SEARCH

SEARCH BY CITATION

Keywords:

  • Acyrthosiphon pisum;
  • aphid resistance;
  • experimental evolution;
  • parasitoid adaptation;
  • secondary symbionts

Abstract

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

Insects harbour a wild diversity of symbionts that can spread and persist within populations by providing benefits to their host. The pea aphid Acyrthosiphon pisum maintains a facultative symbiosis with the bacterium Hamiltonella defensa, which provides enhanced resistance against the aphid parasitoid Aphidius ervi. Although the mechanisms associated with this symbiotic-mediated protection have been investigated thoroughly, little is known about its evolutionary effects on parasitoid populations. We used an experimental evolution procedure in which parasitoids were exposed either to highly resistant aphids harbouring the symbiont or to low innate resistant hosts free of H. defensa. Parasitoids exposed to H. defensa gained virulence over time, reaching the same parasitism rate as those exposed to low aphid innate resistance only. A fitness reduction was associated with this adaptation as the size of parasitoids exposed to H. defensa decreased through generations. This study highlighted the considerable role of symbionts in host–parasite co-evolutionary dynamics.


Introduction

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

There is no species living isolated from other organisms, but they rather establish frequently close associations with other species. Symbiosis is the term coined for a prolonged and intimate relationship between members of different species, in which a ‘symbiont’ occupies a habitat provided by a ‘host’ (Begon et al., 2006). Invertebrate species possess an extraordinary diversity of symbiotic microorganisms, especially bacteria (Chaston & Goodrich-Blair, 2010). These symbionts can be classified into two distinct groups according to the level of dependence on the host. Primary symbionts are necessary for host survival and reproduction, whereas secondary (or facultative) symbionts are not essential for their host, although they generally cannot survive outside the host environment.

Secondary symbionts are predominantly vertically transmitted (Oliver et al., 2010), and any effect conferred to the host that increases its survival or reproduction relative to the uninfected congeners may then enhance their spread and persistence within the host population. During the last decades, there has been increasing evidence for effects of secondary symbionts on host phenotypes. These include host reproduction manipulation (e.g. Hurst et al., 1999) and beneficial effects on host fecundity or survival. In the latter case, recent works on diverse invertebrates have revealed in particular that secondary symbionts can protect their hosts against microbial diseases, parasites or predators (e.g. Gil-Turnes et al., 1989; Gil-Turnes & Fenical, 1992; Kellner, 1999; Jaenike et al., 2010).

Among insects, the symbionts of aphids are perhaps the best studied (Oliver et al., 2010). In addition to their essential nutrient-providing symbiont, Buchnera aphidicola, aphids may also carry one or more secondary symbionts. These heritable symbionts exert diverse effects on their host. In the pea aphid Acyrthosiphon pisum, symbionts have an impact on the host ecology through a wide range of effects, e.g. host plant utilization (Tsuchida et al., 2004), body colour (Tsuchida et al., 2010), protection against heat stress (Montllor et al., 2002) and enhanced protection against natural enemies (Oliver et al., 2003; Scarborough et al., 2005).

