Integrative analysis of the West African Ceraceosorus africanus sp. nov. provides insights into the diversity, biogeography, and evolution of the enigmatic Ceraceosorales (Fungi: Ustilaginomycotina)

The order Ceraceosorales (Ustilaginomycotina) currently includes the single genus Ceraceosorus, with one species, Ceraceosorus bombacis, parasitic on Bombax ceiba in India. The diversity, biogeography, evolution, and phylogenetic relationships of this order are still relatively unknown. Here, a second species of Ceraceosorus is described from West Africa as a novel species, Ceraceosorus africanus, infecting Bombax costatum in Benin, Ghana, and Togo. This species produces conspicuous fructifications, similar to corticioid basidiomata when mature, but sorus-like in early stages of ontogenetic development. The fructifications cover much of the leaf surface and resemble leaf blight. This contrasts with the inconspicuous fructifications of C. bombacis comprising small spots scattered over the lower leaf surface that resemble leaf spot. Both species of Ceraceosorus differ in several micromorphological traits, infect different host plant species in widely separated geographical areas, and are separated by a considerable genetic distance in 28S rDNA and RPB2 genes. The distinct corticioid fructification of C. africanus is a unique morphological trait within the Ustilaginomycotina. Molecular phylogenetic analyses of a single gene dataset (D1/D2 28S rDNA) supported the monophyly of the two Ceraceosorus species and the Ceraceosorales and their placement within the Ustilaginomycotina. Molecular phylogenetic analyses of a multigene dataset (18S/5.8S/28S rDNA/RPB2/TEF1) revealed Exobasidium rhododendri (Exobasidiales) as the closest relative of Ceraceosorus, both clustering together with Entyloma calendulae (Entylomatales), indicating affinities to the Exobasidiomycetes. This phylogenetic placement is in agreement with ultrastructural characteristics (presence of local interaction zone and interaction apparatus) reported for the Ceraceosorales, Entylomatales, and Exobasidiales.


Introduction
Historically, smut fungi were defined as phytoparasites producing (usually) dusty masses of dikaryotic teliospores within plant tissues, germinating to develop basidia with basidiospores, which grow as a saprobic yeast stage in the haplophase. They were included in the single order Ustilaginales (Schröter 1889;Clinton 1906;Zundel 1953). Parasites with such a life strategy and organization are now referred to as classical smuts (Vánky 2001), true smuts (Oberwinkler 2012), or teliosporic smuts . This idea of smut fungi and their classification remained almost unchanged until the end of the twentieth century. Comprehensive studies by  on hyphal septal pores and interaction zones between parasites and host cells using transmission electron microscopy revealed enormous structural diversity that led to the new classification and changed the concept of smut fungi and their relationships to other fungal parasites. These findings were supported by the molecular phylogenetic studies of Begerow et al. (1997). These two complementary studies demonstrated that the classical smut fungi (at that time assigned to the single order Ustilaginales) represented several distinct lineages regarded as the orders Doassansiales, Entorrhizales, Entylomatales, Georgefischeriales, Tilletiales, Urocystidales, and Ustilaginales. These orders were interspersed by nonteliosporic taxa forming two main lineages referred to as the orders Exobasidiales (incl. Graphiolales, Begerow et al. 2002a) and Microstromatales. The new concept of smuts and related nonteliosporic fungi resulted in the description of the new class Ustilaginomycetes , that was subsequently elevated to the level of a subphylum, the Ustilaginomycotina . It became apparent that several genera and species should be removed from the smut fungi since they showed relationships to rust fungi. These are now referred to as false smuts (Oberwinkler 2012) and are classified in the Pucciniomycotina, order Microbotryales Begerow et al. 1997). Recently, two additional genera, Entorrhiza C.A. Weber and Talbotiomyces Vánky, R. Bauer & Begerow, were removed from the smut fungi and classified in the new phylum, Entorrhizomycota, distinct from the phyla Ascomycota and Basidiomycota Riess et al. 2015).
