Patterns of dengue virus transmission are influenced by the abundance, survival, and behavior of the principal mosquito vector, Ae. aegypti. Two main emphases for Ae. aegypti control exist: (1) reduction of the larval stage through environmental management (source reduction), larvicides and biological control; and (2) reduction of the adult stage using fumigation and/or residual spray of insecticides. Since the early 1900s [3], [4], it has been known that the most cost-effective means of preventing mosquito-borne disease is to target the adult vector, which transmit the pathogen. However, the prevailing paradigm for suppressing Ae. aegypti targets immature mosquitoes, the vast majority of which will not survive long enough to transmit virus [5]. For emergency interventions during dengue outbreaks, targeting the adult vector population by outdoor ultra-low-volume (ULV) application of insecticides and/or indoor thermal fogging remain the methods of choice [6], [7]. However, most control interventions that apply adulticides by space-spraying achieve relatively low effectiveness [8]–[13]. One reason for this reduced effectiveness can be attributed to vector behavior. Aedes aegypti is extensively adapted to exploit the human environment. The female almost exclusively takes blood from humans [14] and most commonly feeds and rests indoors. This species will also lay eggs in available oviposition and larval developmental sites inside the home [15]. This extensive use of the human indoor environment poses unique challenges to traditional adult control methods since chemical applied through outdoor and peridomestic ULV methods must pass through house portals to reach the interior space where the vector can make contact with the insecticide. This approach results in the loss of some chemical prior to reaching the interior space. Control in buildings usually accomplished with indoor residual or space spray are often hampered by limited access into homes and resource limitations [5]. On the other hand, Ae. aegypti's high affinity for the human indoor environment also provides opportunities for innovative approaches to control the adult vector [5].

Aedes aegypti has been characterized as having specific resting preferences based on visual cues (i.e., dark colors) [16], [17], and to be significantly attracted by black [18], yellow, orange and red colors [19]. Studies that have exploited Ae. aegypti's attraction to color contrast (i.e. simultaneous presentation of two colors, one which mosquitoes are attracted to in order to direct them to a target) have led to the development of host-seeking adult traps such as the Fay-Prince [20], counterflow geometry trap [21], and the BG Sentinel™ trap [22]. Previous studies in Thailand [23] demonstrated the utility of exploiting the resting preference of Ae. aegypti to develop attractant resting boxes for quickly sampling the indoor-resting population of this species. However, in relation to world-wide dengue burden, relatively few laboratory-controlled studies have been performed to quantify these behavioral patterns, and minimal research has been conducted to determine how to exploit this knowledge to reduce Ae. aegypti mosquito densities inside homes where man-vector contact is high [23].

A full description of mosquito behavior provides important information on their role as disease vectors and could serve as the basis for their control. There is growing consensus that the scarce resources available for mitigating tropical public health problems should be utilized in an evidence-based and cost-effective manner [24]. Historically, adult mosquito control using fumigation and indoor residual spray has focused mainly on the lethal actions of chemicals [6], [7]. However, research shows that there are other chemical actions that break vector-human host contact [25]–[32]. Two such actions are initiating a spatial repellent or deterrent effect, thereby preventing mosquito entry into a treated space (house); and a contact irritant effect, causing an escape response from a treated space prior to mosquitoes biting humans [32]–[35].

Such non-lethal chemical approaches are being evaluated in the development of a Push-Pull strategy for Ae. aegypti control currently in the proof-of-concept stage in both Peru and Thailand. “Push-Pull” is defined here as a strategy that aims to (1) prevent mosquito entry into homes through repellency and/or promote their early exit from homes through contact irritancy (Push); and (2) trap repelled and/or irritated mosquitoes in the outdoor environment using peridomestic traps (Pull). The goal of the strategy is to target preferred mosquito house entry portals and/or indoor resting sites with standard vector control chemicals (i.e. chemicals approved by World Health Organization Pesticide Evaluation Scheme for use in vector control) to make them unsuitable. The approach is to use minimum effective chemical dose and treated surface area coverage to reduce indoor densities of host-seeking (i.e., female adult) Ae. aegypti populations. One component to achieving this goal includes quantifying the patterns of resting behavior of Ae. aegypti exposed to chemical-free and chemical-treated surfaces to define the impact of untreated surfaces on irritancy behavior -is there a shift to resting on “safe-sites” resulting in an attenuated escape response?