Many organisms use aphids as resources, and among them, insect parasitoids (i.e. insects whose larvae develop by feeding on (ectoparasitoids) or within (endoparasitoids) an arthropod host, eventually killing it (Eggleton & Belshaw, 1992)) exert a strong pressure on aphid populations. Parasitism rates can reach 70% (Hufbauer, 2002) but resistance to parasitoids exists in pea aphid populations (Henter & Via, 1995; Hufbauer & Via, 1999; Ferrari et al., 2001). Aphids may counter parasitism with a combination of several mechanisms. They can resist to a parasitoid attack through behavioural defences, but once oviposition has occurred, parasitoid eggs face physiological defences based on host intrinsic immunity and on symbiosis with bacteria (Henter & Via, 1995; Hufbauer & Via, 1999; Le Ralec et al., 2010). Little work has yet been carried out to characterize the pea aphid immune response, but recent studies show that this species is missing several genes central to insect immune functions, conferring consequently only a partial resistance against parasitism (Gerardo et al., 2010). Actually, A. pisum largely relies on secondary symbionts as far as protection against parasitoids is concerned. The failure of the parasitoid development is mainly because of the presence of symbiont members of the Enterobacteriaceae, Serratia symbiotica and especially Hamiltonella defensa (Oliver et al., 2003, 2005; Ferrari et al., 2004). H. defensa has been reported to provide protection to A. pisum against both Aphidius ervi and Aphidius eadyi parasitoids (Oliver et al., 2003, 2005; Ferrari et al., 2004; Bensadia et al., 2006), referred to as ‘symbiont-mediated resistance’ (Oliver et al., 2005). This resistance appears to be largely correlated with the presence of a bacteriophage of H. defensa, A. pisum secondary endosymbiont (APSE) (Oliver et al., 2009), which encodes toxins that may be responsible for prematurely arresting the development of parasitoid larvae (Degnan & Moran, 2008; Oliver et al., 2009). The effectiveness and the specificity of the symbiont-mediated resistance have been studied in the pea aphid. For instance, multiple strains of H. defensa were subsequently examined in a common aphid genotypic background, and all conferred partial protection against A. ervi (Oliver et al., 2005). Additionally, similar levels of resistance were found for a single strain of H. defensa established in multiple A. pisum clonal backgrounds. These empirical studies corroborate the role of this symbiont in protection against parasitoids.

Although major advances have been made on the mechanisms associated with these symbiotic beneficial effects on the infected host, little is known about their evolutionary outcomes on the targeted organisms, the enemies. The frequency of A. pisum infected by H. defensa varied between aphid populations (Frantz et al., 2009) but an empirical study showed that this frequency increased dramatically after repeated exposure to parasitism by the A. ervi parasitoid (Oliver et al., 2008). The level of resistance conferred by the symbiont varies depending on the H. defensa strain, from small reduction in succesfull parasitism to nearly complete protection (Oliver et al., 2010). Given these results, aphid symbiont-mediated resistance is likely to exert strong selective pressures on natural populations of parasitoids because they maintain a particularly intimate relationship with the aphids as a single host harbours the parasitoid’s offspring until maturity (Godfray, 1994). The intimate nature of this relationship promotes co-evolutionary dynamics, while developing adaptation to counter the combined effect of the aphid intrinsic immunity and the symbiont-mediated resistance.

With the use of an experimental evolution approach under controlled laboratory conditions, we investigated the adaptive potential of the aphid parasitoid A. ervi facing either population of aphids harbouring H. defensa, combining innate- and symbiont-mediated resistance, or population of aphids without secondary symbiont, presenting only the innate resistance. The possible effects associated with a potential adaptation to exploit highly resistant hosts were also estimated by measuring the size of the parasitoid along the selection process.

Materials and methods

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

Organisms

Twelve green clones of the pea aphid Acyrthosiphon pisum (Harris) (Hemiptera: Aphididae) were chosen for this study (see Table 1). As secondary symbiont community in A. pisum strongly depends on the host biotype (Frantz et al., 2009), A. pisum clones used in our experiment were chosen among a collection of plant-specialized genotypes with known secondary symbionts (see Frantz et al., 2009; Peccoud et al., 2009 for details on clone collection and symbiotic detection). Four of the 12 clones were free of secondary symbionts (i.e. clones called ‘Hamiltonella-free’ clones). Among them, one was originated from alfalfa (Medicago sativa), whereas the others came from spiny restharrow (Onosis spinosa) and meadow vetchling (Lathyrus pratensis). The eight other clones, collected on alfalfa, harboured the secondary symbiont Hamiltonella defensa (i.e. clones called ‘Hamiltonella’ clones). All aphids were maintained on broad bean plants, Vicia faba.

