SEARCH

SEARCH BY CITATION

Keywords:

  • cost of resistance;
  • Hamiltonella defensa;
  • longevity;
  • parasitoid;
  • symbiosis;
  • trade-off

Abstract

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

Maternally transmitted symbionts can spread in host populations if they provide a fitness benefit to their hosts. Hamiltonella defensa, a bacterial endosymbiont of aphids, protects hosts against parasitoids but only occurs at moderate frequencies in most aphid populations. This suggests that harbouring this symbiont is also associated with costs, yet the nature of these costs has remained elusive. Here, we demonstrate an important and clearly defined cost: reduced longevity. Experimental infections with six different isolates of H. defensa caused strongly reduced lifespans in two different clones of the black bean aphid, Aphis fabae, resulting in a significantly lower lifetime reproduction. However, the two aphid clones were unequally affected by the presence of H. defensa, and the magnitude of the longevity cost was further determined by genotype × genotype interactions between host and symbiont, which has important consequences for their coevolution.


Introduction

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

Insects and other arthropods are frequently infected with heritable microbial endosymbionts. Such symbionts can increase in frequency in the host population by reproductive manipulations that favour their transmission (e.g. the induction of cytoplasmic incompatibility, feminization, male-killing or parthenogenesis by Wolbachia) (Stouthamer et al., 1999) or by providing a net fitness benefit to their hosts (e.g. Jaenike et al., 2010). Aphids harbour a wide variety of bacterial endosymbionts (Oliver et al., 2010). The obligate endosymbiont Buchnera aphidicola is required for aphid survival and provides a nutritional benefit by synthesizing essential amino acids (Douglas, 1998). In addition to B. aphidicola, aphids commonly harbour facultative or secondary endosymbionts that may be beneficial but are not strictly required for aphid survival. One such symbiont belonging to the Enterobacteriaceae, Hamiltonella defensa (Moran et al., 2005), has been shown to increase aphid resistance to parasitoids (Oliver et al., 2003, 2005; Ferrari et al., 2004; Vorburger et al., 2009). Symbiont-conferred resistance provides a strong selective advantage in the presence of parasitoids (Herzog et al., 2007; Oliver et al., 2008), yet H. defensa only occurs at low to intermediate frequencies in natural population of aphids (Tsuchida et al., 2002; Simon et al., 2003; Oliver et al., 2006; Vorburger et al., 2009). This suggests that harbouring H. defensa also entails costs that select against infected aphids when selection by parasitoids is weak. Indeed, a study on pea aphids, Acyrthosiphon pisum, found that H. defensa-infected aphids declined in population cages when competing with uninfected aphids of the same clone in the absence of parasitoids (Oliver et al., 2008). However, the reasons for this decline remained unclear because H. defensa had largely positive effects on aphid life-history traits: infected lines had significantly shorter generation times and a slightly higher fecundity (Oliver et al., 2008). Comparisons of naturally infected and uninfected clones of the black bean aphid, Aphis fabae, also indicated additional benefits rather than costs of possessing H. defensa. In a sample of 24 different clones collected in Switzerland, nine were found to harbour H. defensa and they exhibited a higher daily fecundity on average than the 15 clones without H. defensa (Vorburger et al., 2009). Similarly, a comparison of life-history traits including a somewhat reduced set of 21 clones of A. fabae (seven harbouring H. defensa) revealed that adult size and offspring production were higher on average in the H. defensa-infected clones (Castañeda et al., 2010). However, these studies only focused on young adults, and the correlative evidence from natural infections does not prove a causal link between H. defensa infection and aphid fecundity. Thus, identifying the elusive costs of harbouring these defensive symbionts will require the experimental separation of symbiont-conferred effects from genetic variation of the hosts and a comprehensive assessment of fitness-relevant traits at all life stages. Both are readily possible in aphids. Their clonal mode of reproduction and the possibility to experimentally infect clones with facultative symbionts by microinjection permit the production of sublines with and without symbionts in the same genetic background. We used this approach to introduce six different isolates of H. defensa into each of two naturally uninfected clones of A. fabae. A life-table experiment using these lines revealed that infection with H. defensa strongly decreased aphid lifespan, resulting in lower lifetime reproduction, and that the magnitude of this longevity cost was determined by genotype × genotype interactions between host and symbiont.

