The relative fitness of Drosophila melanogaster (Diptera, Drosophilidae) that have successfully defended themselves against the parasitoid Asobara tabida (Hymenoptera, Braconidae)


Fellowes Department of Biology and NERC Centre for Population Biology, Imperial College at Silwood Park, Ascot, Berks. SL5 7PY, UK.; tel: 44 (0) 1344 294358; fax: 44 (0) 1344 294339.


Drosophila melanogaster larvae defend themselves against parasitoid attack via the process of encapsulation. However, flies that successfully defend themselves have reduced fitness as adults. Adults which carry an encapsulated parasitoid egg are smaller and females produce significantly fewer eggs than controls. Capsule-bearing males allowed repeated copulations with females do not show a reduction in their number of offspring, but those allowed to copulate only once did. No differences were found in time to first oviposition in females, or in time to first copulation in males. We interpret the results as arising from a trade-off between investing resources in factors promoting fecundity and mating success, and in defence against parasitism. The outcome of this investment decision influences the strength of selection for defence against parasitism.


Encapsulation is a cellular immune response through which arthropods defend themselves against macroscopic foreign bodies ( Salt, 1970) and which may also play a role in defence against viruses ( Washburn et al., 1996 ). Specialized haemocytes recognize invading bodies as non-self and cause other haemocytes to aggregate and form a capsule. In insects, this process is central to defence against endoparasitoid wasps and flies whose larvae develop in the haemocoel ( Godfray, 1994). A cascade of reactions involving the tyrosine–phenyloxidase pathway causes the melanization of the capsule. This results in the death of the immature parasitoid by asphyxiation ( Fisher, 1963) or through the production of necrotizing compounds ( Nappi et al., 1995 ).

Parasitoids have evolved a variety of ways to counter these host defences, including concealment in host tissues away from circulating haemocytes, avoidance of recognition and through debilitating the immune response by toxins or symbiotic viruses ( Strand & Pech, 1995). Drosophila melanogaster larvae exhibit both within- and between-population genetic variation in ability to encapsulate the eggs of two of their main parasitoids, Asobara tabida (Hymenoptera, Braconidae) and Leptopilina boulardi (Hymenoptera, Eucoilidae) ( Carton & Boulétreau, 1985; Kraaijeveld & van Alphen, 1995). In Drosophila, the capsule formed by the larva is retained during metamorphosis and is often visible through the abdominal wall of the adult fly. Leptopilina spp. appear to avoid encapsulation by coating their eggs with ‘virus-like particles’ which prevent circulating haemocytes from recognizing them as foreign ( Dupas et al., 1996 ). In contrast, Asobara tabida can avoid encapsulation by embedding its eggs in host tissue ( Kraaijeveld & van Alphen, 1994; Eslin et al., 1996 ), and there is an inverse relationship across populations between the frequency of encapsulation and the fraction of eggs embedded in host tissue ( Kraaijeveld & van Alphen, 1994). Additive genetic variation for resistance against both A. tabida and L. boulardi has been documented in D. melanogaster and increased resistance against each parasitoid has been obtained by artificial selection ( Kraaijeveld & Godfray, 1997; Fellowes et al., 1998 a). Recently, it has been demonstrated that reduced larval competitive ability is a correlated response to selection for resistance against A. tabida ( Kraaijeveld & Godfray, 1997; Fellowes et al., 1998 a).

Where appropriate genetic variation exists, as it appears to do so in all populations studied so far, host resistance to parasitoid attack will evolve as long as the benefits of surviving parasitism outweigh the costs of maintaining the defences. Any fitness benefits are reduced if hosts that survive parasitism through encapsulation have reduced survival or longevity, or if their attractiveness as mates is impaired. Moreover, even if benefits do outweigh costs, the speed with which resistance traits can respond to selection will be influenced by a reduction in fitness of flies that survive parasitoid attack. The formation of the capsule involves the mobilization of resources that might otherwise be used for growth, and this may affect the performance of the adult insect. Additionally, attack by adult parasitoids or the action of the immature parasitoids may damage and hence reduce the fitness of hosts, even if they do eventually manage to destroy the parasitoid. With the exception of a study on D. melanogaster and L. boulardi ( Carton & David, 1983), which we return to in the discussion, the effect of parasitoid attack on the fitness of survivors has not been studied. Here we compare the size, fecundity and mating success of D. melanogaster that have successfully encapsulated A. tabida with control flies that have not suffered parasitoid attack.

