Neotropical Clavulina: Two new species from Mexico and a re-evaluation of Clavulina floridana

ABSTRACT Clavulina comprises ca. 90 described species distributed worldwide in both tropical and temperate regions. However, only one species (C. floridana) has been described so far from tropical North America. We used morphological and molecular data from three DNA loci (nuc rDNA internal transcribed spacer region ITS1-5.8S-ITS2 [ITS], a portion of nuc 28S rDNA [28S], and a fragment of DNA-directed RNA polymerase II second largest subunit [RPB2]) from basidiomata and ectomycorrhizas collected in tropical ecosystems from three biogeographic provinces of Mexico and one tropical province in the USA to investigate the phylogenetic and taxonomic diversity of Clavulina in the region. Nine new species-level clades were discovered, two of which are proposed as new species (C. arboreiparva and C. tuxtlasana). Specimens of C. floridana recently collected in Florida were included in our analyses, for which a modern description is provided. In addition, C. floridana is a new record for Mexico. The diversity of Clavulina in tropical North America is comparable to that found in lowland tropical South America. However, some of the species found in tropical deciduous forests produce small, rare, and inconspicuous basidiomata, which easily go unnoticed, and therefore are poorly represented in collections. Many species remain undescribed in tropical regions of North America.

Morphological features of basidiomata among most temperate species are rather homogeneous in color and shape. Most species have monopodial to branched basidiomata with a whitish to gray or ochraceous coloration, and the basidiomata rarely have bright colors (Corner 1970;He et al. 2016;Olariaga et al. 2009;Petersen 1988). In contrast, the neotropical species, particularly those from northern South America, feature infundibuliform, effuso-coralloid, resupinate-erumpent, or cerebriform basidiomata with either dull or notably bright coloration (Henkel et al. , 2005Thacker and Henkel 2004;Uehling et al. 2012bUehling et al. , 2012c. Both tropical and temperate species of Clavulina have microscopically similar basidia and basidiospore shapes, frequently have postpartal septae, and they share a monomitic hyphal system with or without clamp connections (Corner 1950(Corner , 1970Donk 1964;Petersen 1988). However, they may differ greatly in basidiospore size, sterigmata number, and cystidial characters (when present). For example, several neotropical species feature 2-5(6) sterigmata (Uehling et al. 2012b(Uehling et al. , 2012c. Delimitation of Clavulina species using morphological characters has been particularly difficult in species with simple to branched basidiomata and dull coloration due to the scarce variation in macroscopic morphology and narrow variation of microscopic characters (e.g. basidiospore and basidia size) that often overlap (Olariaga et al. 2009).
Most phylogenetic studies of Clavulina have typically analyzed a single locus, primarily nuc rDNA internal transcribed spacer region ITS1-5.8S-ITS2 (ITS) (He et al. 2016;Kennedy et al. 2012;Olariaga et al. 2009;Phookamsak et al. 2019;Tibpromma et al. 2017), but other loci have also been used, including nuc 28S rDNA (28S) (Thacker and Henkel 2004;Uehling et al. 2012b), and fragments of DNA-directed RNA polymerase II second largest subunit (RPB2) (Henkel et al. , 2005Uehling et al. 2012bUehling et al. , 2012c. These studies have shown that the ITS region is informative for species delimitation (Olariaga et al. 2009). However, this locus has poor resolution at deep internodes. Recent studies (Crous et al. 2019;Pérez-Pazos et al. 2019;Wu et al. 2019) have used some combination of these three loci to propose new species, but many species lack 28S and/or RPB2 data.
Mexico, as a transition zone between the Nearctic and Neotropics, is home to great biological diversity (Morrone 2014). Despite its relevance as a major diversification center for many taxa, only nine Clavulina species have been recorded from Mexico: C. amethystina (Bull.) Donk, C. cinerea (Bull.)  In Mexico, tropical ecosystems belong to the biogeographic Mesoamerican Domain, which is composed of four provinces: Pacific Lowlands, Balsas Basin, Veracruzan, and Yucatan Peninsula (Morrone 2014;Morrone et al. 2017). In the USA, the relatively small region of the Southern Florida Coastal Plain is unique among any other continental ecoregions of the USA by the humid environment and tropical climate (i.e., free of snow or frost; Wiken et al. 2011). The tropical environments of these regions vary from evergreen tropical forests to subdeciduous forests, which offer a great diversity of plant associates for Clavulina and other genera of ectomycorrhizal (ECM) fungi (Alvarez-Manjarrez et al. 2018).