Materials and methods

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

Aphid lines

Aphis fabae is an important pest aphid that is widely distributed in temperate regions of the northern hemisphere. It reproduces by cyclical parthenogenesis, with one sexual, oviparous generation over winter followed by many asexual, viviparous generations between spring and autumn. The two clones used in this study, A06-405 and A06-407, were collected during the asexual phase in summer 2006 from the same site in Switzerland. These clones possess different multilocus genotypes based on eight microsatellite loci (Coeur d’Acier et al., 2004), and they were diagnosed as uninfected with facultative endosymbionts by diagnostic PCR (Sandström et al., 2001; Russell et al., 2003; Tsuchida et al., 2006; Vorburger et al., 2009; McLean et al., 2011). Since their collection, they were maintained in the laboratory on broad beans (Vicia faba) under environmental conditions that ensure continued reproduction by apomictic parthenogenesis (16-h photoperiod at 20 °C). We generated H. defensa-infected sublines of these clones using a microinjection protocol as described in Vorburger et al. (2010), transferring symbiont-containing haemolymph from six different clones of A. fabae that were naturally infected with H. defensa. Collection details and microsatellite genotypes of the six donor clones as well as the two recipient clones are provided in Table 1. All of the donor clones exhibit complete or partial resistance to Lysiphlebus fabarum, the most important parasitoid of A. fabae (Vorburger et al., 2009; R. Rouchet & C. Vorburger, unpublished data).

Table 1.   Collection information and genotypes at eight microsatellite loci (Coeur d’Acier et al., 2004) for the eight clones of Aphis fabae used in this study. The two recipient clones used in the life-table experiment were experimentally infected with Hamiltonella defensa by microinjection of haemolymph from each of the six donor clones.
Sample IDCollection siteCollection dateHost plantFacultative symbiontMicrosatellite locus
AF-48AF-50AF-82AF-85AF-86AF-181AF-betaAF-F
Recipient clones
 A06-405St. Margrethen, Switzerland1 July 2006Chenopodium album315 317257 257167 177220 220217 219311 311280 282127 127
 A06-407St. Margrethen, Switzerland1 July 2006Chenopodium album315 315272 272177 177218 220215 215309 309280 282127 127
Donor clones
 A06-9La Spezia, Italy8 May 2006Vicia fabaHamiltonella defensa315 321257 272171 177220 222219 219309 309280 282127 127
 A06-30Sarzana, Italy8 May 2006Vicia fabaHamiltonella defensa315 315257 257177 177220 220219 219311 311266 282132 136
 A06-76La Grande Motte, France17 May 2006Chenopodium albumHamiltonella defensa315 315257 272192 204220 222217 219311 313280 280127 127
 A06-323Aesch, Switzerland27 June 2006Vicia fabaHamiltonella defensa315 315257 272177 177220 222215 215309 309280 280134 136
 A06-402St. Margrethen, Switzerland1 July 2006Chenopodium albumHamiltonella defensa315 315257 257177 177220 220219 219309 313280 280127 127
 Af6Zürich, Switzerland25 May 2004Euonymus europaeusHamiltonella defensa315 315257 257177 177218 222219 219311 317280 282127 134

