SEARCH

SEARCH BY CITATION

Keywords:

  • adaptive evolution;
  • host–parasite coevolution;
  • obligate killer;
  • optimal killing;
  • Red Queen hypothesis;
  • resistance;
  • virulence

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Standard epidemiological theory predicts that parasites, which continuously release propagules during infection, face a trade-off between virulence and transmission. However, little is known how host resistance and parasite virulence change during coevolution with obligate killers. To address this question we have set up a coevolution experiment evolving Nosema whitei on eight distinct lines of Tribolium castaneum. After 11 generations we conducted a time-shift experiment infecting both the coevolved and the replicate control host lines with the original parasite source, and coevolved parasites from generation 8 and 11. We found higher survival in the coevolved host lines than in the matching control lines. In the parasite populations, virulence measured as host mortality decreased during coevolution, while sporeload stayed constant. Both patterns are compatible with adaptive evolution by selection for resistance in the host and by trade-offs between virulence and transmission potential in the parasite.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Parasites are a major selective force shaping several aspects of host populations, such as the degree of genetic diversity (Haldane, 1949), and cyclic population dynamics (Hudson et al., 1998; Little, 2002). On a higher level, parasites also affect biodiversity, ecosystem functioning and community structure (Hudson et al., 2006). Often, the underlying ecological interactions lead to persistent antagonistic coevolution between hosts and parasites (Woolhouse et al., 2002), which entails constant reciprocal selection for resistant host and infective parasite genotypes. Such a coevolutionary process is also thought to lead to local adaptation of both parties because the exact ecological conditions vary among sites and the course of evolution therefore takes its specific route. Parasites typically have multiple generations per host and reach large population sizes, making it more probably that the parasite is locally adapted to its host rather than vice versa (Kaltz & Shykoff, 1998; Gandon & Michalakis, 2002; Greischar & Koskella, 2007). Data shows that this is not always the case, with cases of local mismatch between host and parasite (Imhoof & Schmid-Hempel, 1998) and a number of studies showing no adaptation of either host or parasite (Unruh & Luck, 1987; Paterson, 2005; Greischar & Koskella, 2007).

A measure, which is often used to quantify parasite adaptation, is virulence. When using virulence it is difficult to conclude that parasites are locally adapted, as parasites are faced a trade-off between different components of parasite fitness, notably between virulence and transmission (Anderson & May, 1982; Frank, 1996). For example, higher parasite numbers within the host would increase transmission at the expense of shortening the lifespan of the infection by damaging the host (virulence) (Day, 2001). Intuitively, it seems that this model does not hold for obligate killers where transmission is mechanically coupled to host death and transmission stages are only released after a parasite killed its host. The parasite potentially pays a large cost for not being virulent enough to kill its host and so enable its own transmission. Fitness might be maximized with early transmission when the parasite kills its host rapidly, but optimal virulence will ultimately depend on the background mortality of the host (Ebert & Weisser, 1997) and within-host multiplication for the production of transmission stages. Indeed, it has been shown experimentally that parasite fitness (spore production in an obligately killing bacterium, Pasteuria ramosa of Daphnia magna) was the highest for intermediate levels of virulence (measured as time to death) (Jensen et al., 2006).

In this study, we want to combine both of the above aspects and investigate virulence evolution of an obligately killing parasite in relation to local adaptation of both host and parasites in the course of experimental antagonistic coevolution using the time-shift method (Buckling & Rainey, 2002; Decaestecker et al., 2007; Koskella & Lively, 2007; Gaba & Ebert, 2009). We used the parasite Nosema whitei (Milner, 1972a) and its natural host, the Red Flour Beetle Tribolium castaneum, as a model system for experimental evolution. Nosema whitei infects early larval stages of the beetle that feed on medium (i.e. flour) containing spores (Milner, 1972a; Blaser & Schmid-Hempel, 2005). Nosema whitei is an obligately intracellular parasite that infects internal organs, such as the fat body, and generally causes the host to die in the late larval or early pupal stage (Dunn & Smith, 2001; Blaser & Schmid-Hempel, 2005). Nosema whitei is considered an obligate killer, since transmission depends on host death such that spores are released to the surrounding environment after the host has died and its body decayed, or when the dead larva (or adult) is cannibalized by others. In fact, dead hosts are highly enriched with spores (Blaser & Schmid-Hempel, 2005). In the rare cases where infected larvae manage to pupate and eclose, the adults usually have lower mating frequency, reduced longevity and up to 50% lower fecundity (Milner, 1972a, b; Armstrong & Bass, 1986; Blaser & Schmid-Hempel, 2005).