With current suggestions that sub-lethal chemical approaches to vector control (i.e. contact irritancy) may pose viable options to reduce disease [32]–[35], [44], it is important to characterize minimal effective doses of irritant chemicals and the relationship between surface area coverage of these doses and the behavioral responses that they elicit (i.e., rapid escape from inside homes). Current adult vector control approaches such as insecticide treated bed nets, and/or clothing rely on human hosts as the attractant or bait to lure mosquitoes into contact with the treated material long enough to deliver the lethal dose of the insecticide [45]. However, when relying on the treatment of resting sites, such as the interior house walls, to reduce man-vector contact through an irritant response, interaction of the vector with these treated surfaces is facultative. Untreated areas in the house or safe-sites may be available and/or preferred for resting [12], [46] thus minimizing the impact of the intervention. It is vital therefore, to understand the drivers of these resting preferences in order to exploit and maximize the effects of irritant chemicals on vector escape responses. Such strategies will guide development of cost-effective tools for the future.

The present study quantified the resting patterns of two Ae. aegypti female populations (THAI and PERU) under both chemical-free and treatment conditions using a simple laboratory assay. During the chemical-free baseline trials, several variables were evaluated to include material type (cotton and polyester), dark : light color surface area coverage (SAC), and fabric configuration (horizontal, vertical). Not surprisingly, results indicate that both mosquito strains were observed resting preferentially on dark versus light colored material against both material types. These patterns were consistent using both the vertical and horizontal configuration study designs. The magnitude of this response was measured as greater than expected proportions of resting observations on the dark material even at the 25% SAC ratio despite the availability of alternate resting sites, or other behavioral responses such as flight. Similar findings have been described during our experimental hut validation studies in Thailand [Thainchum et al. unpublished data] and Peru [Castro et al. unpublished data] where most Ae. aegypti preferred to rest on dark material rather than light, regardless of fabric type even at 25% and 50% SAC.

Although horizontal configuration enhanced resting on both dark cotton and polyester material strips in the current study, as well as under field conditions in experimental huts [Thainchum et al. unpublished data], no consistent preference was observed between upper and lower locations of the dark material within the laboratory assay chamber. This may be due to the relatively small size of the assay that created a spatial bias for the test system – i.e., the laboratory assay dimensions may have precluded substantial differences in height between upper and lower wall portions. However, similar observations have been made in our experimental hut studies where based on observations from upper and lower wall heights, greater proportions of female Aedes aegypti populations were observed resting on lower portions of the wall when exposed to cotton material whereas against polyester, upper wall portions were preferred (unpublished data).

An explanation for the variation in resting patterns between the two material types in the current study may include the variation in the microclimate within the test box. Previous studies under laboratory conditions have reported similar findings using Anopheles and Culex mosquitoes in which they preferred to rest on lower portion of a test box that was cooler than that of the upper portion [47]. Unfortunately, it is not possible to validate this theory using the datasets of the current study because environmental parameters were only measured from a central location inside the bioassay room rather than along the wall surfaces within the test box. Future experiments should integrate microclimate data to better understand behavioral responses.

