Despite control efforts, malaria remains a leading cause of worldwide morbidity and mortality [1], [2]. The rate of contact between vertebrate hosts and mosquito Anopheles vectors has long been recognised as a crucial determinant of malaria transmission [3]–[5], and successful malaria control depends on understanding the interactions between mosquitoes and humans (e.g. [6]–[9]). Predictions of malaria transmission usually assume that all individuals are at equal risk from Anopheles vector bites. However, there is now strong evidence that humans vary in their attractiveness to malaria mosquitoes [10]–[16] and hence, that host-vector contact is far from random [17], [18]. Recent studies have highlighted the importance of heterogeneous biting in determining the prevalence of malaria infection [19], [20], with one example showing that 20% of individuals account for 80% of all infections among African children [19]. Therefore it becomes of great strategic importance to identify the cause of variation in human attractiveness, and to develop malaria control targeting those who are bitten the most [19], [20].

Anopheles gambiae Giles sensu stricto (henceforth An. gambiae) is the primary malaria vector in Africa [21]. The tremendous vectorial capacity of this species is mainly determined by its strong preference for feeding on humans [22]. Besides factors such as heat and moisture [23], females of An. gambiae locate and orientate toward their human host primarily through olfactory cues [24], [25]. Each person has a distinctive body odour that results from the emission of several hundred volatile organic compounds present in the breath and produced by the skin (gland secretions in interaction with resident skin bacteria) [26], [27]. Additional factors such as diet, general health condition, or reproductive status can also act upon this distinct odour signature and determine the odour profile of an individual [26], [28], [29]. Because of their strong effects on odours, these factors have been considered as causes of the observed variation in human attractiveness to malaria mosquitoes. For instance, pregnant women are twice as attractive to Anopheles vectors as their non-pregnant counterparts and are thus at a greater risk for malaria [30]–[32].

Despite potential consequences on exposure to malaria mosquitoes, there is a serious lack of empirical evidence describing how diet affects human attractiveness to disease vectors. While one study suggested that people are more attractive to laboratory bred Aedes mosquitoes after beer consumption [33], the effect of beer consumption on attractiveness to malaria mosquitoes from natural populations remains untested. As alcohol consumption is rising in most endemic malaria areas [34], it is becoming urgent to assess its effects on human attractiveness to malaria vectors.

By using an experimental setting designed to accommodate entire body odour and breath as stimuli, we investigated the human attractiveness to a natural population of An. gambiae before and after beer or water consumption in a malaria endemic area in south-western Burkina Faso (West Africa).

Beer consumption, as opposed to water consumption, significantly increased both the activation and orientation of An. gambiae. Beer consumption (‘After Beer’ (AB) treatment) activated significantly more of the mosquitoes (47%, GLMM; Odds Ratio (OR) = 1.63; 95% Confidence Interval (CI) = [1.41, 1.89]; P<0.001) than the three other treatments, ‘Before Beer’ (BB, 35%), ‘Before Water’ (BW, 37%) and ‘After Water’ (AW, 38%) (figure 2A). The proportion of activated mosquitoes retrieved from the volunteers' odour-baited trap was 50, 53 and 47% for the treatments BB, BW and AW respectively, indicating that mosquitoes did not orientate preferentially towards human odours relative to outdoor air (figure 2B). In contrast, 65% of the An. gambiae flying upwind, orientated toward the odour trap after the volunteers consumed beer, indicating a significant increase in orientation following beer consumption (OR = 1.77; CI = [1.36, 2.30]; P<0.001; figure 2B).