In recent studies, it was shown that both resistance to N. whitei and infection intensity (spore load) is governed by nonadditive components, epistatic interactions and maternal effects, in single and multiple infections (Wegner et al., 2008, 2009), fulfilling necessary prerequisites for antagonistic coevolution. Ongoing selection by coevolving N. whitei leads to increased recombination rates in the host (Fischer & Schmid-Hempel, 2005), which lends support to the Red Queen hypothesis as it relates to the evolution and maintenance of sex. It remains unknown how and to what extent the traits crucial for such a scenario of antagonistic coevolution, i.e. parasite virulence and host resistance, evolve under these conditions. Note that the ability of a host to withstand the effects of infection can be based on resistance (e.g. clearing the infection) or on tolerance (the ability to buffer the negative effects; Raberg et al., 2007). We refer to ‘resistance’ throughout because the experiments were not designed to test for this difference explicitly. Furthermore, tolerance is an ambiguous concept for an obligate killer such as N. whitei.

We here quantify the extent of coevolutionary change in this system by comparing virulence (reduction of host survival) of original parasites (i.e. samples from the start of the experiment) with coevolved parasites (i.e. after a history of selection) on, both, their sympatric coevolved host lines (i.e. after a history of selection) and noncoevolved host lines (i.e. controls that have not encountered the parasite) from the same genetic background. Assuming that hosts can evolve higher resistance when exposed to parasites, we expect that coevolved host lines will show higher survival rates than control host lines when exposed to parasites. Vice versa, assuming that parasites can adapt to their hosts by becoming more virulent with a history of selection, we expect that parasites adapted to their ‘own’, sympatric coevolved hosts (i.e. those host lines they coevolved with) show similar or higher virulence on these host lines compared to the virulence of the original parasites (parasites from the start of the experiment). By comparison, the virulence of the coevolved parasites should be lower when infecting foreign hosts (control beetles that have not encountered the parasite). To check for a trade-off between virulence and infection intensity (an indicator of transmission potential, i.e. parasite fecundity as a token of parasite fitness), we also checked whether in this obligate killer fitness is simply maximized with higher virulence? These questions are crucial to understand to what extent hosts and parasites are able to adapt through antagonistic coevolution and to what extent trade-offs between virulence and transmissive potential influence the evolutionary trajectory in the parasite.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Experimental evolution protocol

Naïve beetles from seven natural isolates of T. castaneum (stock line nr. 32: San Bernadino, U.S.A., nr. 37:Puranpur, India, nr. 39: Beer Sheva, Israel, nr. 41: Palmira, Colombia, nr. 42: New South Wales, Australia, nr. 43: Kyushu Island, Japan, nr. 44: Chiang Mai, Thailand) were used in the experiment. These lines had been kept isolated in large population sizes (> 200 unsexed adults) on standard medium (type 550 ‘Knospe’ organic flour containing 5% dried yeast) under standard environmental conditions (24 h dark, 32 °C, 70% humidity) for at least 50 generations. To increase genetic variability of the hosts we first set up population crosses between these stock lines (Table 1) by mass-matings involving 50 virgin females from one line and 50 virgin males from the respective partner line, with some stock lines used in more than one cross. To generate a given experimental line, reciprocal crosses were set up with equal numbers, to balance maternal effects in the offspring, and the resulting fully hybrid F1-adults were pooled as the starting breeder population of each line. A total of eight such populations were used to form eight experimental lines (nrs. 1–8, c.f.Table 1). Furthermore, one half of the beetles of each line were then assigned to control treatment and the other half to coevolution treatment; thus control and coevolution treatments were paired within each experimental line, defining the same genetic background for a given pair.