The new concept of smuts and related fungi was continuously augmented by further phylogenetic studies that revealed an enormous diversity of life strategies and structural organization of sporulation within the lineages of Ustilaginomycotina. Thus, in addition to classical smut fungi and nonteliosporic species, this subphylum includes lipophilic yeasts associated with warm-blooded animals or marine environments, the species of Malassezia Baill. (Begerow et al. 2000;Amend 2014;Cabañes 2014;Wang et al. 2014), mite-associated yeasts (Boekhout et al. 2003), and saprobic yeasts isolated from diverse ecosystems (Begerow et al. 2000;Boekhout et al. 2006;Nasr et al. 2014;Piątek et al. 2015;Wang et al. 2014). Several species were also isolated from healthy plant tissues and may act as endophytes (Paz et al. 2007;Amin et al. 2010;Takahashi et al. 2011;Padhi and Tayung 2013;Rush and Aime 2013). The most notable of them is the recently described Violaceomyces palustris Albu, Toome & Aime that represents a distinct lineage, the order Violaceomycetales (Albu et al. 2015). Sporulation in the Ustilaginomycotina is notably diverse, including budding cells, simple hyphae with conidiogenous cells producing conidia, teliospores embedded between plant cells or replacing generative or vegetative plant organs, and loose fascicles or hymenial layers of basidia emerging from stomata or the epidermis.
To date Ceraceosorus B.K. Bakshi is a monotypic genus and contains only Ceraceosorus bombacis (B.K. Bakshi) B.K. Bakshi, a leaf pathogen of the silk cotton tree Bombax ceiba L.
(Malvaceae) in Uttar Pradesh, India. This fungus was found and initially described as Dicellomyces bombacis B.K. Bakshi by Bakshi (in Bakshi et al. 1972). Only 3 years later, it was transferred, also by Bakshi, to the new genus Ceraceosorus, distinct from Dicellomyces L.S. Olive (Olive 1945(Olive , 1951 in having sorus-like fructifications, indeterminate hymenial thickenings, intracellular hyphae, basidia with internal thickenings, and lacking cystidia (Cunningham et al. 1976). Ceraceosorus was placed in the Brachybasidiaceae (Cunningham et al. 1976;Kirk et al. 2001). C. bombacis may be an important phytopathogenic species in India as reported in the original publications (Bakshi et al. 1972;Cunningham et al. 1976), though in recent years, it was not included as such in any original phytopathological report. C. bombacis is unique in respect to the formation of intracellular hyphae, a character not known in any other order of Exobasidiomycetes, stressing the isolated phylogenetic position of this species outside all main lineages revealed by molecular phylogenetic analyses . Thus, to accommodate the genus Ceraceosorus, Begerow et al. (2006) added the novel order Ceraceosorales, assigned to the class Exobasidiomycetes. The family Ceraceosoraceae was subsequently described by Denchev and Moore (2009).
The single genus and species currently residing in the Ceraceosorales implies that different aspects such as evolution, biogeography, diversity, and phylogenetic relationships of this order are still obscurely known. Some of these missing aspects could be improved now from the discovery of a second member of Ceraceosorales infecting a native woody species, the silk cotton tree Bombax costatum Pellegr. & Vuillet (Malvaceae), in the Sudanian savanna biome in West Africa. This unknown fungus was consistently found on most specifically inspected B. costatum trees in Benin, Ghana, and Togo during four independent surveys at the end of the rainy period in the years 2011-2013 and 2015. In this study, polyphasic analyses were conducted to establish the systematic and phylogenetic placement of this fungus, studying its structural (dis-)similarity and genetic divergence compared to Ceraceosorus bombacis, as well as phylogenetic relationships to other orders of the Ustilaginomycotina. Furthermore, environmental information assembled in field studies is given to contribute to the knowledge of the ecology of this fungus.

