Introduction
Lichens are symbiotic associations of heterotrophic fungi and photoautotrophic green algae and/or cyanobacteria. In these close associations, interactions between the symbionts have resulted in highly specialized structures such as dual asexual propagules and secondary metabolites present only in the lichenized state (Hawksworth Reference Hawksworth1988). Lichenization occurs only when a fungal partner, the mycobiont, meets a compatible photosynthetic partner, the photobiont (Ahmadjian Reference Ahmadjian1993). Lichen-forming fungi have often been shown to be highly discriminating in their selection of photobiont species, both in algal (Rambold et al. Reference Rambold, Friedl and Beck1998; Beck et al. Reference Beck, Kasalicky and Rambold2002; Yahr et al. Reference Yahr, Vilgalys and DePriest2004) and in cyanobacterial associations (Stenroos et al. Reference Stenroos, Högnabba, Myllys, Hyvönen and Thell2006; Myllys et al. Reference Myllys, Stenroos, Thell and Kuusinen2007). In some cases, however, the same lichen fungus can associate with different photobiont species, that is one mycobiont can either form morphologically identical thalli with different photobiont species or genotypes (Friedl & Büdel Reference Friedl, Büdel and Nash2008; Piercey-Normore Reference Piercey-Normore2006), or two or more photobionts can co-exist in a single thallus (Casano et al. Reference Casano, del Campo, García-Breijo, Reig-Armiñana, Gasulla, del Hoyo, Guéra and Barreno2011). Likewise, even in cases where a given mycobiont has been shown to be selective with respect to a given photobiont, that same photobiont might be shared among unrelated fungi belonging to the same lichen community (Helms et al. Reference Helms, Friedl, Rambold and Mayrhofer2001; Piercey-Normore & DePriest Reference Piercey-Normore and DePriest2001; Rikkinen et al. Reference Rikkinen, Oksanen and Lohtander2002; O'Brien et al. Reference O'Brien, Miądlikowska and Lutzoni2005; Myllys et al. Reference Myllys, Stenroos, Thell and Kuusinen2007).
It has been suggested that some modes of dispersal in lichens may be better adapted than others to maintain symbiotic associations from one generation to the next (Nelsen & Gargas Reference Nelsen and Gargas2008; Wornik & Grube Reference Wornik and Grube2010). In sexually reproducing lichens, the fungal partner produces meiotic spores that need to encounter a compatible photobiont to re-establish the symbiosis. Such lichen systems offer an opportunity for horizontal transfer of the algal or cyanobiont partner. Suitable partners could be obtained by several means, including recruitment from the vegetative propagules (soredia, isidia, thallus fragments) of other lichens (Beck et al. Reference Beck, Friedl and Rambold1998), association with free-living algae or cyanobacteria (Mukhtar et al. Reference Mukhtar, Garty and Galun1994; Sanders & Lücking Reference Sanders and Lücking2002; Hedenås et al. Reference Hedenås, Blomberg and Ericson2007), and parasitization of another lichen (Friedl Reference Friedl1987). By contrast, asexually reproducing lichens reproduce by means of vegetative propagules that contain both partners, usually in the form of soredia or isidia. In principle, such lichens might be expected to retain the same photobiont from one generation to the next, potentially resulting in a high level of specificity in both symbiotic partners (Nelsen & Gargas Reference Nelsen and Gargas2008). Continuous association between the symbionts may lead to co-speciation (i.e. synchronous speciation events in different taxa), and even to parallel cladogenesis (i.e. correlated evolution along lineages), which does not necessarily involve concomitant speciation (Futuyma Reference Futuyma1998). Recent studies, however, have shown that dispersal from vegetative propagules is not invariably linked to retention of the same photobiont even throughout a single lichen life cycle, much less over evolutionary time scales (Nelsen & Gargas Reference Nelsen and Gargas2008; Wornik & Grube Reference Wornik and Grube2010).