Table 1.   Crossing scheme to generate the experimental lines.
Experimental lineStock line 1OriginStock line 2Origin
  1. All crosses were set up as mass matings between 50 males from stock line 1 and 50 females from stock line 2, and the reciprocal cross with equal numbers. Some stock lines have been used in more than one cross. The resulting fully hybrid offspring have been used to start the experiment, and were subjected to both treatments in a paired design.

132San Bernardino, USA37Puranpur, India
232San Bernardino, USA43Kyushu Island, Japan
332San Bernardino, USA44Chiang Mei, Thailand
437Puranpur, India43Kyushu Island, Japan
537Puranpur, India44Chiang Mei, Thailand
643Kyushu Island, Japan44Chiang Mei, Thailand
739Beer Sheva, Israel42New South Wales, Australia
842New South Wales, Australia41Palmira, Colombia

We applied two selection regimes: coevolution and control. In the coevolution treatment the experimental lines were subjected to selection by coevolving N. whitei (see below), whereas the same lines in the control treatment were raised on standard medium free of the parasite. We used the same mixture composed of equal spore numbers from eight different N. whitei isolates to infect all lines in generation 1 of the coevolution experiment and to ensure a sufficient amount of standing genetic variation within the parasite population. The next host generation was started by collecting the surviving adult beetles from the previous generation, which were allowed to lay eggs for 7 days on 200 g of fresh medium. The adults were then removed and the larvae allowed development to keep generations discrete. Host population size was kept constant by always collecting 500 (unsexed) adult beetles from the previous generation as breeders for the next generation. For the parasite population we collected the dead, infected larvae that are enriched with spores, and kept them in the refrigerator until the next host generation was started. To avoid selection on development time alone, we collected dead larvae in a flexible time window between 35 and 50 days post-egg laying, matching the intrinsic differences in development time between the lines and thus balancing the relative time of death across experimental lines. This window was chosen to avoid collecting larvae of the next generation while, at the same time, getting larvae that were old enough to contain enough spores for the preparation of the next generation inoculum. The collected larvae were grounded and sieved to make a homogenous spore-containing powder, which was added to the medium of the new host generation to obtain a final concentration of 2 × 104 spores g−1 of medium. The remainder of the grounded larvae was kept at 4 °C where they can be stored for extended periods of time without losing infectivity (Milner, 1972a). The use of spores that were able to infect the previous host generation and hosts that survived the previous parasite generation ensured that both hosts and parasites were exerting direct reciprocal selection pressures on each other.

Infection and survival experiment

After 10 generations of coevolution, we relaxed the coevolved lines from selection by letting 500 surviving beetles from each of the eight selection lines lay eggs in separate vials on control flour (with no parasites added), following the same protocol described above. This relaxation over one generation before the infection treatment in generation 11 was done to avoid potential maternal (Sadd et al., 2005) or paternal (O. Roth O, G. Joop, H. Eggert, J. Hilbert, J. Daniel, P. Schmid-Hempel & J. Kurtz, unpublished data) priming of resistance, which have been recently found to play a role in insect immunity. For each of the eight control lines, by contrast, we maintained discrete generations on their usual control flour for 11 generations. From each of these experimental lines of generation 11 (control lines, plus coevolved lines relaxed for one generation), we collected 50 unsexed beetles to produce offspring that were used in the survival experiments reported below.