Table 1.   List of the pea aphid clones from various French localities tested in the experiment 1.
CloneSampling siteSampling dateHost plantHamiltonella defensa
H1Domagné (West)2008Medicago sativaYes
H2Domagné2008Medicago sativaYes
H3Domagné2008Medicago sativaYes
H4Pacé (West)2008Medicago sativaYes
H5Pacé2008Medicago sativaYes
H6Pacé2008Medicago sativaYes
H7Le Rheu (West)2008Medicago sativaYes
H8Le Rheu2008Medicago sativaYes
Hf9Lusignan (Centre)1987Medicago sativaNo
Hf10Bugey (East)2006Ononis spinosaNo
Hf11Bugey2006Ononis spinosaNo
Hf12Le Rheu2006Lathyrus pratensisNo

To favour evolutionary responses, genetically diverse starting populations for Aphidius ervi parasitoids were used. The initial culture was obtained by sampling about 200 A. pisum mummies (i.e. dead aphid containing a parasitoid few days before its emergence) in different alfalfa fields around Rennes (Western France) in November 2008. Adults emerging from these collected mummies were split and maintained in three distinct Plexiglas cages (40 × 30 × 50 cm) containing a high genetic and symbiotic diversity of red and green pea aphids originating from six different alfalfa crops, located in two cities that are 40 km apart from one another. To limit genetic drift in the parasitoid populations, immigration was simulated by regular addition of newly sampled individuals or exchanges between the three cages. To obtain parasitoids for experiments, mummies were collected daily from cultures and placed individually in gelatine capsules. Newly emerged females were enclosed in plastic tubes (22 × 1 cm) containing moistened cotton, droplets of honey and one male for mating. Females between 2 and 3 days old were used only once.

All insect rearing and experiments were performed in climate rooms at 20 ± 1 °C, 70 ± 10% relative humidity and L : D 16 : 8 h photoperiod.

Experiment 1: Aphid clone resistance measurement

Experimental set-up

This experiment aimed to evaluate the level of resistance to parasitism in the 12 aphid clones to select clones for the evolution experiment. Their level of parasitism resistance was estimated by using a classical parasitism design (modified from Oliver et al., 2003). Just prior to the experiment, the parasitoid females were allowed to oviposit in a healthy aphid to provide them an oviposition experience. Any female that did not oviposit within the first 5 min was discarded. The female was then introduced in a glass Petri dish containing 15 third-instar aphid larvae of a given clone, placed on a broad bean leaf for an hour. After each oviposition (i.e. the female has stung one aphid individual with its ovipositor), the attacked aphid was removed from the experimental arena and placed on a new broad bean plant. The experiment ended when the parasitoid had attacked 10 aphids successively. During the following days, the number of parasitoids emerging from the 10 attacked aphids was daily counted to determine the effective parasitism rate (i.e. the number of parasitoids emerging from a given host population). Five replicates were carried out for each aphid clone tested.

Statistical analyses

The dependent variable analysed here was the effective parasitism rate, suggesting that the response was binomial (i.e. a parasitoid emerged or not from an aphid). The first analysis aimed to test the effect of the presence of H. defensa on effective parasitism rate. In our experimental design, several clones harbouring or not harbouring the secondary symbiont were tested and this was considered as a random independent variable in our statistical modelling. The effects of the high resistance because of H. defensa (i.e. a fixed independent variable with two modalities) on the effective parasitism rate was then tested by fitting generalized linear mixed models (GLMM) with binomial errors and a logit-link using the lme4 package (Bates & Maechler, 2009) in r 2.8.1 (R Development Core Team, 2008). The second analysis consisted of the comparison of the effective parasitism rate between the different clones containing or not the symbiont. For this purpose, the response variable was tested against the aphid clone by using generalized estimation equations (GEE) that considered repeated data (e.g. a female attacking 10 successive aphids), with binomial errors and a logit-link using the geepack package (Yan, 2002) in r 2.8.1.