Based on a combination of diagnostic PCR and of sequencing the amplicons of PCRs using the general bacterial primers 10F and 35R for the 16S ribosomal RNA gene (Sandström et al., 2001; Russell & Moran, 2005), H. defensa was the only facultative endosymbiont present in the donor clones. The six H. defensa isolates are labelled H 9, H 30, H 76, H 323, H 402 and H Af6, in reference to their clone of origin. Although the different donor clones were collected from as far apart as southern France and Switzerland (Table 1), they should not be regarded as coming from different, isolated populations. Aphids have a high dispersal ability (Llewellyn et al., 2003), and a population genetic survey using microsatellites found very low levels of genetic differentiation in A. fabae across Europe (C. Sandrock, J. Razmjou & C. Vorburger, unpublished data). We have no genetic information about the relatedness among the H. defensa isolates used here and their relatedness to known defensive isolates in other aphid species (e.g. Oliver et al., 2005), but phylogenetic analyses suggest that horizontal transmission among species occurs at least occasionally (Sandström et al., 2001; Russell et al., 2003).

Successful transmission of H. defensa by microinjection normally results in stable infections of clonal lines, as vertical transmission under laboratory conditions is virtually perfect. We confirmed the presence of H. defensa in the recipient lines by diagnostic PCR for the first three generations after transfection as well as immediately before use in the experiment. For one of the two recipient clones (A06-407), we also verified that protection against parasitoids by H. defensa is still expressed in the new genetic background (R. Rouchet & C. Vorburger, unpublished data). The transfected lines carried their H. defensa infections for between 20 and 40 generations prior to the experiment described later.

Life-table experiment

To estimate potential effects of the infection with H. defensa on aphid life-history traits, we carried out a life-table experiment similar to the one described in Vorburger (2005). The experiment took place in a climatized room under fluorescent light with a 16-h photoperiod at 20 °C. All 14 aphid lines (one uninfected and six infected from each clone) were split into eight replicates that were maintained on caged V. faba seedlings growing in plastic pots of 0.07 l volume. One replicate per line was assigned to a random position in eight different plastic trays (randomized complete blocks). To avoid the potential inflation of among-line variation by maternal or grand-maternal environmental effects carried over from the stock culture, we maintained the replicates for two generations (each generation on a fresh plant) before we assayed the life-history traits in the third generation after the split. The test generation was initiated by placing four adult females from the second generation on a new seedling, allowing them to reproduce for 4 h, and then removing the adults and all but one newborn nymph from the plant. These individuals represented the experimental cohort, which was checked daily for survival. After 6 days, we started monitoring the animals every 8 h to determine the time of their final moult (adult ecdysis), from which we calculated development time (duration from birth to adult ecdysis). We weighed all newly moulted adults to the nearest microgram on a Mettler MX5 microbalance (Mettler Toledo GmbH, Greifensee, Switzerland) to determine their fresh mass as an estimate of body size and then returned them on their plants. After that, their offspring were removed and counted daily until they died. To ensure that the aphids developed under favourable conditions and that they remained easy to find every day, we transferred the adults to new seedlings every 5 days. From the number of offspring produced over the first 7 days of reproduction, we calculated the daily fecundity (mean number of offspring produced per day) as an estimate of reproductive performance of young adults. This estimate could not be obtained for individuals that survived for <7 days after adult ecdysis, which was then treated as missing data. We also determined the lifetime reproductive output (total number of offspring produced from adult ecdysis until death) and the age at death. Finally, to obtain an overall fitness estimate for each individual, we used the complete life-table data to calculate inline image, following Service & Lenski (1982):

  • image(1)

where Sxi is the survival of individual i to age class x (one or zero), Bxi is the number of daughters produced by individual i in age class x, and FN is the finite rate of increase in the entire experimental cohort over the duration of one age class (i.e. 1 day in this experiment). FN is obtainable from the stable-age equation (Lenski & Service, 1982; equ. 4), which we solved iteratively. We found FN to be 1.37, which corresponds to the mean of the inline image (Lenski & Service, 1982). inline image is generally interpreted as the lifetime contribution of individual i to population growth, which is a useful measure of individual fitness (Lenski & Service, 1982). Two individuals were accidentally killed in a transfer during the experiment and had to be excluded from all analyses.

