1 Introduction

The yellow fever mosquito, Aedes aegypti (Linnaeus) (Diptera: Culicidae) has played a major role as both a nuisance and vector species in the United States (US) for centuries. While the first description of Ae. aegypti in the continental US was published in 1828 [1], it has likely been present since the seventeenth century as evidenced by intermittent outbreaks of yellow fever and dengue throughout the eastern US [2]. While yellow fever transmission has ceased in the US, local transmission of dengue [3,4,5], chikungunya [6], and Zika virus [7, 8] have occurred in recent decades. In addition to the significant public health threat Ae. aegypti poses, they are also a noteworthy nuisance species as well. Aedes aegypti larvae inhabit artificial and natural containers around human habitation, and the adults are crepuscular feeders [9, 10] with a preference for human bloodmeals [11, 12]. These characteristics also enhance their importance in pathogen transmission because they are competent arboviral vectors with close associations to humans. Another container mosquito, Aedes albopictus (Skuse), is also widely distributed in Florida [13] and will often oviposit in the same containers as Ae. aegypti. Both mosquito species are considered domestic and are part of operational nuisance mosquito control in Florida counties [14].

Using insecticides to control adult populations of mosquitoes, also known as adulticiding, is a critical tool for public health vector control. Adulticiding is considered essential during mosquito-borne disease outbreaks to interrupt transmission by killing infected adult mosquitoes [8, 15,16,17]. However, the efficacy of these missions is largely dependent on the insecticide susceptibility status of the target population.

Mosquito insecticide resistance can be classified by several mechanisms including behavioral, penetration, metabolic, and target-site resistance [18,19,20]. Behavioral resistance is characterized by avoidance of a toxin [19] while penetration resistance is characterized by reduced penetration of an insecticide, often due to thickening or altered composition of the insect cuticle [20]. More well characterized for Ae. aegypti are target-site and metabolic resistance mechanisms. Target-site resistance results from a confirmational change in the insecticide target-site, often as a result of a mutation [18]. Knockdown resistance (kdr) is a type of target-site resistance associated with the sodium channel and has been documented in Ae. aegypti in Florida [21] and elsewhere [22,23,24]. Metabolic resistance is the result of upregulation of metabolic genes that detoxify xenobiotics, such as insecticides, and this has been implicated as a significant mechanism in several Ae. aegypti populations from Florida [25].

Adulticides primarily used for public health vector control in the US belong to the pyrethroid and organophosphate chemical classes. Resistance to both insecticide groups has been detected in the US in several mosquito species [26,27,28,29], including Ae. aegypti [21, 30, 31]. In Ae. aegypti, pyrethroid resistance is more prevalent than organophosphate resistance, and is ubiquitous across the tested populations from Florida [21, 31]. This trend is, in part, driven by the heavy use of pyrethroids in vector control since the 1960s [14, 32] and prior to that, the use of DDT as cross resistance has been documented between pyrethroids and DDT [33, 34].

With only two chemical classes available for adulticiding, failure of one chemical class to elicit mortality in mosquitoes makes rotation near impossible. To regain the ability to use pyrethroids effectively against Ae. aegypti, alternative control strategies must be implemented. Resistance to insecticides has often been associated with fitness costs [35, 36]. Therefore, removal of insecticide pressure, particularly from pyrethroids, may influence the resistance status of the mosquito population. This has been investigated previously in laboratory studies at both the phenotypic and genotypic level [36,37,38]. Both Chang et al. [37] and Brito et al. [36] observed a return to near-susceptible levels within 15 generations after relaxation or removal of insecticide exposure. Similarly, Grossman et al. [38] reported that a highly resistant field-derived strain of Ae. aegypti displayed a significant reversion towards susceptibility after 10 generations. These laboratory studies confirm that there is potential for susceptibility to be regained in pyrethroid-resistant Ae. aegypti populations.

While these results are promising, field (operational) attempts to return resistant populations to a susceptible state will be influenced by ecological factors as well as other sources of insecticides. For example, use of household insecticides can increase pyrethroid resistance levels in Ae. aegypti [39]. A survey conducted in Mexico found that aerosolized sprays were almost exclusively pyrethroid-based products and applied by most homeowners one to three times a day [39]. When exposed to these products, the frequency of kdr mutations increased in these populations. Therefore, it is plausible that even in the absence of an ultra-low-volume treatment with a pyrethroid, there will likely still exist selection pressure for pyrethroid resistance from other sources.

