Current theory suggests that cost–benefit relationships govern the evolution of parasite virulence. The cost of virulence is expected to be high for fungal viruses, which are obligate parasites and completely dependent on their hosts. The majority of fungal viruses infect their hosts without any apparent symptoms. Cryphonectria hypovirus 1 (CHV-1), in contrast, is virulent and debilitates its host, Cryphonectria parasitica. However, the virulence of CHV-1 is associated with high costs for virus transmission, such as an attenuated fungal growth and reduced production of the fungal spores spreading the virus. In this study, we tested the hypothesis that virulence may not only have costs but also benefits for transmitting CHV-1 across vegetative incompatibility barriers between fungi. We investigated viruses with low, medium, and high virulence, and determined their transmission rate per host-to-host contact (transmissibility). The average transmission rate across all combinations tested was 53% for the most virulent virus, 37% for the virus with intermediate virulence, and 20% for the virus with lowest virulence. These results showed that increased virulence was strongly correlated with increased transmissibility, potentially counterbalancing virulence costs. This association of virulence and transmissibility may explain why CHV-1 spread widely and evolved higher virulence than most other fungal viruses.

Why do parasites harm their hosts when—at the same time—they depend on hosts for survival and fitness? Current theory suggests that the evolution of virulence (i.e., harm to the host) depends on a cost–benefit relationship. The benefits of host exploitation on parasite multiplication within hosts are traded off against the cost of this exploitation for parasite transmission between hosts (Anderson and May 1982; Bull 1994; Frank 1996). Highly virulent parasites multiply rapidly and exploit their hosts efficiently, but at the cost of increased host mortality or debilitation, and thus limit the hosts’ ability to transmit the parasite.

Several models have been developed to describe the evolution of virulence in host–parasite systems (Levin and Pimentel 1981; Frank 1992; Antia et al. 1994; Day 2001; Koella and Restif 2001; Day 2002). These generally suggest that the optimal degree of virulence maximizes the parasite's fitness and, thus, depends on the particular cost–benefit relationship in the host–parasite interaction. Some parasites rely on relatively normal host functions for transmission and, hence, cannot be highly virulent, whereas other parasites may debilitate their host without strongly limiting transmission (Read 1994; Ewald 1995). Vertically (to offspring) transmitted parasites, for example, depend on the reproduction of their infected hosts. For them, the cost of reduced host reproduction is much higher than for horizontally (to other hosts) transmitted parasites, which do not require their host to reproduce (Bull et al. 1991). Similarly, the cost of host debilitation is much higher for parasites that cannot survive outside their hosts and depend on a live host than for parasites that are able to survive without their hosts (Walther and Ewald 2004).

Direct investigations of virulence and transmission are required to define the cost–benefit relationship within a host–parasite system. For instance, higher virulence was associated with reduced host fitness, but also with increased parasite replication, in: microsporidian parasites of the yellow fever mosquito (Agnew and Koella 1997), protozoan parasites of the monarch butterfly (De Roode et al. 2008), bacterial parasites of water fleas (Jensen et al. 2006), malaria parasites of mice (Mackinnon and Read 1999), myxoma virus of rabbits (Fenner 1983), and the human immunodeficiency virus (Fraser et al. 2007).

Fungal viruses are obligate parasites, which infect their host persistently (Nuss 2005). They generally lack an extracellular stage outside the host and therefore completely depend on their host for survival and transmission. According to the theory described above, the cost of their virulence is, thus, predicted to be high. Unfortunately, little research has been done on fungal viruses and this field is still in its infancy (Pearson et al. 2009). The vast majority of fungal viruses characterized so far, however, infect their host without any apparent symptoms (Nuss 2005). Cryphonectria hypovirus 1 (CHV-1) is an exception among these fungal viruses and causes marked and quantifiable symptoms in its host Cryphonectria parasitica (Milgroom and Cortesi 2004). This may explain why CHV-1 is one of the fungal viruses that has been best studied.

Cryphonectria parasitica is the causal agent of the tree disease chestnut blight, and hyperparasitation by CHV-1 results in debilitation of the fungus. CHV-1 does not kill C. parasitica but inhibits its sexual reproduction, strongly attenuates its growth and asexual sporulation, and reduces the pathogenicity of C. parasitica toward the chestnut tree (Milgroom and Cortesi 2004). CHV-1 is therefore used for the biological control of chestnut blight (Heiniger and Rigling 1994). Several subtypes of CHV-1 exist, which greatly differ in their virulence toward C. parasitica (Gobbin et al. 2003; Bryner and Rigling 2011), providing variation in virulence in this pathosystem.

