Interspecific competition between entomopathogenic nematodes (Steinernema) is modified by their bacterial symbionts (Xenorhabdus)

Background Symbioses between invertebrates and prokaryotes are biological systems of particular interest in order to study the evolution of mutualism. The symbioses between the entomopathogenic nematodes Steinernema and their bacterial symbiont Xenorhabdus are very tractable model systems. Previous studies demonstrated (i) a highly specialized relationship between each strain of nematodes and its naturally associated bacterial strain and (ii) that mutualism plays a role in several important life history traits of each partner such as access to insect host resources, dispersal and protection against various biotic and abiotic factors. The goal of the present study was to address the question of the impact of Xenorhabdus symbionts on the progression and outcome of interspecific competition between individuals belonging to different Steinernema species. For this, we monitored experimental interspecific competition between (i) two nematode species: S. carpocapsae and S. scapterisci and (ii) their respective symbionts: X. nematophila and X. innexi within an experimental insect-host (Galleria mellonella). Three conditions of competition between nematodes were tested: (i) infection of insects with aposymbiotic IJs (i.e. without symbiont) of both species (ii) infection of insects with aposymbiotic IJs of both species in presence of variable proportion of their two Xenorhabdus symbionts and (iii) infection of insects with symbiotic IJs (i.e. naturally associated with their symbionts) of both species. Results We found that both the progression and the outcome of interspecific competition between entomopathogenic nematodes were influenced by their bacterial symbionts. Thus, the results obtained with aposymbiotic nematodes were totally opposite to those obtained with symbiotic nematodes. Moreover, the experimental introduction of different ratios of Xenorhabdus symbionts in the insect-host during competition between Steinernema modified the proportion of each species in the adults and in the global offspring. Conclusion We showed that Xenorhabdus symbionts modified the competition between their Steinernema associates. This suggests that Xenorhabdus not only provides Steinernema with access to food sources but also furnishes new abilities to deal with biotic parameters such as competitors.


Background
Symbioses between the entomopathogenic nematodes Steinernema spp. and the enterobacteriacae Xenorhabdus spp. are associations in which both partners receive benefits from each other [1][2][3]. In the soil, the infective juveniles (IJs) of the nematodes act as vectors dispersing the bacteria from insect host to insect-host and in turn, the bacteria increase the nematode's fitness within the insects hosts [3,4]. Previous studies showed that these symbioses were highly specific and that no Steinernema spp. was able to associate with a Xenorhabdus spp. genetically distant from its natural one [2,5,6]. As the bacterial dispersion is totally dependent upon the fitness of the nematode within the insect-host, it is possible that Xenorhabdus spp. might select special traits in order to enhance their vector's fitness. It is known that Xenorhabdus spp. are beneficial to their nematodes in providing the latter with a better ability to kill the insect and feed on it [1,7,8]. Previous studies that focused on two different Steinernema species (S. carpocapsae and S. scapterisci) have provided us with insights into the association characteristics [2,5,6,[9][10][11]. Although the two nematode species demonstrated increased fitness when they parasitized insect-hosts with their own native symbiont, S. scapterisci appeared less dependent upon its native symbiont (X. innexi [12]) than S. carpocapsae (associated with X. nematophila). Thus, S. scapterisci's symbiont increased the reproductive rate of its naturally associated nematode by a factor of 1.3, whereas X. nematophila increased the reproductive rate of its naturally associated nematode by sevenfold (i.e. S. carpocapsae) [3]. Moreover, S. scapterisci, transported 700-fold fewer cells of its Xenorhabdus than S. carpocapsae (i.e. ~50 bacteria per nematode for S. carpocapsae and ~0.07 bacteria per nematode for S. scapterisci) [3]. These two Steinernema species also differed in their ability to deal with non-native Xenorhabdus strains in case of co-infection in an insect [5,6]. While S. scapterisci reproduced in co-infection situations with all the tested Xenorhabdus strains (even if its reproduction was better with its native one than with others), S. carpocapsae could not reproduce at all with most of them in the same situation [5,6]. Despite these specific differences, the global trend emerging from these previous experiments was that non-naturally associated Xenorhabdus strains tend to be antagonist against nematodes species which cannot disperse them. One can easily think that this antagonistic effect of Xenorhabdus strains on the fitness of nematodes naturally associated with others Xenorhabdus strains could be selected in case of frequent interspecific competition. In such an evolutionary context, we can postulate that each Xenorhabdus strain should try to provide its own nematode-vector with competitive advantages by producing antagonistic molecules against foreign nematodes. Indeed, a previous study has shown with the association S. carpocapsae-X. nematophila as a model-system that, in insects co-infected by antagonistic Xenorhabdus, X. nemat-ophila partly counteracted their antagonistic effect on the nematodes fitness most probably by the mean of bacteriocins [13,14].
The goal of the present study was to address the question of the impact of Xenorhabdus symbionts on the progression and outcome of interspecific competition between individuals belonging to different Steinernema species. For this, we monitored experimental interspecific competition between (i) two nematode species: S. carpocapsae and S. scapterisci and (ii) their respective symbionts: X. nematophila and X. innexi within an insect-host (Galleria mellonella). In this study, three conditions of competition between nematodes were tested: (i) infection of insects with aposymbiotic IJs (i.e. without symbiont) of both species (ii) infection of insects with aposymbiotic IJs of both species in presence of variable proportion of their two Xenorhabdus symbionts and (iii) infection of insects with symbiotic IJs (i.e. naturally associated with their symbionts) of both species.

