UC Davis UC Davis Previously Published Works Title Phylogenetics and Taxonomy of the Fungal Vascular Wilt Pathogen Verticillium , with the Descriptions of Five New Species

Knowledge of pathogen biology and genetic diversity is a cornerstone of effective disease management, and accurate identification of the pathogen is a foundation of pathogen biology. Species names provide an ideal framework for storage and retrieval of relevant information, a system that is contingent on a clear understanding of species boundaries and consistent species identification. Verticillium, a genus of ascomycete fungi, contains important plant pathogens whose species boundaries have been ill defined. Using phylogenetic analyses, morphological investigations and comparisons to herbarium material and the literature, we established a taxonomic framework for Verticillium comprising ten species, five of which are new to science. We used a collection of 74 isolates representing much of the diversity of Verticillium, and phylogenetic analyses based on the ribosomal internal transcribed spacer region (ITS), partial sequences of the protein coding genes actin (ACT), elongation factor 1-alpha (EF), glyceraldehyde-3-phosphate dehydrogenase (GPD) and tryptophan synthase (TS). Combined analyses of the ACT, EF, GPD and TS datasets recognized two major groups within Verticillium, Clade Flavexudans and Clade Flavnonexudans, reflecting the respective production and absence of yellow hyphal pigments. Clade Flavexudans comprised V. albo-atrum and V. tricorpus as well as the new species V. zaregamsianum, V. isaacii and V. klebahnii, of which the latter two were morphologically indistinguishable from V. tricorpus but may differ in pathogenicity. Clade Flavnonexudans comprised V. nubilum, V. dahliae and V. longisporum, as well as the two new species V. alfalfae and V. nonalfalfae, which resembled the distantly related V. albo-atrum in morphology. Apart from the diploid hybrid V. longisporum, each of the ten species corresponded to a single clade in the phylogenetic tree comprising just one ex-type strain, thereby establishing a direct link to a name tied to a herbarium specimen. A morphology-based key is provided for identification to species or species groups. Citation: Inderbitzin P, Bostock RM, Davis RM, Usami T, Platt HW, et al. (2011) Phylogenetics and Taxonomy of the Fungal Vascular Wilt Pathogen Verticillium, with the Descriptions of Five New Species. PLoS ONE 6(12): e28341. doi:10.1371/journal.pone.0028341 Editor: Alexander Idnurm, University of Missouri-Kansas City, United States of America Received August 5, 2011; Accepted November 6, 2011; Published December 7, 2011 Copyright: 2011 Inderbitzin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Financial support for this research was provided by the California Leafy Greens Board (www.caleafygreens.ca.gov/) and the United States Department of Agriculture National Institute of Food and Agriculture (USDA-NIFA) (www.csrees.usda.gov/fo/specialtycropresearchinitiative.cfm). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: kvsubbarao@ucdavis.edu


Introduction
The genus Verticillium comprises a small group of plantpathogenic fungi that cause billions of dollars of damage annually to a variety of agricultural crops in many parts of the world [1]. Verticillium species are soil-borne and cause Verticillium wilt, a plant disease that affects the vasculature of many different hosts [1], and can cause significant crop losses [2].
Control of Verticillium wilt is difficult and costly [3,4]. In the absence of a suitable plant host, Verticillium species can remain dormant in the soil for years by means of small, melanized resting structures that are extremely durable, and will only germinate in the proximity of a suitable host [5].
Verticillium has a long taxonomic history. The first species of Verticillium was first found in 1816 [6], and approximately 190 species have since been described [7]. The species share the characteristic Verticillium conidiophore that is comprised of sporeforming cells that are narrowly flask-shaped, and are assembled into whorls (verticils) and attached along a main axis. The advent of molecular systematics confirmed that Verticillium was composed of several distantly related and ecologically diverse groups which were subsequently removed from Verticillium [8,9] and placed in other genera. These include Lecanicillium, containing insect and fungus pathogens [10,11,12], Pochonia and Haptocillium comprising nematode parasites [13,14], and Gibellulopsis and Musicillium containing plant pathogens [15]. The reduced genus Verticillium, also referred to as Verticillium sensu stricto, thus consisted of only five species of plant associates and plant pathogens [15], and was retypified with V. dahliae [16]. Verticillium is placed in the family Plectosphaerellaceae [15] that is closely related to Colletotrichum in the Glomerellaceae [17], another important group of plant pathogens. Both Plectosphaerellaceae and Glomerellaceae are families of uncertain phylogenetic position in the Hypocreomycetidae, a subclass within the fungal phylum Ascomycota [17,18]. Gibellulopsis and Musicillium are also part of the Plectosphaerellaceae [15], whereas Lecanicillium, Pochonia and Haptocillium are placed in different families in the order Hypocreales of the Hypocreomycetidae [19]. Verticillium species reproduce only asexually, no sexual state is known [20].
Besides playing an important role in the biology of Verticillium species, resting structures are also taxonomically important. Resting structures were first recognized in V. albo-atrum as brown, pigmented hyphae described as 'Dauermycelien' [21], a term translated to 'resting mycelium' by Isaac [22]. Other types of melanized resting structures in Verticillium are chlamydospores that consist of short chains of brown, rounded cells, whereas microsclerotia are rounded, brown cells that occur in clusters. Resting mycelium, chlamydospores and microsclerotia are collectively referred to as resting bodies or resting structures [23]. Resting structures were traditionally used as the primary characteristic to distinguish Verticillium species. Verticillium alboatrum was defined based on the presence of resting mycelium, V. nubilum formed chlamydospores, V. dahliae and V. longisporum produced microsclerotia, and V. tricorpus formed resting mycelium, chlamydospores and microsclerotia simultaneously [22,23,24]. Supporting earlier studies that cast doubt on the usefulness of resting structures as a taxonomic character [25,26], phylogenetic analyses suggested that resting structure morphology may not be a suitable character to identify species. In a recent phylogenetic tree [15], instead of clustering into separate, well-defined groups, four out of five Verticillium species overlapped and only one species was confined to a separate group in the tree.
The goal of this study was to create a solid taxonomic framework for Verticillium, and to determine whether resting structure morphology is a suitable character for species delimitation in Verticillium. The scheme that we developed attaches names to all major, species-level phylogenetic groups in Verticillium, and provides a means for their identification. The new taxonomic system allows for a more reliable and consistent identification of species, and will lead to a significant improvement of our knowledge of Verticillium biology. Potential practical applications are many, and may include more efficient and effective disease management strategies such as pathogen exclusion and successful quarantine.
Our approach was as follows. We first assembled a diverse collection of Verticillium strains to cover much of the known Verticillium diversity. We then studied evolutionary relationships and species boundaries using multigene phylogenetic analyses and morphological investigations. Finally, we determined the correct names for the species recovered by comparison to ex-type strains, herbarium material and the literature, and described new species for groups where no names were available.