To test our questions, we applied the following infection treatments (all done on hosts of generation 11; G11): (i) exposure to spores from the original parasites (used to infect all lines in generation zero), (ii) exposure to spores from the pool of spores collected three generation ago (i.e. generation 8; G8), (iii) exposure to spores from the current generation (G11) and (iv) controls with no exposure to parasite spores. The original parasites were the same for all lines, while G8- and G11-spores always originated from the host line that was currently tested. Spores were used to infect larvae from, both, the coevolved selection regime and the paired control regime. For these infections, parasite-inoculated flour was made using the same protocol as used for the 11 generations of coevolution. We filled sterile glass vials (40 × 13 mm; VWR, Dietikon, Switzerland) with 0.1 g of either the inoculated medium or control medium. From each of the eight lines from the paired control and coevolution treatments (‘selection regimes’), 72 freshly hatched larvae each were distributed equally among any of the four ‘infection treatments’– control (standard medium), original parasites, G8 and G11 – but such that each vial contained a single larvae. As infection probability by N. whitei is age dependent, we used 3-day-old larvae for all lines. In total, 1152 larvae could be used (8 replicate lines × 2 selection regimes × 4 infection treatments × 18 larvae replicates). After their assignment to treatment and vial the larvae were kept under standard environmental conditions (24 h dark, 32 °C, 70% humidity). The vials were checked for survival on a daily basis until 15 days after distribution, and every 2 days thereafter for a total of 59 days.

Spore load measurement

From each experimental block (replicate line × selection regime × infection treatment), we collected two alive and two dead beetles for spore load analysis. DNA was extracted using 96-well plate extraction kits (Qiagen, Basel, Switzerland) and diluted to 5 ng μL−1. Spore load was measured using quantitative real time PCR of a 220-bp product of N. whitei 16sRNA using methods as described in Wegner et al. (2008). To quantify spores, we used a duplicated four-fold dilution series of the same eight standard samples in every run. Relative spore numbers were log transformed for analysis.

Statistical analysis

Time to death was analysed using Cox regression survival analysis with selection regime, infection treatment and line as fixed effects, further testing for an interaction between selection regime and infection treatment. Mean mortalities of the control lines under the three parasite sources were correlated with mean mortalities of the paired coevolved lines. Within both coevolved and control lines, mean mortalities when exposed to the original parasites, G8 and G11 were correlated. As both variables are subject to measurement error, major axis regression was used to test for significant correlations. Sporeload was analysed for the total dataset, using a mixed effect linear model with log-transformed spore count as the response variable. The model contained selection regime, infection treatment, interaction between selection regime and infection treatment, and outcome Dead/Alive as fixed factors whereas line was analysed as a random effect. Within each subset of dead and alive beetles, we repeated the analysis with a model containing all of the remaining factors except for Dead/Alive. The correlation between average mortality and average sporeload per experimental block (8 replicate lines × 2 selection regimes × 3 infection treatments) was analysed using polynomial regression. All statistical analyses were conducted with the r statistical package (R_Development_Core_Team, 2007).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

Survival

Overall, survival was generally higher in the coevolved beetles than in the control beetles (Fig. 1, χ2 = 13.75, d.f. = 1, P < 0.001). Survival varied significantly among infection treatments (χ2 = 173.03, d.f. = 3, P < 0.001) and host lines (χ2 = 89.48, d.f. = 7, P < 0.001). Treatment effects were however not uniform as indicated by the significant interaction between selection regime and infection treatment, i.e. the direction of change varied according to the interaction (χ2 = 8.32, d.f. = 3, P = 0.04).

image

Figure 1.  Cumulative survival of beetle hosts in the infection experiment. Coevolved beetle lines are shown in gray lines, whereas the control lines are shown in black. Infection treatments are control (dot-dashed line), parasites from the current generation (G11, dotted line), parasites from three generations ago (G8, dashed line) and parasites from the original parasites used to infect all lines at the beginning of the coevolution experiment (original, solid line).