Clavulina floridana was originally described as Clavaria floridana based on a single collection by Singer (1945) and transferred to Clavulina by Corner (1950). The name C. floridana has generated confusion due to the ambiguity used for some characters in the original description. For instance, it has been cited from distant localities (e.g. Brazil and New Zealand) based on a handful of macroscopic and microscopic characters (Petersen 1988;Wartchow 2012). The original description lacked an illustration and detailed morphological data regarding basidiomata color and variation as well as some key features regarding micromorphology and potential plant associations.
In this work, we explored the diversity of Clavulina from tropical environments in North America. We included samples collected from three tropical provinces of Mexico and one in the USA. To discriminate among species, we used macro-and micromorphological data, as well as DNA sequences from three loci. We propose two new species and report the first record of C. floridana for Mexico. We also provide a modern description for C. floridana.

Collections
and morphological analyses.-Specimens were collected in four localities of tropical forests in the Mexican states of Campeche, Jalisco, Oaxaca, and Veracruz, located in the Mesoamerican Domain provinces of Yucatan Peninsula (Campeche), Pacific Lowlands (Jalisco and Oaxaca), and Veracruzan (Veracruz). Additional specimens were collected at three localities in the Southern Florida Coastal Plain ecoregion of the USA (TABLE 1, FIG. 1). Specimens were collected during the rainy seasons of 2002, 2003, 2008, 2011-2013, 2018, and 2019. Collecting effort included 50 days of sampling at the different localities with two to four collectors per site. We also included sequences from ECM samples of Alvarez-Manjarrez et al. (2018) that were obtained in Jalisco, Mexico, and ECM samples of Põlme et al. (2017) that were obtained in Florida.
Morphological characters were analyzed for 20 new specimens. Basidioma size, shape, coloration, consistency, smell, and taste were recorded from fresh specimens. Branching rank (b.r.) was also recorded on fresh material following the terminology of Olariaga et al. (2009). Color codes were recorded according to Kornerup and Wanscher (1981). Basidiospore, basidia, and hyphal characters were recorded by preparing microscopic slides from dehydrated specimens using 5-10% KOH under light microscopy on an Olympus BX51 microscope (Tokyo, Japan) with differential interference constrast illumination at 1000× magnification. Terminology used for the description of microscopic characters followed Petersen (1988). Details of basidiospore surfaces and hilar appendices were observed using a JEOL JSM-5310LV scanning electron microscope (SEM) (Tokyo, Japan). Quantitative data were obtained using a minimum of 30 observations per character. Basidiospore characters were coded as follows: Q = length/width, Q = length average/width average, L = length average, W = width average. Mexican specimens were deposited in the "Herbario de la Facultad de Ciencias, UNAM" (FCME) and "Herbario Nacional de México, UNAM" (MEXU), whereas USA specimens were deposited at the University of Florida Herbarium (FLAS) and University of South Florida Herbarium (USF).
We studied the type specimen of C. floridana (Singer F 733) from the Farlow Herbarium (FH), as well as specimens from the University of Tennessee Herbarium (TENN), and new collections from Florida and Mexico.