DNA sequence data
In order to investigate the phylogenetic relationships between Verticillium species, we generated DNA sequence data for 64 isolates, 317 DNA sequences were submitted to GenBank (Accessions ITS: JN187963-JN188023; ACT: JN188088-JN188151; EF: JN188216-JN188279; GPD: JN188152-JN188215; TS: JN188024-JN188087). An attempt to obtain DNA sequence data from the V. dahliae type specimen failed, as the DNA extract generated from a small part of the Dahlia sp. stem containing V. dahliae microsclerotia did not yield any PCR products (data not shown).

Single-locus analyses
To investigate whether the five single-locus datasets (ITS, ACT, GPD, EF, TS) contained similar phylogenetic information, we first analyzed each dataset individually using parsimony. For each single-locus analysis, we included only one representative of each allele. See Table 1 for descriptive statistics of the single-locus analyses. An ITS alignment, and an alignment of the combined ACT, GPD, EF and TS datasets, were submitted to TreeBase (http://purl.org/phylo/treebase/phylows/study/TB2:S11756).
We did not detect any significant conflict between the most parsimonious trees from the five single-locus datasets on a 70% bootstrap support level (Figures S1, S2, S3, S4, S5), with the following exceptions. Verticillium nubilum was sister group to the clade of V. isaacii, V. klebahnii, V. tricorpus and V. zaregamsianum in the GPD tree with 89% bootstrap support ( Figure S4), but in the EF and TS trees, V. nubilum was sister group to the clade of V. alfalfae, V. dahliae, V. longisporum and V. nonalfalfae with 97% support in both trees ( Figures S3, S5). In the majority of trees, V. zaregamsianum was sister group to the V. isaacii, V. klebahnii and V. tricorpus clade with 100% support whereas in the EF tree ( Figure S3), V. zaregamsianum was equally distantly related to all other Verticillium species. Also, the EF tree differed from the TS tree in the position of Species A1, an unknown ancestral species of the diploid hybrid V. longisporum [27]. Species A1 was a sister group to the clade of Species D1, another unknown ancestral species of V. longisporum [27] and V. dahliae in the TS tree with 76% support ( Figure S5), whereas in the EF tree, Species A1 was sister to the monophyletic group of V. alfalfae, V. nonalfalfae, V. dahliae and Species D1 that were supported by 99% of the bootstrap replicates ( Figure S3). In the remaining single-locus trees, the position of Species A1 was not fully resolved (Figures S1, S2, S4).

Combined analyses
With the goal of improving the phylogenetic resolution, we combined the ACT, EF, GPD and TS datasets into a single alignment for combined analysis. We did not include the ITS dataset since for V. longisporum, the ITS phylogeny does not retrace the evolution of that species [27]. The resulting combined fourlocus alignment comprised 77 taxa and 2658 characters, and was submitted to TreeBase (http://purl.org/phylo/treebase/phylows/ study/TB2:S11756). There were a total of 32 unique multilocus haplotypes ( Table 1). The Bayesian consensus tree is illustrated in Figure 1, it was congruent with the most likely tree (2ln likelihood = 12816.48) and with the 48 most parsimonious trees ( Table 1) that differed at poorly supported branches within V. dahliae and the outgroup Gibellulopsis nigrescens (data not shown, but see support values in Figure 1).
We analyzed the four single-locus datasets jointly despite topological conflicts between them ( Figures S2, S3, S4, S5). To evaluate whether single-locus datasets should be concatenated for combined analyses, a conditional combinability approach is often used which states that datasets should not be combined if there are significant differences between them [28,29]. There is no agreement how much the single-locus datasets are allowed to differ, but topological differences supported by 70-90% of the bootstrap replicates have been used as cutoffs [28,30]. In our case, there were topological differences supported by up to 100% of the parsimony bootstrap replicates between the single-locus datasets, involving the positions of V. nubilum, Species A1 and V. zaregamsianum. However, we found that the four-locus phylogeny comprised 35 groups with .70% support, higher than any of the single-locus trees (Table 1). Also, for V. nubilum, Species A1 and V. zaregamsianum, the combined analyses resulted for each species in the topology that had strongest overall support from the singlelocus phylogenies. But the phylogenetic affinities of V. nubilum and Species A1 remain uncertain (Figure 1), and more data is needed to conclusively determine the closest relatives of these two species in Verticillium. In the combined analyses and in all single-locus datasets but the EF dataset, V. zaregamsianum formed a wellsupported clade with V. isaacii, V. klebahnii and V. tricorpus ( Figure 1). One possibility that could explain this divergence is origin by horizontal transfer of the EF gene in V. zaregamsianum. In conclusion, combined analysis of the single-locus datasets generated a phylogeny with higher overall support than any of the single-locus phylogenies, but did not conclusively settle the phylogenetic positions of V. nubilum and Species A1.