Statistical analyses

Aphid life-history traits were analysed with general linear models using the open-source statistical software R 2.9.2 (R Development Core Team, 2009). We tested for the effects of experimental block, aphid clone, subline and the clone × subline interaction. The variance among sublines and the variance explained by the clone × subline interaction was further partitioned into contributions from the variance between uninfected (H−) and infected (H+) sublines and the variance within H+ sublines (i.e. among different H. defensa isolates), using linear orthogonal contrasts. Because the block effect was far from significant in all analyses, we pooled the variance among blocks into the residual term. Survival data were analysed with a Cox proportional hazards regression, testing for the effects of aphid clone, subline and their interaction.

Results

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

Infection with H. defensa had no detectable effect on aphid development time, but development was significantly slower in clone A06-407 than in clone A06-405 (Table 2, Fig. 1a). Aphid body size measured as adult fresh mass did not differ significantly between clones, nor was there a significant difference among sublines, but the contrast between the means of uninfected (H−) and infected (H+) sublines indicated a slight but significant reduction in body size in the presence of H. defensa (Table 2, Fig. 1b). The fecundity of young adult aphids was similar for both clones but exhibited variation among sublines (Table 2, Fig. 1c). The marginally significant contrast between H− and H+ suggests that this was at least partly due to a slight reduction in the fecundity of infected aphids. However, this effect differed between the two aphid clones as indicated by the significant aphid clone × subline interaction (Table 2), which largely reflected the inconsistent effects of the different isolates of H. defensa on the two clones (Table 2, Fig. 1c).

Table 2.   General linear model results for the six life-history traits measured.
Source of variationd.f.MSFP
Development time
 Aphid clone10.5094.7960.031
 Subline60.1261.1900.319
 Between H− and H+10.1881.7770.186
 Among H+50.1141.0760.379
 Aphid clone × subline61.2721.2000.314
 Between H− and H+10.0220.2040.653
 Among H+50.1481.3950.234
 Residual890.106  
Adult mass
 Aphid clone10.0171.1940.278
 Subline60.0221.5610.168
 Between H− and H+10.0805.6390.020
 Among H+50.0110.7630.579
 Aphid clone × subline60.0130.9030.496
 Between H− and H+10.0281.9770.163
 Among H+50.0100.6680.649
 Residual890.014  
Daily fecundity
 Aphid clone11.4412.4090.125
 Subline61.3782.3040.043
 Between H− and H+12.4254.0550.048
 Among H+51.1851.9820.092
 Aphid clone × subline61.7722.9640.012
 Between H− and H+10.6081.0170.317
 Among H+51.9883.3240.009
 Residual710.598  
Age at death
 Aphid clone1643.215.050<0.001
 Subline6725.916.984<0.001
 Between H− and H+13981.793.248<0.001
 Among H+575.91.7770.125
 Aphid clone × subline6315.27.375<0.001
 Between H− and H+11143.326.775<0.001
 Among H+5148.53.4780.006
 Residual9642.7  
Lifetime reproduction
 Aphid clone17677.823.663<0.001
 Subline61578.54.865<0.001
 Between H− and H+16004.018.502<0.001
 Among H+5692.12.1330.068
 Aphid clone × subline61591.44.905<0.001
 Between H− and H+11895.05.8400.018
 Among H+51532.24.722<0.001
 Residual96324.5  
inline image (finite rate of increase)
 Aphid clone12.2208.5940.004
 Subline61.3735.314<0.001
 Between H− and H+14.59717.790<0.001
 Among H+50.7282.8160.020
 Aphid clone × subline60.8253.1940.007
 Between H− and H+10.0950.3680.546
 Among H+50.9723.7610.004
 Residual960.258  
image

Figure 1.  Life-history traits of infected and uninfected sublines of two clones of Aphis fabae: development time from birth to adult ecdysis (a), adult mass measured as fresh weight of newly ecdysed adults (b), daily fecundity averaged over the first 7 days of reproduction (c), age at death (d), lifetime number of offspring (e) and overall fitness measured as inline image, an estimate of individual contribution to population growth (f). The bars left of the gap contrast the means (+1 SE) of the uninfected and all infected sublines, the bars right of the gap depict the means of infected sublines separately for each isolate of Hamiltonella defensa.

