In the phylum Glomeromycota, the largest group is represented by species producing glomoid spores, which arise blastically at tips of cylindrical or funnel-shaped sporogenous hyphae, as spores of Glomus macrocarpum (see Supplementary Table 1 for species authors), the type species of Glomus and the Glomeromycota (Clements and Shear, 1931; Schüßler and Walker, 2010; Oehl et al., 2011). Of the 330 known species of the Glomeromycota, approximately 60 were originally described to form glomoid spores in epigeous or hypogeous unorganized glomerocarps, i.e., fruit bodies with randomly distributed spores inside them (Gerdemann and Trappe, 1974; Morton, 1988; Jobim et al., 2019). Importantly, of the glomerocarpic species, only seven (Diversispora epigaea, Funneliformis mosseae, Glomus arborense, G. macrocarpum, Glomus pallidum, Glomus tenerum, and Glomus warcupii) were managed to grow in culture, and only eight (D. epigaea, Diversispora sporocarpia, G. macrocarpum, Redeckera megalocarpa, Redeckera fulva, Redeckera pulvinata, Sclerocarpum amazonicum, and Sclerocystis sinuosa) were provided with molecular data (Jobim et al., 2019), which essentially makes this group of fungi ascribed to the Glomeromycota difficult to characterize and classify.

Among many unorganized glomerocarps, previously found from our studies in the Brazilian Northeast, there were two whose glomoid spores first seemed to be identical in morphology to each another but differed from glomoid spores of described species of the Glomeromycota. Further studies of the phenotypic features of the spores revealed very small differences and, consequently, suggested that the glomerocarps represent two semicryptic undescribed species. BLAST searchers using initial molecular data about these fungi confirmed this hypothesis and showed that these species are probably related to the genera Dominikia and Kamienskia in the family Glomeraceae.

“Crypticity” in fungal taxonomy is an aspect known for several phyla (Matute and Sepúlveda, 2019). Also, molecular analyses of some morphological species of the Glomeromycota suggest that they contain cryptic phylogenetic species, in which differences in morphology do not occur, are very small and inconclusive, or are invisible using traditional research methods and tools. Examples of such species are Rhizoglomus irregulare and Septoglomus constrictum (Stockinger et al., 2014; Błaszkowski, pers. observ.).

The disclosure of cryptic species strongly depends on the molecular markers used in phylogenetic identification. Currently, of the many markers that have been used in phylogenetic studies of the Glomeromycota (Błaszkowski et al., 2018a, b, 2019a, b; Jobim et al., 2019; Magurno et al., 2019), those comprising the partial small subunit (18S), internal transcribed spacer (ITS1-5.8S-ITS2 = ITS), and partial large subunit (28S) nuc rDNA segment (= 18S-ITS-28S), as well as the largest subunit of RNA polymerase II (rpb1) gene, have the highest resolution power and characterize the highest proportion of described species within the Glomeromycota (Krüger et al., 2012; Stockinger et al., 2014; Al-Yahya’ei et al., 2017). The first marker allowed to separate, for example, Rhizoglomus dunense from Rhizoglomus clarum, whose morphologies are almost identical, but sequence divergence is very high (13.3%; Al-Yahya’ei et al., 2017). The sequence resolution of the rpb1 gene is higher and indicating that R. irregulare likely contains at least one cryptic species (Stockinger et al., 2014) and strongly confirming the uniqueness of R. dunense with a dissimilarity of 8.1% from R. clarum. According to literature data and our analyses, 18S-ITS-28S and rpb1 sequences show interspecies differences when they differ by ≥3% and approximately 1%, respectively (Krüger et al., 2012; Stockinger et al., 2014; Błaszkowski et al., 2019b; Corazon-Guivin et al., 2019b). The large proportion of species provided with molecular data contained in the two markers allows both to compare a relatively large number of species and to better define the phylogenetic divergences between species belonging to different higher taxonomic ranks. Moreover, it is widely recognized that the concatenation of sequences of the ribosomal RNA-coding genes with sequences of protein-coding genes significantly increases phylogenetic resolving power for revealing relationships between analyzed taxa (Miadlikowska et al., 2014).

