept of a hopeful monster. The occurrence of a monster embodies the idea that a change during early developmental processes may produce a fundamentally altered phenotype. This monster would be hopeful should the change be viable and permit the occupation of a new environmental niche. We can apply this concept to G. laricinus interpreting its secotioid form as a developmental arrest (3). With this view the lack of divergence in the ITS region relative to S. grevillei and its restriction to a single collecting site suggests that this monster is of very recent origin and thus far has not been particularly successful.
If we accept the view that G. laricinus is a recent mutant of S. grevillei then the history of the site allows us to estimate a maximum age of approximately 60 yr for the origin of G. laricinus . This estimate is derived from the planting date of Larix decidua Mill. (E. Both, pers. comm.), the apparent mycorrhizal host of both taxa at the Krull Park location (14). It is possible that G. laricinus was derived somewhere else and migrated into the Krull park site after its establishment, but no appropriate habitats border the park and G. laricinus has never been found in repeated searches of similar habitats in Western New York (E. Both, pers. comm.).
We speculate that most or all extant secotioid boletes are also relatively recent and are unlikely to be the ancestors of the false truffles. We hypothesize this because virtually all of the secotioid boletes are also rare and limited in range, and many have morphologies that suggest close relationships to extant species groups or sections of genera (18, 20). The phylogenetic diversity of their apparent bolete ancestors is great, but most do not appear to be related to Suillus (18, 20). In contrast, many of the false-truffles are related to Suillus (2, 3; T. D. Bruns and T. M. Szaro, unpubl. data). Furthermore, two examples unrelated to the boletes exist in which secotioid forms are interfertile with normal agarics. In both cases the underlying genetic difference was shown to be very slight, perhaps as slight as a single gene (12, 13). Taken together we view these observations as evidence that secotioid forms are easy to produce but are often strongly selected against. Developmentally they may represent very simple errors, and ecologically they probably represent poorly adapted intermediates that are well suited for neither wind nor animal spore dispersal. Thus, we view the secotioid stage as a common, but severe, bottleneck through which the false- truffles probably passed without leaving a trace.
What about the taxonomic status of G. laricinus ? Unique features such as its disoriented and only partially exposed tube walls, its lack of spore discharge, its possession of both apo- and autobasidia, its lack of a well developed veil, and its dry pileus surface are the reasons that Singer and Both felt it was distinct from S. grevillei (14). We would now suggest that these differences are not independent of each other, but instead are likely to be caused by a very small number of developmental genes that have pleiotropic effects. Even if one agrees with this interpretation the real issue is reproductive isolation. The identical ITS sequences do not necessarily indicate that G. laricinus and S. grevillei are not reproductively isolated. ITS variation is insufficient to separate several mating defined taxa within the Armillaria mellea and Laccaria laccata species groups, and ITS variation also fails to distinguish some genera within the Sordariaceae (1, 8,17). Thus, the recognition of G. laricinus as a distinct species remains a matter of opinion. In our opinion, however, it simply represents a local mutant of S. grevillei and is therefore a synonym of it.
We thank Ernst Both for kindly providing samples from his collections of C. laricinus , S. grevillei and additional information on the species; Jim Anderson, Kerry O'Donnell, and Steve Lee for sharing their prepublication results on ITS evolution with us; and Mary Berbee for useful comments on the manuscript. This work was supported in part by NSF grant BSR-8700391 to T. D. Bruns.