Endless forms most frustrating: disentangling species boundaries in the Ramalina decipiens group (Lecanoromycetes, Ascomycota), with the description of six new species and a key to the group

Abstract

Oceanic islands have been recognized as natural laboratories in which to study a great variety of evolutionary processes. One such process is evolutionary radiations, the diversification of a single ancestor into a number of species that inhabit different environments and differ in the traits that allow them to exploit those environments. The factors that drive evolutionary radiations have been studied for decades in charismatic organisms such as birds or lizards, but are lacking in lichen-forming fungi, despite recent reports of some lineages showing diversification patterns congruent with radiation. Here we propose the Ramalina decipiens group as a model system in which to carry out such studies. This group is currently thought to be comprised of five saxicolous species, all of them endemic to the Macaronesian region (the Azores, Madeira, Selvagens, Canary and Cape Verde islands). Three species are single-island endemics (a rare geographic distribution pattern in lichens), whereas two are widespread and show extreme morphological variation. The latter are suspected to harbor unrecognized species-level lineages. In order to use the Ramalina decipiens group as a model system it is necessary to resolve the group's phylogeny and to clarify its species boundaries. In this study we attempt to do so following an integrative taxonomy approach. We constructed a phylogenetic tree based on six molecular markers, four of which are newly developed and generated competing species hypotheses based on molecular (species discovery strategies based on both single locus and multilocus datasets) and phenotypic data (unsupervised clustering algorithms based on morphology, secondary chemistry and geographic origin). We found that taxonomic diversity in the Ramalina decipiens group has been highly underestimated in previous studies. In consequence, we describe six new species, most of them single-island endemics and provide a key to the group. Phylogenetic relationships among species have been reconstructed with almost full support which, coupled with the endemic character of the group, makes it an excellent system for the study of island radiations in lichen-forming fungi.

Citation: Blázquez M, Pérez-Vargas I, Garrido-Benavent I, et al. 2024. Endless forms most frustrating: disentangling species boundaries in the Ramalina decipiens group (Lecanoromycetes, Ascomycota), with the description of six new species and a key to the group. Persoonia 52: 44–93. https://doi.org/10.3767/persoonia.2024.52.03 .

INTRODUCTION

Oceanic islands, those that never have been connected to a continental landmass as a consequence of their volcanic origin, have been regarded as simplified versions of the natural world and used as natural laboratories in which to carry out studies on evolutionary biology (Whittaker & Fernández-Palacios 2007, Warren et al. 2015). Ever since Darwin used the finches of the Galápagos Islands to support his theory of evolution by natural selection, under which “endless forms most beautiful and most wonderful have been, and are being evolved" ( Darwin 1882), island systems have been used to study speciation (Whittaker & Fernández-Palacios 2007). Speciation in oceanic islands can be anagenetic, when trait modification occurs within a given linage and does not give rise to a branching pattern ( Stuessy et al. 2006) or cladogenetic (i.e., radiations; Gould & Eldredge 1977), when a branching pattern does exist ( Simões et al. 2016).

The taxonomy of groups of organisms that have originated in recent radiations has been notoriously difficult to resolve, both by morphological ( Lack 1983) and molecular approaches ( Rokas & Carroll 2006, Zink & Vázquez-Miranda 2019). There are three main factors behind these difficulties. First, lineage accumulation under radiation is often fast ( Kim et al. 2008, Lim & Marshall 2017), which can result in incomplete lineage sorting ( Koblmüller et al. 2010). Second, phenotypic convergence is frequent among species of the same radiation (e.g., the repeated evolution of the same ecomorphs in the radiation of the Anolis lizards of the Greater Antilles driven by ecological features repeated at discrete geographic units, the islands; Losos 2009, Mahler et al. 2013). Third, barriers for gene flow are often weak and hybridization among species originated within the radiation can be common ( Salzburger et al. 2002, Palmer & Kronforst 2015). Particularly relevant to radiations taking place in islands is the ‘surfing syngameon hypothesis' ( Caujapé-Castells 2011) that posits that islands can act as melting pots of genetic variation. This occurs via multiple colonizations of closely related compatible species that generate syngameons (sensu Grant 1957, “the sum total of species or semispecies linked by frequent or occasional hybridization in nature").

Although island radiations have intrigued scientists since the 19th century no studies were carried out in this regard on lichen-forming fungi until quite recently ( Leavitt et al. 2011, Sérusiaux et al. 2011, Lücking et al. 2017b, Simon et al. 2018, Blázquez et al. 2022). This is likely a consequence of the ‘everything small is everywhere' paradigm ( Fontaneto 2011) that has been traditionally applied to these organisms. Lichen-forming fungi have been divided into four biogeographical regions at world level: pantropical, holartic, oceanian and subantartic/Australian ( Feuerer & Hawksworth 2007), with very few endemisms (e.g., Jørgensen 1977, Spjut 1996, Kashiwadani & Moon 2002). Recent studies are challenging this view, with new endemic species being discovered in oceanic islands ( Sérusiaux et al. 2011, Moncada et al. 2014, 2021, Dal Forno et al. 2017, Simon et al. 2018). One of the most striking examples of endemic species from oceanic islands can be found in the ascomycete lichenized fungal genus Ramalina (e.g., Krog & Østhagen 1978, Krog 1990).

Ramalina is a subcosmopolitan genus comprised by c. 230 species ( Lücking et al. 2017a) that typically forms more or less large, straw-colored, fruticose thalli. It belongs to the Ramalinaceae family, which is predominantly crustose ( Kistenich et al. 2018). It is one of the five fruticose genera of the Ramalinaceae, together with Cenozosia, Namibialina, Niebla and Vermilacinia ( Spjut et al. 2020). These genera appear in two separate lineages of the family. One of them is formed by Ramalina and Namibialina, while the other is formed by Niebla, Vermilacinia and microlichens of the genus Cliostomum s.str. Together, the two lineages are sister to Cliostomum griffithii and thus are nested within Cliostomum s.lat. ( Spjut et al. 2020). The phylogenetic placement of Cenozosia is still ambiguous, but it is suspected to cluster with Niebla and Vermilacinia ( Spjut et al. 2020). There are five geographic areas in which Ramalina shows great taxonomic diversity and endemism: East Africa ( Krog & Swinscow 1976), the Andes ( Marcano et al. 2021), Australasia ( Stevens 1987, Blanchon et al. 1996), Baja California ( Nash et al. 2002) and Macaronesia (Krog & Østhagen 1980a, b, Krog 1990). Macaronesia is a phytogeographical region formed by five volcanic archipelagos situated on the Atlantic Ocean (Azores, Madeira, Selvagens Islands, Canary Islands and Cape Verde) and a part of western Africa known as the Macaronesian Enclave ( Fernández-Palacios et al. 2011). The genus Ramalina in the region received great attention in the second half of the last century, especially at the hands of Norwegian lichenologists Hildur Krog and Haavard Østhagen ( Østhagen & Krog 1976, Krog & Østhagen 1978, 1980a, b, Krog 1990). Since then, two new species have been described from the Azores archipelago ( Aptroot & Schumm 2008), one from the Canary Islands (Pérez-Vargas & Pérez-Ortega 2014) and another occurring in the Azorean, Madeiran and Canarian archipelagos ( Spjut et al. 2020). So far, all of them are considered Macaronesian endemics.

Among the characters most commonly used for species differentiation in Ramalina, thallus anatomy is one of the most important. Krog & Østhagen (1980a) classified the Canarian Ramalina species in four anatomical types (farinacea, decipiens, bourgaeana and duriaei). They describe the genus as having a paraplectenchymatous cortex and prosoplectenchymatous ‘chondroid tissue', that appears as a cylinder underneath the cortex in some species or as strands embedded in the medulla or adjoining the cortex in others. They noted that the chondroid tissue is absent in some species. Roughly at the same time, Bowler (1981) distinguished Ramalina from other similar fruticose genera as having a two-layered cortex. The outer layer consists of a network of interwoven paraplectenchymatous hyphae with highly gelatinized cell walls and narrow lumina. The inner layer, in contrast, is formed by longitudinally aligned, periclinally disposed hyphae. Bowler (1981) refers to the latter layer as ‘supportive tissue' and it is most likely the chondroid cylinder described by Krog & Østhagen (1980a). He describes this tissue as prosoplectenchymatous (i.e., ‘periclinally arranged elongate cells'), but also notes that in many species, hyphae in this layer branch irregularly and not necessarily periclinally. He also notes that in some species this supportive tissue appears not as a uniform sheet adjoining the outer layer, but as an irregular system of ‘internal ribs' that often appear in contact with the external layer but that in a few saxicolous species can appear as ‘isolated rafts' in the medullary region. The anatomical types of Krog & Østhagen (1980a) were defined on the basis of the development and situation of these two layers. The farinacea type is characterized by a thin cortex, that can be indistinct or even absent and a continuous chondroid cylinder immediately underneath; the decipiens type is characterized by the presence of chondroid strands adjoining a well-developed cortex; the bourgaeana type is characterized by the presence of a well-developed cortex and chondroid strands embedded in the medulla; and the duriaei type is characterized by the lack of chondroid tissue. Engels (1995) in a pioneer study using isoenzyme patters reconstructed the phylogenetic relationships of many Macaronesian Ramalina endemics. She found that the two types with chondroid strands, the decipiens and bourgaeana types, formed separate clades for most enzymes. She argued that this supported the inclusion of the species with bourgaeana anatomy in Niebla and speculated about segregating those with decipiens anatomy in a new genus. Recent studies based on DNA data, however, clearly place both clades inside Ramalina ( Sérusiaux et al. 2010, Spjut et al. 2020) and refer to them as the Ramalina bourgaeana and R. decipiens groups.

As currently understood, the Ramalina decipiens group is com- prised by five species endemic to Macaronesia: Ramalina decipiens, R. erosa, R. hamulosa, R. maderensis and R. portosantana. There is also R. subwebbiana, a species that Krog & Østhagen (1980a) synonymized with R. decipiens but that has been treated as a separate species in more recent studies ( Aptroot & Schumm 2008, Sérusiaux et al. 2010, Van den Boom & Ertz 2012, Van den Boom et al. 2015, Spjut et al. 2020). Some of these species are restricted to one or two geographically close islands, while others are widespread. Ramalina erosa and R. portosantana are endemic to the island of Porto Santo in the Madeira archipelago ( Krog 1990, Sparrius et al. 2017). Ramalina hamulosa is known from El Hierro, La Gomera, Tenerife, Fuerteventura and Lanzarote in the Canary Islands (Krog & Østhagen 1980a, Hernández-Padrón & Pérez-Vargas 2010, Van den Boom & Ertz 2012). Ramalina maderensis is the species of the group with the largest distribution range. It is known from all the Canary Islands except La Gomera (Krog & Østhagen 1980a, Van den Boom 2015), the islands of Madeira and Porto Santo in the Madeira archipelago ( Motyka 1960, Sparrius et al. 2017) and the island of São Vicente in the Cape Verde archipelago ( Motyka 1960). It has also been reported from St. Helena ( Aptroot 2008), which makes it the only species of the group present in both hemispheres. Given their complex history it is difficult to distinguish the distributions of R. decipiens and R. subwebbiana. Krog & Østhagen (1980a), who regarded R. subwebbiana as a synonym of R. decipiens, reported the presence of the species in all the Canary Islands. Ramalina decipiens has also been reported from the island of Porto Santo in the Madeira archipelago ( Sparrius et al. 2017). As defined by Krog & Østhagen (1980a), R. decipiens and R. maderensis show extreme morphological variation and were suspected to be undergoing speciation, which prompted them to propose wide species concepts. In the last decade we have been studying the Macaronesian Ramalina and we have found that species within the R. decipiens group are rather difficult to differentiate. Also, we suspect that the widespread species could be concealing undescribed species-level lineages. For these reasons, a re-evaluation of species boundaries in the group following modern approaches was needed.

It has been argued that species delimitation in lichen-forming fungi should be carried out under the unified species concept ( De Queiroz 1999, Lücking et al. 2021), which defines species as segments of metapopulation lineages. The unified species concept allows taxonomists to choose different sources of evidence in species delimitation studies, not restricting delimitation to a single feature or property ( De Queiroz 2007). This approach results in more robust delimitations when multiple lines of evidence are considered ( Stegenga & Menon 2017). However, finding the right set of characters is a considerable challenge in and of itself (Lumbsch & Leavitt 2011). Compared to other groups of organisms, lichen-forming fungi display few taxonomically useful characters ( Printzen 2009). Species have traditionally been circumscribed on the basis of differences in morphology, chemistry and ecology ( Printzen 2009), but the different degrees of intraspecific variation in these characters can further confuse taxonomic circumscriptions. Over the last decades the incorporation of molecular data in species delimitation of fungal species has proven remarkably useful (Crespo & Pérez-Ortega 2009, Lumbsch & Leavitt 2011, Lücking et al. 2021). Molecular data can help the taxonomist detect putative species, assign individual specimens to those species, validate them as evolutionarily distinct lineages and infer the relationships between them ( Leavitt et al. 2015). However, molecular data alone is not enough to robustly circumscribe species ( Leavitt et al. 2015, Lücking et al. 2021); an integrative taxonomic framework ( Dayrat 2005, Goulding & Dayrat 2016) that brings together multiple and complementary lines of evidence (e.g., phylogenetics, comparative morphology, chemistry, bio-geography) is needed.

All in all, we consider that the R. decipiens group represents an excellent model system to study the process of evolutionary radiation in lichen-forming fungi. However, before further exploring the drivers of diversification in the group, an in-depth taxonomic revision is needed to clarify species boundaries and phylogenetic relationships among them. Thus, the objectives of the present study are: i) to test the current species delimitations in the R. decipiens group; ii) reconstruct the phylogenetic relationships among species using multilocus sequence data; and iii) revisit the taxonomy of the group using an integrative taxonomic approach.

MATERIAL AND METHODS

Sampling

Between 2010 and 2021 we collected more than 800 thalli of the R. decipiens group in a total of 76 localities from the Canarian, Madeira, Azores and Cape Verde archipelagos (distribution maps for all the species are available as Fig. S1–S15). Originally, we sampled the entire morphological variability of species monographed by Krog & Østhagen (1980a) as having decipiens anatomy, but given that R. polymorpha and R. krogiae did not cluster with the other species in previous phylogenies ( Sérusiaux et al. 2010, Spjut et al. 2020), they were not considered further in the present study. In addition, within the framework of a larger study of the phylogenetic relationships of the Macaronesian Ramalina, we found that R. nematodes and R. pluviariae, two species not studied in Sérusiaux et al. (2010) nor in Spjut et al. (2020), clustered within the R. decipiens group. Hence, specimens of these species were also included in the present study. In total, 306 specimens representing the morphological diversity of the group were selected for subsequent molecular studies (Table S1). Unless otherwise stated, all specimens are deposited in the herbarium of the Real Jardín Botánico-CSIC (MA-Lich).

In silico development of novel molecular markers

In this study we developed novel molecular markers from new genomes of four species of the genus: R. breviuscula, R. clementeana, R. fastigiata and R. rosacea (data not published). Handling of NGS data and marker development were carried out using the bioinformatic pipeline DOMINO ( Frías-López et al. 2016), a highly customizable tool designed for discovering molecular markers at different levels of taxonomic resolution. Briefly, DOMINO workflow consists of three main steps. First, DOMINO performed the pre-processing step of the raw reads to trim regions of low-quality and to clip adapters sequences. Second, the high-quality reads of the four species were independently mapped back to the genome of R. intermedia ( Wang et al. 2018). Then, DOMINO applied some ad hoc filters to reduce putative false nucleotide variants; more specifically: to exclude redundant sequences, remove sequences with statistical differences in coverage, with multi-mapping events and estimate the occurrences of polymorphic positions under a statistical framework. Third, DOMINO conducted the SNP calling using implemented Perl scripts ad hoc to translate the filtered data generated in the previous step (BAM and pileup files) in order to determine the variable regions flanked by conserved positions. Finally, DOMINO generated a list of candidate sequences and their corresponding FASTA file to allow developing PCR primers shared among the taxa of interest and other parameters selected by the user, such as as the minimum level of variation between taxa, the range length of the conserved and variable region, etc. Specifically, we used the parameters listed in Table S2.

Identification of novel molecular markers

To determine the presence of protein-coding regions in the discovered marker sequences, we conducted a BlastP search (v. 2.2.30; Altschul et al. 1997) using as query the set of predicted peptides of R. rosacea (data not published) with an E-value cutoff of 1e^-5. We performed a second BlastN (v. 2.2.30; Altschul et al. 1997) search against the genome of R. rosacea using as query the marker sequences with positive BlastP hits of the previous step and then we extracted the coordinates of the coding and non-coding regions listed in the GFF (General Feature Format) file. Furthermore, we applied InterProScan (v. 5.4.47; Jones et al. 2014) analysis against the predicted peptides of R. rosacea to characterize the functional annotation of the sequence markers with positive BlastP hits. Here we refer to the four markers as ‘unknown gene marker Ramalina' (UGMR): UGMR7_22, UGMR33_20, UGMR70_14 and UGMR197_101.

DNA extraction, PCR and sequencing

Thallus fragments of the apical zone (c. 10 mm²) free of epiphytic microalgae and fungi were selected for DNA extraction under a stereomicroscope (Nikon SMZ800). Samples were washed with acetone to remove secondary metabolites and stored at -80 °C. After one hour of freezing, they were pulverized using TissueLyser II (Qiagen) with two glass beads. Genomic DNA was extracted using E.Z.N.A.^® Forensic DNA Kit (Omega Bio-Tek), following the instructions of the manufacturer. We amplified six loci: the nuclear internal transcribed spacer (ITS), elongation factor 1-α (efa) and the four novel markers identified in the present study. ITS was amplified using primers ITS1F-KYO2 and ITS4-KYO2 ( Toju et al. 2012). For efa we designed new internal primers with Primer-BLAST ( Ye et al. 2012) to improve amplification efficiency. These and the primers used to amplify the four novel markers appear in Table 1. PCR reactions were carried out in a total volume of 15 μL, containing 2 (ITS) or 3 (remaining five markers) μL of template DNA, 0.5 (ITS) or 1 (remaining five markers) μL of each primer (10 μM), 6.5 μL of MyTaq™ Red Mix (Bioline) and 5.5 (ITS) or 3.5 (remaining five markers) μL of distilled water. The amplification programs were (1) initial denaturation at 95 °C for 2 min; 8 cycles of 94 °C for 1 min, 58 °C for 1 min (decreasing 0.5 °C each cycle) and 72 °C for 1 min 30 s; 28 cycles of 94 °C for 1 min, 52 °C for 1 min and 72 °C for 1 min 30 s; with a final extension at 72 °C for 7 min for ITS and (2) initial denaturation at 95 °C for 4 min; 6 cycles of 95 °C for 1 min, 62 °C for 1 min (decreasing 1 °C each cycle) and 72 °C for 1 min 30 s; 30 cycles of 95 °C for 1 min, 56 °C for 1 min and 72 °C for 1 min 30 s; with a final extension at 72 °C for 7 min for the remaining five markers. PCR products were run in 1 % agarose gels stained with SYBR™ Safe DNA Gel Stain (Thermo Fisher Scientific). DNA was sequenced in Macrogen Spain (Madrid, Spain) with the same primers used in the PCRs. Sequences were accessioned in GenBank (Table S1).

Table 1.

Primer sequences used to amplify efa and the four novel markers identified in the present study.

