The eleven strains of M. anisopliae used in our experiment were obtained from the icipe's Arthropod Germplasm Center (Mweke et al., 2018; Akutse et al., 2020). The strains were cultured on Sabouraud Dextrose Agar (SDA) medium using 90-mm Petri dishes and maintained in the darkness at 25°C. Two weeks after the start of the culture, we harvested conidia of each strain by scraping the surface of the sporulated cultures using a sterile spatula. We suspended conidia of the different strains in 10 ml of distilled water with 0.05% Triton X-100 in universal bottles containing 3–5 glass beads (3 mm in diameter per bottle) each. The mixture was then vortexed for 5 min at 700 rpm to homogenize the suspension. Using an improved Neubauer haemocytometer under the light microscope, we determined the conidia concentration of each strain suspension following the protocol described by Lacey (2012). Before each bioassay, we tested the ability of conidia to germinate by spreading 100 μl of each strain suspension (titrated at 3 × 10⁶ conidia ml⁻¹) on SDA plate. We sealed the inoculated plates with Parafilm membrane and incubated them in complete darkness at 25°C. At 18 h post-incubation, we flooded the plates with lactophenol aniline cotton blue to stop the germination process and stain the spore to ease their visibility for counting. Following this, we determined the number of conidia that germinated by counting 100 randomly selected conidia beneath each coverslip under a light microscope (400×). Conidium was considered as germinated if the length of its germ-tube was at least twice its diameter (Lacey, 2012). For each strain, we used five plates as replicates.

Individuals of S. calcitrans used in all our experiment were obtained from the “International Center of Insect Physiology and Ecology (icipe)” Animal Rearing and Quarantine Unit (ARCU) in Nairobi, Kenya (1° 13′ 12″ S, 36° 52′ 48″ E; ≈ 1,600 m above sea level) colony. This colony was established and maintained as described in Baleba et al. (2019). Briefly, wild individuals of S. calcitrans were captured at icipe campus using Vavoua traps (Laveissière and Grebaut, 1990), maintained inside a cage (75 cm × 60 cm × 45 cm) and fed twice per day (800 and 1,600 h) on defibrinated bovine blood poured on moistened cotton to initiate reproduction. Once gravid females were obtained, we exposed them to rabbit dung (fermented in a plastic bag for 1 week) placed in plastic containers (21.5 cm × 14.5 cm × 7.4 cm) for oviposition. After 24 h, we transferred the exposed containers to another cage (75 cm × 60 cm × 45 cm), and we monitored the development of the larval and pupal stages until adult emergence. We fed emerged adults with bovine blood and repeated the previously described above. We reared all the insects and performed our experiments in a laboratory under buffered conditions of 25 ± 5°C 65 ± 5% relative humidity, and 12L:12D photoperiod.

We determined the pathogenicity and virulence of 11 strains of M. anisopliae on S. calcitrans adults following the contamination protocol used by Wamiti et al. (2018) on Glossina fuscipes fuscipes (Diptera: Glossinidae). In this protocol, the contamination device (Supplementary Figure 1A) is comprised of a cylindrical plastic tube (95 mm × 48 mm) which has an inner part covered by a velvet carpet material impregnated with fungal dried conidia. Adult flies were gently introduced into the contamination device and allowed to pick conidia. After this period of exposure, flies were gently removed, and transferred in another cylindrical plastic tube free of conidia. In all our experiment, we used 0.1 g of conidia evenly spread on the velvet carpet, and exposed flies to conidia for 10 min. After transferring the fungus-exposed flies in a clean cylindrical plastic tube, we provided them with blood and recorded the number of dead flies daily for 7 days. We removed cadavers found inside the plastic tube using sterilized forceps and incubated them in Petri dishes containing moistened filter paper to assess the outgrowth of the applied fungal conidia (Supplementary Figure 1B). We used fungus-free flies as control. For each treatment, we used 10 flies and replicated the experiment five times.