Proportion of each Xenorhabdus in insect's hemolymph 72 h post-infection with IJs
We observed that both bacterial strains were able to multiply and co-exist within the hemolymph of the insect. Nevertheless, X. nematophila was clearly less represented within the hemolymph 72 h post-infection with nematodes when an initial injection of a suspension containing 50% of each bacterium in the insect was performed (Fig  1). When 70% and 90% of X. nematophila were injected, the two bacteria were found in a very variable proportion and co-existed (Fig 1). In the competition resulting from infection of insects with symbiotic IJs of both nematode species, as well as when 100% of X. nematophila were injected into insects infected with aposymbiotic IJs of both nematode species, no X. innexi were detected within the hemolymph (Fig 1).

Assessment of nematodes maturation in competition
To know if GFP labelling of X. nematophila, employed to discriminate the two nematode species within the global offspring, triggered differences on the progress of competition between nematodes, the data obtained with or without GFP labelling of X. nematophila in each competition situation were compared with a Mann Withney test. We showed that GFP labelling had no statistically significant effect on both (i) the ratio of S. carpocapsae among all females found in the insect 150 h after infection (N = 4; 4 <U < 14; 0.072 <P < 0.449) and (ii) the ratio of S. carpocapsae among all males found in the insect 150 h after infection (21 <N < 33; 214,000 <U < 398,00; 0.172 <P < 0.780). Because of the non-significant differences observed in these comparisons, the results obtained in each situation with and without GFP labelled X. nemat-ophila were pooled in further analyses. The ratios of S. carpocapsae among all males found in the insects were highly heterogeneous for the different competition conditions (KW: H = 164.052; P < 0.0001). The Noether test showed that no significant difference in the ratio of S. carpocapsae males occurred under four competition conditions: (i) competition resulting from infection of insects with aposymbiotic IJs of both nematode species without injection of bacterium, (ii) competition resulting from infection with aposymbiotic IJs of both nematode species in insects injected with 100% of X. innexi, (iii) competition resulting from infection with aposymbiotic IJs of both nematode species with 50% of each symbiont injected in the insects and (iiii) competition resulting from infection with aposymbiotic nematodes in insects injected with 70% of X. nematophila and 30% of X. innexi (24 < N < 63, 0.464 <z < 2.568, 0.441 <P < 1). In these four conditions, the ratios of S. carpocapsae among males were significantly lower than in the three other conditions of competition: (i) competition resulting from infection with aposymbiotic IJs in insects injected with 90% of X. nematophila and 10% of X. innexi, (ii) competition resulting from infection between aposymbiotic IJs in insects injected with 100% of X. nematophila and (iii) competition resulting from infection with symbiotic IJs (28 < N < 63; 3.418 <z < 9.143; 0.0001 <P < 0.0264) (see Fig. 2). These results showed that an increase in the proportion of X. nematophila within bacteria injected into the insect triggered an increase of S. carpocapsae within adult worms found in the insects. The ratios of S. carpocapsae among females found in the insects were also highly heterogeneous for the different competition conditions (KW: H = 24.946; P < 0.001). The Noether test showed that the ratios of S. carpocapsae within females were lower in competition resulting from infection with aposymbiotic IJs in insects injected with 100% of X. innexi than (i) in competition resulting from infection with aposymbiotic IJs in insects injected with 100% of X. nematophila and (ii) in competition resulting from infection with symbiotic IJs (5 < N < 8; 3.260 <z < 4.028; 0.002 <P < 0.046). In competition resulting from infection with aposymbiotic IJs in insects injected with 100% of X. innexi, no S. carpocapsae females were observed.
Percentile distribution of the ratio of S. carpocapsae within the total number of (A) nematode males and (B) nematode females found in the insect 120 h post-infection Figure 2 Percentile distribution of the ratio of S. carpocapsae within the total number of (A) nematode males and (B) nematode females found in the insect 120 h post-infection. Results are given as box plots, where the horizontal line indicate the median (50 th of the data), the bottom and the top of the box indicate the first quartiles (25 th of the data) and the third quartiles (75 th of the data). The whiskers the range of the data (10 th of the data and 90 th of the data). Others dots are outliers which are under the 10 th percentile of the data or over 90 th percentile of the data. The abscissa shows the condition of the initial infection.

Assessment of the nematode offspring Total number of infective juveniles (total nematode offspring)
The total number of IJs emerging from insects in the different competition situations for S. carpocapsae and S. scapterisci was heterogeneous under the different competition conditions (K.W. H = 71.041, P < 0.0001). The Noether test showed that competition resulting from infection with aposymbiotic IJs without injection of bacteria into the insect resulted in the production of significantly fewer IJs than (i) competition resulting from infection with symbiotic IJs (z = 7.296, P < 0.001), (ii) competition resulting from infection with aposymbiotic IJs in insects injected with 50% of each symbiont (z = 6.108, P < 0.0001), (iii) competition resulting from infection with aposymbiotic IJs in insects injected with 70% of X. nematophila and 30% of X. innexi (z = 5.174, P < 0.0001) and (iiii) competition resulting from infection between aposymbiotic IJs in insects injected with 90% of X. nematophila and 10% of X. innexi (z = 4.304, P < 0.001).

