Description of study areas
The general characterization of the study areas was described previously [21]. The Shazai and its control site, Xiaohu Island, are in Nansha District, Guangzhou, China. The Dadaosha Island and its control site, Guanlong Island, are in Panyu District, Guangzhou, China. During this study, the residential environment had been changed in Shazai Island (release site 1) and Xiaohu Island (control site 1). Specifically, in the northern part of Shazai Island (previously described as zone 20–22), all the residential buildings were torn down in early 2018. Therefore, these areas were excluded from this study. In early 2019, residents of Xiaohu Island started to move out gradually due to the impact of the nearby chemical plant explosion in 2018. In June 2019, residential buildings on Xiaohu Island began to be torn down, and several main roads were blocked in September, so we had to stop mosquito surveillance in early October. The previous release site 2 (Dadaosha Island) and its control sites (Dadaosha Island, control site 2a; Guanlong Island, control site 2b) had minor residential environment changes during this study. But one small area, between a road and zone 8 in release site 2, had changed significantly. The western part of this area used to have poor sanitation, and had a large number of mosquito breeding sites; the eastern part of this area was a buffer zone (by releasing HC males) during mosquito release, to prevent the mosquitoes from immigrating from its western part. But at the end of 2017, the western part was transformed into an outdoor activity place for villagers, thereby eliminating those mosquito breeding sites. Besides, the intensity of local mosquito control activities increased substantially in 2019, as those villages committed to building “beautiful countryside”, which was promoted by the municipal government.
Mosquito colonies
The wild-type A. albopictus Guangzhou Line was newly established by collecting about 200 A. albopictus larvae from each of 10 scattered locations in Guangzhou City and then pooling them together and rearing them into adults. The first generation and descendants of these mosquitoes were named as GUA. The A. albopictus HC line was previously described and mass-reared at Guangzhou Wolbaki Biotech Co., Ltd (hereafter referred to as Wolbaki), which is infected by wPip, wAlbA, and wAlbB [21]. To better represent its quality in the field, different from the previous studies [38], we outcrossed the HC line with the above GUA, which was reared in the laboratory for (< 2) generations after field collection, for four generations, and then self-crossed for 1 generation, designated as oHC. The first generation of the oHC line was used for further experiments.
Two wild-type A. albopictus lines used for the female mating preference experiment (see below) were derived either from Shazai Island or Xiaohu Island and herein designated as szGUA and xhGUA, respectively. To establish these colonies, at the end of 2017 (3 years after mosquito releases), pupae or larvae were collected from 30 to 40 scattered ovitraps (Zhikehongrun Environmental Protection Technology Co., Ltd, Zhengzhou, China) and then pooled together, and reared to adults (G0). Offspring of G0 (G1) were used in the experiment, and they were reared under standard conditions and standard protocol as described before [38]. Briefly, larvae and adults were maintained in a standard mosquito-rearing room at 27 ± 2 (^{circ })C and 75 ± 5% relative humidity, with a 12∶12-hour light: dark photoperiod. Eggs were hatched with hatching solution [1% (w/v) yeast powder in dHO], and the newly hatched larvae were reared in trays (38 cm (times) 25 cm (times) 5 cm) at a density of 300 larvae per 1.0 L tap water per tray. Dry bovine liver powder (weighted and suspended in 5–10 ml of water before use) was supplied to the 1 st and 2nd instar larvae at 0.5 mg per larva per day and 1.0 mg per larva per day for the 3rd and 4th instar larvae. Adult mosquitoes were provided with fresh 10% sucrose solution and allowed to feed on a mouse to lay eggs.
Mosquito survivability under stressed conditions
To study the survival rate of larvae under malnutrition, 100 newly hatched 1 st instar larvae of GUA, HC, and oHC were placed into plastic boxes (17 cm (times) 11.8 cm (times) 5.5 cm) with 500 ml of water. A bovine liver powder solution (prepared by suspending 5 g of dry bovine Liver powder in 100 ml of tap water) was used as larval food. To create a malnutrition condition, 200 ml of Liver powder solution was provided every 3 days (on average, 0.33 mg per larva per day), until all larvae developed into adults or died. There were three replicates for each mosquito line, and the food supplement in the control group was as described above. To study the larvae survival under high temperatures, first instar larvae of GUA, HC, and oHC (50 larvae in plastic boxes, 17 cm (times)11.8 cm (times) 5.5 cm, with 300 ml tap water) were reared in a climate chamber with cyclical temperatures: 39 (^{circ })C during the 12-h Light period and 29 (^ {circ })C during the 12-h dark period. Larvae were reared under such conditions for 5 days before being transferred to standard rearing conditions as described above. Then count the number of larvae that successfully developed into pupae until all larvae died or pupated. To study female adult longevity under stressed conditions, 30 newly emerged female adults of GUA, HC, and oHC were placed in cages (30 cm (times) 30 cm (times) 30 cm), and a bottle of tap water without sugar was provided. Dead adults were counted every 24 h until all mosquitoes died.
