A new order, Entrophosporales, and three new Entrophospora species in Glomeromycota

As a result of phylogenomic, phylogenetic, and morphological analyses of members of the genus Claroideoglomus, four potential new glomoid spore-producing species and Entrophospora infrequens, a new order, Entrophosporales, with one family, Entrophosporaceae (=Claroideoglomeraceae), was erected in the phylum Glomeromycota. The phylogenomic analyses recovered the Entrophosporales as sister to a clade formed by Diversisporales and Glomeraceae. The strongly conserved entrophosporoid morph of E. infrequens, provided with a newly designated epitype, was shown to represent a group of cryptic species with the potential to produce different glomoid morphs. Of the four potential new species, three enriched the Entrophosporales as new Entrophospora species, E. argentinensis, E. glacialis, and E. furrazolae, which originated from Argentina, Sweden, Oman, and Poland. The fourth fungus appeared to be a glomoid morph of the E. infrequens epitype. The physical association of the E. infrequens entrophosporoid and glomoid morphs was reported and illustrated here for the first time. The phylogenetic analyses, using nuc rDNA and rpb1 concatenated sequences, confirmed the previous conclusion that the genus Albahypha in the family Entrophosporaceae sensu Oehl et al. is an unsupported taxon. Finally, the descriptions of the Glomerales, Entrophosporaceae, and Entrophospora were emended and new nomenclatural combinations were introduced.


Introduction
The genus Entrophospora with the type species E. infrequens was erected in the family Endogonaceae based on a species originally described as Glomus infrequens (Hall, 1977), but later recognized to form spores differently (inside the neck of a sporiferous saccule; Ames and Schneider, 1979) than species producing glomoid spores sensu Morton and Redecker (2001). Glomoid spores arise at the tips of sporogenous hyphae as in G. macrocarpum, the type species of the genus Glomus and the phylum Glomeromycota, including arbuscular mycorrhizal fungi (AMF; Clements and Shear, 1931;Schüßler and Walker, 2010;Oehl et al., 2011a). Subsequently, Entrophospora was enlarged by E. schenckii (Sieverding and Toro, 1987) and E. baltica (Błaszkowski et al., 1998), producing two-walled spores with a thick, smooth wall 2 as in E. infrequens, as well as E. colombiana (Schenck et al., 1984), and E. kentinensis (Wu et al., 1995) having spores with three spore walls, of which the innermost wall 3 is relatively thin and usually ornamented ("beaded"), as in most Acaulospora species. Morton and Benny (1990) transferred the entrophosporoid [name coined by Sieverding and Oehl (2006)] species mentioned above to the family Acaulosporaceae in the order Glomales. Sieverding and Oehl (2006) accommodated E. baltica and E. infrequens in a newly erected family, Entrophosporaceae, in the class Glomeromycetes; E. colombiana and E. kentinensis in a new genus, Kuklospora, in the Acaulosporaceae; and E. schenckii in Intraspora gen. nov. in the family Archaeosporaceae sensu Morton and Redecker (2001). Schüßler and Walker (2010) introduced Claroideoglomeraceae fam. nov. with Claroideoglomus gen. nov. into the Glomerales (orthographically corrected Glomales; Schüßler et al., 2001). The type species of the monogeneric Claroideoglomeraceae, Claroideoglomus, was C. claroideum (Schüßler and Walker, 2010), originally described as Glomus claroides (Schenck and Smith, 1982). Entrophosporaceae and Entrophospora were recognized as taxa of uncertain systematic position in the order Diversisporales, Kuklospora synonymized with Acaulospora, and I. schenckii renamed as Archaeospora schenckii. Kaonongbua et al. (2010), based on a phylogeny reconstructed from nuc 28S rDNA (28S) sequences, also rejected Kuklospora as a valid monophyletic group, therefore, integrated it into Acaulospora, and concluded that the mode of spore formation is not a genus-specific character. Oehl et al. (2011b) synonymized Claroideoglomeraceae with Entrophosporaceae based on phylogenetic analyses of nuc 18S rDNA (18S) and 28S sequences of E. infrequens, three species of Claroideoglomus, other representatives of Glomerales, and members of eleven (in the 18S analyses) and nine (28S) other previously described families of Glomeromycota. The genus Claroideoglomus remained in Entrophosporaceae, and Entrophospora with E. infrequens became the type genus of this family. The same analyses also prompted Oehl et al. (2011b) to transfer C. drummondii and C. walkeri, originally described as G. drummondii and G. walkeri (Błaszkowski et al., 2006), to a newly erected genus, Albahypha, with A. drummondii and A. walkeri, and G. viscosum to a new genus, Viscospora, in the Entrophosporaceae. In addition, Oehl et al. (2011b) accommodated E. baltica in Sacculosporaceae fam. nov., as S. baltica in the Diversisporales.
Except for the introduction of the Sacculosporaceae, Redecker et al. (2013) rejected the other changes discussed above and resurrected Claroideoglomeraceae. They found that both 18S and 28S sequences of E. infrequens nest among sequences of Claroideoglomus, and the incongruence between morphological and molecular characters of these taxa could not be explained from either dataset.
The main argument for rejecting Albahypha was linked to analyses conducted by Krüger et al. (2012), which indicated sequences of A. drummondii and A. walkeri clustering among Claroideoglomus sequences. In addition, Redecker et al. (2013) underlined that the generic clade Albahypha in Oehl's et al. (2011b) 28S tree was not supported and, therefore, rendered Claroideoglomus paraphyletic.