The genera Dominikia and Kamienskia were erected in the phylum Glomeromycota mainly due to phylogenetic analyses of 18S-ITS-28S sequences of six species originally described in the genus Glomus and the newly described Dominikia disticha (Błaszkowski et al., 2015a). The type species of these new genera became Dominikia minuta and Kamienskia bistrata. The main characters shared by the two fungi were hyaline or pale-colored, small spores (10- to 65-μm diameter when globose) formed in hypogeous clusters, which were produced in pot trap cultures, but were not found in the field-collected rhizosphere samples used to inoculate the cultures. Therefore, it was suggested that species of Dominikia and Kamienskia sporulate rarely or seasonally and were not found in the field soils because their delicate spores quickly break down before soil sampling (Błaszkowski et al., 2015a).

Later, new species of Dominikia and Kamienskia were described, whose spores were either (i) similar in color and size (Dominikia lithuanica, Dominikia litorea, Kamienskia divaricata), (ii) darker-colored (Dominikia aurea, Dominikia bernensis, Dominikia compressa, Dominikia duoreactiva, Dominikia emiratia), (iii) larger (D. compressa, D. duoreactiva, D. emiratia), and (iv) formed mainly singly (Dominikia difficilevidera) (Oehl et al., 2014, 2015; Błaszkowski et al., 2015b, 2016, 2018c; Al-Yahya’ei et al., 2017) compared to the spores of the species listed above.

Recently, Kamienskia perpusilla and K. divaricata were transferred to a newly erected genus, Microkamienskia with a newly described species, Microkamienskia peruviana (Corazon-Guivin et al., 2019a), and D. litorea and D. emiratia became members of the new genera Microdominikia and Orientoglomus, respectively (Corazon-Guivin et al., 2019b). In addition, Corazon-Guivin et al. (2019b) described a new, monospecific genus, Nanoglomus, with Nanoglomus plukenetiae, sp. nov., with very similar morphology to K. perpusilla, but phylogenetically closely related to species of clades previously treated as Dominikia. All the taxa were erected mainly based on phylogenetic analyses of 18S-ITS-28S sequences and divergences of the sequences from other most closely related species, although Corazon-Guivin et al. (2019b) also showed morphological differences in the spore wall and subtending hypha between these taxa.

All described species of Dominikia and Kamienskia and their closest molecular relatives possess morphological characters distinguishing them from each other. However, none of these species has a morphological synapomorphy that would clearly demonstrate a generic separateness.

Recently, a new AMF, namely, Rhizophagus neocaledonicus, was described, which in the field lived in symbiosis with metallophytes in ultramafic soil in New Caledonia (Crossay et al., 2018). It was soon reassigned to the genus Rhizoglomus, as Rhizoglomus neocaledonicum (Turrini et al., 2018), because the generic name Rhizophagus should not be assigned to any species in the Glomeromycota (Sieverding et al., 2014). The assignment of this species into the genus Rhizoglomus (treated as Rhizophagus) was made based on a maximum likelihood phylogenetic analysis of 18S-ITS-28S sequences, even though the Rhizoglomus clade with the basally positioned R. neocaledonicum obtained only 49% bootstrap support. Moreover, the genetic divergence of R. neocaledonicum from the six Rhizoglomus species considered in the analysis was in the order of values separating clades of the Glomeromycota at the rank of genus. The sister species to the Rhizoglomus clade was a member of the genus Sclerocystis, S. sinuosa (Crossay et al., 2018). Molecular phylogenetic analyses performed by Turrini et al. (2018) confirmed the basal position and strong genetic divergence of R. neocaledonicum in the Rhizoglomus clade. However, they did not include S. sinuosa in their analyses. Further phylogenetic analyses of 18S-ITS-28S sequences of all Rhizoglomus species with available DNA barcodes and representatives of the other genera of the Glomeraceae accommodated R. neocaledonicum between the recently described monospecific genus Halonatospora, represented by Halonatospora pansihalos, and S. sinuosa, of which the latter was located basally to the other species of Rhizoglomus (Błaszkowski et al., 2018b, 2019b). Thus, the information presented above suggested that the New Caledonian fungus does not belong to Rhizoglomus but represents a new clade at the rank of genus in the Glomeraceae.

