Fungi inhabiting attine ant colonies: reassessment of the genus Escovopsis and description of Luteomyces and Sympodiorosea gens. nov.

Escovopsis is a diverse group of fungi, which are considered specialized parasites of the fungal cultivars of fungus-growing ants. The lack of a suitable taxonomic framework and phylogenetic inconsistencies have long hampered Escovopsis research. The aim of this study is to reassess the genus Escovopsis using a taxonomic approach and a comprehensive multilocus phylogenetic analysis, in order to set the basis of the genus systematics and the stage for future Escovopsis research. Our results support the separation of Escovopsis into three distinct genera. In light of this, we redefine Escovopsis as a monophyletic clade whose main feature is to form terminal vesicles on conidiophores. Consequently, E. kreiselii and E. trichodermoides were recombined into two new genera, Sympodiorosea and Luteomyces, as S. kreiselii and L. trichodermoides, respectively. This study expands our understanding of the systematics of Escovopsis and related genera, thereby facilitating future research on the evolutionary history, taxonomic diversity, and ecological roles of these inhabitants of the attine ant colonies. Supplementary Information The online version contains supplementary material available at 10.1186/s43008-021-00078-8.


Introduction
The kingdom Fungi comprises organisms with wide morphological and genetic diversity (Mueller and Schmit 2007;Hawksworth and Lücking 2017). Through systematic approaches, taxonomists have developed mechanisms to categorize organisms based on their phenotypic and genetic characters (Komarek and Beutel 2006;Pavlinov 2018). Notwithstanding, taxonomic and phylogenetic incongruities like non-monophyly preclude the study of many fungal groups, as is the case for the genus Escovopsis (Ascomycota: Hypocreales, Hypocreaceae).
Escovopsis is a diverse group of fungi, members of which are presumed to have evolved parasitizing the mutualistic fungus of fungus-growing ants (Formicidae: Myrmicinae: Attini: Attina, the "attines") (Yek et al. 2012). Besides, Escovopsis has only been found associated with fungus-growing ant colonies, suggesting that the genus has evolved in relation to these ants' system, potentially for millions of years. Despite its biological importance in relation to a canonical system for the study of coevolution and symbiosis, the paucity of taxonomic studies and unresolved phylogenetic inconsistencies have prevented a comprehensive understanding of the systematics, ecology, and evolution of these hypocrealean fungi.
More than a century has passed since Möller (1893) observed a group of "fungi with strong conidial shapes" (i.e. more prevalent fungi) in attine gardens, which Open Access IMA Fungus *Correspondence: quimivimo@gmail.com; andre.rodrigues@unesp.br 1 Department of General and Applied Biology, São Paulo State University (UNESP), Avenida 24-A, n. 1515, Bela Vista, Rio Claro, SP 13.506-900, Brazil Full list of author information is available at the end of the article 80 years later would be named Phialocladus (Kreisel 1972). Almost two decades later, Phialocladus was considered an invalid name because of the lack of the type specimen for its type species, P. zsoltii, and consequently, it was renamed as Escovopsis with E. weberi designated as the type species of the latter genus (Muchovej and Della Lucia 1990). Although the macroscopic characters of E. weberi were not fully described by the authors, the description of this species established the foundation for Escovopsis taxonomy. In 1995, Seifert et al. (1995) described Escovopsis aspergilloides in a detailed taxonomic study. However, after this study, the taxonomy of Escovopsis was set aside for 18 years.
Unlike systematics studies, the relationship between Escovopsis and the attine ants' mutualistic fungi has been the topic of numerous studies (Currie et al. 2003;Gerardo et al. 2006a, b;Taerum et al. 2007;Folgarait et al. 2011;Elizondo Wallace et al. 2014;Marfetán et al. 2015;Birnbaum and Gerardo 2016;de Man et al. 2016;Heine et al. 2018). A great diversity was attributed to the genus Escovopsis through these studies, yet the morphology and phylogenetic placement of most of the strains named as Escovopsis were not properly assessed. The first ideas about the phylogenetic position of Escovopsis were gradually emerging (Currie et al. 2003;Gerardo et al. 2006b;Augustin et al. 2013;Masiulionis et al. 2015;Meirelles et al. 2015a, b). Initially, some authors suggested that Escovopsis belonged to the Hypocreales, although, no phylogenetic evidence was provided to support that hypothesis at the time (Currie et al. 1999a, b). The first phylogenetic analysis of Escovopsis confirmed the genus to be placed in the Hypocreales, as a sister clade of the Hypocreaceae (Currie et al. 2003). However, a more extensive phylogenetic analysis of Escovopsis strains associated with fungus gardens of Apterostigma ants indicated that the genus belonged to the Hypocreaceae (Gerardo et al. 2006b). Augustin et al. (2013) were the first authors to combine morphological and phylogenetic approaches to study Escovopsis. Based on internal transcribed spacer (ITS) and large subunit ribosomal RNA (LSU) sequences, the authors described Escovopsis lentecrescens, E. microspora, and E. moelleri, which formed a monophyletic clade with E. weberi and E. aspergilloides. The most remarkable character of these newly named species was the presence of conidiophores with vesicles, as previously described by Muchovej and Della Lucia (1990) and Seifert et al. (1995). Nonetheless, while phylogenetic analyses of Augustin et al. (2013) based on ITS and LSU sequences suggested that Escovopsis formed a monophyletic clade, their analyses based on translation elongation factor 1-alpha (tef1) (including all strains treated as Escovopsis at that time except for the ex-type culture of E. weberi) suggested that the genus may not be monophyletic.
