Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
• Here, cyanobiont selectivity of epiphytic lichen species was examined in an old growth forest area in Finland.
• Samples of the eight lichen species were collected from the same aspen (Populus tremula) and adjacent aspens in the same stand. The cyanobionts of these samples were compared with free and symbiotic Nostoc obtained from other habitats and geographic regions.
• Our results, based on the phylogenetic analysis of a partial small subunit of the ribosomal DNA (16S rDNA) and the rbcLX gene complex did not show any correlation with the geographic origin of the samples at any spatial scale. Instead, there was a correlation between the cyanobionts and the alleged taxonomy of their mycobionts.
• The results indicate that the lichen species examined are highly selective towards their cyanobiont partners. Only Lobaria pulmonaria proved to be more flexible, being able to associate with a wide range of Nostoc. A same Nostoc strain was found to form associations with taxonomically unrelated lichens indicating that the cyanobiont–mycobiont associations as a whole were not highly specific in the examined species.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Old aspens (Populus tremula) support a characteristic epiphyte flora in eastern Fennoscandia with many species that are rare or absent from the dominant forest trees, Picea abies, Pinus sylvestris and Betula spp. (Kuusinen, 1994, 1996). The most conspicuous epiphytes on aspen are large, mainly foliose cyanobacterial lichens (i.e. those having cyanobacteria as major photobionts or cyanobacterial cephalodia), such as Lobaria pulmonaria, Nephroma spp. and Peltigera spp. These species, being confined to P. tremula exclusively or more frequently in old growth forest than in managed stands (Kuusinen, 1996; Hedenås & Ericson, 2000), have been used as indicators of forest continuity and conservation value in boreal forests of Fennoscandia (Kuusinen et al., 1995; Nitare, 2000).
Kuusinen and Penttinen (1999) explored the spatial occurrence and frequency of eight cyanobacterial epiphytic lichens: L. pulmonaria, Nephroma bellum, Nephroma laevigatum, Nephroma parile, Nephroma resupinatum, Parmeliella triptophylla, Peltigera leucophlebia and Peltigera pratextata. They found that these species showed a significant preference to large aspens and tended to co-occur on the same trees. A potential explanation for the observed pattern may be the similar habitat requirements of the mycobionts and/or cyanobionts. The possibility that the species may even share the same cyanobacterial strain should also be considered. In this study we focus on the last hypothesis.
The objective of the present study was to examine the cyanobiont selectivity of eight epiphytic lichens: L. pulmonaria, N. bellum, N. laevigatum, N. parile, N. resupinatum, P. triptophylla, P. leucophlebia and P. pratextata. Two of the species examined, L. pulmonaria and P. leucophlebia are tripartite, consisting of a mycobiont, a photosynthetic green alga and a cyanobiont, Nostoc, in internal and external cephalodia, respectively (Jordan, 1970). The remaining six species are bipartite with a Nostoc as a primary photobiont. Our goal was to test whether these species, when occurring on the same aspens, share a common cyanobiont or have different, taxon-specific strains. We focused our study on one old-growth forest stand in Finland and collected samples at different spatial scales (i.e. from the same host tree, and adjacent trees in the same stand) to examine if there are Nostoc strains that are specialized into this forest area, or perhaps even into one phorophyte. To address these questions we use information from three gene loci; partial 16S rDNA and partial rbcL and complete rbcX genes of the rbcLX gene cluster in a phylogenetic context.
Materials and Methods
A total of 78 terminals were included in the analysis (see the Supplementary Material, Table S1). Almost half of them, 37, were collected and sequenced for this study from one forest stand within the Kotinen Nature Reserve, situated in the Evo area in Lammi parish, southern boreal Finland. Within this 25-ha study area (referred to as ‘Evo’ hereon) a total of 1245 P. tremula trees has been recorded (Kuusinen & Penttinen, 1999). We focused our collection on five aspens (aspens 1–5; see the Supplementary material, Table S1) where cyanobacterial lichens L. pulmonaria, N. bellum, N. laevigatum, N. parile, N. resupinatum, P. triptophylla, P. leucophlebia and P. pratextata, or at least most of them, had been reported to co-occur (Kuusinen, 1996). In addition, we included samples from aspens where only a few or only one of these species occurred (aspens 6–10; see Table S1). The 10 aspens were carefully inspected for the occurrence of cyanolichens. Binoculars were used to survey the lichens on upper parts of the trunks. All cyanolichens found on the 10 trunks were collected and analysed. Furthermore, we also collected samples from other than epiphytic habitats for comparison.
