Structure and temporal shifts in virulence of Pseudoperonospora cubensis populations in the Czech Republic

Authors


E-mail: ales.lebeda@upol.cz

Abstract

The structure and temporal dynamics of the virulence of Pseudoperonospora cubensis (causal agent of cucurbit downy mildew) were studied in pathogen populations in the Czech Republic from 2001 to 2010. A total of 398 P. cubensis isolates collected from Cucumis (Cm.sativus, Cm. melo, Cucurbita (Cr.maxima, Cr. pepo, Cr. moschata and Citrullus lanatus were analysed for variation in virulence (pathotypes). Virulence was evaluated on a differential set of 12 genotypes of cucurbitaceous plants. All isolates of P. cubensis were characterized by their level of virulence (classified according the number of virulence factors, VF; low VF = 1–4, medium VF = 5–8, high VF = 9–12): high (75%), medium (24%) and low (1%). The structure and dynamics of virulence in the pathogen populations were expressed by pathotypes using tetrad numerical codes and a total of 67 different pathotypes of P. cubensis were determined. The most susceptible group of differentials was Cucumis spp., while the lowest frequency of virulence was recorded on Cr. pepo ssp. pepo, Ci. lanatus and Luffa cylindrica. A high proportion (c. 90%) of isolates was able to infect cucurbit species Benincasa hispida and Lagenaria siceraria, which are not commonly cultivated in the Czech Republic or elsewhere in central Europe. In the recent pathogen populations (2008–2010) there was prevailing frequency (70–100%) of isolates with high numbers (9–12) of virulence factors. ‘Super pathotype’ 15.15.15 was often observed in the study within the pathogen populations and was one of the four most frequently recorded pathotypes. Pseudoperonospora cubensis populations shifted to a higher virulence over time. From 2009 the pathogen population changed dramatically and new pathotypes appeared able to establish natural and serious infection of Cucurbita spp. and Ci. lanatus, which was not observed in 2001–2008. Generally, virulence structure and dynamics of P. cubensis populations are extremely variable in the Czech Republic.

Introduction

Cucurbit downy mildew, caused by Pseudoperonospora cubensis, is the most important foliar disease of cucurbit crops worldwide (Lebeda & Cohen, 2011). The disease is most often recorded in tropical, subtropical and warm temperate areas of the world. However, in the 1980s it was also observed in cooler areas, such as Scandinavia (Tahvonen, 1985; Forsberg, 1986). Currently, the pathogen is very destructive in all humid areas of the world, as well as some temperate areas (Lebeda & Cohen, 2011). Pseudoperonospora cubensis can overwinter in areas with mild winter temperatures or in protected cultivation, as active mycelium in either cultivated or wild species of cucurbits (Lebeda & Cohen, 2011). In the USA, the pathogen is thought to be reintroduced annually in areas with hard winters via long-distance transport of inoculum from warmer areas (Holmes et al., 2004; Ojiambo & Holmes, 2011). Survival by oospores is the subject of ongoing debate; there are reports of oospore formation from Russia, China, Japan, India and Italy (for details see Cohen & Rubin, 2012; Lebeda & Cohen, 2011), and recently, infectious oospores were produced experimentally under laboratory conditions (Cohen & Rubin, 2012). Because of the polycyclic nature of P. cubensis, the disease spreads rapidly both in open fields and protected environments (Lebeda & Cohen, 2011). Infection is caused by zoospores, which requires free leaf moisture, with sporulation occurring at relative humidity >90% (Cohen & Rotem, 1971; Cohen, 1977). The disease is often devastating and requires frequent fungicide application for control (Lebeda & Cohen, 2012). However, resistance has rendered several fungicides ineffective. Pseudoperonospora cubensis was the first oomycete to develop resistance to metalaxyl and reduced sensitivity to mancozeb, and there are now a broad spectrum of fungicides that are ineffective against P. cubensis (Lebeda & Cohen, 2011, 2012).

Pseudoperonospora cubensis is considered a ‘risky’ pathogen (sensuMcDonald & Linde, 2002) with a high evolutionary potential and broad adaptability (Lebeda & Urban, 2007). The spectrum of known host species is enlarging (Lebeda & Widrlechner, 2003; Lebeda & Cohen, 2011) and there have been reports of new hosts of P. cubensis in many countries (e.g. Cohen et al., 2003; Colucci et al., 2006; Choi & Shin, 2008; Ko et al., 2008; Baiswar et al., 2010; Salati et al., 2010a; Pavelkováet al., 2011). Palti (1974) reported that differences in host species’ responses to the pathogen were probably the result of different physiological races and/or pathotypes in various countries. A detailed survey of virulence in P. cubensis demonstrated the existence of a large number of pathotypes (c. 100) and potential races around the world (Lebeda et al., 2006).