The fact that cotton enhanced resting on the dark strips as compared to polyester indicates that: 1) the green color of the polyester did not provide as much contrast to the white background as the black color of the cotton; 2) the weave or texture of the cotton provides enhanced tactile cues; or 3) material-specific moisture absorption properties exist under the conditions in which the assay was conducted (i.e. cotton retains more moisture than polyester). When evaluating green cotton versus green polyester simultaneously, cotton still enhanced resting on the dark strips. This finding suggests that the differential resting preference observed between cotton and polyester may not be due to variations in color contrast between material types, but rather is the result of their texture and/or moisture absorption properties. Cotton exhibits greater moisture absorption than that of polyester [48]. It is interesting to note that studies under field conditions in Thailand and Peru are also indicating an overall general decrease in resting when polyester is used versus cotton under chemical-free conditions [Thainchum et al. ; Castro et al. unpublished data]. Such information could be vital in optimizing various vector control tools and could be most beneficial for products designed to target attraction/resting behaviors.

Observations made within the treatment metal boxes during chemical trials indicate that, knockdown responses in all test assays were low (≤5%) even at high chemical dose and treatment area coverage (i.e., 75% and 100%). Low KD even at test doses higher than WHO recommended field application rate for alphacypermethrin (≈7 nm/cm²) is probably due to a reduced resting on the treated material and consequentially an increase in proportion of mosquitoes flying (irritated/agitated). It must be noted that test populations were only exposed to the treated surfaces for a total of 10 min, well below the standard 1 hour used in toxicity assays [49]. Also, as the THAI Ae. aegypti strain has been characterized as pyrethroid tolerant and DDT resistant [50], [51], it was expected that KD/mortality would be low in these test populations. More importantly, the THAI strain still exhibited a contact irritant response (indicated by increased flying) when exposed to both alphacypermethrin and DDT. These results indicate that sub-lethal approaches to vector control may be effective in resistance management.

Perhaps most important for operational significance is the observation that was made in the test chamber during chemical trials indicating that both mosquito strains continued to rest in greater proportions on dark chemical-treated material versus safe-sites (i.e. chemical-free light material, assay lids and floor) when any dose of either alphacypermethrin or DDT were used. Even under test conditions in which shifting to safe-sites were expected (i.e., 25% SAC), results show no consistent increase in resting counts on chemical-free material. As expected, however, when observations were compared between treatment and matched control assays, significantly fewer mosquitoes were observed resting overall on the dark material treated with chemical.

For all chemical evaluations, the proportion of mosquitoes observed flying was significantly increased in the treatment assay as compared to matched control regardless of the material type used, surface area coverage and configuration of the treated areas within the box. Again, these findings indicate that under current test conditions, Ae. aegypti did not simply move to safe-sites (untreated areas) following contact with chemical-treated material but were clearly agitated as measured by an increased flight response. It is this contact irritant response that may elicit escape behavior from a treated space and can be exploited for reducing man-vector contact inside homes. Any residual chemical that is applied to indoor surfaces and has sufficiently strong irritant properties would potentially disrupt the normal resting and may affect the feeding pattern of a vector. These actions could consequently reduce vector – human contact because of rapid escape from inside human dwellings [35]. Such a contact irritant response is well documented in previous field experimentation [32], [52], [53]. While the designs were different in these studies, results from each indicate a rapid escape of mosquitoes from inside experimental huts in response to irritant chemical applications and is the basis for the current laboratory conclusion. The challenge is to ensure that agitation, observed in the current study, does not increase biting on humans prior to escape as this would be counterproductive to intervention impact. Ongoing laboratory studies using the box assay are evaluating escape responses under similar current test conditions to measure the effect of focal treatment on mosquito movement away from a treatment source.

It should be noted that it was not the aim of the current study to compare resting behavior patterns between THAI versus PERU Ae. aegypti strains. Each strain was evaluated independently as results from each are currently being validated under field conditions at strain-specific locales (i.e, Kanchanaburi and Iquitos, respectively). However, future studies could investigate the relationship between behavioral phenotype and genetic characteristics of each geographical strain to explore differences that may exist in the resting behavior in response to chemical actions. This information would be useful in understanding the varying challenges in successful implementation of sub-lethal vector control strategies designed to have impact on mosquito populations from different geographic locations.

Table S1

Table S2

Table S3

Table S4

Table S5