DNA extraction and PCR amplification.-DNA was extracted from herbarium vouchers and ECM samples using the REDExtract-N-Amp Plant PCR Kit (Sigma-Aldrich, St Louis, Missouri) following the manufacturer's instructions, except for C. floridana vouchers from Florida where an E.Z.N.A. Fungal DNA Mini Kit (Omega Bio-tek, Norcross, Georgia) was used instead. We amplified three loci: ITS, a portion containing the D1-D3 regions of 28S, and a fragment of the RPB2 gene with no known introns. Primers used for ITS amplification were ITS1F (Gardes and Bruns 1993) and ITS4 (White et al. 1990), with the following polymerase chain reaction (PCR) conditions: 94 C for 5 min, 34 cycles at 94 C, 55 C, and 72 C, all for 1 min, and 72 C for 8 min. For 28S, we used primers LROR (Hopple and Vilgalys 1994) and LR5 (Vilgalys and Hester 1990), using the same PCR conditions as those of ITS region except for annealing temperature, which was set to 58 C. For RPB2, we used the primers clav96F and clav938R (Uehling et al. 2012b), with the following PCR conditions: 94 C for 5 min, 25 cycles at 94 C for 30s, 63 C for 30s, and 72 C for 2 min, 11 additional cycles at 94 C for 30s, 45 C for 30s, and 72 C for 2 min, and 72 C for 7 min. PCR products were visualized in 1% agarose electrophoresis gels run in 1× TBE buffer (89 mM Tris-Borate, 2mM EDTA pH 8.3) buffer, stained with GelRed (Biotium, Hayward, California), and cleaned up with ExoSAP-IT (Applied Biosystems, Foster City, California). PCR products were sequenced in both directions with PCR primers using Sanger technology.
We made multiple attempts to generate nuc rDNA sequences following a modified cetyltrimethylammonium bromide (CTAB) protocol described in Farid et al. (2017) from an isotype collection (FLAS-F-51151) of C. floridana, but all were unsuccessful. Therefore, DNA sequences were obtained from a specimen collected recently in Florida approximately 6 km from the type locality. The two C. floridana vouchers from Florida were sequenced with a 3730 DNA analyzer (Applied Biosystems) at the DNA Laboratory at Arizona State University, and the rest with an ABI PRISM 3100 sequencer (Applied Biosystems) at Instituto de Biología, Universidad Nacional Autónoma de México (UNAM). Due to technical problems, not all specimens were sequenced for all loci (TABLE 1).
Phylogenetic analyses.-Representative reference sequences of Clavulina for three loci were obtained from recent literature and BLAST searches in GenBank using our newly generated sequences as templates. From these searches, we kept sequences obtained from vouchered materials of representative species and a few environmental sequences that were sampled from tropical environments or the geographic regions of our study (see SUPPLEMENTARY TABLE 1). Sequences of Hydnum spp. were included as outgroups (Moncalvo et al. 2006). Sequences were aligned using MAFFT 7 (Katoh and Standley 2013) with default parameters. Sites that were not present in all sequences at both 5′ and 3′ regions were pruned. Although we acknowledge that ITS and 28S are linked, we decided to analyze them separately because of the uneven sampling for these loci. The final matrix for the ITS data set contained 112 sequences with 972 sites. The 28S matrix consisted of 51 sequences with 1103 sites, and the RPB2 matrix consisted of 47 sequences with 783 sites. We also produced two concatenated matrices: one with ITS+28S (38 sequences with 1749 sites) and one with ITS+28S +RPB2 (52 sequences with 2562 sites). The length of all alignments includes indels. For the ITS+28S matrix, we only included samples with sequences for both loci, and for the ITS+28S+RPB2 matrix we included samples with at least two out of three loci.