Materials and methods

General methods

The D. melanogaster used originated from Lyon, France, and had been kept in culture for 2½ years in a large outbred population. The A. tabida strain used had been collected from Sospel, France, and had been cultured in the laboratory for 13 years on D. subobscura, a species which never encapsulates this wasp. Both fly species were reared at intermediate densities in bottles containing a baker’s yeast/sugar medium. This strain of D. melanogaster successfully encapsulates ≈55% of the eggs laid by the Sospel wasp strain ( Fellowes et al., 1998b ). Parasitized flies were obtained by allowing five A. tabida to search for 24 h in 8-oz bottles containing ≈200 second instar larvae. Between 70 and 90% of the larvae were attacked under these conditions, with the large majority receiving a single egg as superparasitism is rare in this parasitoid when unattacked hosts are available. Capsules are clearly visible in pupae, and this allowed the collection of those flies which had successfully encapsulated. The flies were sexed by the presence or absence of sex combs which are visible in mature pupae. All experiments were conducted at 20 °C under a 16 h:8 h light:dark regime. Larval competition is often severe in wild populations of D. melanogaster ( Atkinson, 1979), increasing variability in development rates and body size. These confounding effects were avoided by maintaining the populations with a large excess of food.

Body size

The length of the left wing (from wing tip to the major costal break) and the thorax was measured to the nearest 0.01 mm under a ×40 binocular microscope. Thirty individuals of each sex, with and without capsules, were measured. The data were analysed using t-tests with the assumption of unequal variances.

Adult fecundity

To assess whether the presence of a capsule influenced female fecundity, or the fecundity of a male’s mate, nine replicates of each of the following four pairs of flies were set up: (a) both male and female with no capsule; (b) both male and female with capsule; (c) female without and male with a capsule; and (d) female with and male without a capsule. On emergence, each pair of flies was placed in a small vial (25 × 75 mm) with standard Drosophila culture medium and a smear of baker’s yeast. The flies were transferred daily to new vials and the date of first oviposition and the number of eggs laid each day counted. On days 7–10, the eggs were allowed to hatch and hatching success was calculated. The data were analysed using a three-way repeated measures ANOVA with day, presence of a capsule in the female and presence of a capsule in the male as factors. Wing and thorax length were measured, using the methods described above. The experiment was conducted over a 10-day period as it has been shown that mean life expectancy of adult flies is from 1.3 to 6.2 days in natural populations ( Rosewell & Shorrocks, 1987).

Male fecundity in the absence of remating

Any effect of the presence of a capsule on ejaculate size or quality might not be detected in the previous experiment if females compensated for this by remating. We thus explored if the fecundity of a female was influenced by whether or not her mate contained a capsule when mating occurred once. Three-day-old virgin females were allowed to mate once with virgin males with a capsule, or virgin control males that had not been attacked. Sixteen replicates of each treatment were performed. Individual females were placed in vials and treated in the same way as the previous experiment, except that all females were allowed to continue ovipositing until they died. This study was carried out for a longer period as it was thought that fecundity differences may only become apparent as sperm became depleted. Analysis was performed using a two-way repeated measures ANOVA with presence of a capsule and day as factors. Additionally, time to copulation with the control or capsule-bearing males was noted.


Body size

Females with capsules that had survived parasitism had significantly shorter wings and shorter thoraxes than control females. Males with capsules also had significantly shorter thoraxes, but there was no significant difference in wing length ( Table 1).

Table 1.   Comparison of size (in mm) between flies containing a capsule and flies that were not exposed to parasitoids. The differences between the means were analysed by t-tests. Thumbnail image of

Adult fecundity

The daily egg output of females changed markedly and significantly over the course of the 10-day experiment ( Table 2). Females that survived parasitism had lower fecundity than control females ( Fig. 1). The capsule status of the female’s mate had no effect on her fecundity and no higher interactions were significant. Differences in fecundity were not related to body size. In an analysis of covariance of the total number of eggs laid by each female, the main effect of body size was not significant (F1,33 = 0.766, > 0.35) but there was a significant effect of capsule presence (F1,33 = 7.96, < 0.01). The time to first oviposition (mean = 26.6 h) and the probability of egg hatching (mean = 90.5%) were not influenced by the presence of a capsule in the male or female fly.

Table 2.   The effect of defence against parasitoid attack on offspring number with repeated mating, analysed using a three-way repeated measuring ANOVA. Thumbnail image of
Figure 1 The number of eggs laid per day (with standard error) by female D. melanogaster in the first 1.

Figure 1 The number of eggs laid per day (with standard error) by female D. melanogaster in the first 1.

0 days of oviposition. Solid lines represent flies bearing capsules and broken lines represent control flies.

Male fecundity in the absence of remating

There was no difference in the time to first copulation (mean = 72 min) between males carrying or not carrying a capsule (Mann–Whitney test: U=184.5; > 0.65). In a two-way repeated measures ANOVA of the eggs laid per day by a female, the main effect of capsule status of the male was significant, as was the interaction between day and the capsule status ( Table 3). These data are summarized in Fig. 2.

Table 3.   The effect of defence against parasitoid attack on offspring number of males following a single copulation, analysed using a two-way repeated measures ANOVAThumbnail image of
Figure 2.

  The number of eggs laid per day (with standard error) by female D. melanogaster in the 33 days following a single insemination by a male bearing a capsule (solid line) or by a control male (broken line).