Bryoria Brodo & D. Hawksw., with some 80 described species, is a lichenized euascomycete genus in the family Parmeliaceae (http://www.indexfungorum.org). It has a mainly circumpolar distribution in the Northern Hemisphere, though some species occur in mountainous regions of the Southern Hemisphere (Brodo & Hawksworth Reference Brodo and Hawksworth1977). Bryoria is a fruticose genus, easily recognized by its fine, hair-like, pendent or shrubby, grey, brown or black stems, which are repeatedly branched. Most species reproduce mainly asexually by soredia or thallus fragmentation, whereas sexual states are rare or even unknown in some species. In the absence of fruiting bodies, species delimitation has traditionally been based on morphological characters such as branching pattern, presence and type of soralia, presence and form of pseudocyphellae, thallus colour, as well as chemical characters, that is secondary metabolic compounds produced by the lichen thallus (Brodo & Hawskworth Reference Brodo and Hawksworth1977; Krog Reference Krog1980). In a recent phylogenetic study on Bryoria, however, Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011) found that both morpho- and chemospecies concepts are problematic, especially in section Implexae (Gyeln.) Brodo & D. Hawksw. where none of the currently recognized species are discriminated using existing genetic markers, Bryoria glabra (Motyka) Brodo & D. Hawksw. being the only exception. Furthermore, in their study on Bryoria fremontii (Tuck.) Brodo & D. Hawksw., Velmala et al. (Reference Velmala, Myllys, Halonen, Goward and Ahti2009) showed that the concentration of vulpinic acid, a secondary substance unique for this taxon in Bryoria, is not correlated with the phylogeny of the mycobiont. While this suggests that the mycobiont may not be decisive in this regard, it leaves open the question as to whether a correlation could exist between the production of specific secondary compounds and the identity of the photobiont.
The objective of this study was to examine photobiont identity, diversity and selectivity in the genus Bryoria using photobiont ITS and COX2 DNA sequence data. We compared photobiont and mycobiont phylogenies to examine patterns of selectivity and specificity among the symbionts, and more specifically to determine whether photobiont identity correlates with the occurrence of specific secondary metabolites.
Materials and Methods
Taxon selection
Eighty-one specimens from 16 different Bryoria species were included in our analyses (Table 1). For the most part, the material used in this study formed the basis of an earlier study by Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011). Taxon selection focused on Bryoria fremontii and section Implexae sensu Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011). To examine whether lichen substances have a role in photobiont choice, we included additional specimens from section Implexae. In Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011), section Implexae was amended, in comparison with its previous circumscription, to include most members of section Bryoria. The following six species from section Implexae were included in this study: B. capillaris (Ach.) Brodo & D. Hawksw., B. fuscescens (Gyeln.) Brodo & D. Hawksw., B. glabra, B. implexa (Hoffm.) Brodo & D. Hawksw. (including five chemotypes, see Table 1 and Holien Reference Holien1989), B. lanestris (Ach.) Brodo & D. Hawksw., and B. subcana (Nyl. ex Stizenb.) Brodo & D. Hawksw. Multiple samples of each taxon were used to capture potential geographical variation. Bryoria fremontii was used to determine whether metabolite concentration, here vulpinic acid, might correlate with photobiont phylogeny. In an earlier study, Velmala et al. (Reference Velmala, Myllys, Halonen, Goward and Ahti2009) detected only minor genetic variation within this species, prompting us to limit our mycobiont sampling to only four thalli, each representative of a single subclade. For photobiont sampling, we included an additional ten specimens, divided evenly between thalli that lack vulpinic acid (except in soralia and apothecia) and thalli in which this yellow pigment occurs throughout (=former B. tortuosa (G. Merr.) Brodo & D. Hawksw.; see Brodo & Hawksworth Reference Brodo and Hawksworth1977; Velmala et al. Reference Velmala, Myllys, Halonen, Goward and Ahti2009).
Table 1. List of taxa, voucher information, GenBank accession numbers and secondary chemistry.
Accession numbers marked in bold are new for this study. Abbreviations of secondary metabolites: ale=alectorialic acid, atr=atranorin, bar=barbatolic a., cfum=confumarprotocetraric a., cnsti=connorstictic a., fum=fumarprotocetraric a., gyr=gyrophoric a., nsti=norstictic a., pro=protocetraric a., pso=psoromic a., vul=vulpinic a., ( )=present in small amounts.