Experiment 2: Experimental evolution of parasitoid populations exposed to symbiont-mediated resistance

Experimental evolution procedure

The experimental evolution procedure is illustrated in Fig. 1. It was performed for 10 parasitoid generations using two selection treatments that were applied in exactly the same way except the presence or absence of the symbiont-mediated resistance in host population. A host population consisted here of a broad bean plant (about 15 cm height) infested with 32 third-instar aphids and covered with cellophane bags to avoid their escape. To reduce the aphid genotype effect, a host population contained either individuals from the four ‘Hamiltonella-free’ aphid clones or individuals from four clones chosen among the nine harbouring H. defensa. The selection of these ‘Hamiltonella’ clones was based on results from experiment 1. In a host population, each clone was equally represented (i.e. eight individuals per clone). For each parasitoid generation, eight ‘Hamiltonella-free’ and eight ‘Hamiltonella’ host populations were similarly constituted and then exposed to parasitoids. Consequently, parasitoids faced fixed host populations over time.

image

Figure 1.  Experimental evolution design.

Download figure to PowerPoint

For the initial population of parasitoids used in this experimental evolution design, 150 mummies were randomly extracted from the initial cultures. All emerging females were allowed to mate during 1 day and then split for the two selection treatments: (i) the ‘Nonexposed’ parasitoid line was exposed to ‘Hamiltonella-free’ host populations, meaning the wasps were submitted to a low selection exerted by the aphid intrinsic immune responses; and (ii) the ‘Exposed’ parasitoid line was exposed to ‘Hamiltonella’ host populations exerting a strong selection characterized by the combination of both innate- and symbiont-mediated resistances. The two treatments were both followed during 10 generations.

The experimental evolution procedure consisted in introducing four parasitoid females in each host population. One parasitoid male was also added into the cellophane bag to ensure mating and females fecundation. Parasitoids were allowed to exploit the host population for 4 days. After this delay, they were removed from the design and stored in 95 °C ethanol for subsequent measurements. The host population was then maintained in climate chambers during 20 days and examined daily to record and extract aphid mummies isolated in gelatine capsules. To minimize the loss of genetic diversity because of inbreeding and reduced population size, all parasitoid males and females emerging from each host population of a selection treatment were pooled together during 1 day and 32 females (i.e. four females for each of the eight host populations) were randomly chosen to constitute the next generation. This experimental procedure was continuously conducted for 10 successive parasitoid generations (from February 2009 to November 2009).

At each parasitoid generation, the following variables were measured, as they provide an approximation of the fitness components (Roitberg et al., 2001): (i) the effective parasitism rate represented by the number of parasitoids emerging upon the overall host population (i.e. this was the number of parasitoids emerging from the 32 aphids) and (ii) the left tibia size of the individuals that had survived, stored in 95 °C ethanol. The potential effects on parasitoid fitness associated with the adaptation to the high level of resistance conferred by secondary symbionts were estimated by measuring the size of parasitoid females. On the basis of previous works (Godfray, 1994; Nicol & Mackauer, 1999; De Conti et al., 2008), this measurement is a good proxy for parasitoids fecundity, as, in Hymenopteran parasitoids, the reproductive fitness in terms of fecundity, parasitism, searching rate and longevity is often positively correlated with the body size (Godfray, 1994; He & Wang, 2006). Left tibia length was measured under a binocular using the Archimède software (Microvision Instruments, 1997–2007, Evry, France, http://www.microvision.fr) on 356 females for five parasitoid generations (G1, G2, G5, G7 and G10).