We used newly emerged (24 h old) males and females [differentiated based on the size of the two compound eyes that are smaller and more widely separated in females (dioptic) than in males (holoptic)] to see whether the pathogenicity of M. anisopliae varied between the sex of S. calcitrans. In the earlier experiment, we identified ICIPE 30 as the most virulent M. anisopliae strain against S. calcitrans (see results section). Here, we used this strain (0.1 g of dried conidia) to contaminate 10 males and 10 females following the protocol previously described above (Wamiti et al., 2018). As a control, we used unexposed males and females. For the age effect bioassay, we used only 10 exposed female flies (to account for any bias resulting from sex effect) of 1, 7, and 14 days old. The control groups consisted of fungus-free flies of 1, 7, and 14 days old. In both bioassays, we provided each group with bovine blood and recorded individual mortalities daily for 7 days. To confirm whether the death of the flies was caused by M. anispoliae ICIPE 30 infection, we placed dead flies in Petri dishes (9 cm) containing moistened filter paper to initiate fungal sporulation on the cadaver surfaces. We replicated each experiment five times.

Before testing whether M. anisopliae ICIPE 30-exposed males and females could transmit conidia to their conspecific mates, we aimed to determine whether the number of conidia carried by S. calcitrans individuals could vary between sex and across time. To do so, we chilled 5 males and 5 females (2 days old) in ice for 2–3 min to induce a coma. Using fine sterilized forceps, we gently placed the immobilized flies inside the cylindrical plastic tube (on top of velvet carpet containing 0.1 g of ICIPE 30). After recovered from the coma, we allowed the flies to walk on conidia for 30 min, then individually introduced them inside universal bottles containing 2–5 glass beads and 1 ml of sterile distilled water with 0.05% Triton X-100. The bottles with the exposed flies were thereafter vortexed for 5 min (to remove conidia from the insect's body) and estimated the number of conidia carried by each individual using the Neubauer haemocytometer. To assess how the number of conidia carried by each individual be across time, we transferred the exposed male and female flies in a cleaned cage (15 cm × 15 cm × 20 cm), waited for 2, 4, 6, and 8 h before proceeding with the conidia quantification as previously described.

With a slight modification, we followed the protocol described by Maniania et al. (2013) to perform the horizontal transmission assay. We contaminated 5 males (donors) with 0.1 g of M. anisopliae ICIPE 30 conidia for 10 min then transferred them into another clean cage (15 cm × 15 cm × 20 cm). Four hours after this process, we transferred these males inside a clean cylindrical plastic tube and paired them with 5 fungus-free females (receivers). We use the same protocol to pair fungus-exposed females (donors) with fungus-free males (receivers). We considered fungus-free males and females as control. In all the treatments, we provided our flies with blood and recorded their mortality daily for 7 days. To confirm whether the dead in both sexes was induced by M. anisopliae infection, we placed separately the dead bodies of male flies in Petri dishes (9 cm) containing moistened filter paper to later assess fungal growth on the cadaver surfaces. We used five replicates in all the bioassays.

Here, using the previous contamination device, we exposed female S. calcitrans (2 days old) with the M. anisopliae strain ICIPE 30. To determine the effect of this fungal infection on the feeding propensity of S. calcitrans, we recorded three parameters, namely (1) the feeding duration, (2) the proportion of blood-fed, and (3) the amount of blood consumed. We determined the feeding duration by recording the time taken by an individual fly to get engorged after inserting its proboscis into the blood source (Supplementary Figure 2B). The proportion of blood-fed corresponded to the number of flies (in a group of 10 individuals) that managed to take blood after 60 s of their exposure to the blood exposition. The amount of blood consumed per fly was estimated as the difference in their weight, after (Supplementary Figure 2C) and before (Supplementary Figure 2A) the blood meal. As a control, we used fungus-free individuals. We collected all the data 2, 3, and 4 days after fungal infection. The feeding duration and the amount of blood consumed data were obtained from 30 fungus-exposed and fungus-free female flies; while the proportion of blood-fed data were from 5 groups of 10 individuals each.