Photobiont ITS sequences acquired from Bryoria specimens were compared with 12 Trebouxia Puymaly sequences obtained from GenBank to find out the possible identities of photobionts of the genus Bryoria. We also included two Trebouxia sequences from the vulpinic acid-containing species Vulpicida juniperinus (L.) J.-E. Mattsson & M. J. Lai and V. pinastri (Scop.) J.-E. Mattsson & M. J. Lai; both sequences are original in this study. Gowardia arctica Halonen et al. and Pseudephebe pubescens (L.) M. Choisy were selected as outgroup species for the mycobiont analyses. After testing several outgroup candidates for the photobiont ITS analysis, we selected Trebouxia impressa Ahmadjian because it did not group among any of the ingroup taxa in our preliminary analyses. In the combined photobiont ITS and COX2 analysis, only photobiont sequences from Bryoria were used since there were no matching COX2 sequences available in GenBank from the Trebouxia sequences used as a reference in the ITS analysis. Based on the ITS photobiont analysis, we used the photobiont of B. smithii (Du Rietz) Brodo & D. Hawksw. as an outgroup in the combined ITS and COX2 analysis.
Molecular techniques
Total genomic DNA was extracted from dried lichen thalli using Qiagen's DNeasy Blood & Tissue Kit following the protocol described in Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011). Both photobiont and mycobiont sequences were obtained from the same DNA extractions. A total of 162 mycobiont sequences (Table 1) were obtained from the previous studies of Velmala et al. (Reference Velmala, Myllys, Halonen, Goward and Ahti2009) and Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011). For 18 additional specimens, three DNA regions were amplified and sequenced: 1) internal transcribed spacer regions of the nuclear ribosomal DNA (ITS), 2) partial sequence from the protein-coding gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and 3) partial sequence of the small subunit of the mitochondrial ribosomal DNA (mtSSU). PCR amplification was carried out with primers described by Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011) and under conditions as described in Velmala et al. (Reference Velmala, Myllys, Halonen, Goward and Ahti2009).
For the photobiont analyses, the internal transcribed spacer regions of the nuclear ribosomal DNA (ITS) and part of the cytochrome oxidase subunit 2 gene (COX2) were amplified and sequenced. The PCR amplification of the ITS region was carried out with primers ITS1AKL (Dahlkild et al. Reference Dahlkild, Källersjö, Lohtander and Tehler2001) and ITS4 (White et al. Reference White, Bruns, Lee, Taylor, Innis, Gelfand, Sninsky and White1990), for COX2 new primers, COXIIf2 (ttaacgcctaacgagggaac) and COXIIr (atacgaaatcccgttcctga), were designed based on ten COX2 sequences published in Fernández-Mendoza et al. (Reference Fernández-Mendoza, Domaschke, Garcia, Jordan, Martín and Printzen2011) and using the primer design software Primer3Web version 3.0.0 (Rozen & Skaletsky Reference Rozen, Skaletsky, Krawetz and Misener2000). PCR amplification was performed in 25 µl reaction volumes using PuRe Taq Ready-To-Go PCR beads (GE Healthcare) under the following conditions for ITS: initial denaturation of 5 min at 95°C followed by five cycles of 30 s at 95°C, 30 s at 58°C and 1 min at 72°C. For the remaining 30 cycles, the annealing temperature was decreased to 56°C. PCR ended with a final extension of 7 min at 72°C. For COX2, the initial annealing temperature was 57°C, decreasing to 55°C, and the number of cycles was 35 instead of 30.
PCR products were run on a 1% agarose gel stained with ethidium bromide and visualized under UV light. Purification of the PCR products was performed using GE Healthcare's illustra GFXTM PCR DNA and Gel Band Purification Kit following the manufacturer's protocol and eluted with 30 µl of elution buffer. Sequencing was carried out as described in Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011). The same primers used for PCR were also used for sequencing in all samples. Sequences were assembled with SeqMan II 4.00 (DNASTAR). All photobiont sequences obtained in this study were subjected to a BLAST search (Altschul et al. Reference Altschul, Gish, Miller, Myers and Lipman1990) to confirm that they really were algal sequences and to find out the identity of the algal species.
Sequence alignment and phylogenetic analyses
Sequences of each DNA region were aligned with MUSCLE v3.7 (Edgar Reference Edgar2004), located at CSC – IT Center for Science (http://www.csc.fi/english), using default parameters. The matrices obtained were edited manually and the three mycobiont matrices were combined into a concatenated matrix in MacClade 4.08 (Maddison & Maddison Reference Maddison and Maddison2005). From the mycobiont matrix, the hypervariable region of the mtSSU corresponding to positions 808–860 of HQ402637 GenBank accession (Table 1) was removed. The photobiont ITS and COX2 gene regions were first analyzed separately and then combined into a concatenated matrix. All photobiont and mycobiont data sets were subjected to maximum parsimony analysis and Bayesian phylogenetic analysis. Parsimony analyses were performed in TNT v1.1 (Goloboff et al. Reference Goloboff, Farris and Nixon2008) using a traditional search of 100 replicates with random addition of sequences and TBR branch swapping. Ten trees were saved for each replicate and gaps were treated as missing data. Node support was estimated using the bootstrapping method with 1000 replicates.
