MBE Advance Access originally published online on November 23, 2005
Molecular Biology and Evolution 2006 23(3):574-586; doi:10.1093/molbev/msj063
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Article |
Testing "Species Pair" Hypotheses: Evolutionary Processes in the Lichen-Forming Species Complex Porpidia flavocoerulescens and Porpidia melinodes
,1
* Committee on Evolutionary Biology, University of Chicago and Botany Department, Field Museum of Natural History, Chicago
E-mail: j.buschbom{at}holz.uni-hamburg.de.
 |
Abstract |
|---|
 TOP Abstract IntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Pairs of taxa are commonly found in lichen-forming ascomycetes that differ primarily in their reproductive modes: one taxon reproduces sexually, the other vegetatively. The evolutionary processes underlying such "species pairs" are unknown. The species pair formed by Porpidia flavocoerulescens (sexual) and Porpidia melinodes (vegetative) was chosen to investigate four previously proposed hypotheses. These hypotheses posit that species pairs are either two monophyletic, independently evolving species with contrasting reproductive mode; a single outcrossing species polymorphic with regard to its reproductive modes; a sexual mother lineage frequently giving rise to asexual spin-offs; or a complex of cryptic species. The phylogenetic patterns observed within the species pair in the present study were analyzed using stringent hypothesis testing and visualizations of relationships and conflict based on tree and network reconstructions. DNA sequences at the three analyzed loci revealed the same four to five deeply divergent lineages. A detailed analysis of DNA-sequence variability revealed closely linked gene loci, but high levels of conflict within each of the gene fragments, as well as between observed genetic lineages. The observed patterns of phylogenetic relationships, linkage, and conflict are not congruent with any of the previously proposed species pair hypotheses. Rather, it is proposed that the observed results can be explained by conflicting reproductive and nutritional requirements imposed by an obligate symbiotic lifestyle. These interacting constraints produce recurring selective sweeps within predominantly vegetatively reproducing lineages and are the main forces that shape the evolution within the investigated species pair.
Key Words: Ascomycota • hypothesis testing • molecular phylogenetics • networks • population-level processes • Porpidia
|
Introduction |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Closely related (putative sister) taxa are found that show contrasting reproductive modes in many groups of lichen-forming ascomycetes (obligate fungal symbionts of green algae and cyanobacteria). The members of these "species pairs" (Poelt 1970
) differ primarily in their reproductive modes: one taxon reproduces sexually, the other vegetatively. Despite the fact that species pairs have been known and investigated for over 80 years, the evolutionary processes underlying this dichotomy are unknown, and the value of the reproductive system as a character for species delimitations has been questioned.
In the first paper discussing the possible processes shaping patterns found within species pairs, Du Rietz (1924) already mentioned most of the scenarios and concepts that have been subsequently considered. The most prominent of these is the "Artenpaar" (species pair) concept, which has received wide support (Du Rietz 1924; Poelt 1963, 1970, 1972; Hale 1965; Culberson 1973; Bowler and Rundel 1975). It assumes a sexual primary species from which a monophyletic vegetative lineage (secondary species) arises through a rare transition event. The vegetative lineage is thought to be successful due to its superior ability to colonize and survive in marginal habitats. All molecular phylogenetic studies of species pairs performed to date, however, suggest that neither sexual nor vegetative lineages are monophyletic (Lohtander, Källersjö, and Tehler 1998; Lohtander et al. 1998; Myllys et al. 1999; Kroken and Taylor 2001a; Myllys, Lohtander, and Tehler 2001). Rather, phylogenetic trees show sexual and vegetative lineages to be intermingled.
Such a pattern would arise in a single polymorphic (with regard to reproductive mode) and recombining species (Robinson 1975). However, contrary to what would be expected if the species pairs were simply a single outcrossing population, phylogenetic trees within the species pairs are well structured and show well-supported groups of specimens.
Tehler (1982) proposed a process underlying the species pairs in which a sexual primary species repeatedly and frequently produces short-lived vegetative spin-offs. Tehler's hypothesis requires that there is a sexual primary species and, thus, that the ancestor to a species pair is sexually reproducing.
Kroken and Taylor (2000, 2001a, 2001b; Högberg et al. 2002) employed a total evidence phylogenetic species concept to address the species pair problem. In this approach, species are based on the findings of combined analyses of multilocus data sets to avoid spurious results due to incomplete lineage sorting in individual loci. Investigating a species pair within the genus Letharia, they proposed that every well-supported clade within the tree represents a "cryptic" species expanding the species pair into six cryptic species. The proposed species are cryptic because morphological characters associated with the lineages are only "subtle" (Kroken and Taylor 2001a) and outweighed by environmental modifications. Reproductive modes correlate with the lineages insofar that four lineages are sexually reproducing while the other two are vegetative. In contrast to Tehler's hypothesis, they suggest that these vegetative lineages are viable in the long term due to the presence of recombination. At least two transitions are necessary to explain the distribution of the reproductive modes in the unrooted tree. The authors did not provide an explanation for which evolutionary processes might give rise to the frequent speciation events producing the proposed six cryptic species.
All molecular phylogenetic investigations of species pairs conducted to date show the same trend in relationship patterns, that is, well-structured gene trees with highly supported lineages. Unfortunately, published studies have not explicitly tested proposed hypotheses about the processes underlying the species pairs. The studies, therefore, could not reject any of the proposed hypotheses to clear the field of possible explanations.
We chose the "orange species pair" of Porpidia flavocoerulescens (Hornem.) Hertel & A. J. Schwab (sexual) and Porpidia melinodes (Körb.) Gowan & Ahti (vegetative) for detailed population-level investigations into the species pair phenomenon to explicitly test the proposed hypotheses. Both taxa are widespread and common throughout the boreal-arctic zones where they form crustose thalli on siliceous rock surfaces that are snow covered during at least part of the winter and, depending on latitude, more or less shaded and humid. The distribution range of the pair is circumpolar in the northern hemisphere (Hertel 1977; Inoue 1983; Hertel and Knoph 1984; Schwab 1986; Gowan 1989; Gowan and Ahti 1993). Fieldwork revealed that the geographic ranges of both taxa are mostly sympatric (Buschbom 2003). The orange species pair has been shown to form a highly supported monophyletic group with 100% bootstrap support and posterior probabilities in our genus-level phylogenetic study (Buschbom and Mueller 2004).
