Rostania revised: testing generic delimitations in Collemataceae (Peltigerales, Lecanoromycetes)

Abstract Here, we test the current generic delimitation of Rostania (Collemataceae, Peltigerales, Ascomycota) utilizing molecular phylogeny and morphological investigations. Using DNA sequence data from the mitochondrial SSU rDNA and two nuclear protein-coding genes (MCM7 and β-tubulin) and utilizing parsimony, maximum likelihood and Bayesian phylogenetic methods, Rostania is shown to be non-monophyletic in the current sense. A new generic delimitation of Rostania is thus proposed, in which the genus is monophyletic, and three species (Rostaniacoccophylla, R.paramensis, R.quadrifida) are excluded and transferred to other genera. Rostaniaoccultata is further non-monophyletic, and a more detailed investigation of species delimitations in Rostania s. str. is needed. The new combinations Leptogiumparamense and Scytiniumquadrifidum are proposed.


Introduction
Collemataceae is a large group of predominantly foliose lichenized fungi commonly known as the "jelly lichens" due to their gelatinous habit. This is caused by a polysaccharide matrix around the Nostoc cyanobacterial photobionts that swells and becomes extremely gelatinous when wet. Until very recently, the generic classification of the Collemataceae s. str. was very unnatural and based solely on one character, presence (Leptogium) or absence (Collema) of a cellular cortex (Degelius 1954(Degelius , 1974Jørgensen 2007). Already Degelius (1954) questioned the monophyly of Collema and Leptogium. This was also supported by molecular phylogenies (Wiklund and Wedin 2003;Miadlikowska and Lutzoni 2004;Miadlikowska et al. 2014), and somewhat surprisingly, gelatinous genera with one-septate spores that earlier were classified in Collemataceae, were shown to belong to the Pannariaceae (Wedin et al. 2009;Otálora et al. 2010;Ekman et al. 2014;Weerakoon et al. 2018) or Arctomiaceae . Not until Otálora et al. (2013aOtálora et al. ( , 2013b investigated the family in detail was a modern classification of Collemataceae s. str. proposed. Collema and Leptogium were confirmed as highly non-monophyletic, and Otálora et al. (2013b) instead suggested accepting 10 more or less morphologically distinct monophyletic groups from their tree, as genera. In addition to Collema and Leptogium in restricted senses, six old generic names were resurrected (Blennothallia Trevis., Enchylium (Ach.) Gray, Lathagrium (Ach.) Gray, Pseudoleptogium Müll. Arg., Rostania Trevis., and Scytinium (Ach.) Gray), and two new genera were described (Callome Otálora & Wedin and Paracollema Otálora & Wedin).
Rostania, the focus of the present study, corresponds to the Occultatum-group of Collema (Degelius 1954(Degelius , 1974. It is a comparatively small genus with eight currently accepted, mainly epiphytic species, characterised by very small to medium sized (ca 0.3-5 cm in diam.) subcrustaceous to subfoliose thalli with very small apothecia (ca 0.2-0.8(-1) mm in diam.) and cuboid to oblong muriform spores. All five species included in the Occultatum group by Degelius were treated in Rostania by Otálora et al. (2013b); Rostania callibotrys (Tuck.) Otálora Degelius (1954) divided Collema occultatum into two varieties: var. populinum which was characterised by a squamulose, somewhat lobate thallus, and which almost exclusively grew on the bark of Populus, and var. occultatum with a granulose thallus and which occurred on many deciduous trees, including Populus. Perlmutter and Rivas Plata (2018) Otálora et al. (2013aOtálora et al. ( , 2013b included only three species (R. ceranisca, R. multipunctata and R. occultata) in their phylogenies, and thus the taxonomical position of most species has not been tested by molecular methods. As there is a substantial variation in shape and size of the lobes, apothecia and ascospores, as well as the hyphal arrangement in the thallus among the Rostania species, and as several former Collemataceae taxa have been shown to belong outside the family, the delimitation of the whole genus needs investigation. Here, we will test the generic delimitation of Rostania and investigate the relationships of any species falling outside Rostania s. str. Finally, we will note and comment on any indication of species non-monophyly, in this genus.