One way to quantify the virulence of CHV-1 is by assessing the reduction in fungal growth (Peever et al. 2000; Bryner and Rigling 2011), which is a common and biologically relevant measure of virulence in fungal viruses (Pearson et al. 2009). In CHV-1, this growth reduction is highly correlated with other virulence factors, in particular with a reduction in asexual sporulation (Chen and Nuss 1999; Peever et al. 2000; Bryner and Rigling 2011). The spread of CHV-1 occurs mainly by transmission to virus-free individuals of C. parasitica (Hoegger et al. 2003). CHV-1 is transferred into the asexual spores of the fungus, dispersed in these spores and then transmitted from the outgrowing spores to other fungal individuals by hyphal fusion (Milgroom and Cortesi 2004). Virus-infected spores do not, however, seem to be a major source of new chestnut blight infections (Hoegger et al. 2003). The virus, thus, depends on the combination of vertical (into spores) and subsequent horizontal (from outgrowing spores to new hosts) transmission. Virus-infected individuals of C. parasitica are white, whereas virus-free individuals are orange when grown in the laboratory (Choi and Nuss 1992a). Therefore, the change from orange to white (Fig. 1) can be used to determine successful virus transmissions (Peever et al. 2000; Bryner and Rigling 2011).

Figure 1.

Morphological determination of virus transmission between vegetatively incompatible strains of Cryphonectria parasitica. Successful transmission of Cryphonectria hypovirus 1 (left): the virus-free recipient culture (orange) took on the color of the virus-infected donor culture (white). No transmission (right): the virus-free recipient culture remained orange and thus virus-free.

Similar to the other parasites cited above, virulent strains of CHV-1 have been shown to have a higher virus replication rate in the host and a higher virus-infection rate of asexual fungal spores (Suzuki et al. 2003; Lin et al. 2007). However, this occurs at a very high cost for CHV-1 because virulence is associated with the inhibition of fungal growth and of production of the asexual fungal spores carrying the virus (Milgroom and Cortesi 2004). This, therefore, raises the question why highly virulent strains of CHV-1 should be so prevalent (Taylor 2002; Milgroom and Cortesi 2004), in particular because the spreading of CHV-1 depends on the reproduction of its host. In accordance with the theory described above, a strong selective pressure favoring low virulence would be expected in CHV-1, especially under conditions where the transmission potential of CHV-1 is further reduced by the presence of increased transmission barriers within fungal populations (Milgroom 1995; Taylor 2002).

The transmission of CHV-1 is restricted by vegetative incompatibility between fungal host genotypes (Anagnostakis and Day 1979; Anagnostakis and Waggoner 1981; Cortesi et al. 2001; Choi et al. 2012). Vegetative incompatibility is a self/nonself recognition system in fungi that prevents transmission of viruses and other cytoplasmic elements (reviewed in Glass and Kaneko 2003). Vegetative incompatibility, thus, presents an additional transmission barrier in fungi. In C. parasitica, vegetative incompatibility is controlled by at least six biallelic unlinked vegetative incompatibility (vic) loci (Cortesi and Milgroom 1998), some of which have recently been characterized at the molecular level (Choi et al. 2012). Compatible individuals share the same alleles at all six vic loci, whereas incompatible individuals differ in their alleles at one or more loci. Hence, the vic haplotype defines the vegetative compatibility (vc) type (Cortesi and Milgroom 1998). Between individuals of C. parasitica that share the same vc type, CHV-1 is transmitted in virtually 100% of host-to-host contacts. Between different vc types, virus transmission sometimes occurs, but at reduced rates (Liu and Milgroom 1996; Cortesi et al. 2001; Papazova-Anakieva et al. 2008). The presence of these vegetative incompatibility barriers in fungi is another factor that has most likely promoted the evolution of low virulence in most fungal viruses (Milgroom 1999). In a recent study of CHV-1, however, no evidence for a trend toward lower virus virulence was found in fungal populations with high levels of vegetative incompatibility (S. F. Bryner and D. Rigling, unpubl. ms.). Virulence and prevalence of CHV-1 were equally high in C. parasitica populations with and without transmission barriers. This indicates that there must be certain benefits of virulence for transmission that allow CHV-1 to be virulent without compromising its effective transmission.

In the Cryphonectria literature, it has been commonly assumed that the rate of transmission per host-to-host contact (transmissibility) is not influenced by the virus strain (Cortesi et al. 2001). However, more virulent viruses might be more transmissible, and thus, have a higher transmission rate per host-to-host contact than less virulent viruses. This virulence benefit would counterbalance the virulence cost for transmission, that is, the lower number of asexual fungal spores carrying the virus would be counterbalanced by the higher infectivity of these spores. We therefore suggest clearly distinguishing between two main virus-dependent factors that potentially influence virus transmission: first, the number of virus-infected asexual spores that are produced by the fungus and that carry the virus to other hosts, and second, the rate at which the virus is finally transmitted after a successful host-to-host contact. Although the effect of virulence on asexual sporulation of the fungus is well known (Peever et al. 2000; Taylor 2002; Milgroom and Cortesi 2004; Bryner and Rigling 2011), its effect on virus transmissibility has remained unexplored.