The sexual reproductive mode is comprised of apothecia (the sexual fruiting bodies of the fungus), within which ascospores are produced. In the orange species pair P. flavocoerulescens reproduces exclusively sexually. Vegetative (asexual) reproduction is by soredia, microscopically small packets of algal cells wrapped together by a sheath of fungal hyphae. These propagules are formed in specialized organs called soralia that represent openings in the upper thallus cortex. When apothecia are observed in P. melinodes, they are only found co-occurring with soralia. Both, sexual fruiting bodies (apothecia) and soralia are perennial structures that persist and are functional on the thalli for years. The unicellular green algal partner of all taxa of Porpidia investigated so far belong to the genus Trebouxia. Only the mycobiont was analyzed in this study because no sexual reproductive structures are formed by the photobiont when part of the lichen symbiosis in Porpidia.
As in all species pairs, the orange species pair represents a discontinuum between exclusively sexual thalli and exclusively asexual thalli. However, as in most species pairs this discontinuum is not absolute. Specimens are occasionally found that have both apothecia and soralia. These specimens traditionally have been defined as belonging to the vegetative taxon. In the orange species pair the frequency of such mixed specimens depends on geographic region and ranges from extremely rare (e.g., in Quebec, Baffin Island, and on Greenland) to common (e.g., in Northern Scandinavia).
The main goal of the present study was to test the hypotheses that have been proposed in the literature to understand the potential processes shaping the relationship patterns observed in the P. flavocoerulescens and P. melinodes species pair. Monophyly of sexual and vegetatively reproducing specimens was tested as well as an alternative hypothesis of a random distribution of the contrasting reproduction modes throughout the species pair. Conflict between the three sequenced loci was tested. Finally, homoplasy within and between the loci was graphically investigated using median networks to shed light onto the hypothesis of cryptic species. The results are discussed with regard to existing species pair concepts, and a new hypothesis about the evolutionary processes underlying the phenomenon is proposed.
|
Methods |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Taxon and Character Sampling
Thalli of P. flavocoerulescens and P. melinodes were collected in four geographic regions (fig. 1 and Supplement 1, Supplementary Material online): at sites around Schefferville (Québec, Canada), Iqaluit on Baffin Island (Nunavut, Canada), Qeqertarsuaq on Disko Island (Greenland), and in Northern Scandinavia, including localities around Abisko (Norrbotten, Sweden), Kevo (Lappi, Finland), and along the coast in Finnmark (Norway). A total of 110 specimens were included in the study, with 13 (Greenland) to 37 (Baffin Island) collections investigated per region (Supplement 2, Supplementary Material online). Overall, 48 samples of P. flavocoerulescens and 62 samples of P. melinodes were analyzed. Porpidia tuberculosa and Porpidia carlottiana were chosen as outgroups to root the intraspecific network based on the results of Buschbom and Mueller (2004)
. Obtained sequences were submitted to GenBank under the accession numbers DQ314889–DQ314994 (LSU), DQ315069–DQ315155 (ß-tubulin), and DQ314995–DQ315068 (RPB2). Voucher specimens are deposited at the Field Museum of Natural History in Chicago (F).
|
Phylogenetic analyses were based on DNA sequence data from three loci: a 600-bp fragment at the 5' end of the nuclear large subunit ribosomal RNA gene (LSU), a 1-kb fragment at the 3' end of the nuclear protein-coding gene for ß-tubulin, and a 1.2-kb fragment of the nuclear protein-coding gene coding for the second largest subunit of DNA-dependent RNA polymerase II (RPB2), spanning regions 5–7 as described by Liu, Whelan, and Hall (1999). Small fragments of the haploid thalli were sufficient for DNA extraction. Instruments, for example, razor blades and forceps, were flame sterilized between each sample to avoid cross-contamination.
Locus Selection
Previous molecular studies of lichen-forming ascomycete populations exclusively or primarily were based on multicopy RNA genes (mostly ITS1 and ITS2) or randomly amplified DNA fragments for which no information on the size of their gene families exists (Beard and DePriest 1996; Crespo et al. 1997; Zoller, Lutzoni, and Scheidegger 1999; Murtagh, Dyer, and Crittenden 2000; Kroken and Taylor 2001a; Printzen and Ekman 2002). Two of the three loci employed in the present population-level study of P. flavocoerulescens and P. melinodes are known to be one- or two-copy genes. Only a single copy of RPB2 is known to exist within the fungal kingdom (Liu, Whelen, and Hall 1999; Oxelman and Bremer 2000). A gene duplication of ß-tubulin has occurred within the Ascomycota (Landvik, Eriksson, and Berbee 2001). However, both copies have diverged since the duplication event and are easily distinguishable in the analyzed group of lichen-forming ascomycetes (Buschbom and Mueller 2004). The LSU is a member of the nuclear RNA gene tandem repeat and exists in large copy numbers (100–200 copies) within fungal genomes. A preliminary study of sequences of six clones originating from a mixture of 10 polymerase chain reaction (PCR) products of a single extraction revealed no nucleotide differences among cloned sequences. No data are available on the localization of the three loci in the genome of filamentous ascomycetes to date.
Laboratory Procedures
DNA extraction, amplification and sequencing procedures, and conditions were conducted as described in Buschbom and Mueller (2004). PCR and sequencing primers for the LSU and ß-tubulin were the same as in the phylogenetic study of the genus, with the exception that only primers LROR and LR3 were used for sequencing the LSU. The RPB2 gene fragment was amplified using primers fRPB2-5F and fRPB2-7R (Liu, Whelen, and Hall 1999). In addition to the PCR primers, both strands of the region were sequenced with RPB2-6F and RPB2-6R (Liu, Whelen, and Hall 1999) and newly designed primers R2B-1719R (AGCKGGGTCYCGATGMACACC, cited in 5' to 3' direction) and R2B-2176R (AWGRWTYTCVCARTGRGYCMATG). The following PCR cycle for RPB2 was employed: 95°C for 5 min, 35x (95°C for 1 min, 48°C for 2 min, 0.2°C increase per second to 72°C, 72°C for 2 min), 72°C for 10 min. Contigs and initial alignments were assembled using Sequencher 3.7-4.1 (Gene Codes Corp., Ann Arbor, Mich.). Sequence alignments were checked and optimized by eye in MacClade 4.0 (W. P. Maddison and D. R. Maddison 2000). All substitutions that only appeared in a single sequence were verified based on the original chromatograms.
Phylogenetic Analyses: Tree Reconstructions
Phylogenetic analyses included maximum parsimony (MP), maximum likelihood (ML), and Bayesian (B(MC)3) approaches. All tree optimizations were first conducted separately using all available sequences (i.e., specimens) for each locus. Subsequently, the sampling was reduced to the set of specimens for which sequences of all three genes were obtained. Two combined data sets were analyzed: one included the LSU and RPB2 loci, the other one included sequences from all three loci. Combined analyses were conducted as for the single gene data sets.
