Molecular phylogeny and symbiotic selectivity of the green algal genus Dictyochloropsis s.l. (Trebouxiophyceae): a polyphyletic and widespread group forming photobiont-mediated guilds in the lichen family Lobariaceae
Francesco Dal Grande,
Biodiversity and Conservation Biology, Swiss Federal Research Institute WSL, Birmensdorf, Switzerland
Biodiversity and Climate Research Centre (BiK-F), Senckenberg Gesellschaft fuer Naturforschung, Frankfurt am Main, Germany
Dictyochloropsis s.l. is an ecologically important, common but little-studied genus of green algae. Here, we examined the diversity and host selectivity of algae attributed to this genus at both species-to-species and species-to-community levels.
We conducted a molecular investigation of 15 cultured strains and several lichen photobionts, using 18S rRNA, rbcL and ITS sequence data. We further used seven alga-specific microsatellite markers to study algal sharing among fungi of the family Lobariaceae in two populations in Madeira and Taiwan (454 lichens).
We found that the genus Dictyochloropsis s.l. is polyphyletic. Dictyochloropsis clade 1 comprises only free-living algae whereas Dictyochloropsis clade 2 includes lichenized algae as well as free-living algae. Fungal selectivity towards algae belonging to Dictyochloropsis clade 2 is high. Selectivity varies geographically, with photobionts being restricted to a single region. Finally, we showed that Dictyochloropsis clade 2 individuals are shared among different fungal hosts in communities of lichens of the Lobariaceae.
As for other green algal lineages, there is a high amount of cryptic diversity in Dictyochloropsis. Furthermore, co-evolution between Dictyochloropsis clade 2 algae and representatives of the Lobariaceae is manifested at the community level, with several unrelated fungal species being horizontally connected by shared photobiont clones.
Lichens are obligate symbiotic associations between a fungus (mycobiont) and a population of algae and/or cyanobacteria (photobionts) (Ahmadjian, 1993; Honegger, 2012). Although the symbiotic state is obligate for the formation of the fungal thallus, it is not clear whether green algal photobionts are also obligatorily symbiotic. Scattered observations support the hypothesis that members of Trebouxia and Asterochloris, the most common lichen photobionts, may grow in the free-living state (Bubrick et al., 1984; Mukhtar et al., 1994). Nevertheless, free-living, locally available symbiotic green algae seem to be mostly rare and inconspicuous (Wornik & Grube, 2010).
The genetic structure of the lichen symbiosis is profoundly influenced by the way photobionts are dispersed and transmitted to their fungal host (Fernández-Mendoza et al., 2011; Dal Grande et al., 2012). On the one hand, vertical transmission of photobionts occurs when fungal and algal symbionts are co-dispersed from one generation to the next, in the form of asexual propagules, mostly over short distances (Büdel & Scheidegger, 2008; Dal Grande et al., 2012; Werth & Scheidegger, 2012). Vertical photobiont transmission leads to congruent mycobiont–photobiont genetic structures at the intrapopulation and, in some cases, continental levels (Dal Grande et al., 2012; Werth & Scheidegger, 2012; Widmer et al., 2012). Sexual propagation, on the other hand, breaks symbiosis and usually fungal spores are dispersed separately from their photobiont. Sexual dispersal results in the reshuffling of the symbiotic association – horizontal photobiont transmission – because the symbiosis has to be reconstituted. This process is called relichenization. It can significantly influence mycobiont–photobiont relations, even in predominantly vegetative species (Ott, 1987; Nelsen & Gargas, 2008, 2009; Dal Grande et al., 2012). Several strategies have been proposed to explain how the process of relichenization can occur in nature. For example, spores may capture the photobiont from other lichens as reported by Friedl (1987), Rikkinen (1995) and Beck et al. (2002), especially from juvenile stages or senescent/damaged thalli of closely related species (DeBach, 1964; Beck et al., 1998). Faecal pellets of lichenivorous orobatid mites or snails containing viable photobiont cells represent another source of algae for reconstituting the symbiosis (Meier et al., 2002; Boch et al., 2011).
It has been hypothesized that, in species with predominantly sexual reproduction, lichen symbiotic association is less specific, and mycobionts tend to associate with a wide range of photobionts. In the last decades, molecular markers and phylogenetic analyses have been applied extensively to determine the level of specialization in mycobiont–photobiont associations. Different association patterns have been described, from high host selectivity where the mycobiont accepts only a single algal strain (e.g. Psoroglaena stigonemoides associated with the green alga Auxenochlorella protothecoides and P. epiphylla/Chlorella luteoviridis; Nyati et al., 2007), to generalism where the host accepts multiple photobionts (Beck et al., 2002; Romeike et al., 2002; Blaha et al., 2006; Guzow-Krzeminska, 2006; Otálora et al., 2010; Muggia et al., 2013; O'Brien et al., 2013). The degree of host selectivity (range of compatible photobionts for a given lichen fungus) can vary even among closely related species or within a species range (Yahr et al., 2004, 2006; Fernández-Mendoza et al., 2011). High specificity is found in lichen symbioses where both partners are strictly selective towards each other (Beck et al., 2002). On the photobiont side, it has been shown that identical algal strains (green algae or cyanobacteria) can generally form symbiotic associations with several lichen-forming fungal species (Beck et al., 1998; Rikkinen et al., 2002; Wirtz et al., 2003; O'Brien et al., 2005; Honegger, 2008, 2012).
