THE BRYOLOGIST

AJOURNAL OF BRYOLOGY AND LICHENOLOGY

VOLUME 106 FALL 2003 NUMBER 3

The Bryologist 106(3), pp. 349 364 Copyright ᭧ 2003 by the American Bryological and Lichenological Society, Inc.

Phylogenetic Inferences Based on nrDNA Sequences Support Five Morphospecies Within the didactyla Complex (Lichenized )

BERNARD GOFFINET Department of Ecology and Evolutionary Biology, 75 North Eagleville Road, University of Connecticut, Storrs CT 06268-3043, U.S.A. e-mail: gof®[email protected]

JOLANTA MIADLIKOWSKA Department of Biology, Duke University, Box 90338, Durham NC 27708, U.S.A. and Department of Plant Tax- onomy and Nature Conservation, Gdansk University, Al. Legionow 9, 80-441 Gdansk, Poland; e-mail: jolantam@ duke.edu

TREVOR GOWARD Herbarium, Department of Botany, University of British Columbia, Vancouver, British Columbia V6T 2B1, Canada. Mailing address: Edgewood Blue, Box 131, Clearwater, British Columbia V0E 1N0, Canada

Abstract. The Peltigera didactyla complex comprises species of section Peltigera with laminal and submarginal soredia. Three species (P. didactyla, P. lambinonii, and P. ulcerata) and one atypical variety (P. didactyla var. extenuata) are currently recognized within this complex. Phy- logenetic inferences of the entire Internal Transcribed Spacer region (ITS) and the 5Ј half of the gene encoding the large subunit of the rRNA reveal a robust structure within the complex. Under both the maximum parsimony and the maximum likelihood criterion, P. didactyla is resolved as a polyphyletic entity, whereas P. ulcerata, P. lambinonii, and P. didactyla var. extenuata are delimited as monophyletic entities. Peltigera didactyla var. extenuata appears basal within the group, where- as var. didactyla is nested within a clade that also comprises P. lambinonii and P. ulcerata. The polyphyly of P. didactyla is further characterized by the existence of populations that resemble var. extenuata, but differ by their brownish upper cortex. These populations, all from the boreal zone of Canada, compose a monophyletic group sister to the P. didactyla-P. lambinonii clade. For P. didactyla to satisfy a phylogenetic species concept, the var. extenuata is reinstated at the species level, and a new species, P. castanea, is described. Three populations sampled are characterized by unique sequences that may indicate the presence of additional cryptic taxa within the complex. A key to the accepted species is provided. The presence of P. lambinonii in Australia is con®rmed and P. ulcerata is reported as new for Chile.

Lichenized fungi of the Peltigera Willd. nomically important anatomical characters are few, form foliose thalli with either a green alga or a and most had been much overlooked until recent cyanobacterium, or both. Although some species critical studies (Holtan-Hartwig 1993; MartõÂnez are rather small (monophyllous thalli less than one Moreno 1999; Miadlikowska & Lutzoni 2000). cm in diameter), most species are conspicuous, with Most morphological characters seem to cover broad thalli reaching 20 or more cm in width. The genus ranges of expression. Early workers accommodated is notorious for its taxonomic challenges. Species many of the extreme forms as new taxa, but as are essentially de®ned by a combination of char- regional monographs progressed, students faced acters and rarely by distinct unique traits. Taxo- with populations covering much of the morpholog-

