Disentangling Peronospora on Papaver: Phylogenetics, Taxonomy, Nomenclature and Host Range of Downy Mildew of Opium Poppy (Papaver somniferum) and Related Species

Based on sequence data from ITS rDNA, cox1 and cox2, six Peronospora species are recognised as phylogenetically distinct on various Papaver species. The host ranges of the four already described species P. arborescens, P. argemones, P. cristata and P. meconopsidis are clarified. Based on sequence data and morphology, two new species, P. apula and P. somniferi, are described from Papaver apulum and P. somniferum, respectively. The second Peronospora species parasitizing Papaver somniferum, that was only recently recorded as Peronospora cristata from Tasmania, is shown to represent a distinct taxon, P. meconopsidis, originally described from Meconopsis cambrica. It is shown that P. meconopsidis on Papaver somniferum is also present and widespread in Europe and Asia, but has been overlooked due to confusion with P. somniferi and due to less prominent, localized disease symptoms. Oospores are reported for the first time for P. meconopsidis from Asian collections on Papaver somniferum. Morphological descriptions, illustrations and a key are provided for all described Peronospora species on Papaver. cox1 and cox2 sequence data are confirmed as equally good barcoding loci for reliable Peronospora species identification, whereas ITS rDNA does sometimes not resolve species boundaries. Molecular phylogenetic data reveal high host specificity of Peronospora on Papaver, which has the important phytopathological implication that wild Papaver spp. cannot play any role as primary inoculum source for downy mildew epidemics in cultivated opium poppy crops.


Introduction
The genus Papaver (Papaveraceae) comprises about 80 annual, biennial and perennial herbs distributed in Central and southwestern Asia, Central and Southern Europe and North Africa [1]. The most well-known species is opium or oilseed poppy (P. somniferum), an ancient crop and medicinal plant cultivated for its edible seed as well as for the production of opium, the source for important pharmaceutical drugs including morphine, thebaine, codeine, papaverine, and noscapine [2].
One of the most important diseases of P. somniferum is downy mildew caused by Peronospora spp., which is responsible for substantial crop losses world-wide (e.g. [2][3][4][5][6]). Several Papaver species have been reported to be hosts of Peronospora [7], and four species have been described from various Papaver species [8]. However, their taxonomic status, synonymy as well as host range have been much disputed, leading to substantial confusion in the literature about the number of species present on Papaver and their correct naming.
Peronospora cristata, the second species described from Papaver, was reported to infect P. argemone, P. hybridum, P. rhoeas and P. somniferum [5,8,13,14,17], but also Meconopsis betonicifolia [24] and M. cambrica [14]. Remarkably, P. cristata has only been reported on host species that are also recorded hosts of P. arborescens, raising the question about correct species identification and whether one or two species are involved. In the description of P. cristata, Tranzschel [25] reported verrucose oospores which are remarkably distinct from the smooth oospores of P. arborescens, but following Reid [14] who did not mention the oospore characteristics this important character has been largely ignored, and accessions from various Papaver and Meconopsis species were attributed to P. cristata primarily on conidial sizes that are distinctly larger than those of P. arborescens.
From M. cambrica, a third species, P. meconopsidis, has been described [26], which, however, has not received much attention in the plant pathology literature and has been commonly synonymised with P. cristata due to conidia of similar size, or ignored, following Reid [14] who did not even mention P. meconopsidis. The fourth species, P. argemones, was described from Papaver argemone and has conidial sizes in the range of P. cristata and P. meconopsidis. Therefore, Reid [14] synonymised it with P. cristata, ignoring the fact that the oospores of P. argemones were described as smooth, in contrast to the verrucose oospores of P. cristata. Following the approach of Reid [14], two Peronospora species, P. arborescens and P. cristata, have been accepted on Papaver until recently, which were primarily distinguished on their different conidial sizes, and more recently, on distinct ITS sequences [5,20].
For risk assessment of infections of the economically important opium poppy (Papaver somniferum) crop, it is crucial to clarify the host ranges of the pathogens involved. Furthermore, since high numbers of wild Papaver spp. are coincident with the phenology of the cultivated opium poppy, if there is a host overlap, those Papaver spp. could act as alternative hosts and be potential sources of primary inoculum for the disease contributing to disseminating the pathogen within opium poppy crops. In the current study, we report the results of extensive molecular and morphological investigations on Peronospora accessions from various Papaver species and from Meconopsis cambrica to clarify nomenclature, species boundaries and host ranges of the species involved.

