Spontaneous loss of Yr2 avirulence in two lineages of Puccinia striiformis did not affect pathogen fitness



Fitness costs associated with the emergence of virulence (loss of avirulence) have been a subject of much debate in plant pathology. Here, differences in fitness between two pairs of wild types and spontaneous virulence mutants in Puccinia striiformis were studied. The mutants differed from their respective wild types in virulence corresponding to the Yr2 resistance gene in wheat. The wildtype and mutant pairs represented different genetic lineages and virulence phenotypes. Colony size, latent period, lesion growth rate and uredinium density and size were assessed on susceptible and Yr2 resistant wheat varieties. No significant difference was observed for any of the evaluated parameters within wildtype and mutant pairs, i.e. the results did not support a hypothesis of a fitness cost for acquiring virulence in P. striiformis. In contrast, significant differences were found for most of the parameters when isolates were compared according to genetic lineage, which confirm that the applied methodologies were appropriate for detecting differences in pathogen fitness.


The potential fitness cost for host adaptation is a central theme in the evolutionary biology of plant pathogens. The general ideas are that trade-off exists between traits that are important for host infection and pathogen transmission (Sacristán & García-Arenal, 2008; Asplen et al., 2012), and that such trade-off effects play an important role in the host–pathogen coevolution (Leonard, 1977; Thrall & Burdon, 2003; Brown & Tellier, 2011). A better understanding of these trade-off effects may have direct implications for the development of disease control strategies in agricultural systems.

The gene-for-gene model has been one of the most important concepts for the understanding of host–pathogen coevolution (Flor, 1955; Thompson & Burdon, 1992). In this model, the host–pathogen interaction is controlled by specific alleles at corresponding loci in the pathogen and the host. The resistance phenotype is expressed only when the resistance allele (R gene) is present in the host and the corresponding avirulence allele is present in the pathogen. Virulence represents the compatible interaction, which may arise as a result of inactivation of the avirulence allele so that the resulting product no longer elicits resistance in the host (Kang et al., 2001; Sacristán & García-Arenal, 2008). Deployment of crop varieties with R genes for which the corresponding virulence frequency is low or absent in the pathogen population has played a central role for disease management in agriculture. R genes have most often remained effective for disease control in a limited period of time as a result of strong selection favouring pathogen individuals with the matching virulence (e.g. Bayles et al., 2000; de Vallavieille-Pope et al., 2012). When an R gene is widely deployed, pathogen individuals with the matching virulence have a clear fitness advantage and the virulence frequency can increase dramatically in the pathogen population within a short time span (e.g. Hovmøller et al., 1993). An important question is, however, what may happen to the virulence frequency when the frequency of the corresponding R gene is significantly reduced in the host population? Based on observations in flax rust (Melampsora lini) Flor (1953) found that isolates with the least number of virulence genes necessary for survival dominated populations of this pathogen. Later, Van der Plank (1968) introduced the concept of ‘stabilizing selection’, where he hypothesized that in a gene-for-gene system there will be a cost of virulence in the absence of a corresponding host R gene. Following these ideas, the concept of ‘cost of virulence’ has been much debated and the subject of many investigations in viral, bacterial and fungal pathogens (Leach et al., 2001; Pietravalle et al., 2006).

Both observational and experimental approaches have been followed to study a possible cost of virulence in populations of fungal plant pathogens. Observational studies have often been based on population survey data, which may be difficult to interpret because of confounding effects such as ecological and genetic factors and interactions between specific host and pathogen genotypes (Hovmøller & Østergård, 1991; Brown & Tellier, 2011; Sommerhalder et al., 2011). In experimental studies, more direct comparisons of virulent and avirulent isolates have been carried out as competition studies over several reproduction cycles in the field or through the assessment of pathogen fitness components during a single reproduction cycle. In a classical competition experiment, isolates of Puccinia coronata carrying only a few virulences increased in frequency compared to isolates with many virulences (Leonard, 1969). A more recent competition study on the yellow rust pathogen Puccinia striiformis indicated that the cost may vary depending on the gene and the experimental settings (Bahri et al., 2009). The assessment of fitness components for individual fungal isolates to investigate cost of virulence has only been done in a few experiments. Pringle & Taylor (2002) discussed that, even though several life history traits may contribute to the fitness of a fungal pathogen, epidemiological parameters such as infection frequency, latent period and spore production may be sufficient to study fitness differences. In most studies using epidemiological parameters for assessment of pathogen fitness, differences have been ascribed to the origin of the isolate rather than to differences in virulence (Pariaud et al., 2009). An exception is a study on Phytophthora infestans where fitness differences were found in some cases when isolates were grouped according to individual virulences and avirulences (Montarry et al., 2010).