To date, no field studies have been conducted on the reversion of resistant Ae. aegypti back to an insecticide susceptible state under different treatment regimens. Here, we characterized the genotypic and phenotypic responses of 3 Ae. aegypti populations exposed to different operational vector control regimens from 2016 to 2019. The results from this study provide a promising foundation for future research and for the prospects of pyrethroid use for the control of Ae. aegypti.

In the next section, the materials and methods are described (Sect. 2). In Sect. 3, the results of the study are presented with associated figures. The discussion (Sect. 4) places the findings in context of other studies and the conclusions (Sect. 5) summarize the main findings and reiterate the importance of this research.

2 Materials and methods

2.1 Field sites and sampling

Four field locations were identified in Pasco County, Florida to represent four different operational treatment regimens. Treatment sites fell into one of the following treatment regimens: (1) adulticide with pyrethroid only (PYR site), (2) adulticide with organophosphate only (OP site), (3) larvicide only (LARV site), or (4) no treatment (NT site) (Table 1, Fig. 1). A 500–800 m buffer was created around each centroid to create the treatment area. Pasco County Mosquito Control District (PCMCD) uses spray zones to delineate the boundaries of treatment missions. In a routine operational framework, treatment within these zones would have been dictated by mosquito species abundance (through surveillance), resident complaints, disease transmission risk, and weather events. Based on this information, mosquito control would choose the appropriate application method, product, and treatment zones. Prior to this study, the selected treatment areas would have been part of this routine operational framework. At the time this study was conducted, there had not been any local transmission of viruses transmitted by container mosquitoes (i.e. Zika, dengue, or chikungunya viruses). Each treatment area for this study was composed, in part, by 3 spray zones. Any larvicide or adulticide mission conducted in any of the 3 zones was recorded and considered as a treatment to the above listed treatment areas, even though only part of the treatment area may have been treated. Treatment regimens were implemented beginning on January 1, 2017 and remained in place until July 31, 2019. Ovicups were deployed in September 2016 in each site to collect container mosquito eggs. The resulting populations were used for baseline insecticide susceptibility assays. Egg collections were made up of at least a single collection during each mosquito season between April and September. Treatment regimens were adhered to by PCMCD as closely as possible. However, nuisance levels and public health concerns resulted in treatment outside these criteria and those instances were recorded. The final egg collections were conducted on July 31, 2019.

Table 1 The central point, city, and site description of each treatment area in Pasco County, Florida
Full size table
Fig. 1
figure 1

Approximate locations of treatment sites in Pasco County, Florida. Map created using Maptive software

Full size image

Egg collections were made using several ovicups throughout each of the treatment areas. Ovicups were constructed using a 16 oz plastic cup and attaching seed germination paper to the inside of the cup using binder clips [40]. During the mosquito season, ovicups were deployed throughout the treatment areas and filled with water to encourage oviposition by container mosquitoes. The germination paper in the ovicups was collected and replaced every 5–7 days for a period of 5 weeks and ovicups were refilled with water as needed. Eggs collected throughout each 5-week period were considered to be representative of the corresponding treatment area at that time point. Pooling of eggs collected from multiple collection dates has been utilized in resistance studies previously [21] and studies conducted in the Florida Keys and Brazil suggest collections made over a 5-week period should be genetically similar [41, 42].

2.2 Laboratory handling and rearing

Rearing protocols described in [40] were followed when handling field collected samples. Briefly, the number of eggs on the field collected papers were assessed for viability and quantified to determine the number of viable eggs on each paper. Eggs from field collected papers (F0) were then hatched in rearing trays containing 2 L of water at a density of ~ 250 viable eggs per tray. Larval diet consisting of equal parts by weight of lactalbumin and brewer’s yeast was added to the trays at the hatching stage (~ 0.2 g). During the larval rearing process, additional larval diet was added to trays ad libitum. Upon pupation, mosquitoes were transferred to a ‘mosquito breeder’ (Bioquip, Rancho Dominguez, CA, U.S.A) and were sight identified once the adult emerged. Aedes albopictus were also found at these sites but were not used in this study. Aedes aegypti adults (both male and female) were placed in a cage (30.5  × 30.5  × 30.5 cm) (Bioquip) and provided a 10% sucrose solution.