In Schüßler and Walker's (2010) classification, Claroideoglomeraceae was a sister taxon of Glomeraceae in the Glomerales. Two phylogenies have recently been published suggesting other relationships of Claroideoglomeraceae within Glomeromycota sensu Redecker et al. (2013). Beaudet et al. (2018), using spore transcriptomic data for analyses of nine taxa from seven families, recovered Claroideoglomus as sister to Ambispora and Paraglomus, belonging to the orders Archaeosporales and Paraglomerales, respectively. Montoliu-Nerin et al. (2021), using single nuclei sequencing to obtain new genomes of 15 species belonging to seven families, indicated Frontiers in Microbiology 02 frontiersin.org that Claroideoglomeraceae represents an autonomous, wellsupported group, sister to a clade formed by Glomeraceae and Diversisporales. Species of Claroideoglomus, Glomeraceae, and Paraglomus produce solely glomoid spores, while Ambispora species are dimorphic, forming acaulosporoid and glomoid morphs (Bills and Morton, 2015). Of members of the genera of the Diversisporales, only those of Corymbiglomus, Desertispora, Diversispora, Sieverdingia, Redeckera, and Pacispora produce glomoid spores (Oehl et al., 2011c;Błaszkowski, 2012;Symanczik et al., 2018;Błaszkowski et al., 2019). Glomoid sporeproducing species of the genera mentioned above have onewalled spores, except for species of Pacispora and Paraglomus, which produce spores with two spore walls (Mello et al., 2013). Schüßler and Walker (2010) and Oehl et al. (2011a) stated that the spores of Claroideoglomus species are morphologically distinguished by the composition of their spore wall and the phenotypic features of the spore wall components, as well as the characters of the spore subtending hypha at the spore base. According to Schüßler and Walker (2010), the spore wall of the members of Claroideoglomus contains a flexible, thin, colorless innermost layer. Oehl et al. (2011a) distinguished Claroideoglomus spores as having conspicuously funnel-or bill-shaped and colorless subtending hyphae regardless of whether the spore wall is colored or not. However, the above-mentioned morphological features not only characterize all Claroideoglomus species but also define glomoid spores of members of several other genera of Glomeromycota, for example, Diversispora species and Orientoglomus emiratium, a member of Glomeraceae (Błaszkowski et al., 2022). So, a reliable separation of Claroideoglomus species from other Glomeromycota species producing glomoid spores based exclusively on their morphology would be biased.
The data presented above confirm the widely accepted opinion that a more accurate identification and classification of AMF, including the Claroideoglomeraceae/Entrophosporaceae members, needs to be based mainly on reliable molecular data. Short sequences are usually the source of weak signals, resulting in insignificant support values for topologies of phylogenetic trees (Redecker et al., 2013). Even full-length 28S sequences failed to resolve closely related AM fungal species (Redecker et al., 2013). The  following  hypotheses  might  explain  the  cause  of  the  divergent  conclusions  about the Claroideoglomeraceae/Entrophosporaceae, as well as the phylogenies of E. infrequens and the two species considered as C./A. drummondii and C./A. walkeri. First, as suggested by Montoliu-Nerin et al. (2021), the conclusions of Beaudet et al. (2018) were biased by the low number of taxa analyzed. Second, we considered the vague phylogenies of E. infrequens, C./A. drummondii, and C./A. walkeri that resulted from the sequence under-sampling of the taxa and the use of sequences with too low taxonomic resolution. Therefore, we assumed that (i) genome-scale phylogenetic analyses will shed light on the phylogenetic placement of the family Claroideoglomeraceae/Entrophosporaceae and (ii) increased sampling and the use of sequences covering the 18S (partial), ITS1-5.8S-ITS2, and 28S (partial) nuc rDNA (=45S) segment concatenated with partial sequences of the largest subunit of RNA polymerase II (rpb1) will clarify the E. infrequens, C./A drummondii, and C./A. walkeri phylogenies. Sequences of each of the two loci generally separated even very closely related species (Krüger et al., 2012;Kohout et al., 2014;Stockinger et al., 2014). However, phylogenies reconstructed from concatenated sequences of the two unlinked loci (45S+rpb1) were more robust and revealed relationships unexposed when sequences of these two loci were analyzed separately (Błaszkowski et al., 2019(Błaszkowski et al., , 2021a(Błaszkowski et al., , 2015bYu et al., 2022).
We grew E. infrequens and four potential new glomoid spore-producing AMF species in trap and single-species cultures. The morphological characteristics of these glomoid fungi, particularly the colorless subtending hyphae of their pigmented spores, suggested that they may represent Albahypha/Claroideoglomus or Diversispora.
The aims of this study were (i) to determine which of the known hypotheses about the status and relationship of Claroideoglomeraceae/Entrophosporaceae is reliable, (ii) to clarify the position of E. infrequens in the Glomeromycota with the support of rpb1 as an additional phylogenetic marker, (iii) to characterize in detail the morphology of the four potentially new AMF species and determine their phylogenetic positions among sequenced species of Albahypha/Claroideoglomus based on 45S and rpb1 sequences, and (iv) to settle the conflict over the taxonomic status of Albahypha using previously available and newly generated morphological and molecular (45S and rpb1) data.

Materials and methods
Origin of the study material Entrophospora infrequens and the potential new species, initially named Claroideoglomus 1-4, were characterized based on spores extracted from single-species pot cultures. These cultures were established from spores extracted from trap pot cultures inoculated with field-collected mixtures of rhizosphere soils and root fragments of the following plant species. The field host of E. infrequens was Juniperus communis L. growing in a pine forest on inland sand dunes of Kampinos National Park (52 o 19 N, 20 o 45 E), Poland. The field sample was collected by J. Błaszkowski on 26 July 1986.
The regional climate is influenced by the proximity of the Atlantic Ocean (Paruelo et al., 1998). The mean air temperature is about 0 • C in July (winter) and reaches 10 • C in January (summer). The annual rainfall ranges from about 300 to 400 mm along a NE-SW gradient and is evenly distributed throughout the year. The field sample was collected by R. Mendoza, Museo Argentino de Ciencias Naturales (MACN-CONICET) "Bernardino Rivadavia, " Buenos Aires, Argentina, during the period from 1 to 15 March 2009.