Download figure to PowerPoint

The only really striking effect of harbouring H. defensa we observed was a reduction in longevity. An inspection of the survivorship curves (Fig. 2) shows clearly that in both clones, mortality rates differed among sublines, with uninfected aphids living longer on average than aphids harbouring H. defensa. This resulted in a significant subline effect in a Cox proportional hazards regression (LR χ2 = 78.1, d.f. = 6, < 0.001). Interestingly, the two clones were unequally affected by the presence of H. defensa (clone × subline interaction, LR χ= 38.3, d.f. = 6, < 0.001). The reduction in longevity by the different isolates of H. defensa was much more severe in clone A06-407 than in A06-405 (Fig. 2). This is also evident when survivorship is analysed as age at death (Fig. 1d). Individuals of both clones died younger when they were infected with H. defensa. This seems to be a rather general effect of this symbiont because the contrast analysis showed that the significant subline effect was largely due to the difference between H− and H+ (Table 2). However, clone A06-407, which produced longer-lived individuals when uninfected (44.0 days ± 1.3 SE vs. 32.3 ± 3.9 days in A06-405), suffered a reduction by almost two-thirds to 16.5 ± 0.9 days average lifespan, whereas A06-405 only suffered a reduction of about one-fourth to 23.6 ± 1.0 days average lifespan, reversing the order of their performance in the presence of H. defensa (Fig. 1d). This was reflected in the significant clone × subline interaction on the age at death, to which the contrast between H− and H+ contributed most of the variation. Yet the contrast analysis also showed that the effects of the different isolates of H. defensa on longevity depended on the aphid clone (Table 2).

image

Figure 2.  Survivorship curves of sublines of Aphis fabae clone A06-405 (a) and A06-407 (b), which are either uninfected (black lines) or experimentally infected with one of six different isolates of the defensive endosymbiont Hamiltonella defensa (coloured lines).

Download figure to PowerPoint

The marked differences in longevity we observed translated directly into differences in the most inclusive fitness estimates we obtained, namely the lifetime reproductive output and inline image, a life-table-based measure of an individual’s contribution to population growth, of which the means can be interpreted as an estimate of the finite rate of increase for each subline (Lenski & Service, 1982; Service & Lenski, 1982). Both measures varied significantly among sublines (Table 2, Figs. 1e, f), and the contrast between H− and H+ explained much of this variation (Table 2). In accordance with the stronger reduction in longevity, the negative effect on fitness was more pronounced in clone A06-407. However, different isolates of H. defensa contributed unequally to this fitness reduction: clone A06-407 suffered most from the presence of isolates H 30 and H 76, for example. This was supported by a significant clone × subline interaction for both traits, much of which is explained by the interaction of the different isolates within the H+ group and the two aphid clones (Table 2).

Discussion

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

By demonstrating substantial fitness costs of harbouring H. defensa, this experiment supports the notion that infected aphids are competitively inferior in the absence of parasitoids (Oliver et al., 2008), thus preventing the fixation of this facultative symbiont in natural populations. Our study is the first to provide a mechanistic understanding of these costs: infection with H. defensa shortens an aphid’s life. The shorter lifespan was the main reason for the reduced lifetime reproduction of infected aphids in our experiment, because the observed reductions in fecundity were small. Under the benign conditions of our laboratory experiment, the costs of harbouring H. defensa in terms of lifetime reproductive output were quite substantial (Fig. 1e), yet this result should be interpreted with caution. In the field, aphids are unlikely to live their full potential lifespans due to extrinsic sources of mortality such as predation, and in periods of populations growth (e.g. the exponential growth phase of aphid populations in spring), early reproduction contributes more to fitness than late reproduction (Stearns, 1992). Therefore, the fitness costs of harbouring H. defensa may be less pronounced under field conditions.

