Three new lichen species from Macaronesia belonging in Ramalinaceae, with the description of a new genus

Tylocliostomum is described as a new genus and T. viridifarinosum as a new species. Two Bacidina species, B. pallidocarpa and B. violacea, are also described as new to science. They all occur in Macaronesia (Azores, Madeira).

. Fifteen species of these genera are recorded in the checklist of the Azores, with none of them being endemic (Aptroot et al. 2010).
In this study we describe two new Bacidina species: B. violacea, known from the Azores and Madeira, and B. pallidocarpa, endemic to the Azores. Tylocliostomum is described as a new genus and T. viridifarinosum as a new species, known only from the type locality in Madeira.

Material acquisition
This study is mainly based on extensive collections of Bacidia s.l. made by the first author on several Macaronesian islands in 2017 and 2019 and hosted in LG and the private herbarium of the first author. Hand-cut apothecial sections and squashed thallus preparations were examined with a compound microscope (Olympus BX2).
Ascospores and other anatomical details were studied and measured in water or in 10% potassium hydroxide (K) if features were otherwise unseparated. In each collection (~5 well-developed ascospores representing the observed variation of size and shape), conidia and paraphyses were measured to 0.1 µm accuracy.
Chemical spot tests were performed under a compound microscope using sodium hypochlorite (C) and K (Orange et al. 2001). Pigments were determined following the system of Meyer & Printzen (2000). The crystals from all studied specimens were investigated in tissue sections by using a compound microscope with polarization filters. For a detailed comparison of the main diagnostic features of the two new Bacidina species and the most similar Bacidina species, see Table 1.

Molecular data
DNA was extracted from two specimens of Bacidina pallidocarpa and two specimens of B. violacea, and two DNA extractions were performed from the same material of Tylocliostomum viridifarinosum (Table S1). Extraction of DNA followed the protocol of Cubero et al. (1999). We sequenced the ribosomal nuclear loci ITS using primers ITS1F (Gardes & Bruns 1993) and ITS4 (White et al. 1990), the mitochondrial ribosomal locus mtSSU with primers SSU1 and SSU3R (Zoller et al. 1999), and part of the protein-coding gene RPB1 with RPB1AF (Stiller & Hall 1997) and VH6R (Hofstetter et al. 2007).
For the three loci, PCR conditions were as follows: initial denaturation at 95°C for 1 min, followed by 35 cycles of denaturation at 95°C for 45 sec, a different annealing phase for each locus, then an elongation phase of 1 min at 72°C, followed by a final elongation phase of 10 min at 72°C. For ITS the annealing phase was 52°C for 45 sec; for RPB1, 1 min at 56°C; and for mtSSU, 1 min at 58°C. PCR fragments were purified using EXO-SAP (VWR International, West Chester, PA, USA) and sequenced by Macrogen Europe (Amsterdam, The Netherlands). Sequence fragments were assembled using Geneious v. 10.2.3 (Biomatters LTD). Matrices were assembled in Mesquite v. 3.11 (Maddison & Maddison 2016), aligned using MAFFT v. 7.305b with default parameters (Katoh & Standley 2013) and further inspected by eye. Ambiguous regions were delimited manually and excluded from the analyses.