Therefore, the aims of this study were (i) to document that the two Brazilian fungi are new species of the Glomeromycota, (ii) to determine the molecular phylogenetic position of these fungi among their generic relatives of known phylogenies, and (iii) to describe in detail the morphology of these fungi. Moreover, we tested the hypothesis that R. neocaledonicum is a member of an undescribed genus in the Glomeraceae using already available and newly collected morphological and molecular data about this species.

As we expected based on exploratory morphological and molecular phylogenetic analyses, our extended phylogenetic analyses of sequence alignments containing representatives of all genera of the Glomeraceae, using 18S-ITS-28S and rpb1 data, unambiguously and strongly confirmed our hypothesis that the two glomerocarps found in Northeast Brazil are two new species of the Glomeromycota (Figures 1–3). Surprisingly, these morphologically almost identical fungi represent strongly phylogenetically divergent taxa in the Glomeraceae, one belonging to Dominikia and the second neighboring the monospecific genus Kamienskia (Figures 1–3). The divergences of 18S-ITS-28S and rpb1 sequences between the new species, D. glomerocarpica and E. crypticum, were 18.5–19.0% and 11.0–11.8%, respectively.

Another interesting finding in our analyses was that the sister species of D. glomerocarpica was D. aurea (Figures 1–3), which presents significant morphological differences. D. glomerocarpica is distinguished morphologically by (i) the formation of a compact, epigeous glomerocarp with a peridium; (ii) the production of small, colorless to light yellow (4A4), glomoid spores; (iii) the possession of a spore wall consisting of three permanent, smooth layers, of which the colorless to light yellow (4A4) laminate layer 2 is distinctly thick compared to the spore size and stains dark in Melzer’s reagent; and (iv) the formation of an invaginated septum continuous with spore wall layer 3, which is usually located at half the length of the channel connecting the lumen of the spore subtending hypha with the spore interior (Figures 4A–H).

D. aurea was originally described to produce hypogeous, peridium-free, 450–1,200 μm × 510–1,600 μm diameter, glomerocarps (called sporocarps), consisting of tightly packed spores. However, the microphotographs presented in the article of Oehl et al. (2003) and our specimens testify that D. aurea spores arise mainly in compact, hypogeous clusters (Błaszkowski, 2012) rather than in typical unorganized glomerocarps, as the much larger, epigeous glomerocarp with a peridium formed by D. glomerocarpica. Also, we often extracted loose clusters with few spores and single spores of D. aurea. Moreover, D. aurea spores are much darker colored and never are colorless at maturity, have a 1.2- to 2.4-fold thinner and only two-layered spore wall, of which layer 1 is impermanent, usually completely sloughed off in mature spores, and stains in Melzer’s reagent. Finally, the pore of the spore subtending hypha is closed by a curved septum, which is continuous with the innermost laminae of the laminate spore wall layer 2 and is located at or up to 5 μm below the spore base (Oehl et al., 2003; Błaszkowski, 2012). The clear distinctiveness of D. aurea and D. glomerocarpica also proved the large molecular divergences between their 18S-ITS-28S and rpb1 sequences, which were 8.5–10.2% and 4.3–4.8%, respectively.