For almost 30 years, the genus Escovopsis was morphologically defined by the presence of conidiophores with vesicles that support the phialides (i.e., enteroblastic conidiogenous cells), from which conidia are produced. However, that changed with the introduction of Escovopsis trichodermoides (Masiulionis et al. 2015) and E. kreiselii (Meirelles et al. 2015a). These species have conidiophores without vesicles and with poorly differentiated conidiogenous cells (i.e., holoblastic determinate conidiogenous cells with synchronous arrangement; E. trichodermoides) and sympodial conidiogenous cells (i.e., holoblastic proliferous conidiogenous cells; E. kreiselii) instead of phialides. Therefore, Meirelles et al. (2015a) amended the morphological description of Escovopsis to insert the morphological features of E. kreiselii. However, because Masiulionis et al. (2015) and Meirelles et al. (2015a) were published at nearly the same time, the new definition did not include E. trichodermoides.
The insertion of E. trichodermoides and E. kreiselii within Escovopsis not only meant changes to the morphological circumscription of the genus but also intensified the phylogenetic uncertainties showed by Augustin et al. (2013). In the phylogenies produced by Masiulionis et al. (2015) and Meirelles et al. (2015a, b), it was clear that vesiculate Escovopsis were more closely related to Escovopsioides (Augustin et al. 2013) than to E. trichodermoides and E. kreiselii. Nonetheless, both Masiulionis et al. (2015) and Meirelles et al. (2015a, b) preferred to maintain E. trichodermoides, E. kreiselii and the vesiculate Escovopsis as placing in the same genus.
Recently, Montoya et al. (2019) used the ITS, LSU, and tef1 markers in a multilocus phylogenetic approach to describe Escovopsis clavata and E. multiformis. The authors noticed that disagreements in the Escovopsis taxonomy occurred among vesiculate Escovopsis, E. trichodermoides and E. kreiselii. Therefore, they highlighted the need to utilize new molecular markers to resolve the phylogeny of the genus. However, subsequent description of five new Escovopsis species (Marfetán et al. 2018) further complicated Escovopsis taxonomy because interpretation of the phylogenetic analyses made by Marfetán et al. (2018) had some limitations: (1) it was based on the LSU and tef1 genes separately; (2) the tef1 sequences obtained in the study do not align with those sequences in previously published studies; (3) some LSU sequences do not have similarity with Escovopsis; and (4) some of the new species (Escovopsis atlas, E. catenulata, and E. pseudoweberi) fall in the same clade, but strains of E. atlas fall in different (non-monophyletic) clades.
Given this complicated and piecemeal research history, the aim of this study is to reassess the genus Escovopsis by using a comprehensive multilocus phylogeny based on five molecular markers. Our results fill an important gap in mycology and will help future researchers to access the taxonomy, diversity and the evolutionary history of Escovopsis and related genera that inhabit the colonies of fungus-growing ants.

Strains
A total of 102 strains of Escovopsis, i.e., vesiculate (n = 64) and non-vesiculate (n = 38) species, were included in this study (Additional file 1: Table S1). Of these, 30 strains were obtained from previous studies (Augustin et al. 2013;Masiulionis et al. 2015;Meirelles et al. 2015b;Montoya et al. 2019), and the remaining (n = 72) were isolated from three regions in Brazil (Novo Airão and Camp 41, state of Amazonas; Botucatu, state of São Paulo-Additional file 1: Table S1). The process of isolation, purification, and preservation of the strains followed methods outlined in Montoya et al. (2019). Briefly, from each attine colony, 21 garden fragments (0.5-1 mm 3 ) were inoculated on potato dextrose agar (PDA, Neogen Culture Media, Neogen, Bury) plates (seven fragments per plate) supplemented with chloramphenicol (150 μg mL −1 , Sigma-Aldrich, St. Louis). The plates were incubated at 25 °C in darkness and monitored daily for 7 d. When Escovopsis mycelia grown out, they were transferred to new PDA plates without chloramphenicol. Axenic cultures were prepared by single conidial isolation and stored in sterile distilled water kept at 8-10 °C (Castellani 1963), and in 10% aqueous solution of glycerol at − 80 °C. Both the strains isolated in this study and those obtained from other studies are deposited at the Laboratory of Fungal Ecology and Systematics (LESF-Department of General and Applied Biology, São Paulo State University (UNESP), Rio Claro, SP, Brazil) and at the UNESP-Microbial Resources Center (CRM-UNESP), Rio Claro, Brazil, under the same conditions.