For Bayesian analyses, a suitable substitution model was selected by calculating AIC (Akaike Information Criterion) scores in jModelTest2 v2.1.1 (Guindon & Gascuel Reference Guindon and Gascuel2003; Darriba et al. Reference Darriba, Taboada, Doallo and Posada2012). Models with the lowest AIC scores were selected for the analyses. The substitution model of the ITS region was estimated separately for each partition. For the mycobiont, data model GTR+G was used for ITS1, ITS2, mtSSU and GAPDH gene regions and the model K80 was used for 5.8S. The model GTR+G was also used for the photobiont ITS1 and ITS2 regions, and GTR for the 5.8S. Model HKY was selected for the COX2 gene region. Bayesian analyses were then conducted for all four data sets using MrBayes v3.2.1 (Huelsenbeck & Ronquist Reference Huelsenbeck and Ronquist2001). For the mycobiont data, two parallel runs of 10000000 generations were performed using four chains and sampling every 500th tree. The temperature parameter was set to 0·05. After the analysis, 5000 samples per run were discarded as burn-in. For the separate and combined analysis of photobiont ITS and COX2 data, two parallel runs of 2000000 generations were performed, also using four chains but sampling every 100th tree. Again, a default burn-in fraction of 25% was used after the analysis and the number of discarded samples was 5000 per run. The program Tracer v1.5 (Rambaut et al. Reference Rambaut, Suchard, Xie and Drummond2013) was used for all four data sets to check that the runs had reached convergence.
Results
Mycobiont data
Fifty-two new sequences were generated for this study. The combined alignment of mycobiont ITS, mtSSU and GAPDH gene regions consisted of 2346 characters, of which 435 were variable and 93 parsimony-informative. The ITS alignment comprised 481 characters, of which 150 were variable and 73 parsimony-informative. The mtSSU alignment on the other hand included 927 characters, of which 67 were variable and none parsimony-informative. The GAPDH alignment was 938 characters long and consisted of 218 variable characters, of which 20 were parsimony-informative.
The parsimony analysis of combined mycobiont ITS, mtSSU and GAPDH data of 71 taxa and 2346 characters produced 7 equally parsimonious trees, of which a strict consensus tree is shown in Figure 1. All Bryoria sections observed in Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011), that is, sections Americanae Myllys & Velmala, Bryoria, Divaricatae (DR.) Brodo & D. Hawksw., Implexae and Tortuosae (Bystr.) Brodo & D. Hawksw., also appeared as strongly supported monophyletic groups in our analysis. Within section Implexae, B. glabra is the only monophyletic species. All other taxa in the section are unresolved and are divided into two subclades, referred to here as the North American subclade and the European subclade (see also Myllys et al. Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011), although the latter subclade also includes some North American specimens. The Bayesian phylogeny of the combined mycobiont data is consistent with the parsimony phylogeny (Bayesian phylogeny not shown).
Fig. 1. Phylogenetic relationships among Bryoria species. Strict consensus tree obtained from ITS, mtSSU and GAPDH data. Bootstrap support values are shown above nodes and Bayesian posterior probability values are shown below nodes. Abbreviations of secondary metabolites: BAR=barbatolic acid, FUM=fumarprotocetraric a., GYR=gyrophoric a., NSTI=norstictic a., PSO=psoromic a., VUL=vulpinic a., -=no substances. Only main diagnostic secondary substances are shown.
Photobiont data
A total of 141 new Bryoria photobiont sequences were produced for this study. The photobiont ITS alignment consisted of 709 characters, of which 233 were variable and 56 parsimony-informative. The combined alignment of photobiont ITS and COX2 gene regions comprised 1074 characters, of which 191 were variable and 46 parsimony-informative. The ITS portion of the alignment consisted of 685 characters, of which 186 were variable and 43 parsimony-informative, whereas the COX2 alignment was 389 characters long and had five variable characters, of which three were parsimony-informative.