Claroideoglomus 2 was harbored in the field within an initial community consisting of cryptogamic species and the following vascular plants: F. vivipara (L.) Sm., Po. alpina L., Salix herbacea L., S. polaris Wahlenb., and Silene acaulis (L.) Jacq. This community grew in a glacier foreland of Isfallglaciarän located in Tarfala valley, N Sweden (67 • 54 58 N, 18 • 35 E). The average annual air temperature and the annual sum of rainfall in this site are 3.4 • C and 503.3 mm, respectively (The Swedish Agency for Marine and Water Management, available at: www.smhi.se). The soil contained 19.28 and 3.17% of silt and clay, respectively, and its chemical properties were as follows: pH (in H 2 O), 6.3; P 2 O 5 (mg/100 g), 4.63; Na (cmol/kg), 0.04; N (%), 0.00; and C (%), 0.08. The field sample was collected by Paulina Wietrzyk-Pełka and Michał Węgrzyn (Institute of Botany of the Jagiellonian University) on 23 July 2019.
Claroideoglomus 4 was isolated from samples collected in the Sultanate of Oman and in Poland. In Oman, Claroideoglomus 4 was associated with roots of Prosopis cineraria (L.) Druce. It grew in an undisturbed habit adjacent to a date palm (Phoenix dactylifera L.) plantation of an experimental station (22 • 14 N, 59 • 10 E) belonging to the Ministry of Agriculture, Fisheries and Water Resources of Oman. Summer temperatures exceed 48 • C (Glennie and Singhvi, 2002). Based on the aridity index defined by the United Nations Environmental Program (UNEP, 2006), this site is characterized as Hyperarid (AI < 0.05), with annual rainfall not exceeding 100 mm (Fisher and Membery, 1998). The soil of this site was sandy loam with the following chemical properties: pH (in H 2 O), 8.2; contents of P (mg kg −1 ), 41.4; N (%), < 0.1; and organic matter (%), 1.5 (Al-Yahya'ei et al., 2011). The field sample was collected by Mohamed N. Al-Yahya'ei in August 2006.
The site in Poland where Claroideoglomus 4 was found is a field near Lubliniec in the Silesian Upland (50 • 37 N, 18 • 43 E) from where Dominikia bonfanteae was isolated (Błaszkowski et al., 2021b). Details about plant community and the physicochemical properties of soil are available in Malicka et al. (2020). The climate of the region, where the study material was collected, is characterized by Błaszkowski et al. (2019).
Spores of the species originally described as G. walkeri and C. hanlinii, the latter of which was also included in our analyses (see below), originated from single-species cultures grown by J. Błaszkowski. Establishment and growth of trap and single-species cultures, extraction of spores, and staining of mycorrhizal structures Methods used to establish trap and single-species cultures, growing conditions, and methods of spore extraction and staining of mycorrhizal structures were as those described previously . Five to ten spores of uniform morphology of each AMF species were used to establish single-species cultures.

Microscopy and nomenclature
Morphological features of spores and phenotypic and histochemical characters of spore wall layers of the new species presented here were characterized based on at least 50-100 spores of each species mounted in water, lactic acid, polyvinyl alcohol/lactic acid/glycerol (PVLG, Omar et al., 1979), and a mixture of PVLG and Melzer's reagent (1:1, v/v). The preparation of spores for study and photography was as described previously (Błaszkowski, 2012;Błaszkowski et al., 2012). The types of spore wall layers were defined by Walker (1983) and Błaszkowski (2012). Color names were from Kornerup and Wanscher (1983). The nomenclature of fungi and the authors of fungal names are from the Index Fungorum website http://www. indexfungorum.org/AuthorsOfFungalNames.htm. The term "glomerospores" was used for spores produced by AMF as proposed by Goto and Maia (2006

DNA extraction, PCR, cloning, and DNA sequencing
Genomic DNA for obtaining 45S and rpb1 sequences was extracted from eight single spores of E. infrequens and each of the four potential new Claroideoglomus species. Details of the treatment of the spores prior to PCR, conditions and primers used for PCR, as well as cloning and sequencing of PCR products to obtain 45S sequences of the species are available in Krüger et al. (2009), Symanczik et al. (2014, and Błaszkowski et al. (2015b,c).
To obtain rpb1 sequences of these five species, as well as C. hanlinii, C. luteum, and the species originally described as G. walkeri, a new set of primers was developed. The primer set consisted of two forward primers (RPB1-3F: GTC TTC GTG CAG TTT GGG A; RPB1-4F: CTA GGC CTG ATT GGA TGA T) and two reverse primers (RPB1-5R: ACG ATT TGT TTT GGT ACC AT; RPB1-5RN: TTC ATC TCA CCA TCA A). Characteristics of the primers and the PCR conditions are available in Błaszkowski et al. (2021a). To increase the chance to obtain a successful amplification of unknown sequences, the following approach was used to design the primers. All the possible 19-21 bases oligos obtained from the rpb1 sequence available for C. claroideum were assessed according to the following criteria: (i) a perfect match with the C. etunicatum and C. claroideum sequences, (ii) at least one C/G in the 3 end, and (iii) the highest number of matches among all the AMF rpb1 sequences available (about 200) with maximum three mismatches tolerated if not in the last five residues at the 3 end. Candidate oligos were evaluated with OligoAnalyzer for Tm and the presence of secondary structures.

Phylogenomic analyses
Previous genome and transcriptome data phylogenies have shown two conflicting topologies with respect to Claroideoglomeraceae (Beaudet et al., 2018;Montoliu-Nerin et al., 2021). Therefore, we compiled three different datasets based on previously published data (Supplementary Table 1). The first dataset included data used by Beaudet et al. (2018) and Montoliu-Nerin et al. (2021), the second dataset included only Beaudet et al.'s (2018) data, and the third dataset included only genomes from the Montoliu-Nerin et al. (2021) analyses. Single-copy orthologs (SCOs) were identified with OrthoMCL (Li et al., 2003) using default settings, and only SCOs that were present in at least 50% of the taxa were included in the phylogenetic analyses. Individual SCOs were aligned with MAFFT 7.407 (Katoh and Standley, 2013). Poorly aligned regions were trimmed with trimAl 1.4.1 (Capella-Gutiérrez et al., 2009) with a gap threshold of 0.1. The individual alignments were concatenated using the script geneStitcher.py (Ballesteros and Hormiga, 2016). An ML phylogeny was inferred with IQ-TREE 2.0 (Nguyen et al., 2015) using ModelFinder (Kalyaanamoorthy et al., 2017) and searching for the best partition scheme. In addition, the individual SCO alignments were used to infer individual gene trees with IQ-TREE and then a phylogenetic inference with ASTRAL-III 5.7.3 (Zhang et al., 2018) was performed. The dataset that included only genome data used by Montoliu-Nerin et al. (2021) was then used to test the hypothesis that Claroideoglomeraceae is sister to Diversisporales and Glomeraceae. For this, we evaluated the support among individual gene trees for alternative branching orders and carried out a polytomy test in ASTRAL-III .

