OUP user menu

The genetic structure of the cosmopolitan three-partner lichen Ramalina farinacea evidences the concerted diversification of symbionts

del Campo Eva M., Catalá Santiago, Gimeno Jacinta, del Hoyo Alicia, Martínez-Alberola Fernando, Casano Leonardo M., Grube Martin, Barreno Eva
DOI: http://dx.doi.org/10.1111/j.1574-6941.2012.01474.x 310-323 First published online: 1 February 2013

Abstract

The epiphytic lichen Ramalina farinacea is distributed throughout the northern hemisphere in which the same two algal Trebouxia species (provisionally named TR1 and TR9) coexist in every thallus. Ramalina farinacea symbionts were characterized based on the two fungal nuclear loci (nrITS and rpb2) along with the primary and secondary structures of nrITS from each Trebouxia species in the Iberian Peninsula and Canary Islands. The results indicated a noticeable genetic differentiation between mycobionts from these two geographic areas and also suggested concerted changes in the three partners of a lichen symbiosis toward two clearly distinguishable ‘holobiont’ lineages. Modeling of ITS2 RNA secondary structures suggested their temperature sensitivity in TR1 but not in TR9, which was consistent with the observed superior physiological performance of TR9 phycobionts under relatively high temperatures. Both TR1 and TR9 phycobionts have been also found in a variety of taxonomically distinct lichens with a preferably Mediterranean distribution, being TR1 much more widespread than TR9. Our observations support a model in which ecological diversification and speciation of lichen symbionts in different habitats could include a transient phase consisting of associations with more than one photobiont in individual thalli. Such diversification is likely to be promoted by different physiological backgrounds.

Keywords
  • ITS
  • lichen symbiosis
  • phycobiont coexistence
  • Ramalina
  • rpb2
  • Trebouxia

Introduction

Symbiotic life styles constitute an important evolutionary force that is driven by different patterns of associations (Paracer & Ahmadjian, 2000). Variations in symbiotic partnerships depend on multiple factors, including specificity, selectivity, and the mode of symbiont transmission within a selective environment. Specificity refers to the possible taxonomic range of acceptable partners, whereas selectivity denotes the frequency of association between compatible partners (Yahr et al., 2004). Here, we report an interesting variation in the symbiont patterns of lichens that might contribute to their evolutionary and ecological success. Lichens represent major radiations of ascomycetes in a symbiotic stage, characterized by a unique symbiogenetic phenotype of specific biological organization in the thallus (Chapman & Margulis, 1998; Margulis & Barreno, 2003; Barreno, 2004).

The associations of lichens involve at least two very different organisms: a heterotrophic fungus (mycobiont) and a photosynthetic (photobiont) cyanobacterium (cyanobionts) or unicellular green algae (phycobionts). Lichenization allows the mycobiont and photobiont(s) to thrive in habitats that would otherwise be unavailable to either one on its own (Kranner & Lutzoni, 1999). Thus, lichens are found in the most extreme environments, including deserts and high mountains. The important functional interactions between photobionts and mycobionts, and possibly with other symbionts such as bacteria (Grube & Berg, 2009), suggest that these partners evolved simultaneously. This assumption is easily understandable for lichens in which there is a vertical transmission of the photobiont. In such cases, the photobiont is dispersed within vegetative propagules comprising both symbiotic partners. Alternatively, the new photobiont can be incorporated within a fungal-germinating spore or vegetative structure, resulting in its horizontal transmission (Zoller & Lutzoni, 2003; Yahr et al., 2004, 2006; Nelsen & Gargas, 2008). Vertical transmission of photobionts has been hypothesized for asexual and nearly asexual lichen-forming fungal species (Cassie & Piercey-Normore, 2008), and horizontal transmission for sexual lichen-forming fungal species (Honegger, 1991; Werth & Sork, 2010).

Lichens with vertically transmitted photobionts would be expected to show genetic structures that tightly match those of their symbiotic partners. However, in some cases, joint dispersal does not imply the maintenance of association patterns, as suggested by Wornik & Grube (2010). For example, changes in a symbiotic organism's reproductive mode may be governed by trade-offs in the fitness of the symbiosis, as proposed by Buschbom & Mueller (2006). Thus, when the relationship between mycobiont and photobiont is optimal in a given environment, the preponderant reproductive mode would be asexual for species with mixed asexual and sexual propagation systems. When changes in the environment make the established symbiosis suboptimal, thus decreasing global fitness, sexual reproduction would be preferred as it provides access to new partners and produces variability through genetic recombination. Successful relichenizations therefore depend upon the mycobiont's ability to form alternative partnerships within a given habitat (Romeike et al., 2002). According to this view, moderate habitats favor the proliferation of a higher diversity of algal genotypes, while more restrictive habitats reduce the available pool of algae owing to the greater selection pressure (Doering & Piercey-Normore, 2009). The ability of lichens to establish new fungus-photobiont associations has enabled them to colonize varied and wide-ranging habitats, assuming that the specificity and ecological selection of the photobionts is supported by a high degree of flexibility. (Grube, 2010).