0007-2745/03/$1.75/0 350 THE BRYOLOGIST [VOL. 106 ical gradient preferred recognizing few broadly de- era by Miadlikowska and Lutzoni (2000) based on ®ned species (Thomson 2003) with some of the combined molecular, morphological, and chemical variation partitioned among infraspeci®c taxa data showed that P. lambinonii, P. didactyla s.str., (Lambinon 1969; Ozenda & Clauzade 1970; Thom- and P. didactyla var. extenuata form well supported son 1950). monophyletic groups within section Peltigera. As is the case of many lichenized fungi, Peltig- More recently based on the combined large subunit era is characterized by a diversity of secondary me- (LSU), the entire ITS region of the nuclear ribo- tabolites that accumulate on the outer surface of the somal DNA and coded characters (INAASE and the hyphae. Critical study of these chemicals, mainly ITS1 hypervariable region), Miadlikowska et al. tridepsides or triterpenoids, offered new characters. (2003) suggested that individuals of var. extenuata The presence of some compounds appeared corre- form a monophyletic group, non-sister to popula- lated with that of certain morphological characters tions assigned to var. didactyla. They further found [e.g., P. polydactylon (Neck.) Hoffm. sensu lato; that populations of var. extenuata were placed out- White & James 1985], but such correlations be- side of a monophyletic group comprising var. di- tween chemotype and morphotype are often more dactyla, P. lambinonii, and even P. ulcerata, a spe- dif®cult to discern [e.g., P. aphthosa (L.) Willd. cies restricted to temperate climate of tropical and complex; Holtan-Hartwig 1993]. austral regions. All studied morphospecies, phylo- The lack of extracellular secondary metabolites genetically de®ned as monophyletic, were also has been considered a characteristic of the P. can- characterized by unique sequences of a hypervari- ina group (Vitikainen 1994). Although the recently able ITS1 region (Miadlikowska et al. 2003). These described P. retifoveata Vitikainen (Vitikainen observations led these authors to suggest that var. 1985), a species characterized by the production of extenuata deserves to be recognized at the species several triterpenoids and tridepsides, has recently level. been excluded from the P. canina group (Holtan- In the present study we test taxonomic and phy- Hartwig 1993; Miadlikowska & Lutzoni 2000), this logenetic hypotheses within the P. didactyla com- complex remains heterogenous in terms of its plex, by sampling a 1.4-kb fragment at the 5Ј end chemical composition. Indeed, some populations of of the nuclear large subunit rDNA (LSU rDNA) the sorediate species P. didactyla (With.) Laund. and the entire intergenic transcribed spacer (ITS1, sensu lato are known to produce tridepsides, espe- 5.8S, and ITS2). Speci®cally we tested 1) whether cially if not exclusively, in the soralia. Whereas the phylogenetic hypothesis by Miadlikowska et al. Swinscow and Krog (1988) regarded the occasional (2003) regarding the af®nities of P. didactyla var. occurrence of these compounds as taxonomically extenuata would withstand additional sampling ; 2) insigni®cant, Gof®net and Hastings (1995) found a whether a morphological species concept as applied nearly perfect correlation of the presence of these to apotypic populations from Canada would also compounds and a discrete combination of morpho- withstand a phylogenetic criterion; and 3) whether logical characters. These authors recognized paleo- the additional sampling within the P. didactyla tropical thalli with dark brown mat-forming rhizi- group would affect a species-speci®c pattern of var- nes and with gyrophoric acid and methylgyrophor- iation found in the ITS1 hypervariable region ate in the soralia as a new species, P. lambinonii (ITS1-HR; Miadlikowska et al. 2003). Gof®net. In the Northern Hemisphere, the presence of chemicals seemed correlated with broader lobes MATERIAL AND METHODS with abundantly branched whitish rhizines, and a Morphological study. Two hundred and fourteen col- mesophytic habitat. To accommodate this taxon lections of P. didactyla sensu lato currently held at the they resurrected var. extenuata (Vainio) against the Herbarium of the National Nature Museum of Canada and typical and sympatric variety of P. didactyla. Their about 25 additional specimens collected by TG in British Columbia (UBC) were examined for their morphological taxonomic concept soon gained wide acceptance characters. (e.g., Berger 1996; Bergsten 1999; Esslinger 1997; Chemical analysis. All examined specimens were Halonen et al. 2000; MartõÂnez et al. 1997; MartõÂnez tested for the presence of gyrophoric acid using the stan- Moreno 1999; van den Boom 1998), although skep- dard C test. Additional investigation relied on thin layer ticism prevailed at ®rst (Vitikainen 1994), probably chromatography following the method developed by Cul- berson (1972) and White and James (1985), whereby ac- owing to the existence of occasional specimens etone extracts are eluted in solvent G (toluene-ethyl ace- with the morphology of one taxon but the chemis- tate-formic acid in proportion 139:83:8) and C (toluene- try of the other. Gof®net and Hastings (1995) had acetic acid in proportion 100:15). tentatively attributed such oddities to hybridization Taxon sampling for DNA analysis. For the phyloge- netic analyses we selected 29 individuals from seven Pel- or thallus fusion, though no formal evidence was tigera taxa (Table 1). Twenty-six specimens represented presented (see also Miadlikowska et al. 2003). three recognized species (P. didactyla, P. lambinonii, and A recent phylogenetic study of the genus Peltig- P. ulcerata, including one intraspeci®c taxon±P. didactyla 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 351 AY266031 AY266033 AY266035 AY266037 AY266038 AY266040 AY266042 AY266044 AY266019 AY266021 AY266023 AY266025 AY266027 AY266029 (AY257940) (AY257933) (AY257934) (AY257934) (AY257936) (AY257890) (AY257891) (AY257892) (AY257957) (AY257930) (AY257929) (AY257931) (AY257937) (AY257938) (AY257939) outgroup specimens; [ ] ϭ GenBank accession number Ð Ð Ð Ð AY266032 AY266034 AY266036 AY266039 AY266041 AY266043 AY266045 AY266020 AY266022 AY266024 AY266026 AY266028 AY266030 (AF286803) (AF286777) (AF286779) (AF286778) (AF286804) (AF286806) (AF286809) (AF286808) (AF286810) (AY257941) (AY257958) (AY257932) LSU nrDNA ITS nrDNA ) L ) - ) ) UKE ) ) D UGDA ) ( UBC ) UBC OFFINET ( ( L OFFINET - ) ) H H .G ( ( OD .G ) OFFINET ) ) (L L ) ) ) ) ) ) ) ) - .G HERB ) ) ( HERB LG OFFINET AND UBC ( ( ( ) UBC UBC UBC UBC UBC UBC UBC ( ( ( ( ( ( ( OFFINET ) UBC UBC ) UGDA ) HERB .G ( ( ) LG ( L ( ( H - LG 1050 500 ( .G ( / / ) sequences obtained from GenBank (Miadlikowska & Lutzoni 2000; Miadlikowska et al. 2003); all KTC H ( HERB ( ( Rw 788 Rw ϭ / / / I02 1111 UGDA HERB / / ( ( Z Z / / Gof®net 3851 Goward 98-56 Goward 01-635 Goward 94-548 Goward 01-869 Goward 01-858 Goward 99-41a Goward 95-35 Goward 01-653 Goward & Findlay 83-506 Goward 97-129 Gof®net & Goward 97-289 Goward 81-1718 Huneck 88-151 Tibell 12401 Lambinon 71 Lambinon 74 Lambinon 72 Gof®net 6302 Cieslinski 1296 Butkus 5236 Miadlikowska 5233 Faltynowicz & Miadlikowska 5235 Lesiak & Czyzewska 9202 Marcelli, Ahti & Yano 28333 Marcelli, Ahti & Yano 28385 Gof®net 7126 Lambinon 72 Lambinon 71 Canada, BC, Poland, Rwanda, Canada, BC, Canada, BC, Rwanda, Poland, Zaire, Poland, Canada, BC, Zaire, Canada, BC, Canada, BC, Canada, BC, Brazil, Poland, Canada, BC, Australia, Canada, BC, Canada, BC, Mexico, Brazil, Canada, Alberta, Canada, BC, Poland, Rwanda, Canada, BC, Chile, Mongolia, ] ] ] 1 1 2 P. didactyla P. extenuata P. didactyla 1 2 3 4 5 1 2 3 4 5 6 7 1 2 3 4 2 3 4 1 2 3 Taxon Voucher sp. 1 [ sp. 2 [ sp. 3 [ 1. Voucher specimen information and GenBank accession numbers for 25 LSU and 29 ITS nrDNA sequences included in this study. * ABLE specimen identi®cation based on morphological species concept; ( ) T P. continentalis P. kristinssonii P. kristinssonii P. ulcerata P. ulcerata P. ulcerata Peltigera Peltigera Peltigera ϭ Peltigera castanea P. castanea P. castanea P. castanea P. didactyla P. didactyla P. didactyla P. didactyla P. extenuata P. extenuata P. extenuata * * other sequences were generated by this study. P. extenuata P. extenuata P. extenuata P. extenuata P. lambinonii P. lambinonii P. lambinonii P. lambinonii P. lambinonii * 352 THE BRYOLOGIST [VOL. 106 var. extenuata) and one putative species (hereafter referred sets (ML1: ti/tv ratio ϭ 3.5326, gamma distribution shape to as P. castanea) from boreal Canada. All taxa are char- parameter ϭ 0.3804; ML2: ti/tv ϭ 3.4563, gamma distri- acterized by orbicular laminal or submarginal soralia and bution shape parameter ϭ 0.4430); the HKY85 (Hasegawa were shown to belong to the P. didactyla group, section et al. 1985) nucleotide substitution model was selected for Peltigera (Miadlikowska & Lutzoni 2000; Miadlikowska the LSU (ML3) data set and the combined ITS ϩ LSU et al. 2003). Two species, P. continentalis and P. kristins- (ML4) data set (ML3: A ϭ 0.2750, C ϭ 0.2085, G ϭ sonii, from the section Peltigera, were chosen as outgroup 0.2812, T ϭ 0.2353, ti/tv ϭ 1.4037, gamma distribution taxa for all phylogenetic analyses, based on phylogenetic shape parameter ϭ 0.016; ML4: A ϭ 0.2694, C ϭ 0.2112, reconstructions by Miadlikowska and Lutzoni (2000) and G ϭ 0.2694, T ϭ 0.2500, ti/tv ϭ 2.4208, gamma distri- Miadlikowska et al. (2003). bution shape parameter ϭ 0.0096). Maximum likelihood Two species, P. lambinonii and P. ulcerata, are restrict- analyses were implemented as heuristic searches with 100 ed to tropical regions and the temperate zone of the South- random-addition-sequence replicates, TBR branch swap- ern Hemisphere; these taxa were represented by ®ve and ping, MulTrees option in effect, saving all trees and col- three populations, respectively. The cosmopolitan P. di- lapsing branches with maximum branch length equal to dactyla was represented by six specimens morphological- zero. ly consistent with P. didactyla var. didactyla, and eight Constant sites were removed from all maximum parsi- specimens of var. extenuata. The collection from Mexico mony analyses. The unambiguously aligned parts of the (Peltigera sp. 1, Table 1), composed mainly of mono- LSU and ITS nrDNA alignments were subjected to sym- phyllous plane to deeply concave thalli, was tentatively metric step matrices constructed as outlined in Miadli- assigned to P. didactyla sensu stricto, although some mor- kowska et al. (2002). The maximum parsimony analysis phological features (e.g., glabrous upper surface, submar- of the ITS data set (MP1 and MP2) involved three sepa- ginal soralia on young lobes, few simple to somewhat rate step matrices corresponding to the ITS1, 5.8S, and fasciculate rhizines restricted to central portion) may be ITS2 regions. The aligned sites from the LSU nrDNA considered atypical for this taxon. One individual of P. (MP3) were subjected to one speci®c step matrix. didactyla var. didactyla (namely Peltigera sp. 3; Table 1) Phylogenetic signal from ambiguously aligned portions collected in Poland possesses some atypical morphologi- of the alignments was integrated into maximum parsi- cal features (see discussion) including the presence of tri- mony analyses without violating positional homology, us- depsides in the thallus. The specimen of P. extenuata 2is ing the program INAASE 2.3b (Lutzoni et al. 2000). The a young thallus with shell-shaped lobes, dif®cult to distin- weights for substitutions in INAASE were set as follow- guish from P. didactyla based solely on morphological ing: transitions ϭ 1.0, transversions and indels ϭ 2.0. For features. The sample ``P. extenuata 1'' has a broadly lobed maximum parsimony analyses, gaps from the unambigu- thallus with thick tomentum on its the upper surface and ous portions of the alignments were treated as a ®fth char- lacks soralia; however, the lower side of the thallus has acter state. All maximum parsimony analyses (MP1±4) pale ¯occulent rhizines typical typical of var. extenuata. were performed as heuristic searches with 1,000 random- Molecular data. Genomic DNA was obtained from addition-sequence replicates, TBR branch swapping, fresh samples and herbarium specimens. DNA isolation, MulTrees option in effect, saving all trees and collapsing symmetric PCR ampli®cations, asymmetric PCR sequenc- branches with maximum branch length equal to zero. ing products puri®cation, and automated sequencing were Branch support for MP trees was estimated by bootstrap performed as explained in Miadlikowska and Lutzoni analyses (BS) (Felsenstein 1985) with full heuristic (2000) and Miadlikowska et al. (2002). The LSU and ITS searches, 1,000 bootstrap replicates and 2 random-addi- nrDNA sequences were subjected to BLAST searches for tion-sequence per bootstrap replicate. For ML analyses, the veri®cation of their identity. They were assembled us- posterior probability (PP) for each node using the Bayes- ing Sequencher 4.1 (Gene Codes), and aligned with ian Metropolis coupled Markov chain Monte Carlo sam- MacClade 4.01 (Maddison & Maddison 2001). Delimita- pling (B/MC3) method was calculated as implemented in tion of the internal spacers and the 5.8S gene was obtained MrBayes 2.01 (Huelsenbeck 2000). All B/MC3 analyses by comparison with complete ITS sequences from closely were run with four chains simultaneously, each initiated related taxa within the Peltigerineae (Gof®net & Goward with a random tree and ¯at prior. One of every 20 trees 1998; Goward & Gof®net 2000; Miadlikowska et al. was sampled for a total of 1,000,000 generations with 2003). DNA substitution parameters estimated during the search. Phylogenetic analyses. Phylogenetic analyses were The majority rule consensus tree was computed with performed using maximum likelihood (ML) and maxi- PAUP* on the last 15,000 trees out of 50,000 sampled mum parsimony (MP) optimization criteria as implement- with B/MC3. Bipartitions were considered statistically sig- ed in PAUP*4.0b4a (Swofford 1998). We obtained LSU ni®cant when PP values were Ն95%. Congruence between sequences for only one of the ®ve samples of P. lambi- the ITS partition and the LSU partition for 25 OTUs was nonii, namely P. lambinonii 1. Therefore, for the purpose tested by inspecting BS values Ͼ70% on topologies de- of combinability, phylogenetic analyses were carried out rived from separate MP analyses (Mason-Gamer & Kellog on two ITS data sets: 1) phylogenetic analyses on 29 ITS 1996) as explained in Miadlikowska and Lutzoni (2000) sequences (ML1 and MP1) including P. lambinonii 2±5 and PP values Ͼ95% on topologies derived from separate and 2) phylogenetic analyses on 25 ITS sequences (ML2 ML analyses as outlined in Kauff and Lutzoni (2002). and MP2) without P. lambinonii 2±5. Maximum likeli- Con¯icts between data partitions would be considered sig- hood and maximum parsimony searches were implement- ni®cant only if two different relationships (one monophy- ed on 25 LSU sequences (ML3 and MP3) and on the letic, the other non-monophyletic) for the same set of taxa combined ITS ϩ LSU data set (ML4 and MP4) for the were supported by BS Ն70% or PP Ն 95%. same 25 individuals. Because the placement of one specimen, P. castanea 2, Using the Hierarchical Likelihood Ratio Test, as imple- was in con¯ict based on the separate analysis of ITS and mented in Modeltest 3.04 (Posada & Crandall 1998), K80 LSU data sets (part of the monophyletic P. castanea group two-parameter nucleotide substitution model (Kimura versus P. didactyla s.st. group, respectively) exclusively 1980) with equal base frequencies was selected for both when using the Bayesian criterion (PP Ն 95%), we were the 29 OTUs (ML1) and the 25 OTUs (ML2) ITS data concerned about the reliability of the signi®cant posterior 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 353