The aim of this study was to find out more about the effect of virulence on transmissibility and how, if at all, it benefits virus transmission by counterbalancing the known virulence costs in CHV-1. To test the hypothesis that increased virulence is associated with increased transmissibility, we investigated three virus strains with different virulence (determined as the reduction in fungal growth), and assessed their transmissibility among fungal hosts.

Materials and Methods


We assessed the transmission rate of three virus strains between different pairs of fungal donor and recipient strains. The three virus strains represented three subtypes of CHV-1 (I, D, and F1;Gobbin et al. 2003), which are known to differ significantly in virulence (Chen and Nuss 1999; Peever et al. 2000; Robin et al. 2010; Bryner and Rigling 2011). The vc type (Cortesi and Milgroom 1998) of the C. parasitica isolates used as donor strains was EU-2. Virus-free recipient strains of three different vc types, all heteroallelic to vc type EU-2 at one vic locus, were used: EU-4 heteroallelic at vic1, EU-1 heteroallelic at vic2, and EU-14 heteroallelic at vic6. To include some genetic background variation in the fungal host, we used two genetically different isolates originating from different populations as donor strains (isolate “baw” from Germany and isolate “var” from France; Bryner and Rigling 2011), and two different isolates for each of the three vc types as reciepent strains (isolates with suffix “A” from Italy and isolates with suffix “B” from Switzerland; Cortesi et al. 1998). As a control, we also assessed virus transmission rates between vegetatively compatible donor and recipient strains, namely from the virus-infected donor strains “baw” and “var” two virus-free cultures of “baw” and “var.”


The transmission rate of the three viruses was assessed using a previously described method (Liu and Milgroom 1996; Cortesi et al. 2001; Papazova-Anakieva et al. 2008). The six donor cultures (two fungal strains infected with each of the three viruses) were paired with the six recipient cultures (two fungal strains from each of the three vc types), giving a total of 36 different combinations, plus the vegetatively compatible control pairings. For each combination, 25 replicates were tested. A 9-cm petri dish (84-mm inner diameter) was used, containing 25 mL of potato dextrose agar (PDA; Difco Laboratories, Detroit, MI) for each pairing of a virus-infected donor with a virus-free recipient culture. Plates were inoculated with two mycelial plugs (6-mm diameter) from the growing edge of five-day-old precultures, one from the virus-infected donor strain and one from the virus-free recipient strain. The two plugs were placed approximately 4 mm apart and approximately 5 mm from the edge of the plate. The cultures were incubated at 24°C and 70% relative humidity for seven days in the dark and subsequently for 14 days under low light conditions. After 21 days, virus transmission was assessed and double checked by a second person. In a few ambiguous cases, the success or failure of virus transmission was verified by subculturing the recipient culture. Virus transmission was considered successful if the recipient culture took on the morphology and the color of the donor, that is, unpigmented, white mycelium (Fig. 1). Successful transmissions were coded as “1” and unsuccessful transmissions as “0,” which resulted in a binary dataset.


To obtain a measure for the virulence of the three virus strains, we assessed their effect on the growth of the two fungal donor strains “baw" and “var" under the same experimental conditions as we assessed transmission, that is, at 24°C and 70% humidity. The same stock cultures as in the virus transmission experiments were used. We performed two virulence experiments with identical design, one on PDA and one on dormant chestnut stems. Although virulence experiments on dormant chestnut stems provide a more natural setting to obtain measures of virus virulence, experiments on PDA can be better controlled (higher standardization). The power to detect significant differences in virulence among viruses is therefore much higher in experiments on PDA (Bryner and Rigling 2011). However, on PDA, C. parasitica cultures infected with viruses of low virulence often grow faster than virus-free cultures (Chen and Nuss 1999; Chen et al. 2000; Robin et al. 2010; Bryner and Rigling 2011). Assessments on PDA can therefore only be used for comparisons among viruses and not to obtain absolute measures of virus virulence (Bryner and Rigling 2011).

Here, we used six replicates of each fungus-virus combination and six replicates of the virus-free cultures of both fungal strains in each experiment. Replicates were assigned to one of six blocks and randomized within blocks. The virulence of each virus strain on each fungal strain was determined by calculating the growth difference between the virus-infected and the virus-free culture of the same fungal strain. The difference (i.e., the virus effect) was given in proportion (%) to the growth of the virus-free culture, indicating the percentage to which the growth of the fungus was changed due to virus infection.