DNA region	Primer	Sequence (5' → 3')
efa	RamEFA2-For	AGACCCYTCATTTTCTGAAG
	RamEFA2-Rev	CTGACCRTCCTTGGAGATAC
UGMR7_22	UGMR7_22-For	CATTCCYGCCATCAACATC
	UGMR7_22-Rev	GGATAGATTCTCCTCRAAGTC
UGMR33_20	UGMR33_20-For	TAAGCTTGCCAACCCTACCA
	UGMR33_20-Rev	CCGAGRAGCAGCGCAAAYAC
UGMR70_14	UGMR70_14-For	GTAAGGCTGGCCCRGTATC
	UGMR70_14-Rev	ATGCATGAATAGTGCAAGAACC
UGMR197_101	UGMR197_101-For	CCRGCAATCAAATGGGTAGC
	UGMR197_101-Rev	GGTCACYGACTTCGAGACTAC

Sequence alignment and phylogenetic analysis

We performed multiple sequence alignment for each marker using MAFFT v. 7.308 ( Katoh et al. 2002) as implemented in Geneious^® v. 9.1.8. with the following parameters: the FFT-NS-I ×1000 algorithm, the 200PAM / k = 2 scoring matrix, a gap open penalty of 1.53 and an offset value of 0.123. The alignment process was carried out in nucleotide mode. Once aligned, we divided the ITS into three regions, namely ITS1, 5.8S and ITS2. Similarly, efa and the four novel markers were divided into their intronic and exonic regions. When exons contained gaps, which happened in all markers except UGMR70_14, the first UGMR197_101 exon and the second efa exon, care was taken to respect codon positions when gaps were inserted. Given the reported, Maximum Likelihood (ML) and Bayesian approaches were used for inferring phylogenetic relationships for each marker. Single-locus ML trees were calculated using RAxML ( Stamatakis 2014) and 1 000 rapid bootstrap pseudoreplicates were performed to evaluate nodal support. Nodes with bootstrap values equal or higher than 70 % were considered signifi-cant. Bayesian trees were calculated using MrBayes v. 3.2.7 ( Huelsenbeck & Ronquist 2001, Ronquist & Huelsenbeck 2003). Starting with a random tree, two simultaneous, parallel four-chain runs were executed over 1 × 10⁷ generations and sampled after every 1 000th step. The first 20 % of data was removed as burn-in. The 50 % majority-rule consensus tree was calculated from the remaining trees. Nodes with posterior probabilities equal or higher than 95 % were considered significant. ML and Bayesian single-locus phylo genies were used to check for topological incongruences. Lastly, we generated a multi-locus phylogenetic tree using BEAST v. 1.8.1 ( Suchard & Rambaut 2009, Drummond et al. 2012). We used PartitionFinder v. 1.1.1 ( Lanfear et al. 2012) to estimate optimal substitution models and partition schemes using the subdivisions of the six markers as input. We performed BEAST analyses using a lognormal uncorrelated relaxed clock for all partitions inferred by Partition-Finder. We tested the adequacy of the uncorrelated clock model for each marker using MEGA v. 5.2 ( Tamura et al. 2011) on the RAxML and MrBayes topologies. We used substitution models found in PartitionFinder but changing GTR for HKY to reduce model complexity, given previous problems of convergence due to analysis overparameterization. A concatenated RAxML phylogeny was calculated and used here as a starting tree. After Bayes Factors comparisons of different tree priors, this parameter was set to ‘Coalescent: Constant Size'. All clock priors were set to ‘Uniform', with a range between 0.1 and 30.0. The prior constant.popSize was set to ‘Uniform', with a range between 0.1 and 10.0. A total of 6 runs of 2 × 10⁸ generations each, sampling every 10 000 steps, were combined using Log-Combiner v. 1.8.1. ( https://beast.community/logcombiner ). The first 50 % trees of each run were discarded as burn-in. Given that the output file was too large for TreeAnnotator v. 1.8.1 ( https://beast.community/treeannotator ) to handle, we developed a custom R script to generate files with 10 000 randomly selected trees. We generated 10 such files and processed them with TreeAnnotator v. 1.8.1 to generate ten annotated maximum clade credibility trees. Tree parameters reported in this study are the result of averaging over these ten maximum clade credibility trees. We used Tracer v. 1.7.1 ( Rambaut et al. 2018) to check for convergence of chains. RAxML, MrBayes and BEAST inferences were carried out in CIPRES Science Gateway ( Miller et al. 2011). Based on the results of Spjut et al. (2020), we used R. bourgaeana (coll. num.: Pérez-Ortega 7582), R. maciformis (coll. num.: Pérez-Ortega 6393), R. lacera (coll. num.: Blázquez 55), R. hierrensis (coll. num.: Pérez-Ortega 5903) and R. pusilla (coll. num.: Pérez-Ortega 5744) as outgroup to root the phylogenetic trees.

Species discovery strategies based on single locus data-sets

We used five sources of information to build competing species hypotheses. These were based on molecular (species discovery strategies based on both single locus and multilocus datasets) and phenotypic data (unsupervised clustering algorithms based on morphology, secondary chemistry and geographic origin). We employed four species discovery algorithms to each individual marker. These were the Automatic Barcode Gap Discovery (ABGD; Puillandre et al. 2012), the Assemble Species by Automatic Partitioning (ASAP; Puillandre et al. 2021), the Generalized Mixed Yule Coalescent model (GMYC; Pons et al. 2006, Fujisawa & Barraclough 2013) and the Bayesian Poisson Tree Processes model (bPTP; Zhang et al. 2013). ABGD and ASAP are species discovery algorithms that calculate a pairwise genetic distance matrix and attempt to find the so called ‘barcode gap' between the small intraspecific genetic distances and the larger interspecific ones. Both algorithms generate multiple species hypotheses accompanied with a p value and, in the case of ASAP, a measure of the barcode gap width. ABGD was run at https://bioinfo.mnhn.fr/abi/public/abgd/ and ASAP at https://bioinfo.mnhn.fr/abi/public/asap/ . For both algorithms we used the K80 model to calculate pairwise genetic distances. The required transition to transversion ratios (TS/TV) were calculated in MEGA v. 5.2 ( Tamura et al. 2011). The remaining model parameters were left at default values, with the exception of P_max, which was set to 0.01 (see Puillandre et al. 2012). We used three values for the relative gap width (X), 0.1, 1.0 and 1.5, to assess the consistency of the inferred groups under varying gap width values. In contrast with ABGD and ASAP, GMYC and bPTP are species discovery algorithms that need a phylogenetic tree as input. They estimate the branching rates in different regions of the tree and try to identify which part follows a speciation model and which part follows a coalescent model. The species hypothesis is generated by the identification of a threshold that maximizes the transition between the two branching rates. GMYC was run using the gmyc function of the splits R package ( https://rdrr.io/rforge/splits/ ). Species hypotheses following single and multiple threshold models were computed. As input, we generated an ultrametric tree for each marker using BEAST v. 1.8.1 ( Drummond et al. 2012) as implemented in CIPRES Science Gateway ( Miller et al. 2011). Briefly, we partitioned the ITS into the ITS1, 5.8S and ITS2 regions and the remaining markers into their intronic and exonic regions and carried out the BEAST analysis using a lognormal uncorrelated relaxed clock, the GTR + I + G substitution model and a random starting tree. We selected the ‘Coalescent: Constant Size' Tree Prior, following Talavera et al. (2013) and having previously compared different tree priors with Bayes Factors. bPTP was run on https://species.h-its.org/ with 5 × 10 ⁵ MCMC generations and a burn-in of 30 %. We used the previously generated ML phylogenetic trees of each locus as input.

Species discovery strategies based on multilocus datasets

We followed two strategies to derive species hypotheses from the multilocus dataset: genetic clustering and the multispecies coalescent approach. We explored genetic clustering in our dataset as evidence of putative species based on the hypothesis that specimens belonging to a given species should be genetically closer to one another than to specimens belonging to other species. However, the speciation process is quantitative in nature and in evolutionary radiations it often resembles a continuum of divergence that has been referred to as the ‘speciation continuum' ( Nosil 2012). To take this into consideration, we employed the Hierarchical Bayesian Analysis of Population Structure (hierBAPS; Cheng et al. 2013) to cluster DNA sequence data and reveal nested population structure. The hierarchical output of hierBAPS provides a way to explore genetic clustering at multiple resolutions, an approach that can be used to identify putative species at the early stages of the ‘speciation continuum' of an island radiation, such as that of the Ramalina decipiens group. hierBAPS takes a multi-locus DNA alignment as input and transforms it into a single nucleotide polymorphism (SNP) matrix. We excluded in this analysis the 30 samples for which ITS was the only marker available. hierBAPS was run using the function hierBAPS of the rhierbaps R package ( Tonkin-Hill et al. 2018). We set it to cluster individuals in three hierarchical categories and run until it converged into a local optimum. Following the developer recommendation of choosing a n.pops value significantly larger than the expected number of clusters we set this parameter to 70. For the multi-species coalescent approach we used the STACEY ( Jones 2017) BEAST2 ( Bouckaert et al. 2014) package. STACEY is a Bayesian method that analyzes a large set of species trees to delimit species with no a priori assignments, taking evolutionary processes such as incomplete lineage sorting into account. This approach has been successfully applied to other taxa in the Ramalinaceae ( Spjut et al. 2020). The analysis was run in the cluster Drago (CSIC–Ministry of Science and Innovation), using the partition schemes and substitution models returned by PartitionFinder and lognormal uncorrelated relaxed clocks. The analysis was run for 1 × 10⁹ generations, sampling every 100 000 and with the collapse height parameter ε = 10^-4. We discarded the first 50 % of the trees as burn-in and passed the 5 000 remaining ones to SpeciesDelimitationAnalyzer ( Jones et al. 2015) to calculate the most likely number of clusters.

Morphological characterization

Specimens were examined with a Nikon SMZ800 dissecting microscope and a Zeiss Primo Star compound light microscope. We characterized morphological features previously reported in the literature as relevant for the group (Krog & Østhagen 1980a, Krog 1990, Pérez-Vargas & Pérez-Ortega 2014). We studied in detail, for subsequent analyses, 14 qualitative features and eight quantitative traits. Qualitative features were: 1) apothecia (presence / absence); 2) apothecia position (marginal / terminal / marginal and terminal); 3) black ring surrounding the hamathecium (presence / absence); 4) soralia (presence / absence); 5) pseudocyphellae (presence / absence); 6) pseudocyphellae position (marginal / laminar / laminar and marginal / laminar and marginal near the base / laminar near the base); 7) pycnidia (presence / absence); 8) pycnidia color (black / pale); 9) pycnidia position (laminar / marginal / laminar and marginal); 10) pycnidia tuberculate (yes / no); 11) contorted branches e (presence / absence); 12) fenestrations (presence / absence); 13) surface sheen (matt / shiny) and; 14) cortex (presence / absence). Quantitative traits were: 1) pseudocyphellae length; 2) apothecia diameter; 3) branch length; 4) branch width; 5) ascospore length; 6) ascospore width; 7) conidia length; and 8) conidia width. For quantitative traits, we measured five structures per thallus. Measured ascospores and conidia were observed from more than one apothecium and pycnidium, respectively. The number of measured specimens of each species is available in Table S3. Ten additional features were studied in up to eight specimens of each species to complement the species descriptions: anatomical type, cortex thickness, apothecia color, apothecia shape, presence of pruina in the apothecia, thalline margin, branching pattern of the paraphyses, paraphyses width, asci length and asci width. Quantitative traits are reported in the species descriptions as (smallest absolute measurement–) 10th percentile – 90th percentile (–largest absolute measurement).

Chemical characterization

We characterized lichen secondary chemistry by means of thin-layer chromatography (TLC) following Culberson (1972) and Orange et al. (2001). Briefly, we immersed thallus fragments in acetone to extract lichen substances. The acetone extracts were applied to TLC Silica gel 60F254 plates (Merck KGaA, Germany) and run in solvent system C ( Orange et al. 2001). Once they were completely dry, we applied 10 % sulfuric acid to the plates and placed them on a heated plate (120 °C). We examined the developed plates in a CN-6 darkroom cabinet (Vilber Lourmat Sté, France) under a wavelength of 365 nm.

Geographic categorization

We categorized the geographical origin of samples in four categories: Madeira archipelago, Cape Verde archipelago and western and eastern Canary Islands. The approach of dividing the Canarian archipelago has previously been taken in biogeographic studies ( Sanmartín et al. 2008) and it accounts for the age of the islands ( Ancochea et al. 2006) and their present-day climatic differences (Del Arco Aguilar & Rodríguez Delgado 2018).

Scanning electron microscopy

Representative branches from dried specimens were transversally cut with a razor and subsequently coated with gold in a SCD 004 Sputter Coater Balzers (Leica) and examined in a Hitachi S3000N scanning electron microscope at the Unidad Técnica de Apoyo a la Investigación of the Real Jardín Botánico-CSIC.

Light microscopy

Representative branches from dried specimens were fixed with FAA ( Johansen 1940) for 24 h and then dehydrated at room temperature in a graded series of ethanol, starting at 70 % and increasing to 96 % and 100 % for 30 min each step. The fixed and dehydrated samples were embedded in subsequent mixes of 2 : 1, 1 : 1 and 1 : 2 ethanol/LR White resin (Sigma-Aldrich, USA) for 30 min. Finally, the samples were embedded in pure resin and allowed to polymerize in the oven for 24 h at 60 °C. Semi-thin sections (4 μm) were made with a CUT 5062 rotary microtome from SLEE medical GmbH (Germany), stained with toluidine blue and examined under an Olympus BX-51 microscope.

Unsupervised clustering of morphological, chemical and geographic data

We combined the data on morphology, chemistry and geographic origin and used it to generate competing species hypotheses by cluster analysis methodology. The rationale behind this was to include a species discovery strategy that did not rely on molecular data. The analysis was performed in Python v. 3.8.12 using the numpy v. 1.22.3 ( Harris et al. 2020), pandas v. 1.4.2 ( McKinney 2010), matplotlib v. 3.5.1 ( Hunter 2007), seaborn v. 0.9 ( Hunter 2007), scikit-learn v. 1.1.2 ( Pedregosa et al. 2011) and sklearn-pandas v. 2.2.0 ( Wamanse & Patil 2022) libraries. The input dataset (Table S3) comprised 265 specimens and included 32 variables: the 22 morphological traits, the presence/absence of the eight secondary metabolites and two variables summarizing the geographic origin at the archipelago (splitting the Canarian archipelago in western and eastern islands) and island levels. Data pre-processing was performed using the sklearn.preprocessing module ( Wamanse & Patil 2022) and consisted of three steps. First, in order to better understand the structure of our dataset, an exploratory descriptive analysis consisting on the generation of histograms for the quantitative variables and barplots for the qualitative variables, was performed. Second, missing data resulting from the nested structure of some variables (e.g., data on apothecia position, apothecia diameter and length and width of ascospores could not be recorded for the specimens lacking apothecia) were imputed through Sklearn's SimpleImputer tool using the mean strategy for quantitative variables and the constant strategy for qualitative variables. Third, in order to prevent scale effects and improve clustering, quantitative features were normalized using the MinMaxScaler module ( Raju et al. 2020). Then, two different methodologies were used to generate competing species hypotheses: k-Means ( Ding & He 2004) and agglomerative hierarchical unsupervised clustering ( Murtagh & Contreras 2017). The k-Means clustering method works by assigning the number of centroids based on the number of clusters given (k). For the implementation of this method we used the module k-Means from sklearn.cluster using parameters by default. Prior to their inclusion in the k-Means algorithm, categorical data were transformed into numerical data using the OneHotEncode tool from sklearn.preprocessing ( Bisong 2019). Then, we used Principal Component Analysis (PCA) to reduce the dimensionality and correlation between variables. This was done using the sklearn. decomposition module. After performing the PCA, the results were finally passed for the unsupervised clustering K-means. The optimal number of clusters, which corresponds to the species hypothesis derived from the K-means method, was chosen through heuristics employing the elbow method and silhouette scores, which were calculated from k = 1 to k = 30. Further, to visualize patterns in the data we plotted the clusters assigned through k-means algorithm for each PCA using seaborn.pair-plot. The second unsupervised method, hierarchical agglomerative clustering, was performed using the scipy.cluster.hierarchy module. We ran this algorithm using two different input matrices. The first one was calculated from the pre-processed data using Gower's distances (Gower 1971) which allow to handle mixed data (qualitative and quantitative). The second was calculated following Folk et al. (2019). Briefly, we built a Jaccard distance matrix for the qualitative traits and a Euclidean distance matrix (divided by the maximum pairwise Euclidean distance found) for quantitative traits. These two matrices were then summed and the resulting matrix was divided by two. For both matrices we tested two different linkage methods: 1) average linkage; and 2) complete linkage (farthest neighbors). Finally, to visualize clustering results we plotted the assigned groups using scipy.cluster.hierarchy.dendogram tool. The optimal number of clusters for each matrix and linkage method was the one in which the distance between possible clusters was maximized.

Comparison of competing species hypotheses

We intended to evaluate the resulting competing species hypotheses (ABGD, ASAP, the first two hierarchical levels of hierBAPS, STACEY, K-meand and agglomerative clustering using Gower's distances under the average linkage method) using the Bayes factor delimitation approach (BFD, Grummer et al. 2014). For each of the competing species hypotheses, we performed a Bayesian reconstruction of the species tree using *BEAST ( Heled & Drummond 2010, Drummond et al. 2012). Due to previous chain convergence issues, we only included the 259 samples for which at least four markers were available. No outgroup was included. For each hypothesis we performed 6 *BEAST runs of 2 × 10⁸ generations each, sampling every 20 000 steps. All parameters were identical to those used for the BEAST multi-locus phylogeny except the species tree prior, that was set to ‘Species Tree: Yule Process' and the population size model, that was set to ‘Piecewise linear & constant root’. The priors species.popMean and species.yule.birthRate were set to ‘Uniform', with a range between 0.1 and 10.0. Marginal likelihood estimates (MLEs) were then calculated from the Bayesian posterior distributions using the Path Sampling (PS; Lartillot & Philippe 2006) and Stepping-Stone (SS; Xie et al. 2011) methods, using default settings. Chain convergence was checked in Tracer v. 1.7.1 ( Rambaut et al. 2018). We found a widespread lack of convergence, with nearly all runs having post-burn-in ESS < 200 in most parameters. We took three alternative approaches to resolve this lack of convergence: 1) we generated 24 additional runs with the same parameters and length for each hypothesis; 2) one run with identical parameters but with ten times more generations (2 × 10⁹); and 3) one run implementing Metropolis coupled MCMC ( Altekar et al. 2004) and Bayesian model averaging ( Wasserman 2000). For the last analysis, we used the BEAST2 ( Bouckaert et al. 2014) packages CoupledMCMC ( Müller & Bouckaert 2020) and bModelTest ( Bouckaert & Drummond 2017). The shorter runs were computed in the cluster Trueno (CSIC– Ministry of Science and Innovation).

We suspected that these convergence problems were inherent to the nature of the study system (i.e., a recent evolutionary radiation). It has been theorized that at the onset of a radiation, when the ecological space is being explored, speciation is fast. However, as the number of available niches decreases, it soon slows down. This pattern is known as ‘early burst’, and it has been associated with adaptive radiations by theoretical models ( Glor 2010, Landis et al. 2013). Also, species in evolutionary radiations often display introgression ( Salzburger et al. 2002, Palmer & Kronforst 2015) and can even constitute synga meons ( Caujapé-Castells 2011). Because of these reasons, it is possible that speciation in the R. decipiens group is not completely tree-like yet, which could be the source of the convergence problems we encountered ( Posada & Crandall 2002). We used phylogenetic split networks and the Pairwise Homoplasy Index (PHI) ( Bruen et al. 2006) to further explore this possibility. Phylogenetic networks depict incompatible phylogenetic signals present in the dataset as additional edges ( Huson et al. 2010), which makes them useful to visually assess the treelikeness of a DNA alignment ( Bryant & Moulton 2004, Kennedy et al. 2005, Kück et al. 2010). The PHI test has been widely used to detect recombination or introgression in various taxonomic groups, such as plants ( Joly & Bruneau 2006), lizards ( Pinho et al. 2008), birds ( Cabanne et al. 2008) and fungi ( Croll & Sanders 2009, Pino-Bodas et al. 2011, Garrido-Benavent et al. 2021). Both analyses were conducted in SplitsTree v. 4.10 ( Huson 1998): a neighbor-net was computed using the concatenated alignment and PHI tests were carried out for both the concatenated alignment and the individual alignments.