A bloodmeal from a live chicken was offered to 3–5-day old mosquitoes for a period of 45 min (IACUC Protocol # 201,807,682) to amplify mosquito populations. Eggs were collected 2–3 days following the bloodmeal by placing moist germination paper inside the rearing cage. When necessary, the process of hatching, rearing, and amplifying populations was also conducted with F1 populations to obtain a greater number of individuals for the Centers for Disease Control and Prevention (CDC) bottle bioassay [43]. However, populations beyond second generation were not utilized for any genotypic or phenotypic resistance assays.

2.3 kdr genotyping assay

A melt curve analysis was utilized to assess genotype frequency in field populations of Ae. aegypti. Mutation of the 1016 allele from valine to isoleucine (V1016I) and of the 1534 allele from phenylalanine to cysteine (F1534C) were targeted [44, 45]. Melt curve analysis was conducted using methods previously described by [21]. A pyrethroid-resistant Puerto Rico population of Ae. aegypti [46] was used as a positive control and a pyrethroid-susceptible Orlando population of Ae. aegypti [47] was used as a negative control. An artificial heterozygote control was created by adding a single individual from both the Orlando and Puerto Rico populations to a single well.

Genotyping was performed on mosquitoes from each treatment area. These mosquitoes were obtained directly from the rearing cage and were never used in bottle bioassays. Field samples and controls were homogenized in a 96-deep well plate that contained 200 μl of nuclease free water and a glass grinding bead. Homogenization was done for 60 s at 30 wave cycles per second. The master mix was prepared using SYBR Select Master Mix, nuclease free water, and either 1016 or 1534 primers. A mixture of 2 μl of homogenized sample and 8 μl of master mix were subjected to the following cycling conditions: 3 min at 95 °C and 40 cycles at 95 °C for 3 s and 60 °C for 15 s. Fluorescence data was collected continuously from 60 °C to 95 °C.

Melting temperature peaks (Tm) were used to determine genotype. The Tm indicates the presence of the certain alleles for the 1016 and 1534 single nucleotide polymorphisms and the Tm previously described by Estep et al. [21] was used. For the 1016 mutation, amplicon coding for valine has a Tm of 86 ± 0.3 °C and amplicon coding for the resistant isoleucine mutation has a Tm of 77.3 ± 0.3 °C. For the 1534 mutation, phenylalanine has an amplicon Tm of 79.8 ± 0.3 °C and the resistant cysteine mutation has an amplicon Tm of 84.7 ± 0.3 °C. Samples with a single Tm were classified as homozygous for the associated amplicon coding while samples with multiple Tm peaks were classified as heterozygotes.

2.4 Phenotypic resistance assay

The CDC bottle bioassay [43, 48] was used to assess the phenotypic resistance of field populations of Ae. aegypti. Technical grade active ingredients (AI) (ChemService, West Chester, PA) were diluted using acetone to create a stock solution. Up to 6 pyrethroid and 3 organophosphate AIs were used in the CDC bottle bioassay. The pyrethroid active ingredients tested were cypermethrin, deltamethrin, etofenprox, lambda-cyhalothrin, permethrin, and sumithrin. The organophosphate active ingredients tested were chlorpyrifos, malathion, and naled. For each assay, four 250-ml glass bottles (DWK Life Sciences, Millville, NJ) were treated with a diagnostic dose (Table 2) of stock solution. The interior of each bottle was coated by tilting the bottle to either side and rotating until all interior surfaces of the bottle had been coated. Treated bottles were then uncapped and rolled on a table for 2–3 min or until all acetone had evaporated. One control bottle was used for each assay and was treated similarly to test bottles but was treated with acetone only.