Morphological Analysis
Conidiophores and conidia were removed from the underneath of infected leaves, transferred to a drop of anhydrous lactic acid on a slide, carefully torn apart using forceps and needles, shortly heated using an alcohol burner and covered with a cover slip. For oogonia, host tissue was soaked in 2% KOH on a slide, carefully torn apart with forceps and needles and covered with a cover slip. Slides were examined and photographed using a Zeiss Axio Imager.A1 (Zeiss, Jena, Germany) microscope equipped with a Zeiss AxioCam ICc3 digital camera. Measurements are reported as maxima and minima in parentheses and the mean plus and minus the standard deviation of a number of measurements given in parentheses.

Sample Sources
Information on the samples used for sequencing and phylogenetic analyses is given in Table 1. Details on the specimens used for morphological analysis are given in the description of the species. The herbarium acronyms are given according to Thiers [27].

DNA Extraction, PCR and Sequencing
For DNA extraction, infected dry host tissue was placed in 2 ml reaction tubes together with six sterile 2 mm glass beads and ground in a Retsch 200 mixer mill for 10 min. Alternatively, conidiophores were scraped off the leaf surface of the hosts, put in 1.5 ml reaction tubes and ground with sterile quartz sand and a conical micropestle. DNA was extracted using the modified CTAB protocol described in Riethmüller et al. [28] or using the Macherey-Nagel NucleoSpin Plant II extraction kit according to the manufacturer's instructions.

Phylogenetic Analysis
All alignments were produced with Muscle version 3.6 [36]. For evaluation of species status, a combined analysis of cox1 and cox2 was performed, adding a representative selection of Peronospora species according to the phylogenetic tree of Göker et al. [37], with two Pseudoperonospora species as outgroup to root the tree according to the phylogenies of Göker et al. [38]. Because the ITS-LSU rDNA did not show much phylogenetic information to separate closely related species, it was not included in the phylogenetic analyses but the sequences were deposited in GenBank (Table 1); in addition, for complete reference the GenBank accession numbers of ITS sequences of Peronospora accessions included in the present study that have already been deposited by the authors in the course of previous studies [4,39] are also listed in Table 1.
Maximum parsimony (MP) analysis was performed with PAUP* version 4.0 b10 [40], using 1000 replicates of heuristic search with random addition of sequences and subsequent TBR branch swapping (MULTREES option in effect, COLLAPSE = MAXBR-LEN, steepest descent option not in effect). All molecular characters were unordered and given equal weight; analyses were performed with gaps treated as missing data. Bootstrap analysis with 1000 replicates was performed in the same way, but using 5 rounds of random sequence addition and subsequent branch swapping during each bootstrap replicate.
For maximum likelihood (ML) and Bayesian analyses, the HKY substitution model [41] was selected for both cox1 and cox2 by Modeltest 3.6 [42] using the hierarchical likelihood ratio tests, with invariant sites and gamma distribution for the remaining sites (HKY+I+G). In the combined analyses, substitution parameters were estimated separately for each region. For ML analyses, 10 runs with 100 thorough bootstrap replicates each were computed with RAxML [43] as implemented in raxmlGUI 1.3 [44] using the GTRCAT substitution model. For Bayesian analyses using MrBayes version 3.1.2 [45], three parallel runs of four incrementally heated simultaneous Markov chains were performed over 1 million generations from which every 100th tree was sampled in each run, implementing the HKY+I+G substitution model.

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants, and hence the new names contained in the electronic publication of a PLOS ONE article are For institution codes of herbarium vouchers see Thiers [27]; asterisks (*) denote ITS sequences published in Landa et al. [4] and Montes-Borrego et al. [39]; all others were newly generated in the present study (formatted in bold). doi:10.1371/journal.pone.0096838.t001 effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies. In addition, new names contained in this work have been submitted to MycoBank from where they will be made available to the Global Names Index. The unique MycoBank number can be resolved and the associated information viewed through any standard web browser by appending the MycoBank number contained in this publication to the prefix http://www.mycobank. org/MB/. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.
Of the 1262 characters of the combined cox1 -cox2 alignment, 234 were parsimony informative (120 in cox1, 114 in cox2). MP analyses revealed 16 MP trees 1035 steps long, one of which was selected and presented in Figure 1, with MP and ML bootstrap support above 50% and posterior probabilities above 90% given at first, second and third position, respectively, above/below the branches. Topologies of the MP trees slightly differed in the phylogenetic positions of P. sordida, P. corydalis and P. chrysosplenii. The three Bayesian runs revealed almost identical posterior probabilities (PP) and were fully compatible with the MP strict consensus tree. The Peronospora-accessions from Papaver/Meconopsis were contained within three highly supported clades, one comprising Peronospora cristata from Papaver hybridum (clade 1 in Figure 1); a second clade containing P. apula from Papaver apulum, P. argemones from Papaver argemones and P. meconopsidis from Meconopsis cambrica, Papaver pavoninum and P. somniferum (clade 2 in Figure 1); and a third clade with P. arborescens from Papaver rhoeas, P. somniferi from Papaver somniferum, P. sp. 1 from Papaver dubium and P. sp. 2 from Papaver sp. (clade 3 in Figure 1). Whereas within the Peronospora arborescens clade the ITS did not resolve accessions from the different hosts, in the cox1 -cox2 trees the accessions from Papaver dubium, P. rhoeas and P. somniferum were placed in three distinct monophyletic clades, the former two with high and the latter with medium to high support ( Figure 1).