Proportion of each nematode species in the total nematode offspring
In the competition situation resulting from infection with aposymbiotic IJs of both nematode species without injection of bacteria into the insect, the proportion of each nematode species in the total offspring emerging from four insects was assessed with PCR-RFLP made after global extraction of DNA from pools of 500 IJs. In control samples, the presence of 25% and 10% of S. carpocapsae IJs within pools of 500 IJs containing respectively 75% and 90% of S. scapterisci were easily detected (i.e. three bands were observed after incubation of the PCR product with Hind III) (Fig. 3). In all the four pools of 500 IJs tested here, only one band was observed after incubation of the PCR product with Hind III (Fig. 3). This result suggested that after competition between aposymbiotic nematodes without injection of bacteria, more than 90% of all IJs emerging from the insect belonged to S. scapterisci. In other competition situations, the proportion of each nematode species was evaluated by counting the proportion of nematodes harbouring GFP labelled X. nematophila within the global offspring. GFP-fluorescence of X. nematophila is a good indicator of S. carpocapsae's IJs because pilots experiments showed that GFP labelled bacteria were still observable in IJs that were stored during 6 months at 8°C and that after several generations within different insects, at each generation, 95% of the IJs emerging from the insects habored GFP labelled bacteria. Nevertheless, in order to check that the GFP expression was not lost all along our experiments, we checked GFP expression for all the X. nematophila isolated from pools of 500 IJs and confirmed that none of them did not express GFP. This result confirmed the perfect stability of GFP labelled X. nematophila all along our experiments. Competition resulting from infection with aposymbiotic IJs in insects injected with (i) 50% of each symbiont, (ii) 70% of X. nematophila and 30% of X. innexi and (iii) 90% of X. nematophila and 10% of X. innexi gave offspring in which almost no IJs harboured GFP labelled X. nematophila (Fig. 4). These observations showed that S. scapterisci was highly dominant in Percentile distribution of the percentages of IJs harbouring GFP labelled X. nematophila  10% S.C. Apo Apo Apo Apo the total offspring emerging from insects after such competitions. For competition resulting from infection with aposymbiotic IJs in insects injected with 100% of X. nematophila, the proportion of IJs harbouring GFP within the total offspring was very variable from 0 to 100% with almost all the combinations (for the distribution, see Fig.  4). For competition resulting from infection with symbiotic IJs, most of the IJs emerging from the insects (90 ± 22%) harboured GFP labelled X. nematophila (Fig. 4). The latter result suggested that in competition resulting from infection with symbiotic IJs, most of the offspring emerging from the insects belonged to the species S. carpocapsae.