Statistical analyses

The first set of analyses consisted of testing the effect of the parasitoid generation number and the selection treatment on each dependent variable. In this experimental design, some of the response data were correlated. Concerning the effective parasitism rate, the 32 aphids of one host population were exposed to the same four parasitoid females. These data clusters must be considered as if correlation was not taken into account then the standard errors of the parameter estimations would not be valid (for details, see Liang & Zeger, 1986). The analysis of these dependent variables was then carried out by means of generalized estimating equations (GEE) to estimate parameters for correlated data. The effective parasitism rate was tested against the parasitoid generation (from 0 to 10), the selection treatment (‘Nonexposed’ vs. ‘Exposed’ parasitoids line) and the interaction between these two independent variables by GEE assuming a binomial error and a logit-link function using the geepack package (Yan, 2002) in r 2.8.1. The effect of the parasitoid generation, the selection treatment and the interaction between these two independent variables on the left tibia length of females was estimated using a standard analysis of variance. For all statistical analyses, the selection treatment was coded as a factor (i.e. a class-independent variable with two levels) and the parasitoid generation as a covariate (i.e. a continuous independent variable).

The second set of analyses consisted of analysing how the difference between the two selection treatments changed with the parasitoid generation number. Pairwise comparison between the ‘Nonexposed’ and ‘Exposed’ parasitoid lines was then made at each parasitoid generation with the function ‘esticon’ in the ‘doBy’ package (Højsgaard, 2009). For each response, graphics representing the difference between the mean of both selection treatments according to the parasitoid generation number were drawn to illustrate how the two parasitoid lines differed in time.

Experiment 3: Assessment of selection efficiency

Experimental procedure

This experiment aimed to assess how the ability of parasitoids to use aphids harbouring H. defensa as hosts evolved following experimental evolution. Using the parasitism design of the first experiment, we measured the effective parasitism rate of the offspring of both lines on two ‘Hamiltonella’ clones. They were randomly selected among the four clones used in the experiment 2. For each parasitoid line, 10 females were tested on each clone.

Statistical analyses

In this analysis, three parasitoid populations were compared: the initial population before the experimental evolution, the offspring of the ‘Nonexposed line’ and the offspring of the ‘Exposed line’ after the experimental evolution. The effective parasitism rate was then tested against the aphid clone (i.e. a class-independent variable with two levels), the parasitoid population (i.e. a class-independent variable with three levels) and the interaction between these two factors by fitting GEE models with binomial errors and a logit-link function.

Results

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

Experiment 1: Aphid clone resistance measurement

Figure 2 represents the effective parasitism rate measured on the 12 aphid clones tested. The presence of H. defensa in aphids strongly reduced the probability of parasitoid emergence (GLMM, inline image = 10.55, < 0.001). Overall, 143 parasitoids emerged from the 200 ‘Hamiltonella-free’ aphids attacked by A. ervi, whereas only 33 emerged from the 400 ‘Hamiltonella’ aphids exposed to the parasitoid wasp. This confirmed that the presence of H. defensa conferred a stronger level of protection against A. ervi in this particular set of aphid clones than the single innate resistance. Significant difference in parasitism rates was found between ‘Hamiltonella-free’ clones (GEE, inline image = 20.40, < 0.001), and the level of symbiont-mediated resistance varied significantly between aphid clones harbouring the secondary symbiont (GEE, inline image = 70.94, < 0.001).

image

Figure 2.  Effective parasitism rate (mean ± standard error) of Aphidius ervi females on aphid clones harbouring (open bars) or not (black bars) Hamiltonella defensa. Columns with different letters are significantly different (< 0.05). *Clones harbouring the secondary symbiont selected for the experimental evolution.

Download figure to PowerPoint

For the evolution experiment, aphid clones harbouring H. defensa were chosen in such a way that the selection pressure was relatively important, but not too strong to ensure the production of new parasitoid generations. The aphid clones H2, H3, H6 and H8 were then selected to constitute the ‘Hamiltonella’ treatment.

