Background

Carbon dioxide (CO2), a major constituent of vertebrate breath, plays an important role in the host-seeking process of mosquitoes [1–6]. Therefore, the compound is commonly added to traps used for mosquito surveillance [7–9]. Among malaria vectors, opportunistic, zoophilic as well as anthropophilic mosquito species are affected by CO2[2, 4, 6, 10–13]. In Anopheles gambiae sensu stricto, an important vector of human malaria in sub-Saharan Africa and considered to be highly anthropophilic [14], CO2 augments the attractiveness of human odour [6, 12] and it is an essential cue to lure the female mosquitoes into the vicinity of mosquito traps [5, 13].

Even though CO2 has a positive effect on the number of mosquitoes that are caught by suction traps, in resource-poor areas, like sub-Saharan Africa, it is hard to obtain CO2 sources that are reliable, cheap, easy to manage and durable. Propane-powered traps that produce CO2[15] are difficult to obtain, heavy and expensive. The same is true for industrially-acquired CO2, which, packaged in steel cylinders, has the advantage that the release rate of CO2 can be regulated, but leakage at the connections may occur. In addition, flow meters may be costly and sensitive to dust and high humidity. Dry ice, an alternative source of CO2, is cheap and easier to handle than pressurized CO2 cylinders, but is difficult to obtain and transport in the tropics, besides the need for replenishment on a regular basis. Moreover, dry ice has the disadvantage that the release rate of CO2 is highly variable and diminishes over time [2, 16].

Saitoh et al[16] developed an easy and cheap method to produce CO2 by using a yeast-sugar solution in plastic bottles. Under anaerobic conditions, yeast (synonym for strains of Saccharomyces cerevisiae or baker's yeast) converts sugar into CO2 and ethanol [17–20]. In Japan, traps baited with yeast-generated CO2 caught higher numbers of Aedes and Culex spp. than unbaited traps. The objective of the present study was to investigate, under laboratory, semi-field and African field conditions, whether this method is valuable to lure An. gambiae s.s. females towards suction traps, as an alternative for industrial-acquired CO2.

Methods

Mosquitoes

Female mosquitoes used for the laboratory experiments were collected from a culture of Anopheles gambiae s.s. (hereafter referred to as An. gambiae) (Suakoko strain) kept at Wageningen University, The Netherlands. The culture has been reared by blood-feeding on human arms since 1988. Larvae were kept in tap water and fed on Tetramin® baby fish food. Pupae were collected daily and transferred to 30 cm cubic gauze cages for emergence. Adult mosquitoes were kept at 27°C, 80% RH and a photo:scotophase of 12:12 h, respectively. A 6% glucose solution was provided ad libitum on filter paper.

The semi-field experiments were conducted using the Mbita strain of An. gambiae. The mosquitoes have been reared under ambient climatic conditions at insectaries belonging to the Thomas Odhiambo campus of the International Centre of Insect Physiology and Ecology (ICIPE) located at Mbita Point, western Kenya, since 2001. Adult insects were kept in 30 cm cubic gauze cages and provided with a 6% glucose solution ad libitum. Blood feeding took place on human arms. Larvae were kept in filtered water from Lake Victoria and fed on Tetramin® baby fish food. Upon pupation, insects were transferred to adult cages for emergence.

The age of the female mosquitoes used for the laboratory experiments was 5-8 days; the An. gambiae females used for the semi-field experiments were 3-7 days old. The females, previously not blood-fed, were randomly collected from their cage and placed in a release cage (d = 8 cm, h = 20 cm in the laboratory experiments or d = 11-13 cm, h = 15 cm in the semi-field experiments) 16 (laboratory) respectively 8 (semi-field) h before the experiments were started. To prevent dehydration the mosquitoes were offered water-moistened cotton wool on top of the release cage.

Traps

Mosquito Magnet-X counter flow geometry traps (MM-X; American Biophysics Corp., USA, [21], see also [22, 23]), were suspended from metal or wooden stands, with the odour outlet 15 cm above ground level [12, 13]. The bullet-shaped cartridges within the lower end of the odour outlet tube of the traps were removed. The electric ventilators in the MM-X traps operated on 12 V batteries. During the experiments performed in the MalariaSphere [24] also CDC miniature light traps (Model 512; John W. Hock Company, USA, [25]) were used. These traps were run on 6 V batteries (Gaston Battery Industrial Ltd, China). After removing the caught mosquitoes, each trap was cleaned with 10% ethanol.

