Morphological, chemical and species delimitation analyses provide new taxonomic insights into two groups of Rinodina

The genus Rinodina (Physciaceae), with approximately 300 species, has been subject to few phylogenetic studies. Consequently taxonomic hypotheses in Rinodina are largely reliant on phenotypic data, while hypotheses incorporating DNA dependent methods remain to be tested. Here we investigate Rinodina degeliana/R. subparieta and the Rinodina mniaraea group, which previously have not been subjected to comprehensive molecular and phenotypic studies. We conducted detailed morphological, anatomical, chemical, molecular phylogenetic and species delimitation studies including 24 newly sequenced specimens. We propose that Rinodina degeliana and R. subparieta are conspecific and that chemical morphs within the R. mniaraea group should be recognized as distinct species. We also propose the placement of the recently described genus Oxnerella in Physciaceae.


Introduction
Species classification and delimitation within the fungi received a strong impetus when molecular phylogenetic methods emerged (Blackwell 2011;Hibbett & Taylor 2013). Molecular methods have helped to solve several long-standing questions about higher-level taxonomic relationships among lichen-forming fungi (e.g. see James et al. 2006;Thell et al. 2012;Miadlikowska et al. 2014;Resl et al. 2015). However, applying molecular methods also creates new challenges and discussions are ongoing about how to crosslink phenotypic (including morphology, anatomy and secondary chemistry) information with DNA sequence data when drawing taxonomic conclusions such as the description of new lineages (Grube & Kroken 2000;Crespo & Lumbsch 2010;Hibbet & Taylor 2013).
The issue of reconciling molecular and phenotypic evidence to develop taxonomic hypotheses for individual species may be simplified to three different scenarios: 1) molecular and phenotypic evidence both support taxonomic conclusions; 2) taxonomic conclusions are supported by molecular but not phenotypic characters; and 3) phenotypic evidence supports taxonomic conclusions while molecular methods do not.
Not all characters may coincide with a taxonomic hypothesis. Nevertheless, a broad consensus exists that multiple independent lines of evidence in support of taxonomic conclusions should be the desired goal of every modern taxonomic study. Lichenologists, however, are often faced with the challenge of phylogenetic and phenotypic data being inconsistent (e.g. Baloch & Grube 2009;Altermann et al. 2014;Lücking et al. 2014;Singh et al. 2015). With more sequences becoming available, species boundaries traditionally circumscribed by phenotypic characters were challenged and it was proposed that extant taxa numbers were systematically underestimated when relying on phenotypic characters alone (Crespo & Lumbsch 2010;Lumbsch & Leavitt 2011). However, there are also cases where molecular methods have revealed that phenotypic differences might overestimate species diversity (Xanthoparmelia: Leavitt et al. 2011a, b; Bryoria section Implexae: Myllys et al. 2011;Velmala et al. 2014).
While phenotypic differences between species can be subtle (Spribille et al. 2014) they still represent a valuable source of information with which to formulate hypotheses on how to delimit species, especially in groups where phylogenetic data remain fragmentary. One such group, which has been subjected to rather few molecular phylogenetic studies, is the family Physciaceae (e.g. Lohtander et al. 2000Lohtander et al. , 2009Wedin et al. 2000;Grube & Arup 2001;Helms et al. 2003;Crespo et al. 2004;Kaschik 2006;Nadyeina et al. 2010;Kondratyuk et al. 2014a). In this family, taxa are delimited mainly on the basis of secondary chemistry, asci and ascospores. The latter particularly have been emphasized as important and, consequently, ascospores are usually summarized into different 'types' (Poelt & Mayrhofer 1979;Mayrhofer & Moberg 2002;. A thorough phylogenetic treatment of the Physciaceae including members of all known genera and