Interestingly, it has been suggested that fungal hosts may express their photobiont specificity at a community level (photobiont-mediated lichen guilds; Rikkinen et al., 2002; O'Brien et al., 2013). Lichen guilds are communities of lichens growing in the same habitat that are horizontally linked by sharing the same photobiont. Green algal lichen guilds have been reported for lichens of the genera Lecidella and Xanthoria associated with Trebouxia (Beck et al., 1998) and more recently for Lepraria and Stereocaulon associated with the green algal genus Asterochloris (Peksa & Škaloud, 2011). However, due to the lack of highly variable molecular markers, the lichen guild hypothesis has thus far only been tested at the species level of mycobiont–photobiont interactions. Also, marker resolution becomes critical when studying highly clonal organisms such as lichens (Arnaud-Haond et al., 2007; Dal Grande et al., 2012). In a recent study on a green algal lichen symbiosis associated with the green algal genus Dictyochloropsis, however, highly variable microsatellite markers enabled us to track down the photobiont transmission to the individual/thallus level and to identify clonal thalli in a population/guild (Dal Grande et al., 2012).
One of the most challenging aspects of lichen biology has been unveiling the identity of the photobionts and the reciprocal selectivity of the symbionts (Beck et al., 2002; Rikkinen et al., 2002; Rikkinen, 2003; Otálora et al., 2010). Photobiont selectivity studies rely on the phylogenetic characterization of algal taxa involved in the symbiosis. However, there is a general consensus that the diversity of green microalgae at all levels is still far from fully understood. A number of recent studies revealed the polyphyly of many morphologically defined genera in Trebouxiophyceae, the most common class of lichenized green algae (Friedl & Büdel, 2008). Studies on green algal lichen photobionts have focused mainly on the genera Trebouxia and Asterochloris (Trebouxiophyceae), although some studies have begun investigating many of the other lineages of lichen-associated Trebouxiophyceae (Diplosphaera: Thüs et al., 2011; Fontaine et al., 2012; Coccomyxa and Pseudococcomyxa: Lohtander et al., 2003; Zoller & Lutzoni, 2003; Muggia et al., 2011). Dictyochloropsis is a common but less-studied, ecologically important genus of aerophytic and lichenized algae. The genus was introduced by Geitler (1966) with the free-living type species D. splendida, and later emended by Tschermak-Woess (1980, 1984) with the addition of lichenized taxa (from here on called Dictyochloropsis s.l.). Representatives of Dictyochloropsis s.l. have been reported to form associations with many lichens of the family Lobariaceae (Tschermak-Woess, 1984, 1988, 1995; Moncada et al., 2013), Ramalinaceae, Catillariaceae and Megalosporaceae (Nakano et al., 1991; Rambold et al., 1998). Species of Dictyochloropsis s.l. have also been reported as free-living, for example, soil and epiphytic algae (Geitler, 1966). One species of Dictyochloropsis s.l., D. reticulata, is the photobiont of the frequently studied lichen species Lobariapulmonaria, and has been extensively studied regarding its population genetics and mode of transmission (Werth et al., 2006, 2007; Dal Grande et al., 2012; Grube & Spribille, 2012; Scheidegger et al., 2012; Widmer et al., 2012). Previous investigations on a global scale have shown that its fungal host is highly selective towards this widely distributed algal species (Walser et al., 2004; Widmer et al., 2012). Conversely, D. reticulata was reported to form symbioses with different lichen-forming fungi. However, the bulk of literature concerning the degree of selectivity for this organism relies on morphological identifications of algal cultures (Tschermak-Woess, 1951, 1984, 1995). Recently, Widmer et al. (2010) showed with molecular methods that the photobiont of L. pulmonaria is found in other Lobariaceaen lichens (L. macaronesica, L. tuberculata) and that the culture isolated from the lichen Brigantiaea ferruginea (Müll. Arg.) Kashiw. & Kurok. (Takeshita et al., 1991), used in many studies as a reference culture for D. reticulata (Friedl, 1995; Walser et al., 2003), belongs to a different algal lineage. Recent studies presented evidence that Dictyochloropsis s.l. is polyphyletic (Škaloud et al., 2007; Widmer et al., 2010; Thüs et al., 2011). Moreover, Škaloud et al. (2005) emphasized that the present taxonomic diversity of the genus does not cover the overall variability of its representatives. However, the scarcity of molecular data has so far prevented an accurate assessment of the taxonomy and phylogenetic position of the representatives of this genus.