Very little is known about the mechanisms shaping the associations of lichen symbioses or the relative contributions of phylogenetic history and environmental variation. This is partly owing to poor knowledge of symbiont composition and diversity, as few tools are available for distinguishing and classifying lichen symbionts. However, the refinement of molecular methods has facilitated the detection of potentially novel species of phycobionts and mycobionts, which can then be confirmed on the basis of subtle differences in macromorphological or ultrastructural characters, discovered on their re-evaluation (Casano et al., 2011). This approach to previously hidden species has been successfully applied to lichen mycobionts (Crespo & Pérez-Ortega, 2009) and has brought numerous ‘cryptic species’ to light. The concept of ‘cryptic species’ is particularly relevant for algal partners within lichen thalli, given the lack in the lichenized stage of diagnostic characters that could be used to identify distinct photobionts by optical microscopy.

Internal transcribed spacers from the nuclear ribosomal cistron 18S-5.8S-26S rRNA gene (nrITS) have been extensively exploited as a molecular tool to study partner selectivity in lichens (Grube & Muggia, 2010) and to carry out phylogenetic reconstructions of mycobionts (Kelly et al., 2011; Schmull et al., 2011) and photobionts (Piercey-Normore, 2006; Nelsen & Gargas, 2008; Skaloud & Peksa, 2010). Moreover, studies based on nrITS have not been limited to their primary sequences but have been extended to ITS secondary structures, which show strong conservation in both lichen fungi and phycobionts (Beiggi & Piercey-Normore, 2007). In fact, the RNA secondary structure of ITS2 is the most conserved among eukaryotes. It includes canonical domains forming the helices of a ‘four-fingered hand’ among several conserved features (Mai & Coleman, 1997). Among the four helices, helix II is the most conserved, and almost always shows at least one pyrimidine–pyrimidine mismatch; helix III is usually the longest and often includes a YGGY motif; and helix IV contains a CAGG sequence at its base (Schultz et al., 2005). The maturation of large subunit rRNA includes a series of cleavages that result in the removal of the internal transcribed spacer (ITS2) and thus separates mature 5.8S from 25/28S rRNAs. Several investigations have demonstrated that the formation of a higher-order secondary structure within the assembling pre-ribosomal particle is a prerequisite for accurate and efficient pre-rRNA processing (Cote et al., 2002), and in sum, for functional ribosomes.

Patterns of lichen symbiotic associations within an ecological and a historical framework are highly diverse. They range from a mycobiont species that cultivates the alga species as a monoculture within its thallus (Piercey-Normore & Deduke, 2011) to a fungal species that associates with more than one algal lineage [e.g. Lecanora (Blaha et al., 2006) and Tephromela (Muggia et al., 2008, 2010)]. In some cases, different algal genotypes coexist within the same thallus [e.g. Rinodina (Helms et al., 2001) and Evernia (Piercey-Normore, 2006)].

The lichen Ramalina farinacea constitutes a particular case in which a fungal species specifically associates with two different Trebouxia phycobionts (provisionally named TR1 and TR9) coexisting within each lichen thallus (del Campo et al., 2010; Casano et al., 2011). The two phycobionts differ in their physiological responses to different environmental conditions (Casano et al., 2011), including oxidative stress (del Hoyo et al., 2011). According to Krog & Osthagen (1980), R. farinacea originated in the Macaronesian–Mediterranean region and thereafter gradually increased its area to include most of the temperate and boreal regions of the northern hemisphere, with the Canary Islands as its probable southernmost limit in the Atlantic region.

The primary aim of the present study was to investigate a peculiar symbiotic association found in R. farinacea and the possible co-diversification events resulting from a long-term continuous symbiotic association in populations from the Iberian Peninsula and the Canary Islands. We therefore examined the genetic variability of the TR1 and TR9 phycobionts within R. farinacea in relation to mycobiont genotypes and geographic location. The symbiont composition and genetic diversity of each sample was evaluated based on the primary and secondary RNA structures of the nrITS from each phycobiont (TR1 and TR9), along with the sequence of fungal nrITS and rpb2 genes. The obtained algal nrITS sequences corresponding to the TR1 and TR9 algae from R. farinacea and other Ramalina species were then used to search sequence databases to determine the presence of these phycobionts in other lichen taxa.

Materials and methods

Taxon sampling

For this study, 31 thalli from nine populations (Table 1) of R. farinacea were analyzed. Four populations were from the Iberian Peninsula: Asturias, Puerto de Maravio (As1–2 samples) and Puerto de Ventana (As3–5 samples); León (Le1–5 samples); Castellón (CS1–5 samples) and Ciudad Real (CR1–5 samples). Two populations were from the Canary Islands (Tenerife): Pinar de la Guancha (Gu1-5 samples) and Los Realejos (Re1-5 samples). Two populations were from California: Mount Hamilton, Santa Clara Co. (CA1 sample) and San Francisco Botanical Garden (CA2 sample). Additional details on the geographic coordinates, altitude, bioclimatic belt and phorophytes are provided in Casano et al. (2011). All of the studied materials were deposited in the Herbarium of the Department of Botany, Universitat de Valencia (Valencia, Spain), stored at a temperature of −20 °C.

View this table:

Locations of the collections of Ramalina farinacea samples and algae from data banks used in this study and their GenBank accession numbers

DNA isolation, amplification, and sequencing

All thalli were surface washed by immersion for 10 min in aqueous sodium hypochlorite (final concentration 3% w/v) and then with sterile distillated water prior to DNA isolation. Total DNA was isolated from each collected lichen thallus following the procedure of Cenis (1992), which was developed to rapidly extract fungal DNA for PCR amplification. The algal nrITS were PCR-amplified using the primers of Kroken & Taylor (2000) and Casano et al. (2011). Fungal nrITS were amplified using the primers ITS1F (Gardes & Bruns, 1993) and ITS4 (White et al., 1990). A portion of the fungal rpb2 gene, encoding DNA-directed RNA polymerase II polypeptide 2, was amplified with the primer pair RpolRfF: 5′-TGG TTT GTC CAG CAG AGA CC-3′ and RpolRfR 5′-TCT CAA GAT CTT CCG GAG TC-3′.

