Macro- and microevolutionary perspectives on Seynesiella juniperi, a fungus in the Venturiales (Dothideomycetes, Ascomycota)

The present work represents the first comprehensive phylogenetic study of the dothideomycete genus Seynesiella. The genus belongs into the family Cylindrosympodiaceae within the order Venturiales, based on a phylogeny reconstructed with five loci. The high genetic diversity found within the type species, S. juniperi, points towards cryptic speciation, with up to five distinct species that might be associated to different Juniperus hosts. Combining phylogenetics and multi-locus delimitation analyses, together with more detailed measurements of ascospores, will be fundamental for a better understanding of species boundaries and the biogeographic history of the delimited species, as well as for revealing more specific fungal-plant association patterns.


Introduction
The largest class of Kingdom Fungi is Dothideomycetes (Ascomycota), with ca. 19000 species that are distributed in two subclasses (Pleosporomycetidae and Dothideomycetidae), 1261 genera, 110 families and 23 orders (Wijayawardene et al. 2017). Members of this class are characterized by producing bitunicate asci with fissitunicate dehiscence and are extremely diverse ecologically, as they comprise endophytic, epiphytic, saprobic, phytopathogenic, zoopathogenic (including human pathogenic), lichenized, lichenicolous, nematode trapping, and rock-inhabiting taxa (Schoch et al. 2009;Hyde et al. 2013;Ametrano et al. 2019;Hongsanan et al. 2020). The number of known taxa is increasing at a steady pace as more geographic areas and habitats are investigated (e.g. Valenzuela-López et al. 2018;Pem et al. 2019;Hongsanan et al. 2020a). For example, a recent study on Venturiales Y. Zhang, C.L. Schoch & K.D. Hyde, which encompasses saprobes and pathogens of plants and animals (including humans) analyzed 115 taxa representing 30 genera and introduced a new family, eight new genera, and 12 new species (Shen et al. 2020). The use of molecular data within a multi-locus phylogenetic context has been crucial to revealing most of the known diversity. Recently, whole-genome sequence data generated for more than one hundred dothideomycete species have been used to refining phylogenetic relationships and understanding other evolutionary aspects, such as adaptation to stress, host specificity and the origin of pathogenicity (Haridas et al. 2020). Although these advances have allowed the resolution of several long-standing questions on Dothideomycetes diversity and evolution, many questions remain open about the evolutionary history of particular taxa, especially genera and species for which molecular data are not yet available.
One such genus is Seynesiella G. Arnaud, currently comprising five species worldwide (Sivanesan & Shivas 2002) and whose type species is S. juniperi (Desm.) G. Arnaud (Arnaud 1918). Ascomata of this genus are superficial thyriothecia, more or less conical to hemispheric, tapered with a rounded ostiole, and with a peridium formed by isodiametric cells towards the center and more rectangular cells towards the periphery; microscopically, asci produce eight spores, spores are bicellular, smooth, hyaline first and turning brownish at maturity. Although Müller & von Arx (1962) placed the genus in Venturiaceae E. Müll. & Arx ex M.E. Barr, other authors suggested its placement in Microthyriaceae Sacc. based on ascoma morphology and peridium structure (Luttrell 1965;Barr 1968;Lumbsch & Huhndorf 2010). A close relationship with the genera Tothia Bat. and Arnaudiella Petr. was also hypothesized (Wu et al. 2011). So far, the absence of molecular data for any of the known species in the genus has prevented us from answering these questions.
The species Seynesiella juniperi seems to be widespread across the Northern Hemisphere (GBIF 2019), from Ontario, British Columbia (Canada) and Wisconsin (USA) growing on Juniperus communis L., J. communis var. depressa Pursh and J. scopulorum Sarg. (Greene 1967(Greene -1968Wu et al. 2011;Lee & Golinski 2020) to the United Kingdom, France and Scandinavia, growing on several Juniperus spp., and even Iran (Petrak 1949). In the Iberian Peninsula, there are a considerable number of occurrence reports from Catalunya and Huesca (Sierra López 2006) and Segovia (Olariaga 2018), growing on J. communis, J. phoenicea L. and J. thurifera L. The question arises whether this fungal species occurs in other areas of the Iberian Peninsula and on additional Juniperus L. hosts, such as J. oxycedrus L. and J. sabina L., and also whether there is any signal of genetic differentiation between specimens growing on different hosts.
The present study first proposes a macroevolutionary framework to test the placement of the genus Seynesiella in known high level taxa within Dothideomycetes using molecular data from several loci, and second, a microevolutionary framework that uses morphological and ecological traits, as well as phylogenetics and species delimitation analyses, to examine genetic differentiation in S. juniperi and determine whether there are signs of concordant evolutionary history between the fungi and their plant hosts.