The present study sets out to achieve three goals. The first is to identify genetically distinct monophyletic groups of strains of green algae assigned to the genus Dictyochloropsis s.l. The second goal is to determine the degree of fungal selectivity within this algal lineage. To achieve these objectives, we used a molecular multi-locus phylogenetic approach to investigate the genetic variability of free-living as well as symbiotic algae collected across a broad geographical spectrum. The third goal is to test the existence of photobiont-mediated lichen guilds at a high-resolution level, using microsatellite fingerprinting to access the taxonomic range of compatible hosts and the amount of intrapopulation horizontal transmission for one Dictyochloropsis s.l. species. Studying patterns of symbiont diversity and selectivity is crucial for the understanding of fundamental ecological processes such as dispersal and establishment in the family Lobariaceae whose representatives are related to old-growth forest structures and are widely used as indicators of ecological continuity in conservation practice (Scheidegger & Werth, 2009).
Materials and Methods
Species sampling and algal cultures
Strains were sampled to represent the diversity of the algae attributed to the genus Dictyochloropsis s.l. Furthermore, algal symbionts were sequenced from lichen fungi in the Lobariaceae (genera Crocodia, Dendriscosticta, Lobaria, Lobariella, Pseudocyphellaria, Ricasolia, Sticta s.l.) collected in Europe, North, Central and South America, Russia, China, Taiwan and New Zealand. Details of the material, area of collection and GenBank accession numbers are presented in Table 1.
Table 1. Algal strains and lichen photobionts used in the present study and 18S, rbcL and internal transcribed spacer (ITS) accession numbers
Algal cultures and uncultured lichen photobionts were sequenced at two loci: part of the nuclear small subunit ribosomal RNA gene (18S rRNA; 43 specimens) and part of the plastid-encoded large subunit of the ribulose-bisphosphate carboxylase-RuBisCO gene (rbcL; 52 specimens). For a subset of specimens and additional 19 specimens of lichens of the genera Lobaria, Dendriscosticta and Ricasolia (see Table 1), we also obtained sequences of the nuclear internal transcribed spacer (ITS). Most likely ITS2 secondary structures of the RNA transcript were determined by depicting the highly conserved start and end region of the four helices (Mai & Coleman, 1997). The structure of these sequence sections has been calculated using the RNAfold WebServer (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi).
Total lichen DNA was isolated from dried thallus material (35–50 mg) or from sterile algal cultures (50 mg) using the DNeasy 96 Plant Kit (Qiagen) according to the manufacturer's protocol. All amplifications were performed in a total volume of 50 μl with 2 μl DNA-extract using JumpStart™ REDTaq® ReadyMix™ (Sigma-Aldrich) and 100 nM of each primer. For primer sequences and PCR cycling conditions, see Table 2. All PCR products were purified with the MinElute PCR Purification Kit (Qiagen) and labeled with Big Dye Terminator v3.1 Kit (Life Technologies, Carlsbad, CA, USA). Cycle sequencing was performed as follows: 25 cycles of: 20 s 96°C, 5 s 50°C, 2 min 60°C. Post reaction cleanup was performed using Performa DTR Gel Filtration Cartridges (Edge BioSystems, Gaithersburg, MD, USA) following the manufacturer's protocol. Forward and reverse strand sequences were detected in an ABI PRISM 3100 Avant (Life Technologies) and checked with BLASTN (megablast) in GenBank. Sequence contigs were trimmed, assembled, aligned and manually edited in CLC DNA Workbench software (CLC-Bio, Mühltal, Germany). Regions containing gaps or poorly aligned sites were manually removed from the dataset.
Table 2. PCR cycling conditions and primer sequences for 18S, rbcL and internal transcribed spacer (ITS) primers
Life Technologies Veriti thermal cycler (Life Technologies).
Primer nomenclature for the newly described primers follows that of Gargas & DePriest (1996) and Nelsen & Gargas (2006), with primer locations corresponding to those of Chlamydomonas reinhardtii (GenBank Accession M32703).
This primer was initially described as one which excludes algal amplification, while preferentially amplifying fungi (Gargas & DePriest, 1996); however, we were able to successfully use this primer to amplify algae from some lichen thalli. It is possible that with an increased annealing temperature, this primer would preferentially amplify fungi over algae. This primer was also used to amplify the 18S from algae by Schmitt & Lumbsch (2001) and Muggia et al. (2011).