Another surprising finding of our efforts to establish the taxonomic status of the Brazilian fungi was the placement of the previously named “Species 2” in a sister clade to the monospecific, generic clade represented by K. bistrata (Figures 1–3). However, the two species differ fundamentally in morphology and are separated from one another by a very large molecular distance (11.0–15.5%). Therefore, we erected a new genus for “Species 2,” named Epigeocarpum, typified by E. crypticum, despite the fact that the morphological characters distinguishing and clearly separating this species from K. bistrata are shared by other members of the Glomeraceae, as discussed below.

The distinctive morphological features of E. crypticum are (i) the production of small, colorless or light yellow (4A5), glomoid spores in a compact, epigeous glomerocarp with a peridium; (ii) a spore wall composed of three permanent, smooth layers, of which the much thicker laminate layer 2 stains dark in Melzer’s reagent and, most importantly, contracts and transfers into a crown-like structure in spores crushed in PVLG and PVLG + Melzer’s reagent; and (iii) the formation of a septum by spore wall layer 3 of characters and location similar to those of the septum of D. glomerocarpica spores (Figures 5A–H).

K. bistrata does not produce epigeous glomerocarps with a peridium, but its spores are grouped in loose to compact hypogeous clusters (Błaszkowski et al., 2009a; Błaszkowski, 2012). Most importantly, the spore wall of K. bistrata is 2.2- to 4.4-fold thinner than the E. crypticum’s, consists of only two colorless layers, of which the uniform layer 1 is thicker than the laminate layer 2, none of these layers stains in Melzer’s reagent, and the channel connecting the lumen of the spore subtending hypha with the spore interior is not closed by any septum.

As mentioned above, one of the distinctive features of the two here newly described species is the location of the septum separating the lumen of the spore subtending hypha from the spore interior (Figures 4C,F,G, 5G). The septum usually occurs at half the thickness of the laminate spore wall layer 2 as similarly described for Rhizoglomus maiae (Błaszkowski et al., 2019b), whereas for all other glomoid spore-producing species, the septum is formed at or slightly below the spore base. Moreover, R. maiae also produces compact, epigeous clusters with spores, whose spore wall structure and phenotypic properties of its components are like those of D. glomerocarpica and E. crypticum, except for the contractility of spore wall layer 2 of E. crypticum (Figures 5D–F). Thus, the location of the septum in these three species, belonging to three far related genera of the Glomeraceae, clearly proves that this feature is phylogenetically uninformative.

The formation of the septum much above the spore base in D. glomerocarpica, E. crypticum, and R. maiae probably is associated with the spore wall ontogeny and particularly with the differentiation of the distinctively thick laminate spore wall layer 2 by these fungi. As ontogenetic studies of spores of Claroideoglomus claroideum revealed, the innermost spore wall layer 4, forming an invaginated septum at or slightly below the spore base, arises after the full differentiation of the laminate spore wall layer 3 (Stürmer and Morton, 1997). Most likely, a similar sequence of events occurs during the development of the spore wall of D. glomerocarpica, E. crypticum, and R. maiae. However, the laminate spore wall layer 3 of C. claroideum is a relatively thin component of the entire spore wall, and consequently, the full differentiation of this layer is faster than that of the much thicker laminate spore wall layer 2 of the three main species discussed above. Therefore, the formation of the invagination of spore wall layer 4 of C. claroideum, shaped by the available space (the channel connecting the spore subtending hyphal lumen with the spore interior) in which the development of this layer is initiated, starts relatively earlier and closer to the spore base.

Another structure distinguishing E. crypticum is the contracting laminate spore wall layer 2, which takes the shape of a crown in spores crushed in PVLG-based mountants (Figures 5D–F). A spore wall layer of similar properties was also found in Microkamienskia perpusilla, originally described as Glomus perpusillum (Błaszkowski et al., 2009b), although the contracting spore wall layer 2 of M. perpusilla is a unit structure (not divided into visible sublayers) and does not transform into a crown following crushing of spores in PVLG-based mountants. Therefore, it seems that the contractility of spore wall layer 2 of E. crypticum is also not a generic synapomorphy of the new genus Epigeocarpum.