Polymorphism statistics and neutrality tests

We used the software DnaSP v. 6.12.01 to calculate DNA polymorphism statistics for each dataset. Four indices were calculated: nucleotide diversity (π, calculated using the Jukes and Cantor correction; Nei 1987, Lynch & Crease 1990), number of polymorphic sites (s), number of haplotypes (h) and haplotype diversity (Hd; Nei 1987). Additionally, we tested for deviations from neutrality with Tajima's D (Tajima 1989) and Fu's Fs (Fu 1997) statistics, also in DnaSP v. 6.12.01, using the number of segregating sites. The significance of both tests was assessed based on 10⁴ coalescent simulations.

RESULTS

Development of novel molecular markers

We obtained a total of 306.098.980 genomic raw reads across all four Ramalina specimens. After conducting the preprocessing step, the number of high-quality reads kept was 249 567 359. Of those reads, a total of 155 973 840 were mapped back to the genome of R. intermedia. Then, through the SNP calling procedure implemented in DOMINO, we identified 235 molecular markers shared across the five taxa. Eleven pairs of primers were preliminarily selected and successfully amplified. Four of them were selected for use in the present study based on their phylogenetic informativeness and amplification success.

Identification of novel molecular markers

BlastP searches against the protein sequence database of R. rosacea showed that all markers enclosed predicted coding regions for nuclear genes. We characterized the functional domains of two of them using InterProScan. UGMR7_22 exhibited the characteristics PFAM and IPR identifiers related to an YjeF-related protein, whereas UGMR197_101 might encode an ABC transporter. We extracted the gene coordinates using the genomic sequences and the structural annotation data (GFF file) of R. rosacea. We obtained that the full sequences of the UGMR7_22 and UGMR33_20 markers corresponded to a single exon. In contrast, only ~20 % of the sequence of UGMR70_14 was aligned to a coding region, where the marker region between positions 409 and 545 belonged to the first exon and the rest of the sequence to a non-coding region. We also identified that the marker sequence of UGMR197_101 encoded for two exons (positions 1–121 and 175–489) and one intronic region (122–174) (Fig. 1).

Fig. 1.

Position of the four novel markers in the aligned genomes. Protein coding regions are depicted in blue and non-coding regions in yellow.

Sequence data and phylogenetic analysis

In this study we generated 1 461 new sequences of six different regions (Table S1). Alignment lengths where 787 bp for the ITS, 655 bp for UGMR7_22, 607 bp for UGMR33_20, 621 bp for UGMR70_14, 552 bp for UGMR197_101 and 738 bp for efa. We inspected the single-marker trees (Fig. S16–S21) looking for incongruences among them. Incongruence was considered significant if supported clades of the different phylogenies contradicted each other ( Mason-Gamer & Kellogg 1996). Clades were considered to be supported if they had bootstrap values higher than 70 and posterior values higher than 0.95. Following this procedure, we found significant incongruences across all markers. The partition scheme and substitution models returned by PartitionFinder for the concatenated BEAST analysis are reported in Table 2. The adequacy of the uncorrelated clock model was corroborated for all markers using the RAxML and MrBayes topologies (Table S4). The resulting runs were topologically congruent but showed convergence issues, having post-burn-in ESS < 200 in the following parameters: posterior, prior, treeModel.rootHeight, tmrca(ingroup), constant.popSize, coalescent and various parameters related to the clocks and tree likelihoods. After combining the six runs most of these parameters reached ESS > 200, with the exception of the clock and tree likelihood parameters. We generated ten maximum clade credibility trees through random sampling of the tree space resulting from the post-burn-in combination of the six runs, which contained 60 000 trees. One of these maximum clade credibility trees is shown in Fig. 2, with support values being averaged across the ten maximum clade credibility trees. The 10 trees were almost topologically identical, differing only in the placement of the undescribed Ramalina sp. 1. In some of the trees this species appeared as the sister species of R. decipiens and in others as the sister species of the clade formed by R. fortunatarum, R. maderensis and R. portosantana, in both cases without support. The trees showed an unsupported relationship between R. nematodes and R. pluviariae.

Table 2.

Partition scheme and substitution models returned by Partition-Finder for the concatenated BEAST analysis.

Partition	Components	Substitution model
1	ITS1	GTR + I + G
	UGMR7_22 1st
	UGMR7_22 2nd
	UGMR7_22 3rd
	UGMR197_101 (exonic) 2nd
2	5.8S	GTR + G
3	ITS2	GTR + G
	UGMR33_20 1st
	UGMR33_20 2nd
	UGMR33_20 3rd
	UGMR70_14 (exonic) 1st
	UGMR70_14 (exonic) 2nd
	UGMR70_14 (exonic) 3rd
	UGMR197_101 (exonic) 1st
	UGMR197_101 (exonic) 3rd
4	UGMR70_14 (intronic)	GTR + G
5	UGMR197_101 (intronic)	GTR + I + G
	efa (exonic) 1st
	efa (exonic) 3rd
	efa (intronic)
6	efa (exonic) 2nd	GTR + I + G

Fig. 2.

Maximum clade credibility tree from the BEAST analysis. Branches with posterior probability ≥ 0.95 are highlighted in bold. The table on the right side provides further information for all accessions: chemical characterization by TLC (column 1), geographic origin (column 2) and the more biologically sensible species hypotheses returned by ABGD (UGMR197_101, column 3), ASAP (UGMR197_101, column 4), GMYC (ITS, column 5), bPTP (ITS, column 6), hier-BAPS I (column 7), hierBAPS II (column 8), STACEY (column 9), K-means (column 10) and hierarchical agglomerative clustering (based on Gower’s distances using average linkage, column 11). The meaning of the color schemes of the first two columns is explained at the bottom left of the figure. Color schemes of the remaining colums bear no meaning and have been selected to facilitate visualization. Relationships among the outgroup species are not displayed.

Species discovery strategies based on single locus data-sets

ABGD returned between 3 (UGMR7_22) and 62 (UGMR33_20) putative species. No distinct barcode gap could be observed in the distance histograms of most markers, the only exceptions being UGMR7_22 and UGMR197_101. ABGD retrieved the most biologically sensible partition under UGMR197_101, consisting in 17 putative species. Similar results were obtained in the ASAP analysis, which grouped the specimens in between 11 (ITS) and 154 (efa) putative species, showing a clear barcode gap only in UGMR7_22 and UGMR197_101. The most biologically sensible partition under ASAP was returned by UGMR197_101, consisting in 19 putative species. GMYC returned between 2 (UGMR7_22) and 112 (UGMR197_101) putative species under the single threshold method and between 58 (efa) and 129 (ITS) under the multiple threshold method. The most biologically sensible partition was found under ITS and the single threshold model, consisting in 17 putative species. The inclusion of outgroup sequences did not modify the number of inferred putative species. bPTP returned between one (UGMR33_20) and 36 (efa) putative species. bPTP retrieved the most biologically sensible partition under ITS, consisting in 19 putative species. The inclusion of outgroup sequences did not modify the number of inferred putative species. The most biologically sensible hypotheses returned by the four methods are depicted on Fig. 2. The remaining ones are summarized on (Table S5).

Species discovery strategies based on multilocus datasets

The first hierBAPS hierarchical category consisted of seven genetic clusters. These were subdivided into 21 clusters in the second hierarchical category and 56 in the third. Cluster assignment probabilities of the studied Ramalina specimens ranged from 0.50 to 1.00 in the first hierarchical category, from 0.53 to 1.00 in the second and from 0.69 to 1.00 in the third. Mean cluster assignment probabilities were 0.99, 0.98 and 0.99, respectively. STACEY returned 26 putative species. Interestingly, specimens from one putative species had zero posterior probability of belonging to another one, except in the clusters belonging to the two widespread species (R. decipiens and R. maderensis) and, to a lesser extent, to the recently diverged R. papyracea, R. sabinosana and Ramalina sp. 2 (Fig. 2).

Chemical characterization

Five major lichen substances were detected by TLC (Fig. 3): 4-O-demethylbarbatic acid, divaricatic acid, lecanoric acid, protocetraric acid and salazinic acid. Usnic acid probably occurs in all the species, but sometimes its concentration is too low to be detected by TLC (Krog & Østhagen 1980a). Salazinic acid was the most common, being present in all species except R. delicata, R. erosa, R. fortunatarum and R. sampaioana. 4-O-demethylbarbatic acid was the rarest, being only present in R. fortunatarum and a R. maderensis chemotype restricted to the island of São Vicente. Additionally, three different triterpene patterns were detected in the species containing divaricatic acid.

Fig. 3.

TLC chromatogram representation of acetone extracts of the species of the Ramalina decipiens group, after being run with solvent C and revealed applying 10 % sulfuric acid and heating. Roman numbers after the species names represent different chemotypes. Control contains atranorin and norstictic acid. a. Normal light; b. UV light (365 nm). Secondary metabolites depicted are atranorin (At), 4-O-demethylbarbatic acid (dem), divaricatic acid (Div), lecanoric acid (Lec), norstictic acid (Nor), protocetraric acid (Prot), salazinic acid (Sal) and three different triterpene patterns (visible below divaricatic acid under UV light).

Geographic categorization

Most of the species (46 %) were restricted to the Canary Islands, followed by the Madeira archipelago (26 %). The only species present in the Azores and Cape Verde archipelagos was R. maderensis which, together with R. decipiens and Rama linasp. 2, were the only species occurring in more than one archipelago. Geographic origin at the island level was revealed to be of great taxonomic value, as many of the species were single-island endemics (Fig. 2).

Unsupervised clustering of morphological, chemical and geographic data

The exploratory analysis revealed some outliers in the distributions of the quantitative variables. These outliers were not removed since they corresponded to the actual morphological diversity present in the studied organisms (e.g., considerably larger branch length in the R. nematodes samples). After normalization, pseudocyphellae length and ascospore length were the quantitative variables with the lowest variability (sd = 0.07). The first four axes of the PCA accounted for 51.9 % of total explained variation, reaching 77.47 % within the first ten axes. The number of clusters (k) according to the elbow method and the Silhouette score was ten; and the maximum Silhouette score (0.43) was reached at k = 15. Overlap could be observed among most of the clusters. Although some groups could be visually distinguished in some pair combinations of principal components, this separation was not maintained for the rest of the combinations. The agglomerative method using Gower's distances found that the most suitable number of clusters was six both by average linkage and by complete linkage, although the putative species of the two hypotheses comprised different specimens. When the distance matrix of Folk et al. (2019) was used the most suitable number of clusters was five under the average linkage method and eleven under the complete linkage method. Putative species in all the species hypotheses derived from morphological, chemical and geographic data were not monophyletic (Fig. 2).

Comparison of competing species hypotheses

We took three complementary approaches to resolve the lack of convergence of the *BEAST analyses. First, we generated 30 short (2 × 10⁸ generations) *BEAST runs for each competing species hypothesis, i.e., ABGD, ASAP, the first two hierarchical levels of hierBAPS, STACEY, K-meand and agglomerative clustering using Gower's distances under the average linkage method. We found a generalized lack of convergence, with c. 90 % of all the runs having post-burn-in ESS < 200 in most parameters. The remaining runs also showed convergence issues, although most were related to clock prior parameters. In addition, the parameters with ESS > 200 reached different local maxima. Therefore, we could not combine them and the averaging of MLEs and Bayes factors calculation were not carried out. On the second approach we generated a longer (2 × 10⁹ generations) run for each hypothesis. This did not improve ESS values compared with the previous approach, with all runs showing convergence issues in nearly all parameters. On the third approach, we generated one run implementing Metropolis coupled MCMC and Bayesian model averaging for each hypo thesis. This, however, did not resolve the convergence problems either. The phylogenetic split network and the PHI tests suggest that these problems stem from speciation in the R. decipiens group not being completely tree-like yet. The result from the neighbor-net analysis based on the multilocus dataset is depicted in Fig. 4. The network clusters were mostly congruent with the supported clades of the BEAST tree (Fig. 2), the main difference being R. portosantana, that appeared nested inside R. decipiens. However, the relationships among them, particularly in the inner part of the network, are represented by parallel edges, indicating alternative connections and thus uncertainty. These results are congruent with those of the PHI tests, which found significant recombination in the entire dataset, UGMR7_22 and efa (p < 0.001), although not in ITS (p = 0.372), UGMR33_20 (p = 0.615), UGMR70_14 (p = 0.440) and UGMR197_101 (p = 1.000). As a compromise solution, we cautiously based species boundaries on the supported clades of the phylogenetic tree and circumscribed them on the basis of the morphological, chemical and geographic data. We understand that this is not the desirable solution under an integrative taxonomy approach, but given the thorough characterization of the studied material we propose 15 species-level lineages in the R. decipiens group, seven of which correspond to previously known species ( Østhagen & Krog 1976, Krog & Østhagen 1980a, b, Krog 1990), six are newly described and two will remain undescribed until more material is available.

Fig. 4.

SplitsTree Neighbor-net network of the Ramalina decipiens group based on the multilocus dataset, with main clusters color-labelled by species.

Polymorphism statistics and neutrality tests

The combined matrix consisted of 3 960 bp. efa was the marker with the greatest number of polymorphic sites. π ranged between 0.0179 in UGMR33_20 and 0.06086 in UGMR7_22, while Hd ranged between 0.932 in UGMR7_22 and 0.9892 in efa. Values for the remaining statistics are summarized in Table 3. We did not find deviations from neutrality in any of the datasets or species according to Tajima's D. However, Fu's Fs was signifi-cant for the whole ITS dataset (Table 3) and five species: R. hamulosa (efa), R. gomerana (ITS, UGMR7_22, UGMR70_14 and UGMR197_101), R. portosantana (UGMR33_20), R. sabinosana (UGMR33_20 and UGMR197_101) and R. sp. 1 (ITS, UGMR7_22 and UGMR197_101) (Table S6). Aside from R. sabinosana, most of these results seem to be caused by the low number of samples of those species in the datasets.

Table 3.

Polymorphism statistics and neutrality tests results for each marker.

Marker	n	bp	Gaps/missing	s	h	Hd	π (JC)	Tajima's D	Fu's Fs
ITS 306	754	466	75	71	0.953	0.03066	−0.07553 (ns)	−1.20189 (**)
efa 188	737	348	131	120	0.9892	0.05687	−0.09397 (ns)	−1.64174 (ns)
UGMR7_22	241	654	443	65	47	0.932	0.06086	−0.10144 (ns)	−1.22052 (ns)
UGMR33_20	217	607	262	73	65	0.965	0.0179	−0.10005 (ns)	−1.17429 (ns)
UGMR70_14	243	611	368	85	99	0.979	0.03725	−0.11191 (ns)	−1.46822 (ns)
UGMR197_101	244	552	223	85	77	0.967	0.02627	−0.10312 (ns)	−1.30317 (ns)

Note: number of sequences (n), alignment length (bp), number of gaps and missing data, number of polymorphic sites (s), number of haplotypes (h), haplotype diversity (Hd), nucleotide diversity calculated using the Jukes and Cantor correction (π), Tajima’s D and Fu’s F. Statistical significance of Tajima’s D and Fu’s Fs is depicted as: ns > 0.05;

* < 0.05;

** < 0.01;

*** < 0.001.

TAXONOMY

Ramalina decipiens Mont., in Webb & Berthelot, Hist. Nat. Îles Canar. (Paris) 3(2): 100. 1840 — Fig. 4–7

Fig. 7.

Ramalina decipiens (coll. num.: Pérez-Ortega 6121). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Synonyms. Ramalina scopulorum subsp. decipiens (Mont.) Nyl. (Published as R. scopulorum * R. decipiens), Bull. Soc. Linn. Normandie, sér. 3 4: 157. 1870.

Ramalina scopulorum subsp. subwebbiana Nyl. (published as R. scopulorum ** R. subwebbiana), Bull. Soc. Linn. Normandie, sér. 3 4: 158. 1870.

Ramalina subwebbiana (Nyl.) Hue, Nouv. Arch. Mus. Hist. Nat., Paris, 3 sér. 2: 267. 1890.

Ramalina subdecipiens Steiner, Öst. Bot. Z. 54(10): 353. 1904.

Typus. In Canariin, ex herb. Webbianum, sine coll. (Original material BM000890348!).

Thallus saxicolous, rigid, erect or subpendulous, sparingly to richly branched. Laciniae stramineous, green-grey or pale yellow-brown, usually complanate, in some specimens canaliculated, often shiny but sometimes matt, (10–)19–61(–117) × 1–4(–11) mm, sometimes contorted, dichotomously branching, arising from a single holdfast. Pseudocyphellae frequently present, laminar, usually near the base, 1–4(–5) mm in length. Fenestrations present in some specimens. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed, (28–)45–73(–95) μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia mostly marginal, in some thalli subterminal, 1–7(–12) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, occasionally showing a black ring, circular, crenate or with radial cracks. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, (32–) 33–42(–43) × (6–)8–10 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (8–)10–12(–15) × 4–5 µm. Pycnidia marginal, at times also laminar, sometimes tuberculated, with black or pale ostioles. Conidia bacilliform, (3–)4–5(–6) × 1–2 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution — Ramalina decipiens is present in all the Canary Islands (Krog & Østhagen 1980a) as well as in Porto Santo in the Madeira archipelago ( Sparrius et al. 2017). We have collected it between altitudes of 165 and 1 043 m, in localities with frequent fogs. It grows on all types of rocky substrates, from vertical cliffs to volcanic scoria. Ramalina decipiens has a broad ecological niche. It can be found growing with all the other species of the R. decipiens group; as well as other Ramalina species: R. bourgaeana, R. confertula, R. crispatula, R. cupularis, R. deminuta, R. hierrensis (which we have found growing on rock in Lanzarote), R. jamesii, R. krogiae, R. nodosa, R. parva, R. pitardii, R. subfarinacea, R. timdaliana, R. tortuosa and R. webbii.