The parsimony analysis of photobiont ITS data of 89 taxa and 709 characters produced 49 equally parsimonious trees, of which a strict consensus tree is shown in Figure 2. In the ITS parsimony phylogeny, seven clades with moderate to high bootstrap support are distinguished. Clade I consists of the photobionts of B. smithii as well as Trebouxia asymmetrica Friedl & Gärtner and T. decolorans Ahmadjian. Clade II contains photobionts of B. bicolor (Ehrh.) Brodo & D. Hawksw. and B. tenuis (Å. E. Dahl) Brodo & D. Hawksw. In Clade III are photobionts of B. furcellata (Fr.) Brodo & D. Hawksw., together with two T. simplex Tscherm.-Woess sequences obtained from Lecanora conizaeoides Nyl. ex Cromb. Clade IV has only two sequences: photobionts of one B. americana (Motyka) Holien specimen and one B. fremontii specimen. In Clade V are photobionts of six B. fremontii specimens and one T. jamesii (Hildreth & Ahmadjian) Gärtner sequence obtained from Letharia lupina Altermann & Goward ined. Clade VI consists of photobionts of eight B. fremontii specimens and one T. jamesii sequence from Letharia vulpina (L.) Hue. Photobionts of vulpinic acid-containing B. fremontii specimens did not form a distinct clade in either of the photobiont phylogenies, but instead were evenly distributed among the three clades containing B. fremontii specimens. Clade VII contains all photobionts of species in section Implexae, as well as the photobionts of B. nadvornikiana (Gyeln.) Brodo & D. Hawksw., B. nitidula (Th. Fr.) Brodo & D. Hawksw., B. simplicior (Vain.) Brodo & D. Hawksw. and two B. americana specimens. The phylogenetic relationships of photobionts inside this clade remain mostly unresolved with the exception of photobionts of B. nadvornikiana, which form a strongly supported small monophyletic subclade. Interestingly, all North American specimens of section Implexae fell into one subclade inside Clade VII. Also in Clade VII are photobiont sequences from Flavocetraria nivalis (L.) Kärnefelt & A. Thell, Pseudevernia furfuracea (L.) Zopf, Umbilicaria antarctica Frey & I. M. Lamb, Vulpicida juniperinus and V. pinastri. The parsimony analysis of COX2 data of 66 taxa and 389 characters produced one tree (phylogeny not shown). COX2 contained less variation than ITS, with only three parsimony-informative sites, and this was also evident in the phylogeny as only B. nadvornikiana distinguished as a monophyletic species.
Fig. 2. Strict consensus tree of ITS data obtained from photobionts in Bryoria and from reference sequences obtained from GenBank. Bootstrap support values are shown above nodes and Bayesian posterior probability values are shown below nodes.
The parsimony analysis of combined photobiont ITS and COX2 data of 75 taxa and 1074 characters produced 30 equally parsimonious trees, of which a strict consensus tree is shown in Figure 3. The combined ITS and COX2 phylogeny included only photobiont sequences from Bryoria but was otherwise mostly consistent with the ITS phylogeny, with the exception of B. americana L199 specimen and species B. fremontii, B. bicolor, B. tenuis and B. furcellata. In the ITS phylogeny, photobionts of B. fremontii formed three clades (Clades IV, V and VI, Fig. 2), and one of these clades (Clade IV) included the photobiont from specimen L199 of B. americana. However, in the combined ITS and COX2 phylogeny, photobionts of all B. fremontii specimens and the photobiont of B. americana specimen L199 grouped together, although this clade was not supported. Specimens of B. bicolor and B. tenuis formed a distinct clade in the ITS phylogeny (Clade II, Fig. 2) and grouped together with B. fucellata in the combined ITS and COX2 phylogeny, but this clade was also not supported. The Bayesian phylogenies of ITS, COX2 and combined ITS and COX2 data were consistent with corresponding parsimony phylogenies (Bayesian phylogenies not shown).
Fig. 3. Strict consensus tree of combined ITS and COX2 data obtained from photobionts in Bryoria. Bootstrap support values are shown above nodes and Bayesian posterior probability values are shown below nodes.