FIGURE 1. Uncorrected pairwise differences among ITS nrDNA sequences included in phylogenetic analyses (MP and ML). Above diagonal: total number of differences including optimal number of changes within ambiguously aligned regions as estimated with INAASE. Indels with multiple consecutive gaps were counted as one change. A ``ϩ'' sign indicates that the given number of differences can be higher due to the missing sites within ambiguous regions. Below diagonal: character differences among unambiguously aligned regions only. Boxes indicate pairwise sequence differ- ences within recognized taxa, including two newly proposed species, P. extenuata and P. castanea. Shadow areas indicate pairwise sequence differences among specimens with uncertain taxonomic af®liation (Peltigera sp. 1, 2, and 3) and their putatively closely related species (P. extenuata and P. didactyla, respectively) as revealed from the ML1 (Fig. 3A), MP1 (Fig. 5), ML4 and MP4 (Fig. 4) analyses. probability of the very short internode (length of internode in length for all P. lambinonii representatives and 0.0005 substitutions per site) supporting the P. didactyla similar in their range within P. didactyla s.st., ex- s.st. group in the ML3 tree. In order to test if this PP value was overestimated due to chance, we generated ten non- cept two specimens (namely Peltigera sp. 1 and sp. parametric bootstrap data matrices derived from the LSU 2; Table 1) with much longer ITS1. All specimens using Seqboot as implemented in Phylip 3.5c (Felsenstein; of P. extenuata have almost identical lengths of the http://evolution.genetics.washington.edu/phylip/get- ITS1 and ITS2 regions. me.html). Bayesian analysis with the same search options as for the original LSU data set (outlined above) was com- The number of differences in nucleotide com- pleted on each of the pseudoreplicates. Topologies and position of the ITS between taxa is distinctly higher posterior probabilities for all internodes were recorded and than among specimens from the same taxon. Within compared. If a PP value of Յ95% was obtained from more all putative species, except one, P. ulcerata, there than one analysis, we could not reject the null hypothesis that the initial high posterior probablity was due to chance is only slight variation among ITS sequences, in- and, therefore we could conclude that the apparent con¯ict cluding ambiguously aligned parts of the alignment between the two partitions was not signi®cant. (Fig. 1). A total number of changes for the ITS sequences, including ambiguous regions, ranged RESULTS from zero to two within P. extenuata; zero to four ITS and LSU characters. The ®nal alignment among atypical populations from British Columbia for the 29 ITS and 25 ITS sequences consisted of (i.e., P. castanea); and from four to ®ve changes 656 sites. Twenty-®ve ambiguously aligned regions within P. didactyla s.st. and P. lambinonii, respec- were delimited, resulting in the exclusion of 210 tively. The most variable ITS was found among sites giving a total of 446 sites (353 constant and members of P. ulcerata, which differ from each 93 variable) included in the ML1, ML2, and ML4 other by a total of 13 to 23ϩ changes, out of which analyses. For MP1, MP2, and MP4, the 353 con- six to 13 are within the unambiguous parts of the stant sites were excluded but the 25 ambiguous re- alignment (Fig. 1). Peltigera extenuata is separated gions provided 25 additional INAASE coded char- from P. didactyla by 40±44 differences (17±21 acters for a grand total of 118 non-constant char- within unambiguous portions of the alignment). acters included in the MP analyses. Out of these, Peltigera castanea shares the least number of 93 and 86 characters were parsimony informative changes, 16±51, with Peltigera sp. 1 and sp. 2; only in the MP1 and MP2 analyses, respectively. six to 20 of them occur in the non-ambiguous parts The ITS sequences varied in total length from of the alignment. Among all taxa included in this 560 to 589 nucleotides (ITS1: 208±240; 5.8S: 158± study, P. castanea is most similar to Peltigera sp. 159; ITS2: 188±207) among the sequences included 1 and sp. 2 (at least 16±18 changes). The most dis- here. The ITS1 and ITS2 sequences were identical tant species exhibiting the greatest number of total 354 THE BRYOLOGIST [VOL. 106

FIGURE 2. Uncorrected pairwise differences among LSU nrDNA sequences included in phylogenetic analyses (MP and ML). Above diagonal: total number of differences including optimal number of changes within ambiguously aligned regions as estimated with INAASE. Indels with multiple consecutive gaps were counted as one change. Below diagonal: character differences among unambiguously aligned regions only. Boxes indicate pairwise sequence differences within recognized taxa, including two newly proposed species, P. extenuata and P. castanea. Shadow areas indicate pairwise sequence differences among specimens with uncertain taxonomic af®liation (Peltigera sp. 1, 2, and 3) and their puta- tively closely related species (P. extenuata and P. didactyla, respectively) as revealed from the ML4 and MP4 (Fig. 4) analyses and between the specimen of P. castanea 2 and its putatively closely related species (P. didactyla)as revealed from the ML3 (Fig. 3B) analysis. changes (34 to 70ϩ) compared to the remaining was identical with specimen 1 of P. didactyla and ingroup species is P. ulcerata. differs only by a single change from other members The ®nal alignment for the 25 LSU sequences of P. didactyla (Fig. 2). consisted of 1,334 sites. Two ambiguously aligned Phylogenetic relationships. When analyzing regions were determined, resulting in the exclusion the 29 OTUs ITS data set alone, a single most like- of 27 sites giving a total of 1,307 sites (1,267 con- ly tree was obtained from the ML1 (ln likelihood stant and 40 variable) included in the ML3 and ϭϪ1216.38158; Fig. 3A), and three most parsi- ML4 analyses. For MP3 and MP4, the 1,627 con- monious trees resulted from the MP1 (tree length stant sites were excluded but the two ambiguous ϭ 480.59 steps; Fig. 5). The ML1 and MP1 topol- regions provided two additional INAASE coded ogies are identical and both show ®ve Peltigera characters for a grand total of 42 non-constant char- species, including the two putative species, P. ex- acters included in MP analyses. Out of these, 21 tenuata and P. castanea, as monophyletic and sig- were parsimony informative. ni®cantly supported (PP Ͼ 95% and BS Ͼ 97). In A pattern of variation similar to that observed both analyses, P. extenuata is well separated from among the ITS sequences was present within and P. didactyla, forming a sister group to a clade in- among taxa for the LSU sequences. However, the corporating the rest of the studied taxa. Relation- number of changes among species was generally ships among taxa within this major clade are well lower compared to the ITS data (Fig. 2). The var- supported only in MP1 due to the phylogenetic sig- iation within unambiguous parts of LSU sequences nal provided by coded INAASE characters (Figs. within species ranged from zero to two for P. di- 3A & 5). Peltigera didactyla and P. lambinonii dactyla, zero to four for P.extenuata, zero to seven share a common ancestor (BS ϭ 76%) and are the for P. castanea and three to eight for P. ulcerata. closest relatives to a new species, P. castanea. Pel- The highest total number of changes, when includ- tigera ulcerata is sister to the P. didactyla ϩ P. ing ambiguous regions, occurred among P. ulcerata lambinonii ϩ P. castanea clade (PP ϭ 100%; BS (six to 11) and P. castanea (two to eight) speci- ϭ 93%). Af®liation of three Peltigera individuals, mens. Among-species variation was similar for all Peltigera sp. 1±3, remains uncertain. ML2 and the taxa including the outgroup species, except P. MP2 analyses on the 25 individuals (results not ulcerata. Peltigera ulcerata was the most distant shown) were carried out mainly for combinability species, exhibiting the greatest difference in nucle- purposes, however, the topologies and internode otide composition with P. extenuata (20±21). An supports were similar to the results obtained from unexpectedly high degree of similarity (zero to one the ML1 and MP1 searches. total change) was found between sequences of P. Phylogenetic analyses on the LSU nrDNA data didactyla and P. castanea 2. Peltigera castanea 2 set alone, ML3 and MP3, revealed two equally 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 355