Phylogenetic relationships between the specimens were reconstructed using MP and ML as implemented in PAUP* 4.0b10 (Swofford 1998). Under MP, transition and transversion costs between nucleotides were estimated from the alignment and incorporated into the optimization through stepmatrices (W. P. Maddison and D. R. Maddison 1992). Heuristic searches were conducted with Tree Bisection-Reconnection (TBR) branch swapping, the "MULTREES" option in effect, and 1,000 random addition sequence replicates. Support for relationships was estimated using nonparametric bootstrapping with 1,000 pseudoreplicates in MP analyses. Trees were optimized for each replicate using the original settings, but with only 10 random addition sequence replicates per pseudoreplicate.
Phylogenetic reconstructions under ML were conducted using heuristic searches with TBR branch swapping, the MULTREES option in effect, and 100 random addition sequence replicates. Nucleotide-substitution models and parameter values were selected using the Akaike information criterion as implemented in Modeltest 3.06 (Posada and Crandall 1998). Branch support was (1) estimated using nonparametric bootstrapping (100 pseudoreplicates with each two random sequence additions) and (2) through estimation of posterior probabilities in a Bayesian framework. Markov chain Monte Carlo runs using MrBayes 3.0b3 (Huelsenbeck and Ronquist 2001) were conducted assuming a general time reversible (GTR) model with rate heterogeneity among sites and a proportion of sites being invariant. Default priors were used for all other model parameters. Chains were run for two million generations with every 100th generation sampled. After inspection, the first one million generations were discarded as burn-in. The results are thus based on the last 10,000 trees.
Congruence Between Gene Trees
Gene trees at the intraspecific level can provide conflicting information due to population processes. Incongruence between loci can be due to incomplete lineage sorting, recombination, or admixture (gene flow). Congruence between gene trees of unlinked loci thus can be used as an indicator for the extent of genetic isolation within and between lineages. The reduced specimen sets were used for the congruence tests, which included only those specimens for which all three loci could be sequenced. Congruence between trees was tested under MP using the ILD test (incongruence length difference test, Farris et al. 1995), implemented in PAUP* as a partition homogeneity test. One hundred replicates, each with 10 random sequence additions, were run under the same settings as described above.
A Bayesian approach was implemented for testing congruence among ML trees. Confidence envelopes obtained through B(MC)3 analyses around ML gene trees were used to test how often gene trees obtained from other loci or data partitions are compatible with trees in the confidence envelope. For this, the B(MC)3 plateaus of each of the loci were separately filtered for topologies congruent with constraints correlating to the topologies resulting from the other data sets using PAUP*. Two sets of constraints were tested: (1) the constraints incorporate all supported internodes (
95% posterior probability) of the observed gene trees from analyses of the single and combined matrices and (2) the constraints only reflect supported relationships in the backbone of the trees, that is, relationships between lineages and not within lineages. It can be assumed that two gene trees represent different samples, that is, evolutionary histories, if less than 5% of the trees in the confidence envelope around one of the trees are in congruence with the gene tree (i.e., constraint) of the other data partition.
Phylogenetic Analyses: Network Reconstruction
Relationships among individuals on the intraspecific level are expected to be tree-like only under certain conditions (e.g., asexuality). It can generally be expected that relationships within recombining populations form networks. However, phylogenetic reconstruction methods based on Wagner parsimony (Kluge and Farris 1969; Farris 1970; Fitch 1971) or ML algorithms (Cavalli-Sforza and Edwards 1967; Felsenstein 1981) will force such relationships into the potentially misleading shape of trees. Thus, the extent of incompatibility within the alignments and the appropriateness of tree-reconstruction methods were explored graphically using median networks (Bandelt 1994; Bandelt et al. 1995).
Median networks were constructed using SPECTRONET (Huber et al. 2002). In this approach all incompatibilities present in the alignment are taken into account and depicted in a graph. Median networks were based on actual nucleotide site patterns, that is, "direct encoding." A correction for parallel evolution and multiple hits using the Hadamard conjugation (i.e., spectral analysis, Hendy and Penny 1993) did not change the results. Sites with more than two types of nucleotides were recoded as purines and pyrimidines, thus only taking the transversion between those two nucleotide classes into account. Weights for purine/pyrimidine splits were left at the default of 1.0.
Because SPECTRONET currently cannot handle the number of sequences present in the data sets, each alignment was reduced so that each haplotype present was represented only by a single sequence. For this procedure, Collapse 1.1 (D. Posada, http://dario.uvigo.es/software/collapse.html) was implemented. Those sequences that differed only in the presence of "N" from another sequence were subsequently removed by hand.
Phylogenetic Analyses: Hypothesis Testing
To stringently test the four previously proposed species pair hypotheses, testable aspects had to be identified for each of the hypotheses. During this process it became obvious that Tehler's hypothesis of vegetative spin-offs from a sexual mother lineage had to be tested at the genus and family level (J. Buschbom and D. Barker, unpublished data, see Discussion). For the hypothesis of cryptic species no such testable aspect could be found, however, as explained in the Discussion, we do not accept this hypothesis. The remaining two hypotheses were tested using phylogenetic approaches. These hypotheses and the aspects that facilitate testing are:
- I) The label "species" pair is appropriate: sexual and vegetative lineages form two monophyletic sister entities.
- II) The species pair represents a single species with a polymorphic reproductive system. Here, vegetative and sexual lineages are not monophyletic, and character state changes are randomly distributed throughout the tree, that is, not correlated with phylogenetic relatedness.
- II) The species pair represents a single species with a polymorphic reproductive system. Here, vegetative and sexual lineages are not monophyletic, and character state changes are randomly distributed throughout the tree, that is, not correlated with phylogenetic relatedness.
