Introduction

In host–parasite or host–pathogen interactions, antagonistic selective pressures can lead to a coevolutionary arms race based on adaptations and counter-adaptations in each partner. However, this process requires occurrence of genetic variation for traits involved in the final outcome of the interaction (Thompson, 1994; Sorci et al., 1997). Such variation has often been described in plant–pathogen interactions (Frank, 1994), but little is known regarding interactions between insect parasitoids and their hosts. In this kind of interaction, where larval stages of the parasitoid are dependent on their arthropod host (Godfray, 1994), reciprocal selective pressures are particularly strong as only one partner can survive infestation.

To defend themselves against parasitoid attacks, host species have set up diverse behavioural and/or immune defences (Gross, 1993; Strand and Pech, 1995). In insects, the most well-known immune defence against parasitoids is the formation of a multi-cellular, melanized capsule around the parasitoid egg. This encapsulation reaction, when successful, leads to parasitoid death and host survival (Carton and Nappi, 1997). To counteract host immune defences, parasitoids have developed various strategies, based mainly on the use of virulence factors (Strand and Pech, 1995; Pennacchio and Strand, 2006). Surprisingly, despite their relevance for the study of coevolution, little data are currently available on host–parasitoid interactions with regard to the genetic bases of resistance and virulence traits (Kraaijeveld et al., 1998; Dupas et al., 2002).

Variability in both resistance (ability to neutralize the parasitoid egg using immune defences) and virulence (ability to overcome the host immune defences) has been documented in only three host–parasitoid interactions, two of them involving the host species Drosophila melanogaster (Kraaijeveld and van Alphen, 1994; Dupas et al., 2002) and the other, the pea aphid Acyrthosiphon pisum (Henter, 1995; Henter and Via, 1995). In these models, genetic variation for both resistance and virulence has been observed and resistance and virulence components are thus predicted to coevolve under antagonistic selective pressures.

This interpretation is, however, unlikely to explain all resistance and virulence patterns in natural populations (Dupas et al., 2002). Indeed, as stressed by Thompson (1999), selective pressures exerted by other parasitoid and host species will also influence the geographic patterns of resistance–virulence in a given pairwise interaction. Unfortunately, data on resistance–virulence traits in complex interactions involving at least one parasitoid and different host species or one host species and its parasitoids are lacking. To estimate potential trade-offs or genetic constraints between resistance and virulence traits, it is now necessary to analyse several pairwise systems involving different interacting host–parasitoid species.

Drosophila species and their parasitoids represent one of the best models to analyse multiple species host–parasitoid interactions and to characterize resistance and virulence patterns in the field. Variability in resistance to two different parasitoids, Asobara tabida (Kraaijeveld and van Alphen, 1995) and Leptopilina boulardi (Dupas et al., 2002), has been observed in D. melanogaster, and variability in virulence has been reported in the parasitoid species L. boulardi against two Drosophila host species (Dupas and Boscaro, 1999). The genetic determinism of these traits has, however, only been studied in the D. melanogaster/L. boulardi pairwise interaction, using crosses between laboratory isofemale lines. In this system, simple genetic determinism was found, with one diallelic locus explaining resistance and one diallelic locus being responsible for virulence (Carton et al., 1992; Dupas et al., 1998). Interestingly, in tropical Africa, L. boulardi can infest several other species of the melanogaster subgroup of Drosophilidae, including D. yakuba (Carton and Nappi, 1991). This species is widespread on the African mainland (Lachaise et al., 1988) and was also described recently on some islands (Lachaise et al., 2000). Variability in virulence against D. yakuba has been observed in natural populations of L. boulardi and is also determined by a single diallelic locus (Dupas and Boscaro, 1999; Dupas and Carton, 1999).

The aim of this work was to quantify genetic variation for resistance in D. yakuba in natural populations and to analyse its genetic basis. We compared the genetic systems for resistance to L. boulardi in D. melanogaster and D. yakuba. Besides providing genetic information on a second pairwise host–parasitoid interaction, these data enable us to compare for the first time a parasitoid interaction with two different host species, thereby providing a complete picture of virulence–resistance traits in a multi-species host–parasite system.