Amplification reactions were carried out in a total reaction volume of 50 μL and using Illustra Hot Start Mix RTG (GE Healthcare, NJ). Negative controls, without DNA template, were included in every round of PCR amplifications to eliminate false-positive results caused by contaminants in the reagents. Cycling conditions were as follows: one cycle at 94 °C for 2 min and 40 cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min and a final extension at 72 °C for 7 min. Amplification products were subjected to electrophoresis through 1.5% agarose gels. The corresponding bands were excised from the gel, purified with a DNA gel extraction kit (Qiaex II; Qiagen GmbH, Hiden, Germany) and directly used for sequence analysis. Purified amplicons were sequenced using the Big DyeTM Terminator Cycle Sequence Ready Reaction kit II (Applied Biosystems Co. Foster City, CA), separated by automated multicapillary electrophoresis, and further analyzed on an ABI Prism 3730 Genetic Analyzer (Applied Biosystems Co.).

Alignments, phylogenetic analyses, RNA secondary structure constructions, database searches, and genetic differentiation estimates

All the sequences determined in this study were aligned using the default parameters implemented in Muscle 3.6 (Edgar, 2004), by using seaview v4 (Gouy et al., 2010), and then adjusted with bioedit 7.0 (Hall, 1999). Data sets were subjected to maximum-parsimony (MP), maximum-likelihood (ML) using paup 4.0b10 (Swofford, 2003), and examined with a Bayesian approach by using beast (Heled & Drummond, 2010). Trees were displayed with figtree v1.3.1. (Rambaut, 2008). The model for nucleotide substitution was chosen according to the best score obtained in tests performed with jmodeltest 0.1 (Posada, 2009). In the MP and ML analyses, a heuristic search was carried out using random stepwise addition, tree bisection-reconnection (TBR), and a branch-swapping algorithm to determine the best tree. For the ML analyses, the transition/transversion ratio was estimated from the minimum evolution tree as implemented in paup. Tree length (TL) was used for selecting a true tree that corresponded with the minimum evolution criterion (Pauplin, 2000). For Bayesian analyses, the Markov chain Monte Carlo method from a random starting tree was initiated in the Bayesian inference and run for 108 generations. Sampling was every 100th generation. After chain convergence analysis, all samples obtained during the first 50 000 generations were discarded as burnin. Posterior probabilities for each node were inferred from the resulting consensus tree. Bootstrap probabilities (Felsenstein, 1985) were calculated to estimate the robustness of the clades from 1000 replicates of the data. Statistical parsimony networks were reconstructed using tcs 1.18 (Clement et al., 2000). This program takes into account certain population-level phenomena, such as recombination and the possible persistence of ancestral haplotypes in the population, which are not addressed by traditional methods (MP, NJ, and ML). The inclusion of haplotype networks allows the natural incorporation of the frequently non-bifurcating genealogical information associated with population-level divergences. Database searches were performed by using blast (Altschul et al., 1997). RNA secondary structures of algal nrITS sequences were constructed with c software (Matzura & Wennborg, 1996) and modeled at different temperatures (−10, 15, and 37 °C). Temperatures were selected after evaluating their effects of temperature during culture on growth and photosynthesis in isolated TR1 and TR9 phycobionts [partial results are published in Casano et al. (2011)]. To estimate genetic differentiation between populations, the dnasp software was used (Librado & Rozas, 2009). This program is able to compute several measures of DNA sequence variation within and between populations. FST measures the effect of population subdivision, which is the reduction in heterozygosity in a subpopulation owing to genetic drift. FST values range from zero, indicating panmixis or the absence of genetic divergence, to one, indicating complete isolation or extreme subdivision. Generally, FST values up to 0.05 imply negligible genetic differentiation, whereas with values > 0.25 very great genetic differentiation within the population analyzed can be expected. Snn is a measure of how often the nearest neighbors of sequences are found in the same locality. If a population is strongly structured, then the nearest neighbor of a sequence can be expected in the same locality. Thus, the value of Snn will be close to one when the populations at the two localities are highly differentiated. The Snn statistic is more powerful than other statistics such as those of Hudson et al. (1992) in cases with small sample sizes (10 or 15), especially in the presence of recombination and low to moderate levels of variation.