Discussion
Many invertebrates are associated with prokaryotes that provide them with the ability to live in extreme environments [15][16][17]. The most documented symbioses are those where symbionts provide their hosts with new abilities to take advantage of poorly nutritive resources [15,[18][19][20]. Nevertheless, symbionts can play other important roles on diverse life history traits of their hosts, for example, the bacterium Vibrio which provides its sepiolids host with a better discretion towards its predators [21,22]. We already knew that Xenorhabdus symbionts provided Steinernema nematodes with diverse adaptations to their environment: (i) they produce virulence factors in order to kill the insect [23][24][25], (ii) they lead to a better access to the insect biomass and are themselves food for the nematodes [3,4], (iii) they produce antibiotics which counteract the multiplication of both distant and closely related bacteria [26][27][28][29]. The goal of this study was to infer the abilities of Xenorhabdus to modulate the progression and outcome of interspecific competition between Steinernema species. For this, we made experimental competitions between S. carpocapsae and S. scapterisci and their bacterial symbionts. We assessed several parameters of these competitions: (i) the proportion of each symbiont (i.e. X. nematophila and X. innexi) within the hemolymph of the insect after delay of multiplication, (ii) the ratio of S. carpocapsae among the adults found in the insects and (iii) the offspring of each nematode species after competition. The assessment of bacterial multiplication within insect hemolymph showed that X. innexi appeared more competitive than X. nematophila. Moreover, in insects infected with aposymbiotic IJs and in which 50% of each Xenorhabdus strains were initially injected, we found X. innexi highly dominant within the hemolymph after 48 h of multiplication (Fig. 1). Nevertheless, the ability of X. innexi to outcompete X. nematophila was not sufficient enough in the symbiotic competition where S. scapterisci IJs brought 700-fold less of their symbionts than S. carpocapsae in the insects [3]. In competition resulting from infection of insects with symbiotic IJs, our results showed that such a competition took place within the insects where only X. nematophila proliferated.
In this study, we studied both (i) the development of nematodes within the insects and (ii) their ability to reproduce in them. Regarding the first aspect, we showed that the nematode's development into adults differed significantly for (i) competition resulting from infection with aposymbiotic IJs of both nematode species without injection of bacteria and (ii) competition resulting from infection between symbiotic IJs of both nematode species. The competition between aposymbiotic nematodes (without injection of bacteria) allowed evaluation of the intrinsic ability of each nematode species to outcompete the other one without the help of its symbiotic Xenorhabdus. In such a situation, we showed that (i) significantly fewer nematodes developed into adults and that (ii) S. scapterisci was highly dominant in both males and females found in the insects. These results showed that S. scapterisci had a better competitive ability without its Xenorhabdus symbiont than did S. carpocapsae (Fig. 2). On the contrary, in the competition between symbiotic nematodes, S. carpocapsae was highly dominant in adults (Fig.  2). Experiments in which the amount of each Xenorhabdus sp. initially introduced in the insect was experimentally modified showed that the proportion of adults from each nematode species in the insects was directly linked to the proportion of their bacterial symbionts within the insect hemolymph. (Fig. 2). The more X. nematophila was dominant within the insect hemolymph, the more the proportion of S. carpocapsae in adult worms found in the insects increased. The analyses of the offspring emerging from the insects showed that competition resulting from infection with aposymbiotic IJs without injection of bacteria almost only produced IJs belonging to the species S. scapterisci. On the contrary, and in accordance with the results obtained on adults development, almost all the IJs emerging from the competition resulting from infection with symbiotic IJs belonged to the species S. carpocapsae.