The cyanobionts of Evo samples were compared with free and symbiotic Nostoc obtained from other habitats and geographic regions, some of them collected by us, some obtained from GenBank. For example, we included specimens of P. leucophlebia from litophytic habitats (specimens Haikonen 21834, Haikonen 21759, Haikonen 21860 and Myllys & Kuusinen 463) to examine whether these share the same or different cyanobiont than epiphytic P. leucophlebia specimens. Microcystis aeruginosa and Synechocystis sp. from Chroococcales and two Scytonema species from Nostocales were used as outgroups.
In our analyses, we included only those terminals where all three gene regions had been sequenced to allow comparison and simultaneous analysis of the data sets without introducing large amounts of missing data. Consequently, many potential 16S sequences available in GenBank were excluded because they were not sequenced for rbcL and rbcX gene regions. The only exception was the inclusion of 16S sequence of cyanobiont from specimen L. pulmonaria Myllys & Kuusinen 376a, although rbcL and rbcX sequences for this taxon could not be obtained. This sequence was included only in the separate 16S analysis to allow a comparison between two separate extractions obtained from the same host specimen (see later).
DNA of the lichen samples was isolated from thalli of bipartite lichens and from pieces of cephalodia of tripartite lichens using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). From one L. pulmonaria specimen (Haikonen 23259), two separate extractions were performed from different internal cephalodia. From L. pulmonaria specimen Myllys & Kuusinen 376, one extraction was performed from cephalodia (Myllys & Kuusinen 376a), and the other one was obtained by scratching the lower surface of the thallus (Myllys & Kuusinen 376b). The extraction procedure was performed according to the manufacturer's instructions except that the liquid nitrogen phase was omitted. Instead, thallus or cephalodial fragments of approx. 1–3 mm in diameter were ground with a mini-pestle in 50 µl of the lysis buffer provided with the kits. The extracted DNA was eluted in 120 µl of the elution buffer included in the kits.
Three loci, a partial small subunit of the ribosomal DNA (16S rDNA) and the 3′-end of the rbcL gene and complete rbcX gene of the rbcLX gene cluster were amplified and sequenced from the cyanobionts using the following primers: 106F 5′-CGGACGGGTGAGTAACGCGTGA-3′, 781R 5′-GACTACTGGGTATCTAATCCCATT-3′ (Nübel et al., 1997) and CW 5′-CGTAHCTTCCGGTGGTATCCACGT-3′, CX 5′-GGGGCAGGTAAGAAAGGGTTTCGTA-3′ (Rudi et al., 1998), respectively. The partial 16S region covers approximately the first half of the gene, where most of the variable sites reside (Lohtander et al., 2003). Polymerase chain reaction (PCR) reactions were performed using Ready To Go PCR beads (Pharmacia Biotech, Cambridge, UK) following the manufacturer's instructions with the following reaction conditions: denaturation of 2 min at 95°C, then 30 cycles of 1 min at 95°C (denaturation), 1 min at 55°C or at 60°C in the case of rbcLX gene complex (annealing), and 1 min at 72°C (extension), followed by an elongation cycle of 7 min at 72°C. The PCR products were purified with Qiagen's QIAquick PCR Purification Kit and eluted with 43 µl AE Buffer.
The sequencing reactions were prepared using the BigDye Terminator Cycle Sequencing Reaction Kit vs 2.0 (PE Applied Biosystems, Foster City, CA, USA). Two alternative schedules were used: denaturation for 1 min at 96°C in both schedules, then 25/30 cycles with 10/30 s at 96°C, annealing for 5/15 s at 50°C, and extension for 4 min at 60°C. The samples were run on an ABI Prism 377 automatic sequencer from Perkin Elmer (Norwalk, CT, USA).