In the experiment on PDA, we used an individual sterile 9-cm petri dish (84-mm inner diameter), containing 25 mL of PDA for each fungal colony. The center of the plates were inoculated with a mycelial plug (6-mm diameter) taken from the growing edge of five-day-old precultures. The plates were wrapped with Parafilm, arranged in adjacent blocks on a shelf and illuminated at approximately 3330 lx (Illuminance Meter, Minolta, Japan) for a 14-h photoperiod. We measured two cardinal diameters of each colony through two orthogonal axes previously drawn on the bottom of each plate after two, three, and four days at 24-h intervals. As the shape of the colonies was not a perfect circle, we calculated the geometric mean diameter of an ellipse. Linear regression implemented in Microsoft Excel 2007 was performed to determine the rate of linear growth on PDA (increment of the fitted regression line, R2= 0.980).

For the experiment on dormant chestnut stems, healthy C. sativa stems (50 cm length, 5–10 cm diameter) were cut from six different chestnut sprout clusters in Ticino (Switzerland) after the end of the vegetation period, a few days before the start of the experiment. Stems originating from the same sprout cluster (i.e., isogenic stems) were put into the same container for the experiment. Six opaque plastic containers (57 cm × 37 cm × 13 cm) were used, each representing one of six experimental blocks. The ends of the stems were sealed with paraffin and four circular wounds were made along the axis of the stem. The wounds were arranged 12 cm apart from each other and 7 cm from the two ends. Each was filled with two mycelial mats (6-mm diameter) obtained from the growing edge of the five-day-old precultures. After 18 days of incubation, we determined the lesion diameter on the chestnut stems, using a millimeter scale. Two diameters of each lesion were measured, one along the longitudinal and a second along the lateral axis of the stem. As the shape of the lesions resembled an ellipse, we calculated the geometric mean diameter.


To describe virus transmission, we used a logistic regression model. This is a common model for describing the presence or absence of a trait (i.e., binary data), and has previously been applied to describe the probability of virus transmission in C. parasitica (Cortesi et al. 2001). We used the generalized linear mixed model procedure in R version 2.6.2 (R Development Core Team (2008), function glmer with family binomial from the lme4 package) to fit the logistic regression. The probability of virus transmission between pairs of fungal isolates was modeled as a function of the virus and the vic locus at which the fungal donor and recipient isolates were heteroallelic. The effects of the virus and the vic locus were therefore included as fixed factors. The effects of the donor and recipient, on the other hand, were defined as genetic background effects, and thus, included as random factors. Recipient was nested within heteroallelic vic locus. As a base model, the following logistic regression was employed:



  • pijkl is the transmission probability of virus i from fungal donor k to fungal recipient l, which are heteroallelic at vic locus j;

  • μ is the intercept of the logistic regression;

  • βi is the effect of virus i, i= F1, D, or I;

  • γj is the effect of the vic locus j at which the fungal donor and recipient isolates are heteroallelic, j=vic1, vic2, or vic6;

  • dk is the effect of the fungal donor isolate k, N(0, σ2d);

  • rl(j) is the effect of the fungal recipient isolate l within the heteroallellic vic locus j, N(0, σ2r).

The function anova implemented in R was employed to identify significant model terms. The base model (Table 1, Model 0) was compared to several simpler models with fewer factors (Table 1, Models 1–4), and to a model that also included an interaction effect between virus and vic locus (Table 1, Model 5). The model with the lowest Akaike's Information Criterion (AIC) value was preferred and applied to estimate the parameter values.

Table 1.  Comparison of the base model (0) with alternative logistic regression models describing virus transmission.
No.ModelAICEst. dfχ2
  1. *P≤ 0.05; **P≤ 0.01; ***P≤ 0.001; ****P≤ 0.0001.

0μ+βij+dk+rl(j)420.80 7N.A.
1μ+γj+dk+rl(j)608.88 5192.08****
2μ+βi+dk+rl(j)428.80 5 12.004**
3μ+βij+rl(j)435.99 6 17.195****
4μ+βij+dk488.00 6 69.198****
5μ+βij+dk+rl(j)+ (β×γ)ij426.5311 2.2684

To analyze the virus effect (virulence), we applied a general linear model (GLM) in SPSS 19.0 (SPSS, Somers) with the fixed factors “Virus,”“Fungus,” and “Virus × Fungus” and with “Block” as a random factor. Tukey's test implemented in the GLM was performed to detect significant (at α≤ 0.05) differences among viruses. Pearson's correlation coefficients were calculated in SPSS to test for a linear relationship between the virus effects measured on PDA and on dormant chestnut stem, as well as between the virus effects and virus transmissibility.



The lowest AIC value was reached using the base model (Model 0, Table 1), and it was therefore considered the appropriate model to use for parameter estimation. The chi-square values of the comparisons between the base model and the models that included fewer factors (Models 1–4, Table 1) revealed that these simpler models were significantly different from the base model. This indicated that all factors in the base model, that is, virus, vic locus, donor, and recipient, had a significant effect on the probability of virus transmission. The model that included an interaction effect between the virus and vic locus (Model 5, Table 1) was not significantly different from the base model. This indicates that there was no significant interaction and that the effect of the virus was independent of the vic locus at which the donor and recipient isolates were heteroallelic.