Interestingly, combined data show that resistance patterns differ in closely related Drosophila host species and that different specificity levels can be observed in a single parasitoid species. Altogether, the L. boulardi/D. melanogaster interaction can be well explained by a ‘gene-for-gene’ (‘incompatibility’) pattern, whereas interaction of L. boulardi with D. yakuba would be better described by a ‘compatibility’ pattern. Occurrence of genetically based variations of virulence and resistance in L. boulardi interactions with two different Drosophila host species makes this model unique for understanding the evolution of complex host–parasitoid relationships and addressing the question of the specificity of resistance and virulence traits.

Materials and methods

Insects

Reference lines of L. boulardi

L. boulardi isofemale lines ISy and ISm have been described by Dupas et al. (1998). ISm (Gif stock, no. 431) derives from a single female collected in the Nasrallah oasis (Tunisia). ISy (Gif stock, no. 486) derives from a single female originating from a mass culture of a Brazzaville (Congo) population. ISm parasitoids are highly virulent against D. melanogaster but are always encapsulated in D. yakuba. On the contrary, parasitism success of ISy parasitoids in D. melanogaster depends on the resistance/susceptibility genotype of the flies. The ISy line was used by Carton et al. (1992) to study the genetics of resistance to L. boulardi in D. melanogaster.

Both ISm and ISy lines were reared on a susceptible D. melanogaster strain (Gif stock, no. 1333), at 25°C. After emergence, adults of both lines were kept at 18°C on agar medium with honey.

Reference lines of D. yakuba

The isofemale lines 1880-D (R1 line) and 1907 (R2 line) were selected from two populations in Tanzania and further used to analyse the genetic basis of resistance. R1 and R2 show opposite immune capacities against the ISy line of L. boulardi (see Table 1). These isofemale lines were obtained from a single inseminated founder female collected in the field (David et al., 2005) that was used to initiate a full sib line. The sib lines had been maintained in the laboratory for 8 years at the time of experiments, and were thus probably homozygous for all the loci potentially involved in resistance to the parasitoid wasp. The R1 line was used by Dupas et al. (1998) to study the genetics of virulence of L. boulardi against D. yakuba.

Table 1 Collection sites of the D. yakuba reference lines used for genetic analyses and percentage of encapsulation of L. boulardi eggs from the ISy and ISm lines
Full size table

Natural populations of D. yakuba

Seven D. yakuba populations were collected throughout the species distribution area, in the Afrotropical region, where D. melanogaster is also present (Lachaise et al., 1988). Several isofemale lines or multi-female strains were obtained from these populations. Occurrence of geographic variation in resistance to L. boulardi was tested using either multi-female strains or pools of 6–24 isofemale lines (Table 2). This method was previously shown to preserve the genetic variability of a population (David et al., 2005). All D. yakuba strains were raised at 25°C using a standard Drosophila cornmeal–yeast–agar medium.

Table 2 D. yakuba strains collection sites, collection dates, status and percentage of encapsulation of L. boulardi eggs from the ISy and ISm lines
Full size table

Bioassay procedure

For bioassays, batches of 30 second instar D. yakuba larvae (48 h old) were submitted to parasitism by three L. boulardi females for 4 h. All the bioassays were performed at 25°C. Encapsulation ability was estimated 48 h later by dissecting late third instar larvae. At this time, a melanized capsule is found in resistant larvae, but not in susceptible ones. The encapsulation rate was calculated as the ratio of encapsulated parasitoid eggs to the number of parasitized hosts, using data from only monoparasitized larvae.

Genetic analysis

Crossing procedure

Two generations of reciprocal crosses (Table 3) between the R1 and R2 lines, respectively susceptible and resistant to the parasitoid ISy line, were performed. The two parental lines, the two F1 hybrids and the two F2 hybrids were tested and compared for their encapsulation ability.

Table 3 Percentage of encapsulation of L. boulardi ISy eggs by D. yakuba larvae originating from different crosses involving the R1 and R2 lines
Full size table

Statistical analysis

The mode of inheritance of D. yakuba resistance to the L. boulardi ISy line was assessed using analyses of variance (ANOVA) with the generalized linear model procedure, assuming binomial error term distribution. Analyses were performed according to the method described by De Belle and Sokolowski (1987), which was previously used to analyse the genetic determinism of resistance of D. melanogaster to the parasitoid species L. boulardi and A. tabida (Carton et al., 1992; Benassi et al., 1998). The following comparisons were made (cross numbers refer to those described in Table 3): difference between parental strains (1 vs 2), dominance or additive effects (1+2 vs 3+4), deviation from an autosomal mode of inheritance (3 vs 4).