). Constraints mirroring the proposed hypotheses of relationships were constructed in MacClade. Using PAUP*, the B(MC)3 plateaus of the single gene runs (10,000 trees sampled) were filtered and the frequency of the trees congruent with a specific constraint recorded. Specimens were coded as sexual or vegetative. The occasional specimens showing both reproductive modes were coded as vegetative in the monophyly tests as well as the randomization (J. Buschbom and D. Barker, unpublished data). If the species pair represents a single outcrossing taxon, reproductive modes should be distributed randomly and not be correlated with phylogeny. Reproductive modes were randomized across taxa using MacClade to test if thalli showing contrasting reproductive modes are distributed randomly throughout a gene tree, that is, their reproductive mode is not correlated with phylogeny (hypothesis II). The gene trees resulting from the single locus ML analyses were used as topologies. The reproductive character states were randomly reassigned to the leaves of the tree 1,000 times. The states were reconstructed applying a MP approach with equal transition costs between states. The number of steps of the reproductive character reconstructed on the trees was recorded. The randomization provided the null distribution against which the number of steps reconstructed in the original data sets was tested.
|
Results |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Gene Sequences, Alignments, and Haplotypes
The LSU fragment could be amplified from basically every specimen extracted. However, sequences for the ß-tubulin and RPB2 loci could only be obtained from a subset of those specimens, probably reflecting a decreased fit of primers. No amplification products could be obtained for RPB2 from taxa of the sister group to the orange species pair (including P. tuberculosa). However, a change from P. tuberculosa to P. carlottiana (member of the next closest subgroup in Porpidia s.l.) did not drastically change the number of variable sites in either the LSU or ß-tubulin. With P. tuberculosa as outgroup, the first intron in ß-tubulin could be aligned and included, while both introns had to be excluded with P. carlottiana as outgroup. MP analyses of the LSU and ß-tubulin were performed as described above with both outgroups or with only P. tuberculosa or P. carlottiana as outgroups and showed no differences in resolution, observed relationships, or support within the species pair (but see rooting below). Thus, analyses were subsequently performed with P. carlottiana as outgroup to enable parallel analyses with all three loci.
The LSU alignment consists of 110 ingroup sequences of a 600-bp region at the 5' end of the gene. Final sequences included in the analyses are 581 bp long and show no length differences throughout the alignment. With P. carlottiana as outgroup the LSU matrix included 30 variable sites of which 20 were found to be parsimony informative (Supplement 3, Supplementary Material online).
The ß-tubulin matrix consists of 96 ingroup sequences extending over a length of 945 bp. Three gaps were introduced by the P. carlottiana sequence within the introns. Interestingly, one P. melinodes specimen (number 259) showed a single nucleotide insertion within the second intron, extending a run of (CT)TTT to TTTTT that was not present in any other ingroup or outgroup. Introns in ß-tubulin were excluded due to ambiguities introduced by length and nucleotide differences between in- and outgroup. After exclusion of the introns, 843 exon sites were included, of those, 122 sites were variable and 47 sites were parsimony informative with P. carlottiana as outgroup (Supplement 3, Supplementary Material online).
RPB2 sequences were obtained for a total of 73 specimens. The alignment is 1,171 sites long and includes a single intron of 52 bp at the 3' end. No length differences were observed among sequences. In addition, the intron could be aligned and was included in the analyses. With P. carlottiana the matrix contained 132 variable sites of which 44 were parsimony informative (Supplement 3, Supplementary Material online).
Considering only the ingroup, ß-tubulin has nearly twice as many variable and parsimony informative sites as RPB2 and LSU. This ratio is also found if only the exons in ß-tubulin and RPB2 are considered. The increase in variability is mainly due to the third positions in ß-tubulin, which are much more polymorphic than in RPB2. The higher amount of variability in ß-tubulin is also reflected in the number of sites affected by multiple hits based on the number of nucleotide types: while ß-tubulin has four third positions showing three nucleotide types, neither of the other two loci have any sites with more than two nucleotide types within the ingroup.
In contrast to the varying amounts of variability present in the data sets, the number of haplotypes found in the three loci (excluding the outgroup) is similar. The LSU shows 15, RPB2 13 and ß-tubulin 16 haplotypes among the ingroup specimens analyzed.
Reduced data sets were assembled including only those 69 specimens for which all three loci could be amplified and sequenced. All major phylogenetic groups are still represented by at least one specimen in those data sets. Combining the LSU and RPB2 matrices resulted in an alignment of 1,752 bp length, including 160 variable sites and 63 parsimony informative sites (Supplement 3, Supplementary Material online). If all three loci are combined, the matrix is 2,699 characters long and contains 282 variable sites and 108 parsimony informative sites.
Phylogenetic Relationships: Trees
Phylogenetic trees for each locus resolve the same relationships within the orange species pair independent of the employed reconstruction method (MP, ML, and B(MC)3) and of taxon set analyzed (all available sequences per gene vs. the specimen set common for all three genes). One to three most parsimonious and most likely trees were found in the analyses. Differences between best trees are due to the placement of specimens within major lineages. Five generally well-supported lineages were found in the trees (lineages I–V). The LSU and RPB2 track exactly the same relationships among the analyzed specimens (fig. 2 and Supplement 4, Supplementary Material online). In the unrooted quartet, lineages I and II were most closely related while lineage III and IV + V were grouped together on the other end of the central internode. MP and ML nonparametric bootstrap support and posterior probabilities for all lineages are 100% in the RPB2 tree and well above 75% bootstrap support and the 95% significance level for lineages I and III in the LSU. Lineages II and IV + V show mixed support in the LSU: they are supported above 70% by MP bootstrap values, while ML bootstrap and posterior probabilities are (slightly) below the cutoff point of 70% or significance level of 95%, respectively.
|
Three of the five lineages are also present in the ß-tubulin trees (lineages I–III; fig. 2 and Supplement 4, Supplementary Material online). However, here group IV + V splits into two parts: 12 specimens represent a well-supported divergent lineage (lineage V) that forms the sister group to both lineages III and the remainder of the former group IV + V (lineage IV). Lineage IV in ß-tubulin is most closely related to lineage III, while in the LSU and RPB2 analyses it merges with lineage V. Lineages I, II, and V are well supported in ß-tubulin. In contrast to the other two loci, posterior probabilities are high in the backbone of the ß-tubulin tree. Both lineages I and II, as well as lineages III and IV are significantly grouped together. However, support for lineages III and IV was mixed.
With the exception of posterior probabilities in ß-tubulin, no support was found for the backbone of trees generated in the analyses. Overall, different branch support measures (MP and ML bootstrap, posterior probabilities) show the same trends and do not conflict.
The rooting of the species pair in a MP framework is ambiguous and depends on the outgroup (P. carlottiana or P. tuberculosa) and the gene locus. The LSU and ß-tubulin loci show the root to be attached to the central internode, grouping lineages I and II, as well as lineages III–V. In RPB2, however, the root attaches to the branch leading to lineage III. The picture becomes clearer when model-based reconstructions are considered. In most ML and B(MC)3 analyses and loci, P. carlottiana attaches at the central internode, separating lineages I and II from lineages III–V. The only exception is the ML tree for ß-tubulin, in which P. carlottiana roots the tree within lineage III, clearly an artifact. However, if the more closely related P. tuberculosa is used as outgroup, the root is located at the central internode. The location of the root at the central internode is consistent with the placement of the root to the orange species pair in the genus- and family-level phylogeny (Buschbom and Mueller 2004).