Odour stimuli

Yeast-produced carbon dioxide was produced by mixing dry yeast (Dr. Oetker, The Netherlands, used in the laboratory experiments carried out in Wageningen or Angel Yeast Co. Ltd., China, used in the semi-field and field experiments in Kenya), sugar (Van Gilse Kristalsuiker, Suiker Unie, The Netherlands, in the laboratory experiments or Sony Sugar, South Nyanza sugar Co. Ltd., Kenya, in the (semi-)field experiments) and tap water [16] in two plastic bottles of 1.5 L or 5 L, connected with each other by silicon tubing, or one plastic container of 25 L. Mixing took place 1-1½ h before mosquitoes were released, at ambient temperature, until the dry yeast was dissolved. No additional stirring or mixing took place during the experiments. A 0.5 L respectively 1 L bottle was put in between the 1.5 L respectively 5 L bottles with the mixtures and the MM-X trap to prevent foam produced by the mixtures entering the trap (Figure 1A-C). Holes were drilled into the original screw caps of the bottles and into the side of the small bottles; silicon tubing (Ø 7 mm; Rubber B.V., The Netherlands) fitted through these holes to connect the bottles. The smaller bottle was connected to the MM-X trap using the original MM-X tubing (micron filter and orifice removed) and the Luer connection at the underside of the trap's top lid. The connections were sealed by Teflon tape and held under water to check for leakage. Several combinations of bottle size and amount of yeast, sugar and water were used. The carbon dioxide output was estimated by measuring the volume of water displaced from a submerged measuring cylinder (Table 1). For this purpose, the tubing that was attached to the MM-X traps during the mosquito trapping experiments was now led into a measuring cylinder which was held in a bucket of water (Figure 1D).

Figure 1
figure 1

Pictures showing the different setups used to apply the yeast-sugar solutions and to measure the CO 2 production. A. Two 1.5 L bottles; B. One 25 L container; C. Two 5 L bottles; D. CO2 production measurement.

Full size image
Table 1 Carbon dioxide flow rate (ml/min) produced by different yeast-sugar solutions
Full size table

Industrial carbon dioxide (≥ 99.9%) was released from pressurized gas cylinders (Linde Gas Benelux B.V., The Netherlands in laboratory experiments or Carbacid Investments Ltd., Kenya, in (semi-)field experiments) and supplied to the MM-X traps through silicon tubing (Ø 7 mm; Rubber B.V., The Netherlands). The Luer connection at the underside of the trap's top lid was used to release the gas directly into the odour outlet tube of the trap. A flow meter (Sho-Rate model GT1350 or GT1355, used in laboratory and semi-field experiments; Brooks Instruments, The Netherlands) or an orifice (American Biophysics Corp., USA; used in field experiments) regulated the flow rate of CO2. During the laboratory experiments, CO2 was led through a 0.5 L bottle before it was released into a MM-X trap. This bottle was filled for 50% with a 10% sugar solution.

Human foot odour was released from nylon socks (40 Den, 100% polyamide, HEMA, The Netherlands) worn by WHS (laboratory experiments) or KJvR (semi-field and field experiments) for 12 h prior to the experiments [6, 12, 13, 26–29]. A clean nylon sock served as a control. Socks were placed along the odour outlet tube of the MM-X trap without blocking the airflow and held in position by odourless tape (3M™ Double Coated Tape 400; used in laboratory experiments) or by a small metal wire (in (semi-)field experiments).

Laboratory experiments

Two MM-X traps were placed in a textile screen cage (330 × 250 × 233 cm; Howitec Netting BV, The Netherlands, [30]) at approximately 2.5 m distance from each other inside a climate-controlled room (22.2 ± 1.6°C and 52.6 ± 7.8% RH). The CO2 cylinder and the yeast-produced CO2 bottles were positioned within the sluice of the cage. Either two 1.5 L bottles or one 25 L container contained the yeast-sugar solution. In each 1.5 L bottle, 7 g of dry yeast and 100 g of refined household sugar were dissolved in 1 L of tap water. In the 25 L container, a mixture of 70 g of dry yeast, 1 kg of sugar and 10 L of tap water was prepared. In contrast to the 1.5 L bottles, which were used during a single experiment only, the 25 L container was used during two consecutive days without adding additional yeast, sugar or water. During the time it was not used, the container was closed and stored at room temperature. Worn socks were used to test the effect of human emanations on the attractiveness of yeast-produced CO2. The flow rate of industrial CO2 was set at 15 ml/min, a flow rate within the range (up to 20 ml/min) that was previously measured to be produced by two 1.5 L bottles each containing a mixture of 7 g of dry yeast, 100 g sugar and 1 L of tap water.

