Aedes vector–host olfactory interactions in sylvatic and domestic dengue transmission environments

Abstract

Interactions between Aedes (Stegomyia) species and non-human primate (NHP) and human hosts govern the transmission of the pathogens, dengue, zika, yellow fever and chikungunya viruses. Little is known about Aedes mosquito olfactory interactions with these hosts in the domestic and sylvatic cycles where these viruses circulate. Here, we explore how the different host-derived skin odours influence Aedes mosquito responses in these two environments. In field assays, we show that the cyclic ketone cyclohexanone is a signature cue for Aedes mosquitoes to detect the NHP baboon, sykes and vervet, whereas for humans, it is the unsaturated aliphatic keto-analogue 6-methyl-5-hepten-2-one (sulcatone). We find that in the sylvatic environment, CO₂-baited traps combined with either cyclohexanone or sulcatone increased trap catches of Aedes mosquitoes compared to traps either baited with CO₂ alone or CO₂ combined with NHP- or human-derived crude skin odours. In the domestic environment, each of these odourants and crude human skin odours increased Aedes aegypti catches in CO₂-baited traps. These results expand our knowledge on the role of host odours in the ecologies of Aedes mosquitoes, and the likelihood of associated spread of pathogens between primates and humans. Both cyclohexanone and sulcatone have potential practical applications as lures for monitoring Aedes disease vectors.

1. Introduction

The Aedes-borne pathogens that cause dengue, zika, yellow fever and chikungunya in humans originated from zoonotic cycles [1], although evolutionary adaptation has allowed for mosquito transmission exclusively between humans. Recently, there have been frequent outbreaks of these diseases affecting humans in various parts of the world [2]. However, non-human primates (NHPs) are also recognized as important hosts in the interepidemic maintenance of these pathogens, vectored by forest-associated Aedes mosquitoes in sylvatic transmission cycles [1,3,4]. Spillover human infections can result from infective bites of bridge vectors (i.e. vectors that bite both reservoir host and susceptible hosts) when humans encroach into forested areas or through movement of infected sylvatic vectors into rural areas occupied by humans [5,6]. However, knowledge of the evolutionary adaptations that guide such Aedes vectors to exploit both humans and wild primates remain poor.

Mosquitoes including Aedes (Stegomyia) rely on olfactory cues to locate vertebrate hosts for a blood meal. Previous work has identified vertebrate host chemicals mainly from humans that attract mosquitoes [7,8]. The attractive cues are of practical and epidemiologic relevance and have been exploited in disease vector control and/or surveillance [9–11]. Vector surveillance is critical for any vector-borne pathogen risk assessment. This entails effective vector trapping to reveal trends in abundance, pathogen infection rates, and host-feeding patterns for appropriate and timely interventions including evaluating their impact.

Aedes (Stegomyia) vectors also exploit NHPs for a blood meal suggesting that olfactory cues may also mediate their host-seeking behaviour. In this paper, we describe experiments investigating the response of Aedes species to NHP and human odours and their components in the sylvatic and domestic environments. The goal of the research is two-pronged: (i) to increase our understanding of Aedes vector–host interactions in both environments, and (ii) to develop lures that can be deployed in existing surveillance traps for effective monitoring of adult populations of sylvatic and domestic Aedes (Stegomyia) vectors.

2. Material and methods

(a). Study sites

Experiments in the sylvatic environment were conducted in the Oloolua Forest located in the outskirts of Nairobi (electronic supplementary material, figure S1). It is a natural tropical forest reserve with indigenous woodland and measures about 400 acres. It houses the Institute of Primate Research (IPR), a constituent of the National Museums of Kenya. Common wildlife present include vervet and colobus monkey, baboon, duiker, bush pig, water buck, and hyenas. Also common are small mammal species including rodents and bats [12]. Average monthly temperature ranges are much lower compared to the values of 22°C and 28°C in Nairobi. Field trials in the domestic environment were performed in Rabai, Kilifi County (electronic supplementary material, figure S1) in coastal Kenya, endemic for dengue [13] and located on the outskirts of Mombasa. The houses are built with either cement, stone or mud walls and iron sheeting or grass thatch roofing system. Monthly temperature averagely ranges between 27°C and 31°C.