In the present study, the hypothesis that the acquisition of virulence in P. striiformis is associated with a fitness cost was investigated. Two Yr2 avirulent wildtype isolates of different lineages and corresponding spontaneous virulence mutants were inoculated on susceptible and Yr2 resistant host varieties. Fitness was assessed at the macroscopic level by latent period, lesion growth and uredinium size and density. At a microscopic level, the sizes of colonies in infected leaves were assessed at different time points after inoculation. The experimental material in this study offered a unique opportunity to study the cost of virulence in a biotrophic fungal pathogen.

Materials and methods

Pathogen isolates and hosts varieties

Two pairs of wild types and mutant isolates of Puccinia striiformis were used. Mutants were derived in 2005 and 2006 from experimental field plots aiming to detect virulence mutants for Yr2 resistance in wheat. Different rows of a susceptible wheat cultivar, Anja, were inoculated with wildtype isolates of northwest European origin, DK24/95 and UK75/30 (Hovmøller & Justesen, 2007a). DK24/95 (lineage 2 in this study) had been kept as freeze-dried urediniospores in sealed ampoules since the year of collection in Denmark (1995), with occasional multiplication on susceptible wheat seedlings to ensure viability. UK75/30 (lineage 1 in this study) had been stored in liquid nitrogen from 1975 to 2005 in the yellow rust isolate collection hosted by the John Innes Centre, Norwich, UK, after which it was transferred to Denmark, revived on plants and resulting urediniospores were freeze dried and kept in sealed ampoules at 4°C until use. Prior to field inoculation the isolates were recovered on seedlings of Anja and identity and purity was confirmed by virulence phenotyping according to Hovmøller & Justesen (2007a). Field inoculation was carried out by rubbing live infected seedlings on field plants of Anja. The field rows of Anja were flanked by rows of cultivar Skater (Yr2, Yr32) (Hovmøller, 2007), which was resistant to the population of P. striiformis in Denmark at that time. In addition to differences in five virulence genes (Table 1), the two wildtype isolates differed by 13 AFLP fragments among a total of 30 AFLP polymorphisms detected in the northwest European P. striiformis population at that time. To detect whether P. striiformis infections emerging on Skater were virulence mutants arising from a wildtype isolate, or exotic immigrants from outside the field experiment, single lesions were transferred to the laboratory for multiplication, virulence phenotyping and AFLP fingerprinting according to Justesen et al. (2002). Based on these tests, Yr2 virulence mutant isolates were identified from the two wildtype isolates. No polymorphism was detected within wildtype and mutant pairs when screened by 20 AFLP primer combinations producing c. 1400 AFLP fragments. From the time of collection in the field to the time of experiment (summer 2007) wildtype and mutant isolates were reinoculated onto new host plants every 2–3 months. Purity and identity of the isolates was confirmed by virulence phenotyping at the time of multiplication of urediniospores for the present experiment.

Table 1. Virulence phenotype of wildtype (W) and virulence mutant (M) isolates of lineage 1 and 2
Isolate designationVirulence phenotypea
  1. a

    Figures and symbols designate virulence corresponding to specific yellow rust resistance genes. –, avirulence; Sd, Strubes dickoff; Sp, Spalding prolific.

UK75/30Wildtype (W1)12532SdSp
Mut15/05Mutant (M1)122532SdSp
DK24/95Wildtype (W2)23462532Sd
Mut21/06Mutant (M2)223462532Sd

Fitness parameters of wildtype and mutant isolates were evaluated on four host cultivars: Avocet S and Cartago which were susceptible to all four isolates, and two cultivars carrying Yr2 resistance, Skater (Yr2, Yr32) and Heines VII (Yr2, Yr25, +). Seeds were sown in 7·5 cm pots filled with Pindstrup substrate, a standard peat-based mix with slow release nutrients (Pindstrup Mosebrug A/S). Pots were placed in trays covered with plastic lids to ensure high humidity during germination in spore-proof greenhouse cabins. Artificial light of 50–100 μE m−2 s−1 was applied when daylight was <10 000 lux outside. Light was applied 18 h a day and with 17°C day/12°C night.

