Pea aphids

Pea aphids (Acyrthosiphon pisum, Homoptera, Aphididae) are cyclical parthenogens that feed on a range of legume species. The two most common parasitoid species that attack pea aphids are the braconids, Aphidius ervi and A. eadyi. The former also attacks a number of related aphid species, while A. eadyi is a specialist on pea aphids (Starý, González & Hall, 1980; Müller et al. 1999). In addition, pea aphids are subject to infection by a number of fungi, of which Erynia neoaphidis (Entomophthorales, Entomophthoraceae) is the most common, at least in southern England.

How pea aphids defend themselves against parasitoids and fungi is unknown. What is clear, however, is that encapsulation, the prime cellular defence mechanism in insects, plays no role (Milner, 1982; Henter & Via, 1995). At single sites, there is substantial between-clone variation in resistance against A. ervi, A. eadyi and E. neoaphidis (Henter & Via, 1995; Ferrari et al. 2001). Among populations, the picture is complicated by the host plant specialisation of pea aphids. The species feeds on a wide range of legume species, but individual genotypes are very specialised on different plant species (Via, 1991a, b, 1999; Sandström, 1994; Sandström & Petterson, 1994; Via, Bouck & Skillman, 2000). In two separate areas of the United States, pea aphids specialised on Medicago sativa were more resistant to A. ervi than aphids from Trifolium pratense (Hufbauer & Via, 1999; Hufbauer, 2001).

Within one field of M. sativa, Henter & Via (1995) did not observe an increase in the frequency of resistant clones in a population over the course of a field season despite variation in resistance among clones and substantial rates of parasitism. One explanation for this is that costs of resistance are important. So far, however, the nature and magnitude of the costs of resistance in pea aphids has proved elusive. We measured the resistance of over 28 pea aphid clones to A. ervi, A. eadyi and E. neoaphidis and, for a subset of clones, life-time fecundity on good and poor quality plants (Ferrari et al. 2001). No correlation across clones between resistance to any of the three natural enemies and life-time fecundity was found (Fig. 1 shows the results for A. ervi).

Fig. 1. Correlation amongst Acyrthosiphon pisum clones between susceptibility to Aphidius ervi and offspring production on (a) good and (b) poor quality host plants. See Ferrari et al. (2001) for experimental details.

Across clones, resistance against A. ervi and A. eadyi is positively correlated (Ferrari et al. 2001). Some clones are almost completely susceptible to both species, while others are completely resistant. However, several clones are more resistant to the specialist A. eadyi than to the generalist A. ervi (Fig. 2a). These results were complicated by the observation that some clones which survived attack by A. eadyi (our definition of resistance) produced virtually no offspring after successful defence. Parasites and parasitoids often shut down host investment in reproduction which is wasteful from the parasite's point of view. This may be why some of the clones produced no offspring, even though they did defend themselves successfully against the parasitoid. Cross-resistance between parasitoids and the fungus, E. neoaphidis, is not significantly correlated, though there is a positive trend (Fig. 2b; Ferrari et al. 2001).

Fig. 2. Correlation amongst Acyrthosiphon pisum clones between susceptibility to Aphidius ervi and susceptibility to (a) A. eadyi or to (b) Erynia neoaphidis. The open circles in (a) are the clones that do not mummify after attack by A. eadyi which fail to reproduce. The susceptibility values to E. neoaphidis are an index. See Ferrari et al. (2001) for experimental details.

We have limited evidence that the relative resistance of different aphid clones remains the same when challenged by different parasitoid genotypes. Resistance of British pea aphid clones to a British strain of A. ervi was highly and positively correlated with resistance against an American strain (unpublished data). Furthermore, there is no evidence that parasitoids collected on different host plant species survived better on aphids from their own host plant species compared with aphids from other host plant species (Hufbauer, 2001).

Drosophila melanogaster

D. melanogaster is among the more common Drosophila species breeding in fermenting fruits. Its larvae are attacked by several parasitoid species, of which the most common in Europe are the braconid, Asobara tabida, and the figitids, Leptopilina heterotoma and L. boulardi (Carton et al. 1986). Like many insects and other invertebrates (Nappi, 1975; Lackie, 1988a, b), D. melanogaster larvae are able to mount a cellular immune response, called encapsulation, against parasitism (Nappi, 1975; Rizki & Rizki, 1984; Strand & Pech, 1995). First, the parasitoid egg is recognised as non-self and haemocytes (blood cells) aggregate around the egg, forming a multi-layered capsule. Cytokines and other proteins are thought to be involved in mediating haemocyte aggregation behaviour (Strand & Pech, 1995). The subsequent deposition of melanin around the haemocyte-egg aggregate involves crystal cells (a separate class of haemocytes) and enzymes of the phenoloxidase-cascade (Strand & Pech, 1995). If deposition of melanin leads to a closed, blackened capsule, the parasitoid egg dies, possibly as a result of starvation or asphyxiation, or possibly due to toxic substances emanating from the capsule (Nappi et al. 1995; Nappi & Vass, 1998). When the host survives, the encapsulated egg is usually visible in the abdomen as a black dot.