Additional specimens examined. Portugal, Madeira archipelago, Madeira, Ponta de São Lourenço, N32°44'57" W16°41'36", on volcanic rock, 83 m alt., 6 Aug. 2018, S. Pérez-Ortega 7260, 7266; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6978. – Spain, Canary Islands, El Hierro, Guarazoca, alrededores del mirador de la Peña, N27°48'27" W17°58'48", on volcanic rock, 648 m alt., 29 July 2011, S. Pérez-Ortega 5814, 5819, 5820, 5824, 5830 & M. Arróniz Crespo; Canary Islands, El Hierro, Guarazoca, alrededores del mirador de la Peña, N27°48'27" W17°58'48", on volcanic rock, 648 m alt., 3 Nov. 2021, I. Pérez-Vargas & M. Blázquez 350, 351, 352, 353, 354; Canary Islands, El Hierro, road margin in Sabinar de la Dehesa, N27°44'48" W18°07'35", on volcanic rock, 605 m alt., 16 Mar. 2018, S. Pérez-Ortega. 5951, 5954, 6066, 6072, M. Blázquez 356 & I. Pérez-Vargas (MA-Lichen); Canary Islands, Fuerteventura, Malpaís de la Arena, N28°37'50" W13°55'28", on volcanic rock, 310 m alt., 19 Mar. 2018, S. Pérez-Ortega 6402, 6407, 6429 & I. Pérez-Vargas; Canary Islands, Fuerteventura, Montaña de la Muda, N28°38'44" W13°57'32", on volcanic rock, 166 m alt., 19 Mar. 2018, S. Pérez-Ortega 6455, 6457, 6463, 6483, 6484, 6491, 6497, 6513 & I. Pérez-Vargas; Canary Islands, Fuerteventura, Pico del Aceitunal, N28°32'50" W13°57'24", on volcanic rock, 650 m alt., 20 Mar. 2018, S. Pérez-Ortega 6546, 6556, 6558; Canary Islands, La Gomera, Amalahuigue reservoir, N28°11'21" W17°59'05", on volcanic rock, 510 m alt., 1 Nov. 2018, S. Pérez-Ortega 7617, 7710, 7717, 7728, 7733, 7742, 7783, 7796, 7799, 7800 & I. Pérez-Vargas; Canary Islands, La Gomera, mirador de El Santo, N28°07'46" W17°19'30", on volcanic rock, 829 m alt., 1 Nov. 2018, S. Pérez-Ortega 7594, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, near road GM-1, N28°11'26" W17°13'44", on volcanic rock, 659 m alt., 1 Nov. 2018, S. Pérez-Ortega 7553, 7574, 7575, 7576, 7578, 7579, 7580, 7586, 7638, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, road between Vallehermoso and Epina, N28°09'41" W17°17'16", on volcanic rock, 696 m alt., 1 Nov. 2018, S. Pérez-Ortega 7954, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, Road margin in Alajeró, N28°04'29" W17°15'52", on volcanic rock, 924 m alt., 1 Nov. 2018, S. Pérez-Ortega 8347, 8350, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, Roquedos al sur de Imada, N28°04'31" W17°14'13", on volcanic rock, 1 042 m alt., 2 Nov. 2018, S. Pérez-Ortega 7477, 7498, 7803, 7806, 7847, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, Roquedos y paredones en el risco de Heredia, N28°07'52" W17°19'23", on volcanic rock, 811 m alt., 1 Nov. 2018, S. Pérez-Ortega 7644, 7649, 7669, 8114, 8122, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Palma, volcán San Antonio, N28°29'10" W17°51'02", on volcanic rock, 574 m alt., 3 Nov. 2018, M. Blázquez 158, 165, 167, 168, 170, 171, 172, 174, 228 & I. Pérez-Vargas; Canary Islands, La Palma, Volcán Teneguía, N28°28'38" W17°51'00", on volcanic rock, 391 m alt., 3 Nov. 2018, M. Blázquez 59, 60, 61, 63, 65 & I. Pérez-Vargas; Canary Islands, Lanzarote, Caldero Riscado, N28°55'55" W13°45'19", on volcanic rock, 372 m alt., 2 Nov. 2021, M. Blázquez 308, 309 & I. Pérez-Vargas; Canary Islands, Lanzarote, malpaís near bodegas El Grifo, N29°00'04" W13°38'36", on volcanic rock, 312 m alt., 17 Mar. 2018, S. Pérez-Ortega 6317, 6321 & I. Pérez-Vargas; Canary Islands, Lanzarote, Peñas del Chache, N29°07'16" W13°30'59", on volcanic rock, 611 m alt., 17 Mar. 2018, S. Pérez-Ortega 6107, 6111, 6113, 6116, 6121, 6123, 6124, 6144, 6156, 6173 & I. Pérez-Vargas; Canary Islands, Lanzarote, Peñas del Chache, N29°07'16" W13°30'59", on volcanic rock, 611 m alt., 1 Nov. 2021, I. Pérez-Vargas & M. Blázquez 325, 326, 327; Canary Islands, Lanzarote, Volcán de la Corona summit, N29°11'5" W13°28'53", on volcanic rock, 520 m alt., 18 Mar. 2018, S. Pérez-Ortega 6202, 6223, 6263 & I. Pérez-Vargas (MA-Lichen); Canary Islands, Lanzarote, Volcán de la Corona summit, N29°11'5" W13°28'53", on volcanic rock, 520 m alt., 1 Nov. 2021, I. Pérez-Vargas 323 & M. Blázquez; Canary Islands, Tenerife, Mirador Altos de Baracán, N28°19'42" W16°51'21", on volcanic rock, 808 m alt., 30 Oct. 2018, S. Pérez-Ortega 7996, 8000, 8006, 8018, 8027, 8037, 8069, 8073, 8149, M. Blázquez 342, 343 & I. Pérez-Vargas; Canary Islands, Tenerife, mirador de Aguaide, N28°33'53" W16°17'41", on volcanic rock, 593 m alt., 29 Oct. 2018, S. Pérez-Ortega 7924, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, montaña de Taco, N28°22'27" W16°50'07", on volcanic rock, 295 m alt., 23 Oct. 2021, M. Blázquez 329, 330, 331, 333, 334 & I. Pérez-Vargas; Canary Islands, Tenerife, recreative area near road TF-82, N28°21'38" W16°45'58", on volcanic rock, 556 m alt., 31 Oct. 2018, S. Pérez-Ortega 7694, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, risco ‘El Mogote' on the road to Taganana, N28°33'29" W16°12'20", on volcanic rock, 427 m alt., 29 Oct. 2018, S. Pérez-Ortega 7455, 7457, 7473, 7491, 7902, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, riscos de Chinamada, N28°33'49" W16°17'45", on volcanic rock, 534 m alt., 29 Oct. 2018, S. Pérez-Ortega 7734, 7812, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, Roque Vitor, N28°20'50" W16°50'53", on volcanic rock, 528 m alt., 30 Oct. 2018, S. Pérez-Ortega 7461, 7993, 8126, 8127, 8143, 8147, 8162, M. Blázquez 337, 339 & I. Pérez-Vargas.

Notes — Ramalina decipiens is the sister species of the clade formed by R. fortunatarum, R. maderensis and R. portosantana (Fig. 2). It is a widespread species that can be distinguished from the other species of the R. decipiens group in a species-by-species basis. Pycnidia color, surface sheen, the presence of a black ring on the apothecia and the presence of salazinic acid are generally the most taxonomically useful characters. As previously reported by Krog & Østhagen (1980a) R. decipiens shows extreme morphological variation. Two contrasting morphotypes can be distinguished in R. decipiens, although all manner of intermediate forms between them can be found. The occurrence of those intermediate forms and the lack of phylogenetic structure within the species prevented us to recognize them at the species level. According to Krog & Østhagen (1980a), Montagne's original material in the Natural History Museum (BM000890348, digitized in https://plants.jstor.org/ ) corresponds with a morphotype showing pycnidia with pale ostioles and matt, broad branches (Fig. 5). As noted by Krog & Østhagen (1980a), we found that this morphotype is uncommon and restricted to Lanzarote and Fuerteventura. In our phylogeny, specimens belonging to this morphotype showed up in more than one subclade inside R. decipiens. The remaining specimens correspond with the morphologies of R. subwebbiana (thallus shiny, pycnidia with black ostioles) and R. subdecipiens (thallus shiny, pycnidia tuberculated with black ostioles, apothecia with a black ring around the hamathecium). Again, these specimens appear in more than one of the species' subclades. For this reason, we agree with Krog & Østhagen (1980a) in considering R. subwebbiana and R. subdecipiens as R. decipiens synonyms. This morphotype is characterized by having pycnidia with black ostioles which may in some cases be tuberculated, flat and shiny branches with straight margins and by showing a black ring on their apothecia (Fig. 6). In Porto Santo we have found two thalli (TFC-Lich 14961 and TFC-Lich 14965) that cluster inside R. decipiens but are morphologically indistinguishable from R. portosantana. Most likely these samples correspond to R. portosantana and their phylogenetic placement inside R. decipiens is an artifact caused by incomplete lineage sorting or hybridization in the ITS, a marker that recovered not less than eight supported R. decipiens subclades, one of them corresponding to R. portosantana. The ITS was the only marker available for TFC-Lich 14965 (sample IPV297 on Fig. S16) and one of two available for TFC-Lich 14961 (alongside UGMR70_14, sample IPV293 on Fig. S16, S19). However, further studies should be carried out to test this hypothesis. Krog & Østhagen (1980a) reported the existence of two R. decipiens chemotypes. One, including Montagne's original material (BM000890348), containing salazinic acid and an other containing protocetraric acid. In our phylogeny, the specimens showing the latter chemotype clustered within R. maderensis and are here included under that species, which leaves salazinic acid as the only chemo-type in R. decipiens. Most R. decipiens specimens are easily distinguished from other species of the R. decipiens group. The pycnidia with black ostioles are of great diagnostic value, as the only other species with black pycnidia on the group are R. portosantana (Krog & Østhagen 1980b) and R. sampaioana. However, R. portosantana is never fertile, has matt, subterete or more or less complanate branches and is endemic to Porto Santo (Krog & Østhagen 1980b). Ramalina decipiens can be easily distinguished of R. sampaioana by its chemistry, as R. decipiens reacts K+ yellow, then orange to red, C-, Pd+ orange, UV- (containing salazinic acid) while R. sampaioana reacts K-, C-, KC-, Pd-, UV+ blue-white (containing divaricatic acid). The rare specimens belonging to the morphotype with pale pycnidia could be superficially confused with R. maderensis (Krog & Østhagen 1980a), but are easily separated by chemistry, as R. decipiens only has salazinic acid while R. maderensis in the Canary Islands has either lecanoric (K-, C+ red, KC+ carmine red, Pd-, UV-) or protocetraric acid (K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-). However, given the great morphological variability of R. decipiens some intermediate forms lacking pycnidia with black ostioles could be locally problematic to separate from micro-endemic species such as R. fortunatarum, R. gomerana, R. hamulosa, R. papyracea or R. sabinosana. Differences with those species are provided under their respective discussions. We observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 7b). These granules are similar to those observed by Stevens (1987) in other Ramalina species, which she suspected to be usnic acid crystals. We have observed them in other species of the R. decipiens group: R. gomerana, R. hamulosa, R. maderensis, R. sabinosana and R. sampaioana. Future studies should be carried out to explore the ubiquity and function of these granules in Ramalina.

Fig. 5.

Ramalina decipiens (morphotype with pale pycnidia, TFC-Lich 11971). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 6.

Ramalina decipiens (morphotype with black pycnidia, coll. num.: Pérez-Ortega 6221). Macroscopic and microscopic characters. a. Habit; b. laciniae detail; c. pseudocyphellae; d, e. apothecia; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Ramalina delicata Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846691; Fig. 8, 9

Fig. 8.

Ramalina delicata (holotype, TFC-Lich 17094). Macroscopic and microscopic characters. a. Habit; b. laciniae detail; c. laciniae detail showing a tremelloid gall; d. branch cracks; e. apothecium (coll. num.: Blázquez 279); f. thallus fractures; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 9.

Ramalina delicata (isotype, MA-Lich 26072). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Etymology. The specific epithet ‘delicata', from the Latin delicatus, refers to the extremely fragile laciniae of the species, compared to those of the remaining species of the R. decipiens group.

Typus. Spain, Canary Islands, Lanzarote, Volcán de la Corona summit, N29°11'5" W13°28'53", on volcanic rock, 520 m alt., 18 Mar. 2018, S. Pérez-Ortega (SPO 6226) & I. Pérez-Vargas (TFC-Lich 17094 holotype, isotype in MA-Lich 26072).

Diagnosis. Thallus saxicolous, laciniae subterete, extremely fragile, vegetative propagules (soralia, isidia and pycnidia) absent. Chondroid tissue very scarce, appearing as strands imbedded in the medulla and never adjoining the cortex. Secondary chemistry: divaricatic acid and a unique triterpene pattern.

Thallus saxicolous, fragile, erect, richly branched. Laciniae stramineous or green-grey, subterete to slightly complanate, matt, (10–)13–21(–25) × (0.5–)0.7–1.3(–2.3) mm, non-contorted, dichotomously branched, arising from a single holdfast. Pseudo-cyphellae absent. Fenestrations rare, occasionally present in some thalli. Anatomy bourgaeana-type (Krog & Østhagen 1980a). Cortex well developed, (15–)17–28(–33) μm thick. Chondroid tissue very scarce, never adjoining the cortex and appearing as strands imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia rare, marginal, (1–)1.1–1.4(–1.5) mm diam. Disc pale orange colored, concave, pruinose; thalline margin present, persistent. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elon-gate-clavate, 8-spored, 39–42(–43) × (4–)5 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, 10–13 × 4–5 µm. Pycnidia not seen.

Chemistry — Medulla K-, C-, KC-, Pd-, UV+ blue-white; TLC divaricatic acid and triterpene pattern type 1 (Fig. 3).

Ecology & Distribution — We have collected R. delicata in two localities in Lanzarote, Volcán de la Corona and Caldero Riscado, and in one locality in Fuerteventura, growing on rock crevices exposed to fog. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. maderensis (protocetraric acid chemotype) and R. pluviariae; as well as other Ramalina species: R. bourgaeana, R. crispatula, R. cupularis, R. hierrensis (which we have found growing on rock in Lanzarote) and R. krogiae.

Additional specimens examined. Spain, Canary Islands, Lanzarote, Caldero Riscado, N28°55'55" W13°45'19", on volcanic rock, 372 m alt., 2 Nov. 2021, M. Blázquez 312, 313, 314, 315, 316, 317, 318 & I. Pérez-Vargas; Canary Islands, Lanzarote, Volcán de la Corona summit, N29°11'5" W13°28'53", on volcanic rock, 520 m alt., 5 Sept. 2020, M. Blázquez 278, 279, 280, 305 & I. Pérez-Vargas.

Notes — Ramalina delicata is the basal species of the R. decipiens group (Fig. 2). It is an easily recognizable species, mainly because of the brittle consistency of its thallus and abundant fragmentation scars. Broken branches are common, leaving more or less circular fractures superficially resembling pseudocyphellae or even soralia. This feature could lead to confusion with certain morphotypes of R. krogiae, a species that does not belong to the group although it also has chondroid strands embedded in the medulla (Krog & Østhagen 1980a, Spjut et al. 2020). Ramalina krogiae also shares with R. delicata the presence of divaricatic acid (K-, C-, KC-, Pd-, UV+ blue-white). However, the triterpene patterns are diagnostic for each species. The rarity of reproductive structures and the presence of these fracture zones could be indicative of this species reproducing mainly through thallus fragmentation. The species is certainly rare in the Canarian archipelago, restricted to a few localities in the most eastern islands where it ranges from rare to locally abundant. Considering that some of the known localities have a high anthropogenic impact, conservation measures should be considered for its protection. Interestingly, historesin-embedded samples observed under the light microscope provided a picture of a rather lax medullar layer (Fig. 9b), quite different from the dense medulla observed in SEM pictures (Fig. 9a). One possible explanation is that the medulla may be densely packed with secondary products that disappear in historesin inclusion protocols. This pattern was also observed in R. erosa, but not in R. sampaioana, although both species have the same chemistry as R. delicata.

Ramalina erosa Krog, Lichenologist 22(3): 242. 1990 — Fig. 10, 11

Fig. 10.

Ramalina erosa (TFC-Lich 14932). Macroscopic and microscopic characters. a. Habit; b–c. laciniae detail; d. pseudocyphellae; e. branch showing solaria; f. detail of a soralium; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 11.

Ramalina erosa (coll. num.: Pérez-Ortega 7097). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 300 µm, b = 20 µm.

Typus. Portugal, Madeira, Porto Santo, Pico do Facho, N33°05' W16°19', on acidic rock, 500 m alt., 30 Apr. 1987, H. Krog & E. Timdal, (O-L-1215 holotype).

Thallus saxicolous, rigid, erect, sparingly to moderately branched. Laciniae stramineous, green-grey or pale yellow-brown, subterete to complanate, slightly angular, tapering gradually to blunt apices, shiny, (19–)27–49(–69) × 1–3(–5) mm, sometimes contorted, dichotomously branching, arising from a single hold-fast. Pseudocyphellae present in most thalli, laminar, usually near the base, 1–4(–7) mm in length. Fenestrations present in some thalli. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed, (30–)32–56(–60) μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia marginal, oval, delimited, to 1 mm in length. Pro- ducing coarse granules c. 200 μm diam. Frequently producing short branchlets with hook-shaped apices. Apothecia not seen. Pycnidia absent.

Chemistry — Medulla K-, C-, KC-, Pd-, UV-; TLC divaricatic acid and triterpene pattern type 2 (Fig. 3).

Ecology & Distribution — Ramalina erosa is only known from Porto Santo ( Krog 1990). We have collected it near the summit of four volcanic cones between 354 and 496 m altitude. This expands the distribution of the species, that was only known from near the type locality ( Sparrius et al. 2017). It grows on vertical cliffs influenced by mists. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. maderensis (lecanoric acid chemotype), R. nematodes, R. portosantana, R. sampaioana and the undescribed lineage Ramalina sp. 2; as well as other Ramalina species: R. confertula, R. crispatula, R. jamesii, R. krogiae and R. timdaliana.

Additional specimens examined. Portugal, Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11", on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7190, 7223; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6816, 6817; Madeira archipelago, Porto Santo, Pico do Facho, N33°05'02" W16°19'25", on volcanic rock, 474 m alt., 4 Aug. 2018, S. Pérez-Ortega 7017, 7029, 7030, 7033, 7040, 7043, 7070, 7074, 7108.

Notes — Ramalina erosa is the sister species to the clade formed by R. decipiens, R. fortunatarum, R. maderensis, R. porto santana and R. sampaioana (Fig. 2). It differs from all other species of the R. decipiens group by producing soralia ( Krog 1990). Some specimens with complanate branches could be confused with narrow-lobed forms of R. krogiae, a sorediate species that does not belong to the group but shares the same thallus anatomy and the presence of divaricatic acid (K-, C-, KC-, Pd-, UV+ blue-white, Fig. 3). In this case, the triterpene pattern is a useful diagnostic character to distinguish them ( Krog 1990). As in R. delicata, historesin-embedded samples observed under the light microscope provided a picture of a rather lax medullar layer (Fig. 11b), very different from the dense medulla observed in SEM pictures (Fig. 11a). This is further discussed under the ‘Notes’ section of R. delicata.

Ramalina fortunatarum Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846692; Fig. 12, 13

Fig. 12.

Ramalina fortunatarum (holotype, TFC-Lich 11829). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium; f. pycnidia (isotype, MA-Lich 26073); g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 13.

Ramalina fortunatarum (isotype, MA-Lich 26073). a. SEM micrography showing thallus anatomy; b. cortex detail, including a ×2 zoomed area showing extracellular granules. — Scale bars: a = 500 µm, b = 20 µm.

Etymology. The specific epithet ‘fortunatarum', refers to the term ‘Fortunatae Insulae’, one of the names given to the Canarian Archipelago by the Romans, from which the species in endemic.

Typus. Spain, Canary Islands, Tenerife, near Mirador de Aguaide, N28°33'49" W16°17'45", on volcanic rock, 566 m alt., 29 Oct. 2018, S. Pérez-Ortega, M. Blázquez & I. Pérez-Vargas (TFC-Lich 11829 holotype, isotype in MA-Lich 26073).

Diagnosis. Morphologically similar to R. maderensis, but with laciniae mostly shiny, up to 103 mm in length, containing 4-O-demethylbarbatic acid.

Thallus saxicolous, rigid, erect to subpendulous, moderately branched. Laciniae stramineous or pale yellow-brown, compla-nate, sometimes canaliculate, shiny, (10–)21–73(–103) × (1–)2– 4(–5) mm, contorted, dichotomously branching, arising from a single holdfast. Pseudocyphellae laminar, usually near the base, 1–4(–5) mm in length. Fenestrations not seen. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed, (33–)37–69(–83) μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia mostly marginal, rarely also subterminal, usually spurred, (1–)2–6(–8) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular, crenate or with radial cracks, occasionally turning vegetative and producing branchlets. Paraphyses simple, enlarged apically, 2–3 μm thick. Asci elongate-clavate, 8-spored, (30–)35–49(–50) × (6–)7– 9(–10) µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (9–)10–12(–15) × 4–5 µm. Pycnidia present in most thalli, marginal, at times also laminar, sometimes tuberculate, with pale ostioles. Conidia bacilliform, 4–5 × 1 µm.