Thomas et al. (1987) proposed the first differential set for virulence determination based on three host genera (Cucumis [Cm.], Cucurbita [Cr.], Citrullus) and distinguished five different pathotypes of P. cubensis. Lebeda & Křístková (1993) noted that host–pathogen specificity between Cr. pepo and P. cubensis is probably controlled by race-specific factors. In contrast, no virulence variation in P. cubensis (originating from cucumber) was detected on Cm. sativus and wild Cucumis species (Lebeda, 1992a,b; Lebeda & Prášil, 1994). In 2003 a new pathotype of P. cubensis was described in Israel (Cohen et al., 2003). A differential set was developed for P. cubensis pathotype determination based on 12 cucurbitaceous differential genotypes (Lebeda & Widrlechner, 2003). Pathotypes are determined by the interactions observed on each of the 12 hosts and assigned unique tetrad codes to describe the interaction (Lebeda & Widrlechner, 2003). This system allowed characterization of the virulence variability of P. cubensis at the individual and population levels (Lebeda et al., 2006; Lebeda & Urban, 2007). Recently, studies on the genetic variation of P. cubensis (Sarris et al., 2009; Mitchell et al., 2011) and its populations (Runge et al., 2011; Quesada-Ocampo et al., 2012) were published. However, detailed studies focusing on virulence variation of this pathogen in areas of its natural distribution are lacking. This paper is a contribution to this knowledge gap. The objectives of this study were: (i) to measure variation in virulence (at the level of a pathotype) of P. cubensis in the Czech Republic in the period 2001–2010; and (ii) to illustrate the dynamics of pathogen populations from the viewpoint of temporal and spatial changes in virulence structure.

Materials and methods

Study area and survey period

The distribution, occurrence and damage caused by P. cubensis on cucurbitaceous vegetables were evaluated in 12 of 14 regions and 37 of 77 districts of the Czech Republic. Three main surveys were conducted per year (late July to late August) in two main areas of the Czech Republic (Moravia, central and southern parts; and Bohemia, eastern and central parts; Fig. 1) from 2001 to 2010 (Lebeda et al., 2011). The main cucurbitaceous vegetable production areas were visited (e.g. south and central Moravia, east Bohemia and Polabí) but some marginal areas for cucurbit cultivation (e.g. areas of Jeseníky, Beskydy, Českomoravská Vrchovina and Podkrkonoší) were also surveyed (Fig. 1). The occurrence of P. cubensis was monitored during the main harvest period in hobby gardens, small private fields and large production fields. Disease prevalence and severity were evaluated annually at c. 80–100 locations (Lebeda et al., 2011).

Figure 1.

 Main collection areas of cucurbit downy mildew (Pseudoperonospora cubensis) isolates in the Czech Republic and origin of the isolates collected on new host plants (Cucumis melo, Cucurbita spp., Citrullus lanatus).

Source of Pseudoperonospora cubensis isolates

Extensive monitoring carried out from 2001 to 2010 confirmed that P. cubensis is widespread and occurs annually across the whole studied area of the Czech Republic. The plant material infected by P. cubensis was most frequently collected from Cm. sativus (2001–2010) (Table 1). Infection of Cm. melo was recorded only at two locations in 3 years (2003: Oplocany, Olomouc region; 2004 and 2009: Olomouc-Holice, Olomouc region; Lebeda et al., 2011). From 2009, rare infection of Cr. moschata was recorded at two locations in the Czech Republic (Nový Jičín-Kojetín, Silesia region of Moravia, 2009–2011; Olomouc-Holice, central Moravia, 2010) (Lebeda et al., 2011; Pavelkováet al., 2011). Other cucurbitaceous plants were without natural infection of P. cubensis until 2010. From 2010 P. cubensis infection was also recorded on Cr. pepo, Cr. maxima, Cr. ficifolia, Citrullus lanatus and Lagenaria siceraria, mostly in south Moravia (Fig. 1). Pseudoperonospora cubensis isolates originating from these crops were also subjected to pathotyping.

Table 1. Survey of screened Pseudoperonospora cubensis isolates in the period 2001–2010
YearNo. of screened isolatesOriginal host plant
TotalBohemiaMoraviaCSCMeCPCMCMoCL
  1. CS: Cucumis sativus; CMe: Cucumis melo; CP: Cucurbita pepo; CM: Cucurbita maxima; CMo: Cucurbita moschata; CL: Citrullus lanatus.

