Production and shipment of wolbachia-infected eggs allows the control of aedes albopictus through the incompatible insect technique on a remote island

Abstract

Climate and land-use changes are accelerating the spread of the mosquito Aedes albopictus, a major arbovirus vector, leading to the emergence and autochthonous transmission of dengue or chikungunya viruses in temperate regions such as Italy and France. This has prompted the development of alternative control strategies to counteract growing insecticide resistance. The Incompatible Insect Technique (IIT) uses mass releases of Wolbachia-infected males that sterilize wild females through Cytoplasmic Incompatibility (CI). We conducted a six-month IIT suppression trial on a remote island in the Western Indian Ocean using a novel Aedes albopictus line with a single Wolbachia infection inducing bi-directional CI. This ensures both released males and accidentally released females are incompatible, avoiding population replacement, a common limitation of standard IIT. The trial was conducted under operational conditions, with release numbers adjusted to wild population densities. Eggs were produced in a central insectary over 1000 km away, shipped by plane and boat to the release site. Our results show: (i) over 95% suppression within a few weeks of treatment, (ii) effective prevention of population replacement, and (iii) successful long-distance egg shipment. These findings support the feasibility, scalability, and field-readiness of this environmentally friendly vector control approach.

Production and shipment of Wolbachia-infected eggs for controlling a mosquito (Aedes albopictus) population on a remote island in the Western Indian Ocean through the Incompatible Insect technique.

Introduction

Aedes albopictus has for a long time been considered as a poor disease vector because of a zoophilic behaviour and a limited competence for the transmission of some arboviruses¹. This mosquito species is being increasingly scrutinized because it has been the major vector of a number of dengue or chikungunya epidemics in tropical regions within the last decades. Most importantly, the plasticity and invasive behaviour of this vector, together with accelerating global changes, are leading to a rapid expansion of its geographic distribution, including in temperate countries where Ae. albopictus has proven efficient for autochthonous transmission of dengue virus^2,3. In the absence of available vaccines, transmission is routinely reduced by controlling vector populations using insecticides. However, more environmental-friendly methods, such as the Sterile Insect Technique (SIT) or the Incompatible Insect Technique (IIT), are being developed to bypass the selection of insecticide resistance and reduce the impacts of pesticides on non-target species. Both methods rely on the release of sterile or incompatible males to reduce the population size of the targeted species, offering the advantage of species-specific control allowing sterilization of females in cryptic habitats or effective control of insecticide-resistant mosquito populations⁴.

IIT leverages the properties of Wolbachia pipientis, a maternally inherited endosymbiotic bacterium that infects approximately 40% of arthropod species^5,6. Wolbachia can manipulate mosquito reproduction through Cytoplasmic Incompatibility (CI), a form of conditional embryonic lethality that occurs when a Wolbachia infected male mates with a female that is either uninfected or infected by a different, incompatible Wolbachia strain⁵. CI can be understood as resulting from a modification/rescue mechanism (mod/resc), wherein the mod function modifies the sperm during spermatogenesis and the resc function, expressed in the egg cytoplasm laid by Wolbachia-infected females, rescues the embryo through an interaction with the modified sperm^7,8. Wolbachia genes involved in CI induction are organized into an operon-like genetic element that consists of two genes, which are named cifA and cifB^9–11.

CI can be hijacked for the development of the IIT, which involves the release of Wolbachia-infected males that are incompatible with resident females^12,13. Recent studies have demonstrated the effectiveness of standalone IIT in controlling Ae. aegypti populations in the U.S¹⁴, Australia¹⁵, Singapore¹⁶, and Puerto Rico¹⁷ as well as Ae. albopictus populations in China¹⁸. Importantly, all Wolbachia-transinfected lines used in the aforementioned field assays exhibit unidirectional CI with the resident females, meaning that accidentally released females are compatible with resident males, thus increasing the risk of population replacement with Wolbachia transinfected mosquitoes. To mitigate this risk, which may lead to IIT failure, it has been proposed to combine incompatible and sterile insect techniques, utilizing Wolbachia infection to induce male sterility and a low irradiation dose to sterilize those females that the sexing processes fail to eliminate^19–21. An alternative approach is the exploitation of a bidirectional pattern of CI (Bi-CI), occuring when the native Wolbachia infection is removed and replaced by a foreign Wolbachia inducing CI^22,23. In this case, transinfected females are only compatible with co-released transinfected males, significantly reducing the invasive potential of Wolbachia²⁴. Indeed, assuming a similar fitness of natural and Wolbachia transinfected mosquitoes, theory predicts that the Wolbachia infection frequency must exceed 50% to allow invasion²⁵, compared to >0% in the case of uni-CI. In practice, invasion thresholds depend on the strength of CI, the efficiency of Wolbachia maternal transmission, and on fitness costs associated with infection^26,27. For example, in an Ae. aegypti wMel transinfected line, which induces uni-CI and incurs relatively small fitness costs, population replacement requires releases exceeding 30% of the local population²⁸. Under these circumstances, population replacement could occur only if the objective of IIT deployment is complete elimination, which, to our knowledge, has never been achieved in any SIT/IIT mosquito program⁴.

When Bi-CI occurs, models suggest that immigration rates of approximately 2% of wild type individuals in the release area would be sufficient to mitigate the risk of population replacement in the long term²⁹. Thus, the use of standalone IIT with a Wolbachia strain inducing Bi-CI is expected to achieve high levels of population suppression with limited risk of population replacement, as has been demonstrated in Australia to control a wMel-Ae. aegypti population¹⁵. However, in Ae. albopictus, this approach has been discussed primarily in theoretical terms^30,31, and no field trials have yet demonstrated sustained, high-level suppression using stable bi-directional CI to control this vector over the long term.