Specimen sampling and documentation
This study is based on phylogenetic and morphological analyses of a leaf parasite on Bombax costatum (Malvaceae) originating from eight locations in three different countries and diverse habitats in West Africa (Benin, Ghana, Togo; Figs. 1 and 2). The voucher specimens were deposited in the fungal collection of the W. Szafer Institute of Botany of the Polish Academy of Sciences, Kraków, Poland (KRAM F). Additionally, the ex-type culture of Ceraceosorus bombacis isolated from Bombax ceiba (Cunningham et al. 1976) was anew sequenced for phylogenetic analyses. Detailed specimen information is given in Table 1. Nomenclatural novelty was registered in MycoBank (www.mycobank.org).

Morphological analyses
Morphological characters of fructifications were studied using dried herbarium material. Observations of the macroscopic appearance of fructifications were made with the naked eye and under a stereoscopic microscope Nikon SMZ-2T. Micromorphological characters were studied by light microscopy. For this purpose, thin freehand sections of fructifications were made under the stereoscopic microscope using a razor blade; mounted in 3 % (wt/vol) aqueous potassium hydroxide, 1 % (wt/vol) aqueous phloxine, Melzer's reagent, and 0.1 % Cotton Blue (wt/vol) in 60 % (wt/vol) lactic acid; and examined under a Nikon Eclipse E-400 light microscope. In presenting the spore size variation, 5 % of measurements were excluded from each end of the range and are given in parentheses. In the species description, the following abbreviations were used: L = mean length of all spore measurements, W = mean spore width, Q = length to width range ratio, Q m = mean Q value, and n = number of measurements. Q values were obtained from dividing the average basidiospore length by width. Line drawings were made from slides mounted in aqueous potassium hydroxide.

Molecular phylogenetic analyses
For DNA isolation, approximately 0.25 cm 2 of dried leaves of Bombax costatum with disease symptoms and a 0.25 cm 2 agar block containing a living culture of Ceraceosorus bombacis, respectively, were selected, deep-frozen in liquid nitrogen, and ground several times with a plastic pestle. Total genomic DNA was subsequently extracted using the InnuPREP Plant DNA Kit (Analytik Jena, Jena, Germany) following the standard protocol. C. bombacis was processed again since no 5.8S, RPB2, or TEF1 sequences were available, and only a short 28S sequence was available in the NCBI's GenBank database (www.ncbi.nlm.nih.gov).
All DNA regions were aligned separately with MAFFT 7.147b using the E-INS-i option (Katoh et al. 2005;Katoh and Standley 2013). In the rDNA datasets, ambiguously aligned regions were removed using Gblocks 0.91b (Castresana 2000) with gaps allowed at alignment positions at which gaps were present in more than half of the sequences, while the protein coding sequences were modified manually with the help of annotated sequences of RPB2 (DQ408132) and TEF1 (X73529). The final alignment length for dataset 1 (D1/D2 28S) was 447 bp (15 % of the initial alignment length before Gblocks), and for dataset 2 (18S + 5.8S + 28S + RPB2 + TEF1) 1505 bp for partial 18S (54 %), 163 bp for complete 5.8S, 1272 bp for partial 28S (48 %), 972 bp for the complete exon 6 of RPB2, and 984 bp for partial TEF1. The single-gene alignments were then concatenated into one sequence alignment of a total of 4896 bp (dataset 2). Phylogenetic analyses were computed for both datasets using maximum likelihood (ML) with combined rapid bootstrapping under the GTRCAT model from 1000 runs Table 1 List of Ceraceosorus species, host plant species, GenBank accession numbers (D1/D2 28S), and voucher information for specimens used for phylogenetic (see Fig. 3) and morphological analyses using RAxML 8.0.17 (Stamatakis 2014). Additional posterior probability nodal support values were determined in a Bayesian phylogenetic MCMC search using MrBayes 3.2.2 (Ronquist et al. 2012) under the general time reversible model with gamma-distributed rate variation (GTR + G) as suggested in the MrBayes version 3.2 manual. Each search comprised two runs of four chains each for 10 × 10 6 generations sampled every 100 generations with the first 2.5 × 10 6 generations discarded as burn-in. Genetic divergences (uncorrected p-distance) within the Ceraceosorales were calculated for 18S, 28S, and RPB2 using Mesquite 2.75 (Maddison and Maddison 2011) and the unmodified alignments.

