The results of this study showed that Anopheles arabiensis is the predominant anopheline species in the area, and it feeds mainly on cattle. An. arabiensis has already developed resistance to the available pyrethroid insecticides and alternative insecticides may be needed for the treatment of cattle. Houses close to the main mosquito breeding site harboured more freshly fed An. Arabiensis and those fed on human blood.
Earlier studies from Ethiopia have examined the blood meal origins of An. arabiensis from animal sheds and human dwellings during the main malaria transmission seasons only[3, 5, 6], neglecting the dry months. A strength of our study is that the blood meal origins of freshly fed An. arabiensis were determined by collecting mosquitoes from outdoor pit shelters and inside houses throughout a year, as was recommended by Garrett-Jones. Mosquitoes were sampled from 30 collection sites every two weeks each month and, hence, their blood meals are representative of human contact with the mosquito vector. Our data compare well with those of Loha and Lindtjørn, who studied the incidence of malaria in the same village and reported the highest incidence of malaria in the nearest village to Lake Abaya (sub-village 3), where we found the highest densities of freshly fed and human-fed An. arabiensis.
One limitation of our study was the inability to determine the cryptic mixed blood meals of malaria vectors that had fed on different individuals of the same species. This might have led to underestimation of human–vector contact and pathogen transmission intensity, as was reported by Norris et al. and Scott and Takken. Another limitation is that we could not identify other animal sources of blood meals for malaria vectors in addition to humans and cattle. Such information may be important in the planning of vector control options. The failure to determine the blood meal origins of some freshly fed An. arabiensis may have occurred because we lacked antibodies for other hosts, or it could have resulted from enzymatic degradation of the blood.
Many zoophilic An. arabiensis were collected indoors using space spray catches after they had fed on cattle outdoors, which provides clear evidence for preference of a bovine blood meal over human. The zoophilic behaviour of An. arabiensis observed in this study is consistent with other findings from Ethiopia[3, 6]. The HBI (38%) of An. arabiensis from space spray catches was lower than the HBI from southern Zambia (92.3%), the Kenyan coast (91%), Konso in southern Ethiopia (55.2%) and the Gambia (82%), but higher than from Eritrea (20%) and western Kenya (23%). The percentage of mixed blood meals for indoor resting An. arabiensis (21.0%) was comparable with that found in other studies[13, 34]. No mixed blood meals were identified in resting An. arabiensis from inside houses in Kenya.
The An. arabiensis collected using CDC light traps had higher HBI than those from indoor resting and outdoor pit shelters. Fornadel et al. reported an HBI of 94% for An. arabiensis from southern Zambia collected using CDC light traps. Interestingly, a high proportion of An. arabiensis from indoor CDC light traps had mixed blood meals (65.2%). This suggests that they were interrupted while feeding outdoors on cattle and moved into houses to complete their feeding in a single night or on consecutive nights[29, 36]. The lowest HBI was found for An. arabiensis from pit shelters located near cattle that are kept outdoors. This reveals that the accessibility of hosts influences the feeding behaviour of this species, as also reported by others. This is the first report of the HBI of An. marshalli and An. garnhami. Future studies should be conducted to examine the sporozoite rate of these species to determine their possible role in malaria transmission.
The few An. funestus collected from outdoor pit shelters was found with cattle blood meal. Unfortunately, we did not identify the species group using molecular method. However, the occurrence of some species from larval identification is known in Ethiopia. Of the members of the group, An. parensis and An. rivulorum are regarded to be zoophilic elsewhere in Africa[39, 40]. An. funestus has been incriminated as an anthropophilic and endophilic malaria vector in many countries in Africa. Therefore, the An. funestus group identified morphologically in this study could be either An. rivulorum or An. parensis or both .
In this study area, the distribution of the malaria vector was seasonal. The maximum number of freshly fed and human blood meal-engorged An. arabiensis was recorded one month after the peak rainfall. Possible reasons are that the rainfall in the previous month may have provided more breeding sites and increased the relative humidity, which contributes to a high density and longevity of the vectors and consequently increases human–vector contact. In particular, the longevity of the vector is crucial for disease transmission because it increases the chance of an infectious bite occurring. Kristan et al. have also shown a one month lag after rainfall as a predictor of vector density in the African highlands. A study from Eritrea also has shown an increase in the An. arabiensis population one month after the start of rainfall. Moreover, the distribution of An. arabiensis was influenced mainly by the location of breeding sites on the shore of Lake Abaya. A study from the same area and one from Northern Tanzania showed a higher risk of malaria infection in a population living near to mosquito breeding sites. To locate and identify households at greater risk of malaria is, therefore, crucial in the planning and implementation of vector control approaches.