20014219234200000
20025411435400000
20035614425600000
2004404364000000
2005257182500000
20062912172900000
20073912273900000
20083214183200000
20094416284210010
20103713242804311
Total39812227638714321

Isolation and maintenance of P. cubensis

The collected infected leaf samples were incubated on wet filter paper in plastic pots (110 × 85 × 45 mm) (Lebeda & Urban, 2010). Inoculum was prepared by cutting out one lesion from each infected leaf and agitating in distilled water to detach the spores. This water was then sprayed over the abaxial surface of a leaf of the highly susceptible cucumber cv. Marketer 430 and placed in a Petri dish (100 mm in diameter) on wet filter paper. Inoculated leaves were incubated in a growth chamber under standard conditions as previously described (Lebeda & Urban, 2010). The pathogen usually produced spore-bearing sporangiophores 7–8 days after inoculation.

Cultures of P. cubensis on cucumber leaf discs were stored in Petri dishes at −80°C. The spores were viable for about 6 months, after which it was necessary to inoculate a new set of plants to maintain pathogen viability. Some of the P. cubensis isolates used in this research are deposited in the Czech National Collection of Microorganisms at Palacký University in Olomouc.

Plant material

Cucumis sativus cv. Marketer 430 was used for multiplication of the pathogen isolates. The virulence of isolates was established on a differential set of 12 cucurbit taxa (Lebeda & Widrlechner, 2003) (Table 2). Plants were grown in the greenhouse at 25/15°C day/night temperature, under natural lighting with daily watering and the addition of fertilizer (Kristalon Start, applied by watering) once per week. Plants were not treated with any pesticides. The plants (leaves) were used for testing when 5–8 weeks old (three to six true leaves present).

Table 2. Differential set of Cucurbitaceae taxa for the determination of pathogenic variability in Pseudoperonospora cubensis (Lebeda & Widrlechner, 2003)
No.TaxonAccession numberCultivar nameCountry of origin
DonorEVIGEZa
  1. aEVIGEZ – Czech genebank number.

1 Cucumis sativus   H39-0121 Marketer 430USA
2 Cucumis melo ssp. melo PI 292008 H40-1117 Ananas Yoqne′amIsrael
3 Cucumis melo ssp. agrestis var. conomon CUM 238/1974 H40-0625 Baj-GuaJapan
4 Cucumis melo ssp. agrestis var. acidulus PI 200819 H40-0611  Myanmar
5 Cucurbita pepo ssp. pepo PI 171622 H42-0117 DolmalikTurkey
6 Cucurbita pepo ssp. texana PI 614687 H42-0130  USA
7 Cucurbita fraterna PI 532355 H42-0136  Mexico
8 Cucurbita maxima   H42-0137 GoliášCzechoslovakia
9 Citrullus lanatus   H37-0008 MalaliIsrael
10 Benincasa hispida BEN 485 H15-0001  USA
11 Luffa cylindrica   H63-0010  Unknown
12 Lagenaria siceraria   H59-0009  Unknown

Determination of variation in virulence

A total of 398 isolates of P. cubensis were recovered and screened for virulence variation (pathotypes) from 2001 to 2010 (Table 1). A leaf-disc method (Lebeda & Urban, 2010) was used to determine P. cubensis pathotypes (Lebeda & Widrlechner, 2003). Each genotype of a differential set (Lebeda & Widrlechner, 2003) was represented by five leaf discs (15 mm diameter) in three replicates (one replicate per plant, i.e. altogether 15 leaf discs were screened); replicates were carried out at the same time. A highly susceptible cucumber (Cm. sativus) cv. Marketer 430 was used as a control. Discs were inoculated with a spore suspension (105 spores mL−1) using a glass sprayer and incubated in a growth chamber under standard conditions (Lebeda & Urban, 2010). A visual 0–4 scale (Lebeda, 1991; Lebeda & Urban, 2010) was used to evaluate sporulation intensity over a 2-day period from 6 to 14 days after inoculation as follows: 0 = no sporulation, 1 = 0 to ≤25% of the leaf disc covered by sporulation, 2 = >25% to ≤50% of the leaf disc covered by sporulation, 3 = >50% to ≤75% of the leaf disc covered by sporulation, and 4 = >75% of the leaf disc covered by sporulation.

Sporulation intensity (SP) was expressed as the percentage of maximum sporulation intensity (Lebeda, 1992a):

image

where = the number of discs in every category of infection, = the category of infection, = the maximum level of sporulation and = the total number of evaluated discs.