Results

Genetic variability of each of the two coexisting phycobionts within R. farinacea in relation to mycobiont genotypes

The genetic diversity of the fungal partners in R. farinacea lichens was analyzed based on the nrITS and the rpb2 gene, as described earlier (see Table 1 for information about specimens and GenBank accessions). Phylogenetic reconstructions were performed on 62 new sequences using 31 individuals of R. farinacea: 38 sequences from the Iberian Peninsula, 20 from the Canary Islands, and 4 from California. Initial estimations of fungal haplotype diversity (Hd) based on a combination of the two different molecular markers, by implementing the tcs 1.21 program, rendered a single genealogy, thus suggesting that all tested thalli belonged to the same species (data not shown). The obtained haplotype network revealed a clearly differentiated clade including fungal haplotypes of lichens collected in the Canary Islands. The clustering of fungal haplotypes from the Canary Islands was confirmed in phylogenetic reconstructions based on the same molecular markers (Fig. 1). The analysis of the two combined markers included 837 characters: 59 were variable and 11 were parsimony informative. Sequences from Ramalina complanata were used as an out-group (accession numbers: FJ356152 for nrITS and DQ883762 for rpb2) because of the relative proximity of this species to R. farinacea and the availability in GenBank of sequences for the two markers used to construct the phylogeny shown in Fig. 1. MP analyses yielded a tree length (TL) of 62, a consistency index (CI), excluding uninformative characters, of 0.9677, and a retention index (RI) of 0.9512. In the tree depicted in Fig. 1, two main clades are distinguished. Clade I contains several samples from the Iberian Peninsula (Asturias, León, Castellón, and Ciudad Real) and is well supported (bootstrap values of 1000 replicates). Clade II consists of samples from all populations (Iberian Peninsula, Canary Islands, and California) and is supported by a bootstrap value of 74. Samples from León, Castellón, and Ciudad Real had fungal haplotypes distributed throughout the tree of Fig. 1. Samples from Asturias had fungal haplotypes exclusively included in clade I. Samples from the Canary Islands and California had fungal haplotypes exclusively included in clade II. Additional genetic differentiation estimates, obtained with dnasp v5, rendered values of 1.000 and 0.623 for the Snn and FST statistics (all were P ≤ 0.0001) carried out for the nrITS, and 0.927 and 0.678 for the Snn and FST statistics (all were P ≤ 0.0001) for the rpb2 gene. These values suggested a high degree of divergence between fungal genotypes from the Iberian Peninsula and the Canary Islands.

Phylogram based on the analysis of fungal nrITS and rpb2 gene sequences in specimens of Ramalina farinacea, with Ramalina complanata as out-group (accession numbers FJ356152 and DQ883762 for nrITS and rpb2, respectively). Vertical bars on the right indicate different fungal haplotypes of F1–F13. Algal haplotypes are indicated at the right of each sample: TR1 haplotypes based on nrITS sequences (TR1a–TR1e) and TR9 haplotypes based on nrITS sequences (TR9a–TR9b) are indicated at the left and the right of the slash-marks, respectively. The predominant phycobiont associated with each fungal haplotype is noted in bold letters. Subclade (i) includes mycobionts from the Canary Islands. The tree was obtained using a Bayesian approach with beast. Values at the left and right of the slash-mark on the tree branches correspond to posterior probabilities inferred with Bayesian and MP methods, respectively. Below the branches are the bootstrap values recorded only for those lineages supported by 50% of all (1000) replicates. Subpopulations within the studied populations (the Iberian Peninsula, the Canary Islands and California) and their coordinates are shown in the inset.

Similar genetic differentiation estimates were obtained when dnasp v5 was applied to each separate phycobiont, with values of 0.672 and 0.222 for the Snn and FST statistics (all them with 0.001 < P < 0.01) for the TR1 phycobionts and 0.655 and 0.500 for the Snn and FST statistics (all 0.001 < P < 0.01) for the TR9 phycobionts. These results suggested a high degree of divergence between algal genotypes from the Iberian Peninsula and the Canary Islands, as previously reported for mycobionts.

In relation to the genetic diversity of phycobionts, the values of Hd, estimated with the dnasp v5 software, were as follows: absence of polymorphism for TR1 and 0.526 for TR9 phycobionts in the Iberian Peninsula; 0.644 for TR1 and without polymorphism for TR9 phycobionts in the Canary Islands. Regarding the genetic diversity of phycobionts with respect to mycobionts, Fig. 1 shows that the mycobionts from the Canary Islands form a sub-clade (i) within clade II, which is associated with only a single TR9 haplotype (TR9a) but with five algal TR1 haplotypes (TR1a–TR1e). Conversely, all samples from the Iberian Peninsula contained a single TR1 haplotype (TR1a) and two different TR9 haplotypes (TR9a-TR9b). If algal haplotype combinations (TR1 plus TR9) within each lichen thallus are considered, then seven different combinations are obtained (I–VII, Table 2). Among these, only combination II was present in the Iberian Peninsula and the Canary Islands, whereas the remaining combinations were only found either in the Iberian Peninsula (combination I), the Canary Islands (III–VI), or California (VII).

View this table:

Geographic distribution of haplotype combinations of Ramalina farinacea phycobionts based on algal nrITS sequences

Searching for TrebouxiaTR1 and T. TR9 phycobionts beyond R. farinacea

After defining the different algal haplotypes for each sampled thallus of R. farinacea, we searched for haplotypes similar to those of either TR1 or TR9 in other lichen taxa. Our search initially focused on other members of the Ramalina genus analyzed using the same generic primers for algal nrITS as used for R. farinacea. Sequence analyses identified at least five species different from R. farinacea but with either TR1 or TR9 Trebouxia algae as the phycobiont: R. calicaris,R. fastigiata,R. fraxinea,R. lusitanica, and R. siliquosa (see Fig. 3 for accessions). In all of these species, TR1 was the phycobiont, except R. fastigiata, in which TR9 was the phycobiont. Next, we searched for nrITS haplotypes similar to those reported for TR1 and TR9 algae in the NCBI nucleotide database. While approximately 100 sequences very similar to the TR1 haplotypes were found, only two were similar to the TR9 haplotypes in Tephromela atra (accessions EU551521 and EU551522), apart from the sequences of R. farinacea and R. fastigiata as obtained by our group.