Each 18S and rbcL sequence was BLASTed against the entire GenBank database. For each sequence, the top 100 BLAST hits belonging to the class Trebouxiophyceae or Ulvophyceae were included in the dataset. In total, our 18S and rbcL datasets included 415 and 431 sequences, respectively. After the exclusion of identical sequences, datasets were reduced to 284 (18S) and 278 (rbcL) unique sequences. Datasets were analyzed separately. For each dataset, the nucleotide sequences were aligned with the program MAFFT v5 (Katoh & Standley, 2013) and introns were removed manually. Phylogenetic relationships and their confidence values were inferred using RAxML (1000 bootstrap pseudo-replicates), and Bayesian Markov chain Monte Carlo (BM) method (Metropolis et al., 1953; Hastings, 1970) as implemented in MrBayes 3.1 (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003). All ML searches followed a GTRGAMMA model of molecular evolution. For BM analysis, the best-fit model was selected with the corrected Akaike Information Criterion and the Bayesian Information Criterion as implemented in jModelTest 0.1.1 (TrNef + I + G, −loge = 10 464.3541 for 18S rRNA; 012212 + I + G + F, −loge = 23491.1044 for rbcL; Guindon & Gascuel, 2003; Posada, 2006, 2008). All trees were rooted and BM phylograms including posterior probabilities were computed with five million generations, one out of every 100 trees was sampled and the first 12 500 trees were discarded as burn-in (likelihoods below stationary level). In all analyses, representatives of Ulvales were used as outgroup. ML trees of the two loci were graphically displayed with FigTree (http://tree.bio.ed.ac.uk/software/figtree/). Several clades were collapsed for clarity of presentation.
We used alternative hypothesis testing to evaluate whether our data are sufficient to reject monophyly of Dictyochloropsis s.l. Using the GTR + I + G nucleotide substitution model, constrained and unconstrained trees were inferred, as implemented in Tree-PUZZLE 5.2 (Schmidt et al., 2002). We used two tests to compare the different topologies: the Shimodaira–Hasegawa (SH) test (Shimodaira & Hasegawa, 1999) and the expected likelihood weight (ELW) test (Strimmer & Rambaut, 2002).
Haplotype networks were inferred for 18S and rbcL alignments of the photobionts of the host genera Crocodia, Dendriscosticta, Lobaria, Lobariella, Pseudocyphellaria, Ricasolia, Sticta and for the ITS alignment of 24 photobionts of samples that amplified at seven microsatellite loci using statistical parsimony with multibase indels coded as single characters with TCS v1.21 (Clement et al., 2000).
Analysis of specificity and microsatellite population genetics
Seven highly variable microsatellite markers (SSRs) specific for the algal strain AB06.006A2 (Dal Grande et al., 2010; Widmer et al., 2010) were tested on all samples.
In order to test the existence of photobiont-mediated guilds sharing the algal strain AB06.006A2, we used SSR markers to further determine the algal genotype in two old-growth forests harbouring rich Lobariaceaen communities, that is, a primary laurel forest (laurisilva) in Madeira (Portugal; 32°45′37.0″N, −17°00′57.8″E, 824 m above sea level (a.s.l.)) and a subtropical mountainous broadleaved evergreen forest in Taiwan (Hualien County; 24°10′05.6″N, 121°17′29.8″E, 2938 m a.s.l.). Our sampling design was conceived to maximize the likelihood of collecting an identical photobiont genotype in different fungal host species. From the geographic center of the population, ten trees within an area of c. 900 square meters were checked for Lobariacean species. Thalli of Lobaria and Dendriscosticta species for each cardinal point were collected over the whole height of each tree. When several thalli of the same lichen were found at the same height of a tree, a maximum of three thalli per species was collected, for a total of 249 specimens for Madeira, and 205 for Taiwan.
Microsatellite genotyping was performed on total lichen DNA following Dal Grande et al. (2012). Fragment lengths were determined on a 3730 DNA Analyzer (Life Technologies), and electropherograms were analyzed with GENEMAPPER 3.7 (Life Technologies) using LIZ-500 as internal size standard. Microsatellite data were deposited in the Dryad Digital Repository: http://doi.org/10.5061/dryad.4q6b9. For each sampling site, photobiont populations were defined according to the fungal host (Table 3). For each population, the number of multi-locus genotypes (MLGs), unbiased genetic diversity (h) and number of alleles were calculated in GenAlex v6.5 (Peakall & Smouse, 2012). Samples having identical MLGs were treated as clones. The extent of interlocus recombination for the photobiont of each population was characterized using Multilocus 1.2 (Agapow & Burt, 2001). The occurrence of identical algal MLGs in different lichen fungi was represented as I × J contingency tables using the function visweb in R, where each cell depicts the number of interactions recorded between a clonal algal MLG and a specific combination of at least two different lichen fungi. Algal genotype occurrences within fungal species (presence–absence matrices) were also represented as a network using the function plotweb (package bipartite; Dormann et al., 2008, 2009) and the package pheatmap in R (R Development Core Team, 2012). Algal genotypes and fungal species correspond to the nodes of the network, and genotype occurrences in lichens correspond to the links between the nodes. Host specialization was further inferred by grouping photobiont MLGs by fungal host species and calculating pairwise linearized FST in Arlequin v3.01 (Excoffier et al., 2005). Only groups with > 10 samples were included in the comparisons. Significance was assessed with permutation tests (9999 replicates). Rarefaction curves for the cumulative number of unique algal MLGs and the number of samples analyzed in the two sites were calculated in MOTHUR v1.22.2 (Schloss et al., 2009).