Phylogenetic analyses
To clarify the status of Albahypha and Entrophospora, as well as to infer the position of the four potential new species, two sequence datasets were produced. The first consisted of sequences of the 45S segment, or part thereof, of our E. infrequens and the AMF, initially recognized as new species of Claroideoglomus, as well as all other sequenced species of these genera sensu Oehl et al. (2011b). Claroideoglomus hanlinii, a species described later (Błaszkowski et al., 2015a), was also included. The second dataset contained rpb1 sequences available for the species of the first dataset and sequences from the genome of C. candidum. The sequences covered part of the fourth and fifth exons of the rpb1 gene and the intron between them. In each of these datasets, the outgroup was represented by Diversispora species, and the datasets were aligned separately using MAFFT 7 with the E-INS-i strategy. Subsequently, the two alignments were used to manually produce a concatenated alignment, 45S+rpb1. The 45S sequences lacking the rpb1 part were not removed from the alignment. An additional alignment was produced as follows: 45S+rpb1_G with three Glomus instead of Diversispora species in the outgroup to test the impact of this outgroup representing Glomeraceae, the sister family of Claroideoglomeraceae in the Glomerales sensu Schüßler and Walker (2010), on the topology and measures used in comparison of the trees generated in the analyses (described below).
The phylogenetic position of Albahypha, E. infrequens, and the four potential new species was reconstructed based on Bayesian inference (BI) and maximum likelihood (ML) phylogenetic analyses of the 45S, rpb1, 45S+rpb1, and 45S+rpb1_G alignments, performed via CIPRES Science Gateway 3.1 (Miller et al., 2010). The 45S alignment was divided into five partitions: 18S, ITS1, 5.8S, ITS2, and 28S. Five additional rpb1 partitions were added in the alignments with the concatenated genes, as described in Błaszkowski et al. (2021b).
As suggested by Abadi et al. (2019), in both BI and ML analyses, GTR+I+G was chosen as the nucleotide substitution model for each nucleotide partition. Substitution models selected by ModelTest-NG 0.1.5 (Darriba et al., 2020) were also tested in ML analyses, but the trees obtained final loglikelihood values lower compared to those where GTR+I+G was used.
Four Markov chains were run over two million generations in MrBayes 3.2 (Ronquist et al., 2012), sampling every 2,000 generations, with a burn-in at 3,000 sampled trees. The ML phylogenetic tree inference was performed with RAxML-NG 1.0.1 (Kozlov et al., 2019) using a maximum likelihood/1,000 bootstrapping run and an ML estimated proportion of invariable sites and base frequencies. The alignments and tree files were deposited as Supplementary material.
We assumed that clades were supported when the Bayesian posterior probabilities and the ML bootstrap values were ≥0.95 and ≥70%, respectively. The generated phylogenetic trees were visualized and edited in TreeGraph 2 (Stöver and Müller, 2010). To evaluate possible conflicts between the genes, the topologies of the ML trees (collapsed at bootstrap values <70%) were compared. In addition, the trees were compared based on three measures, referred to the ingroup, which are as follows: (i) the number of species clades supported with BI ≥ 0.95 and ML ≥ 70%, (ii) mean BI and ML values for species clades when supported, (iii) mean supports of nodes with BI ≥ 0.95 and ML ≥ 70%, and (iv) the amount of resolution of each tree. The amount of resolution was calculated for the ingroup of the tree by dividing the number of significantly supported internal branches by the size of the ingroup (Thorley and Wilkinson, 2000).
To detect other possible findings of the potential new species, their 45S sequences were used as queries in BLASTn to retrieve more than 300 nucleotide sequences from GenBank. The sequences were selected according to the percentage of identity > 96.5, with at least one of the queries. The possible relatedness with any of the four species was assessed using an evolutionary placement algorithm-based (EPA) approach, performed with RAXML 8.2.10 (Berger et al., 2011). The sequences were aligned in MAFFT together with the 45S dataset, and the 45S+rpb1 ML tree was used as reference support for the phylogenetic affiliation. GTRGAMMA model, the maximum likelihood estimation for the base frequencies, and the partition scheme previously described for the 45S alignment were used for the phylogenetic inference. The jplace output was used in Gappa (Czech et al., 2020) for the placement mass (likelihood weight ratio) accumulation of the placements of each sequence upward from the reference tree with the threshold = 0.9.

Results
General data and phylogeny The first dataset used for the phylogenomic analyses, which included data from Beaudet et al. (2018) and Montoliu-Nerin et al. (2021), contained 78 SCOs and recovered Claroideoglomeraceae as sister to Diversisporales and Glomeraceae (Supplementary Figure 1). Since Beaudet et al. (2018) showed Claroideoglomeraceae to be sister to Ambisporaceae and Paraglomeraceae based on analyses of spore transcriptomic data, we produced and analyzed a dataset that contained only these data. The analysis yielded 122 SCOs, and the phylogeny showed again Claroideoglomeraceae as sister to Diversisporales and Glomeraceae (Supplementary Figure 2). Finally, to obtain a higher number of SCOs and to perform a topology test, we excluded the Beaudet et al.'s (2018) dataset and analyzed only AM genome data used by Montoliu-Nerin et al. (2021). For this dataset, we obtained 1,260 SCOs present in at least 50% of the taxa, and the phylogenomic analysis also recovered Claroideoglomeraceae as sister to Diversisporales and Glomeraceae (Figure 1). Other phylogenomic analyses on every genome dataset consistently recovered Glomerales as polyphyletic, with Claroideoglomeraceae as sister to Diversisporales and Glomeraceae (data not shown). The quartet gene frequencies inferred with ASTRAL supported this relationship (q1 = 0.43, q2 = 0.26, q3 = 0.31) (Figure 2).