An. arabiensis showed a high level of resistance to knockdown and mortality in response to pyrethroid insecticides (deltamethrin, alphacypermethrin, lambdacyhalothrin and cyfluthrin) and DDT. The knockdown resistance was most likely due to the possession of various detoxifying enzymes. Studies from East and Central Africa[45, 46] have reported the occurrence of high levels of mono-oxygenase enzymes in resistant An. arabiensis. Elevated levels of mixed function oxidases and β-esterases were also reported in resistant An. arabiensis in Tanzania. Moreover, the West African kdr mutation (L1014F) detected in high frequencies in South-West and Northern Ethiopian An. arabiensis populations[18, 48] could be another reason for high knockdown resistance in the study area. The KDT50 of lambdacyhalothrin (39 minutes) was higher than that of the other pyrethroid insecticides, but shorter than that reported from Senegal (43.6 minutes) in An. gambiae. Compared with studies from Ethiopia, the KDT50 values of 25.3 minutes for An. arabiensis from Gorgora and 37.6 minutes from Ghibe were higher than that we observed for deltamethrin (21 minutes), but similar to that reported from Sodere (21.9 minutes). The impact of knockdown resistance is that it can allow the vectors to bite humans even inside the long lasting insecticide treated nets (LLINs) because the vector can withstand a long duration of exposure without being knocked down.
The high level of resistance of An. arabiensis to deltamethrin and DDT is not surprising because of the long history of the use of DDT for IRS and the widespread use of deltamethrin for LLINs and IRS, and cross-resistance may occur. The high level of DDT (90%) resistance in An. arabiensis was expected because 60% resistance was reported from South-Western Ethiopia 14 years ago. The mortality rate (10%) due to DDT was slightly higher than that reported byhttp://Yewhalaw and his colleagues[17, 48] but lower than that observed by Balkew et al.[18, 52]. The mortality rate due to deltamethrin (47%) was lower than that observed in other studies in Ethiopia[17, 18, 48].
The resistance of An. arabiensis to alphacypermethrin, lambdacyhalothrin and cyfluthrin was unexpected because they have not been used for vector control. This implies that the use of insecticides with similar modes of action could shorten the duration of efficacy of other insecticides of the same class once resistance has developed in the mosquito population. The most likely explanation is the presence of cross-resistance between insecticides of the same group, which might limit the choice of alternative insecticides for vector control. Cross-resistance between DDT and permethrin has been reported in Ethiopia in An. arabiensis. No information is available in Ethiopia about the resistance of An. arabiensis to alphacypermethrin, lambdacyhalothrin and cyfluthrin. A study from Ghana has shown high survival rates of An. gambiae s.s after exposure to cyfluthrin and lambdacyhalothrin.
The results obtained in this study have implications for vector control. An. arabiensis showed a tendency to feed more frequently on cattle than on humans. In similar settings, Mahande and colleagues and Rowland et al. reported the success of treatment of cattle with pyrethroid insecticides in controlling zoophilic malaria vectors. Moreover, the preference of An. arabiensis to rest indoors after feeding on cattle outdoors in an area that practises indoor-based vector control activities could explain the low efficacy of LLINs and IRS, owing to the resistance of An. arabiensis to pyrethroid insecticides. Previously, N’Guessan et al. reported a low efficacy of LLINs and IRS in areas with resistant malaria vectors. On the other hand, the indoor resting preference of An. arabiensis is an opportunity to use current indoor based antivector strategies because mosquitoes inside houses are easily targeted, but appropriate management of insecticide resistance needs to be implemented.
The possible explanation for the higher HBI and presence of mixed blood meals in An. arabiensis from indoor CDC light traps may be that most people are bitten indoors before they go to bed, or that protection from indoor antivector interventions is reduced by the presence of pyrethroid-resistant An. arabiensis. In the same setting, Loha and Lindtjørn described the personal protection role of LLINs, with no impact on community members who did not use the nets. It is the killing capacity that provides protection from the infectious bites of malaria vectors for people in the community who do not use bed nets. In an area with pyrethroid-resistant malaria vectors, even the combination of LLINs and IRS has a low impact on the prevalence of malaria, and in other settings an increase in malaria cases has been reported. Asidi et al. showed that the treatment of bed nets with pyrethroid insecticides provides additional protection from mosquito bites only if the vectors are susceptible to the chemicals. Our findings also show that the density of freshly fed and human blood-fed An. arabiensis increased in April 2010 despite the mass distribution of bed nets in March 2010. Hence, it is advisable to introduce additional vector control strategies that target a reduction in the entry of blood-searching vectors into houses and diversion to alternative hosts available outdoors. However, we should not underestimate the fact that malaria transmission can occur outdoors via human-biting mosquitoes, even if the HBI is low.
In addition, the finding of the lowest HBI and percentage of mixed blood meals in An. arabiensis from outdoor pit shelters suggests that An. arabiensis is less likely to leave houses after feeding indoors on humans, or that people are bitten outdoors less frequently in the area. Therefore, IRS and LLINs can provide successful protection from malaria infection if the vectors are susceptible to the available pyrethroid insecticides.