Experimental procedures

Three separate experiments were conducted on seedlings for assessment of (i) latent period and lesion growth rate, (ii) uredinium size and density, and (iii) colony size, respectively. In all three experiments, wheat seedlings were inoculated 16 days after sowing, when the second green leaf was fully expanded. High and uniform inoculum quality was developed by multiplying all isolates for three generations on cultivar Cartago. Plants were further shaken 48 h prior to harvest of spores for the actual experiment. Fresh spores were mixed with talc (1:19 w/w) and applied to the leaves with a camelhair paintbrush (size 1). The inoculated plants were incubated in a dark cold chamber (10°C) for 20 h. Following incubation, plants were transferred to spore-proof chambers and pots were randomized within and between chambers.

For assessment of latent period and lesion growth rate, a 0·5 cm long area on the adaxial side of the second leaf was inoculated c. 11 cm below the leaf tip. Fourteen pots (replicates), each with one plant, were used per isolate–variety combination (treatment). Latent period was defined as the number of hours from inoculation until the first spores were visible in next generation uredinia on the leaf surface. Plants were observed with a ×10 hand lens at 12 h intervals starting 8 days after inoculation. Four days after the first spores were observed on each leaf, the first measurement of lesion length was done with a ruler to the closest mm. Any plants with infections outside the inoculated area were excluded from the experiment at this time point. The lesion was defined as the area with visible and erupted pustules. The lesion expands along the leaf during infection as a result of semi-systemic growth, and the second lesion measurement was done 4 days after the first. Lesion growth rate was calculated as the difference between the two measurements divided by number of days between them.

In the set-up for determination of uredinium size and density, 10 pots, each with two plants, were used per treatment and an adaxial area of 2 cm approximately 11 cm below the tip of the second leaf was inoculated. For assessment of uredinium size and uredinium density, one inoculated leaf segment was cut off per pot 16 days after inoculation and prepared for microscopy.

For estimation of colony size, 10 seeds were sown per pot and reduced to five uniform plants per pot prior to inoculation. Ten pots were used per treatment and an adaxial area of 2 cm about 11 cm below the tip of the second leaf was inoculated. One leaf segment per pot was sampled from different plants at 3, 5 and 7 days post-inoculation (dpi). Leaf samples were prepared for microscopy. All three experiments were repeated twice.

Histological assessments

Prior to microscopic observation, leaf samples for assessment of uredinium size, uredinium density and colony size were stained according to Moldenhauer et al. (2006). Leaf segments were fixed and cleared in ethanol:chloroform (3:1, v/v) + 0·15% (v/w) trichloroacetic acid for at least 24 h. After being washed twice in 50% ethanol, they were left in 0·05 m NaOH for 30 min. Specimens were then rinsed in water before being submerged in 0·1 m Tris-HCl buffer (pH 5·8) for 30 min. They were then stained for 5 min in 0·1% (w/v) Uvitex 2B (Polysciences Inc.) in Tris-HCl buffer (pH 5·8). Following staining, specimens were washed four times in water, one time in 25% glycerol and left overnight in water. They were stored in 50% glycerol for later observation. Whole mounts were prepared at the time of observation and microscopy was done with a Leica DMR equipped with optics for epifluorescence. Structures were visualized with a UV-1D filter (excitation filter 355–425 nm, barrier filter 455 nm).

Five randomly chosen uredinia per leaf sample were measured for the estimation of uredinium size. Uredinium density was estimated by counting the number of uredinia within three random areas of 2 × 2 mm per leaf sample. Colony size was estimated by assessing between two and five randomly selected colonies for each leaf sample, depending on the number of infections. The number of infections was highly variable and some leaves had no visible infections. A colony was defined as a consistent mycelium resulting from a single substomatal vesicle originating from a single spore. Dimensions of both colonies and uredinia were measured by recording the largest length and largest width with a calibrated eyepiece micrometer. Colony size and uredinium size was calculated as: largest length × largest width × π/4 (Baart et al., 1991). The mean was calculated for measurements done on the same leaf and the leaf mean value was used as experimental unit in the statistical analysis.


