Friend or Foe: Symbiotic Bacteria in Bactrocera dorsalis–Parasitoid Associations

Simple Summary Host-associated gut bacteria influence the eco-physiological functions of their hosts, for example, how they interact with natural enemies in their ecosystems, and hence the outcome of biological pest management regimes. Here, we assessed the effect of three common gut bacteria of the oriental fruit fly, Bactrocera dorsalis, on the biological control of this pest using parasitoid wasps. We found that some gut bacteria increased parasitoid emergence and positively impacted the size and fecundity of the emerging parasitoid offspring. We, therefore, conclude that some bacteria can be used as probiotics in the mass rearing of parasitoids to boost the biological control of the oriental fruit fly. Abstract Parasitoids are promising biocontrol agents of the devastating fruit fly, Bactrocera dorsalis. However, parasitoid performance is a function of several factors, including host-associated symbiotic bacteria. Providencia alcalifaciens, Citrobacter freundii, and Lactococcus lactis are among the symbiotic bacteria commonly associated with B. dorsalis, and they influence the eco-physiological functioning of this pest. However, whether these bacteria influence the interaction between this pest and its parasitoids is unknown. This study sought to elucidate the nature of the interaction of the parasitoids, Fopius arisanus, Diachasmimorpha longicaudata, and Psyttlia cosyrae with B. dorsalis as mediated by symbiotic bacteria. Three types of fly lines were used: axenic, symbiotic, and bacteria-mono-associated (Lactococcus lactis, Providencia alcalifaciens, and Citrobacter freundii). The suitable stages of each fly line were exposed to the respective parasitoid species and reared until the emergence of adult flies/parasitoids. Thereafter, data on the emergence and parasitoid fitness traits were recorded. No wasps emerged from the fly lines exposed to P. cosyrae. The highest emergence of F. arisanus and D. longicaudata was recorded in the L. lactis fly lines. The parasitoid progeny from the L. lactis and P. alcalifaciens fly lines had the longest developmental time and the largest body size. Conversely, parasitoid fecundity was significantly lower in the L. lactis lines, whereas the P. alcalifaciens lines significantly improved fecundity. These results elucidate some effects of bacterial symbionts on host–parasitoid interactions and their potential in enhancing parasitoid-oriented management strategies against B. dorsalis.


Introduction
The oriental fruit fly, Bactrocera dorsalis (Hendel) (Diptera: Tephritidae), is a major horticultural pest that infests numerous fruit and vegetable crops across Asia, Africa, Oceania, and the Americas, resulting in hefty losses to the horticultural sectors of affected countries [1][2][3][4]. The management of this invasive pest is largely dependent on chemical insecticides, with far-reaching consequences on human and environmental health [5]. Furthermore, the overreliance on these chemicals very often results in the development of The Bactrocera dorsalis flies used in this study were reared on ripe, insecticide-free mangoes, as previously described [22], in the insectary of the International Centre of Insect Physiology and Ecology (icipe). The flies were derived from the eggs collected from a 15thgeneration, laboratory-reared colony, which was initiated with field collections and was regularly infused with field collections and maintained, as previously described [53]. The eclosed adult flies were fed on yeast and water ad libitum [15]. The parasitoids used in this study, F. arisanus, D. longicaudata, and P. cosyrae, were also reared in the insectary at icipe at a temperature range of 25-27 • C, 60-70% relative humidity, and a 12:12 day:light photoperiod. Psyttalia cosyrae was reared on a laboratory colony of Ceratitis cosyra Walker (Diptera: Tephritidae) [54], while D. longicaudata and F. arisanus were reared on B. dorsalis as described by Mohamed et al. [9] and Mohamed et al. [18], respectively. The parasitoids D. longicaudata, F. arisanus, and P. cosyrae used in this study were at the 183rd, 164th, and 177th generations in the laboratory, respectively.