Sporulation was used as a measure of virulence of individual isolates. Leaf discs with no, or only a low level of sporulation (SP ≤ 35%), were considered to show a resistant response (usually a resistant reaction (−); values 0 or 1 on the above scale); and those with a medium or high level of sporulation (SP > 35%) to be from susceptible genotypes (susceptible reaction (+); values 2–4 on the scale). The reproducibility of results was very high (i.e. 95%), there were no or only small differences (mostly only one degree on the evaluation scale) between individual discs (of which there were 15 for each differential genotype) in each assay. The virulence level of isolates was determined on the basis of the number of virulence factors (VF 1–12, numbered according the differential genotypes, Table 2), i.e. the number of susceptible reactions within the differential set of cucurbitaceous taxa (Table 2). Based on a binary evaluation of susceptible/resistant reaction patterns (+ or −) of a certain isolate of P. cubensis, a numeric tetrad code was created for each isolate, on the basis of differential set groupings. The code comprised three parts, each corresponding to one of three groups of four differentials (Table 2, nos 1–4, 5–8, 9–12). Within each group, numerical values of 1, 2, 4 or 8 were assigned to + results and then summed. The three sums were then presented as a code, in the format (sum of group 1).(sum of group 2).(sum of group 3), which served as a identifier for each pathotype (Tables 3–5). The numeric composition of the code gives a clear picture of the virulence of an isolate (Lebeda & Widrlechner, 2003).

Table 3. Variability of the most frequently recorded pathotypes of Pseudoperonospora cubensis in the period 2001–2010
PathotypeNo. of differential genotypes/pathotype reaction patternNor.aFrequency (% of total)
123456789101112
  1. aNor., number of records (total of 398 isolates of P. cubensis, Table 1).

  2. bThe two most frequently observed pathotypes in the period 2001–2010.

  3. +, compatible/virulent reaction of P. cubensis isolates on cucurbit differential genotypes.

  4. −, incompatible/avirulent reaction of P. cubensis isolates on cucurbit differential genotypes.

15.14.10b+++++++++7619·1
15.14.11b++++++++++5112·8
15.15.11+++++++++++297·3
15.15.15++++++++++++266·5
15.14.14++++++++++215·3
15.15.10++++++++++194·8
15.10.10++++++++164·0
15.10.14+++++++++153·8
15.15.14+++++++++++153·8
15.2.10+++++++123·0
7.14.10++++++++123·0
Total29273·4
Table 4. Fluctuation of the most frequent pathotypes in the Pseudoperonospora cubensis population in the period 2001–2010
PathotypeYear/Frequency of pathotype (%)a
2001200220032004200520062007200820092010
  1. aFrequency was calculated from the total number of collected and analysed (for pathotypes) isolates of P. cubensis in individual years (Table 1).

  2. bThe two most frequently recorded pathotypes in the period 2001–2010.

15.14.10b 14·8  36·034·528·225·045·5 
15.14.11b  19·6 12·013·853·8   
15.15.11   25·0     18·9
15.15.15  10·7    21·9 27·0
15.14.149·5 10·7       
15.15.10   20·0      
15.10.10 18·5        
15.10.147·114·8        
15.15.144·8 10·7       
15.2.10        13·6 
7.14.10     13·8    
Table 5. Characterization of Pseudoperonospora cubensis isolates originating from new host species (sampled in the period 2009–2010)
Host plantIsolateOriginPathotypeNo. of differential genotypesVF
123456789101112
  1. CMe: Cucumis melo; CP: Cucurbita pepo; CM: Cucurbita maxima; CMo: Cucurbita moschata; CL: Citrullus lanatus; VF: number of virulence factors.

  2. +, compatible/virulent reaction of P. cubensis isolates on cucurbit differential genotypes.

  3. −, incompatible/avirulent reaction of P. cubensis isolates on cucurbit differential genotypes.

CMe89/09OL/OL Olomouc-Holice15.14.11++++++++++10
CP58/10JM/HO Mutěnice15.15.3++++++++++10
CP61/10JM/HO Ratiškovice15.15.11+++++++++++11
CP72/10ZL/ZL Napajedla15.15.3++++++++++10
CP73/10ZL/ZL Napajedla15.15.3++++++++++10
CM12/10JM/BO Moravské Bránice15.15.7+++++++++++11
CM67/10JM/HO Veselí nad Moravou15.15.11+++++++++++11
CM81/10OL/OL Olomouc-Holice15.6.0++++++6
CMo86/10MS/NJ Nový Jičín-Kojetín15.15.15++++++++++++12
CMo88/09MS/NJ Nový Jičín-Kojetín4.15.0+++++5
CL83/10OL/OL Olomouc-Holice15.15.11+++++++++++11

