Introduction

The Phytopathological Society of Japan (2000) recorded four powdery mildew fungi on Cucumis sativus in Japan: Erysiphe polygoni, Oidiopsis sicula (anamorph of Leveillula taurica; Saito and Kurata 1975), Oidium sp. of polygoni-type (Oidium subgenus Pseudoidium; Sato et al. 1996), and Sphaerotheca cucurbitae (=Podosphaera xanthii; anamorph: Oidium subgenus Fibroidium). Because the anamorph of E. polygoni belongs to Oidium subgenus Pseudoidium, the powdery mildews of C. sativus may be reduced to three. Because the occurrence of Oidiopsis sicula and Oidium subgenus Pseudoidium has been reported only once in Japan, the major causal agent of the cucumber powdery mildew seems to be S. cucurbitae. Based on morphological observations of the conidial stage and on molecular phylogenetic analyses, the genus Sphaerotheca was merged into the genus Podosphaera (Braun and Takamatsu 2000), and the name of the cucumber mildew was revised to Podosphaera xanthii (Braun et al. 2001). We therefore named the Podosphaera species on cucumber in the present study as Podosphaera xanthii and its anamorph as Fibroidium.

In 2002, three powdery mildews with conidial stages of different morphologies were found on cucumber plants in a greenhouse in Hiratsuka-shi, Kanagawa Prefecture, Japan. Two of the fungi were identified as Oidium subgenus Fibroidium and Oidium subgenus Pseudoidium. The third mildew with catenate conidia and lacking fibrosin bodies belonged to Oidium subgenus Reticuloidium, anamorph of the genus Golovinomyces. Golovinomyces orontii is a common powdery mildew of cucurbits worldwide, including Europe, North America, South America, Africa, and West and Central Asia (Amano 1986). In East Asia, although E. cucurbitacearum (=G. orontii) has been reported to occur in China, there is no record of this fungus in Korea and Japan (Braun 1987; Shin 2000; Phytopathological Society of Japan 2000). Therefore, we characterized the fungus through morphological observations and molecular phylogenetic analyses.

Materials and methods

Fungal isolation

Leaf disks with a diameter of 9 mm were cut from cotyledons of C. sativus (cv. Tokiwa Hikari-3P) with a cork borer. One hundred leaf disks were put on a wet filter paper in a Petri dish (150 mm in diameter) and exposed for 3 h at two locations in a greenhouse with diseased cucumber plants. Then, the leaf disks were moved to small Petri dishes (60 mm in diameter; a single leaf/dish) and incubated for 6 days at 20°C under 2,000–2,500 lux with a photoperiod of 12/12 h (day/night) in an incubation room. A single spore was picked up with an endodontic file from a colony that developed on the leaf disk and placed on a healthy cucumber cotyledon, then incubated under the same conditions. Powdery mildew of Lamium amplexicaule in the same greenhouse, was also isolated by the same procedure. Cucumber seedlings (cultivar Tokiwa Hikari-3P) at the 2–3 leaf stage were sprayed with conidia of the isolates every 2 weeks and kept in a plastic box (20 × 20 × 30 cm) until the pathogenicity test.

Morphological observations

Hyphae, conidiophores and conidia were stripped off the surface of a detached leaf with clear adhesive tape, mounted on a microscope glass slide with the fungal mycelium uppermost and examined in water. To observe conidial germ tubes, we inoculated the inner cell layer of onion scales using the method of Hirata (1942, 1955).

Pathogenicity test

An isolate of Fibroidium B-7SF and two isolates of Reticuloidium No. 2–7 (from cucumber) and B-7H2 (from Lamium amplexicaule) were used to test pathogenicity. Potted cucumber plants (cvs. Sharp 1, Haru-no Megumi and Tokiwa Hikari-3P) were grown in Kureha horticultural compost (Kureha Chemical, Tokyo, Japan) at 25°C under 3,000–5,000 lux with a photoperiod of 12/12 h (day/night) in a phytotron for 2–3 weeks. These plants were then sprayed with a conidial suspension [ca. 105 conidia/ml, amended with 1/5,000 (v/v) Tween 20] of an isolate. The inoculated plants were put into a plastic box and incubated for 21 days at 20°C under 2,000–2,500 lux at the photoperiod of 12/12 h (day/night). The percentage of infected area of each of the powdery-mildewed cucumber leaves was determined by the following scale: 0 = no colony, 0.5 = percentage of area infection ≤10%, 1 = >10–20%, 2 = >20–40%, 3 = >40–60%, 4 = >60–80%, 5 = >80%. Disease severity was calculated as [(5A + 4B + 3C + 2D + E + 0.5F)/5G] × 100, where A, B, C, D, E and F are the number of leaves corresponding to the scale at 5, 4, 3, 2, 1 and 0.5, respectively, and G is the total number of leaves assessed.