The parasitoid species attacking D. melanogaster have evolved different counter-defence mechanisms to avoid their eggs becoming encapsulated. The egg chorion of A. tabida has proteinaceous filaments, which cause the eggs to stick to and subsequently become hidden in host tissue, away from circulating haemocytes (Kraaijeveld & van Alphen, 1994; Eslin et al. 1996). In contrast to this passive counter-defence mechanism of A. tabida, the Leptopilina species actively suppress the host's immune system. Ovipositing females inject virus-like particles into the host (Rizki & Rizki, 1990; Dupas et al. 1996), which are believed to enter the host's haemocytes and cause apoptosis (Rizki & Rizki, 1990).

The encapsulation ability of D. melanogaster and the level of counter-defence of A. tabida shows considerable variation, both at a geographic level and within populations (reviewed in Kraaijeveld & Godfray, 1999). The evidence so far (based mostly on the interaction between D. melanogaster and A. tabida; Kraaijeveld & Godfray, 2001) does not indicate that this variation can be explained by local adaptation (where hosts are better adapted to sympatric than to allopatric parasitoids or vice versa; see Kaltz & Shykoff, 1998, for a review of local adaptation in host-parasite systems). There is also no evidence that specific genotypes of the host are resistant to specific genotypes of the parasitoid. Rather, it seems that both defence of D. melanogaster and counter-defence of A. tabida are traits analogous to the running speeds of a predator and its prey. The relative resistance of different host populations appears to be independent of the parasitoid population attacking them and, similarly, the relative counter-resistance of different parasitoid populations is independent of which host population they parasitise. Genotype-specificity might play a role, however, in the interaction between D. melanogaster and L. boulardi, as there is some evidence that L. boulardi survives better in sympatric than in allopatric host populations (Carton, 1984; but see Kraaijeveld et al. 1998).

The pupae of D. melanogaster are attacked by a small number of parasitoid species, of which the pteromalid, Pachycrepoideus vindemiae, is the commonest in Europe (Carton et al. 1986). P. vindemiae lays its eggs in the space between the actual pupa and the puparium. As such, it is an ectoparasitoid and its eggs do not come into contact with the host's immune system. However, the parasitoid has to drill through the puparial wall to lay its eggs. As in the house fly example mentioned above, the puparial wall is the main defence barrier of D. melanogaster pupae against pupal parasitoids.

Costs of actual defence

Oviposition of a parasitoid egg into a D. melanogaster larva leads to an increase in the number of circulating haemocytes, followed by the production of a melanised capsule (Carton & Kitano, 1979; Nappi, Carton & Frey, 1991). Whether capsule formation is successful or not, these activities are likely to require resources.

Carton & David (1983) showed that successful encapsulation of L. boulardi eggs by D. melanogaster leads to smaller adult flies and reduced female fecundity. Encapsulation of A. tabida eggs also results in a reduction of adult size and lower fecundity in females (Fellowes, Kraaijeveld & Godfray, 1999a). This study also looked at the costs of actual defence against A. tabida in males. When only given the opportunity to mate once, males which had successfully encapsulated a parasitoid egg obtained less offspring from this single mating than males which had not been parasitised. However, no difference was found when males were allowed to mate more than once with females. Hoang (2001) showed that flies which had successfully encapsulated an egg of A. tabida had reduced resistance to desiccation and starvation. Flies surviving parasitism lived 12% shorter than unparasitised control flies when they were maintained under desiccating conditions and 14% shorter when deprived of food. As stress resistance is correlated with body size in D. melanogaster (Djawdan et al. 1998), the decreased resistance to desiccation and starvation is likely to be an additional consequence of the reduction in body size of flies surviving parasitism.

Activation of the immune system after parasitism by A. tabida, whether successful in encapsulating the parasitoid egg or not, results in a reduction in larval feeding rate (Tiën et al. 2001). Feeding rate, the frequency of retractions of the cephalopharyngeal skeleton, is an important determinant of larval competitive ability (Joshi & Mueller, 1988, 1996). The reduction in feeding rate in D. melanogaster larvae after parasitism may result in parasitised larvae having a lower competitive ability than unparasitised larvae. We found no reduction in feeding rate when D. subobscura larvae were parasitised by A. tabida (Tiën et al. 2001). D. subobscura is one of the most common Drosophila species on fermenting fruits and has no immune response against parasitoids (Kraaijeveld & van der Wel, 1994). The lack of a reduction in feeding rate in parasitised larvae of this species indicates that parasitism per se does not reduce feeding rate. This observation increases the likelihood that the reduction of feeding rate in D. melanogaster after parasitism reflects a cost of actual defence.