Logistic regression revealed that the probability of virus transmission significantly differed among all three viruses tested (Table 2). In the model, virus D was used as the reference category for the term virus, and its parameter value was therefore set to 0. The transmission probability for this virus was intermediate, whereas the transmission probability of virus F1 was significantly higher and the transmission probability of virus I significantly lower. Table 3 shows the transmission rate of each virus for each heteroallelic vic locus between each donor and recipient strain. Averaged over all vic loci and all combinations of donor and recipient strains, the mean transmission rate was 53% for virus F1, 37% for virus D, and 20% for virus I (Table 3).

Table 2.  Estimates of the logistic regression model for virus transmission probability. Transmission probability was estimated for the three viruses representing one of the Cryphonectria hypovirus 1 subtypes D, F1, and I between different pairings of Cryphonectria parasitica strains heteroallelic at one vegetative incompatibility (vic) locus.
TermParameterEstimateStd. errorZ
  1. *P≤ 0.05; **P≤ 0.01; ***P≤ 0.001; ****P≤ 0.0001.

Vic locusγvic60N.A.N.A.
Table 3.  Virus transmission rates. Transmission rates of three virus strains were calculated, each representing one of the Cryphonectria hypovirus 1 subtypes F1, D, and I, between different pairings of Cryphonectria parasitica strains heteroallelic at one vegetative incompatibility (vic) locus.
VirusHeteroallelic vic locusFungal donorFungal recipientNo. transmissionsNo. trialsTransmission rateAverage transmission rate
F1vic1varEU-4 A2525100% 
  varEU-4 B2525100% 
  bawEU-4 A2525100% 
  bawEU-4 B2525100%100%
 vic2varEU-1 A1425 56% 
  varEU-1 B 625 24% 
  bawEU-1 A 320 15% 
  bawEU-1 B 020 0% 24%
 vic6varEU-14 A 625 24% 
  varEU-14 B1420 70% 
  bawEU-14 A 319 16% 
  bawEU-14 B 414 29% 35%
Dvic1varEU-4 A2525100% 
  varEU-4 B2425 96% 
  bawEU-4 A2525100% 
  bawEU-4 B2125 84% 95%
 vic2varEU-1 A 325 12% 
  varEU-1 B 125 4% 
  bawEU-1 A 125 4% 
  bawEU-1 B 125 4% 6%
 vic6varEU-14 A 2258% 
  varEU-14 B 725 28% 
  bawEU-14 A 025 0% 
  bawEU-14 B 025 0% 9%
Ivic1varEU-4 A2325 92% 
  varEU-4 B 625 24% 
  bawEU-4 A2425 96% 
  bawEU-4 B 625 24% 59%
 vic2varEU-1 A 025 0% 
  varEU-1 B 025 0% 
  bawEU-1 A 025 0% 
  bawEU-1 B 0250% 0%
 vic6varEU-14 A 025 0% 
  varEU-14 B 018 0% 
  bawEU-14 A 025 0% 
  bawEU-14 B 021 0% 0%

The probability of virus transmission also depended on the heteroallelic vic locus. In the model, vic6 was used as the reference category for the term vic locus, and its parameter value was therefore set to 0 (Table 2). Heteroallelism at vic2 compared to heteroallelism at vic6 did not have a significantly different impact on transmission probability. However, transmission probability was significantly higher when vic1 was heteroallelic. Averaged across all three viruses and all combinations of donor and recipient strains, the mean transmission rate was 85% when vic1 was heteroallelic, 15% when vic6 was heteroallelic, and 10% when vic2 was heteroallelic (Table 3). In the control experiment, where the same two fungal strains were used as donors and as recipients, the transmission rate was 100% for all pairings. To some extent, the probability of virus transmission was also influenced by the genetic background of the donor and recipient strains sharing the same vc type. The estimated variances attributable to specific donor and recipient strains were greater than zero (Table 2).


GLM results are displayed in Table 4. The factor “Virus” had a significant effect on the virus effect, but none of the other factors did. In both experiments, on PDA and on dormant chestnut stems, virus F1 had the strongest effect on fungal growth, virus I the mildest effect, and virus D an intermediate effect (Fig. 2), as expected. The virulence measurements on PDA and on dormant chestnut stems were highly correlated (r= 0.914, P≤ 0.05). Differences in virulence among all three viruses were significant in the experiment on PDA. In the experiment on dormant chestnut stems, virus I was significantly less virulent than the other two viruses, whereas the difference in virulence between virus D and F1 was not significant at α≤ 0.05.