We used PartitionFinder 2.1.1 (Lanfear et al. 2012(Lanfear et al. , 2017 to find the best-fitting evolution model and partition scheme for all matrices. We set one partition each for the ITS and 28S matrices and two partitions for RPB2, one for the first two codon positions and a second one for the third codon position. For the two concatenated matrices, we defined one partition for each locus. In all cases, model search in PartitionFinder was limited to those implemented in MrBayes, analyzing all possible partition schemes (command search = all). The selected model for ITS, 28S, and the ITS+28S matrices was GTR+I+G (Tavaré 1986). For the ITS+28S+RPB2 matrix, the selected model was GTR+I+G for the two ribosomal loci and K80+I+G (Kimura 1980) for RPB2. For the RPB2 alignment, two partitions were optimal: GTR+I+G for the first two codon positions and HKY+I +G (Hasegawa et al. 1985) for the third position. We accordingly assigned these models of evolution to alignments and obtained a consensus tree in MrBayes 3.2.7a (Ronquist et al. 2012) with unlinked priors for each partition, 20-50 million generations in two independent runs of four chains each, and using the first 25% of generations as the burn-in. After each run, we assessed their adequacy by checking the potential scale reduction factor (PSRF), average standard deviation of split frequencies (SDSF) between independent runs, and swap rate (SR) between immediate chains. We also used Tracer 1.7.2 (Rambaut et al. 2018) to visually check for convergence and adequacy of burn-in proportions. We recorded the posterior probabilities (PPs) for all branches in the consensus trees for all analyses. In addition, maximum likelihood trees were obtained for all matrices using RAxML 8.2.12 (Stamatakis 2014) with GTRGAMMA as the model of evolution and assessing support with 1000 bootstrap (BS) replicates. In all cases, the matrices were partitioned as described above. All MrBayes and RAxML analyses were performed using the CIPRES Science Gateway (Miller et al. 2010).

RESULTS
We collected four C. floridana specimens from the Southern Florida Coastal Plain and 16 specimens of undescribed Clavulina species from four locations in the Mesoamerican Domain in Mexico (TABLE 1, FIG. 1), two of which were later identified as C. floridana. Two C. floridana specimens were collected in Miami-Dade County, one from Everglades National Park near Coccoloba diversifolia and one from the campus at Florida International University under planted Coccoloba uvifera. Two of these samples were sequenced. Two ITS sequences from ectomycorrhizas reported in previous works were identified as C. floridana, one from a sample associated with Coccoloba uvifera collected in Miami, Florida (Põlme et al. 2017; GenBank accession KX398712; 99.85% similarity), and one associated with Guapira sp. collected in Chamela, Jalisco (Alvarez-Manjarrez et al. 2018; GenBank accession KP896291; 99.45% similarity).
Of the remaining 14 Mexican specimens, 10 were collected from a tropical evergreen forest in the Veracruzan Province, two from two locations in tropical subdeciduous forests in the Pacific Lowlands Province, and two from a tropical subdeciduous forest in the Yucatan Peninsula Province (TABLE 1, FIG. 1). In addition, three ectomycorrhiza samples from the Pacific Lowlands of Jalisco (Alvarez-Manjarrez et al. 2018) were identified using ITS sequence data as Clavulina spp. Details of all samples used in this work, GenBank accessions, sampling locations, and other metadata can be found in TABLE 1.
Two new species were recognized using both molecular and morphological information from 11 specimens, one species from subdeciduous tropical forests (C. arboreiparva, sp. nov.) and another from a tropical evergreen forest (C. tuxtlasana, sp. nov). TABLE 2 provides a synopsis of the morphological features of the newly described species and a comparison of specimens of C. floridana from Florida (USA) and Jalisco (Mexico).
Phylogenetic analyses.-We generated 20 ITS, 16 28S, and 15 RPB2 new sequences from the above specimens. These sequences were analyzed together with 92 ITS, 34 28S, and 32 RPB2 reference sequences obtained from GenBank (SUPPLEMENTARY TABLE 1). Alignments and trees can be downloaded from SUPPLEMENTARY MATERIAL (SUPPLEMENTARY FILES 1-5). For all Bayesian analyses, PSRF values were equal to 1, average values of SDSF between independent runs were below 0.006, and SR values between immediate chains were above 20%. Distribution of parameters converged, and a 25% burn-in was deemed adequate as visualized in Tracer.
Phylogenetic analyses of ITS showed that Mexican specimens corresponded to nine previously unknown species-level clades within Clavulina, which agreed with their morphological characters (where available) and, to some extent, geographic distribution (TABLE  1, FIG. 2). In general, the ITS tree showed good support values near the tips but poor values at deeper branches as reported previously using the same locus (He et al. 2016;Kennedy et al. 2012;Olariaga et al. 2009;Tibpromma et al. 2017). The nine newly recognized clades and new sequences of C. floridana belong to an unsupported but more inclusive clade, which contains a mixture of temperate species from the Northern Hemisphere, some tropical samples from South America (Brazil, Colombia, Ecuador), and samples from Australasia (FIG. 2).