Phylogenetic groups obtained
We were able to infer a robust phylogenetic tree of Verticillium. The majority of the branches received maximal support, species were distinct and well defined, and the relationships between species were generally well resolved. As expected, branches with lower support were mainly present within species [31].
We recognized ten different species based on a phylogenetic species concept [31]. Except for the diploid hybrid V. longisporum [27], species were defined as terminal or subterminal clades receiving maximal support in the phylogenetic analyses based on the combined four-locus dataset ( Figure 1). Each of the nine species level clades contained a single ex-type strain representing herbarium material, and thus linking each clade to one of the following nine species names (Figure 1). Verticillium albo-atrum, V. alfalfae, V. dahliae, V. isaacii, V. klebahnii, V. nonalflalfae, V. nubilum, V. tricorpus and V. zaregamsianum. Alleles of the diploid hybrid V. longisporum were present in the three different clades Species A1, Species D1 and V. dahliae (Figure 1), reflecting the hybrid origin of V. longisporum [27]. Verticillium longisporum evolved at least three different times from four different parental lineages in Species A1, Species D1 and V. dahliae [27]. Species A1 and Species D1 were not linked to any type material and could not be officially described, since Species A1 and Species D1 have never been found [27].
The evolutionary relationships among the Verticillium species was overall well resolved, the species fell into two major clades reflecting morphological similarity. The major clades were Clade Flavexudans containing species producing yellow-pigmented hyphae including V. albo-atrum, V. isaacii, V. klebahnii, V. tricorpus and V. zaregamsianum, and Clade Flavnonexudans with species devoid of yellow-pigmented hyphae, including V. alfalfae, V. dahliae, V. nonalfalfae, and V. longisporum (Figure 1). The exception was V. nubilum whose placement in Clade Flavnonexudans agreed with morphological data, but was only supported in the parsimony analyses ( Figure 1). The other exception was the position of the V. longisporum ancestor Species A1 whose placement in the Bayesian consensus tree (Figure 1) contradicted phylogenetic analyses by Inderbitzin et al. [27] who used a different dataset.
Mating type distribution in V. alfalfae and V. nonalfalfae All seven V. alfalfae and nine V. nonalfalfae isolates were screened for presence of MAT1-1 and MAT1-2 idiomorphs which are the two mating compatibility alleles in ascomycetes [32]. All V. alfalfae isolates showed the MAT1-1 specific PCR band whereas all V. nonalfalfae isolates lacked that band. All V. alfalfae isolates lacked the MAT1-2 specific band, whereas the MAT1-2 specific band was present in all V. nonalfalfae isolates ( Figure 2). Thus, all V. alfalfae isolates likely have MAT1-1 idiomorphs whereas V. nonalfalfae isolates have MAT1-2 idiomorphs.

