Introduction

Oomycetes cause commercially important downy mildew disease in numerous crop and ornamental plants worldwide. Recently, as the market size of ornamental plants grows larger and plants of new areas become available for trade, the risk of the introduction of new downy mildew diseases is increasing. Thus the importance of accurate diagnosis and identification of these pathogens is crucial. As a result of efforts into this direction over the past decade, several new downy mildew species on economically relevant ornamental plants have been described, e.g. Peronospora belbahrii on basil (Thines et al. 2009), P. apula and P. somniferi on opium poppy (Voglmayr et al. 2014), P. salviae-plebeiae and P. salviae-officinalis on sages (Choi et al. 2009), and Plasmopara destructor and Pl. velutina on impatiens (Görg et al. 2017).

Members of the genus Aquilegia (Ranunculaceae) are widely grown in public and private gardens throughout the world (Nold and Nelson-Nold 2003; Stace 2010). Downy mildew disease has been first found on Aquilegia plants throughout England and Wales in 2013, but before on Semiaquilegia in Korea (Denton et al. 2015) and China (Yu et al. 1998). While it seems possible that Aquilegia and Semiaquilegia downy mildew (ASDM) is caused by a previously undescribed species of Peronospora (Denton et al. 2015), a decision regarding the conspecificity or distinctiveness of the species could not be reached due to uncertainty regarding the relationships of the downy mildew pathogens from Aquilegia and Semiaquilegia, and their relationships with other pathogens, in particular of other members of the Ranunculales. In the phylogenetic analyses based on ITS and LSU rDNA sequences presented by Denton et al. (2015), it was revealed that the ASDM collections formed a monophyletic cluster, but further clustered with several different species of Peronospora parasitic on Euphorbiaceae, Papaveraceae, Ranunculaceae, and Saxifragaceae, with high support values (Denton et al. 2015). As the ITS region often exhibits insufficient variability for phylogenetic distinction in closely related species, e.g. in Peronospora (Choi et al. 2007; Voglmayr et al. 2014), Pseudoperonospora (Choi et al. 2005), and Phytophthora (Goodwin et al. 1999; Cooke et al. 2000; Jung and Burgess 2009), it was uncertain, if the pathogens from Aquilegia and Semiaquilegia should be considered conspecific or not. Therefore, in the current study phylogenetic analyses were conducted using four nuclear (ITS, LSU, heat shock protein 90 [hsp90], β-tubulin [β-tub]) and five mitochondrial loci (cytochrome c oxidase subunit II and I [cox2 and cox1], a spacer region between cox2 and cox1 genes [coxS], NADH dehydrogenase subunit I [nad1], ribosomal protein S10 and its flanking region [rps10]). In addition, detailed morphological comparisons were carried out in order to clarify the taxonomic identity of the downy mildew pathogens of Aquilegia and Semiaquilegia.

Materials and methods

Oomycete specimens

Five downy mildew specimens originating from Aquilegia and Semiaquilegia were analysed in this study. In addition, two Peronospora species parasitic on Euphorbiaceae, four on Papaveraceae, six on Ranunculaceae, and two on Saxifragaceae were included in the phylogenetic reconstructions. Pseudoperonospora cubensis was used as an outgroup. Information on the specimens sequenced in this study is shown in Table 1.

Table 1 Information on downy mildew specimens used in this study
Full size table

Morphology

Morphological characteristics of conidiophores, conidia, and oospores were investigated using dried herbarium specimens. A piece of infested leaf tissues was mounted on a drop of 70% lactic acid on a slide, gently warmed, covered with coverslips, and examined using an Olympus BX53 microscope (Olympus, Tokyo, Japan). Measurements were performed at 100–200× for conidiophores and at 400× for conidia, ultimate branchlets, and oospores, using the TCapture software (Tucsen Photonics, Fuzhou, China). The measurements were reported as follows; (minimum) – standard deviation towards the minimum – mean – standard deviation towards the maximum – (maximum) (n = 50). DIC micrographs were captured using an Olympus BX53 microscope equipped with a DigiRetina 16 M camera (Tucsen Photonics, Fuzhou, China) or using a Zeiss Imager M2 AX10 microscope with an AxioCam MRc5 camera (Carl Zeiss, Göttingen, Germany).