To elucidate whether M. anisopliae ICIPE 30 could affect the reproduction of S. calcitrans, we used (1) egg-laying decision, (2) fecundity and (3) fertility as proxies. To test the effect of M. anisopliae ICIPE 30 on S. calcitrans egg-laying decision, we conducted two oviposition choice bioassays (Supplementary Figure 3A). In the first bioassay, we exposed 10 gravid female S. calcitrans to two Petri dishes (Diameter: 5.5 cm) containing each, only 50 g of rabbit dung to see whether they will lay the same number of eggs on both Petri dishes. For the second bioassay, we presented rabbit dung supplemented with 0.1 g of M. anisopliae conidia (strain ICIPE 30) and rabbit dung only (control) to 10 gravid female S. calcitrans to see whether these females will select either substrate preferentially. In both bioassays, we used 10 replicates, and for each replicate, we counted the number of eggs laid on each substrate after 24 h and determined their ability to hatch 5 days after egg deposition (by counting the number of larvae found on each substrate).

To assess the effect of M. anisopliae ICIPE 30 on the fecundity (number of eggs laid) and fertility (number of eggs hatched) of S. calcitrans, we performed two no-choice oviposition bioassays (Supplementary Figures 3Bi,ii). To do so, following the previously described protocol, we exposed 30 females (4 days old) to M. anisopliae strain ICIPE 30 and transferred them individually inside cages (15 cm × 15 cm × 20 cm) containing 2 males to allow mating. We supplied these flies with blood daily and once females become gravid, we provided them with a Petri dish containing rabbit dung for oviposition. As a control, we used fungus-free gravid female S. calcitrans. We recorded the number of eggs laid on each substrate daily until the female succumbs from fungal infection. To assess the fertility of eggs laid by infected and uninfected gravid female S. calcitrans, we determined their hatchability by counting the number of larvae found on each substrate 5 days after the egg deposition.

It has been reported previously that S. calcitrans larvae are not susceptible to M. anisopliae infection (Moraes et al., 2008). To test this, we infected second larval instar of this fly with the same 11 strains of M. anisopliae as described above. For each strain, we sprayed 10 larvae (placed on a Petri dish) with 10 ml of suspension at the concentration of 2 × 10⁸ conidia ml⁻¹ using a Burgerjon's spray tower (Burgerjon, 1956). After spraying, we transferred the contaminated larvae in transparent plastic cups of 200 ml prior containing 50 g of rabbit dung. As a control group, we used larvae treated with sterile distilled water containing 0.05% Triton X-100. We recorded the number of dead larvae daily until pupation. We carried out all the treatments in five times. We observed that most infected S. calcitrans larvae (90 %) managed to reach the pupal stage. Therefore, in a subsequent bioassay, we aimed to challenge these larvae with higher concentrations of M. anisopliae. As previously described, we contaminated 10 S. calcitrans larvae with four increasing conidia concentrations (3 × 10⁸, 4 × 10⁸, 5 × 10⁸, and 6 × 10⁸ conidia ml⁻¹) of the strain ICIPE 30 and recorded the number of dead larvae daily until pupation.

To test whether the tolerance to M. anisopliae ICIPE 30 infection in S. calcitrans larvae could impact their fitness parameters, we contaminated 10 individuals of each S. calcitrans larval instar (L1, L2, and L3) with 10 ml of ICIPE 30 concentrated at 2 × 10⁸ conidia ml⁻¹. It is indicated that life stages that undergo metamorphosis (occasioning the change in size and morphology) should be treated independently when studying their responses to biotic stresses (McCormick and Gagliano, 2009; Kingsolver et al., 2011; Ezeakacha and Yee, 2019). We followed the contaminated larvae daily until the adult stage by recording the following life-history fitness parameters: (1) pupation time, (2) larval weight, (3) pupation rate, (4) pupal weight, (5) emergence percentage, (6) emergence time, and (7) adult weight. Larval, pupal, and adult weight data were collected as described in Baleba et al. (2020). For the weight parameter, we weighed all the larvae individually, as well as pupae and adults that emerged from contaminated larvae. We recorded larval weight 2, 4, and 6 days after fungal contamination in the individuals from L1 and L2 instars. While in individuals from the L3 instar (close to the pupal stage), we recorded weight only 2 days after contamination. As a control, we used L1, L2, and L3 individuals sprayed with sterile distilled water containing 0.05% Triton X-100. We replicated this experiment five times.