Table 3. (a) Information on two photobiont populations (Madeira, Taiwan) structured by mycobiont species (b) Pairwise FST values for photobionts, structured by mycobiont species, for two populations in Madeira and Taiwan (only species with > 10 specimens are included)
MLG, multi-locus genotypes; h, unbiased genetic diversity; IA, Index of Association (P < 0.01); rD, rBarD (P < 0.01).
(b) All values are significant (P < 0.001) as assessed by Bonferroni corrected permutation tests.
L. crassior (n = 41)
L. isidiophora (n = 34)
L. japonica (n = 26)
D. platyphylloides (n = 81)
D. praetextata (n = 16)
We obtained 15 18S rRNA and 14 rbcL sequences of green algal cultures attributed to the genus Dictyochloropsis s.l. and 28 18S rRNA and 38 rbcL lichen photobiont sequences (Table 1). Sequencing reactions of uncultured lichen photobionts produced clean reads at all loci, indicating that a single photobiont genotype was predominant in each thallus. The final 18S rRNA alignment contained 1022 sites, whereas the rbcL alignment contained 994 sites (data available from the Dryad Digital Repository: http://doi.org/10.5061/dryad.4q6b9). Although the position of some clades varied in the phylograms inferred from the two loci data, the overall topological differences of highly supported clades (ML > 75%, PP > 0.95) did not represent significant conflict (Fig. 1, Supporting Information Fig. S1). Many backbone nodes of the 18S and rbcL trees were poorly supported, as already reported in Nakada et al. (2008) and Leliaert et al. (2012). The nuclear internal transcribed spacer (ITS) sequences were too divergent to produce a reliable alignment large enough for phylogenetic analysis. However, ITS2 secondary structure analysis may be used for the discrimination of algal strains at the subgeneric level (Fig. S2).
Alternative hypothesis test using both ELW and SH-tests showed that the hypothesis of Dictyochloropsis s.l. to be monophyletic can be rejected (P < 0.001). This genus is partitioned into two independent and well-supported trebouxiophyceaen clades which likely represent distinct genera (ML = 98%, PP = 1 for 18S, ML = 88%, PP = 0.99 for rbcL; Figs 1, S1), and are here referred with the names Dictyochloropsis clade 1 and clade 2. Dictyochloropsis clade 2 grouped as sister to Viridiella fridericiana with moderate support in the 18S phylogeny (ML = 78%, PP = 0.96), whereas it is currently impossible to determine a sister group for Dictyochloropsis clade 1.
Dictyochloropsis clade 1 is composed of free-living, aerophytic strains only. Dictyochloropsis clade 2 is represented by lichen photobionts (Figs 1, S1; in green) and by epiphytic, free-living strains from Austria (SAG2036), Japan (NIES-378, SAG2069, SAG2154) and Malaysia (CAUPH603 and CAUPH8602). Among the lichenized strains, Dictyochloropsis clade 2 showed a low selectivity for lichen genera, being associated with members of the Lobariaceae (Crocodia, Dendriscosticta, Lobaria, Lobariella, Pseudocyphellaria, Ricasolia, Sticta), Mycocaliciaceae (Chaenothecopsis, SAG 27.81), Brigantiaeaceae (Brigantaea, e.g. CCHU 5616), Megalosporaceae (Megalospora), Coniocybaceae (Chaenotheca, SAG 244.80), Phlyctidaceae (Phlyctis, UTEX LB 2599/2612) and Ramalinaceae (Biatora) (Fig. 1).
The photobiont of a specimen of Catillaria chalybeia collected near Zürich (Switzerland) belonged to the Elliptochloris clade which contains, amongst others, symbiotic algae of lichens and sea anemones (Zoochlorellae), whereas a specimen collected near Lunz (Austria) was reported to be lichenized with Dictyochloropsis reticulata sensu Tschermak-Woess (1984).
Photobiont selectivity in the Lobariaceae
All 18S and rbcL photobiont alleles of the studied lichens of the Lobariaceae belonged to Dictyochloropsis clade 2 except for the photobiont of two unknown species of Sticta and Pseudocyphellaria (Taiwan, unknown alga).
According to 18S rRNA and rbcL data, the same photobiont was found in the following lichen fungi: Sticta canariensis/S. aff. neopulmonaria (Sym3), Lobariella sp./Ll. pallidocrenulata (Sym1), Crocodia aurata/Lobaria patinifera (Sym4), Dendriscosticta platyphylla /D. wrightii /L. macaronesica /L. pulmonaria /R. amplissima (Sym5), P. lividofusca/P. multifida/S. latifrons (Sym7), P. lindsayi/S. subcaperata (Sym8), P. fimbriata/P. homeophylla (Sym9), and Sticta sp./Pseudocyphellaria sp. (Taiwan, Yilan County, Ming Ch'ih forest) (Figs 1-3).