Experiments were conducted in the last 4 h of the dark phase when An. gambiae is normally searching for a blood host [31–33]. For each replicate, 50 mosquitoes were released from the centre of the screen cage and left in it for 4 h. After this period, the release cage and the traps were closed, the mosquitoes killed by freezing, and counted. The dual-choice experiments conducted are listed in Table 3. Treatments were alternated between the two positions to rule out any positional effect. In addition, experiments with two unbaited MM-X traps were conducted to test for positional effects. Each dual-choice experiment was replicated 6-8 times. Surgical gloves were worn by the operator to avoid contamination of equipment with human volatiles.

Semi-field experiments

General

The semi-field experiments were conducted under ambient temperature and humidity (26.6 ± 0.9°C and 92.1 ± 8.9% RH) at the Thomas Odhiambo campus of ICIPE, Mbita Point, Kenya. Each semi-field experiment started at 9:30 pm by connecting the CO2 tubing and powering the traps, followed by releasing the mosquitoes. At 6:30 am the following morning the experiments were terminated by closing the traps and disconnecting the carbon dioxide and power supplies. The MM-X traps and collection bags of the CDC traps were placed in a freezer to kill the caught mosquitoes prior to counting. In addition, at 11 am the number of mosquitoes resting inside the house in the MalariaSphere was determined by way of actively searching for mosquitoes. In dual-choice MM-X experiments treatments were alternated between the two positions to rule out any positional effect. In addition, experiments with two unbaited MM-X traps were conducted to test for positional effects. Surgical gloves were worn to avoid contamination of equipment with human volatiles.

Effect of CO2 flow rate on trap catches

Experiments with industrial CO2 were conducted to establish the minimal CO2 flow rate needed to catch An. gambiae females using MM-X traps. For this purpose, a cage made of mosquito netting (2 × 2 × 6 m) was constructed inside a greenhouse (Cambridge Glass House Co. Ltd., UK) at Mbita Point, western Kenya. The greenhouse had a glass-panelled roof, gauze covered side walls, and sand on the floor [24, 29]. Two MM-X traps were placed at opposite ends of the cage at a distance of approximately 5½ m of each other. Carbon dioxide was provided from a gas cylinder positioned outside the cage. During each experiment the CO2 cylinder was connected to one of the MM-X traps (for details see above). The other MM-X trap was unbaited. Five CO2 flow rates were tested: 25, 60, 100, 250 and 500 ml/min. These flow rates were chosen because they are commonly used to bait traps in mosquito surveillance exercises and/or are close to flow rates previously measured to be produced by the yeast-sugar solutions that had been tested in the laboratory. Each flow rate was tested four times. In each experiment 100 female mosquitoes were released from the centre of the cage.

Effect of yeast-produced CO2 on trap catches

Two MM-X traps were placed in the opposite corners of a screen-walled greenhouse (11.4 × 7.1 × 2.5 m, Cambridge Glass House Co. Ltd.) with a large mosquito-netting cage (10 × 6 × 2.5 m; mesh width 3 mm) suspended from the ceiling to the floor (screen house; [29]). This resulted in a distance of approximately 12½ m between the traps placed at 1½ m from the corner. A CO2 cylinder was placed next to each trap and CO2 was led to the trap using silicon tubing (for details see above). During the experiments traps were either unbaited, baited with industrial or yeast-produced CO2 or/and a worn sock. Industrial CO2 was applied at a flow rate of 100 or 250 ml/min. Yeast-produced CO2 was also applied at two different flow rates, using either a mixture of 17.5 g of dry yeast (Angel), 250 g sugar (Sony) and 2½ L of tap water or 35 g of dry yeast (Angel), 500 g sugar (Sony) and 2½ L of tap water in each 5 L bottle. The flow rates for industrial and yeast-produced CO2 were chosen based on the results obtained in the previously described experiments (see Table 4) and the flow rates measured to be produced by different yeast-sugar solutions (see Table 1), taking into account that temperatures are lower during the night than during the day, resulting in a lower production by the yeast-sugar solution.

In addition, the effectiveness of yeast-produced CO2 was tested 24 h and 48 h after mixing the ingredients. Each dual-choice experiment was done four times, each with 200 female mosquitoes released from the centre of the screen house. See Table 5 for an overview of the experiments performed.

Effect of CO2 baited traps on house entry behaviour

The MalariaSphere described by Knols et al [24] was used to test the potential of MM-X traps baited with either industrial or yeast-produced CO2 to reduce house entry by An. gambiae females [34, 35]. The MalariaSphere consists of a screen-walled greenhouse (11.4 × 7.1 × 2.5 m, Cambridge Glass House Co. Ltd.) in which a traditional African house (3.2 × 2.8 × 1.7 m) has been built and crops planted.