sequences of multiple gene fragments to test these concepts is still missing. We focus here on two examples of closely related taxa with unclear taxonomy from the genus Rinodina with Physciaand Physconia-molecular characters also allows us to provide integrated hypotheses of species taxonomy and nomenclatural status. The first case was presented to us in the course of studying corticolous Rinodina specimens collected by GT and TT from Japan. Fertile, esorediate collections were found showing close morphological, anatomical and chemical similarities to the sorediate species Rinodina degeliana Coppins. Rinodina degeliana is scattered throughout the Northern Hemisphere. It occurs in Europe (Coppins 1983;Tønsberg 1992;Mayrhofer & Moberg 2002;Giavarini et al. 2009), North America Lendemer et al. 2014) and Japan (GT, TT, J.W. Sheard, pers. obs.). The esorediate specimens were subsequently found to be phenotypically identical to the holotype of Rinodina subparieta (Nyl.) Zahlbr., a species considered to be endemic to Japan.
The treatment of otherwise identical sorediate or esorediate morphs has been discussed by Tønsberg (1992), and more recently by Brodo & Lendemer (2012), among others. For a full literature list and exhaustive discussions, please consult the references in these two papers. Suffice it to say that there is no consensus on how different morphs of lichen taxa should be recognized. Most instructive has been the molecular study of Tehler et al. (2009), of a single species aggregate within the genus Roccella where different taxa were found to be either variable or uniform with respect to their reproductive morphs. Therefore, it appears that all apparent pairs of fertile and vegetative morphs must be studied in detail in order to decide their taxonomic status. This leads us to question if the names Rinodina degeliana and R. subparieta should both continue to be recognized as distinct species distinguished by the presence or absence of soralia.
The second example is Rinodina mniaraea (Ach.) Körb., a terricolous taxon that has been studied in detail by Timpe (1990) and Ertl (2000). It is found in cold and moist alpine dwarf shrub and grassland habitats on bare soil, bryophytes and plant remnants (Mayrhofer & Moberg 2002;. Within R. mniaraea several infraspecific taxa have been recognized, differing in chemistry (Mayrhofer & Moberg 2002). All these taxa regularly form apothecia with Lecanora-type asci, possess large Physcia-type ascospores and frequently contain variolaric acid. They have been accepted as varieties in a treatment of the genus Rinodina (Mayrhofer & Moberg 2002), although  did not accept the use of varietal status. However, the names R. cinnamomea (Th. Fr.) Räsänen and R. mniaraeiza (Nyl.) Arnold already existed at the species rank for two of the varieties. How should these entities be referred to and which taxonomic rank is appropriate for them?
Here we present the results of a study using collections of R. degeliana and R. subparieta from multiple localities within their known Northern Hemisphere range and material of the three varieties of R. mniaraea from the European Alps and western North America. We performed detailed morphological, anatomical and chemical analyses of material from those species. Additionally, we reconstructed maximum likelihood and Bayesian phylogenies of numerous Rinodina taxa including R. degeliana, R. subparieta, R. mniaraea var. mniaraea, R. mniaraea var. mniaraeiza and R. mniaraea var. cinnamomea based on sequences from nuclear and mitochondrial gene fragments. We performed species delimitation methods using a Bayesian implementation of the General Mixed Yule Coalescence model (Reid & Carstens 2012) to find evidence for our two aims: 1) to examine species delimitation within these closely related sets of taxa in order to re-evaluate esorediate and sorediate morphs (in the case of R. degeliana/subparieta) and different chemotypes (R. mniaraea s. lat.), and 2) to decide on the status of the names R. degeliana, R. subparieta and the varieties of R. mniaraea s. lat.