We conducted all the statistical analysis in the R environment for statistical computing (version 3.6.3) (R Core Team., 2020) and grouped all the graphs in Adobe Illustrator CC 2017(version 21.0). Before conduct the analysis, we subjected mortality data to Abbot's correction (Abbot, 1925).

For the bioassay aiming to study the effect of the 11 strains of M. anisopliae on the S. calcitrans survival, we performed Kaplan–Meier survival analysis with the Mantel–Cox log-rank chi-squared test using the R package “survival” (Therneau, 2015) to see how the survival of S. calcitrans adults varied as a result to exposure to the different fungus strains. Owing to the normal distribution (Shapiro–Wilk test: P > 0.05) and the homoscedasticity (Bartlett's test: P > 0.05) of the median lethal time data, we ran the analysis of variance (ANOVA) followed by the Student–Neuman–Keuls (SNK) post-hoc multiple comparison tests to see how this parameter varied across the 11 strains. For the same reason, we performed the ANOVA followed by the SNK post-hoc tests to compare the proportion of alive S. calcitrans (at 7th day of our bioassay) across the 11 strains.

In the experiment testing the effect of sex and age on M. anisopliae infectivity, we used the Kaplan–Meier survival analysis with the Mantel–Cox log-rank chi-squared test to elucidate how these factors affected the infectivity of M. anisopliae. We employed the unpaired t-test to compare the median lethal time between the sexes of S. calcitrans. To determine whether this parameter could vary across the three ages of S. calcitrans (1, 7, and 14 days), we performed the ANOVA followed by the SNK post-hoc tests. Using the same analysis, we compared the number of alive S. calcitrans (at 7th day of our bioassay) across the sex and the ages.

For the experiment aiming to test whether, in S. calcitrans, M. anisopliae conidia could be transferred from one sex to another, we used the Kaplan–Meier survival analysis with the Mantel–Cox log-rank chi-squared test to see whether the survival of the fungus-donor, fungus-receiver, and fungus-free (control) S. calcitrans could significantly vary. We ran the unpaired t-test to compare the median lethal time between fungus-donor and fungus-receiver flies. We performed the ANOVA followed by the SNK post-hoc tests to compare the number of fungus-donor, fungus-receiver, and fungus-free S. calcitrans that were still alive at the end of our experiment (7th day).

In the feeding propensity test, we used the unpaired Wilcoxon test to compare the feeding time of infected and uninfected flies. Owing to the binary nature of the feeding proportion data (engorged vs. not engorged) we performed a generalized linear model (GLM) with binomial distribution followed by the analysis of deviance (with the chi-squared test) to see how the proportion of blood-fed flies varied between infected and uninfected flies. We executed the unpaired t-test to compare the amount of blood taken by infected and uninfected flies.

For the experiment testing the effect of M. anisopliae on S. calcitrans reproduction, we used the paired t-test to compare the number of eggs laid by gravid females S. calcitrans on the two Petri dishes containing only rabbit dung. We used the same statistical analysis to compare the number of eggs laid by these females on Petri dishes with and without M. anisopliae ICIPE 30 dried conidia. The Egg hatchability data were binary (hatched vs. unhatched); therefore, we used a GLM with binomial distribution and analysis of deviance (with chi-squared test) to see how this parameter varied between substrates with and without conidia. To compare the number of eggs laid by infected and uninfected gravid females S. calcitrans, we used an unpaired t-test. We compared the hatchability the eggs produced by these females, using a GLM with binomial distribution and analysis of deviance (with chi-squared test).