Taxonomy
The genus Verticillium sensu stricto corresponds to a monophyletic group of taxa comprising V. dahliae that has been conserved as the type of Verticillium [15,16]. We recognize ten species in Verticillium sensu stricto that are listed below in alphabetic order. The information provided for each species was obtained from morphological examination of cultures and herbarium specimens (Figure 3), literature surveys and phylogenetic analyses ( Figure 1).
Distribution and host range. Currently known from Canada, Germany, UK and USA (WI). Substrates include Irish potato and soil collected from Irish potato fields.
The original description of V. albo-atrum by Reinke and Berthold [21] was based on observations from decaying potato stems, and is congruent with our observations from the V. albo-atrum isolates examined in this study. The exception was the presence of yellow pigment associated with hyphal cells (Figure 4l) not seen by Reinke and Berthold [21]. However, Klebahn [35] reported that V. albo-atrum mycelium on Salep Agar medium was white with a yellow tinge (p. 64), whereas V. dahliae mycelium was described as white (p. 65).
In addition to resting mycelium, Verticillium albo-atrum also forms microsclerotia (Figures 4i, 4j, 4k). Microsclerotia are very 'small, firm, frequently rounded masses of hyphae with or without the addition of host tissue or soil.' [36]. The V. albo-atrum microsclerotia were described and illustrated in the protolog on pages 74 and 75, and in Figures 1 and 2 of Plate 9 [21], a translation from the German original is provided by Isaac [22]. The microsclerotia consist of aggregations of brown-pigmented, thick-walled hyphae, no lateral cell divisions are involved in their formation [21]. This is opposed to the microsclerotia of V. dahliae where an increase in width is achieved by the lateral divisions of hyphal cells as described by Klebahn [35] on pages 56 and 57, and illustrated in Klebahn's [35] Figure 8. Microsclerotia were only observed on WA-p and PLYA media, they were absent from strains cultured on PDA. Verticillium albo-atrum has frequently been confused with V. alfalfae and V. nonalfalfae that form resting mycelium but no microsclerotia.
The name 'V. albo-atrum' is correct with or without hyphen (Art. 23.1), the hyphenated form is more commonly encountered in the literature.
Distribution and host range. Currently known from Canada, Japan and the USA (PA), only from alfalfa.
Commentary. Verticillium dahliae is the type of Verticillium and was described by Klebahn [35] from Dahlia sp. cv. Geiselher in Germany (Figure 3a). Verticillium dahliae is not the oldest species of the genus, but it has the largest impact as a pathogen, is common and genetically relatively homogenous, and has thus been conserved as the type of the genus [16,34]. Since a viable exholotype culture is no longer available [33], and DNA extraction attempts from the holotype specimen failed, we designated a V. dahliae epitype with an ex-epitype culture that serves as an interpretive type for molecular studies.
The original description of V. dahliae by Klebahn [35] was based on material from Dahlia sp., and from cultures on Salep Agar medium which is a mixture of polysaccharides contained in orchid tubers [37]. The composition of Klebahn's medium is unknown, but as a reference, Noël [38] isolated fungal symbionts of orchids using a clear, weak decoction of salep containing 2% agar. We examined the V. dahliae holotype material which contains an approximately 50 cm long stalk of Dahlia sp. 'Sorte Geiselher' and several leaves (Figure 3a). The microsclerotia present on the stem  Figure 6i). No conidiophores were observed, these are difficult to detect on Dahlia sp. [35], but are illustrated as part of the protolog [35].
The description of V. dahliae based on V. dahliae strains PD322, PD327 and PD502 agreed with the original description by Klebahn [35] except that we failed to detect strands of erect, hyphal aggregates containing conidia and microsclerotia. Klebahn [35] reported the presence of a slightly wider cell (foot cell) at the base of conidiophores. Since foot cells were absent in culture and we did not inoculate live plants, we were unable to confirm the presence of foot cells in V. dahliae. The dimensions provided by Klebahn [35] for microsclerotia, conidiophores, conidiogenous cells and conidia were at the lower end of the range of dimensions that we observed. Our dimensions were similar to reports in the literature for conidia [22] and microsclerotia [22,39,40,41]. Short brown-pigmented hyphae attached to microsclerotia were illustrated by Klebahn [35], the ones that we observed resemble immature microsclerotia as illustrated by Klebahn [35] and Isaac [22]. Verticillium dahliae resembles V. longisporum but has smaller conidia.
Distribution and host range. Currently known from Canada, UK and USA (CA, WA). Substrates include garden tomato, globe artichoke, hairy nightshade, lettuce, spinach and soil.
Distribution and host range. Currently only known from the USA (CA, WA) from lettuce.
Commentary. Verticillium longisporum is a diploid hybrid that originated at least three different times from four different parental lineages in three different species, including V. dahliae, Species A1 and Species D1 (Figure 1) [27]. Verticillium dahliae is the only known parent of V. longisporum, Species A1 and Species D1 have never been collected [27]. The holotype of V. longisporum represented by ex-holotype strain PD687 belongs to V. longisporum lineage A1/D3 that is one of the three lineages of V. longisporum, and V. longisporum is thus polyphyletic [27]. There is general agreement that fungal species should be monophyletic. However, we decided that V. longisporum should remain a polyphyletic species, because it seems impractical to name each lineage of V. longisporum. We currently know of three lineages of V. longisporum that represent three independent hybridization events, but there might be many more.
Little is known about fungal hybrids, but in plants, hybrids can evolve frequently over short periods in small areas [43].
We included the ex-holotype isolate V. longisporum strain PD687 in our studies, strain PD687 did not form any microsclerotia. But microsclerotia were present in the holotype that is a dried culture of strain PD687 (CBS 124.64) (Figure 3b). The microsclerotia in the holotype documented in Figure 9g were similar to the ones described by Stark [42] on page 500 for 'Typ X' as V. longisporum was referred to prior to its description. Thus, V. longisporum strain PD687 likely lost its ability to produce microsclerotia due to prolonged culturing.
Karapapa et al. [24] compared V. longisporum to the morphologically similar V. dahliae, and found that V. longisporum microsclerotia and conidia were longer than the ones in V. dahliae, and that V. longisporum conidiophores had fewer phialides in each whorl than V. dahliae.
We evaluated those characters and found that for the isolates used in this study grown on PDA, microsclerotia and conidia size might be useful to distinguish V. longisporum from V. dahliae. In Verticillium longisporum strain PD356, the majority of microsclerotia were elongate (Figure 9d), but rounded microsclerotia were still present (Figure 9e), and in some sectors of the colony, rounded microsclerotia were in the majority (Figure 9e). In V. longisporum strain PD348, there were roughly as many elongate microsclerotia as there were rounded microsclerotia. Verticillium dahliae microsclerotia were mostly rounded, but in some areas elongate microsclerotia were prevalent. The short brown-pigmented hyphae that were frequently attached to microsclerotia (Figure 9f) are possibly immature microsclerotia as illustrated by Isaac [22]. Similar structures were seen in this study in V. dahliae (Figure 6k). The third strain of V. longisporum investigated here, the ex-holotype strain PD687 did not form any microsclerotia. Conidia of V. longisporum were on average 8.563.5 mm (Figure 9c) and conidia of V. dahliae 6.563.0 mm (Figure 6h). However, conidia lengths might also at times be misleading, as the size ranges overlap, standard errors were 2.5 and 1.5 mm, respectively. We found that both V. longisporum and V. dahliae had similar numbers of phialides in each whorl, 2-4 for V. dahliae, and 2-5 for V. longisporum, this is unlike that proposed by Karapapa et al. [24] who reported 4-5 in V. dahliae and mostly 3 in V. longisporum. In our hands, Verticillium longisporum strain PD348 very frequently had 5 phialides per whorl.
Thus, a combination of conidia length and microsclerotia morphology might in many cases yield correct species identifications, but the two characters will also be misleading at times.
Distribution and host range. Currently known from Canada, Cuba, Japan, Slovenia and UK. Substrates include common hop, Irish potato, petunia and spinach.
Commentary. Verticillium nonalfalfae is morphologically indistinguishable from V. alfalfae, but the two species differ in pathogenicity. Verticillium nonalfalfae causes disease on a variety of different hosts, whereas V. alfalfae causes disease mainly on lucerne [44]. Other differences include vegetative compatibility groups [45], mating types (Figure 2), as well as the DNA characters listed in the species descriptions. Verticillium alfalfae and V. nonalfalfae were described as new species because no synonyms of the morphologically similar V. albo-atrum were available (www. indexfungorum.org, accessed on September 30, 2011).
Verticillium nonalfalfae and V. alfalfae have long been recognized as two genetically distinct groups referred to as non-lucerne and lucerne pathotype, respectively [46,47,48].
Distribution and host range. Currently known from the UK. Substrates include mushroom compost, Irish potato and soil.
Commentary. Verticillium nubilum was described by Pethybridge [50] from the surface of a potato tuber attacked by Phytophthora infestans. The protolog of V. nubilum contains descriptions of the V. nubilum morphology and a photograph of chlamydospores, but no reference is made to type material. We inquired at Kew (K), CABI Bioscience (IMI) and Dublin (DBN), none of which has any V. nubilum type material in its possession. Isaac [23] who studied V. nubilum in detail did not mention any herbarium material. We did not find any V. nubilum cultures by Pethybridge in any of the major culture collections (CBS, IMI, DSMZ, ATCC). Thus, in absence of any original fungal material, we designated the illustration from the V. nubilum protolog, Figure 5 on Plate 4 in Pethybridge [50], as the lectotype for V. nubilum.
Isaac [23] studied V. nubilum in detail and submitted several strains to CBS, of which we selected a dried culture of strain PD742 (CBS 457.51) as epitype. Our observations of V. nubilum agreed with the accounts by Pethybridge [50] and Isaac [23]. Pethybridge [50] noted that V. nubilum conidia were larger than those of V. albo-atrum. We found that V. nubilum condia were on average 7.562.5 mm (Figure 11c), the largest in Verticillium, with the exception of V. longisporum conidia that were on average 8.563.5 mm (Figure 9c). Differing from both Pethybridge [50] and Isaac [23], small numbers of brown-pigmented hyphae not directly associated with chlamydospores were sometimes present (Figure 11i), but these were lighter colored than the resting mycelium in other species (eg Figures 4g, 4h).
All the V. nubilum isolates that we examined formed very few conidia and conidiophores, which prevented us from conclusively assessing conidiophore morphology and dimensions. However, the few conidiophores and phialides that we saw were similar to other Verticillium species, in both morphology and dimensions (Figures 11c, 11d). Verticillium nubilum can be differentiated from other Verticillium species by the near exclusive formation of chlamydospores as resting structure (Figures 11f, 11g, 11h), in combination with the relatively large conidia (Figure 11c), but can be confused with Gibellulopsis nigrescens that forms distinctly smaller chlamydospores [15,23].
Distribution and host range. Currently known from Japan, Netherlands and UK. Substrates include carnation, garden tomato, Irish potato and larkspur. Commentary. Isaac [23] (p. 194) deposited V. tricorpus type material at IMI, K and CBS. We were only able to locate specimen IMI 51602, a dried V. tricorpus culture on PDA medium labeled 'isotype ?' (Figure 3c). Specimen IMI 51602 is likely derived from ex-type strain IMI 51602 deposited at CBS by I. Isaac as strain CBS 447.54. Thus, since the holotype appeared to be missing, we designated specimen IMI 51602, a likely isotype, as the lectotype of V. tricorpus (Art. 9.2, Art. 9.9, Art. 9.10). Specimen IMI 51602 did not display typical V. tricorpus morphology. Whereas verticillate conidiophores and microsclerotia were present (Figure 12g) in agreement with the description provided by Isaac [23], yellowpigmented hyphae, chlamydospores and resting mycelium were absent. However, according to the ICBN, lectotypes have to be chosen from among isotypes if they exist (Art. 9.10). Verticillium tricorpus specimen IMI 51602 is a likely isotype and was thus designated as lectotype. Upon initial culturing, Verticillium tricorpus colonies on agar medium are yellow to orange (Figures 12a, 12b) due to the presence of yellow-pigmented hyphae (Figure 12h). Resting mycelium, chlamydospores and microsclerotia are also formed simultaneously (Figures 12d, 12e, 12f, 12g). The yellow to orange coloration is typically less intense after prolonged culturing, or if obscured by resting structures. Verticillium tricorpus is morphologically indistinguishable from V. isaacii and V. klebahnii. All three species are characterized by the formation of resting mycelium (Figures 7f, 8e, 12d), chlamydospores (Figures 7g, 8f, 12e) and microsclerotia (Figures 7h, 8h, 12f, 12g), as well as yellowpigmented hyphae (Figure 7i, 8h, 12h) that confer agar cultures yellow to orange coloration (Figure 7b, 8b, 12b).
Distribution and host range. Currently only known from Japan. Substrates include lettuce and tenweeks stock.