In this context, we implemented a 6-month IIT program aiming at controlling an Aedes albopictus population on a 9-hectare island in the Seychelles, utilising a mono-transinfected mosquito line that induces bi-directional CI with the local population. We demonstrate that IIT deployment achieves population suppression reaching 95% within 3 months of the initial release, and a suppression level maintained above 50% until the end of the release period. Moreover, we show that while achieving a high suppression level over several months, the contamination percentage of incompatible individuals in the field never exceeds 7% and drops to 0% 2 months after the end of the releases. Finally, we demonstrate the feasibility of a mass production model in which mosquito eggs are produced at a central facility and subsequently distributed to a satellite rearing center dedicated to producing incompatible males only^32,33. Altogether these results support the viability of this mass production model for industrial deployment.

Results

CI penetrance of the SEY-wPip line

Incompatible male releases were conducted using an Ae. albopictus wPip transinfected line²² introgressed with a wild genetic background from the Seychelles, resulting in the SEY-wPip line. Reciprocal crosses between wild lines and SEY-wPip confirmed that wPip induced bidirectional CI in Ae. albopictus regardless of the host genetic background^22,34. Indeed, when SEY-wPip males were crossed with wild females, they induced nearly complete CI, with a hatching rate ranging from 0.37 to 1.05%. Symmetrically, when wild males were crossed with SEY-wPip females, the hatching rate ranged from 0.47 to 9.41% (Table 1).

Table 1.

Hatch rate obtained in control and reciprocal crosses between the SEY-wPip and three different wild lines

Crosses	nb of hatched eggs	nb of unhatched eggs	Hatch rate (%) (±SD)	CI (%)	CIcorr (%)
♂ SEY-wPip x ♀ Le Port	78	9053	1.05 (±1,67)	98.95	92.93
♂ SEY-wPip x ♀ S-RUN	45	8007	0.59 (±0,70)	99.41	97.38
♂ SEY-wPip x ♀ La Providence	17	4683	0.37 (±0,07)	99.63	98.33
♂ Le Port x ♀ SEY-wPip	61	7491	0.76 (±0.31)	99.24	96.34
♂ S-RUN x ♀ SEY-wPip	25	5401	0.47 (±0.21)	99.53	97.74
♂ La Providence x ♀ SEY-wPip	465	4562	9.41 (±2.96)	90.59	54.69
♂ Le Port x ♀ Le Port	10,468	1701	85.14 (±7.30)
♂ S-RUN x ♀ S-RUN	5416	1652	77.51 (±4.92)
♂ La Providence x ♀ La Providence	1918	563	77.88 (±5.80)
♂ SEY-wPip x ♀ SEY-wPip	9875	2483	79.23 (±7.94)

Life history traits (LHTs) and mating competitiveness of the SEY-wPip line in laboratory conditions

To assess the fitness of the SEY-wPip line following introgression, we compared different LHTs with those of two wild lines, S-RUN and Le Port. The traits analyzed included longevity, number of laid eggs and hatching rate. For both sexes, Le Port line exhibited greater longevity than the S-RUN and SEY-wPip lines (log-rank test; P < 0.001 for both comparisons), and SEY-wPip also had a longer longevity than S-RUN, although this difference was significant in males only (Supplementary Fig. 1A, B). We also observed a significant difference in the number of laid eggs between Le Port line, recording an average of 47.25 (±14.65) eggs laid per female, and both SEY-wPip and S-RUN lines, for which similar number of eggs were laid, 40.52 (±12.30) and 42.19 (±16.01), respectively (GLMM; quasi-poisson family; P = 0.627) (Supplementary Fig. 1C). We observed a similar hatch rates in Le Port and S-RUN lines, with a mean of 83.97% (±19.25) and 82.94% (±21.05), respectively. These hatch rates were significantly higher than those observed for the SEY-wPip line showing a mean value of 75.89% (±25.75) (Supplementary Fig. 1D).

Finally, we assessed the mating competitiveness of SEY-wPip males compared to wild-type males from the S-RUN and Le Port lines, by crossing virgin wild females with increasing ratios of incompatible (SEY-wPip) to compatible (S-RUN or Le Port) males. Observed hatch rates at 1:0 and 0:1 ratios (where the first and second numbers correspond to the proportions of compatible and incompatible males, respectively) were compared with the expected hatch rates under the assumption that both male types had equal mating competitiveness. As expected, hatch rates decreased with increasing proportions of incompatible males, in crosses involving wild females from S-RUN or Le Port lines (Supplementary Fig. 1E). Of note, when different combinations of SEY-wPip and S-RUN males were used, observed hatch rates were consistently and significantly lower than expected, indicating that incompatible males were more competitive than S-RUN males. In crosses involving SEY-wPip and Le Port males, observed and expected hatch rates were not different at 1:5 and 1:10 ratios. However, at a 1:1 ratio, the observed hatch rate was 20% higher than expected. Altogether, SEY-wPip appears as a mono-transinfected line suitable for mass-production and subsequent use in an IIT program, as the potential fitness cost associated with wPip infection and/or genetic introgression are absent or minimal.