Table 2 Chemical class, diagnostic dose, and diagnostic time for Ae. aegypti to 9 active ingredients used in the bottle bioassay
Full size table

Diagnostic times were calibrated by conducting the CDC bottle bioassay with the ORL 1952 insecticide susceptible population of Ae. aegypti [47] (Table 2). The diagnostic time was the time point at which 100% mortality was achieved in the susceptible population. Fifteen to 25 unfed 3- to 5- day old mosquitoes were introduced to the four treated bottles and the control bottle. Mortality was recorded at 0, 5, 10, 15, 30, 45, 60, 75, 90, 105, and 120-min. Mosquitoes were counted as dead if they could no longer stand or fly. CDC definitions were used to categorize field populations as susceptible (> 97% mortality at diagnostic time), developing resistance (90–96% mortality), or resistant (< 90% mortality) [43]. At the conclusion of the 2-h bottle bioassay, mosquitoes were transferred to clean holding cages to allow for a 24-h mortality reading, described previously [40]. Holding cages were covered with a fine mesh and mosquitoes were provided a 10% sucrose-soaked cotton ball until the 24-h mortality reading.

2.5 Data analysis

Genotype frequency over time was analyzed using a Chi-square analysis for the OP and LARV site separately using R statistical software [49]. Genotype frequency from the NT was not analyzed because data was only available from 2017. Confidence intervals (95%) were constructed for the response of Ae. aegypti to organophosphate and pyrethroid active ingredients.

3 Results

3.1 Operational mosquito control treatments

At the PYR site, several unplanned, but necessary adulticide missions with organophosphate-based products took place within the first 12 months of the study. Therefore, results from this site were discarded, and no future egg collections were made in that treatment area.

The OP site received 56 ground adulticide treatments of Fyfanon® (malathion) and one aerial adulticide treatment with Dibrom® (naled) during the study period. A total of four aerial larvicide applications were conducted within the OP site, but these were confined to lakes in the treatment area and likely did not have an impact on container mosquito populations. Two of these aerial missions used Metalarv® (s-methoprene) and two used Bacillus thuringiensis israelensis (Bti). Two residual applications were made using Envion (synergized permethrin) to two different individual private residences during the study period due to the lack of residual treatments available for this application in other chemical classes.

The LARV site received 26 aerial larvicide applications that treated approximately 5–25% of the treatment area on each application. Products used were Metalarv® (s-methoprene) or Bacillus thuringiensis israelensis (Bti) and also included one release of Gambusia affinis, a larvivorous fish, into a neglected/ non-chlorinated pool; 6 truck-based ground applications of Altosid® or MetaLarv® (s-methoprene); 7 ground applications of BVA oil (larvicide oil); and 42 ground applications of Abate® (temephos). A deviation occurred in this treatment area during the study period when a residual application of Envion (synergized permethrin) to a single property was conducted due to the lack of alternative residual treatment options.

The NT site received 26 aerial and 31 ground larvicide applications, but similar to the OP site, these treatments were targeted to lakes within the treatment area and likely had little effect on container mosquitoes. A single ground adulticide mission was conducted in the treatment area using Kontrol 30–30 (synergized permethrin) during the study period.

3.2 kdr genotyping assay

There were 9 possible genotype combinations for V1016I and F1534C: IICC, (homozygous resistant for both 1016 and 1534), IIFC, IIFF, VICC, VIFC, VIFF, VVCC, VVFC, and VVFF (homozygous susceptible for both 1016 and 1534). A total of 330 mosquitoes were used in the kdr assay and during the study period, genotypes observed in our samples included: IICC, VICC, VIFC, VVCC, and VVFC (Fig. 2) and the frequency at which each genotype was detected within the OP site and LARV site changed over time. In the NT site, Ae. aegypti were only obtained in 2017 for genotypic testing due to low numbers of eggs collected in ovicups. Genotype analysis on mosquitoes from the 2017 NT site collection (N = 38) revealed a majority (94.7%) had the IICC (homozygous resistant) genotype and 5.3% had the VICC genotype.

Fig. 2
figure 2

Genotype distribution for V1016I and F1534C over time at 3 different field sites in Pasco County, FL. Darker colors indicate a more resistant genotype while lighter colors indicate a less resistant genotype. IICC = homozygous resistant for both 1016 and 1534 alleles; VICC = heterozygous for 1016 allele and homozygous resistant for 1534 allele; VIFC = heterozygous for both 1016 and 1534 allele; VVCC = homozygous susceptible for 1016 allele and homozygous resistant for 1534 allele; VVFC = homozygous susceptible for 1016 allele and heterozygous for 1534 allele