Experiment 2: Experimental evolution of parasitoid populations exposed to symbiont-mediated resistance

Effective parasitism rate

The effective parasitism rate in the two treatments is shown in Fig. 3. The proportion of emerged parasitoids from the host population varied from 0.2 to 0.6, and this response fluctuated in the course of the experimental selection producing a high level of variance (Fig. 3a). Neither the parasitoid generation number (GEE: inline image = 3.22, = 0.072) nor the selection treatment (GEE: inline image = 3.56, = 0.059) significantly influenced the mean effective parasitism rate. There was, however, a significant interaction between the selection treatment and the generation number (GEE: interaction term, inline image = 7.63, = 0.005). Parasitoids from ‘Exposed’ lines produced significantly lower offspring when exposed early to ‘Hamiltonella’ host populations compared to parasitoids exposed to ‘Hamiltonella-free’ host populations. However, after a few generations, both parasitoid lines produced the same number of offspring in the presence or in the absence of H. defensa in host populations. Pairwise comparisons showed that four generations were sufficient to observe similar parasitism performance between ‘Exposed’ and ‘Nonexposed’ parasitoid lines (Fig. 3b). This convergence of the parasitism rates was only because of the significant increase in the ‘Exposed’ line (GEE: inline image = 12.4, < 0.001), as the parasitism rate in the ‘Nonexposed’ line was not significantly affected by the generation number (GEE: inline image = 0.307, = 0.58). It should be noted that the effective parasitism rate of the ‘Exposed’ parasitoid line surpassed the ‘Nonexposed’ one at the fifth generation.

image

Figure 3.  (a) The effective parasitism rate for each generation (mean ± standard errors) of Aphidius ervi females exposed to populations of aphids harbouring (white dots, ‘Exposed’ parasitoid line) or not (black dots, ‘Nonexposed’ parasitoid line) Hamiltonella defensa. (b) Differential effective parasitism rate between the two selection treatments: mean of the ‘Nonexposed’ parasitoid line minus the mean of the ‘Exposed’ parasitoid line. Stars represent significant differences between both treatment means (see Materials and methods) (***< 0.001, **< 0.01 and *< 0.05).

Download figure to PowerPoint

Female size

Overall, parasitoids from ‘Exposed’ line had similar tibia lengths than those from ‘Nonexposed’ line (Fig. 4a, GEE: inline image = 1.19, = 0.276). The body size, however, declined towards the end of the experiment in the two lines (GEE: inline image = 9.51, = 0.002), and it did not depend on the selection treatment (GEE: interaction term, inline image = 0.25, = 0.618). The analysis of mean difference between the two selection treatments at each parasitoid generation showed that parasitoids from the ‘Exposed’ and ‘Nonexposed’ lines significantly differed in their tibia size after 10 generations of selection (Fig. 4b). After 10 generations, the parasitoids exposed to ‘Hamiltonella’ host populations had smaller left tibias compared with parasitoids exposed to ‘Hamiltonella-free’ host populations.

image

Figure 4.  (a) Length of the left tibia for each generation (mean ± standard errors) of Aphidius ervi females exposed to populations of aphids harbouring (white dots, ‘Exposed’ parasitoid line) or not (black dots, ‘Nonexposed’ parasitoid line) Hamiltonella defensa. (b) Differential parasitoid left tibia length between the two selection treatments: mean of the ‘Nonexposed’ parasitoid line minus the mean of the ‘Exposed’ parasitoid line. Stars represent significant differences between both treatment means (see Materials and methods) (***< 0.001, **< 0.01 and *< 0.05).

Download figure to PowerPoint

Experiment 3: Efficiency of selection to symbiont-mediated resistance

When comparing the effective parasitism rate of parasitoids before and after the experimental evolution on two ‘Hamiltonella’ clones, the statistical model indicated a selective response. The ‘Exposed’ parasitoid line had more parasitoids successfully emerging from ‘Hamiltonella’ aphids than both initial parasitoid population and ‘Nonexposed’ parasitoid line (Fig. 5, GEE: inline image = 6.52, = 0.038), and this result did not depend on the aphid clone tested (GEE: interaction term, inline image = 0.166, = 0.920).

image

Figure 5.  Effective parasitism rate (mean ± standard errors) of the initial Aphidius ervi parasitoid population before the experimental evolution (grey dots), of the ‘Nonexposed’ parasitoid line (black dots) and of the ‘Exposed’ parasitoid line (white dots) after the experimental evolution on the two clones H2 and H6 harbouring Hamiltonella defensa.

