This Article Abstract Full Text (PDF) Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Miadlikowska, J. Articles by Posada, D. Search for Related Content PubMed Articles by Miadlikowska, J. Articles by Posada, D. Agricola Articles by Miadlikowska, J. Articles by Posada, D.

New approach to an old problem: Incorporating signal from gap-rich regions of ITS and rDNA large subunit into phylogenetic analyses to resolve the Peltigera canina species complex

     Department of Biology, Duke University, Durham, North Carolina 27708-0338 and Department of Plant Taxonomy and Nature Conservation, Gdansk University, Al. Legionow 9, 80-441 Gdansk, Poland

François Lutzoni

     Department of Biology, Duke University, Durham, North Carolina 27708-0338

Trevor Goward ²

     Herbarium, Department of Botany, University of British Columbia, Vancouver, British Columbia V6T 2B1, Canada

Stefan Zoller

     Department of Biology, Duke University, Durham, North Carolina 27708-0338

David Posada

     Variagenics, 60 Hampshire Street, Cambridge, Massachusettes 02139

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 

The Peltigera canina species complex consists of foliose lichenized bitunicate ascohymenial discomycetes forming section Peltigera within the genus Peltigera (Lecanoromycetes, lichen-forming Ascomycetes). To test the circumscription of highly polymorphic species and to resolve relationships among putative members of the P. canina complex, part of the nuclear ribosomal DNA large subunit (LSU rDNA) and the entire internal-transcribed spacer (ITS rDNA) were sequenced for 84 individuals representing 33 putative Peltigera taxa. Seventeen of the 25 taxa from the P. canina complex are well established and widely accepted. The remaining eight taxa have been proposed recently but are undescribed. A hypervariable region in ITS1 (ITS1-HR, sites 111–237 in our alignment) showed remarkable variation in length, especially in the P. canina complex, ranging from 8 to 126 bp, and contained several microsatellites. We describe here an alignment-free method to code such large gap-rich hypervariable regions for phylogenetic analyses. Variation among ITS1-HR sequences greatly contributed to species delimitation and species identification and can be a major asset to future population studies for specific species within section Peltigera. Sequences of ITS1-HR alone were sufficient to identify all existing species of Peltigera from the P. canina species complex and related sections Retifoveatae and Horizontales included in this study. However, only when INAASE (for short ambiguously aligned regions) and ITS1-HR coded characters were added to the combined analysis of nonambiguous LSU and ITS sites was it possible to reach the level of phylogenetic resolution and support necessary to disentangle the P. canina complex. We report here complete concordance between phylogenetically based and morphologically based species delimitation for 15 of the 17 species from the P. canina complex (P. canina, P. cinnamomea, P. degenii, P. evansiana, P. frigida, P. kristinssonii, P. laciniata, P. lambinonii, P. lepidophora, P. membranacea, P. monticola, P. ponojensis, P. praetextata, P. rufescens and P. ulcerata). Four of the eight newly proposed but undescribed taxa most likely represent new species (P. "fuscopraetextata", P. "neocanina", P. "neorufescens" and P. "scotteri") within the P. canina complex. We found that morphologically and chemically distinct P. didactyla s. str. and P. didactyla var. extenuata form two non-sister monophyletic entities, therefore the latter taxon should be recognized at the species level (P. extenuata). The North American and European populations of the morphologically uniform P. degenii might represent two sibling species because they were found to be genetically distinct and monophyletic. Two major monophyletic groups within the P. canina complex (CICADE = CInnamomea + CAnina + DEgenii group and PORUDI = POnojensis + RUfescens + DIdactyla group) seem to be correlated with different humidity preferences. Although some authors previously have suggested interspecies recombination within the P. canina complex, we did not find statistically significant evidence for this phenomenon based on LSU and ITS sequences.

Key words: Bayesian inference, gap-rich alignments, INAASE characters, internal transcribed spacer (ITS1, 5.8S, ITS2), lichenized Ascomycota, maximum likelihood, maximum parsimony, microsatellites, morphospecies, nuclear large-subunit ribosomal DNA (LSU rDNA), Peltigera canina species complex, Peltigerales, phylogenetics, section Peltigera

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Species complexes are abundant among lichen-forming fungi. The Peltigera canina species group comprises many worldwide taxa that often are difficult to identify. These primarily terricolous and muscicolous foliose macrolichens inhabit a wide range of environments, from somewhat xeric, exposed, often unstable localities (e.g., P. didactyla var. didactyla, P. ponojensis and P. rufescens) to humid, sheltered, usually wooded sites (e.g., P. praetextata, P. degenii and P. cinnamomea).

From the broadly defined type species for the genus Peltigera, Lichen caninus L. (Linnaeus 1753, Peltigera canina), several taxa have been segregated, most of which were members of the P. canina group. According to Gyelnik (1927 and 1933), P. canina and allied species are defined by the "caninaeform" type of venation, i.e., veins that are relatively thin, elevated, with rather wide interspaces on the lower surface of the thallus and by an arachnoid tomentum (with hyphal tips appressed or entangled) developed on the upper surface of the thallus (Figs. 1–5 in Vitikainen 1994). In 1993, Holtan-Hartwig, based on Norwegian collections, proposed a more precise circumscription of the P. canina group. According to this author, members of the P. canina group (eight species) are well characterized by: 1) the internal and external structure of veins on the lower side of the thallus "... cylindrical veins with the central parts composed of parallel hyphae in which the walls are conglutinated. The outer parts of the veins are composed of branched hyphae or hyphae branching in right angles from the veins"; 2) the mostly tomentose upper cortex of the thallus "... appressed, thin-walled and branched hairs on the upper side ..."; and 3) the lack of secondary metabolites detectable by thin layer chromatography (TLC), except in P. didactyla. This morphological and chemical circumscription of the P. canina group subsequently has been widely accepted since then (e.g., Goffinet and Hastings 1994, Martínez et al 1997, Martínez Moreno 1999, Vitikainen 1994). However, Vitikainen (1994) suggested that P. retifoveata (placed by Holtan-Hartwig [1993] in a group of its own), as well as some other non-European tomentose species with secondary metabolites (e.g., P. laciniata) and glabrous secondary metabolite-deficient species (e.g., P. austroamericana and P. subhorizontalis), also should be included in the P. canina group.