The phylogenetic position of Albahypha, Claroideoglomus, E. infrequens, and the four potential new species was reconstructed based on 115 sequences of the 45S segment or part thereof (28S) and 36 rpb1 sequences. Of these, 57 were newly obtained (39 45S, 18 rpb1). The 45S and rpb1 sequences characterized 14 and 10 species, respectively, of Albahypha/Claroideoglomus, including our four new species, as well as three Diversispora (Glomus) species that served as an outgroup. These sequences were used to prepare four alignments (45S, rpb1, 45S+rpb1, and 45S+rpb1_G) for BI and ML analyses. Data about the numbers of variable and parsimony informative sites of each alignment are presented in Table 1.
Four phylogenetic trees summarizing the BI and ML analyses were reconstructed, named 45S, rpb1, 45S+rpb1, and 45S+rpb1_G (Figure 3 and Supplementary Figures 3-5, respectively). Small differences that occurred in the topologies of the trees were described below. In all trees, the number of supported species clades and the mean support values of species clades and nodes were identical or similar (Supplementary Table 2). Overall, the trees with concatenated genes performed slightly better than the 45S tree. Best Maximum Likelihood phylogeny inferred with IQ-TREE from a concatenated alignment of 1260 single copy orthologs shared by at least 50% of the taxa. The same topology was recovered with ASTRAL III. Values near the branches correspond to bootstrap support (BS) from the ML analysis, local posterior probabilities (LPP), and quartet support from the ASTRAL analysis (BS/LPP/quartet support). Asterisks indicate maximum support (100 or 1.0). Mucoromycota and Mortierellomycota were used as outgroups. Bar indicates 0.2 expected change per site per branch.
In all trees, the clades with Claroideoglomus 1-4 obtained full or almost full BI (= 0.99-1.0) supports (Figure 3 and Supplementary Figures 3-5). Also, the ML supports of these species in all trees were significant (≥70%), and most of them were very high or full.
In all trees, the glomoid Claroideoglomus 3 formed a mixed clade with the entrophosporoid E. infrequens, whose 45S and rpb1 sequences were obtained from spores originating from our collection (Figure 3 and Supplementary  Figures 3-5). In the 45S+rpb1 and 45S trees, the glomoid Claroideoglomus 2 also grouped with E. infrequens (Figure 3 and Supplementary Figure 3), whose 28S sequences were obtained by Oehl et al. (2011b); no rpb1 sequence of the fungus was available (pers. comm.). Hereafter, the two mixed clades will be referred to as Claroideoglomus 2 and Claroideoglomus 3. Evaluation of support among 1260 individual gene trees for alternative hypotheses of the phylogenetic placement of Claroideoglomeraceae. Bar graphs represent the gene tree quartet frequencies for three possible branching orders. The dashed horizontal line indicates the expectation of a hard polytomy.   A 50% majority-rule consensus tree from the Bayesian analysis of 45S nuc rDNA sequences concatenated with rpb1 sequences of Claroideoglomus 1, 2, 4 (newly described as Entrophospora argentinensis, E. glacialis, and E. furrazolae, respectively), Claroideoglomus 3 (a glomoid morph of the E. infrequens epitype), eight other species of Claroideoglomus sensu C. Walker and A. Schüßler, and three Diversispora species serving as outgroup. The former species names included in Entrophosporales are reported between brackets. The new species and the accession numbers of the sequences obtained in this study are in bold. The Bayesian posterior probabilities ≥0.90 and ML bootstrap values ≥50% are shown near the branches, respectively. Bar indicates 0.002 expected change per site per branch. The two basal branches were shortened to 20% in length to improve visibility (indicated by //).

Taxonomy
The results of the phylogenomic and phylogenetic analyses described above prompted us (i) to propose a new order, Entrophosporales, for Entrophosporaceae in place of Claroideoglomeraceae, (ii) to emend the descriptions of Glomerales, Entrophosporaceae, Entrophospora, and E. infrequens, (iii) to designate an epitype of E. infrequens with the morphology and phylogeny determined from analyses of the entrophosporoid E. infrequens morph and the glomoid morph preliminary named Claroideoglomus 3, which are grown in our collection, (iv) to introduce new nomenclatural combinations, and (v) to describe Claroideoglomus 1, 2, and 4 as E. argentinensis sp. nov., E. glacialis sp. nov., and E. furrazolae sp. nov., respectively.