Morphology: ontogenetic development
The fungus on Bombax costatum developed intracellular hyphae in host cells that agglomerated, as intercellular hyphae, in substomatal cavities and emerged through stomata forming sorus-like fructifications with a densely packed and strongly agglutinated hyphal subhymenial layer (that could be referred to as sterile external stroma), and an outer hymenial layer with basidioles and basidia. The adjoining sorus-like fructifications coalesced to form a continuous, compact, widely effused layer on the host leaves. The single sorus-like fructifications were visible best at the edge of the continuous layer, whereas after coalescence, they were undifferentiated and whole fructifications resembled those of corticioid fungi. The predominant parts of the fructifications were sterile, composed only of basidioles, waxy and pinkish in color. The fertile areas were scattered over the sterile areas and were visible under lens as whitish and slightly pruinose areas. The detailed morphological characterization of the fungus on B. costatum is included in the species description and depicted in Figs. 5a-f, 6, 7, and 8.

Molecular phylogenetic analyses
D1/D2 28S sequences of the six specimens of the leaf parasite on Bombax costatum were identical. Their phylogenetic Table 2 List of species and GenBank accession numbers used for the combined 18S/5.8S/28S/RBP2/TEF1 analyses (see Fig. 4) GenBank accession numbers Accession numbers of sequences generated in this study are given in italics. Specimen information can be inferred from the respective GenBank accession numbers relationships, analyzed with 90 sequences representing sampled genera of the Ustilaginomycotina (plus two outgroup species from Pucciniomycotina), are shown in Fig. 3. In all analyses, the specimens of the leaf parasite on B. costatum   both Bayesian and ML analyses. However, the phylogenetic position of the Ceraceosorales within the Exobasidiomycetes was not r esolved, and the Exobasidiomycetes were not supported as monophyletic (Fig. 3).   The phylogenetic analyses of the concatenated 18S/5.8S/ 28S/RPB2/TEF1 dataset (Table 2) revealed the representative of the Exobasidiales, Exobasidium rhododendri (Fuckel) C.E. Cramer, as the closest relative of Ceraceosorus bombacis supported by 83 ML bootstrap and 0.89 posterior probability (Fig. 4). The Ceraceosorales, Exobasidiales, Entylomatales, and Doassansiales were monophyletic, however, without statistical support. In congruence with the topology derived from D1/D2 28S sequences, the monophyly of the Exobasidiomycetes was not supported.  Description: Parasitic on Bombax costatum. Fructifications on the leaves, infection local, only single leaves infected. Fructifications hypophyllous, (100-)200-500 μm thick, resupinate, widely effused, closely adnate, waxy, cracked in old specimens, initially sorus-like and forming circular patches up to 0.5 mm in diam. (best visible in the marginal parts of the fructifications), later coalescent forming continuous, compact areas up to 6.5 cm long, covering a part of the hypophyllous surface, usually about 1/2 or 2/3 of the leaf blade, typically starting from the petiole, exceptionally from other parts of the leaf blade, rarely covering the whole hypophyllous area; the corresponding epiphyllous area clearly discolored, usually yellowish or brownish, finally blackish and necrotic; the fructifications sometimes appear on the upper side of leaves but then only in the form of thin, linear parallel rows on both sides of central nerves. Hymenial surface smooth, white and slightly pruinose (visible under lens) in fertile parts, pinkish to pinkish brown in sterile parts. Hyphal system monomitic. Hyphae 1-3 μm wide, lacking clamps, thin-to thick-walled, mostly hyaline or pale yellow in substomatal cavities, sometimes with very fine, scattered oil drops inside; hyphae sparsely branched, straight or slightly sinuous, at first intracellular, later agglomerating, as intercellular hyphae, in substomatal cavities and emerging through the stomata to form a dense, compact structure in all parts of fructifications, including subhymenium. Hymenium forming a dense palisade of basidioles and basidia. Basidioles 3-4 μm wide, cylindrical, usually thin-walled, sometimes thick-walled with even walls up to 0.5 μm thick, often with granular oil drops inside. Paraphysoid hyphae sometimes present among basidioles, 0.4-0.8 μm wide, thin-walled, sparsely branched or unbranched. Basidia 48-82 × 3-5 μm (measured without sterigmata), dacrymycetoid, thin-walled, with cylindrical distal end, tapering toward basal septa, sometimes anastomozing in basal parts (below basal septa, H-shaped in outline), with two prominent sterigmata up to 15 μm long; sterigmata emerge from the apical part of basidioles, at first as semicircular, later as elongated and obtuse, and finally as subulate (pointed) projections,   16-4.33, Q m = 3.14, n = 90), elongated, lacrymoid, usually with curved apical part, obtuse at distal end, thin-walled, hyaline, with somewhat darkened and refractive apical part, inamyloid, acyanophilous. Conidiophores and conidia not observed.