(b). Collection of odours

We collected skin odours from the NHPs baboon (Papio anubi), lowland sykes (Cercopithecus albogularis) and vervet (Chlorocebus pygerythrus). The Aedes-borne viruses have been detected or isolated from species belonging to these genera [1,14–16]. The NHPs are maintained in captivity at the Institute of Primate Research. Mosquitoes are guided to their hosts by volatiles released from the skin [10,17]. Skin odours from male and female NHPs (aged 2–7 years old) were trapped on solvent- and oven-cleaned cotton material as previously described [10,17]. Briefly, the entire body, excluding the anal areas, of each animal was rubbed separately with the cotton material (23 cm × 23 cm, Lux Premium, Bidhannagar, West Bengal, India) for 10–15 min. Before odour collection, the NHPs were anaesthetized using ketamine mixture (0.5 ml xylazine in 10 ml ketamine) injected intramuscularly. The cotton material was handled with odourless latex gloved hands to minimize contamination from human skin. Human odours comprised foot odours trapped on worn socks (overnight for 12 h from male volunteers aged 30–50 years old) and used for trapping experiments immediately after collection. The sock material was made of cotton (Lux Industries Ltd 39 K.K Tagarest, Kolkata, India) and similarly baked and cleaned with solvent as described previously for the cotton material used on the primates. Aliquot of the cotton material or socks with trapped odours were wrapped in aluminium foil, transported on dry ice for subsequent odour trapping at the Duduville Campus, International Centre of Insect Physiology and Ecology (ICIPE) Nairobi, as described below.

(c). Field evaluation of non-human primate and human-scented materials on Aedes mosquito catches in the sylvatic environment

We assessed the response of crude skin odours trapped on cotton materials from each of the NHPs and humans (socks) on Aedes (Stegomyia) mosquito catches using Biogents (BG) Sentinel traps (BioQuip Products, Rancho Dominiguez, CA, USA). This trap is designed to target Aedes (Stegomyia) mosquitoes [18]. The treatments were evaluated in a completely randomized design comprising the traps baited with: (i) CO₂ only (control), (ii) CO₂ + baboon-scented cotton, (iii) CO₂ + sykes-scented cotton, (iv) CO₂+vervet-scented cotton, and (v) CO₂ + human-worn socks (experiment I; electronic supplementary material, table S1). Two separate experiments were conducted. Each experiment comprised five treatments making a total of 10 traps daily (i.e. two replicates for each treatment). However, each experiment was set in separate areas, more than 100 m apart in the forest. Each treatment was used only once at each site. Overall, each treatment was replicated twice daily. The experiment was conducted over 12 consecutive days in 24 locations or microsites. Each experiment was repeated using fresh skin odours. Carbon dioxide was delivered daily in the form of dry ice by placing approximately 1.5 kg in each Igloo thermos container (approx. 2 l) (John W. Hock, Gainesville, FL, USA). This was delivered via a 13 mm hole at the bottom as described previously [10]. An inter-trap distance of 40–50 m was observed. For each experiment, the treatments were randomly assigned to traps placed in the selected site. Traps were deployed at 07.30 and emptied at 18.30 on the same day. The captured mosquitoes were anaesthetized using triethylamine before being identified based on their morphology to species level [19,20]. The daily counts per treatment were recorded.

(d). Extraction of volatiles from human and non-human primate-scented cotton material

We collected headspace volatiles (24 h) from the cotton- and sock-adsorbed skin samples using two Super Q adsorbents (30 mg, Alltech, Nicholasville, KY, USA) per animal type. We eluted each adsorbent with 150 µl GC/GC–MS-grade dichloromethane (Burdick and Jackson, Muskegon, MI, USA) and the total eluent (300 µl) was concentrated to 200 µl under nitrogen before chemical analyses as described below. We used the volatile entrainment system for headspace trapping from equal-sized materials taken per animal [10,17]. We passed charcoal-filtered air over the enclosed scented materials in a glass jar at a flow rate of 260 ml min⁻¹. We also collected and extracted volatiles from unused cotton material following identical procedures as described for NHPs or humans. Additionally, we used identical entrainment and sample preparation parameters for this sample before analysis.

(e). Chemical analyses

We used coupled gas chromatography–electroantennographic detection (GC–EAD) analysis to isolate EAD-active components from human and NHP skin-scented cotton as described previously [10]. Antennae of laboratory-reared 3–5-day-old adult females of Aedes aegypti were used. The mosquitoes were provided 6% glucose ad libitum up to 12 h before use when the glucose was removed. Aliquots of the extracts (2 µl) were analysed using a Hewlett-Packard (HP) 5890 Series II gas chromatograph fitted with an HP-5 column (30 m × 0.25 mm i.d. × 0.25 µm film thickness, Agilent, CA, USA). Nitrogen was used as the carrier gas at 1 ml min⁻¹. Details of the analysis, mode and GC oven conditions, data capture and processing are described in Tchouassi et al. [10]. The whole mosquito was mounted alive while restrained with an adhesive tape. The base of the head was connected to the reference electrode and the tip of the antennae to the recording electrode. Respective host volatiles that elicited antennal responses in at least three replicates using fresh female antennae were designated as EAD-active components.