For the data from the bioassay testing the effect of M. anisopliae on the survival of S. calcitrans larvae, we performed the ANOVA test to compare the proportion of larvae that pupated across the different M. anisopliae strains and the ICIPE 30 concentrations. In the experiment testing the effect of M. anisopliae infection on the life-history parameters of three different larval instars of S. calcitrans, we subjected data from pupation time, larval weight, pupal weight, emergence time, and adult weight to the normality and homogeneity tests. In case the data of a particular parameter were normally distributed (Shapiro–Wilk test: P > 0.05) and their variances were homogeneous (Bartlett's test: P > 0.05), we used the unpaired t-test to see how this parameter varied between infected and uninfected larvae. When these two assumptions were not fulfilled, we used the unpaired Wilcoxon test. We analyzed the pupation and emergence percentage data using a GLM with binomial distribution and analysis of deviance (with chi-squared test).

Statistical significance was noted at P < 0.05 and its strength was represented with asterisks (^*P < 0.05; ^**P < 0.01; ^***P < 0.001, and ^****P < 0.001).

All the 11 strains of M. anisopliae used in our study possessed a germination percentage above 90% (Figure 1A). As time progressed, the proportion of S. calcitrans surviving from M. anisopliae infection reduced with a significant difference across the strains (Figure 1B; log-rank test, χ² = 50.3, df = 11, P < 0.0001). The median lethal time [Figure 1C; One-way ANOVA: F_(10−44) = 7.79, P < 0.0001] and the proportion of alive S. calcitrans after the 7 days post-infection [Figure 1D; One-way ANOVA: F_(11−47) = 17.5, P < 0.0001] significantly differed across the 11 strains of M. anisopliae. Of all the S. calcitrans individuals infected with 11 strains of M. anisopliae, only those infected with the strain ICIPE 30 had simultaneously, lower median lethal time (Figure 1C) and lower proportion of alive individuals (20%) at the end of our experiment (Figure 1D). Although at the end of our experiment, the proportion of alive S. calcitrans infected with the strain ICIPE 7 was similar to that of S. calcitrans infected with the strain ICIPE 30 (Figure 1D), the strain ICIPE 7 took the longest time (>5 days) to kill half individuals of S. calcitrans as compared to the strain ICIPE 30 (<4 days) (Figure 1C). Therefore, we considered the strain ICIPE 30 as the most potent and virulent for S. calcitrans.

In both sex of S. calcitrans, the survival over time of uninfected (males and females) and infected (males and females) individuals significantly varied (Figure 2Ai; log-rank test, χ² = 11.8, df = 3, P = 0.008). But the pairwise comparison using the log-rank test revealed that the survival of infected males and infected females over time did not significantly differ (P = 0.26). This was also true for uninfected males and uninfected females (P = 0.39). The mean median lethal time of infected female S. calcitrans was significantly higher than that of infected males (Figure 2Aii; U = 22, P = 0.04). At the end of our experiment, infected flies of both sexes had a significantly lower proportion of alive individuals as compared to that uninfected flies [Figure 2Aiii; One-way ANOVA: F₍₃₋₁₆₎ = 27.27, P < 0.0001]. But this proportion was similar between infected males and females; and uninfected males and females.

The survival of S. calcitrans over time significantly changed across the different age of flies (log-rank test, χ² = 16.9, df = 5, P = 0.005); with 14-days old infected possessing the lower survival rate (Figure 2Bi). As compared to 1 and 7 days old, 14-days old infected flies had a smaller median lethal time [Figure 2Bii; One-way ANOVA: F₍₂₋₁₂₎ = 16.65, P < 0.001]. Regardless of the age, the proportion of alive flies obtained 7 days after contamination was significantly lower in infected flies and as compared to uninfected flies [Figure 2Biii; One-way ANOVA: F₍₅₋₂₄₎ = 59.39, P < 0.0001]. In infected flies, this proportion was significantly lower in 14 days old flies followed by 7 and 1-day old flies (Figure 2Biii).