The single-gene model with complete dominance of the resistance allele was tested using Mendelian analysis with the ratio of susceptible to resistant larvae, as described by Carton et al. (1992). The observed and expected ratios of susceptible to resistant were analysed using χ2 analysis. All statistical analyses were performed using Systat 10 software (SPSS Inc.).

Results

Occurrence of variability in resistance of D. yakuba to L. boulardi

Two D. yakuba isofemale lines, R1 and R2, originating from Tanzanian populations, were selected from a range of candidate lines for their high (92.9±2.4%) and low (18.8±3.9%) encapsulation levels of eggs from the L. boulardi wasp ISy line, respectively. Both lines encapsulated almost 100% of the eggs of the reference ISm line (99.4 and 100%, respectively) (Table 1). Based on these results, R1 and R2 were considered as susceptible and resistant, respectively, to the ISy line of L. boulardi.

Genetic determinism of resistance

Encapsulation rates of ISy parasitoid eggs by D. yakuba larvae originating from different crosses performed with the R1 and R2 isofemale lines, and results of the contrast analysis of variance are presented in Tables 3 and 4, respectively. As expected, the R1 and R2 lines differed significantly in their resistance to infestation by ISy parasitoids (F=436.79, P<0.001). F1 hybrids exhibited significantly higher resistance than the mean parents, showing that resistance was dominant (F=252.74, P<0.001). Reciprocal F1 hybrids were not significantly different from each other, excluding the involvement of non-autosomal inheritance (F=1.05, P=0.305). Larvae from the different crosses were classified into two classes (resistant and susceptible) according to their ability to encapsulate ISy eggs. The single gene model hypothesis was tested by comparing susceptible vs resistant proportions (S:R) with expected Mendelian proportions, using χ2 analysis. The results (Table 5) were consistent with the hypothesis of a single major segregating locus with two alleles, the resistant one being completely dominant. The locus responsible for resistance of D. yakuba to ISy parasitoids was named RlbISy (for resistance to L. boulardi of the ISy type) and the resistant and susceptible alleles, RlbISy+ and RlbISy−, respectively.

Table 4 Contrast ANOVA for encapsulation of L. boulardi ISy eggs by D. yakuba larvae originating from crosses involving the R1 and R2 lines (generalized linear model with binomial error term)
Full size table
Table 5 Deviation of the proportions of (susceptible vs resistant) (S:R) larvae in the progeny of different crosses from the proportions expected under a one-locus model with dominance of the resistance allele
Full size table

Genetic variability in natural populations

Encapsulation rates of the two reference L. boulardi lines, ISy and ISm, were recorded in seven D. yakuba populations (Table 2). Figure 1 presents the geographic pattern of resistance to ISy parasitoids.

Figure 1
figure 1

Geographical distribution of D. yakuba resistance to ISy parasitoids in tropical Africa. The resistance level (in black) was estimated from the percentage of ISy eggs encapsulated by D. yakuba larvae.

Full size image

A high level of variation in resistance to the ISy line was observed in the field, which confirmed the results obtained with the Tanzanian R1 and R2 isofemale lines: resistance to ISy in D. yakuba shows significant variability. Encapsulation rates varied from 6 to 98% but most populations encapsulated ISy eggs at a high frequency (from 77.6 to 97.9%). Only populations from the São Tome island were susceptible to ISy infestation (6.5 and 7.0% of ISy eggs encapsulated). Resistance was thus found either at a very high or a very low frequency in D. yakuba Afrotropical populations. By contrast, there was no variability between D. yakuba strains for resistance to the ISm parasitoid line, with 100% of eggs encapsulated in each strain (Table 2). Despite intense investigations, no D. yakuba population susceptible to the ISm line of L. boulardi has ever been found.

Discussion

Genetic determinism of parasitoid resistance in Drosophila

The present study reports occurrence of variation in the ability of D. yakuba to encapsulate eggs of the ISy line of L. boulardi. Using crosses between lines having contrasted resistance levels, we demonstrate that resistance is determined by a single major locus, which we termed RlbISy.