Regarding the offspring production of each nematode species, the experimental modification of the bacterial symbiotic environment within the insects led to surprising results. In all insects where X. innexi was injected [even at the lowest dose (10% of the bacteria injected], almost no S. carpocapsae were found in the offspring. The inability of S. carpocapsae to reproduce in such situations was certainly due to the antagonistic effect of X. innexi toward this nematode which has already been demonstrated. Indeed, previous studies had shown that aposymbiotic S. carpocapsae were unable to reproduce in insects where only X. innexi was injected [5] but that the co-injection of X. nematophila with X. innexi led to a partially re-established reproduction of S. carpocapsae [13]. In the present study, even competition situations where X. nematophila co-infected the insects with X. innexi gave almost no offspring for S. carpocapsae. In such competition situations, it seems that X. nematophila was not able to counteract the effect. We can thus postulate that the intrinsic ability of S. scapterisci to outcompete by itself S. carpocapsae reinforces the antagonistic effect of X. innexi on S. carpocapsae's reproduction. In such conditions, S. scapterisci and X. innexi acted in synergy to outcompete S. carpocapsae and X. nematophila leading to the absence of reproduction for the latter nematode. On the contrary, in the competition situation which was closest to the natural situation (i.e. symbiotic nematodes), the association S. carpocapse-X. nematophila totally outcompeted the association S. scapterisci-X. innexi. The outcome of the competition between symbiotic nematodes in favour of S. carpocapsae is quite understandable since this competition occurred in a bacterial environment constituted exclusively of X. nematophila. This dominance of X. nematophila within the insect hemolymph is due to the fact that S. scapterisci is poorly associated with its symbiont compared to S. carpocapsae (700-fold less cell transported) [3]. Thus, dramatically less cells of X. innexi than X. nematophila were released by the worms within the insect hemolymph.

Conclusion
This study shows that the bacterial symbionts Xenorhabdus modify the progress and outcome of interspecific competitions between their Steinernema associated nematodes. This suggests that Xenorhabdus not only provides Steinernema with access to food sources but also furnishes new abilities to deal with biotic parameters such as competitors. Thus, the evolution of the specific association between Steinernema and Xenorhabdus in environments where interspecific competitions occurred frequently should lead to the selection for an increased number of bacterial cells retained by each nematode in order to outcompete other specific nematodes-bacteria complexes. A previous study has reported that there is dramatically different bacterial retention between some species of Steinernema [3]. This bacterial retention variability could be linked to differences in the biotic environments where each symbiotic association evolves. Indeed, S. carpocapsae is a worldwide distributed nematode contrary to S. scapterisci which is only found in South-America [30][31][32]. Moreover, S. carpocapsae is supposed to have a wider insect-host range than S. scapterisci [33]. These specific ecological traits of each Steinernema species suggest that S. carpocapsae could have evolved under a higher competition pressure than S. scapterisci. This evolutionary constraint could have induced a selection toward a tighter relationship (i.e. more bacterial cells by IJ) in the S. carpocapsae-X. nematophila association than in the S. scapterisci-X. innexi association.