Data for all variables were analysed using mixed models (West et al., 2006). Variance structure and normal distribution were evaluated by a plot of model residuals. Data for latent period and colony size were log-transformed before analysis to normalize the distribution of residuals and results were back-transformed for presentation (Jørgensen & Pedersen, 1998). The fixed effect in the model was the combinations of isolate and cultivar. Random effects included experiment and the combinations of experiment, isolate and cultivar. Data for all the compatible interactions (all four isolates on Cartago and Avocet S and the two mutant isolates on Skater and Heines VII) were analysed in the same model. Contrasts were formulated in order to test fixed effects of isolate, cultivar and their interaction for data collected on susceptible and resistant cultivars, respectively. Selected pairwise comparisons between isolates or pair of isolates were estimated based on the effects for the combinations of isolate and cultivar. Degree of freedom for the denominator was calculated using the Kenward–Roger's method (Kenward & Roger, 1997). No adjustments of P values for multiple comparisons were applied and the significance levels were set at α = 0·05. The analysis was carried out using the mixed procedure in sas v. 9.2 (SAS Institute Inc.).

The effect of replicating the experiments was analysed using general linear models. Experiment, isolate, cultivar and combinations of these were included as fixed effects. Results for latent period and colony size were log-transformed before analysis. Analysis was carried out with the glm procedure in sas v. 9.2.


All four isolates used in this study showed full compatibility with the two susceptible wheat cultivars, Avocet S and Cartago, whereas only the two mutant isolates were able to reproduce on the Yr2 resistant cultivars, Skater and Heines VII. Statistical results for latent period, lesion growth rate, uredinium size and uredinium density are shown in Tables 2 and 3, and results for colony size are presented in Tables 4 and 5. No significant interaction between isolate and cultivar was found for any of the assessed parameters whereas main effects of both isolate and host cultivar was common (Tables 2 and 4).

Table 2. Main effect of Puccinia striiformis isolate and wheat cultivar on the four fitness parameters latent period, lesion growth rate, uredinium density and uredinium size. Results for the susceptible cultivars (Cartago and Avocet S, susceptible to all isolates) and the Yr2 resistant cultivars (Heines VII and Skater, susceptible only to the mutant isolates) were analysed separately
Fixed effectd.f.Latent periodaLesion growth rateUredinium sizeUredinium density
F valueP valueF valueP valueF valueP valueF valueP value
  1. a

    Analysis performed on log-transformed data.

All isolates on Cartago and Avocet S
Isolate × Cultivar30·820·51040·510·67600·390·76040·670·5900
Mutants on Heines VII and Skater
Isolate × Cultivar10·120·73310·000·96502·550·13931·710·4250
Table 3. Pairwise comparisons of wildtype and mutant isolates of Puccinia striiformis for latent period and lesion growth rate on the susceptible wheat cultivars (Avocet S and Cartago) by the use of statistical contrasts
Contrastsd.f.Latent periodd.f.Lesion growth rate
F valueP valueF valueP value
  1. a

    ‘Wild types’ represents the combined results for W1 and W2 whereas ‘Mutants’ is the combination of M1 and M2.

  2. b

    ‘Lineage 1’ represents the combination of W1 and M1 whereas ‘Lineage 2’ is the combination of W2 and M2.

W1 vs M110·74·430·05971962·510·1115
W2 vs M210·21·810·20751960·440·5093
Wild types vs Mutantsa10·40·330·57881962·520·1139
M1 vs M211·113·70·00351964·480·0356
W1 vs W29·720·110·742919610·10·0018
Lineage 1 vs Lineage 2b10·48·480·014919613·80·0003
Table 4. Effects of isolate, cultivar and their interaction on the size of Puccinia striiformis colonies in wheat leaves at four different times after inoculation (days post-inoculation, dpi). Results for the susceptible cultivars (Cartago and Avocet S, susceptible to all isolates) and the Yr2 resistant cultivars (Heines VII and Skater, compatible only with mutant isolates) were analysed separately
Fixed effectd.f.Colony size
3 dpi5 dpi7 dpi
F valueP valueF valueP valueF valueP value
All isolates on Cartago and Avocet S
Isolate × Cultivar30·510·68000·370·77640·420·7387
Mutants on Heines VII and Skater
Isolate × Cultivar10·260·64550·030·86710·710·4130
Table 5. Pairwise comparisons for colony size of wildtype and mutant isolates on susceptible cultivars (Avocet S and Cartago) with the use of statistical contrasts, at 7 days post-inoculation
Estimatesd.f.F valueP value
  1. a

    ‘Wild types’ represents the combined results for W1 and W2 whereas ‘Mutants’ is the combination of M1 and M2.