Download figure to PowerPoint

After 11 generations of separate evolutionary trajectories, we still found a significant correlation between the mean mortality of the control and the paired coevolved host line from the same original experimental line (nrs. 1–8; Fig. 2). This pattern was particularly obvious when exposed to the original parasites (Fig. 2a) and to the current parasites (Fig. 2c), but not when exposed to parasites from three generation ago (Fig. 2b).

image

Figure 2.  Correlation of mean mortality of the control and coevolved host lines when exposed to (a) original parasites used to infect all lines at the start of the coevolution experiment (r = 0.89, P = 0.003), (b) parasites from three generations ago (G8, r = 0.56, P = 0.148) and (c) parasites from the current generation (G11, r = 0.75, P = 0.032). Every symbol indicates a pairwise comparison within a single replicate line of the same origin. The dashed line indicates the expectation if mortality would be equal for both control and coevolved host lines. Solid lines are the fitted least square regressions. Mean mortality of coevolved and host lines under all three parasite infection treatments correlated (r = 0.88, P = 0.003).

Download figure to PowerPoint

Temporal dynamics of parasite virulence

Parasite-induced host mortality decreased strongly during the course of the experiment from 41.0% mortality rate when exposed to the original parasite mix, down to 19.9% when exposed to the current generation (G11). Within the coevolved lines, there was no positive correlation between mortality when exposed to the original parasites on one hand, and to coevolved parasites from G8 (r = 0.51, P = 0.197) and G11 on the other hand (r = 0.20, P = 0.631). However, there was a positive correlation between mortality when exposed to parasites from G8 and G11 (r = 0.81, P = 0.014). In the control lines, the pattern was inversed, as there was a positive correlation of mortality when exposed to the original parasites on one hand, and to parasites from G8 (r = 0.80, P = 0.017) and G11 (r = 0.74, P = 0.034) on the other hand. However, there was no correlation between resistance against parasites from G8 and G11 (r = 0.57, P = 0.135). These patterns suggest that experimental coevolution rapidly drives susceptibility to parasites over time scales of a few generations, while with no coevolution susceptibility might fluctuate around a trend line of change over short time spans.

Spore load

Spore load was significantly higher in dead individuals (3.79 ± 0.09 SE) than in alive individuals (0.59 ± 0.05 SE, Table 2). In alive beetles there was no significant effect of selection regime, treatment or line (Tables 2 and 3). In dead beetles, coevolved and control host lines showed an inverse response to the different infection treatments (Tables 2 and 3). In the coevolved host lines average spore load of G8 and G11 was lower than when exposed to the original parasites, whereas in the control host lines spore load of G8 and G11 were higher than when exposed to the original parasites. There were no main effects of either selection regime, treatment or line (Table 2).

Table 2.   Mean sporeload (log-transformed spore counts per beetle) for alive and dead coevolved and control beetles when exposed to three parasite sources.
Parasite sourceDead coevolved beetlesDead control beetlesAlive coevolved beetlesAlive control beetles
  1. Entries are means of log transformed spore load ± standard error. Sample sizes per category are shown between brackets.

Original4.06 ± 0.07 (n = 17)3.76 ± 0.25 (n = 16)0.67 ± 0.16 (n = 16)0.68 ± 0.11 (n = 16)
G83.41 ± 0.38 (n = 13)4.09 ± 0.05 (n = 18)0.45 ± 0.08 (n = 16)0.52 ± 0.08 (n = 16)
G113.23 ± 0.40 (n = 14)4.00 ± 0.06 (n = 13)0.56 ± 0.15 (n = 17)0.66 ± 0.19 (n = 16)
Table 3. anova table of sporeload.
 D.F.SSF-valueP-value
  1. Significant effects are given in bold.