Discussion
We have generated a solid taxonomic framework for Verticillium that recognizes ten species, five of which are new to science. Our results show that resting structure morphology, traditionally the most important morphological character to differentiate Verticillium species still plays a part in species identification, but the nearcomplete reliance on resting structure morphology to identify Verticillium species will have to be abandoned.
As other recent studies of fungal diversity [54,55,56,57,58,59,60], our approach combined phylogenetic analyses, literature research and morphological comparisons, and established that each Verticillium species, except the hybrid V. longisporum, corresponded to a single group in the phylogenetic tree. We included ex-type strains that are derived from herbarium type material to which fungal names are permanently linked according to the International Code of Botanical Nomenclature (ICBN). All species-level phylogenetic groups contained a single ex-type strain that thus conferred a species name to all current and future group members guaranteeing taxonomic stability.
This study recognized all previously known species of Verticillium [15]. These were V. albo-atrum [21], V. dahliae [35], V. longisporum [24], V. nubilum [50] and V. tricorpus [23]. In order to stabilize the application of names, we selected several new types. For V. alboatrum and V. nubilum, we designated illustrations as lectotypes since no herbarium material was available, and for V. tricorpus an isotype was designated as lectotype. For V. dahliae and V. albo-atrum, epitypes were selected based on our morphological comparisons, and a V. nubilum epitype was chosen among strains deposited by Isaac [20], who studied V. nubilum in detail.
The sister species Verticillium alfalfae and V. nonalfalfae (Figure 1) were previously referred to as the respective lucerne and nonlucerne pathotypes of 'V. albo-atrum' [46], and have long been recognized as two genetically distinct groups [47,48]. The two species are morphologically indistinguishable, but differ in pathogenicity. Verticillium nonalfalfae causes disease on a variety of hosts whereas V. alfalfae causes disease on lucerne [44]. Other differences include vegetative compatibility groups [45], mating types ( Figure 2), as well as the DNA characters listed in the species descriptions. Molecular data have previously been included in species descriptions [55,58,59,61,62]. We did not detect any genetic variation within V. alfalfae and V. nonalfalfae ( Figure 1). However, variation within V. nonalfalfae has been demonstrated using AFLP markers and a proteomics approach [63,64,65,66].
Verticillium alfalfae and V. nonalfalfae are related to V. dahliae and V. longisporum (Figure 1), but differ morphologically by the formation of resting mycelium. However, resting mycelium is also present in the distantly related V. albo-atrum with which V. alfalfae and V. nonalfalfae have frequently been confused [15].
The remaining three new species proposed here, V. isaacii, V. klebahnii and V. zaregamsianum are related to V. tricorpus (Figure 1). Verticillium isaacii, V. klebahnii and V. tricorpus are morphologically indistinguishable, they are characterized by the formation of resting mycelium (Figures 7f, 8e, 12d), chlamydospores (Figures 7g, 8f, 12e) and microsclerotia (Figures 7h, 8h, 12f), as well as the presence of yellow-pigmented hyphae (Figure 7i, 8h, 12h), providing the colonies on agar medium with a yellow or orange coloration (Figure 7b, 8b, 12b). The three species are a monophyletic group (Figure 1), and they could have been considered as three different lineages within just one species, V. tricorpus. However, compared with other Verticillium species, V. tricorpus, including what are now V. isaacii and V. klebahnii, was known to be very diverse, both in terms of ITS sequence data [48] and vegetative compatibility groups [72]. There is evidence for differences in pathogenicity. Verticillium tricorpus was only pathogenic on tomato [23], V. klebahnii was pathogenic on lettuce [51], and V. isaacii was not pathogenic on either lettuce or artichoke [41]. Further research is needed to determine the host ranges of these species.
Verticillium zaregamsianum, the third new species related to V. tricorpus (Figure 1), is morphologically distinct from all other Verticillium species, V. zaregamsianum forms predominantly microsclerotia (Figure 13i), as well as yellow-pigmented hyphae (Figures 13k, 13l). Verticillium zaregamsianum is a pathogen of lettuce in Japan [73]. Table 2 provides an overview of the taxonomic changes made in this paper and relates the new taxonomic system to previously described species.