Chemistry — Medulla K-, C+ orange, KC+ orange, Pd-, UV-; TLC 4-O-demethylbarbatic acid (Fig. 3).

Ecology & Distribution — We have collected R. fortunatarum in three localities in Tenerife, between 300 and 593 m altitude. We found it growing in vertical cliffs in Anaga but also in small rocks in Montaña de Taco – in both cases under the clear influence of mists. The species appears to have suffered a geographic range reduction in the last 40 years. We have collected R. fortunatarum only in Tenerife, but Krog & Østhagen (1980a) reported its presence in Lanzarote and the TFC-Lich herbarium harbors a specimen (TFC-Lich 137) collected in Gran Canaria in 1976. This range reduction could be an effect of climate change, a phenomenon already observed in other cryptogams of the Macaronesian region ( Patiño et al. 2016) and narrow endemic species of lichen-forming fungi elsewhere ( Allen & Lendemer 2016). It can be accompanied by another species of the R. decipiens group, R. decipiens, as well as other Ramalina species: R. bourgaeana, R. crispa tula, R. cupularis, R. krogiae, R. nodosa and R. pitardii.

Additional specimens examined. Spain, Canary Islands, Gran Canaria, Tamadaba, 14 Apr. 1976, B. Mond (TFC-Lich 137); Canary Islands, Tenerife, mirador de Aguaide, N28°33'53" W16°17'41", on volcanic rock, 593 m alt., 29 Oct. 2018, S. Pérez-Ortega 7931, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, montaña de Taco, N28°22'27" W16°50'07", on volcanic rock, 295 m alt., 23 Oct. 2021, M. Blázquez 335 & I. Pérez-Vargas; Canary Islands, Tenerife, risco ‘El Mogote’ on the road to Taganana, N28°33'29" W16°12'20", on volcanic rock, 427 m alt., 29 Oct. 2018, S. Pérez-Ortega 7464, 7899, 7912, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, riscos de Chinamada, N28°33'49" W16°17'45", on volcanic rock, 534 m alt., 29 Oct. 2018, S. Pérez-Ortega 7826, M. Blázquez 344, 345, 346 & I. Pérez-Vargas.

Notes — Ramalina fortunatarum is the sister species of R. maderensis (Fig. 2) and it was previously interpreted as a chemotype of R. maderensis containing 4-O-demethylbarbatic acid (Krog & Østhagen 1980a). However, in our phylogeny all specimens form a monophyletic group sister to R. maderensis which was identified as a separate species by most of the species discovery strategies, including the multispecies coalescent approach implemented with STACEY. Thus, the taxon is here described as a new species. Ramalina fortunatarum could be confused with R. maderensis on the basis of the presence of pseudocyphellae, commonly spurred apothecia and pycnidia with pale ostioles (Krog & Østhagen 1980a), but both species can be easily differentiated by chemistry, as R. fortunatarum always contains 4-O-demethylbarbatic acid (K-, C+ orange, KC+ orange, Pd-, UV-), while R. maderensis contains lecanoric (K-, C+ red, KC+ carmine red, Pd-, UV-) or protocetraric acids (K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-). This is true for the studied populations from the Canary Islands and Madeira archipelago. However, in Cape Verde (São Vicente island), we have found R. maderensis specimens with 4-O-demethylbarbatic acid. In this case, the diagnostic character to distinguish both species is surface sheen, as branches in Ramalina fortunatarum are shiny while all examined thalli of R. maderensis containing 4-O-demethylbarbatic acid are matt. Ramalina fortunatarum could also be confused with some specimens of R. decipiens (Krog & Østhagen 1980a). In this case, chemistry is diagnostic, as R. decipiens only contains salazinic acid (K+ yellow, then orange to red, C-, Pd+ orange, UV-). We observed granules in the cortex as well as in the chondroid strands of this species (Fig. 13b). These granules differ of those found in R. decipiens, R. gomerana, R. hamulosa, R. maderensis, R. sabinosana and R. sampaioana in that they do not appear as concentric rings around the hyphae, but are haphazardly distributed across the intercellular matrix. Because of their extracellular occurrence we suspect that they may correspond to accumulations of the 4-O-demethylbarbatic acid that this species contains. Further studies should be undertaken to explore this possibility. In Blázquez et al. (2022) we referred to this species as Ramalina fortunata nom. prov.

Ramalina gomerana Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846693; Fig. 14, 15

Fig. 14.

Ramalina gomerana (holotype, MA-Lich 26074). Macroscopic and microscopic characters. a. Habit; b–d. laciniae detail; e. apothecium; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 15.

Ramalina gomerana (isotype, TFC-Lich 17096). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Etymology. The specific epithet ‘gomerana' refers to La Gomera island, from which the species is endemic.

Typus. Spain, Canary Islands, La Gomera, Mirador de los Roques, N28°6'33" W17°12'58", on volcanic rock, 1120 m alt., 2 Nov. 2018, S. Pérez-Ortega (SPO 7855), M. Blázquez & I. Pérez-Vargas (MA-Lich 26074 holotype, isotype in TFC-Lich 17096).

Diagnosis. Morphologically similar to R. decipiens, but lacking pseudo-cyphellae: pycnidia are marginal, not tuberculate, with pale ostioles.

Thallus saxicolous, rigid, erect, moderately branched. Laciniae stramineous, green-grey or pale yellow-brown, more or less complanate, sometimes canaliculate, matt or, more rarely, shiny, showing vein-like ridges that are in fact the external appearance of the abundant chondroid strands that adjoin the cortex, 30–37(–44) × 2–3 mm broad, sometimes contorted, dichotomously branching, arising from a single holdfast. Pseudocyphellae absent. Fenestrations not seen. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well-developed, (55–)62–80(–83) μm thick. Chondroid tissue very abundant, forming a discontinuous cylinder or separate strands adjoining the cortex, with abundant strands filling almost completely the medulla layer. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia terminal, 4–5(–6) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, (33 –) 34 – 39 × (6 –)7– 9 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, 10–11 × 4–5 µm. Pycnidia present in most thalli, marginal, non-tuberculated, with pale ostioles. Conidia bacilliform, 5 × 1 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution — Ramalina gomerana is only known from four localities in La Gomera. We have collected it between 809 and 1134 m altitude, growing in vertical cliffs exposed to mist. It can be accompanied by other species of the R. decipiens group, R. decipiens and R. hamulosa, as well as other Ramalina species: R. bourgaeana, R. cupularis, R. krogiae, R. nodosa, R. subfarinacea and R. tortuosa.

Additional specimens examined. Spain, Canary Islands, La Gomera, Mirador de los Roques, N28°06'33" W17°12'57", on volcanic rock, 1 134 m alt., 2 Nov. 2018, S. Pérez-Ortega 7829, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, Roquedos al sur de Imada, N28°04'31" W17°14'13", on volcanic rock, 1 042 m alt., 2 Nov. 2018, S. Pérez-Ortega 7534, 7536, 7540, 7548, M. Blázquez & I. Pérez-Vargas; Canary Islands, La Gomera, Roquedos y paredones en el risco de Heredia, N28°07'52" W17°19'23", on volcanic rock, 811 m alt., 1 Nov. 2018, S. Pérez-Ortega 7938, M. Blázquez & I. Pérez-Vargas.

Notes — Ramalina gomerana is the sister species of R. hamulosa (Fig. 2). It can be differentiated from other species of the R. decipiens group by the large proportion of the cross-section that is filled by chondroid tissue, which gives the species a characteristic rigidity. Ramalina gomerana could be confused with some forms of R. decipiens. Ramalina decipiens thalli showing pycnidia with black ostioles can be easily differentiated from R. gomerana as this species’ pycnidia always have pale ostioles. Fortunately, this was the case of all the R. decipiens specimens collected in the localities in which R. gomerana was found. It can also be easily separated from R. decipiens forms possessing apothecia with a black ring, regardless of pycnidia, as this ring only occurs in R. decipiens. Unfortunately, there are R. decipiens specimens that lack pycnidia, pseudocyphellae and do not have a black ring on their apothecia. Ramalina gomerana shows a considerable morphological overlap with those specimens. As both species have salazinic acid as their only secondary metabolite, chemistry is of no use either. In such instances, a molecular approach would be necessary, as the two species can be unambiguously separated by DNA sequence data. We have observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 15b). These granules are further discussed under the ‘Notes’ section of R. decipiens.

Ramalina hamulosa Krog & Østh., Norweg. J. Bot. 27(3): 275. 1980 — Fig. 16, 17

Fig. 16.

Ramalina hamulosa (holotype, O-L-918). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium; f. hook-shaped structures (arrow); g. thallus anatomy (TFC-Lich 12597); thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 17.

Ramalina hamulosa (coll. num.: Pérez-Ortega 5934). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 200 µm, b = 20 µm.

Typus. Spain, Canary Islands, El Hierro, Barranco Hondo above Sabinosa, 350–400 m alt., 2 Apr. 1978, Krog & Østhagen (O-L-918 holotype!, isotypes in BM-001106894, UPS-L-078706).

Thallus saxicolous, fragile, shrubby to subpendulous, moderately to richly branched. Laciniae stramineous, subterete or flattened, shiny, (23–)27–87(–108) × 1–2(–3) mm, mostly not contorted, dichotomously branching, arising from a single hold-fast. Often with fragmentation areas in thin branches. With numerous, short, divergent branchlets terminating in hook-shaped structures or nodules. Pseudocyphellae marginal, (1–)2 – 5(–10) mm in length. Fenestrations sometimes present. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well-developed, (30–)35–55(–63) μm thick, at times eliminated in laciniae margins due to marginal pseudocyphellae. Chondroid tissue in younger parts forming continuous upper and lower layers, becoming disrupted in older parts, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia extremely rare, marginal, 1 mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, 32–39 × 6–9 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, 12–13 × 4–5 µm. Pycnidia absent.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution — Ramalina hamulosa had been reported from El Hierro, La Gomera, Tenerife, Lanzarote and Fuerteventura (Krog & Østhagen 1980a, Hernández-Padrón & Pérez-Vargas 2010, Van den Boom & Ertz 2012). However, we did not find the species in Lanzarote and Fuerteventura. We have collected it between 504 and 1 043 m altitude, growing in exposed cliffs affected by mists. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. gomerana, R. maderensis, R. papyracea and R.sabino sana, as well as other Ramalina species: R. bourgaeana, R. crispatula, R. cupularis, R. krogiae, R. nodosa, R. pitardii, R. subfarinacea and R. tortuosa.

Additional specimens examined. Spain, Canary Islands, El Hierro, road margin in Sabinar de la Dehesa, N27°44'48" W18°07'35", on volcanic rock, 605 m alt., 16 Mar. 2018, M. Blázquez 357, 358 & I. Pérez-Vargas; Canary Islands, La Gomera, between San Sebastián and Hermigua, N28°7' W17°9', on volcanic rock, 460 m alt., Jan. 2012. I. Pérez-Vargas (TFC-Lich 12614); Canary Islands, La Gomera, Roquedos al sur de Imada, N28°04'31" W17°14'13", on volcanic rock, 1 042 m alt., 2 Nov. 2018, S. Pérez-Ortega 7848, M. Blázquez & I. Pérez-Vargas; Canary Islands, Tenerife, Roque Vitor, N28°20'50" W16°50'53", on volcanic rock, 528 m alt., 23 Oct. 2021, M. Blázquez 338, 341 & I. Pérez-Vargas.

Notes — Ramalina hamulosa is the sister species of R. gomerana (Fig. 2). It can be differentiated from other species of the R. decipiens group by the presence of numerous hook-shaped branchlets. Krog & Østhagen (1980a) speculated that the fragile branchlets and nodules of this species could serve as vegetative propagules, which would explain why apothe cia in R. hamulosa are so infrequent. Ramalina hamulosa is easily distinguished from the other species of the R. decipiens group. It can only be confused with R. sabinosana and narrow forms of R. decipiens. However, the presence of hook-shaped branchlets (Fig. 16f) in R. hamulosa makes the separation rather straightforward in both cases. Ramalina hamulosa could also be superficially confused with some forms of R. crispatula which also present terete branches with hook-like structures at their ends. The two species can be easily differentiated by their anatomy, because while R. hamulosa has its chondroid filaments attached to the cortex, in R. crispatula they are embedded in the medulla. We have observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 17b). These granules are further discussed under the ‘Notes’ section of R. decipiens.

Ramalina maderensis Motyka, Fragm. Florist. Geobot. (Kraków) 6(4): 705. 1961 — Fig. 18–20

Fig. 18.

Ramalina maderensis collected in Porto Santo, Madeira archipelago (TFC-Lich 14906). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 20.

Ramalina maderensis (coll. num.: Pérez-Ortega 7246). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Typus. Portugal, Madeira, Mandon (H-NYL 36945 holotype!).

Thallus saxicolous, rigid, erect to subpendulous, sparingly to moderately branched. Laciniae stramineous, green-grey or pale yellow-brown, palmate or sublinear, sometimes canaliculate, matt or shiny, (6–)15–61(–180) × (1–)2–6(–22) mm, often contorted, dichotomously branching, arising from a single hold-fast. Pseudocyphellae laminar, rarely marginal, longitudinally arranged, (1–)2–5(–124) mm in length. Fenestrations present in some thalli. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well-developed, (35–)38–70(–80) μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla lax and highly hydrophobic. Isidia absent. Soralia absent. Apothecia mostly marginal, in some thalli terminal, often spurred, (1–)2–9(–15) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular, crenate or with radial cracks, occasionally turning vegetative and producing branchlets. Paraphyses simple, enlarged apically, 2–3 μm thick. Asci elongate-clavate, 8-spored, (31–)32–40(–50) × (6–)7–10 µm broad. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (7–)9– 12(–16) × (3–)4–5(–6) µm. Pycnidia marginal and laminar, mostly not tuberculate, with pale ostioles, sometimes mixed with tubercles that may be incipient pycnidia or apothecia. Conidia bacilliform, (3–)4–5(–6) × 1(–2) µm broad.

Chemistry — Four chemotypes have been detected: I) Medulla K-, C+ red, KC+ carmine red, Pd-, UV-; TLC lecanoric acid; II) medulla K-, C-, KC+ pink (reaction often weak and ephe- meral), Pd+ orange-red, UV-; TLC protocetraric acid; III) medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid; and IV) medulla K-, C+ orange, KC+ orange, Pd-, UV-; TLC 4-O-demethylbarbatic acid (Fig. 3).

Ecology & Distribution —Ramalina maderensis is the species of the R. decipiens group with the largest geographical range. We have collected it in the Canary Islands, the Madeira archipelago, the Cape Verde archipelago and the Azores archipelago, growing in all kind of volcanic rocky substrates influenced by sea mists. The records from São Miguel represent the first record for the Azores archipelago. Likewise, it is recorded for the first time from Sal in the Cape Verde archipelago and La Gomera in the Canary Islands. Ramalina maderensis has also been reported from St. Helena ( Aptroot 2008), being the only species of the R. decipiens group present in both hemispheres. It can be accompanied by all the other species of the R. decipiens group, as well as other Ramalina species: R. bourgaeana, R. confertula, R. crispatula, R. cupularis, R. hierrensis (which we have found growing on rock in Lanzarote), R. jamesii, R. krogiae, R. nodosa, R. parva, R. pitardii, R. rubrotincta, R. subfarinacea, R. timdaliana and R. webbii.

Additional specimens examined. Lecanoric acid: Portugal, Madeira archipelago, Madeira, Ponta de São Lourenço, N32°44'57" W16°41'36", on volcanic rock, 83 m alt., 6 Aug. 2018, S. Pérez-Ortega 7245, 7249, 7250, 7253, 7254, 7261, 7263, 7265, 7273, 7277; Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11", on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7161; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6825, 6914, 6922, 6924, 6955; Canary Islands, El Hierro, Gorona del Viento, N27°47'41" W17°55'19", on volcanic rock, 735 m alt., 15 Mar. 2018, M. Blázquez 355 & I. Pérez-Vargas; Canary Islands, El Hierro, rocky outcrops and rock fences along a rural path, N27°49'37" W17°55'05", on volcanic rock, 520 m alt., 15 Mar. 2018, S. Pérez-Ortega 5904, 5905, 5914, 5915, 5921, 5923, 5931 & I. Pérez-Vargas; Canary Islands, Fuerteventura, Jandía peninsula, Pico de Cuchillo del Ciervo, N28°05'25" W14°22'39", on volcanic rock, 492 m alt., 20 Mar. 2018, S. Pérez-Ortega & I. Pérez-Vargas; Canary Islands, Fuerteventura, Pico del Aceitunal, N28°32'50" W13°57'24", on volcanic rock, 650 m alt., 20 Mar. 2018, S. Pérez-Ortega 6549; Canary Islands, La Palma, between Las Caletas and Faro de Fuencaliente, N28°29'28" W17°49'48", on volcanic rock, 441 m alt., 3 Nov. 2018, M. Blázquez 79, 82, 83, 89, 90, 93 & I. Pérez-Vargas; Canary Islands, La Palma, Fuencaliente, Montaña del Viento, N28°28'16" W17°50'11", on volcanic rock, 215 m alt., 3 Nov. 2018, M. Blázquez 98 & I. Pérez-Vargas; Canary Islands, La Palma, Volcán Teneguía, N28°28'38" W17°51'00", on volcanic rock, 391 m alt., 3 Nov. 2018, M. Blázquez 64 & I. Pérez-Vargas; Canary Islands, Lanzarote, Caldero Riscado, N28°55'55" W13°45'19", on volcanic rock, 372 m alt., 2 Nov. 2021, M. Blázquez 310, 311 & I. Pérez-Vargas; Canary Islands, Lanzarote, Montaña Roja, N28°52'11" W13°51'13", on volcanic rock, 177 m alt., 2 Nov. 2021, M. Blázquez 303 & I. Pérez-Vargas; Canary Islands, Lanzarote, Peñas del Chache, N29°07'16" W13°30'59", on volcanic rock, 611 m alt., 17 Mar. 2018, S. Pérez-Ortega 6147 & I. Pérez-Vargas. Protocetraric acid: Portugal, Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11", on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7229. – Spain, Canary Islands, Fuerteventura, Jandía peninsula, Pico de Cuchillo del Ciervo, N28°05'25" W14°22'39", on volcanic rock, 492 m alt., 20 Mar. 2018, S. Pérez-Ortega 6535 & I. Pérez-Vargas; Canary Islands, Fuerteventura, Malpaís de la Arena, N28°37'50" W13°55'28", on volcanic rock, 310 m alt., 19 Mar. 2018, S. Pérez-Ortega 6401, 6411, 6415, 6417, 6420, 6421, 6424, 6440, 6441, 6442, 6445, 6446, 6449, 6453 & I. Pérez-Vargas; Canary Islands, Fuerteventura, Pico del Aceitunal, N28°32'50" W13°57'24", on volcanic rock, 650 m alt., 20 Mar. 2018, S. Pérez-Ortega 6560; Canary Islands, La Gomera, mirador de El Santo, N28°07'46" W17°19'30", on volcanic rock, 829 m alt., 1 Nov. 2018, S. Pérez-Ortega 8089, M. Blázquez & I. Pérez-Vargas; Canary Islands, Lanzarote, barranco de la Poceta, N29°06'51" W13°32'15", on volcanic rock, 249 m alt., 17 Mar. 2018, S. Pérez-Ortega 6084, 6087, 6089, 6092, 6102 & I. Pérez-Vargas; Canary Islands, Lanzarote, Ermita de las Nieves, N29°06'24" W13°31'44", on volcanic rock, 596 m alt., 17 Mar. 2018, S. Pérez-Ortega 6270, 6272, 6282 & I. Pérez-Vargas; Canary Islands, Lanzarote, malpaís de la Corona, N29°12'12" W13°25'41", on volcanic rock, 20 m alt., 18 Mar. 2018, S. Pérez-Ortega 6335, 6338, 6339, 6342, 6343, 6345, 6348, 6350, 6355, 6359, 6366, 6368, 6369, 6371, 6372, 6373, 6375, 6380, 6385, 6387 & I. Pérez-Vargas; Canary Islands, Lanzarote, malpaís near bodegas El Grifo, N29°00'04" W13°38'36", on volcanic rock, 312 m alt., 17 Mar. 2018, S. Pérez-Ortega 6286, 6287, 6291, 6294, 6309, 6310, 6313, 6319, 6328 & I. Pérez-Vargas; Canary Islands, Lanzarote, malpaís on the road between Masdache and La Vegueta, N29°00'39" W13°39'88", on volcanic rock, 316 m alt., 17 Mar. 2018, S. Pérez-Ortega 6127, 6136 & I. Pérez-Vargas; Canary Islands, Lanzarote, Peñas del Chache, N29°07'16" W13°30'59", on volcanic rock, 611 m alt., 17 Mar. 2018, S. Pérez-Ortega 6145, 6146, 6168, 6170 & I. Pérez-Vargas; Canary Islands, Lanzarote, Volcán de la Corona summit, N29°11'5" W13°28'53", on volcanic rock, 520 m alt., 18 Mar. 2018, S. Pérez-Ortega 6206, 6213, 6244, 6258, 6264, 6265 & I. Pérez-Vargas. Salazinic acid: Cape Verde, Cape Verde archipelago, Sal, top of Monte Grande, N16°49'15" W22°54'35", on volcanic rock, 374 m alt., 7 Aug. 2017, S. Pérez-Ortega 4773, 4774, 4825, 4828, 5177, 5179, 5182, 5183, 5191, 5210, 5212, 5213, 5214, 5216, 5218, 5219, 5220, 5221, 5223, 5224, 5225, 8660, 8662, I. Garrido-Benavent & I. Pérez-Vargas. 4-O-demethylbarbatic acid: Cape Verde, Cape Verde archipelago, Sal, top of Monte Grande, N16°49'15" W22°54'35", on volcanic rock, 374 m alt., 7 Aug. 2017, S. Pérez-Ortega 4777, 5184, I. Garrido-Benavent & I. Pérez-Vargas; Cape Verde archipelago, São Vicente, rocky outcrops in a small hill in the way from Mindelo to Salamansa, N16°53'16" W24°57'16", on volcanic rock, 233 m alt., 31 July 2017, S. Pérez-Ortega 4762, 4859, 4861, 4900, 5046, 5072, 5080, 5082, 5084, 5095, I. Garrido-Benavent & I. Pérez-Vargas.