Based on a broad phylogenetic study of taxa mainly from North America and Europe, Miadlikowska and Lutzoni (2000) redefined the P. canina group using monophyly as the grouping criterion and recognized this group as one of eight sections (Peltigera section Peltigera) within the genus Peltigera. A total of 16 species were included in this section. Eight species (P. cinnamomea Goward, P. continentalis Vitik., P. evansiana Gyeln., P. frigida R. Sant., P. laciniata (G. Merr. ex Riddle) Gyeln., P. lambinonii Goffinet, P. monticola Vitik. and P. ponojensis Gyeln.) were added to the set of taxa selected by Holtan-Hartwig (1993; P. canina [L.] Willd., P. degenii Gyeln., P. didactyla [With.] J. R. Laundon, P. kristinssonii Vitik., P. lepidophora [Vain.] Bitter, P. membranacea [Ach.] Nyl., P. praetextata [Sommerf.] Zopf, and P. rufescens [Weiss] Humb.). Most of these species are widely distributed in Europe and North America, with the exception of P. lambinonii (Africa and Australia), P. continentalis (Asia), P. frigida and P. laciniata (Central and South America).

Morphological and chemical attributes of the newly circumscribed P. canina complex remain essentially unchanged compared to Holtan-Hartwig's delimitation of this species group. The triterpenoid zeorin, present in P. laciniata, and the glabrous upper cortex of the thallus and horizontal apothecia of P. frigida are new features expanding the morphological and chemical delimitation for this species group sensu Holtan-Hartwig. Three nonmolecular synapomorphic character states define this section: 1) the absence of depsides and terpenoids (followed by two reversals in lineages leading to P. laciniata and P. didactyla var. extenuata); 2) the frequent observation of pycnidia on specimens from this section; and 3) the veins composed of hyphae, which are mostly to completely parallel and conglutinated. The characters complete the description of section Peltigera: exclusively bimembered, with cyanobacteria (Nostoc) as the photobiont; rhizines simple or branched (fibrillose, flocculent or penicillate); apothecia pale brown to dark brown, usually vertical and saddle-shaped; vegetative propagules (isidia, phyllidia, and soredia) frequent; if present, terpenoids do not co-occur with depsides within the same individual. A large number of lichenicolous fungi that are associated with the genus Peltigera (peltigericolous fungi) are found predominantly on members of the P. canina complex (Miadlikowska 1999).

Comparatively few studies have been directed toward members of the P. canina complex; most have examined structure and development of vegetative propagules in selected species (Darbishire 1926; Lindahl 1953, 1960; Strato 1921; Thomson 1948) to elucidate their diagnostic use for species delimitation. However, some of the conclusions from these studies were conflicting, e.g., using phyllidia as a valid character to separate P. praetextata from the morphologically similar P. canina. A summary of the features of pycnidia and conidia observed in some Peltigera species, including three members of section Peltigera (P. canina, P. rufescens and P. laciniata), was presented by Lindahl (1959).

Many of the phenotypic characters traditionally used to resolve taxonomic units within the P. canina complex are unreliable, owing to a high degree of morphological variability among the species (Goffinet and Hastings 1994, Goffinet et al 1994, Goward et al 1995, Miadlikowska and Lutzoni 2000, Vitikainen 1994). This has resulted in the description of many species and intraspecific taxa that, upon closer re-examination, proved to be variants of previously described species (Vitikainen 1994). In addition, intermediate pheno- and chemotypes frequently are identified within a population or an individual thallus of species from the P. canina complex. Goffinet and Hastings (1995) hypothesized such patterns of variation observed in P. didactyla var. extenuata and P. didactyla s. str. by an introgressive hybridization or thallus fusion. Morphometric work on selected traits (spore size, thallus thickness, and vein width and height) was conducted on P. canina, P. membranacea and P. praetextata populations using statistical methods to better delimitate these species (Martínez and Burgaz 1996).

Despite successful in vitro resynthesis of Peltigera praetextata and P. didactyla thalli (Stocker-Wörgötter and Türk 1988, 1990, 1991), experimental studies (including mating tests) are not practical or feasible for resolving taxonomic questions within the P. canina complex and lichen-forming fungi in general. Therefore, morphological continuity within species and morphological discontinuity among species has been used as the main criteria for delimitating lichen-forming fungal species. Taxonomy at the species level within the P. canina complex relies mainly on vegetative features of the thallus, e.g., tomentum type and pattern, vein and rhizine morphology and arrangement and vegetative propagule presence (type and shape)/absence.

For the past 10 years, a phylogenetic species concept based on molecular characters (ITS sequences alone and LSU + ITS combined; except for Kroken and Taylor [2001] who used multiple loci) was employed to circumscribe species within lichen species complexes and to evaluate a posteriori which phenotypic characters are associated with each distinct monophyletic species (Pseudocyphellaria crocata complex, Miadlikowska et al 2002; Peltigera section Horizontales, Goffinet and Miadlikowska 1999; Lasallia, Niu and Wei 1993; Ramalina, Groner and LaGreca 1997; and Xanthoria, Franc and Kärnefelt 1998). In all these cases, monophyletic groups of individuals were supported by distinct morphological traits. Grube and Kroken (2000) reviewed the implication of molecular phylogenetic approaches to resolve species complexes in lichenized fungi.

When a large fraction of the variable sites of an alignment is ambiguous due to length variation among sequences, recovering phylogenetic signal from these regions can greatly contribute toward increasing phylogenetic resolution and confidence. In 2000, Lutzoni et al described a new method (INAASE) to integrate signal from ambiguously aligned regions of DNA sequences into phylogenetic analyses without violating positional homology. Because this method relies in part on pairwise alignments for all pairs of sequences within an ambiguous region and involves step matrices, INAASE can be implemented only on relatively small ambiguous regions (rarely up to 50 sites long). A method is needed that would allow the incorporation of large ambiguous regions into phylogenetic analyses without requiring alignments or imposing that all sites (ambiguously and nonambiguously aligned sites) be optimized directly (e.g., POY, Wheeler 1996) when reconstructing phylogenies.

Our detailed reassessment of morphological, chemical and molecular data for the genus Peltigera (Miadlikowska and Lutzoni 2000) was designed mainly to address questions at the intrageneric level. Although the P. canina complex was well sampled, our previous study did not focus on this section and included only one molecular dataset—the nuclear LSU rDNA. The aims of this molecular phylogenetic study were: 1) to test the monophyletic recognition of the Peltigera canina complex (section Peltigera) sensu Miadlikowska and Lutzoni (2000) and its sister relationship to section Retifoveatae; 2) to evaluate the delimitation of existing morphospecies within the P. canina complex; 3) to assess the validity of newly proposed putative morphospecies; 4) to detect recombination events among members of the P. canina complex; and 5) to evaluate the potential contribution of INAASE characters (Lutzoni et al 2000), and a new method developed here that generates coded characters from a large hypervariable region of 126 sites long within ITS1 (ITS1-HR), toward resolving this species complex.