Diagnosis: Producing entrophosporoid and glomoid spores or only one of these morphs. Entrophosporoid spores formed singly within the necks of sporiferous saccules, in soil, rarely in roots. Spores with two spore walls. Spore wall 1, forming the spore surface, composed of two short-lived to semi-permanent layers continuous with the neck and sporiferous saccule wall layers and a permanent, laminate, pigmented layer continuous with a wall of a cylindric to a funnel-shaped structure supporting the wall of the part of the neck directly connected with the sporiferous saccule and the wall of the saccule at its base; not extending into the part of the neck located distally to the sporiferous saccule, hence spores have only one persistent cicatrix, proximal to the sporiferous saccule, created when the opening connecting the interior of the spore with the lumen of the sporiferous saccule neck is closed by a plug made of the contents of the spore when fully formed; layer 1 stains dark in Melzer's reagent. Spore wall 2, having no physical contact with spore wall 1, composed of three, hyaline, smooth layers: two thin, flexible layers enclosing a much thicker, semi-flexible layer; none of the layers stains in Melzer's reagent. Glomoid spores formed at tips of sporogenous hyphae either branched from the sporiferous saccule wall and/or the sporiferous saccule neck of the entrophosporoid morph or continuous with extraradical mycorrhizal hyphae, occasionally intercalarily inside hyphae, rarely in roots, usually with one subtending hypha. Extraradical spores produced singly, rarely in clusters with few spores. Spores with one spore wall consisting of two to five layers; when multilayered, the innermost layer frequently is flexible to semi-flexible, <1.0 µm thick, colorless or brightly colored, and loosely associated with the inner surface of the penultimate, usually laminate, layer. One or two spore wall layers may stain in Melzer's reagent. Subtending hypha with a hyaline, rarely brightly colored wall, always strikingly much lighter than the spore wall in colored spores, frequently conspicuously funnel-or bill-shaped at the spore base; subtending hyphal wall composed of layers continuous with spore wall layers. Pore closed by a septum connecting the inner surfaces of either the subtending hyphal wall or the spore wall, or septa connecting both these structures, at or slightly below the spore base, and frequently additionally by a septum continuous with the innermost flexible to semi-flexible spore wall layer. Entrophosporoid and glomoid species producing mycorrhiza with arbuscules, vesicles, as well as intra-and extraradical hyphae staining dark in Trypan blue. Diagnosis: Spores formed at tips of sporogenous hyphae, occasionally intercalarily inside hyphae, and in roots, usually with one subtending hypha. Extraradical spores produced singly, in clusters with few to dozens of spores, compact unorganized glomerocarps (= sporocarps) with randomly distributed spores or organized glomerocarps with spores born terminally from hyphae developed radially from a central plexus of hyphae. Glomerocarps with or without a peridium. Spores with one spore wall consisting of one laminate layer or many layers; when multilayered, the innermost layer usually is laminate and thicker than the others. Subtending hypha with a wall concolorous with or slightly lighter than the spore wall, funnel-shaped, cylindrical, or constricted at the spore base; subtending hyphal wall composed of layers continuous with spore wall layers. Pore closed by thickening spore wall and subtending hyphal wall toward the center of the subtending hyphal lumen connecting it with the spore interior, a septum connecting the inner surfaces of the subtending hyphal wall slightly below the spore base, or the innermost spore wall layer at or slightly above the spore base. Mycorrhiza with arbuscules, vesicles, as well as intra-and extraradical hyphae staining dark in Trypan blue. Diagnosis: Differs from other Entrophospora species forming glomoid spores in spore morphometric features, phenotypic and histochemical properties of spore wall layers, as well as in nucleotide composition of sequences of the 45S nuc rDNA region and the rpb1 gene.
Ecology and distribution: Entrophosporoid morph found associated with roots of many cultivated and uncultivated plant species growing in soils highly differing in pH and the content of nutrients, organic matter, and water (Sieverding and Oehl, 2006;Błaszkowski, 2012). Glomoid morph physically connected with an entrophosporoid morph grown in a singlespecies culture with the host plant P. lanceolata (Figures 4F,G) and probably lived in arbuscular mycorrhizal symbiosis with roots of L. sinuatum in the field, but no molecular analysis was performed to confirm this hypothesis. Reported as C. cf. claroideum, E. infrequens, Glomeromycota sp., and uncultured Glomus from Czech Republic, France, Peru, Poland, Spain, Switzerland, and USA (Supplementary Table 3).  ≡ Albahypha walkeri (Błaszk. and Renker) Sieverd., Oehl, B.T. Goto, and G.A. Silva. Mycotaxon 117:309. 2011.
Diagnosis: Differs from other species of Entrophospora in having a laminate spore wall layer uneven in thickness in its different regions and nucleotide composition of sequences of the 45S nuc rDNA region and the rpb1 gene.
Ecology and distribution: In the field, associated with roots of D. flexuosa and Po. rigidifolia in Tierra del Fuego, Argentina. Forming mycorrhiza with arbuscules, vesicles, as well as intraradical and extraradical hyphae in single-species cultures with P. lanceolata as the host (Figures 7A,B); structures stained violet white (16A2) to deep violet (16E8) in 0.1% Trypan blue. Found as Glomus cf. claroideum associated with a spore of E. infrequens grown in trap culture inoculated with soil from an agricultural grassland of Switzerland and as trap-cultured glomeromycotan spore in the USA (Supplementary Table 3).
Etymology: Latin, glacialis, referring to the glacial habitat, in which this new species was originally found.
Diagnosis: Differs from other Entrophospora species with glomoid spores in producing spores with a spore wall consisting of five permanent layers and nucleotide composition of sequences of the 45S nuc rDNA region and the rpb1 gene.
Ecology and distribution: In Oman, E. furrazolae probably lived in the field in arbuscular mycorrhizal symbiosis with Pr. cineraria in a site characterized by hyper aridity, but no molecular analysis was performed to confirm this hypothesis. In Poland, E. furrazolae was associated with Po. trivialis in an unpolluted natural site, as indicated by NGS sequencing of partial 28S PCR amplicons obtained from DNA extracted from rhizosphere soil and root samples of this plant species (Malicka et al., 2022). In single-species cultures with P. lanceolata as the host plant, E. furrazolae formed mycorrhiza with arbuscules and intra-and extraradical hyphae that stained bluish-white (20A2) to light blue (20A5) in 0.1% trypan blue. Found as C. drummondii, as well as uncultured Claroideoglomeraceae, Glomus, and Glomeromycota in undisturbed and degraded soils of Czech Republic, China, Spain, and Switzerland (Supplementary Table 3).

Discussion
Overall, the results of phylogenomic, phylogenetic, and morphological analyses presented in this study led to the following statements. First, the glomoid spore-producing species of the genus Claroideoglomus with the type species C. claroideum in the monogeneric family Claroideoglomeraceae sensu Schüßler and Walker (2010) represent a new order in the Glomeromycota, here described as Entrophosporales. Second, the entrophosporoid morph morphologically conspecific with E. infrequens represents a group of cryptic and potential dimorphic species widely distributed in the world. Third, our hypothesis was confirmed: the glomoid spore-producing AMF, preliminary named as Claroideoglomus 1, 2, and 4, originating from Argentina, Poland, Oman, and Sweden, are actually new species, described here as E. argentinensis, E. glacialis, and E. furrazolae. Finally, Redecker et al.'s (2013) conclusion that Albahypha is an unsupported taxon was confirmed (Figure 3 and Supplementary Figures 3-5).
