Malaria in the highlands of western Kenya is characterized by unstable transmission that is closely related the ecosystem type and weather variability. The two ecosystems (U and V-shaped) have different vulnerabilities to malaria epidemics. Current interventions using long lasting insecticide treated nets (LLINs) and in some sites indoor residual spraying (IRS) are suppressing vector populations and also selecting the less anthropophilic member of the An. gambiae sl. complex, An. arabiensis. Decreasing transmission due to reduced vector abundance and dominance of less efficient species will result in reduced exposure to malaria parasites and fewer individuals with immunity to malaria  with subsequent increased vulnerability to epidemics and severe malaria. These changes in transmission will require careful monitoring and surveillance as part of early detection of transmission risks and the evolution of epidemics.
Heavy rains associated with the 2009/2010 El Nino event increased vector population in the entire study area resulting in peaks in January and February 2010. Plasmodium falciparum sporozoites were detected in December and January at Iguhu and February and March at Emutete and none in the V-shaped ecosystems (Marani and Fort Ternan). However, in October 2009 the prevalence of antibodies increased in all the sites suggesting exposure of the population to the parasite. In Emutete the peak of antibody prevalence was observed in December 2009 while sporozoites were first detected in February 2010. Analysis of the relationship between the indoor resting densities of female An. gambiae and the prevalence of antibodies revealed a significant correlation between the two variables at Iguhu and Fort Ternan but not at Emutete and Marani. This may have resulted from under-sampling of the vectors in the former sites. The human population sampled had a wider dispersion compared to the houses sampled for vectors.
Assessment of the abundance showed that the U-shaped ecosystem had 3-fold higher vector densities in the V-shaped valleys and this is supported by the higher prevalence of antibodies in the U-shaped ecosystem.
These results indicate that surveillance of CSP-MSP antibodies provided an earlier and more responsive indicator to malaria transmission than EIR. The rapid diagnostic kit (RDT) is cheap and easy to use. A large human population can be assessed in a very short time and results are available immediately. The RDT provides a new opportunity for early detection of hyper-transmission and may complement results of the climate based early malaria epidemic prediction models .
In 1998, An. gambiae was the only member of the An. gambiae complex reported in the western Kenya highlands at 1500 m asl . In another study An. arabiensis was not detected in areas above 1400 m asl in the western Kenya highlands . From June 2003-June 2004; no An. arabiensis were detected at Marani while the proportion of this species at Iguhu was 0.8% . Our results indicate that the proportion of An. arabiensis at Iguhu has increased to 6.7% and at Marani to 3.6%. More remarkable was the high proportion of this species that was observed at Fort Ternan (66.7%). The ownership of LLINs in Marani and Fort Ternan are similar and therefore difference in the LLIN ownership cannot explain the differences in the proportion of An. arabiensis observed in the two sites. A contemporary Larval ecology study at Fort Ternan reported a 71% proportion of An. arabiensis in the An. gambiae s.l. population at Fort Ternan . In the lowlands of western Kenya An, gambiae has been largely replaced by An. arabiensis and this has been attributed to the wide scale use of LLINs [9, 10]. The continued use of LLINs in the western Kenya highlands and continued warming due to climate change will favor the more zoophilic and possibly exophilic An. arabiensis. While this species is less amenable to vector control using insecticides it is a much less efficient vector. It is capable of maintaining low level transmission but has been known to be the only vector causing epidemics particularly in the semi arid ecosystems. To date An. arabiensis is the only reported vector in the central Kenya highlands .
While the CSP-MSP antibody profiles in the human population were very similar in Marani and Fort Ternan, the vector profile was not similar with For Ternan. This is indicative of lower indoor resting densities. One possibility is that a large population of An. arabiensis females in Fort Ternan was not resting indoors.