During the experiments, a male African volunteer (aged 27) slept inside the house on a bed, protected by an untreated bed net. Two CDC miniature light traps were hung at a height of 140 cm (bottom at 80 cm) above ground level beside the bed net on the foot-side end of the sleeping volunteer, with its shield touching the side of the bed net [36]. An odour-baited MM-X trap was hung outdoors under the overhanging part of the thatched roof of the house, 15 cm above ground level [12, 13]. Either industrial CO2 at a flow rate of 100 ml/min or yeast-produced CO2 produced by 17.5 g dry yeast (Angel) + 250 g sugar (Sony) + 2½ L tap water in each 5 L bottle was tested. Also the effect of the addition of human emanations to CO2 was examined by putting a worn sock in the MM-X trap (see Table 6). Each treatment was tested six times, and in each experiment 200 female mosquitoes were released 5 m away from the house (Figure 2).

Figure 2
figure 2

Diagram showing the placement of the three traps inside (two CDC traps) and outside (a MM-X trap) an African house during the experiments conducted in the MalariaSphere [24].

Full size image

Field experiments

The field experiments were conducted in Lwanda, a rural village at an altitude of 1169 m above sea level in the basin region of Lake Victoria, Nyanza Province, western Kenya. The area has a main rainy season from March to May and a short rainy season from October to December. Experiments were conducted at the end of the short rainy season, in December 2008. Lwanda has a variety of mosquito breeding habitats [37, 38].

Based on several criteria (household, location of cooking site, roof construction, vegetation around the house and all houses at walking distance from each other) four approximately similar houses in Lwanda were selected. The occupants of the houses were sleeping under bed nets. Each house was provided with a MM-X trap, a car battery and a CO2 cylinder. The MM-X traps were hung outdoors, 15 cm above ground level, under the overhanging thatched roof, at the window side of the house [13]. Vaseline petroleum jelly was used around the tubing, suspension cable and electrical cables to prevent ants from reaching the mosquitoes caught in the MM-X trap.

Two series of each eight nights (i.e. two blocks of a 4 × 4 Latin square) were run. In the first series the following four treatments were tested: unbaited, industrial CO2 at a flow rate of 250 ml/min (the amount of CO2 released by a human, [2]), and yeast-produced CO2 at two different flow rates, using either a mixture of 17.5 g of dry yeast (Angel), 250 g sugar (Sony) and 2½ L of tap water or 35 g of dry yeast (Angel), 500 g sugar (Sony) and 2½ L of tap water in each 5 L bottle. In the second series the effect of the combination of CO2 and human emanations on the trap catches was examined by testing industrial CO2 at a flow rate of 250 ml/min with or without the addition of a worn sock, and yeast-generated CO2 produced by 35 g of dry yeast (Angel), 250 g sugar (Sony) and 2½ L of tap water in each 5 L bottle with or without the addition of a worn sock. Each experiment ran from 8:30 pm until 6:30 am, after which the mosquitoes in the traps were killed by placing the traps in a freezer and counted. Surgical gloves were worn to avoid contamination of equipment with human volatiles.

The mosquitoes caught in each trap during one night were morphologically identified and counted. Culicines were identified to genus, anophelines to species. Female An. gambiae sensu lato mosquitoes were placed in a 2 ml Eppendorf tube with dry silica gel and a piece of cotton wool. These mosquitoes were transported to the Laboratory of Entomology of Wageningen University for species identification. The Bender buffer method [39] was used to extract DNA from a mosquito leg and part of the abdomen of each mosquito, followed by polymerase chain reaction (PCR) analysis [40].

Yeast-produced CO2 concentration measurements

The concentration of yeast-generated CO2 produced by 17.5 g of dry yeast, 250 g sugar and 2½ L of tap water in each 5 L bottle and flowing from a MM-X trap was measured in the laboratory using a Xentra 4100 CO2 analyser (Servomex, The Netherlands). The data were transferred to a PC using Das Wizard 2.0 software (Measuring Computing Corporation, USA). The analyser measured at 1 Hz and was programmed to shift to the next measuring point after 60 successive readings. The average of these 60 readings was plotted in a graph. The analyzer measured over a range of 0-1030 ppm with an accuracy of 0.1 ppm.

Three series of readings, each for a duration of 20 min, were taken at different times after mixing the yeast-sugar solution (1½, 25½ and 49½ h). For each series of readings, three measuring points were positioned at different distances from the MM-X trap (Table 2). To make a comparison with concentrations flowing from a MM-X trap baited with industrial CO2 another three series of readings, at different distances from the MM-X trap, were made (Table 2). For this comparison the human equivalent of CO2 percentage present in breath (5%) and the amount released (250 ml/min) were chosen [1, 2].