Estimation of the natural population size before releases

The density of mosquito population at the release site was evaluated using mark-release-recapture (MRR) experiments. A total of 4389 and 5682 wild males marked with a fluorescent pigment were released during the first and second MRR experiment, conducted in October and November 2022, respectively performed 9 and 8 months before the first release. For both MRRs, the recapture rate of marked males was 1.5% over 4 consecutive days (68/4389 for the first and 87/5682 for the second MRR). Based on the first MRR, we estimated a total number of wild males on the whole island of 9994 (±1984), while the second yielded an estimate of 15,540 (±3743). We used the data of the second MRR to calibrate the number of incompatible males to be released to ensure a minimum incompatible male ratio of 5:1. This number was estimated to ~80,000 males per week (15,540 * 5).

Shipping of eggs, release of incompatible males and suppression of the Ae. albopictus population

The control site consists of a 7-hectare area in the Southern part of Saint Anne island, while the release site encompassed the entire 9-hectare area of Moyenne island. These sites are separated by 1.2 km (Fig. 1).

Fig. 1. Control and release sites, located within the Sainte Anne Marine National Park in the Seychelles.

The control site (7 ha) is a part of Sainte Anne island, while the release site corresponds to the entire surface of Moyenne island.

The mosquito population dynamics in both sites was monitored weekly before male releases, using ovitraps from 15th February, 2022 to 27th June, 2023. During this period, population trends were similar (Supplementary Fig. 2A–C), as supported by the absence of significant difference in the number of hatched eggs between the two sites (Wilcoxon rank sum test; P = 0.608), and confirming the suitability of the selected control site. Likewise, no significant difference was observed in egg hatching rates between the release (95.27 ± 5.09%) and control (94.87 ± 5.79%) sites during the prerelease phase (GLMM-quasi-binomial family; P = 0.947) (Supplementary Fig. 2B), nor the number of laid eggs (Supplementary Fig. 2C).

Three months before the start of incompatible male releases, eggs of the SEY-wPip line, continuously produced in Saint-Denis (La Réunion), were shipped weekly to the Seychelles in bubble wraps via commercial airlines. Over 9 months of shipping, a total of approximately 145 g of eggs (~4 g per week) was transported. Delivery time ranged from 7 to 15 days, depending on the regulatory inspection process at the Seychelles airport. Eggs were sent out in 32 shipments with an overall sending success rate of 90% (29/32), meaning that the observed hatching rate was close to the expected value.

Releases of incompatible males started on 13th July 2023, and were maintained for 6 consecutive months. The release period was eventually separated in three phases, with distinct numbers of released males (Fig. 2D). Approximately 80,000 males were released every week for 12 weeks during phase I. We observed an immediate suppression in the 1st week following the first release, with a 30% decrease in the hatch rate at the release site. By the 2nd week, the hatch rate had dropped by 60% (Fig. 2A). At that time, using BG traps, we measured a male release ratio of 23:1 (SEY-wPip to wild males) (Fig. 2E). Then the hatch rate continued to decline throughout phase I, reaching 18.65% at the end of phase I (3 months after the first release), compared to 93.10% in the control site (Fig. 2A). Although the number of hatched eggs temporarily decreased in the control site during the first few weeks after the first release, it remained significantly lower in the release site throughout phase I (13th July- 29th September, 2023) (P < 0.001) (Fig. 2B). The impact of IIT deployment became particularly obvious from the 8 week following the first release, even if the male release ratio dropped to 1.5:1 (SEY-wPip to wild males) (Fig. 2E), with a male mating competitiveness index close to 1, meaning that incompatible males were as competitive as the wild males (Fig. 2E) At this point, the average number of hatched eggs in the control site began to increase weekly, reaching a mean of 49.15 (±33.92) by the end of phase I. In contrast, the number of hatched eggs remained consistently low in the release site, with a mean of 3.08 (±8.27) during the same period. Additionally, at the end of the phase I, 12 out of 26 ovitraps in the release site contained no hatched eggs. At this point the male release ratio increased to 7:1 (SEY-wPip to wild males), with a male mating competitiveness index of 0.6 (Fig. 2E), After 12 weeks of releases, this resulted in an overall population suppression level of 95% (Fig. 2B).

Fig. 2. Suppression of the wild population.

a Average hatching rate per week in the release and control sites. b The average number of hatched eggs per week and suppression efficiency. The red dashed line represents the suppression efficiency on the release site compared to the control site. Vertical blue and green lines represent standard deviation. c The weekly temperature and the total amount of precipitation before, during and after the release period (data source: NASA POWER Data Access Viewer⁶³). d Total number of SEY-wPip males released each week during the release period. e Incompatible male ratios and associated male mating competitiveness index calculated during the release period.

After achieving a population suppression of 95%, we reduced the number of incompatible males released to approximately 50,000 per week, for 6 consecutive weeks, corresponding to phase II (Fig. 2D). During this phase, we were able to maintain control over the population size, as evidenced by a significant difference in the number of hatch eggs between the release and the control sites (Wilcoxon rank sum test; P < 0.001). However, towards the end of phase II, we observed an increase in the number of hatched eggs in the release site during the final 3 weeks, coinciding with a lower male release ratio of 1:1 (SEY-wPip to wild males) (Fig. 2E). Between the end of phase I and phase II the hatch rate increased by 23%, while the suppression level decreased by 35% (Fig. 2A, B).

In response to the decline in effectiveness, we increased the number of incompatible males released to approximately 140,000 per week for the last 4 weeks of releases, corresponding to phase III. This adjustment significantly restored control efficiency, stabilizing the hatch rate and increasing the population suppression level, which reached 80% by the end of releases. During phase III, we also observed a significant difference in the number of hatched eggs between the release and control sites (Wilcoxon rank sum test; P < 0.001).