Total model
 Dead/Alive1478.62844.806< 0.0001
 Selection regime11.783.1360.078
 Treatment21.251.1010.335
 Selection regime × treatment23.022.6630.073
 Error174   
Dead beetles
 Selection regime12.7283.4210.068
 Treatment21.4360.9000.410
 Selection regime × treatment22.6273.2940.042
 Error78   
Alive beetles
 Selection regime10.0730.2460.621
 Treatment20.6341.0610.351
 Selection regime × treatment20.03780.06330.939
 Error84   

Average spore load (a token for transmission success in an obligate killer) of both alive and dead individuals correlated negatively with survival in each experimental block (R2 = 0.115, F1,46 = 5.98, P = 0.018). However, a quadratic polynomial regression (Fig. 3, R2 = 0.263, F2,45 = 8.015, P = 0.001) could explain significantly more of the variation (P < 0.001), thus indicating the presence of intermediate optimal virulence associated with highest transmission capacity. Considering coevolved and control lines (Fig. 3) separately we observed the same qualitative pattern. The optimal virulence (i.e. maximum spore load) shifted from a mortality of 53.6% survival in the control lines to 59.5% in the coevolved lines and for any given mortality spore load was higher in control beetles (Fig. 3).

image

Figure 3.  Correlation of average survival and spore load. Each dot indicates an experimental block. Control lines are shown as triangles and coevolved lines as circles. The solid line is the significant quadratic polynomial regression for all data points (R2 = 0.263, F2,45 = 8.015, P = 0.001). The dotted line is the quadratic polynomial regression for the coevolved host lines (R2 = 0.216, F2,21 = 2.891, P = 0.078) and the dashed line is the significant quadratic polynomial regression for the control host lines (R2 = 0.316, F2,21 = 4.857, P = 0.018).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

The evolution of host resistance

We found that experimental coevolution has led to the evolution of resistance in the host. Host survival after exposure to parasites was higher in coevolved than in control host lines (Fig. 1). This suggests that antagonistic coevolution also has a component of directional selection in addition to, and beyond purely negative frequency dependent selection where host and parasite types would be persistently recycled with no directional adaptation. However, the timescale of our experiment might have been too short to formally distinguish between directional and negative frequency dependent selection. On the other hand, it is also conceivable that the resolution of sampling points was not sufficient to detect the underlying processes.

Previous studies show that resistance against N. whitei has additive as well as substantial nonadditive genetic effects (Wegner et al., 2008, 2009). Because a response to selection was observed, an additive component must be present. Furthermore, the correlation of observed mortalities between coevolved and corresponding control lines after 11 generations (Fig. 2) furthermore suggests that the genetic background represented by each line was a strong determinant of general resistance and for the observed changes during the history of selection in the experiment. As far as these backgrounds represent the additional, nonadditive genetic effects, the observed response to selection underline the importance of the genetic architecture for resistance as a determinant of survival. Our experiment also showed that coevolved host lines did not only become more resistant, but that resistance had both a general (defined by the genetic background of each line) and a specific (defined by the time lag within the host line) element. Mortality of coevolved host lines exposed to parasites from three generations ago (G8) correlated with mortality induced by recent parasites (G11) whereas no such correlation could be observed between original parasites (from the start of the experiment) and parasites from either G8 and G11.

Parasite virulence decreased during coevolution

Whereas the above discussion focused on the evolution of host resistance, the experiment also provided information about parasite virulence. In contrast to our expectation, parasite virulence decreased during coevolution, not only when exposed to control lines but also on the coevolving host lines (Fig. 1). Comparing the survival curves for control hosts, parasites from three generations ago (G8) were almost as virulent as the original parasites, and a reduction of virulence was only observed for parasites from the current generation (G11, Fig. 1). In the coevolved host lines, on the other hand, the survival curves suggest that virulence of G8 and G11 parasites were both lower than the original parasites, but there was no difference between recent (G8) and current (G11) parasites. These patterns suggest that the parasite virulence does not attenuate until some time into the experiment (almost no difference in virulence between original parasites and G8, but large difference in virulence between G8 and G11), and that virulence is also expressed rather specifically in relation to the coevolved host line (correlation between G8 and G11). Likely, the parasites showed a directional change towards lower overall virulence combined with specific coadaptations to the coevolving host line. A possible explanation for the reduced virulence might be high relatedness and hence kin selection between parasites (Frank, 1992). As we started with a cocktail of parasite spores, parasite relatedness might increase during coevolution due to e.g. lineage sorting, which can lead to reduced virulence (Nowak & May, 1994; Buckling & Brockhurst, 2008). Although a sound possibility, it seems not to be likely in our system, because a previous study showed that virulence did not differ between multiple infections with unrelated strains compared to single strain infections (Otti, 2007).