Insects, nematodes and bacteria
The two Steinernema species naturally associated with their Xenorhabdus (naturally symbiotic nematodes) were established in the laboratory as soon as they were sampled by successive experimental infections of the last instar of the wax moth Galleria mellonella. The aposymbiotic and symbiotic IJs used for experiments were always freshly emerged from insects. Insect hosts were reared in the dark in aired plastic boxes at 28°C, 65% RH, on a diet of pollen and wax. Aposymbiotic IJs (i.e. without symbiont) of the species S. carpocapsae and S. scapterisci were obtained by disinfecting nematode eggs with a bleach solution as described previously [34]. The symbiotic IJs of S. carpocapsae associated with GFP (Green Fluorescent Protein) labelled X. nematophila (strain F1D3) were produced as described previously [34]. The GFP labeling is based on a plasmid expression which is very stable even if it is not recombined in the bacterial chromosome.
Bacterial suspensions were prepared by transferring a single colony of one bacterial strain to 5 ml of Luria-Bertani broth for liquid culture incubated at 28°C for 15 h. 100 μl of this liquid subculture was used to perform a culture to reach an optical density of 0.7 (600 nm wavelength). Before inoculation into the insect, the number of bacterial cells in each culture was counted with a Thoma cell and diluted in order to obtain a suspension of 100 cells/μl. A control of the actual number of bacteria in the injected suspension was measured by plating it onto three NBTA plates [35]. These plates were incubated at 28°C for 48 h. Then, the colonies that grew on these plates were counted. An experiment was kept only if the number of colonies ranged from 1500 to 2500.

General settings
We performed three different competition experiments: (i) competition resulting from infection with aposymbiotic IJs without injection of bacteria into the insects, (ii) competition resulting from infection with aposymbiotic nematodes in insects followed by injection of variable proportions of their respective bacterial symbiont (X. nematophila and X. innexi) and (iii) competition resulting from infection with symbiotic IJs (i.e. IJs of nematodes naturally containing Xenorhabdus in their guts). In each of these competition experiments, we monitored (i) the proportion of each bacterial symbiont within the hemolymph 72 h post-infection with IJs, (ii) the proportion of males and females from each species within adult nematodes found in the insect and (iii) the proportion of each nematode species within the offspring (IJs). In order to be able to specifically distinguish IJs emerging from the cadaver, we labelled S. carpocapsae with its symbiotic bacteria, X. nematophila, hosting a plasmid expressing GFP. Previous studies showed that 96% of S. carpocapsae IJs were associated with X. nematophila and that S. scapterisci was unable to associate with this bacterial strain [34,36,37]. To be able to evaluate S. carpocapsae's offspring within the total number of IJs emerging from an insect, we made two different sets of insects for each type of competition (except the competition between aposymbiotic nematodes without injection of bacteria): (i) one set of insects was infected with GFP labelled X. nematophila and (ii) one set with wild type of X. nematophila. (for sample sizes, see Table 1).

Experimental infection
Forty IJs of each species (aposymbiotic or symbiotic depending on the type of experiment) were counted and deposited into 1.5-ml Eppendorf tubes containing a filter paper. A last instar of G. mellonella was then introduced into each Eppendorf, and the nematodes and the insecthost were incubated together at 24°C during 24 h. For competition resulting from infection with aposymbiotic and symbiotic IJs, no bacteria were experimentally introduced into the insect. For competition resulting from infection with aposymbiotic IJs in insects with a controlled and variable bacterial symbiont environment, 24 h after infection with aposymbiotic IJs, insects were injected with different proportions of the two bacterial symbionts (X. nematophila and X. innexi) [13]. The different proportions tested were (i) 100% of X. innexi, (ii) 50% of each symbiont, (iii) 70% of X. nematophila and 30% of X. innexi, (iiii) 90% of X. nematophila and 10% of X. innexi and (iiiii) 100% of X. nematophila (for sample sizes, see table 1).