Bacteria Culturing
The bacteria P. alcalifaciens, L. lactis, and C. freundii used in this study were retrieved from glycerol stocks (stored at −80 • C) previously isolated from B. dorsalis [46]. Using a sterile tip, individual species, each 5 µL, were retrieved from the glycerol stocks, spread on a nutrient-rich medium, i.e., brain heart infusion (BHI) agar (Thermo Scientific™ Oxoid, Hampshire, UK), and incubated at 37 • C for 14 h. They were then re-cultured in BHI broth (Thermo Scientific™ Oxoid, Hampshire, UK) at 37 • C for 14 h in a shaking incubator.
The bacteria were harvested by centrifuging at 12879 RCF for five minutes, and the precipitate containing the bacteria was diluted in distilled water to 1 × 10 8 concentration.

Generation of Axenic, Symbiotic, and Mono-Associated Bacterial Lines
The generation of bacteria-free fly lines was performed as described by Gichuhi et al. [46]. Briefly, freshly laid B. dorsalis eggs were collected from perforated mango domes (used as oviposition sites), surface-sterilized in 70% ethanol, dechorionated in a 7% v/v sodium hypochlorite solution for three minutes and rinsed in distilled water thrice. To confirm the elimination of bacteria from the flies, the dechorionated embryos were inoculated in autoclaved liquid larval diets [55] (modified by excluding streptomycin and nipagen) and left to develop until the 2nd instar. Consequently, 10 2nd instar larvae were randomly selected and individually homogenized in 100 µL of phosphate-buffered saline (PBS), and the homogenate was plated on BHI agar plates. The plates were incubated at 37 • C for 14 h and observed for bacterial growth. There was no bacterial growth observed on the plates, and this confirmed the axenic state of the larvae.
From these dechorionated embryos, negative (axenic) and positive (symbiotic) controls, as well as the bacterial mono-associated lines (BMALs) (L. lactis or P. alcalifaciens or C. freundii), were generated. Axenic lines were generated by inoculating the dechorionated embryos on autoclaved artificial liquid larval diets [55] (modified by excluding streptomycin and nipagen) supplemented with distilled water only. The BMALs were generated by directly inoculating the dechorionated eggs on autoclaved artificial liquid larval diets [55] (modified by excluding streptomycin and nipagen) supplemented with 1000 µL of 1 × 10 8 CFU/mL inoculum of the respective bacterial isolates, individually. Symbiotic fly lines were generated by directly inoculating the dechorionated embryos on non-autoclaved larval diets previously fed upon by larvae with an intact microbiome, i.e., from a non-dechorionated embryo fly line reared under normal laboratory conditions. This unsterilized/recycled diet was selected because it would contain microbiota from the frass of the previously reared larvae. For each type of fly line, inoculations per replicate were carried out as previously described [46], with a slight modification in the size of tubes used: flat-based glass tube cylinders (75 × 25 mm). The tubes were capped with autoclaved cotton wool and maintained in a microbiological safety cabinet) at 25-27 • C,  60-70% relative humidity, and 12:12 day:light photoperiod. To minimize confounding effects, all fly lines were generated from the same batch of collected embryos that were dechorionated in the same batch before the respective inoculations and then reared under the exact same conditions.