Table 4.  General linear models on virus virulence. The effect of virus infection on fungal growth on potato dextrose agar (PDA) (growth rate in colony diameter during the phase of linear growth) and on dormant chestnut stems (lesion diameter after 18 days of incubation) were assessed. The virus effect is the difference in growth between the virus-infected and the corresponding virus-free strain of Cryphonectria parasitica as a proportion (%) of the virus-free strain. In the model “Virus” (n= 3), “Fungus” (n= 2), and “Fungus × Virus” (n= 6) were used as fixed factors. “Block” (n= 6) was included as a random term.
SourcedfGrowth rate on PDALesion diameter on dormant chestnut stems
Fungus 10.008 2.1970.1510.061 0.9300.344
Virus 21.067294.292≤0.0011.77527.178≤0.001
Fungus×Virus 20.003 0.7780.4700.080 1.2330.309
Block 50.005 1.4910.2280.005 0.8390.353
Error250.004  0.065  
Figure 2.

Effect of three Cryphonectria hypovirus 1 strains, F1, D, and I, on the growth of the two Cryphonectria parasitica strains var (open circles) and baw (closed circles): (A) on potato dextrose agar (growth rate in colony diameter during the phase of linear growth) and (B) on dormant chestnut stems (lesion diameter after 18 days of incubation). The virus effect is the difference in growth between the virus-infected and the corresponding virus-free strain of C. parasitica as a proportion (%) of the virus-free strain. Error bars represent standard errors (n= 6).


Virulence and transmissibility were highly correlated (P≤ 0.05) at r=−0.870 when virulence was assessed on PDA, and at r=−0.854 when virulence was assessed on dormant chestnut stems, as displayed in Figure 3. The sign of the correlation was negative as virulence was expressed by the reduction in fugal growth due to virus infection (negative value). The highest transmissibility was obtained with virus F1, which had the highest virulence. The lowest transmissibility was obtained with virus I, which had the lowest virulence. Finally, virus D was intermediate for both transmissibility and virulence. These differences in transmissibility were consistent across all pairings of fungal donor and recipient strains (Table 3), and thus independent of the degree of vegetative incompatibility between fungal hosts strains (Fig. 3).

Figure 3.

The virus transmission rate (transmissibility) plotted in relation to the virus effect (virulence) on fungal growth. Here, the virus effect on fungal growth on dormant chestnut stems is shown. The assessments on dormant chestnut stems and on potato dextrose agar were highly correlated (r= 0.914, P≤ 0.05). The transmission rate is the proportion (%) of successful transmissions. The virus effect is the difference in growth between the virus-infected and the corresponding virus-free strain of Cryphonectria parasitica as a proportion (%) of the virus-free strain. Transmission was assessed under different degrees of vegetative incompatibility between the fungal virus donor and recipient, that is, when the donor and recipient were heteroallelic at the vegetative incompatibility locus vic1 (diamonds, dashed trend line), vic2 (triangles, dotted trend line), and vic6 (circles, solid trend line), respectively.


The results of this study suggest that virulence has not only costs for transmission of the fungal virus CHV-1 but also benefits. In the present experiment, more virulent viruses, which strongly debilitated the fungus, had higher rates of transmission per host-to-host contact (higher transmissibility). Unlike previously assumed (Cortesi et al. 2001), transmissibility was strongly influenced by the virus strain.

So far, most research on virus transmission in the Cryphonectria-hypovirus pathosystem has focused on the impact of vegetative incompatibility between fungal individuals (Liu and Milgroom 1996; Cortesi et al. 2001; Papazova-Anakieva et al. 2008). Cortesi et al. (2001) established a logistic regression model to predict the transmission of CHV-1. According to their model, the probability of virus transmission depends primarily on the vic loci at which the fungal donor and the fungal recipient are heteroallelic and to a minor extent also on the genetic background of the donor and the recipient strains. Our study confirmed the effect of the heteroallelic vic loci and of the genetic background of donor and recipient. In addition, we extended the model of Cortesi et al. (2001) by including virus strain as a factor. We investigated three virus strains that significantly differed in virulence, and assessed their transmissibility between different combinations of fungal donor and recipient strains (heteroallelic at one vic locus in all cases). The model selection showed that the factor virus had a significant impact on transmission probability, and that there was no interaction between virus and heteroallelic vic locus. The three virus strains tested not only significantly differed in virulence, but also in transmissibility: higher virulence was strongly associated with increased transmission rate per host-to-host contact. This is the first time that the possibility that high virulence could be associated with higher transmissibility in CHV-1 has been considered.