Discussion
Photobiont identity and selectivity
A major finding of this study is that the photobionts of all but one Bryoria species subjected to ITS and COX2 sequence analyses can be assigned to the Trebouxia simplex group [also referred to as Trebouxia jamesii; see Opanowicz & Grube (Reference Opanowicz and Grube2004) and Doering & Piercey-Normore (Reference Doering and Piercey-Normore2009) for discussion]. Previous studies have shown that this group consists of several phylogenetic species, many of which, however, have not been formally described due to a lack of cultured strains (Kroken & Taylor Reference Kroken and Taylor2000; Piercey-Normore Reference Piercey-Normore2006; Hauck et al. Reference Hauck, Helms and Friedl2007). Below we discuss the possible identities of photobionts in the genus Bryoria while acknowledging the taxonomic uncertainty of many of these phylogenetic species. Only in the case of B. bicolor and B. tenuis (Clade II, Fig. 2) and B. americana and B. fremontii appearing in Clade IV (Fig. 2) do the identities of the photobionts remain unclear, owing to a lack of GenBank reference sequences.
In the study of Hauck et al. (Reference Hauck, Helms and Friedl2007), Lecanora conizaeoides was found to associate with Trebouxia simplex. In our study, photobionts of B. furcellata grouped with the photobionts of L. conizaeoides (Clade III, Fig. 2) sequenced by Hauck et al. (Reference Hauck, Helms and Friedl2007), and we suggest that these most probably represent T. simplex as well.
Photobionts of all but one specimen of Bryoria fremontii grouped with either T. jamesii from Letharia ‘lupina’ (Clade V, Fig. 2) or with T. jamesii from Letharia vulpina (Clade VI, Fig. 2). Kroken & Taylor (Reference Kroken and Taylor2000) referred to these two algal clades as T. jamesii ‘letharii’ and T. jamesii ‘vulpinae’ respectively, and suggested they be regarded as separate phylogenetic species. Based on similarities in chloroplast morphology, the latter most likely represents T. jamesii subsp. angustilobata A. Beck. Hauck et al. (Reference Hauck, Helms and Friedl2007) provisionally named T. jamesii ‘letharii’ as T. lethariae and proposed that the name T. angustilobata be used in place of T. jamesii subsp. angustilobata. However, they refrained from formally describing either species, owing to a lack of cultured strains.
All photobionts from section Implexae, as well as the photobionts of B. nadvornikiana, B. nitidula, B. simplicior and two B. americana specimens, are closely related to the photobionts of Pseudevernia furfuracea, Umbilicaria antarctica, Flavocetraria nivalis, Vulpicida juniperinus and V. pinastri (Clade VII, Fig. 2). Judging from the low level of sequence divergence, these photobionts probably represent a single algal lineage for which Hauck et al. (Reference Hauck, Helms and Friedl2007) introduced the name Trebouxia hypogymniae Hauck & Friedl ined. However, as pointed out by Hauck et al. (Reference Hauck, Helms and Friedl2007), a formal description of the species is not possible until the samples used in the analyses have been cultured. In the meantime, it is interesting to note phylogenetic structure within Clade VII (Fig. 2), where the photobionts of B. nadvornikiana formed a strongly supported subclade both in the ITS and in the combined ITS and COX2 phylogenies. In addition, the photobionts of the North American specimens were found only in one subclade.
The photobiont of B. smithii appears to be unique within Bryoria in being more closely related to T. asymmetrica and T. decolorans than to any of the lineages in the T. simplex group (Clade I, Fig. 2). However, the exact identity of this alga remains unsolved. Bryoria smithii is an oceanic species distributed in South-East Asia, the Himalayas, and central and northern Europe (Hawksworth Reference Hawksworth1972; Jørgensen Reference Jørgensen1972), with an outlier in the mountains of Hawaii (Smith Reference Smith1984). In Fennoscandia, the species is rare and confined to suboceanic localities. Our sample, based on two specimens from widely disjunct portions of the species range, is consistent with the hypothesis that this species may associate with the same photobiont throughout its range. One of the reasons for the current distribution pattern of the species could be the association with a photobiont adapted to oceanic habitats. However, this needs to be investigated further with a broader sampling of specimens including photobiont sequences from other lichen species and from other areas.