  2. b

    ‘Lineage 1’ represents the combination of W1 and M1 whereas ‘Lineage 2’ is the combination of W2 and M2.

W1 vs M116·31·870·1897
W2 vs M217·40·250·6209
Wild types vs Mutantsa16·90·360·5571
M1 vs M216·41·260·2785
W1 vs W217·38·690·0089
Lineage 1 vs Lineage 2b16·98·340·0103

Latent period and lesion growth rate

Both latent period and lesion growth were significantly affected by replication (< 0·0001), no interaction between experiment and main effects was found (data not shown).

Latent period and lesion growth was significantly affected by isolate and cultivar (Table 2). This was the case for both susceptible and Yr2 resistant cultivars. However, a comparison of wild types and corresponding mutants based on statistical contrasts revealed no significant differences in these parameters (Table 3). The significant effect of isolate was caused by differences between lineages defined by AFLP. The mutant of lineage 1 had a significantly longer latent period than the mutant of lineage 2 and the wild type of lineage 2 had a significantly higher lesion growth rate than the wild type of lineage 1. As no significant difference was detected when comparing the wild types with their corresponding mutants, the results for each wild type and mutant pair were combined to provide a stronger statistical test. This test confirmed the previous results, i.e. isolates of lineage 2 had a significantly shorter latent period and a higher lesion growth rate than isolates of lineage 1. These differences observed on the susceptible cultivars were supported by the corresponding results of the virulence mutants infecting the Yr2 resistant cultivars, i.e. a shorter latent period and a higher lesion growth rate for the mutant of lineage 2 compared to the mutant of lineage 1 (Table 2). The effects of cultivar could be ascribed to a significantly shorter latent period and a higher lesion growth rate on Cartago as compared to Avocet S. On the Yr2 resistant cultivars, the latent period was shortest on Skater and the lesion growth rate was highest on Heines VII. Least square means of latent period and lesion growth rate for all treatments are presented in Figure 1.

Figure 1.

Least square means (± SE) of (a) latent period, and (b) lesion growth rate for Puccinia striiformis wildtype (W1, W2) and mutant isolates (M1, M2) on two susceptible (Avocet S, Cartago) and two Yr2 resistant wheat cultivars (Skater, Heines VII).

Uredinium size and density

Uredinium size and density were significantly affected by experimental replication with bigger uredinium size and higher uredinium density in the first replication (< 0·0001). On the susceptible cultivars, uredinium size was not significantly affected by isolate or cultivar, whereas uredinium density was higher on Avocet S compared to Cartago (Table 2). Because of the insignificant effect of isolate, no post hoc comparisons between the isolates were performed. On the Yr2 resistant cultivars, the virulence mutant of lineage 2 produced significantly larger uredinia and had a higher uredinium density than the mutant of lineage 1. Uredinia were larger on Skater compared to Heines VII. Least square means for all treatments are presented in Figure 2.

Figure 2.

Least square means (± SE) of (a) uredinium size, and (b) uredinium density of Puccinia striiformis wildtype (W1, W2) and mutant isolates (M1, M2) on two susceptible (Avocet S, Cartago) and two Yr2 resistant wheat cultivars (Skater, Heines VII).

Colony size

A significant effect of experiment replication was found for colony size, with significantly larger colonies in the first experiment at all time points (< 0·0001). On susceptible cultivars a significant effect of isolate was observed at 7 dpi (Table 4). This was due to the larger colonies of the wild type of lineage 2 compared to the wild type of lineage 1. Lineage 2 had significantly larger colonies than lineage 1 (Table 5). As for the other parameters, no significant influence of the virulence mutations corresponding to Yr2 was observed. The results observed on susceptible cultivars were confirmed by significantly larger colonies on the Yr2 resistant varieties at 5 and 7 dpi of mutant of lineage 2 compared to mutant of lineage 1. Least square means for colony size of all treatments at 7 dpi are presented in Figure 3.

Figure 3.