DNA extraction, PCR amplification, and sequencing

From herbarium specimens, 5–10 mg of infected leaf tissue was taken and disrupted in a mixer mill (MM2, Retsch, Germany), using two iron beads of 3 mm and five beads of 1 mm diameter per sample and shaking twice at maximum speed for 15 min. Genomic DNA was extracted using the BioSprint 96 DNA Plant Kit (Qiagen) on a KingFisher Flex (Thermo Scientific) robot. Two nuclear (ITS, hsp90) and five mitochondrial markers (cox2, cox1, coxS, nad1, rps10 and its flanking region) were amplified using oomycete-specific primers, as outlined previously by Choi et al. (2015b). In addition, two regions of LSU rDNA, D1–5 and D6–8, were amplified using two primer sets, LR0R (Vilgalys and Hester 1990) /LR6-O (Riethmüller et al. 2002) and LR6-OR2C (designed here; TAAGCTCGTCTGGCGATG)/LR9 (Hopple and Vilgalys 1999), respectively. In addition, as a forth nuclear gene β-tub was amplified using TUBUF2 (Kroon et al. 2004) and bTub1024R-O (Göker et al. 2007). In some cases, the cox2 gene and the spacer region had to be separately amplified with two primer sets, cox2-F (Hudspeth et al. 2000) & cox2-RC4 (Choi et al. 2015a) for the gene, and FM79 & FMPh-10b (Martin et al. 2004) for the spacer region. Amplicons were sequenced at the Biodiversity and Climate Research Centre (BiK-F) laboratory using primers identical to those used for amplifications.

Phylogenetic analysis

Sequences were edited using the DNAStar software package (DNAStar, Inc., Madison, Wis., USA), version 5.05. An alignment of each locus was performed using MAFFT 7 (Katoh and Standley 2013) employing the Q-INS-i algorithm (Katoh and Toh 2008). SequenceMatrix 1.7.8 (Vaidya et al. 2011) was used for concatenating individual gene sequences and for checking unusually similar or divergent sequences. After making sure that no conflicting support was present in individual loci, three concatenation alignments for nuclear (ITS, LSU, hsp9, β-tub), mitochondrial (cox2, cox1, coxS, nad1, rps10 and its flanking region), and all nine loci were constructed. Phylogenetic inference was done using two different methods, Maximum Likelihood (ML) and Minimum Evolution (ME). For ML analyses, 1000 rounds of random addition of sequences as well as 1000 fast bootstrap replicates were performed using RAxML 7.0.3 (Stamatakis 2006) as implemented in raxmlGUI 1.3 (Silvestro and Michalak 2012) using the GTRCAT variant. ME analysis was done using MEGA 7.0 (Kumar et al. 2016), with the default settings of the program, except for using the Tamura-Nei model instead of the maximum composite likelihood model.

Results

Morphological analysis

Based on morphology of observable structures the downy mildew pathogens on Aquilegia and Semiaquilegia could not be distinguished. However, they could be discriminated from all other related pathogens (Table 2). The dimensions of conidia and oospores were observed to be the most useful characters distinguishing between ASDM and other Peronospora species. The mean conidial length in ASDM was of 17.6 μm and, thus, smaller than in other Peronospora species (more than av. 20 μm) previously recorded on members of the Ranunculaceae (Gäumann 1923; Gustavsson 1959). Similarly, the diameter of oospores (av. 24.0 μm) was also smaller than any other report for a Peronospora species on this family (Gäumann 1923). In the morphological comparison to four phylogenetically closely related species, P. bulbocapni and P. meconopsidis from the Papaveraceae, as well as P. chrysosplenii and P. saxifragae from the Saxifragaceae, ASDM was distinguishable from the former two species again by smaller conidia (av. 17.6 μm in ASDM but av. 22 μm in P. bulbocapni (Beck 1886) and av. 24.8 μm in P. meconopsidis (Voglmayr et al. 2014). In addition, the diameter of its oospores (av. 24.0 μm) was also smaller (Beck 1886; Voglmayr et al. 2014). Similarly, the dimensions of conidia were smaller in ASDM than in P. chrysosplenii (17–28 × 15–20 μm in (Kochman and Majewski 1970); 22–30 × 16–23 μm in (Stanyavichene 1984)) and P. saxifragae (av. 24.5 × 20.8 μm in (Gäumann 1923); (20–)22–30 × 17–25 μm in (Kochman and Majewski 1970)). Moreover, ASDM exhibited smaller oospores than those of P. chrysosplenii (Kochman and Majewski 1970; Stanyavichene 1984). Two Peronospora species parasitic to the Euphorbiaceae differed from ASDM by larger conidia of (17–)21–23(−27) μm in P. cyparissiae and 19–24 μm in P. esulae (Stanyavichene 1984). In addition, P. cyparissiae exhibited larger oogonia (37–46 μm) and oospores (30–35 μm) (Stanyavichene 1984).