Full size image

In 2016, the majority of Ae. aegypti individuals collected from the OP site exhibited the IICC genotype, which is homozygous resistant for both the 1016 and 1534 alleles. Of the individuals sampled from the 2016 OP site (N = 41), 85.4% were the IICC genotype and 14.6% were VICC, which is heterozygous for the 1016 allele and homozygous resistant for the 1534 allele (Fig. 2). The prevalence of genotypes changed significantly over the study period (df = 8, χ2 = 44.728, P < 0.0001). When the same site was sampled in 2017, the prevalence of the IICC genotype among sampled individuals (N = 42) had decreased to 30.9% while the prevalence of the VICC genotype increased to 45.2% classified. An additional 3 genotypes were detected at the OP site in 2017 that were not detected in 2016. The VIFC genotype (heterozygous for both alleles) made up 9.5% of sampled individuals, VVCC (homozygous susceptible for 1016 and homozygous resistant for 1534) made up 4.8%, and 9.5% were VVFC (homozygous susceptible for 1016 and heterozygous for 1534). In 2018, the VIFC and VVFC were not detected and the prevalence of the IICC genotype increased to 40.5% among the sampled individuals (N = 42). The VICC genotype decreased to 42.9% VICC while the VVCC genotype increased to 16.7% of the population.

During the study period, 3 genotypes were detected from the LARV site: IICC (homozygous resistant for both alleles), VICC (heterozygous for 1016 allele and homozygous resistant for 1534 allele), and VVCC (homozygous susceptible for 1016 allele and homozygous resistant for 1534 allele). Compared to the OP site, the sampled population from the LARV site (N = 83) had a lower percentage of individuals with the IICC genotype (38.6%) and a higher percentage with the VICC (60.2%) and VVCC (1.2%) genotype in 2016 (Fig. 2). The prevalence of genotypes changed significantly over the study period (df = 4, χ2 = 33.854, P < 0.0001). In 2017, the prevalence of the IICC genotype (38.1%) was similar to 2016), but the frequency of the VICC genotype decreased to 45.2% and the VVCC genotype increased to 16.7% (N = 42). In 2018, only two genotypes were detected. The percentage of sampled individuals (N = 42) with the IICC genotype increased to 76.2% and the VICC genotype decreased to 23.8%.

3.3 Phenotypic resistance assay

A total of 74 bottle bioassays were completed with an average of 22 mosquitoes per bottle (~ 8,100 mosquitoes used in assay). The response of Ae. aegypti to the different active ingredients listed in Table 2 were considered collectively based on chemical class. Therefore, the percent mortality at the diagnostic time, the 2-h, and the 24-h readings were an average of the response of all pyrethroid AIs or all organophosphate AIs. Collections of Ae. aegypti were made in 2019 only for the NT site, but at least one population of Ae. aegypti was collected for both the OP site and the LARV site from 2016 to 2019.

At the time of the baseline assay in 2016, the Ae. aegypti populations from both the OP site and the LARV site demonstrated susceptibility to the organophosphate active ingredients with ~ 98–100% mortality at the diagnostic time and 100% mortality at both the 2-h and 24-h mortality reading (Fig. 3). The response of Ae. aegypti populations from the OP site and the LARV site maintained similar levels of susceptibility throughout the study period to organophosphates, except for 2018. In April 2018, mortality for Ae. aegypti from the OP site at the diagnostic time decreased to 74% for organophosphates with 99 and 97% mortality at the 2 h and 24-h time points, respectively. For the 2018 collection from the LARV site, mortality at the diagnostic time was only 47.3% at the diagnostic time and increased to 91% at 2 h and 95% at 24 h. Outside of these 2018 collections at the OP and LARV site, susceptibility to organophosphates was indicated by all phenotypic mortality readings from 2016 to 2019.