Axillary temperature decreased slightly after alcohol consumption (starting mean ± SE  = 36.3±0.09 C° to 36.1±0.08 C°; paired t-test, t = 3.2, df = 23, P = 0.004) and water consumption (from 36.3±0.11 C° to 36.2±0.11 C°; t = 2.2, df = 16, P = 0.045), but volunteer temperature did not affect mosquito activation (OR = 1.15; CI = [0.7, 1.5]; P = 0.7) or orientation (OR = 1.05; CI = [0.95, 1.15]; P = 0.6). The mean carbon dioxide CO₂ concentration of outdoor air was 346±4 ppm. No significant difference in mean CO₂ concentration exhaled by volunteers was found before and after beer (582±37 ppm and 568±38 ppm) or water consumptions (563±38 ppm and 589±42 ppm). The level of CO₂ exhaled by volunteers in the Y-olfactometer had no effect on mosquito activation (OR = 1; CI = [0.99, 1.1]; P = 0.9), but higher levels of exhaled CO₂ were associated with a lower degree of orientation (OR = 0.998; CI = [0.997, 0.999]; P<0.001). These findings indicate that the increased human attractiveness observed following beer consumption cannot be explained by increased carbon dioxide emission or body temperature.

The number of mosquitoes retrieved from the two traps increased over the evening, indicating an effect of release time on mosquito activation (OR = 1.25; CI = [1.14, 1.36]; P<0.001). Release time did not alter mosquito orientation (OR = 1.1; CI = [0.9, 1.32]; P = 0.3) and volunteer position (i.e. whether volunteer odour was released from the left or right arm of the olfactometer) did not affect mosquito activation (OR = 1.07; CI = [0.85, 1.29]; P = 0.57) or orientation (OR = 1.05; CI = [0.66, 1.44]; P = 0.8).

Activation and orientation varied with the volunteer tested (figure 3). There was no relationship between activation and orientation (before drink consumption: r² = −0.016, P = 0.56; and after dolo consumption: r² = 0.07, P = 0.11; or after water consumption: r² = −0.05, P = 0.7). Volunteers that induced high mosquito activation did not systematically induce high mosquito orientation.

Individual volunteers' odours affected mosquito behaviour consistently across trials (activation: r² = 0.31, P<0.001, figure 4A; orientation: r² = 0.37, P<0.001, figure 4B). Volunteers that induced high activation or orientation before drinking also induced high activation or orientation after drinking.

Overall, our findings indicate that, despite the individual differences of volunteers, beer consumption consistently increased volunteers' attractiveness to mosquitoes.

We explored the effect of beer consumption on human attractiveness to a natural population of An. gambiae using a Y-olfactometer designed to accommodate total body emanations as a source of odour stimuli. We found that beer consumption not only enhanced the number of mosquitoes that engage in odour-mediated upwind flight (mosquito activation) but also enhanced the strength of their odour-mediated anemotactic response (mosquito orientation). Mosquito activation and orientation are important parts of the natural host-seeking process of An. gambiae and any increases in these behaviours will facilitate vector-human contacts [23], [24]. Water consumption did not affect these mosquito behavioural responses, demonstrating that beer was solely responsible for increased human attractiveness. To the best of our knowledge, this study provides the first evidence that beer consumption increases human attractiveness to An. gambiae, which is the principal vector of malaria in Africa.

The proximate reasons for why people are more attractive after beer consumption are currently unclear. The increased attractiveness of pregnant women has been attributed to increased body temperatures and exhaled breath [30]. Here, higher body temperatures were not associated with higher attractiveness and beer consumption actually resulted in decreased body temperatures. Our results also indicate that increased exhaled breath cannot account for the observed increase in human attractiveness following beer consumption. Higher levels of exhaled breath (as measured with a CO₂ analyser) were not associated with higher attractiveness and beer consumption did not affect CO₂ expiration rate. We postulate that the metabolism of alcohol following beer consumption induces changes in breath and odour markers (i.e. increases the production of kairomones such as 1-octen-3-ol) that increases attractiveness to An. gambiae. Beyond this coincidental side effect of beer consumption, mosquitoes may have evolved preferences for people who recently consumed beer - possibly due to reduced host defensive behaviours or highly nutritious blood-meals. This hypothesis is appealing but requires further investigations.