DNA extraction, PCR and sequencing
The genomic DNA of the strains was extracted using a modified CTAB method (Möller et al. 1992). Briefly, fungal aerial mycelia, grown for 7 d at 25 °C on PDA, were crushed with the aid of glass microspheres (Sigma) in lysis solution and incubated at 65 °C for 30 min. The organic phase was separated using a solution of chloroform-isoamyl alcohol (24:1). Then, the material was centrifuged (10,000g for 10 min), and the supernatant with the genomic DNA was collected. This extract was precipitated with 3 M sodium acetate and isopropanol and purified with two successive washes of 70% ethanol. The DNA was suspended in 30 μL of Tris-EDTA solution and stored at − 20 °C. Five molecular markers were amplified for all newly isolated Escovopsis strains: the internal transcribed spacer (ITS), the large subunit ribosomal RNA (LSU), the translation elongation factor 1-alpha (tef1), and the RNA polymerase II protein-coding genes (rpb1 and rpb2, Additional file 1: Table S2). For strains from previous publications, we utilized previously published ITS, LSU and tef1 sequences, when available, and generated missing sequences for other molecular markers (Additional file 1: Table S1). Sequences of rpb1 and rpb2 for 23 strains in the genus Escovopsioides were also generated in this study to complete our dataset (Additional file 1: Table S1).
Final amplicons were purified with the Wizard SV Gel and PCR Clean-up System (Promega, Madison) following the manufacturer's protocol. Sequences (forward and reverse) were generated on an ABI3500 (ThermoFisher Scientific, Waltham), and the consensus sequences were assembled in BioEdit v. 7.1.3 (Hall 1999) or Geneious (Kearse et al. 2012). All sequences are deposited in Gen-Bank (Additional file 1: Table S1 for accession numbers).

Phylogenetic analyses
In order to have a complete perspective of the Escovopsis' phylogenetic incongruences and their possible solutions, we performed phylogenetic analysis to: (1) know the phylogenetic placement of all strains currently treated as Escovopsis, and (2) provide a phylogenetic framework that establishes the foundations of the genus' systematics.

Phylogenetic placement of fungi treated as Escovopsis
We reconstructed a phylogenetic tree combining all tef1 sequences of fungi treated as Escovopsis in the literature (Currie et al. 2003;Gerardo et al. 2004Gerardo et al. , 2006bTaerum et al. 2007Taerum et al. , 2010Augustin et al. 2013;Masiulionis et al. 2015;Meirelles et al. 2015a, b;Montoya et al. 2019) with our tef1 data set and the tef1 sequences from other genera of the Hypocreales (Additional file 1: Table S3). It was not possible to perform a multilocus analysis because few strains on the literature were sequenced for more than one molecular marker. Nonetheless, the tef1 gene was the one used in most of the studies published already. The final data set contained a total of 440 tef1 sequences (754 bp), that included vesiculate-Escovopsis (n = 274 strains), non-vesiculate Escovopsis (n = 105 strains), 60 strains from five Hypocreaceae genera, i.e., Escovopsioides, Hypomyces (along with species under its anamorphic genus Cladobotryum) (Põldmaa 2011), Protocrea, Sphaerostilbella, and Trichoderma, and Lecanicillium antillanum CBS 350.85 as the outgroup (Additional file 1: Table S3). The tef1 sequences of species described in Marfetán et al. (2018) were not included in this analysis because they do not align with the Escovopsis sequences from the previous studies.
The data set was first aligned in MAFFT v.7 (Katoh and Standley 2013), and phylogenetic trees were reconstructed using maximum likelihood (ML) and Bayesian inferences (BI) in RAxML (Stamatakis 2014) and MrBayes v.3.2.2 (Ronquist et al. 2012), respectively. The nucleotide substitution model was GTR for ML and K80 + G for BI and was calculated in jModelTest 2 (Darriba et al. 2012), using the Akaike Information Criterion (AIC) with 95% confidence intervals. For ML analysis, 1000 independent trees and 1000 bootstrap replicates were performed, while for BI two separate runs (each consisting of three hot chains and one cold chain) were carried out. In the last case, five million generations of the Markov Chain Monte Carlo (MCMC) were enough to reach convergence (standard deviation (SD) of split frequencies fell below 0.01). To generate final BI tree, the first 25% of trees were discarded as burn-in. The final tree was edited in FigTree v.1.4 (http:// tree. bio. ed. ac. uk/ softw are/ figtr ee/) and Adobe Illustrator CC v.17.1.