Morphological studies
Leaves of Juniperus spp. were examined under a Leica S8APO dissecting microscope, and macroscopic photographs of ascomata were taken with a Leica EC3 image capture system. Preparations of entire ascomata were mounted in water, and microscopic details were observed using a Zeiss Axioplan 2 microscope fitted with Nomarski differential interference contrast (DIC). Photographs were taken with a Zeiss AxioCam digital camera. Microscopic measurements were made by means of the Zeiss Axiovision 4.8 image analyzer system and are given as the average and its standard deviation. Specimens are deposited in the Real Jardín Botánico of Madrid (MA herbarium).

DNA extraction, PCR amplification and DNA sequencing
Up to five ascomata per leaf showing no contamination by hyphae from other fungi were isolated from each specimen and deposited into an Eppendorf tube. After grinding the ascomata with the help of micropestles, genomic DNA was extracted using the Speedtools Tissue DNA kit (Biotools® B&M Labs., S.A) following the manufacturer's recommendations. To reconstruct a phylogeny of orders in Dothideomycetes, the following five markers were selected and amplified by PCR: the nuclear small (nuSSU) and large (nuLSU) subunits of the ribosomal RNA, the RNA polymerase II largest (RPB1) and second largest (RPB2) subunits, and the elongation factor 1-alpha (tef1-α) gene. The nuclear ribosomal Internal Transcribed Spacer (nrITS), which includes the subregions nrITS1, 5.8S, and nrITS2, was additionally amplified to examine intraspecific variation in the focal species. This marker has been generally used for barcoding fungi due to its high nucleotide variability (Schoch et al. 2012); for the same reason, the nrITS is also useful for studying variation below the species level. The primers used as well as PCR conditions were the same as in Pérez-Ortega et al. (2016). PCR reactions were carried out using the Illustra Ready-To-Go GenomiPhi V3 DNA amplification kit (GE Healthcare Bio-Sciences, Pittsburgh, Pennsylvania, USA) following the manufacturer's instructions. Amplicons were purified and cleaned using the QIAGEN quick spin columns (Qiagen ® ), and both complementary DNA strands were sequenced at MACROGEN (Madrid). Raw electropherograms were manually checked, trimmed and assembled using SeqmanII v.5.07 © (Dnastar Inc.). GenBank accession numbers are provided in section Results.

Phylogenetic analyses at ordinal level
Following previous considerations of the genus Seynesiella being a member of the ascomycete class Dothideomycetes (Lumbsch & Huhndorf 2010;Wu et al. 2011;Hyde et al. 2013;Wijayawardene et al. 2017;Hongsanan et al. 2020b), a five-locus dataset was compiled with publicly-available DNA sequence data for 86 species representing most known orders in Dothideomycetes (Zhang et al. 2011 (Katoh et al. 2002;Katoh & Standley 2013). The resulting alignments were visualized in Geneious ® v.9.0.2 to identify and manually remove in all five markers intronic regions, which were the ones that encompassed most ambiguously-aligned positions. The first, second and third codon positions of the protein-coding tef1-α, RPB1 and RPB2 markers were also delimited. Before producing a concatenated dataset, Maximum Likelihood phylogenies were constructed for each marker independently with the CIPRES' online RAxML-HPC2 v.8.2 version (Miller et al. 2010;Stamatakis 2006;Stamatakis et al. 2008), using 1000 bootstrap pseudoreplicates and the GTRGAMMA model, to test for topological incongruence among them, assuming bootstrap values ≥70 % as significant for conflicting relationships among the same set of taxa (Mason-Gamer & Kellogg 1996). Subsequently, RAxML-HPC2 was again used to build a multi-locus phylogeny, with the same parameters. Additionally, the software BEAST 1.8.1 (Drummond et al. 2012) was chosen to infer a phylogeny under a Bayesian framework, using an uncorrelated lognormal relaxed molecular clock and a Birth-Death tree prior. The ucld.mean parameter of each marker was given a uniform prior (lower= 0.0; upper= 5.0; initial= 1). A run using a chain length of 2 × 10 8 steps was implemented and parameters were logged every 2 × 10 4 steps. Resulting log files were checked in Tracer 1.7 to ensure that all parameters had ESS above 200 after removing the first 20% of saved trees as burn-in. Then, the median heights of the 1 × 10 4 post-burnin tree samples were annotated with TreeAnnotator 1.8.1., and the tree was drawn with FigTree 1.4. The latter three programs are available at http://tree.bio. ed.ac.uk/. Optimal substitution models used in the above analysis for the five markers and their inner partitions were inferred with PartitionFinder v.1.1.1 (Lanfear et al. 2012) considering a model with linked branch lengths and the Bayesian Information Criterion (BIC). However, to avoid overparameterization and chain convergence issues, less complex substitution models, such as HKY, were used instead of GTR.