Phylogenetic relationships of Verticillium species
In agreement with previous studies [15,74], we identified two major groups in Verticillium that we named Clades Flavexudans and Flaxnonexudans, respectively ( Figure 1). Clade Flavexudans comprised all species that produced yellow-pigmented hyphae that were absent in all members of Clade Flavnonexudans. Whereas Clade Flavexudans was well supported by the phylogenetic analyses, Clade Flavnonexudans, in particular the monophyly of V. nubilum with the remaining members of Clade Flavnonexudans, only received support in the parsimony analyses ( Figure 1). More research is needed to conclusively determine the phylogenetic placement of V. nubilum within Verticillium.
The phylogenetic relationships within the major clades were well resolved (Figure 1). Within Clade Flavexudans, the branching order of V. albo-atrum, V. isaacii, V. klebahnii, V. tricorpus and V. zaregamsianum had maximal support in all analyses. The topology of Clade Flavnonexudans was also well resolved, except for the placement of Species A1, an ancestor of the diploid hybrid V. longisporum. Species A1 that is unknown and has never been collected [27], is basal to the clade of V. alfalfae, V. dahliae, V. nonalfalfae as well as Species D1, another unknown species and second ancestor of V. longisporum [27], but only supported by the Bayesian analyses ( Figure 1). Inderbitzin et al. [27] studied the evolutionary history of V. longisporum in detail, they found that V. longisporum evolved at least three different times from four different parental lineages representing three different species. The results of Inderbitzin et al. [27] differ from the current study with regard to the placement of Species A1 that formed a clade with Species D1 and V. dahliae, whereas in this study, Species A1 was a sister group to the clade of V. alfalfae, V. dahliae, V. nonalfalfae and Species D1. The topological divergence involving Species A1 might be due to differences in taxon sampling and the use of an additional locus for phylogenetic analyses in Inderbitzin et al. [27].