Previous phylogenomic analyses have suggested that the order Glomerales is polyphyletic (Beaudet et al., 2018;Montoliu-Nerin et al., 2021), but the phylogenetic position of Claroideoglomeraceae was not clearly established. When data from Beaudet et al. (2018) and Montoliu-Nerin et al. (2021) were combined, we recovered the same topology as in Montoliu-Nerin et al. (2021). The topology, showing Claroideoglomeraceae sister to Diversisporales and Glomeraceae, was consistently recovered in all other analyses, even when only the Beaudet et al.'s (2018) data were used (Figure 1 and Supplementary Figures 1, 2). However, when combining all the data, we were able to extract only 78 SCOs present in at least 50% of the taxa (vs. 1260 SCOs), which could be due to the low genome completeness of the transcriptomic data presented in Beaudet et al. (2018).
In all phylogenomic analyses, we included only a few members of Mortierellomycota and Mucoromycota as outgroups. This was because the phylogenetic relationship among the three phyla was not the focus of the present work, but rather the phylogenetic placement of the former Claroideoglomeraceae with respect to the rest of Glomeromycota. This (i) resulted in better taxonomic resolution at the phylum level (Glomeromycota), (ii) provided consistent support for the phylogenetic placement of the former Claroideoglomeraceae as sister to a clade formed by Diversisporales and Glomeraceae, and, consequently, (iii) supported the need to transfer the Claroideoglomeraceae from Glomerales to a new order, Entrophosporales. The name of the order followed the synonymization of Claroideoglomeraceae with Entrophosporaceae because E. infrequens was described before C. claroideum (Ames and Schneider, 1979;Schenck and Smith, 1982). Through this, the order Glomerales remained with one family, Glomeraceae. Furthermore, the comparisons of the phylogenetic trees reconstructed from analyses with concatenated genes using Glomus (Glomerales after emendation) or Diversispora (Diversisporales) as alternative outgroups, despite some differences in the topology (Figure 3 and Supplementary Figures 3, 5), did not lead to different conclusions about the status of the new species, and Albahypha was not a valid generic taxon.
Both 45S, 45S+rpb1, and 45S+rpb1_G trees showed a group of taxa, formerly Claroideoglomus sensu Oehl et al. (2011b), characterized by short branches and a few deep-supported nodes (Figure 3 and Supplementary Figures 3, 5), a pattern that resembles an evolutionary radiation event, and a group of three sister taxa, referable to the former Albahypha sensu Oehl et al. (2011b), characterized by longer branches and different relationships depending on the dataset used. In the former Claroideoglomus clade, the species E. claroidea and E. lamellosa were always unresolved, as also shown in Delavaux et al. (2022) for the former species, a situation coherent with the evolutionary radiation hypothesis. The Albahypha clade occurred only in the 45S+rpb1_G phylogeny, where the former Claroideoglomus clade was affected by insufficient BI and ML supports. The presence of a supported Albahypha clade in the 45S+rpb1_G phylogeny probably resulted from long-branch attraction (LBA) rather than true close relationships.
Our research significantly enriched the knowledge of Glomeromycota by showing the dimorphism and cryptic nature of the species called E. infrequens, previously known only for its entrophosporoid morph with a distinctive and uniform morphology. First, both nuc rDNA and rpb1 sequences of one of the analyzed morphs, here designated as an epitype of E. infrequens from spores found in Poland, clustered with sequences of a Spanish glomoid morph preliminary named Claroideoglomus 3 (Figure 3 and Supplementary Figures 3-5). The sequences from the two isolates were obtained in different years and from two different laboratories, so we ruled out the possibility that our findings could be biased by PCR contamination. Second, sequences of a Swedish glomoid morph (Claroideoglomus 2/E. glacialis) formed a mixed clade with sequences obtained from E. infrequens identified in Switzerland (Oehl et al., 2011b). Third, the FJ461806, FJ461807, MT832169, and MT832170 sequences of E. infrequens originating from the U.S. Indiana and California states were placed in unresolved positions of the trees. Finally, the entrophosporoid morph of the E. infrequens epitype was shown to be physically connected with glomoid spores (Figures 4F,G). Consequently, the discovery of the dimorphic and cryptic nature explained the early inexplicable localization of sequences obtained from this entrophosporoid morph in different phylogenetic tree clades and unraveled the inconsistent opinion on the taxonomic status of E. infrequens sensu Ames and Schneider (1979).
According to Lücking et al. (2021), cryptic speciation may be caused by the delay of phenotypic divergence relative to phylogenetic divergence because, e.g., (i) species may have diverged too recently to have also diverged in phenotype and (ii) the forces at work were selective and prevented phenotypes from diverging. This may explain the phenotypic indistinctness of the entrophosporoid morph representing different glomoid spore-producing Entrophospora species (Figures 3, 5-8 and Supplementary Figures 3-5). On the other hand, the massive construction of entrophosporoid spores, consisting of two thick walls with four permanent components, may have acted as a physical barrier that prevented the production of other entrophosporoid phenotypes due to the influence of the external environment.
Dimorphism is not the persistent behavior of dimorphic AMF species which, for undefined reasons, produced either both morphs (e.g., acaulosporoid and glomoid by Am. leptoticha) or only one of them (glomoid by another strain of Am. leptoticha) even over many propagation cycles (Bills and Morton, 2015). More (1998) concluded that the dimorphism in even extensively studied fungal groups was influenced or governed by a wide range of unspecific metabolic and environmental factors.
Thus, the cryptic nature and the dimorphic abilities of the entrophosporoid morph implied treating the glomoid spore-producing species of the genus Claroideoglomus sensu Schüßler and Walker (2010) as members of the newly emended genus Entrophospora, in accordance with the new combinations introduced in the section "Taxonomy".