Phylogenetic analyses at the family level
Once the most likely placement of Seynesiella at ordinal level within Dothideomycetes was known, the work of Shen et al. (2020) on Venturiales was followed to test the placement of our focal genus in any known family. To do this, we used a different five-locus dataset comprising data of 5.8S, nuLSU, tef1-α, RPB2 and the protein-coding β-tubulin (β-tub) gene. The same procedure and programs were chosen to construct a concatenated alignment and to infer Maximum Likelihood and Bayesian phylogenies (see previous section). To root the tree, the following species were chosen as outgroup: Schismatomma decolorans (Arthoniomycetes) and the dothideomycetes Tricho-

Intraspecific diversity and species boundaries
The alignment of nine nrITS sequences of Seynesiella juniperi was carried out with MAFFT v.7.308, and was manually optimized in Geneious v. 9.0.2 to trim alignment ends of longer sequences that included a section of the 18S-28S ribosomal subunits. The software MEGA 5.2 (Tamura et al. 2011) was employed to calculate several indices of nucleotide diversity and to infer a Maximum Likelihood phylogeny, using the T92 nucleotide substitution model and 500 bootstrap replicates. Then, the Automatic Barcode Gap Discovery method (ABGD, Puillandre et al. 2012), a model-based approach which automatically infers a one-sided confidence limit for intra-and interspecific divergence based on the distribution of all pairwise distances, was chosen to examine species boundaries in the dataset. The program was remotely run at http://wwwabi.snv.jussieu.fr/public/abgd/abgdweb. html, using the Kimura two parameters (K2P) model to calculate genetic distances between individuals, a TS/TV value of 2.01 obtained from MEGA 5.2, the Pmax value set to 0.01 (Puillandre et al. 2012), the relative gap width (X) to 1 and the remaining model parameters to default.   Morphological description. Ascomata thyriothecia, shiny black to black, solitary, scattered or grouped in clusters, on the upper side of leaves (Juniperus spp. with squamiform leaves) or either in the upper or lower surface (Juniperus spp. with needle-like leaves), 238.5 ± 52.3 μm (n= 28) in diameter, first flattened and later conical to hemisphaerical, ending in an ostiole ca. 16-25 μm wide; peridium consisting of radial hyphae with deep brown, isodiametric cells towards the ostiole, and more rectangular cells towards the periphery, the latter with a size of 7-10 × 4-9 μm (n= 5). Hamathecium embedded in mucilage. Pseudoparaphyses present and abundant, branched, cylindrical, with a diameter up to 4.5 μm (n= 4). Asci 67.70 ± 15.79 μm long and 23.65 ± 3.18 μm wide (n= 40), claviform to subglobose, pedicellate, with eight spores arranged in an overlapping biseriate manner. Ascospores ellipsoidal to long ellipsoidal, septate, formed by two cells with different sizes (heteropolar), each with one guttule, with ends more or less rounded, wall smooth, surrounded by a mucilaginous layer, hyaline or rarely yellowish brown when very mature or injured, 23 ± 1.88 μm in length, and width of small and large cells 8.18 ± 0.82 μm and 9.91 ± 0.91 μm, respectively (n= 66).

Taxonomy and description
Ecology and distribution. In the Iberian Peninsula, Seynesiella juniperi occurs on five Juniperus species: J. communis, J. oxycedrus, J. phoenicea, J. sabina and J. thurifera. Symptoms of injury due to parasitism were seldom observed in host leaves; in a few cases, however, a yellowish brown areole surrounding groups of thyriothecia was noticed. Based on the collections reported in the present study, this fungus displays a wide bioclimatic tolerance, being present in Mesomediterranean zones in the eastern Iberian Peninsula (e.g. low mountains near the coast in València and Alacant), as well as in more continental (Supramediterranean) areas of the inner Iberian Peninsula (Albacete, Soria) and mountains up to 1300-1600 m high (Castelló, Teruel, Madrid