Hosts and geographic distribution of Verticillium species
The isolates used in this study represent only a small fraction of the vast literature on Verticillium [1] and therefore do not paint a complete picture on geographic distribution and host range. However, the data provided here and in Inderbitzin et al. [27] are associated with correctly identified isolates and constitute an initial approximation of the distributions and host associations of Verticillium. For V. dahliae, V. longisporum and V. nubilum distribution and host range data are in general agreement with the literature [1,23,50,75]. Verticillium dahliae is known from four continents and fourteen host families, and is by far the most widespread Verticillium species. This contrasts with V. nubilum that is only known from Irish V. albo-atrum is more closely related to V. tricorpus than to V. dahliae, whereas V. alfalfae and V. nonalfalfae are closely related to V. dahliae ( Figure 1). b V. zaregamsianum was referred to as V. tricorpus at least once [73], but differs morphologically. doi:10.1371/journal.pone.0028341.t002 potato in the UK, and with V. longisporum that occurs in Europe, Japan and North America but is restricted mainly to hosts in the Brassicaceae. More work is needed to expand our knowledge on the distributions and host ranges of the remaining species, including the newly described V. alfalfae, V. isaacii, V. klebahnii, V. nonalfalfae and V. zaregamsianum, as well as V. albo-atrum and V. tricorpus that are now more narrowly defined.

Identification of Verticillium species
Correct and consistent identification is crucial for effective and efficient disease control [76], but we found Verticillium species may frequently have been misidentified. Based on DNA sequencing and phylogenetic analyses, we determined that at least 34 of the 293 isolates used in this study and the study by Inderbitzin et al. [27], were not correctly identified ( Table 3). Given that the majority of Verticillium strains were from Verticillium research labs, the error rate among non-specialists is likely to be higher.
Verticillium is difficult to separate from similar genera as it lacks morphological characters that are unique. The most conspicuous characters of Verticillium, the conidiophores bearing whorls of conidiogenous cells, as well as the resting structures, are also present in other genera including Gibellulopsis and Musicillium. This problem is illustrated by the fact that 26 of the 34 misidentified isolates belonged to genera other than Verticillium (Table 3).
Within Verticillium, resting structure morphology, conidia size, conidiophore size and pigmentation, the number of phialides per whorl, and the formation of yellow-pigmented hyphae has been used to differentiate species [23,24,35,50]. It is known that resting structure morphology may vary depending on culture medium [50,77] and other environmental conditions [78], and that yellowpigmented hyphae may be lost after prolonged culturing [23]. We did not investigate the influence of environmental conditions on Verticillium morphology in detail, but found that differences in resting structures between V. albo-atrum, V. alfalfae and V. nonalfalfae were more readily observed on WA-p and PLYA media than on PDA medium. However, the yellow coloration conferred to the agar medium by species in the Clade Flavexudans (Figure 1) was most prominent on PDA. Thus, based on our results, we recommend the combined use of PDA and WA-p for species identification.
In Figure 14 we provide a key to Verticillium species based on morphological characters. However, given the morphological variability of Verticillium species as discussed above, the key is more intended as an overview of Verticillium morphology than as an authoritative means for species identification. All results obtained using the key should be confirmed by DNA sequencing and phylogenetic analyses with ex-type isolates.

Conclusions
The new taxonomic system presented here is based on a multifaceted approach that included phylogenetic and morphological investigations, herbarium and literature research, and allows for a more reliable and consistent identification of Verticillium species. We envision that over time, this taxonomic system will lead to a significant improvement of our knowledge of Verticillium biology. Potential practical applications are many, and may include more efficient and effective disease management strategies and quarantine regulations.

Future Research
Future research will focus on the determination of host ranges of some of the new species of Verticillium as well as V. albo-atrum. Also, inclusion of more isolates from non-agricultural systems in studies of Verticillium diversity would be desirable.

Materials and Methods
Taxon selection, origin of fungal strains and DNA sequences retrieved from GenBank Taxa were selected to cover the known diversity of Verticillium [46], and included 74 strains representing V. tricorpus, V. nubilum, V. albo-atrum, V. longisporum and V. dahliae as well as the outgroup Gibellulopsis nigrescens based on results from Zare et al. [15]. We previously clarified the phylogenetic relationship of V. dahliae and V. longisporum [27], and for V. dahliae and V. longisporum only included six taxa representing the main lineages of the two species. The isolates were obtained from a variety of different sources (Table S1), and initially identified based on morphology. Common names used for hosts were obtained from www.ITIS.gov accessed on June 6, 2010.
For the following 13 isolates DNA sequence data by Inderbitzin et al. [27] was retrieved from GenBank. V. dahliae strains PD322 (HQ206718, HQ206921, HQ414624, HQ414719, HQ414909) PD-327 (HQ206723, HQ206925, HQ414628, HQ414723, HQ414913), Table 3. Names of misidentified isolates received are given in top row, approximate correct names based on DNA sequencing and comparison to GenBank are in left column, numerals refer to numbers of isolates in each category. Species recognition, description and naming Species were defined as terminal or subterminal clades inferred from multigene phylogenetic analyses in accordance with the Genealogical Concordance Phylogenetic Species Recognition approach outlined by Taylor et al. [31] and named by the inclusion of ex-type strains. Except for the diploid hybrid V. longisporum, each species-level clade contained a single ex-type strain. New species were described for all species-level clades for which no existing names were available. Existing, readily available names include synonyms that are listed in Index Fungorum (www. indexfungorum.org). We were unable to search for additional synonyms among the 266 described Verticillium species listed in Index Fungorum (accessed September 30, 2011).
Morphological descriptions were based on cultures grown on PDA, WA-p and PLYA media. Microscopy was performed using a Leica DM5000 B microscope (Leica Microsystems CMS GmBH, Wetzlar, Germany), with bright field (BF), differential interference contrast (DIC) and phase contrast (PC) illumination of specimens mounted in water. Photographs were taken with a Leica DFC310 FX camera, using Leica Application Suite Version 3.6.0 software. Culture photographs were generated with a desktop document scanner (DS). The terminology used in the species diagnoses follows Kirk et al. [36]. For conidia dimensions, standard deviations are given. Nucleotide substitutions in the species diagnoses included all derived substitutions shared by all members of a species, except the substitutions that were in alignment regions of low complexity (single or multi-nucleotide repeats) or near gaps in regions of ambiguous alignment.