In summary, the sole characters convincingly rendering the genus Epigeocarpum unique are those residing in its DNA. Importantly, the genetic differences between the 18S-ITS-28S (14.5–15.6%) and rpb1 (11.0–11.6%) sequences of E. crypticum and K. bistrata are similar to those between sequences of directly neighboring species of most of the genera present in the tree depicted in Figures 1, 2. Several examples of pairwise comparisons are displayed in Supplementary Table 3. Of the described species producing unorganized glomerocarps with glomoid spores, whose molecular phylogeny is uncertain or unknown, D. glomerocarpica and E. crypticum most closely resemble Glomus microcarpum and Glomus nanolumen; the spores of these species are similar in color and size. The main differences readily separating the four species reside in their spore wall structure, the phenotypic and histochemical features of the spore wall components, and the characters of the pore connecting the subtending hyphal lumen with the spore interior.

Glomus microcarpum was originally characterized as having one, thick spore wall, which was also found by Gerdemann and Trappe (1974) and Berch and Fortin (1984). Instead, Błaszkowski (2012) defined and illustrated this species as producing spores with a spore wall consisting of two layers: a short-lived, quickly sloughing off, or semipermanent, slowly decomposing, thin layer 1, forming the spore surface, and a permanent, thick, laminate layer 2, staining dark red in Melzer’s reagent. Similar specimens identified as G. microcarpum with a spore wall as that characterized by Błaszkowski (2012) were presented by Oehl et al. (2011) and repeatedly found in Northeast Brazil (unpublished data). Of the Glomeromycota species grown in culture, only Sieverdingia tortuosa produced spores with a one-layered spore wall (Błaszkowski et al., 2019a), and in most of glomoid and glomoid-like spore-producing species, their laminate spore wall layer is covered with a relatively thin, usually short-lived, rarely permanent layer (Oehl et al., 2011; Błaszkowski, 2012). In field-collected spores, this outer spore wall layer is usually strongly or completely sloughed off, and therefore, it is difficult to detect or no longer present. On top of our knowledge, G. microcarpum has not yet been grown in culture, which makes impossible to examine spores at different developmental stages. However, regardless of the number of layers existing in the spore wall of the true G. microcarpum, the structure of the spore wall and the phenotypic properties of its components in D. glomerocarpica and E. crypticum (Figures 4B–H, 5B–H) differ clearly from those characterized above for G. microcarpum. The sequence identity, obtained by BLAST, of 18S-ITS-28S sequences of E. crypticum compared to 38 18S-ITS-28S sequences of G. microcarpum deposited in GenBank is only 87.5–92.02%. Unfortunately, no data were published about the morphology of the “donor” of the sequences ascribed to G. microcarpum. Thus, there is an urgent need for unambiguous morphological and molecular characterization of G. microcarpum, one of the first two described species currently classified in the phylum Glomeromycota and until 2010 regarded as the type species of Glomus (e.g., Gerdemann and Trappe, 1974).

Regarding G. nanolumen, its glomerocarps do not contain a peridium, and most importantly, their spores have a two-layered spore wall only, of which the laminate layer 2 is not uniform in thickness but differs two to three times in its different regions, and none of these layers stains in Melzer’s reagent (Koske and Gemma, 1989; Błaszkowski, 2012). Moreover, the subtending hypha of G. nanolumen is 1.6- to 2.5-fold wider at the spore base, and the channel connecting the subtending hyphal lumen with the spore interior is open.