To investigate the duration of suppression following the end of releases, we continued to monitor the population for 10 additional weeks. During this post-release survey, we observed a significant difference in the number of hatched eggs between sites (Wilcoxon rank sum test; P < 0.001). However, starting from the 7th week after the last release, the hatch rate and the number of hatched eggs in the release site became comparable to those observed in the control site.

Spatial heterogeneity of mosquito suppression

An analysis of spatial population dynamics revealed that egg hatch was impacted by the release of incompatible males at all surveyed ovitraps. However, we found that this impact was heterogeneous, with some areas being more difficult to maintain at low densities throughout the entire release period, particularly in the southwestern part of the release site (Fig. 3).

Fig. 3. Spatial-temporal dynamics of hatched eggs at the release and control sites for each release phase.

The different color indicates the average number of hatched eggs.

Absence of population replacement

The experimental set-up allowed addressing whether standalone IIT using a transinfected line exhibiting bi-CI could strongly suppress a field population without causing population replacement. To assess this, we firstly quantified weekly the female contamination rate in the male pupae production following mechanical sex separation. The female contamination rate ranged from 0.015 to 0.08% throughout the entire mass production period (Fig. 4A). During this quality control of the production, all pupae identified as females were removed. However, since not all male pupae were verified and because human verification is inherently imperfect, the release of accidental females did occur as shown by the presence of wPip females at the study site (see below).

Fig. 4. Temporal dynamics of wPip females in laboratory male production and at the release site.

a Female contamination rates measured in SEY-wPip males during the mass production. For each week of mass production, we indicated the total number of pupae individually inspected under a binocular loop and between brackets the percentage of pupae individually inspected by the total number of adult released males. b wPip-positive individual rate detected at the release site measured using ovitraps. For each screening we indicated the number of wPip-positive individuals and the total number of screened adults. Between brackets: total number of ovitraps for which at least one wPip-positive individual was detected.

Thus, to monitor the risk of SEY-wPip mosquito establishment in the field, we collected and reared all larvae from eggs found in ovitraps at the release site every 3 weeks, from phase I through the end of the post-release phase. Larvae were reared until adulthood to determine the percentage of wPip-positive individuals per ovitrap. In total, this monitoring was conducted seven times during the releases and four times during the post-release period, representing 1196 screened adults.

We observed a wPip-positive rate ranging from 0 to 6.3% during the release period, with a maximum of three positive ovitraps (Fig. 4B). Although wPip-positive individuals were detected several weeks after the last release, their frequency never exceeded 7%. Importantly, no positive samples (N=176) were found in the final screening conducted more than 2 months after the last release. Moreover, analysis of the spatial distribution of positive ovitraps revealed that they were located in different areas across screening, with only one ovitrap testing positive on two consecutive occasions (Supplementary Fig. 3). These findings support the hypothesis of an absence of local establishment and instead suggest sporadic releases, involving one or a few females released at different locations from week to week.

Discussion

Aedes albopictus is playing a major role in the emergence or re-emergence of dengue and chikungunya in tropical regions and has also been implicated in outbreaks in newly-colonized temperate regions such as Italy in 2007^35–38. Although proof-of-concept examples remain limited, the Incompatible Insect Technique (IIT) and the Sterile Insect Technique (SIT) have already demonstrated their effectiveness in controlling Ae. albopictus populations.

One of the main advantages of IIT over SIT is that males are “ready to use”. In SIT programs, the rearing facility producing sterilized insects must be located near the release sites, or sterile males must be transported over long distances, in which case handling conditions (such as chilling and compaction) need to be optimized and transport minimized to prevent adverse effects on males³⁹. However, SIT offers the sustained advantage over IIT that accidentally released female are sterile. This limitation of IIT is particularly salient in the case of uni-directional CI, which to our knowledge has been thus far used in all long-term published IIT studies aiming at suppressing Ae. albopictus populations. IIT deployments based on bi-directional CI to control Ae. albopictus were either limited to a low number of released males (<5000 males/week) over a short period⁴⁰ or involved releases from a single point location⁴¹.

To overcome this limitation, we implemented a long term standalone IIT program for the control of Ae. albopictus using a transinfected line exhibiting bi-CI with the local population. We conducted this trial in “real-life” conditions, meaning through the monitoring of mosquito dynamics during suppression and subsequent adjustment of release conditions. The release of SEY-wPip males led to a 95% population suppression within only 12 weeks. This level of suppression was reached although the estimated male release ratio was low, 1.5:1 and 7:1 (SEY-wPip to wild males), 8 and 12 weeks after the first release, respectively. These results suggest that the mono-transinfected line used in this study allowed efficient suppression even when using low incompatible to wild male ratios. Such a performance is in keeping with previously published data showing that the introgression of a genetic background from the targeted population does limit fitness costs associated with Wolbachia transinfection^34,42–44. A recent IIT deployment using a triple-Wolbachia strain infection wAlbA, wAlbB, wPip also demonstrated a rapid suppression of an Ae. albopictus population in an urban area of Changsha (China) with immediate suppression of 49% after the 1st week of release and reaching 91% after 9 weeks¹⁸. However, when the same transinfected line was used in an IIT program in Guangzhou (China), and with comparable incompatible male release ratios, the suppression effect was limited to 65% 2 months after the first release. In this case, a combined IIT/SIT strategy was subsequently implemented, which allowed suppression levels exceeding 90%¹⁹.