The nucleotide sequences of 16S rDNA, rbcL and rbcX genes were provisionally aligned with clustalx (Thompson et al., 1997) using default parameters. The hypervariable intergenic region between rbcL and rbcX genes (Rudi et al., 1998) was removed from the analysis. The rbcL and rbcX alignments were manually adjusted with reference to the amino acid code. However, the nucleotides coding for 24 last amino acids in Synechocystis rbcX gene were treated as missing data, since no similarity was found either to outgroup or ingroup taxa for these regions. The aligned sequences were subjected to parsimony analysis as implemented in paup*, version 4.0b10 (Swofford, 2002) with the following settings: heuristic search, random addition sequence with 100 replicates and TBR branch swapping. No more than 40 trees were saved for each replicate to save computation time. Gaps were treated as missing data.
Because the sequences, especially those from the rbcX gene, were variable in length, it was impossible to obtain an unambiguous alignment despite using amino acid code as the reference. This is problematic, since phylogenetic hypotheses are dependent on primary homology assumptions made during alignment (e.g. different alignments may produce different cladograms). In order to find the most optimal solution to our data, we decided, in addition to clustal alignment + paup analysis, to perform analysis with an optimization alignment or direct optimization (Wheeler, 1996) as implemented in the programme poy, version 3.0.6 (Gladstein & Wheeler, 1997). We used parsimony as an optimality criterion in our analysis. Direct optimization requires no separate alignment step prior to analysis, but multiple alignment and cladogram searching are done simultaneously. This is done by comparing different topologies with each other and trying to find one that minimizes transformations (i.e. nucleotide substitutions and insertion/deletion events that have to be assumed to have happened between the terminal nodes).
Three different data matrices were analysed in poy: (1) 16S data; (2) the combined rbcL and rbcX data; and (3) the combined 16S and rbcLX data. Before the analyses, the sequences were aligned using clustalx (Thompson et al., 1997) and then split into shorter fragments within conserved regions to save computation time and memory. As in the paup analysis, the hypervariable intergenic region between rbcL and rbcX genes (Rudi et al., 1998) was removed from the analysis. After cutting, the gaps were removed and unaligned sequence fragments were submitted to poy.
The poy analyses were performed using the IBMSC parallel supercomputer, located at CSC (Scientific Computing Ltd, Espoo, Finland). This is an IBM eServer Cluster 1600 system, constructed of 16 pSeries 690 nodes (each equipped with 32 POWER4 processors) and a SP Switch2 (Colony). Eight processors of one node were used for our analysis. The commands and a brief explanation of the commands (Janies & Wheeler, 2002) are given in the Supplementary Material, Appendix S1.
It is possible to assign different weights to the different types of transformations but, in our analyses, we preferred to treat them all as equal following the argumentation by Frost et al. (2001) and Grant and Kluge (2003) (i.e. equally weighted parsimony maximizes congruence over all data by minimizing the total number of hypothesized transformations). By contrast, any weighting scheme necessarily increases the number of events required to explain the data, making the preferred hypothesis less parsimonious (Grant & Kluge, 2003).
Various methods have been proposed for estimation of ‘reliability’ or ‘support’ of multiple phylogenetic hypotheses included in each cladogram. We have shown Bremer (Bremer, 1994) to facilitate the comparison of different portions of the tree. However, we agree with Kluge (1997) and Grant & Kluge (2003) who have challenged the use of these metrics altogether. Real tests of the current hypotheses will be provided only by the addition of empirical data. In our study, Bremer support values were calculated for the internal nodes using poy. However, in poy, these values are calculated on the basis of a branch-swapping (TBR) search, not on collapsing ever-more catholic searches, and therefore the values may be overestimations (Janies & Wheeler, 2002). However, all the values were calculated in the same way and thus are comparable with each other.
The poy analysis of the combined data set with 77 terminals resulted in three equally parsimonious alignments (and trees) of 1079 steps. The analysis took approx. 25 h and during this process over 66 million trees were evaluated. In the highly resolved strict consensus, free-living and symbiotic Nostoc samples form a well-supported monophyletic group with one free-living Nostoc specimen collected from USA as basal (Fig. 1). The cyanobionts of epiphytic lichens collected from Evo are distributed into the two major clades appearing in the tree (Fig. 1, I and II).