The seven alga-specific microsatellite markers amplified DNA of photobiont Sym5 only, for a total of four Dendriscosticta, 16 Lobaria and one Ricasolia species. Their allele lengths were in the same range or matched the allele lengths of the photobiont of L. pulmonaria (Widmer et al., 2010, 2012; Dal Grande, 2011; Dal Grande et al., 2012).
A 97–100% identity in ITS sequence of the photobiont of these samples further indicated that Sym5 is a single, coherent species.
The 7-loci microsatellite analysis of Sym5 photobionts from 249 (Madeira) and 205 (Taiwan) thalli of Lobariaceaen lichens resulted in a total of 79 (Madeira) and 110 (Taiwan) photobiont multi-locus genotypes (MLGs). Microsatellite markers showed considerable variation at the population level (h Madeira = 0.614, h Taiwan = 0.756; Table 3). Of these, 35 (Madeira) and 78 (Taiwan) were unique MLGs, occurring only once in the population. Nineteen MLGs in 72 thalli (Madeira) and eleven MLGs in 32 thalli (Taiwan) occurred multiple times with the same fungal host (i.e. were clonal within the same lichen being the result of potential vertical transmission). Twenty-one MLGs for a total of 142 (Madeira) and 25 MLGs for a total of 95 thalli (Taiwan) were shared among different fungal hosts (Figs 4, S3). When we compared photobionts associated with different fungal hosts using FST, none of the species had significantly different photobiont populations from all other species (FST Madeira = 0.03−0.09; FST Taiwan: 0.03−0.07), except for L. pulmonaria in Madeira which displayed the highest number of private alleles (PA = 15, Table 3) and FST = 0.13−0.21. No identical algal MLGs were found between the two localities. Algal rarefaction curves computed for the whole dataset of Madeira (249 samples) and Taiwan (205 samples) did not reach saturation (Fig. 4a), suggesting that further genotyping would have revealed more algal MLGs in each population.
Polyphyly of the genus Dictyochloropsis s.l.
The taxonomy of algae of the genus Dictyochloropsis s.l. has been questioned based on recent morphological studies and molecular data (Škaloud et al., 2005; Widmer et al., 2010; Thüs et al., 2011). Partial nuclear 18S rRNA and rbcL cpDNA sequences were used to infer phylogenetic relationships among several lineages of both free-living and lichenized green algae attributed to this genus. Our results indicate that the genus Dictyochloropsis s.l., as currently circumscribed, is polyphyletic. The genus is split into two distinct, highly supported lineages. The morphology of algae of one lineage, Dictyochloropsis clade 1, is congruent with the original description of the genus, and includes the type species, D. splendida (Geitler, 1966; Škaloud et al., 2007). This lineage is composed of free-living, aerophytic algae that have a peculiar chloroplast ontogeny with a distinct parallel arrangement of lobes, and reproduce only via autospores (Škaloud et al., 2007). By contrast, Dictyochloropsis clade 2 is composed of free-living as well as of lichenized algae. These algae have one evenly perforated chloroplast and produce both aplanospores as well as zoospores with a typical insertion of the flagella (Tschermak-Woess, 1980, 1984; Nakano & Isagi, 1987; Škaloud et al., 2007). The high support from both nuclear and plastid loci and the differences in chloroplast morphology indicate that these clades most likely represent different genera.
Both Dictyochloropsis clade 1 and clade 2 contain several undescribed taxa whose phylogenetic position cannot be fully resolved with the loci used. Considering the small number of isolates available for each species, a more extensive taxon sampling and sequencing data of more variable loci is needed before drawing taxonomic conclusions on the taxa of the two genera. Preliminary analyses of ITS rRNA sequencing data show that molecular signatures based on unique ITS secondary structures can be used for delimiting particular algal species, as recently shown for other algal groups (e.g. Asterochloris, Škaloud & Peksa, 2010; see Fig. S2; Coleman, 2009). In particular, the surprising diversity of photobionts of Sticta and Pseudocyphellaria s.l. species collected from South America and New Zealand is an issue certainly worthy of further study.
Our results are in concordance with molecular data for other algal groups that have revealed extreme polyphyly and cryptic diversity of morphologically simple genera of diatoms (Lundholm et al., 2006; Amato et al., 2007; Sato et al., 2008; Vanormelingen et al., 2008), free-living green algae (Lewis & Flechtner, 2004; Vanormelingen et al., 2007) and lichenized algae (Trebouxia, Kroken & Taylor, 2000; Coccomyxa, Friedl et al., 2007).
Patterns of association of lichenized algae
In Dictyochloropsis clade 2, many lineages were lichenized. Algae of this genus are found in association with members of the Lobariaceae (Crocodia, Dendriscosticta, Lobaria, Pseudocyphellaria, Ricasolia, Sticta), Mycocaliciaceae (Chaenothecopsis), Brigantiaeaceae (Brigantaea), Megalosporaceae (Megalospora), Coniocybaceae (Chaenotheca), Phlyctidaceae (Phlyctis) and Ramalinaceae (Biatora).