Table 2 Morphological comparison between Peronospora sp. and other phylogenetically or morphologically close species
Full size table

Multi-locus phylogeny

Trees based on the concatenated alignment of four nuclear loci compared to that of five mitochondrial loci each showed no strong conflicting support with a phylogeny based on the concatenated alignment of all nine loci. Thus, only the phylogeny based on concatenation of all loci is shown in Fig. 1. The final concatenated alignment had 8122 total characters, including 1464 variable characters, 887 of which were parsimony-informative. Since the dataset revealed no significant conflicts in the topologies derived from ML and ME analyses, only the tree from the ML inference is shown in Fig. 1, with bootstrap support from both analyses.

Fig. 1
figure 1

Maximum Likelihood tree of a concatenated alignment of four nuclear (ITS, LSU, heat shock protein 90 [hsp90], β-tubulin [β-tub]) and five mitochondrial loci (cytochrome c oxidase subunit II and I [cox2 and cox1], NADH dehydrogenase subunit I [nad1], ribosomal protein S10, and its flanking region [rps10]). Bootstrapping support values of Maximum Likelihood and Minimum Evolution methods higher than 60% are given above or below the branches. The scale bar equals the number of nucleotide substitutions per site. Specimens originating from different host families are marked with grey (Ranunculaceae), pink (Euphorbiaceae), yellow (Papaveraceae), and green (Saxifragaceae) bars

Full size image

In the multi-gene tree, all ASDM sequences were identical in sequence and formed a monophyletic group with maximum support, demonstrating the genetic homogeneity of the pathogen, despite originating from two different host genera, Aquilegia and Semiaquilegia, and having been collected from two distant countries, U.K. and Korea. The grouping of ASDM with two Peronospora species from the Saxifragaceae (P. chrysosplenii and P. saxifragae) and two species from the Papaveraceae (P. bulbocapni and P. meconopsidis), was well resolved in the present study. The further grouping with two Peronospora species from the Euphorbiaceae (P. cyparissiae and P. esulae) was weakly supported in ML and not supported in ME analyses. Interestingly, except for ASDM, all other species of Peronospora parasitic on Ranunculaceae formed another large clade with maximum support that was further split into several different lineages, corresponding to a particular host genus or species.

Taxonomy

Peronospora aquilegiicola Thines, G. Denton & Y.J. Choi, sp. nov. [MB 827342] (Fig. 2).

Fig. 2
figure 2

Peronospora aquilegiicola sp. nov. parasitic on Aquilegia vulgaris. a–c Conidiophores (bar = 100 μm); d & e Ultimate branchlets (bar = 10 μm); f & g Conidia (bar = 10 μm); h & i Resting organs inside a leaf (h) and a rhizome (i) (bar = 20 μm). Source: GLM-F116093

Full size image

Etym.: ‘aquilegiicola‘ refers to the host plant, aquilegia.