To estimate the Hd for TR1 in relation to the taxa of its fungal partner, the genealogy shown in Fig. 2 was constructed with the tcs 1.21 program. As shown in Fig. 2, nrITS haplotypes similar to TR1 were present in a large diversity of lichens from eight different genera accounting for a total of 16 species: Lecanora (L. bicincta,L. cenisia,L. lojkaeana and L. rupicola), Ramalina (R. calicaris,R. farinacea,R. fraxinea,R. lusitanica and R. siliquosa), Tephromela (T. atra and T. grumosa), Amandinea (A. punctata), Anaptychia (A. runcinata), Evernia (E. prunastri), Schaereria (S. tenebrosa), and Lecidea (L. roseotincta). In the genealogy of TR1 haplotypes, a parent node included three different lichen species from different genera (L. rupicola,L. roesotincta, and T. grumosa) from which three main clades could be derived. Lineage L1 included 61 samples from 12 different species from four genera (Evernia,Lecanora,Ramalina, and Tephromela). Lineage L2 included 25 samples from five species from four different genera (Amandinea,Anaptychia,Lecanora and Tephromela). Lastly, lineage L3 included three samples from the species T. grumosa.

Parsimony network of algal TR1 haplotypes based on nrITS sequences associated with different fungi. The network was built using the tcs 1.21 program according to the 95% parsimony interval. Each circle corresponds to a particular haplotype. The sizes of the circles are proportional to the number of individuals bearing this particular haplotype. Numerals above circles correspond to algal haplotypes (ITSA) of Fig. 3. Each bar in the network represents a single mutation step and each small empty circles an additional change. Sequences with uncertainness at any nucleotide position have been excluded.

To study the relationships among TR1-like phycobionts, 433 characters were aligned with bioedit 7.0, using several sequences of Trebouxia decolorans as out-group (accessions: AJ969550, FJ705190, FJ705199, FJ705201, and FJ705206). MP analyses of the algal nrITS sequences resulted in a total of 65 variable characters, of which 42 were parsimony informative. MP analyses resulted in a TL of 95, a CI of 0.7895 and a RI of 0.9736. The phylogram depicted in Fig. 3 included 121 algal nrITS sequences similar to either TR1 (96 sequences) or TR9 (20 sequences) haplotypes and five corresponded with four T. decolorans and one Trebouxia sp. as an out-group. As expected, TR1 and TR9 haplotypes appeared in two separate clades. Clade I comprised TR1 haplotypes and clade II, TR9 haplotypes, with high bootstrap values of 88 and 100, respectively. The tree topology was consistent with the genealogy depicted in Fig. 2.

Phylogram based on the analysis of algal nrITS from TR1 and TR9-like phycobionts associated with different lichen taxa, using several sequences of Trebouxia decolorans as out-group (accession numbers: AJ969550, FJ705190, FJ705199, FJ705201 and FJ705206). Vertical bars on the right indicate different lichen species containing a phycobiont whose nrITS sequence is very similar to that of either TR1 or TR9 phycobionts from Ramalina farinacea. The tree was obtained using a Bayesian approach with beast. Values at the left and right of the slash-mark correspond to posterior probabilities inferred with Bayesian and MP methods, respectively. Below the branches are the bootstrap values recorded only for those lineages supported by 50% of all (100) replicates. The scale bar indicates substitutions/site. Each sample is named according to the number of algal haplotype (ITSA) followed by their GenBank accession number.

Algal nrITS2 RNA secondary structures of TrebouxiaTR1 and T. TR9 phycobionts

To further characterize algal diversity of the two Trebouxia phycobionts, which coexist in R. farinacea and are also present in other lichen species, we analyzed the RNA secondary structures of the nrITS2 region. This region comprises the intergenic sequence between 5.8S rDNA and 26S rDNA within the algal nrITS in all specimens noted in Fig. 3. Using the rnadraw program, we obtained two structural variants of ITS2, ITS2-A, and ITS2-B, when modeled at 15 °C. As shown in Fig. 4, TR1 and TR9 shared these two variants of the ITS2 RNA secondary structure, in which four distinct domains or helices form a ‘four-fingered hand’, as described in the Introduction. The main difference between ITS2-A and ITS2-B was the presence of an additional helix (IIa) in the ITS2-A structure, resulting in a shortening of helix III.

Haplotype genealogies and RNA secondary structures of Ramalina farinacea phycobionts constructed on the basis of nrITS2 RNA sequences. (a) Haplotype genealogy and RNA secondary structures of ITS2 RNA from samples collected in the Canary Islands. (b) Haplotype genealogy and RNA secondary structures of ITS2 RNA from samples collected in the Iberian Peninsula and California. Haplotype genealogies were built using the tcs 1.21 program. The depicted RNA secondary structures were constructed on the basis of the sequences of TR1 phycobionts with the rnadraw program. The ITS2-A and ITS2-B RNA secondary structures based on the sequences of TR9 phycobionts (not shown) were identical to those of TR1 phycobionts at each given temperature. Subpopulations within the studied populations (the Iberian Peninsula, the Canary Islands, and California) and their coordinates are shown in the insets.