Phylogenetic analyses
We downloaded ITS, mtSSU and RPB1 alignments from Kistenich et al. (2018). We pruned them to keep  Kistenich et al. (2018), and favouring type species, then taxa with all or most of the three loci, except in clades where sequences exhibited high similarity with our sequences, where all representatives were kept. Catillaria contristans, Catillaria erysiboides, Glyphopeltis ligustica, Protoblastenia rupestris and Psora decipiens were selected as outgroups and other outgroup taxa were further removed (Table S1). We further selected additional sequences showing high similarity to our collections by performing BLASTn searches in the GenBank nucleotide collection (Wheeler et al. 2007), and added them to our dataset. In particular, numerous sequences from Switzerland (Mark et al. 2016) and Central and Eastern Europe (Czarnota & Guzow-Krzemińska 2012, 2018 were added. Single-locus matrices consisted of 100 sequences for ITS, 71 sequences for mtSSU, and 57 sequences for RPB1. The best ML tree was inferred for each locus using RAxML v. 8.2.9 (Stamatakis 2006), searching for the best ML tree and performing 1000 bootstrap pseudoreplicates in the same run, with the GTR-GAMMA model and default parameters. Congruence was examined by eye and no significant conflict between loci was observed. The data were therefore concatenated using the combine.pl script from the PLexus package (Magain 2018). The data were partitioned in 7 subsets as follows: ITS1, 5.8S, ITS2, mtSSU, and the three codon positions of RPB1. The best partitioning scheme was inferred using PartitionFinder2 (Lanfear et al. 2016) on the CIPRES portal (Miller et al. 2010) using the BIC criterion and the greedy algorithm. The three-locus concatenated alignment consisted of 115 taxa and 1850 characters. For the three-locus matrix, the best likelihood tree was inferred with RAxML v. 8.2.12 (Stamatakis 2006;Stamatakis et al. 2008) as implemented on the CIPRES portal with the default settings and the GTRGAMMA model, and bootstrapped in the same run, letting RAxML automatically stop the bootstrap, and using the best partition obtained with PartitionFinder. We further ran, for the same matrix, a Bayesian analysis using MrBayes v. 3.2.6 (Huelsenbeck & Ronquist 2001) on the CIPRES portal using the partition and the substitution models determined by PartitionFinder, running for 20 million generations and sampling every 1000 th generation. The analysis consisted of two runs of three heated chains and one cold one. We assessed the convergence using Tracer v. 1.6 (Rambaut & Drummond 2007). The first 25% of trees were discarded as burn-in. Branches of the phylogenetic trees were considered strongly supported if bootstrap support was higher than 70 for maximum likelihood analyses and posterior probabilities higher than 0.95 for the Bayesian analysis.

Molecular data
We obtained 3 ITS, 3 mtSSU and 4 RPB1 sequences. We obtained ITS and RPB1 sequences from one specimen of Bacidina violacea, whereas we could not obtain any sequences from the other specimen. We obtained mtSSU and RPB1 sequences from one B. pallidocarpa specimen, and only mtSSU for the other specimen. We obtained ITS and RPB1 sequences from the two DNA extractions of Tylocliostomum viridifarinosum (Table S1).