GC–EAD-active peaks were identified by GC-mass spectrometry (MS).

For GC–MS analysis, 1 µl of the volatile extract from the different animals was analysed on a 7890 gas chromatograph (Agilent Technologies, CA, USA) coupled to an inert XL EI/CI with Triple-Axis mass selective detector (MSD) mass spectrometer (5975C, electron energy 70 eV, Agilent) in a splitless injection mode. The GC was equipped with an HP-5 Agilent fused silica column (30 m × 0.25 mm i.d. × 0.25 µm film thickness, Agilent), with helium as the carrier gas and a column head pressure of 7.07 psi at a flow rate of 1.2 ml min⁻¹. The oven temperature was held at 35°C for 5 min, then programmed to increase at 10°C min⁻¹ to 280°C which was held at this temperature for 10 min. GC–EAD-active components were identified by comparing their mass spectral data with library data (Adams2.L, Chemecol.L and NIST05a.L) and confirmed with those of authentic standards where available, and analysed similarly by GC–EAD and GC–MS. For GC–MS, the absolute areas of each constituent as calculated by the NIST05a.L software were used to estimate their amounts using an external calibration equation generated from known amounts of authentic compounds. Additionally, trapped odours from unused cotton material and unused socks were included as controls in all GC–MS analysis. Solvent blanks (dichloromethane) were concentrated and analysed using identical parameters. Compounds present in the unused cotton or sock extracts including blank analyses (to identify contaminants) were excluded from the composition percentages of compounds in the samples.

We quantified important signature cues based on the presence/absence in humans and the NHP volatile collections. We considered four to six independently acquired volatile samples per host type. The ratio of their corresponding peak areas was compared to that of the external standard, cyclohexanone (authentic sample). We recorded the peak areas for different concentrations of cyclohexanone covering the expected analyte concentration range. From this, a calibration curve and linear equation (R² = 0.97; Y = 1E + 06x−8E + 06) was used to quantify the selected components in the GC–MS volatile profiles.

(f). Field evaluation of host signature cues on Aedes captures in the sylvatic environment

Cyclohexanone was reproducibly detected in GC–EAD profiles of the three NHP species (see §3b), whereas sulcatone was identified in human odours as previously reported [21]. We estimated the emission rates of these two compounds by dividing the amounts calculated from the calibration curves with total time the respective collections were made. Of note is that the emission rates relate to volatiles captured on the cotton materials (including socks) and not the actual release rates from a live host. Using these values, we upscaled by a common factor (10×) to obtain different doses which were tested in the field. The lowest doses obtained were 0.01 mg ml⁻¹ and 0.02 mg ml⁻¹ for sulcatone and cyclohexanone, respectively. Then, two consecutive 10-fold higher doses were prepared in hexane viz: 0.1 and 1 mg ml⁻¹ for sulcatone and 0.2 and 2 mg ml⁻¹ for cyclohexanone. Two types of dispensers were evaluated, rubber septa and sponge cloth (Wipex, Supa Brite Ltd, Nairobi, Kenya) (18.5 mm × 20.5 mm) (electronic supplementary material, figure S2). Solutions (200 µl each) were applied and left for 30 min at room temperature to allow the solvent to evaporate. Thereafter, they were placed immediately into the odour dispensing pocket of the BG Sentinel trap for field experiments. However, the sponge cloth was enclosed in a low-density polyethylene sachet folded twice (0.15 mm thickness, 25 cm² surface area) and sealed with a hot rod sealer before deployment.

We performed preliminary trials in the sylvatic environment in Oloolua Forest (electronic supplementary material, figure S1) to determine the effectiveness in dispenser type to dispense each of the compounds (electronic supplementary material, S2). We monitored total mosquito catches in traps having CO₂ and different doses of the individual compounds in three replicate trials over 3 days. Rubber septa was selected to dispense cyclohexanone and sponge cloth for sulcatone, based on higher mean catches (n = 3) (electronic supplementary material, S2). This dispenser–compound combination was retained in all subsequent field trials. The individual compounds comprising the blend (CS) were dispensed using the respective dispensers: sulcatone (sponge cloth) and cyclohexanone (rubber septa).