To achieve these goals, a 1.4-kb fragment at the 5' end of the nuclear large-subunit rDNA (LSU rDNA) and the entire internal transcribed spacer (ITS1, 5.8S and ITS2) were sequenced for 25 putative species (17 described, including P. ulcerata, and eight undescribed species) selected from the Peltigera canina complex and eight additional species from sections Horizontales, Phlebia and Retifoveatae. Each data partition was analyzed separately and together using neighbor joining, maximum likelihood and maximum parsimony. Coded characters derived from ambiguously aligned parts of the alignments (including a hypervariable region within the ITS1 [ITS1-HR]) were added to unambiguously aligned sites for maximum-parsimony searches. Bootstrap support (BS, Felsenstein 1985) and Bayesian Metropolis coupled Markov chain Monte Carlo tree sampling (BMCMCMC, Huelsenbeck 2000, Larget and Simon 1999) were used to estimate levels of confidence associated with phylogenetic relationships revealed by this study. The contribution of INAASE and recoded ITS1-HR characters was evaluated in light of results from the BS and BMCMCMC analyses without these characters.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Taxon sampling – A total of 84 specimens representing 33 putative taxa were sampled. Of these, 25 taxa (71 specimens) belong to the P. canina complex and eight (13 specimens) are members of other sections within the genus Peltigera (Table I). Seventeen of the 25 taxa from the P. canina complex are well established and widely accepted. Affiliation of P. ulcerata (occurring outside Europe and North America) to section Peltigera was anticipated, based solely on morphological features, and therefore was tested in this study. The remaining eight taxa from the P. canina complex, represented here by 17 specimens, are potential species currently under study by T. Goward. Taxa selected for this study cover a wide spectrum of morphological and chemical variation within the P. canina complex. They represent all known members of the complex in Europe and North America and selected members from other continents. Several Peltigera species with tomentose thalli (e.g., P. dilacerata [Gyeln.] Gyeln., P. erioderma Vain., P. fibrilloides [Gyeln.] Vitik., P. pindarensis Awashthi & M. Joshi, P. rufescentiformis [Gyeln.] Dodge, P. patagonica Räsänen., P. soredians Vitik., and P. spuriella Vain.) or with glabrous but acid-deficient thalli (e.g., P. austroamericana Zahlbr. and P. subhorizontalis Gyeln.) occurring outside Europe and North America were not included in this study. Based on those morphological traits, these species very likely are to be part of section Peltigera, however their putative inclusion within section Peltigera remains unconfirmed.

DNA isolation, amplification and sequencing – Well-preserved lichen thalli lacking any visible symptoms of fungal infection were selected for DNA isolation. All external pieces of plant material attached to the targeted part of the thallus were removed under a dissecting microscope. Small thallus fragments from terminal parts of lobes from freshly collected or herbarium specimens were sampled for DNA isolation. The oldest specimen from which DNA was successfully isolated was collected in 1984. Total DNA was isolated with the Purgene Kit (GENTRA Systems) from fragments containing both cyano- and mycobionts, following the manufacturer's protocol for filamentous fungi. DNA concentration was estimated by spectrophotometry or by visual comparison with a positive control ( 100 ladder, concentration 10, 20, 40 Ng) on an ethidium-bromide-stained 1.5% TBE agarose gel and accordingly was diluted with distilled water (usually 10 times) before amplification.

Symmetric polymerase chain reactions (PCR) amplified a 1.4-kb fragment at the 5' end of the nuclear LSU rDNA and the entire ITS region (ITS1, 5.8S and ITS2). The LSU rDNA PCR amplification and both the LSU and the ITS sequencing was done as outlined in Miadlikowska and Lutzoni (2000) and Miadlikowska et al (2002). The amplification reaction for ITS was prepared for a final 50-µL volume containing 32.7 µL of sterile double-distilled water, 5.0 µL of 10x Taq polymerase reaction buffer (Boehringer-Mannheim), 5.0 µL dNTP, 1 µL of 100x Bovine Serum Albumin (BSA; BioLabs), 0.3 µL Taq DNA polymerase (Boehringer-Mannheim), 2.5 µL for each of the 10 µM primers ITS1F or NS24R (5'-TAAAAGTCGTAACAAGGTTT-3') and ITS4 (Gardes and Bruns 1993, White et al 1990), and 1.0 µL of template genomic DNA. PCR was performed on Peltier Thermal Cyclers PTC-200 (MJ Research) under these conditions: denaturation at 95 C for 1 min; 25 cycles at 95 C for 1 min, annelation at 53 C for 45 s, extension at 72 C for 2 min; 14 cycles at the same temperature conditions as above but with time extension of 5 s per cycle. Samples then were held for an additional 10 min at 72 C to complete primer extensions, after which they were kept at 4 C until electrophoresis on 1.5% TBE agarose gel was performed. Sequence fragments were subjected to BLAST searches to verify their identities.

A total of 82 sequences for the ITS and 79 sequences for the LSU were gathered for this study (Table I). Sequences for each molecule were assembled and aligned by using Sequencher 3.0 (Gene Codes). Alignments were optimized by eye and corrected manually when necessary. The secondary structure of the LSU rDNA of Saccharomyces cerevisiae (Larsen et al 1993) was used to verify the LSU alignment and to provide an additional criterion to delimit ambiguously aligned regions (Kjer 1995, Lutzoni et al 2000).

Phylogenetic analyses – All phylogenetic analyses were performed using PAUP* 4.0b 8a (Swofford 2001). Because we were unable to complete LSU and ITS sequences for the exact same set of specimens, these datasets with different numbers of specimens were analyzed: LSU, 79 and 77 specimens; ITS, 82 and 77 specimens; and combined LSU + ITS, 77 specimens.

Datasets were subjected to weighted maximum-parsimony (MP), maximum-likelihood (ML) and neighbor-joining (NJ) searches (Table II). The MP and NJ analyses were carried out on data matrices containing only variable characters. Symmetric step matrices were created for unambiguous portions of the LSU and ITS alignments using the computer program STMatrix 2.1 (written by S. Zoller and available upon request to S.Z. or F.L.), as outlined in Miadlikowska et al (2002). ITS1, ITS2 and 5.8S each were subjected to a specific symmetric step matrix. Gaps were used as a fifth character state for unambiguous portions of the alignment. All ambiguously aligned regions were excluded. However, these regions were recoded with the program INAASE (Lutzoni et al 2000) and then re-integrated into the dataset for neighbor-joining (NJ1) and maximum-parsimony analyses (MP1, MP2, MP4 and MP5). Sixteen of these 20 INAASE-coded characters were subjected to a specific symmetric step matrix implemented during maximum-parsimony searches.