Phylogenetic framework for Escovopsis' systematics
Phylogenetic analyses were performed at order and family levels. The objective of the analyses at the order level was to investigate whether all clades formally described as Escovopsis belong to the Hypocreaceae close to the Cordycipitaceae as previously observed by Augustin et al. (2013). The objective of the analyses at the family level was to investigate whether the vesiculate species of Escovopsis form a monophyletic clade, separating them from the non-vesiculate species, as previously observed by Montoya et al. (2019). Besides, we wanted to know if the monophyly of all those clades remains constant considering the Genealogical Concordance Phylogenetic Species Recognition (GCPSR) concept (Taylor et al. 2000), although it was applied to the consideration on the generic differentiation.
For all analyses, datasets were first aligned separately for each gene in MAFFT v.7 (Katoh and Standley 2013). The nucleotide substitution model for each alignment was calculated in jModelTest 2 (Darriba et al. 2012), using the Akaike Information Criterion (AIC) with 95% confidence intervals. Then, the datasets were concatenated in Winclada v.1.00.08 (Nixon 2002). All phylogenetic trees were constructed using ML in RAxML v.8 (Stamatakis 2014) and BI in MrBayes v.3.2.2 (Ronquist et al. 2012). For ML, we estimated 1000 independent trees and performed 1000 bootstrap replicates using the GTR + I + G model for each partition independently. For BI analyses, we carried out two separate runs (each consisting of three hot chains and one cold chain) using the GTR + I + G model for each partition independently; for all analyses, two million generations of the Markov Chain Monte Carlo (MCMC) were enough to reach convergence [standard deviation (SD) of split frequencies fell below 0.01]. To generate BI trees, the first 25% of trees were discarded as burn-in. Trees were edited in FigTree v.1.4 (http:// tree. bio. ed. ac. uk/ softw are/ figtr ee/) and Adobe Illustrator CC v.17.1.

Morphology
We examined the microscopic structures of nine ex-type cultures of Escovopsis species (E. aspergilloides, E. clavata, E. kreiselii, E. lentecrescens, E. microspora, E. moelleri, E. multiformis, E. trichodermoides, and E. weberi), representing the known diversity of the genus, to determine if those morphological features support the results observed in phylogenetic analyses.
To assess and compare the microscopic structures (i.e., conidiophores, conidiophore branches, vesicles, conidiogenous cells, conidia, and chlamydospores) and their features (i.e., shape and pattern), we carried out slide culture preparations on PDA. To do so, we placed three cylinders of PDA (ca. 5 mm in diameter × 5 mm in height) on a sterilised microscopic slide, and we then inoculated each fragment with conidia of the fungus. Each inoculated fragment was covered with a coverslip and incubated at 25 °C for 4-7 d in the dark. After that, the fragments of PDA were removed and the coverslips with fungal mycelia were placed on slides with a drop of lactophenol. The slides were examined under a light microscope (DM750, Leica, Wetzlar), and the microscopic fungal structures were photographed using the software LAS EZ v.4.0 (Leica, Wetzlar).

Phylogenetic placement of fungi previously known as Escovopsis
The tef1 phylogenetic tree showed nine well supported monophyletic clades within the Hypocreaceae ( Fig. 1 and Additional file 3: Fig. S2). The four previously recognized genera of the Hypocreaceae, i.e., Escovopsioides, Hypomyces (along with species under its anamorphic genus Cladobotryum), Protocrea, and Trichoderma, preserved their monophyly. On the other hand, the genus Escovopsis (clades A-E) was found polyphyletic and scattered throughout the family ( Fig. 1 and Additional file 3: Fig. S2).
