Hidden fungal diversity from the Neotropics: Geastrum hirsutum, G. schweinitzii (Basidiomycota, Geastrales) and their allies

Taxonomy of Geastrum species in the neotropics has been subject to divergent opinions among specialists. In our study, type collections were reassessed and compared with recent collections in order to delimit species in Geastrum, sect. Myceliostroma, subsect. Epigaea. A thorough review of morphologic features combined with barcode and phylogenetic analyses (ITS and LSU nrDNA) revealed six new species (G. neoamericanum, G. rubellum, G. brunneocapillatum, G. baculicrystallum, G. rubropusillum and G. courtecuissei). In additon, the presence of hairs on the exoperidium, a commonly used feature to diagnose Geastrum species, proved to be ineffective because it is a derived character within subsect. Epigaea.


Introduction
The Neotropical biogeographic realm, or Neotropics, comprises Central America, most of South America (except Patagonia), and the southern portion of North America. It is considered the most diverse region for well-studied terrestrial taxa, mainly animals (amphibians, reptiles, birds and mammals) and plants (Angiosperms) [1,2,3,4,5]. However, knowledge regarding neotropical fungi is still insufficient. Indeed, the neotropical region represents a priority area for taxonomic studies since it encompasses megadiverse countries (Brazil, Colombia, Costa Rica, Ecuador, Mexico, Peru, Venezuela) with hotspot areas, such as the Atlantic Rainforest, Cerrado and Caribbean Islands; and tropical ecosystems where many potentially new taxa are threatened by human impacts [6,7,8,9].
Geastrum Pers., is a genus of gasteroid fungi in which the outer layers of the fruitbodies (basidiomata) open in a stellate pattern once the spores are mature, which makes them Colour descriptions were based on [34,35]. Sample observations and macro morphological image capturing were done using a stereomicroscope Nikon H600L coupled with a Nikon DS-Ri camera. Micro morphological studies were carried out using a Nikon Eclipse Ni light microscope (LM) coupled with a Nikon DS-Ri camera, and scanning electron microscope (SEM) analysis was done under a Shimadzu SSX-550. For light microscopy of basidiospores, eucapillitium, rhizomorphs and exoperidial hyphae, the samples were mounted in 5% KOH (w/v). Also, samples were mounted in Congo Red to observe the basidia; and Melzer's reagent was used to test the exoperidium, subiculum and rhizomorph hyphae. Preparation of the material examined under SEM followed Silva et al. [36]. At least thirty randomly selected basidiospores were measured using LM at 1000× magnification, including surface ornamentation; and the height of the ornamentation was also measured. Basidiospore abbreviations follow [37]: n = number of randomly measured basidiospores; x = mean ± standard deviation of basidiospore diameter and height (including ornamentation); Q m = mean height/width quotient. Geographic distributions of delimited taxa followed Biogeographic Realms, Biomes and Ecoregions proposed by Dinerstein et al. [5].

Molecular analyses
For UFRN-Fungos, INPA-Fungos and CEPEC samples, the extractions of DNA were performed utilizing 10 mg of gleba from dry basidiomata, preferably mature gleba. For the DNA isolation, DNeasyTMPlant Mini Kit (Qiagen, Valencia, CA) was used following manufacturer's instructions; except that the incubation in the lysis buffer was done at 55-60˚C overnight. For all other samples, fungal DNA was extracted from fragments of dried fruitbodies by using the Wizard Genomic Purification kit (Promega, Charbonnière les Bains, France) according to the manufacturer's recommendations, and the final pellet resuspended in 40 μl of sterile water. Internal Transcribed Spacer (ITS) region of the nuclear ribosomal gene, including the 5.8S subunit (ITS nrDNA), and Large Subunit region of nuclear ribosomal DNA (LSU nrDNA) were the loci selected for molecular analyses. DNA amplification, purification and sequencing protocols are deposited in protocols.io (dx.doi.org/10.17504/ protocols.io.wpdfdi6).
Sequences obtained in this study were submitted to Genbank under the accession numbers indicated in Table 1. The newly-generated ITS and LSU sequences, and homologous sequences retrieved from EMBL/GenBank/DDBJ databases, mainly from [10,12,16,38], were separately aligned in MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 [39]. Alignment gaps were marked with "-" and unresolved nucleotide positions were indicated with "N". Geastrum velutinum was included as outgroup since this species is in section Myceliostroma, subsection Velutina J.C. Zamora, sister clade of subsect. Epigaea [10].
The maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference analyses are also deposited in protocols.io under the doi indicated above.