Fig. 3
figure 3

Mean response and 95% confidence interval of Pasco County Aedes aegypti field populations in response to organophosphate active ingredients in the CDC bottle bioassay. For each population and each time point, the percent mortality at the diagnostic time, at 2 h, and 24 h is displayed

Full size image

Mortality in the CDC bottle bioassay to pyrethroids was highly variable throughout the study period. Similar to the organophosphate active ingredients, a decrease in mortality at the diagnostic time also occurred in 2018, but the decrease was not as pronounced as it was with the organophosphate active ingredients. However, resistance was observed in the same populations to the pyrethroid active ingredients. Aedes aegypti from the OP and LARV site collected in 2016 only achieved 36% and 28% mortality, respectively, at the diagnostic time. Mortality did increase by the 2-h reading, but decreased by the 24-h reading, indicating recovery from insecticide exposure and knockdown resistance (Fig. 4). The trend of increased mortality from the diagnostic time to the 2-h mortality reading and a decrease by the 24-h mortality reading was consistent for all collections from all sites made from 2016 to 2019. Mortality from pyrethroid AIs fluctuated at the diagnostic time for Ae. aegypti from the OP site from 44% in 2017, to 24–28% in 2018, and increased in 2019 to 58%. At the LARV site, 21% mortality was observed at the diagnostic time in 2017, 17% in 2018, followed by an increase to 72% mortality in 2019 (Fig. 4).

Fig. 4
figure 4

Mean response and 95% confidence interval of Pasco County Aedes aegypti field populations in response to pyrethroid active ingredients in the CDC bottle bioassay. For each population and each time point, the percent mortality at the diagnostic time, at 2 h, and 24 h is displayed

Full size image

4 Discussion

A significant amount of research has been conducted on insecticide resistance in mosquito populations throughout the world [18]. This research has indicated (1) that mosquito resistance to the major chemical classes used for vector control is widespread [50, 51], (2) resistance can directly impact the efficacy of operational vector control treatments, and (3) strategies are needed to combat resistance to maintain the use of a limited adulticide chemical toolbox. While the research on insecticide resistance and the mechanisms that confer resistance are prevalent in the literature, studies on the effectiveness of various control strategies on reversing resistance in mosquito populations is not as common. A small number of studies have assessed reversion to insecticide susceptibility in previously resistant populations [36, 37, 52,53,54]. Of these, none have evaluated reversion to susceptibility in a field scenario. Here, we provide data on the genotypic and phenotypic response of Ae. aegypti populations to different mosquito control regimens. Our results provide a field-based foundation that supports what has been found in laboratory studies: the level of resistance in field mosquito populations can be reduced when pressure from one insecticide class is removed.

Under pyrethroid selection pressure, knockdown resistance develops rapidly in field populations of Ae. aegypti [55]. Increases in kdr alleles are negatively associated with the number of detoxifying enzymes present in the mosquito [45]. Kdr alleles play a role in pyrethroid and DDT resistance, while the metabolic detoxification mechanism can affect all major classes of chemicals used in vector control [56]. Therefore, the co-existence of metabolic and target-site resistance mechanisms in Ae. aegypti provides a ‘broad-spectrum’ resistance protection against the major insecticide classes used for mosquito control. However, these resistance mechanisms do not always contribute equally. For example, in a study with Ae. aegypti where the kdr alleles were fixed, relaxation of pyrethroid pressure decreased oxidase activity, but not kdr frequency [57]. Another study suggests that selection pressure on metabolic genes may be weaker in Ae. aegypti populations that have kdr alleles [58]. It is possible that Ae. aegypti populations under selection pressure from pyrethroids become more resistant to those pyrethroids, specifically through target-site mutations, and the expressed metabolic detoxification mechanisms decrease. However, insecticide resistance is a complex system fueled by multiple mechanisms and what occurs in one mosquito population is not necessarily true for all mosquito populations. The response of the Pasco County, FL Ae. aegypti populations to organophosphates (Fig. 3) indicates that populations were classified as susceptible for most of the study period, which suggests these populations did not have elevated levels of detoxification enzymes. In Florida, pyrethroid resistance in Ae. aegypti is widespread [31] and the role of kdr alleles has been well characterized [21]. Finally, the ace-1 target-site mutation has not been implicated as a major resistance mechanism in Florida populations of Ae. aegypti [21]. Based on this information a rotational strategy between multiple chemical classes for combatting resistance is necessary.