Parasitic Wasp Infection of BMALs and Axenic and Symbiotic Fly Lines
For D. longicaudata and P. cosyrae parasitization assays, B. dorsalis larvae were used. Briefly, B. dorsalis embryos were dechorionated and raised as described above for each experimental group until the larvae reached the 2nd instar stage. Using soft forceps, a set of a hundred 2nd instar B. dorsalis larvae were randomly selected from each fly line and individually transferred to larval oviposition units containing autoclaved carrot diets [56] (modified by excluding nipagen). The carrot diets for the BMALs and axenic and symbiotic lines were supplemented with 500 µL (1 × 10 8 ) of the respective bacterial species, distilled water, or a liquid diet previously fed on by larvae with an intact microbiome, respectively. The oviposition units were transferred to a Perspex cage (12 × 12 × 12 cm) holding 10 D. longicaudata or P. cosyrae mated female parasitoids for parasitization. In a recent study, Gwokyalya et al. [22] demonstrated that D. longicaudata achieved oviposition above 90%, whereas P. cosyrae achieved similar oviposition rates when host exposure was carried out for six hours. Consequently, in this study, exposure to D. longicaudata was implemented for two hours, while exposure to P. cosyrae was carried out for six hours. Post-parasitization, the host larvae were retrieved from the oviposition units and transferred to the autoclaved carrot diets. The carrot diets were supplemented with 500 µL of the respective bacteria inoculum for the BMALs, a regular liquid larval diet for the symbiotic lines, or distilled water for the axenic lines. The host larvae were left to feed and subsequently pupariate. The puparia were transferred to four-liter transparent lunch boxes (18 × 11 × 15 cm) covered with a cotton mesh and monitored until fly and/or parasitoid emergence.
Fopius arisanus parasitization assays were conducted using B. dorsalis eggs. Briefly, 150 dechorionated embryos were transferred to a filter paper laid on an agar plate and moistened with a 1000 µL (1 × 10 8 ) inoculum of the respective bacteria species for each BMAL. For the axenic lines, distilled water was used to moisten the filter paper, whereas, for the symbiotic lines, 1000 µL of the liquid diet used to raise normal laboratory flies was used. The agar plates were incubated for one hour (to allow for efficient bacterial infection in mono-associated and symbiotic lines), after which the oviposition units were transferred to individual Perspex cages (12 × 12 × 12 cm) containing 10 mated F. arisanus females for parasitization, for six hours. Post-parasitization, B. dorsalis eggs were retrieved from the oviposition units and transferred to autoclaved liquid larval diets [55] (excluding streptomycin and nipagen) supplemented with 500 µL of the respective bacteria inoculum for the BMALs, a normal larval liquid diet for the symbiotic lines, or distilled water for the axenic lines and left to feed and subsequently pupariate. Consequently, the puparia were transferred to transparent 4-liter lunch boxes (18 × 11 × 15 cm) covered with a cotton mesh and monitored for fly and/or parasitoid emergence.
The parasitization experiments for each fly line and parasitoid were conducted individually in sterile conditions. All the wasps used in these assays underwent parasitization by exposing them to bacteria-free B. dorsalis larvae or eggs prior to parasitization assays. The wasps that did not exhibit oviposition behavior within 15 min were removed and replaced with new ones. All fly lines and parasitoids were kept at 25-27 • C, 60-70% humidity, and 12:12 day:light photoperiod.

Effect of Bacterial Symbionts on Host Acceptability for Oviposition by Parasitoid Females
To check for possible variations in acceptability by parasitoids across the different fly lines, host larvae and/or eggs from all fly lines were exposed to parasitoids, as described in Section 2.4, in no-choice assays and dissected to check for oviposition success. Diachasmimorpha longicaudataand P. cosyrae-parasitized fly lines were dissected six hours post-parasitization, while those exposed to F. arisanus were dissected after four days. The host larvae containing parasitoid egg(s)/larva(e) were recorded and expressed as a percentage of the total number of exposed hosts to reflect host acceptability. These experiments were replicated five times.

Parasitoid Emergence and Fitness Assays
To investigate the effect of bacterial symbionts on parasitism success and fitness of the emerging parasitoids, the host fly lines were generated and parasitized, as described in Section 2.4. The parasitized fly lines were reared until pupation, and the pupae were collected, transferred to transparent four-liter lunch boxes (18 × 11 × 15 cm) covered with a cotton net cloth, and monitored for host fly and parasitoid emergence. The emerging host flies were collected, counted, and killed by freezing, while the parasitoids were transferred to separate transparent four-liter lunch boxes (18 × 11 × 15 cm) and fed on 50% honey solution ad libitum. The emerged wasps from each line for each parasitoid species were counted and sexed, and the number of each sex was recorded. On day 28 of post-exposure to parasitoid wasps, the host cadavers (unhatched host puparia) were dissected to check for the presence of dead parasitoid offspring. If found, these were recorded as parasitoid deaths.
The total number of emerging parasitoids was expressed as a percentage of the total recovered puparia to reflect parasitoid emergence. The total number of emerging host flies was expressed as a percentage of the total recovered puparia to reflect fly emergence. The total number of host puparia containing a dead parasitoid was expressed as a percentage of the total recovered puparia to reflect parasitoid death. Sex ratios were calculated as the proportion of female wasps out of all the emerged parasitoids.