FIGURE 3. Phylogenetic relationships among individuals from three known Peltigera species (rounded-cornered boxes), two putative new species, P. extenuata (former P. didactyla var. extenuata) and P. castanea (square-cornered boxes), and the outgroup taxa (P. continentalis and P. kristinssonii) as revealed by maximum likelihood analyses of the ITS (A) and LSU (B) data sets alone. PP values Ն95% are shown above internodes. A star indicates con¯icting placement of P. castanea 2 as revealed from ML1 and ML3 analyses. Ð A. ML1 analysis. Sequences of P. lambinonii 2±5, which are included only in the ML1 and MP1 analyses, are shown in bold. BS values Ն70% on MP1 are shown below internodes. Thicker lines indicate internodes with PP Ն 95% on ML1 and BS Ͼ 70% on MP1 and MP2. Vertical bars indicate internodes with PP Ͻ 95% on ML1 but BS Ն 70% on MP1 and MP2. Ð B. ML3 analysis. BS values Ն 50% are shown below internodes. Thicker lines indicate internodes with PP Ն 95% on ML3 and BS Ͼ70% on MP3. Vertical oval indicate internodes with PP Ն 95% on ML3 but BS Ͻ 70% on MP3. most likely trees (ln likelihood ϭϪ2225.63757; dence. Both reconstructions, the placement of P. Fig. 3B) and 296 equally most parsimonious trees castanea 2 within the monophyletic P. castanea (tree length ϭ 126.37 steps; results not shown) re- group in ML1 (Fig. 3A) and ML2 (results not spectively. Only P. extenuata and P. ulcerata were shown) and within the otherwise monophyletic P. reconstructed as monophyletic entities, signi®cantly didactyla in ML3 (Fig. 3B), were signi®cantly sup- supported by both PP and BS values. Note that in ported (PP Ͼ 95%). this analysis, P. lambinonii is represented by a sin- Our test showed that the internode leading to the gle exemplar. Peltigera castanea 2 was nested monophyletic P. didactyla with P. castanea 2 within the P. didactyla sensu stricto group in both (ML3; Fig. 3B) obtained signi®cant support (PP Ն ML3 and MP3 reconstructions. The monophyly of 97%) in only three out of the 10 B/MC3 analyses this composite clade was, however, only well sup- on the LSU pseudoreplicates. This internode was ported by a high posterior probability value (PP ϭ not present at all in the 50% majority rule consen- 98% on ML3; Fig. 3B). Relationships among spe- sus trees revealed by the seven remaining pseudo- cies are mostly not resolved based on the LSU data replicate analyses. Other internodes signi®cantly set alone. supported in the ML3 tree (Fig. 3B) were recovered Although the LSU data set alone provided a low- with signi®cant support values (PP Ն 95%) from er level of resolution and statistical con®dence for each of the 10 LSU pseudoreplicates, except one intra- and interspeci®c relationships (B/MC3 and pseudoreplicate, where P. extenuata was delimited BS support) compared with the separate ITS data as a monophyletic entity excluding specimen 4. All set, a con¯ict was detected only when using the these internodes, except one (sister relationship of 95% PP value for estimating phylogenetic con®- Peltigera sp. 2 and sp. 1) were also highly sup- 356 THE BRYOLOGIST [VOL. 106

FIGURE 4. Phylogenetic relationships among 25 individuals from three known Peltigera species (round-cornered boxes), two putative species P. extenuata (former P. didactyla var extenuata) and P. castanea (square-cornered boxes), and the outgroup taxa (P. continentalis and P. kristinssonii) as revealed by maximum likelihood analysis (A) and maximum parsimony analysis (B) on the combined ITS and LSU nrDNA data set. Ð A. ML4 analyses. PP values Ն 95% are shown above internodes (thicker lines). Ð B. MP4 analyses ]CI (excluding uninformative characters) ϭ 0.8637, RI ϭ 0.9271]). BS values Ն70% are shown above internodes (thicker lines). ported by ML and MP bootstrap values (BS Ͼ resolved and more robust when using a parsimony 70%). This simple test indicated that the high PP criterion that allowed us to incorporate INAASE value obtained with the LSU data set for the con- coded characters (Fig. 4B). A non-sister relation- ¯icting internode (P. didactyla with P. castanea 2) ship between P. didactyla var. didactyla and P. di- was due to chance. Because the ITS and LSU data dactyla var. extenuata was revealed from both com- partitions were congruent based on the bootstrap bined analyses. Peltigera castanea shares a com- 70% criterion we combined the 25 sequences and mon ancestor with P. didactyla sensu stricto and P. analyzed them simultaneously with maximum like- lambinonii group (BS ϭ 84%), and represents a lihood (ML4) and maximum parsimony (MP4). well de®ned species within the P. didactyla group. The single most likely tree resulting from ML4 Peltigera ulcerata is closely related to the P. di- analysis (ln likelihood ϭϪ3524.01324; Fig. 4A) dactyla sensu stricto ϩ P. lambinonii ϩ P. castanea was identical to the single most parsimonious tree monophyletic group. Phylogenetic placement of P. obtained from the MP4 analysis [tree length ϭ lambinonii, which is represented in the combined 602.53 steps, CI (excluding uninformative charac- data set by only a single sequence, is the same as ters) ϭ 0.8637, RI ϭ 0.9271; Fig. 4B] and the most in the ML1 (Fig. 3A) and MP1 (Fig. 5) trees. There parsimonious tree obtained from the MP2 analysis are three specimens, named Peltigera sp. 1±3, with (results not shown). All tested species, including P. uncertain taxonomic status and phylogenetic af®li- extenuata and P. castanea, are monophyletic and ations. signi®cantly supported (PP ϭ 100% and BS Ն Sites between position 92 and 147 of the ITS 97%). The relationships among species are better alignment were extremely variable in length and 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 357

FIGURE 5. Comparison between species delimitation based on the phylogenetic relationships among 29 individuals as revealed from the analysis on ITS data set alone (MP1; revealed also from ML1, ML4, and MP4 analyses) and sequence variation within the ITS1 hypervariable region (sensu Miadlikowska et al. 2003) included in the MP analysis as one out of 25 INAASE coded characters. Strict consensus tree of three maximum parsimonious trees (CI [excluding uninformative characters] ϭ 0.8973, RI ϭ 0.9517). BS values Ͼ 50% are shown above each internode. Thicker branches indicate BS Ͼ 70%. Sequence from the ITS1 hypervariable region is provided for each individual following a character state resulting from the INAASE coding. The number in square brackets indicates the length of the sequence (number of nucleotides). Boxes delimit Peltigera species recognized here and their ITS1 marker sequences; square-cornered boxes represent the newly delimited species, P. extenuata and P. castanea. Sequences of the ITS1 hypervariable region are aligned within each species. nucleotide composition across species, but relative- within the P. didactyla complex. Phylogenetic in- ly conserved and unique within each delimited spe- ferences suggest: 1) that most individuals assigned cies (Fig. 5). Variation in length of the ITS1 hy- to P. didactyla var. extenuata form a monophyletic pervariable region (sensu Miadlikowska et al. group distinct from P. didactyla var. didactyla;2) 2003) among the recognized members of the P. di- that most populations of the latter compose a dis- dactyla group ranged from 15 (P. lambinonii)to41 tinct monophyletic group; 3) that var. didactyla is (Peltigera sp. 1) nucleotides, whereas this region more closely related to P. lambinonii and other so- was much longer, 51±57 nucleotides, in the out- rediate taxa than to var. extenuata; 4) that apotypic group taxa. This hypervariable ITS1 region was populations from (oro-)boreal zones of Canada identical for all specimens of P. extenuata and P. share a common recent ancestor; and 5) that they lambinonii and almost identical (single nucleotide are sister to a clade comprising P. didactyla, P. ul- insertion) for individuals from P. didactyla s.st. cerata, and P. lambinonii. Variation among se- Peltigera ulcerata, and P. castanea show more var- quences of the ITS1 hypervariable region con- iation among the sequences of this region. The se- ®rmed each of the ®ve monophyletic morphotaxa, quences of Peltigera sp. 1, 2, and 3 are different and revealed the genetic distinctness of samples from their respective closely related species (P. ex- Peltigera sp. 1±3 (Fig. 4). In summary, P. didactyla tenuata and P. didactyla s.st.) and each of them is sensu Vitikainen (1994) is polyphyletic due to the unique. nested position of P. ulcerata and P. lambinonii. For taxonomic units to represent natural groupings, DISCUSSION and hence satisfy the monophyletic criterion, the Analysis of the variation observed among se- concept of P. didactyla must be either broadened quences of the ITS and the LSU loci of the rDNA to accommodate P. ulcerata and P. lambinonii,or tandem repeat reveals a distinct genetic structure narrowed to allow for retention of the latter species. 358 THE BRYOLOGIST [VOL. 106