Like the phylogenetic analyses, morphological comparisons also showed that the relationships between the three new glomoid spore-producing Entrophospora species and the glomoid morph of the epitype of E. infrequens versus all previously sequenced Entrophospora species are bland or very weak. None of the known Entrophospora species produces spores (i) with a laminate spore wall layer, which is uneven in thickness in its different regions, as in E. argentinensis (Figures 6B-E,G), (ii) so strongly resembling spores of species of Dominikia and related genera in their frequent formation in loose clusters, appearance, size, pigmentation, spore wall structure, as well as in the phenotypic and histochemical properties of components of this wall (Błaszkowski et al., 2021b), as the glomoid spores of E. infrequens (Figures 4F,G, 5A-F), (iii) with a spore wall consisting of five permanent layers, as in E. glacialis (Figures 8B-H), and (iv) with a spore wall containing three permanent layers, of which two stain in Melzer's reagent, except for E. furrazolae (Figures 9B-F,H).
The poorly and ambiguously supported relationship of E. glacialis to E. claroidea and E. candida (Figure 3 and Supplementary Figures 3, 5) somewhat confirms their morphological similarity. Spores of E. glacialis and E. claroidea (i) are surrounded by a halo produced by colorless or slightly colored spore wall layers surrounding the main structural, much darker colored, laminate spore wall layer and (ii) have a spore wall with an innermost layer that is flexible to semi-flexible, colorless, thin, and usually separates from the lower surface of the laminate layer in crushed spores (Figures 8A-H; Błaszkowski et al., 2021b). However, in E. claroidea, the halo is produced by two impermanent layers, frequently highly or completely sloughed off in mature specimens, of which the outermost layer 1, forming the spore surface, stains in Melzer's reagent. The innermost E. claroidea spore wall component is layer 4. Instead, in E. glacialis the halo is formed by three permanent layers (layers 1-3), remaining intact even in old spores, and none of the fivelayered spore wall layers reacts in Melzer's (Figures 8A-H). In addition, E. claroidea spores are clearly lighter in color, not reaching the shade of brown of E. glacialis spores at maturity.
Spores of E. candida also have a halo when all spore wall layers are present (Furrazola et al., 2010). However, according to the original description, spores of this species compared to those of E. glacialis are much lighter (white, rarely pale yellow) and 1.2-to 2.2-fold larger when globose, have only a two-layered and ∼ 1.4-fold thinner spore wall, as well as may have an up to 1.5-fold wider subtending hypha at the spore base.
Although unsupported, the appearance of a sister relationship of E. argentinensis to E. etunicata in the 45S tree (Supplementary Figure 3) and the neighborhood of these species in the 45S+rpb1, rpb1, and 45S+rpb1_G trees (Figure 3 and Supplementary Figures 4, 5) call for a comparison of their morphology. As mentioned above, none of the glomoid spore-producing Entrophospora species, including E. etunicata, has a laminate spore wall layer that is uneven in thickness (Błaszkowski, 2012;Błaszkowski et al., 2015a). In addition, the spore wall of E. etunicata consists of two layers only, lacking the semi-flexible spore wall layer 3 of E. argentinensis (Figures 6B-D,F-H), and the spore wall layer 1 of E. etunicata stains in Melzer's reagent (Becker and Gerdemann, 1977;Stürmer and Morton, 1997;Błaszkowski, 2012). In E. argentinensis, none of the spore wall layers reacts in this reagent ( Figure 6B). Moreover, the pore of the spore subtending hypha of E. etunicata is closed by a septum continuous with the innermost laminae (sublayers) of the laminate spore wall layer 2, and that of E. argentinensis is closed by the innermost laminae of the laminate spore wall layer 2 and a septum continuous with the innermost spore wall layer 3 (Figures 6G,H). Finally, globose E. etunicata spores may be 1.4-fold larger.
Morphologically, E. furrazolae may be confused with E. drummondii, a sister species indicated in the 45S and 45S+rpb1_G trees (Figure 3 and Supplementary Figure 3). In both species, the flexible, thin innermost spore wall layer, forming a septum in the subtending hypha, stains in Melzer's reagent (Figures 9D-H; Błaszkowski et al., 2006;Błaszkowski, 2012), a property unknown in other glomoid spore-producing species. However, in E. drummondii, this staining reaction occurs only in spore wall layer 3, and in E. furrazolae it appears in spore wall layer 4, absent in the spore wall of the former species, and spore wall layer 1 (Figures 9D-H). In addition, the spore wall of E. drummondii is 1.4-to 1.9-fold thinner, and the mature spores of this species are always slightly darker colored.
We omitted in our analyses the genus Viscospora included in the Entrophosporaceae defined by Oehl et al. (2011b) because (i) its sole species, V. viscosa, originally described as G. viscosum (Walker et al., 1995), is provided with only one 18S nuc rDNA sequence and (ii) Redecker et al. (2013) found that V. viscosa belongs to the genus Septoglomus. However, morphologically V. viscosa does not match any Septoglomus species. Therefore, more robust molecular proofs are needed to clarify the phylogenetic position of V. viscosa. In addition, our analyses included the MT733209 and MT733210 rpb1 sequences erroneously ascribed to C. cf. drummondii by Błaszkowski et al. (2021a) instead of to the new E. furrazolae.
We recommend for the upcoming works involving species in the Entrophosporales that (i) E. infrequens-like culture collections should be checked carefully to detect the possible presence of dimorphism, (ii) the morphology of E. infrequens-like and/or glomoid-like spores should be carefully described to highlight any possible difference not yet detected, and (iii) descriptions of new species based on E. infrequens-like cultures should be based on strong molecular data.

Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI MH590060, MH590061, MT722021-MT722024, ON950363-ON950390, MT722034-MT722038, MT733201-MT733210, and ON938320-ON938327.

Funding
This work was supported by the Polish National Centre of Science, grant no. 2020/37/N/NZ9/00509, as well as by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) that provided research grants to BT Goto (proc. 311945/2019-8). We also thank Universidade Federal do Rio Grande do Norte for covering the costs of B.T. Goto's stay as a collaborative researcher in the West Pomeranian University of Technology in Szczecin in the period from December 2019 to January 2020. The research was funded, in part, by the Institute of Botany at the Jagiellonian University, project no. N18/DBS/000002.