Download figure to PowerPoint

Discussion

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

We used experimental evolution to determine whether there was heritable variation for parasitoid response to high host protection linked to a combination of symbiont- or innate-mediated resistance and whether this response was associated with changes in fitness components. Experimental evolution is particularly suited for the study of adaptation and is also a powerful tool for the detection of fitness trade-offs related to an evolved trait (Fuller et al., 2005). However, a limitation of experimental evolution is that not all biological systems are equally suited to these types of studies as in many host–parasite interactions, one or both of the antagonists cannot be prevented from evolving: the genotype of either host or parasite cannot be preserved over time (Gaba & Ebert, 2009). In our context, parthenogenetic reproduction of aphids ensured the transmission and the preservation of the same genetic and symbiotic background (Oliver et al., 2010) and A. ervi parasitoids faced host population of the same resistance at each generation. Finally, this experimental approach allowed us to select parasitoid adaptation and demonstrated the important role of secondary symbionts in the host–parasite co-evolutionary dynamics.

In this study, measurements of parasitism resistance in the pea aphid clones showed that the presence of H. defensa explained most of the variations in parasitoid protection, validating results from earlier studies (for a review, see Le Ralec et al., 2010 and Oliver et al., 2010). Interestingly, the clones infected with the secondary symbiont varied in their degree of resistance. One source of such variations could be the nature of toxins produced by the different type of APSE bacteriophages harboured by H. defensa (Degnan & Moran, 2008). Another explanation being that genotype-by-symbiont interactions exist in the pea aphid host. This is in contradiction with Oliver et al. (2005) who did not find any interaction between symbiont isolate and genetic background. However, in our study, there were significant variations in the resistance of ‘Hamiltonella-free’ clones, suggesting interclonal variations in aphid innate resistance components. This reinforces Vorburger et al. (2009) suggestion that aphid parasitoids are confronted with two lines of aphid defences: ‘innate defences’ and ‘acquired defences’ provided by secondary symbionts, which likely differ in their effectiveness and specificity. During this first experiment, we had the opportunity to study the evolution of parasitoids facing two types of resistance: (i) a strong selection pressure because of a combination of Hamiltonella-mediated resistance and intrinsic immunity and (ii) a low selection pressure exerted only by the aphid immune response.

Interestingly, this study showed how resistance components in insect hosts are potential agents in the selection of their parasitoids through changes in fitness over the course of experimental evolution. From the fifth parasitoid generation, the population fitness of both selection treatments, estimated by the effective parasitism rate, converged. The third experiment confirmed the beneficial effect of the experimental selection on parasitoid population as ‘Exposed’ parasitoid line was more able to counter the resistance of the aphids harbouring H. defensa compared to the initial parasitoid population. Our results provide evidence that mechanisms involved in ‘virulence’, defined as wasp’s ability to develop within host and to emerge from it (Henter, 1995; Antolin et al., 2006), exist in parasitoid populations. Parasitoid virulence is generally mediated by surface attributes of parasitoid eggs and eggs associated with biological materials injected into the host by the wasp (Bensadia et al., 2006). The main bioactive protein isolated from A. ervi female venom is called γ-glutamyl transpeptidase (γGT) and triggers the apoptosis of the germaria and ovarioles cells of the pea aphid (Digilio et al., 2000; Falabella et al., 2007). At least two different alleles of the γGT have been identified (C. Anselme, unpublished). The venom effect is completed by two proteins, FABP and ENO, released by the teratocytes (cells derived from the parasitoid egg membrane). All those putative virulence factors may induce host-resource hijacking, favouring the parasitoid development (Falabella et al., 2005, 2009). Consequently, the response of the selected female could rely on a variation in the virulence factors in the initial population, traits under directional selection in our experiment, and transmitted by the females to the next generations. Further molecular analysis is needed to determine the mechanisms associated with this improved virulence, for instance, by comparing the venom composition of the ‘Exposed’ parasitoid line that responded to the strong selective pressure with ‘Nonexposed’ parasitoid line that did not respond significantly to the low host resistance.