Table 2 Position of CO2 measurements; CO2 either produced by a yeast-sugar solution (17.5 g yeast+250 g sugar+2½ L water in each 5 L bottle) or released from a CO2 cylinder (5%, 250 ml/min)
Full size table

Statistical analysis

For each dual-choice test (laboratory and semi-field experiments) a χ2-test was used to test whether the distribution of the total number of mosquitoes caught in the treatment or control trap over all replicates differed from a 1:1 distribution. A χ2-test was also used to compare the total number of mosquitoes found inside (total number caught by the two CDC light traps + found resting inside the house) and caught outside (by the MM-X trap) the house in the MalariaSphere. Effects were considered to be significant when P < 0.05.

Of the mosquitoes caught during the field experiments, the rarely caught male mosquitoes were discarded from the data. Due to many zeros, the numbers of anopheline and Aedes females were transformed (natural logarithm of (x+1)) before subjection to a Generalized Linear Model (GLM; Genstat® release 12.1; Normal distribution, fitted terms: night, house, treatment, and when significant the interaction between house and treatment). Mansonia, Culex and total counts were not transformed before subjection to a GLM (Poisson distribution, linked in log, dispersion estimated to account for heterogeneity, fitted terms: night, house, treatment, and when significant the interaction between house and treatment). Two-sided t-probabilities were calculated to test pairwise differences of means. Effects were considered to be significant when P < 0.05.

Results

Laboratory experiments

Experiments with two unbaited MM-X traps revealed no positional effect within the cage (P = 0.24, n = 297; Table 3). In total, 15.5% of the mosquitoes were caught by the two traps. A trap baited with a worn sock caught significantly more mosquitoes than a trap baited with a clean sock (P < 0.001, n = 277). The two traps together caught on average 41.6% of the mosquitoes that flew out of the release cage.

Table 3 Effect of yeast-produced CO2 on trap catches during laboratory experiments
Full size table

A trap baited with yeast-produced CO2, produced by a mixture of 7 g of dry yeast, 100 g sugar and 1 L of tap water in each 1.5 L bottle, caught significantly more mosquitoes than an unbaited trap (P < 0.001, n = 279). During these experiments, on average 77.0% of the mosquitoes released were caught. Also when the other trap was baited with industrial CO2 (15 ml/min) led through two 1.5 L bottles each filled with 1 L of sugar water, the trap baited with yeast-produced CO2 (two 1.5 L bottles with each 7 g dry yeast + 100 g sugar + 1 L water) caught significantly more mosquitoes (P < 0.001, n = 298, in total 51.6% caught).

Significantly more mosquitoes were caught by traps baited with yeast-produced CO2 combined with a worn sock than traps baited with a worn sock only. This was observed when two 1.5 L bottles each containing 7 g dry yeast+100 g sugar+1 L water were used for the production of yeast-produced CO2 and when one 25 L container with 70 g of dry yeast, 1 kg of sugar and 10 L of tap water was used (P = 0.007, n = 278 respectively P < 0.001, n = 371). In total a mean of 55.5% and 78.8%, respectively, of the mosquitoes that had left the release cage were caught during these experiments.

Semi-field experiments

Effect of CO2 flow rate on trap catches

No positional effects were found in the cage when both traps were left unbaited (P = 0.33, n = 200; Table 4); the two unbaited traps together caught on average 19.0% of the mosquitoes released. A trap baited with 25 ml/min of industrial CO2 caught similar numbers of mosquitoes as an unbaited trap (P = 0.07, n = 400). In total, a mean of 37.5% of the mosquitoes was trapped. Traps baited with industrial CO2 at a flow rate of 60, 100, 250 or 500 ml/min caught significantly more mosquitoes than unbaited traps (P < 0.001, < 0.001, < 0.001 and 0.03, respectively, n = 400). The traps caught on average 30.5, 54.8, 39.5, and 29.5% of the females that left the release cage, respectively.

Table 4 Effect of CO2 flow rate on trap catches during screen house experiments
Full size table

Effect of yeast-produced CO2 on trap catches

Experiments in the screen house with unbaited traps revealed no bias for either side (P = 0.64, n = 800; Table 5). The two unbaited traps together caught only 5.1% of the mosquitoes that were released. A worn sock attracted significantly more mosquitoes than a clean sock (P < 0.001, n = 800); on average 43.1% of the mosquitoes were trapped.

Table 5 Effect of yeast-produced CO2 on trap catches during screen house experiments
Full size table

Significantly more mosquitoes were caught by traps baited with yeast-produced CO2 than unbaited traps, independent of the ratio used for the yeast-sugar solution (P < 0.001, n = 800). Traps baited with yeast-produced CO2 also caught significantly more mosquitoes when tested against traps baited with industrial CO2, independent of the flow rate tested (P < 0.001, n = 800). On average, between 32.8 and 71.3% of the females were caught (Table 5).