Diapause induction
The diapause induction procedure was as previously reported with modifications [39]. Briefly, for diapause induction, mosquitoes at all developmental stages were kept at 21 (^{circ })C with a 6∶18 light: dark cycle, and mosquitoes reared under 21 (^{circ })C with an 18∶6 light: dark cycle served as the control group. Approximately 300 first-instar larvae of GUA, HC, or oHC were reared in a plastic tray (as described above), and dry bovine liver powder was provided as larvae food on demand. Pupae were collected and moved to a cage (30 cm (times) 30 cm (times) 30 cm) provided with 10% sucrose solution. Two weeks later, blood-fed those mosquitoes with mice, and then 3 days later, an oviposition cup was provided to collect eggs. These eggs were kept wet under the same conditions for 10 days to allow for embryo maturation. To judge the egg hatch rate and diapause rate, the eggs were hatched under their corresponding conditions continuously for 24 h, then the egg papers were dried for 1 week, and then the eggs were hatched for another 24 h. Egg hatch rate was calculated by dividing the de-capped eggs by the total eggs immersed in water. Since during winter, the eggs of A. albopictus that can be hatched are biologically dead, we focused on the unhatched eggs, which could be either in a diapause state or undeveloped. To differentiate between these two types of unhatched eggs, a bleaching process was performed using a 5% sodium hypochlorite solution (Zhongding Biotech, Tianjin, China). The determination of dead or diapause status was carried out according to a previously described method [40]. Three replicates were performed, and each replicate involved randomly collecting and bleaching around 30 eggs in each group. The rate of unhatched eggs was calculated by dividing the number of unhatched eggs by the total number of eggs. The diapause rate was calculated by the number of diapause eggs divided by the total number of unhatched eggs. The proportion of diapause eggs was calculated by multiplying the rate of unhatched eggs by the diapause rate.
Mosquito mass production, transportation, and field release
Male mosquito pupae of HC line were mass-produced in Wolbaki following the standard protocol as described previously [21]. Then about 40,000 male pupae were placed in an eclosion cage (120 cm (times) 30 cm (times) 30 cm) with cotton balls soaked with 10% sucrose water placed on the top of it. After pupae emergence, the containers for pupae were taken out, and then those cages were moved to a chilling room with a temperature of (8 pm 2) (^circ)C, to knock down the adult mosquitoes so that they can be packaged. About 40,000 mosquitoes were packaged in one plastic box with a cap (27 cm (times) 19 cm (times) 3 cm). Packaged mosquitoes were then stored in a portable refrigerator (Alpicool, Foshan, China, 75 L in volume) at a temperature of (10 pm 2) (^circ)C for transportation, and released immediately after they arrived at the field site (Shazai Island). During the whole process, mosquitoes were kept at low temperatures for about 2.5 h. The handling process has been proven to have a minor effect on males’ fitness [41]. To release the chilling-paralyzed mosquitoes, packages were taken out of the refrigerator, caps were removed, and mosquitoes gradually woke up by sensing the rising temperature and flew away. Mosquitoes were released while walking, but preferably released near vegetation. In 2018 and 2019, HC males were released twice a week (on Wednesday and Saturday) in Shazai Island. On average, about 400,000 and 370,000 males were released weekly in 2018 and 2019, respectively. During this study, no mosquito was released on Dadaosha Island, and at the end of 2019, release activity was stopped on Shazai Island.
Monitoring mosquito populations in the field
The population of A. albopictus, as well as C. quinquefasciatus, was monitored from March to October between 2016 and 2019 by BG-Sentinel Traps (Biogents, Regensburg, Germany) in Shazai and Xiaohu Island, but in Dadaosha Island and its control sites, the mosquito population was monitored by ovitraps only. The operation procedures of BG traps and ovitraps were the same as described before [21]. Briefly, BG traps continuously run for 24 h every Tuesday. The captured adults, both Aedes and Culex, were sent back to the lab for species identification and quantification. Ovitraps were placed in the field for a week (usually from Monday to Monday), and those with eggs present were collected and taken back to the lab to count the hatched larvae. In 2018, in the release sites and control sites, the distributions and locations of BG traps and ovitraps were the same as previously described (2 BG traps and 5 ovitraps in each zone). But in 2019, in order to reduce manpower, the distribution density of BG traps and ovitraps was decreased. Specifically, 20 BG traps were distributed in Shazai Island (generally 1 BG trap in each zone) and 15 BG traps in Xiaohu Island; 20 ovitraps were placed on Dadaosha Island (2 to 3 ovitraps in each zone) and its control sites.
Release ratio calculation
To reduce the cost, the release ratio (HC male to wild-type male) was calculated by the numbers of mosquitoes captured by BG traps in both the control site and release site, rather than PCR assay used from 2018 to 2019. Firstly, calculate the ratio r of male to female according to the numbers of male and female A. albopictus in the BG traps in the control site. Let n and m denote the numbers of males and females in the release site, respectively. Then the release ratio R is given by (R=frac{n}{rtimes m}-1). If no female was caught in either the release site or the control site, the denominator becomes 0. In such cases, this particular data was disregarded.