Taxon sampling and morphological studies
We sampled 52 specimens of Collemataceae for the molecular study, including six of the eight currently accepted Rostania species and representatives of all genera within the family Collemataceae, including type species. Sequences originating from the study of Otálora et al. (2013a) were downloaded from GenBank (https://www.ncbi. nlm.nih.gov/) and all sequences used in this work are summarized in Table 1. Our own collections were deposited in UPS and S, and additional herbarium material from the herbaria PRA, GZU, UPS and S was also included (Table 1). Additional herbarium type material from the herbaria H and O was studied morphologically only (listed on the end of the manuscript). Herbarium acronyms follow Thiers (2018). Three species of Rostania not included in earlier studies were successfully added (R. callibotrys, R. quadrifida and R. paramensis). The sampling of Rostania occultata included specimens of both varieties. To enable testing of generic monophyly and family placement of taxa potentially to be excluded from Rostania, we added secondary outgroups including newly produced sequences of two species from the sister family Placynthiaceae (Placynthium nigrum and P. rosulans) and sequences available in GenBank of two from the more distantly related Pannariaceae (Pannaria rubiginosa and Staurolemma omphalarioides). Finally, Peltigera aphthosa was used as outgroup to root the tree.
We studied morphological and anatomical characters under the light microscope and dissecting microscope. We used hand-cut longitudinal sections of apothecia to observe internal and microscopic characteristics, in water. Microscopic examinations of the thalli were conducted on transversal cross-sections of lobes in water, or lactic blue.

Data generation
Two apothecia or (in the case of sterile samples) a thallus fragment, were selected for extraction. We extracted total DNA using the Plant DNA Mini Kit (Qiagen, Hilden, Germany) following the manufacturers' instructions. We amplified ca 0.6 kb of the small subunit of the mitochondrial rDNA (mtSSU), ca 0.6 kb of the two proteincoding genes DNA replication licensing factor mini-chromosome maintenance complex component 7 (MCM7) and the β-tubulin gene (b-tub) using the same primer combinations and PCR settings as in previous studies (Otálora et al. 2013a;Košuthová et al. 2016). We assembled and edited DNA sequences using Geneious version R8 (http://www.geneious.com; Kearse et al. 2012). Table 1. Sequences utilized in this study (newly produced sequences in bold, remaining sequences produced by Otálora et al. (2013a) and some of the outgroup sequences are taken from Wiklund and Wedin (2003), Buschbom and Mueller (2004), Otálora et al. (2010), Prieto et al. (2013)). In case of Rostania species, origin of both, state and provinces are given.

Sequence alignment and analysis
To identify and avoid contaminants among the new sequences, we used Megablast high similarity matches in Geneious version R8 (http://www.geneious.com; Kearse et al. 2012). Alignments were constructed using AliView 1.09 (Larsson 2014) with the "ClustalW/Multiple alignment" option and subsequent manual adjustments. All ambiguously aligned regions (sensu Lutzoni et al. 2000) were excluded from analysis. The mitochondrial and the two protein-coding datasets were analysed separately before concatenation using parsimony jackknifing (JK) in WinClada (Nixon 1999(Nixon -2002 with 100-200 replicates and otherwise default settings. As no significant (JK support above 70%) incongruence was detected, the alignments were concatenated. Final alignments have been deposited in TREEBASE (http://www.treebase.org) with accession number (http://purl.org/phylo/treebase/phylows/study/TB2:S23889). After concatenation, we inferred phylogenetic relationships using parsimony, maximum likelihood and Bayesian phylogenetic methods with indels treated as missing data. Partitions scheme and optimal model of nucleotide substitution for Bayesian analysis were selected using PartitionFinder2 (Guindon et al. 2010;Lanfear et al. 2012Lanfear et al. , 2016. PartitionFinder was set as follow: linked branch lengths, data blocks according to each codon position of each genetic region (mtSSU, MCM7, b-tub), the greedy search scheme, the Bayesian information criterion as selection metric and only models that are implemented in MrBayes. The selected substitution model schemes are provided in Table 2.
We performed parsimony JK in WinClada (Nixon 1999(Nixon -2002 with 2000 replicates and otherwise default settings. For maximum likelihood and ML bootstrapping we used RAxML 8 (Stamatakis 2014) implementing a general time reversible (GTR) model of nucleotide substitution with gamma distributed rate heterogeneity GTR+G (GTRGAMMA)following recommendations in the user manual. We used 4 parti-  tions determined by PartitionFinders (Table 2). 1000 bootstrap (BS) replicates were completed using the parametric (BS) algorithm of RAxML-HPC2 on the Cipres Web Portal (Miller et al. 2010). Bayesian phylogenetic analysis was inferred using MrBayes 3.2.5 (Huelsenbeck and Ronquist 2001;Ronquist and Huelsenbeck 2003;Ronquist et al. 2011) with the evolutionary models following the partitioning scheme from PartitionFinder (Table 2). We estimated posterior probabilities (PP) by running one cold and two heated chains for 2 130 000 generations in parallel mode, saving trees every 100 th generation. To test whether the Markov chain converged, we monitored the average standard deviation of split frequencies (ASDSF), which should fall below 0.01 when comparing two independent runs. We discarded the 25% of generations before the point where the ASDSF fell below 0.01 as burn-in. All remaining trees were summarized as a Bayesian 50% majority rule (MR) consensus tree with PP calculated for each clade.