Although beer consumption significantly increased the volunteer attractiveness relative to the outdoor air control (mosquito orientation), the absence of orientational bias toward volunteer odours in the three other treatments (BB, BW and AW) is intriguing. Similar results have been recently obtained using the same methodology and mosquito population [39]. In field situations, odour-mediated anemotaxis is an effective strategy to locate a vertebrate. Upon arrival in the vicinity of the host, additional cues such as warm, moist convective currents and host movement are exploited by the insect to orientate toward the host. Therefore, the absence of orientational bias for the volunteer odour-baited trap may have stemmed from the fact that our bioassay removes some of these host-related short-range stimuli.

Unsurprisingly An. gambiae activation increased over the evening (17:00 to 21:30). Like most anopheline species, An. gambiae is a night biter. In this species, the biting cycle starts at sunset and rises to peak between 24:00 and 1:00 [24]. Thus, the positive relationship between mosquito activation and time simply reflects the natural circadian flight activity, the process by which hungry females of An. gambiae engage in non-oriented flight to optimise their chance of encountering host stimuli [44]. Upon contact with odour stimuli, An. gambiae then switches from this appetitive search to the actual host location behaviours.

We found that higher levels of expired CO₂ were associated with lower mosquito orientation. This result is in apparent contradiction with the general acceptance that blood-feeding insects are attracted to carbon dioxide. Despite thorough investigations, the role of CO₂ in host-finding by An. gambiae mosquitoes remains equivocal [24]. At least three lines of arguments can be advanced to resolve this apparent contradiction. First, it is increasingly recognized that while CO₂, a compound exhaled by all mammals, induces mosquito activation, it does not provide accurate orientational cues to the anthropophilic An. gambiae [24], [42]. Second, compounds which are termed attractants can also act as repellents at high concentrations [45]. In natural conditions, human exhalations are dispersed and strongly diluted before contacting host-seeking mosquitoes [42]. Accordingly, the CO₂ concentration released in the volunteer trap (575 ppm on average) may have been too strong, resulting in a negative effect on mosquito orientation [15]. Finally, the fine-scale structure of the CO₂ plume (continuous vs. pulsed stream) is known to affect the orientation, with a continuous plume reducing the behavioural responses of mosquitoes [24], [46]. We do not know the structure of the CO₂ plume in this bioassay.

Alcohol consumption is a widespread phenomenon throughout the world and represents one of the most pressing global health priorities [47]. The alcoholic beverage used in this experiment is a very popular drink in West Africa [36]. Therefore, the increased attractiveness following beer consumption found here raises crucial issues regarding strategic planning for malaria control. Recent models have stressed that local malaria control can only be reached if people who are bitten the most can be identified [20]. By ascertaining beer consumption as a risk factor, our study has identified a potential underlying cause of heterogeneous biting, and hence provides insights into the feasibility of targeted interventions.

The outlook may be even worse if we consider that alcohol contributes substantially to the global burden of diseases [48], especially by compromising the host immune defence against parasites. Numerous studies have demonstrated that moderate and chronic alcohol consumptions can have strong immunosuppressive effects [49]. Therefore, people who drink beer are not only at higher risk of exposure to malaria mosquitoes but could also be more vulnerable to the Plasmodium parasites. Given the importance of beer consumption in the populations that are most at risk from malaria, this is a possibility that requires attention.

To eliminate the possibility that other active ingredients in beer apart from alcohol could be driving the observed effects, future studies are needed to test whether consumption of other alcoholic beverages also increases the risk of being bitten by An. gambiae, and hence being infected with malaria parasites in natural situations. Finally, it is crucial to investigate the effect of alcohol consumption on the success of gametocyte (the Plasmodium infective stage for mosquitoes) production. Expanding experiments and observations on the attractiveness of people consuming alcohol to malaria mosquitoes and Plasmodium sp. development in these hosts should allow us to gauge the role of alcohol consumption on the transmission dynamic of malaria.