After achieving a substantial level of population suppression (phase I), we choose to decrease the number of released males from ~80,000 to 50,000 per week (phase II). Indeed, it has been suggested to adapt the number of incompatible males released to the ever-decreasing wild population, thus reducing the risk of releasing Wolbachia-infected females while reducing costs⁴⁵. Although at this time the number of wild mosquitoes was low due to efficient suppression, the number of hatching eggs rapidly increased as exemplified by two consecutive measures following releases with low SEY-wPip to wild males ratio (1:1). This unexpected trend may be due to high rainfall observed at the end of phase I.

Finally, we increased the number of incompatible males to ~140,000 per week (phase III) for the last four releases, allowing restoring 80% suppression. Altogether, these data indicate that, despite a significant reduction of the wild population size, a minimal incompatible male ratio should be conserved for a long term population control.

Interestingly, we showed that the suppression effect was heterogeneous at the release site, potentially resulting from habitat heterogeneity, as reported in other IIT program¹⁸. We observed that ovitrap M01, located in the southwest part of the release site, displayed consistently lower suppression than surrounding ovitraps. We hypothesized that this difference resulted from favorable breeding conditions around this trap, neighboring a bamboo area (where cut bamboo stems provide larval sites unless they are refilled). This interesting pattern suggests that constant monitoring of the treated areas should be associated with local tuning of the number of released males, with increasing numbers being used in areas identified as resisting to mosquito suppression. The combination of high rainfall, during the last 2 weeks of releases, together with the presence of “resistant” areas, could explain why the number of hatched eggs became similar between the release and control sites only 7 weeks after the end of releases.

While presented data show the effectiveness of a standalone IIT for the control of an Ae. albopictus population in an insular context, we demonstrate that a significant population suppression can be achieved without population replacement. Indeed, although we detected wPip-positive individuals during and shortly after the release period (ranging from 0 to 6.3%), no incompatible individuals were found in our final screening conducted 2 months after the last release. We acknowledge that visual inspection of a large number of pupae performed in this study is time-consuming and can hardly be implemented at substantially larger scales. To overcome this limitation, it has been proposed to implement an individual automated verification at the adult stage, allowing to reduce drastically the female contamination rate^14,15,17. Classically, female pupae contamination after classic mechanical sex separation ranges from 0.5 to 3%^14,46. Here, we demonstrated that female contamination can be reduced to 0.02%, collecting pupae on 2 consecutive days for each production batch.

Lastly, this program was carried out on a remote island, representing one of the most challenging scenarios for the deployment of IIT as population replacement cannot be mitigated by the migration of wild mosquitoes surrounding the treated area. Presented data highlight another advantage of the deployment of a standalone IIT: the possibility of developing a mass production model based on a unique central facility for egg production associated with satellite units dedicated to the production of incompatible males. Such satellite units are relatively easy to set-up as there is little need for space and labor as equipment for blood-feeding or egg collection is unnecessary³³. As compared to the shipping of chilled adults from a central facility to satellite units, the shipping of eggs, facilitated by the resistance of Ae. albopictus eggs to desiccation, and their low weight ensures the production of high quality males while avoiding a quality control on males post-shipping. This model of mass production can also be developed in SIT programs under the strict condition that an irradiator is present at each satellite unit, which is clearly challenging but should be considered in the case of a mosquito extinction program.

Methods

Mosquito lines

We firstly constructed a transinfected line with a wild genetic background from the Seychelles. For this, we used a laboratory line transinfected with wPip-IV²², a Wolbachia strain naturally found in Culex pipiens⁴⁷. Females from this line were backcrossed with males from a wild line sampled using ovitraps on Ile Longue (Seychelles), located about 400 m from the release site. Backcrossing was performed using 15-day-old wild males (using the F₁ post sampling) with 4–5 day-old virgin females from the transinfected line for four consecutive generations, resulting in a line named SEY-wPip, expected to display 93.75% of the wild nuclear genetic background.

We used three other wild lines, named S-RUN, Le Port, and La Providence, to conduct laboratory experiments aimed at quantifying CI level induced by the SEY-wPip line and at measuring possible fitness costs following introgression. The S-Run and Le Port lines were originally derived from a natural population on Reunion Island^34,48, and were maintained for 50 and 10 generations, respectively, before experimentation. The La Providence line was sampled from the Providence neighborhood on Mahé island (the largest island in the Seychelles archipelago) and maintained for 7 generations before experiments.

Adults of all lines were maintained at a temperature of 27 ± 1 °C, a relative humidity of 70 ± 5% and a 12 : 12 h light:dark photoperiod in the insectary located at Saint-Denis (Reunion Island). Females were blood-fed using the Hemotek system (Hemotek Ltd, United Kingdom), bovine blood provided by the regional slaughterhouse (Saint-Pierre, Reunion Island), and supplemented with EDTA (0,1%).