In clade I, Nostoc muscorum collected from France, and the cyanobionts of P. leucophlebia and P. praetextata specimens collected from Evo, group together with three of the four P. leucophlebia samples obtained from litophytic habitats from other parts of Finland. In clade II, the two cyanobionts sequenced from Nephroma laevigatum are restricted to one subclade (IIa) together with the cyanobionts of Degelia plumbea, L. scrobiculata, N. helveticum, N. tangeriense and Pseudocyphellaria crocata. This subclade includes also four cyanobionts of L. pulmonaria collected from different localities. All the cyanobionts of N. bellum, N. parile, N. resupinatum and P. triptophylla, irrespective of their geographic origin, are confined to subclade IIb together with three cyanobionts of L. pulmonaria and with one, possibly free-living Nostoc attached to the lower part of L. pulmonaria thallus (specimen Myllys & Kuusinen 376b). Within subclade IIb N. parile samples, all of them collected from Evo, form a separate group together with two cyanobionts of L. pulmonaria.
The clustal alignment of the combined data set produced 1367 aligned sites of which 297 were parsimony informative. The paup analysis generated 1840 equally parsimonious trees of 980 steps, a consistency index (CI) of 0.663 and a retention index (RI) of 0.826. The strict consensus tree (not shown) was in total agreement with the consensus tree obtained from the poy analysis: for example, both major clades I and II and the subclades IIa and IIb were present. As in the poy analysis, the cyanobionts of P. praetextata and P. leucophlebia specimens group together in Clade I and the cyanobionts of Nephroma parile form a separate group within subclade IIb.
The poy analysis of the 16S data set with 78 taxa resulted in five equally parsimonious optimizations (and hence trees) of 280 steps. In the strict consensus (Fig. 2) clade I is nonmonophyletic, although the clade including P. leucophlebia and P. praetextata specimens is recovered. In a poorly supported clade II only subclade IIa is recovered. However, it now includes also a cyanobiont obtained from L. pulmonaria Myllys & Kuusinen 376a, the specimen used only in the separate 16S analysis.
The rbcLX data set with 77 taxa resulted in three trees of 280 steps, the strict consensus of which is given in Fig. 3. In a monophyletic clade I P. leucophlebia and P. pratextata group together, a result also obtained from the 16S and from the combined analyses. The clade II is almost recovered; only the basal taxon appearing in the combined tree (Nostoc sp. 1) is left outside. This clade is poorly supported and both subclades IIa and IIb are absent.
Our results, based on the phylogenetic analysis of 16S rDNA and rbcLX sequences, are in accord with the earlier studies (e.g. Rikkinen et al., 2002; Lohtander et al., 2003) and show that the cyanobionts of the examined lichen genera Lobaria, Nephroma, Parmeliella and Peltigera belong to the genus Nostoc. In our analysis, the subgroups obtained within the two main clades do not show any correlation with the geographic origin of the samples at any spatial scale. For example, we did not find any Nostoc strains specialized into the Evo forest stand or to separate aspens within the area. Instead, there was a clear correlation between the cyanobionts and the alleged taxonomy of their mycobionts, since in all but one case cyanobionts of the epiphytic species examined each was confined to one subclade only. The result clearly demonstrates that the mycobionts of these lichen symbioses are highly discriminative in their choice of cyanobiont. Similar results (i.e. strong selectivity of the cyanolichens towards their photobionts) have been obtained in several studies (Paulsrud et al., 1998, 2000; Rikkinen et al., 2002; Summerfield et al., 2002; Stenroos et al., 2006). However, our results contrasts with those of O’Brien et al. (2005) who reported, based on separate analyses of 16S, rbcLX and trnL sequences, that cyanobacteria from the same lichen species were often more closely related to strains from other species or to nonsymbiotic strains than to each other.