Materials and Methods
SC, TS, GT, TT and James Lendemer provided fresh collections for molecular analysis. For Rinodina degeliana and R. subparieta, procedures for morphological and anatomical examination, and photography, follow those of Sheard et al. (2014). For R. mniaraea s. lat., we provide revised descriptions based on Timpe (1990). Spore measurements for R. subparieta/R. degeliana are quoted as the 25%-75% range around the median with the 5% and 95% outliers in brackets. For R. mniaraea s. lat the average spore size and the minimum and maximum range are given. Thin-layer chromatography was carried out according to the methods of Culberson & Kristinsson (1970) and Culberson (1972) with later modifications.
As a basis for our study we selected sequences from the dataset of Nadyeina et al. (2010) containing a broad range of Physciaceae species with different ascospore types and secondary metabolites. We were especially interested in the relationships of R. degeliana and R. subparieta with other Rinodina species possessing Physcia-type ascospores and the secondary compound atranorin. In addition to the ten samples from R. degeliana and two from R. subparieta, we also included eight specimens of R. mniaraea from its three varieties, all possessing Physcia-type ascospores and sometimes atranorin. Furthermore, we sequenced R. freyi H. Magn. and R. efflorescens Malme with Physcia-type spores, but both lacking atranorin. To complete our dataset we included samples of the previously unsequenced or rarely sequenced species R. trevisanii (Hepp) Körb. and Phaeorrhiza nimbosa (Fr.) H. Mayrhofer & Poelt, and also the hitherto unplaced Oxnerella safavidiorum S. Y. Kondr. et al.

DNA isolation, PCR amplification and sequencing
Thallus squamules or apothecia were placed in 1·5 ml Eppendorf tubes, frozen at −80 °C and ground to powder in a Retsch cell grinder. We directly applied lysis buffer to the sheared cells. DNA extraction followed the QIAmp DNA Investigator Kit protocol according to the manufacturer's instructions. We eluted the nucleic acids in 50 μl elution buffer and used them undiluted for subsequent polymerase chain reactions (PCR). For each sample we sequenced the internal transcribed spacer regions 1 and 2 with the embedded 5.8S region of the ribosomal DNA (ITS) and the mitochondrial ribosomal small subunit (mtSSU). PCR was performed with PuReTaq Ready-To-Go PCR beads using primer pairs ITS1F/ITS4 (Gardes & Bruns 1993;White et al. 1990) and mtSSU1:mtSSU3R (Zoller et al. 1999) for ITS and mtSSU, respectively. PCR products were checked for the correct size on ethidium bromidestained agarose gels and purified using the Omega E.Z.N.A Cycle Pure Kit according to the manufacturer's instructions. Purified PCR products were sequenced by Microsynth (Vienna).

Alignment and phylogenetic reconstruction and tree selection
Raw sequences were subjected to manual quality control with BioEdit 7.2.5 (Hall 1999) and only high quality sequences were used for phylogenetic analyses. Sequences were aligned and concatenated using our in-house script pipeline (Resl et al. 2015;Resl 2015). We aligned sequences with MAFFT v.7 (Katoh & Standley 2013) employing the -genafpair algorithm. Alignments were checked manually with BioEdit 7.2.5 and only obvious errors (when MAFFT placed single nucleotides on either end of the alignments introducing long gaps) were corrected. These alignments were then used, without excluding any sites. Data matrices and trees are deposited at Treebase.org under study ID: 19205. We created gene trees for the two gene fragments with RAxML v.8 (Stamatakis 2014). For each gene fragment we set RAxML to generate 500 fast bootstrap replicates. We partitioned the ITS dataset into the ITS1, 5.8S and ITS2 components and used PartitionFinder 1.1.1 (Lanfear et al. 2012) with a greedy search strategy under a BIC criterion to determine the optimal number of partitions and substitution models for RAxML. Accordingly, the dataset was divided into a 5.8S and ITS1/ITS2 partition with GTRGAMMA respectively. For the mtSSU dataset we used one partition and the GTRGAMMA substitution model. We identified topological conflicts among gene trees with compat.py (Kauff & Lutzoni 2002) based on a support threshold of 70. Conflict-free single locus datasets were combined in a concatenated RAxML analysis. Again we used PartitionFinder to identify the optimal number of partitions and substitution models. Accordingly we used two partitions (ITS1/ ITS2/mtSSU and 5.8S) with a GTRGAMMA substitution model and generated 500 fast bootstrap replicates.
We also performed Bayesian phylogenetic reconstruction of the ITS dataset using BEAST 2.2.1 (Bouckaert et al. 2014). We determined the best site model and partitioning scheme for the ITS dataset (ITS1: GTR+G, 5.8S: TrNef+G, ITS2: GTR+G) with PartitionFinder 1.1.1 according to the BIC criterion and used a relaxed log-normal clock model for each partition. We linked the tree model and subjected it to a Yule tree prior. The MCMC chain was run for 10 7 generations and every 1000th tree was retained. Convergence of model parameters was assessed using Tracer 1.6 (Rambaut et al. 2014) and considered sufficient for ESS values larger than 200. After discarding the first 20% of trees from the posterior tree sample as burn-in, we created a maximum clade credibility tree with TreeAnnotator 2.2.1 included in the BEAST package and randomly selected 100 trees using bash scripts. The maximum clade credibility tree and the set of 100 random trees were used in all further analyses.