The possibility that improved virulence is accompanied by some changes in other fitness components is suggested by a decrease in the size of the left tibia in females exposed to the two types of resistance in aphid populations. This proxy for fecundity in hymenopterans (He & Wang, 2006, 2008) decreased in the two treatments at the tenth generation, so noticeable fitness modifications did not occur until parasitoids had spent between 7 and 10 generations on hosts. This delay of the response excludes the influence of a possible variation in nutritional quality between aphids harbouring or not H. defensa as the effect on parasitoid size would have appeared at the first generation. The size reduction was much lower in parasitoids exposed to ‘Hamiltonella-free’ clones than to ‘Hamiltonella’ clones, suggesting a cost for aphid resistance adaptation in parasitoid populations. The existence of such trade-offs supports the notion of a genotypic response of the parasitoids to the selective agent. Trade-offs associated with the improvement in host resistance have been largely described in Drosophila (Kraaijeveld et al., 1998; Vijendravarma et al., 2009) and also in aphids (Vorburger et al., 2008), but the costs associated with modifications in virulence of parasitoids are less well documented. The possibility of fitness costs to improved virulence is suggested by our observations but the study of their ecological and evolutionary consequences would be particularly helpful to understand the aphid-parasitoid co-evolutionary processes.

Intense selective pressure resulted in diverse adaptations in parasitoid strategies to successfully infest a host (Poiriéet al., 2009). Our experimental evolution provides evidence for rapid evolution in A. ervi populations when faced with high aphid resistance and highlights the adaptive potential of parasitoid populations (Kraaijeveld et al., 2001; Henry et al., 2008; Gaba & Ebert, 2009). This experimental work is consistent with previous studies revealing the potential of A. ervi populations showing H. defensa/parasitoid genotype specificities in the aphid Lysiphlebus fabarum/A. ervi system (Vorburger et al., 2009), or populations increasing their performances on particular target host species and becoming locally adapted to their natal host population (Henry et al., 2008, 2010). Henter (1995) also described the genetic adaptive potential of the parasitoid wasp populations. Up to now, 11 pea aphid host plant biotypes have been genetically determined (Peccoud et al., 2009) and these different races differ in their prevalence of secondary symbionts (Frantz et al., 2009). In nature, they may present a spatial heterogeneity in resistance to A. ervi in association with the prevalence of symbionts conferring resistance. Consequently, the differential resistance levels may lead to the adaptation of particular A. ervi genotypes, promoting local specialization and then diversification in parasitoid populations (Fry, 1996; Laine, 2009). It remains unclear to what extent host resistance may affect the population structure or divergence at the parasitoid level, and it may trigger codivergence in their parasitoids. However, host-associated differentiation in parasitoids is increasingly described (Stireman et al., 2006; Smith et al., 2007), giving suggestions regarding the issues of co-evolutionary processes between hosts and parasitoids, and maybe explaining the high richness of the hymenopteran parasitoid group (LaSalle & Gauld, 1992).

Acknowledgments

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

The authors like to thank Caroline Anselme, Marylène Poirié, Yvan Rahbé and Manuel Plantegenest for all stimulating discussions and suggestions about this work. Thanks also to Jean Peccoud for help with pea aphid collection, Bernard Chaubet for the rearing, Ludovic Guyonvarch for the ‘enormous’ technical support, Benjamin Agogue and Alan Walton for language corrections. Two anonymous referees are also acknowledged for their constructive comments on an earlier draft. This study was supported by the French INRA Santé des Plantes et Environnement (SPE) Department and the French ‘Ministère de l’Enseignement Supérieur et de la Recherche’.

References

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