The only exception to the observed pattern is L. pulmonaria, whose cyanobionts were distributed all over the clade II. This suggests that L. pulmonaria is more flexible than the other species, being able to associate with a wide range of cyanobacteria. Based on our observations and on a report by Zoller et al. (1999), the thalli of L. pulmonaria do not always contain cephalodia. The occasional absence of these structures could be an indication of a loose relationship between the two symbionts, allowing the mycobiont to associate with several different Nostoc strains. The low selectivity may also lead to the presence of different Nostoc strains in a single lichen thallus. We examined this hypothesis in one case (specimen Haikonen 23259) and found that the cyanobionts of different cephalodia were genetically identical. By contrast, two genetically distinct Nostoc strains were observed in the L. pulmonaria specimen Myllys & Kuusinen 376, as shown by the separate analysis of 16S sequences (Fig. 2). However, while one of the extractions was obtained from an internal cephalodium, the other originated outside of this structure and may either represent a free-living strain or come from a symbiotic propagule of some cyanolichen species.
Despite the high species-level selectivity of the mycobionts towards their cyanobionts, there is no evidence of cospeciation of symbionts for the entire data set, since the cyanobionts of different mycobiont species, genera or even of different mycobiont families were found to belong to the same subclade. This indicates that cyanobionts have a low degree of selectivity towards their mycobiont partners and that the cyanobiont–mycobiont associations as a whole are not highly specific in the species examined (for further information on the definitions of the terms selectivity and specificity see Beck et al., 2002). Our results are in agreement with earlier reports from Beck (1999), Helms et al. (2001), Rikkinen et al. (2002), Piercey-Normore & DePriest (2001) and O’Brien et al. (2005) that the same photobiont taxon may form associations with taxonomically unrelated lichens.
The two main clades in our study (I and II, Fig. 1) correspond to those already found and discussed in several previous studies (Rikkinen et al., 2002; Lohtander et al., 2003; Rikkinen, 2003, 2004; Oksanen, 2004). Rikkinen et al. (2002) suggested that the division correlates with the ecology of the lichen species, and introduced the terms ‘Peltigera guild’ and ‘Nephroma guild’ for the main groups. ‘Peltigera guild’, which corresponds to the Clade I in our study, includes predominately terricolous cyanolichens but also symbiotic bryophytes and cycads, while the ‘Nephroma guild’ (Clade II in our study) comprises epiphytic lichen species confined to old-growth forests. However, as shown also by Stenroos et al. (2006), the Nephroma clade does not correlate with ‘epiphytic’ conditions as suggested by Rikkinen et al. (2002), since it includes also cyanobionts collected from terrestrial and litophytic habitats.
In our study, a moderately specific association (i.e. one cyanobiont strain being restricted to the mycobionts of the same genus) was found only in clade I. Here, the bipartite P. praetextata and tripartite P. leucophlebia were restricted to one subclade only and they probably share a common cyanobiont. The two species are not closely related, as shown by the combined analysis of morphological, chemical and LSU rDNA data (Miadlikowska & Lutzoni, 2000, 2004), but share similar habitat preferences being confined, when epiphytic, to the very base of the tree trunks. The Nostoc strain shared by P. leucophlebia and P. praetextata, however, is by no means restricted to tree bases, since three of the four P. leucophlebia specimens collected from litophytic habitats appear in the same clade. Interestingly, the cyanobiont of one P. leucophlebia specimen (Myllys & Kuusinen 463) growing on a rock face in north-eastern Finland was left out this clade. Further studies with more extensive sampling of P. leucophlebia specimens are needed to examine whether this cephalodiate species could in fact be as flexible in its cyanobiont choice as L. pulmonaria.
Six remaining epiphytic species, confined to clade II in our study, tend to grow higher up on the basal trunk typically between heights of c. 0.5 and 2 m in boreal forests of Fennoscandia (Kuusinen, 1996). Rikkinen et al. (2002) suggested the cyanolichens of the ‘Nephroma guild’ including N. bellum, N. parile, N. resupinatum, P. triptophylla and possibly also L. pulmonaria exploit a common pool of cyanobacteria. Sexually reproducing species (i.e. N. bellum, N. laevigatum and N. resupinatum) that disperse via ascospores would especially benefit from a common cyanobacterial genotype (Rikkinen, 2003; Oksanen, 2004). According to our study, however, only three of these species, N. bellum, N. resupinatum and P. triptophylla share the same cyanobiont pool (strain or group of strains), whereas N. laevigatum and perhaps also N. parile both associate with different, although closely related Nostoc strains. Lobaria pulmonaria, as discussed above, is a generalist being able to associate with more than one Nostoc strain found in clade II. As both subclades also include asexually reproducing species, our result supports the theory of Rikkinen et al. (2002) that sexual ‘fringe species’ depend on the asexual ‘core species’ for the dispersal of cyanobionts. While N. bellum and N. resupinatum would benefit from the diaspores of L. pulmonaria and P. triptophylla, N. laevigatum could in turn obtain its cyanobiont from the soredia of L. pulmonaria.
