Xylopsora canopeorum (Umbilicariaceae), a new lichen species from the canopy of Sequoia sempervirens

Abstract Xylopsora canopeorum Timdal, Reese Næsborg & Bendiksby is described as a new species occupying the crowns of large Sequoia sempervirens trees in California, USA. The new species is supported by morphology, anatomy, secondary chemistry and DNA sequence data. While similar in external appearance to X. friesii, it is distinguished by forming smaller, partly coralloid squamules, by the occurrence of soralia and, in some specimens, by the presence of thamnolic acid in addition to friesiic acid in the thallus. Molecular phylogenetic results are based on nuclear (ITS and LSU) as well as mitochondrial (SSU) ribosomal DNA sequence alignments. Phylogenetic hypotheses obtained using Bayesian Inference, Maximum Likelihood and Maximum Parsimony all support X. canopeorum as a distinct evolutionary lineage belonging to the X. caradocensis–X. friesii clade.


Introduction
The squamulose lichen genus Xylopsora Bendiksby & Timdal consists of two species, X. caradocensis (Nyl.) Bendiksby & Timdal and X. friesii (Ach.) Bendiksby & Timdal. The two species were formerly placed in Hypocenomyce M. Choisy and referred to as the H. friesii group (Timdal 1984(Timdal , 2001 until Bendiksby and Timdal (2013) showed that Hypocenomyce was highly polyphyletic. Xylopsora is the phylogenetic sister of the clade consisting of the two foliose genera Lasallia Mérat and Umbilicaria Hoffm. Those three genera make up the sister clade of the genus Fulgidea Bendiksby & Timdal, another Hypocenomyce segregate. The four genera together constitute the Umbilicariaceae (Bendiksby and Timdal 2013). Fulgidea consists of two species, F. oligospora (Timdal) Bendiksby & Timdal and F. sierrae (Timdal) Bendiksby & Timdal. Fulgidea and Xylopsora are morphologically, anatomically and ecologically very similar and differ mainly in secondary chemistry; alectorialic acid and thamnolic acid occur in the former, friesiic acid (= "friesii unknown") in the latter (Timdal 2001(Timdal , 2002. All species of Fulgidea and Xylopsora grow on bark and wood and, with the exception of X. caradocensis, show preference for burnt stumps and trunks of conifers. Coast redwood (Sequoia sempervirens) forests are an important component of California's ecosystems. Spanning more than six degrees of latitude along the Pacific coast (Van Pelt 2001) and containing individual trees that can live for more than 2000 years (Rocky Mountain Tree Ring Research 2017), these forests provide important habitats for many terrestrial species (Sawyer et al. 1999). However, biodiversity occupying the redwood forest canopies remains relatively under-explored because access into the tree crowns, which often grow to over 100 m in height, is challenging. The epiphytic lichen flora in oldgrowth redwood forests appears to be particularly species rich; an epiphyte survey of just nine large redwood trees yielded 137 lichen species including a new species of Calicium Sillett 2007, Williams andTibell 2008).
Recent epiphyte surveys in the crowns of additional large coast redwood trees in the southern part of the geographic range (Reese Naesborg 2017) revealed a previously undescribed species of Xylopsora. Here the authors have provided detailed morphologic, anatomic, chemical and molecular description of this new species, as well as characterising the habitat and substrates it occupies.
Establishing a multiple DNA sequence alignment (MSA) of non-coding loci, which often have unequal lengths due to indels, can be both time-consuming and highly subjective with regard to structural correctness. There has been great activity in recent years in the development of multiple sequence alignment tools (reviewed by Kamena and Notredame 2009). Moreover, there is no single recommendation as to what phylogenetic algorithm to use to transform the MSA into a reliable phylogenetic hypothesis. The current, moderately sized dataset has been used to test whether a less time-consuming and more objective approach, SATé-II (Liu et al. 2012), provides similar and meaningful results. Both manual and automatic approaches have been used to establish a concatenated MSA of three loci, of which one is non-coding and highly variable (nrITS). Different methods representing three classes of phylogenetic inference (Bayesian, likelihood and parsimony) have also been used.