In the latter case, var. extenuata and the material chen-forming fungi in general and within the P. from boreal parts of Canada must be accorded spe- canina species complex in particular (e.g., Miadli- cies rank. kowska et al. 2003). Because our test was designed The level of divergence between var. didactyla to evaluate the level of con®dence obtained from and other putative taxa is similar to, if not greater the Bayesian analysis on the separate LSU data set, than those observed among other closely related which revealed considerable uncertainty about species of Peltigera, e.g., within the P. aphthosa monophyletic delimitation of P. didactyla s.st. in- complex (Gof®net & Bayer 1997; Goward & Gof- cluding P. castanea 2 (non-monophyly resulted ®net 2000), among members of the section Hori- from Bayesian analysis on seven out of 10 LSU zontales (Gof®net & Miadlikowska 1999), or pseudoreplicates), we concluded that the PP value among members of the P. canina group (Miadli- for this con¯icting node was due to chance (Fig. kowska et al. 2003). The incongruence between our 3B). Furthermore this is the only internode in the initial identi®cation and the phylogenetic af®nities ML3 tree that obtained signi®cant PP support but of some populations (Peltigera sp. 1 to 3 in Figs. low (Ͻ70%) BS values resulting from both maxi- 3±4) may be seen as evidence that falsi®es the hy- mum likelihood and maximum parsimony analyses pothesis of multiple species composing this mor- (Fig. 3B). Our results are in agreement with recent phological complex. However, given their genetic simulation studies showing that in some cases the divergence from other population clusters, we in- posterior probability is unreasonably high on very terpret these populations as representatives of ad- short but wrong branches (Alfaro et al. 2003; Su- ditional species as yet undescribed (see below). zuki et al. 2002). Suzuki et al. (2002) suggested The monophyly of P. extenuata, P. castanea, P. that, in such cases, bootstrap probabilities are more lambinonii, P. ulcerata, and P. didactyla sensu suitable for assessing the reliability of phylogenetic stricto is well supported by ITS data alone or in trees than posterior probabilities. combination with the LSU sequences, regardless of Broadening the concept of P. didactyla would the optimality criterion used. The data also offer require abandoning P. lambinonii and P. ulcerata, maximum support for the sister group relationships which are morphologically clearly distinct (Gof®net between P. extenuata and the remaining ingroup & Hastings 1995). Finally, the populations com- taxa, but relationships among the latter are ambig- posing the two most basal clades within the com- uous (Figs. 3±4) . When analyzed alone, sequences plex can be de®ned by morphological characters, of the LSU locus provide similar support for P. even if the differences may appear subtle at best, extenuata, P. ulcerata, and P. didactyla, but no at least to the untrained eye. support for the monophyly of P. castanea. It should The lack of support for branches de®ning the re- be noted, however, that topological incongruence lationships among the clades within the P. didac- between the two character sources do not seem sig- tyla complex (Fig. 3A; except for the sister-group ni®cant as alternative relationships are not each relationship of P. extenuata to the rest of the in- characterized by high support values. This is not group) may suggest that we are dealing with a sin- surprising considering that the LSU gene is consid- gle taxon, with highly partitioned genetic diversity. ered to accumulate mutations at a lower rate than Although this is clearly a possibility, and actually the non-coding intergenic spacer region, hence little the null hypothesis, we argue that the pattern in homoplasy is needed to break down a monophyletic molecular differentiation overlaps with trends in group. The only exception to this pattern concerns morphological changes. It should be noted that the the af®nities of the sample P. castanea 2, which is lack of support characterizes analyses from which resolved as part of the P. didactyla sensu stricto characters of ambiguous homology have been ex- clade based on LSU characters (Fig. 3). This close cluded (Fig. 3A±B, and Fig. 4A), but that the in- relationship between P. castanea 2 and members of tegration of these characters, following the methods P. didactyla sensu stricto was also indicated by the developed by Lutzoni et al. (2000), results in ad- high similarity of their LSU sequences (Fig. 1). ditional phylogenetic signal and increased support One possible explanation for this relationship could values for these branches (Figs. 1 & 4B) without involve an incomplete lineage sorting for P. cas- affecting the topology (results not shown). Finally, tanea and P. didactyla detected for the LSU locus the lack of strong support for most cladogenic (Kroken & Taylor 2001) and not seen in the ITS- events within the complex may simply re¯ect rapid based phylogeny where the two putative taxa were speciation events, with much of the molecular dif- well delimited and supported (Fig. 3A). We also ferentiation occurring subsequently. Our data cer- considered recombination as the potential factor for tainly provide strong support for the monophyly of this grouping (e.g., Posada 2000; Guttman 1997), the various ``terminal'' clades, and the number of however the occurrence of interspeci®c hybridiza- uncorrected pairwise differences (patristic distance) tion has not been conclusively demonstrated in li- compare well with those reported among other spe- 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 359 cies in Peltigera (Miadlikowska et al. 2003). We ported from scattered localities throughout predict that additional characters from other loci (Berger 1996; Bergsten 1999; Esslinger 1997; Hal- will resolve and provide strong support for rela- onen et al. 2000; Martinez et al. 1997; Martinez tionships among these species. Moreno 1999; Miadlikowska 1999; van den Boom Based on these arguments, we propose to rein- 1998). EmmanueÈl SeÂrusiaux (LG; pers. comm.) col- state P. didactyla var. extenuata at the species rank lected the species in Papua New Guinea. and recognize the specimens from Canada as a dis- tinct species. Hence, we recognize ®ve species 2. PELTIGERA ULCERATA MuÈll. Arg., Flora 63: 261. within the Peltigera didactyla complex. 1880. TYPE: ``Prope Apiahy Brasilae merid. Cres- cit: Puiggari n. 1023'' (G, not seen). 1. PELTIGERA EXTENUATA (Vainio) Lojka, - oth. Univ., Fasc. V, no. 222. 1886. Chemistry. Methyl gyrophorate and gyrophoric var. extenuata Vainio, Medeland. Soc. acid are present in the soralia. According to Viti- Fauna Fl. Fennica 2: 49. 1878. TYPE: FINLAND. Ta- kainen (pers. comm.) the type was too small to vastia australis, Asikkala, Kaitas, SileÂn & Norrlin (H, check for a C reaction. None of the specimens we lectotype). Peltigera canina var. extenuata Norrlin, No- have examined lack the above tridepsides, and we tiser SaÈllsk. Fauna Fl. Fennica FoÈrhandl. 11: 178. 1870. (nomen nudum). are not aware of any mention of chemical poly- Peltigera canina [subsp.] extenuata (Vainio) Vainio, Med- morphism within this species. del. Soc. Fauna Fl. Fennica 6: 129. 1881. Differentiation. This species is easily identi®ed Peltigera didactyla (With.) Laund. var. extenuata (Vainio) by submarginal soralia and glabrous upper cortex. Gof®net & Hastings, Lichenologist 27: 48. 1995. Fertile populations occur throughout the range. A Note. In his exsiccati, Lojka was the ®rst to formal description can be found in Galloway recognize Vainio's var. extenuata at the species lev- (1985) and Swinscow and Krog (1988); the latter el, but did not provide a rationale for his new com- authors also provide an illustration of the species. bination. However, neither his nor Vainio's con- Ecology. In Africa, P. ulcerata grows on ``soil cepts were followed by later lichenologists. The va- or over mosses, on earth banks, rocks and fallen riety was recently resurrected by Gof®net and Has- tree trunks in montane forests and the ericaceous tings (1995). zone...'' (Swinscow & Krog 1988) whereas in Chemistry. Methyl gyrophorate and gyrophoric New Zealand it occurs on soil or among mosses acid are often present in the soralia. above the tree line (Galloway 1985). Differentiation. Peltigera extenuata typically Distribution. Peltigera ulcerata is widespread forms spreading thalli with a gray to light brown, in temperate regions of the Southern Hemisphere dull, minutely roughened, laminally sorediate upper (Galloway 1985; Swinscow & Krog 1988). The cortex that is tomentose especially toward the lobe species was hitherto not known from Chile (Gal- tip, and a lower surface with white to pale veins loway & Quilhot 1998), and is here reported as new and ¯occulent rhizines. The species rarely bears to this country from the Province of Osorno (W of apothecia or pycnidia. Gof®net and Hastings (1995) Parque Nacional Puyehue, along road to Antillanca, considered the presence of tridepsides in the soralia about 1, 2 km E of junction with Rd 215; seepy to be diagnostic of P. extenuata with regard to P. road bank, dominated by mosses, below bamboos didactyla. Our more recent studies of this species and tree ferns, elev. 380 m, March 20, 2000, Gof- have revealed that these tridepsides are not constant ®net 7126 & Cox (herb. B. Gof®net). among populations; hence, rather than invoking hy- bridization to account for this polymorphism (see 3. PELTIGERA CASTANEA Goward, Gof®net, & above), it may be more parsimonious to consider Miadlikowska, sp. nov. FIG.6 the presence of tridepsides in this species as vari- able. A similar hypothesis is proposed for P. lam- Thallus foliosus, Peltigerae extenuatae similis sed pagina superna nitida et castanea differt. Apothecia ignota. binonii and P. didactyla. A description along with Pycnidia frequentes. Habitat in supibus muscosis. photographs of P. extenuata are provided in Gof- TYPE: CANADA: BRITISH COLUMBIA. Clearwater River ®net and Hastings (1995). Basin. Philip Creek, Battle Mountain Road, elev. 1,500 Ecology. The species grows in rather mesic m, 51Њ52Ј N, 119Њ55Ј W, on mossy boulder, in open forest on south-facing hillside, 1 October 2001, Goward forests, usually over or among terricolous or epix- 01±858 (UBC, holotype; CANL, H, isotypes). ylic mosses. Unlike P. didactyla, P. extenuata rare- ly grows directly over soil. Thallus foliose, loosely appressed, small, to 6±8 Distribution. Peltigera extenuata was reported cm across; lobes stiff, fragile, to 1.0±1.5 cm wide, by Gof®net and Hastings (1995) from North Amer- not much longer than wide, strongly concave or ica, Central Europe and China. The species (as P. sometimes plane, loosely overlapping, irregularly didactyla var. extenuata) has subsequently been re- branched; lobe tips rounded, upturned; lobe mar- 360 THE BRYOLOGIST [VOL. 106