Most of the species examined, which appeared as distinct species in the mycobiont phylogeny (i.e. B. bicolor, B. furcellata, B. nadvornikiana, B. simplicior, B. smithii and B. tenuis), appear to be highly selective with regard to their photobiont. Likewise, all members of section Implexae examined appear to associate with a single photobiont lineage. We caution, however, that these findings must be treated with care as in most cases they are based on only a small number of sequences. The only exceptions to this pattern were B. americana and B. fremontii, which associate with two and three lineages of the T. simplex group, respectively. Even if most Bryoria species appeared as highly selective towards their photobionts, few of the photobionts were found to be selective. Only B. smithii associated with a specific photobiont not found in other Bryoria species. Furthermore, in most cases Bryoria shared its photobionts with other genera in the Parmeliaceae such as Pseudevernia, Flavocetraria and Letharia, as well as with genera in the Umbilicariaceae and Lecanoraceae. Our visual inspection of the phylogenies revealed little evidence of parallel cladogenesis between the symbionts. The only exceptions are B. bicolor and B. tenuis, which are closely allied (Fig. 1) and most probably share the same photobiont. The recent studies of Nelsen & Gargas (Reference Nelsen and Gargas2008) and Wornik & Grube (Reference Wornik and Grube2010) found no correlation between algal and fungal phylogenies in spite of joint dispersal and concluded that sorediate species do not necessarily maintain the symbiotic partner, but may obtain their photobionts from vegetative propagules of other individuals of the same species, or even from other lichen species or from free-living algae. In contrast to these studies, our results suggest that the same photobiont lineage is maintained over consecutive generations in Bryoria. As already mentioned, the Bryoria species included in this study reproduce predominantly asexually from soredia or from thallus fragments, which could be one of the reasons for the high levels of fungal selectivity observed. However, it must be noted that there are also other factors beyond co-dispersal which may explain the high fungal selectivity. Irrespective of the dispersal mode, the mycobiont may achieve higher fitness by increasing specialization in the selection of a photobiont. Although production of ascomata is rare in most of the species sampled, even rare sexual reproductive events would reduce the level of selectivity that was observed in some species (B. americana and B. fremontii) in our study.
Photobiont identity and secondary chemistry
In section Implexae, secondary chemistry plays a major role in species delimitation and the presence of a specific secondary compound may be the only character separating one species from another. However, this chemospecies concept is controversial and has led different authors to treat these taxa at different taxonomic ranks. Brodo & Hawksworth (Reference Brodo and Hawksworth1977) treated Bryoria pseudofuscescens (which contains norstictic acid), B. friabilis (gyrophoric acid) and B. implexa (psoromic acid) as separate species, whereas Holien (Reference Holien1989) found little morphological variation and considered them to be conspecific. Our results are in agreement with the recent phylogenetic analysis of Bryoria by Myllys et al. (Reference Myllys, Velmala, Holien, Halonen, Wang and Goward2011), and show that the genetic diversity in section Implexae is extremely low, with only B. glabra appearing as a distinct species (see Fig. 1). Instead, the results suggest that North American B. capillaris and B. implexa are genetically distinct from European B. capillaris and B. implexa, respectively. Within these two subclades genetic variation is non-existent, suggesting that all the species inside them might be conspecific. In the light of these results, it is not surprising that species from this section all associated with the same photobiont, Trebouxia hypogymniae Hauck & Friedl ined.
Only a few studies have examined the role of the in vivo photobiont in the production of lichen secondary metabolites. For instance, Blaha et al. (Reference Blaha, Baloch and Grube2006) examined photobiont diversity in Lecanora rupicola (L.) Zahlbr. and found no correlation between different chemotypes and the associated photobiont. In this study, we investigated whether chemical diversity within Bryoria section Implexae correlates with photobiont identity. According to our results, this does not seem to be the case. Otherwise, individuals with different chemistries would have formed their own separate clades in the photobiont phylogeny. Similarly, our study did not find any correlation between the presence/concentration of vulpinic acid and photobiont identity, as photobionts of B. fremontii specimens containing this substance did not form their own clade in the photobiont phylogeny. As discussed by Hauck et al. (Reference Hauck, Helms and Friedl2007), most lichen species associating with T. simplex and T. hypogymniae Hauck & Friedl ined. are found on acidic substrata. They suggested that the substratum rather than the chemistry of the lichen would explain the photobiont identity. The occurrence of the same photobiont lineages in taxonomically diverse but ecologically similar lichens has also been reported for Asterochloris-associating species (Peksa & Škaloud Reference Peksa and Škaloud2011).
The authors thank Laura Häkkinen and Satu Laitinen for help with laboratory work and Pradeep Divakar for help with Bayesian phylogenetic analyses. Three anonymous reviewers are thanked for their valuable comments on the manuscript. This study was financially supported by the Academy of Finland (grant 1133858).