Parasitoid Progeny Developmental Time
In a separate set of experiments, B. dorsalis larvae and eggs were exposed to D. longicaudata and F. arisanus for two and six hours, respectively, and treated in a similar manner as described in Section 2.4. The host puparia were transferred to four-liter lunch boxes and monitored until the onset of parasitoid emergence upon which the number of emerging male and female parasitoids was recorded daily until the last parasitoid emerged. These data were used to calculate the developmental time of the parasitoids, which was computed using the formula below [14]. These experiments were replicated 10 times.
where i signifies an individual of a total of n insects; nFi is the daily individual insect (parasitoid) emergence; and nDi is the duration in days for the development of the ith insect (parasitoid) from oviposition to adult emergence.

Parasitoid Progeny Body Size
The hind tibia of the F1 parasitoid progeny of D. longicaudata and F. arisanus were used as a proxy to measure the parasitoid progeny size. Briefly, one hind leg of each of the 10-day-old F1 parasitoids was removed, mounted on a microscope slide using entellan TM rapid mounting medium for microscopy (Merck, Darmstadt, Germany), and layered with coverslips. The hind tibiae (one tibia from each parasitoid) were measured under a stereomicroscope (Zeiss Stemi 508, Zeiss, Oberkochen, Germany). From each fly line, 10 individuals (five males and five females) of each parasitoid species were randomly selected for the assays, and this was replicated five times.

Longevity of Host-Deprived F1 Parasitoid Progeny
From the same groups of emerging parasitoids described in Section 2.7.1 above, five replicates were randomly selected for the parasitoid progeny survival analysis. From each selected replicate of each F1 parasitoid line, 10 male and 10 female parasitoids were ran-domly selected, transferred to sterile transparent four-liter lunch boxes (18 × 11 × 15 cm), and fed on 50% honey solution ad libitum, but they were not provided with hosts for oviposition. The parasitoids were monitored daily for mortality until the last parasitoid died. For each F1 progeny line within a parasitoid species, the survival data were obtained for 50 males and 50 females.

Fecundity of F1 Female Parasitoid Progeny
In a separate set of assays, 100 host larvae or 150 eggs from the axenic line, symbiotic line, and BMALs were exposed to 10 mated female D. longicaudata or F. arisanus for 2 and 6 hours, respectively, and incubated until parasitoid emergence, as described in Section 2.5. The emerging F1 parasitoids of each species from each fly line were used to investigate the symbiont-mediated effects on F1 parasitoid fecundity. Briefly, 10 randomly selected 5-day-old, mated females of D. longicaudata or F. arisanus from each fly line were transferred to four-liter lunch boxes (18 × 11 × 15 cm) and exposed to axenic larvae or eggs for two or six hours, respectively. The exposure was carried out once every day for 15 consecutive days. A 1:10 parasitoid: host ratio was maintained throughout the experiment. If a female parasitoid died, the number of host larvae/eggs offered daily was adjusted to maintain this ratio. The Diachasmimorpha longicaudata-parasitized larval groups were dissected 48 h after parasitization, while the F. arisanus-parasitized groups were dissected after four days, under a stereomicroscope (Zeiss Stemi 508, Zeiss, Oberkochen, Germany). The total number of parasitoid eggs and/or larvae in each host larva was recorded daily, and this was presented as the number of eggs laid per female parasitoid per day. These assays were performed in four independent replicates for each parasitoid species from each fly line.