Additional specimens examined ( Comments: The description above is based on the holotype specimen from Ghana. The materials from other localities were roughly similar, with the exception of the specimen from Togo (KRAM F-57389) that differs in having 1-sterigmate basidia and longer basidiospores, which were up to 25 μm long. Such morphology of basidia and size of basidiospores are considered an abnormality or variation rather than the typical morphology of Ceraceosorus africanus. Basidiospores in 1-sterigmate basidia may be larger than in 2-sterigmate basidia because there are more nutrients available. Similar abnormalities in basidiospore sizes produced on 1-2-sterigmate and 4-sterigmate basidia were already observed in other fungi (Lebel and Castellano 2002;Kautmanová et al. 2012). The specimens collected in September 2015 from Benin were not fully mature.
Distribution and ecology: Ceraceosorus africanus was found at eight locations in Benin, Ghana, and Togo, within the Sudanian savanna biome. The northernmost location was found near the Tanougou Waterfalls in Benin, at about 10°48′ 21″ N; the southernmost location ca. 13.5 km W of Fufulsu in Ghana, at about 09°09′ 05″ N; the westernmost location ca. 18 km W of Fufulsu in Ghana, at about 01°25′ 47″ W; and the easternmost location near Bassa in Benin, at about 02°43′ 41″ E. These localities are situated in the core geographical range of Bombax costatum (Fig. 1), but it is likely that C. africanus occurs with the host plant over its distribution in West Africa. The habitat conditions for five of the eight locations where C. africanus was found were similar: open Sudanian savanna with scattered shrubs and trees, with infected trees of B. costatum fully exposed to the sun. At the locality near the Tanougou Waterfalls, the infected tree was found in the transition zone between gallery forest and rocky savanna in a semishaded place, at the locality in the Forêt Classée de Béléfoungou in the transition zone between gallery forest and the Sudanian woodland, and at the locality in Tébou on the top of the inselberg (Fig. 2). The first infection symptoms with immature fructifications were observed at the end of September (observations in 2015 in Benin). Infected leaves with mature fructifications were found at the end of October and at the beginning of November, which corresponds to the beginning of the dry season in West Africa. The phenological appearance of C. africanus is in agreement with phenology of plant pathogenic fungi in other tropical regions (Piepenbring et al. 2015). The level of infection was very low, ranging from a single infected leaf to about 1 % of infected leaves on one particular tree. At the type locality in Ghana, two trees of B. costatum were growing side by side, at a distance of ca. 5 m, and only one of them was infected by C. africanus, though the level of infection was ca. 1 %, which was the highest observed infection among all detected locations of C. africanus. The locality near the Tanougou Waterfalls in Benin was visited in 2011 and 2012, and a weak infection (one leaf) of the small tree was observed only in 2011, while in the subsequent year, the same tree was healthy. This suggests that infection by C. africanus may not appear on the same tree/population every year.