The specimens
Five specimens of an unknown Xylopsora species were collected from Sequoia sempervirens trees in the southern part of the geographic range of coast redwood. The new species was documented on five trees in Big Basin Redwoods State Park, Santa Cruz County and on another five trees in Armstrong Redwoods State Natural Reserve, Sonoma County, California. The morphology, anatomy, chemistry and DNA sequences of these newly collected specimens have been studied and then compared to existing descriptions of Xylopsora and relatives (Bendiksby and Timdal 2013). A total of 50 accessions (Umbilicariales + Fuscideaceae) and their respective DNA sequences were reused from Bendiksby and Timdal (2013). Vouchers of the newly collected specimens are deposited at JEPS, NY and O.

Anatomy
Microscope sections were cut on a freezing microtome and mounted in water, 10 % KOH (K), 50 % HNO 3 (N), lactophenol cotton blue and a modified Lugol's solution in which water was replaced by 50 % lactic acid. Amyloid reactions were observed in the modified Lugol's solution after pretreatment in K. Ascospore measurements are given as X ± 1.5×SD, rounded to 0.5 μm, where X is the arithmetic mean and SD the standard deviation.

Secondary chemistry
Thin-layer chromatography (TLC) was performed in accordance with the methods of Culberson (1972), modified by Menlove (1974) and Culberson and Johnson (1982).

DNA extraction, PCR, and sequencing
DNA was extracted from the apothecia of four of the five newly collected specimens. The DNA extraction, PCR amplification (nrITS and mtSSU), PCR product purification, cycle sequencing and DNA sequence assembly and editing were performed as described by Bendiksby and Timdal (2013), including a subset of the oligonucleotide primers used (i.e. the forward primers ITS5 and mtSSU1 and the reverse primers ITS4 and mtSSU3R). The four DNA isolates were deposited in the DNA collection at O (Natural History Museum, University of Oslo).

DNA sequence analysis
The newly produced DNA sequences (mtSSU and nrITS) were aligned manually using BioEdit 7.2.3 (Hall 1999) into a trimmed version of the DNA sequence alignments used by Bendiksby and Timdal (2013). The resultant concatenated alignment comprised three genetic regions (nrLSU, mtSSU and nrITS) and a subset of 54 accessions representing the Elixiaceae, the Fuscideaceae, the Ophioparmaceae and the Umbilicariaceae. In addition to this alignment, hereafter referred to as "MSAmanual", the software SATé-II version 2.2.7 (Liu et al. 2012) was also used to establish an automated alignment, referred to as "MSAsate". Both alignments were analysed phylogenetically using Bayesian Inference (BI), Maximum Likelihood (ML) and Maximum Parsimony (MP) algorithms. The outgroup was defined as a clade consisting of two accessions representing the Fuscideaceae, which were also used for rooting. SeqState v.1.36 (Müller 2005) was used to convert alignments between different formats and FigTree 1.4.0 (Rambout 2006-2012) for visualising and editing output trees.
The BI analyses were performed as described in Bendiksby and Timdal (2013), but with only six million generations due to the smaller dataset. All trees saved prior to the point where the average standard deviation of split frequencies (ASDSF) fell below 0.01 were discarded as burn-in. For the sake of comparability of results, the evolutionary models GTR+G were used for both loci in the BI analysis (only a limited number of evolutionary models are available in SATé-II).
The software SATé-II simultaneously estimates multiple sequence alignments and ML phylogenetic trees. Prior to analyses, MSAmanual was divided into non-orphan (no empty sequences), single-locus datasets and were de-aligned (i.e. all gaps deleted). The MSAsate and its corresponding ML tree were estimated as a multilocus dataset in SATé-II using MAFFT (Katoh et al. 2005, Katoh andToh 2008) as the aligner, MUSCLE (Edgar 2004a, b) as the merger and RAxML v.7.2.8 (Stamatakis 2006) as the tree estimator with the GTRCAT model. The limit of iterations was set to 50 and otherwise default settings were used. For comparison, MSAmanual was also analysed phylogenetically using SATé-II under the same settings.
For the MP analyses, NONA (Goloboff 1999) was used in combination with Win-Clada 1.0 (Nixon 1999), applying the heuristic search option with 2000 replicates and maxtrees set to 10000 and otherwise default settings. Parsimony jack-knifing (JK; Farris et al. 1996) with 2000 replicates was performed and otherwise default setting. Parsimony jackknifing was also performed on single-locus datasets for assessing potential gene-tree incongruence prior to estimating phylogenetic hypotheses based on all three loci.