Three evidences unambiguously justified the erection of the new monospecific genus Silvaspora in the Glomeraceae for the New Caledonian fungus, originally described as R. neocaledonicus (Crossay et al., 2018) and here renamed S. neocaledonica. First, our phylogenetic analyses of the 18S-ITS-28S and 18S-ITS-28S + rpb1 alignments, containing sequences of all sequenced Rhizoglomus species, placed the New Caledonian fungus in a clade at the rank of genus, and importantly, the clade was located in the sister relation to the genus Sclerocystis, which was outside the Rhizoglomus clade (Figures 1, 3). Thus, assigning the New Caledonian fungus to Rhizoglomus would make this genus polyphyletic. Moreover, although the analyses of rpb1 sequences clustered the New Caledonian fungus with Rhizoglomus spp., this grouping did not obtain either BI or ML support (Figure 2). Second, the divergences of 18S-ITS-28S and rpb1 sequences of the New Caledonian fungus from those of S. sinuosa were 22.7–22.9% and 11.4%, respectively. Thus, these divergence values were similar to the sequence divergences of other neighboring genera in the Glomeraceae (see above). Third, the mode of formation and morphology of spores of S. sinuosa differ fundamentally from those of S. neocaledonica (Gerdemann and Bakshi, 1976; Błaszkowski, 2012; Crossay et al., 2018; pers. observ.). All Sclerocystis species sensu Schüßler and Walker (2010) produce organized glomerocarps, in which glomoid-like spores arise radially from a central sterile plexus of mycelium, and the spores have a one-layered spore wall. Instead, S. neocaledonica produces spores singly or in clusters of 2 to 3, and their spore wall consists of three layers (Figures 6A–D; Crossay et al., 2018; pers. observ.).

We did not include into our phylogenetic analyses the monospecific genus Simiglomus with Simiglomus hoi because it was erected based on two sequences of the 18S gene (Oehl et al., 2011), which in our alignments is represented only by approximately 240 base pairs, and the origination of these sequences is uncertain, as stated by Redecker et al. (2013). Thus, further studies are needed to clarify the status of Simiglomus in the Glomeraceae.

The known glomerocarpic species probably constitute a small part of those existing in nature, mainly due to the difficulty in finding and characterizing these fungi. Collecting glomerocarps by raking and searching the upper soil layer and buried plant fragments is much more time-consuming and tiring than the wet-sieving and decanting method (Gerdemann and Nicolson, 1963) commonly used to find specimens of the Glomeromycota. Interestingly, Gerdemann and Trappe (1974) stated that finding glomerocarps is also related to the acquisition of a so-called hypogeous instinct. Attempts to grow glomerocarpic fungi in culture rarely succeeded, and therefore their characterization, defined only on the basis of field-collected specimens, may lack accurateness. Besides, the morphological and molecular characterization of these fungi may be impossible due to the degradation of their physical structures and DNA (pers. observ.). Importantly, the few described glomerocarpic species with known molecular phylogeny belong to various taxonomic groups in the Glomeromycota, and frequently they miss genus-specific morphological characters. Moreover, molecular phylogenetic analyses have shown that glomerocarpic-like fruit bodies with glomoid-like spores are also produced by fungal taxa not belonging to the Glomeromycota (Yamamoto et al., 2020). Also, the trophic status of glomerocarpic and glomerocarpic-like species is unclear, particularly that of epigeocarpic species. In order to know the real representativeness of glomerocarpic species in the Glomeromycota and determine their trophic and taxonomic status, the collection of specimens of this group of fungi and the study of their ecology and phylogeny should be significantly intensified.

The datasets generated for this study can be found in online repositories. The names of the repository/repositories and accession number(s) are as follows: GenBank MW507148-MW507157, MW541060-MW541067, and MW249072-MW249073.

BG, FM, JB, PN, and SZ were responsible for conceptualization and funding acquisition. BG, FM, JB, and PN were responsible for data acquisition. FM, JB, PN, TB, and WB performed analyses. JB and FM wrote the first draft. BG, EM, FM, JB, KJ, LC, PN, RM, PM, SZ, TC, and WB wrote, reviewed, and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.