Congruence Between Gene Trees
A simultaneous partition homogeneity test of all three gene data sets predicted significant incongruence between the loci (table 1). Pairwise tests of the three data sets showed that the observed conflict is due to ß-tubulin. Tests involving ß-tubulin reject congruence between the ß-tubulin topology and RPB2 or both LSU and RPB2. However, congruence between LSU and RPB2 topologies cannot be rejected (they are basically identical). The ILD test of LSU and ß-tubulin data sets returned a nonsignificant result.
|
Tests of congruence between ML trees using a Bayesian approach (table 2) showed that this type of analysis is very sensitive and returns significant results in cases in which taxa in the tips of the tree show conflicting relationships, even if the rest of the tree topology is identical. Accordingly, congruence between gene trees was rejected in most tests if all supported branches were incorporated into the test constraint. The nonsignificant results obtained between RPB2 and most other gene partitions are due to the lack of any support for relationships between and within the four lineages found by that gene. In contrast, the highly supported grouping of P. flavocoerulescens 113 and 234 in the LSU results in significant test statistics, despite the general lack of support for relationships in the backbone of that gene tree. The outcomes of the Bayesian approach to congruence testing were identical to the results of the ILD test if the tests were focused on the backbone only, that is, if compatibility was only tested with regard to the relationships between the five lineages. Thus, the Bayesian approach can be used to identify in which part of a tree conflicting topological relationships among different gene partitions occur.
|
In summary, conflicting relationships were found both among lineages and within lineages in the present study. While no conflict was found among the backbones of LSU, RPB2 and combined LSU and RPB2 data sets, some conflict among those gene partitions was detected within lineages. However, comparisons involving ß-tubulin (either separately or in the combined LSU, RPB2 and ß-tubulin data set) suggested that ß-tubulin is tracking different evolutionary relationships compared to the other two loci.
Phylogenetic Relationships: Networks
When phylogenetic relationships are not forced into a tree structure, multidimensional median networks of increasing complexity for each of the three loci from LSU, RPB2 to ß-tubulin are reconstructed (fig. 3). Parallel edges represent the same split in the networks. The simplest network was obtained for the LSU, showing a single central cube from which lineages I to V originate. Reticulations interconnect lineages I and II and connect specimen P. melinodes 251 in lineage III to the central cube.
|
In RPB2 the central network consists of several interconnected cubes from which the four lineages originate. The network of a combined LSU and RPB2 matrix resembles the one for RPB2. In all three partitions, the edges that lead to the four lineages are comparatively long, that is, each of the splits is supported by several sites. Interestingly, conflict in these data sets was found only in the backbone that connects the lineages identified during tree reconstructions. With the exception of P. melinodes 251, no incompatibility among sites was detected that involved the tips of the lineages, either between different lineages or within lineages. In P. melinodes 251, the single mutation that underlies the reticulation might have been a multiple hit at this site or a PCR mutation.
In ß-tubulin, and thus also the combined matrix of all three loci, the central portion of the network becomes highly interconnected and multidimensional. The five lineages previously identified in the MP and ML analyses were supported. The only specimen that cannot be placed clearly into any of the lineages in the network is P. melinodes 259. It appears as an internal node between lineages III and IV. Central reticulations in these two partitions extend into the lineages and some of the tip haplotypes are involved in incongruencies. However, it appears that these reticulations do not represent conflicts limited to relationships within a lineage (an exception might be the circle involving P. flavocoerulescens 111 and 112 and P. melinodes 273). On the contrary, these reticulations are part of large split systems (see the parallel edges) and thus are due to incongruence that is also involved in conflict among lineages.
The low amount of structure in lineages I, II, and III is probably due to the low level of variability within the lineages (rather than a lack of sampled specimens). However, lineage IV in the LSU and RPB2, as well as lineages V and III in ß-tubulin show structure within the lineages, but no reticulations. One exception, as discussed before, is P. melinodes 251 in the LSU data set, a specimen that is connected to the core by a reticulation. In ß-tubulin, P. melinodes 251 and P. melinodes 259 introduce incompatibility into lineages III and IV.
In the network analyses as in the tree reconstructions, the species pair is rooted centrally. Porpidia carlottiana attaches to the networks generally in a position that groups lineages I and II to one side and lineages III–V to the other side of the networks. The networks thus provide additional support for the placement of the root. They can provide information especially in cases in which only a distantly related outgroup is available.
Distribution of Reproductive Modes Among Lineages
Observed reproductive modes are not strictly correlated with phylogenetic relationships within and among lineages. For example, in the most inclusive LSU data set (fig. 4), lineage I consists of two sorediate specimens, its sister lineage II contains exclusively sexual specimens. Lineage III consists of a single specimen (P. flavocoerulescens 242) with only apothecia, 31 collections with only soralia, and 4 specimens with both apothecia and soralia. Lineage IV + V represents a mixture of sexual and vegetatively reproducing specimens. In ß-tubulin, groups IV and V form separate lineages. Lineage V consists of 12 specimens, most of which reproduce exclusively sexually. However, four sorediate collections are also members of this lineage. Lineage IV and V in the LSU and RPB2 trees and lineages III and V in the ß-tubulin tree show relative high amounts of within-lineage variability (fig. 2 and Supplement 4, Supplementary Material online).
|
Repeatedly, the same haplotype is shared by specimens that show contrasting reproductive modes or specimens with both reproductive modes present on one thallus (called "mixed" reproduction). However, all specimens showing the mixed reproductive mode share haplotypes that are either found only in sorediate specimens or in collections with soredia plus apothecia. They never had a haplotype that was exclusively found in sexual collections.
Hypothesis Testing: Relationships Between Sexual and Vegetative Specimens
The hypothesis of sexual and vegetatively (including mixed reproduction) reproducing specimens forming two independently evolving lineages with a single switch between reproductive modes at the base of the species pair (hypothesis I) was rejected for each locus. Tests were performed evaluating hypotheses of monophyly of vegetative specimens, of sexual specimens, and of reciprocal monophyly of both groups of specimens for each locus. In none of the plateaus investigated did a single tree occur in which a separation between sexual and vegetatively reproducing specimens was observed (in all cases: posterior probability 
0.0001).