Molecular datasets
The first five-locus dataset (nuSSU, nuLSU, tef1-α, RPB1 and RPB2), which was used for elucidating the placement of Seynesiella at ordinal level in Dothideomycetes, consisted of 92 taxa and 6610 bp of which 3146 were variable, 2486 parsimony informative and 660 sites had singletons. The second compiled five-locus dataset (5.8S, LSU, tef1-α, RPB2 and β-tub) that was used for establishing a closer phylogenetic placement of the focal genus within the order Venturiales was composed of 56 taxa and 1991 nucleotides; 904 of them were variable, 738 parsimony informative, and 166 sites with singletons. Finally, the nine nrITS sequences of Seynesiella juniperi generated a 506 bp-long alignment, with 23 and 8 variable and parsimony informative sites, respectively, together with 15 positions with singleton substitutions. The alignments are available in FigShare (doi: 10.6084/m9. figshare.14206964).

Phylogenetic relationships at the order and family levels
The ML phylogeny of Dothideomycetes orders calculated with RAxML had a lnL = -70080.4361. All parameters estimated in the BEAST analysis had ESS well above 200. Because these two phylogenies showed no supported topological conflicts, the ultrametric tree estimated in BEAST is shown in Fig. 2. The genus Seynesiella was placed at the base of the order Venturiales and this relationship was supported by both phylogenetic analyses (PP= 0.98; BS= 81%). Venturiales was composed of three well supported clades but relationships among them remained elusive. The two most closely related lineages to Venturiales were members of the family Phaeotrichaceae Cain and Microthyrium microscopicum (Microthyriales G. Arnaud). These lineages together with Venturiales formed a statistically supported monophyletic group within Dothideomycetes (PP= 1; BS= 78%). At the family level, the ML phylogeny of Venturiales calculated with RAxML had a lnL = -21754.7185 and all parameters estimated in BEAST reached ESS above 200. As before, the topology inferred with the Bayesian program is shown in Fig. 3. The genus Seynesiella was placed in the family Cylindrosympodiaceae P. Crous, M. Shen and Y. Zhang, and formed a well-supported clade (PP= 1; BS= 96%) sister to a representative of Sympodiella acicola W.B. Kendr; however, this relationship only received support from the BEAST analysis (PP= 0.96). The family Cylindrosimpodiaceae formed a monophyletic group supported only by the Bayesian analysis (PP= 1), and its phylogenetic relationships with the other two families within Venturiales, Venturiaceae and Sympoventuriaceae Y. Zhang, C.L. Schoch & K.D. Hyde, were not resolved with certainty in our reconstructions. The order Venturiales was again supported only by results of the Bayesian analysis (PP= 1).

Diversity within the focal species Seynesiella juniperi
The ML phylogeny of Seynesiella juniperi nrITS sequences revealed three statistically-supported monophyletic clades Fig. 4B. One encompassed four specimens collected on Juniperus oxycedrus in the eastern and central Iberian Peninsula (BS= 71%). A sample collected on J. thurifera leaves in El Rincón de Ademuz (València, I1186) and another from Lisbon on J. phoenicea (I1183) were placed close to those four specimens, with a BS value of 93%. Two other specimens collected on J. phoenicea in València and Alacant at approximately the same latitude than Lisbon were genetically more divergent (see samples I1201 and I1209 in the alignment Fig. 4D). Finally, the most genetically distinct specimen was collected on J. communis in Madrid (central Iberian Peninsula; see sample I1223 in Fig. 4D). Although the ABGD analysis did not reveal a clear barcode gap in the distribution of genetic distances (data not shown), the number of partitions (or putative species) inferred ranged between five (at values of P= 0.001292 to 0.001668), three (P= 0.002154 to 0.007743) and two (P= 0.010000). In Fig. 4B only the two-and three-putative species solutions are represented. If two species were assumed, specimens growing on J. communis would correspond to a distinct species to those occurring on J. oxycedrus, J. thurifera and J. phoenicea. Accepting the existence of three different species would mean separating the specimens growing on València and Alacant J. phoenicea into a third species. Considering the least conservative hypothesis of five species would imply that specimens occurring on different Juniperus species are also distinct; in that case, specimens occurring on western and eastern J. phoenicea would be also different taxa.