Species delimitation using General Mixed Yule Coalescence
We performed species delimitation with the statistical programming language R (R Development Core Team 2013) using the R package bGMYC (Reid & Carstens 2012). This package provides a Bayesian implementation of the general mixed Yule-coalescence model (Pons et al. 2006). Based on a given tree sample, bGMYC models the contribution of between-species divergence (Yule model component) and within-species coalescence events (Coalescent model component) (Pons et al. 2006), which give a hypothesis for species delimitation. To account for phylogenetic uncertainty in our dataset we performed all bGMYC analyses on a set of 100 randomly selected trees from the BEAST posterior tree sample (see above). We used the function bgmyc. multiphylo() to sample 50 000 steps with a burn-in of 40 000 and the thinning parameter set to 100 for each tree. We then created a coassignment probability heat map as implemented in the bGMYC function spec.probmat().
To determine consensus partitions from the probability heat map, we employed a k-medoid clustering approach as described in Ortiz-Álvarez et al. (2015). We used the function pamk() in the R package fpc (Hennig 2013) to create groups from the bGMYC co-assignment matrix. Pamk() uses a partitioning around medoid (PAM) algorithm, which randomly selects data points from the co-assignment matrix as centres and clusters other data points around those centres according to their distance. The number of clusters is determined by optimum average silhouette width (Hennig 2013).

Phylogenetic analyses
We obtained 40 sequences from 24 isolates. Together with the sequences downloaded from GenBank, our dataset consisted of 49 samples and a total of 84 sequences. All sequences used, including those newly published, are summarized in Table 1. Results from our maximum likelihood analyses are depicted in Fig. 1A-C. We identified one conflict between the ITS (Fig. 1B) and mtSSU ( Fig. 1C) gene trees coming from specimen P280. As ITS is an important fungal barcode (Schoch et al. 2012) and our species delimitations were based on the ITS dataset, we excluded mtSSU sequences of these specimens from the concatenated phylogenetic analysis.
Sister to the above group, the "Physcia" clade 2 also contained a moderately supported clade (69% BS) consisting of Rinodina  The BEAST analyses of the ITS dataset led to stable parameter values (ESS > 200) according to Tracer. We only present the BEAST maximum clade credibility tree here (Fig.  2), with posterior probability node support indicated by different colours.

Species delimitation analysis
Based on the Bayesian mixed Yule-coalescent analysis for species delimitation (Reid & Carstens 2012), we detected significant levels of genetic structure in Rinodina degeliana and R. subparieta, as well as in R. mniaraea s. lat. The pamk clustering approach (see above) detected 21 clusters in the whole dataset. Rinodina degeliana and R. subparieta samples form a total of three clusters (cluster numbers 1-3; see Fig. 2). The two specimens of R. subparieta are assigned to cluster 1, together with three R. degeliana samples.