Results
The ITS dataset included 48 sequences of Geastrum specimens, of which 34 were generated in this study (Table 1) and 14 obtained from EMBL/GenBank/DDBJ databases. In addition to the new species, sequences were obtained from type collections of five previously described species (Table 1). The ITS alignment resulted in 657 unambiguously aligned nucleotide positions (317 constant, 103 parsimony-uninformative, and 237 parsimony-informative). MP analysis resulted in one most parsimonious tree (S1 Fig)  The evolutionary models chosen by jModel-Test for Bayesian inference were HKY+G for ITS dataset (according to all criteria: AIC, BIC, AICc and DT), and TIM3+I+G for LSU dataset (according to AIC and AICc criteria-BIC and DT suggested TIM3+I, so we decided to use the more thorough algorithm). In the Bayesian analyses, the first 2,000 trees from the non-stationary phase were discarded. Maximum Clade Credibility tree and Posterior Probabilities (PP) were calculated from the 18,002 remaining trees. The summarized MCC tree has lnL = -2962.852. Concatenated ITS/LSU analysis gave us a MCC tree with lnL = -4329.812 (S5 Fig), which has been summarized from 18,002 trees after excluding 2,000 initial samples from the non-stationary phase.
Tree topologies generated by maximum parsimony (S1 and S3 Figs), maximum likelihood (S2 and S4 Figs), and Bayesian analyses (Fig 1 and S5 Fig) were similar, showing equivalent clustering patterns at terminal nodes. Except in ML analyses, G. schweinitzii isotype K(M)   Table 1. Numbers at the nodes indicate the Maximum Parsimony bootstrap, bootstrap values obtained from Maximum likelihood, and Posterior Probabilities from Bayesian analysis (MPbs /MLbs/ PP). Thick-lined branches are those with higher support (MPbs and MLbs higher than 85%, and PP higher than 0.85). Asterisk denotes fully supported branches in all three analyses.
Clade II. The two sequences identified as Geastrum aff. mirabile Mont. obtained from Gen-Bank (seqs. JN845108/JN845226 and JN845109/JN845227); [16]) formed a strongly supported group (MPbs = 100, MLbs = 100, PP = 1). According to the authors, specimens were collected in Asia (Bonin Island-Japan). In the protologue, these specimens of G. mirabile, are described as having small basidiomata, sessile endoperidial body, basidiospores 3.5-5.0 μm diam., and lignicolous habit. Leprieur's revision of the original type collections of G. mirabile from French Guiana held at PC (PC0084351, Leprieur 849, 2 plates), not successfully sequenced, suggests that G. mirabile is a morphological synonym of G. schweinitzii. Since we have not been able to analyze the Japanese collections of Kasuya et al. [16] we cannot ensure whether their morphology matches that of the PC vouchers of G. mirabile. However, all available photographs of the samples alleged to be G. mirabile at on-line TNS fungarium databases (including source materials of the sequences used in our analysis) are from specimens with small mycelial tufts in a slightly hirsute exoperidium (S6 Fig). Thus, Japanese material from TNS claimed to be G. mirabile needs reassessment because they did not group in any of the clades presenting 'schweinitzii-like' morphology, and do not exhibit this morphological pattern. On the other hand, original PC vouchers can be either a synonym of G. schweinitzii or a cryptic species, and only molecular assessment of this material can clarify its real identity.
Subiculum composed of hyaline, filamentous, slender hyphae, 0.9-1.2 μm diam, dextrinoid, sinuous. Rhizomorphs composed of hyaline, slender hyphae, lumen not evident, surface covered by crystals, with coarser and more irregular oblique prism shape, 1. Remarks. This species has small basidiomata (7-9 mm wide when expanded), exoperidium light brown, tomentose to rugulose when mature; pseudoparenchymatous layer reddish when fresh to light brown when mature; peristome slightly depressed on the endoperidium; basidiospores subglobose to oval (Q m = 1.07) with 4.2-5.9 μm diam, warts short (0.1-0.5 μm high) with planar to rounded tips. Based on morphology, G. rubropusillum is very similar to G. schweinitzii, and this explains some misidentifications, such as the collection MA-Fungi 36141 (seqs. KF988438/KF988568), which was previously identified as G. schweinitzii [10] but it belongs to the species G. rubropusillum. According to our data, G. schweinitzii differs from G. rubropusillum by having lighter pseudoparenchymatous layers (whitish when fresh), peristome non-depressed on the endoperidium, and globose basidiospores. Another species similar to G. rubropusillum is G. pusillipilosum, which is distinguished by its densely hairy exoperidium and globose basidiospores (Q m = 1.00).