Taxonomy
As a result of the morphological and molecular phylogenetic investigations, six taxa are here recognised as occurring on Papaver, two of which are described as new. In addition, detailed descriptions of the other four already described species are provided. All type specimens cited were morphologically investigated in the present study.
Description. Infection commonly systemic, more rarely localized, when systemic whole plants or leaves stunted, strongly distorted, chlorotic, dwarfed, when localized producing polyangular to confluent lesions with distinct margins. Down hypophyllous, greyish, consisting of dense and felt-like conidiophores.  Phylogram showing phylogenetic relationships of Peronospora accessions from Papaver and Meconopsis. One of 16 most parsimonious trees 1035 steps long inferred from the combined cox1-cox2 sequence data matrix; parsimony and likelihood bootstrap support above 50% and posterior probabilities above 90% are given at first, second and third position, respectively, above/below the branches. The tree was rooted with two species of Pseudoperonospora according to Gö ker et al. [38]. doi:10.1371/journal.pone.0096838.g001  Comments. Peronospora arborescens appears to be confined to Papaver rhoeas, on which it is very common. Its disease symptoms are highly distinctive, infected plants being distorted and showing light yellowish green discolouration. The various accessions included in our molecular phylogenetic analyses covered most of Europe (Austria, Croatia, France, Germany, Hungary, Italy, Romania, Spain), and were molecularly highly homogeneous. Accessions from other Papaver species were placed in distinct clades and represent genetically distinct lineages. The accessions from Spain were partly sampled within cultivated opium poppy fields.   Comments. Peronospora argemones appears to be confined to Papaver argemone. Gä umann [13] separated P. argemones from P. arborescens based on larger conidia. Our measurements fit well those reported by Gä umann [13] (21.1 6 18.1 vs. 21 6 18.6). P. argemones is not closely related to P. arborescens but to P. meconopsidis which has larger conidia. Its closest relative, however, is P. apula, which differs by slightly smaller conidia and its host Papaver apulum. Comments. Peronospora cristata has been confirmed only from Papaver hybridum, with which it is apparently co-occurring. The type and authentic specimens of Tranzschel [23] from the Crimean Peninsula morphologically fully resemble the recent Spanish collections for which DNA data are available. The irregularly verrucose oospores are highly distinctive for the species, all other Peronospora species on Papaver having smooth oospores. In addition, it has the largest conidia of all Peronospora species on Papaver. Peronospora meconopsidis, which has conidia of similar size, differs by smooth oospores and different hosts (Papaver somniferum, Meconopsis cambrica).
Replaced synonym. Description. Infection localized, producing more or less distinct polyangular to confluent lesions. Down mostly hypophyllous, greyish, consisting of mostly scattered conidiophores, very rarely dense and felt-like. observations and upon examination of herbarium specimens. It has commonly been misidentified as Peronospora arborescens, which mainly differs by conspicuous disease symptoms relating to mostly systemic infection which has never been observed for P. meconopsidis. Based on similar conidial sizes, P. meconopsidis has been referred to as Peronospora cristata in recent literature, which goes back to Reid [14] who classified accessions from Meconopsis cambrica under that species. However, P. cristata is easily distinguishable by its verrucose oospores, which are unique in Peronospora on Papaver. The Australia records of Peronospora cristata from Papaver somniferum [5] therefore actually represent Peronospora meconopsidis, which is also corroborated by sequence data (Figure 1).
Oospores of Peronospora meconopsidis are reported here for the first time, and they have only been found in young infected plants of Papaver somniferum collected in Asia. No oospores were found in specimens of Papaver somniferum from Europe or from Meconopsis cambrica despite thorough investigations.
The species was first described as Peronospora gaeumannii by Mayor [46]. However, because this is a younger homonym of P. gaeumannii Mundk., Mayor [26] proposed the new name P. meconopsidis. Three authentic specimens mentioned in the original description of P. gaeumannii Mayor are present at NEU, of which the largest, best developed and preserved is here selected as lectotype. This folder also contains the original drawings and spore statistics published in Mayor [46]. According to the herbarium label, the lectotype specimen consists of several collections from the same place collected from 1945 to 1952, which were subsequently mixed and cannot be separated any more. The mean spore sizes recorded by Mayor [46] agree well with those of the current study (23.5 6 21.15 vs. 24.8 6 20.6 mm).
Molecular diagnosis. Peronospora somniferi differs from its closest phylogenetic neighbour, P. arborescens, by unique fixed alleles in two tree loci (cox1, cox2) based on alignments of the separate loci deposited in TreeBASE as study S15609: Comments. Peronospora somniferi appears to be confined to Papaver somniferum. It is closely related to P. arborescens which differs by smaller conidia and a different host, Papaver rhoeas. Peronospora somniferi cannot be reliably distinguished from P. arborescens by ITS data alone because there is only a single consistent nucleotide difference at the beginning of the ITS1, but cox1 and cox2 are distinctive and good barcode markers for the species (11 and 10 diagnostic substitutions, respectively). Peronospora somniferi has been reported as an economically important pathogen of Papaver somniferum throughout Europe [4]. Peronospora meconopsidis, which also occurs on Papaver somniferum, differs mainly by a non-systemic infection which is characterised by typical polyangular spots (see Figure 8).