Morphological investigation
We undertook an a posteriori examination of ascospore size and structure, and soredium size between the R. subparieta/R. degeliana clades (Fig. 2) distinguished by the molecular analysis. This was necessarily incomplete due to the absence of ascospores in non-fertile specimens.
We found that specimens from all clades may possess soredia. Soredia were measured under the compound microscope to an accuracy of 5 μm but showed only a limited variation in size. Specimens from Norway and Washington (P102 and P233 respectively, pamk cluster 2) developed slightly smaller (median 25 μm diam.) soredia compared to 30 μm in other regions of the world ((20-)25-30(-40) μm, n = 115). However, this small size difference is accommodated in the 25-75 of the above percentile range and is therefore not meaningful. Soralia develop marginally (somewhat labriform; Fig. 3F & G) on the areoles, and are of similar size in the different clades.
Concerning Rinodina mniaraea, Sheard (2010) noted that he searched in vain for morphological, anatomical and ascospore characters to separate the different chemotypes. Numerous collections by Imshaug (MSC) in the southern Rocky Mountains offered the opportunity to test ascospore sizes of different chemotypes from the same collections or Resl et al. Page 8 Lichenologist. Author manuscript; available in PMC 2018 February 01.
Europe PMC Funders Author Manuscripts from the same localities. Some significant differences were found but they were not consistent between the chemotypes. Therefore,  was unable to recognize the chemical variants as being taxonomically distinct. Descriptions of the morphology of the taxa based on the chemical variants are found in the following section.
Secondary metabolites: atranorin (very rarely absent) and zeorin. Cortical atranorin typically abundant as indicated by strength of P+ spot test.
Notes. The revised description provided here is based on our studies of Rinodina degeliana from Europe (Coppins 1983;Tønsberg 1992;Mayrhofer & Moberg 2002), North America (Sheard 2010) (with the exception of the ascospore dimensions), the type of R. subparieta and fertile specimens from Japan collected by GT and TT. Rinodina subparieta is a variable species. In addition to having fertile sorediate, fertile esorediate (rare) and non-fertile sorediate morphs (the most frequent state), ascospore size and the development stage attained are also variable. Such ascospore variability is typical of many Rinodina species possessing vegetative diaspores (J. W. Sheard, pers. obs.).
The very detailed protologue of Rinodina degeliana (Coppins 1983) agrees with our observations of the types of that species. We found only one important difference: that fully developed ascospores belong to the Physconia-type, rather than the Dirinaria-type or Physcia-type as reported by Mayrhofer & Moberg (2002) and , respectively.  questioned the distinction between the Physciaand Physconia-ascospore types, citing the many Rinodina species that have intermediate ascospore types. Rinodina subparieta is another such example, with specimens exhibiting the Physcia-spore type having arrested ascospore development, a very common feature of the species in Europe, North America and Scandinavia. The ascospore dimensions cited above are from the total of specimens examined in the present study.
The species is easily distinguished from other sorediate Rinodina species (such as R. efflorescens, R. griseosoralifera Coppins, R. sheardii Tønsberg and R. stictica Sheard & Tønsberg) by its typically light grey areoles with marginal, labriform soralia and whitish soredia, and chemically by the presence of abundant cortical atranorin (P+ strong lemon yellow) and zeorin in its medulla. Very rarely the thallus and the soredia may have a greenish tinge (e.g. specimen in Fig. 3A). We have noted two thalline morphotypes. Most common is the morph with a plane and somewhat shining (waxy) surface, the areoles more or less angular and 0·6-0·8 mm wide, to which the type specimens of R. degeliana belong.
Very rarely such areoles coalesce to form a rimose-areolate thallus. The second morph, which may be more common in fertile specimens, has smaller areoles, rarely exceeding 0·5 mm in width, with a matt, slightly convex surface and minutely sublobate margins before breaking into soredia. It is notable that the esorediate form of R. subparieta has only been found at very high elevations on Honshu, Japan ( Pycnidia not observed. Chemistry. Orange to yellow pigment especially in the lower part of the medulla reacting K + purple to violet; chemotype A: Skyrin, ± variolaric acid, ± cinnamomea unknown (5/3/5), ± atranorin; chemotype B: 1-O-methylemodin or 8-O-methylemodin, ± variolaric acid, ± cinnamomea unknown, ± atranorin.
Ecology and distribution. On soil, bryophytes and decaying plants often above slightly calcareous and acid bedrock in arctic-alpine habitats. Widespread across the Northern Hemisphere, scattered in more southerly mountains such as the Rocky Mountains, northern Scotland, Alps, Carpathian Mountains, Balkan Peninsula, Anatolian Mountains, Caucasus, Tibet, Himalayas (Mayrhofer & Moberg 2002;Strasser et al. 2015).
Notes. This species is characterized by a yellow to orange medulla. Chemotype

Europe PMC Funders Author Manuscripts
Europe PMC Funders Author Manuscripts latter in the appendix. The index was given after the appendix, meaning that it was published at the same time in one issue. We therefore retain the name mniaraea, which was used earlier in Rinodina.