On the other hand, the hypothesis that both reproductive modes are distributed randomly throughout the species pair (hypothesis II) could also be rejected for each of the loci. On the LSU topology, 11 steps were required to account for the observed distribution of both reproductive modes in the tips of the tree; on the RPB2 topology, 3 steps were required; and on the ß-tubulin topology, seven steps were needed. However, the null distributions obtained through the randomizations of character states predicted 34–48 (mean = 45.07, P 0.001); 5–10 (mean = 8.74, P 0.001); and 6–15 (mean = 11.47, P = 0.003) steps for a random distribution of reproductive modes on LSU, RPB2, and ß-tubulin ML topologies, respectively. The observed number of steps was in each case significantly lower than what would be expected if reproductive modes were randomly distributed. The MP reconstructions of reproductive modes on each of the gene trees reconstructed the vegetative modes as ancestral in all cases independently of how mixed specimens were coded.
|
Discussion |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
The primary goal of the present investigation was to test a priori proposed evolutionary hypotheses about the processes and histories shaping populations of lichen-forming ascomycetes. The existing hypotheses on the species pair phenomenon have been widely referred to and discussed for over 80 years. That the concepts are still open to debate shows how little progress has been made with regard to questions of basic population biology and evolution in these fungi. Employing stringent hypothesis testing, the conclusions of the present study are that the traditionally proposed hypotheses of Poelt (1963
; hypothesis I), Robinson (1975; hypothesis II), and Tehler (1982) can be rejected for P. flavocoerulescens and P. melinodes. It can be expected that Poelt's hypothesis and Robinson's hypothesis will also be rejected for previously investigated species pairs if they were tested with available molecular data because phylogenetic patterns found within all studied species pairs have been similar (see below). We also carefully considered whether or not the identified lineages represented cryptic species (e.g., Kroken and Taylor 2001a) and decided that the cryptic species hypothesis does not fit our results. While testing the proposed hypotheses, the present study provided important insights into the evolutionary processes and history underlying the orange species pair that should focus our attention and efforts on new developments and theories, which can be tested in the future.
Hypothesis I: Sexual Primary Species Giving Rise to a Monophyletic Vegetative Lineage
As mentioned in Introduction, the results of all molecular studies to date show that this hypothesis can be rejected as neither sexual nor vegetative lineages are monophyletic.
Hypothesis II: Species Pairs Represent Single Polymorphic, Recombining Species
This has been rejected as all published studies document that the species pairs are well structured and show well-supported groups of specimens.
Hypothesis III: Sexual Primary Species Spin-Off Short-Lived Vegetative Lineages
J. Buschbom and D. Barker (unpublished data) showed with high probability that the ancestors to the species pairs in Porpidia were predominantly vegetatively reproducing. In addition, underlying Tehler's species pair concept is a transition model in which transitions only occur from the sexual to the vegetative reproductive state, reversals are not possible. However, in Porpidia the reconstructions showed a transition model in which the rate of transitions away from the vegetative state is several times higher than vice versa. Thus, hypothesis III can be discarded for species pairs within the genus Porpidia.
Hypothesis IV: Cryptic Species
Kroken and Taylor (2001a) proposed that observed patterns of deeply isolated lineages represent groups of cryptic species. The species are called cryptic because they are apparent only in genetic analyses, that is, they are not distinguishable morphologically and/or biochemically. Considering the different aspects of this hypothesis closely, we, however, conclude that a process of cryptic speciation does not adequately explain the evolution of the orange species pair (discussed in detail below).
A New Hypothesis: Selective Sweeps in Predominantly Asexual Populations
We propose that symbiont interactions characterized by successive selective sweeps of highly fit symbiont combinations in a generally vegetatively reproducing fungal population underlie the observed deeply divided genetic lineages within the orange species pair. Lichen-forming ascomycetes need to balance conflicting constraints: the symbiosis is essential for the survival of these obligate symbionts. Reproducing vegetatively (dispersing fungus and alga together) ensures the continued propagation of the symbiosis. However, population-genetic modeling suggests that exclusive asexuality is generally not viable in the long term. If the fungi reproduce sexually, they reap the advantages of recombination, but they loose their photobionts because the fungal propagules are dispersed independently of the alga. The new individual will need to acquire the appropriate photobiont to survive.
Generally, the fungus will reproduce vegetatively so that both partners are dispersed together. Here, the probability to switch partners is low and the fungus is "stuck" with its algal symbiont even if the particular fungus-alga combination is not advantageous. In such a disadvantageous situation the fungus would have a selective advantage to reproduce sexually because this would increase the chance to "escape" from its current partner. It is assumed that this step in the lifecycle of the fungus represents a severe bottleneck because of the infrequency of forming a new fit fungus-alga partnership. However, once a highly fit symbiont combination has been founded, this particular combination of fungus and alga will expand in numbers and sweep through the population. In this situation it is advantageous for the fungus to not switch algae and to reproduce vegetatively, rapidly dispersing both partners together. After some time, either a more successful symbiont combination might appear sweeping itself through the population, or the originally advantageous lichen symbiosis might degrade. Degradation could occur either through mutation or recombination breaking up well-integrated genotypes. At this point selection will favor the fungus' switch to sexual reproduction and the cycle starts again.
The present investigation documented that diverse genotypes occur within the fungal populations. No such information exists for the photobiont in the orange species pair. However, in the species pair Letharia columbiana-Letharia vulpina, the photobiont was also represented by a diversity of genotypes (Kroken and Taylor 2000). Furthermore, several culture studies have found that the success and resulting morphology of lichen symbioses in cultures depend on the combination of specific fungal and photobiont clones (Ahmadjian 1993).
We prefer this hypothesis over the cryptic species hypothesis for a number of reasons.
Apart from the reproductive system, several additional character systems have been proposed to be of importance for the evolution within species pairs. These are morphology, secondary metabolites, ecology, and geography. Similar to other molecular phylogenetic investigations of species pairs, some of the lineages that we identified have been found to correlate with a specific morphological character, chemotype, or environment (fig. 4; see detailed discussion in Buschbom 2003). However, more often than not the phylogenetic lineages would not have been predicted by any of these characters or their combinations. In neither the present nor any previous study could a character system be found that correlates closely with the observed genetic variation, that is, the potential cryptic species.