Female mating preference assay
A. albopictus were collected from the field in Shazai Island 3 years after release and in Xiaohu Island, which experienced no release. Virgin female adults from the first-generation (F1) offspring of the field-collected mosquitoes were used in the dual-choice experiment to mate with GUA males (F1) from Shazai and HC males, following a protocol as previously described with modifications [42]. Specifically, 2–3 day-old GUA and HC male mosquitoes were first transferred into the 30 cm (times) 30 cm (times) 30 cm cage in three different ratios, 1∶1, 1∶5 and 1∶10, and on the 2nd day, females were introduced into the cages (Table S1). All cages maintained a similar mosquito density, with 350–360 mosquitoes per cage, to avoid a density-dependent effect on mating. Mosquitoes were allowed to mate for 24 h, then male mosquitoes were removed by an aspirator. Three days later, females were blood-fed with mice. Mosquitoes that took blood were collected and reared individually in 50 ml conical centrifuge tubes supplied with wet filter paper at the bottom for egg-laying. Ten days later, those eggs were hatched by adding hatching solution into each tube and hatched for 48 h. As validated in the control crosses (Table S1), GUA females mated with HC males do not produce viable progeny, while those mated with GUA males produce mostly viable eggs. Therefore, we used egg viability as a proxy to determine whether GUA females mated with GUA or HC males. The female mating choice indices (FMCI) towards HC males or GUA males were calculated as: FMCI to HC males = ((N / R) / (Y + N / R)); FMCI to GUA males = (Y / (Y + N / R)), in which N is the number of GUA female individuals producing unviable eggs, Y is the number of GUA female individuals producing viable eggs, and R is the release ratio of GUA male vs HC male.
Based on the hypothesis that females with a preference for mating might be more resistant (but not necessarily reject) mating with unqualified males, it is expected that males would take longer to successfully engage females in stable mating pairs. We performed another experiment to assess the strength of female mating preference. We put 5 virgin males (1–2 days old, of the same line) in a 30 cm (times) 30 cm (times) 30 cm cage and supplied 10% sucrose water solution. Then, 24 h later, one virgin female mosquito (1–2 days old) was released into the cage, and immediately started timing to record how much time a male would take to seize the female to form a stable mating pair (the mating pair will rest on the cage surface). This experiment was repeated 10 times for each mating group, and here we performed three mating groups: szGUA female (times) HC male, szGUA female (times) xhGUA male, and xhGUA female (times) HC male.
Monitoring wPip infection frequency in the field
During this field trial, wPip infection frequencies were monitored in larvae collected by ovitraps or female adults collected by the human landing method. From 2018 to 2020, about 95 ovitraps were distributed on Shazai Island, and the positions of those ovitraps were the same as previously located [21]. Egg-positive ovitraps were sent back to the lab for wPip detection weekly. Because the release activity was stopped at the end of 2017 on Dadaosha Island, so we did not work on the wPip detection of the larvae in the ovitraps. Starting in July 2019, larva samples from ovitraps on Dadaosha Island were collected and analyzed via PCR assays to determine whether population replacement had occurred. Furthermore, female adults were collected on Shazai Island, in 3 independent human-landing catch activities in 2020, to monitor the wPip infection. Selecting locations for conducting human landings was the same as described previously [21]. In brief, we selected collection sites in close proximity to houses, often in shaded areas, where mosquitoes were readily accessible. Mosquito collection was conducted between 9:00 and 11:00 or between 16:00 and 18:00 using a mosquito aspirator. At each location, a staff remained present for 15–30 min to capture female mosquitoes. The captured females were taken to the lab for PCR assay. The PCR assay of wPip infection in larvae and adults was the same as described previously [21].
Statistical analysis
All data were statistically analyzed by SPSS 13.0 (IBM, Chicago, USA) and GraphPad Prism 6.0 software (GraphPad Software, San Diego, USA). The significance of mosquito population size between release sites and control sites, and the release ratio between different years, was determined using the two-tailed Mann Whitney U test. Indices of female mating choice to different mosquito males were compared by two-tailed t-test. Mating pair formation time between groups, the mosquito population size across different years, and the mosquito population recovery level across different zones were compared by one-way ANOVA. The significance of larval survival rate, egg hatch rate, diapause rate, and proportion of diapause eggs of different mosquito lines was analyzed by one-way ANOVA followed by Tukey’s multiple comparisons test. The log-rank (Mantel-Cox) test was used to compare mosquito longevity. Pearson correlation test was used to compare the C. quinquefasciatus population dynamics. The temporal distribution of wPip-positive larvae pools was compared by Fisher’s exact test. The difference in yearly egg suppression efficiency in 2017 and 2018 across different zones was analyzed by a paired t-test. When the P-value is less than or equal to 0.05, we consider that there was a significant difference between the compared data.