Samples from different individuals identified as the same species formed highly supported clades. However, the relationships of the new lineages with previously reported species were not clear in most cases. Unnamed  (FIG. 2).
The phylogeny obtained from 28S showed a similar pattern (SUPPLEMENTARY FIG. 1), with an unsupported clade consisting mostly of samples from the Northern Hemisphere and clades toward the root of the tree, with varying support, grouping species from Guyana. As in the ITS tree, well-represented species formed highly supported monophyletic groups: C. arboreiparva (PP = 0.91; BS = 99), C. floridana (PP = 1; BS = 100), and C. tuxtlasana (PP = 0.98; BS = 100).
The phylogeny inferred using RPB2 (SUPPLEMENTARY FIG. 2) was congruent with the other two loci and recovered a moderately supported clade (PP = 0.92; BS = 71) including the new species and most Northern Hemisphere samples. Clavulina floridana and C. tuxtlasana formed independent monophyletic groups with high support (PP = 1; BS = 97 and PP = 1; BS = 100, respectively). Like the phylogenies for the other two loci, species from Guyana formed several clades with varying support but many taxa were on long branches. Two Chinese samples identified as C. ornatipes (Peck) Corner and C. purpurascens P. Zhang were also resolved among some of the Guyanese taxa but without support.
Ecology and distribution: Gregarious to cespitose (i.e., in clusters, but not fused), in mesic rockland hammocks and disturbed, cultivated rockland soils associated with Coccoloba diversifolia and Coccoloba uvifera. Usually on the ground or occasionally on small rotten sticks or leaves, in tropical hammocks in Florida and the Bahamas under Coccoloba species, Jun-Oct. In tropical subdeciduous forests of the Pacific lowlands of Jalisco, Mexico, forming ectomycorrhizas with Achatocarpus gracilis, Guapira petenensis, and other unidentified Nyctaginaceae, Aug-Sep. Notes: The specimens selected in this work for the reevaluation of C. floridana (USF 301181, FLAS-F-67287) differ from the type material by the slightly smaller spores (6.5-8(-8.5) × 5-6(-7)), but all other characters coincide with the original Singer description (Singer 1945). As far as we know, our collections in Florida in 2018 were the first new collections since the type collection in this region. This is probably because this species is easily overlooked, and little mycological field work has been done in south Florida. During three years of field work (2018-2020), we found C. floridana to be quite common in Miami-Dade County under large mature trees of Coccoloba diversifolia and Coccoloba uvifera in both cultivated and natural populations.
Clavulina floridana was reported from New Zealand by Petersen (1988). However, this specimen (TENN-F-044109) has larger spores (9-10.4 × 7.6-9 µm), much larger basidia (70-80 × 8 µm), and clamp connections. In addition, Petersen (1988) noted that several species of Clavulina from New Zealand were associated with Nothofagus forests. Therefore, this specimen likely represents a different taxon. Clavulina floridana was also cited from Sao Paulo, Brazil (Baltazar and  Gibertoni 2009), but molecular data should be used to verify this observation. The Mexican specimens of C. floridana from the Pacific Lowlands are similar to the material from Florida in coloration, basidiospore size, hyphal characters, and the absence of clamp connections (FIGS. 5 and 6). However, the Mexican specimens have smaller basidiomata (8-35 × 1 mm), differ in vegetation type (tropical subdeciduous forest), and have different putative plant associates (A. gracilis, G. petenensis, and other unidentified Nyctaginaceae). Nevertheless, molecular analyses showed that all C. floridana samples from Mexico and the USA are monophyletic (FIGS. 2 and 3; SUPPLEMENTARY FIGS. 1, 2, and 3) and are similar in both the ITS and 28S sequences. The disjunction in the geographic distribution, habitat, and plant associates could be enough to argue that the specimens from Mexico and the USA belong to different species. However, given the morphological similarities and shared monophyly, we prefer to recognize them as a single species.  Etymology: tuxtlasana (Latin), in reference to Los Tuxtlas, an area in which this species is common and frequently collected.