Population history is another factor that has been proposed to have shaped species pairs through allopatric speciation followed by the subsequent merger of partially reproductively isolated populations. Taxa of the orange species pair are boreal-arctic in their distribution, thus, for them Pleistocene habitat restrictions will not have occurred. The most recent earth age that might have led to drastic distribution range constrictions was the Miocene (Tertiary, 5–23 MYA), when the vegetation zones of the northern hemisphere showed a pronounced northward shift. It remains unclear how Tertiary refugia could have produced such divergent genetic lineages without those species also having significantly diverged morphologically and biochemically over the past millennia. Thus, none of the alternative forces that might drive reproductive, and thus genetic, isolation appear to be the causes of, or closely associated with, the processes that underlie the patterns of lineages that were revealed during this study.
Additionally, recombination, rather than lineage sorting, seems to be a more likely explanation for the pattern of conflict observed in the orange species pair. Conflict between gene loci has been found in all published multilocus studies that conducted tests of incongruence between the employed loci (Kroken and Taylor 2001a; Myllys, Lohtander, and Tehler 2001; Högberg et al. 2002, this study). Conflicting relationships for individual specimens (Myllys, Lohtander, and Tehler 2001), or groups of specimens (Kroken and Taylor 2001a), were reconstructed based on different loci. In the present study, only two cases were found in which reconstructed relationships between lineages conflicted between loci: P. melinodes 259 appears in lineage IV in the LSU and RPB2, but basal to lineages III and IV in the ML analysis of ß-tubulin. This specimen does not seem to fit well into either of the two lineages. Conflict among data sets regarding the relationships among observed lineages is found for lineages III, IV, and V. In the LSU and RPB2 lineages IV and V are not differentiated, while in ß-tubulin lineage IV represents a well-resolved sister group to lineage III.
Myllys et al. (1999) proceeded to combine the loci despite conflicts based on a total evidence argument. However, evolutionary processes at the intraspecific level differ from those at the interspecific level, so combining conflicting data sets when addressing population-level questions would preclude addressing microevolutionary processes that give rise to the conflicts. Kroken and Taylor (2000, 2001a, 2001b) and Högberg et al. (2002) resolved to exclude homoplasious sites and specimens that introduced incongruence before they combined data sets for further analyses. Their combined analyses had increased resolution and nodal support than their single loci analyses, but information important for testing mechanistic hypotheses were lost in the process. In the present study, we decided that a combined analysis of the three loci which forces the relationships into a tree-like structure would not be informative for addressing the species pair question, especially if there is a possibility that recombination underlies the distinct relationship patterns tracked by the LSU and RPB2 versus ß-tubulin sequence data. Even though the combined topology does provide more resolution and higher branch support (data not shown), those relationships are likely misleading and may obscure patterns that can provide important information on evolutionary processes.
Previously, conflict in phylogenetic reconstructions between loci has either been used to stress that recombination exists in lichen-forming ascomycetes (Kroken and Taylor 2001b) or was explained as being due to incomplete lineage sorting between genetically isolated lineages (Kroken and Taylor 2001a). Lineages IV and V in the orange species pair are each identified by several sites and highly supported by bootstrap values and posterior probabilities in the ß-tubulin tree, but are not resolved in the LSU and RPB2 trees. Members of lineage V are scattered throughout a large clade that contains specimens grouped into lineages IV and V in ß-tubulin and do not show any affinity to each other. Rather, they share several divergent and derived haplotypes with members of lineage IV. Because the lineages are so distinct in ß-tubulin, it might be expected that even if lineage sorting in the LSU and RPB2 was incomplete, signs of an independent evolutionary history would be evident as synapomorphies and that coalescence of at least some haplotypes of lineage V and more haplotypes of lineage IV would have been reached. However, signs of an independent evolutionary history for these specimens were not observed in either LSU or RPB2 data sets. The probability that all three loci are tracking the same (cryptic) species tree appears to be low under the observed divergence scenario and with the number of specimens sampled in the present study (Rosenberg 2002).
Finally, DNA sequencing of continuous 0.6–1.1-kb stretches of gene fragments in the present study allowed not only the investigation of conflict between loci but also the evaluation of incongruent site patterns within loci. Phylogenetic approaches, such as median networks, that represent conflicting signals in the form of reticulating multidimensional networks and do not force evolutionary relationship into tree-shaped graphs, proved to be of great use in identifying and delimiting the parts of the relationship network within the orange species pair that were affected by incongruence.
Incongruence in the present data set was found to be more dominant within genes than between loci. Network analyses of combined matrices of the LSU and RPB2, and LSU, RPB2 and ß-tubulin did not show drastic increases in, or new patterns of, incongruence compared to the individual loci, but rather identified complexities that could reasonably be expected from the individual loci. Thus, each of the loci in the P. flavocoerulescens-P. melinodes species pair tracks the main aspects of the evolution of the group.
The pattern of conflict characteristic for cryptic species as proposed by Kroken and Taylor (2001b) with recombination within lineages, but not between lineages was not observed in the present investigation. Each of the analyzed gene fragments provides evidence for more or less extensive conflict in the backbone, which connects the different lineages. No, or only low, amounts of conflict were observed within the identified lineages, despite the fact that some lineages showed greater genetic variability than others.
A scenario of cryptic species would explain deeply divided lineages in a group of frequently outcrossing organisms. However, the patterns of conflict found in the orange species pair do not fit this hypothesis well (cf., contrasting amounts of conflict within and between loci and lineages). Furthermore, while many processes can be responsible for a pattern of cryptic, isolated lineages, so far no character system or historical factor could be identified that is closely associated with the genetic diversity within the species pairs. An assumption of a complex of cryptic species would require repeated sympatric speciation events. Thus, the question remains why would sympatric populations in P. flavocoerulescens and P. melinodes speciate repeatedly to form lineages that are not morphologically distinguishable and that reproductively and chemically are subject to parallel evolution and/or numerous reversals?
At this point it seems more reasonable to assume that the diversity found within the orange species pair primarily represents intraspecific variation. Besides the points made above, an additional field observation supports this conclusion: on rock surfaces on which thalli of different lineages within the orange species pair co-occur, pycnidia (assumed to play a role in sexual reproduction in these fungi) are mainly produced along those thalli margins that are shared between thalli belonging to the species pair but not along margins that are bordered by other lichen species. Suggesting, too, that the lineages are not reproductively isolated, but belong to a single species.