Molecular phylogenetic analyses

Whole-cell DNA was extracted from mycelia by the chelex method (Walsh et al. 1991) as described in Hirata and Takamatsu (1996). The rDNA ITS region and the 5′ end of the 28S rDNA, including the D1 and D2 regions, were separately amplified two times by a polymerase chain reaction using nested primer combinations, and then sequenced directly as described in Matsuda and Takamatsu (2003).

The sequences were initially aligned using the Clustal_X program (Thompson et al. 1997) and further analyzed to improve the alignment with a word processing program with colour-coded nucleotides and deposited in TreeBASE (http://www.treebase.org/) with accession number S2265. Phylogenetic trees were obtained from the data by the maximum-parsimony method using the heuristic search option in the program PAUP* 4.0b8 (Swofford 2001). This search was repeated 100 times with different random starting points, using the stepwise addition option to increase the likelihood of finding the most parsimonious tree. Transversions and transitions were treated as equal weight. All sites were treated as unordered, with gaps treated as missing data. The branch-swapping algorithm was the tree bisection and reconstruction method, the MulTrees option was in effect, and zero-length branches were collapsed. The strength of the internal branches from the resulting trees was tested by bootstrap analysis (Felsenstein 1985) using 1,000 replications.

Results

Morphological observation

To investigate the powdery mildew species occurring in a greenhouse of cucumber of Hiratsuka-shi, Kanagawa Prefecture, we examined all 106 powdery mildew colonies that appeared on 52 naturally infected leaves of 80 cucumber plants (cv. Sharp-1) in the greenhouse in June 2002. As a result, 83, 22 and 1 colonies were identified as Oidium subgenera Reticuloidium, Fibroidium and Pseudoidium, respectively, based on Braun et al. (2002). The morphology of Reticuloidium was as follows: white mycelium on leaves, amphigenous, powdery-like. Leaf surface with colony slightly swelled out (Fig. 1a). Conidiophores erect, foot-cells slightly curved at the base, conidia produced in chains following 2–4 mother cells, semicylindrical or doliform, germ tubes arising from the middle portion of the conidia, tips unlobed, slightly enlarged with nipple-shaped appressoria (cichoracearum-type) (Table 1; Fig. 1c–e). The morphology of single-spore isolate No. 2–7 was consistent with these characteristics.

Fig. 1
figure 1

Symptoms of powdery mildew on cucumber and Lamium amplexicaule and causal fungus Oidium subgenus Reticuloidium. Symptoms of cucumber (a) and L. amplexicaule (b). Conidiophore with catenate conidia (c), conidia (d), and conidial germ tube (e) of Oidium subgenus Reticuloidium (bar 20 μm)

Full size image
Table 1 Morphology of the Reticuloidium species found in a greenhouse on Cucumis sativus and Lamium amplexicaule
Full size table

Powdery mildews also occurred on Lamium amplexicaule (Fig. 1b) and Oxalis carniculata, common weeds both inside and outside of the greenhouse where cucumber powdery mildew was found. Powdery mildew on Oxalis carniculata belonged to subgenus Pseudoidium based on the characteristics of having noncatenate conidia and polygoni-type conidial germ tubes and is regarded as an anamorph of Erysiphe russellii (Nomura 1997). The fungus on Lamium amplexicaule belonged to subgenus Reticuloidium, from which a single-spore isolate, B-7H2, was established (Table 1).