Results and discussion
We produced 61 new sequences (Table 1) for the phylogenetic analyses (24 mtSSU, 15 MCM7, 22 b-tub) including 57 taxa and 1947 nucleotide positions (735 for mtSSU and 582 for MCM7 and 630 for b-tub) for the final matrix. The alignment contained 618 parsimony-informative characters (177 for mtSSU, 237 for MCM7 and 204 for b-tub). The most likely tree from the RAxML analysis is presented in Figure 1 with likelihood BS, Bayesian PP and parsimony JK support superimposed. The analyses resulted in a topology (Fig. 1) very similar to the results of Otálora et al. (2013aOtálora et al. ( , 2013b. Some of the backbone topology, however, has unfortunately no or low support. In Otálora et al. (2013b) Callome was the sister to Rostania, but in our study this relationship is not formed. All Rostania species are nested within Collemataceae, but Rostania in the sense of Otálora et al. (2013b) is non-monophyletic. Three species form a core group, which we here treat as Rostania s. str. Rostania s. str. is well supported and includes Rostania occultata ( Fig. 2A), R. ceranisca, and R. multipunctata. We can conclude that R. occultata as currently circumscribed is non-monophyletic. Rostania multipunctata (Fig. 2B) shares the cuboid shape and size of the spores with R. occultata s. lat. (Fig. 3A), but the thallus differs in size (the lobes are generally larger, up to ca 2.5 cm long in R. multipunctata, while in R. occultata s. lat. they are up to ca 3 mm long). It has also accessory lobules developing from the wrinkles (Fig 2B), which do not occur in R. occultata s. lat. The delimitation of the two varieties of R. occultata is unclear, as is the separation from R. multipunctata. Our study is not designed to study species-delimitations and we will extend our investigation of this species complex in a larger study currently in preparation.
Rostania ceranisca, the only terricolous Rostania, is sister to the group consisting of R. multipunctata and R. occultata s. lat. In addition to its terricolous ecology, it is easily recognized by the erect accessory finger-like lobules (Fig. 2C), which grow from the edge of the main lobes. The spores in R. ceranisca differ in shape from the cuboid spores in R. multipunctata and R. occultata s. lat. (Fig. 3A) in being oblong (Fig. 3B). Although Degelius (1954) noted only four spores in the ascus, we have usually observed eight spores, even if four of them may be aborted or are at least not clearly visible when mature (Fig. 3B).
Rostania callibotrys does not group with Rostania s. str. (Fig. 1), but forms an unsupported group with Enchylium. Rostania callibotrys has a comparatively distinct thalline apothecium margin, similar to some species of Enchylium. However, this is a widespread feature in the family including some species of Rostania s. str. The thallus with characteristic accessory lobules in R. multipunctata (Fig. 2B) and R. laevispora (Fig. 2D) is very similar to R. callibotrys (Fig. 2E). Rostania callibotrys also has spores that are very similar to the typical cuboid to oblong Rostania-spores in R. multipunctata and R. occultata s. lat. (Fig. 3A, B), but the spores in R. callibotrys have fewer cells (Fig. 3C) than in these species. Rostania laevispora (Fig. 2D), a rarely collected species that we did not manage to get sequences from, is very similar and likely very closely related to R. callibotrys (Fig. 2E). As there is no support for excluding these species, and no distinct morphological evidence suggests any other relationship, we tentatively leave both R. callibotrys and R. laevispora in Rostania.
We did not manage to get molecular data from R. coccophylla (Fig. 4A), a tropical and rarely collected species where the available material was too old. Although R. coccophylla is similar to R. callibotrys and R. multipunctata, the apothecia in R. coccophylla are very different in that they are convex and stipitate when mature (compared to concave and initially immersed and later sessile, in Rostania) and considerably larger compared to other Rostania species. The apothecia of R. coccophylla are similar to several species in Collema sensu Otálora (2013b), where this species originally was placed. Although we preferably would want molecular data to test the correct placement of this species, we suggest that it is re-instated in Collema, where the name Collema coccophyllum Nyl. is available.