The secondary structures of both ITS2-A and ITS2-B RNA changed differentially in TR1 and TR9 when the temperature was shifted to higher or lower values (Fig. 4). When the temperature was lowered from 15 to −10 °C, the same two RNA secondary structures, ITS2-A and ITS2-B were obtained in TR1 and TR9 (Fig. 4a and b). However, with an increase in temperature from 15 to 37 °C, a third RNA secondary structure, ITS2-C (Fig. 4b), appeared in TR1, whereas TR9 maintained the ITS2-A structure.

Discussion

Mycobionts and phycobionts from R. farinacea are differentially distributed and genetically differentiated between the Iberian Peninsula and the Canary Islands

One of the most remarkable finding of our study is the clear genetic differentiation of mycobionts from R. farinacea between the Iberian Peninsula and the Canary Islands. This differentiation was determined by three different and complementary approaches based on the combination of two nuclear-encoded molecular markers (nrITS and the rpb2 gene). First, a clearly differentiated clade including all samples from the Canary Islands was identified in the haplotype network (data not shown). Second, the clustering of fungal specimens from the Canary Islands within a monophyletic sub-clade (i) was confirmed in the phylogenetic reconstructions shown in Fig. 1. Third, estimates based on FST and Snn values obtained using the dnasp v5 program corroborated our initial finding of a high degree of genetic divergence between fungal genotypes from the Iberian Peninsula and Canary Islands.

Studies on symbiosis have generally concentrated on a single primary symbiont and its host. Consequently, the focus of studies of heritable changes in development has been on the host genome and, occasionally on the genome of a specific primary symbiont (often in the context of co-evolution studies). Nowadays, the hologenome theory of evolution considers the holobiont with its hologenome, acting in consortium, as a unit of selection in evolution (Rosenberg & Zilber-Rosenberg, 2011). The holobiont has been defined as the host organism together with all of its symbiotic microbiota (Gilbert et al., 2010). Consideration of the hologenome brings forth certain modes of variation in the holobiont, which are unique to this entity. One such variation is ‘microbial amplification’, in which there is an increase in one group of symbionts relative to others, which can occur when conditions change. An increase in the abundance of a particular microbe is actually equivalent to gene amplification (Rosenberg & Zilber-Rosenberg, 2011). In our study, when ‘universal primers’ for Trebouxia algae were used, a single algal nrITS sequence per thallus, which clearly corresponded to either TR1 or TR9 phycobionts, was consistently obtained, and never a mixed nrITS sequence. We interpret these results as reflecting the occurrence of ‘microbial amplification’, with an increase in either TR1or TR9 as suggested in previous studies (Casano et al., 2011). Accordingly, in this study we define the ‘predominant phycobiont’ to be the alga whose nrITS sequence is obtained after amplification with ‘universal or generic primers’ (Kroken & Taylor, 2000) for Trebouxia algae and the ‘non-predominant phycobiont’ to be the alga whose nrITS sequence is obtained only after amplifying with primers specific for either TR1 or TR9 (Casano et al., 2011).

After analyzing all the obtained algal nrITS sequences of the predominant and non-predominant phycobionts, we found that the differentiation of R. farinacea lichens between the Iberian Peninsula and the Canary Islands was also reflected in their TR1 and TR9 phycobionts. TR1 was the predominant phycobiont in all samples from the Iberian Peninsula and California, whereas TR9 was the predominant phycobiont in all samples from the Canary Islands, as previously reported (Casano et al., 2011).

Additionally, the predominant TR9 phycobiont in the Canary Islands was found to be linked with a series of fungal haplotypes that represent a clearly distinguishable sub-clade (i), shown in Fig. 1. If this finding implies an incipient specialization of R. farinacea fungal partners (enclosed in sub-clade i) in different geographic areas, then this is an important contribution to our understanding of lichen evolution. The differential specialization of symbionts is accompanied by a higher genetic diversity of both the mycobiont and the non-predominant phycobionts in the Canary Islands (Hd = 0.822 and Hd = 0.644, respectively) than in the Iberian Peninsula (Hd = 0.432 and Hd = 0.526). The higher diversity and the apparent specialization of Canary Islands populations may be explained by the island immaturity-speciation pulse model of island evolution, which posits that the opportunities for speciation have a broadly predictable relationship to the life cycle of oceanic islands (Whittaker et al., 2007). Opportunity drives the speciation rate and is greatest at a relatively early stage in the life cycle of an island, when the intrinsic carrying capacity exceeds species richness by the greatest margin. Tenerife is one of the youngest islands within the Canarian Archipelago, which is generally thought to have emerged initially as two or possibly three separate islands, the oldest dating back to c. 11 Ma. These proto-islands fused intro the single present-day island only within the last 3.3 Myr (Fernández-Palacios et al., 2011). Moreover, during their life span, volcanic oceanic islands are often subject to a series of catastrophic events. Thus, in terms of their geological dynamics, the Canary Islands are clearly different from other island groups (Whittaker et al., 2007).