Phylogenetic analyses
The final 3-locus alignment consisted of 131 taxa and 1872 characters. The best partitioning scheme retrieved by PartitionFinder consists of a single set. The best substitution model recovered was GTR+I+G. Our best ML tree retrieves the monophyletic family Ramalinaceae as circumscribed by Kistenich et al. (2018), with high support [ML Bootstrap (BS) = 75, Bayesian posterior probabilities (pp) = 1; Fig. 4]. In both the ML and the Bayesian analyses, we recovered clades D (BS = 72, pp = 1), E (BS = 100, pp = 1) and G (BS = 85, pp = 1) sensu Kistenich et al. (2018) as monophyletic with high support, clade F (BS = 77, pp = 0.77) with high support in the ML analysis only, and clade C (BS = 65, pp = 0.90) as monophyletic without support. We recovered the same sister relationships of clades C and D without support (BS = 49, pp = 0.84) in the ML analysis, and clades C, D and E formed a well-supported group (BS = 79, pp = 1) in the Bayesian analysis. We did not recover a sister relationship between clades F and G. Instead, we recovered clade F as sister to clades C, D and E, highly supported in the Bayesian analysis (BS = 69, pp = 1).
The genus Bacidina appears paraphyletic in our analyses, as it appeared in Kistenich et al. (2018). Bacidina pallidocarpa and B. violacea belong to a well-supported clade corresponding to part of the genus Bacidina, labelled Bacidina clade II (Fig. 4, BS = 98, pp = 1) within clade D. The type species of the genus, B. phacodes, belongs to the other clade, labelled Bacidina clade I in Figure 4. Within Bacidina clade II, B. violacea is part of a highly supported group resulting from the first split within the genus (BS = 98, pp = 1), along with four accessions from Switzerland labelled Bacidina arnoldiana aggr. (Mark et al. 2016). However, the true Bacidina arnoldiana, including a representative from Kistenich et al. (2018) and one from the AFTOL fungal tree of life, belongs to another clade within Bacidina. The four accessions from Switzerland may belong to B. violacea but the morphology of these specimens could not be examined. Bacidina pallidocarpa appears as sister species of B. neosquamulosa (represented by two accessions from the Netherlands) with high support in the ML analysis (BS = 88, pp = 0.85). If the genus Bacidina was to be split in two, the names Woessia (type species corresponding to B. sulphurella) and Lichingoldia (type species corresponding to B. inundata) could accommodate Bacidina clade II, containing B. pallidocarpa and B. violacea.
Tylocliostomum viridifarinosum belongs to clade G, and more specifically forms a well-supported group with Tylothallia biformigera and two accessions of Cliostomum haematommatis (BS = 0.99, pp = 1). Cliostomum haematommatis accomodates the pycnidiate state of a species initially described as Lichenophoma haematommatis (Hawksworth et al. 2006). However, these two accessions likely do not represent the genus Cliostomum sensu stricto, as Cliostomum corrugatum, the type species of the genus, belongs to a different subclade within clade G.
Etymology. The epithet refers to the habitus of the apothecia, especially the always white to pale cream colour.
Distribution and ecology. Azores, known from four localities, where it grows always abundantly, on trunks of trees and on wood of fences. At the type locality no additional lichens have been found, not even in the collection (56937) from Pico.
Notes. The most similar species, Bacidina chloroticula, for which it can be mistaken in the field, differs by the whitish, pale pinkish, yellowish, or pale grey-brown apothecia, the scurfy greenish to pale brownish thallus, ± cracked to areolate, its much smaller apothecia, up to 0.2 mm diam, and ascospores somewhat wider (1-1.8 µm), 0-3-septate. For further main diagnostic features of similar Bacidina species, see Table 1.
The new species somewhat resembles Bacidia assulata s.l. because the apothecia are pale brick-red to very pale orange-brown, but the ascospores are much longer and 1.5-2 µm wide.
Etymology. The epithet refers to the colour of the apothecial margin.
Distribution and ecology. Azores and Madeira, known from two localities only, where it grows abundantly, on trunks of an unidentified tree (type locality) and branches of a Michelia doltsopa tree. No accompanying lichen species have been found.  Notes. It is most likely to be confused in the field with Bacidina caligans, but that species has a scurfy granular-sorediate thallus, the granules are somewhat wider (20-50 µm), the margin of the apothecia is darker than the disc, and it has white to pink immersed pycnidia. Pycnidia are unknown in the new species. A similar species, Bacidina sulphurella, for which it also can be mistaken in the field, differs by the mainly grey brown apothecia which have a raised margin and are often white pruinose; the excipulum has ellipsoid lumina ~ 3-7 × 1.5-5 µm and the ascospores are 1-3-septate.
No chemical compounds detected.
Etymology. The epithet refers to the habitus of the thallus, which is farinose and clearly green.  This new species is abundantly present in the type locality, growing on twigs of Erica shrubs. Cliomegalaria symmictoides and Endohyalina ericina, whose presence is revealed by many or few apothecia, respectively, share this habitat.
Notes. In the field, Tylocliostomum viridifarinosum is easily mistaken for Cliomegalaria symmictoides because the apothecia are cream-coloured and slightly convex and appressed, but the thallus differs, being areolate, not farinose, with a somewhat shiny upper surface, and the ascospores are thick-walled, 1-septate and much wider (4.5-6.5 µm). The habitus of the new genus somewhat resembles Cliostomum, especially if the ascomata are pale in the latter. The hyaline ascospores in Cliostomum griffithii are usually also 1-septate, but sometimes wider, reaching 3.5 µm in width versus 3.2 µm maximum in T. viridifarinosum. Cliostomum species have usually numerous, conspicuous black pycnidia, which seem to be absent in the new species. The relationship with the genus Tylothallia is unexpected, because T. biformigera, known from the temperate zone of the Northern Hemisphere, develops a thick rimose thallus with dark brown apothecia, on acidic rock (Smith et al. 2009).