Rostania quadrifida and R. paramensis are not closely related to Rostania s. str. Rostania quadrifida was described by Stone and McCune (2010) as Collema quadrifidum, and was later included in Rostania based on spore shape and thallus morphology (McCune et al. 2014). It differs from Rostania s. str. by having spores with fewer septa  ( Fig. 5A). Here it forms the sister group to Scytinium (Fig. 1), within a well-supported group consisting of Blennothallia, Lathagrium and Scytinium. Rostania quadrifida has a thallus composed by densely interwoven hyphae, and with a pseudocortex (Fig. 6A), features that do not occur in Rostania s. str., but in some species of Scytinium (similar to e.g. Scytinium intermedium and S. magnussonii ;Jørgensen 1994). These similarities support including it in Scytinium, which we do below.
The generic position of R. paramensis has been complicated to assess. Jørgensen and Palice (2012) described it as Collema paramense, based on the holotype (Palice 2796) and another sample from a second locality in Ecuador (Palice 2273). As the thallus has a pseudocortex, Otálora et al. (2013b) transferred it to Scytinium. Jørgensen and Palice (2015) later studied another sample from the second locality (Palice 2274). They concluded that the spores in the holotype must have been unusually developed, and transferred it to Rostania based on the oblong spores (similar to R. ceranisca) found in Palice 2274. Our re-examination of these three specimens, including the holotype, shows that Palice 2273 and Palice 2274 contain two distinct Collemataceae species (Fig. 4C, D). One of these (Fig. 4D), present in small amounts only in both samples, is identical with holotype of Collema paramense and is characterised by a matt dark olive thallus with a pseudocortex (Fig. 6B), and hyaline, muriform, ellipsoid spores with acute ends (Fig.5B). This is very different from the spores in Rostania, but typical for species in Leptogium s. str. (Fig. 5C). We sequenced the holotype, and we can conclude that among the Leptogium species we have sampled, it forms a group with Leptogium azureum (the conserved type of Leptogium; Jørgensen et al. 2013) and L. denticulatum (Fig. 1). It has a thallus which is appressed to the substrate and composed by relatively small lobes (Fig. 4B) which is rare in other Leptogium s. str., and in section it has straight and unbranched hyphae which are perpendicular to the surface (Degelius 1954 ; Fig. 6B). This character is present in several groups in Collemataceae. It was observed by Degelius (1954) in some Collema species, and has also been noted in the newly described Leptogium antarcticum by Kitaura et al. (2018) who used the term "columnar hyphae" for the same hyphal arrangement. We have observed this hyphal arrangement in Leptogium azureum (Fig. 6C) and L. denticulatum too, but it is apparently not present in Rostania. The second species present in Palice 2273 and Palice 2274, apparently confused Jørgensen and Palice (2015) as their observation of oblong spores (Fig. 5D, E) refer to this species, which has a shiny brown thallus (Fig. 4C) and not a matt dark olive thallus as in "Rostania" paramense (Fig. 4B). The second species differs from Rostania by having a proper eucortex (Fig. 6D), and by producing isidia along the apothecium margin (Fig. 4C). The thallus is paraplectenchymateous throughout (Fig. 6D). This hyphal arrangement is present in several groups in Collemataceae, including Rostania occultata s. lat. (Fig. 6E). Already Degelius (1954) noted this hyphal arrangement in his Occultatum-group and Otálora et al. (2013b) observed the same in Blennothallia, Pseudoleptogium and in Scytinium. We sequenced also this species and we can confirm that both samples belong in Scytinium, but the species remains to be identified.

Conclusions
Here we have tested the current generic concept of Rostania and conclude that at least three of the species should be excluded and that the position of R. callibotrys and R. laevispora in Rostania is uncertain. Rostania is characterized by crustose to subfoliose thallus with initially immersed apothecia (Fig. 2D, E), which only later become sessile. The disc is concave when young and plane when older, but never convex. The spores are muri- form with at least 5 cells, cuboid to oblong, but never fusiform to ellipsoid (Fig. 3A-C). Most species are comparatively small, and all lack cortex, rhizines and isidia.