As expected, traps baited with the combination of yeast-produced CO2 (17.5 g of dry yeast (Angel), 250 g sugar (Sony) and 2½ L of tap water in each 5L bottle) and a worn sock caught significantly more mosquitoes than unbaited traps (P < 0.001, n = 800), catching in total 53.0% of the released mosquitoes (Table 5). This combination attracted also significantly more mosquitoes than a worn sock alone (P < 0.001, n = 800), resulting in a total trapping efficacy of 79.5%. Significantly fewer mosquitoes were caught by traps baited with a combination of industrial CO2 (100 ml/min) and a worn sock than traps baited with yeast-produced CO2 and a worn sock (P = 0.002, n = 800, in total 75.9%).

Twenty-four hours after mixing the ingredients, significantly more mosquitoes were trapped using yeast-produced CO2 than when no bait was used (P < 0.001, n = 800), catching a mean total of 18.8% (Table 5). However, significantly more mosquitoes were caught by traps baited with industrial CO2 (100 ml/min) than by traps baited with yeast-produced CO2 prepared 24 h before the start of the dual-choice trapping experiments (P < 0.001, n = 800; 78.0%). After 48 h, traps baited with yeast-produced CO2 caught similar numbers of mosquitoes as unbaited traps (P = 0.11, n = 800, 12.3%), and significantly fewer mosquitoes than traps baited with industrial CO2 (P < 0.001, n = 800, 46.5%).

Effect of CO2-baited traps on house entry behaviour

The number of mosquitoes trapped by a MM-X trap baited with yeast-produced CO2 hanging outside the house in the MalariaSphere was significantly higher than the total number of mosquitoes that entered the house when unoccupied (total number caught by the two CDC light traps + found resting inside the house) (P < 0.001, n = 800; Table 6). In total, 58.5% of the mosquitoes that were released were either caught by the three traps (one MM-X, two CDC light traps) or found resting inside the house. In contrast, when the house was occupied by a human sleeping under a bed net, significantly more mosquitoes entered the house than were caught by the yeast-produced CO2-baited MM-X trap (P < 0.001, n = 800). This was also the case when the MM-X trap was baited with industrial CO2 (100 ml/min; P < 0.001, n = 800). Together, 47.8% (yeast-produced), respectively 53.5% (industrial) of the mosquitoes were retrieved, 'outdoors' plus 'indoors'.

Table 6 Effect of CO2 baited traps on house entry behaviour in the MalariaSphere
Full size table

When a worn sock was added to the MM-X trap baited with either yeast-produced or industrial CO2, significantly more mosquitoes were trapped outdoors than caught in the CDC traps and found resting indoors where a human was present (P < 0.001, n = 800): of all mosquitoes trapped, 68.9% (with industrial CO2) to 82.5% (with yeast-produced CO2) were caught in the CO2 + human odour baited MM-X trap. In total, 68.8% (yeast-produced), respectively 73.9% (industrial) of the mosquitoes released were recovered from inside the three traps and the house together.

Field experiments

In the first series of field experiments 392 and in the second series 486 female mosquitoes were caught over eight nights in traps hanging next to the four selected houses. The majority consisted of Mansonia spp. mosquitoes: 48.7% and 66.0% in series 1 and 2, respectively. Also Culex spp. females were caught in high proportions: 34.7% respectively 23.3% of the total number of female mosquitoes found in the traps. Of the anophelines (12.2% and 9.7%, respectively) 3.8% respectively 5.1% were An. gambiae s.l. females. PCR tests revealed that all (except five specimens that could not be identified) of the An. gambiae s.l. specimens were Anopheles arabiensis. The majority of the anophelines were Anopheles coustani females; only a few Anopheles funestus (1%) were found in the traps. In addition, 4.3% respectively 1.0% of the mosquitoes caught were Aedes spp.

GLM analysis showed that both in series 1 and 2 the average number of mosquitoes caught by the four traps hardly varied during the eight nights, whereas the location of the trap (i.e. house) often significantly affected the number of mosquitoes trapped during a night (Table 7). In the first series a significant effect of treatment was found for Culex and Mansiona spp., as well as for Culex spp. in the second series.