CI expression of the transinfected line

To quantify CI penetrance of the SEY-wPip line following introgression, we performed reciprocal en masse crosses using S-RUN, Le Port and La Providence lines. Although en masse crosses can mask individual variability, especially in the case of moderate CI, complete or nearly complete CI were expected²², which led us to favor en masse rather than individual crosses in order to screen a larger number of mosquitoes. Crosses involved 2–5-day-old virgin females and males (100 from each sex) in 15 × 15 × 15 cm cages (Bugdorm, Taiwan). Three replicates were performed for each cross. Females were given a blood meal 48 h after mating, and eggs were collected 5 days later. After 7 days of drying, eggs were allowed to hatch for 24 h (in a jar containing 250 mL tap water supplemented with 50 mg of TetraMin (TETRA) and the number of hatched and unhatched eggs was counted. A hatch rate was calculated as follows: hatch rate = (number of hatched eggs/total number of eggs) × 100. To account for the embryonic mortality not related to CI, we used a corrected CI index (CI_corr)^49,50 calculated as follows: CI_corr = [(CI_obs − CCM)/(100 − CCM)] × 100, where CI_obs is the percentage of unhatched eggs observed in a given incompatible cross, and CCM is the mean mortality observed in the control crosses.

Life history traits of the transinfected line

Longevity and fecundity of the SEY-wPip line were compared to those of the S-RUN and Le Port lines, both showing high fitness under laboratory-controlled conditions^22,34. Larvae-rearing conditions were standardized between lines for all experiments. Specifically, eggs from each line were allowed to hatch for 24 h at 31°C in containers containing 250 ml of water supplemented with 50 mg TetraMin (TETRA). For each line, 2500 L1 were manually counted and transferred to a tray (53 × 325 × 65 cm, MORI 2 A) and fed with a controlled quantity of food (TetraMin (TETRA), day 1: 0.45 g, day 2: 1 g, day 3: 1.25 g, day 4: 1 g, day 5: 1 g, day 6: 0.75 g, day 7: 0.5 g) until pupal stage. Male and female pupae were initially separated using a pupae sex sorter (Wolbaki, WBK-P0001-V1 model), and then individually inspected under a binocular loupe.

For each line, longevity was measured by introducing 100 newly emerged males or females separately in 15 × 15 × 15 cm cages (Bugdorm, Taiwan) in which sugar meal (5%) was changed weekly. Three replicates were performed at different times for each line and sex, and longevity was determined by recording the number of dead mosquitoes for 45 days.

For each line, the number of laid eggs was measured by placing 200 male and 200 female pupae (one cage per line) in 30 × 30 × 30 cm cages (Bugdorm, Taiwan) and left for 3 days following emergence. A blood meal was then provided, and engorged females were placed in a separate cage. Five days later, 50 females were randomly selected and placed individually in a small plastic cup for egg laying. Cups with at least one laid egg were conserved for the analyses. After 7 days of drying, eggs were counted, allowed to hatch for 24 h and hatch rates were measured. Three replicates were performed at different times for each line.

Mating competitiveness of incompatible males in laboratory-controlled conditions

We evaluated the mating competitiveness of incompatible males by mixing virgin females from the S-RUN or Le Port line with increasing ratios of males from the SEY-wPip line. Pupae from both lines were allowed to emerge in separate 30 × 30 × 30 cm cages (Bugdorm, Taiwan), and adults were provided sugar meal (5%). Mating competitiveness was monitored using 2–3-day old virgin females and males. A hundred males were first placed inside cages, followed by the release of an equal number of females, and mating was allowed for 48 h. All males were then removed, a blood meal was provided to females and eggs were collected by oviposition en masse 5 days later. After 7 days of drying, eggs were allowed to hatch for 24 h, and the hatching rate was measured and used as a proxy of mating competitiveness. Five ratios of males were tested: 1 : 1 (50♂ S-RUN or Le Port : 50♂ SEY-wPip), 1 : 5 (17♂ S-RUN or Le Port : 85♂ SEY-wPip), 1 : 10 (9♂ S-RUN or Le Port: 90♂ SEY-wPip) and two control ratios, 1 : 0 (100♂ S-RUN or Le Port: 0♂ SEY-wPip) and 0 : 1 (0♂ S-RUN or Le Port: 100 SEY-wPip). Three replicates in different times were performed for each cross and ratio.

Description of study areas

Both study sites were located within the Sainte Anne Marine National Park in the Seychelles. The control site consists of a 7-hectare area in the Southern part of Saint Anne island, while the release site encompassed the entire 9-hectare area of Moyenne island (Fig. 1). These sites, separated by 1.2 km, presented comparable ecological conditions, with similar vegetation and geology, providing suitable climatic conditions for Ae. albopictus populations. The region is classified as having a tropical maritime climate, with year-round temperatures ranging from 22 °C to 33 °C. The rainy season extends from November to April, peaking in January and February, while the dry season lasts from May to October. Relative humidity is generally high, often exceeding 80%. Experiment approvals were provided by the Seychelles Bureau of Standards (SBS) and the National Biosecurity Agency (NBA), in accordance with local regulations on insect population release and monitoring.

Estimating Ae. albopictus natural population size at the release site

The natural population size of Ae. albopictus at the release site was estimated through two mark-release-recapture (MRR) experiments conducted prior releases. Adults used for MRR were obtained from eggs collected in the wild via ovitraps, which were deployed to monitor the temporal dynamics of Ae. albopictus population size at the release site (see below). Larvae were reared under laboratory-controlled conditions until pupal stage. Male and female pupae were initially separated using a pupae sex sorter (Wolbaki, WBK-P0001-V1 model), and all male pupae were individually inspected under a binocular loupe.