Nomenclature
The electronic version of this document in itself does not represent a published work according to ICBN, and hence the new names contained in the electronic version are not effectively published under that Code from the electronic edition alone. Therefore, a separate edition of this document was produced by a method that assures numerous identical printed copies, and those copies were simultaneously distributed (on the publication date noted on the first page of this article) for the purpose of providing a public and permanent scientific record, in accordance with Article 29 of the Code. Copies of the print-only edition of this article were distributed on the publication date to botanical or generally accessible libraries of the following institutions, BPI, CBS, CUP, DAOM, HMAS, IMI, IRAN, NY, SFSU, TNS, UBC, and UC. The separate print-only edition is available on request from PLoS (Public Library of Science) by sending a request to PLoS ONE, Public Library of Science, 1160 Battery Street, Koshland Building East, Suite 100, San Francisco, CA 94111, USA along with a check for $10 (to cover printing and postage) payable to ''Public Library of Science''. This article is digitally archived in PubMed Central and LOCKSS.
DNA extraction, PCR amplification for direct sequencing and DNA sequencing conditions DNA was extracted according to Inderbitzin et al. [27]. For extraction of DNA from the V. dahliae type material, the same protocol as for extractions from mycelium recovered from agar plates was used.

Loci used for phylogenetic analyses and primer design
Five loci were used in this study, including actin (ACT), elongation factor 1-alpha (EF), glyceraldehyde-3-phosphate dehydrogenase (GPD), tryptophan synthase (TS) and the ribosomal internal transcribed spacer region ITS. Primers used to PCR amplify and sequence the ITS region were ITS1-F [81], ITS4 and ITS5 [82]. The TS region of Verticillium albo-atrum was at times PCR amplified and sequenced with primer pair VTs5f (59-ACC TAT GTC ACT GCC GGC T-39) and VTs4r (59-CAA TGA AGC CGT TGA CGC C-39). For more details on TS as well as the remaining loci, PCR conditions and DNA sequencing, see Inderbitzin et al. [27].

Phylogenetic analyses
Besides the single-locus ACT, EF, GPD, TS and ITS datasets, a combined, four-locus dataset comprised of concatenated ACT, EF, GPD and TS datasets was analyzed.
Most parsimonious trees were inferred using 30 random addition replicates. Otherwise, default settings were used, including treating insertion/deletion gaps as missing data.
Bootstrap support values were based on 500 replicates. Maximum likelihood analyses were done using default settings and 30 random addition replicates, bootstrap supports were based on 415 replicates. Bayesian analyses were performed with default settings, running four chains over 10 million generations and sampling each 100 th tree. The first 1000 of the 10,000 saved trees were omitted and the consensus tree was based on the remaining 9,000 trees. Maximum likelihood and Bayesian analyses implemented an optimal model of DNA sequence evolution determined using Modeltest 3.7 [86]. All analyses were run with a single representative of each haplotype. Figure S1 Phylogenetic tree of Verticillium based on the ITS dataset comprising 74 taxa and 514 characters. Shown is the single most parsimonious tree, 94 steps in length. Isolates are represented by a strain identifier; species are delimited by a vertical bar followed by a name. Branches with 100% bootstrap support are in bold, other support values above 70% are given by the branches. (TIF) Figure S2 Phylogenetic tree of Verticillium based on the ACT dataset comprising 77 taxa and 638 characters. Shown is one of the nine equally parsimonious trees, 427 steps in length. Isolates are represented by a strain identifier; species are delimited by a vertical bar followed by a name. Branches with 100% bootstrap support are in bold, other support values above 70% are given by the branches. (TIF) Figure S3 Phylogenetic tree of Verticillium based on the EF dataset comprising 77 taxa and 614 characters. Shown is one of the 12 equally parsimonious trees, 599 steps in length. Isolates are represented by a strain identifier; species are delimited by a vertical bar followed by a name. Branches with 100% bootstrap support are in bold, other support values above 70% are given by the branches. (TIF) Figure S4 Phylogenetic tree of Verticillium based on the GPD dataset comprising 77 taxa and 781 characters. Shown is one of the 2 equally parsimonious trees, 430 steps in length. Isolates are represented by a strain identifier; species are delimited by a vertical bar followed by a name. Branches with 100% bootstrap support are in bold, other support values above 70% are given by the branches. (TIF) Figure S5 Phylogenetic tree of Verticillium based on the TS dataset comprising 77 taxa and 625 characters. Shown is one of the 396 equally parsimonious trees, 565 steps in length. Isolates are represented by a strain identifier; species are delimited by a vertical bar followed by a name. Branches with 100% bootstrap support are in bold, other support values above 70% are given by the branches. (TIF) Table S1

Supporting Information
(DOC)