Several species of Lobariaceae associate with Dictyochloropsis clade 2 photobionts (Tschermak-Woess, 1984). Two undescribed Pseudocyphellaria (SCH22290/97) and Sticta (SCH22386/79) species from Taiwan associate with a photobiont from a different algal group related to an endophyte of tropical trees, Heveochlorella hainangensis. It is unlikely that these divergent sequences represent contaminating epibionts, as the sequencing of DNA extracts of isolated photobiont layers obtained by scraping the lichens using a sterile scalpel resulted in identical sequences.
It has been suggested for lichens as well as for other symbiotic systems that reproductive strategies and ecological factors are the driving forces leading to specificity (Yahr et al., 2006; Elvebakk et al., 2008; Rikkinen & Virtanen, 2008; Piercey-Normore, 2009; Otálora et al., 2010; O'Brien et al., 2013). Lichen fungi of Lobariaceae are generally associated with old-growth forests and display both dispersal mechanisms – that is, asexual dispersal leading to vertical photobiont transmission, and sexual dispersal leading to horizontal photobiont transmission (Scheidegger & Werth, 2009; Dal Grande et al., 2012; Werth & Scheidegger, 2012). For algal species with wide distributions, we did not find any case of high reciprocal selectivity between algal haplotype groups and single fungal species. Therefore, even in predominantly vegetative species such as L. pulmonaria, low-frequency sexual reproduction can effectively contribute to photobiont switching (Ohmura et al., 2006; Dal Grande et al., 2012).
For Nostoc-containing lichen symbioses it has been suggested that geographic patterns do not guide the evolution of specificity in lichens (O'Brien et al., 2005, 2013; Stenroos et al., 2006; Elvebakk et al., 2008; Otálora et al., 2010). In our study, we have evidence to both support and reject this hypothesis, depending on the algal clade analyzed. On the one hand, we found that lichens growing in different geographic regions may share identical 18S and/or rbcL photobiont haplotypes. This is evident for the alga Sym4 found in one Crocodia species from New Zealand, Europe and North America and one Lobaria species sampled on the Galápagos islands, and for the alga Sym5 found in three Lobaria, two Dendriscosticta and one Ricasolia species sampled in Europe, as well as East Asia and North America (Figs 2, 3, Table 1). On the other hand, we found unique photobiont haplotypes in lichens with limited distributions. For example, algae Sym2 and Sym3 were only found in Lobariaceae of Central and South America (Costa Rica, Colombia, Brazil). We also identified several photobiont clades that were exclusive to New Zealand (Sym7, Sym8 and Sym9) or the distinct photobionts from Taiwan.
Interestingly, we did not find any generalist Lobariaceaen taxa, that is, fungi associated with more than one photobiont haplotype group/species, except for C. aurata. The ability to associate with diverse photobionts, possibly adapted to different ecological niches may allow the mycobiont to colonize a wider range of habitats (Yahr et al., 2006; Nelsen & Gargas, 2008). It has been shown in other lichen systems (e.g. Antarctic cyanolichens, Wirtz et al., 2003; Protoparmeliopsis muralis, Muggia et al., 2013) that mycobionts having broader and varied ecological distribution have low photobiont selectivity. However, the high selectivity for certain Dictyochloropsis clade 2 strains presented in this study is consistent with the fact that fungi of the Lobariaceae have narrow ecological niches. Representatives of this family are known to be susceptible to high light (Gauslaa & Solhaug, 1999) and to favor habitats with high air humidity (Pannewitz et al., 2003), and they colonize forest areas of high ecological continuity (Scheidegger, 1995).
Photobiont-mediated guild hypothesis
The results of microsatellite fingerprinting show that the alga Sym5 is a common photobiont among Lobariacean lichens of the genera Lobaria (16 species) and Dendriscosticta (4 species) and Ricasolia (1 species, see Table 1). Rikkinen et al. (2002) hypothesized that co-occurring lichen fungi may exhibit a high level of selectivity at the community level. This was shown at the species level for lichens associated with Asterochloris and Nostoc photobionts (Paulsrud et al., 1998; Škaloud & Peksa, 2010). In this study, we used highly variable microsatellite markers developed for Sym5 to show that high selectivity at the community level can be linked to horizontal transmission of photobiont clones among lichens with similar microhabitat requirements.
Population genetics analyses using microsatellites in two long-lived forests harboring rich Lobariaceaen communities (Madeira, Taiwan) showed a very high genetic diversity of the photobiont species. Rarefaction analysis of algal MLGs suggested the presence of high levels of intraspecific cryptic diversity. A denser sampling is therefore needed in order to reveal the full algal diversity associated with fungi of the Lobariaceae in these communities.