To mark adult males with fluorescent pigment, we constructed a self-marking unit adapted from a previously described prototype⁵¹. A total of 2500 male pupae were placed in a square petri dish (12.5 × 12.5 cm) with approximately 0.5 cm water, which was previously positioned inside a 30 × 30 × 30 cm cage (Bugdorm-1 model) (Bugdorm, Taiwan). A removable exit grid, consisting of wool yarn lines spaced approximately 0.5 cm apart and pre-coated with fluorescent pigment (Radiant colour), was then placed above the petri dish. Upon emergence, adult males were self-marked as they passed through the grid and were subsequently conserved in the cage with a sugar solution 5% until release. The self-marking unit was then removed from the cage, and the number of males that had not crossed the grid was counted. Just before the release, dead adults were removed from the cage to ensure an accurate count of the alive marked males being released.

Altogether, 4389 and 5682 marked males were released during the first and second MRR experiments, conducted in October and November 2022, respectively. For both MRRs, males were released at four equidistant spots, surrounded by 27 BG traps (Supplementary Fig. 4), corresponding to an area of ~4.5 ha. Just after release, each BG trap was activated for 4 days of capture, with a collection bag being replaced every 24 h. All captured individuals were frozen at −20 °C for 24 h, species and sex of each trapped adult were determined under a binocular loupe. All Ae. albopictus males were then examined under ultraviolet light to detect the presence of fluorescent dust. The population size was estimated as follows: N = n * (M/m), where N is the population size per day, n is the number of captured mosquitos, M the number of released marked mosquitoes, and m the number of recaptured marked mosquitos^52–54.

Mass Production and release of incompatible males

Mass production and subsequent release of SEY-wPip males involved five steps: (i) egg production at the CYROI insectary (Saint-Denis, Reunion island), (ii) transportation of eggs to Seychelles using a commercial airline company, (iii) larvae rearing and sex separation at insectary of the Ministry of Health of Seychelles (Grand Anse, Seychelles), (iv) transportation of male pupae by boat to the release site, and (v) release of adult males.

For egg production, we maintained adult cages (45 × 45 × 45 cm) (Bugdorm, Taiwan) continuously, each containing approximately 6000 female pupae and 2000 male pupae. Adults were fed with a 5% sugar solution. Females were blood-fed three times a week with bovine blood, and eggs were collected weekly for 3 consecutive weeks. Eggs were matured for 1 week before being used either for the production of adults used for egg production or sent to Seychelles for the production of incompatible males. The procedures for egg hatching, larvae rearing and sex separation were similar in both distant insectaries. Briefly, the eggs were brushed and transferred in 120 ml plastic containers with a hatching solution (tap water with TetraMin (TETRA) at 1 g/L) and left for 8 h. Approximately 3000 larvae were added to each tray (53 × 325 × 65 cm, MORI 2 A) and fed with a controlled quantity of food (TetraMin (TETRA), day 1: 0.5 g, day 2: 1 g, day 3: 1.25 g, day 4: 1.5 g, day 5: 0.75 g, day 6: 1 g, day 7: 0.75 g, day 8: 0.5 g) until pupal stage. Male and female pupae were separated using a pupae sex sorter (Wolbaki, WBK-P0001-V1 model).

During the release phase, after the initial mechanical sex separation, quality control was performed daily on all the necessary number of pupae (or nearly all, depending on the production phase; see below) collected pupae considered as male pupae, by measuring female contamination under a binocular loupe.

Using a volumetric template, daily male pupae production was divided into several lots of approximately 2250 male pupae, placed into 500 mL containers (Nalgene) with ~2 cm water depth. The following day, all pupae were transported by boat to the release site, and each container was transferred into a release tube (10-cm diameter × 35-cm height). The tubes were placed in trays containing ~2 cm water depth, and a source of 10% sugar solution was placed at the top. The following day, the release tube was removed from the water and adult males were released. All male pupae produced weekly were released over two consecutive days with male pupae collected on Monday released on Thursday, and those collected on Tuesday released on Friday, with the release evenly distributed across the 40 release spots, spaced approximately 40 m apart (Supplementary Fig. 5).

Incompatible males were released over a period of 6 months, divided into three phases, each distinguished by the number of released males. Phase I involved the release of approximately 80,000 males/week over 12 weeks. Phase II involved the release of around 50,000 males/week over 6 weeks. Finally, phase III involved the release of approximately 140,000 males/week over the last four weeks (Fig. 2D). For phases I and II, quality control of the male pupae production after mechanical sex separation was performed on all collected pupae, while in phase III, quality control was conducted on only two-thirds of the production.

Measure of the incompatible vs. wild male ratios and the mating competitiveness of transinfected males during the release period

The incompatible vs. wild male ratios were measured monthly throughout the entire release period (i.e. five measurements), from July 2023 to November 2023. For this, 9 BG-Sentinel traps (Biogents, Regensburg, Germany) were distributed across the release site, spaced approximately 90 m apart, resulting in one trap per hectare (Supplementary Fig. 6). BG traps were activated shortly after each release and for 24-h of capture. Captured mosquitoes were then morphologically identified under a binocular loupe to determine sex and species of each specimen. All Ae. albopictus adult males were individually placed in 1.5 ml Eppendorf tubes and stored at −20 °C before being shipped to Reunion Island for Wolbachia typing through PCR analysis as previously described⁵⁵. We also computed a competitiveness index (C)⁵⁶ to compare the performance of incompatible and wild males, which is defined as follows: C = (N/S)*[(Hc − Hi)/(Hi − Hs)], where N is the number of “compatible” males, S is the number of incompatible males, Hc is the mean hatch rate observed in the control site, Hi is the mean hatch rate observed in the release site, and Hs is the mean residual fertility calculated in laboratory control crosses (see above) involving incompatible males and wild females.