Table 7 Mean ± SD mosquitoes caught during field experiments by MM-X traps baited with different test odours
Full size table

In the case of An. gambiae s.l., the effect of the different baits (treatment) on the number of mosquitoes caught depended on the location of the trap (Pinteraction = 0.04 in series 1 and Pinteraction = 0.03 in series 2). In the first series, traps baited with industrial CO2 caught significantly more An. gambiae s.l. than unbaited traps (P = 0.02), but similar numbers as traps baited with yeast-produced CO2 (P = 0.14 and 0.33, respectively; Table 7). The second series of experiments showed that, overall, adding a worn sock to either yeast-produced or industrial CO2 significantly increased the number of mosquitoes caught (P = 0.003 and 0.002, respectively). Traps baited with yeast-produced CO2 plus a worn sock also caught more mosquitoes than industrial CO2 alone (P = 0.02). The majority of the An. gambiae s.l. females were trapped next to house #1 (P < 0.05).

Taking all mosquito species caught during the first series together, unbaited traps caught significantly fewer mosquitoes than odour-baited traps (P < 0.05). Traps baited with yeast-produced CO2 at the lowest flow rate caught significantly fewer mosquitoes than traps baited with yeast-produced CO2 at the highest flow rate and traps baited with industrial CO2 (P = 0.009 and 0.003, respectively). Traps baited with the latter two baits caught similar numbers of mosquitoes (P = 0.74). In the second series the location of the traps determined the total numbers of mosquitoes caught (P = 0.003), independent of treatment.

Yeast-produced CO2 concentration measurements

The carbon dioxide concentrations measured at different distances from a MM-X trap are summarized in Figure 3. It shows clearly the distance effect on the concentration of CO2, the further away from the MM-X trap the lower the CO2 concentration, independent of its source (CO2 cylinder or yeast-sugar solution 1½, 25½ or 49½ h post mixing). Concentrations measured at a distance of 200 cm or at a height of 100 cm were between 400 and 500 ppm. Measurements taken 1½ hours after mixing the yeast-sugar solution, within or close to the trap (0 and 30 cm from the trap, 5 cm above ground level) also showed CO2 levels between 400 and 500 ppm.

Figure 3
figure 3

Diagram summarising industrial and yeast-produced CO 2 concentrations measured at different distances of a MM-X trap. Blue circle: 400-500 ppm; green rectangular: 500-600 ppm; red triangle: > 600 ppm; 1, 2 and 4: 1½, 25½ and 49½ h post mixing the yeast-sugar solution (17.5 g yeast+250 g sugar+2½ L water in each 5 L bottle); C: industrial CO2 (5%, 250 ml/min); A: all (yeast-produced and industrial) CO2 sources.

Full size image

Carbon dioxide concentrations produced by yeast-sugar solutions 25½ or 49½ h post mixing, measured inside or 10 cm below the trap outlet, was higher (600-850 ppm) than when industrial CO2 (5%, 250 ml/min) was used to bait the trap (500-600 ppm). At 30 cm from the trap and 5 cm above ground level, CO2 levels were similar for industrial and yeast-sugar solutions 25½ and 49½ h after mixing (450-550 ppm).

Discussion

Based on the results, CO2, and possibly other volatiles, produced by fermenting baker's yeast appears a promising alternative for industrial CO2 supplied from expensive and cumbersome cylinders to lure An. gambiae females towards traps. Trap catches were similar or even significantly higher when yeast-produced CO2 was used to bait MM-X traps compared to industrial CO2. This finding presents an important step in the development of a cheap and easily applicable CO2 source that could be used for mosquito surveillance or removal in rural settings.

The indoor and semi-field trapping experiments showed that yeast-produced CO2, produced by yeast-sugar solutions in different ratios, significantly increased the number of An. gambiae females caught by MM-X traps. Traps baited with yeast-produced CO2 also caught significantly more mosquitoes than traps baited with industrial CO2 at a similar or probably higher flow rate. Yeast-produced CO2 also significantly increased the catches of traps baited with human odour collected on nylon socks (Tables 3 and 5). These finding are in agreement with previous research with industrial CO2, showing the importance of this compound in the trapping of this mosquito species [5, 11–13].

The experiments conducted in the MalariaSphere revealed that a trap baited with yeast-produced CO2 hanging outdoors next to a house caught significantly more An. gambiae than entered the unoccupied house. This was not the case when a human was sleeping inside the house, regardless of the use of yeast-produced or industrial CO2 as only bait in a trap placed outdoors. However, when adding human foot volatiles to either yeast-produced or industrial CO2, significantly fewer mosquitoes were found inside the occupied house than in the MM-X trap placed under the eaves outdoors (Table 6), proving that the combination of human skin odour + CO2 effectively prevented a large proportion of mosquitoes entering the house. These encouraging results suggest that it is possible to develop traps that can be placed outdoors, baited with CO2 and a synthetic blend mimicking human odour, to reduce the number of malaria mosquitoes entering houses through the eaves. Jawara et al[13], however, showed that in The Gambia human odour-baited traps placed either next to or inside an experimental house did not decrease the number of wild mosquitoes entering the house. Other measures, like house screening or application of repellent odours, to prevent house entry may therefore be necessary to apply in addition to odour-baited traps [41]. Semi-field and field experiments are ongoing to explore this further.

