These are the first experimental data describing the distribution of mosquito activity at a human-occupied bed net. Using a simple technique, results demonstrated that the majority of An. gambiae approach the host and land at the top surface of a human-baited bed net, and that this activity is localized within a relatively discrete area directly above the torso and head of the bed net occupant. The observed behaviour was consistent across studies carried out in two laboratories, with different experimental set-ups, and with multiple human volunteers.
While bias arising from the experimental procedures was a concern, especially the possibility that the horizontal top surface might have caught disproportionately more mosquitoes than the vertical side surfaces, three key findings indicated that this was unlikely. First, in the absence of host bait, significantly fewer mosquitoes were caught on the top surface of a standard net (Figure 5A). Second, the pitched net experiment eliminated any potential influence of surface orientation yet revealed a behaviour pattern equivalent to that on standard nets. Third, observation studies indicated that the number and frequency of visits were similar at both vertical and horizontal surfaces (Table 1) and that horizontal surfaces were less effective in capturing mosquitoes. The fact that significantly more horizontal than vertical contacts were needed before capture occurred, suggests the results might even have underestimated activity at the top of the net.
It is proposed therefore that the activity described is a true representation of An. gambiae behaviour. As such, this provides evidence for the earlier proposition that mosquitoes are attracted to the top surface of a bed net by a hypothetical rising plume of putative attractants emanating from the prone human sleeper  and funnelled upwards by the net itself . The exact source of these attractants remains elusive. Studies with seated human hosts suggested that foot odour was a key attractant for An. gambiae, though Dekker et al. later showed that when the subject was lying down with the legs raised, significantly fewer mosquitoes bit the legs and feet compared to the rest of the body . More recent studies suggested that body sweat was attractive [17–19] while breath had both attractant and repellent components . Since nothing is known of how these different odours disperse or mix after emission by a prone human, it is impossible to draw conclusions as to which is the most important cue from the results presented here. Nonetheless, the remarkable ‘focus’ of behavioural activity within a discrete area directly over the sleeper’s head and torso has never been reported previously and offers potential for further exploration of host attractants as well as for exploitation in the design of LLINs.
Interestingly, mosquito release height or location did not significantly affect the distribution of mosquitoes caught on the sticky-nets, suggesting that arrival at the host by An. gambiae might be similar whether mosquitoes enter through windows, doors or eaves. However, the sticky net’s inability to capture mosquitoes on first contact clearly was a limitation because more detailed investigation of arrival patterns and foraging behaviour at the bed net-human interface were not possible. Further work is needed to investigate these important events.
A next step from this study is to measure responses of this and other mosquito species to commercially available LLINs, to determine if arrival patterns are similar on the various insecticide-treated bed nets currently available. This is particularly important for two-in-one type nets. For example, where pyrethroid-treated side netting is combined with a non-pyrethroid treated top surface: preferential contact with the top surface would expose vectors to the non-pyrethroid (the desired effect); but if activity predominated at the sides, two-in-one nets would perform no differently to existing LLINs. In a worst-case scenario, infrequent, limited, sublethal dose exposure to the insecticide on top might even promote resistance.