Data Analysis
All analyses were performed using R software (version 4.0.1) [57]. The data obtained from host acceptability bioassays were analyzed using a generalized linear model (GLM) with quasi-Poisson distribution considering parasitoid species and host infection status as fixed factors. Data from the host suitability (parasitoid and fly emergence as well as parasitoid mortality) bioassay and the parasitoid progeny fitness trait (sex ratio, developmental time, body size, fecundity, and longevity) bioassay were analyzed separately for each parasitoid species. Since they were collected from the same batch of host fly lines, parasitoid and fly emergence, as well as parasitoid mortality data, were analyzed using multivariate analysis of variance (MANOVA), with host infection status as a fixed factor. In cases in which a significant effect of host infection status was detected, a univariate one-way analysis of variance (ANOVA) was conducted to check for significant differences within the response variables (parasitoid emergence, fly emergence, and parasitoid death), and this was followed by Student-Neuman-Keuls (SNK) post hoc multiple comparisons tests.
The sex ratios of D. longicaudata and F. arisanus F1 progeny were analyzed using a GLM with a binomial error distribution, considering host infection status as a fixed factor. Parasitoid progeny developmental time was analyzed using a GLM assuming a Gamma error distribution with host infection status as a fixed factor. Parasitoid progeny body size (hind tibia length) was analyzed using a linear mixed-effect model from the 'lme4' package (version 1.1.27.1) [58], considering host infection status and the sex of the parasitoid progeny as fixed and random factors, respectively. A generalized, linear mixedeffect model (using the 'glmer' function from the 'lme4' package [58]), with a Gamma error distribution, was fitted to analyze the fecundity of the female parasitoid progeny of F. arisanus and D. longicaudata, considering host infection status and replicate population as fixed and random factors, respectively. A Cox proportional hazards model using the "coxme" function from the "survival" package (version 3.2.13) [59] was fitted to determine the effect of bacterial symbionts on the longevity of the F1 progeny of both parasitoid species. In the model, both host infection status and replicate population were included as fixed and random factors, respectively. The significance of all these models was determined using analysis of deviance with chi-square tests, and mean separation was determined using Tukey's multiple comparison tests at α ≤ 0.05.
tribution, was fitted to analyze the fecundity of the female parasitoid progeny of F. arisanus and D. longicaudata, considering host infection status and replicate population as fixed and random factors, respectively. A Cox proportional hazards model using the "coxme" function from the "survival" package (version 3.2.13) [59] was fitted to determine the effect of bacterial symbionts on the longevity of the F1 progeny of both parasitoid species. In the model, both host infection status and replicate population were included as fixed and random factors, respectively. The significance of all these models was determined using analysis of deviance with chi-square tests, and mean separation was determined using Tukey's multiple comparison tests at α ≤ 0.05.

Effect of Bacterial Symbionts on Host Suitability for Development of the Immature Stages of the Parasitoids
We recorded a significant effect of symbiotic bacteria on parasitoid emergence, fly emergence, and parasitoid mortality in the F. arisanus-parasitized fly lines (MANOVA: F 12,135 = 5.913, p < 0.001). Further, analysis using ANOVA showed significant effects of bacterial symbionts on parasitoid emergence (F 4,45 = 24.122, p < 0.001; Figure 2a) as well as host fly emergence (F 4,45 = 33.708, p < 0.001; Figure 2b) but not parasitoid mortality (F 4,45 = 0.872, p = 0.488; Figure 2c). The flies inoculated with L. lactis yielded the highest percentage of parasitoid wasps and the lowest percentage of eclosed host flies, while the reverse was true for those inoculated with P. alcalifaciens (Figure 2a,b). parasitoid wasps and the lowest percentage of eclosed host flies, while the reverse was true for those inoculated with P. alcalifaciens (Figure 2a, b).  We did not record any parasitoid emergence from P. cosyrae-parasitized larval groups irrespective of the infection status. However, we found a significant multivariate effect of bacterial symbionts on host fly emergence and parasitoid mortality (MANOVA:  We did not record any parasitoid emergence from P. cosyrae-parasitized larval groups irrespective of the infection status. However, we found a significant multivariate effect of bacterial symbionts on host fly emergence and parasitoid mortality (MANOVA: F 8,90 = 4.667, p < 0.001). Further analysis revealed that host fly emergence significantly var-

Parasitoid Progeny Sex Ratio
The bacterial symbionts of B. dorsalis did not alter the sex ratios of the progeny of F. arisanus (