Discussion
This study uncovered a hidden richness of subiculose neotropical Geastrum species. From our analyses 12 species were recovered, mainly collected in South and Central America, of which six were species unknown to science. Thus, we confirm the underestimated biodiversity of the genus Geastrum in the Neotropical region, illustrated by two cases of species complexes involving G. schweinitzii and G. hirsutum.
The barcode sequence (ITS) of the isotype of G. schweinitzii (K (M) 180187) grouped with only one sequence (INPA 143435) in Clade 1. It is interesting to note that these two collections are from the same biogeomorphological region: the Guiana Shield, a pristine Amazonian area with minor modifications during landscape evolution [42] revealing a possible scenario for allopatric or parapatric speciation dynamics in which G. schweinitzii may be endemic, since the Pakaraima Mountains are an ecological barrier for many organisms [44,48,49]. Keeping this in mind, the ten synonyms of Geastrum schweinitzii proposed by Ponce de León [27] need to be reassessed. Furthermore, some of the putative synonyms are from very distinct ecoregions [5] Fifteen samples with the morphology traditionally associated to the name G. schweinitzii, collected from widespread areas of the neotropical region, appear in clade XII in Fig 1, illustrating a case of semi-cryptic species [17,18,20,21] and evolutionary convergence in their morphology. Another case of semi-cryptic species of G. schweinitzii is G. baculicrystallum, the two species could be distinguished only by details in basidiospore size and ornamentation.
Three ITS sequences from GenBank previously identified under the name G. schweinitzii, KF988437, KF988438, and KF988439, are in fact three different species: G. courtecuissei, G. rubropusillum, and G. pusillipilosum, respectively. In these cases, morphological features distinguish each of these species.
The presence of hairs on the exoperidium is a recurrent feature of subiculose species. However, presence or absence of hairs as a single decisive feature for taxonomic identification could result in misidentification. Five of the species recognized here included this same feature. These semi-cryptic species are distinguished by molecular data, but discriminatory morphological features are unremarkable. Geastrum brunneocapillatum, G. rubellum and G. hirsutum are semi-cryptic species, and are not even sibling/sister species [17,18,20], reinforcing the statement that the presence of hairs on the exoperidium alone is not a suitable feature for species delimitation in Geastrum, but, instead, it represents an evolutionary convergence.
Recently a synonymization of Geastrum trichiferum to G. hirsutum was proposed [28]. Geastrum trichiferum is a mysterious species involved in taxonomic and nomenclatural problems in recent years [28,29,30]. Trying to better understand the nomenclature and taxonomic status of this species, we analyzed the collections PACA 15970 (packet labeled holotype in PACA), BPI 706088 (Rick's original collection alleged by Zamora & Parra [30] and BPI 706086 (lectotype designated by Trierveiler-Pereira & Silveira [28] (S7 Fig). It was possible to distinguish BPI and PACA collections from other species studied in this paper. By morphological analysis, we realized that these two exsiccates are notably different from each other, and they should probably be treated as distinct species: PACA 15970 has a basidiome with non-delimited peristome and small basidiospores (2.7-4 μm diam); while, BPI 706086 has delimited-peristome and larger basidiospores (4.4-6.5 μm diam). Thus, besides the nomenclatural problems involving its protologue, G. trichiferum has ambiguous type collections, since no voucher was indicated in the original description.
When compared to G. hirsutum, the collection BPI 706086 of G. trichiferum shows that specimens have larger basidiospores (4.4-6.5 μm diam) and lighter hairs (light brown), while morphologic analysis of the collection PACA 15970 shows that the single basidioma is distinct from G. hirsutum by the non-delimited peristome and short hairs on the exoperidium.
The type collections of G. trichiferum (S7 Fig) were also compared to G. pusillipilosum, a morphologically similar species, which also exhibits a hairy exoperidium. Our morphological analysis demonstrated that they can be differentiated mainly by basidiospore size and ornamentation. The collection PACA 15970 is distinguished from G. pusillipilosum by its nondelimited peristome and smaller basidiospores (2.7-4 μm diam) with inconspicuous ornamentation under LM, composed of small (0.1-0.6 μm high) warts with rounded tips under SEM; while the specimens of the BPI 706086 collection grow on a developed subiculum, the hairs on the exoperidium are longer (0.8-1.3 mm high), basidiospores have similar ornamentation: inconspicuous under LM, small (0.2-0.6 μm high) warts with rounded wart tips under SEM. Unfortunately, DNA extraction was not allowed by the herbarium and definitive conclusions are not possible.
We note that it is necessary to be extremely careful with species synonymization and consequent under-estimating of biodiversity. The integration of molecular and bioinformatic approaches for taxonomic and systematic studies seems to be essential for species delimitations in Geastrum, especially when dealing with species complexes.