During the field experiments in the present study, mosquito numbers were low and no An. gambiae s.s., the subject of our study, was caught. Its sibling species, An. arabiensis, however, was present and collected significantly more with human skin odour + CO2 than with CO2 alone (Table 7). Similar comparative results for An. gambiae s.s. and An. arabiensis with odour baits were also reported by Okumu et al[42], suggesting that both important malaria vectors can be collected with odour-baited traps. Also, yeast-produced CO2 seems to be as good as industrial CO2 as bait for several other vector and nuisance mosquito species (Table 7).

In the laboratory and screen house significantly more mosquitoes were caught in traps baited with yeast-produced CO2 than in traps baited with industrial CO2 when tested directly against each other. Since the flow rates were either comparable or more likely lower for yeast-produced CO2 (e.g., due to lower temperatures during the night), possible differences in flow rates between industrial and yeast-produced CO2 did not result in differences in attractiveness. It is, however, known that growing yeast produces additional compounds besides CO2[20]. Preliminary analyses of headspaces of yeast-sugar solutions (70 g Y + 1000 g S + 10 L W in 25 L container), two and 28 h post mixing, revealed that yeast produces volatile organic compounds (VOCs) previously found in human emanations and which may therefore play a role in the host-seeking behaviour of An. gambiae s.s. [43–46] (Table 8). These additional VOCs may explain the differences found in catches between traps baited with yeast-produced CO2 compared to traps baited with industrial CO2 and should be further examined.

Table 8 Preliminary data of volatile organic compounds found to be more present in headspace samples of yeast-sugar solutions (2 or 28 h post mixing) than in background samples (order of compounds based on retention time on a DB-5 column)
Full size table

Measurements of CO2 concentrations at different distances from a MM-X trap showed that, at close range of the trap, CO2 concentrations produced by yeast-sugar solutions were higher than from cylinders containing 5% CO2 (equal to the concentration in human breath). Further away from the trap, at 30 cm, concentrations of industrial and yeast-produced CO2 had dropped to a comparable low level (Figure 3). Even though this was measured in a laboratory where no wind was present, it is very likely that also in the field packets of CO2 are produced by yeast-sugar solutions with concentrations similar to or higher than what is produced by humans [47–50]. Since mosquitoes respond to small changes in CO2 concentration above ambient, this will be sufficient to induce upwind flight [1, 50–52].

In Japan and Malaysia, traps baited with dry ice caught more Culex and Aedes mosquitoes than traps baited with yeast-produced CO2[16, 53]. However, the advantages, such as low costs and feasible logistics, of the yeast-method clearly outweigh the logistic disadvantages and relatively high costs associated with both dry ice and CO2 cylinders. Variable CO2 output may occur when using yeast-sugar solutions, probably depending on the ambient temperature. This issue, however, is not problematic, since the current results show that mosquitoes are attracted to yeast-produced CO2, regardless of the concentrations used. In addition, indications have been found that fluctuating concentrations of CO2 above the ambient level induce upwind orientation of mosquitoes [50, 52], although the laboratory and field experiments of the present study indicate that higher concentrations are favourable.

Both laboratory and semi-field experiments showed that yeast-produced CO2 is still 'attractive' 24-34 h post mixing the ingredients (Tables 3 and 5), although less than industrial CO2 (which is released with a constant flow rate and concentration), showing that this bait is at least applicable during one sampling night. In the screen house, yeast-produced CO2 lost its attractiveness somewhere between 34 and 48 h post-mixing the ingredients (Table 5). Carbon dioxide flow rates dropped from 60 ml/min after 30 h to 0 ml/min within 51 h. In contrast, the CO2 measurements showed that even after 49 h CO2 concentrations should be sufficiently high to activate mosquitoes (Figure 3) and simultaneous CO2 output measurements showed a flow rate of 30 ml/min. These differences may have been due to temperature differences or tap water of different sources.

Conclusion

Carbon dioxide and possibly additional volatiles produced by yeast-sugar solutions are attractive to An. gambiae and, therefore, these solutions can be used as baits for the surveillance or possibly removal of this important malaria vector. The results suggest that CO2 is the most important constituent of these VOCs, because addition of human foot volatiles enhanced attraction of mosquitoes similar as with industrial CO2. As long as CO2 production will be sufficient for at least one night, the smaller the bottle and the cheaper and easier accessible the ingredients, the better for implementation in rural areas. This technology could represent a new solution for sampling An. gambiae and other human-biting mosquito species in remote areas, with low financial and technological demands.