Discussion
Ceraceosorus africanus differs somewhat from the generic characters outlined in the diagnosis of Ceraceosorus (Cunningham et al. 1976), notably in lacking internal thickenings of basidia and in having a distinct hymenium produced on an external subhymenial layer (that could be referred to as sterile external stroma). Cunningham et al. (1976) reported in the generic diagnosis that Ceraceosorus shows an indeterminate hymenial thickening, but it is difficult to conclude what they meant by this term. Moreover, Cunningham et al. (1976) differentiated Ceraceosorus from Dicellomyces as lacking a well-defined external stroma, but this character was not included in the generic diagnosis, only mentioned in the text.
Considering the ontogenetical development and other aspects of morphology (intracellular hyphae, sorus-like fructifications at least in the initial stage of development, waxy and pinkish fructifications, morphology of basidia and basidiospores), as well as the same host genus (Bombax L.), both C. africanus and Ceraceosorus bombacis are however similar and should be considered con-generic. These divergent characters may therefore not be relevant at the generic level but useful to distinguish between C. africanus and C. bombacis. In addition, C. africanus is morphologically different from C. bombacis (Cunningham et al. 1976) in having somewhat longer and thinner basidia [48-82 × 3-5 μm vs. (20-)35-50(−85) × 3-6.5 μm], shorter sterigmata (up to 15 μm long vs. up to 22 μm long) and slightly longer and wider basidiospores [(6.5-)9-16(−18) × (3-)3.4-4.8(−5.2) μm vs. (5.5-)8.5-14(−18.5) × 2-4.5 μm)] and, most importantly, in having a different macroscopic symptoms of infection. The infection of leaves is apparently local in C. africanus, i.e., only single leaves on one tree are infected, but fructification covers large parts of the leaf surface and resembles leaf blight. In contrast, the infection of C. bombacis resembles leaf spot and comprises small light pink to brown spots scattered over the lower leaf surface (Bakshi et al. 1972;Fig. 5g-i). Although it is premature to generalize this observation based on two known species, this difference in macroscopic symptoms resembles those in Exobasidium Woronin (Begerow et al. 2002a;Piątek et al. 2012), where the same host plant may be infected by species producing leaf spots or leaf blights (as surculicolous or systemic infection), e.g., Exobasidium arescens Nannf. and Exobasidium myrtilli Siegm., respectively, on Vaccinium myrtillus L. (Ericaceae), and many others (Nannfeldt 1981).
The prominent and compact corticioid fructifications of Ceraceosorus africanus characterized by well-developed subhymenia and hymenia composed of dense palisades of basidioles and basidia, and resembling, for example, agaricomycotinous genera Gloeocystidiellum Donk, Phanerochaete P. Karst., or Phlebia Fr., represent a unique morphological trait in the subphylum Ustilaginomycotina. Indeed, some species of Exobasidium, like Exobasidium vaccinii (Fuckel) Woronin, form corticioid hymenia (Nannfeldt 1981), but not as prominent and well developed as in C. africanus. The fructification of other nonteliosporic Ustilaginomycotina species is rather small or loose and delicate. Obviously, this character cannot be considered an autapomorphy for C. africanus since this type of fructification evolved multiple times in different Basidiomycota lineages, most commonly in different clades of Agaricomycotina (corticioid fungi, Larsson 2007), but also in some species of Pucciniomycotina, e.g., in the rust Stereostratum corticioides (Berk. & Broome) H. Magn. (Cummins and Hiratsuka 2003), or in the atractiellomycete Saccoblastia farinacea (Höhn.) Donk ). However, it expands the structural diversity of sporulation within the Ustilaginomycotina.
The two currently known Ceraceosorus species are fully allopatric, separated both by geographic distribution and host plant species. Ceraceosorus africanus is known only on Bombax costatum in West Africa and Ceraceosorus bombacis occurs only on Bombax ceiba in India. The natural range of both host plants is widely separated and does not overlap in any area. B. costatum is naturally distributed in West Africa, while B. ceiba in tropical southern and eastern Asia, Indonesia, Philippines, Papua New Guinea, and northern Australia. Assuming that both Ceraceosorus species are strictly host species specific, it could be hypothesized that the speciation of C. africanus and C. bombacis was triggered by geographical isolation and ecological factors (host plant species).