Results
Four nrITS and three mtSSU sequences were generated (GenBank accession numbers MG309307-MG309313; Table 1). Preliminary parsimony jack-knife analyses of the individual three loci, regardless of the alignment approach, produced congruent gene- trees that were resolved to various extents (not shown). In subsequent analyses, the three loci were analysed in concert. The MSAmanual alignment (i.e. three loci, manually aligned) was 10 characters shorter than MSAsate (i.e. three loci, automatically aligned) and had 12 fewer parsimony informative characters (PIC; Table 2). Both alignments are provided as Suppl. material 1, 2 (MSAmanual.nex, MSAsate.nex). The MSAmanual dataset produced 1220 most parsimonious trees (MPTs) of length 1479, whereas MSAsate produced 10 MPTs of length 1485. Homoplasy measures (Farris 1989) differ negligibly between the two (RC: 46.6 vs 46).The likelihood scores from the RAxML analyses in SATé-II were very similar ( Table 2). The ASDSF fell below 0.01 faster in the BI analysis of MSAsate (at generation 820) than in the BI analysis of MSAmanual (around generation 1300). All significantly supported clades were congruent amongst the BI, ML and MP analyses, regardless of the dataset analysed (MSAmanual vs MSAsate). Only results from analyses of the MSAsate dataset are shown (Fig. 1). The authors regarded clade support of at least 60% jack-knife (JK) and at least 0.9 posterior probability (PP) as significant.  The four accessions of the tentatively new species group with significant support and showed themselves as sister to a clade consisting of three accessions of Xylopsora friesii (2, 3 and 4; Fig. 1). Four characters varied amongst the four accessions of the tentatively new species (three in the nrITS and one in mtSSU), none of which were parsimoniously informative within the group. Xylopsora cf friesii 1 differed from X. friesii 2, 3 and 4 in eight characters in the nrLSU, at least four in the mtSSU and at least 16 in the nrITS. The MP analyses supported Elixia as monophyletic (JK MSAmanual = 94 %; JK MSAsate = 87 %) with Meridianelia as sister (JK MSAmanual = 95 %; JK MSAsate = 93 %). In the ML and BI analyses, Elixia monophyly was not supported, as Meridianelia grouped with accessions of E. flexella and Elixia sp. and excluded Elixia cretica. This topology was not significantly supported by any analyses. The sister-relation between the Elixiaceae and the Ophioparmaceae was significantly supported only by Bayesian PP (Fig. 1).

Discussion
Forest canopies in general are relatively understudied because accessing the tree crowns requires technical expertise and equipment (Lowman et al. 2012). Therefore, the potential for encountering new species is relatively high compared to more easily accessible forest floor environments. The new species presented here, Xylopsora canopeorum, has so far only been registered from the crowns of large coast redwood trees, but other similar habitats, like the fibrous bark of other large members of Cupressaceae, should be explored for the species. The collected Xylopsora canopeorum specimens occurred on stable bark surfaces of old, large redwood trees together with several species previously classified in the genus Hypocenomyce (e.g. Carbonicola anthracophila, Fulgidea oligospora, F. sierrae, and H. scalaris), which was recently shown to be highly polyphyletic (Bendiksby and Timdal 2013). Together, these species covered a substantial portion of the trunk surface. Figure 1. Hypothesis of the phylogenetic relationships and placement of the potentially undescribed species of Xylopsora based on DNA sequence data. The depicted topology is based on an automated alignment (MSAsate) of two nuclear (ITS and LSU) and one mitochondrial (SSU) ribosomal loci and is the "best tree" from a RAxML analysis using SATé-II. Clade support over certain values from Bayesian inference (posterior probability; PP) and parsimony jackknifing (JK) analyses are superimposed: PP >0.9 and JK>60% (PP/JK). Clades receiving maximum PP support (1.0) and at least 90% JK support are indicated with a black dot. Multiple accessions of the same taxon are numbered according to Table 1 (also corresponding to the numbering in Bendiksby & Timdal 2013). Family and genus circumscriptions are indicated. Abbreviation: Fusc. = Fuscideaceae. One accession in the tree (Meridianelia maccarthyana) appeared on a very long branch that is manually shortened (arrow tipped) to reduce the size of a broad figure.