Infected Aquilegia spp. leaf tissue first yellowish or chlorotic, later often darkening to become purplish, vein-delimited if not fully systemic, resulting in a polyangular, mosaic appearance on leaves. Down present on the lower leaf surface of leaves, dense, felt-like, purplish or beige due to the colour of conidia. At systematic infection lesions more uniform, rims of infected leaves often curling outwards. Flowers are water-soaked, distorted, decolourised to brown. Overall, flower development is negatively affected. Mycelium in leaves, shoots and rhizomes. Conidiophores protruding from stomata on the underside of infected leaf tissue, erect, hyaline, monopodially branched up to 7 orders, (163–)189–235–278(−389) μm long, (4.1–)4.6–7.6–10.6(−20.1) μm wide, sometimes swollen at the base, up to 12.5 μm; trunk (46–)59–90–121(−188) μm. Ultimate branchlets slightly curved to substraight, mostly in pairs (90%), but rarely in single (10%), branched at an angle of 80–110°, with different lengths, (5.4–)6.59–9.08–11.57(−16.8) μm for the longer ones, (2.5–)4.91–7.23–9.55(−12.32) μm for the shorter ones, a ratio of longer to the shorter ultimate branchlets of (0.82–)1.08–1.30–1.52(−2.01) μm, (0.80–)1.37–1.75–2.13(−2.6) μm wide at the base, tip obtuse or pointed. Conidia broadly ellipsoidal, (13.7–)16.3–17.6–18.9(−20.6) μm long, (11.3–)13.6–14.9–16.1(−17.3) wide, with a length to width ratio of (1.11–)1.15–1.22–1.29(−1.38), directly germinating with a germ tube. Resting organs present in the necrotic leaf tissue but also in rhizome lesions with brownish dots. Oogonia light brown, (24.0–)28.0–32.0–36.0(−37.05) μm in diameter. Oospores (19.0–)21.0–24.0–27.0(−32.5) μm in diameter, wall 2.0–4.5 μm (mean = 3.0 μm) thick, surface smooth.

Typus: UK; Cymru, Swansea, Killay, Cyne Vallay Cottages, Touchwood, in a public garden, on living leaves of Aquilegia vulgaris affected by downy mildew disease, April 2015, Carrie Thomas (GLM-F116093– holotypus).

Habitat: On living leaves of Aquilegia alpina, A. buergeriana, A. flabellata, A. viridiflora, A. vulgaris and Semiaquilegia adoxoides (Ranunculaceae).

Distribution: South Korea, the UK, and China (?)

Additional specimens examined: see Table 1.

Note: This pathogen seems to have been first found on Semiaquilegia adoxoides in China (Yu et al. 1998) as “Peronospora ficariae”. In a simple description of the pathogen on S. adoxoides, the measurements of conidiophores (196–511 μm in length, 5.7–8.5 μm in width) and oospores (22.3–25 μm in diameter) are in line with measurements for P. aquilegiicola performed in this study. However, larger conidia of 11.4–42.6 × 12.8–34 μm have been reported than the present study, requiring further investigation, before a conspecificity can be ascertained.

Discussion

Downy mildew of Semiaquilegia was found in natural host populations in South Korea and presumably in China, which precedes the year 2013, when Aquilegia downy mildew emerged in the UK. However, because Semiaquilegia plants are found only for a short time in spring, and the infested plants are often almost indistinguishable from the healthy ones, the disease may be easily overlooked. Thus, it is most likely that P. aquilegiicola is an indigenous species in northeast Asia. The genetic similarity of P. aquilegiicola collections from Korea and the UK suggests that this pathogen has been quite recently introduced to the UK, presumably by trade with infected plants or seeds from East Asia, the origin of Semiaquilegia. There it might have jumped onto non-indigenous Aquilegia plants or the indigenous species, A. buergeriana, which are grown for ornamental purpose. The two genera, Aquilegia and Semiaquilegia, are morphologically similar (Tucker and Hodges 2005; Damerval and Nadot 2007) and phylogenetically close (Wang and Chen 2007), and the host plant S. adoxoides was often classified under Aquilegia.

The market size for Aquilegia is increasing due to growing demand for this ornamental plant in gardens and parks. Peronospora aquilegiicola should be seen as a high-risk pathogen due to its aggressiveness, its production of oospores that outlast unfavourable conditions, and the production of air-borne conidia which are produced abundantly. In the past, some of the most devastating oomycete diseases have emerged by a transfer of infected plants and seeds. For example, basil downy mildew caused by Peronospora belbahrii, resulted in significant losses in global sweet basil production (Belbahri et al. 2005; Thines et al. 2009; Wyenandt et al. 2015). When the basil downy mildew was first discovered, the causal agent was attributed to Peronospora lamii, which is widely distributed throughout the world together with its host, Lamium purpureum. Inevitably, few countries have taken quarantine measures against infested basils or seeds, resulting in its spread over most of the world (Thines and Choi 2016). At the moment, as Aquilegia downy mildew seems to be still restricted to the UK, P. aquilegiicola should be considered to be a quarantine organism, to block further spread of this pathogen. If measures are taken quickly, it might be possible to avoid the advent of another global epidemic and render P. aquilegiicola the first downy mildew to be successfully halted from spreading.