The negative effects of virulence on the spread and persistence of CHV-1 have, however, been widely discussed (Milgroom 1995, 1999; Taylor 2002; Milgroom and Cortesi 2004; Robin et al. 2010). Infection with CHV-1 debilitates C. parasitica and reduces its growth and asexual sporulation. At the same time, CHV-1, like all fungal viruses, relies completely on fungal growth and sporulation for spread and transmission (Taylor 2002). In debilitating its host, CHV-1 should therefore directly reduce its own transmission potential. Based on these considerations, various mathematical models (Morozov et al. 2007; Brusini et al. 2011) and empirical observations (Robin et al. 2010) suggest that the ability of CHV-1 to establish and/or persist in populations decreases with increasing virulence. Our study, however, indicates that this scenario is too simplistic.

Our findings suggest that the positive effects of virulence in CHV-1 on virus replication in the fungus and on the infection rate of asexual fungal spores are not its only benefits. Virulence also seems to have positive effects on the virus transmission between hosts, and we found that virulent strains of CHV-1 had a higher transmission rate per host-to-host contact. These positive effects on transmission may counterbalance the known negative effects of high virulence on virus transmission between hosts, that is, inhibiting fungal growth and the production of asexual fungal spores.

The correlation between virulence and transmissibility we detected may have to do with the process of virus transmission, which occurs after hyphal fusion between two fungal individuals (Nuss 2005). However, when vegetatively incompatible individuals (i.e., individuals that are hereoallelic at one or more vic loci) fuse, cell death is induced by the fungal self/nonself recognition system (Glass and Kaneko 2003). A comparative genomic study in C. parasitica recently confirmed the functional role of the vic loci for restricted virus transmission between incompatible individuals (Choi et al. 2012).

Virus transmission depends on the strength of the incompatibility reaction and negatively correlates to the rate at which cells die after fusion (Biella et al. 2002). CHV-1 moves rapidly within the fungal mycelium (Martin and Van Alfen 1991), and may thus cross-hyphal bridges between fungal individuals before the fused cells collapse. Furthermore, it has been shown that infection with CHV-1 can influence the frequency of cell death after hyphal fusion (Biella et al. 2002). Given the finding that transmissibility increased with greater virulence, virulent viruses may interfere with the fungal self/nonself recognition and delay the initiation of cell death, thus increasing their chances of being transmitted between incompatible individuals. The possibility that CHV-1 may interact with antiviral defense strategies in the fungus (such as the vegetative incompatibility reaction) is also mentioned by Choi et al. (2012). In addition, the chances of transmission of virulent viruses might be even further increased by their typically higher replication rate (Suzuki et al. 2003; Lin et al. 2007) and the resulting higher virus concentration in the hyphal tissue.

In our experiment, the transmission rate between vegetatively compatible hosts was 100% for all viruses, irrespective of their virulence. Between vegetatively incompatible hosts, however, the average transmission rate was 53% for the most virulent virus, 37% for the virus with intermediate virulence, and 20% for the virus with lowest virulence (Table 3 and Fig. 3). These differences in transmissibility between the virus strains were consistent across all combinations of vegetatively incompatible fungi tested. Virulent strains of CHV-1 may, thus, not only impair the general fitness of C. parasitica and negatively impact transmission, but also impede the incompatibility reaction and, thus, positively impact transmission.

This hypothesis corresponds with a more mechanistic view of virulence evolution, according to which the virulence–transmission relationship is determined by the mechanisms through which the parasite interferes with host functions (Frank and Schmid-Hempel 2008). A study in bacterial parasites of mice revealed that the same bacterial effector proteins that caused morbidity and mortality in mice were required for the transmission of bacteria (Wickham et al. 2007). Virulent bacterial strains, which possessed these effector proteins, were therefore favored. Similar coupling of transmission with virulence was found in a nucleopolyhedrovirus of caterpillars (Szewczyk et al. 2006).

To explore the potential for the coupling of virulence and transmissibility in CHV-1, the differential impact of CHV-1 strains with different virulence on the frequency of cell death after hyphal fusion could be assessed with light microscopy (Biella et al. 2002). This was attempted recently in a yet unpublished study at the laboratory of Myron Smith in Ottawa, Canada. Smith et al. discovered that the highly virulent virus CHV-1/EP713 suppressed cell death more than the less virulent virus CHV-1/Euro7 (M. Smith, pers. comm.). Further studies, both on the genomic and on the phenotypic level, should be conducted to understand how these different virulence factors are associated in CHV-1. The genomic regions that enable CHV-1 to affect the incompatibility reaction in C. parasitica may be determined by specific gene mutations (Choi and Nuss 1992b).