Pathogenicity test

Reticuloidium as well as Fibroidium are pathogenic on the three cucumber varieties used in this study. However, the disease severities with Reticuloidium isolates No. 2–7 (disease severity = 16–45%) and B-7H2 (22–42%) were lower than that of Fibroidium isolate B-7SF (64–83%) (Table 2). The latent period of Reticuloidium was about 7 days longer than that of Fibroidium . In this study, we estimated disease severity 21 days after inoculation. The different disease severities between Fibroidium and Reticuloidium may be derived from the different latent period of these fungi. Reticuloidium isolate No. 2–7 from cucumber was pathogenic on Lamium amplexicaule.

Table 2 Disease severity on three cucumber varieties after inoculation with conidia of three powdery mildew isolates
Full size table

ITS phylogeny

The rDNA ITS region and the D1/D2 domains of the 28S rDNA were sequenced for each isolate from C. sativus (cv. No. 1–16) and Lamium amplexicaule (cv. B-7H2). The sequences were deposited in the DDBJ DNA database with accession numbers AB427187 and AB427188, respectively. These two sequences were identical in both ITS and 28S rDNA regions. The two ITS sequences obtained in this study were aligned with 70 ITS sequences retrieved from the DDBJ database. The ITS data set consists of 72 taxa, of which 70 are members of the genus Golovinomyces (including Oidium subgenus Reticuloidium). Of the 70 Golovinomyces sequences, five sequences (shown by solid circle in Fig. 2) are from cucurbit powdery mildew. The ITS sequences of the isolates from C. sativus and Lamium amplexicaule were identical to the sequences of G. orontii from Arabidopsis thaliana (AF031282) and Veronica arvensis (AB077652) and differed by one base from the sequences of G. orontii from Cucurbita pepo (AF229017 and AB077670), Cucurbita sp. (AB077696), Nicotiana tabacum (AB022413); G. cichoracearum from Sonchus sp. (AB077669), Cichorium intybus (AB077695, AB077666 and AF011294), Lactuca serriola (AB077688), and Mycelis muralis (AB077661); and Oidium sp. on Dahlia pinnata (AB077677).

Fig. 2
figure 2

Phylogenetic analysis of the nucleotide sequences of the ITS region including the 5.8S rDNA for 70 sequences from the genus Golovinomycetes (including Oidium subgenus Reticuoidium) and two outgroup sequences. The tree is a phylogram of the maximum likelihood tree among the 640 most parsimonious trees with 303 steps, which was obtained by a heuristic search employing the random stepwise addition option in the program PAUP*. Gaps were treated as missing data. Horizontal branch lengths are proportional to the number of nucleotide substitutions that were inferred to have occurred along a particular branch of the tree. Percentage bootstrap support (1,000 replications; ≥70%) are shown on the branches. Solid circle before a taxon label indicates a sequence from a cucurbit powdery mildew

Full size image

The data set consisted of 534 characters, from which 39 sites were removed to circumvent ambiguous alignments. Of the 495 remaining characters, 138 sites were variable, and 116 sites were phylogenetically informative for parsimony analyses. Two sequences from Arthrocladiella mougeotii were used as the outgroup based on Mori et al. (2000). The parsimony analysis using PAUP* generated 640 equally parsimonious trees of 303 steps (consistency index [CI] = 0.6469, retention index [RI] = 0.8981, rescaled consistency index [RC] = 0.5809). One of the 640 trees with the highest log-likelihood value is shown in Fig. 2, which shows nine clades differentiated within the Golovinomyces taxa investigated in this study. Each clade was strongly supported by bootstrap values of 94% or more. Of the nine clades, four clades, I, V, VI and VII, comprise isolates from each of the host tribes of the Asteraceae: clade I from tribe Cardueae, clade V from Anthemideae, clade VI from Astereae, and clade VII from Heliantheae. Clade II comprises isolates from Lamiaceae and Ranunculaceae. Clade III comprises a single isolate from Lapsana (Asteraceae), clade IV comprises isolates from Lycopersicon and Solanum (Solanaceae), and clade VIII comprises isolates from Phlox (Polemoniaceae) and Physalis (Solanaceae). Clade IX is large and comprises isolates from the tribe Lactuceae of the Asteraceae, and many other plant families such as Plantaginaceae, Solanaceae, Brassicaceae and Cucurbitaceae. Sequences from the isolates No. 1–16 (from C. sativus) and B-7H2 (from Lamium amplexicaule) belonged to the clade IX. All the nine clades were commonly appeared in all 640 parsimonious trees, although the branching order of the clades differed among the trees.