In the sylvatic environment, we assessed the attractiveness of these compounds by monitoring mosquito trap collections compared to those caught in crude skin odours from the individual NHP and humans. Experimental replicates (experiment II, electronic supplementary material, table S1) in a complete randomized design consisted of the traps baited with: (i) CO₂ only (control), (ii) CO₂ + baboon-scented cotton, (iii) CO₂+sykes-scented cotton, (iv) CO₂ + vervet-scented cotton, (v) CO₂ + human-worn socks, (vi) CO₂ + different doses of individual compounds, and (vii) CO₂ + CS (comprising a blend of cyclohexanone (0.2 mg ml⁻¹) and sulcatone (0.1 mg ml⁻¹), at optimum doses). In experiment II, we tested 12 treatments making a total of 12 traps, daily. This experiment was conducted in separate areas (approx. 100 m apart) in the forest. Each treatment was used only once at each site and replicated once daily. The experiment was conducted daily for 14 consecutive days. The experimental design, timing of trap deployment and retrieval, processing, mosquito identification and analysis were as described previously.

(g). Field evaluation of host signatures cues in the domestic environment

This experiment (experiment III, electronic supplementary material, table S1) was conducted in surrounding vegetation around homesteads in Rabai, Kilifi County (electronic supplementary material, figure S1). The treatments evaluated in the BG Sentinel trap included: (i) CO₂ only (control), (ii) CO₂ + human-worn socks, (iii) CO₂ + different doses of individual compounds (same doses evaluated in sylvatic environment), and (iv) CO₂ + CS. The same design was employed as described earlier. Here, the experiment comprised nine treatments making a total of nine traps evaluated daily. The experiment was conducted in separate areas with each treatment being used only once at each site. Overall, the experiment was conducted daily for 12 consecutive days in 12 separate locations.

(h). Chemicals

The chemicals used included: hexanal (Sigma Aldrich, 99%), benzaldehyde (Sigma Aldrich, 99.5%), cyclohexanone (Sigma Aldrich, 99%), 6-methyl-5-hepten-2-one (sulcatone) (Sigma Aldrich, 99%), heptanal, octanal, nonanal and decanal (Sigma Aldrich, 99%).

(i). Statistical analysis

We recorded the daily mosquito counts by species for each treatment. We tested the effect of treatment and day as predictor variables on species-specific adult female mosquito abundance (Aedes chaussieri, Ae. aegypti, Aedes pembaensis, Aedes cumminsii). These served as response variables in separate generalized linear models (GLM) with a negative binomial error structure. Additionally, we included the effect of trap location as a covariate in analysis of field data involving crude odours in the sylvatic environment only (experiment I). All GLMs were implemented in R v. 3.3.1 [22] with the MASS package at 95% significance level. Model validity was assessed by inspection of residuals. Quantitative differences in the amounts of cyclohexanone between the primate species and sulcatone among the different human volunteers were compared using univariate analysis of variance.

3. Results

(a). Sylvatic Aedes mosquitoes are weakly attracted to human and non-human primate-scented cotton material

Trap captures with crude skin odours comprising human and NHP-scented cotton materials yielded 4131 mosquitoes in 24 replicate trials. Aedes (Stegomyia) chaussieri dominated the total captures (79.6%) followed by Ae. aegypti (11.2%) and Ae. pembaensis (6.1%). The mosquito composition recorded in the experiment is represented in the electronic supplementary material, figure S3A. Aedes chaussieri catches varied by treatment (χ²_4,92=133.91; p < 0.0001), day (χ²_11,108 = 295.81; p < 0.0001) and microsite (trap location) (χ²_12,96 = 208.48; p < 0.0001). The addition of cotton-scented material from the individual NHPs and humans (socks) to CO₂₋baited traps resulted in a significant reduction in catches of Ae. chaussieri relative to the control CO₂ (figure 1; electronic supplementary material, table S1; experiment I). In fact, there was a two- to fivefold reduction in Ae. chaussieri catches in CO₂-baited traps containing skin odours of the respective hosts compared to control traps having only CO₂. The abundance of Ae. pembaensis varied by day (χ²_11,108 = 156.77; p < 0.0001) and microsite (χ²_12,96 = 128.27; p = 0.005) but not among the treatments (χ²_4,92 = 121.16; p = 0.13). A similar pattern was observed for Ae. aegypti where variation in catches was influenced by day (χ²_11,108 = 122.52; p < 0.0001) and microsite (χ²_12,96 = 101.21; p = 0.046) but not between the treatments (χ²_4,92 = 99.63; p = 0.90) (electronic supplementary material, table S1; experiment I).

Figure 1.

Mosquito abundance pattern in crude NHP and human skin odours in the sylvatic environment. (a) Aedes chaussieri abundance, (b) Ae. aegypti abundance and (c) Ae. pembaensis abundance. Boundaries of the dot plot whiskers represent the minimum and maximum of all the count data; closed dots represent data points and those outside the boundaries are outliers; replicate values as open dots are also indicated for each treatment; black bars represent the median number of catches; **significance at p < 0.01 from the control (CO₂ only); n.s. denotes non-significance from the control at p < 0.05; the host odours comprised skin volatiles trapped on cotton material (NHPs) or worn socks (humans); number of replicates, n = 24. All treatments were less attractive than the control. (Online version in colour.)