Overall, the genetic structure of the photobiont at both sites was that of a clonal organism (IA Madeira = 0.13, P = 0.01; IA Taiwan = 0.39, P < 0.01). Analyses at the level of the single algal clones (i.e. identical multi-locus genotypes at seven SSR loci) indicate a surprising amount of symbiont switching (probably involving dissociation and relichenization). All fungal species analyzed associate with more than one genotype shared between at least two fungal hosts. In many cases, we observed recurrent photobiont clones shared between two or more host species, for example, algal clones aM, bM, cM, dM found in three Lobaria species in the population of Madeira, and clones aT, bT, cT, dT associated with two Lobaria and two Dendriscosticta species in the population of Taiwan (see Figs 4, S3). Thus, horizontal interspecific transfer of the photobiont was common in both green-algal lichen communities, as recently found for cyanobacterial lichen communities (O'Brien et al., 2013). This suggests that a diverse range of fungal partners plays a role in increasing the ecological range of the alga. Thus, like mycobionts, switching partners is a way for the alga to widen its habitat range and to colonize different ecological niches, thus becoming able to survive under new selective pressures. On the fungal side, when algal shifts are common enough, the selection–drift balance should attenuate the effect of detrimental algal mutations on fungal fitness (Buschbom & Mueller, 2006; Mikheyev et al., 2007; Nelsen & Gargas, 2008, 2009; Queller & Strassmann, 2009). Furthermore, according to the photobiont-mediated guild hypothesis (Rikkinen et al., 2002), the dispersal ecology of the guild may be shaped by different reproductive strategies of its members. Ideally, a guild has two kinds of members: predominantly asexual core species and predominantly sexual fringe species. Core species (L. immixta, L. macaronesica, L. pulmonaria in Madeira, and L. isidiophora, L. spathulata and Dendriscosticta praetextata in Taiwan) can effectively co-disperse photobionts at the local scale by producing large numbers of symbiotic diaspores. These photobionts, as shown in our study, can be incorporated via horizontal transmission in fringe species (L. sublaevis, L. virens in Madeira; L. crassior, L. orientalis, Dendriscosticta platyphylla and D. platyphylloides in Taiwan). The common benefit for the guild members relies on the fact that, so long as even a few photobiont cells survive, they can significantly promote the re-colonization of the area by the whole fungal guild (Piercey-Normore & De Priest, 2001; Rikkinen et al., 2002; Piercey-Normore, 2006, 2009; Rikkinen, 2009). The guild concept was recently challenged for cyanobacterial lichen symbioses by the discovery that fungal hosts (Peltigera spp.) previously thought to be highly specialized on one cyanobacterial cluster, are in fact associated with at least two divergent groups of Nostoc (O'Brien et al., 2013). In our study, however, the most extensively sampled species, L. pulmonaria, associates with a single species of Dictyochloropsis clade 2 (Sym5), and all other species associated with Sym5 were never found associated with other photobiont species, too. The investigated lichen communities therefore likely represent a horizontally linked system of fungi highly specialized on a single, shared photobiont.
In conclusion, our results support the hypothesis that, from the fungal perspective, co-evolution between photobionts and their fungal hosts occurs at a deeper phylogenetic level. Thus by ignoring the guild scale, the extent of interactions and co-evolution in lichens may be underestimated. The photobiont-mediated lichen communities described in our study were in accordance with a model in which the algal photobiont acts as a landscape for a community of potentially competing fungi. Such patterns have been observed in other symbioses, for example, the intracellular bacterial genus Wolbachia and aphids, fungus-growing insects, corals and root symbioses (Chapela et al., 1994; Okuma & Kudo, 1996; Wulff, 1997; Rowan, 1998, 2004; Herre et al., 1999; Stanton, 2003; Pochon & Pawlowski, 2006; Aanen et al., 2007; Mikheyev et al., 2007; Porras-Alfaro & Bayman, 2007). Future studies should aim to clarify the role of different reproductive strategies of guild members (core and fringe species) in the evolution of photobiont-mediated guilds.
We thank Bibiana Moncada and Robert Lücking for specimens from South America, Soili Stenroos and Filip Högnabba (Helsinki) for specimens from New Zealand, Frank Bungartz (Charles Darwin Research Station, CDRS) and Karen Dillman (Tongass National Forest Service) for specimens from Galapagos and Alaska, Jeng-Tze Yang and Jeng-Long Tsai National Chung Hsing (University of Taichung, Taiwan) for organizing collecting permits in Taiwan and for field assistance, Steve Leavitt for specimens from North America, Carrie Andrew and Robert Lücking for sequences, and Philipp-André Schmidt for rarefaction analysis. This work was supported by the Swiss National Science Foundation (projects 31003A-105830 and 31003A-127346 to C.S.), the Botanical Society of America and the University of Chicago (to M.P.N.), a fellowship from the German Academic Exchange Service (DAAD) (to G.S.).