Discussion
The methodological approach conducted in the present work has allowed the phylogenetic placement of the genus Seynesiella in the Dothideomycetes Tree of Life. Based on data from five loci, the genus is placed in the family Cylindrosympodiaceae within the order Venturiales. This family was recently introduced in Shen et al. (2020) Wu et al. 2011). Therefore, the concept of Cylindrosympodiaceae may be expanded to include probable parasite species as well. The placement of Seynesiella in the order Venturiales contradicts earlier placements in the family Microthyriaceae (order Microthyriales) due to similarities in the type of ascomata and peridium structure (Luttrell 1965;Barr 1968;Lumbsch & Huhndorf 2010). Therefore, our results are more in line with the proposal of Müller & von Arx (1962) about a closer affinity with members of the family Venturiaceae, and also with Wu et al. (2011), who suggested a close similarity of Seynesiella with the thyriothecia-forming genus Tothia. However, our phylogenetic reconstruction suggests that Seynesiella and Tothia are not the closest genera within the family Cylindrosympodiaceae; instead, Seynesiella is sister to Sympodiella, whose type species, S. acicola, was described by Kendrick (1958) to introduce a hyphomycete genus inhabiting decaying needles of Pinus sylvestris L. Molecular data of additional species within the various genera in Cylindrosympodiaceae, including the five Seynesiella species described so far, are needed for deepening our understanding of the diversity and phylogenetic relationships within this family. The analysis of intraspecific genetic diversity in Seynesiella juniperi based on Iberian Peninsula specimens points towards the existence of more than one species. Although the sampling only covered a reduced portion of the species' distribution, the considerably high number of variable positions in the nrITS alignment together with the observation of several well-supported clades in the ML phylogeny and the ABGD outcome supports the split of the species into two, three or even five distinct taxa. This result seems to indicate a case of cryptic speciation, i.e. two or more distinct species classified as a single species (Bickford et al. 2007), which is a common phenomenon in tiny fungi (e.g. Crous et al. 2004;Pringle et al. 2005;Alves et al. 2008). Although the studied specimens cannot be discriminated by macroscopical characters, microscopical and ecological traits (e.g. host identity) could be combined to provide a more straightforward species delimitation. Thus, central Iberian specimens on Juniperus communis are phylogenetically well separated from the remaining clades and produce larger asci and longer spores compared with the remaining specimens on other Juniperus spp. (Tab. 1). Similarly, phylogenetic isolation of S. juniperi specimens growing on J. thuri-fera may be supported by the production of smaller asci and spores. In contrast, the separation into distinct species of specimens growing on J. oxycedrus and J. phoenicea only seems feasible on the basis of phylogenetics, as the measured asci and spore size values are largely overlapping. However, the fact that genetic differences are relatively high among specimens occurring on J. phoenicea from Lisbon and eastern Iberian Peninsula suggests that genetic differentiation at the geographical scale may be playing an important role as well. The lack of a clear barcode gap found in the ABGD analyses might align with the latter observation, which would indicate that the overall genetic differences observed are just intraspecific and controlled by geography. In any case, to further evaluate whether morphological characters are informative enough for species separation, we re-commend performing more detailed measurements of spores and asci for specimens growing on different Juniperus hosts and test any differences statistically. In particular, measurements of the length and width of each of the two spore cells would be recommended (see Fig. 1L for reference).
The hypothesis of evolutionary events in Seynesiella juniperi recapitulating divergence events in the phylogeny of their Juniperus hosts has not been validated in the present work. Based on the reconstructed Juniperus spp. phylogenies of Mao et al. (2010) and Adams & Schwarzbach (2013), J. communis is close to J. oxycedrus and both belong in Juniperus section Juniperus, whereas J. thurifera and J. phoenicea are placed in Juniperus section Sabina. The phylogeny of the fungus, however, places specimens growing on J. communis away from those occurring on J. oxycedrus. This lack of concordance between fungal and host phylogenies could be an artifact of the molecular markers used, though. Moreover, our knowledge about the range of likely Juniperus hosts in the Northern Hemisphere to which S. juniperi associates is still limited, and therefore host switching may be a likely phenomenon occurring in this peculiar fungal-plant association.

Conclusion
The present work provides crucial information to understand the evolutionary history of Seynesiella juniperi, a tiny but fairly common ascomycete fungus growing on Juniperus spp. leaves. A denser sampling of specimens, localities and molecular markers is needed for further evaluating species boundaries in this fungus, as well as gaining a deeper understanding of the evolutionary events that shaped current geographic distributions and associations with different Juniperus hosts. In particular, it would be interesting to test whether the origin of the fungus occurred in a temporal window that coincides with the origin of its plant hosts.