Colony size least square means (± SE) for Puccinia striiformis wildtype (W1, W2) and mutant isolates (M1, M2) on two susceptible (Avocet S, Cartago) and two Yr2 resistant wheat cultivars (Skater, Heines VII) at 7 days post-inoculation.


The experimental material in the present study provided an opportunity to study the cost of virulence in a fungal pathogen. In contrast to many previous studies of cost of virulence in fungal pathogens (e.g. Pariaud et al., 2009), the current study had access to stocks of wildtype and mutant pathogen isolates. The mutant isolates were derived from a field experiment specifically designed to screen for spontaneous virulence mutants. The experiment was based on the availability of two atypical wildtype isolates of different P. striiformis lineages (based on AFLP) that shared virulence to Yr32 and avirulence to Yr2. They were inoculated on a susceptible spreader host, which was grown in between rows of Skater (Yr2, Yr32) and Robigus (Yr2, Yr32), which were resistant to the two wildtype isolates (Hovmøller, 2007). The field experiments took place in 2005 and 2006 in Denmark, where yellow rust was scarce and Yr32 resistance provided 100% disease control (Hovmøller & Henriksen, 2008). The virulence phenotype of the two wildtype isolates had been absent from the Danish P. striiformis population for more than 10 years. During the two experimental growing seasons, putative virulence mutants were collected on Skater and Robigus and they were subsequently assayed for virulence phenotype in the greenhouse according to Hovmøller & Justesen (2007a). Additional DNA fingerprinting for confirmation of identity was made by AFLP primer pairs selected to reveal maximum variability in the northwest European P. striiformis population. The two virulence mutant isolates shared both AFLP fingerprint and virulence phenotype with their respective wildtype isolate, except for Yr2 virulence. The virulence phenotype was confirmed for all isolates prior to the experiment and experimental results did not indicate any change in the Yr2 virulence of mutants as a result of storage and several cycles of multiplication on plants. The identification of the same phenotypic shift in virulence in two different genetic lineages offered the opportunity to study a possible trade-off between virulence and fitness (cost of virulence). Eliminating the effect of genetic background for assessment of a possible fitness penalty of individual virulence genes has previously been considered highly important, e.g. by using near-isogenic isolates or a sexual recombining population (Østergård, 1987).

None of the fitness parameters evaluated in the present study suggested a cost of virulence in P. striiformis for Yr2 resistance in wheat. This result was observed in two different genetic lineages in which virulence for Yr2 had been acquired by independent mutation events. Nevertheless, the generality may be difficult to predict because several changes in an avirulence gene may give rise to a certain virulence phenotype (Schürch et al., 2004; Rouxel & Balesdent, 2010). In the present case, the exact nature of the genetic shift from wild type to mutant is not known. Nonetheless, the lack of a fitness penalty in pathogens acquiring a new virulence is not uncommon in plant pathogens (Leach et al., 2001). Only a few other studies have dealt with cost of virulence in P. striiformis. Lannou et al. (2005) observed no significant difference in rate of change in frequency between complex and simple pathotypes when inoculated in pure stands, which implied no cost of virulence. In contrast to these results, Bahri et al. (2009) reported a cost of virulence in P. striiformis corresponding to Yr6 and Yr4 resistance in wheat, but not for Yr9. In the oomycete, Phytophthora infestans, no cost of virulence was detected in most cases, but results indicated a negative effect of accumulating virulence (Montarry et al., 2010). Watson & Singh (1952) carried out some of the earliest competition studies on cost of virulence in the stem rust pathogen, Puccinia graminis, where an isolate of a particular combination of virulences had a reduced fitness. In Blumeria graminis, another common biotropic fungus, Bronson & Ellingboe (1986) showed that fitness segregated independently of virulence, and for this pathogen it has also been suggested that avirulence can be lost without a fitness cost as a result of compensatory functions of avirulence paralogues (Ridout et al., 2006). Further, it has been discussed that compensatory mutations in other parts of the genome may alleviate a potential cost of virulence (Sacristán & García-Arenal, 2008).