The two currently known Ceraceosorus species are separated by a considerable genetic distance (uncorrected p values: 28S = 4.03 % or 83 bp, RPB2 = 18.5 % or 196 bp). Such a considerable genetic distance between Ceraceosorus africanus and Ceraceosorus bombacis could, e.g., be an indication for an ancient separation of the lineages under a model of neutral sequence evolution, or an indication for a selection of the loci under a nonneutral model of sequence evolution. Assuming an ancient separation of the lineages, the currently recognized species (1) may represent living descendants of a larger but extinct Ceraceosorus diversity, or (2) undiscovered species still may exist in poorly studied tropical countries, especially on the remaining Bombax species (Bombax buonopozense P. Beauv., Bombax insigne Wall., Bombax mossambicense A. Robyns; Malvaceae). The discovery of C. africanus, a common species producing conspicuous symptoms of infection on the common tree, Bombax costatum, but occurring in the phytopathologically unstudied Sudanian savanna biome of West Africa could support the latter assumption. Moreover, similar to other lineages of Ustilaginomycotina, saprobic yeast species may contribute to unknown Ceraceosorus diversity. In fact, one yeast species assigned to this genus was announced but not yet described (Kijpornyongpan and Aime 2014;Albu et al. 2015).
The discovery of Ceraceosorus africanus in West Africa raises the number of smut species parasitic on members of the Malvaceae. It also highlights the significance of scrutiny of the tropical ecosystems in order to increase the knowledge of fungal diversity and to find evolutionary unique Bmissing^fungal taxa in general (e.g., Isaac et al. 1993;Aime and Brearley 2012;Gazis et al. 2012) and smut fungi in particular. Previously, eight smut species were recognized on hosts of that family, including five teliosporic species from the genera Entyloma de Bary, Geminago Vánky & R. Bauer, and Pericladium Pass. (Table  3; Vánky 1996Vánky , 2011Vánky , 2012, and three nonteliosporic species from the genera Ceraceosorus and Volvocisporium Begerow, R. Bauer & Oberw. (Table 3; Cunningham et al. 1976;Ritschel et al. 2008). All of these genera are known exclusively on hosts in the Malvaceae with the exception of Entyloma, which members infect hosts in multiple dicot families (Begerow et al. 2002b;Vánky 2012;Lutz and Piątek 2016), and only Entyloma sidae-rhombifoliae Cif. (Ciferri 1928, but doubtful smut species according to Zundel 1939) is parasitic on the Malvaceae. Interestingly, Ceraceosorus and Geminago are restricted to host species from different subfamilies, the Bombacoideae and the Helicteroideae, respectively, while Pericladium and Volvocisporium are both parasitic on members of the Grewioideae (Table 3). However, these smut genera are not phylogenetically related, indicating that in the evolution of smut fungi, hosts in Malvaceae were colonized several times from different ancestors. Interestingly, all smut species on hosts in the Malvaceae, except North American En. sidae-rhombifoliae (Vánky 2012), are confined to the African-Indian-Australian part of Gondwanaland, and Ceraceosorus and Volvocisporium have allopatric species in Africa and India. The disjunctive occurrence of smuts between Africa and India was already reported for some species (e.g., Piątek et al. 2014), but the mechanisms responsible for such a distribution pattern are not yet known.
The phylogenetic affinities of the Ceraceosorales, represented by Ceraceosorus bombacis, have previously been analyzed only few times , the same analyses later repeated by Begerow et al. 2014;Albu et al. 2015;Sharma et al. 2015). In the multigene analyses, including ITS, 28S, ATP6, and β-tubulin sequences, conducted by Begerow et al. (2006), C. bombacis was nested within the Exobasidiomycetes occupying a common branch with Tilletiopsis albescens Gokhale and members of the Entylomatales, however without statistical support. In the multigene analyses, including 18S, ITS, 28S, TEF1, and βtubulin sequences, conducted by Albu et al. (2015), the Ceraceosorales was weakly supported as member of the Ustilaginomycetes. Thus, the affinities of the Ceraceosorales were not resolved by these analyses Albu et al. 2015).