Considering the high number of genes involved in insect immune pathways (Irving et al., 2001; Zettervall et al., 2004), variation in resistance to parasitoid wasps was expected to be determined by multigenic systems (Sorci et al., 1997). However, in all cases in which genetic determinism of resistance to parasitoids has been studied – D. melanogaster/L. boulardi, D. melanogaster/A. tabida and now D. yakuba/L. boulardi – differences between lines resistant and susceptible to a given parasitoid have always been explained by a single diallelic locus. In D. melanogaster, two loci, named Rat and Rlb (labelled as Rst(2)Lb in Flybase; ID number: FBgn0016729), localized 35 centimorgans apart, are responsible for resistance to the parasitoids A. tabida and L. boulardi, respectively (Carton et al., 1992; Benassi et al., 1998; Poirié et al., 2000). The use of isofemale lines, which are not representative of the extent of genetic variation in the field, might favour the recovery of simple genetic systems (Kraaijeveld et al., 1998). However, studies by Orr and Irving (1997), dealing with genetic variation of resistance to A. tabida in D. melanogaster populations from different parts of Europe, also concluded on a simple genetic basis for resistance. The advantage of performing analyses with well-characterized strains is that it allows the precise localization and identification of resistance loci (Hita et al., 1999; M Poirié, unpublished data). These loci are expected to contain genes involved in the response to parasitoid attacks, showing enough polymorphism to respond to selection pressures and thus potentially evolving under coevolutionary processes.

The genetic bases of interactions between two different lines of L. boulardi and the two host species D. melanogaster and D. yakuba are summarized in Figure 2. It is the first complete report of genetic interactions between a parasitoid and two different host species.

Figure 2
figure 2

Genetic interactions matrix in the Drosophila–L. boulardi reference system. The interactions between the reference lines ISy and ISm of the parasitoid L. boulardi and the reference lines of the host species D. melanogaster and D. yakuba are illustrated. Resistance in each host species is conferred by one major locus, and two independent loci are responsible for virulence of the parasitoid against the two host species. In the L. boulardi–D. melanogaster system (cells 1a to 1d), the host alleles Rlb+ and Rlb− are responsible for resistance and susceptibility to ISy infection, and the parasitoid alleles ISm+ and ISm− are responsible for virulence and avirulence of the parasitoid against the host. Parasitism success is the rule in D. melanogaster (cells 1a, 1b and 1c), except when a resistance allele Rlb+ is present without any virulence allele ISm+ (cell 1d). In the L. boulardi–D. yakuba system (cells 2a to 2d), the host alleles RlbISy+ and RlbISy− are responsible for resistance and susceptibility to the ISy infection, and the parasitoid alleles ISy+ and ISy− are responsible for virulence and avirulence of the parasitoid against the host R1 line. In this system, parasitism failure is the rule (cells 2a, 2b and 2d), except when two virulence alleles ISy+ are present with two susceptibility alleles RlbISy− (cell 2c).

Full size image

In the L. boulardi–D. melanogaster system, the success of parasitism is the rule; failure only occurs when the parasitoid has no virulence alleles (ISm−/ISm−, ISy line) and when the host is resistant (at least one Rlb+ allele, resistant strain) (Dupas et al., 2002). This system resembles the ‘gene-for-gene’ (or ‘incompatibility’) model of plant–pathogen interactions (Briggs and Johal, 1994). Here one parasitoid genotype (ISm line) has a ‘universal virulence’ (Frank, 1994), which means that it can infest all D. melanogaster flies, whatever their genotypes.

By contrast, the general outcome in the L. boulardi–D. yakuba interaction is the encapsulation of the parasite egg; parasitism is successful only if the parasitoid is homozygous for virulence alleles (ISy+/ISy+, ISy line) (Dupas et al., 1998) and if the host is homozygous for susceptible alleles (RlbIsy−/RlbIsy−, R1 line). This genetic pattern of interactions, different from the one described above, resembles the ‘compatibility model’ of Briggs and Johal (1994). In this model, parasitoid success would require a specific match between host and parasite ‘compatible factors’ and each modification of the host target would prevent development of the parasite.

Genetic variations in resistance and virulence have rarely been described in pairwise host–parasitoid interactions and had never been analysed before in multi-species interactions. Our data show that there are different resistance patterns to the parasitoid species L. boulardi in D. melanogaster and D. yakuba, suggesting complex ecological interactions in the field.