To explore the impact of INAASE characters (five in the LSU and 15 in the ITS datasets) and 24 noncorrelated characters derived from the hypervariable ITS1 region (ITS1-HR) in recovering phylogenetic relationships within the P. canina group, phylogenetic analyses (NJ, MP and ML) were carried out on LSU and ITS datasets, separately and combined, with and without coded characters (Table II). The LSU 79 OTU dataset (NJ1) incorporated an additional sequence for P. ulcerata and P. "latopraetextata", whereas the ITS 82 OTU dataset (NJ2) contained four additional sequences (three for P. lambinonii and one for P. praetextata). The purpose of the NJ1 and NJ2 analyses was to confirm the validity of taxa represented by only one LSU or ITS sequence in the 77 OTU combined dataset. The goal of the separate MP and ML analyses was to test for topological congruence between the two data partitions. PAUP* settings for all weighted maximum-parsimony analyses were the same as in Miadlikowska and Lutzoni (2000). A hierarchical likelihood ratio test (Modeltest 3.04PPC, Posada and Crandall 1998) was used to select models and parameters for NJ2 and ML searches. Each type of datasets containing 77 OTUs was subjected to Modeltest. All maximum-likelihood analyses were done with heuristic searches of 100 random-addition sequences (RAS); 20 000 TBR rearrangements per RAS; MulTrees option in effect; reconnection limit equal 8; and all branches of effectively zero length were collapsed. Constant sites were excluded from all ML analyses.

Branch support, congruence test, sequence comparison and recombination analyses – Branch support for NJ, MP and ML trees was estimated by bootstrap analyses (Felsenstein 1985), using the same parameter settings as for the initial search. To evaluate the robustness of the MP and NJ tree bipartitions, 1000 bootstrap replicates with 2 RAS per bootstrap replicate were performed. For ML1–ML3 analyses, 500 bootstrap replicates were conducted with 2 RAS per bootstrap replicate.

For ML1, ML2, and ML3 datasets, in addition to bootstrapping, posterior probabilities (PP) were calculated as implemented in MrBayes 2.01 (Huelsenbeck 2000). All BMCMCMC analyses were run with four chains simultaneously, each initiated with a random tree, and flat prior; one out of every 20 trees was sampled for 1 000 000 generations with substitution parameters estimated during the search. The first 20 000 sampled trees from each run were discarded before calculating the majority-rule consensus tree to ensure that all chains had converged at the same level. Majority-rule consensus trees were assembled with PAUP* for the remaining 30 000 BMCMCMC sampled trees. Posterior probabilities for topological bipartitions were considered statistically significant when P 0.95.

Within an MP framework, congruence between LSU and ITS data partitions was assessed by inspecting bootstrap scores 70% resulting from the separate MP analyses (Mason-Gamer and Kellogg 1996), as outlined in Miadlikowska and Lutzoni (2000). Conflict between ML trees derived from LSU and ITS, when analyzed separately, was detected when two different relationships (one monophyletic and the other nonmonophyletic) for the same set of taxa were supported by posterior probabilities 0.95.

To detect recombination, four methods, based on patterns of substitution, discordance in phylogenies and phylogenetic incongruence on a site-by-site basis (Posada and Crandall 2001) were used: PLATO (Grassly and Holmes 1997), "Reticulate" (Jakobsen and Easteal 1996), RDP (Martin and Rybicki 2000) and GENECONV 1.81 (Sawyer 1989). Each heterogeneous region of the alignments detected by PLATO was treated as a separate data partition, and conflict between these partitions and the remaining part of the LSU or ITS was tested with a BMCMCMC approach as for the combinability test used for ML trees (see above).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
DNA sequence variation and alignments – LSU sequences were similar in length (1316–1334 nucleotides), whereas the ITS sequences varied remarkably in length, from 503 to 660 nucleotides (ITS1: 158–294; 5.8S: 158–161; ITS2: 181–263). The hypervariable ITS1 region alone, located between sites 111 and 237 of our ITS alignment, showed differences in length ranging from 8 bp for Peltigera lepidophora to 126 bp for one P. membranacea individual.

Separate phylogenetic analyses on LSU and ITS data (NJ1 and NJ2; ML1 and ML2) – For the NJ1 analysis, mean character differences among sequences were used to allow the inclusion of five INAASE characters for a grand total of 116 variable characters (Table II). Peltigera ulcerata 1, included exclusively in the NJ1 analysis, confirmed the validity of LSU sequence for P. ulcerata 2; they form a highly supported monophyletic group (BS = 100%; tree not shown). The relationships among P. "latopraetextata" specimens, including P. "latopraetextata" 1, which was present only in the NJ1 dataset, were unresolved; therefore we were unable to validate sequences for this putative species. NJ2 analysis was performed using the K80 substitution model (Kimura 1980) with these parameters: ti/tv ratio = 2.6996, and equal rates for all sites. Three additional specimens of P. lambinonii (specimens 2–4), included exclusively in the NJ2 search, formed a monophyletic assemblage with P. lambinonii 1. The relationships among P. praetextata specimens, including P. praetextata 4 present only in the NJ2 dataset, were unresolved; therefore we were unable to validate sequences for this species.

The ML1 analysis was performed using the TrN substitution model (Tamura and Nei 1993) with this rate matrix: [A–C] = 1.0000, [A–G] = 3.1119, [A–T] = 1.0000, [C–G] = 1.0000, [C–T] = 5.1928, and [G–T] = 1.0000. Fourteen equally most-likely trees resulted from the ML1 search (ln likelihood = -1010.68249; result not shown). In the ML1 tree, the Peltigera canina complex is defined as a monophyletic and significantly supported group (PP = 1.00; BS = 78%). Most of the resolution and high support values were found at the base and along the "backbone" of the tree, including a sister relationship between the P. canina complex and section Retifoveatae.

The ML2 analysis was performed using the K80 substitution model (Kimura 1980) with ti/tv ratio = 2.6712. Eight equally most-likely trees resulted from this analysis (ln likelihood = -672.00570; tree not shown). Monophyly of the Peltigera canina complex did not receive significant support (PP = 0.78 and BS = 50%). None of the discrepancies between the ML2 analysis of the ITS were in conflict with the ML1 analysis of the LSU. In general, ML analysis of the ITS dataset resulted in a slightly more resolved tree (eight equally most-likely trees for ML2 versus 14 trees for ML1) but with far less support (especially for the "backbone") than the ML analysis of the LSU dataset.

Maximum-likelihood analyses on LSU and ITS combined dataset (ML3) – Because no conflict was detected when examining ML1 and ML2 topological bipartitions that received significant posterior probabilities (0.95), the LSU and ITS rDNA datasets were analyzed together. ML3 analysis was performed using the K80 substitution model (Kimura 1980) with ti/tv ratio = 2.1604. Eight equally most-likely trees part of four islands hit 92 out of 100 RAS were revealed by the ML3 search (ln likelihood = -1790.54212; Fig. 1).