Although Rikkinen et al. (2002) found genetic variation in the ‘Nephroma’ clade, the subgroups obtained within the clade could not be linked to any specific patterns of cyanobiont selectivity. This contrasts with our study, where cyanobionts of the species examined were each confined to one subclade only. These slightly contradictory results may arise not only from different taxon sampling but also from the different gene regions used. In Rikkinen et al. (2002), the phylogenetic analysis of near-complete 16S rDNA sequences revealed no correlation between fungal taxa and their cyanobionts, most probably because of lack of resolution. In our study partial 16S rDNA as well as rcbLX sequences alone offered poor resolution (Figs 2, 3), but when analysed in combination, they grouped cyanobionts according to the species identities of their fungal hosts. It has also earlier been reported that 16S rDNA may be too conservative to distinguish between different Nostoc strains (Fox et al., 1992; Summerfield et al., 2002). Our results clearly demonstrate that phylogenetic hypotheses based on simultaneous analysis of multiple data sets have highest explanatory value (Farris, 1983; Nixon & Carpenter, 1996).
Combination of the data had clearly a positive effect on the support values. Some of the nodes, however, are still poorly supported. This kind of poor support seems to be common in phylogenies of cyanobacteria associated with lichenized fungi. Low support values have been obtained regardless of data sampling or the phylogenetic method used (Rikkinen et al., 2002; Lohtander et al., 2003; Wirtz et al., 2003; O’Brien et al., 2005; Stenroos et al., 2006). By contrast, in phylogenies of green algae associated with lichens both deep and low-level relationships are often highly supported (see for example Lohtander et al., 2003). Clearly, more variable gene loci are needed to examine the evolutionary relationships of cyanobacteria and their symbiotic interaction with lichenized ascomycetes.
Finally, it should be noted that free-living or other symbiotic Nostocs are entirely absent from clade II, except for one unidentified Nostoc associated with Cycas at the base of the clade and for one Nostoc attached to Lobaria pulmonaria (specimen Myllys & Kuusinen 376b). Consequently, the different subclades cannot be assigned to any known strain of Nostoc with certainty. We suggest that clade II consists of several, or at least two different Nostoc strains (subclades IIa and IIb) for the following reasons: first, cyanobionts in clade II are not genetically uniform, as shown by the highly resolved phylogeny and second, cyanobionts are regularly distributed in specific subclades according to their fungal hosts.
To conclude, our results indicate that the co-occurrence of the epiphytic examined cyanolichens cannot always be explained by a common specific cyanobiont. Even thalli of different species occurring in close contact on the same aspen may have their own, taxon-specific cyanobionts. All the eight epiphytic species examined are rare in the remaining small patches of suitable habitats in the managed landscape (Kuusinen & Penttinen, 1999). For N. laevigatum and P. leucophlebia, in particular, old aspens may provide practically the only suitable epiphytic habitat in the boreal forests of Finland (Kuusinen, 1994). Our results imply, however, that the distribution of these species is not limited by the distribution of their cyanobionts. For example, N. laevigatum, which has been classified as an endangered species in Finland, shares the same Nostoc strain as L. pulmonaria (specimen Haikonen 23259) collected from an aspen, where none of the other lichens examined occurred. Furthermore, the same Nostoc was found even from other than epiphytic habitats, for example, from Lobaria scrobiculata collected from a rock face and from Pseudocyphellaria crocata growing on dirt in southern Patagonia.
We thank Veera Sivonen and Eija Virtanen for technical assistance and Orvo Vitikainen for help in identification of Peltigera specimens. The study was financially supported by the Academy of Finland Grant 204201.