ITS2 RNA temperature-dependent secondary structure variations provide new physiological clues regarding the persistence of TR1 and TR9 phycobionts within R. farinacea

The thermal variations of the ITS2 RNA secondary structure provided new physiological clues regarding the persistence of TR1 and TR9 phycobionts within R. farinacea. The RNA secondary structure of ITS2 is highly conserved among eukaryotes showing canonical domains forming a ‘four-fingered hand’ and several conserved features (Mai & Coleman, 1997). Maturation of the large subunit rRNA includes a series of cleavages that result in the removal of ITS2, separating mature 5.8S and 25/28S rRNAs. Several investigations have demonstrated that the formation of a higher-order secondary structure within the assembling pre-ribosomal particle is a prerequisite for accurate and efficient pre-rRNA processing (Cote et al., 2002). It has long been known that temperature-related alterations in rRNA processing greatly inhibit the production of large ribosomal subunits in the cytoplasm (Toniolo et al., 1973). To find a possible link between the different nrITS sequences of TR1 and TR9 phycobionts and distinct physiological traits, models of the RNA secondary structures of the ITS2s from all the TR1 and TR9 haplotype variants at different temperatures were created. This approach yielded important physiological insight into the persistence of TR1 and TR9 phycobionts within R. farinacea thalli. As reported in the Results, the RNA secondary structure of this key intergenic region seems to be highly thermally stable in TR9 phycobionts but thermally labile in TR1 phycobionts. In this study, two variants of the ITS2 RNA secondary structures were noted to occur at −10 and 15 °C in TR1 and TR9 phycobionts (Fig. 4). These two structures, ITS2-A and ITS2-B, may be comparable with the two alternative secondary structures previously proposed for ITS2 (Yeh & Lee, 1990; Joseph et al., 1999). Elements of both structural models are important in efficient rRNA processing (Cote et al., 2002). However, when the same structures were modeled at 37 °C, we obtained a third variant in the TR1 phycobionts (ITS2-C) but not in the TR9 phycobionts. This ITS2-C RNA structure lacked the elemental features necessary for the effective removal of ITS2 after transcription. These structural models are hypothetical, and biochemical experimentation is necessary to support their configurations. Nevertheless, it may be that the predicted thermal dependence of the ITS2 RNA secondary structure and its probable consequences on nuclear ribosomal cistron processing contribute to explain the superior physiological performance of TR9 phycobionts under relatively high temperatures, in contrast to TR1, which thrives at relatively low–moderate temperatures (Casano et al., 2011). It is therefore possible that the R. farinacea fungus utilizes the two algae to broaden its physiological behavior in response to changing environmental conditions.

Together, these results can be interpreted as an incipient specialization of mycobionts along with a bias in phycobiont predominance (TR1 in the Iberian Peninsula and TR9 in the Canary Islands). This specialization is probably imposed by gross ecological traits related to latitude (see inset of Fig. 1) and/or the continentality/insularity of geographic locations. The latitude of the two Californian localities was similar to that of the localities from the Iberian Peninsula (see inset of Fig. 1) and likewise subject to a Mediterranean climate. Interestingly, fungal haplotypes from California (F6 and F7) were included within clade II but outside sub-clade (ii) in Fig. 1. These two samples had TR1 as predominant phycobiont (as in all samples from the Iberian Peninsula) but the TR1b haplotype (as several samples from the Canary Islands). The adjacent positions of the Californian and Canarian samples (F6) in Fig. 1 and their phycobiont composition may suggest an intermediate diversification status of holobionts. However, a higher number of samples from California must be analyzed to confirm this assumption.

Changes in fungal genomes seem to be encompassed with changes in each algal genome, resulting in concerted changes in the three genomes. Such changes may illustrate a transient phase toward specialization by the fungal and algal partners in the different habitats and seems to be facilitated by the association with more than one phycobiont in each thallus.

The coexistence of two or perhaps more physiologically different Trebouxia taxa within the same lichen thallus migth be more common than previously thought

It has been shown that Ramalina species from different climates prefer distinct lineages of Trebouxia algae (Cordeiro et al., 2005). In Tephromela atra and T. grumosa, a number of lichen thalli were shown to have TR1 as the phycobiont (named Trebouxia spec. 2 in Muggia et al., 2008, 2010) while in only two samples of T. atra was TR9 the phycobiont. Interestingly, these two samples were collected from Ischia, a volcanic island lying at the northern end of the Gulf of Naples, in Italy. Preliminary studies in our laboratory with specimens of R. fastigiata (Catalá S, unpublished results) and T. atra (del Campo et al., unpublished results) from different localities of the Iberian Peninsula, using the same molecular markers as in R. farinacea, revealed the coexistence of TR1 and TR9 in all the analyzed samples. Therefore, the TR1/TR9 binomial may be more frequent than previously thought suggesting its successful colonization of at least Mediterranean and Macaronesian regions.

Lecanora rupicola also has TR1 its phycobiont referred to as Trebouxia spec. 2 in Blaha et al. (2006). Recently, an association between Lecidea roseotincta and two different Trebouxia species was reported (Schmull et al., 2011). One of the Trebouxia species has been named T. roseotinctae (accession HQ667310) but seems to be TR1, based on its nrITS sequences. In addition, at least six different Trebouxia phycobionts have been detected, including one that seems to be TR1, in R. fraxinea (Catalá et al., unpublished results). Interestingly, more than two of these algae occupied (by TEM) the same thallus. These considerations along with the results of this study suggest that the coexistence of at least two physiologically different Trebouxia strains or species within the same lichen thallus is a not uncommon occurrence.