Discussion
Augmentative biological control is a widely used approach for the management of fruit flies in many countries [11][12][13]16,20]. This approach does not only require the production of parasitoids on a large scale, but the produced parasitoid wasps also need to be of high quality in terms of fitness, to ensure their optimal performance in the field. Sym-

Discussion
Augmentative biological control is a widely used approach for the management of fruit flies in many countries [11][12][13]16,20]. This approach does not only require the production of parasitoids on a large scale, but the produced parasitoid wasps also need to be of high quality in terms of fitness, to ensure their optimal performance in the field. Symbiotic bacteria could be exploited to enhance the quality of mass-reared parasitoids [41]. Indeed, symbiotic bacteria have been reported to influence the physiological quality of their hosts, inevitably influencing how they are perceived by parasitoids and their suitability for the development of parasitoid immatures and, eventually, the fitness of the emerging parasitoid progeny [39,41,42,60].
The comparable acceptability of the three parasitoid species (F. arisanus, P. cosyrae, and D. longicaudata) across the B. dorsalis fly lines reported in this study might be an indication that these symbiotic bacteria had no influence of the host cues used by the parasitoid females to locate their hosts. Indeed, Sochard et al. [27] reported that Hamiltonella defensa and Regiella insecticola did not alter the acceptability of Acyrthosiphon pisum (Harris) (Hemiptera: Aphididae) by its parasitoid, Aphidius ervi Haliday (Hymenoptera: Braconidae). Likewise, Xie et al. [32] showed that Spiroplasma sp. had no effect on the parasitism rates of Leptopilina heterotoma Förster (Hymenoptera: Figitidae) in Drosophila melanogaster Meigen (Diptera: Drosophilidae). However, the findings by Attia et al. [39] and Frago et al. [29] revealed that symbionts influence the host acceptability by parasitoid wasps, a phenotype that was attributed to the symbiont-mediated cues that influence host finding and ovipositing behavior by the parasitoids.
Nonetheless, despite the similar outcome in host acceptability, parasitoid emergence varied among the B. dorsalis fly lines, suggesting that bacterial symbionts had a profound effect on parasitoid development. Most facultative bacteria described to date possess protective roles where, in some instances, complete host protection against parasitoids is realized [30]. These protective phenotypes are largely attributed to the symbiont-encoded viruses that impair parasitoid development [35,36], the ability of the symbionts to alter the host immune defenses [37], or their ability to compete for nutrients, which are needed for the successful development of the parasitoids [38]. However, the high parasitism rates recorded in the L. lactis-inoculated fly lines in the current study suggest that L. lactis increases the susceptibility of B. dorsalis to attack by parasitoids. This susceptibility could be associated with the suppression of immune defense mechanisms of B. dorsalis and/or the L. lactis-mediated pathogenicity mechanism, previously highlighted in the B. dorsalis-L. lactis-Metarhizium anisopliae interaction [46]. However, this mechanism remains poorly understood.
Surprisingly, one evaluated symbiont, P. alcalifaciens, was found to confer protection to B. dorsalis against F. arisanus but not D. longicaudata, suggesting that bacteria-mediated protection against parasitoids is not universal across host-parasitoid systems. These two parasitoids attack different stages of their host, with F. arisanus using the eggs and D. longicaudata using the larvae as hosts, and this might have contributed to the discrepancy in P. alcalifaciens-mediated protection. Alternatively, it is possible that P. alcalifaciens produces toxins or engages pathogenic mechanisms to which only F. arisanus is susceptible. Similar symbiont-specific host-protective phenotypes have been demonstrated in other host-symbiont-parasitoid tripartite interactions [31,40,61,62].
Similar to earlier studies [40,42], we found that symbiotic bacteria did not affect the sex ratios of the progeny of F. arisanus and D. longicaudata. It has been demonstrated that parasitoids discriminate between good-and poor-quality hosts and, consequently, allocate more female offspring to good-quality hosts [63][64][65][66]. The results obtained in the current study showed that the F1 progeny of both parasitoid species from all fly lines were female-biased. The bias may be caused by the parasitoids perceiving hosts across all fly lines as good quality, thus choosing to oviposit more females in all of them. Furthermore, the female parasitoids used in this study were newly mated and in their early stages of oviposition, which may have contributed to this sex bias. Studies have shown that mated female parasitoids are likely to lay more females than males since most eggs at this stage are diploid (fertilized) eggs that develop into female progeny [67,68].