Other specimens examined: MÉXICO. VERACRUZ: San Andrés Tuxtla municipality, Estación de Biología de Los Tuxtlas, Darwin pathway, 18.5833 N, 95.0736 W, 18 Sep 2011, Villegas Ríos M. 2825 Notes: Clavulina tuxtlasana is similar to C. floridana in that both species form simple to fasciculate basidiomata that are scarcely branched with a grayish hymenium and a clearly differentiated stipe. Both species lack clamp connections, but the basidiospores differ slightly in size. They also can be differentiated by stipe coloration-white in C. floridana but grayish orange to golden brown in C. tuxtlasana (TABLE 2)

DISCUSSION
The ITS phylogeny revealed nine different species-level clades out of 17 specimens and five ECM samples from Mexico and Florida (TABLE 1). These correspond to nine putative species based on the phylogenetic species criterion (Taylor et al. 2000). Here we describe two new species, Clavulina arboreiparva and C. tuxtlasana, and disclose the first record of C. floridana for Mexico. These new species also represent the first confirmed records of Clavulina from tropical Mexico. The recent record of C. fuscolilacina from the Yucatan Peninsula (Uitzil-Colli et al. 2021) still needs phylogenetic evidence for confirmation. Clavulina arboreiparva and C. tuxtlasana are clearly differentiated from any other known Clavulina species based on morphology, molecular characters, and geographic distribution. The incorporation of new samples of C. floridana from the Pacific Lowlands in Jalisco, Mexico, greatly expands the known distribution of this species. Citations of this species from New Zealand are questionable. Comparisons of C. floridana samples from Mexico and the USA show very similar morphology but with differences in geographic distribution, vegetation type, and ECM plant associates (TABLE 2). The ITS and 28S alignments revealed fixed differences between the populations, nine at the ITS locus and one at the 28S locus. These differences are larger than those observed for some species with wide distribution or recognized geographic structure (Crous et al. 2019;Olariaga et al. 2009). There is only one substitution in the ITS between samples of C. reae collected in Spain and Mexico, six substitutions between C. rugosa from Spain-Andorra and other sites in Europe, and two substitutions between the two variants of C. iris from Cyprus and France (SUPPLEMENTARY TABLE 2). The average ITS nucleotide similarity between C. floridana from Florida and Mexico (98.27%) is somewhat less than the within-species averages for C. reae (99.42%) and C. iris (99.66%), but greater than those of C. rugosa (94.65%) and C. sprucei (Berk.) Corner (ca. 93%), which likely represent species complexes  ; see SUPPLEMENTARY  TABLE 2 for more comparisons and details of how these estimates were obtained). Our data suggest that specimens from the Pacific Lowlands and samples from south Florida of C. floridana are likely conspecific, despite some consistent molecular and morphological differences. Additional DNA loci and collections from across Mexico, the southeastern USA, and the Caribbean are needed to evaluate this hypothesis further.
Our ITS phylogeny and previous studies (He et al. 2016;Kennedy et al. 2012;Olariaga et al. 2009;Tibpromma et al. 2017) consistently recover South American specimens as branches diverging close to the root of the tree, with an unsupported branch containing specimens from the Northern Hemisphere and Australasia (FIG. 2). The North American tropical representatives of Clavulina that we added to the ITS phylogeny did not alter this pattern.
Trees inferred separately for each locus (SUPPLEMENTARY FIGS. 1 and 2) and trees inferred from concatenated alignments (FIG.  3;  SUPPLEMENTARY FIG. 3) generally show a pattern concordant with the one described above for the ITS phylogeny. All loci seem suitable for species recognition, but resolution for the genus phylogeny is still poor. Trees inferred from concatenated matrices slightly increase support for deeper branches and resolved a handful of divergent species from Brazil and Guyana near the root of the tree. However, support values for the ITS+28S tree are somewhat better than those of the ITS +28S+RPB2 tree. The uneven sampling of these loci across taxa impeded further comparisons, but a more complete sampling may improve future inferences.