Results

Host–pathogen specificity

The development of infection on individual differential genotypes differed substantially during the studied period (Fig. 2a). From the screening, Cucumis spp. genotypes were most frequently susceptible, i.e. a large number of isolates were virulent on them. Genotypes of Cucurbita spp. exhibited a large race-specific-like variation in susceptibility to different P. cubensis isolates. Benincasa hispida and La. siceraria were frequently susceptible to the Czech isolates of P. cubensis (except for an isolate from 2010, for which the frequency of compatible reactions on La. siceraria was lower – only 69%). Luffa cylindrica expressed high variability in susceptibility/resistance (over 60% susceptible reactions in 2001 and 2010 and only about 3% in 2006 and 2007). Citrullus lanatus showed a high frequency of resistant reactions. However, compatible interactions with more than 40% of all tested isolates were recorded in 2003, 2004, 2007 and 2008, and 74% in 2010.

Figure 2.

 Variation in susceptibility of cucurbit differential genotypes (1–12, Table 2) to (a) 398 Pseudoperonospora cubensis isolates (Table 1; mean for the period 2001–2010) and (b) 11 P. cubensis isolates from new hosts from 2009 to 2010 (Table 5 and Fig. 1).

The reaction of 11 P. cubensis isolates (originating from Cucurbita spp. and Ci. lanatus; Table 1) on differential genotypes 1–10 (Table 2) were almost identical (Fig. 2b). However, there were only two susceptible reactions with Lu. cylindrica and five with La. siceraria, which was the lowest frequency of susceptible reactions detected. A high frequency of susceptibility to these isolates was recorded on Cr. pepo ssp. pepo and Ci. lanatus.

Variation in virulence at the isolate level

In total, the 398 isolates were classified as 67 different pathotypes. The number of pathotypes recorded ranged from 33 in 2001 to five in 2007 (Fig. 3). The most frequent pathotypes detected varied by year, but overall, pathotypes 15.14.10 and 15.14.11 were the most frequently recorded (Table 3). One pathotype (15.15.15; i.e. virulent to every differential genotype) was named ‘super pathotype’ and detected frequently in 2001, 2003, 2004, 2008 and 2010, and belonged to the four most frequently recorded pathotypes (Table 3), especially in the years 2003, 2008 and 2010 (Table 4). At the isolate level, 73·4% of isolates were classified into 11 P. cubensis pathotypes (Table 3).

Figure 3.

 Temporal shift of virulence variation (number of pathotypes/year) of Pseudoperonospora cubensis in the Czech Republic in the period 2001–2010.

Eleven isolates, originating from Cm. melo, Cr. pepo, Cr. maxima, Cr. moschata and Ci. lanatus (Table 1), sampled in 2009 and 2010, were classified as seven different pathotypes (Table 5). Two of the pathotype designations (4.15.0 and 15.6.0) were unique in the whole period of study.

Variation in virulence at the population level

The virulence profiles of the studied pathogen populations were highly variable. At the population level, a large proportion of the screened isolates could be considered as highly virulent (i.e. 9–12 VF; Fig. 4). Only pathotypes with high virulence were detected in 2007. In contrast, isolates with low virulence (1–4 VF) were recorded only in 2001. Isolates with 9–10 VF had a considerably higher frequency than isolates with other numbers of VF (Fig. 5). Although isolates with moderate and high virulence predominated in the pathogen populations from 2001 to 2010, the ratio between these two pathogenic groups varied. It was about 1:1–1:3 in 2001, 2002, 2005, 2006, 2008 and 2009, but changed to 1:8 in 2003, 1:12 in 2004 and 1:17 in 2010 (Fig. 4).

Figure 4.

 Structure of Pseudoperonospora cubensis populations according to their virulence level (i.e. total number of virulence factors (VF) able to overcome differential genotypes, Table 2). Three basic categories were distinguished: low virulence level (1–4 VF), moderate (5–8 VF) and high (9–12 VF).

Figure 5.

 Frequency of Pseudoperonospora cubensis isolates by number of virulence factors (mean for the period 2001–2010).

Of the isolates originating from a new host species (Cm. melo, Cucurbita spp., Ci. lanatus), 82% were highly virulent pathotypes (9–12 VF) and only 18% expressed moderate virulence (5–8 VF).