Notes — Ramalina maderensis is the sister species of R. fortunatarum (Fig. 2). It is a widespread species that can be distinguished from the other species of the R. decipiens group mainly by its abundant laminar pseudocyphellae and chemical characters. Based on material collected in the Canary Islands Krog & Østhagen (1980a) interpreted R. maderensis as having two chemotypes, one containing lecanoric acid (to which the type specimen belongs, Krog & Østhagen 1980a) and other containing 4-O-demethylbarbatic acid. Based on our molecular data, the 4-O-demethylbarbatic acid chemotype present in the Canary Islands is here segregated as a new species, R. fortunatarum (for differences, see the ‘Notes’ section under R. fortunatarum). However, R. maderensis from Cape Verde shows two additional chemotypes, one of which contains 4-O-demethylbarbatic acid. All but one thallus (which contained lecanoric acid) collected in São Vicente contain 4-O-demethylbarbatic acid and all but two specimens (which contained 4-O-demethylbarbatic acid) collected in Sal contain salazinic acid. Krog & Østhagen (1980a) also reported a R. decipiens chemotype, restricted to the eastern Canary Islands, containing protocetraric acid. The studied specimens showing this chemotype are morphologically and molecularly indistinguishable from R. maderensis, so we consider them as clearly belonging to this species. Thus, R. maderensis is the species of the R. decipiens group showing the highest chemical diversity. This diversity, however, is not phylogenetically structured (see Fig. 2). Interestingly, two of the chemotypes contain depsides (lecanoric and 4-O-demethylbarbatic acids) and the other two contain depsidones (salazinic and protocetraric acids), two groups of substances that are structurally different ( Huneck & Yoshimura 1996). Like R. decipiens (the other widespread species in the group) R. maderensis shows extreme morphological variability (Fig. 10, 11). This fact, together with the presence of well-supported intraspecific clades showing geographic structure (see Fig. 2), could be indicative that R. maderensis is currently in the process of speciation, a possibility already speculated by Krog & Østhagen (1980a) based only on morphological and chemical evidence. With the data at hand, however, we think it best to keep interpreting the species in a wide sense. Ramalina maderensis shows remarkable variability on its size. Some specimens from Porto Santo surpass the length of 10 cm, which makes them the largest thalli in the R. decipiens group. Ramalina maderensis could be confused with R. fortunatarum and with some forms of R. decipiens. In both cases chemistry is diagnostic (Krog & Østhagen 1980a). Ramalina decipiens always contains salazinic acid (K+ yellow, then orange to red, C-, Pd+ orange, UV-) but does not occur in the Cape Verde archipelago, where the R. maderensis salazinic acid chemotype is restricted. Ramalina fortunatarum always contains 4-O-demethylbarbatic acid (K-, C+ orange, KC+ orange, Pd-, UV-). Ramalina maderensis also has a chemotype containing 4-O-demethylbarbatic acid, but it is restricted to São Vicente. Ramalina maderensis can also be confused with R. papyracea. We have found these species occurring together only in La Palma, but they can be morphologically differentiated because R. papyracea has very thin, soft, shiny and abundantly fenestrated branches. They can also be separated by chemistry, as R. papyracea contains only salazinic acid while the salazinic-producing R. maderensis chemotype is restricted to Sal in the Cape Verde archipelago. We have observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 20b). These granules are further discussed under the ‘Notes’ section of R. decipiens.

Ramalina nematodes (Nyl.) Krog & Østh., Norweg. J. Bot. 27(3): 187. 1980 — Fig. 21, 22

Fig. 21.

Ramalina nematodes (coll. num.: Pérez-Ortega 6802). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium (coll. num.: Pérez-Ortega 6986); f. ascospore (coll. num.: Pérez-Ortega 6986); g. thallus anatomy; thalline layers are indicated as Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–e = 1 mm, f = 10 µm, g = 200 µm.

Fig. 22.

Ramalina nematodes (coll. num.: Pérez-Ortega 6802). a. SEM micrograph showing thallus anatomy; b. chondroid strand detail. — Scale bars: a = 200 µm, b = 20 µm.

Basionym. Ramalina scopulorum var. nematodes Nyl., Bull. Soc. Linn. Normandie, sér. 2 4(2): 157. 1870.

Synonym. Ramalina siliquosa var. nematodes (Nyl.) Tav., Portugaliae Acta Biol., Sér. B 3: 381. 1952.

Typus. Ins. Canar. [The Madeira Islands], Portu Sanctu [Porto Santo] (H-NYL-36988 lectotype!, PC-0000405 isolectotype).

Thallus saxicolous, rigid, subpendulous to pendulous, moderately to richly branched. Laciniae stramineous or green-grey, subterete or more or less complanate, matt, (29–)39–101(–125) × (0.2–)0.4–1.2(–2.5) mm broad, dichotomously branching, ari- sing from a single holdfast. Pseudocyphellae-like structures (see notes) prominent, longitudinally arranged, 1–6(–11) mm in length. Rarely fenestrated. Cortex absent. Strands of chondroid tissue intermingled with a dense, highly hydrophobic medulla. Isidia absent. Soralia absent. Apothecia very rare, marginal, 2–7(–10) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, 30–39 × 8(–9) µm. Asco-spores 1-septate, hyaline, broadly ellipsoid to slightly kidney- shaped, 10–11(–13) × 4–5 µm. Pycnidia not seen.

Chemistry — Two chemotypes have been detected: I) Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid; and II) Medulla K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-; TLC protocetraric acid (Fig. 3).

Ecology & Distribution — Ramalina nematodes is endemic to the island of Porto Santo in the Madeira archipelago (Krog & Østhagen 1980b). We have collected it in three of the island’s volcanic cones, between 204 and 395 m altitude, but it is known to form dense populations on all cliff tops of the island ( Sparrius et al. 2017). It grows on all manner of rocky substrates influenced by sea mists. It can be accompanied by other species of the R. decipiens group, namely R. decipiens, R. erosa, R. maderensis (of its lecanoric acid strain) and R. sampaioana, as well as other Ramalina species: R. confertula, R. crispatula, R. jamesii, R. krogiae and R. timdaliana.

Additional specimens examined. Salazinic acid: Portugal, Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6847, 6933, 6965, 6966, 6980, 6983, 6987, 6988, 6994, 6997, 7002, 7004, 7006, 7007. Protocetraric acid: Portugal, Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11", on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7230, 7234; Madeira archipelago, Porto Santo, Pico de Ana Ferreira, N33°02'49" W16°22'04", on volcanic rock, 172 m alt., 4 Aug. 2018, S. Pérez-Ortega 7166, 7182, 7201; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6814, 6820, 6931, 6944, 6969, 6976, 6977, 6981, 6995, 6999; Madeira archipelago, Porto Santo, Pico do Facho, N33°05'02" W16°19'25", on volcanic rock, 474 m alt., 4 Aug. 2018, S. Pérez-Ortega 7020, 7026, 7066, 7072, 7096.

Notes — Ramalina nematodes was originally described as having pseudocyphellae (Krog & Østhagen 1980b), but given that the species lacks a cortex these structures cannot be described as true pseudocyphellae (i.e., breaks in the cortex where the medullar hyphae extend to the surface), but rather as the outward appearance of the longitudinally arranged chondroid strands interspersed with the medulla. Ramalina nematodes is sister to R. pluviariae (Fig. 2). It can be differentiated rather easily from most of the species of the R. decipiens group because it lacks a cortex. Morphologically it is only slightly similar to its sister species, R. pluviariae, as both share the lack of a cortex (Krog & Østhagen 1980a, b). The two can be distinguished from each other by the lack of a delimited holdfast in R. pluviariae, which is always present in R. nematodes; the prostrate thallus of R. pluviarieae, pendulous in R. nematodes; and thicker walled ascospores in R. pluviariae. In addition, R. nematodes is endemic to Porto Santo, whereas R. pluviariae is only known from the easternmost Canary Islands. Ramalina nematodes could be superficially confused with R. rubrotincta, a species endemic to the Cape Verde archipelago. These two species can be easily distinguished, however, by the presence of norstictic and connorstictic acids in R. rubrotincta. Based on their ecological and morphological similarities, Mies & Lösch (1995) hypothesized that R. nematodes, R. pluviariae, R. rubrotincta and R. implexa (as R. arabum) were endemic vicariant species sharing a common ancestor. Our results do not support this hypothesis as only R. nematodes and R. pluviariae cluster into the R. decipiens group, while R. rubrotincta and R. implexa appear elsewhere in the Ramalina phylogeny (Pérez-Ortega et al. unpubl. data).

Ramalina papyracea Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846694; Fig. 23, 24

Fig. 23.

Ramalina papyracea (holotype, MA-Lich 26075). Macroscopic and microscopic characters. a. Habit; b. laciniae detail; c. pseudocyphellae; d. apothecium; e. fenestrations; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 24.

Ramalina papyracea (holotype, MA-Lich 26075). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Etymology. The specific epithet ‘papyracea', from the Latin papyrus, refers to the papery consistency of the species laciniae when moist.

Typus. Spain, Canary Islands, El Hierro, road HI-500 between Pozo de la Salud and El Verodal, N27°45'58" W18°7'50", on volcanic rock, 30 m alt., 16 Mar. 2018, S. Pérez-Ortega (SPO 6021) & I. Pérez-Vargas (MA-Lich 26075 holotype, isotype in TFC-Lich 17092).

Diagnosis. Morphologically similar to R. decipiens, but with thin laci niae, abundantly fenestrated; pycnidia are laminar, not tuberculate, with pale ostioles; pseudocyphellae mainly laminar but also marginal.

Thallus saxicolous, rigid (becoming very soft when moist), erect, sparingly to moderately branched. Laciniae stramineous or green-grey, more or less complanate, very thin, shiny, (7–)12– 57(–84) × 2–7(–12) mm, sometimes contorted, dicho tomously branching, arising from a single holdfast. Pseudocyphellae laminar and marginal, (1–)2–5(–6) mm in length. Fene strations very abundant. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well-developed, (20 –)23 – 43(– 48) μm thick. Chondroid tissue very abundant, forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia marginal and/or terminal, 2–7(–10) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular, crenate or with radial cracks. Paraphyses simple, not capitate, 1 μm thick. Asci elongate-clavate, 8-spored, (30–)33–38(–39) × 7–10 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, 10–14(–15) × 4–5 µm. Pycnidia very frequent, laminar, non-tuberculated, with pale ostioles. Conidia bacilliform, 4–5 × 1 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution — Ramalina papyracea has been collected in El Hierro, Tenerife and La Palma. We have found it in localities between 37 and 255 m altitude in El Hierro and La Palma, while in Tenerife it was found at 504 m. All localities were influenced by mists. The species grows on cliffs, but also in small rocks and volcanic scoria in El Hierro. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. hamulosa and R. maderensis, as well as other Ramalina species: R. bourgaeana, R. cupularis, R. krogiae and R. pitardii.

Additional specimens examined. Spain, Canary Islands, El Hierro, malpaís near road HI-500 between Pozo de la Salud and El Verodal, N27°45'58" W18°07'49", on volcanic rock, 37 m alt., 16 Mar. 2018, S. Pérez-Ortega 5989, 5991, 5995, 6004 & I. Pérez-Vargas; Canary Islands, El Hierro, near road HI-500 on the way up to Sabinar de la Dehesa, N27°45'13" W18°08'38", on volcanic rock, 255 m alt., 16 Mar. 2018, S. Pérez-Ortega 6007 & I. Pérez-Vargas; Canary Islands, La Palma, Fuencaliente, Montaña del Viento, N28°28'16" W17°50'11", on volcanic rock, 215 m alt., 3 Nov. 2018, M. Blázquez 96 & I. Pérez-Vargas; Canary Islands, Tenerife, Roque Vitor, N28°20'50" W16°50'53", on volcanic rock, 528 m alt., 23 Oct. 2021, M. Blázquez 336, 340 & I. Pérez-Vargas.

Notes — Ramalina papyracea is the sister species of R. sabinosana (Fig. 2). It is an easily recognizable species within the R. decipiens group, mainly because of its thin, abundantly fenestrated branches. Ramalina papyracea could be confused with R. maderensis since they share some morphological features. Both have relatively broad branches, laminar pseudocyphellae, marginal and /or terminal apothecia and laminar, non-tuberculate pycnidia with pale ostioles (Krog & Østhagen 1980a). Ramalina papyracea can be morphologically differentiated from R. maderensis by its very thin, soft, shiny and abundantly fenestrated branches. Its pseudocyphellae are laminar but also marginal, while R. maderensis rarely has marginal pseudocyphellae (Krog & Østhagen 1980a). Chemically, R. papyracea always has salazinic acid (K+ yellow, then orange to red, C-, Pd+ orange, UV-) while R. maderensis contains either lecanoric (K-, C+ red, KC+ carmine red, Pd-, UV-) or protocetraric acids (K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-). This is true for the localities in which the two species grow together. Ramalina maderensis also has a chemotype with salazinic acid, but it seems to be restricted to the island of Sal in the Cape Verde archipelago. Ramalina papyracea could also be confused with some forms of R. decipiens, but it can again be differentiated by the softness and thinness of its branches and its abundant fenestrations. The location of the pseudocyphellae can also be helpful to distinguish them; in R. papyracea they are both laminar and marginal, while in R. decipiens (when present) they are only laminar (Krog & Østhagen 1980a).

Ramalina pluviariae Krog & Østh., Norweg. J. Bot. 23(4): 236. 1976 — Fig. 25, 26

Fig. 25.

Ramalina pluviariae (isotype, O-L-933). Macroscopic and microscopic characters. a. Habit; b. laciniae detail; c. pseudocyphellae; d. apothecia; e. pycnidia; f. ascospore (coll. num.: Pérez-Ortega 6555); g. thallus anatomy (coll. num.: Pérez-Ortega 6555); thalline layers are indicated as Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–e = 1 mm, f = 10 µm, g = 200 µm.

Fig. 26.

Ramalina pluviariae (coll. num.: Pérez-Ortega 6204). a. SEM micrography showing thallus anatomy; b. chondroid strand detail. — Scale bars: a = 200 µm, b = 20 µm.

Typus. Spain, Canary Islands, Lanzarote, Famara, Bco. de la Poceta, 400–600 m alt., Krog & Østhagen (O-L-934 holotype, isotypes at O-L-933!, BM-001106925, UPS -L-078716, TFMC).

Thallus saxicolous, rigid, erect to prostrate, richly branched. Laciniae stramineous or green-grey, subterete, matt, (16–)18– 57(–93) × (0.3–)0.4–0.9(–1.2) mm, dichotomously branching, without a delimited holdfast. Pseudocyphellae-like structures (see notes on R. nematodes) longitudinally arranged, (0.3–)0.4–1.1(–1.7) mm in length. Very rarely fenestrated. Cortex absent. Strands of chondroid tissue intermingled with a dense, highly hydrophobic medulla. Isidia absent. Soralia absent. Apothecia marginal, (1–)2–4(–5) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, (32–)34–37(–38) × 7–10 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (6–)9–11 × 4–5 µm, with thick spore walls. Pycnidia laminar, non-tuber-culated, with pale ostioles. Conidia bacilliform, (4–)5 × 1 µm.

Chemistry — Medulla K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-; TLC protocetraric acid (Fig. 3).

Ecology & Distribution — Ramalina pluviariae is only found in Lanzarote and Fuerteventura, where it can be locally abundant (Krog & Østhagen 1980a). Hernández-Padrón & Pérez de Paz (1995) reported this species from La Palma island in a rocky area of humid Monteverde in the Reserve of ‘El Canal y Los Tilos’. This is an area ecologically completely different from the localities in Fuerteventura and Lanzarote. This location has been visited several times by us and no thalli of this species were found. We collected R. pluviariae between 20 and 648 m altitude, in localities affected by mist. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. maderensis (usually of its protocetraric acid chemotype) and R. delicata, as well as other Ramalina species: R. bourgaeana, R. crispatula, R. cupularis, R. deminuta, R. krogiae, R. parva, R. pitardii, R. subfarinacea and R. webbii.

Additional specimens examined. Spain, Canary Islands, Fuerteventura, Pico del Aceitunal, N28°32'50" W13°57'24", on volcanic rock, 650 m alt., 20 Mar. 2018, S. Pérez-Ortega 6553; Canary Islands, Lanzarote, malpaís de la Corona, N29°12'12" W13°25'41", on volcanic rock, 20 m alt., 1 Nov. 2021, M. Blázquez 319, 321 & I. Pérez-Vargas.

Notes — Ramalina pluviariae is the sister species of R. nematodes (Fig. 2). It can be differentiated rather easily from most of the species of the R. decipiens group because it lacks a cortex. The most morphologically similar species is R. nematodes, which also lacks the cortical layer (Krog & Østhagen 1980a, b), but both species can be easily differentiated (see the ‘Notes’ section under R. nematodes). In addition, R. pluviariae is only known from the easternmost Canary Islands, whereas R. nematodes is endemic to Porto Santo.

Ramalina portosantana Krog, Lichenologist 22(3): 244. 1990 — Fig. 27, 28

Fig. 27.