While clades A-E are treated as Escovopsis, clades A (vesiculate Escovopsis; the type species of Escovopsis, i.e., E. weberi belongs to this clade), B (E. trichodermoides), and E (E. kreiselii) are the only clades containing formally described Escovopsis species, so far. Although clades A and B are sister clades, they are far away separated from the clade E (paraphyletic clade; Fig. 1
Analyses on the family level based on GCPSR (Taylor et al. 2000) revealed that Hypomyces, Escovopsioides, E. kreiselii, E. trichodermoides, and vesiculate-Escovopsis each form separate, monophyletic clades (Fig. 3). The phylogenetic placement of the five clades varies depending on the molecular marker used for the analysis. The analyses made separately with ITS, LSU, and rpb1 (Fig. 3A, B, D, respectively), as well as the concatenated analysis based on the five markers (Fig. 3F) indicate vesiculate-Escovopsis, and E. kreiselii as sister clades. The analyses made separately with tef1 and rpb2 (Fig. 3C, E, respectively), and the concatenated analysis based on four markers (LSU, tef1, rpb1, and rpb2-Fig. 3G) indicate vesiculate-Escovopsis and E. trichodermoides as forming sister clades. In addition, the analyses performed with ITS and the five concatenated-markers showed Escovopsioides and E. trichodermoides forming a monophyletic clade (PP = 1, MLB = 98- Fig. 3A and PP = 1, MLB < 70- Fig. 3F, respectively). The analysis based on Page 6 of 18 Montoya et al. IMA Fungus (2021) 12:23 rpb1, however, indicated Escovopsioides forming a monophyletic clade with Hypomyces (Fig. 3D), while analysis based on tef1 placed Hypomyces between Escovopsioides and E. kreiselii (Fig. 3C). Finally, the analyses carried out with LSU, tef1, and rpb2, separately (Fig. 3B, C, E, respectively), and the concatenated analysis (LSU, tef1, rpb1 and rpb2- Fig. 3G)  LESF 979) and (LESF 969, LESF 997, LESF 1003 and LESF 996) formed well-supported monophyletic clades, that were clearly separated from one another (Fig. 3G). Nonetheless, the ex-type cultures of E. weberi and E. microspora grouped together with 45 other isolates in the same well-supported clade (Fig. 3G). Regarding the five species described by Marfetán et al. (2018), our analysis of the available LSU data indicates that they form two different clades closely related to E. aspergilloides and E. lentecrescens (Additional file 2: Fig.  S1). Specifically, the four strains of E. primorosea formed a well-supported monophyletic clade (Additional file 2: Fig. S1), however E. atlas, E. catenulata, E. pseudoweberi, and five strains named as E. weberi by the authors formed a single monophyletic clade (Additional file 2: Fig. S1). This result apparently does not support the new species hypothesis. Future research will be required to clarify the existence of these species.
As the first group, vesiculate Escovopsis spp. present conidiophores with one apical vesicle (mono-vesiculate) or with two to more vesicles (poly-vesiculate) (Fig. 4). Mono-vesiculate conidiophores (Fig. 4A, B) emerge from aerial mycelia in an alternating and opposite pattern. Poly-vesiculate conidiophores (Fig. 4C-H) also emerge from the aerial mycelia and can present short branches composed of one or two cells or long branches with multiple cells. Some poly-vesiculate conidiophores have a swollen cell at the apex from where branches emerge (Fig. 4F). Conidiophores that have swollen cells form their branches only from swollen cells. Vesicles, mostly composed of a single cell (non-septate vesicles- Fig. 4I-T), and rarely of two cells (septate vesicles- Fig. 4U, V). Vesicles can emerge from the apices and axes of the conidiophore and its branches (usually only one vesicle in the short branches and two or more in long branches) (Fig. 4G, H). Vesicles are connected to the hypha from where they emerge by a basal septum or by a peduncle made up of one or two cells. Vesicles can exhibit different shapes (globose, subglobose, capitate, obovoid, prolate, spatulate, clavate, cymbiform, lanceolate, subulate, cylindric, filiform, clavate-septate, cylindric-septate) (Fig. 4I-V) and sizes, depending on the species. Phialides (i.e., enteroblastic conidiogenous cells; Fig. 4W-Z) emerge mainly from the vesicles and less frequently from the aerial mycelia. These structures are lageniform and ampulliform, forming conidia. Main differences of phialides among the different species are mostly related to the sizes of their base, the widened part and the neck (Fig. 4W-Z). Conidia are single-celled, produced in chains from the phialides (enteroblastic conidiogenesis), and can vary in shape (globose, subglobose, ellipsoidal, oblong, and oblong-ornamented) (Fig. 4a-e). Conidia can be smooth or with ornamentations on cell walls (Fig. 4e). Chlamydospores are rarely observed.
The two non-vesiculate groups form conidiophores remarkably different from those of the vesiculate group. Conidiophores of E. trichodermoides lack vesicles and are mostly pyramidal (Trichoderma-like), with one to six short levels of branches arising at more or less right angles from the conidiophore axis (Fig. 5A, B). In contrast to vesiculate Escovopsis, this species has poorly differentiated conidiogenous cell (i.e., holoblastic determinate conidiogenous cells with synchronous arrangement; Fig. 5C-F) with ampulliform shapes producing either solitary (Fig. 5C) or up to three conidia ( Fig. 5D-F). The conidia are subglobose to obovate, yellow-brown, mostly ornamented (verrucose; Fig. 5G) and rarely smooth (Fig. 5H). Unlike the vesiculate Escovopsis spp., E. trichodermoides regularly forms chlamydospores (Fig. 5I).