Temporal and spatial variation in virulence

Temporal changes in the virulence of P. cubensis populations in the Czech Republic during the 10-year period were observed (Fig. 3). This phenomenon was demonstrated by the frequency with which specific pathotypes were detected in individual years (Table 4). The highest number of pathotypes (33) was detected in the first year of the study (2001), decreasing dramatically to 17 pathotypes in 2002, and then shifting to between 5 and 14 (Fig. 3) from 2003 to 2010. The distinct temporal shift to reduced variation in virulence was most apparent from 2001 to 2007, after which the number of pathotypes increased to 10–15, as before (Fig. 3). Virulence of isolates varied by location, but this was not substantial and it is difficult to make any concrete conclusions regarding this phenomenon. Starting in 2009, new P. cubensis pathotypes were found annually and it seems that the virulence structure of the pathogen population has been changing substantially during the last 3 years (Table 4, Fig. 2b; A. Lebeda & J. Pavelková, unpublished results).

Discussion

Previous studies have shown that P. cubensis is a highly variable pathogen from the viewpoint of host-specificity, race-specificity and virulence (Lebeda et al., 2006; Lebeda & Cohen, 2011). These findings are also supported by recent molecular studies (Sarris et al., 2009; Mitchell et al., 2011; Runge et al., 2011; Quesada-Ocampo et al., 2012). Nevertheless, understanding the variation in virulence of populations at the scale of country or continent, including spatiotemporal shifts, is still very poor. This paper may be the first comprehensive contribution to this topic from the viewpoint of pathotypes.

The comparison of P. cubensis isolates collected from 2001 to 2010 in the Czech Republic confirms that the virulence structure of Czech pathogen populations is very broad and dynamic in time and space. It is evident that P. cubensis populations are extremely variable in host specificity and virulence. Variation in virulence within Czech populations is even broader than was previously reported in 2002 (Lebeda & Gadasová, 2002). Virulence structure showed a temporal shift from 2001 to 2007 to a higher number of virulence factors and a lower number of pathotypes (Lebeda & Urban, 2007). However, variation in virulence changed and increased again in the few years after that (2008–2010) (Table 4; Fig. 4).

Although this variation is consistent with some other downy mildews (e.g. Lebeda & Schwinn, 1994), such a broad spectrum of variation in virulence has not been reported before for P. cubensis from any other country (e.g. Bains & Sharma, 1986; Thomas et al., 1987; Cohen et al., 2003; Salati et al., 2010b) where mostly only a few pathotypes were recorded. Highly virulent isolates were observed in Israel (Cohen et al., 2003) originating from Cr. moschata and Cr. pepo ssp. pepo. European isolates (mostly originating from the Czech Republic) are highly variable and differ substantially (Lebeda & Gadasová, 2002; Lebeda et al., 2006, 2010) in comparison to the five pathotypes described previously from Japan, Israel and the USA (Thomas et al., 1987; Cohen et al., 2003). However, the previously reported pathotypes were described based on a different set of differential host genotypes, i.e. it is likely that the variability was underestimated. Nevertheless, the high variability in European isolates was hypothesized and expected (Lebeda, 1991), and agrees with previous experimental results that demonstrated substantial differences in pathotype structure among various European locations (Lebeda & Gadasová, 2002; Lebeda & Urban, 2007). Unfortunately, these results cannot be compared directly with the data from other countries/continents (e.g. USA, Asia), where mostly only molecular polymorphism of P. cubensis populations has been studied (e.g. Quesada-Ocampo et al., 2012). Nevertheless, the results of the present work indicate that the virulence structure of Czech P. cubensis populations is very variable and probably also different from that of other parts of the world (Lebeda & Gadasová, 2002; Lebeda et al., 2006; Colucci, 2008; Lebeda & Cohen, 2011). This is also supported by recent molecular studies (Quesada-Ocampo et al., 2012). In addition, P. cubensis was shown to be quite variable in time and space (Tables 4 & 5; Fig. 3). This explains in part why it is not possible to predict reliably its development and structure.

Downy mildew pathogens are variable and able to overcome new disease-resistant host genotypes rapidly (Drenth & Goodwin, 1999). There are good examples (e.g. Bremia lactucae) that show a temporal shift in virulence (Lebeda & Zinkernagel, 2003). McDonald & Linde (2002) hypothesized that the evolutionary potential of a pathogen population is reflected in its population genetic structure, and pathogen populations with a high evolutionary potential are more likely to overcome genetic resistance than those with a low evolutionary potential. The findings of the present study indicate that P. cubensis has a high evolutionary potential and, according to the terminology of McDonald & Linde (2002), it should be considered a ‘risky’ pathogen (Lebeda & Urban, 2007). The evolutionary forces operating in cucurbit–P. cubensis interactions and responsible for pathogen population diversity and breakdown of host resistance are not well known (Lebeda & Widrlechner, 2003). The high potential for genetic variation could depend on many variables which are the subject of the following discussion.