(b). Gas chromatography–electroantennographic detection analysis and analysis of volatiles of human and non-human primate-scented cotton material

Aedes aegypti detected a total of eight components in the scented cotton/socks of NHPs and humans. Representative GC–EAD traces are presented for odours to the NHPs (figure 2a–c) and human (figure 2d). These components were identified and confirmed by GC–MS analysis as the aldehydes heptanal, octanal, nonanal, decanal, benzaldehyde, and ketones, 3-hydroxy-2-butanone, 6-methyl-5-hepten-2-one (sulcatone) and cyclohexanone. All the compounds were detected by Ae. aegypti antennae in at least three GC–EAD runs.

Figure 2.

Representative GC–EAD profiles using adult female Ae. aegypti to (a) baboon, (b) vervet, (c) sykes and (d) human, odours. Upper traces are flame ionization detector (chemical profile) of the respective host odours and lower traces are EAD responses; number of runs, n = 3; scale bar in all the GC–EAD runs, 5 mV div; red asterisks are unidentified compounds. (Online version in colour.)

Further, cyclohexanone and sulcatone were important qualitative signature cues of NHPs and humans, respectively. The amounts of cyclohexanone did not vary between the individual NHP species (F₂ = 0.06, p = 0.94) with the mean emission rates being 19.4 ± 1.9 ng h⁻¹ per NHP-scented cotton. Likewise, the amounts of sulcatone did not vary between the human-scented socks from the volunteers (F₂ = 0.5, p = 0.72) averaging 10.3 ± 0.4 ng h⁻¹ per volunteer-scented material.

(c). Sylvatic Aedes are attractive to components of crude primate and human skin odours

We assessed whether the identified signature cues, cyclohexanone (for NHPs) and sulcatone (for humans) influenced attractiveness of Aedes mosquitoes in standard BG Sentinel traps. The experiment yielded 11 346 mosquitoes in 14 replicate trials. The distribution of species in this experiment is shown in the electronic supplementary material, figure S3B. Aedes chaussieri was dominant (75.7%, n = 8590), followed by Ae. cumminsii (18.6%, n = 2106) and Ae. pembaensis (4.5%, n = 512). Catches of these species varied by day and treatment. Compared to the control, only addition of cyclohexanone at 0.2 mg ml⁻¹ or sulcatone at 0.1 mg ml⁻¹ significantly increased Ae. chaussieri catches when combined with CO₂. Traps at the indicated doses in combination with CO₂, captured about two times more of this species than control traps (figure 3; electronic supplementary material, table S1; experiment II). Whereas a dose effect on mosquito trap collections for each of the compounds was observed, the addition of skin odours of the individual NHPs (except baboon) or human caused a reduction in trap catches relative to the control, as previously encountered. A blend of both compounds at their optimal attractive doses (CS) did not improve trap catches relative to the individual compounds (electronic supplementary material, table S1; experiment II). A dot plot display of the daily catches of specific species in the treatments compared to the control is presented in figure 3.

Figure 3.

Comparison in mosquito abundance pattern in NHP and human signature cues and their crude skin odours in the sylvatic environment. (a) Aedes chaussieri abundance, (b) Ae. cumminsii abundance and (c) Ae. pembaensis abundance. Boundaries of the dot plot whiskers represent the minimum and maximum of all the count data; closed dots represent data points and those outside the boundaries are outliers; replicate values as open dots are also indicated for each treatment; black bars represent the median number of catches; *,** significance at p < 0.05 and p < 0.01, respectively, from the control (CO₂ only); the host odours comprised skin volatiles trapped on cotton material (NHPs) or worn socks (humans); number of replicates, n = 14. Sulcatone 1 mg ml⁻¹ (sulca1), sulcatone 0.1 mg ml⁻¹ (sulca0.1), sulcatone 0.01 mg ml⁻¹ (sulca0.01), cyclohexanone 2 mg ml⁻¹ (cyclo2), cyclohexanone 0.2 mg ml⁻¹ (cyclo0.2), cyclohexanone 0.02 mg ml⁻¹ (cyclo0.02), CS (a blend of cyclohexanone and sulcatone at optimum doses of 0.2 mg ml⁻¹ and 0.1 mg ml⁻¹, respectively). (Online version in colour.)