Conidiophores of E. kreiselii also lack vesicles and are formed on the aerial mycelium in an alternated or opposite pattern (Fig. 6A, B), and are more branched (with Fig. 2 Phylogenetic tree indicating the placement of Escovopsis (brown branches), Sympodiorosea (pink branches) and Luteomyces (yellow branches) within the Hypocreales. The tree shown was inferred using Bayesian Inference (BI). Highlighted clades in different colours represent six different families (Bionectriaceae, Clavicipitaceae, Cordycipitaceae, Hypocreaceae, Nectriaceae, and Ophiocordycipitaceae) in the Hypocreales. The analysis was based on concatenated sequences of LSU, tef1, rpb1 and rpb2 (See Additional file 1: Table S5 for all strains and their associated metadata used to infer this phylogenetic tree). Numbers on branches indicate BI posterior probabilities (PP) and Maximum Likelihood bootstrap support values (MLB), respectively. Hyphens (--) indicate MLB < 70%. We used eight species from the family Stachybotryaceae (dark green box) to root the tree. ET indicates ex-type cultures (See figure on next page.) Montoya et al. IMA Fungus (2021) Fig. 6C-H), and they are formed on the apex and on both the main and branch axes of the conidiophore (Fig. 6B). Conidia of E. kreiselii are solitary, globose to subglobose, smooth but with thick-walled, light brown to dark brown, and usually with a denticle (Fig. 6I) or with a lesion (when the denticle remains on the conidiogenous cell; Fig. 6J). Like E. trichodermoides, E. kreiselii also forms chlamydospores regularly (Fig. 6K).

Taxonomy
Phylogenetic analyses in this study unambiguously demonstrate that species previously treated as Escovopsis form three distinct monophyletic clades (Fig. 3). These groups also differ significantly from one another in morphological characters. The first clade (Fig. 3G) Table S1 for all strains and their associated metadata used to infer these phylogenetic trees Page 10 of 18 Montoya et al. IMA Fungus (2021)   , which are all vesiculate species. We recognize these species as true Escovopsis because they form a monophyletic clade including the type species of the genus and preserve the main character (presence of vesicles on conidiophores) that gave origin to the generic name (Muchovej and Della Lucia 1990). Therefore, we redefine and restrict Escovopsis to include only vesiculate species (Fig. 3G) and the description of the genus is emended based on the criteria adopted in this study.
Notes: Escovopsis is phylogenetically placed within the Hypocreaceae as a sister clade of Luteomyces (Fig. 3G). Escovopsis exhibits faster growth and different colony colour (brown) than Luteomyces (yellow). Unlike Luteomyces, which presents poorly diferenciated conidiogenous cells (i.e., holoblastic determinate conidiogenous cells with synchronous arrangement), Escovopsis forms phialides (i.e., enteroblastic conidiogenous cells). The main feature of this genus is the presence of conidiophores with terminal vesicles that differentiate it from its sister clade and from all other known genera in the Hypocreaceae.
Etymology: "Luteomyces" based on the colour exhibited by the colonies of the type species.
Notes: Luteomyces is phylogenetically placed in the Hypocreaceae as a sister clade of Escovopsis (Fig. 3G). Nonetheless, Luteomyces grows slower and has different colony colour (mainly yellow) than Escovopsis (mainly brown), and forms conidiophores without vesicles and large number of chlamydospores (rarely observed in Escovopsis).  ( Fig. 6).
Etymology: "Sympodio" refers to the sympodial conidiogenous cells, and "rosea" to the colony colour of the type species.
Diagnosis: Similar to Escovopsis and Luteomyces in the way it begins to grow, i.e., dense germination and forming stolon-like mycelia. However, Sympodiorosea differs from these genera and other known genera in the Hypocreaceae by its holoblastic sympodial proliferous conidiogenous cells. Description: Monophyletic genus in the Hypocreaceae. Colonies form inconspicuous to floccose, white, palebeige, pink, brown aerial mycelia. Conidiophores formed on aerial mycelia (Fig. 6A), alternate or opposite, usually at right angles, with irregular branching conformation (Fig. 6A, B). Conidiogenous cells holoblastic, sympodial, proliferous ( Fig. 6C-H), in pairs or in verticils on the apices of conidiophores and their branches, and solitary, alternate or opposite, on both the axes of the conidiophore and their branches (Fig. 6B). Conidia formed solitary, globose to subglobose, smooth or rough (thickwalled), light-brown to dark-brown, with denticles or lesions like holes (Fig. 6I, J). Chlamydospores commonly formed (Fig. 6K).