It is tempting to conclude that the reduction in longevity caused by H. defensa is mechanistically linked to the protection it provides against parasitoids. The protection results from the presence of toxin-encoding bacteriophages within H. defensa’s genome (Degnan & Moran, 2008a,b; Oliver et al., 2009). These toxins appear to kill the eggs or early larval stages of parasitoids, but they may also have negative effects on the aphids themselves, thus reducing their longevity. This hypothesis is yet to be tested. It will be particularly important to know whether the toxin genes are expressed constitutively or only upon attack by parasitoids.

Our experiment not only identified clear costs of harbouring H. defensa resulting from early mortality, it also showed that the magnitude of these costs depends on the host’s genetic background. One aphid clone was much more affected than the other. Furthermore, the costs depended on the exact combination of host clone and symbiont isolate, reflecting a genotype × genotype interaction between A. fabae and H. defensa. This may have important consequences for the frequencies of H. defensa in natural aphid populations as well as for the dynamics of host–symbiont coevolution. It suggests that the cost–benefit ratio of possessing H. defensa would differ among aphid genotypes. For certain host–symbiont combinations (e.g. aphid clone A06-407 with H. defensa isolate H 30 in the present experiment; Figs 1 and 2), the net effect on aphid fitness resulting from the symbiosis is likely to be negative, despite increased resistance to parasitoids. Such combinations are unlikely to be encountered in the field. In other combinations (e.g. clone A06-405 with H76), the benefits of increased resistance are likely to exceed the longevity cost. Obviously, the cost–benefit ratio will also be affected by the risk of attack by parasitoids, which will vary in space as well as in time. We can thus expect that the H. defensa-infected aphids we observe in the field do not represent random combinations of host and symbiont genotypes, but rather well-matching combinations that were favoured by natural selection because the protective effect of H. defensa comes at comparatively low costs. A potential test of this hypothesis would include a similar experiment using lines from which natural infections with H. defensa were removed by antibiotic curing (e.g. McLean et al., 2011), the prediction being that the gain in longevity would then be relatively modest. If this was indeed the case, it would help explain why comparisons of naturally infected and uninfected clones of A. fabae did not reveal any evidence for costs (Vorburger et al., 2009; Castañeda et al., 2010). Another explanation could be that just like other vertically transmitted symbionts, H. defensa relies on host reproduction for its own transmission. That is why its ability to protect aphids against parasitoids evolved in the first place, but the same would apply to the symbiont’s own effects on host survival. Upon successful infection of a host lineage, H. defensa should evolve to be less ‘virulent’ in that it shows reduced effects on host survival. However, evolution of reduced virulence towards the host could be constrained if it entailed reduced protection against parasitoids. Whether such a trade-off exists remains to be investigated.

To conclude, we show that in the black bean aphid, A. fabae, the defensive symbiont H. defensa is only helpful when required, i.e. when aphids are under strong selection by parasitoids. In the absence of parasitoids, harbouring H. defensa is associated with costs that are mostly due to a reduction in host longevity. The magnitude of the negative effect on host survival is to a large extent determined by genotype × genotype interactions between hosts and symbionts, which have important consequences for their coevolution.

Acknowledgments

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

We thank Daniel Bopp, Paula Rodriguez and Romain Rouchet for help with microinjections, Julia Ferrari for help with endosymbiont screening and Luis Cayetano for comments on the manuscript. This work was supported by the Swiss National Science Foundation (SNSF).

References

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

Data deposited at Dryad: doi: 10.5061/dryad.9120