Key to known Species of Peronospora on Papaver and Meconopsis
Note: Morphological identification requires well developed material as well as knowledge of the host. For conidial measurements, it is crucial that a sufficient number of mature conidia are included. Because some species are highly similar, they often cannot be unequivocally identified by morphology alone, and sequence data (cox1 or cox2) are essential in case of poorly developed specimens, if the host species is unknown or represents a host species from which no sequence data are available or from which several Peronospora species are known.

Molecular Phylogenetic Investigations
The current investigations clearly show that the biodiversity of Peronospora on Papaver is higher than previously perceived, which is in line with other investigations on Peronosporaceae (e.g. [37,[47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63]), demonstrating that high biodiversity is commonly the result of high host specificity. Peronospora from Papaver are distributed amongst three clades (Figure 1), of which P. cristata from clade 1 is phylogenetically isolated from the other Peronospora species from Papaver, but clades 2 and 3 appear to be closely related (Figure 1). Of special interest is clade 3, the species of which were formerly classified under Peronospora arborescens, and which is here referred to as P. arborescens sensu lato clade. Whereas the ITS data are highly similar within this clade and therefore do not allow for unequivocal distinction (data not shown), the cox1 and cox2 data are highly distinctive for the accessions from various hosts, in which each form genetically homogeneous lineages irrespective of the geographic origins. This is evident for the Peronospora accessions from Papaver rhoeas and P. somniferum which were sampled from various regions all over Europe, and which form distinct uniform genetic lineages. In addition, also differences in conidial sizes could be documented for both lineages. Therefore, the accessions from Papaver somniferum, previously classified under P. arborescens, are here described as a distinct species, P. somniferi. Apart from these two lineages, another two genetically distinct entities were present within the P. arborescens s. l. clade (Figure 1), Peronospora sp. 1 from Papaver dubium, and Peronospora sp. 2 from an unidentified Papaver species. However, because only few accessions were available for these, we currently refrain from describing them as new taxa. Considering the numerous additional Papaver species for which Peronospora have been recorded (see [4]) but for which no accessions were available for molecular phylogenetic investigations, additional genetically distinct entities may turn up in the future.

Molecular Barcoding
cox1, chosen as barcoding locus for higher animals and considered to be the primary barcoding marker for organisms unless shown to be unsuitable (http://www.barcodeoflife.org), has been demonstrated to be an appropriate barcoding locus for oomycetes [32], which is confirmed in the current study. However, cox2 shows similarly good resolution and consequently is an equally good barcoding marker; it even has some advantages over cox1, as it usually amplifies better especially in cases of low DNA quantity or old herbarium samples (as also shown in [64]), and thus cox2 sequences are available for many more species in Peronosporales. The current study provides additional data for equally good discriminative power of cox1 and cox2; e.g. Peronospora somniferi is distinct from Peronospora arborescens by shared 11 and 10 substitutions in cox1 and cox2, respectively, clearly showing a similar significant barcode gap between both species in both markers. Therefore, both cox1 and cox2 are considered good markers for reliable identification of Peronospora species on Papaver, and it is recommended to sequence both loci in studies of Peronosporales whenever possible to obtain representative robust data for molecular barcoding. On the other hand, the ITS region does not resolve closely related species like P. arborescens and P. somniferi, which has been observed also in other groups of Peronosporaceae (e.g. [47,54]).