ISm eggs are rarely encapsulated in D. melanogaster and always encapsulated in D. yakuba lines. Despite intense investigations, we have not been able to recover D. melanogaster flies that are resistant or D. yakuba flies that are susceptible to this parasitoid line. By contrast, encapsulation of ISy eggs can take place in both host species depending on the genotype of the fly (ie the resistance status of the host strain). Hence, the success of the ISm line is ‘species-dependent’, whereas the success of the ISy line is ‘host-genotype-dependent’. We thus report the occurrence of two different levels of host specificity in a single parasitoid species. As suggested in other host–parasite systems, the success or failure of a parasitoid seems to be due to neither parasitoid virulence alone nor host resistance ability alone, but rather determined by complex interactions between host and parasitoid species (Kraaijeveld and Godfray, 2001; Dupas et al., 2002; Little et al., 2005).

Geographic patterns of resistance of D. yakuba to L. boulardi

L. boulardi eggs of the ISm line were encapsulated in all D. yakuba strains tested, whatever their geographic origin, thus confirming the absence of variation in resistance to this parasitoid line. Occurrence of variability in resistance to ISy females, discovered in Tanzanian populations (origin of the R1 and R2 lines), was further confirmed using other natural populations. The question remains whether this variability in resistance can also be attributed to the RlbISy locus.

A high resistance level was found in Kenya, Gabon, Ivory Coast and in the Principe Island, whereas populations from São Tome were susceptible. Interestingly, resistance to ISy parasitoids is also high in D. melanogaster populations, even in areas where this type of parasitoid is absent (Dupas et al., 2002). It is generally postulated that resistance to a pathogen or a parasite is a life-history trait whose polymorphism is maintained by a balance between positive selection forces (the presence of the pathogen) and counter-selecting forces (the cost of maintaining resistance in the absence of the pathogen) (Kurtz et al., 2002; Rolff and Siva-Jothy, 2003). In this model, positive selective forces should have acted in every locality where resistance to ISy parasitoids is observed at high frequencies. So far, L. boulardi populations corresponding to the ‘ISy type’ (able to develop on D. yakuba) have only been found in Congo and, to a lesser extent, in Ivory Coast (Dupas and Boscaro, 1999). If the ‘ISy type’ distribution area was the same in the past, other biotic or abiotic pressures might have selected resistance in the majority of D. yakuba populations. Whether individuals resistant to ‘ISy parasites’ can also resist other parasitoids/pathogens remains to be determined, but preliminary results suggest that they are at least able to resist another parasitoid species found in these localities, Leptopilina freyae (Allemand et al., 2002; Y Carton, unpublished data). Resistance in D. yakuba populations could be considered in this way to be a ‘generalist resistance’ rather than a ‘specific resistance’.

According to the same hypothesis, the absence of resistance in São Tome island could result from the absence of agents selecting resistance in mainland populations, combined with counter-selection of resistance alleles, leading to the loss of resistance. Unfortunately, the questions whether L. boulardi parasitoids can be found in São Tome and whether D. yakuba resistance is costly remain to be answered. A second hypothesis could be that under a high diversity of attack or a high prevalence of parasites, tolerance or non-investment in immune defence mechanisms is an optimal strategy (Sasaki and Godfray, 1999; Jokela et al., 2000). We lack information on the abundance of D. yakuba pathogens or parasitoids on this island but considering classical theories in biogeography, this hypothesis seems rather unlikely (MacArthur and Wilson, 1967). Finally, the susceptibility of São Tome populations to L. boulardi might be explained by a founder effect having occurred during colonization of the island by a small number of flies (Cariou et al., 2001).

Observation of variability in resistance and virulence in host–parasitoid communities and determination of their genetic bases are important in understanding diffuse coevolutionary processes. This can be achieved by analysing and comparing pairwise interactions involving the same parasitoid and different host species or the same host and different parasitoid species. We draw here the first complete picture of resistance–virulence genetic interactions between a parasitoid and its two host species, and suggest that multi-species interactions may indeed greatly influence coevolutionary processes. Deciphering the physiological and molecular bases of these traits will finally help to address the question of specificity in host–parasitoid interactions and will provide precious information on the ongoing coevolutionary arms races.