Discussion
In the present study we have investigated the status of Rinodina degeliana, R. subparieta and R. mniaraea s. lat. By evaluating evidence derived from molecular phylogenetic, chemical and anatomical studies we provide hypotheses about the taxonomy and updated nomenclature for these crustose lichen lineages.

The status of Rinodina subparieta and R. degeliana
We recovered samples of Rinodina degeliana and R. subparieta in two well-supported clades in our concatenated maximum likelihood phylogeny (Fig. 1A). According to pamkclustering based on the bGMYC results (Fig. 2), the two Rinodina subparieta samples were assigned to cluster 1 together with samples of R. degeliana. However, species delimitation also recovered Rinodina degeliana in two additional clusters (pamk cluster 2-3; Fig. 2), indicating multiple previously undiscovered lineages in R. degeliana.
The only reliable morphological character to separate Rinodina degeliana and R. subparieta was the lack of soralia in R. subparieta. We took an a posteriori approach to identify other possible synapomorphies informed by our phylogenetic hypothesis. Our morphological investigations, however, did not reveal any reliable differences between the R. degeliana clades or between R. degeliana and R. subparieta.
Fertile specimens were rarely observed in the material investigated and occasionally they lacked soralia. Ascospores from this material are variable in size and structure, even when immature and overmature ascospores were excluded from measurement. Such variable ascospores are typical for Rinodina species that tend to rely on vegetative reproduction (J. W. Sheard, pers. obs.). Young, not fully pigmented ascospores have Physcia-type lumina and mature into darkly pigmented Physcia-type ascospores or into Physconia-type ascospores also with darkly pigmented walls, a prominent torus and with rather heavily ornamented walls, although the latter character is sometimes less well developed. Asynchronous ascospore development was also frequent among specimens with Physconia-type spores; rarely only four ascospores in the ascus reach maturity.
The frequent development of Physcia-type ascospores into Physconia-type would seem to confirm the opinion of Sheard (2010) that these ascospore types should be regarded as developmental phases of a single ascospore type. Both are accompanied by a distinctive redbrown epihymenium colour in all clades, as are other species possessing these ascospore types (Ropin & Mayrhofer 1993). This coloration is derived from a dispersed red-brown pigment in the epihymenial gelatin together with the dark brown or black capitate pigmentation of the terminal cells of the paraphyses.
Conclusively, R. degeliana and R. subparieta (and also the different R. degeliana clades) were indistinguishable in respect to morphology, anatomy and chemistry. However, molecular results indicate multiple, although in part poorly supported, clades and clusters (Figs. 1A & 2). This could imply multiple species-level lineages in Rinodina degeliana. The specimens of R. subparieta available to us were nested in one of the R. degeliana clades and form a cluster with R. degeliana ITS sequences in bGMYC. Our molecular results and the lack of phenotypic characters to separate the two species thus lead us to conclude that R. subparieta and R. degeliana are conspecific.
We recognize that our interpretation of this species is broad and also includes specimens assigned to possible species-level lineages containing only R. degeliana specimens (pamk clusters 2 & 3; Fig. 2). This may represent an oversimplification of the taxonomic status of these lineages. At this point, the formal introduction of new names for any of the three clades would have to be based on molecular evidence only. It is generally possible to describe species based mainly on clade affinity in phylogenetic trees (Leavitt et al. 2013a), however we refrain from doing so in the case of Rinodina degeliana/R. subparieta.
Another solution to this problem, although probably impossible, would be to sequence the type specimens of R. subparieta and R. degeliana, ascertain which names apply to individual clades and adjust taxonomy accordingly. Since R. subparieta is nested in a clade (Fig. 1A) and cluster (Fig. 2) with R. degeliana samples and we lack morphological, anatomical and chemical characters to distinguish these lineages, we have been conservative in our interpretation. Further studies, particularly in view of a possible geographical separation, may help reveal the evolutionary history of the clades. Until additional data are available, we propose to use the name R. subparieta, which takes priority over R. degeliana.
Differences between R. mniaraea var. cinnamomea, R. mniaraea var. mniaraeiza and R. mniaraea var. mniaraea are restricted to secondary chemistry (Mayrhofer & Moberg 2002): var. mniaraea possesses either no secondary compounds or variolaric acid while var. mniaraeiza possesses atranorin and sometimes variolaric acid; var. cinnamomea produces the anthraquinones skyrin and 1-O-methylemodin or 8-O-methylemodin, sometimes with atranorin and variolaric acid or both, and one additional unknown. Räsänen (1931) had already established R. cinnamomea (Th. Fr.) Räsänen and Arnold (1870) transferred Lecanora "mniaroeiza" described by Nylander (1870) to Rinodina. It remains unclear if the lineages discovered by us are truly distinct species. We cannot rule out the alternative hypothesis that R. mniaraea varieties all belong to a single species. In the mtSSU gene tree (Fig. 1C) Rinodina mniaraeiza does not form a monophyletic group and we thus detected a topological conflict between gene trees. A topological conflict could indicate gene flow. However, at present the size of our molecular dataset of R. mniararea s. lat.
prevents us from investigating this issue in detail. An alternative explanation could be recent speciation with still incomplete lineage sorting. Additional data are needed to resolve this issue. We failed to find anatomical or morphological characters to separate taxa in R. mniarea s. lat. in a situation similar to that of R. subparieta/R. degeliana. Secondary chemistry, however, is characteristic and corresponded to clade (Fig. 1A) and cluster (Fig. 2) affinity in all R. mniaraea s. lat. specimens in our dataset. This is similar to other studies where morphologically indistinguishable taxa were proposed to be chemically distinct species (Lendemer 2012;Leavitt et al. 2013b). Our consistent chemical, phylogenetic (but Arnold and propose that the varieties of R. mniaraea s. lat. should be recognized at the species level.