The results obtained in this present study are in agreement with recent virulence phenotype dynamics in the northwest European populations of P. striiformis (Hovmøller et al., 2002; de Vallavieille-Pope et al., 2012). Virulence for Yr2 has been present in high frequencies in the northwest Europe population (e.g. wheatrust.org) despite the limited use of varieties with Yr2 resistance by farmers (Hovmøller, 2007). Restoration of Yr2 avirulence has been reported in Australia (Wellings & McIntosh, 1990). However, virulence for Yr2 is very common in most parts of the world (wheatrust.org; Sharma-Poudyal et al., 2013), so there is no indication of selection for a restoration of Yr2 avirulence under natural conditions. In Europe, restoration of avirulence at the phenotypic level has been reported for Yr32 and Yr9, but this process appears to be much less frequent than the loss of avirulence (Hovmøller & Justesen, 2007b). An increase in the number of virulences carried by individual P. striiformis isolates has been reported in many parts of the world (Yahyaoui et al., 2002; Chen, 2005; Enjalbert et al., 2005; Ochoa et al., 2007), suggesting that the potential cost of virulence in this pathogen is either insignificant or absent. However, the interpretation of data from virulence surveys should proceed with caution because of confounding factors, for example, non-random associations of alleles and a resulting genetic hitch-hiking. Quite strong non-random associations (gametic disequilibria) are often seen in agricultural systems, where strong and diverging selection forces are caused by crop varieties with different R genes present across large areas (Østergård & Hovmøller, 1991). Subsequent selection for one allele in such a system may result in unpredictable dynamics of other alleles as a result of hitch-hiking, i.e. decreasing frequencies if the linkage disequilibrium parameter is negative and increasing frequencies if the sign is positive. The linkage disequilibria are expected to emerge in sexual as well as asexual pathogen populations, although recombination is expected to reduce existing disequilibria. Recent discoveries have revealed that P. striiformis is capable of sexual reproduction on species of barberry and mahonia (Jin et al., 2010; Wang & Chen, 2013) and in some parts of Asia recombining populations have been found (Mboup et al., 2009; Duan et al., 2010; Ali, 2012). In other parts of the world, e.g. Europe and Australia, low genetic diversity may suggest mainly asexual reproduction (Wellings & McIntosh, 1990; Hovmøller et al., 2002; Enjalbert et al., 2005).

The parameters applied here have often been used to evaluate fitness of fungal plant pathogens (Pariaud et al., 2009) and to study the effect of resistance in host varieties (e.g. Cromey, 1992; Broers, 1997). Simulation studies using models for prediction of epidemic development have demonstrated that these parameters are highly important factors for disease severity (Luo & Zeng, 1995). However, it is important to notice that most studies about cost of virulence only deal with fitness parameters representing certain parts of a pathogen life cycle. The current study found no negative effect of loss of avirulence based on a number of epidemiological parameters, whereas other life history traits such as off-season survival and reproduction were not investigated. In some cases, looking at individual fitness parameters also has a limited scope for understanding the outcome of competition between isolates. This was demonstrated in a study on Potato virus Y where an avirulent strain had a competitive advantage, which could not be predicted by single inoculations (Janzac et al., 2010).

Most of the evaluated fitness components are highly influenced by environment (McGregor & Manners, 1985), which was evident in the present study by a significant effect of experimental replication. However, interactions between experiment and other factors were only observed in a few cases in this study and they did not affect the overall conclusions. The significant differences between isolates of different lineages demonstrate the utility and robustness of the methods applied. Similar or even larger differences in fitness between pathogen isolates representing different lineages have often been demonstrated (Pariaud et al., 2009). The significance of increased pathogen fitness is emphasized by the recent emergence and worldwide spread of two aggressive, and high-temperature tolerant, strains of P. striiformis which have caused serious epidemics in many wheat growing areas (Hovmøller et al., 2008; Milus et al., 2009).

In the present study it was possible to study the cost of virulence at the level of an individual gene independent of the genetic background. The results did not support the general hypothesis of cost of virulence but the study showed that difference in fitness exists even between isolates of related lineages of the same geographical origin. The effect of such differences on the dynamics in pathogen population is something that needs further evaluation.


The authors thank Dr Steffen Madsen, Institute of Biology, University of Southern Denmark for access to the microscope. They also thank Kristian Kristensen for statistical advice and Ellen Frederiksen and Steen Meier for expert technical assistance, and Lesley Boyd, John Innes Centre, Norwich, UK, for supplying isolate UK75/30. The development of virulence mutant isolates was supported by European Commission through the 6th Framework Programme, BIOEXPLOIT, and the subsequent research was funded by Aarhus University, Denmark.

Conflicts of interest

The authors have no conflicts of interest to declare.