Monitoring population suppression

Temporal dynamics of the Ae. albopictus population size in both the control and release sites was monitored using ovitraps before, during, and after the release periods. Ovitraps consisted of small black 800-mL buckets containing 500 mL of tap water and one egg-laying paper (Sartorius^TM) fixed with a piece of plexiglas. From February 2022 to February 2024, 26 and 20 ovitraps were distributed throughout the release and control sites, respectively, with ovitraps placed approximately 50 m apart, resulting in approximately 3 ovitraps per hectare (Supplementary Fig. 7A, B). Ovitraps were retrieved and replaced weekly. Positive egg-laying papers, defined as those containing at least one egg, were preserved, and the number of eggs was recorded. Eggs were then stored at room temperature for 6 days, and then allowed to hatch for 24 h in a jar containing 250 mL tap water supplemented with 50 mg of TetraMin (TETRA). Hatch rate was then calculated as described previously³⁴.

To measure the suppression efficiency, we calculated the suppression percentage using the following equation, where Hc and Hr represent the average number of hatched eggs per trap at the control and release sites, respectively¹⁸:

$$SE\left(\%\right)=\frac{Hc-Hr}{Hc}x100$$

At the beginning of the ovitrap monitoring, we also deployed BG traps equipped with a CO₂ generator based on yeast fermentation (Biogents, Germany) to monitor adult female densities before, during and after the releases of incompatible males. Starting in February 2022, nine BG traps were installed in the release and seven in the control sites (one per hectare). These traps were operated for 3 consecutive days every 2 weeks, and the BG-CO₂ powder were replaced daily during each capture session. However due to the very low number of females collected and the lack of correlation with seasonal variation (Supplementary Fig. 8), this approach was discontinued, as it proved unsuitable for monitoring female densities.

Monitoring the risk of population replacement

The risk of population replacement at the release site, caused by accidental releases of transinfected females, was monitored from the beginning until 2 months after the last release (from 2 August 2023 to 20 February 2024). During this period, every 3 weeks, all larvae from the 26 ovitraps were conserved and reared until adult stage. Adults were then frozen for 1 h at −20 °C, individually transferred to 1.5 mL Eppendorf tube and shipped to Reunion island laboratory facilities for PCR screening. The dynamics of wPip-positive adults were monitored at eleven different time points, with a total of 1196 individuals checked for the presence or absence of wPip. Two indicators were used to monitor the risk: (i) the proportion of ovitraps containing wPip-positive adults and (ii) the proportion of wPip-positive individuals emerging from each positive ovitrap.

DNA extraction and wPip screening

Genomic DNA was extracted from adult mosquitoes using the cetyltrimethylammonium bromide (CTAB) method⁵⁷. The PCR reaction mixture consisted of 2 μl gDNA template, 8.5 μl nuclease free water, 1 μl of each primer (10 μM), and 12.5 μl of GoTaq^® G2 Hot Start Master Mixes (Promega corporation). We targeted an ankyrin domain protein by amplifying the ank2 gene using the following primers: F-CTTCTTCTGTGAGTGTACGT and R2-TCCATATCGATCTACTGCGT⁵⁸. PCR amplification was performed under the following conditions: 95 °C 5 min followed by 35 cycles of 94 °C 45 s, 53 °C 45 s, 72 °C 45 s, and 72 °C for 7 min. Amplified fragments were run in agarose gel (1.5%) electrophoresis.

Statistical analysis

Longevity data were analysed using a log-rank test. Fecundity data (count data) were analysed using a generalized linear mixed model (GLMM, quasi-Poisson family, log link) in which the different mosquito lines were included as a fixed effect and the repetitions (individual egg laying) as a random factor. We used a GLMM to analyse egg hatch rate (binary data, quasi-binomial family, logit link). The overdispersion of the data was checked using a R code proposed by Ben Bolker and others (https://bbolker.github.io/mixedmodels-misc/glmmFAQ.html). A F-test for the GLMM-quasi-binomial and quasi-poisson models was used to analyse deviance. Exact binomial tests were used to compare the observed and expected hatch rates in the mating competitiveness experiment (proportional data). Analyses were performed in R version 4.3.1⁵⁹, using lme4 package for all mixed models⁶⁰, MASS package⁶¹ for using the quasi-binomial and quasi-poisson families in the GLMMs, and survival package for longevity data analyses⁶². For all data, the significance level was set to α  =  0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

J.C. contributed to the study concept and design, conducted laboratory experiments and mass mosquito production, analyzed and interpreted the data, and wrote and prepared the manuscript and figures. B.G. contributed to the study design, analyzed and interpreted the data and was responsible for mass production of incompatible males and entomological surveys in Seychelles. M.D., D.S., A.J-B., and S.D. contributed to the mass production and release of incompatible males, as well as to entomological monitoring in Seychelles. M.A., H.D., H.P., K.S., and S.S. contributed to the entomological surveys. L.B. assisted in obtaining the necessary approvals to implement the field trial in Seychelles. J.F. and G.R. contributed to the logistical and technical administrative aspects. Ms S.S., Q.L., and J.E. contributed to laboratory experiments, mass egg production in La Réunion, and molecular biology analyses. P.M. contributed to the initiation of the project in Seychelles. P.T. also contributed to the project’s initiation and to the preparation of the manuscript.

Peer review

Peer review information

Communications Biology thanks Zhiyong Xi, Nigel Beebe and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Hannes Schuler and Mengtan Xing. A peer review file is available.

Data availability

All relevant raw data associated with the manuscript are located in the Supplementary data file.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09269-0.