Discussion
Here, we provide the basis for the systematics of Escovopsis and related genera using a set of morphological characters, and comprehensive multilocus phylogenetic analyses. Our results supported the separation of species previously treated as Escovopsis into three distinct genera. Accordingly, we redefine and restrict Escovopsis to vesiculate species, and we describe Sympodiorosea to accommodate E. kreiselii and Luteomyces to accomodate E. trichodermoides. This study provides a long-awaited revision of Escovopsis systematics and related genera, thus helping future researchers to assess the diversity and evolutionary history of these fungus-growing ant associates. Genera of the Hypocreaceae have morphological features that differentiate them from one another (Jaklitsch et al. 2008;Jaklitsch 2009;Põldmaa 2011). Due to the morphological plasticity of fungi (Slepecky and Starmer 2009;Wrzosek et al. 2017), variations in the shades of colours expressed by these organisms are highly expected. Nonetheless, the prevalent brown colour of Escovopsis colonies is a unique feature of this genus within the family. Curiously, other genera in the same family also exhibit unique colours, as is the case for the genus Trichoderma, which is characterized by its mostly green colonies (Jaklitsch 2009), and Escovopsioides, which is characterized by its white colour (Augustin et al. 2013). While Sympodiorosea and Luteomyces are currently monotypic genera, phylogenetic analysis suggests there are more Sympodiorosea species waiting to be described (Figs. 2,3G). Colours exhibited by these genera (pink and yellow, respectively) are also unique within the Hypocreaceae. Interestingly, the separation of these clades by the colour patterns was previously observed by other authors (Gerardo et al. 2006b;Meirelles et al. 2015b), but the lack of a deep morphological analysis prevented reaching the conclusion that they were different genera.
Microscopic features also differentiate Escovopsis from other genera in the Hypocreaceae. Conidiophores with terminal-vesicles producing phialides, present in Escovopsis, are a unique feature of the genus within this family. Escovopsioides also presents vesicles, however they are formed intercalary on aerial mycelia and solitary on the apex of the conidiophores. Furthermore, the vesicles of this genus are smaller and have fewer phialides than those of Escovopsis (Augustin et al. 2013). On the other hand, Sympodiorosea and Luteomyces are the only genera within the Hypocreaceae that present sympodial and poorly differentiated holoblastic conidiogenous cells, respectively. Interestingly, only a distant group of entomopathogenic fungi, i.e., Beauveria (Ascomycota: Hypocreales, Cordycipitaceae), has sympodial conidiogenesis (Rehner et al. 2011) like Sympodiorosea, and there are no other groups of fungi within the Hypocreaceae that form poorly differentiated holoblastic conidiogenous cells like Luteomyces. Future studies will hopefully shed light on the evolutionary pressures that led Escovopsis, Sympodiorosea, and Luteomyces to form these unique microscopic characters.
The consideration of all fungi producing brown conidia in the attine ant's colonies as Escovopsis made of this genus polyphyletic and paraphyletic (Fig. 1, Additional file 3: Fig. S2). Lack of a comprehensive phylogenetic analysis has precluded resolving the phylogenetic uncertainties of Escovopsis (Montoya et al. 2019). In light of our results, we considered two hypotheses to solve the phylogenetic disagreements of this group of fungi: first, Escovopsis, Luteomyces, and Sympodiorosea belong to the same genus. In this case, Escovopsioides (sister clade of Luteomyces, Fig. 3A, F) and Hypomyces (closely related with Sympodiorosea and Escovopsioides, Fig. 3C, D) would have to belong to the same genus to enforce monophyly (Baum and Smith, 2013). However, both Escovopsioides and Hypomyces are well supported, separate monophyletic clades (Figs. 1, 2, 3) and present unique morphological characters that differentiate them from other genera in the Hypocreaceae. Escovopsioides, for instance, is the only genus within the Hypocreaceae that forms phialides on tiny intercalary vesicles on the aerial mycelium. In contrast, many species of Hypomyces forms septate conidia and some species form sexual structures (despite being a physiological or genetic character not easily observed within the group), which are not observed in Escovopsis, Luteomyces, Sympodiorosea or Escovopsioides. Second, Escovopsis, Luteomyces, and Sympodiorosea represent taxa within different genera. Moreover, while different, Escovopsioides and Hypomyces (some species) are the only genera within the Hypocreaceae that form plural types of conidia (i.e., one type of conidia from phialides and another direct from aerial mycelia without conidiogenous cells (Põldmaa et al. 1999;Augustin et al. 2013). In this case, considering that: (1) the same genes, in different genera, follow different evolutionary paths (Gompel and Prud'homme 2009), and (2) regardless of the molecular markers used in this study, each clade preserves its monophyly within the Hypocreaceae; the variation of the phylogenetic position of the three clades (Fig. 3) could be better explained if they are different genera. In light of this evidence, the combination of morphological and phylogenetic data, using various molecular markers in a multilocus analysis (Taylor et al. 2000) suggests the second hypothesis as the most parsimonious. Future research, using genome-based phylogenetic methods, may resolve the relationship of these genera to one another.