(d). Aedes aegypti is attracted to volatiles from crude human-scented material and constituents of human and non-human primate odours in the domestic environment

In the domestic environment in coastal Kenya, we compared adult Aedes mosquito catches in traps baited with CO₂ only (control) and each compound at different doses combined with CO₂. We recorded 1968 female mosquitoes in 12 replicate trials. Aedes aegypti was dominant (84.0%, n = 1653), followed by Ae. simpsoni s.l. (6.0%, n = 118). ‘Other’ species comprised culicines and anophelines recorded in low numbers (10.0%, n = 197) (electronic supplementary material, figure S3C). We found a significant variation in Ae. aegypti catches between the treatments (χ²_8,99=124.6; p = 0.002) (figure 4). Both cyclohexanone and sulcatone when individually combined with CO₂ increased catches of this species compared to CO₂ alone. However, the effect was dose dependent with pronounced catches at higher doses than previously recorded in the sylvatic environment (figure 4; electronic supplementary material, table S1; experiment III). We recorded about two- to threefold increase in catches of Ae. aegypti following the addition of cyclohexanone (2 or 0.2 mg ml⁻¹) or sulcatone (1 mg ml⁻¹) in CO₂-baited traps than the control. Interestingly, in this environment, human-worn socks when combined with CO₂ recorded threefold catches of Ae. aegypti relative to the control trap (figure 4; electronic supplementary material, table S1; experiment III). This pattern is in stark contrast to the reduced mosquito captures observed for human odours in the sylvatic environment (electronic supplementary material, table S1; experiments I, II). Furthermore, Ae. aegypti catches in the blend of both compounds at their optimal attractive doses (CS) were comparable to those of the individual doses and the control (figure 4; electronic supplementary material, table S1; experiment III).

Figure 4.

Aedes aegypti abundance pattern in NHP and human signature cues and human-scented socks in the domestic environment. Boundaries of the dot plot whiskers represent the minimum and maximum of all the count data; closed dots represent data points and those outside the boundaries are outliers; replicate values as open dots are also indicated for each treatment; black bars represent the median number of catches; *, ** significance at p < 0.05 and p < 0.01, respectively, from the control (CO₂ only); human odours comprised skin volatiles trapped on worn socks; number of replicates, n = 12. Sulcatone 1 mg ml⁻¹ (sulca1), sulcatone 0.1 mg ml⁻¹ (sulca0.1), sulcatone 0.01 mg ml⁻¹ (sulca0.01), cyclohexanone 2 mg ml⁻¹ (cyclo2), cyclohexanone 0.2 mg ml⁻¹ (cyclo0.2), cyclohexanone 0.02 mg ml⁻¹ (cyclo0.02), CS (a blend of cyclohexanone and sulcatone at optimum doses of 0.2 mg ml⁻¹ and 0.1 mg ml⁻¹, respectively). (Online version in colour.)

4. Discussion

In this study, we assessed NHPs as sources of attractants for sylvatic and domestic Aedes species. We identified cyclohexanone and sulcatone as important signature cues for NHPs and humans, respectively, and demonstrated their attractive effect based on Aedes mosquito catches in field trials. In the presence of CO₂, these two compounds significantly increased Aedes captures in the conventional BG Sentinel traps compared to the traps baited with CO₂ alone (electronic supplementary material, table S1; experiments II, III). Thus, both compounds have potential practical applications as lures that could be deployed in traps to increase mosquito captures to maximize virus detection during periods of enzootic virus transmission of low viral activity [10].

Mosquitoes typically encounter a complex mixture of skin odourants during host seeking for a blood meal. Odour coding to natural odour mixtures often are perceived more distinctly and efficiently than their constituents [23], supporting the widely posited view that a repertoire of host-derived chemicals elicit attraction better than single compounds [24]. We found increased attraction of Ae. aegypti to worn socks representing human odours in the domestic than sylvatic environment (electronic supplementary material, table S1; experiment III). This finding partly supports this hypothesis, although comparable mosquito catches were observed in traps baited with socks and specific doses of the compounds (figure 4; electronic supplementary material, table S1; experiment III). The increased Ae. aegypti catches to odours from human-worn socks in the domestic but not in the forest environment possibly reflects an adaptive and evolutionary preference of this mosquito species for human body odour as previously reported [21]. Notably, we sampled mosquitoes in both forest and sylvatic environments using the odours from the same pool of volunteers.

Interestingly, the single synthetic compounds (cyclohexanone and sulcatone) were more attractive to mosquitoes than the crude host skin volatiles in the sylvatic environment. This finding could be related to the different dispensers used. Dispensers vary in pore size and dispense chemicals differently. Previous work has found improved mosquito catches to specific host constituents as attractants than blends or crude volatiles [8,25], perhaps reflecting the need for additional research to design improved lures.