The amount of M. anisopliae ICIPE 30 conidia carried by S. calcitrans significantly varied across time (P < 0.0001) with no variations between sex (P = 0.051). This amount was higher directly after the contamination process; but 2 h later, it drastically dropped with no significant change even 8 h after the fly's contamination (Figure 3A). We found that 4 h after exposure to M. anisopliae ICIPE 30 conidia, fungus-exposed flies (donors) were still able to contaminate fungus-free flies (receivers). When we contaminated S. calcitrans males (donors) and associated them with fungus-free females (receivers), the survival of these flies over time significantly reduced as compared to those of males and females maintained uncontaminated throughout the bioassay (Figure 3Bi; log-rank test, χ² = 26.2, df = 3, P < 0.0001). The median lethal time of male donors was significantly lower as compared to that of female receivers (Figure 3Bii; t = 3.6, df = 6, P = 0.013). Both male donors and female receivers had reduced proportion of alive individuals at the 7th-day post-contamination as compared to males and females maintained uncontaminated [Figure 3Biii; One-way ANOVA: F₍₃₋₁₄₎ = 16.29, P < 0.0001]. We obtained the same result pattern when we contaminated females (donors) and associated them with fungus-free males (receivers). The survival of female donors and male receivers significantly reduced over time as compared to the survival of uncontaminated males and females (Figure 3Ci; log-rank test, χ² = 16.2, df = 3, P < 0.001). Female donors had a lower mean median lethal time as compared to that of male receivers (Figure 3Cii, U = 2, P = 0.026). The proportion of female donors and male receivers alive at the end of our experiment was significantly lower than those of uncontaminated males and females [Figure 3Ciii; One-way ANOVA: F₍₃₋₁₇₎ = 10.2, P < 0.001].

Infection of S. calcitrans by M. anisopliae ICIPE 30 significantly altered its feeding propensity. Fourty-eight (48) hours after fungal exposure, the time taken by S. calcitrans to get engorged did not vary between infected and uninfected flies (t = 0.06, df = 67.7, P = 0.95). But, 72 (t = 2.18, df = 65.5, P = 0.032) and 96 (U = 466, P < 0.001) h after the conidia exposure occurred, infected flies took significantly more time to get engorged as compared to uninfected flies (Figure 4A). Sixty (60) seconds after blood exposure, 48-h infected flies and uninfected flies had the same proportion flies that managed to take blood (Figure 4Bii; GLM, χ² = 19.92, df = 1, P = 0.50). Nonetheless, this proportion significantly reduced in 72-h (GLM, χ² = 9.5, df = 1, P = 0.042) and 96-h (GLM, χ² = 25.53, df = 1, P = 0.004) infected flies (Figure 4B). The amount of blood taken by infected flies at 48 (t = 4.60, df = 61.5, P < 0.0001), 72 (t = 3.71, df = 66, P < 0.001), and 96 (t = 2.9, df = 39.4, P < 0.01) h after fungal exposure was significantly lower as compared to that of uninfected flies (Figure 4C).

In our dual-test oviposition bioassay, when we presented two fungus-free substrates to S. calcitrans gravid females, they laid the same number of eggs on both substrates (Figure 5Ai, t = 0.83, df = 17.9, P = 0.41). However, when we added dried conidia of M. anisopliae on one of the substrates, these females laid significantly few numbers of eggs on the fungal-embedded substrate (Figure 5Aii; U = 0, P < 0.01). Consequently, the proportion of eggs deposited on the substrate with dried conidia that hatched was significantly less than that of eggs laid on the substrate without dried conidia (Figure 5Aiii, GLM; χ² = 23.07, df = 1, P < 0.0001). In the no-choice bioassay, S. calcitrans gravid females infected with M. anisopliae ICIPE 30 laid a significantly fewer number of eggs than uninfected S. calcitrans gravid females (Figure 5Bi; t = 5.08, df = 11.84, P < 0.001). Also, the hatchability of eggs laid by infected females was significantly lower than that of eggs deposited by uninfected females (Figure 5Bii, GLM; χ² = 26.63, df = 1, P < 0.0001).