A further indication that defence requires limiting resources comes from the observation that the probability of encapsulation of L. boulardi eggs by D. melanogaster larvae decreases with larval crowding (Wajnberg et al. 1990). A major consequence of crowding is increased competition for food amongst the larvae, and reduced food intake may mean less resources available for defence. However, an alternative explanation is that the increased build-up of waste products in crowded conditions harms the larvae and has a negative effect on their immune system. Encapsulation was not reduced in crowded larvae parasitised by A. tabida or L. heterotoma (Tiën et al. 2001). Whether this reflects a difference between the immune responses to L. boulardi and the other two species, or simply a difference in the design of the experiment, is not clear.

Apart from encapsulation, the phenoloxidase cascade is also involved in puparium formation and melanin-precursors are incorporated into the hardening puparial wall (Fraenkel & Rudall, 1947). We may therefore expect activation of the immune system to have an effect on puparium formation: resources used up by the immune system cannot be used again for puparium formation. Indeed, puparia of larvae which have successfully encapsulated A. tabida eggs have thinner walls than those of unparasitised larvae (Fellowes et al. 1998b). A thinner puparial wall is likely to lead to greater vulnerability to physical damage or desiccation, and also to increased susceptibility to the pupal parasitoid, P. vindemiae (Fellowes et al. 1998b).

Female flies can potentially detect whether a male has successfully encapsulated a parasitoid egg, as the capsule is visible through the abdominal wall. Should she discriminate between capsule-bearing and unparasitised males? On the one hand, capsule-bearing male flies advertise having genes for resistance against parasitoids, which would make them more attractive as mates. On the other hand, males with a capsule also advertise the fact that they failed to avoid parasitism in the first place, and they have reduced insemination capabilities (see above), which would make them less attractive mates. These two alternative predictions were tested in experiments where virgin females were released in cages with equal numbers of capsule-bearing and unparasitised males. Matings with both types of males occurred with equal frequencies, and hence there is no evidence that female flies choose for or against capsule-bearing male flies (Kraaijeveld, Emmett & Godfray, 1997).

Costs of standing defences

In order to measure the costs of standing defences, we selected replicate lines of D. melanogaster for increased resistance to A. tabida and, in a separate experiment, to L. boulardi. In five generations, encapsulation ability increased from 5% to 60% in the experiments with A. tabida, and from 0.5% to 45% in the experiments with L. boulardi (Kraaijeveld & Godfray, 1997; Fellowes, Kraaijeveld & Godfray, 1998a). We measured total haemocyte numbers in the A. tabida-selected lines and their controls and found that the increased resistance to A. tabida is associated with a doubling of the number of circulating haemocytes (Kraaijeveld, Limentani & Godfray, 2001b). This is consistent with the positive correlation that Eslin & Prévost (1998) found across species when they measured encapsulation ability and haemocyte numbers in D. melanogaster and five related species. Preliminary results (unpublished data) show no increase in phenoloxidase activity in the selected lines.

We tested whether there was a cost of standing defence by comparing selected and control lines for a number of life-history traits, ranging from egg viability to female fecundity (Kraaijeveld & Godfray, 1997). When flies were reared with excess larval food, we found no difference between control and selection lines in any of these traits. This was not unexpected as trade-offs are more likely to become manifest when organisms are stressed (Stearns, 1992; Bergelson & Purrington, 1996). As competition for food between larvae is important in natural Drosophila populations (Atkinson, 1979), we varied the level of larval competition and compared control and selection lines. Fixed numbers of larvae from either control or selection lines were placed on food patches of varying size together with a fixed number of larvae from a standard reference strain (in our case an eye-colour mutant). Fig. 3 shows the main result: at higher levels of competition, larvae from both the lines selected for resistance against A. tabida and against L. boulardi showed a reduced relative competitive ability compared to larvae from their respective control lines (Kraaijeveld & Godfray, 1997; Fellowes et al. 1998a). We found no significant differences between individuals from control and selection lines in development time, adult size or fluctuating asymmetry.

Fig. 3. Competitive ability (relative to a reference strain of fly) at different levels of larval competition of Drosophila melanogaster lines selected for resistance against parasitoids (hatched bars) and their respective control lines (white bars). Left panel shows results of selection experiments using Asobara tabida; right panel shows results using Leptopilina boulardi. Each bar is mean of four replicate lines±S.E.; competitive ability is calculated as ln(e/(r+1)), where e is the number of surviving experimental (control or selection) flies and r the number of surviving reference flies; ‘no’, ‘low’, ‘int’ and ‘high’ levels of competition refer to 30 larvae on, respectively, 0.4, 0.2, 0.1 and 0.05 ml of a yeast suspension. See Kraaijeveld & Godfray (1997) for experimental details.