Findings from the above-mentioned study on cell death and from our study both indicate that virulence factors, such as the reduction of growth and asexual sporulation of C. parasitica, are coupled with enhanced horizontal virus transmission in CHV-1. This would explain why no trend for virulence to be reduced was found when virus transmission barriers in fungal populations were high (S. F. Bryner and D. Rigling, unpubl. ms.). Previously, it was assumed that transmission barriers (i.e., high levels of vegetative incompatibility in C. parasitica populations) had a stronger impact on virulent viruses, and would thus select against high virulence (Milgroom 1995, 1999). However, if the negative effects of high virulence on fungal growth and sporulation are counterbalanced by higher virus transmissibility across vegetative incompatibility barriers, there may be no selective disadvantage of highly virulent viruses. This would also agree with the observation of high variability in virulence within CHV-1 populations, irrespective of the presence or absence of transmission barriers (S. F. Bryner and D. Rigling, unpubl. ms.).

These findings have positive implications for the biological control of C. parasitica with CHV-1. They suggest that strains of CHV-1 with increased virulence, which substantially reduces the pathogenic potential of C. parasitica toward the chestnut tree, are able to spread and persist in natural populations and exert an effective control of the chestnut blight disease. However, to exactly define the relationship between transmissibility and virulence, further transmission experiments with more viruses that only gradually differ in virulence and transmissibility would be required. In any case, virulence might not increase infinitely. A minimum production of asexual fungal spores might be required for sufficient virus dispersal and for subsequent transmission from the spores to other fungal individuals (Taylor 2002).

Viruses of CHV-1 subtype I, the subtype with the lowest virulence, have successfully spread within and among populations and have thus become very abundant throughout Europe (Gobbin et al. 2003; Milgroom and Cortesi 2004). Viruses of the more virulent subtype D in Germany have been observed to also persist over several decades (F. Peters and B. Metzler, pers. comm.). However, viruses of subtype F1, the subtype with the highest virulence, could not be recovered from the area where they had artificially been released 20 years earlier in France (Robin et al. 2010). Nevertheless, viruses of subtype F1 have repeatedly been found in France (C. Robin, pers. comm.), and also in Spain (Montenegro et al. 2008) during surveys. Therefore, the range of virulence exhibited by viruses of subtype F1 may be just around the mentioned virulence threshold, which could explain why some viruses persist while others do not.

Our study highlights the importance of understanding if and how virulence factors are coupled in parasites. Coupling of virulence factors is likely to shift the cost–benefit relationship of virulence, and thus affect the optimal degree of virulence that will evolve. CHV-1 belongs to the family Hypoviridae, which consists of the single genus Hypovirus (Nuss and Hillman 2011; Rigling and Hillman 2012). Three other members of that genus—CHV-2, CHV-3, and CHV-4—have been described. All hypoviruses are parasites of C. parasitica, but they differ in genome size and organization (Hillman and Suzuki 2004). In contrast to CHV-1, the other three hypovirus species have not been very successful as biocontrol agents, either due to their low virulence and limited ability to reduce the pathogenic potential of C. parasitica (CHV-3 and CHV-4), or due to a lack of ecological fitness and of spread (CHV-2) (Milgroom and Cortesi 2004). The hypothesized coupling of host debilitation with the ability to inhibit cell death upon hyphal fusion in CHV-1 could perhaps explain why CHV-1 has evolved much higher virulence than most other fungal viruses (including CHV-3 and CHV-4), and also why it has spread widely (unlike CHV-2). In the absence of such coupling, virulent viruses should be ecologically less fit than less virulent ones and may therefore not become widely established. This might be the case for the fungal viruses that have evolved low virulence and cause no symptoms in their host.

In conclusion, our study provides strong evidence for an association of high virulence with high transmissibility in the fungal virus CHV-1. This finding indicates that virulence may have negative effects on one factor influencing virus transmission between hosts (the production of virus-infected spores), but also positive effects on another factor influencing this transmission (transmission rate per host-to-host contact, i.e., transmissibility). We suggest that virulence is coupled with transmissibility in CHV-1, and that virulent viruses might interfere with the fungal self/nonself recognition system, and thus facilitate virus transmission between vegetatively incompatible fungi. Nevertheless, virulence may only increase up to a certain threshold above which dispersal, and eventually also transmission, are inhibited. Overall, the results of this study are in line with previous suggestions (Read 1994; Weiss 2002; Ebert and Bull 2003; Frank and Schmid-Hempel 2008), which indicate that the biological specificities of the host–parasite interaction, and the associated cost–benefit relationship in particular, are keys for understanding the evolution of parasite virulence.

Associate Editor: A. Read


We thank R. Holderegger, D. Refardt, S. Alizon, Ch. Rellstab, D. Keller, D. Bühler, and two anonymous reviewers for helpful comments on the manuscript. We acknowledge A. Drewek for statistical advice and S. Dingwall for an English revision of the manuscript. This study is part of the research project “The Role of Genetic Diversity in Host-Pathogen Interactions in Dynamic Environments” (GEDIHAP), funded by the Competence Center Environment and Sustainability (CCES) of the Eidgenössische Technische Hochschule Domain.