As demonstrated in other studies, our study showed that S. calcitrans larvae are not susceptible to infection with M. anisopliae strains used. All the M. anisopliae strains-contaminated larvae and uncontaminated larvae (control) had similar pupation percentage [F_(11−48) = 0.42, P = 0.94]. For each treatment, about 90% of larvae reached the pupal stage (Figure 6Ai). Even at high conidia concentrations, the pupation percentage of contaminated larvae did not reduce [Figure 6Aii; F₍₄₋₂₀₎ = 0.62, P = 0.65]. Even though the high proportion of contaminated larvae did not succumb from M. anisopliae exposure (since some managed to pupate), we demonstrated a negative impact of this contamination on some of their fitness parameters.

We observed that the pupation time was significantly longer in contaminated first (Figure 6Bi1: U = 70, P < 0.0001), second (Figure 6Bii1: U = 255.5, P < 0.0001), and third (Figure 6Biii1, U = 346, P < 0.0001) larval instars of S. calcitrans as compared to that of their corresponding control (uncontaminated larvae). Also, the weight of contaminated first (Figure 6Bi2: Day 2: t = 8.45, df = 14.4, P < 0.0001; Day 4: U = 317, P = 0.002; Day 6: t = 3.2, df = 17.98, P = 0.005), second (Figure 6Bii2: Day 2: t = 5.5, df = 14.6, P < 0.0001, Day 4: t = 3.5, df = 18, P = 0.002; Day 6: t = 2.4, df = 15.9, P = 0.026), and third (Figure 6Biii2; t = 2.97, df = 15.4, P = 0.009) larval instar of S. calcitrans was smaller than those of uncontaminated instars. The pupation percentage did not significantly change between contaminated and uncontaminated individuals from the first (Figure 6Bi3; GLM; χ² = 3.55, df = 1, P = 0.06), second (Figure 6Bii3; GLM; χ² = 3.25, df = 1, P = 0.071), and third (Figure 6Biii3; GLM; χ² = 2.48, df = 1, P = 0.11) larval instars. The pupal weight was significantly reduced in contaminated individuals developed from the first larval instars (Figure 6Bi4: t = 3.5, df = 80, P < 0.001). However, pupae formed from the contaminated second (Figure 6Bii4: t = −0.73, df = 37.66, P = 0.46), and third (Figure 6Biii4: t = −0.89, df = 65.51, P = 0.37) larval instars had similar weight with those developed from uncontaminated second and third larval instars respectively. Pupae developed from contaminated and uncontaminated first (Figure 6Bi5: U = 1002.5, P = 0.12), second (Figure 6Bii5: U = 229.5, P = 0.16), and third (Figure 6Biii5: U = 262.5, P = 0.32) larval instars formed at the same time. The proportion of adults obtained from contaminated and uncontaminated first (Figure 6Bi6: GLM; χ² = 1.46, df = 1, P = 0.22) and second (Figure 6Bii6: GLM; χ² = 0.002, df = 1, P = 0.97) larval instars was not significantly different; but in the third larval instar, this proportion was significantly higher in uncontaminated larvae (Figure 6Biii6: GLM; χ² = 7.04, df = 1, P = 0.008). Adult obtained from contaminated first larval instars had a significant reduced weight as compared to those emerged from uncontaminated first larval instars (Figure 6Bi7: t = −3.27, df = 68.8, P = 0.002). While those obtained from contaminated and uncontaminated second (Figure 6Bii7: t = −1.06, df = 44.2, P = 0.29) and third (Figure 6Biii7: t = 1.27, df = 29.45, P = 0.21) larval instars had the same weight. Independently to the larval instars, we numbered 12 deformed adults from contaminated larvae and no deformed adults from uncontaminated larvae (Figure 6C).