Molecular analyses
Automatic alignment by SATé-II differed only slightly from the manually aligned dataset. Only areas with ambiguous alignment solutions varied between the manually aligned multi-locus alignment (MSAmanual) and the one aligned automatically (MSAsate). Moreover, the two alignments rendered highly similar topologies when analysed using the same algorithm. MSAsate contained slightly more parsimony phylogenetic information and produced fewer MPTs. Although the different algorithms produced variously resolved trees, the same significantly supported clades were present in all output trees. This suggests significant time-savings by using SATé-II and software of similar quality for both automated alignment and phylogenetic analyses.
As expected, the overall tree-topology ( Fig. 1) largely corresponded to previous findings (Bendiksby and Timdal 2013: fig. 2A), the only exception being that Elixia monophyly was not supported by the BI and ML analyses and supported only by MP (JK MSAmanual = 94 %; JK MSAsate = 87 %). The grouping of Meridianelia with accessions of E. flexella and Elixia sp., however, was not significantly supported by any analyses (Fig. 1) and was not considered of taxonomical significance. More importantly, monophyly of the four newly included accessions was significantly supported regardless of alignments or analyses algorithm. Likewise, this clade's sister relation to X. friesii was significantly supported. The low and non-informative genetic variation between the four newly included accessions strongly suggests they belong to a single species. The specimen Xylopsora cf friesii 1, on the other hand, differed significantly from X. friesii 2, 3 and 4. It is hypothesised that X. cf. friesii 1 represents a species distinct from X. friesii, but more material will need to be studied prior to drawing additional taxonomic conclusions.

Taxonomy
Xylopsora canopeorum Timdal, Reese Naesborg & Bendiksby, sp. nov. Mycobank: MB823500 Fig. 2 Diagnosis. The species differs from X. caradocensis and X. friesii mainly in forming more minute, coralloid and sometimes, sorediate squamules and sometimes (the holotype) in containing thamnolic acid in addition to friesiic acid; it also differs from the former in having shorter, non-septate ascospores.
Distribution. Specimens were collected from central coastal California in Big Basin Redwoods State Park (37.1°N, 11 km from the Pacific Ocean) and Armstrong Redwoods State Natural Reserve (38.3°N, 18 km from the Pacific Ocean).
Ecology. Xylopsora canopeorum was observed on coarse, fibrous bark and occasionally on charred bark between 5 and 75 m above ground level along the trunks of large coast redwood trees in old-growth redwood forests. The species commonly co-occurred with Carbonicola anthracophila, Fulgidea oligospora, F. sierrae, Hertelidea botryosa and Hypocenomyce scalaris, which together covered substantial portions of the trunk surface. Xylopsora canopeorum appeared to have an affinity for old and stable bark surfaces on the main trunks of large redwood trees.
Etymology. The specific epithet "canopeorum" refers to the habitat in which the species was encountered ¾ in the canopy of old-growth redwood forests.
In the current Californian lichen checklist (Tucker 2014), Lecidea xanthococcoides Zahlbr. is the only species unknown to the authors that could be assumed to be an earlier name for X. canopeorum. That species was described from conifer trunks at 1700 m alt. in the San Bernardino Mountains, i.e. in an area and habitat where X. canopeorum possibly can occur. The holotype (H.E. Hasse 705) was not found in W upon enquiry. Details in the original description (Zahlbruckner 1900) indicate that it is a different species, however -Apothecia becoming convex and immarginate, hymenium 160-180 μm high and ascospores 12-15 × 5.5-6 μm.
Additional specimens examined. USA. California. Santa Cruz Co.: label data as for holotype, R. Reese Naesborg 1544 (NY); 800 m WNW of North Escape Road up