28S phylogeny

The two 28S rDNA sequences obtained in this study were aligned with 49 28S rDNA sequences retrieved from the DDBJ database. The 28S rDNA data set consists of 51 sequences, of which 50 are from members of the genus Golovinomyces. These 50 sequences include 11 Golovinomyces species from 44 different plant species that represent 13 plant families, of which three sequences shown by solid circle in Fig. 3 are from cucurbit powdery mildews. The two 28S rDNA sequences determined in this study differed one base from G. cichoracearum on Lactuca serriola (AB077687) and Picris hietacioides (AB077686), and two bases from G. cichoracearum on Sonchus sp. (AB077668), Cichorium intybus (AB077667), Mycelis muralis (AB077660) and Lactuca serriola (AB077651), and G. orontii on Veronica arvensis (AB077651). There was no sequence identical to these two 28S rDNA sequences.

Fig. 3
figure 3

Phylogenetic analysis of the divergent domains D1 and D2 sequences of the 28S rDNA for 50 sequences from the genus Golovinomycetes (including Oidium subgenus Reticuloidium) and one outgroup sequence. The tree is one of 48,109 most parsimonious trees with 121 steps and was obtained by a heuristic search using PAUP*. Gaps were treated as missing data. Horizontal branch lengths are proportional to the number of nucleotide substitutions that were inferred to have occurred along a particular branch of the tree. Percentage bootstrap support (1,000 replications; ≥70%) is shown on the branches. Solid circle before a taxon label indicates a sequence from a cucurbit powdery mildew

Full size image

The data set consists of 718 characters, all of which were used for tree construction because all their sites were aligned unambiguously. Of the 718 characters, 65 sites were variable, and 45 were phylogenetically informative for parsimony analyses. A sequence from Arthrocladiella mougeotii was used as an outgroup based on Mori et al. (2000). The parsimony analysis using PAUP* generated 48,109 equally parsimonious trees of 121 steps (CI = 0.6694, RI = 0.8964, RC = 0.6001). Because the tree topologies were almost identical among the 48,109 trees except for branching order of terminal branches, one of the trees is shown in Fig. 3. All trees support the respective clades recognized in the ITS analysis, although clades III and IV were not included in the 28S tree due to the lack of DNA sequence data. The two sequences from isolates No. 1–16 (from C. sativus) and B-7H2 (from Lamium amplexicaule) belonged to the large terminal clade IX.

Discussion

Four powdery mildew species, i.e., Erysiphe cichoracearum (=E. cucurbitacearum, E. polyphaga; present name: G. orontii; anamorph: Oidium subgenus Reticuloidium), E. communis (anamorph: Oidium subgenus Pseudoidium), Sphaerotheca fuliginea (present name: Podosphaera xanthii; anamorph: Oidium subgenus Fibroidium) and Leveillula taurica (anamorph: Oidiopsis siculae) are listed as causal agents of cucumber powdery mildew in the world (Amano 1986). Three powdery mildews, i.e., Oidium subgenus Pseudoidium, Podosphaera xanthii (=S. fuliginea, S. fusca, S. cucurbitae) and Leveillula taurica, have been reported to occur in Japan (Phytopathological Society of Japan 2000; Saito and Kurata 1975; Sato et al. 1996). Of these, Podosphaera xanthii is the main causal agent and the other two fungi are rare. The fungus from Hiratsuka-shi morphologically resembled Podosphaera xanthii in producing catenate conidia, but differed from Oidium subgenus Pseudoidium and Leveillula taurica, which have noncatenate conidia. However, the present fungus was also distinguished from Podosphaera xanthii in not having distinct fibrosin bodies in its conidia. In addition, the conidial germ tube of Podosphaera xanthii is the fuliginea-type, whereas that of the present fungus is the cichoracearum-type. These characteristics reveal that the present fungus differs from Podosphaera xanthii and belongs to Oidium subgenus Reticuloidium. This is the first record of cucumber powdery mildew caused by Reticuloidium in Japan. Because the teleomorph of Reticuloidium belongs to the genus Golovinomyces (Braun et al. 2002), the present fungus may be an anamorphic stage of G. orontii, a common powdery mildew of cucumber in the world except East Asia (Sitterly 1978; Braun 1987). Results of the molecular analysis showing that the fungus belongs to the clade of G. orontii support the assumption.