Ramalina portosantana (holotypus, O-L-1217). Macroscopic and microscopic characters. a. Habit; b–d. laciniae detail; e pseudocyphellae; f. pycnidia; g. thallus anatomy (TFC-Lich 14963); thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 2 mm, g = 500 µm.

Fig. 28.

Ramalina portosantana (coll. num.: Pérez-Ortega 7010). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 300 µm, b = 20 µm.

Typus. Portugal, Madeira, Porto Santo, Pico do Facho, N33°05' W16°19', on acidic rock, c. 500 m alt., 30 Apr. 1987, H. Krog & E. Timdal (O-L-1217 holotype!, isotypes at BM, UPS).

Thallus saxicolous, rigid, erect to subpendulous sparingly branched. Laciniae green-grey or pale yellow-brown, subterete or more or less complanate, tapering gradually towards blunt apices, matt, sometimes with shiny apices, (24–)30–58(–62) × 1–2(–3) mm, non-contorted, dichotomously branching, often eroded, arising from a single holdfast. Pseudocyphellae present in most thalli, laminar, usually near the base, 1–3(–5) mm in length. Fenestrations absent. Cortex well developed, (40–)48– 73(–80) μm thick, becoming almost completely eroded in older parts. Chondroid tissue very abundant, mostly forming numerous strands imbedded in the medulla, to a lesser extent also adjoining the cortex. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia not seen. Pycnidia present, laminar (marginal in flat laciniae), non-tuberculated, with black ostioles. Conidia bacilliform, 4–5 × 1–2 µm.

Chemistry — Two chemotypes have been detected: I) Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid; and II) medulla K-, C-, KC+ pink (reaction often weak and ephemeral), Pd+ orange-red, UV-; TLC protocetraric acid (Fig. 3).

Ecology & Distribution — Ramalina portosantana is only known from Porto Santo. Krog (1990) only reported it from the summit of Pico do Facho. We have also found it in three additional volcanic cones, between 334 and 469 m in altitude. It grows on all manner of rocky substrates influenced by mists. It can be accompanied by other species of the R. decipiens group, such as R. decipiens, R. erosa, R. maderensis and the undescribed lineage Ramalina sp. 2, as well as other Ramalina species: R. confertula, R. crispatula, R. jamesii, R. krogiae and R. timdaliana.

Additional specimens examined. Salazinic acid: Portugal, Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 7001; Madeira archipelago, Porto Santo, Pico do Facho, N33°05'02" W16°19'25", on volcanic rock, 474 m alt., 4 Aug. 2018, S. Pérez-Ortega 7010. Protocetraric acid: Portugal, Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11", on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7194, 7224; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6975.

Notes — Ramalina portosantana is the sister species to the clade formed by R. fortunatarum and R. maderensis (Fig. 2). It is easily distinguished from other species of the R. decipiens group by its eroded branches, abundant black pycnidia and lack of apothecia. Krog (1990) described R. portosantana as having a single chemotype, containing salazinic acid. However, we have also found specimens containing protocetraric acid. Ramalina portosantana could be confused with R. sampaioana and some forms of R. decipiens on the basis of their pycnidia with black ostioles (Krog & Østhagen 1980a). Ramalina portosantana, however, has matt, subterete to more or less complanate branches and it is never fertile ( Krog 1990). It is easily separated from R. sampaioana, as this species always contains divaricatic acid (K-, C-, KC-, Pd-, UV+ blue-white). It could also potentially be confused with sparingly branched specimens of R. nematodes with broad branches; R. nematodes, however, always lacks cortex (Krog & Østhagen 1980b, Krog 1990).

Ramalina sabinosana Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846695; Fig. 29, 30

Fig. 29.

Ramalina sabinosana (holotype, TFC-Lich 17093). Macroscopic and microscopic characters. a. Habit; b. laciniae detail; c. tappering; d. pseudo-cyphellae; e. apothecia (coll. num.: Pérez-Ortega 6051); f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 30.

Ramalina sabinosana (holotype, TFC-Lich 17093). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 200 µm, b = 20 µm.

Etymology. The specific epithet ‘sabinosana' refers to the Sabinosa region in El Hierro, from which the species is endemic.

Typus. Spain, Canary Islands, El Hierro, Sabinar de la Dehesa, N27°45'1" W18°7'36", on volcanic rock, 590 m alt., 16 Mar. 2018, S. Pérez-Ortega (SPO 6060) & I. Pérez-Vargas (TFC-Lich 17093 holotype, isotype in MA-Lich 26076).

Diagnosis. Morphologically similar to R. decipiens, but pycnidia are marginal, tuberculated, with pale ostioles, pseudocyphellae always marginal and apothecia generally absent.

Thallus saxicolous, rigid, erect to subpendulous, moderately to richly branched. Laciniae stramineous or green-grey, compla-nate, tapering gradually towards sharp apices, shiny, 46 – 67(–70) × c. 1 mm, non-contorted, dichotomously branching, arising from a single holdfast. Pseudocyphellae marginal, (4–)5–9(–12) mm in length. Fenestrations present but scarce. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed, (30–)32–48 μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia extremely rare, marginal, 1.3 mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, 30–38 × 7–9 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, 12–13 × 4–5 µm. Pycnidia marginal, tuberculated, with pale ostioles. Conidia bacilliform, (3–)4–5 × 1 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution —Ramalina sabinosana is only known from three close localities in the Sabinar de la Dehesa, which is included in the ‘H-4 Parque Rural de Frontera’ natural protected area, in El Hierro island. We have collected it from 592 to 660 m in altitude. It grows in all manner of rocky substrates influenced by sea mists. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. hamulosa and R. maderensis (of its lecanoric acid strain), as well as other Ramalina species: R. crispatula, R. cupularis, R. krogiae, R. nodosa, R. pitardii and R. subfarinacea.

Additional specimens examined. Spain, Canary Islands, El Hierro, road margin in Sabinar de la Dehesa, N27°44'48" W18°07'35", on volcanic rock, 605 m alt., 16 Mar. 2018, S. Pérez-Ortega 5941, 6051, 6064, M. Blázquez 306, 307 & I. Pérez-Vargas.

Notes — Ramalina sabinosana is the sister species of R. papyracea (Fig. 2). It can be differentiated from other species of the R. decipiens group by its marginal pseudocyphellae and tuberculated pycnidia with pale ostioles. Ramalina sabinosana could be confused with R. decipiens and R. hamulosa, two species with which it occurs. Ramalina hamulosa shows characteristic small branchlets ended in hook-shaped structures (Krog & Østhagen 1980a) which are unique in the group. In addition, R. sabinosana shows abundant pycnidia, absent in R. hamulosa (Krog & Østhagen 1980a). Despite being phylo-genetically rather distant, differentiating R. sabinosana and R. decipiens can be more problematic due to the great intraspecific variability that R. decipiens can show. All examined thalli of R. sabinosana show marginal, tuberculated pycnidia with pale ostioles and marginal pseudocyphellae. Apothecia are extremely rare in the species; only one fertile specimen has been found (coll. num.: Pérez-Ortega 6051). Thus, the species is easily differentiated of fertile forms of R. decipiens or those having pycnidia with black ostioles. However, there are some rare R. decipiens specimens that show marginal, tuberculated pycnidia with pale ostioles and lack apothecia. In this case both species can be differentiated by the position of their pseudocyphellae. In R. sabinosana pseudocyphellae are always marginal whereas in R. decipiens, if present, are laminar (Krog & Østhagen 1980a). We have observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 30b). These granules are further discussed under the ‘Notes’ section of R. decipiens. In Blázquez et al. (2022) we referred to this species as R. sabinosae nom. prov.

Ramalina sampaioana Blázquez, Pérez-Vargas & Pérez-Ort., sp. nov. — MycoBank MB 846696; Fig. 31, 32

Fig. 31.

Ramalina sampaioana (holotype, MA-Lich 26077). Macroscopic and microscopic characters. a. Habit; b–c. laciniae detail; d. pycnidia; e. pseudo-cyphellae; f. apothecium; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Fig. 32.

Ramalina sampaioana (isotype, TFC-Lich 17095). a. SEM micrography showing thallus anatomy; b. cortex detail. — Scale bars: a = 500 µm, b = 20 µm.

Etymology. The specific epithet ‘sampaioana' refers to the Portuguese lichenologist Gonçalo Sampaio, on whose honor the species is named.

Typus. Portugal, Porto Santo, Pico do Facho, N33°5'2" W16°19'16", on volcanic rock, 489 m alt., 4 Aug. 2018, S. Pérez-Ortega (SPO 7083) (MA-Lich 26077 holotype, isotype in TFC-Lich 17095).

Diagnosis. Morphologically similar to R. decipiens, but lacking pycnidia and containing divaricatic acid and a triterpene pattern different to that of R. erosa and R. delicata.

Thallus saxicolous, rigid, erect, moderately branched. Laciniae stramineous or green-grey, complanate, shiny, (22–)29–64 (–90) × 1–3(–4) mm broad, mostly non-contorted, dichotomously branching, arising from a single holdfast. Pseudo-cyphellae present in some thalli, laminar, usually near the base, 2–6(–7) mm in length. Fenestrations sometimes present. Cortex well-developed, 30 – 54(–70) μm thick. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia marginal, (1–)2–4(–6) mm diam. Disc pale orange colored, at first deeply concave, becoming flat at maturity, pruinose; thalline margin present, persistent, circular or with radial cracks. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored, (36–)37–40(–41) × 8–9 µm. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (9–)10–13(–15) × 4–5 µm. Pycnidia rare, marginal, with black ostioles. Conidia bacilliform, 5 × 1 µm.

Chemistry — Medulla K-, C-, KC-, Pd-, UV-; TLC divaricatic acid and triterpene pattern type 3 (Fig. 2).

Ecology & Distribution — Ramalina sampaioana is only known from Porto Santo. We have collected it near the summit of four of the island’s volcanic cones, between 204 and 496 m altitude. It grows on all manner of rocky substrates influenced by sea mists. It can be accompanied by other species of the R. decipiens group, R. decipiens, R. erosa, R. maderensis (lecanoric acid chemotype) and R. nematodes, as well as other Ramalina species: R. confertula, R. crispatula, R. jamesii, R. krogiae and R. timdaliana.

Additional specimens examined. Portugal, Madeira archipelago, Porto Santo, Pico Branco e Terra Chã, N33°05'38" W16°18'11"W, on volcanic rock, 334 m alt., 4 Aug. 2018, S. Pérez-Ortega 7191, 7197, 7221, 7222, 7235; Madeira archipelago, Porto Santo, Pico de Ana Ferreira, N33°02'49" W16°22'04", on volcanic rock, 172 m alt., 4 Aug. 2018, S. Pérez-Ortega 7210; Madeira archipelago, Porto Santo, Pico do Castelo, N33°04'46" W16°20'04", on volcanic rock, 340 m alt., 4 Aug. 2018, S. Pérez-Ortega 6917; Madeira archipelago, Porto Santo, Pico do Facho, N33°05'02" W16°19'25", on volcanic rock, 474 m alt., 4 Aug. 2018, S. Pérez-Ortega 7028, 7031, 7046, 7051, 7056, 7061, 7064, 7065, 7082, 7088, 7106, 7113, 7115, 7116, 7117.

Notes — Ramalina sampaioana is the sister species to the clade formed by R. decipiens, R. fortunatarum, R. maderensis, R. portosantana and Ramalina sp. 1 (Fig. 2). It can be distinguished from the other species of the R. decipiens group mainly by the presence of pycnidia with black ostioles and divaricatic acid. Ramalina sampaioana is morphologically close to R. decipiens specimens with flat, shiny branches that lack pycnidia. The two species are easily separated, however, by chemistry: R. decipiens always has salazinic acid (K+ yellow, then orange to red, C-, Pd+ orange, UV-), while R. sampaioana always has divaricatic acid (K-, C-, KC-, Pd-, UV+ blue-white) and a triterpene pattern. Specimens with black pycnidia could also be superficially confused with R. portosantana; in this case the terete branches and lack of divaricatic acid in the latter species are diagnostic characters. We have observed numerous granules around the cortical hyphae of this species that are not present in the chondroid strands (Fig. 32b). These granules are further discussed under the ‘Notes’ section of R. decipiens. In Blázquez et al. (2022) we referred to this species as R. sampaiana nom. prov.

Ramalina sp. 1 — Fig. 33a, b

Fig. 33.

Undescribed lineages. a, b. Ramalina sp. 1 (TFC-Lich 11902); c, d. Ramalina sp. 2, collected in Tenerife (TFC-Lich 11852); e, f. Ramalina sp. 2, collected in La Gomera (TFC-Lich 12593); g, h. Ramalina sp. 2, collected in Madeira (TFC-Lich 10886). — Scale bars: a, c, e, g = 2 cm, b, d, f, h = 2 mm.

Thallus saxicolous, rigid, erect to subpendulous, richly branched. Laciniae stramineous, complanate, shiny, (44–)48–74(–84) × (2.5–)2.8–4.2(–4.3) mm, sometimes contorted, dichotomously branching, arising from a single holdfast. Pseudocyphellae some- times present, laminar, usually near the base, 2.2–5.6 mm in length. Fenestrations present. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothecia marginal, (2.5–)2.9–4.7(–5.2) mm diam. Disc pale orange colored, concave, pruinose; thalline margin present, persistent, circular. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (9–)10–11 × 4–5 µm. Pycnidia marginal, non- tuberculate, with black ostioles. Conidia bacilliform, 5 × 1 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution —Ramalina sp. 1 was represented in our sampling by only two specimens (TFC-Lich 11902 and coll. num. Pérez-Ortega 7897) collected in Tenerife and Fuerteventura. The species co-occurred with other Ramalina species, two of which belonging to the R. decipiens group: R. decipiens and R. maderensis.

Notes — Ramalina sp. 1 is the only species in our study whose phylogenetic placement was ambiguous: in some trees it appeared as the sister species of R. decipiens and, in others, as the sister species of the clade formed by R. maderensis, R. fortunatarum and R. portosantana. Support was lacking in both cases (Fig. 2). The two specimens were morphologically and chemically indistinguishable from R. decipiens specimens belonging to the morphotype with dark pycnidia, in particular with the concept of R. subwebbiana in that they have relatively flat, shiny laciniae, with none or relatively few pseudocyphellae (restricted to the base) and non-tuberculate pycnidia with black ostioles (in the specimen in which pycnidia are present). We argue, however, that this name should not be applied to Ramalina sp. 1, as multiple specimens with the same morphology appear across the subclades of R. decipiens. We therefore refrain from formally describing this lineage as a new species until more material is available.

Ramalina sp. 2 — Fig. 33c–h

Thallus saxicolous, rigid, erect to subpendulous, scarcely to richly branched. Laciniae stramineous or green-grey, complanate, matt or shiny, (12.3 –)17.8 – 93.2(–123.9) × (1–)1.4 – 4.3(–5.4) mm, sometimes contorted, dichotomously branching, ari- sing from a single holdfast. Pseudocyphellae sometimes present, laminar, rarely also marginal, sometimes near the base, (0.9–)1–3.6(–5.5) mm in length. Fenestrations sometimes present. Anatomy decipiens-type (Krog & Østhagen 1980a). Cortex well developed. Chondroid tissue forming a discontinuous cylinder or separate strands adjoining the cortex, with some strands appearing imbedded in the medulla. Medulla dense and highly hydrophobic. Isidia absent. Soralia absent. Apothe cia marginal and terminal, (1.6–)1.8–3.6(–3.8) mm diam. Disc pale orange colored, concave, pruinose; thalline margin present, persistent, circular. Paraphyses simple, not enlarged apically, 1 μm thick. Asci elongate-clavate, 8-spored. Ascospores 1-septate, hyaline, broadly ellipsoid to slightly kidney-shaped, (10–)11–13(–14) × (3–)4(–5) µm. Pycnidia laminar and marginal, sometimes tuberculated, with pale ostioles. Conidia bacilliform, 4 × 1 µm.

Chemistry — Medulla K+ yellow, then orange to red, C-, Pd+ orange, UV-; TLC salazinic acid (Fig. 3).

Ecology & Distribution —Ramalina sp. 2 was represented in our sampling by seven specimens collected in: Tenerife (TFCLich 11852), La Gomera (TFCLich 12590 and TFCLich 12593), Madeira (TFCLich 10857, TFCLich 10886 and TFCLich 10893) and Porto Santo (coll. num.: Pérez-Ortega 7206). It is accompanied by other Ramalina species of the R. decipiens group such as R. erosa, R. hamulosa, R. maderensis, R. nematodes, R. portosantana and R. sampaioana.

Notes — Ramalina sp. 2 appears as the sister species of the clade formed by R. sabinosana and R. papyracea (Fig. 2). It is unclear if this lineage corresponds to a single species. On the one hand, the seven specimens are phylogenetically very close, with differences in only 12 positions of the ITS. On the other hand, they show striking differences in size and overall morphology: some specimens have subterete branches with abundant pseudocyphellae (Fig. 33c, d); others possess com- planate branches without pseudocyphellae (Fig. 33e, f); and others have rather broad, contorted branches without pseudo-cyphellae and show circular scars (Fig. 33g, h). In addition, they have been collected in four islands of two different archipelagos. This is not a common pattern in the R. decipiens group, dominated by single-island endemics. We refrain from formally describing this lineage as a new species until more material is available.

