The present study clearly revealed, in the absence of organized vector control activities, a dramatic temporal decline in the density of malaria mosquito vectors in a SSA rural area with a previous history of high P. falciparum malaria endemicity. The results indicated that the indoor density of the two main vectors of P. falciparum malaria, An. gambiae and An. funestus, declined considerably during both study periods, which covered a total time span of more than 9 years. The overall decline in the number of Anopheline mosquitoes was within the range of 55-77% over the first study period (1998-2001), whereas the trend continued and became more pronounced over the subsequent period (2003-2009). Hence, during the last year of mosquito collection in Kirare, the annual numbers of collected An. gambiae and An. funestus, as compared to 2004, had declined by more than 99%. Despite the annual employment of more than 2300 traps, a total of only 14 Anopheline mosquitoes were collected in 2009. A valid comparison of the two study periods should be performed with caution, since the data were collected in different communities, which, despite their relative proximity to each other may differ in terms of local factors of relevance for mosquito vector capacity and population dynamics, including variations in quality and number of breeding sites, microclimatic conditions, housing structures or environmental factors.
The P. falciparum infection rates in humans were not assessed in the study communities during the two study periods. However, an almost complete absence of indoor exposure to Anopheline mosquitoes, as observed in Kirare in 2009, would most likely result in a rapid and pronounced decline in intensity of transmission, and thereby the prevalence of malaria infection, morbidity and mortality among the study inhabitants.
Recent reports and publications have described pronounced reductions in the prevalence of P. falciparum infections and malaria-related disease burdens among inhabitants in several African countries over the past decade [1, 4–7]. Interestingly, the study from Pemba, Tanzania, an island in close proximity to Tanga Region, suggested that malaria transmission had started to decrease prior to the onset of a control programme implemented during the period 2003-2005 . Of specific relevance to the present study are also longitudinal investigations carried out in two rural communities (lowland and highland) in Korogwe District, Tanga Region, Tanzania - an area located less than 70 kilometre from Masaika and Kirare . During the period 2003-2007, the P. falciparum prevalence decreased dramatically from 78.4% to 24.0% in the lowland community, and from 24.7% to 6.5% in the highland community, respectively. Likewise, the incidence of febrile malaria episodes in the two communities decreased by almost 85% during the same period. Temporal changes in mosquito vector densities were not assessed by the authors and specific reasons for the observed decline in P. falciparum infections and related morbidity could not be identified. When considering the proximity to the communities included in the present study, the results suggest that it is likely that the decline in malaria infection and disease burden in those communities could have been a result of a decline in the Anopheline population of perhaps a similar magnitude as observed in the present study. If this assumption is correct, the findings suggest that at least in some areas in Africa, the decline in prevalence of human P. falciparum infections and malaria-related disease is driven by a natural decline in the occurrence of mosquito vectors. In tropical Africa, where mosquito abundance often follows the rainy season, precipitation can be regarded as the primary climatic determinant for variation in Anopheline population size. However, the results suggest that this linkage to rainfall may be related in a complex fashion to the patterns of rainfall experienced in a location as well as the population dynamics and breeding habits of the vector species in question. Thus, while declines in An. gambiae numbers in Masaika were related to a declining trend in mean monthly rainfall, variations in none of the Kirare mosquito populations appeared to be related to changes in mean monthly rainfall. The present analysis has highlighted that one intriguing possibility that may underlie the decrease in An. gambiae populations in particular may arise as an outcome of the change from a regular seasonal rainfall pattern as observed in period 1 (Masaika) to a more noisy or variable temporal pattern observed in Kirare. This suggests that climate change leading to the disruption of stable seasonal rainfall in a location could result in highly dynamic vector population dynamics that might increase the probability of mosquito extinction .
This conclusion appears to be further supported by the fact that there has been no organized attempt to control mosquitoes using insecticides or environmental management in the two study communities. During the first study period only very few inhabitants in Masaika used a bed net, while in Kirare, although bed nets were rarely used in the beginning, a survey carried out after the end of the study indicated that the prevalence of bed nets reached 27% and some inhabitants also used ITNs (6.6%). The use of ITNs may have resulted in some decline in the indoor density of Anopheline mosquitoes and could have triggered a change in biting/resting behavior including a shift from an indoor to an outdoor environment . The sibling species composition within the An. gambiae complex and the An. funestus group was not investigated in the present study. It has been suggested that a specific bed net coverage will result in a higher reduction in transmission by An. gambiae ss as compared to reduction in transmission by the more zoophagic and exophilic An. arabiensis . However, even if all the An. gambiae in the present study were senso stricto, it is not likely that ITNs used by less than 7% of inhabitants (as seen during the last year of the study) would result in a 99% decrease in the indoor Anopheline mosquito populations .
The decline in mosquito numbers may also be a consequence of changes in socio-ecological conditions in the study area (e.g. changes in temperature, ability for water to pool, deforestation or land-use , change in the use of agricultural pesticides or insecticide-like compounds not directly applied for targeting malaria vectors , improved house constructions  or changes related to agricultural activities [28, 29]). An increase in predatorily pressure on the mosquito population  (e.g. birds or invertebrates) or an insect pathogen that specifically targeted mosquitoes, e.g. a bacterial , viral  or fungi infection , could also potentially have induced the observed declines. Regardless of the actual cause, the marked decline in the density of Anophelines followed by their almost complete absence at the end of the second study period suggests that the causing factor or factors have placed an extreme pressure on these mosquitoes.
It is interesting to notice that in neither of the two study communities was the population of Cx. quinquefasciatus affected with the same magnitude as compared to the two species of Anopheline mosquitoes. This difference suggests that either the Culex mosquitoes were less sensitive to the causing factor due to direct species-specific differences or that their pre-dominant breeding sites (organic polluted water bodies, pit latrines) were less affected by the changes as compared to the breeding sites of the Anopheline mosquitoes.
Assuming that the intensity of P. falciparum transmission has decreased to very low levels within the last five years, it is possible that an increasing proportion of inhabitants in this study area have only been marginally exposed to infections during this period. This would affect their natural acquisition of immunity, which is of specific importance for children who have been born and raised in the community within the last five years . In a prospective scenario, where the factors which have suppressed the Anopheline mosquito population will cease to play a role, immunologically naïve or partly naïve children and adults may thus have a significant risk of severe malaria related morbidity and death . In order to monitor and predict the risk of sudden epidemics it is, therefore, of outmost importance to identify the underlying causative factor or factors. The findings suggest that the pronounced decline in malaria mosquito vectors over the study periods is not a consequence of bed net use or indoor residual spraying. Although there may be an intriguing link to recent changes in rainfall patterns that requires further investigation, other potential explanations for the observed decline in vector populations in the two study communities should be explored, including an assessment of the potential role of changes in temperature, the ability for water to pool, agricultural activities, land use, vegetation, sibling species composition and impact of toxic substances and insect pathogens. It is likely that similar pronounced declines in malaria vector densities have contributed and still contribute to the decrease in levels of malaria infections and malaria disease seen in other geographical areas of SSA. Health professionals and researchers are strongly encouraged to initiate a systematic collection of the relevant mosquito vectors in order to monitor and assess the causative role of declining malaria vector populations in areas where decreasing levels of malaria infection and disease burden have been reported.