TR1 phycobionts are widespread among lichens, with a preferably Mediterranean distribution

Our search results produced approximately 100 nrITS sequences very similar to those from Trebouxia phycobionts in a large variety of lichen taxa within Lecanorales but in distinct families, such as Lecanoraceae (Schmull et al., 2011) Parmeliaceae (Evernia prunastri, this work), Physciaceae (Helms et al., 2001), Ramalinaceae (several Ramalina species in this work), and Tephromelataceae (Muggia et al., 2008, 2010). As shown in Fig. 2, the majority of Hd in TR1 phycobionts descended from a single haplotype (ITSA-4). This ancestral haplotype, although not highly represented, seems to be widely distributed both phylogenetically and geographically. Phylogenetically, it is shared by three lichen species belonging to three distinct genera (Lecidea,Lecanora, and Tephromela) but it is absent in Ramalina. Geographically, it has been found within median–high latitudes, ranging from 38 to 43° in Spain and Italy to higher than 54° in Denmark and Norway. The observation of multiple pathways, including three distinct lineages and interconnections resulting in a network linking haplotypes, is consistent with recombination. In this network, haplotypes of intermediate ancestry, such as ITSA-1, ITSA-2, and ITSA-3, are associated with a number of fungi. Further down in the genealogy, a radiation of new haplotypes is observed, especially in Tephromela atra. It is noteworthy that the genealogy includes a similar number of samples from R. farinacea and T. atra, which belong to the same order (Lecanorales) and suborder (Lecanorineae) but to different families, that is, Ramalinaceae and Lecanoraceae, respectively. The distribution of these two lichen species is rather cosmopolitan, as they have been studied in samples collected in similar geographic areas in the Mediterranean regions of Italy and Spain. As shown in Fig. 2, in R. farinacea there are only two different haplotypes (ITSA-1 and 2), both rather ancestral and constrained to lineage 1. This contrast with the higher number of different haplotypes in T. atra, most of which belong to lineage 2 (e.g. haplotypes ITSA-7, 9, 11–16). This difference might reflect the different modes of reproduction of T. atra and R. farinacea, sexual and asexual, respectively. However, the presence of diverse and in some cases shared (e.g. ITSA-3 and 5) haplotypes in T. grumosa, which is closely related to T. atra, suggests an alternative explanation since T. grumosa reproduces asexually.

Most of the sequences shown in Figs 2 and 3 were from lichens collected in Mediterranean regions, with TR1 being more widespread than TR9. In the case of TR1, the same algal genotypes are found in association with taxonomically different fungi whose habitats have a preferred Mediterranean distribution (e.g. ITSA-1, ITSA-2, and ITSA-3 in Fig. 3). These observations, along with the physiological differences determined in the two phycobionts (Casano et al., 2011; del Hoyo et al., 2011), are consistent with an eco-physiological adaptation to Mediterranean conditions. Associations of phycobionts enclosed in particular algal clades with taxonomically different but ecologically similar lichens also have been observed in Asterochloris algae (Peksa & Skaloud, 2011). These associations include two distinct genera (Lepraria and Setereocaulon) belonging to a single family (Stereocaulaceae). In the present work, the same phycobionts were found to be shared in several distinct genera from five distinct families (Lecanoraceae,Parmeliaceae,Physciaceae,Ramalinaceae, and Tephromelataceae).

Experimental evidence indicates that the lichen association between photobionts and mycobionts increases tolerance to stress conditions (Kranner et al., 2005; Kosugi et al., 2009; Catalá et al., 2010). By changing the composition (qualitatively and quantitatively) of their algal partners, lichenized fungi may expand their ability to adapt to a broad range of ecological conditions. Sun & Friedmann (2005) noted that the ratio of consumer (fungi) vs. producer (algae) can be adjusted depending on the climatic conditions. According to their ‘community adaptation hypothesis’, the temperature of the habitat correlates with a higher proportion of producers to account for the respiration of both partners. Apart from this mode of adaptation, there is some evidence that widely distributed lichens can adjust their algal partner according to their presence in different habitats, following the concept of ‘habitat-adapted symbiosis’ (Rodriguez et al., 2008). For example, L. rupicola has different photobiont species in alpine vs. Mediterranean habitats (Blaha et al., 2006), and a similar situation was shown to occur in Lecidea fuscoatra (unpublished results). Our results suggest a third possibility of adaptation, in which physiologically distinct strains of the same or sister species are present in the same thallus. Whether two or perhaps more Trebouxia species coexist in the same lichen thallus in other lichen taxa remains the subject of further studies using approaches similar to those applied in R. farinacea. Additional physiological studies are also necessary to confirm our hypothesis that the coexistence of physiologically complementary photobionts within the same lichen thallus increases tolerance to changing and often stressful environments. Another question that remains to be answered is whether the proportion of these strains varies according to the ecological conditions.

Finally, it will be of interest to determine whether the ecological diversification and speciation of lichen symbionts in the different habitats includes a transient phase, consisting of associations between more than one phycobiont in individual thalli. Alternatively, the coexistence of multiple phycobionts may be crucial for increasing eco-physiological fitness under changing environments and therefore might be the rule in lichens living on a wide variety of substrata and in diverse habitats. These considerations along with the results of this study suggest that the coexistence of at least two physiologically different Trebouxia strains or species within the same lichen thallus is a not uncommon occurrence, even though the relative abundance of strains may vary substantially in different species or ecological settings.

Acknowledgements

This study was funded by the Spanish Ministry of Science and Innovation (CGL2009-13429-C02-01/02), the AECID (PCI_A_l024755/09), and the Generalitat Valenciana (PROMETEO 174/2008 GVA).

Footnotes

  • Editor: Ian C. Anderson

References

View Abstract