Porpidia flavocoerulescens and P. melinodes are the sixth species pair investigated with molecular DNA-based information (Lohtander, Källersjö, and Tehler 1998; Lohtander et al. 1998, 2000; Myllys et al. 1999; Kroken and Taylor 2001a; Myllys, Lohtander, and Tehler 2001; Högberg et al. 2002). These species pairs differ from each other in their systematic position and relationships (Arthoniomycetes vs. Lecanoromycetes), growth forms (crustose vs. foliose vs. fruticose), ecology (substrata, habitats), biogeography (vegetation zones and histories), and distribution patterns (restricted vs. holarctic, disjunct vs. continuous ranges). Nevertheless, similar phylogenetic patterns are reported in each study system. The data we present on the orange species pair can be reconciled according to the proposed hypothesis by rare recombination events in a predominantly vegetatively reproducing organism. Here switches between reproductive modes are triggered through symbiont interactions. To what extent the proposed hypothesis will hold up in the future and will be applicable to other species pairs will need to be investigated and tested.
|
Supplementary Material |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Supplements 1–4 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
This study was supported by a National Science Foundation Doctoral Dissertation Improvement grant (DEB-0105024), as well as awards and grants from the American Society of Plant Taxonomists, the Botanical Society of America, the Explorers Club, Scott Polar Research Institute, and the University of Chicago Graduate Student Hinds Fund. The first author received fellowships from the German Academic Exchange Service (DAAD), the Field Museum of Natural History, and the Mycological Society of America.
|
Footnotes |
|---|
1 Present address: Institute of Forest Genetics and Forest Plant Breeding, Großhansdorf, Germany.
|
References |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Ahmadjian, V. 1993. The lichen symbiosis. John Wiley & Sons, New York.
Bandelt, H.-J. 1994. Phylogenetic networks. Verh. Naturwiss. Ver. Hambg. 34:51–71.
Bandelt, H.-J., P. Forster, B. C. Sykes, and M. B. Richards. 1995. Mitochondrial portraits of human populations using median networks. Genetics 141:743–753.[Abstract]
Beard, K. H., and P. T. DePriest. 1996. Genetic variation within and among mats of the reindeer lichen, Cladonia subtenuis. Lichenologist 28:171–182.[CrossRef]
Bowler, P. A., and P. W. Rundel. 1975. Reproductive strategies in lichens. Bot. J. Linn. Soc. 70:325–340.
Buschbom, J. 2003. Evolutionary history and processes underlying contrasting reproductive modes in the lichen-forming genus Porpidia (Ascomycota). Pp. 244. Committee on evolutionary biology. University of Chicago, Chicago.
Buschbom, J., and G. Mueller. 2004. Resolving evolutionary relationships in the lichen-forming genus Porpidia and related allies (Porpidiaceae, Ascomycota). Mol. Phylogenet. Evol. 32:66–82.[Medline]
Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation procedures. Evolution 32:550–570.[CrossRef]
Crespo, A., P. D. Bridge, O. F. Cubero, and D. L. Hawksworth. 1997. Determination of genetic variability in the lichen-forming fungus Parmelia sulcata. Pp. 73–79 in R. Zorer, ed. Progress and problems in lichenology in the nineties. J. Cramer Verlag, Berlin, Stuttgart, Germany.
Culberson, W. L. 1973. The Parmelia perforata group: niche characteristics of chemical races, speciation by parallel evolution, and a new taxonomy. Bryologist 76:20–29.
Du Rietz, G. E. 1924. Die Soredien und Isidien der Flechten. Sven. Bot. Tidskr. 18:371–396.
Farris, J. S. 1970. Methods for computing Wagner trees. Syst. Zool. 34:21–34.
Farris, J. S., M. Källersjö, A. G. Kluge, and C. Bult. 1995. Testing significance of incongruence. Cladistics 10:315–319.[CrossRef][ISI]
Felsenstein, J. 1981. Evolutionary trees from DNA sequences: a maximum likelihood approach. J. Mol. Evol. 17:368–376.[CrossRef][ISI][Medline]
Fitch, W. M. 1971. Toward defining the course of evolution: minimal change for a specific tree topology. Syst. Zool. 20:406–416.[CrossRef]
Gowan, S. P. 1989. The lichen genus Porpidia (Porpidiaceae) in North America. Bryologist 92:25–59.
Gowan, S. P., and T. Ahti. 1993. Status of the lichen genus Porpidia in eastern Fennoscandinavia. Ann. Bot. Fenn. 92:25–59.
Hale, M. E. J. 1965. A monograph of Parmelia subgenus Amphigymnia. Cladistics 36:193–358.
Hendy, M. D., and D. Penny. 1993. Spectral analysis of phylogenetic data. Cladistics 10:5–24.
Hertel, H. 1977. Gesteinsbewohnende Arten der Sammelgattung Lecidea (Lichenes) aus Zentral-, Ost- und Südasien. Khumbu Himal 6:145–458.
Hertel, H., and J. G. Knoph. 1984. Porpidia albocaerulescens. Eine weit verbreitete, doch in Europa seltene und vielfach verkannte Krustenflechte. Mitt. Bot. Staatssamml. Muench. 20:467–488.
Högberg, N., S. Kroken, G. Thor, and J. W. Taylor. 2002. Reproductive mode and genetic variation suggest a North American origin of European Letharia vulpina. Mol. Ecol. 11:1191–1196.[Medline]
Huber, K., M. Langton, D. Penny, V. Moulton, and M. D. Hendy. 2002. Spectronet: a package for computing spectra and median networks. Appl. Bioinform. 1:159–161.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17:754–755.
Inoue, M. 1983. Japanese species of Huilia (Lichens), 1–3. J. Jpn. Bot. 58:113–128, 161–173, 225–236.
Kluge, A. G., and J. S. Farris. 1969. Qualitative phyletics and the evolution of anurans. Syst. Zool. 18:1–32.
Kroken, S., and J. W. Taylor. 2000. Phylogenetic species, reproductive mode and specificity of the green alga Trebouxia forming lichens with the fungal genus Letharia. Bryologist 103:645–660.
———. 2001a. A gene genealogical approach to recognize phylogenetic species boundaries in the lichenized fungus Letharia. Mycologia 93:38–53.
———. 2001b. Outcrossing and recombination in the lichenized fungus Letharia. Fungal Genet. Biol. 34:83–92.[Medline]
Landvik, S., O. E. Eriksson, and M. L. Berbee. 2001. Neolecta—a fungal dinosaur? Evidence from beta-tubulin amino acid sequences. Mycologia 93:1151–1163.
Liu, Y. J., S. Whelen, and B. D. Hall. 1999. Phylogenetic relationships among ascomycetes: evidence from an RNA polymerase II subunit. Mol. Biol. Evol. 16:1799–1808.[Abstract]
Lohtander, K., M. Källersjö, R. Moberg, and A. Tehler. 2000. The family Physciaceae in Fennoscandia: phylogeny inferred from ITS sequences. Mycologia 92:728–735.