The placement of Oxnerella safavidiorum in Physciaceae
The monotypic genus Oxnerella was recently described from the Iranian province Esfahan (Kondratyuk et al. 2014b). Oxnerella safavidiorum possesses a crustose rusty brown thallus, lecanorine apothecia with a Biatora-type ascus and hyaline 1-septate ascospores (Kondratyuk et al. 2014b). Kondratyuk et al. (2014b) conducted a phylogenetic analysis based on ITS and mtSSU sequences but could not resolve the placement of Oxnerella. A possible explanation is that because Oxnerella possesses hyaline 1-septate ascospores, Kontratyuk et al. (2014b) assumed a position close to Lecania in the family Ramalinaceae.
Our phylogenetic analysis now resolves Oxnerella safavidiorum as sister to Rinodina bischoffii with high support (82% BS; Fig. 1A). This placement should be considered tentative as long as fresh material is unavailable to verify sequence identity and rule out contaminations of Oxnerella type material. The differences in ascus type and ascospores of Oxnerella and Rinodina are surprising. A verification of ascus and ascospore characters of Oxnerella also has to be postponed until additional material for study is available to us.  Phylogenetic hypothesis and species delimitation results of the ITS dataset according to the Bayesian General Mixed Yule-Coalesence model. The tree displayed is the maximum clade credibility tree from the BEAST analysis. Node support is provided as coloured circles. Probabilities of the bGMYC probability map are indicative for the chance that tips could be assigned to one species. The pamk clusters refer to grouping based on k-medoids. bGMYC and pamk clustering was based on a tree sample of 100 randomly selected trees from the BEAST posterior distribution of trees.  ID, origin, species and used loci of the present molecular studies. IDs indicate laboratory-tracking numbers and also correspond to numbers used in Fig. 1 and the