Earlier studies suggested that the parasitoid progeny's developmental time and size are a function of host size [42,69], wherein the parasitoids emerging from smaller hosts take longer to develop, especially for koinobiont parasitoid species [70,71]. Indeed, the B. dorsalis pupae from the L. lactis and P. alcalifaciens fly lines were reported to have higher weights than those from the other fly lines [46]. It was, therefore, expected that the parasitoid progeny from these lines would emerge earlier. However, in this study, these parasitoids emerged later than those reared on the, C. freundii, axenic, and symbiotic host fly lines with associated lower puparia weights. This phenotype could be a symbiont-mediated manipulation of host physiology such that the parasitoids developing in these lines delay development and subsequent emergence until the hosts reach the desired physiological state. Alternatively, it is known that the parasitoid rate of the consumption of its host is inversely proportional to the size of the host, which may increase the development time of the parasitoids, especially in tissue-feeding endoparasitoids [69,70]. It is, therefore, possible that the parasitoids from the P. alcalifaciens and L. lactis fly lines had longer developmental times, as they had to completely consume their relatively larger hosts prior to emergence. These findings, however, contrast with a previous report that other Providencia species shorten the developmental time and alter the sex ratio of D. longicaudata progeny [41]. Differences in bacteria strains used, inoculation strategy, and host species may account for the contrast between our findings and the findings of that study, more so since in the latter, bacterial isolates were supplemented in non-axenic diets, thus accommodating additive effects from other native bacterial symbionts in the host larvae.
Moreover, the finding that the parasitoids emerging from the P. alcalifaciens and L. lactis host fly lines were larger than those reared on the axenic, symbiotic, and C. freundii counterparts conforms with the host-parasitoid size theory, according to which larger hosts are predicted to yield larger parasitoid progeny [69][70][71][72][73]. The larger body size of the parasitoid progeny from these lines could, therefore, be an additive effect of these two symbionts on the size of their hosts, a trait that could be attributed to symbiont-mediated effects on the nutritional resources of the hosts reared on these bacterial lines.
Previous studies suggested that the size of a female parasitoid is directly proportional to its fecundity and longevity, e.g., [72,74,75]. While this held true for the F1 parasitoid progeny emerging from the P. alcalifaciens fly lines, the opposite was recorded for those reared on the L. lactis fly lines, which, despite being larger than the parasitoids from the axenic, symbiotic, and C. freundii fly lines, were less fecund. This finding suggests the L. lactis-associated fitness costs in such a way that the female parasitoid progeny from this fly line allocate more resources towards growth and adult maintenance at the detriment of their reproductive potential. This symbiont may, therefore, have cryptic fitness costs on the reproductive potential of parasitoids that supersede the progeny size benefits. Similar symbiont-mediated tradeoffs in parasitoid fitness have been demonstrated in other hostsymbiont-parasitoid systems [39,62,76].

Conclusions
This study reveals a mixed nature of interactions: a general synergism with the parasitoids with L. lactis and a more beneficial role to the host flies in the case of P. alcalifaciens, suggesting that symbiotic-bacteria could have significant consequences on host-parasitoid dynamics. In this regard, these findings underscore the previously unknown role of bacterial symbionts in mediating the host-parasitoid interactions in B. dorsalis by providing information vital for parasitoid mass rearing. For example, L. lactis can be harnessed as a probiotic in mass-rearing programs of D. longicaudata and F. arisanus, hence enhancing the performance of parasitoids in biological control programs targeting this pest. Nevertheless, future studies are needed to elucidate the mechanisms underlying these symbiont-mediated host-parasitoid interactions.