Proportion of each Xenorhabdus in insect's hemolymph 72 h post-infection with IJs
For infections made with aposymbiotic IJs followed by the injection of variable proportions of each symbiont, the proportion of each symbiont was assessed 72 h postinfection with aposymbiotic IJs (i.e. 48 h after the direct injection of Xenorhabdus cells into the insects). For infections made with symbiotic IJs, the same assessment was made 72 h post-infection with symbiotic IJs. To evaluate the proportion of each symbiotic bacterium within the insect, the hemolymph of each Galleria mellonella was collected using a syringe. The sampled hemolymph was diluted by 10 5 and 100 μl of the obtained suspension was plated out onto three NBTA plates. The plates were then incubated at 28°C during 48 h. After incubation, few cells from each colony were sampled with the help of sterile tooth picks and transferred in lines onto two NBTA plates containing control samples of X. nematophila and X. innexi. One of these plates was incubated at 28°C during 48 h and the other one at 37°C during 48 h. These two different incubation temperatures helped us to discriminate between the two Xenorhabdus strains. Indeed, both of these bacteria grew at 28°C but only X. innexi was also able to grow at 37°C [38]. The proportion of X. nematophila in the insect was then obtained by subtracting the number of colonies grown at 28°C from the number of colonies grown at 37°C and then dividing the result by the number of colonies grown at 28°C.

Assessment of nematode maturation in competition
120 h post-infection by IJs, the insects were dissected and adult nematodes were transferred to separate Eppendorfs containing sterile Ringer. Males were then mounted with a drop of water between slide and coverslip and the species was determined after spicule observation [39]. The proportion of S. carpocapsae within all males found in an insect was calculated (for the number of insect analysed, see table 1). For each tested condition, all nematode females (i.e. from 2 to 37 per insect) found in four insects were separately analysed by PCR-RFLP and assigned to one nematode species. In order to do that, DNA from each female was separately extracted as described previously [40]. Then, the ITS region was amplified as described previously [39]. 10 μl of PCR products were then incubated at 37°C with 0.5 μl of the restriction enzyme Hind III, 2 μl of enzyme buffer and 7.5 μl of sterile water. Previous studies showed that Hind III cuts the amplified ITS fragment for S. carpocapsae and not for S. scapterisci [39]. After such analyses, the ratios of S. carpocapsae within nematode females found in each insect were calculated.

Assessment of the nematode offspring
The total number of offspring produced by each infection was separately harvested in 50-ml Falcon flasks two months after infection and stored at 8°C. The total number of IJs produced was then evaluated under binocular microscope, using 1 ml of the suspension taken from the Falcon flask on a grid drawn on a 6-cm Petri dish. Then, for each IJ emergence coming from the competitions performed with GFP labelled X. nematophila, we (i) examined the vesicle of 100 IJs by epifluorescence microscopy and (ii) made isolation of bacteria contained in 500 IJs as previously described by Sicard et al., 2003. The first experiment led to assign each IJ to one nematode species. Indeed, if GFP labelled X. nematophila were observed within their vesicles, the IJs were assigned to S. carpocapsae, if not to S. scapterisci. The second experiment led to check if all the clones of X. nematophila isolated from IJs expressed GFP. In case of competition resulting from infection with aposymbiotic IJs without injection of bacteria into the insects, the presence/absence of each nematode species in the offspring was evaluated by performing PCR-RFLP (as described above) with DNA extraction on four pools of 500 IJs coming from four different insects. Preliminary experiments showed that when IJs of S. carpocapsae were mixed with IJs of S. scapterisci three bands were observed after PCR-RFLP. Such an approach led to detect 10% of IJs of S. carpocapsae within pools containing 90% of S. scapterisci. We used this latter experiment as a standard in our experiment (see Fig 3).

Statistical analyses
In order to compare two non-parametric distributions, we used the Mann Withney test. To compare more than two non-parametric distributions, we performed a Kruskal-Wallis (KW) test followed by the pair wise comparison test of Noether [41].