The high decrease of entry rate (42.85 à 100%) of An. gambiae s.l. natural populations into bedrooms, could be explained by the large scale of IRS with bendiocarb that created a stressful environment, which in turn could lead to impressive reduction of entry rate of mosquitoes. This is consistent with previous studies that showed that the unpleasant atmosphere created by the presence of bendiocarb on the walls inside houses is harmful to the mosquitoes and leads to an increase in the exit rate . Conversely, the strong irritant effects that we observed during the evaluation in experimental huts  for deltamethrin, showed reduction of entry rate was less effective in LLITNs areas. This difference in efficacy could be due to the mass community effect of bendiocarb used on a large scale. Indeed, in IRS areas, 2,623 kg of bendiocarb were sprayed in 142,814 bedrooms during the first intervention and 2,751 kg in 156,233 bedrooms during the second intervention; 90 to 100% of bedrooms were treated, according to RTI (Research Triangle International) who carried out the spraying operation. This is consistent with previous studies that showed that community-wide use of insecticide-treated bednets (ITBN) engenders a mass effect . The data analyzed on a cohort of children, revealed for those not using ITBNs, an increasing level of ITBN usage within the area surrounding each child was associated with a decreased risk of developing malaria. This effect was significant in areas at distances of up to 1.5 km away from each child . In addition, some beneficiaries of LLITNs do not use them, but continue to use the untreated nets they had before. Prior to free distribution of LLITNs, it was shown that heat, choking, beliefs and taboos seem to be barriers to the use of bednets in the study area . It is also possibly due to their poor living standards. People sell LLITNs to address other problems, as was the case in trials sponsored by The WHO in the Congo and Tanzania . For LLITNs to be fully effective, requires that community members are actively involved in the process, to ensure that nets are used even during seasons when such use is unpleasant because of the heat and insect bites do not seem numerous enough to justify it .
The endophily rate of An. gambiae s.l. observed before the IRS and LLITN interventions corroborates previous reports of anopheline behaviour [23, 24]. However, after the IRS and LLITN interventions a dramatic decrease of the endophily rate was observed in IRS area. This could be explained by a strong decrease in the proportion of gravid and half gravid mosquitoes, according to Table 3. The unpleasant atmosphere created by the presence of bendiocarb on the walls inside houses was harmful to the mosquitoes and might be the cause of this shift in behavior. Furthermore, the impressive reduction of entry rate of An. gambiae s.l. could justify this shift. Despite this deterrent effect a low proportion of An. gambiae s.l. enter bedrooms. But once on the walls, they absorb the bendiocarb which kills them and they do not have time to bite and to rest inside to digest their blood meal. Indeed, other studies previously conducted [10, 25, 26] have shown the effectiveness of alternative insecticides such as carbamates to control An. gambiae s.l. resistant to pyrethroids. Furthermore, the present study confirms the absence of An. gambiae s.l. resistance to bendiocarb in southern Benin [16, 27]. Conversely, An. gambiae s.l. resistance to pyrethroids [18, 25, 28] was corroborated and could justify the difference in efficacy of LLITNs impregnated with deltamethrin compared to that using IRS based on bendiocarb.
The findings have also demonstrated that large scale IRS can alter An. gambiae s.l. populations and reduce the epidemiological importance of indoor-biting mosquitoes. This decrease was also observed in the LLITNs area, but in a lower proportion. This is consistent with others studies showing that IRS  and ITNs in Somalia  and Tanzania  can reduce the mean density, survival, infectiousness and fitness of mosquito populations.
Conversely, in the control area An. gambiae s.l. remained endophagic in the period after intervention, whereas it has been more exophagic in intervention areas. In this context, the human-biting behaviour of vectors in Oueme region appears to be independent of population density for these species . Nevertheless, a very plausible case  is presented that correlates community-wide ITN use with significant changes in the biting profile of the principal malaria vectors. This indicates that factors relating to locality and seasonal climatic variations would have little effect on shifts of behaviour of taking human blood.
The use of vector control tools and behaviors of the host would be the main factors that modify the behavior of sucking human blood observed on An. gambiae s.l.. Indeed recent studies [33, 34] showed that the long-term indoor application of residual insecticides contributes towards an increased tendency for outdoor feeding among malaria vector populations. This is expected to erode the efficacy of malaria vector control interventions over time, much as increased insecticide resistance would .
Despite the effectiveness of bendiocarb used for IRS, it has the disadvantage of having a short residual effect [10, 36]. In this case it appears that the LLITNs, although providing modest efficacy against pyrethroid resistant An. gambiae s.l., are necessary to supplement IRS because of the long duration of the action of deltamethrin and the role as a mechanical barrier played by this tool [17, 22], against mosquitoes. Regardless of a shift in host seeking behaviour of An. gambiae s.l., other possibilities for outdoor anti-vector interventions need to be explored, in combination with ongoing IRS and LLITN distribution because of the short residual effect mentioned above.