We found that the addition of cotton-scented materials from the NHPs and humans decreased mosquito catches in the sylvatic environment when compared with control traps. This trend was repeatedly observed in the sylvatic environment even though the experiments were conducted at different times (electronic supplementary material, table S1; experiments I, II). These findings contrast previous reports of enhanced mosquito captures following the addition of odour-scented materials in traps [9,17]. It appears that the type of material used as an adsorbent for skin odours can have profound influence on mosquito catches [26]. A possible reason for the low attractiveness of scented cotton material to mosquitoes in the sylvatic environment might be associated with its matrix. We used cotton material as an adsorbent and assumed that it traps skin odours from the animals to the same degree. This might not have been the case, given that the animals have different skin types (and surface areas). Other material types for odour sampling should be considered and explored in further research. Given that the same cotton material (socks) was used in the domestic environment but traps baited with the human-worn socks had the highest mosquito catches (figure 4; electronic supplementary material, table S1; experiment III) suggests that additional factors may contribute to the differential attraction. For instance, the variation in the performance of crude odours (e.g. from humans) on mosquito catches could be ascribed to differences in the biology of the major species encountered. Aedes chaussieri was the dominant species in the sylvatic environment and Ae. aegypti in the domestic environment. Both species may differ in their sensitivity to constituents of the crude volatiles, which would require additional research. Such sensitivity difference could evoke differential responses or behaviours to host odours as previously highlighted [27,28]. Furthermore, because crude host volatiles comprise a mixture of compounds, some components may be attractants or repellents [7,9]. As such, variation in the amounts and ratios of these components released between different individuals could account for differences in trap catches.

Sulcatone has been implicated in the chemical ecology of various hematophagous insects and previously identified as an important signature cue for humans [21]. This odourant has been reported to either attract or repel Ae. aegypti depending on the dose tested [21]. Our results provide further evidence of its attractiveness on this mosquito species in the field. Field studies capture the natural response of arthropods including mosquitoes to odourants. On the other hand, cyclohexanone is an odourant that has been reported to activate the cpA neuron on the maxillary palps like CO₂ [29]; however, this study did not establish the source or possible connection with skin odour. In a previous study carried out in a mesocosm using a counter-flow design trap, the closely related compound, cyclopentanone, was found to capture similar numbers of Culex quinquefasciatus as CO₂ [30]. Perhaps these cyclic ligands may be sensitive to a broad range of mosquito species, given their chemical similarity (cyclic structure and presence of a ketone as a functional group). In general, our results show that cyclohexanone and sulcatone are both detected by the mosquito antennae and could be used as attractants for field monitoring of mosquito populations targeting a broad range of species. Evidence for this comes from our data showing that although detected by Ae. aegypti in GC–EAD assays, both compounds elicited a behavioural response to different Aedes species in field experiments.

The most attractive doses to mosquitoes were much lower for sulcatone than cyclohexanone. It is unclear whether this finding indicates potential differences in sensitivities between the compounds. Nonetheless, the behavioural sensitivity of Aedes mosquitoes to these signature human and primate odourants expand our knowledge on the role of host odours in the ecologies of these vectors, and the likelihood of associated spread of pathogens between primates and humans. Population growth in Kenya and elsewhere has resulted in human encroachment into wildlife territories in recent times [31]. Here, humans may encounter forest-dwelling Aedes mosquitoes adapted to primates. Such exposure of humans to mosquito infectious bites in the forest–village interface has been the main mechanism driving emergence of some Aedes-borne viruses like yellow fever virus, especially in East Africa [32].

Our study further demonstrates that the most attractive dose for each compound was higher in the domestic than sylvatic environment (cyclohexanone: 2 versus 0.2 mg µl⁻¹; sulcatone: 1 versus 0.1 mg µl⁻¹ in the domestic and sylvatic environment, respectively). This observation could be attributed to environmental differences (e.g. temperature, wind), possibly affecting the release rates of the compounds, and consequently, how mosquitoes perceive them. Data from the Kenya Meteorological Department showed that the mean temperature and wind speed during field testing in the sylvatic environment in Oloolua Forest was 17.2°C and 68.7 km d⁻¹, respectively. In the domestic environment in Rabai during the experimental period, the mean temperature and wind speed was 25.5°C and 143.0 km d⁻¹, respectively. This finding indicates that the ecological environment may influence the optimal performance of a given attractant formulated at a specific dose. Our study also highlights differences in dispenser type for both compounds in enhancing trap catches. In preliminary trials, sulcatone recorded higher mosquito catches when dispensed using sponge cloth sealed in a low-density polyethylene than rubber septa at the optimal dose (electronic supplementary material, figure S2). Inherent difference between the compounds having different volatilities/vapour pressures and emission/release rates may account for this observation. Taken together, our findings highlight important considerations for lure development and field assessment on insect trap catches.