View larger version (42K): [in this window] [in a new window]   FIG. 1. ML3 analysis. One of eight most-likely trees for 64 individuals representing 25 Peltigera taxa from the P. canina species complex (section Peltigera) based on nuclear LSU and ITS rDNA combined dataset (Table II). A total of 13 specimens were selected from eight outgroup species. Bootstrap support values BS 50% are provided below each internode. Posterior probabilities 0.50 are provided above each internode. Internodes with PP 0.95 are shown as thicker lines. Boxes A–G represent main monophyletic groups (containing at least two species) within the P. canina complex. Groups that are circumscribed by continuous lines (boxes A, C, D and E) have posterior probabilities 0.95. Poorly supported groups are delimited with dashed lines. Numbers in small boxes refer to node numbers in Table III. These nodes were selected because they received BS 70% or a posterior probability 0.95 in ML3 or MP5 (Fig. 2). Names in quotation marks indicate putative undescribed species within the P. canina complex

Maximum-parsimony analysis on LSU and ITS combined dataset (MP3–5) – Because no conflict was detected when examining MP1 and MP2 topological bipartitions that received BS support 70%, the LSU and ITS rDNA datasets were analyzed together. MP3 and MP4 analyses were carried out mainly to compare their respective bootstrap support with support values associated with MP5 and ML3 (Table III). We were not able to complete the 1000 replicates for MP3 without limiting the number of TBR swapping. When the re-arrangement limit was set to 50 000 and the reconnection limit equal to eight, the MP3 analysis resulted in 10 509 most-parsimonious trees (tree length = 596.79 steps; CI [excluding uninformative characters] = 0.5957, RI = 0.8840; results not shown). From MP4 analysis, 15 441 most-parsimonious trees (tree length = 936.25 steps; CI [excluding uninformative characters] = 0.6512, RI = 0.8932; results not shown) were obtained. Because we could not reject independence for any of the 24 ITS1-HR characters, all were included in the MP5 search. The MP5 search resulted in 1620 most-parsimonious trees, part of a single island hit 935 out of 1000 RAS (tree length = 1538.91 steps; CI [excluding uninformative characters] = 0.6474, RI = 0.8589; Fig. 2). Topologies from these three MP searches (MP3–5) generally were similar. When different, competing pairwise discrepancies among MP3–5 and ML3 trees never were supported by bootstrap values 70%.

View this table: [in this window] [in a new window]   TABLE III. Variation of bootstrap support values (BS) at selected internodes when comparing MP3-5 to ML3 (bold; FIG. 1). Only internodes with BS or posterior probabilities 50% were compared. Groups and node numbers are shown in FIGS. 1 and 2

View larger version (43K): [in this window] [in a new window]   FIG. 2.  MP5 analysis. One of 1620 most-parsimonious trees for 64 individuals representing 25 Peltigera taxa from the P. canina species complex (section Peltigera) based on nuclear LSU and ITS rDNA combined dataset, including 20 INAASE and 24 ITS1-HR characters (Table III and Appendix 1). A total of 13 specimens were selected from eight outgroup species. Bootstrap support values 50% are provided above each internode, which are shown as thicker lines. Boxes A–E and G represent the same monophyletic groups within the P. canina complex as recognized in ML3 (Fig. 1). Groups that are circumscribed by continuous lines (boxes A–D and G) have BS 70%. Poorly supported Group E is delimited with a dashed line. Numbers in small boxes refer to node numbers in Table III. These nodes were selected because they received BS 70% or a posterior probability 0.95 in MP5 or ML3 (Fig. 1). Names in quotation marks indicate putative undescribed species within the P. canina complex. In the P. degenii (B) group, subgroups I and II correspond to a North American and a European clade, respectively

MP5 provided a high level of resolution and bootstrap support within species groups and, therefore, was the most useful in recognizing monophyletic groups (Fig. 1 versus Fig. 2; ML3 versus MP5; Table III). The P. ponojensis (A) group incorporated specimens representing three taxa: P. monticola, P. ponojensis and the putative P. "scotteri". Peltigera monticola and P. ponojensis each form a monophyletic group (BS = 100%, BS = 71%, respectively). Peltigera "scotteri" 1 might represent a lineage distinct from P. monticola and P. ponojensis.

The P. rufescens (E) group included specimens representing five taxa: three widely accepted species (P. laciniata, P. lepidophora and P. rufescens) and two potentially new taxa (P. "fuscoponojensis" and P. "neorufescens"). Peltigera laciniata, P. lepidophora and P. rufescens (including two specimens initially identified as P. "neorufescens") were recognized as monophyletic entities (BS 98%). Specimens of P. "neorufescens" were found in two different groups; specimens 2 and 3 are closely related to members of P. rufescens, and specimen 1 is sister to P. "fuscoponojensis" (BS = 99%). The phylogenetic affiliation of the latter two specimens to the P. rufescens group is uncertain, while the remaining taxa within this group form a monophyletic entity (BS = 76%).

The Peltigera didactyla (G) group included specimens representing three known species (P. didactyla, P. lambinonii and P. ulcerata) and one intraspecific taxon (P. didactyla var. extenuata). Peltigera didactyla var. extenuata was delimited as a robust monophyletic entity (BS = 100%), sister to other sorediate species that were part of a single highly supported group (BS = 84%). Two of the three specimens of P. didactyla var. didactyla are grouped together, while the third specimen (No. 2) was quite different; its relationship with specimens 1 and 3, as well as with P. ulcerata and P. lambinonii, is uncertain.

The Peltigera degenii (B) group incorporated specimens representing two well-established species (P. degenii and P. membranacea) and the putative P. "scotteri". All specimens of P. membranacea belonged to a single group (BS = 100%). Peltigera degenii was partitioned into two distinct monophyletic groups: a North American P. degenii I group (specimens 2 and 4, including P. "scotteri" 2), and a European P. degenii II group (specimens 1 and 3), both with surprisingly high support (BS = 98% and BS = 99%, respectively).

The Peltigera canina s. str. (C) group incorporated individuals belonging to eight taxa: four known species (P. canina, P. cinnamomea, P. evansiana and P. praetextata) and four potential taxa (P. "boreorufescens", P. "fuscopraetextata", P. "latopraetextata" and P. "pallidorufescens"). When compared to ML3 (Fig. 1), MP3 and MP4, adding 24 coded characters from the ITS1-HR greatly improved the resolution and reduced phylogenetic uncertainty at the interspecific level within this most complex group of species. The monophyletic and highly supported Peltigera canina species (including P. "boreorufescens" 1) and P. praetextata species (including P. cinnamomea 3 and P. "latopraetextata" 3) were sister to each other (BS = 76%). They shared a common ancestor with P. evansiana (BS = 81%). The newly proposed P. "pallidorufescens" and P. "fuscopraetextata" were defined as monophyletic sister taxa (BS = 94%). Their common ancestor was part of the earliest divergence event within the P. canina s. str. group.