Reduction in virulence can also be explained by a trade-off between virulence and transmission. Here, the parasite’s transmission potential, measured as spore load in dead individuals, did not decrease over the course of the experiment. Both past and present coevolved parasites reached lower spore loads in the coevolved hosts than in the control hosts (Table 3; see interaction term ‘parasite treatment’ × ‘selection regime’ in Table 2). Since spore load on the control lines did actually increase (c.f.Table 3), it seems that parasites did not necessarily lose their transmission potential as they lost virulence.

Nosema whitei is faced with a trade-off between virulence and transmission potential

As both high and low virulence lead to lower average spore load and thus fewer transmission stages (Fig. 3), our results are suggestive of a cost of virulence (Anderson & May, 1982). Given the curves of Fig. 3, parasite transmission potential is maximized at 53.6% survival in the control and 59.5% in the coevolved lines, with a similar shape in both the coevolved and the control host lines. This seems to suggest that N. whitei might face a trade-off between exploiting the host resources for reproduction and killing the host, which although hypothesized for obligately killing parasites (Ebert & Weisser, 1997) has been observed only twice experimentally (Jensen et al., 2006; de Roode et al., 2008). Even though we cannot completely rule out the possibility that the observed trade-off is due to the combination between host resistance and parasite infectivity in coevolved hosts, the fact that we also observed an optimum in control hosts suggests that qualitatively our result is mainly dependent on the parasite. Furthermore, the optimal virulence level is lower in the coevolved host lines (higher survival) than for the control host lines. Our results are thus contradicting many serial passage experiments that showed an increased virulence on the sympatric host lines (Sabin et al., 1954; Muskett et al., 1985; Little et al., 2006). Most of our knowledge on the evolution of virulence comes from serial passage studies (Ebert, 1998; Little et al., 2006; Yourth & Schmid-Hempel, 2006). Although there are ample theoretical papers that have investigated under which conditions virulence evolves or attenuates (Lenski & May, 1994; Day, 2001; Gandon et al., 2002), so far, little has been done to experimentally study the evolution of virulence in dynamic systems where both protagonists are allowed to adapt and counter adapt. Hence, experimental studies such as ours are dearly needed to explore the different implications serial passage and coevolution experiments might have on the direction the evolution of virulence might take.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

In summary, antagonistic coevolution has lead to adaptive responses in both hosts and parasites. The coevolved host lines have evolved higher resistance, not only towards noncoevolved parasites but also towards coevolved parasites. In addition, parasites have evolved lower virulence on both coevolved and control hosts. The lower virulence might be explained by the trade-offs between virulence and transmission potential. Obligately killing parasites might thus face similar trade-offs as parasites that continuously shed transmission stages. This could make it difficult to predict the precise evolutionary trajectory of the system (Grech et al., 2006) even though the general process of co-adaptation occurs.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References

The authors thank Geraldine Chew for conducting a pilot study on this subject, Niels Kerstes for help with distributing larvae and Natasha Rossel and Miguel Hon for help with cleaning and autoclaving everything needed for this experiment. Supported by the Genetic Diversity Center of ETH Zurich (GDC) and CCES. Financially supported by SNF grant 31-120451 to KMW and ETH grant nr. TH-09 60-1 to PSH.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. References