over the upper surface (in addition of course to the presence of apothecia). Sterile material must be carefully distinguished by the more pouch-like hab- it and the mostly discrete, non-¯occulent rhizines. Like P. castanea, P. didactyla can be somewhat shiny, but in that case the cortex has a predomi- nantly purplish brown cast quite different from the castaneous hue of P. castanea. Ecology. P. castanea occurs in (oro)boreal for- ests and alpine heaths, where it grows in open sites, possibly exclusively on south facing outcrops. Most of the existing specimens were collected from rath- er xerophytic moss mats. Distribution. Currently known only from Can- ada. Paratypes. CANADA. ALBERTA. Banff National Park, Snow Creek Valley, Beder s.n. (CANL 22128). BRIT- ISH COLUMBIA. Atlin Lake: Warm Bay Springs, Goward 82±599 (UBC); Philip Creek, Goward 02±2173 (UBC), Wildhorse Creek, Goward 01±1064 (UBC). ONTARIO. Penn FIGURE 6. Peltigera castanea (holotype, UBC). Ð A. Island, Kershaw s.n. (CANL 36013). YUKON. Kluane Na- Habit (ϫ1.3). Ð B. Lower surface when dry (ϫ2). Ð C. tional Park: Mount Maxwell, Scotter 20165 (CANL). Fish Lower surface when wet (ϫ2). Creek, Lupine tundra, Cannings s.n. (UBC). gins even; upper surface dark chestnut brown (pale 4. PELTIGERA LAMBINONII Gof®net, The Lichenol- bluish gray where sheltered), more or less shiny, ogist 27: 45, 1995. TYPE: ZAIÈRE: KIVU, Top of glabrous, except tomentose near lobe tips, soredi- Mt. Biega, heath with Erica bequaertii and Phi- ate; soredia granular, brown, loose, in rounded to lippia, 2,740 m, 6 January 1972, Lambinon 72/ irregular soralia to 2±3 mm across; lower surface Z/102 (LG, holotype; herb. B. Gof®net, isotype). densely veined; veins white, grading inwards to brown or black; rhizines copious, tufted and usually Chemistry. Methyl gyrophorate and gyrophoric strongly hedgerow-forming near lobe tips, but be- acid are typically, albeit not always present in the coming ¯occulent toward thallus center, concolo- soralia. So far we have only seen the specimens rous with veins. Cortex 40±60 ␮m thick; photo- examined by Gof®net and Hastings (1995). As in biont layer 50±20 ␮m thick, containing , me- the case of P. extenuata, these authors argued that dulla white, 80±120 ␮m thick. Apothecia unknown. chemical variation in this species may be due to Pycnidia frequent, marginal, pycnoconidia ellip- hybridization with sympatric P. didactyla. Howev- soid, 6±7 ϫ 4±5 ␮m. er, it would be more parsimonious to consider the Etymology. The name castanea describes the species chemically polymorphic until such time as dark, chestnut-colored upper surface characteristic the hybridization hypothesis can be critically tested. for this species. Differentiation. Except for the soralia, P. lam- Chemistry. No compounds were detected by binonii is reminiscent of P. rufescens, both species TLC. having elongate, rather concave lobes with a (typ- Differentiation. Peltigera castanea forms ically) densely tomentose upper surface. The lower spreading lobes, with a smooth, shiny, chestnut surface bears dark (rarely pale) brown, branched brown, mostly non-tomentose upper cortex, and rhizines that form a dense mat. Apothecia are com- darkening veins and ¯occulent rhizines that are of- mon on the tips of erect lobes. The species is illus- ten tufted and hedgerow forming (Fig. 6). Apothe- trated and described in Gof®net and Hastings cia are unknown, but marginal pycnidia are com- (1995). mon. Peltigera castanea closely resembles P. ex- Ecology. Based on the African material, P. tenuata, but that species has less concave lobes and lambinonii is known to ``grow over soil and humus, a dull, abundantly tomentose, minutely scabrid, among mosses in open sites, in the upper tropical, bluish gray or (where exposed) purplish brown up- montane forest between 1,700 and 2,740 m'' (Gof- per cortex usually covered in dense tomentum. The ®net & Hastings 1995). In Australia, Tibell col- veins and rhizines also usually remain pale lected this species on the ground in a low Eucalyp- throughout. tus forest with Banksia asplenifolia. Fertile specimens of P. didactyla are readily rec- Distribution. Gof®net and Hastings (1995) de- ognized by the occurrence of a thick tomentum scribed this species as an endemic of the East-Cen- 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 361 tral African mountains. Miadlikowska and Lutzoni distal portion of the upper surface. Often one ob- (2000) included P. lambinonii in their analysis serves an abrupt transition from the glabrous basal based on a collection from Australia by Leif Tibell. portion of the lobes to the distinctly tomentose dis- Our analysis con®rms the conspeci®city of the Aus- tal portions. The presence of scattered apothecial tralian and the African material, indicating that this initials along the lobe tips is another signal that species has a broader distribution in the paleotrop- syngamy has occurred, as is the development of ics than formerly suspected. Peltigera lambinonii is one or more thickened ``anchor'' rhizines at the here reported as new to Australia (AUSTRALIA. base of such lobes. Soredia do not form on these NEW SOUTH WALES. New England N.P., Point Look- fertile lobes, though they can persist inde®nitely on out, 30Њ30Ј S, 153Њ25Ј E; elev. 1,550 m, April 18, their basal portions. Eventually the fertile lobes be- 1981, Tibell 12401 [H, UPS]). come erect, and the apothecia develop to their ma- ture size, usually about 3±4(±7) mm in diameter 5. PELTIGERA DIDACTYLA (With.) J. R. Laundon, (Vitikainen 1994). Characteristically, the apothecia Lichenologist 16: 217. 1984. are folded lengthwise, and hence appear much lon- Lichen didactylus With., Bot. Arr. Veg. Gr. Brit. 1, 2: 718. ger than wide. Fertile specimens of the somewhat 1776. See Vitikainen (1994) for exhaustive list of syn- similar P. rufescens have somewhat larger apothe- onyms. TYPE: Dillenius, Hist. Musc. tab. 28, ®g. 108. cia (5±8 mm in diameter (MartõÂnez Moreno 1999), 1742. (OXF, holotype, epitype [typotype]; see Laundon and to 10 mm wide according to Gof®net and Has- 1984). tings 1994) that are less conspicuously folded. Pyc- Vitikainen (1994) provided a formal description nidia are also often present along the lobe margins. of this species. Although the original concept of Ecology. Peltigera didactyla is a species of this taxon is here revised, P. didactyla remains a early succession, being found most often over soil taxon of great morphological and chemical vari- or thin, mostly acrocarpous moss mats. ability (see below). Distribution. Despite the much narrower con- Chemistry. Tridepsides are typically absent; but cept of P. didactyla adopted here, this species re- if present, as in several specimens from across the tains a cosmopolitan distribution, being known range, they are restricted to the soralia. We consider from all continents, including Antarctica (évstedal this chemical variation to re¯ect phenotypic poly- & Lewis Smith 2001; Vitikainen 1994). morphism rather than evidence of past hybridiza- tion events. KEY TO THE SPECIES IN THE P. DIDACTYLA COMPLEX Differentiation. Two distinct phases of thallus development can be recognized for P. didactyla: 1. Thallus attached by a single holdfast; upper sur- face glabrous; triterpenoids present in medulla, tri- one asexual, the other sexual. In its asexual phase, depsides lacking ------P. vainioi this species, when very young, forms weakly con- 1. Thallus attached by scattered rhizines; upper sur- cave ``saucers'' that soon develop into deep, mon- face glabrous or tomentose; triterpenoids absent, ophyllous ``pouches.'' Many specimens of P. di- tridepsides present or absent ------2 2. Upper cortex glabrous throughout ------3 dactyla appear to remain in this asexual state in- 2. Upper cortex tomentose at least along margin de®nitely; they eventually form shiny, more or less ------4 elongate lobes with cupped tips. In extreme cases 3. Upper cortex shiny; soralia at least partly elon- the lobes measure as much as 10±15 mm long and gate, situated along lobes margins; soralia con- 4±8 mm wide. Such thalli invariably bear soredia, taining gyrophoric acid and methyl gyrophorate ------P. ulcerata which thus provide the only obvious means of re- 3. Upper cortex dull; soralia mostly in rounded production. The upper cortex is mostly brownish, patches, situated toward thallus center, or at least often dark chestnut or purplish brown, smooth, and not marginal; secondary substances typically ab- somewhat shiny. Prior to syngamy, tomentum is sent ------P. didactyla (young sterile forms) 4. Rhizines dark to pale brown, typically abun- lacking in most specimens, or is at most sparse and dantly branched and forming a dense mat; up- restricted to the vicinity of the lobe margins. Char- per surface often tomentose toward center ---- acteristic for this species is the presence, in the dis------P. lambinonii tal portions of the lobes, of scattered, pale, slender 4. Rhizines whitish, to brown or black toward rhizines that usually taper to a pointed tip. Thalli center, simple to ¯occulent, but not forming dense mat, upper surface with tomentum re- with abundant short and branched rhizines occur stricted to vicinity of lobe margin, rarely ex- throughout the range, but can be distinguished from tending inward (except on fertile lobes) ------5 other tomentose species by the shiny upper cortex. 5. Apothecia and/or apothecial initials present ------6 Roughly half of the 122 specimens of P. didac- 5. Apothecia and apothecial initials absent ------7 6. Apothecia common, located at the tips of up- tyla observed in our study were fertile. The tran- right lobes; secondary substances typically sition to the sexual phase seems to be marked by absent ------P. didactyla the development of a dense tomentose nap over the 6. Apothecia rare, located along lobe margins; 362 THE BRYOLOGIST [VOL. 106