The Peltigera cinnamomea (D) group incorporated specimens from four taxa: one known species (P. cinnamomea) and three potential taxa (P. "boreorufescens", P. "latopraetextata" and P. "neocanina"). Two robust monophyletic groups were reconstructed with MP5. One group represents P. cinnamomea, including P. "latopraetextata" 1 (BS = 100%), and the second group represents a putative taxon composed of P. neocanina (specimens 1 and 2) and P. "boreorufescens" 2.

Relationships among Peltigera continentalis, P. frigida and P. kristinssonii are uncertain in MP5. Each seems to represent independent lineages. With MP5, the distinct P. horizontalis + P. elisabethae and P. neckeri groups formed a well-supported monophyletic section Horizontales (BS = 91%).

ITS1-HR sequence variation within and among monophyletic Peltigera taxa – A total of 51 different sequences of the ITS1 hypervariable region (ITS1-HR) were present among 77 Peltigera individuals in this study. Eighteen of these sequences were found in at least two Peltigera individuals, and 33 were unique. The 24 characters we used to characterize these sequences captured the diversity and pattern of variation among the ITS1-HR sequences within the P. canina complex and outgroup taxa included here (Fig. 3).

View larger version (82K): [in this window] [in a new window]   FIG. 3.  Sequences of the ITS1 hypervariable region (ITS1-HR) for 64 individuals representing 25 Peltigera taxa from the P. canina species complex (above wavy line) and 13 individuals from eight outgroup species (below wavy line). When specimens initially were misidentified, only the revised identifications (name after "=" in Fig. 2) and associated specimen numbers were included here. Continuous lines separate major monophyletic groups (A–E, G) within the P. canina and sections (Retifoveatae, Horizontales and Phlebia) within outgroup, as revealed by MP5 (Fig. 2). Dashed lines separate putative Peltigera species, as shown in MP5 tree (Fig. 2). Sequences were aligned within each putative species category, except for P. membranacea (where the alignment was not possible) and for group A (where sequences were aligned among putative species). Sequence length (number of nucleotides) is provided in square brackets to the left of each sequence. Each different ITS1-HR sequence received a different colored symbol and number found in the right side of each sequence. Twenty-four coded characters describing features of these sequences are shown to the right of this figure (see Appendix 1 and "Phylogenetic analyses" in Materials and Methods for a description of these characters)

Contribution of INAASE and ITS1-HR characters to phylogenetic confidence – The addition of 20 INAASE characters brought the number of internodes with bootstrap values 70% from 21 in MP3 to 33 in MP4 (Table III). When 24 ITS1-HR characters were added to the 20 INAASE characters (MP5), the total number of internodes with BS values 70% increased to 48. The number of internodes that received BS values 70% with ML3 bootstrap was 25, compared to a total of 29 internodes obtaining posterior probabilities 0.95. All internodes that received a BS value 70% in MP3 and MP4 also were supported highly in MP5 or ML3. Only MP5 revealed nodes (a total of four) that were exclusive to this analysis and highly supported. There were two nodes that received bootstrap values 70% only in the ML3 analysis, compared to 10 nodes with high BS values exclusively in MP5. Only four nodes with BS values 70% in the MP5 analyses received a BS value 70% in ML3 or a posterior probability 0.95. This is a very small number of nodes compared to ML3 bootstrap with 26 nodes 70% BS that were highly supported by BMCMCMC or other MP bootstrap analyses. Bayesian BMCMCMC was not much better than ML3 bootstrap, with 22 nodes that obtained posterior probabilities 0.95 and were supported with bootstrap values 70% in at least one ML or MP bootstrap analysis.

Testing for recombination in the P. canina species complex – Our results were highly dependent on the method used. RDP and Reticulate did not detect any clusters of incompatible sites within LSU and ITS alignments, suggesting that there was no recombination event among the included sequences. GENECONV showed some evidence for gene conversion and recombination. However, because of the different amount of variation among sequences (some sequences were very similar whereas others were quite distinct), none of the patterns were found to be significant. PLATO detected heterogeneity for two regions within the LSU and ITS alignments. However, no conflict was detected, when comparing topological bipartitions that received significant posterior probabilities (0.95) in ML1 and ML2 with topologies that resulted from the respective ML analyses on the LSU and ITS data partitions defined by PLATO.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Morphological versus phylogenetic species recognition in the Peltigera canina species complex – We evaluated the morphological delimitation of 17 widely recognized species and eight additional putative species from the Peltigera canina complex (section Peltigera) using a combined LSU and ITS rDNA phylogeny. We found a complete concordance between phylogenetically based and morphologically based species circumscriptions for 15 of the 17 accepted species (P. canina, P. cinnamomea, P. degenii [monophyly not supported], P. evansiana, P. frigida, P. kristinssonii, P. laciniata, P. lambinonii, P. lepidophora, P. membranacea, P. monticola, P. ponojensis, P. praetextata, P. rufescens and P. ulcerata; Fig. 2). Peltigera lambinonii and P. ulcerata, each represented in the combined analyses by a single individual, were shown to be monophyletic based on more sequences included in the separate analyses of ITS and LSU datasets, respectively. Although seemingly morphologically homogenous, populations of P. degenii from North America and Europe are genetically and biogeographically divided into two putative cryptic species (Table I, Figs. 2 and 3). Because no type material for P. degenii was included in this study (collected in Japan, Vitikainen 1994) and because of our limited sampling of P. degenii s. l., we cannot assign names to these two putative cryptic species. Results of further systematic studies on P. degenii s. l. will be presented in a separate publication. We found higher genetic variation among species than among individuals within each of these monophyletic groups.

Peltigera praetextata includes morphotypes without phyllidia (P. praetextata 6), as well as morphotypes with phyllidia (specimens 2 and 3). When present, phyllidia are taxonomically significant and a valuable diagnostic character for P. praetextata (see Lindahl 1960), however, this is not the only diagnostic character distinguishing P. praetextata from its sister species P. canina (see Goffinet and Hastings 1994, Vitikainen 1994).

The phenotypic delimitation of P. didactyla conflicts with its phylogenetic circumscription. The commonly accepted delimitation of P. didactyla encompasses P. didactyla var. didactyla and P. didactyla var. extenuata, which differ both on morphological and chemical grounds (Goffinet and Hastings 1995). Peltigera didactyla var. extenuata represents a monophyletic lineage, nonsister to P. didactyla s. str., and therefore should be recognized at the species level. Because of specimen P. didactyla var. didactyla 2, the taxonomic delimitation of P. didactyla s. str. is uncertain. This specimen is anomalous in containing secondary metabolites (gyrophoric acid and methyl gyrophorate) usually present only in members of P. didactyla var. extenuata. A detailed systematic treatment of all taxa belonging to the P. didactyla (G) group, as delimited here, are addressed in a separate publication (Goffinet et al 2003).