Female flies that escape parasitism are smaller and have reduced fecundity; males have shorter thorax length but their wing length is not significantly affected. Under some circumstances, mates of males that bear an encapsulated parasitoid egg have lower fecundity. We found no evidence that development time or the viability of eggs in the next generation were influenced by parasitoid attack.

A problem with studying the fitness consequences of surviving parasitism is to separate the costs of mounting a successful defence from the effects of unmeasured third variables that may simultaneously influence the probability of survival and the fitness of the fly when adult. Suppose that there is a trade-off between a fly’s ability to avoid parasitism in the first place and correlates of adult fitness such as size or fecundity. Our results could then be explained as a consequence of the survival of insects with poorer adult performance. However, recent artificial selection experiments ( Kraaijeveld & Godfray, 1997) designed to detect genetic trade-offs between resistance against parasitoids and other components of fitness showed no relationships of this sort (a trade-off was found between encapsulation ability and larval competitive ability, but this was only manifest at much higher levels of competition than occur in the experiments described here). While such a trade-off is conceivable, a much more likely confounding effect is a positive association between the ability to survive parasitism and adult fitness. Such a correlation might simply reflect general phenotypic variance in overall condition. If the correlation was present, and there were no costs to surviving parasitism, we would expect flies with capsules to be superior to control flies because poor condition flies would have been eliminated by the parasitoid. If there are costs then they would be underestimated. We thus believe that the costs of surviving parasitism revealed by our experiments are a lower bound to their true cost in nature.

Flies that had successfully escaped parasitism tended to be smaller than control flies. This may be due to the direct negative effects of parasitoid attack, or it may be a consequence of the redirection of resources from growth towards defence. Studies of Drosophila have usually found a positive correlation between size and most components of fitness (e.g. Partridge & Farquhar, 1983; Santos et al., 1994 ). Interestingly, the reduction of fecundity we observed in females that had survived parasitism appeared to be statistically independent of body size, though this is largely due to the effects of capsule status on body size being relatively modest.

No difference was found in the time from eclosion to the beginning of first oviposition indicating that defence against parasitism does not influence development time, but as we recorded eclosion and oviposition every 24 h, our statistical power to detect subtle differences is not great ( but see Fellowes et al., 1998b ). Longer development time is likely to have severe consequences for Drosophila which often compete in patches with many conspecifics. Longer development time might also increase the risk of further parasitoid attack. It is possible that the flies have evolved to maintain development rate at the expense of adult size. The absence of a significant reduction in male wing length may be another example of a selected channelling of costs because wing length plays an important role for males in mating ( Ewing, 1964).

Encapsulation results in a reduction in the total number of offspring fathered if a male is allowed to mate only once. Normal male Drosophila melanogaster transfer 2000–4000 sperm per ejaculate, of which the female retains 500–1000 in the paired spermathecae and seminal receptacle ( Ashburner, 1989). There are a number of potential explanations for the reduction in offspring fathered by males with capsules. The most reasonable is that those with capsules have smaller testes and hence produce smaller numbers of spermatozoa in their ejaculate. Kraaijeveld et al. (1997 ) have shown that there is no difference in the likelihood of capsule-bearing male D. melanogaster obtaining matings when compared to control males. This is supported by the lack of a difference in the time to first copulation between males with capsules and the controls. However, female D. melanogaster preferentially copulate with virgin males ( Markow et al., 1978 ), and differences in male attractiveness may only become evident after the first copulation.

Some of our results parallel those of Carton & David (1983) who worked with a different parasitoid, Leptopilina boulardi. While they found a significant reduction in the overall fecundity of capsule-bearing females, unlike us they did not find a significant difference in the total numbers of eggs laid in the first 10 days after eclosion. They also found a significant reduction in size of both male and female D. melanogaster which had successfully defended themselves against parasitoid attack. The differences between our results and their’s could be a result of different patterns of resource allocation in the fly populations used. However, the results of the female fecundity experiments are similar in direction, if not in strength. Therefore, although the different parasitoid species interact with their hosts physiology in different ways, the outcome is a similar reduction in fitness. Carton & David (1983) did not investigate the effect of encapsulation on male fitness, nor the consequences to a female of mating with a male bearing a capsule.

To conclude, the fitness costs found by us and Carton & David (1983) are of interest because they affect the rate at which defence against parasitoids will evolve. The results suggest that after parasitism the fly has to balance resource allocation between defence and other fitness-related traits. If the fly allocates too few resources to defence it will fail to kill the parasitoid and die. If it allocates too much it may survive, but its reproductive success as an adult is severely diminished. Our results also suggest that the trade-off between competing resource demands may differ in the two sexes, and we speculate that flies may protect those functions most highly correlated with fitness in each sex.


M.F. was supported by NERC award GT4/95/178/T. We also thank Richard Cooke, Emma Wittmann and two anonymous referees for helpful comments on the manuscript.