We found reduced or comparable catches to a blend comprising optimal attractive doses of the odourants (CS) compared to the individual compounds (electronic supplementary material, table S1; experiments I, II, III). It is not clear whether this finding is related to the mosquitoes' peripheral coding of both odourants. Understanding the molecular mechanisms underlying attractancy to these odourants should be the focus of future studies. Worth noting is that the response of an insect to a given compound is known to occur in a dose-dependent manner [10,25,28]. Dose–response studies including the use of different dispensers are needed to confirm the observed response of combining both compounds on trap catches.

Our study has certain limitations. First, we employed cotton material as the adsorbent for host skin volatiles used for evaluation on mosquito captures. This approach appears to be imperfect as cotton material may not simulate animal skin to release odours at the natural rate [9]. Thus, traps baited with living animals could be helpful to shed light on vector–host-specific responses [33,34]. Second, quantitative differences in chemical profile between individuals of the same species and even between sexes are well appreciated [7]. Similar differences are not unexpected among the individual NHP species considered in this study. We found other compounds (some EAG-active, e.g. the aldehydes) with quantitative variation in the volatile profiles of humans and individual NHPs (data not shown). These compounds could be the subject of further studies to evaluate their behavioural impact on mosquito catches and even in blends with the currently identified attractants. Third, not much is known about the epidemiologic significance of some of the Aedes spp. encountered in the sylvatic environment (Ae. chaussieri, Ae. pembaensis) in relation to the Aedes-borne pathogens indicated. Arboviruses such as Lumbo and Rift Valley fever viruses have been isolated from Ae. pembaensis/Ae. cumminsii [35,36], viraemia owing to the former virus having been observed in vervet monkeys [35]. While the mosquito, Ae. aegypti is regarded as the primary vector of the viruses (i.e. dengue, zika, yellow fever and chikungunya viruses) globally, a range of African Aedes species may be competent and epidemiologically significant vectors [5,6], many yet to be fully unravelled owing to poor research or minimal surveillance. To the best of our knowledge, there are no reports on these mosquitoes inhabiting domestic or urban environments. Nonetheless, our data on the performance of the host-derived compounds on overall mosquito catches suggest broad attractiveness on diverse mosquito species worth further investigation in field trials.

5. Conclusion

We provide evidence through chemical analyses and field data that both NHPs and humans are sources of volatile odours that attract Aedes mosquitoes including the dengue vector, Ae. aegypti. Both sulcatone and cyclohexanone improved trapping of Aedes mosquitoes and thus have potential practical applications as lures in traps. The behavioural sensitivity of both signature human and primate odourants, expand our knowledge on the role of host odours in the ecologies of Aedes mosquitoes, and the likelihood of associated spread of pathogens between primates and humans.

Ethics

The use of mice for routine colony maintenance of Ae. aegypti and participation of humans as volunteers to contribute foot odours for field experimentation and setting up of traps around homesteads was sought from the Kenya Medical Research Institute Scientific and Ethics Review Unit (KEMRI-SERU) (project number SERU 2787). Informed consent was obtained from the human subjects to participate in the study including access to environment traps around homesteads in the domestic environment. Approval for the use of primates for odour collection in field trials was sought from the Institutional Review Committee (IRC) of the IPR, National Museums of Kenya (no. IRC/01/2016). The IPR ensures adherence to national guidelines on the care and use of animals in research and education in Kenya enforced by the National Commission for Science, Technology and Innovation (NACOSTI). The Institute is a member of the Association of Assessment and Accreditation of Laboratory Animal Care (AAALAC) in Africa.

Data accessibility

This article has no additional data.

Authors' contributions

D.P.T., R.S. and B.T. designed the research, D.P.T., J.W.J. and E.O.O. collected data, D.P.T. analysed data, D.P.T. and B.T. wrote the first draft of the manuscript, and all authors contributed to revisions.

Competing interests

We declare we have no competing interests.

Funding

This study was funded by National Institutes of Health (NIH), grant no. 1R01AI099736-01A1 to R.S. We gratefully acknowledge the financial support for this research by the following organizations and agencies: UK's Department for International Development (DFID); Swedish International Development Cooperation Agency (Sida); the Swiss Agency for Development and Cooperation (SDC); and the Kenyan Government. The views expressed herein do not necessarily reflect the official opinion of the donors. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.