Populations of P. cubensis have not been sufficiently studied at the pathotype level, using a settled differential set, for there to be enough data to understand spatiotemporal virulence variation (Lebeda et al., 2006). The differential set used by Thomas et al. (1987) for pathotype determination was expanded (Lebeda & Gadasová, 2002; Lebeda & Widrlechner, 2003) to comprehensively unify the effort for characterizing populations of P. cubensis at the level of pathotypes (Lebeda et al., 2006). While the majority of available studies solely define pathotypes (Lebeda & Widrlechner, 2003), the lack of uniform and comparable differential genotypes does not allow for pathotype determinations to be drawn from the pronounced divergences in host-range reactions reported both within and between countries. The situation is further complicated by the lack of differentials to distinguish among races on the most important Cucurbitaceae host taxa (e.g. Cucumis, Cucurbita, Citrullus) (Lebeda et al., 2006). However, different races have been postulated and reported (e.g. Shetty et al., 2002; Lebeda et al., 2006) on various cucurbits, and there are probably different genes involved in resistance to different races (Lebeda et al., 2006), if a gene-for-gene interaction is assumed. Other data indicate that the host–pathogen specificity between Cm. meloP. cubensis and Cucurbita spp.–P. cubensis is controlled by race-specific factors (Lebeda, 1991, 1999; Lebeda & Křístková, 1993, 2000; Lebeda & Gadasová, 2002; Lebeda & Widrlechner, 2003, 2004) which can serve as a force for microevolutionary changes in pathogen populations.

The cucurbit–P. cubensis system is not well known or well defined from the viewpoint of host–pathogen specificity and genetics (Lebeda & Widrlechner, 2003; Lebeda et al., 2006; Lebeda & Cohen, 2011). The virulence structure of the pathogen population is principally a function of the structure of the host population (i.e. host species, number of resistance genes and their dynamics in time and space) (Müller et al., 1996). However, a study on co-evolution in various pathosystems showed that this is not totally correct (Lebeda & Zinkernagel, 2003) and population structure is influenced by other factors. In general, the most important processes that affect the generation and maintenance of genetic diversity within populations of downy mildews include mutation, the reproductive system, cytoplasmic factors, migration and gene flow, genetic drift and selection (Drenth & Goodwin, 1999). Similar processes could also influence variation of P. cubensis populations. Little is known about mutations and mutation rates in oomycetes (Drenth & Goodwin, 1999), including P. cubensis (Lebeda & Cohen, 2011). Nevertheless, this process could contribute to broad variation because of extremely large population sizes and high asexual reproduction potential (Lebeda & Cohen, 2011). Recent laboratory studies of sexual reproduction demonstrated oospore formation in P. cubensis, which may play a crucial role in sexual recombination of this pathogen and the appearance of new pathotypes (Cohen & Rubin, 2012). However, detailed studies of this phenomenon need to be conducted (Lebeda & Cohen, 2011).

Migration and gene flow are considered very important aspects of the population genetics of oomycetes (Drenth & Goodwin, 1999). In US population studies of P. cubensis, it was hypothesized that transport of the pathogen can occur over long distances via atmospheric wind currents (Holmes et al., 2004). A recent detailed spatiotemporal study of P. cubensis spread in the eastern USA clearly showed that infection of cucurbits by P. cubensis appears to be an outcome of a contagion process and that factors occurring on a large spatial scale (c. 1000 km) facilitate the spread of the pathogen (Ojiambo & Holmes, 2011). The study demonstrated that the median nearest-neighbour distance of a spread of new disease cases was c. 110 km, and the epidemic expanded at a rate of 9·2 and 10·5 km day−1 (Ojiambo & Holmes, 2011). In Europe, transport of sporangia from southern to central Europe and to Scandinavia has also been suggested (Lebeda & Schwinn, 1994). However, spatial dispersal depends not only on climatic factors (e.g. temperature, humidity, wind currents), but also on host population density, species and genetic structure. Previous and recent population studies of P. cubensis in the Czech Republic showed that the pathogen population is very diverse (pathogenicity variation, fungicide resistance) and dynamic in time and space (Lebeda & Urban, 2007; Urban & Lebeda, 2007; Lebeda et al., 2010). The data from the present study showed the occurrence of a large number of different pathotypes (67 in total) in the pathogen population over the studied period and their spatiotemporal fluctuations. This rather unique and changeable virulence structure could be the result of pathogen migration in Europe. Because the Czech Republic is located in central Europe and sporangia of P. cubensis are wind-dispersed over a long distance (Ojiambo & Holmes, 2011), the population structure could be primarily (at the beginning of the cucurbit growing season, i.e. May–June) formed by pathogen propagules coming from southern Europe, and secondarily (i.e. during the main growing season, July–August) by propagules from all neighbouring and some other European countries. The main reason for this expectation could be relatively short distances and a high population turnover in countries in Europe, very frequently changing wind currents, and differences in the cucurbits grown, which could contribute to the selection of various pathotypes. These factors could also substantially contribute to the very changeable virulence structure of P. cubensis populations from a temporal viewpoint. On the other hand, infrequent human-mediated transport of P. cubensis is also expected (Quesada-Ocampo et al., 2012).