Status of the putative new species – As reported by Miadlikowska and Lutzoni (2000) and confirmed by this study, the eight newly proposed Peltigera taxa belong to the P. canina species complex (section Peltigera). Four taxa, P. "fuscopraetextata" (part of the P. canina s. str. group), P. "neocanina" (part of the P. cinnamomea group), P. "neorufescens" (part of the P. rufescens group) and P. "scotteri" (part of the P. ponojensis group) most likely represent four new species that need to be described, based on an extensive sampling of populations (Fig. 2). Detailed field studies on these species already were initiated. Unlike P. "neocanina" and P. "fuscopratextata", not all specimens initially identified as P. "neorufescens" and P. "scotteri" were part of a single well-supported monophyletic group with highly distinct ITS sequences. Despite these misidentifications before this study, these four newly circumscribed phylogenetic species are distinguishable morphologically. Formal descriptions of these new Peltigera species will be the subject of a separate publication in collaboration with T. Goward. The epithets appearing in this paper in quotations and originally published by Miadlikowska and Lutzoni (2000) are herbarium names not intended for eventual effective and valid publication.

The taxonomic status of one of the other four putative taxa (P. "pallidorufescens") is uncertain. Peltigera "pallidorufescens" is sister to P. "fuscopraetextata" but is not well supported. For this reason we prefer to await further data before drawing any taxonomic conclusion. The other three putative taxa (P. "boreorufescens", P. "latopraetextata" and P. "fuscoponojensis") are clearly part of other species within Peltigera and should not be recognized as new species. Some individuals were part of the newly defined species P. "neocanina" and P. "neorufescens", whereas the other specimens were nested in the well-established P. canina, P. cinnamomea, P. degenii I, P. rufescens and P. praetextata species (Fig. 2). The differences among these putative taxa can be subtle, apparently owing to phenotypic plasticity.

Contribution of the ITS1 hypervariable region (ITS1-HR) and INAASE characters to evolutionary studies of the Peltigera canina complex – Variation among ITS1-HR sequences greatly contributed to species delimitation and hence species identification and can be a major asset to future population studies for specific species within section Peltigera. Sequences of ITS1-HR alone were sufficient to identify all existing species of Peltigera currently placed in the P. canina complex, as well as related sections Retifoveatae and Horizontales (Fig. 3). This molecular marker thus provides an exceptional tool for reliable identification of a lichen assemblage with few, or subtle, diagnostic morphological and chemical characters. Given that the amplification and sequencing of ITS1 alone would be sufficient for any studies requiring reliable identifications of members of this complex, it is likely that a simple RFLP analysis of the ITS1 would provide the same level of accuracy as sequences for species identification.

By using the ITS1-HR marker, it also was possible to recognize new undescribed species within the P. canina complex. Variation among these sequences was in agreement with each of the monophyletic morphospecies and permitted the recognition of distinct species even when represented by only one specimen in the combined dataset (e.g., P. continentalis, P. lambinonii and P. ulcerata; Figs. 2 and 3). In one case (P. degenii I and II), ITS1-HR sequences were sufficient to detect individuals belonging to putative cryptic species. Because most individuals of Peltigera in this study were collected in the Northern Hemisphere (North America and Europe), the complete intra- and interspecific ITS1-HR variation within the P. canina complex is probably higher than we have documented. The LSU and ITS nrDNA datasets analyzed separately (including INAASE characters and 24 coded characters from ITS1-HR) could not phylogenetically delimit all morphospecies evaluated here (see Kroken and Taylor 2001). Only when INAASE and ITS1-HR characters were added to the combined analysis (MP5) were we able to reach the level of resolution, support and disentanglement of the P. canina complex shown here (Figs. 1 versus 2; Table III). For example, Peltigera monticola (P. ponojensis group), P. canina, P. praetextata, P. "fuscopraetextata" (P. canina s. str. group) and P. "neocanina" (P. cinnamomea group) were circumscribed and highly supported (BS 70%) as monophyletic species only with the MP5 analysis. These ITS1-HR characters greatly contributed to the determination of interspecific relationships, especially among species in the P. canina s. str. group (e.g., the sister relationships between P. canina and P. praetextata and between P. canina + P. praetextata and P. evansiana). These relationships were corroborated partly by the high degree of similarity between P. praetextata and P. canina, based on selected anatomical and morphological features, such as spore size, thallus thickness, width and height of veins. Only differences in vein height were statistically significant between these two species (Martínez and Burgaz 1996).

The hypervariability of the ITS1-HR was due largely to the presence of microsatellites. Sixteen of the 33 taxa recognized here included at least one microsatellite, 15 of which were found in the P. canina complex. Some of these microsatellites were highly variable within species and will be ideal markers for population studies. Peltigera membranacea, P. "neocanina" and P. rufescens are good examples, because none of the specimens sampled had an identical microsatellite length, even when collected from the same area (e.g., P. "neocanina"; Table I, Fig. 3).

Adding only the 20 INAASE characters to the combined dataset (i.e., without ITS1-HR characters) in the MP4 analyses was sufficient for the total number of well-supported internodes (BS 70%) to be greater than the number of internodes with BS 70% and posterior probabilities 0.95 from ML3 analyses. According to a recent simulation study comparing the statistical performance of ML and MP bootstrapping to a Bayesian MCMCMC approach (Alfaro et al 2003), MP bootstrapping often supports fewer internodes (BS 70%) than ML bootstrap and BMCMCMC. Because at the time we conducted this study we could not accommodate multiple models simultaneously within an ML or BMCMCMC framework, including models that can accommodate INAASE and ITS1-HR coded characters, our MP bootstrap provided a higher level of phylogenetic confidence than ML bootstrap and Bayesian approach. Therefore, the addition of INAASE and ITS1-HR characters to our MP analyses had a greater positive impact on the level of phylogenetic confidence than using more powerful methods (i.e., requiring less data to converge on the correct topology; Huelsenbeck and Hillis 1993, Yang 1996) such as ML and BMCMCMC, but with fewer characters (i.e., without INAASE and ITS1-HR characters).