Clavulina systematics is particularly difficult due to convergent morphology among some species and poor phylogenetic resolution of the loci that have been employed (Hibbett et al. 1997;Binder et al. 2005;Larsson 2007;Moncalvo et al. 2006;Uehling et al. 2012a). However, morphological characters of Clavulina are still important for species recognition. For example, basidiomata size and shape, stipe coloration, basidiospore size, and presence/ absence of fibulae are informative enough to distinguish some species (Corner 1950(Corner , 1970Petersen 1967Petersen , 1988Uehling et al. 2012b). In most studies (e.g. He et al. 2016;Henkel et al. 2011Henkel et al. , 2005Pérez-Pazos et al. 2019;Tibpromma et al. 2017;Uehling et al. 2012c;Yuan et al. 2020), morphological and molecular data are used together, allowing a robust delimitation of species and a clear interpretation of characters. However, for a more comprehensive polyphasic taxonomy (Lücking et al. 2020), the addition of reproductive and ecological attributes is still needed (e.g. enzymatic capabilities or plant associate preferences).
Determination of plant associates can help elucidate the diversification and geographic distribution of ECM fungi. Our data suggest that the hosts of C. arboreiparva and C. tuxtlasana are likely co-occurring tropical species of Nyctaginaceae and Polygonaceae (Arroyo-Rodríguez et al. 2009;Martínez and Galindo-Leal 2002). These two families of Caryophyllales are well known as ECM photobionts in the Neotropics (Alvarez-Manjarrez et al. 2018;Corrales et al. 2018). Although presently known only from Florida, the Bahamas, and Mexico, C. floridana may be as widespread as its host species. Coccoloba diversifolia and Coccoloba uvifera occur throughout the Caribbean Basin (Felger 2006), A. gracilis can be found along the Pacific coast of Mexico from Sinaloa to Chiapas, and G. petenensis has a wider distribution from the Pacific coasts of central Mexico to Central America (Douglas and Manos 2007;Hernández-Ledesma 2008;Villaseñor et al. 2013). Another Caribbean ECM species, Scleroderma bermudense Coker, shares the same hosts as C. floridana (Bandou et al. 2006;Séne et al. 2015). Scleroderma bermudense is found in Bermuda and Florida and extending to the southern regions of the Mexican Pacific, Gulf of Mexico, and the Yucatan Peninsula (Guzmán 1986;Guzmán et al. 2013). Scleroderma bermudense is also found with cultivated Coccoloba uvifera, where it has been introduced across the tropics (Séne et al. 2018). Cantharellus coccolobae Buyck, P.-A. Moreau & Courtec also occurs widely with Coccoloba uvifera in Florida, the Yucatan Peninsula, and the Greater and Lesser Antilles (Buyck et al. 2016;Kreisel 1971;Lodge 2002). Clavulina floridana ranges from dirty white to gray to blackish and can be easily overlooked. This likely explains the 75-year lag between when the type was collected and the discovery of our recent specimens.
Our ITS sampling included root tips with ectomycorrhizas from Jalisco, Mexico. However, only four out of 115 samples corresponded to Clavulina species (Alvarez-Manjarrez et al. 2018). One of these ectomycorrhiza sequences was assigned to C. floridana, but the other three are divergent enough to represent three additional undescribed species (FIG. 2). It is noteworthy that during the three years of ectomycorrhiza sampling, no basidiomata from Clavulina were observed. This suggests that these small and inconspicuous Clavulina species may not reproduce sexually every year, and that the simultaneous sampling of ectomycorrhizas and basidiomata (when present) is important to better document Clavulina diversity in the dry tropics.
Our results agree with previous findings that neotropical environments are diverse for some ECM fungi (e.g. Kennedy et al. 2012;Smith et al. 2011). However, the high diversity of the North American tropical species studied here suggests that there may have been diversification events of tropical species outside South America, and that there may be many more unique tropical Clavulina species that require future study. Finally, our results highlight the utility of including new samples of previously described species in the analyses because they provide important morphological and molecular characters that were unavailable at the time of the original descriptions.