soralia containing gyrophoric acid and methyl foveolate appearance. The rhizines are simple and gyrophorate ------P. extenuata ¯aring upon contact with the substrate. The lower 7. Lobes monophyllous, soon strongly concave and pouch-like; outermost rhizines mostly tapered to surface is white throughout. The thallus was void a point; on soil or thin moss ------P. didactyla of any secondary chemicals detectable by TLC. 7. Lobes usually polyphyllous, weakly to strongly The nucleotide sequences of this population (Pel- concave, but not usually pouch-like; outermost tigera sp. 2) are similar to those of P. didactyla but rhizines ¯aring and/or ¯occulent; mostly on thick deviate from the consistency observed among the moss ------8 8. Upper surface smooth, more or less shiny, other samples of this species. This population grew castaneous brown; lower surface darkening near other sorediate thalli (Gof®net 3853, herb. toward thallus center; rhizines often tufted Gof®net) composed of smaller, mostly erect lobes hedgerow-forming toward lobe tips; pycnidia and numerous simple rhizines. The morphology ®ts frequent along lobe margin P. castanea ------our concept of P. didactyla, a hypothesis we con- 8. Upper surface minutely roughened, dull, dove-gray to purplish brown; lower surface ®rmed based on the ITS sequence. Thus, the spec- mostly pale; rhizines more or less ¯occulent imen (Gof®net 3851) differs morphologically and throughout; pycnidia rare ------P. extenuata ``genetically'' from typical and sympatric P. didac- tyla. It is possible that this specimen represents a Aberrant samples. As mentioned above, three distinct haplotype of P. didactyla not worthy of a of the samples included in this study had nucleotide taxonomic recognition, but the level of divergence sequences that were either incongruent with our ini- between this sample and typical P. didactyla sam- tial identi®cation or diverged suf®ciently from ples is similar to that observed between other spe- those of other taxa to warrant a cautious note re- cies. Here too, additional observations are needed garding their taxonomic status. The collection from to elucidate the signi®cance of the above morpho- Veracruz (Mexico; Peltigera sp. 1) grew on soil logical differences. along a road in the botanical garden of Xalapa. The The population from Poland was collected from sample was tentatively identi®ed as P. didactyla, soil (sparse mosses) on the edge of a Pinus forest based on the narrow strongly concave and upright in an oligotrophic habitat. The specimen resembles lobes, the lack of highly branched rhizines, and the P. didactyla in its monophyllous thalli with abun- absence of tridepsides (con®rmed by TLC). The dant apothecia. The upper surface bears laminal lack of cortical hairs and the mainly submarginal soralia that are rather sparsely distributed and position of the soralia would point to P. ulcerata, sometimes fused. On many lobes the tomentum ex- but that species has two tridepsides in its soralia, tends toward the center; it is extremely compact and these compounds are lacking here. The speci- and whitish or the same color as the cortex, and men differs, however, from both taxa, in that the almost unnoticeable due to the dull cortex. On other lobes bear few rhizines, which are mostly produced lobes, the tomentum is sparse to even lacking. The toward the center of the thallus. Peltigera vainioi veins are well de®ned, often convex and broad, and Gyelnik, a sorediate taxon from South America, is have the same beige color as the thallus. The rhi- characterized by a single holdfast, but differs from zines are simple with pointed ends or divided but the Mexican specimen by the presence of triterpe- never ¯occulent or penicillate. They are abundant noids (Marcano et al. 1997). The sequences, which toward the center but sparse along the margin we con®rmed twice, are most similar to those of P. where they are grouped along the veins in short extenuata (Figs. 1±4), but do nevertheless differ rows. This specimen has gyrophoric acid and meth- from the latter in the same order of magnitude as yl gyrophorate. As for the preceding sample, the those of P. castanea and P. didactyla (Figs. 1±2). combination of characters is atypical for P. didac- We have not yet seen other specimens that are mor- tyla, but whether this should warrant formal taxo- phologically similar or that have been collected nomic recognition as may be extrapolated from the from this area, and taxonomic interpretation of this observed sequence divergence, is not clear at this variation should await further study. point. In 1994, Gof®net collected a sorediate species of Peltigera growing appressed over a thin layer of ACKNOWLEDGMENTS mosses over a decaying tree trunk in the North Sas- This study was made possible through ®nancial support katchewan River valley in Edmonton (Alberta, from the University of Connecticut and from the National Canada). The thallus is brownish-gray and lobed Science Foundation (DEB-0089633) to BG. Harry Sipman with upturned margins. The surface is undulate, (B) and Orvo Vitikainen (H) offered help in locating the with numerous soralia, and tomentum that is mostly type of P. ulcerata. Many thanks to Kenneth G. Wright and FrancËois Lutzoni for critical reading of the manu- lacking, although thin and extensive on one lobe. script. Thanks to Pak Yau Wong, CANL and Olivia Lee, The medulla is thin, but the veins are distinct and UBC, for forwarding specimens for study. Finally, we are some are strongly ropy, giving the lower surface a grateful to David Wagner (UCONN) for taking the picture 2003] GOFFINET ET AL.: PELTIGERA DIDACTYLA COMPLEX 363 of P. castanea and to Virge Kask for assembling the fungi found during the XIV Symposium of Baltic My- plates. The sequences were generated in F. Lutzoni's lab, cologists and Lichenologists in JaÈrvselja, Estonia Departement of Biology, Duke University and in B. Gof- [Samblike ja lihhenikoolsete seente leidudest XIV Bal- ®net's lab at the University of Connecticut. ti MuÈkoloogide ja Lihhenoloogide SuÈmpoosiumil JaÈrvseljal, Eestis]. Folia Cryptogamica Estonica 36: 17±21. LITERATURE CITED HASEGAWA, M., H. KISHINO &T.YANO. 1985. Dating the human-ape splitting by a molecular clock of mito- ALFARO, M. E., S. ZOLLER &F.LUTZONI. 2003. Bayes or chondrial DNA. Journal of Molecular Evolution 22: bootstrap? A simulation study comparing the perfor- 160±174. mance of Bayesian Markov chain Monte Carlo sam- HOLTAN-HARTWIG, J. 1993. The lichen genus Peltigera, pling and bootstrapping in assessing phylogenetic con- exclusive of the P. canina group, in Norway. Som- ®dence. Molecular Biology and Evolution 20: 255± merfeltia 15: 1±77. 266. HUELSENBECK, J. P. 2000. ``MrBayes: Bayesian inference BERGER, F. 1996. Neue und seltene Flechten und licheni- È of phylogeny'', Distributed by the author (http://mor- cole Pilze aus OberoÈsterreich, Osterreich II. Herzogia phbank.ebc.uu.se/mrbayes/), University of Rochester. 12: 45±84. KAUFF,F.&F.LUTZONI. 2002. Phylogeny of Gyalectales BERGSTEN, J. 1999. En foÈr Sverige ny ®ltlav, Peltigera and Ostropales (Ascomycota, Fungi): among and with- monticola, och naÊgra andra noteringar om slaÈktet Pel- in order relationships based on nuclear ribosomal tigera. Svensk Botanisk Tidskrift 93: 279±285. RNA small and large subunits. Molecular Phyloge- CULBERSON, C. F. 1972. Improved conditions and new data netics and Evolution 25: 138±156. for the identi®cation of lichen products by a standard- KIMURA, M. 1980. A simple method for estimating evo- ized thin-layer chromatographic method. Journal of lutionary rate of base substitution through comparative Chromatography 46: 85±93. studies of nucleotide sequences. Journal of Molecular ESSLINGER, T. L. 1997. A cumulative checklist for the li- Evolution 16: 111±120. chen-forming, lichenicolous and allied fungi of the KROKEN, S. & J. W. TAYLOR. 2001. A gene genealogical continental United States and Canada. North Dakota approach to recognize phylogenetic species boundaries State University: http://www.ndsu.nodak.edu/instruct/ in the lichenized Letharia. Mycologia 93: 38± esslinge/chcklst/chcklst7.htm (First Posted 1 Decem- 53. ber 1997, Most Recent Update 27 August 2001), Far- go, ND. LAMBINON, J. 1969. Les . Les Naturalistes Belges, Bruxelles, Belgium. FELSENSTEIN, J. 1985. Con®dence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783± LAUNDON, J. R. 1984. The typi®cation of Withering's ne- 791. glected lichens. Lichenologist 16: 211±239. LUTZONI, F., P. WAGNER,V.REEB &S.ZOLLER. 2000. In- GALLOWAY, D. J. 1985. Flora of New Zealand Lichens. P. D. Hasselberg, Government Printer, Wellington. tegrating ambiguously aligned regions of DNA se- quences in phylogenetic analyses without violating po- ÐÐÐ&W.QUILHOT. 1998. Checklist of Chilean lichen- forming and lichenicolous fungi [Lista patron de los sitional homology. Systematic Biology 49: 628±651. liquenes y hongos liquenicolas de Chile]. Gayana Bo- MADDISON, W. P. & D. R. MADDISON. 2001. MacClade: tanica 55 (2): 111±185. analysis of phylogeny and character evolution, version GOFFINET,B.&R.I.HASTINGS. 1994. The lichen genus 4.01. Sinauer, Sunderland, MA. Peltigera (lichenized Ascomycetes) in Alberta. Pro- MARCANO, V., A. MORALE S MEÂ NDEZ,S.MOHALI &E. vincial Museum of Alberta Natural History Occasional PALACIOS-PRUÈ . 1997. Morphological and chemical ob- Paper 21: 1±54. servations on Peltigera vainioi Gyelnik (Lichenized ÐÐÐ & ÐÐÐ. 1995. Two new sorediate taxa of Pel- Ascomycetes, ) from South America. Li- tigera. Lichenologist 27: 43±58. chens 1: 1±10. Â ÐÐÐ&R.J.BAYER. 1997. Characterization of myco- MARTINEZ MORENO, M. I. 1999. TaxonomõÂa del GeÂnero bionts of photomorph pairs in the Peltigerineae (lich- Peltigera Willd. (Ascomycetes Liquenizados) en la enized Ascomycetes) based on ITS sequences of spe- PenõÂnsula IbeÂrica y Estudio de sus Hongos LiquenõÂ- ci®cally ampli®ed fungal ribosomal DNA. Fungal Ge- colas. Ruizia 15: 1±200. netics and Biology 21: 228±237. MARTIÂNEZ, I., A. R. BURGAZ &O.VITIKAINEN. 1997. Stud- ÐÐÐ&T.GOWARD. 1998. Is silvae-veteris ies on the genus Peltigera in the Iberian Peninsula. II. the cyanomorph of Lobaria oregana? Insights from Nova Hedwigia 64: 367±391. molecular, chemical and morphological characters, pp. MASON-GAMER,R.&E.KELLOGG. 1996. Testing for phy- 41±52. In M. G. Glenn, R. C. Harris & M. S. Cole logenetic con¯ict among molecular data sets in the (eds.), Lichenographia Thomsoniana: North American tribe Triticeae (Gramineae). Systematic Biology 45: Lichenology in Honor of John W. Thomson. Myco- 524±545. taxon Ltd, Ithaca, NY. MIADLIKOWSKA, J. 1999. Rodzaj Peltigera (, ÐÐÐ&J.MIADLIKOWSKA. 1999. Peltigera phyllidiosa Ascomycota) na tle jego wspolczesnej systematyki. (Peltigeraceae, Ascomycotina), a new species from the Ph.D. dissertation, Gdansk University, Gdansk, Po- Southern Appalachians corroborated by ITS sequenc- land. es. Lichenologist 31: 247±256. ÐÐÐ&F.LUTZONI. 2000. Phylogenetic revision of the GOWARD, T. & B. GOFFINET. 2000. Peltigera chionophila, genus Peltigera (lichen-forming Ascomycota) based a new lichen (Ascomycetes) from the western cordil- on morphological, chemical, and large subunit nuclear lera of . THE BRYOLOGIST 103: 493± ribosomal DNA data. International Journal of Plant 498. Sciences 161: 925±958. GUTTMAN, D. S. 1997. Recombination and clonality in nat- ÐÐÐ, B. MCCUNE &F.LUTZONI. 2002. Pseudocyphel- ural populations of Escherichia coli. Tree 12: 16±22. laria perpetua, a new lichen from western North HALONEN, P., M. KUKWA,J.MOTIEJUNAITE,P.LOHMUS & America. THE BRYOLOGIST 105: 1±10. L. MARTIN. 2000. Notes on lichens and lichenicolous ÐÐÐ, F. LUTZONI,T.GOWARD,S.ZOLLER &D.POSADA. 364 THE BRYOLOGIST [VOL. 106

