Anopheles arabiensis was the main vector of malaria in the study area and feeds readily on both humans and domestic animals. Malaria transmission by this species was largely undetectable throughout much the study period except in a few wet months when low levels of transmission were detected. Suggesting that transmission of this disease in arid and semi-arid areas of Africa is mainly seasonal. The finding that An. arabiensis is the only sibling species of An. gambiae s.l. present in the study area is consistent with previous cytogenetic studies [22–25]. Some An. funestus were collected during the study period. However, this species is known to thrive well in hot and humid environments as opposed to the hot and dry conditions [26–29]. It is also possible that lack of suitable long-lasting habitats preferred by this species may partly account for the low densities witnessed in the current study . Previous invesigations in the study area reported production of An. arabiensis in Kamarimar is sustained by permanent or semi-permanent larval habitats that included pan dams, marshes and adjoining drainage canals used for irrigation that are less dependent on rainfall .
A significant fraction of anthropophagic mosquitoes were sampled outdoors from the two villages, a clear evidence of house-exiting by vectors after feeding on humans. Conversely, large fractions of zoophilic mosquitoes were collected indoors illustrating the potential of employing zooprophylaxis as a strategy as has happened in some parts of Africa, Europe and USA . The role of human activities in increasing human-vector contact cannot, however be ruled out. Lightly-dressed residents in the study villages stay out late in the evening to irrigate their farms before temperatures sky-rocket during day time, exposing themselves to mosquito bites. Herding is a common practice in Baringo and indeed, most semi-arid areas of Africa. The largest communal grazing field in the study areas is located in Tirion and is used by hundreds of pastoralists from different villages. Increasing herd sizes could be a plausible vector control strategy but may be counterproductive under certain circumstances where high livestock densities lead to an increase in vector densities and high human biting rates [31, 32].
The intensity of malaria transmission by An. arabiensis as measured by EIR was extremely low and seasonal. This could be due to low infectiousness of the human population in the area or because a majority of mosquitoes caught by light traps were newly emerged and had not had an opportunity to acquire an infectious blood meal. Further, it is also likely that adult lifespan of this species was shorter than the extrinsic incubation period for malaria parasites but this possibility was not evaluated in the current study. These findings corroborate those from other semi-arid areas of Africa in Sudan and Eritrea. Shililu and others  established that the risk of exposure to sporozoite-laden An. arabiensis in Eritrea was highly heterogeneous and seasonal. Biting rates and EIRs peaked during the rainy season, but little or no transmission occurred during the dry season. Year-long studies conducted in the Sudan by Dukeen and Omer  in the 1970s showed An. arabiensis adult densities peaking seasonally during certain months when Nile water levels were low, and rising when the water levels rose.
Entry and/or exit of mosquitoes largely depended on house type. The average number of indoor resting mosquitoes collected in both villages varied significantly depending on the type of house occupied by potential human hosts, highlighting the importance of this micro-epidemiological factor in malaria transmission. There was a strong preference for grass-thatched houses, making house modification to limit mosquito flight into houses a plausible control strategy. This finding reinforces the common belief that poverty is a major driver of malaria transmission in Africa. Most communities in the continent's rural and resource constrained areas are largely unable to afford decent housing with adequate screening measures to block mosquito entry into houses.
It is encouraging that an EIR value was realized in only one out of the 22 months in which sampling was done. This is a positive outcome in the sense that increasing uptake of malaria prevention measures through integrated vector management programs could lower the current prevailing EIR levels to below one infective bite per person per month. Such programs have the potential of eliminating malaria in these areas by reducing parasite rates to levels that can interrupt malaria transmission. We believe these findings have important implications and will inform policy on vector control in epidemiological zones of the world where low seasonal malaria transmission patterns are experienced.