Additionally, populations from the OP site had a dramatic drop in the IICC genotype (homozygous resistant for both alleles) from 2016 to 2017 (Fig. 2). This shift was not observed at the LARV site, but Ae. aegypti from that site exhibited a lower frequency of the IICC genotype at the beginning of the trial. Nevertheless, the decrease in the IICC genotype could be due to pressure from an organophosphate. In the black fly, Simulium damnosum (Theobald), the effective use of temephos was regained in a previously resistant population as the result of a rotational larviciding strategy [59]. While this reversion occurred in a different vector, it is promising that it resulted in the preservation of chemical tools used for controlling a significant vector. It is important to note that the dramatic decrease in the IICC genotype did not directly correlate to increased susceptibility in the CDC bottle bioassay in that same year. This highlights the complexity of insecticide resistance and that a genotypic change does not necessarily equal a phenotypic response.

In laboratory studies conducted with Ae. aegypti, reversion to susceptibility is also observed, but this is as a result of an insecticide-free environment [37, 52, 55] versus the rotational pressure that we applied in this study. While laboratory studies that remove exposure to study reversion to susceptibility are useful, they oversimplify the environmental conditions in which exposure occurs. The household use of insecticides [39] as well as inputs from agriculture, lawn care, or aquatic weed management may all influence resistance in field populations of mosquitoes. Therefore, evaluating reversion to susceptibility in a field-setting is appropriate and needed to understand how resistance can be combatted. Laboratory studies suggest that in a truly pyrethroid-free environment, reversion to insecticide susceptibility should occur due to significant fitness costs of pyrethroid-resistance, specifically associated with kdr [36]. However, studies conducted with Culex pipiens (Linnaeus) indicate that these fitness costs may decrease over time with continuous selection pressure [60, 61] and this is further complicated if the kdr alleles are fixed in the population. In a field scenario, this is particularly problematic as decreased pressure from pyrethroids may not have as dramatic of an impact on resistant populations as the fitness costs are decreased.

Within our study period, clear trends in phenotypic reversion to insecticide susceptibility were not observed. However, in the OP site, a shift in genotype frequencies towards more susceptible genotypes correlatively suggest that the rotational pressure of using an organophosphate may eventually have a phenotypic impact on susceptibility. However, due to the dynamics of insecticide resistance, especially in a field setting it is difficult to assess the absolute impact of these genotypes on resistance. Notably, throughout the study period, organophosphates performed well against field populations, making them a useful rotational tool in this particular county against resistant Ae. aegypti.

Mosquito control programs may detect resistance to adulticides, but operational recommendations do not go beyond generic integrated pest management or insecticide rotation strategies. Due to the lack of field-based research on the topic, the implemented rotational strategies are inevitably different from program to program. The limited data available from our study demonstrate that a decrease in resistant genotype can occur when pyrethroid use is ceased and coupled with rotation with another chemical class. This decrease in resistant genotypes is the precursor to improved insecticide susceptibility, but that was not observed in this study, potentially due to the impact of other resistance mechanisms or failure to reach a critical threshold of susceptible genotypes in the population. The magnitude of the impact of rotation was difficult to assess because the PYR site (control) had to be discarded. However, the inability of pyrethroids to control Ae. aegypti in that treatment area that then necessitated the use of organophosphates is a further testament to the ineffectiveness of pyrethroids for the control of Ae. aegypti in this area and the dire need for chemical rotation. This study is the only one to assess field rotational strategies on resistance in mosquitoes and more are needed to draw operationally relevant conclusions for vector control programs.

5 Conclusions

Insecticide resistance in mosquitoes presents a significant challenge for vector control programs. While resistance monitoring is becoming more prevalent, operational strategies for combatting resistance in the field are not clearly defined. As with most field studies, a limitation of this study was the inability to control all sources of insecticide use such as agricultural, pest control, and homeowner use of insecticides. Additionally, as this study was conducted in collaboration with an operational vector control program, their mission to protect public health took priority over study parameters and restrictions. Despite this, valuable data was gathered from the test sites that remained active throughout the study duration.

This study represents the first field study to determine how altering insecticide pressures in resistant mosquito populations affects insecticide resistance. Although a phenotypic change in resistant populations was not observed, the decrease in frequency of the double homozygous resistance genotype could indicate the early steps of a reversion to insecticide susceptibility from a resistant state. Further research could expand on this study to identify phenotypic changes and their intensity or investigate other resistance mechanisms (i.e. metabolic mechanisms). Because Ae. aegypti is a nuisance and vector species, it is imperative that strategies that combat and potentially reverse resistance be investigated and implemented.