The following key is presented to identify the species and un-described lineages in the R. decipiens group:

• 1. Thallus without cortex . . . . . . . . . . . . . . . . . . . . . . . . . . 2

• 1. Thallus with cortex . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

• 2. Thallus without a delimited holdfast, prostrate. Usually fertile species. Ascospores with thick walls. So far only found in the eastern Canary Islands . . . . . . R. pluviariae

• 2. Thallus with a delimited holdfast, pendulous. Rarely fertile species. Ascospores with thin walls. So far only found in Porto Santo . . . . . . . . . . . . . . . . . . . . . . . . R. nematodes

• 3. Soralia present. So far only found in Porto Santo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. erosa

• 3. Soralia absent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

• 4. Pycnidia present . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

• 4. Pycnidia absent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

• 5. Pycnidia with black ostioles . . . . . . . . . . . . . . . . . . . . . . 6

• 5. Pycnidia with pale ostioles . . . . . . . . . . . . . . . . . . . . . . 8

• 6. Medulla K-, C-, KC-, Pd- and UV+ blue-white (containing divaricatic acid) . . . . . . . . . . . . . . . . . . . . R. sampaioana

• 6. With different medullary substances . . . . . . . . . . . . . . . 7

• 7. Thallus sterile. Laciniae subterete, mostly matt, often eroded. Chondroid strands mostly dominating the cross-section. So far only found in Porto Santo R. portosantana

• 7. Thallus often fertile. Laciniae complanate, mostly shiny. Chondroid strands mostly adjoining the cortex. Widespread . . . . . . . . . . . . . . . . . . . . . . R. decipiens/Ramalina sp. 1 . . . . . . . . . . . . . . . . . . (see notes under Ramalina sp. 1)

• 8. Medulla K- (containing lecanoric, protocetraric or 4-O-demethylbarbatic acid) . . . . . . . . . . . . . . . . . . . . . . . . . 9

• 8. Medulla K+ yellow, then orange to red (containing salazinic acid) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

• 9. Medulla C+ and KC+ carmine red and Pd- (containing lecanoric acid) or C- and KC+ pink (reaction often weak and ephemeral) and Pd+ orange-red (containing protocetraric acid) . . . . . . . . . . . . R. maderensis (chemotypes I or II)

• 9. Medulla C+ and KC+ orange and Pd- (containing 4-O-demethylbarbatic acid) . . . . . . . . . . . . . . . . . . . . . . . . . . 10

• 10. Thallus matt. Cape Verde R. maderensis (chemotype IV)

• 10. Thallus shiny. Canary Islands . . . . . . . . . R. fortunatarum

• 11. Thallus mostly matt. Laminar pseudocyphellae usually along the entire length of the laciniae. Apothecia marginal, mostly spurred. Cape Verde . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. maderensis (chemotype III)

• 11. Thallus matt or shiny. Pseudocyphellae marginal, laminar or absent; when laminar rarely along the entire length of the laciniae. Apothecia marginal or terminal, mostly not spurred . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

• 12. Pycnidia tuberculate . . . . . . . . . . . . . . . . . . . . . . . . . . 13

• 12. Pycnidia not tuberculate . . . . . . . . . . . . . . . . . . . . . . . 15

• 13. Mostly sterile species. Laciniae with marginal pseudo-cyphellae. So far only found in El Hierro . . R. sabinosana

• 13. Species often fertile. Pseudocyphellae rare, when present laminar and usually near the base. Widespread . . . . . 14

• 14. Apothecia with a black ring around the hamathecium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. decipiens

• 14. Apothecia without a black ring . . . . . . . . Ramalina sp. 2

• 15. Thallus thin, paper-like, soft especially when moist. Laciniae abundantly fenestrated . . . . . . . . . . . . . . . R. papyracea

• 15. Thallus relatively thick and not paper-like. Laciniae not or sparsely fenestrated . . . . . . . . . . . . . . . . . . . . . . . . . . 16

• 16. Apothecia with a black ring around the hamathecium. Widespread . . . . . . . . . . . . . . . . . . . . . . . . . R. decipiens

• 16. Apothecia without a black ring. . . . . . . . . . . . . . . . . . . 17

• 17. Thallus often matt, thick, medulla almost completely filled by chondroid tissue, pseudocyphellae absent, apothecia 4–5 mm. So far only found in La Gomera . . R. gomerana

• 17. Thallus often shiny, not so thick, with chondroid tissue not so abundant, pseudocyphellae present or absent, apothecia 2–3 mm, present in the Canary Islands and the Madeira archipelago, thallus from La Gomera always shiny . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramalina sp. 2

• 18. Medulla K-, C-, KC-, Pd- and UV+ blue-white (containing divaricatic acid) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

• 18. With different medullary substances . . . . . . . . . . . . . . 20

• 19. Laciniae (10–)13–21(–25) mm long, subterete, with matt surface. Very fragile. So far only found in Lanzarote. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. delicata

• 19. Laciniae 29–64(–90) cm long, complanate, with shiny sur- face. So far only found in Porto Santo. . . R. sampaioana

• 20. Medulla K+ (containing salazinic acid) . . . . . . . . . . . . 21

• 20. Medulla K- (containing protocetraric, lecanoric or 4-O-demethylbarbatic acid) . . . . . . . . . . . . . . . . . . . . . . . . 26

• 21. Laminar pseudocyphellae usually along the entire length of the laciniae. Apothecia marginal, mostly spurred. So far only found in the Cape Verde archipelago. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. maderensis (chemotype III)

• 21. Pseudocyphellae marginal, laminar, or absent; when laminar rarely along the entire length of the laciniae. Apothecia marginal or terminal, mostly not spurred . . . . . . . . . . . 22

• 22. Laciniae with numerous divergent short branchlets terminating in hook-shaped structures or nodule R. hamulosa

• 22. Divergent branchlets absent . . . . . . . . . . . . . . . . . . . . 23

• 23. Laciniae thin, paper-like, soft especially when moist. Laciniae abundantly fenestrated . . . . . . . . . . . R. papyracea

• 23. Laciniae relatively thick and not paper-like. Laciniae not or sparsely fenestrated . . . . . . . . . . . . . . . . . . . . . . . . . . 24

• 24. Apothecia with a black ring around the hamathecium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. decipiens

• 24. Apothecia without a black ring around the hamathecium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

• 25. Laciniae often matt, thick, with chondroid tissue very abundant (> 50 % of the cross section), pseudocyphellae absent, apothecia 4–5 mm. So far only found in La Gomera . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. gomerana

• 25. Laciniae often shiny, not so thick, with chondroid tissue not so abundant (< 50 % of the cross section), pseudocyphellae present or absent, apothecia 2–3 mm, present in the Canary Islands and the Madeira archipelago, thallus from La Gomera always shiny . . . . . . . . . . . . . Ramalina sp. 2

• 26. Medulla C+ and KC+ carmine red and Pd- (containing lecanoric acid) or C- and KC+ pink (reaction often weak and ephemeral) and Pd+ orange-red (containing protocetraric acid). . . . . . . . . . . . . R. maderensis (chemotypes I or II)

• 26. Medulla C+ and KC+ orange and Pd- (containing 4-O-demethylbarbatic acid) . . . . . . . . . . . . . . . . . . . . . . . . . . 27

• 27. Thallus matt. So far found only in the Cape Verde archipelago . . . . . . . . . . . . . . . R. maderensis (chemotype IV)

• 27. Thallus shiny. So far found only in the Canary Islands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. fortunatarum

DISCUSSION

Our results show that the taxonomic diversity in the R. decipiens group has been highly underestimated. Widespread sampling coupled with thorough morpho-anatomical and chemical studies and the use of molecular markers has allowed us to recognize eight previously unrecognized species-level lineages, six of which are here described as new species. In addition, our approach has allowed a more natural and proper species delimitation in the group. Krog & Østhagen (1980a) had in fact reported high morphological and chemical plasticity in some of the species of the R. decipiens group. For example, they were aware of the high morphological variability in R. decipiens and R. maderensis, although they proposed a conservative treatment of this variability. According to our results, the newly described R. gomerana, R. papyracea and R. sabinosana, as well as the two undescribed lineages, would be nested within the previous concept of R. decipiens and R. fortunatarum within R. maderensis. In recent decades, an increasing number of studies have shown that there are a large number of unknown fungal species hiding under the names of widely distributed species (e.g., Crespo & Lumbsch 2010, Simon et al. 2018, Moncada et al. 2021). This has been reported in various species complexes in Ramalina, such as the R. americana ( LaGreca 1999) or R. siliquosa ( LaGreca et al. 2020) groups (but see Stocker-Wörgötter et al. 2004). In spite of the split of newly described species previously included in the concepts of R. decipiens and R. maderensis, these two species continue to be the most widely distributed species of the group, occurring in all the archipelagos and basically all the studied islands. Our phylogenetic data revealed a moderate level of geographic structure underlying these large distributions, a fact that could indicate that these species may be undergoing differentiation due to a lack of gene flow. This possibility was already suggested by Krog & Østhagen (1980b) and will be explored in a future study. Even after the split of the newly described species that were nested inside the previous species concepts, R. decipiens and R. maderensis remain the most morphologically plastic species of the group. Their variability, however, is not clearly correlated with the geographic structure observed in the phylogeny, which prevented us from raising their morphotypes to the species level. Muggia et al. (2014) found a similar issue in Tephromela atra, a widespread species in which they found fifteen intraspecific lineages, some of them well-supported but lacking enough morphological and chemical differentiation to warrant formal descriptions. Cryptic species lineages that can be separated using molecular data but lack diagnostic characters are common in lichen-forming fungi (Crespo & Pérez-Ortega 2009, Divakar et al. 2010, Spribille et al. 2011, Pino-Bodas et al. 2012, Haugan & Timdal 2019). It is worth mentioning that even though some of the diagnostic characters are often subtle, we consider that there are no cryptic species in the R. decipiens group, with the only exception of the undescribed Ramalina sp. 1, which we were not able to differentiate either morphologically or chemically from R. decipiens. Given that Ramalina sp. 1, however, is represented in our dataset by just two specimens, this must be interpreted with caution.

Nevertheless, the absence of cryptic species does not mean that taxonomy and species delimitation in the group is straightforward. In addition to the large plasticity typical of the genus Ramalina and closely related fruticose Ramalinaceae ( Spjut 1996), we consider that the problem of species delimitation in this group is perhaps due to the fact that it represents a relatively recent radiation. The process of speciation is quantitative in nature. It does not happen instantly; but as the result of a more-or-less gradual accumulation of genetic differences and reproductive isolation ( Hendry et al. 2000, Mallet 2008, Nosil et al. 2009). Because of this, it is often difficult to determine when speciation is ‘complete’. It can be more useful to think of it as a gradient of variation between panmictic populations and distinct, reproductively isolated species. This gradient has been referred to as the ‘speciation continuum’ ( Nosil 2012) and can be roughly divided in four stages ( Hendry 2009). These are: 1) continuous variation within a group with no reproductive isolation; 2) partially discontinuous variation with minor reproductive isolation; 3) strong discontinuous variation and strong but potentially reversible reproductive isolation; and 4) complete speciation, with irreversible reproductive isolation. The barriers between these stages are not always clear and transitions between them are not necessarily unidirectional. In recent radiations the lineages under study are usually in the earlier stages of the speciation continuum, with evidence of incomplete reproductive isolation between them (e.g., Salzburger et al. 2002, Grant & Grant 2006, Wegener et al. 2019).

The ‘speciation continuum’ ( Nosil 2012), usually coupled with a background of low extinction ( Rabosky & Lovette 2008, Rabosky & Glor 2010), lies at the root of the challenging taxonomy of recent radiations. Taxonomic issues exist even in textbook examples that have been studied in depth, such as the Anolis lizards and the Darwin finches. Species in Anolis are understood in a wide sense and many of them show intraspecific genetic variation comparable to that found between species ( Losos 2009: 306), with some of the subclades being congruent with a previously rejected taxonomic proposal ( Losos 2009: 308). Species boundaries in the Darwin finches were recently evaluated with genomic data (Zink & Vázquez-Miranda 2019) and they remain unclear, particularly in the genera Geospiza and Camarhynchus. The problems experienced in the molecular delimitation of species in the R. decipiens group are not unique within lichen-forming fungi and although our data set may be limited in terms of the number of molecular markers used, the problems may not be fully solved by increasing the number of analyzed loci. Jorna et al. (2021) recently examined species boundaries in Niebla with genome-scale data. Niebla is a fruticose genus of the Ramalinaceae that, although it does not represent an insular radiation, is well known to have a complex evolutionary history and puzzling taxonomy ( Spjut 1996, Jorna et al. 2021). Jorna et al. (2021) found that even using a molecular approach that utilized tens of thousands of RADseq loci failed to unequivocally delimit species. The speciation continuum is probably also at the root of the overall poor results of the automatic species discovery strategies we used in this study. Automatic species discovery strategies based on single- locus datasets such as ABGD, ASAP, GMYC or bPTP have been widely used to address the taxonomy of diverse groups of lichen-forming fungi ( Kraichak et al. 2015, Del-Prado et al. 2016, Pérez-Ortega et al. 2016, Simon et al. 2018). It is known that these automatic strategies have limitations. For instance, GMYC shows a tendency for oversplitting, while ABGD handles singleton sequences poorly ( Pentinsaari et al. 2017). We have found that for algorithms based on pairwise genetic distances, such as ABGD and ASAP, the speciation continuum translates into a lack of the barcode gap in most markers, which makes their results unreliable. Also, algorithms based on identifying differences in the branching rates of a phylogenetic tree, such as GMYC and bPTP, produce an unrealistically high number of putative species under most DNA regions. This is most likely influenced by the group lineages being in the earlier stages of the speciation continuum, as these methods are negatively influenced by gene flow ( Luo et al. 2018). In agreement with previous studies ( Klimov et al. 2019), we found that species discovery strategies based on multilocus data returned more sensible species hypotheses. This was especially true for STACEY, maybe because its ability to factor in evolutionary processes such as incomplete lineage sorting could help it to overcame some of the challenges posed by the speciation continuum. The problem with the species discovery strategies based on phenotypic and geographical data alone stems not so much from the speciation continuum itself as from the fact that they are unable to weight the importance of the different traits ( Ding & He 2004, Murtagh & Contreras 2017). In other systems that have fewer taxonomic characters, like microalgae, these methods reach more satisfactory results, but their performance has been shown to still be inferior to that of supervised approaches ( Salvesen et al. 2020).

The speciation continuum has prevented the generation of robust taxonomies in evolutionarily complex lineages using reduced molecular data sets (e.g., Monaghan et al. 2006) or even employing large genomic scale data (e.g., Zink & Vázquez-Miranda 2019), which may discourage the undertaking of studies aimed at exploring the factors underlying such complexity ( Raposo et al. 2021). The species has been regarded as the basic unit of biology ( Mayr 1944). Numerous species concepts have been proposed ( Zachos 2016), among which the unified species concept is the most used in lichenology ( Lücking et al. 2021). Under the unified species concept, which defines species as separately evolving segments of metapopulation lineages ( De Queiroz 1999), species are delimited based on a common shared ancestry and the accumulation of supportive evidence from independent data sources ( De Queiroz 2011). Difficulties arise from the operational criteria used to apply the concept in lineages such as recent evolutionary radiations, in which speciation may not be ‘complete’ yet, barriers for gene flow may be weak and hybridization among species can be common ( Salzburger et al. 2002, Palmer & Kronforst 2015), in addition to common sharing of alleles due to incomplete lineage sorting ( Koblmüller et al. 2010). Some authors have argued that, in the face of these problems, the most practical solution is a DNA taxonomy independent of the Linnaean naming system that could be used as a reference system to integrate ecological and other biodiversity data ( Tautz et al. 2003). We, however, agree with Hawksworth (2020) and Lücking et al. (2021) in that taxonomic ranks are useful because they facilitate communication between researchers and a wider public, such as conservation stakeholders, about any given taxa and that a taxonomist's task is to identify and utilize the most objective criteria to reach practical solutions. In consequence, we have carried out an in-depth revision of specimens searching for diagnostic characters. Although it is a relatively common practice in the literature not to make taxonomic proposals when studying problematic and complex groups of species, we have chosen to propose a new taxonomic framework for the R. decipiens group, basing the species boundaries in phylogenetically supported clades and the integration of geography, chemistry and morphological characters. In agreement with Dufresnes et al. (2023), our intention is that this will facilitate communication on relevant taxa, especially the narrow endemic species, that may need priority conservation actions, and make future studies in the R. decipiens group more feasible.

As here circumscribed, most of the species of the R. decipiens group represent narrow endemics. Traditionally, lichen-forming fungi have been thought to display widespread distribution ranges, especially compared to other sessile organisms such as vascular plants. However, the use of molecular approaches to species delimitation has brought the existence and abundance of endemic species into focus. New endemics have been found in oceanic islands in the genus Nephroma ( Sérusiaux et al. 2011), Pseudocyphellaria ( Moncada et al. 2014), Cora ( Dal Forno et al. 2017), Cyphellostereum ( Dal Forno et al. 2017), Dictyonema ( Dal Forno et al. 2017) and Sticta ( Simon et al. 2018, Moncada et al. 2021). This pattern is not restricted to islands, however, as continental endemics have also been described in the genera Rhizoplaca ( Leavitt et al. 2013), Cora ( Lücking et al. 2014) and Parmelia ( Crespo et al. 2020).

Endemic species are certainly abundant in the genus Ramalina compared to other widely distributed genera with high species diversity and this tenet is especially true for oceanic islands. Endemic Ramalina species other than those of the R. decipiens group are known from oceanic islands in many archipelagos: Tristan da Cunha (one species, Jørgensen 1977), Cape Verde (one species, Arechavaleta Hernández et al. 2005), the Canary Islands (11 species, Krog & Østhagen 1980a, Pérez-Vargas & Pérez-Ortega 2014), Madeira (three species, Krog & Østhagen 1980b, Krog 1990), Polynesia (one species, Molho et al. 1981), Hawaii (18 species, www.anbg.gov.au/abrs/lichenlist/HAWAIIAN_ISLANDS_lichen_list.html ), Galápagos (four species, Aptroot & Bungartz 2007), Azores (two species, Aptroot & Schumm 2008), St. Helena (five species, Aptroot 2008), the Greater Antilles (three species, Landrón 1972) and the Scattered Islands (three species, Poncet et al. 2021). Other genera of fruticose Ramalinaceae also show high levels of endemism, with the case of Niebla in Baja California being particularly astonishing ( Spjut et al. 2020, Jorna et al. 2021). As stated above, the taxonomy of this genus is problematic and molecular data suggest that the number of species is larger than previously thought ( Spjut et al. 2020, Jorna et al. 2021). To explain this, a micro-endemism scenario has been proposed in which species with extremely narrow distributions develop similar morphologies through convergent evolution along the coastal fog dessert of the region ( Spjut et al. 2020).

The results obtained in this study raise the need for conservation policies for some species of the R. decipiens group, especially the narrow endemics. Narrow endemics are inherently vulnerable due to their limited geographic ranges and often have small and isolated populations, which usually translates into low levels of genetic diversity ( Ellstrand & Elam 1993, Gitzendanner & Soltis 2000) that can be further reduced by their low competitive ability ( Lavergne et al. 2004) and high habitat specificity ( Babbel & Selander 1974, Karron 1987). Because of this they are commonly included in conservation priority assessments (e.g., Wulff et al. 2013). Sparrius et al. (2017) evaluated the population size and threats of some of the narrow endemic species of the R. decipiens group occurring exclusively in Porto Santo (i.e., R. erosa, R. nematodes and R. portosantana). Using the IUCN criteria, these authors classified all of them as critically endangered.

CONCLUSIONS

The current update of the taxonomy of the R. decipiens group in Macaronesia has resulted in the recognition of 15 lineages at the species level, six of which are described as species new to science. The endemic character of the R. decipiens group in Macaronesia makes it an excellent model system for the study of the process of evolutionary radiations in lichen-forming fungi on oceanic islands.

Fig. 19.

Ramalina maderensis collected in Fuerteventura, Canary Islands (TFC-Lich 12087). Macroscopic and microscopic characters. a. Habit; b, c. laciniae detail; d. pseudocyphellae; e. apothecium; f. pycnidia; g. thallus anatomy; thalline layers are indicated as C (cortex), Ch (chondroid tissue) and M (medulla). — Scale bars: a = 2 cm, b–f = 1 mm, g = 200 µm.

Supplementary material

Fig. S1

Distribution map of Ramalina decipiens, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S2

Distribution map of Ramalina delicata, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S3

Distribution map of Ramalina erosa, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S4

Distribution map of Ramalina fortunatarum, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S5

Distribution map of Ramalina gomerana, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S6

Distribution map of Ramalina hamulosa, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S7

Distribution map of Ramalina maderensis, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S8

Distribution map of Ramalina nematodes, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S9

Distribution map of Ramalina papyracea, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S10

Distribution map of Ramalina pluviariae, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S11

Distribution map of Ramalina portosantana, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S12

Distribution map of Ramalina sabinosana, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S13

Distribution map of Ramalina sampaioana, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S14

Distribution map of Ramalina sp. 1, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S15

Distribution map of Ramalina sp. 2, showing occurrence records on the (A) Azores, (B) Canary, (C) Madeira and (D) Cape Verde archipelagos.

Fig. S16

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the ITS region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Fig. S17

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the UGMR7_22 region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Fig. S18

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the UGMR33_20 region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Fig. S19

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the UGMR70_14 region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Fig. S20

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the UGMR197_101 region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Fig. S21

Phylogenetic tree inferred by maximum likelihood and Bayesian analyses of the efa region. Branches with bootstrap values ≥ 70 and/or posterior probability ≥ 0.95 are highlighted in bold.

Table S1

Specimens selected for the molecular study, including GenBank accession numbers.

Table S2

Parameters used in the DOMINO analysis.

Table S3

Input table used on the “Unsupervised clustering of morphological, chemical and geographic data" analysis.

Table S4

Test for a strict molecular clock for each locus conducted in MEGA 5.2.2. *denotes rejection of the null hypothesis (i.e., equal rates).

Table S5

Results of the species discovery strategies based on genetic distances (ABGD and ASAP) and phylogenetic trees (GMYC and bPTP) for the six markers.

Table S6

Polymorphism statistics and neutrality tests.