2003. Gap-rich regions from ITS and rDNA large sub- of molecular phylogenies obtained by Bayesian phy- unit are incorporated into phylogenetic analyses to re- logenetics. Proceedings of the National Academy of solve the Peltigera canina species complex. Mycolo- Sciences 99: 16138±16143. gia (in press). THOMSON, J. W. 1950. The species of Peltigera of North éVSTEDAL,D.O.&R.I.LEWIS SMITH. 2001. Lichens of America north of Mexico. American Midland Natu- Antarctica and South Georgia. Cambridge University ralist 44: 1±68. Press, Cambridge, UK. ÐÐÐ. 2003. Lichens of Wisconsin. Wisconsin State OZENDA, P. & G. CLAUZADE. 1970. Les Lichens: EÂ tude Herbarium, Madison. Biologique et Flore IllustreÂe. Masson & Cie, Paris, VAN DEN BOOM, P. 1998. Enkele interessante vondsten van France. lichenen en lichenicole fungi in Nederland II. Bux- baumiella 47: 40±43. POSADA, D. 2000. How does recombination affect phylog- VITIKAINEN, O. 1985. Three new species of Peltigera. An- eny estimation. Tree 15: 489±490. nales Botanici Fennici 22: 291±298. ÐÐÐ&K.A.CRANDALL. 1998. Modeltest: testing the ÐÐÐ. 1994. Taxonomic revision of Peltigera (lichen- model of DNA substitution. Bioinformatics Applica- ized Ascomycotina) in Europe. Acta Botanica Fennica tions Note 14: 817±818. 152: 1±196. SWINSCOW,T.D.V.&H.KROG. 1988. Macrolichens of WHITE,J.F.&P.W.JAMES. 1985. A new guide to micro- East Africa. British Museum of Natural History, Lon- chemical techniques for the identi®cation of lichen don, U.K. substances. Bulletin of the British Lichenological So- SWOFFORD, D. L. 1998. PAUP*: Phylogenetic Analysis ciety 57 (suppl.): 1±41. Using Parsimony (* and other methods), Version 4. Sinauer, Sunderland, MA. SUZUKI, Y., G. V. GLAZKO &M.NEI. 2002. Overcredibility ms. received Feb. 19, 2003; accepted April 15, 2003.