Quesada-Ocampo et al. (2012) recently reported on a detailed study of the genetic structure of worldwide P. cubensis populations and identified six different genetic clusters. However, approximately half of the isolates belonged to one of the clusters. There was no direct correspondence between inferred genetic clusters and grouping of isolates by predefined geographic and host categories. Multilocus genotypes were shared across continental, host-of-origin and temporal scales, indicating that some genotypes are widely dispersed and persistent (Quesada-Ocampo et al., 2012). Previous genetic studies found that isolates from the Czech Republic, the Netherlands and France were significantly different from isolates from Greece (Sarris et al., 2009). Also, the study by Quesada-Ocampo et al. (2012) showed that countries in western Europe (France, the Netherlands, Germany) had a similar cluster composition, as did Greece, Italy and Spain. However, the Czech Republic, Turkey and Israel each displayed a different cluster composition. These results show that a high genetic differentiation of P. cubensis populations exists in Europe and surrounding countries and this can also contribute to the virulence variation in these countries, including the Czech Republic. This conclusion is supported by the high genetic diversity estimates for isolates from Greece, Italy and the Czech Republic. In contrast, Quesada-Ocampo et al. (2012) observed that, compared with this, California had low diversity estimates. This may be because of its geographic separation from eastern US states with severe downy mildew outbreaks. Recently, Runge et al. (2011) found potential evidence for a spatiotemporal split between the two major lineages/subspecies of P. cubensis between east Asia and western countries, which seems to have equalized over time through the worldwide spread of both groups of pathogens.

Processes of genetic drift and selection could also influence the virulence structure of P. cubensis populations. Recent phylogenetic studies showed some genetic similarity between P. cubensis and P. humuli (Choi et al., 2005; Sarris et al., 2009). There were also reports (Mitchell et al., 2011; Runge et al., 2011) that there is some limited infectivity of P. cubensis on hop and P. humuli on cucumbers. According to Mitchell et al. (2011), P. humuli is nested within P. cubensis; however, Runge et al. (2011) demonstrated that P. humuli is separate and most likely basal to P. cubensis. Population genetic studies including P. humuli isolates could be key to determining the potential extent of gene flow between these sister species, including the contribution of P. humuli populations to the genetics and variation in virulence of P. cubensis populations. The study of Quesada-Ocampo et al. (2012) showed that there is some genetic diversification between the isolates originating from different host cucurbit species. The diversity estimates were also higher for isolates collected from non-cucumber hosts than from cucumber. This supports a previous idea (Lebeda et al., 2006; Lebeda & Cohen, 2011) that higher genetic diversity of resistance in Cucurbita spp. and some other Cucurbitaceae may substantially contribute to the selection of new P. cubensis pathotypes. These results also suggest that inclusion of isolates from Cm. melo and Cucurbita spp. that show different genetic composition and high genetic diversity is necessary to capture the genetic variation of P. cubensis. These aspects must be also considered from the viewpoint of geography, i.e. regions with high genetic diversity should be of special concern because P. cubensis populations with high levels of genetic variation are likely to adapt more rapidly to resistant hosts (Quesada-Ocampo et al., 2012). This process was recently demonstrated in the Czech Republic by repeated severe infections of Cucurbita spp. (Table 5; Lebeda et al., 2011; Pavelkováet al., 2011). There is an alternative possibility that a new pathotype(s) evolved somewhere else and migrated into the Czech Republic. Future detailed studies may contribute to the understanding of the evolutionary forces responsible for spatiotemporal shifts in pathogenicity and genetic variation within and among populations of P. cubensis.

Acknowledgements

The authors thank G. J. Holmes (Valent USA Corporation, Cary, North Carolina) for his valuable critique of this manuscript. This research was supported by grants MSM 6198959215 (Czech Ministry of Education, Youth and Sports) and QH 71229 (Czech Ministry of Agriculture), and the Internal Grant Agency (Prf-2010-001 and Prf-2011-3).

Ancillary