As mentioned above, larval feeding rate is an important determinant of competitive ability. We measured feeding rates of larvae from the control and selection lines and found that the reduction in competitive ability of the lines selected for high resistance is associated with a decrease in larval feeding rate (Fellowes, Kraaijeveld & Godfray, 1999b). There were no differences between pupae from control and selection lines in fat reserves (measured by ether extraction; unpublished data). The difference in feeding rate between larvae from selected and control lines was much greater than that found between parasitised and unparasitised larvae from the same population (see above). To investigate further the trade-off between resistance and competitive ability, we are currently conducting replicated experiments in which larvae from a population with high resistance are reared in crowded and uncrowded conditions (A. Sanders, unpublished). If the trade-off between resistance and competitive ability is direct and symmetrical, it is expected that resistance will decrease in the crowded lines.

The reason for the negative association between haemocyte number and feeding rate is not clear at the moment. One possibility is that there is a switch in the general energy budget of a larva away from investment in a trophic function to investment in the immune system. An alternative, more specific, explanation involves the early development of the larva: the head musculature and the haemopoietic organ (where haemocytes are produced) both originate from the same part of the embryo (Fullilove, Jacobson & Turner, 1977; Tepass et al. 1994). Increased allocation of tissue to the future haemopoietic organ may be at the expense of future muscle tissue. A third potential explanation (suggested to us by M. Siva-Jothy) is that a doubling of the number of circulating haemocytes increases the viscosity of the haemolymph, leading to lower rates of resource supply (such as glucose) to working muscles.

Given that parasitoid species use a variety of counter-defence mechanisms, trade-offs between resistance to different parasitoid species might occur. We found no evidence for this kind of trade-off, however. Selection for increased resistance to A. tabida and L. boulardi both lead to increased resistance to L. heterotoma (Fellowes, Kraaijeveld & Godfray, 1999c). This is consistent with positive correlations that have been found amongst isofemale lines for resistance against L. boulardi and L. heterotoma (Boulétreau & Wajnberg, 1986; Delpuech, Frey & Carton, 1994).

Cross-resistance is not necessarily symmetric (Fig. 4): selection for increased resistance to A. tabida leads to a very small, non-significant increase in resistance to L. boulardi (left panel of Fig. 4). In contrast, lines selected for increased resistance to L. boulardi are as resistant against A. tabida as the lines that were actually selected for resistance against that species (right panel of Fig. 4; Fellowes et al. 1999c). This suggests that the immune system of D. melanogaster has two components: a general component, effective against parasitoids in general, and a specific component, required in addition for resistance against the more specialised parasitoid L. boulardi.

Fig. 4. Cross-resistance (measured by percentage encapsulation) by Drosophila melanogaster lines selected for resistance against either Asobara tabida or Leptopilina boulardi (S, hatched bars) and their respective control lines (C, white bars). Left panel shows results of selection experiments using A. tabida; right panel shows results using L. boulardi. Within each panel, the left pair of bars shows encapsulation of the parasitoid species used in the selection experiment itself (A.t. in the case of the A. tabida-selected lines, L.b. in the case of the L. boulardi-selected lines), the right pair of bars shows encapsulation of the other parasitoid species. Each bar is mean of four replicate lines±S.E. See Fellowes et al. (1999c) for experimental details.

As mentioned above, encapsulation and puparium formation share a common pool of resources. Therefore, selection for increased encapsulation ability could have an effect on puparium formation and thus on resistance to pupal parasitoids. However, Green (2000) found no difference in the thickness of the puparial wall of the lines selected for increased resistance against A. tabida and their controls (Fig. 5a). He also directly examined the susceptibility of the two classes of hosts to parasitoid attack by offering females of the pupal parasitoid P. vindemiae a choice between pupae from the control and selection lines. Survival of flies after exposure to parasitoids did not differ between control and selection lines (Fig. 5b), confirming equal susceptibility. These results are consistent with the lack of correlation amongst isofemale lines for resistance to larval and pupal parasitoids (Delpuech et al. 1994).

Fig. 5. (a) Puparium thickness (determined by dividing the weight of empty puparia by their surface area, assuming puparia to be ellipsoid) of Drosophila melanogaster lines selected for resistance against Asobara tabida (S, hatched bar) and control lines (C, white bar). Each bar shows mean of four replicate lines±S.E. See Green (2000) for experimental details. (b) Survival of Drosophila melanogaster lines selected for resistance against Asobara tabida (S, hatched bar) and control lines (C, white bar) after exposure of pupae to Pachycrepoideus vindemiae. Each bar shows mean of four replicate lines±S.E. See Green (2000) for experimental details.