The molecular phylogenetic analyses of the single gene dataset (28S) conducted in this study (Fig. 3), including both Ceraceosorus africanus and Ceraceosorus bombacis, supported the monophyly of Ceraceosorus and confirmed the placement of this genus and the order Ceraceosorales within the Ustilaginomycotina in an unresolved relation to the Ustilaginomycetes and Exobasidiomycetes.
The molecular phylogenetic analyses of the multigene dataset (18S/5.8S/28S/RPB2/TEF1; Fig. 4) revealed Exobasidium rhododendri, i.e., member of the Exobasidiales, as the closest relative of Ceraceosorus, both clustering together with Entyloma calendulae, i.e., member of the Entylomatales. Thus, it is plausible that the Ceraceosorales belongs to the Exobasidiomycetes (sensu Begerow et al. 2006). It is noteworthy, however, that in this study, the Exobasidiomycetes were not resolved as monophyletic in both the single gene and the multigene analyses (Figs. 3 and 4). This is in congruence with several previous analyses made by, e.g., Begerow et al. (2006, dataset 18S/ITS/28S/ATP6/β-tubulin, their Supplementary Figure 1) or Wang et al. (2014). Interestingly, considering the ultrastructural characters studied by , the Exobasidiomycetes also have been divided into two groups: orders with species lacking an interaction apparatus (Georgefischeriales, Microstromatales, Tilletiales) and orders with species having an interaction apparatus (Entylomatales, Exobasidiales, Doassansiales), the latter referred to as superorder Exobasidianae. The ultrastructural characters of Ceraceosorus bombacis (Begerow et al. , 2014, such as a local interaction zone with simple interaction apparatus, indicate its affinity to the superorder Exobasidianae. In this superorder, Exobasidiales and Doassansiales are characterized by  Begerow et al. (2001) having a complex interaction apparatus, while Entylomatales is characterized by having a simple interaction apparatus Begerow et al. 2006Begerow et al. , 2014. Thus, the Ceraceosorales shares this ultrastructural trait only with the Entylomatales. This could suggest a close phylogenetic relation of the Ceraceosorales to the Entylomatales, which was also revealed (though statistically not supported) in the multigene analyses of Begerow et al. (2006Begerow et al. ( , 2014, as well as inferred and well supported in the multigene analyses of the current study. The superorder Exobasidianae has been resolved as monophyletic in the ITS/28S/ATP6/β-tubulin analysis of Begerow et al. (2006). In the current multigene analyses, the Exobasidianae formed a monophyletic group but without statistical support (Fig. 4). Thus, we are in favor of the following evolutionary scenario for the Ceraceosorales and the Exobasidiomycetes (sensu Begerow et al. 2006): (1) the order Ceraceosorales belongs to the superorder Exobasidianae and is related to the Entylomatales and the Exobasidiales; and (2) the class Exobasidiomycetes is composed of two major lineages, probably deserving their classification at the class level, defined by ultrastructural characters of cellular interactions. A robust phylogenetic hypothesis for this scenario may not be resolved by multigene phylogenetic analyses, reinforcing the similar conclusion of Begerow et al. (2006) for a different dataset. Probably, only phylogenomics may provide a robust hypothesis for the phylogeny of the Ustilaginomycotina. In recent phylogenomic analyses conducted by Sharma et al. (2015), including genomes of six species of the Ustilaginomycotina, all branches received maximum statistical support. Ceraceosorus bombacis was resolved as sister species to Ustilago maydis (DC.) Corda, Sporisorium reilianum (J.G. Kühn) Langdon & Full., Ustilago hordei (Pers.) Lagerh., and Melanopsichium pennsylvanicum Hirschh., while Malassezia globosa Midgley, E. Guého & J. Guillot was resolved as basal to all sampled species. The sampling by Sharma et al. (2015) included members of only three orders (Ustilaginales, Ceraceosorales, Malasseziales), and the inclusion of representatives of the remaining orders to phylogenomic analyses is a challenge for future studies.