Reticuloidium mildew on cucumber was classified as E. cichoracearum in the monograph of Salmon (1900). E. cichoracearum sensu Salmon was reported from a wide range of plant families including Asteraceae, but considered as a complex of numerous races with narrow host ranges. Hammarlund (1945) inoculated 99 plant species from a wide range of families with an ‘E. cichoracearum’ isolate from cucumber and reported that the fungus infected 62 plant species of 11 families such as Asteraceae, Crassulaceae, Cucurbitaceae, and Solanaceae. Thus, he concluded that this fungus has a wide host range, although it resembled E. cichoracearum in morphology, and called the fungus ‘E. polyphaga’. Blumer (1952, 1967) performed independent inoculation tests and confirmed the presence of such a fungus with a wide host range. Braun (1980) found that the foot cells of the conidiophore of ‘E. polyphaga’ often curved at the base and proposed the name ‘E. orontii’ (the oldest available name for this fungus). He confined the name ‘E. cichoracearum’ to fungi on Asteraceae. The genus Erysiphe was then split into three genera based on the morphology of the anamorph and on molecular analyses, and E. orontii was revised as G. orontii (Braun 1999; Heluta 1988).

We used five cucurbitaceous mildew sequences in the present phylogenetic analysis. There is no sequence identical to the ITS sequence of the present Japanese isolates. AF229017 sequenced in France differs by only one base from the Japanese isolate at the 125th site from 5′-end of the ITS1 region. Recently, a new occurrence of Reticuloidium has been reported on several cultivated plants such as Helianthus tuberosus, Viola tricolor and Zinnia elegans (Hoshi et al. 2007; Tanda and Suga 2002). It is likely that these Reticuloidium mildews were newly imported from abroad.

We found Reticuloidium on Lamium amplexicaule in the same greenhouse in which cucumber Reticuloidium was found. Both the ITS and the 28S rDNA sequences of this isolate are identical to those of the cucumber mildew, and the isolate is pathogenic on cucumber. This result suggests that Lamium amplexicaule provided the primary inoculum for the cucumber mildew in this greenhouse. Neoerysiphe galeopsidis has been the only powdery mildew species to infect Lamium species (Amano 1986; Braun 1987). Blumer (1952) reported that Reticuloidium infects Lamium galeobdolon, and Garibaldi et al. (2007) reported the occurrence of G. orontii on Lamium galeobdolon. Uchida et al. (2002) found Oidium subgenus Fibroidium on Lamium amplexicaule near the cucumber greenhouses in Hiratsuka-shi. This Fibroidium was pathogenic to both cucumber and Lamium amplexicaule, suggesting that common weeds like Lamium amplexicaule and Lamium purpureum serve as hosts to produce primary inoculum of Fibroidium as well as Reticuloidium to cause cucumber mildew.

In Europe and other regions of the world, G. orontii is an important causal agent of cucurbit mildew with Podosphaera xanthii (Bardin et al. 1999; Jahn et al. 2002; Mohamed et al. 1995; Sitterly 1978; Sz Nagy 1970, 1972, 1976; Vakalounakis and Klironomou 1995; Vakalounakis et al. 1994). Based on Sz Nagy (1976), ‘E. cichoracearum’ (=Reticuloidium) can germinate in wider temperature ranges than ‘Sphaerotheca fuliginea’ (=Podosphaera xanthii; Fibroidium), and conidia of the former species are able to germinate at lower humidity. Hoshi et al. (2007) reported that Reticuloidium occurs earlier than Fibroidium in Japan. Thus, the two mildew fungi differ in some physiological and ecological characteristics. Although evidence on cucumber mildew caused by Fibroidium has been accumulating for many years in Japan (Abiko 1978, 1982a, 1982b; Endo 1989; Hosoya et al. 1999, 2000; Kuzuya et al. 2003, 2006), there is no report of cucumber mildew caused by Reticuloidium. Further research on the physiology, epidemiology, chemical control of this fungus and resistant varieties of cucumber is needed.