Nomenclature of Peronospora on Papaver
In the recent literature, accessions infecting Papaver were mainly classified as Peronospora arborescens. However, as the molecular data clearly show, there are several species involved, six of which are recognized and treated in the current publication. However, nomenclature of these species is partly complex mainly due to misleading species concepts of the past, as shall be outlined below.
In a phytopathological perspective, an important result of recent molecular phylogenetic investigations was that downy mildew disease of the crop Papaver somniferum is caused by two distinct species, which were classified as Papaver arborescens and Papaver cristata [4,5]. However, the current detailed investigations show that this classification cannot be retained and has to be substantially modified.
Starting from Gä umann [13], Peronospora accessions from Papaver somniferum have been classified as Peronospora arborescens, which has been consistently followed. However, the current molecular phylogenetic investigations show that these accessions are genetically distinct from accessions from the type host, Papaver rhoeas, and there are also differences in conidial sizes. Consequently, the Peronospora accessions from Papaver somniferum belonging to the Peronospora arborescens sensu lato clade (clade 3; Figure 1) are here classified as a distinct species, P. somniferi.
Recently, the name Peronospora cristata was used for a second Peronospora species on Papaver somniferum that was first recorded from that host by Scott et al. [5] from Tasmania. This name was applied because their ITS sequences matched an English accession from Meconopsis cambrica that had previously been deposited in GenBank (DQ885375) under Peronospora cristata. This naming apparently goes back to Reid [14], who compared conidial measurements of Peronospora from Meconopsis cambrica with those from Papaver argemone and considered them to be conspecific due to similar size. In addition, because conidial sizes were also similar to those reported for Peronospora cristata from Papaver hybridum, he considered accessions from these three hosts to be conspecific, and thus classified them under P. cristata due to priority. However, he ignored that the oospores of Peronospora cristata are distinctly verrucose, being significantly different from the smooth oospores of Peronospora argemones. These substantial morphological differences are also mirrored by a distant phylogenetic position of P. cristata (Figure 1), confirming that P. cristata is a clearly distinct species confined to Papaver hybridum.
In addition, also the accessions from Papaver argemone and Meconopsis cambrica are phylogenetically distinct (Figure 1) and therefore have to be classified under P. argemones and P. meconopsidis, respectively. Consequently, as the Peronospora of M. cambrica and the second Peronospora species from P. somniferum are conspecific, the correct name to be applied for this species is P. meconopsidis.

Host-parasite Relationships
The phylogenetic relationships of Peronospora on Papaver and those of their hosts are only partly congruent. For instance, within Papaver sect. Argemonidium which form a closely related group [1], the Peronospora species from Papaver argemone and P. apulum (Peronospora argemones and P. apula) are closest relatives, whereas the Peronospora from the third species of the section, Papaver pavoninum, falls within P. meconopsidis which is sister group to the former two species (Figure 1). However, Peronospora cristata, parasitising the fourth species of sect. Argemonidium, Papaver hybridum, is phylogenetically and morphologically distinct from all other Peronospora species on Papaver.
Remarkably, Peronospora meconopsidis infects hosts from two genera, Meconopsis and Papaver. However, molecular phylogenetic investigations show that Meconopsis is polyphyletic and embedded within Papaver, M. cambrica being unrelated to the other Meconopsis species [1]. Whereas the two main hosts of Peronospora meconopsidis, M. cambrica and Papaver somniferum are contained within the same clade but not closely related, the third host species, P. pavoninum, is member of the phylogenetically isolated Papaver section Argemoni-dium [1]. This indicates that phylogenetic radiation of Peronospora on Papaver is rather effected by host jumps.
Finally, from a pytopathological point of view the results from this study showed that wild Papaver spp. cannot play any role as primary inoculum for downy mildew epidemics in cultivated opium poppy crops since there is high host specificity. This result was corroborated when different Peronospora specimens (P. somniferi, P. cristata and P. arborescens) were sampled from the same fields of cultivated Papaver somniferum where Papaver hybridum and Papaver rhoeas were growing as weeds. Consequently, to avoid development of downy mildew epidemics in cultivated opium poppy efforts should be placed in controlling airborne P. somniferi sporangia from diseased opium poppy plants as well as avoiding the use infected seed lots [22].