Ecological preferences and diversification within the P. canina complex – None of the phylogenetic analyses provided strong support for deep relationships within the P. canina complex. With MP5, two main monophyletic groups were reconstructed: the B+C+D (CICADE) and A+E+G (PORUDI) groups. The well-supported CICADE group includes closely related (BS = 79%) P. degenii (B) + P. canina s. str. (C) groups and P. cinnamomea (D) group (BS = 72%), whereas the weakly supported PORUDI group, consists of P. ponojensis (A), P. rufescens (E), and P. didactyla (G) groups (BS = 50%). This major division within the canina complex seems to be correlated with humidity preferences. Species from the CICADE group are mesophytic and subhygrophytic (common in woodlands), and rarely occur in xeric sites (P. canina sometimes grows in subxeric conditions); whereas species from the PORUDI group are xerophytic and, to much lesser extent, mesophytic (see also character 69 in Miadlikowska and Lutzoni 2000). Species from the latter group mostly are heliophilous and common on periodically disturbed soils. Three species with uncertain affiliations within section Peltigera (P. continentalis, P. frigida and P. kristinssonii) and the outgroup species selected for this study are common in mesic habitats. Section Peltigera, which seems to be one of the two most-derived sections within this genus (Miadlikowska and Lutzoni 2000), includes most of the known xeric and heliophilous Peltigera species. Therefore, it would be interesting to determine if they all are derived from a single evolutionary event, from multiple independent origins at various times during the evolution of Peltigera or from multiple independent origins at one specific time during the evolution of this genus.

In general, phenotypic similarity seems to be positively correlated with degree of relatedness among species and species groups within the P. canina complex (e.g., close phylogenetic relationships of P. degenii s. l. with P. membranacea, P. canina with P. praetextata, and P. monticola with P. ponojensis; Fig. 2). Such a correlation, however, does not apply in all cases. For example, P. cinamomea (group D) can be very similar to nonphyllidiate, broad lobed, morphotypes of P. praetextata (group C; Fig. 2).

Our reconstructed phylogeny (MP5 tree) provides a framework for further systematic, biogeographical and evolutionary studies on the Peltigera canina species complex. To gain confidence in the deep relationships among major groups in this section, inclusion of additional molecular characters is needed.

Relationships outside section Peltigera – Phylogenies resulting from this study (Figs. 1 and 2) are the first to be reconstructed for this section, based on two molecular markers. Despite the removal of chemical and morphological characters and the addition of the ITS to the LSU dataset, relationships outside the P. canina complex are in agreement with our previous results (Miadlikowska and Lutzoni 2000). This study significantly confirms the monophyletic delimitation of the section Peltigera (including P. ulcerata; PP 0.95 and BS 70%) and its sister relationship with the monospecific section Retifoveatae (Figs. 1 and 2). By incorporating ITS1-HR and INAASE characters in MP5, the bootstrap support for section Horizontales increased from 68% (in Miadlikowska and Lutzoni 2000) to 91% in the MP5 analysis (Fig. 2).

The recombination process within the Peltigera canina complex – The remarkably high intraspecific and low interspecific levels of morphological/chemical variation, as well as the frequent occurrence of pycnidia in the Peltigera canina complex, compared to other sections, has led to the assumption that interspecific recombination is frequent in this group (e.g., Goffinet and Hastings 1995). Despite this belief, evidence of recombination never was found to be statistically significant with ITS and LSU sequences. Sampling strategies for populations, species and loci designed specifically to detect recombination are needed to establish the importance of this process in the evolution of this complex group of species.

	ACKNOWLEDGMENTS

	FOOTNOTES

 
¹ Corresponding author. E-mail: jolantam{at}duke.edu

² Mailing address: Edgewood Blue, Box 131, Clearwater, British Columbia V0E 1N0, Canada

Accepted for publication April 13, 2003.

	LITERATURE CITED

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Alfaro ME, Zoller S, Lutzoni F., 2003 Bayes or bootstrap? A simulation study comparing the performance of Bayesian Markov chain Monte Carlo sampling and bootstrapping in assessing phylogenetic confidence. Mol Biol Evol 20:255-266[Abstract/Free Full Text]

Darbishire OV., 1926 The structure of Peltigera with especial reference to P. praetextata. Ann Bot 40:727-758

Felsenstein J., 1985 Confidence limit on phylogenies: an approach using the bootstrap. Evolution 39:783-791

Franc N, Kärnefelt EI., 1998 Phylogeny of Xanthoria calcicola and X. parietina, based on ITS sequences. Graphis Scripta 9:49-54

Gardes M, Bruns TD., 1993 ITS primers with enhanced specificity for basidiomycetes—application to the identification of mycorrhizae and rusts. Mol Ecol 2:113-118[Medline]

Goffinet B, Hastings RJ., 1994 The lichen genus Peltigera (lichenized Ascomycetes) in Alberta. Provincial Mus Alberta Nat Hist Occas Papers 21:1-54

———, ———. 1995 Two new sorediate taxa of Peltigera. Lichenologist 27:43-58

———, Miadlikowska J., 1999 Peltigera phyllidiosa (Peltigeraceae, Ascomycotina), a new species from the Southern Appalachians corroborated by ITS sequences. Lichenologist 31:247-255

———, ———, Goward T., 2003 Phylogenetic inferences based on mrDNA sequences support five morphospecies within the Peltigera didactyla complex (lichenized Ascomycota). Bryologist 106:349-364

———, Sérusiaux E, Diederich P., 1994 Le genre Peltigera (lichens) en Belgique et au Grand–Duché de Luxembourg. Belg Journ Bot 127:184-206

Goward T, Goffinet B, Vitikainen O., 1995 Synopsis of the genus Peltigera (lichenized Ascomycetes) in British Columbia, with a key to the North American species. Can J Bot 73:91-111

Grassly NC, Holmes EC., 1997 A likelihood method for the detection of selection and recombination using nucleotide sequences. Mol Biol Evol 14:239–247. (http://evolve.zoo.ox.ac.uk/Plato/main.html)

Groner U, LaGreca S., 1997 The "mediterranean" Ramalina pannizzei north of the Alps: morphological, chemical and nrDNA sequence data. Lichenologist 29:441-454

Grube M, Kroken S., 2000 Molecular approaches and the concept of species and species complexes in lichenized fungi. Mycol Res 104:1284-1294

Gyelnik V., 1927 Peltigera-tanulmányok. (Peltigera-studien). Botanikai Kozlemények 24:122-140 +(33–38)

———. 1933 Clavis et enumeratio specierum generis Peltigerae. Rev Bryol Lichénol 5:61-73

Holtan-Hartwig J., 1993 The lichen genus Peltigera, exclusive of the P. canina group, in Norway. Sommerfeltia 15:1-77

Huelsenbeck JP., 2000 "MrBayes 2.01: Bayesian inference of phylogeny". University of Rochester, U.S.A. (http://morphbank.ebc.uu.se/mrbayes/)

———, Hillis DM., 1993 Success of phylogenetic methods in the four-taxon case. Syst Biol 42:247-264

Jakobsen IB, Easteal S., 1996 A program for calculating and displaying compatibility matrices as an aid in determining reticulate evoluti