The circumscription of Escovopsis raises important questions for the genus. How diverse is Escovopsis? What is its host range? How is the genus phylogeographically distributed? And, what is its role in attine gardens? For many years, Escovopsis was considered a diverse group of fungi (Gerardo et al. 2006a, b;Rodrigues et al. 2008Rodrigues et al. , 2011Caldera et al. 2009;Pagnocca et al. 2012;Yek et al. 2012;Meirelles et al. 2015a, b). However, that assumption was based on considering Sympodiorosea spp., L. trichodermoides and the clades C and D (putative new genera; Fig. 1, Additional file 3: Fig. S2) within Escovopsis. Currently, E. aspergilloides, E. clavata, E. lentecrescens, E. microspora, E. moelleri, E. multiformis, and E. weberi are the only species formaly described within the genus Escovopsis analysed using a multilocus approach based on five molecular markers (Fig. 3G). The species introduced by Marfetán et al. (2018) also belong to Escovopsis (Additional file 2: Fig. S1), nonetheless, the phylogeny of these species are still unclear. Therefore, future studies should consider that both the genetic diversity and the number of known species of the genus were overestimated. Since the taxonomic conditions to evaluate the macroscopic features and growth rates of Escovopsis species are still not standardized, the assessment to the morphological diversity of the genus and description of new species are still limited. Accordingly, future studies should access the morphology of the formally described Escovopsis species to create a standardized taxonomic framework of the genus, and strengthen the foundations of its systematics. Finally, several studies provided evidence that some Escovopsis species act as specialized Page 16 of 18 Montoya et al. IMA Fungus (2021) 12:23 mycoparasite (Currie et al. 1999a(Currie et al. , 2003Currie 2001;Gerardo et al. 2004;Little and Currie 2007;de Man et al. 2016). However, a recent study showed that Escovopsis species may act as an opportunistic fungus in attine ant colonies depending on host susceptibility (Jiménez-Gómez et al. 2021). Either way, the parasitic mechanisms of the Escovopsis species that have a parasitic behaviour are still poorly understood, and the hypothesis of specialized mycoparasite for the genus was also raised considering Sympodiorosea spp., L. trichodermoides and the clades C and D ( Fig. 1 and Additional file 3: Fig. S2) within the same genus. Therefore, future studies should unveil the phylogenetic correspondence of Escovopsis with the ants and the mutualistic fungi, and carefully address the mechanisms of the parasitism of this group of fungi. Similar questions as those raised for Escovopsis must be addressed in future studies for Sympodiorosea, Luteomyces and the clades C and D. The genetic and morphological diversity, as well as the geographical distribution of Sympodiorosea, Luteomyces and the two putative new genera ( Fig. 1 and Additional file 3: Fig. S2), are still a mystery. Some authors suggested that Sympodiorosea spp. and L. trichodermoides are more likely to be associated with the colonies of lower attine ants than higher attine (e.g., leaf-cutting) ants (Gerardo et al. 2004(Gerardo et al. , 2006b, but more evidence is necessary to confirm this hypothesis. Recent studies have shown that some strains of S. kreiselii behave as antagonists of the mutualistic fungus of Mycetophylax morschi being able to kill it under laboratory conditions (Custodio and Rodrigues 2019). However, the mechanisms by which it manages to kill the cultivars are completely unknown. On the other hand, Bizarria et al. (2020) demonstrated that L. trichodermoides had little negative impact on the mutualistic fungus of Mycocepurus goeldii, being just able to inhibit the fungus cultivars in-vitro and unable to overcome defenses of the ant colonies. The taxonomy, ecology, and lifestyle of the clades C and D ( Fig. 1 and Additional file 3: Fig. S2) previously treated as Escovopsis are still unknown. Therefore, all assumptions about these groups of fungi should be reconsidered and examined carefully in future studies.

Conclusion
Since Möller (1893) observed the "fungi of the strong conidial shapes" within the fungus gardens of fungusgrowing ants, several groups of fungi that share the same habitat were classified as Escovopsis. Many phylogenetic incongruities have been reported in the last two decades, and the lack of phylogenetic studies for this genus has hampered scientists to recognize the root of the problem. After a detailed systematic study, we conclude that taxonomic disagreements in the genus Escovopsis were caused due to the inclusion of two groups of fungi that belong to different genera (Luteomyces and Sympodiorosea) within the same genus. This discovery not only solves the phylogenetic disagreements of the genus but significantly expands our understanding of the systematics of Escovopsis, and related genera, and provides a stable foundation from which to build future research on the evolutionary history, taxonomic diversity, and ecological roles of these unique fungi.